This article was published in the above mentioned Springer issue. The material, including all portions thereof, is protected by copyright; all rights are held exclusively by Springer Science + Business Media. The material is for personal use only; commercial use is not permitted. Unauthorized reproduction, transfer and/or use may be a violation of criminal as well as civil law. ISSN 0960-3115, Volume 19, Number 8
18
Embed
Intermediary disturbance increases tree diversity in riverine forest of southern Brazil
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
This article was published in the above mentioned Springer issueThe material including all portions thereof is protected by copyrightall rights are held exclusively by Springer Science + Business Media
The material is for personal use onlycommercial use is not permitted
Unauthorized reproduction transfer andor usemay be a violation of criminal as well as civil law
ISSN 0960-3115 Volume 19 Number 8
ORI GIN AL PA PER
Intermediary disturbance increases tree diversityin riverine forest of southern Brazil
Jean Carlos Budke bull Joao Andre Jarenkow bull
Ary Teixeira de Oliveira-Filho
Received 9 August 2009 Accepted 31 March 2010 Published online 16 April 2010 Springer Science+Business Media BV 2010
Abstract Floods are frequently associated with disturbance in structuring riverine forests
and they lead to environmental heterogeneity over space and time We evaluated the
distribution of tree species ecological groups species richness and diversity from the point
bar to the slope of a riverside forest in southern Brazil (Lat 30010S Long 52470W) to
analyze the effects of flooding on soil properties and forest structure A plot of 50 9 200 m
divided in five contiguous transects of 10 9 200 m parallel to the river was installed
where we measured all the individual trees with pbh C 15 cm A detailed topographical
and soil survey was carried out across the plot and indicated significant differences in
organic matter and most mineral nutrients through the topographical gradient The 1229
surveyed individuals belonged to 72 species and 35 families We used Partial CCA and
Species Indicator Analysis to observe the spatial distribution of species Both analyses
showed that species distribution was strongly related to the flooding gradient soil prop-
erties and also by space and pure spatial structuring of species and environmental variables
(spatial autocorrelation) although a large part of variation remains unexplained The
ecological groups of forest stratification plant dispersal and requirements for germination
indicated slight differences among frequently occasional and non-flooded transects
Species richness and diversity were higher at intermediate elevations and were associated
to the increased spatialndashtemporal environmental heterogeneity Across the plot the direct
Electronic supplementary material The online version of this article (doi101007s10531-010-9845-6)contains supplementary material which is available to authorized users
J C Budke (amp)Departamento de Ciencias Biologicas Universidade Regional Integrada do Alto Uruguai e dasMissoesmdashURI Campus de Erechim Av Sete de Setembro 1621 Erechim RS 99700-000 Brazile-mail jeanuriceredubr
J A JarenkowDepartamento de Botanica Universidade Federal do Rio Grande do Sul Av Bento Goncalves 9500Porto Alegre RS 91501-970 Brazil
A T de Oliveira-FilhoDepartamento de Botanica Universidade Federal de Minas Gerais Av Antonio Carlos 6627Belo Horizonte MG 31270-901 Brazil
influence of flooding on tree species distribution created a vegetation zonation that is
determined by predicted ecological traits
Keywords Disturbance Ecological groups Flooding regime Partial CCA Soil properties Species richness and diversity
Introduction
Natural disturbances play an important role in structuring plant communities by leading to
environmental heterogeneity over space and time at different scales Several studies have
demonstrated that disturbance and abiotic stress affect diversity especially at local-scale
(Ferreira and Stohlgren 1999 Weiher 2003) In riparian ecosystems flooding events are
the key factor in shaping community features either by a positive or a negative effect on
the ecosystemrsquos function according to the timing frequency and magnitude of such events
(Neiff 1990) Long-lasting floods represent a major stress and may result in species-poor
plant communities due to restricted productivity in the aquatic phase and high mortality of
non-adapted species (Pollock et al 1998 Guilherme et al 2004 Wittmann et al 2004) On
the other hand periodic and short floods may contribute to the input of nutrients which
increase productivity and diversity (Desilets and Houle 2005)
Once magnitude and duration of flooding are directly associated with local relief (eg
relative elevation inclination) many studies have investigated the relationships among
topography and correlated variables (eg chemical and textural soil properties sedimen-
tation rates) on the distribution of plant species and patterns of richness and diversity
(Oliveira-Filho et al 1994 Ferreira 2000 Rosales et al 2001 Damasceno-Junior et al
2005 Budke et al 2007)
In riparian systems with regular or predicted (seasonal) flood events as Amazonian and
Pantanal floodplains in South America plant species show different strategies to survive
floods including morphological anatomical and physiological adaptations and also phe-
nological timing for both reproductive and vegetative phases Ferreira et al (2009) has
demonstrated that species living in low-lying areas may be ecotypes originated from
surrounding non-flooded forests In contrast riverine forests with unpredictable flooding
pulses are frequently colonized by species of early successional stages of wide geo-
graphical distribution (Walker et al 1995 Budke et al 2007) On this hand Budke et al
(2008) observed that in low order rivers where water column oscillated due to a local and
concentrated rainy period the species richness increased along a gradient from frequently
to occasionally flooded stands Furthermore Robertson (2006) showed that predictability
of species occurrence across different rivers in the south-eastern US Coastal Plain was
directly related to the geomorphic dynamics intermediate level of stream energy (eg
flooding magnitude) and non-altered hydrological regimes
This ubiquitous difference of showing or not a temporally synchronous and expected
disturbance is one of the most interesting in eco-hydrological studies On this way the
structured view of the dynamic-equilibrium model (Huston 1994) shows different patches
from different seral stages result from spatial variation of disturbance frequencies If
disturbance frequencies vary over time a landscape could also contain such patches
(Pollock et al 1998) Indeed as expected in the intermediate disturbance model richness
would be higher when disturbance is neither too rare nor too frequent (Connell 1978)
In this work we focused on species richness and diversity of tree species in a riverside
sequence from the point bar to the lateral slope in a river with an unpredictable flooding
2372 Biodivers Conserv (2010) 192371ndash2387
123
Authors personal copy
regime The inundation regime also varies according to the topographical position fol-
lowing to frequently flooded forests to well drained non-flooded forests Furthermore we
investigated the relationships among tree component structure species and functional
groups distribution and spatialndashenvironmental variables We hypothesized that (1) as
flooding may gradually affect environmental heterogeneity richness and diversity will be
higher at intermediate elevations and directly associated with increased environmental
heterogeneity and (2) both species and functional groups will reflect variations in elevation
andor soil texture and chemistry
Methods
Study area
The study area is a forest remnant of ca 20 ha situated in the riparian fringes of the
Botucaraı river near its confluence with the Jacuı river (Lat 30010S Long 52470W)
(Budke et al 2007) The headwaters of the river lie in the southernmost extent (ca 650 m
asl) of the high planes region locally known as Planalto Meridional which geologically
is part of the Serra Geral formation made up of Cretaceous basalts originated from giant
lava flows that covered the sedimentary lowlands of the Parana Basin (Leinz 1949)
Downstream at its mid-course the Botucaraı river reaches the lowlands (ca 100 m asl)
and the topography is dominated by recently flood-deposited sedimentsmdashmeanders and
point bars At its lower course near the study area flooding events are enhanced by the
confluence with the stronger adjoining stream flow of the Jacuı river therefore promoting
lateral overflow According to Budke et al (2007) soils in such areas reflect not only the
geomorphic features from the basin but they also reflect flooding dynamics which fre-
quently produces non-stratified layers of fine gravel wood debris litter and sediment As a
consequence different soil profiles occur from well structured planosols in the riverside
slopes to recent deposited layers of sediment in the lowlands
The regional climate is moist subtropical without a regular dry season mean tem-
peratures ranges from 249C (hottest month) to 142C (coldest month) with high tem-
perature variation (absolutes values ranges from 42C in the summer to -3C in the
winter) mean annual rainfall is 1594 mm year-1 respectively (IPAGRO 1982) The
predominant soil is a Hydromorphic Planosol with typical stratified layers of depositional
sediments (Streck et al 2002)
Floods in the area are highly unpredictable because there is no marked seasonal rainy
period and rainfall is relatively well distributed throughout the year As a consequence
floods occur at any time of the year with duration of overflow periods varying from some
days to a few weeks (Budke et al 2008)
Regional vegetation is an extent of the Atlantic Forest Domain (Oliveira-Filho et al
2006) and includes overlapping patches of Seasonal Semideciduous Forests and Araucaria
Rain Forests at the river headwaters at Serra Geral formation Seasonal Semideciduous
Forests shows several genera of deciduous Fabaceae trees as Apuleia leiocarpa Myro-carpus frondosus Enterolobium contortisiliquun Parapiptadenia rigida and Erythrinafalcata as well as perennial ones which include Myrtaceae Lauraceae Sapotaceae and
Rubiaceae among others Canopy and emergent tree species can reach 25 m high
although mean vegetation stature is near 12ndash15 m In the lowlands of the river basin
Seasonal Semideciduous Forests is gradually changed by grasslands of the Pampa Domain
Biodivers Conserv (2010) 192371ndash2387 2373
123
Authors personal copy
(Oliveira-Filho et al 2006) and the river basin play a typical role of forest corridor toward
south reaching the Uruguay pampas as forest enclaves or galleries (Budke et al 2006)
Data collection
We carried out a tree survey in a 1 ha plot installed in a toposequence in the lowland areas
from the river margin to the lateral slope and therefore liable to different flooding regimes
The plot was divided in five 10 9 200 m transects and each transect was subdivided in
sampling units of 10 9 10 m All individual living trees having at least one stem and with
perimeter at breast height (pbh) C15 cm were sampled Voucher specimens of the different
species were collected prepared and lodged in the Herbarium ICN of the Universidade
Federal do Rio Grande do Sul (UFRGS)
A detailed topographic survey of the transects was carried out using a 10 m long water-
filled levelling hose 38 in a tape measure and a compass according to Cardoso and
Schiavini (2002) The resulting grid of vertical transects was used to produce contour maps
and to obtain the relative elevation of each sampling unit rather to the river To estimate
flooding frequency in each sampling unit we overlap their relative elevation to the
hydrometer records of the Jacuı river station (data calibrated according to topography)
Through Pulse 111 software (Neiff and Neiff 2003) we estimated the mean number of
floods per year from 1981 to 2004 and we used this variable as a pulse disturbance estimate
to sampling units (hereafter named flooding)
We collected samples of the topsoil (0ndash20 cm depth) from 15 sites distributed in dif-
ferent positions in such a way that its overall topographic variation was encompassed The
soil samples were kept in polyethylene bags and taken to the UFRGS Soil Laboratory for
chemical and textural analyses The variables were pH in water suspension levels of
potential acidity (Al H) bases saturation (V) sum of bases (S) cation exchange
capacity (CEC) organic matter (OM) and levels of clay sand and silt All procedures
followed EMBRAPA (1997) protocol In those plots without a soil subsample we
extrapolated real values by distance-proportional mean of the closest plots (ter Braak
1995) We compared the means of each soil property among transects by using one-way
ANOVA (Zar 1996)
Data analysis
Phytosociological parameters of density frequency and dominance (derived from tree
basal area) were calculated to describe tree community structure (Mueller-Dombois and
Ellenberg 1974) Frequency distributions into classes of diameter for each transect were
prepared and one-way ANOVA was used to compare transects Classes of exponentially
increasing range were used for diameters to make up for the accentuated decline in tree
frequency towards larger diameters (Oliveira-Filho et al 2001)
We applied rarefaction curves for each transect in order to analyse the range of species
richness within the toposequence The rarefaction curve technique generates expected
number of species based on the individualsrsquo density and then providing statistical
assumptions to this comparison (Gotelli and Colwell 2001) We also compared Shannon
diversity indices (H0) of each transect by bootstrap resampling tests with the software
Multiv (Pillar 2006) and depicted diversity and topography in a regression model
To verify topographical ranges of species we used an Indicator Species AnalysismdashISA
(Dufrene and Legendre 1997) which is a direct analysis of association between flooding
2374 Biodivers Conserv (2010) 192371ndash2387
123
Authors personal copy
and species distribution As the aim of this analysis was to assess the association between
species and topographyflooding it was used a non-hierarchical clustering procedure kmeans to produce k groups from the mean elevation of the original sampling units and by
using the resulting groups as the clustering factor required in the ISA (Dufrene and
Legendre 1997 Budke et al 2008) The analysis was performed in the PC-Ord program
(McCune and Mefford 1997)
We partitioned the variance of species distribution over the toposequence accounted by
spatial and environmental variables by successive partial Correspondence Canonical
Analysis (Borcard et al 1992) This approach combines three different matrices to
decompose all species variation in four components pure effect of environment pure
effect of spatial pattern combined variation of environment and spatial pattern and finally
unexplained variation Species assemblages from a determined position are affected by
surrounding sites because of contagious biotic process and environmental variables used to
describe biological processes are also neither randomly or uniformly spatially distributed
(Legendre 1993) In such case it is necessary to incorporate the spatial structure in the
modelling because the independence of observations is not respected (Legendre 1993) The
first matrix or species matrix included the abundances of all species with density C10
individuals The environmental matrix included initially all chemical and granulometric
figures the topographic variable (average elevation) and an ordinal (ranking) variable
labeled lsquolsquoflooding frequencyrsquorsquo We obtained the last variable directly from the topographic
survey summarizing flood occurrences and their intensity in each plot (Budke et al 2008)
The third matrix or spatial matrix included all terms of a polynomial function of geo-
graphical coordinates ie centers of each sampling unit and it was made by adding all
terms of a cubic trend surface regression
f x yeth THORN frac14 x y xy x2 y2 x2y xy2 x3 y3
According to Borcard et al (1992) this ensures the detection of more complex spatial
features as gaps or patches which require the quadratic and cubic terms of the coordinates
and their interactions
The variance partitioning proceeded in two steps First we extracted from each
explanatory matrix (environmental variables and spatial variables) all non-significant
variables by forward stepwise regression using Monte Carlo permutations (999 permuta-
tions P 005) with CANOCO 40 (ter Braack and Smilauer 1998) and performed two
canonical ordinations that are redundant in terms of explained variation over the species
data due to spatial structuring (Borcard et al 1992) Then two partial canonical analyses
were carried out (lsquoenvironmentalrsquo and lsquospatialrsquo) each of them constrained by one of the
sets of explanatory variables to determine the relative contribution of environmental and
spatial variables in accounting for species variation Final partition is possible by using the
sum of all canonical eigenvalues of two canonical ordinations constrained by one set of
explanatory variables and of two partial canonical ordinations each of them constrained
by one set of explanatory variables while controlling for the effect of the others (covari-
ables) (Borcard et al 1992 Titeux et al 2004)
To search for ecological differences in the toposequence we classified the species in
ecological groups of regeneration vertical distribution and dispersal We defined regen-
eration based on the categories proposed by Swaine and Whitmore (1988) The two main
levels are (a) lsquopioneerrsquo which includes the species showing an entirely heliophilous life
cycle a seed bank but no bank of juveniles and (b) lsquolate successional speciesrsquo which are
those able to germinate and establish under some degree of shade to form a bank of
Biodivers Conserv (2010) 192371ndash2387 2375
123
Authors personal copy
juveniles The later was divided into (b1) lsquoshade-tolerantrsquo and (b2) lsquolight-demanding late
successional speciesrsquo which are better seen as the two sides of a continuum of solar
radiation required by the trees to lsquoreleasersquo the bank of juveniles (Oliveira-Filho et al 1994)
We defined the vertical distribution based on the strata commonly reached by the adult
individuals (a) small tree species (b) medium tree species and (c) tall tree species (see
Oliveira-Filho et al 1994) The dispersal was (a) zoochorous species with animal-med-
iated dispersal syndrome (b) anemochorous and hydrochorous those with mechanisms to
facilitate wind-dispersal or flotation and (c) autochorous those dispersed by free fall or
ballistic mechanisms (Pijl 1982) The classification of each species into the ecological
groups was based on observations during fieldwork from 2004 to 2005 and on scientific
literature (Barroso et al 1999 Budke et al 2005 2008) We tested the distribution of trees
into frequency classes according to the ecological group by KruskalndashWallis tests (Zar
1996)
Results
River corridor along the studied area has a typical meandering system with well-defined
geomorphic features The lowest sector encompasses the levee and depression which
interacts directly with river floods Next to these sites we identified the lower-slope the
middle-slope and the ridge according to the relative elevation to the river channel
(Table 1) and these sectors corresponded to our installed transects The lower slope veg-
etation is a sharp transition between lowland and upland forests and only large inundation
floods this sector whereas upland sites present slight differences in vegetation structure
due to absence of flooding and allied effects Nevertheless there is a distinct gradient of
organic matter (OM) clay and cation exchange capacity (CEC) being higher toward upper
sites as also showed by potential acidity (Al H) (Table 1) By other hand sum of bases
(S) and phosphorus contents (P) showed a tendency of decreasing toward upper sites
(Table 1) Furthermore the variance of some soil variables was quite high and demon-
strated the high heterogeneity across transects
The field inventory yielded a total of 1229 individuals belonging to 72 species and 35
families from which Myrtaceae and Fabaceae were the richest families with 11 species
followed by Rubiaceae and Sapotaceae with four species (Table S1) Although Myrtaceae
and Fabaceae presented the highest richness both families appeared generally with low
density or basal area The stand showed a forest of low stature with most individuals
between 5 and 7 m tall and few emergent trees reaching up 15 m The diameter-class
distribution of trees revealed typical inverted-J distribution with most individuals situated
in the first two classes (Fig 1) Across the toposequence higher density was found near the
river (Levee) followed by lower density values in the depression and again an increased
density through lower and middle slope On the other hand the ridgetop transect presented
the lowest density but an increased basal area (Table 2) and several trees with diameter
[40 cm Vertical distribution of trees also showed the predominance of medium-sized
individuals followed by a decreased proportion of small and emergent trees (Fig 2A)
The proportion of light-demanding trees was higher towards the upper sites (Fig 2B)
Pioneer trees presented an opposite pattern being more abundant in low sites Shade-
tolerant trees also showed an increased density at upper sites where flooding is restrict or
absent Within the dispersal groups zoochorous trees presented higher proportion in all
transects Autochorous and hydrochorous trees decreased toward the ridgetop whereas
anemochorous trees followed the inverse pattern (Fig 2C) These structural patterns
2376 Biodivers Conserv (2010) 192371ndash2387
123
Authors personal copy
Tab
le1
So
ilv
aria
ble
so
ffi
ve
tran
sect
so
fri
ver
ine
fore
sto
nth
eB
otu
cara
ıri
ver
so
uth
ern
Bra
zil
Soil
var
iable
sL
evee
Dep
ress
ion
L-s
lope
M-s
lope
Rid
ge
FP
Rel
ativ
eel
evat
ion
(m)
38
plusmn0
5a
54
plusmn0
7a
85
plusmn2
3b
11
8plusmn
35
bc
13
6plusmn
37
c8
03
0
00
1
pH
(H2O
)4
8plusmn
04
47
plusmn0
44
7plusmn
04
47
plusmn0
64
7plusmn
06
08
50
93
ns
Pmdash
Meh
lich
(mg
dm
-3)
71
plusmn2
17
1plusmn
23
63
plusmn1
66
plusmn1
75
7plusmn
14
22
30
07
ns
K(m
gd
m-
3)
76
1plusmn
12
57
96
plusmn1
64
89
plusmn2
62
94
3plusmn
34
87
8plusmn
24
14
36
03
5n
s
Ca
(cm
olc
dm
-3)
62
plusmn4
36
8plusmn
44
7plusmn
47
61
plusmn4
57
plusmn3
73
19
05
2n
s
Mg
(cm
olc
dm
-3)
15
plusmn0
71
6plusmn
07
15
plusmn0
71
5plusmn
06
14
plusmn0
50
36
09
8n
s
Al
H
(cm
olc
dm
-3)
66
plusmn3
1a
71
plusmn3
ab8
5plusmn
4ab
96
plusmn5
5ab
10
plusmn4
2b
97
60
04
S(c
mo
lcd
m-
3)
8plusmn
48
87
plusmn5
18
1plusmn
46
84
plusmn5
17
4plusmn
41
27
70
59
ns
CE
C(c
mo
lcd
m-
3)
15
2plusmn
48
16
2plusmn
38
17
1plusmn
33
17
9plusmn
43
17
3plusmn
34
14
50
22
ns
V(
)5
57
plusmn1
71
49
7plusmn
20
24
59
plusmn2
46
45
5plusmn
23
64
18
plusmn2
07
40
60
39
ns
OM
()
26
plusmn1
1a
28
plusmn1
a3
2plusmn
1ab
37
plusmn1
1b
38
plusmn0
8b
22
9
00
01
Cla
y(
)1
37
plusmn2
4a
15
4plusmn
33
ab1
58
plusmn2
8ab
16
plusmn2
1b
15
5plusmn
17
ab2
49
00
4
San
d(
)2
0plusmn
74
23
1plusmn
22
24
2plusmn
10
12
23
plusmn7
52
23
plusmn6
28
91
00
6n
s
Sil
t(
)6
42
plusmn1
09
60
4plusmn
14
59
8plusmn
12
96
2plusmn
91
62
plusmn7
24
91
02
9n
s
Val
ues
are
mea
ns
plusmnst
and
ard
dev
iati
on
sfr
om
0to
20
cmd
epth
top
soil
sam
ple
s(N
=2
0fo
rea
chtr
anse
ct)
Dif
fere
nt
lett
ers
afte
rv
alu
esin
dic
ate
sign
ifica
nt
dif
fere
nce
sin
AN
OV
Ate
sts
(ns
=n
on
-sig
nifi
can
t)
Biodivers Conserv (2010) 192371ndash2387 2377
123
Authors personal copy
shaped the physiognomic features of different sectors that varied according to the topo-
sequence and consequently in flooding regime The depression sector presented lower
density basal area and also low tree diameters whereas the levee portion presented high
density and basal area
Species distribution across the topographic gradient is presented in Table 3 according
to the Indicator Species Analysis Some species were clearly distributed from lowland to
medium sites as Eugenia uniflora Myrciaria tenella Eugenia uruguayensis whereas
others were restricted to upland areas as Chomelia obtusa and Cordia americana Many
species did not show a specific site distribution and occurred over a wide distribution range
as Gymnanthes concolor and Casearia sylvestris
The relative elevation of each transect reflects the pattern of flooding frequency and
duration in each site then spatial aggregation of trees may indicate preferences or
restriction on the establishment of some species Typical riverine species appeared near the
river margin as Pouteria gardneriana Guettarda uruguensis and others (Table 3) whereas
typical species of well-drained forests as Sorocea bonplandii Parapiptadenia rigida and
Cupania vernalis occurred frequently in the ridgetop transect Furthermore 13 species did
Fig 1 Diameter-class distributions of trees with pbh C 15 cm surveyed in five transects of riverine foreston the Botucaraı river southern Brazil Diameter-classes are used for increasing intervals (see lsquolsquoMethodsrsquorsquosection) Bars and ranges are means and 95 confidence intervals of 100 sampling units respectively
Table 2 Density (ind ha-1) dominance (m2 ha-1) mean height (m) and mean diameter (cm) for differenttransects of the riverside forest of the Botucaraı river southern Brazil
Transect AD ADo Height Diameter
Levee 1655 plusmn 467 a 3927 plusmn 268 a 62 plusmn 25 a 1761 plusmn 1697
Depression 1005 plusmn 369 b 198 plusmn 137 b 67 plusmn 21 b 161 plusmn 1194
L-slope 1120 plusmn 443 ab 229 plusmn 165 b 71 plusmn 26 b 1547 plusmn 1302
M-slope 1415 plusmn 438 a 274 plusmn 165 b 69 plusmn 25 b 1453 plusmn 98
Ridge 950 plusmn 294 b 282 plusmn 194 b 7 plusmn 24 b 1727 plusmn 1555
ANOVA F = 107 F = 129 F = 305 F = 35
Different letters after values indicate significant differences in t tests ( P 005 P 0001)
AD density ADo dominance
2378 Biodivers Conserv (2010) 192371ndash2387
123
Authors personal copy
not present a topographic association due to wide distribution through the gradient On the
other hand the distinction among environmental and spatial effects showed that space
contributes significantly to the distribution of tree species (Fig 3) Environmental variables
selected by forward selection procedure (P 005) are summarized in Table 4 All geo-
graphical terms of the polynomial function were significant (P 005) during spatial CCA
and were add to the model The four CCA analyses provided the following results
1 CCA of the species matrix constrained by the environmental matrix sum of all
canonical eigenvalues = 0944 Monte Carlo tests for overall analysis F = 286
P 0001
2 CCA of the species matrix constrained by spatial matrix sum of all canonical
eigenvalues = 1017 Monte Carlo tests for overall analysis F = 359 P 0001
Fig 2 Ecological groups of vertical distribution (A) regeneration (B) and dispersal (C) in five transects ofriverine forest of Botucaraı river southern Brazil Pi pioneer Ld light-demanding St shade-tolerant Zoozoochorous Auto autochorous Ane anemochorous Hydro hydrochorous
Biodivers Conserv (2010) 192371ndash2387 2379
123
Authors personal copy
3 Environmental partial CCA (after removing the effect of geographical matrix) sum of
all canonical eigenvalues = 0416 Monte Carlo tests for overall analysis F = 135
P 0001
Table 3 Indicator species analysis (ISA) performed for species with density C10 individuals sampled infive transects with 20 sampling units each Botucaraı river southern Brazil
Biplots of environmental variables and species or environmental variables and sampling
units were depicted with the environmental partial analyses results (Fig 4) In this step
species distributions are purely based on speciesndashenvironment relationships The first two
axes accounted respectively for 268 (eigenvalue = 0112) and 181 (eigen-
value = 0075) of the speciesndashenvironment relationships and speciesndashenvironment corre-
lations for these axes were 0742 and 0649 (P 005) respectively The first four axes
accounted for 691 of speciesndashenvironment relationships Table 4 shows the intraset
correlations among variables and canonical coefficients with the first two axes The first
canonical axis was positively correlated with topography and several soil variables that are
Fig 3 Variation partitioning ofthe tree species matrix
Table 4 Intraset correlations among environmental variables selected for the model during lsquolsquoenviron-mentalrsquorsquo partial CCA of the species matrix and canonical coefficients of the first two axes
Environmental variables were selected by forward stepwise selection and included on the model if sig-nificant in Monte Carlo tests (P 005)
Biodivers Conserv (2010) 192371ndash2387 2381
123
Authors personal copy
influenced by flooding events Percentages of organic matter saturation of bases and cation
exchange capacity where higher through upper sites whereas aluminium contents were
higher in lower areas Thus this environmental gradient may affect tree distribution pat-
terns by restricting or facilitating species establishment As related in the Indicator
Analysis (although not accounting for spatial patterns directly) typical species of flooding
areas occurred near the levee and depression and the zonation was sharp once frequent
species of well-drained areas occurred only in the middle slope and ridgetop Sampling
units (Fig 5) also appeared distributed according to the toposequence with some over-
lapping due to species distribution
Rarefaction curves of species revealed significant differences on the expected total
number of species in each transect with higher richness within the middle slope (Fig 6)
A regression model (Fig 6) fitted diversity in a second-order polynomial regression
(y = -00586x2 03594x 291 R2 = 077) that showed the same pattern
Discussion
Environmental and spatial patterns
Tree species distribution throughout the topographical gradient indicated that both envi-
ronmental and spatial features were particularly important in predicting species and
community patterns This agrees with the well-know influence of geomorphic features and
hydrological regimes on riparian forests over different temporal and spatial scales
(Tabacchi et al 1998 Turner et al 2004 Desilets and Houle 2005) even though dis-
tinctions between environmental and spatial effects remain poorly studied (Titeux et al
Fig 4 Ordination biplotdepicting the two axes of theenvironmental partial CCA ofsampling units in a riverine forestin southern Brazil Eachsampling unit was identified bydifferent symbols according tothe respective transectEnvironmental variables arerepresented by their acronyms(see Table S1)
2382 Biodivers Conserv (2010) 192371ndash2387
123
Authors personal copy
2004) Here we employed a routine to partialling out the spatial effects on the analysis of
speciesndashenvironment relationships that also highlight the spatial component embedded in
such analysis (Legendre 1993) Once several biotic processes as growth mortality dis-
persal and predation influence the observed distributions of organisms (resulting in spatial
correlation) or if their distributions are dependent on explanatory variables which are
Fig 5 Ordination biplot depicting the two axes of the environmental partial CCA of species of a riverineforest in southern Brazil Species and environmental variables are represented by their acronyms (seeTable S1)
Fig 6 Rarefaction curves of tree species and Shannon diversity indices from five transects of riverine foreston the Botucaraı river southern Brazil Sampling units are representing different transects
Biodivers Conserv (2010) 192371ndash2387 2383
123
Authors personal copy
themselves spatially structured (Legendre 1993 Titeux et al 2004) spatial structuring is
an intrinsic component of ecosystems In our analysis lsquopurersquo spatial trends were more
attributed to species distribution than to lsquopurersquo environmental effects which link conta-
gious biological processes as important to the tree species distribution Furthermore
species and environmental data have a reasonable proportion of similar spatial structuring
identified by the largest proportion on the species variation due to spatially structured
environmental variation (1246) According to Borcard et al (1992) species and envi-
ronmental variables have in this case the same response to some common underlying
causes as the topographicndashflooding gradient In fact several studies have demonstrated the
direct effect of wetting and desiccation processes on both mineralogy and microbial
ecology of the sediment including nutrient dynamics (Baldwin and Mitchell 2000)
Once sediment or soils are submerged the inundation leads to a decrease in oxygen
contents and then resulting in progressive anaerobic conditions Rapid cycling of litter may
occur due to an increase on microbial activity which generates pulses on nutrient contents
and finally result in highly productive systems (Baldwin and Mitchell 2000) However a
negative effect is the rapid oxygen consumption which quickly leads to soil hypoxia or
anoxia When flood ends the anaerobic zones of sediments are newly oxygenated and
microbiota is replaced gradually to a new phase
Other spatially structured variables are sediment deposition and litter displacement
which are not covered in our study As demonstrated in lsquovarzearsquo forests of Amazonia
(Wittmann et al 2004) sediment deposition decrease toward upper sites and species
colonizing such lower areas show specific adaptations to the new site conditions regarding
to adventitious roots that probably offer mechanical support (Parolin et al 2004 Wittmann
et al 2004) However litter removal or deposition including seed bank may affect
directly species distribution once flooding and allied effects reallocate litter and seeds
among sites (Johansson et al 1996) Moreover studies have showed that flooding timing
frequency and magnitude can be used as indicators of sapling zonation on floodplain
forests (Vreugdenhil et al 2006)
All these processes are included in the unmeasured variables or spatially structuring
processes that have been missed by the geographical terms (Titeux et al 2004) and
accounted to the far unexplained variation (662) As also stressed by these authors the
stochastic spacendashtime fluctuations of each population the lsquounsaturationrsquo pattern (some
species do not use all suitable habitats) and species recording in not appropriated spatial
scales contribute to this unexplained variation too Notwithstanding occurrence data or
species abundances are often noisy (ter Braak 1995) and widespread in ecological studies
(Borcard et al 1992 Titeux et al 2004)
Richness and diversity patterns
Significant transitions occurred from the levee and depression to the following lower slope
transect regarding to stand structure and ecological groups Inversions on the proportion of
pioneershade-tolerant trees and auto-hydrochorous to anemochorous trees occurred in that
small transition and affected not only ecological groups but also species occurrences As a
consequence this zonation transect may consist spatially as a boundary for tolerant and
intolerant trees with regarding to flooding In fact few species occurred over the entire
flooding gradient and the lower slope also appears as an edge for several species
In a temporal scale the lower slope area will probably present more heterogeneous
spans in flooding events and it may consist in the most heterogeneous temporalndashspatial
sector across the topographic gradient which agree with the findings of Pollock et al
2384 Biodivers Conserv (2010) 192371ndash2387
123
Authors personal copy
(1998) from wetlands with different flooding regimes In the structured view of the
dynamic-equilibrium model (Huston 1994) different patches from different seral stages
result from spatial variation of disturbance frequencies If disturbance frequencies vary
over time a landscape could also contain patches of different seral stages (Pollock et al
1998) These authors used that assumption in a model with temporally synchronous dis-
turbance and found that at the community-scale level the results supported many pre-
dictions of the dynamic-equilibrium model especially regarding to species richness
In our study there are two major factors related to disturbance frequency First and
foremost is that once river floods are unpredictable due to a hydrological regime that varies
with occasional long-rainy periods temporal heterogeneity should be higher than in sea-
sonal predicted flood areas and second microtopography must create spatial heterogeneity
during floods events in the local area However well-drained upper sites are probably more
affected by inherent community processes as gap-phase dynamics and direct supply rates
of light (Stevens and Carson 2002) As related by Worbes et al (1992) and Parolin (2001)
hundreds of tree species with different phenological and other ecological traits grow in
seasonal flooded forests In such cases the cyclic alternation on floods and droughts drove
species to life history behavioral and morphological adaptations (Lytle and Poff 2004)
But in the case of unpredictable floods and droughts as assigned in our study bet-hedging
strategies might be evolved for example by persistent seed bank or asynchronous
reproductive phenologies (Brock 2003) although there are no conclusive studies related to
this theory (Lytle and Poff 2004)
Allowed by a transition in the ecological groups from the lower sites to the upper ones
species richness has a maximum at the lower slope transect probably due to higher het-
erogeneity in disturbance events (space and time) and correlated variables As reported by
Desilets and Houle (2005) the spatial gradient provides some evidences for stress toler-
ance and competition as factors structuring species distribution across the topographicndash
flooding gradient also boosted by an unpredictable pattern of floods that vary in frequency
timing and magnitude Lower sectors showed predicted ecological groups already
described for these areas (Budke et al 2007 2008 Junk et al 1989 Lytle and Poff 2004)
and expected structuring changes also occurred toward upper sites where the proportion of
shade-tolerant and small trees increased as well as anemochorous trees
In summary spatialndashtemporal and environmental variables are arranging tree species
distribution across the toposequence of our study site Furthermore predicted ecological
groups reflected the dynamics of disturbance in the topographicndashhydrological gradient
Species richness and diversity also reflected such pattern and were higher in the mid-sector
where occasional floods should prevent competitive exclusion and generate high envi-
ronmental heterogeneity
Acknowledgements We are grateful to the Programa de Pos-Graduacao em Botanica of the UniversidadeFederal do Rio Grande do SulmdashUFRGS for the opportunity to undertake this study and to CAPES Agencyfor the scholarship granted to the first author Our special thanks to Diogo lsquolsquoBagualrsquorsquo Lindenmaier forfieldwork assistance and to Ricardo Braga Eduardo Rossi and colleagues of the Laboratorio de Fitoeco-logiamdashUFRGS for critiques and suggestions We also appreciated the reviewing efforts of anonymouscontributors for providing useful comments to the manuscript
References
Baldwin DS Mitchell AM (2000) The effects of drying and re-flooding on the sediment and soil nutrientdynamics of lowland river-floodplain systems a synthesis Regul River 16457ndash467 doi1010021099-1646
Biodivers Conserv (2010) 192371ndash2387 2385
123
Authors personal copy
Barroso GM Morim MP Peixoto AL Ichaso CLF (1999) Frutos e sementes morfologia aplicada a si-stematica de dicotiledoneas Editora UFV Vicosa
Borcard D Legendre P Drapeau P (1992) Partialling out the spatial component of ecological variationEcology 731045ndash1055 doi1010292006WR005044
Brock MA (2003) Drought and aquatic community resilience the role of eggs and seeds in sediments oftemporary wetlands Freshw Biol 481207ndash1218 doi101046j1365-2427200301083x
Budke JC Athayde EA Giehl ELH Zachia RA Eisinger SM (2005) Composicao florıstica e estrategias dedispersao de especies lenhosas em uma floresta ribeirinha arroio Passo das Tropas Santa Maria RSBrasil Iheringia Bot 6017ndash24
Budke JC Jarenkow JA Oliveira-Filho AT Lindenmaier DS (2006) Padroes de riqueza e diversidade emrios de pequeno porte In Mariath JEA Santos RP (eds) Os avancos da botanica no inıcio do seculoXXI SBB Porto Alegre
Budke JC Jarenkow JA Oliveira-Filho AT (2007) Relationships between tree component structuretopography and soils of a riverine forest Rio Botucaraı southern Brazil Plant Ecol 189187ndash200 doi101007s11258-006-9174-8
Budke JC Jarenkow JA Oliveira-Filho AT (2008) Tree community features of two stands of riverine forestunder different flooding regimes in southern Brazil Flora 203162ndash174 doi101016jflora200703001
Cardoso E Schiavini I (2002) Relacao entre distribuicao de especies arboreas e topografia em um gradienteflorestal na Estacao Ecologica do Panga (Uberlandia MG) Rev Bras Bot 25277ndash289
Connell JH (1978) Diversity in tropical rain forests and coral reefs Science 1991302ndash1310 doi101126science19943351302
Damasceno-Junior GA Semir J Santos FAM Leitao-Filho HF (2005) Structure distribution of species andinundation in a riparian forest of Rio Paraguai Pantanal Brazil Flora 200119ndash135 doi101016jflora200409002
Desilets P Houle G (2005) Effects of resource availability and heterogeneity on the slope of the species-areacurve along a floodplain-upland gradient J Veg Sci 16487ndash496 doi1016581100-9233
Dufrene M Legendre P (1997) Species assemblages and indicator species the need for a flexible asym-metrical approach Ecol Monogr 67345ndash366
EMBRAPA (1997) Manual de metodos de analises de solo Empresa Brasileira de Pesquisa Agropecuariaand Centro Nacional de Pesquisas de Solos Rio de Janeiro
Ferreira LV (2000) Effects of flooding duration on species richness floristic composition and forest structurein river margin habitat in Amazonian blackwater floodplain forests implications for future design ofprotected areas Biodivers Conserv 91ndash14 doi101023A1008989811637
Ferreira LV Stohlgren TJ (1999) Effects of river level fluctuation on plant species richness diversity anddistribution in a floodplain forest in Central Amazonia Oecologia 120582ndash587 doi101007s004420050893
Ferreira C Piedade MTF Franco AC Goncalves JFC Junk WJ (2009) Adaptive strategies to tolerateprolonged flooding in seedlings of floodplain and upland populations of Himatanthus sucuuba aCentral Amazon tree Aquat Bot 90246ndash252 doi101016jaquabot200810006
Gotelli NJ Colwell RK (2001) Quantifying biodiversity procedures and pitfalls in the measurement andcomparison of species richness Ecol Lett 4379ndash391 doi101046j1461-0248200100230x
Guilherme FAG Oliveira-Filho AT Appolinario V Bearzoti E (2004) Effects of flooding regime andwoody bamboos on tree community dynamics in a section of tropical semideciduous forest in south-eastern Brazil Plant Ecol 17419ndash36 doi101023BVEGE000004605197752cd
Huston M (1994) Biological diversity the coexistence of species in changing landscapes CambridgeUniversity Press Cambridge
IPAGRO (1982) Atlas agroclimatico do Rio Grande do Sul Pallotti Porto AlegreJohansson ME Nilsson C Nilsson E (1996) Do rivers function as corridors for plant dispersal J Veg Sci
7593ndash598Junk WJ Bayley PB Sparks RE (1989) The flood pulse concept in river-floodplain systems Can J Fish
Aquat Sci 106110ndash127Legendre P (1993) Spatial autocorrelationmdashtrouble or new paradigm Ecology 741659ndash1673Leinz V (1949) Contribuicao a geologia dos derrames basalticos do Rio Grande do Sul Bol Fac Filos Let
581ndash55Lytle DA Poff NL (2004) Adaptation to natural flow regimes Trends Ecol Evol 1994ndash100 doi
101016jtree200310002McCune B Mefford MJ (1997) PCndashORD Multivariate analysis of ecological data version 436 MjM
Software Design Glaneden BeachMueller-Dombois D Ellenberg H (1974) Aims and methods of vegetation ecology John Wiley New York
2386 Biodivers Conserv (2010) 192371ndash2387
123
Authors personal copy
Neiff JJ (1990) Ideas para la interpretacion ecologica del Parana Interciencia 15424ndash441Neiff JJ Neiff M (2003) Pulso software para analisis de fenomenos recurrentes httpwwwneiffcom
Cited 25 May 2009Oliveira-Filho AT Vilela EA Gavilanes ML Carvalho DA (1994) Effect of flooding regime and understory
bamboos on the physiognomy and tree species composition of a tropical semideciduous forest in south-eastern Brazil Vegetatio 11399ndash124
Oliveira-Filho AT Curi N Vilela EA Carvalho DA (2001) Variation in tree community composition andstructure with changes in soil properties within a fragment of semideciduous forest in south-easternBrazil Edinb J Bot 58139ndash158 doi101017S0960428601000506
Oliveira-Filho AT Jarenkow JA Rodal MJN (2006) Floristic relationships of seasonally dry forests ofeastern South America based on tree species distribution patterns In Pennington RT Ratter JA LewisGP (eds) Neotropical savannas and dry forests plant diversity biogeography and conservation CRCPress Boca Raton
Parolin P (2001) Morphological and physiological adjustments to waterlogging and drought in seedlings ofAmazonian floodplain trees Oecologia 128326ndash335 doi101007s004420100660
Parolin P de Simone O Haase K Waldhoff D Rottenberger S Kuhn U Kesselmeier J Kleiss B SchmidtW Piedade MTF Junk WJ (2004) Central Amazonian floodplain forests tree adaptations in a pulsingsystem Bot Rev 70357ndash380 doi1016630006-8101(2004)070[0357CAFFTA]20CO2
Pijl L (1982) Principles of dispersal in higher plants Springer New YorkPillar VD (2006) Multivariate exploratory analysis randomization testing and bootstrap resampling version
2320 Departamento de Ecologia UFRGS Porto AlegrePollock MM Naiman RJ Hanley TA (1998) Plant species richness in riparian wetlandsmdasha test of biodi-
versity theory Ecology 7994ndash105Robertson KM (2006) Distributions of tree species along point bars of 10 rivers in the south-eastern US
Coastal Plain J Biogeogr 33121ndash132 doi101111j1365-2699200501371xRosales J Petts G Knab-Vispo C (2001) Ecological gradients within the riparian forests of the lower Caura
river Venezuela Plant Ecol 152101ndash118 doi101023A1011411020040Stevens MHH Carson WP (2002) Resource quantity not resource heterogeneity maintains plant diversity
Ecol Lett 5420ndash426 doi101046j1461-0248200200333xStreck EV Kampf N Dalmolin RSD Klamt E Nascimento PC Schneider P (2002) Solos do Rio Grande do
Sul EMATERRS and UFRGS Porto AlegreSwaine MD Whitmore TC (1988) On the definition of ecological species groups in tropical rain forests
Vegetatio 7581ndash86Tabacchi E Correll DL Hauer R Pinay G Planty-Tabacchi AM Wissmar R (1998) Development
maintenance and role of riparian vegetation in the river landscape Freshw Biol 40497ndash516 doi101046j1365-2427199800381x
ter Braack CJF Smilauer P (1998) Canoco reference manual and userrsquos guide to Canoco for Windowssoftware for canonical community ordination (version 40) Microcomputer Power Ithaca
ter Braak CJF (1995) Ordination In Jongman RHG ter Braak CJF van Togeren OFR (eds) Data analysis incommunity and landscape ecology Cambridge University Press New York
Titeux N Dufrene M Jacob JP Paquay M Defourny P (2004) Multivariate analysis of fine-scale breedingbird atlas using a geographical information system and partial canonical correspondence analysisenvironmental and spatial effects J Biogeogr 311841ndash1856 doi101111j1365-2699200401125x
Turner MG Gergel SE Dixon MD Miller JR (2004) Distribution and abundance of trees in floodplainforests of the Wisconsin river environmental influences at different scales J Veg Sci 15729ndash738
Vreugdenhil SJ Kramer K Pelsma T (2006) Effects of flooding duration frequency and depth on thepresence of saplings of six woody species in north-west Europe For Ecol Manage 23647ndash55 doi101016jforeco200608329
Walker KF Sheldon F Puckridge JT (1995) A perspective on dryland river ecosystems Regul River 1185ndash104 doi101002rrr3450110108
Weiher E (2003) Species richness along multiple gradients testing a general multivariate model in oaksavannas Oikos 101311ndash316 doi101034j1600-0706200312216x
Wittmann F Junk WJ Piedade MTF (2004) The varzea forests in Amazonia flooding and the highlydynamic geomorphology interact with natural forest succession For Ecol Manage 196199ndash212 doi101016jforeco200402060
Worbes M Klinge H Revilla JD Martius C (1992) On the dynamics floristic subdivision and geographicaldistribution of Varzea forests in Central Amazonia J Veg Sci 3553ndash564
Zar JH (1996) Biostatistical analysis Prentice-Hall New Jersey
Biodivers Conserv (2010) 192371ndash2387 2387
123
Authors personal copy
ORI GIN AL PA PER
Intermediary disturbance increases tree diversityin riverine forest of southern Brazil
Jean Carlos Budke bull Joao Andre Jarenkow bull
Ary Teixeira de Oliveira-Filho
Received 9 August 2009 Accepted 31 March 2010 Published online 16 April 2010 Springer Science+Business Media BV 2010
Abstract Floods are frequently associated with disturbance in structuring riverine forests
and they lead to environmental heterogeneity over space and time We evaluated the
distribution of tree species ecological groups species richness and diversity from the point
bar to the slope of a riverside forest in southern Brazil (Lat 30010S Long 52470W) to
analyze the effects of flooding on soil properties and forest structure A plot of 50 9 200 m
divided in five contiguous transects of 10 9 200 m parallel to the river was installed
where we measured all the individual trees with pbh C 15 cm A detailed topographical
and soil survey was carried out across the plot and indicated significant differences in
organic matter and most mineral nutrients through the topographical gradient The 1229
surveyed individuals belonged to 72 species and 35 families We used Partial CCA and
Species Indicator Analysis to observe the spatial distribution of species Both analyses
showed that species distribution was strongly related to the flooding gradient soil prop-
erties and also by space and pure spatial structuring of species and environmental variables
(spatial autocorrelation) although a large part of variation remains unexplained The
ecological groups of forest stratification plant dispersal and requirements for germination
indicated slight differences among frequently occasional and non-flooded transects
Species richness and diversity were higher at intermediate elevations and were associated
to the increased spatialndashtemporal environmental heterogeneity Across the plot the direct
Electronic supplementary material The online version of this article (doi101007s10531-010-9845-6)contains supplementary material which is available to authorized users
J C Budke (amp)Departamento de Ciencias Biologicas Universidade Regional Integrada do Alto Uruguai e dasMissoesmdashURI Campus de Erechim Av Sete de Setembro 1621 Erechim RS 99700-000 Brazile-mail jeanuriceredubr
J A JarenkowDepartamento de Botanica Universidade Federal do Rio Grande do Sul Av Bento Goncalves 9500Porto Alegre RS 91501-970 Brazil
A T de Oliveira-FilhoDepartamento de Botanica Universidade Federal de Minas Gerais Av Antonio Carlos 6627Belo Horizonte MG 31270-901 Brazil
influence of flooding on tree species distribution created a vegetation zonation that is
determined by predicted ecological traits
Keywords Disturbance Ecological groups Flooding regime Partial CCA Soil properties Species richness and diversity
Introduction
Natural disturbances play an important role in structuring plant communities by leading to
environmental heterogeneity over space and time at different scales Several studies have
demonstrated that disturbance and abiotic stress affect diversity especially at local-scale
(Ferreira and Stohlgren 1999 Weiher 2003) In riparian ecosystems flooding events are
the key factor in shaping community features either by a positive or a negative effect on
the ecosystemrsquos function according to the timing frequency and magnitude of such events
(Neiff 1990) Long-lasting floods represent a major stress and may result in species-poor
plant communities due to restricted productivity in the aquatic phase and high mortality of
non-adapted species (Pollock et al 1998 Guilherme et al 2004 Wittmann et al 2004) On
the other hand periodic and short floods may contribute to the input of nutrients which
increase productivity and diversity (Desilets and Houle 2005)
Once magnitude and duration of flooding are directly associated with local relief (eg
relative elevation inclination) many studies have investigated the relationships among
topography and correlated variables (eg chemical and textural soil properties sedimen-
tation rates) on the distribution of plant species and patterns of richness and diversity
(Oliveira-Filho et al 1994 Ferreira 2000 Rosales et al 2001 Damasceno-Junior et al
2005 Budke et al 2007)
In riparian systems with regular or predicted (seasonal) flood events as Amazonian and
Pantanal floodplains in South America plant species show different strategies to survive
floods including morphological anatomical and physiological adaptations and also phe-
nological timing for both reproductive and vegetative phases Ferreira et al (2009) has
demonstrated that species living in low-lying areas may be ecotypes originated from
surrounding non-flooded forests In contrast riverine forests with unpredictable flooding
pulses are frequently colonized by species of early successional stages of wide geo-
graphical distribution (Walker et al 1995 Budke et al 2007) On this hand Budke et al
(2008) observed that in low order rivers where water column oscillated due to a local and
concentrated rainy period the species richness increased along a gradient from frequently
to occasionally flooded stands Furthermore Robertson (2006) showed that predictability
of species occurrence across different rivers in the south-eastern US Coastal Plain was
directly related to the geomorphic dynamics intermediate level of stream energy (eg
flooding magnitude) and non-altered hydrological regimes
This ubiquitous difference of showing or not a temporally synchronous and expected
disturbance is one of the most interesting in eco-hydrological studies On this way the
structured view of the dynamic-equilibrium model (Huston 1994) shows different patches
from different seral stages result from spatial variation of disturbance frequencies If
disturbance frequencies vary over time a landscape could also contain such patches
(Pollock et al 1998) Indeed as expected in the intermediate disturbance model richness
would be higher when disturbance is neither too rare nor too frequent (Connell 1978)
In this work we focused on species richness and diversity of tree species in a riverside
sequence from the point bar to the lateral slope in a river with an unpredictable flooding
2372 Biodivers Conserv (2010) 192371ndash2387
123
Authors personal copy
regime The inundation regime also varies according to the topographical position fol-
lowing to frequently flooded forests to well drained non-flooded forests Furthermore we
investigated the relationships among tree component structure species and functional
groups distribution and spatialndashenvironmental variables We hypothesized that (1) as
flooding may gradually affect environmental heterogeneity richness and diversity will be
higher at intermediate elevations and directly associated with increased environmental
heterogeneity and (2) both species and functional groups will reflect variations in elevation
andor soil texture and chemistry
Methods
Study area
The study area is a forest remnant of ca 20 ha situated in the riparian fringes of the
Botucaraı river near its confluence with the Jacuı river (Lat 30010S Long 52470W)
(Budke et al 2007) The headwaters of the river lie in the southernmost extent (ca 650 m
asl) of the high planes region locally known as Planalto Meridional which geologically
is part of the Serra Geral formation made up of Cretaceous basalts originated from giant
lava flows that covered the sedimentary lowlands of the Parana Basin (Leinz 1949)
Downstream at its mid-course the Botucaraı river reaches the lowlands (ca 100 m asl)
and the topography is dominated by recently flood-deposited sedimentsmdashmeanders and
point bars At its lower course near the study area flooding events are enhanced by the
confluence with the stronger adjoining stream flow of the Jacuı river therefore promoting
lateral overflow According to Budke et al (2007) soils in such areas reflect not only the
geomorphic features from the basin but they also reflect flooding dynamics which fre-
quently produces non-stratified layers of fine gravel wood debris litter and sediment As a
consequence different soil profiles occur from well structured planosols in the riverside
slopes to recent deposited layers of sediment in the lowlands
The regional climate is moist subtropical without a regular dry season mean tem-
peratures ranges from 249C (hottest month) to 142C (coldest month) with high tem-
perature variation (absolutes values ranges from 42C in the summer to -3C in the
winter) mean annual rainfall is 1594 mm year-1 respectively (IPAGRO 1982) The
predominant soil is a Hydromorphic Planosol with typical stratified layers of depositional
sediments (Streck et al 2002)
Floods in the area are highly unpredictable because there is no marked seasonal rainy
period and rainfall is relatively well distributed throughout the year As a consequence
floods occur at any time of the year with duration of overflow periods varying from some
days to a few weeks (Budke et al 2008)
Regional vegetation is an extent of the Atlantic Forest Domain (Oliveira-Filho et al
2006) and includes overlapping patches of Seasonal Semideciduous Forests and Araucaria
Rain Forests at the river headwaters at Serra Geral formation Seasonal Semideciduous
Forests shows several genera of deciduous Fabaceae trees as Apuleia leiocarpa Myro-carpus frondosus Enterolobium contortisiliquun Parapiptadenia rigida and Erythrinafalcata as well as perennial ones which include Myrtaceae Lauraceae Sapotaceae and
Rubiaceae among others Canopy and emergent tree species can reach 25 m high
although mean vegetation stature is near 12ndash15 m In the lowlands of the river basin
Seasonal Semideciduous Forests is gradually changed by grasslands of the Pampa Domain
Biodivers Conserv (2010) 192371ndash2387 2373
123
Authors personal copy
(Oliveira-Filho et al 2006) and the river basin play a typical role of forest corridor toward
south reaching the Uruguay pampas as forest enclaves or galleries (Budke et al 2006)
Data collection
We carried out a tree survey in a 1 ha plot installed in a toposequence in the lowland areas
from the river margin to the lateral slope and therefore liable to different flooding regimes
The plot was divided in five 10 9 200 m transects and each transect was subdivided in
sampling units of 10 9 10 m All individual living trees having at least one stem and with
perimeter at breast height (pbh) C15 cm were sampled Voucher specimens of the different
species were collected prepared and lodged in the Herbarium ICN of the Universidade
Federal do Rio Grande do Sul (UFRGS)
A detailed topographic survey of the transects was carried out using a 10 m long water-
filled levelling hose 38 in a tape measure and a compass according to Cardoso and
Schiavini (2002) The resulting grid of vertical transects was used to produce contour maps
and to obtain the relative elevation of each sampling unit rather to the river To estimate
flooding frequency in each sampling unit we overlap their relative elevation to the
hydrometer records of the Jacuı river station (data calibrated according to topography)
Through Pulse 111 software (Neiff and Neiff 2003) we estimated the mean number of
floods per year from 1981 to 2004 and we used this variable as a pulse disturbance estimate
to sampling units (hereafter named flooding)
We collected samples of the topsoil (0ndash20 cm depth) from 15 sites distributed in dif-
ferent positions in such a way that its overall topographic variation was encompassed The
soil samples were kept in polyethylene bags and taken to the UFRGS Soil Laboratory for
chemical and textural analyses The variables were pH in water suspension levels of
potential acidity (Al H) bases saturation (V) sum of bases (S) cation exchange
capacity (CEC) organic matter (OM) and levels of clay sand and silt All procedures
followed EMBRAPA (1997) protocol In those plots without a soil subsample we
extrapolated real values by distance-proportional mean of the closest plots (ter Braak
1995) We compared the means of each soil property among transects by using one-way
ANOVA (Zar 1996)
Data analysis
Phytosociological parameters of density frequency and dominance (derived from tree
basal area) were calculated to describe tree community structure (Mueller-Dombois and
Ellenberg 1974) Frequency distributions into classes of diameter for each transect were
prepared and one-way ANOVA was used to compare transects Classes of exponentially
increasing range were used for diameters to make up for the accentuated decline in tree
frequency towards larger diameters (Oliveira-Filho et al 2001)
We applied rarefaction curves for each transect in order to analyse the range of species
richness within the toposequence The rarefaction curve technique generates expected
number of species based on the individualsrsquo density and then providing statistical
assumptions to this comparison (Gotelli and Colwell 2001) We also compared Shannon
diversity indices (H0) of each transect by bootstrap resampling tests with the software
Multiv (Pillar 2006) and depicted diversity and topography in a regression model
To verify topographical ranges of species we used an Indicator Species AnalysismdashISA
(Dufrene and Legendre 1997) which is a direct analysis of association between flooding
2374 Biodivers Conserv (2010) 192371ndash2387
123
Authors personal copy
and species distribution As the aim of this analysis was to assess the association between
species and topographyflooding it was used a non-hierarchical clustering procedure kmeans to produce k groups from the mean elevation of the original sampling units and by
using the resulting groups as the clustering factor required in the ISA (Dufrene and
Legendre 1997 Budke et al 2008) The analysis was performed in the PC-Ord program
(McCune and Mefford 1997)
We partitioned the variance of species distribution over the toposequence accounted by
spatial and environmental variables by successive partial Correspondence Canonical
Analysis (Borcard et al 1992) This approach combines three different matrices to
decompose all species variation in four components pure effect of environment pure
effect of spatial pattern combined variation of environment and spatial pattern and finally
unexplained variation Species assemblages from a determined position are affected by
surrounding sites because of contagious biotic process and environmental variables used to
describe biological processes are also neither randomly or uniformly spatially distributed
(Legendre 1993) In such case it is necessary to incorporate the spatial structure in the
modelling because the independence of observations is not respected (Legendre 1993) The
first matrix or species matrix included the abundances of all species with density C10
individuals The environmental matrix included initially all chemical and granulometric
figures the topographic variable (average elevation) and an ordinal (ranking) variable
labeled lsquolsquoflooding frequencyrsquorsquo We obtained the last variable directly from the topographic
survey summarizing flood occurrences and their intensity in each plot (Budke et al 2008)
The third matrix or spatial matrix included all terms of a polynomial function of geo-
graphical coordinates ie centers of each sampling unit and it was made by adding all
terms of a cubic trend surface regression
f x yeth THORN frac14 x y xy x2 y2 x2y xy2 x3 y3
According to Borcard et al (1992) this ensures the detection of more complex spatial
features as gaps or patches which require the quadratic and cubic terms of the coordinates
and their interactions
The variance partitioning proceeded in two steps First we extracted from each
explanatory matrix (environmental variables and spatial variables) all non-significant
variables by forward stepwise regression using Monte Carlo permutations (999 permuta-
tions P 005) with CANOCO 40 (ter Braack and Smilauer 1998) and performed two
canonical ordinations that are redundant in terms of explained variation over the species
data due to spatial structuring (Borcard et al 1992) Then two partial canonical analyses
were carried out (lsquoenvironmentalrsquo and lsquospatialrsquo) each of them constrained by one of the
sets of explanatory variables to determine the relative contribution of environmental and
spatial variables in accounting for species variation Final partition is possible by using the
sum of all canonical eigenvalues of two canonical ordinations constrained by one set of
explanatory variables and of two partial canonical ordinations each of them constrained
by one set of explanatory variables while controlling for the effect of the others (covari-
ables) (Borcard et al 1992 Titeux et al 2004)
To search for ecological differences in the toposequence we classified the species in
ecological groups of regeneration vertical distribution and dispersal We defined regen-
eration based on the categories proposed by Swaine and Whitmore (1988) The two main
levels are (a) lsquopioneerrsquo which includes the species showing an entirely heliophilous life
cycle a seed bank but no bank of juveniles and (b) lsquolate successional speciesrsquo which are
those able to germinate and establish under some degree of shade to form a bank of
Biodivers Conserv (2010) 192371ndash2387 2375
123
Authors personal copy
juveniles The later was divided into (b1) lsquoshade-tolerantrsquo and (b2) lsquolight-demanding late
successional speciesrsquo which are better seen as the two sides of a continuum of solar
radiation required by the trees to lsquoreleasersquo the bank of juveniles (Oliveira-Filho et al 1994)
We defined the vertical distribution based on the strata commonly reached by the adult
individuals (a) small tree species (b) medium tree species and (c) tall tree species (see
Oliveira-Filho et al 1994) The dispersal was (a) zoochorous species with animal-med-
iated dispersal syndrome (b) anemochorous and hydrochorous those with mechanisms to
facilitate wind-dispersal or flotation and (c) autochorous those dispersed by free fall or
ballistic mechanisms (Pijl 1982) The classification of each species into the ecological
groups was based on observations during fieldwork from 2004 to 2005 and on scientific
literature (Barroso et al 1999 Budke et al 2005 2008) We tested the distribution of trees
into frequency classes according to the ecological group by KruskalndashWallis tests (Zar
1996)
Results
River corridor along the studied area has a typical meandering system with well-defined
geomorphic features The lowest sector encompasses the levee and depression which
interacts directly with river floods Next to these sites we identified the lower-slope the
middle-slope and the ridge according to the relative elevation to the river channel
(Table 1) and these sectors corresponded to our installed transects The lower slope veg-
etation is a sharp transition between lowland and upland forests and only large inundation
floods this sector whereas upland sites present slight differences in vegetation structure
due to absence of flooding and allied effects Nevertheless there is a distinct gradient of
organic matter (OM) clay and cation exchange capacity (CEC) being higher toward upper
sites as also showed by potential acidity (Al H) (Table 1) By other hand sum of bases
(S) and phosphorus contents (P) showed a tendency of decreasing toward upper sites
(Table 1) Furthermore the variance of some soil variables was quite high and demon-
strated the high heterogeneity across transects
The field inventory yielded a total of 1229 individuals belonging to 72 species and 35
families from which Myrtaceae and Fabaceae were the richest families with 11 species
followed by Rubiaceae and Sapotaceae with four species (Table S1) Although Myrtaceae
and Fabaceae presented the highest richness both families appeared generally with low
density or basal area The stand showed a forest of low stature with most individuals
between 5 and 7 m tall and few emergent trees reaching up 15 m The diameter-class
distribution of trees revealed typical inverted-J distribution with most individuals situated
in the first two classes (Fig 1) Across the toposequence higher density was found near the
river (Levee) followed by lower density values in the depression and again an increased
density through lower and middle slope On the other hand the ridgetop transect presented
the lowest density but an increased basal area (Table 2) and several trees with diameter
[40 cm Vertical distribution of trees also showed the predominance of medium-sized
individuals followed by a decreased proportion of small and emergent trees (Fig 2A)
The proportion of light-demanding trees was higher towards the upper sites (Fig 2B)
Pioneer trees presented an opposite pattern being more abundant in low sites Shade-
tolerant trees also showed an increased density at upper sites where flooding is restrict or
absent Within the dispersal groups zoochorous trees presented higher proportion in all
transects Autochorous and hydrochorous trees decreased toward the ridgetop whereas
anemochorous trees followed the inverse pattern (Fig 2C) These structural patterns
2376 Biodivers Conserv (2010) 192371ndash2387
123
Authors personal copy
Tab
le1
So
ilv
aria
ble
so
ffi
ve
tran
sect
so
fri
ver
ine
fore
sto
nth
eB
otu
cara
ıri
ver
so
uth
ern
Bra
zil
Soil
var
iable
sL
evee
Dep
ress
ion
L-s
lope
M-s
lope
Rid
ge
FP
Rel
ativ
eel
evat
ion
(m)
38
plusmn0
5a
54
plusmn0
7a
85
plusmn2
3b
11
8plusmn
35
bc
13
6plusmn
37
c8
03
0
00
1
pH
(H2O
)4
8plusmn
04
47
plusmn0
44
7plusmn
04
47
plusmn0
64
7plusmn
06
08
50
93
ns
Pmdash
Meh
lich
(mg
dm
-3)
71
plusmn2
17
1plusmn
23
63
plusmn1
66
plusmn1
75
7plusmn
14
22
30
07
ns
K(m
gd
m-
3)
76
1plusmn
12
57
96
plusmn1
64
89
plusmn2
62
94
3plusmn
34
87
8plusmn
24
14
36
03
5n
s
Ca
(cm
olc
dm
-3)
62
plusmn4
36
8plusmn
44
7plusmn
47
61
plusmn4
57
plusmn3
73
19
05
2n
s
Mg
(cm
olc
dm
-3)
15
plusmn0
71
6plusmn
07
15
plusmn0
71
5plusmn
06
14
plusmn0
50
36
09
8n
s
Al
H
(cm
olc
dm
-3)
66
plusmn3
1a
71
plusmn3
ab8
5plusmn
4ab
96
plusmn5
5ab
10
plusmn4
2b
97
60
04
S(c
mo
lcd
m-
3)
8plusmn
48
87
plusmn5
18
1plusmn
46
84
plusmn5
17
4plusmn
41
27
70
59
ns
CE
C(c
mo
lcd
m-
3)
15
2plusmn
48
16
2plusmn
38
17
1plusmn
33
17
9plusmn
43
17
3plusmn
34
14
50
22
ns
V(
)5
57
plusmn1
71
49
7plusmn
20
24
59
plusmn2
46
45
5plusmn
23
64
18
plusmn2
07
40
60
39
ns
OM
()
26
plusmn1
1a
28
plusmn1
a3
2plusmn
1ab
37
plusmn1
1b
38
plusmn0
8b
22
9
00
01
Cla
y(
)1
37
plusmn2
4a
15
4plusmn
33
ab1
58
plusmn2
8ab
16
plusmn2
1b
15
5plusmn
17
ab2
49
00
4
San
d(
)2
0plusmn
74
23
1plusmn
22
24
2plusmn
10
12
23
plusmn7
52
23
plusmn6
28
91
00
6n
s
Sil
t(
)6
42
plusmn1
09
60
4plusmn
14
59
8plusmn
12
96
2plusmn
91
62
plusmn7
24
91
02
9n
s
Val
ues
are
mea
ns
plusmnst
and
ard
dev
iati
on
sfr
om
0to
20
cmd
epth
top
soil
sam
ple
s(N
=2
0fo
rea
chtr
anse
ct)
Dif
fere
nt
lett
ers
afte
rv
alu
esin
dic
ate
sign
ifica
nt
dif
fere
nce
sin
AN
OV
Ate
sts
(ns
=n
on
-sig
nifi
can
t)
Biodivers Conserv (2010) 192371ndash2387 2377
123
Authors personal copy
shaped the physiognomic features of different sectors that varied according to the topo-
sequence and consequently in flooding regime The depression sector presented lower
density basal area and also low tree diameters whereas the levee portion presented high
density and basal area
Species distribution across the topographic gradient is presented in Table 3 according
to the Indicator Species Analysis Some species were clearly distributed from lowland to
medium sites as Eugenia uniflora Myrciaria tenella Eugenia uruguayensis whereas
others were restricted to upland areas as Chomelia obtusa and Cordia americana Many
species did not show a specific site distribution and occurred over a wide distribution range
as Gymnanthes concolor and Casearia sylvestris
The relative elevation of each transect reflects the pattern of flooding frequency and
duration in each site then spatial aggregation of trees may indicate preferences or
restriction on the establishment of some species Typical riverine species appeared near the
river margin as Pouteria gardneriana Guettarda uruguensis and others (Table 3) whereas
typical species of well-drained forests as Sorocea bonplandii Parapiptadenia rigida and
Cupania vernalis occurred frequently in the ridgetop transect Furthermore 13 species did
Fig 1 Diameter-class distributions of trees with pbh C 15 cm surveyed in five transects of riverine foreston the Botucaraı river southern Brazil Diameter-classes are used for increasing intervals (see lsquolsquoMethodsrsquorsquosection) Bars and ranges are means and 95 confidence intervals of 100 sampling units respectively
Table 2 Density (ind ha-1) dominance (m2 ha-1) mean height (m) and mean diameter (cm) for differenttransects of the riverside forest of the Botucaraı river southern Brazil
Transect AD ADo Height Diameter
Levee 1655 plusmn 467 a 3927 plusmn 268 a 62 plusmn 25 a 1761 plusmn 1697
Depression 1005 plusmn 369 b 198 plusmn 137 b 67 plusmn 21 b 161 plusmn 1194
L-slope 1120 plusmn 443 ab 229 plusmn 165 b 71 plusmn 26 b 1547 plusmn 1302
M-slope 1415 plusmn 438 a 274 plusmn 165 b 69 plusmn 25 b 1453 plusmn 98
Ridge 950 plusmn 294 b 282 plusmn 194 b 7 plusmn 24 b 1727 plusmn 1555
ANOVA F = 107 F = 129 F = 305 F = 35
Different letters after values indicate significant differences in t tests ( P 005 P 0001)
AD density ADo dominance
2378 Biodivers Conserv (2010) 192371ndash2387
123
Authors personal copy
not present a topographic association due to wide distribution through the gradient On the
other hand the distinction among environmental and spatial effects showed that space
contributes significantly to the distribution of tree species (Fig 3) Environmental variables
selected by forward selection procedure (P 005) are summarized in Table 4 All geo-
graphical terms of the polynomial function were significant (P 005) during spatial CCA
and were add to the model The four CCA analyses provided the following results
1 CCA of the species matrix constrained by the environmental matrix sum of all
canonical eigenvalues = 0944 Monte Carlo tests for overall analysis F = 286
P 0001
2 CCA of the species matrix constrained by spatial matrix sum of all canonical
eigenvalues = 1017 Monte Carlo tests for overall analysis F = 359 P 0001
Fig 2 Ecological groups of vertical distribution (A) regeneration (B) and dispersal (C) in five transects ofriverine forest of Botucaraı river southern Brazil Pi pioneer Ld light-demanding St shade-tolerant Zoozoochorous Auto autochorous Ane anemochorous Hydro hydrochorous
Biodivers Conserv (2010) 192371ndash2387 2379
123
Authors personal copy
3 Environmental partial CCA (after removing the effect of geographical matrix) sum of
all canonical eigenvalues = 0416 Monte Carlo tests for overall analysis F = 135
P 0001
Table 3 Indicator species analysis (ISA) performed for species with density C10 individuals sampled infive transects with 20 sampling units each Botucaraı river southern Brazil
Biplots of environmental variables and species or environmental variables and sampling
units were depicted with the environmental partial analyses results (Fig 4) In this step
species distributions are purely based on speciesndashenvironment relationships The first two
axes accounted respectively for 268 (eigenvalue = 0112) and 181 (eigen-
value = 0075) of the speciesndashenvironment relationships and speciesndashenvironment corre-
lations for these axes were 0742 and 0649 (P 005) respectively The first four axes
accounted for 691 of speciesndashenvironment relationships Table 4 shows the intraset
correlations among variables and canonical coefficients with the first two axes The first
canonical axis was positively correlated with topography and several soil variables that are
Fig 3 Variation partitioning ofthe tree species matrix
Table 4 Intraset correlations among environmental variables selected for the model during lsquolsquoenviron-mentalrsquorsquo partial CCA of the species matrix and canonical coefficients of the first two axes
Environmental variables were selected by forward stepwise selection and included on the model if sig-nificant in Monte Carlo tests (P 005)
Biodivers Conserv (2010) 192371ndash2387 2381
123
Authors personal copy
influenced by flooding events Percentages of organic matter saturation of bases and cation
exchange capacity where higher through upper sites whereas aluminium contents were
higher in lower areas Thus this environmental gradient may affect tree distribution pat-
terns by restricting or facilitating species establishment As related in the Indicator
Analysis (although not accounting for spatial patterns directly) typical species of flooding
areas occurred near the levee and depression and the zonation was sharp once frequent
species of well-drained areas occurred only in the middle slope and ridgetop Sampling
units (Fig 5) also appeared distributed according to the toposequence with some over-
lapping due to species distribution
Rarefaction curves of species revealed significant differences on the expected total
number of species in each transect with higher richness within the middle slope (Fig 6)
A regression model (Fig 6) fitted diversity in a second-order polynomial regression
(y = -00586x2 03594x 291 R2 = 077) that showed the same pattern
Discussion
Environmental and spatial patterns
Tree species distribution throughout the topographical gradient indicated that both envi-
ronmental and spatial features were particularly important in predicting species and
community patterns This agrees with the well-know influence of geomorphic features and
hydrological regimes on riparian forests over different temporal and spatial scales
(Tabacchi et al 1998 Turner et al 2004 Desilets and Houle 2005) even though dis-
tinctions between environmental and spatial effects remain poorly studied (Titeux et al
Fig 4 Ordination biplotdepicting the two axes of theenvironmental partial CCA ofsampling units in a riverine forestin southern Brazil Eachsampling unit was identified bydifferent symbols according tothe respective transectEnvironmental variables arerepresented by their acronyms(see Table S1)
2382 Biodivers Conserv (2010) 192371ndash2387
123
Authors personal copy
2004) Here we employed a routine to partialling out the spatial effects on the analysis of
speciesndashenvironment relationships that also highlight the spatial component embedded in
such analysis (Legendre 1993) Once several biotic processes as growth mortality dis-
persal and predation influence the observed distributions of organisms (resulting in spatial
correlation) or if their distributions are dependent on explanatory variables which are
Fig 5 Ordination biplot depicting the two axes of the environmental partial CCA of species of a riverineforest in southern Brazil Species and environmental variables are represented by their acronyms (seeTable S1)
Fig 6 Rarefaction curves of tree species and Shannon diversity indices from five transects of riverine foreston the Botucaraı river southern Brazil Sampling units are representing different transects
Biodivers Conserv (2010) 192371ndash2387 2383
123
Authors personal copy
themselves spatially structured (Legendre 1993 Titeux et al 2004) spatial structuring is
an intrinsic component of ecosystems In our analysis lsquopurersquo spatial trends were more
attributed to species distribution than to lsquopurersquo environmental effects which link conta-
gious biological processes as important to the tree species distribution Furthermore
species and environmental data have a reasonable proportion of similar spatial structuring
identified by the largest proportion on the species variation due to spatially structured
environmental variation (1246) According to Borcard et al (1992) species and envi-
ronmental variables have in this case the same response to some common underlying
causes as the topographicndashflooding gradient In fact several studies have demonstrated the
direct effect of wetting and desiccation processes on both mineralogy and microbial
ecology of the sediment including nutrient dynamics (Baldwin and Mitchell 2000)
Once sediment or soils are submerged the inundation leads to a decrease in oxygen
contents and then resulting in progressive anaerobic conditions Rapid cycling of litter may
occur due to an increase on microbial activity which generates pulses on nutrient contents
and finally result in highly productive systems (Baldwin and Mitchell 2000) However a
negative effect is the rapid oxygen consumption which quickly leads to soil hypoxia or
anoxia When flood ends the anaerobic zones of sediments are newly oxygenated and
microbiota is replaced gradually to a new phase
Other spatially structured variables are sediment deposition and litter displacement
which are not covered in our study As demonstrated in lsquovarzearsquo forests of Amazonia
(Wittmann et al 2004) sediment deposition decrease toward upper sites and species
colonizing such lower areas show specific adaptations to the new site conditions regarding
to adventitious roots that probably offer mechanical support (Parolin et al 2004 Wittmann
et al 2004) However litter removal or deposition including seed bank may affect
directly species distribution once flooding and allied effects reallocate litter and seeds
among sites (Johansson et al 1996) Moreover studies have showed that flooding timing
frequency and magnitude can be used as indicators of sapling zonation on floodplain
forests (Vreugdenhil et al 2006)
All these processes are included in the unmeasured variables or spatially structuring
processes that have been missed by the geographical terms (Titeux et al 2004) and
accounted to the far unexplained variation (662) As also stressed by these authors the
stochastic spacendashtime fluctuations of each population the lsquounsaturationrsquo pattern (some
species do not use all suitable habitats) and species recording in not appropriated spatial
scales contribute to this unexplained variation too Notwithstanding occurrence data or
species abundances are often noisy (ter Braak 1995) and widespread in ecological studies
(Borcard et al 1992 Titeux et al 2004)
Richness and diversity patterns
Significant transitions occurred from the levee and depression to the following lower slope
transect regarding to stand structure and ecological groups Inversions on the proportion of
pioneershade-tolerant trees and auto-hydrochorous to anemochorous trees occurred in that
small transition and affected not only ecological groups but also species occurrences As a
consequence this zonation transect may consist spatially as a boundary for tolerant and
intolerant trees with regarding to flooding In fact few species occurred over the entire
flooding gradient and the lower slope also appears as an edge for several species
In a temporal scale the lower slope area will probably present more heterogeneous
spans in flooding events and it may consist in the most heterogeneous temporalndashspatial
sector across the topographic gradient which agree with the findings of Pollock et al
2384 Biodivers Conserv (2010) 192371ndash2387
123
Authors personal copy
(1998) from wetlands with different flooding regimes In the structured view of the
dynamic-equilibrium model (Huston 1994) different patches from different seral stages
result from spatial variation of disturbance frequencies If disturbance frequencies vary
over time a landscape could also contain patches of different seral stages (Pollock et al
1998) These authors used that assumption in a model with temporally synchronous dis-
turbance and found that at the community-scale level the results supported many pre-
dictions of the dynamic-equilibrium model especially regarding to species richness
In our study there are two major factors related to disturbance frequency First and
foremost is that once river floods are unpredictable due to a hydrological regime that varies
with occasional long-rainy periods temporal heterogeneity should be higher than in sea-
sonal predicted flood areas and second microtopography must create spatial heterogeneity
during floods events in the local area However well-drained upper sites are probably more
affected by inherent community processes as gap-phase dynamics and direct supply rates
of light (Stevens and Carson 2002) As related by Worbes et al (1992) and Parolin (2001)
hundreds of tree species with different phenological and other ecological traits grow in
seasonal flooded forests In such cases the cyclic alternation on floods and droughts drove
species to life history behavioral and morphological adaptations (Lytle and Poff 2004)
But in the case of unpredictable floods and droughts as assigned in our study bet-hedging
strategies might be evolved for example by persistent seed bank or asynchronous
reproductive phenologies (Brock 2003) although there are no conclusive studies related to
this theory (Lytle and Poff 2004)
Allowed by a transition in the ecological groups from the lower sites to the upper ones
species richness has a maximum at the lower slope transect probably due to higher het-
erogeneity in disturbance events (space and time) and correlated variables As reported by
Desilets and Houle (2005) the spatial gradient provides some evidences for stress toler-
ance and competition as factors structuring species distribution across the topographicndash
flooding gradient also boosted by an unpredictable pattern of floods that vary in frequency
timing and magnitude Lower sectors showed predicted ecological groups already
described for these areas (Budke et al 2007 2008 Junk et al 1989 Lytle and Poff 2004)
and expected structuring changes also occurred toward upper sites where the proportion of
shade-tolerant and small trees increased as well as anemochorous trees
In summary spatialndashtemporal and environmental variables are arranging tree species
distribution across the toposequence of our study site Furthermore predicted ecological
groups reflected the dynamics of disturbance in the topographicndashhydrological gradient
Species richness and diversity also reflected such pattern and were higher in the mid-sector
where occasional floods should prevent competitive exclusion and generate high envi-
ronmental heterogeneity
Acknowledgements We are grateful to the Programa de Pos-Graduacao em Botanica of the UniversidadeFederal do Rio Grande do SulmdashUFRGS for the opportunity to undertake this study and to CAPES Agencyfor the scholarship granted to the first author Our special thanks to Diogo lsquolsquoBagualrsquorsquo Lindenmaier forfieldwork assistance and to Ricardo Braga Eduardo Rossi and colleagues of the Laboratorio de Fitoeco-logiamdashUFRGS for critiques and suggestions We also appreciated the reviewing efforts of anonymouscontributors for providing useful comments to the manuscript
References
Baldwin DS Mitchell AM (2000) The effects of drying and re-flooding on the sediment and soil nutrientdynamics of lowland river-floodplain systems a synthesis Regul River 16457ndash467 doi1010021099-1646
Biodivers Conserv (2010) 192371ndash2387 2385
123
Authors personal copy
Barroso GM Morim MP Peixoto AL Ichaso CLF (1999) Frutos e sementes morfologia aplicada a si-stematica de dicotiledoneas Editora UFV Vicosa
Borcard D Legendre P Drapeau P (1992) Partialling out the spatial component of ecological variationEcology 731045ndash1055 doi1010292006WR005044
Brock MA (2003) Drought and aquatic community resilience the role of eggs and seeds in sediments oftemporary wetlands Freshw Biol 481207ndash1218 doi101046j1365-2427200301083x
Budke JC Athayde EA Giehl ELH Zachia RA Eisinger SM (2005) Composicao florıstica e estrategias dedispersao de especies lenhosas em uma floresta ribeirinha arroio Passo das Tropas Santa Maria RSBrasil Iheringia Bot 6017ndash24
Budke JC Jarenkow JA Oliveira-Filho AT Lindenmaier DS (2006) Padroes de riqueza e diversidade emrios de pequeno porte In Mariath JEA Santos RP (eds) Os avancos da botanica no inıcio do seculoXXI SBB Porto Alegre
Budke JC Jarenkow JA Oliveira-Filho AT (2007) Relationships between tree component structuretopography and soils of a riverine forest Rio Botucaraı southern Brazil Plant Ecol 189187ndash200 doi101007s11258-006-9174-8
Budke JC Jarenkow JA Oliveira-Filho AT (2008) Tree community features of two stands of riverine forestunder different flooding regimes in southern Brazil Flora 203162ndash174 doi101016jflora200703001
Cardoso E Schiavini I (2002) Relacao entre distribuicao de especies arboreas e topografia em um gradienteflorestal na Estacao Ecologica do Panga (Uberlandia MG) Rev Bras Bot 25277ndash289
Connell JH (1978) Diversity in tropical rain forests and coral reefs Science 1991302ndash1310 doi101126science19943351302
Damasceno-Junior GA Semir J Santos FAM Leitao-Filho HF (2005) Structure distribution of species andinundation in a riparian forest of Rio Paraguai Pantanal Brazil Flora 200119ndash135 doi101016jflora200409002
Desilets P Houle G (2005) Effects of resource availability and heterogeneity on the slope of the species-areacurve along a floodplain-upland gradient J Veg Sci 16487ndash496 doi1016581100-9233
Dufrene M Legendre P (1997) Species assemblages and indicator species the need for a flexible asym-metrical approach Ecol Monogr 67345ndash366
EMBRAPA (1997) Manual de metodos de analises de solo Empresa Brasileira de Pesquisa Agropecuariaand Centro Nacional de Pesquisas de Solos Rio de Janeiro
Ferreira LV (2000) Effects of flooding duration on species richness floristic composition and forest structurein river margin habitat in Amazonian blackwater floodplain forests implications for future design ofprotected areas Biodivers Conserv 91ndash14 doi101023A1008989811637
Ferreira LV Stohlgren TJ (1999) Effects of river level fluctuation on plant species richness diversity anddistribution in a floodplain forest in Central Amazonia Oecologia 120582ndash587 doi101007s004420050893
Ferreira C Piedade MTF Franco AC Goncalves JFC Junk WJ (2009) Adaptive strategies to tolerateprolonged flooding in seedlings of floodplain and upland populations of Himatanthus sucuuba aCentral Amazon tree Aquat Bot 90246ndash252 doi101016jaquabot200810006
Gotelli NJ Colwell RK (2001) Quantifying biodiversity procedures and pitfalls in the measurement andcomparison of species richness Ecol Lett 4379ndash391 doi101046j1461-0248200100230x
Guilherme FAG Oliveira-Filho AT Appolinario V Bearzoti E (2004) Effects of flooding regime andwoody bamboos on tree community dynamics in a section of tropical semideciduous forest in south-eastern Brazil Plant Ecol 17419ndash36 doi101023BVEGE000004605197752cd
Huston M (1994) Biological diversity the coexistence of species in changing landscapes CambridgeUniversity Press Cambridge
IPAGRO (1982) Atlas agroclimatico do Rio Grande do Sul Pallotti Porto AlegreJohansson ME Nilsson C Nilsson E (1996) Do rivers function as corridors for plant dispersal J Veg Sci
7593ndash598Junk WJ Bayley PB Sparks RE (1989) The flood pulse concept in river-floodplain systems Can J Fish
Aquat Sci 106110ndash127Legendre P (1993) Spatial autocorrelationmdashtrouble or new paradigm Ecology 741659ndash1673Leinz V (1949) Contribuicao a geologia dos derrames basalticos do Rio Grande do Sul Bol Fac Filos Let
581ndash55Lytle DA Poff NL (2004) Adaptation to natural flow regimes Trends Ecol Evol 1994ndash100 doi
101016jtree200310002McCune B Mefford MJ (1997) PCndashORD Multivariate analysis of ecological data version 436 MjM
Software Design Glaneden BeachMueller-Dombois D Ellenberg H (1974) Aims and methods of vegetation ecology John Wiley New York
2386 Biodivers Conserv (2010) 192371ndash2387
123
Authors personal copy
Neiff JJ (1990) Ideas para la interpretacion ecologica del Parana Interciencia 15424ndash441Neiff JJ Neiff M (2003) Pulso software para analisis de fenomenos recurrentes httpwwwneiffcom
Cited 25 May 2009Oliveira-Filho AT Vilela EA Gavilanes ML Carvalho DA (1994) Effect of flooding regime and understory
bamboos on the physiognomy and tree species composition of a tropical semideciduous forest in south-eastern Brazil Vegetatio 11399ndash124
Oliveira-Filho AT Curi N Vilela EA Carvalho DA (2001) Variation in tree community composition andstructure with changes in soil properties within a fragment of semideciduous forest in south-easternBrazil Edinb J Bot 58139ndash158 doi101017S0960428601000506
Oliveira-Filho AT Jarenkow JA Rodal MJN (2006) Floristic relationships of seasonally dry forests ofeastern South America based on tree species distribution patterns In Pennington RT Ratter JA LewisGP (eds) Neotropical savannas and dry forests plant diversity biogeography and conservation CRCPress Boca Raton
Parolin P (2001) Morphological and physiological adjustments to waterlogging and drought in seedlings ofAmazonian floodplain trees Oecologia 128326ndash335 doi101007s004420100660
Parolin P de Simone O Haase K Waldhoff D Rottenberger S Kuhn U Kesselmeier J Kleiss B SchmidtW Piedade MTF Junk WJ (2004) Central Amazonian floodplain forests tree adaptations in a pulsingsystem Bot Rev 70357ndash380 doi1016630006-8101(2004)070[0357CAFFTA]20CO2
Pijl L (1982) Principles of dispersal in higher plants Springer New YorkPillar VD (2006) Multivariate exploratory analysis randomization testing and bootstrap resampling version
2320 Departamento de Ecologia UFRGS Porto AlegrePollock MM Naiman RJ Hanley TA (1998) Plant species richness in riparian wetlandsmdasha test of biodi-
versity theory Ecology 7994ndash105Robertson KM (2006) Distributions of tree species along point bars of 10 rivers in the south-eastern US
Coastal Plain J Biogeogr 33121ndash132 doi101111j1365-2699200501371xRosales J Petts G Knab-Vispo C (2001) Ecological gradients within the riparian forests of the lower Caura
river Venezuela Plant Ecol 152101ndash118 doi101023A1011411020040Stevens MHH Carson WP (2002) Resource quantity not resource heterogeneity maintains plant diversity
Ecol Lett 5420ndash426 doi101046j1461-0248200200333xStreck EV Kampf N Dalmolin RSD Klamt E Nascimento PC Schneider P (2002) Solos do Rio Grande do
Sul EMATERRS and UFRGS Porto AlegreSwaine MD Whitmore TC (1988) On the definition of ecological species groups in tropical rain forests
Vegetatio 7581ndash86Tabacchi E Correll DL Hauer R Pinay G Planty-Tabacchi AM Wissmar R (1998) Development
maintenance and role of riparian vegetation in the river landscape Freshw Biol 40497ndash516 doi101046j1365-2427199800381x
ter Braack CJF Smilauer P (1998) Canoco reference manual and userrsquos guide to Canoco for Windowssoftware for canonical community ordination (version 40) Microcomputer Power Ithaca
ter Braak CJF (1995) Ordination In Jongman RHG ter Braak CJF van Togeren OFR (eds) Data analysis incommunity and landscape ecology Cambridge University Press New York
Titeux N Dufrene M Jacob JP Paquay M Defourny P (2004) Multivariate analysis of fine-scale breedingbird atlas using a geographical information system and partial canonical correspondence analysisenvironmental and spatial effects J Biogeogr 311841ndash1856 doi101111j1365-2699200401125x
Turner MG Gergel SE Dixon MD Miller JR (2004) Distribution and abundance of trees in floodplainforests of the Wisconsin river environmental influences at different scales J Veg Sci 15729ndash738
Vreugdenhil SJ Kramer K Pelsma T (2006) Effects of flooding duration frequency and depth on thepresence of saplings of six woody species in north-west Europe For Ecol Manage 23647ndash55 doi101016jforeco200608329
Walker KF Sheldon F Puckridge JT (1995) A perspective on dryland river ecosystems Regul River 1185ndash104 doi101002rrr3450110108
Weiher E (2003) Species richness along multiple gradients testing a general multivariate model in oaksavannas Oikos 101311ndash316 doi101034j1600-0706200312216x
Wittmann F Junk WJ Piedade MTF (2004) The varzea forests in Amazonia flooding and the highlydynamic geomorphology interact with natural forest succession For Ecol Manage 196199ndash212 doi101016jforeco200402060
Worbes M Klinge H Revilla JD Martius C (1992) On the dynamics floristic subdivision and geographicaldistribution of Varzea forests in Central Amazonia J Veg Sci 3553ndash564
Zar JH (1996) Biostatistical analysis Prentice-Hall New Jersey
Biodivers Conserv (2010) 192371ndash2387 2387
123
Authors personal copy
influence of flooding on tree species distribution created a vegetation zonation that is
determined by predicted ecological traits
Keywords Disturbance Ecological groups Flooding regime Partial CCA Soil properties Species richness and diversity
Introduction
Natural disturbances play an important role in structuring plant communities by leading to
environmental heterogeneity over space and time at different scales Several studies have
demonstrated that disturbance and abiotic stress affect diversity especially at local-scale
(Ferreira and Stohlgren 1999 Weiher 2003) In riparian ecosystems flooding events are
the key factor in shaping community features either by a positive or a negative effect on
the ecosystemrsquos function according to the timing frequency and magnitude of such events
(Neiff 1990) Long-lasting floods represent a major stress and may result in species-poor
plant communities due to restricted productivity in the aquatic phase and high mortality of
non-adapted species (Pollock et al 1998 Guilherme et al 2004 Wittmann et al 2004) On
the other hand periodic and short floods may contribute to the input of nutrients which
increase productivity and diversity (Desilets and Houle 2005)
Once magnitude and duration of flooding are directly associated with local relief (eg
relative elevation inclination) many studies have investigated the relationships among
topography and correlated variables (eg chemical and textural soil properties sedimen-
tation rates) on the distribution of plant species and patterns of richness and diversity
(Oliveira-Filho et al 1994 Ferreira 2000 Rosales et al 2001 Damasceno-Junior et al
2005 Budke et al 2007)
In riparian systems with regular or predicted (seasonal) flood events as Amazonian and
Pantanal floodplains in South America plant species show different strategies to survive
floods including morphological anatomical and physiological adaptations and also phe-
nological timing for both reproductive and vegetative phases Ferreira et al (2009) has
demonstrated that species living in low-lying areas may be ecotypes originated from
surrounding non-flooded forests In contrast riverine forests with unpredictable flooding
pulses are frequently colonized by species of early successional stages of wide geo-
graphical distribution (Walker et al 1995 Budke et al 2007) On this hand Budke et al
(2008) observed that in low order rivers where water column oscillated due to a local and
concentrated rainy period the species richness increased along a gradient from frequently
to occasionally flooded stands Furthermore Robertson (2006) showed that predictability
of species occurrence across different rivers in the south-eastern US Coastal Plain was
directly related to the geomorphic dynamics intermediate level of stream energy (eg
flooding magnitude) and non-altered hydrological regimes
This ubiquitous difference of showing or not a temporally synchronous and expected
disturbance is one of the most interesting in eco-hydrological studies On this way the
structured view of the dynamic-equilibrium model (Huston 1994) shows different patches
from different seral stages result from spatial variation of disturbance frequencies If
disturbance frequencies vary over time a landscape could also contain such patches
(Pollock et al 1998) Indeed as expected in the intermediate disturbance model richness
would be higher when disturbance is neither too rare nor too frequent (Connell 1978)
In this work we focused on species richness and diversity of tree species in a riverside
sequence from the point bar to the lateral slope in a river with an unpredictable flooding
2372 Biodivers Conserv (2010) 192371ndash2387
123
Authors personal copy
regime The inundation regime also varies according to the topographical position fol-
lowing to frequently flooded forests to well drained non-flooded forests Furthermore we
investigated the relationships among tree component structure species and functional
groups distribution and spatialndashenvironmental variables We hypothesized that (1) as
flooding may gradually affect environmental heterogeneity richness and diversity will be
higher at intermediate elevations and directly associated with increased environmental
heterogeneity and (2) both species and functional groups will reflect variations in elevation
andor soil texture and chemistry
Methods
Study area
The study area is a forest remnant of ca 20 ha situated in the riparian fringes of the
Botucaraı river near its confluence with the Jacuı river (Lat 30010S Long 52470W)
(Budke et al 2007) The headwaters of the river lie in the southernmost extent (ca 650 m
asl) of the high planes region locally known as Planalto Meridional which geologically
is part of the Serra Geral formation made up of Cretaceous basalts originated from giant
lava flows that covered the sedimentary lowlands of the Parana Basin (Leinz 1949)
Downstream at its mid-course the Botucaraı river reaches the lowlands (ca 100 m asl)
and the topography is dominated by recently flood-deposited sedimentsmdashmeanders and
point bars At its lower course near the study area flooding events are enhanced by the
confluence with the stronger adjoining stream flow of the Jacuı river therefore promoting
lateral overflow According to Budke et al (2007) soils in such areas reflect not only the
geomorphic features from the basin but they also reflect flooding dynamics which fre-
quently produces non-stratified layers of fine gravel wood debris litter and sediment As a
consequence different soil profiles occur from well structured planosols in the riverside
slopes to recent deposited layers of sediment in the lowlands
The regional climate is moist subtropical without a regular dry season mean tem-
peratures ranges from 249C (hottest month) to 142C (coldest month) with high tem-
perature variation (absolutes values ranges from 42C in the summer to -3C in the
winter) mean annual rainfall is 1594 mm year-1 respectively (IPAGRO 1982) The
predominant soil is a Hydromorphic Planosol with typical stratified layers of depositional
sediments (Streck et al 2002)
Floods in the area are highly unpredictable because there is no marked seasonal rainy
period and rainfall is relatively well distributed throughout the year As a consequence
floods occur at any time of the year with duration of overflow periods varying from some
days to a few weeks (Budke et al 2008)
Regional vegetation is an extent of the Atlantic Forest Domain (Oliveira-Filho et al
2006) and includes overlapping patches of Seasonal Semideciduous Forests and Araucaria
Rain Forests at the river headwaters at Serra Geral formation Seasonal Semideciduous
Forests shows several genera of deciduous Fabaceae trees as Apuleia leiocarpa Myro-carpus frondosus Enterolobium contortisiliquun Parapiptadenia rigida and Erythrinafalcata as well as perennial ones which include Myrtaceae Lauraceae Sapotaceae and
Rubiaceae among others Canopy and emergent tree species can reach 25 m high
although mean vegetation stature is near 12ndash15 m In the lowlands of the river basin
Seasonal Semideciduous Forests is gradually changed by grasslands of the Pampa Domain
Biodivers Conserv (2010) 192371ndash2387 2373
123
Authors personal copy
(Oliveira-Filho et al 2006) and the river basin play a typical role of forest corridor toward
south reaching the Uruguay pampas as forest enclaves or galleries (Budke et al 2006)
Data collection
We carried out a tree survey in a 1 ha plot installed in a toposequence in the lowland areas
from the river margin to the lateral slope and therefore liable to different flooding regimes
The plot was divided in five 10 9 200 m transects and each transect was subdivided in
sampling units of 10 9 10 m All individual living trees having at least one stem and with
perimeter at breast height (pbh) C15 cm were sampled Voucher specimens of the different
species were collected prepared and lodged in the Herbarium ICN of the Universidade
Federal do Rio Grande do Sul (UFRGS)
A detailed topographic survey of the transects was carried out using a 10 m long water-
filled levelling hose 38 in a tape measure and a compass according to Cardoso and
Schiavini (2002) The resulting grid of vertical transects was used to produce contour maps
and to obtain the relative elevation of each sampling unit rather to the river To estimate
flooding frequency in each sampling unit we overlap their relative elevation to the
hydrometer records of the Jacuı river station (data calibrated according to topography)
Through Pulse 111 software (Neiff and Neiff 2003) we estimated the mean number of
floods per year from 1981 to 2004 and we used this variable as a pulse disturbance estimate
to sampling units (hereafter named flooding)
We collected samples of the topsoil (0ndash20 cm depth) from 15 sites distributed in dif-
ferent positions in such a way that its overall topographic variation was encompassed The
soil samples were kept in polyethylene bags and taken to the UFRGS Soil Laboratory for
chemical and textural analyses The variables were pH in water suspension levels of
potential acidity (Al H) bases saturation (V) sum of bases (S) cation exchange
capacity (CEC) organic matter (OM) and levels of clay sand and silt All procedures
followed EMBRAPA (1997) protocol In those plots without a soil subsample we
extrapolated real values by distance-proportional mean of the closest plots (ter Braak
1995) We compared the means of each soil property among transects by using one-way
ANOVA (Zar 1996)
Data analysis
Phytosociological parameters of density frequency and dominance (derived from tree
basal area) were calculated to describe tree community structure (Mueller-Dombois and
Ellenberg 1974) Frequency distributions into classes of diameter for each transect were
prepared and one-way ANOVA was used to compare transects Classes of exponentially
increasing range were used for diameters to make up for the accentuated decline in tree
frequency towards larger diameters (Oliveira-Filho et al 2001)
We applied rarefaction curves for each transect in order to analyse the range of species
richness within the toposequence The rarefaction curve technique generates expected
number of species based on the individualsrsquo density and then providing statistical
assumptions to this comparison (Gotelli and Colwell 2001) We also compared Shannon
diversity indices (H0) of each transect by bootstrap resampling tests with the software
Multiv (Pillar 2006) and depicted diversity and topography in a regression model
To verify topographical ranges of species we used an Indicator Species AnalysismdashISA
(Dufrene and Legendre 1997) which is a direct analysis of association between flooding
2374 Biodivers Conserv (2010) 192371ndash2387
123
Authors personal copy
and species distribution As the aim of this analysis was to assess the association between
species and topographyflooding it was used a non-hierarchical clustering procedure kmeans to produce k groups from the mean elevation of the original sampling units and by
using the resulting groups as the clustering factor required in the ISA (Dufrene and
Legendre 1997 Budke et al 2008) The analysis was performed in the PC-Ord program
(McCune and Mefford 1997)
We partitioned the variance of species distribution over the toposequence accounted by
spatial and environmental variables by successive partial Correspondence Canonical
Analysis (Borcard et al 1992) This approach combines three different matrices to
decompose all species variation in four components pure effect of environment pure
effect of spatial pattern combined variation of environment and spatial pattern and finally
unexplained variation Species assemblages from a determined position are affected by
surrounding sites because of contagious biotic process and environmental variables used to
describe biological processes are also neither randomly or uniformly spatially distributed
(Legendre 1993) In such case it is necessary to incorporate the spatial structure in the
modelling because the independence of observations is not respected (Legendre 1993) The
first matrix or species matrix included the abundances of all species with density C10
individuals The environmental matrix included initially all chemical and granulometric
figures the topographic variable (average elevation) and an ordinal (ranking) variable
labeled lsquolsquoflooding frequencyrsquorsquo We obtained the last variable directly from the topographic
survey summarizing flood occurrences and their intensity in each plot (Budke et al 2008)
The third matrix or spatial matrix included all terms of a polynomial function of geo-
graphical coordinates ie centers of each sampling unit and it was made by adding all
terms of a cubic trend surface regression
f x yeth THORN frac14 x y xy x2 y2 x2y xy2 x3 y3
According to Borcard et al (1992) this ensures the detection of more complex spatial
features as gaps or patches which require the quadratic and cubic terms of the coordinates
and their interactions
The variance partitioning proceeded in two steps First we extracted from each
explanatory matrix (environmental variables and spatial variables) all non-significant
variables by forward stepwise regression using Monte Carlo permutations (999 permuta-
tions P 005) with CANOCO 40 (ter Braack and Smilauer 1998) and performed two
canonical ordinations that are redundant in terms of explained variation over the species
data due to spatial structuring (Borcard et al 1992) Then two partial canonical analyses
were carried out (lsquoenvironmentalrsquo and lsquospatialrsquo) each of them constrained by one of the
sets of explanatory variables to determine the relative contribution of environmental and
spatial variables in accounting for species variation Final partition is possible by using the
sum of all canonical eigenvalues of two canonical ordinations constrained by one set of
explanatory variables and of two partial canonical ordinations each of them constrained
by one set of explanatory variables while controlling for the effect of the others (covari-
ables) (Borcard et al 1992 Titeux et al 2004)
To search for ecological differences in the toposequence we classified the species in
ecological groups of regeneration vertical distribution and dispersal We defined regen-
eration based on the categories proposed by Swaine and Whitmore (1988) The two main
levels are (a) lsquopioneerrsquo which includes the species showing an entirely heliophilous life
cycle a seed bank but no bank of juveniles and (b) lsquolate successional speciesrsquo which are
those able to germinate and establish under some degree of shade to form a bank of
Biodivers Conserv (2010) 192371ndash2387 2375
123
Authors personal copy
juveniles The later was divided into (b1) lsquoshade-tolerantrsquo and (b2) lsquolight-demanding late
successional speciesrsquo which are better seen as the two sides of a continuum of solar
radiation required by the trees to lsquoreleasersquo the bank of juveniles (Oliveira-Filho et al 1994)
We defined the vertical distribution based on the strata commonly reached by the adult
individuals (a) small tree species (b) medium tree species and (c) tall tree species (see
Oliveira-Filho et al 1994) The dispersal was (a) zoochorous species with animal-med-
iated dispersal syndrome (b) anemochorous and hydrochorous those with mechanisms to
facilitate wind-dispersal or flotation and (c) autochorous those dispersed by free fall or
ballistic mechanisms (Pijl 1982) The classification of each species into the ecological
groups was based on observations during fieldwork from 2004 to 2005 and on scientific
literature (Barroso et al 1999 Budke et al 2005 2008) We tested the distribution of trees
into frequency classes according to the ecological group by KruskalndashWallis tests (Zar
1996)
Results
River corridor along the studied area has a typical meandering system with well-defined
geomorphic features The lowest sector encompasses the levee and depression which
interacts directly with river floods Next to these sites we identified the lower-slope the
middle-slope and the ridge according to the relative elevation to the river channel
(Table 1) and these sectors corresponded to our installed transects The lower slope veg-
etation is a sharp transition between lowland and upland forests and only large inundation
floods this sector whereas upland sites present slight differences in vegetation structure
due to absence of flooding and allied effects Nevertheless there is a distinct gradient of
organic matter (OM) clay and cation exchange capacity (CEC) being higher toward upper
sites as also showed by potential acidity (Al H) (Table 1) By other hand sum of bases
(S) and phosphorus contents (P) showed a tendency of decreasing toward upper sites
(Table 1) Furthermore the variance of some soil variables was quite high and demon-
strated the high heterogeneity across transects
The field inventory yielded a total of 1229 individuals belonging to 72 species and 35
families from which Myrtaceae and Fabaceae were the richest families with 11 species
followed by Rubiaceae and Sapotaceae with four species (Table S1) Although Myrtaceae
and Fabaceae presented the highest richness both families appeared generally with low
density or basal area The stand showed a forest of low stature with most individuals
between 5 and 7 m tall and few emergent trees reaching up 15 m The diameter-class
distribution of trees revealed typical inverted-J distribution with most individuals situated
in the first two classes (Fig 1) Across the toposequence higher density was found near the
river (Levee) followed by lower density values in the depression and again an increased
density through lower and middle slope On the other hand the ridgetop transect presented
the lowest density but an increased basal area (Table 2) and several trees with diameter
[40 cm Vertical distribution of trees also showed the predominance of medium-sized
individuals followed by a decreased proportion of small and emergent trees (Fig 2A)
The proportion of light-demanding trees was higher towards the upper sites (Fig 2B)
Pioneer trees presented an opposite pattern being more abundant in low sites Shade-
tolerant trees also showed an increased density at upper sites where flooding is restrict or
absent Within the dispersal groups zoochorous trees presented higher proportion in all
transects Autochorous and hydrochorous trees decreased toward the ridgetop whereas
anemochorous trees followed the inverse pattern (Fig 2C) These structural patterns
2376 Biodivers Conserv (2010) 192371ndash2387
123
Authors personal copy
Tab
le1
So
ilv
aria
ble
so
ffi
ve
tran
sect
so
fri
ver
ine
fore
sto
nth
eB
otu
cara
ıri
ver
so
uth
ern
Bra
zil
Soil
var
iable
sL
evee
Dep
ress
ion
L-s
lope
M-s
lope
Rid
ge
FP
Rel
ativ
eel
evat
ion
(m)
38
plusmn0
5a
54
plusmn0
7a
85
plusmn2
3b
11
8plusmn
35
bc
13
6plusmn
37
c8
03
0
00
1
pH
(H2O
)4
8plusmn
04
47
plusmn0
44
7plusmn
04
47
plusmn0
64
7plusmn
06
08
50
93
ns
Pmdash
Meh
lich
(mg
dm
-3)
71
plusmn2
17
1plusmn
23
63
plusmn1
66
plusmn1
75
7plusmn
14
22
30
07
ns
K(m
gd
m-
3)
76
1plusmn
12
57
96
plusmn1
64
89
plusmn2
62
94
3plusmn
34
87
8plusmn
24
14
36
03
5n
s
Ca
(cm
olc
dm
-3)
62
plusmn4
36
8plusmn
44
7plusmn
47
61
plusmn4
57
plusmn3
73
19
05
2n
s
Mg
(cm
olc
dm
-3)
15
plusmn0
71
6plusmn
07
15
plusmn0
71
5plusmn
06
14
plusmn0
50
36
09
8n
s
Al
H
(cm
olc
dm
-3)
66
plusmn3
1a
71
plusmn3
ab8
5plusmn
4ab
96
plusmn5
5ab
10
plusmn4
2b
97
60
04
S(c
mo
lcd
m-
3)
8plusmn
48
87
plusmn5
18
1plusmn
46
84
plusmn5
17
4plusmn
41
27
70
59
ns
CE
C(c
mo
lcd
m-
3)
15
2plusmn
48
16
2plusmn
38
17
1plusmn
33
17
9plusmn
43
17
3plusmn
34
14
50
22
ns
V(
)5
57
plusmn1
71
49
7plusmn
20
24
59
plusmn2
46
45
5plusmn
23
64
18
plusmn2
07
40
60
39
ns
OM
()
26
plusmn1
1a
28
plusmn1
a3
2plusmn
1ab
37
plusmn1
1b
38
plusmn0
8b
22
9
00
01
Cla
y(
)1
37
plusmn2
4a
15
4plusmn
33
ab1
58
plusmn2
8ab
16
plusmn2
1b
15
5plusmn
17
ab2
49
00
4
San
d(
)2
0plusmn
74
23
1plusmn
22
24
2plusmn
10
12
23
plusmn7
52
23
plusmn6
28
91
00
6n
s
Sil
t(
)6
42
plusmn1
09
60
4plusmn
14
59
8plusmn
12
96
2plusmn
91
62
plusmn7
24
91
02
9n
s
Val
ues
are
mea
ns
plusmnst
and
ard
dev
iati
on
sfr
om
0to
20
cmd
epth
top
soil
sam
ple
s(N
=2
0fo
rea
chtr
anse
ct)
Dif
fere
nt
lett
ers
afte
rv
alu
esin
dic
ate
sign
ifica
nt
dif
fere
nce
sin
AN
OV
Ate
sts
(ns
=n
on
-sig
nifi
can
t)
Biodivers Conserv (2010) 192371ndash2387 2377
123
Authors personal copy
shaped the physiognomic features of different sectors that varied according to the topo-
sequence and consequently in flooding regime The depression sector presented lower
density basal area and also low tree diameters whereas the levee portion presented high
density and basal area
Species distribution across the topographic gradient is presented in Table 3 according
to the Indicator Species Analysis Some species were clearly distributed from lowland to
medium sites as Eugenia uniflora Myrciaria tenella Eugenia uruguayensis whereas
others were restricted to upland areas as Chomelia obtusa and Cordia americana Many
species did not show a specific site distribution and occurred over a wide distribution range
as Gymnanthes concolor and Casearia sylvestris
The relative elevation of each transect reflects the pattern of flooding frequency and
duration in each site then spatial aggregation of trees may indicate preferences or
restriction on the establishment of some species Typical riverine species appeared near the
river margin as Pouteria gardneriana Guettarda uruguensis and others (Table 3) whereas
typical species of well-drained forests as Sorocea bonplandii Parapiptadenia rigida and
Cupania vernalis occurred frequently in the ridgetop transect Furthermore 13 species did
Fig 1 Diameter-class distributions of trees with pbh C 15 cm surveyed in five transects of riverine foreston the Botucaraı river southern Brazil Diameter-classes are used for increasing intervals (see lsquolsquoMethodsrsquorsquosection) Bars and ranges are means and 95 confidence intervals of 100 sampling units respectively
Table 2 Density (ind ha-1) dominance (m2 ha-1) mean height (m) and mean diameter (cm) for differenttransects of the riverside forest of the Botucaraı river southern Brazil
Transect AD ADo Height Diameter
Levee 1655 plusmn 467 a 3927 plusmn 268 a 62 plusmn 25 a 1761 plusmn 1697
Depression 1005 plusmn 369 b 198 plusmn 137 b 67 plusmn 21 b 161 plusmn 1194
L-slope 1120 plusmn 443 ab 229 plusmn 165 b 71 plusmn 26 b 1547 plusmn 1302
M-slope 1415 plusmn 438 a 274 plusmn 165 b 69 plusmn 25 b 1453 plusmn 98
Ridge 950 plusmn 294 b 282 plusmn 194 b 7 plusmn 24 b 1727 plusmn 1555
ANOVA F = 107 F = 129 F = 305 F = 35
Different letters after values indicate significant differences in t tests ( P 005 P 0001)
AD density ADo dominance
2378 Biodivers Conserv (2010) 192371ndash2387
123
Authors personal copy
not present a topographic association due to wide distribution through the gradient On the
other hand the distinction among environmental and spatial effects showed that space
contributes significantly to the distribution of tree species (Fig 3) Environmental variables
selected by forward selection procedure (P 005) are summarized in Table 4 All geo-
graphical terms of the polynomial function were significant (P 005) during spatial CCA
and were add to the model The four CCA analyses provided the following results
1 CCA of the species matrix constrained by the environmental matrix sum of all
canonical eigenvalues = 0944 Monte Carlo tests for overall analysis F = 286
P 0001
2 CCA of the species matrix constrained by spatial matrix sum of all canonical
eigenvalues = 1017 Monte Carlo tests for overall analysis F = 359 P 0001
Fig 2 Ecological groups of vertical distribution (A) regeneration (B) and dispersal (C) in five transects ofriverine forest of Botucaraı river southern Brazil Pi pioneer Ld light-demanding St shade-tolerant Zoozoochorous Auto autochorous Ane anemochorous Hydro hydrochorous
Biodivers Conserv (2010) 192371ndash2387 2379
123
Authors personal copy
3 Environmental partial CCA (after removing the effect of geographical matrix) sum of
all canonical eigenvalues = 0416 Monte Carlo tests for overall analysis F = 135
P 0001
Table 3 Indicator species analysis (ISA) performed for species with density C10 individuals sampled infive transects with 20 sampling units each Botucaraı river southern Brazil
Biplots of environmental variables and species or environmental variables and sampling
units were depicted with the environmental partial analyses results (Fig 4) In this step
species distributions are purely based on speciesndashenvironment relationships The first two
axes accounted respectively for 268 (eigenvalue = 0112) and 181 (eigen-
value = 0075) of the speciesndashenvironment relationships and speciesndashenvironment corre-
lations for these axes were 0742 and 0649 (P 005) respectively The first four axes
accounted for 691 of speciesndashenvironment relationships Table 4 shows the intraset
correlations among variables and canonical coefficients with the first two axes The first
canonical axis was positively correlated with topography and several soil variables that are
Fig 3 Variation partitioning ofthe tree species matrix
Table 4 Intraset correlations among environmental variables selected for the model during lsquolsquoenviron-mentalrsquorsquo partial CCA of the species matrix and canonical coefficients of the first two axes
Environmental variables were selected by forward stepwise selection and included on the model if sig-nificant in Monte Carlo tests (P 005)
Biodivers Conserv (2010) 192371ndash2387 2381
123
Authors personal copy
influenced by flooding events Percentages of organic matter saturation of bases and cation
exchange capacity where higher through upper sites whereas aluminium contents were
higher in lower areas Thus this environmental gradient may affect tree distribution pat-
terns by restricting or facilitating species establishment As related in the Indicator
Analysis (although not accounting for spatial patterns directly) typical species of flooding
areas occurred near the levee and depression and the zonation was sharp once frequent
species of well-drained areas occurred only in the middle slope and ridgetop Sampling
units (Fig 5) also appeared distributed according to the toposequence with some over-
lapping due to species distribution
Rarefaction curves of species revealed significant differences on the expected total
number of species in each transect with higher richness within the middle slope (Fig 6)
A regression model (Fig 6) fitted diversity in a second-order polynomial regression
(y = -00586x2 03594x 291 R2 = 077) that showed the same pattern
Discussion
Environmental and spatial patterns
Tree species distribution throughout the topographical gradient indicated that both envi-
ronmental and spatial features were particularly important in predicting species and
community patterns This agrees with the well-know influence of geomorphic features and
hydrological regimes on riparian forests over different temporal and spatial scales
(Tabacchi et al 1998 Turner et al 2004 Desilets and Houle 2005) even though dis-
tinctions between environmental and spatial effects remain poorly studied (Titeux et al
Fig 4 Ordination biplotdepicting the two axes of theenvironmental partial CCA ofsampling units in a riverine forestin southern Brazil Eachsampling unit was identified bydifferent symbols according tothe respective transectEnvironmental variables arerepresented by their acronyms(see Table S1)
2382 Biodivers Conserv (2010) 192371ndash2387
123
Authors personal copy
2004) Here we employed a routine to partialling out the spatial effects on the analysis of
speciesndashenvironment relationships that also highlight the spatial component embedded in
such analysis (Legendre 1993) Once several biotic processes as growth mortality dis-
persal and predation influence the observed distributions of organisms (resulting in spatial
correlation) or if their distributions are dependent on explanatory variables which are
Fig 5 Ordination biplot depicting the two axes of the environmental partial CCA of species of a riverineforest in southern Brazil Species and environmental variables are represented by their acronyms (seeTable S1)
Fig 6 Rarefaction curves of tree species and Shannon diversity indices from five transects of riverine foreston the Botucaraı river southern Brazil Sampling units are representing different transects
Biodivers Conserv (2010) 192371ndash2387 2383
123
Authors personal copy
themselves spatially structured (Legendre 1993 Titeux et al 2004) spatial structuring is
an intrinsic component of ecosystems In our analysis lsquopurersquo spatial trends were more
attributed to species distribution than to lsquopurersquo environmental effects which link conta-
gious biological processes as important to the tree species distribution Furthermore
species and environmental data have a reasonable proportion of similar spatial structuring
identified by the largest proportion on the species variation due to spatially structured
environmental variation (1246) According to Borcard et al (1992) species and envi-
ronmental variables have in this case the same response to some common underlying
causes as the topographicndashflooding gradient In fact several studies have demonstrated the
direct effect of wetting and desiccation processes on both mineralogy and microbial
ecology of the sediment including nutrient dynamics (Baldwin and Mitchell 2000)
Once sediment or soils are submerged the inundation leads to a decrease in oxygen
contents and then resulting in progressive anaerobic conditions Rapid cycling of litter may
occur due to an increase on microbial activity which generates pulses on nutrient contents
and finally result in highly productive systems (Baldwin and Mitchell 2000) However a
negative effect is the rapid oxygen consumption which quickly leads to soil hypoxia or
anoxia When flood ends the anaerobic zones of sediments are newly oxygenated and
microbiota is replaced gradually to a new phase
Other spatially structured variables are sediment deposition and litter displacement
which are not covered in our study As demonstrated in lsquovarzearsquo forests of Amazonia
(Wittmann et al 2004) sediment deposition decrease toward upper sites and species
colonizing such lower areas show specific adaptations to the new site conditions regarding
to adventitious roots that probably offer mechanical support (Parolin et al 2004 Wittmann
et al 2004) However litter removal or deposition including seed bank may affect
directly species distribution once flooding and allied effects reallocate litter and seeds
among sites (Johansson et al 1996) Moreover studies have showed that flooding timing
frequency and magnitude can be used as indicators of sapling zonation on floodplain
forests (Vreugdenhil et al 2006)
All these processes are included in the unmeasured variables or spatially structuring
processes that have been missed by the geographical terms (Titeux et al 2004) and
accounted to the far unexplained variation (662) As also stressed by these authors the
stochastic spacendashtime fluctuations of each population the lsquounsaturationrsquo pattern (some
species do not use all suitable habitats) and species recording in not appropriated spatial
scales contribute to this unexplained variation too Notwithstanding occurrence data or
species abundances are often noisy (ter Braak 1995) and widespread in ecological studies
(Borcard et al 1992 Titeux et al 2004)
Richness and diversity patterns
Significant transitions occurred from the levee and depression to the following lower slope
transect regarding to stand structure and ecological groups Inversions on the proportion of
pioneershade-tolerant trees and auto-hydrochorous to anemochorous trees occurred in that
small transition and affected not only ecological groups but also species occurrences As a
consequence this zonation transect may consist spatially as a boundary for tolerant and
intolerant trees with regarding to flooding In fact few species occurred over the entire
flooding gradient and the lower slope also appears as an edge for several species
In a temporal scale the lower slope area will probably present more heterogeneous
spans in flooding events and it may consist in the most heterogeneous temporalndashspatial
sector across the topographic gradient which agree with the findings of Pollock et al
2384 Biodivers Conserv (2010) 192371ndash2387
123
Authors personal copy
(1998) from wetlands with different flooding regimes In the structured view of the
dynamic-equilibrium model (Huston 1994) different patches from different seral stages
result from spatial variation of disturbance frequencies If disturbance frequencies vary
over time a landscape could also contain patches of different seral stages (Pollock et al
1998) These authors used that assumption in a model with temporally synchronous dis-
turbance and found that at the community-scale level the results supported many pre-
dictions of the dynamic-equilibrium model especially regarding to species richness
In our study there are two major factors related to disturbance frequency First and
foremost is that once river floods are unpredictable due to a hydrological regime that varies
with occasional long-rainy periods temporal heterogeneity should be higher than in sea-
sonal predicted flood areas and second microtopography must create spatial heterogeneity
during floods events in the local area However well-drained upper sites are probably more
affected by inherent community processes as gap-phase dynamics and direct supply rates
of light (Stevens and Carson 2002) As related by Worbes et al (1992) and Parolin (2001)
hundreds of tree species with different phenological and other ecological traits grow in
seasonal flooded forests In such cases the cyclic alternation on floods and droughts drove
species to life history behavioral and morphological adaptations (Lytle and Poff 2004)
But in the case of unpredictable floods and droughts as assigned in our study bet-hedging
strategies might be evolved for example by persistent seed bank or asynchronous
reproductive phenologies (Brock 2003) although there are no conclusive studies related to
this theory (Lytle and Poff 2004)
Allowed by a transition in the ecological groups from the lower sites to the upper ones
species richness has a maximum at the lower slope transect probably due to higher het-
erogeneity in disturbance events (space and time) and correlated variables As reported by
Desilets and Houle (2005) the spatial gradient provides some evidences for stress toler-
ance and competition as factors structuring species distribution across the topographicndash
flooding gradient also boosted by an unpredictable pattern of floods that vary in frequency
timing and magnitude Lower sectors showed predicted ecological groups already
described for these areas (Budke et al 2007 2008 Junk et al 1989 Lytle and Poff 2004)
and expected structuring changes also occurred toward upper sites where the proportion of
shade-tolerant and small trees increased as well as anemochorous trees
In summary spatialndashtemporal and environmental variables are arranging tree species
distribution across the toposequence of our study site Furthermore predicted ecological
groups reflected the dynamics of disturbance in the topographicndashhydrological gradient
Species richness and diversity also reflected such pattern and were higher in the mid-sector
where occasional floods should prevent competitive exclusion and generate high envi-
ronmental heterogeneity
Acknowledgements We are grateful to the Programa de Pos-Graduacao em Botanica of the UniversidadeFederal do Rio Grande do SulmdashUFRGS for the opportunity to undertake this study and to CAPES Agencyfor the scholarship granted to the first author Our special thanks to Diogo lsquolsquoBagualrsquorsquo Lindenmaier forfieldwork assistance and to Ricardo Braga Eduardo Rossi and colleagues of the Laboratorio de Fitoeco-logiamdashUFRGS for critiques and suggestions We also appreciated the reviewing efforts of anonymouscontributors for providing useful comments to the manuscript
References
Baldwin DS Mitchell AM (2000) The effects of drying and re-flooding on the sediment and soil nutrientdynamics of lowland river-floodplain systems a synthesis Regul River 16457ndash467 doi1010021099-1646
Biodivers Conserv (2010) 192371ndash2387 2385
123
Authors personal copy
Barroso GM Morim MP Peixoto AL Ichaso CLF (1999) Frutos e sementes morfologia aplicada a si-stematica de dicotiledoneas Editora UFV Vicosa
Borcard D Legendre P Drapeau P (1992) Partialling out the spatial component of ecological variationEcology 731045ndash1055 doi1010292006WR005044
Brock MA (2003) Drought and aquatic community resilience the role of eggs and seeds in sediments oftemporary wetlands Freshw Biol 481207ndash1218 doi101046j1365-2427200301083x
Budke JC Athayde EA Giehl ELH Zachia RA Eisinger SM (2005) Composicao florıstica e estrategias dedispersao de especies lenhosas em uma floresta ribeirinha arroio Passo das Tropas Santa Maria RSBrasil Iheringia Bot 6017ndash24
Budke JC Jarenkow JA Oliveira-Filho AT Lindenmaier DS (2006) Padroes de riqueza e diversidade emrios de pequeno porte In Mariath JEA Santos RP (eds) Os avancos da botanica no inıcio do seculoXXI SBB Porto Alegre
Budke JC Jarenkow JA Oliveira-Filho AT (2007) Relationships between tree component structuretopography and soils of a riverine forest Rio Botucaraı southern Brazil Plant Ecol 189187ndash200 doi101007s11258-006-9174-8
Budke JC Jarenkow JA Oliveira-Filho AT (2008) Tree community features of two stands of riverine forestunder different flooding regimes in southern Brazil Flora 203162ndash174 doi101016jflora200703001
Cardoso E Schiavini I (2002) Relacao entre distribuicao de especies arboreas e topografia em um gradienteflorestal na Estacao Ecologica do Panga (Uberlandia MG) Rev Bras Bot 25277ndash289
Connell JH (1978) Diversity in tropical rain forests and coral reefs Science 1991302ndash1310 doi101126science19943351302
Damasceno-Junior GA Semir J Santos FAM Leitao-Filho HF (2005) Structure distribution of species andinundation in a riparian forest of Rio Paraguai Pantanal Brazil Flora 200119ndash135 doi101016jflora200409002
Desilets P Houle G (2005) Effects of resource availability and heterogeneity on the slope of the species-areacurve along a floodplain-upland gradient J Veg Sci 16487ndash496 doi1016581100-9233
Dufrene M Legendre P (1997) Species assemblages and indicator species the need for a flexible asym-metrical approach Ecol Monogr 67345ndash366
EMBRAPA (1997) Manual de metodos de analises de solo Empresa Brasileira de Pesquisa Agropecuariaand Centro Nacional de Pesquisas de Solos Rio de Janeiro
Ferreira LV (2000) Effects of flooding duration on species richness floristic composition and forest structurein river margin habitat in Amazonian blackwater floodplain forests implications for future design ofprotected areas Biodivers Conserv 91ndash14 doi101023A1008989811637
Ferreira LV Stohlgren TJ (1999) Effects of river level fluctuation on plant species richness diversity anddistribution in a floodplain forest in Central Amazonia Oecologia 120582ndash587 doi101007s004420050893
Ferreira C Piedade MTF Franco AC Goncalves JFC Junk WJ (2009) Adaptive strategies to tolerateprolonged flooding in seedlings of floodplain and upland populations of Himatanthus sucuuba aCentral Amazon tree Aquat Bot 90246ndash252 doi101016jaquabot200810006
Gotelli NJ Colwell RK (2001) Quantifying biodiversity procedures and pitfalls in the measurement andcomparison of species richness Ecol Lett 4379ndash391 doi101046j1461-0248200100230x
Guilherme FAG Oliveira-Filho AT Appolinario V Bearzoti E (2004) Effects of flooding regime andwoody bamboos on tree community dynamics in a section of tropical semideciduous forest in south-eastern Brazil Plant Ecol 17419ndash36 doi101023BVEGE000004605197752cd
Huston M (1994) Biological diversity the coexistence of species in changing landscapes CambridgeUniversity Press Cambridge
IPAGRO (1982) Atlas agroclimatico do Rio Grande do Sul Pallotti Porto AlegreJohansson ME Nilsson C Nilsson E (1996) Do rivers function as corridors for plant dispersal J Veg Sci
7593ndash598Junk WJ Bayley PB Sparks RE (1989) The flood pulse concept in river-floodplain systems Can J Fish
Aquat Sci 106110ndash127Legendre P (1993) Spatial autocorrelationmdashtrouble or new paradigm Ecology 741659ndash1673Leinz V (1949) Contribuicao a geologia dos derrames basalticos do Rio Grande do Sul Bol Fac Filos Let
581ndash55Lytle DA Poff NL (2004) Adaptation to natural flow regimes Trends Ecol Evol 1994ndash100 doi
101016jtree200310002McCune B Mefford MJ (1997) PCndashORD Multivariate analysis of ecological data version 436 MjM
Software Design Glaneden BeachMueller-Dombois D Ellenberg H (1974) Aims and methods of vegetation ecology John Wiley New York
2386 Biodivers Conserv (2010) 192371ndash2387
123
Authors personal copy
Neiff JJ (1990) Ideas para la interpretacion ecologica del Parana Interciencia 15424ndash441Neiff JJ Neiff M (2003) Pulso software para analisis de fenomenos recurrentes httpwwwneiffcom
Cited 25 May 2009Oliveira-Filho AT Vilela EA Gavilanes ML Carvalho DA (1994) Effect of flooding regime and understory
bamboos on the physiognomy and tree species composition of a tropical semideciduous forest in south-eastern Brazil Vegetatio 11399ndash124
Oliveira-Filho AT Curi N Vilela EA Carvalho DA (2001) Variation in tree community composition andstructure with changes in soil properties within a fragment of semideciduous forest in south-easternBrazil Edinb J Bot 58139ndash158 doi101017S0960428601000506
Oliveira-Filho AT Jarenkow JA Rodal MJN (2006) Floristic relationships of seasonally dry forests ofeastern South America based on tree species distribution patterns In Pennington RT Ratter JA LewisGP (eds) Neotropical savannas and dry forests plant diversity biogeography and conservation CRCPress Boca Raton
Parolin P (2001) Morphological and physiological adjustments to waterlogging and drought in seedlings ofAmazonian floodplain trees Oecologia 128326ndash335 doi101007s004420100660
Parolin P de Simone O Haase K Waldhoff D Rottenberger S Kuhn U Kesselmeier J Kleiss B SchmidtW Piedade MTF Junk WJ (2004) Central Amazonian floodplain forests tree adaptations in a pulsingsystem Bot Rev 70357ndash380 doi1016630006-8101(2004)070[0357CAFFTA]20CO2
Pijl L (1982) Principles of dispersal in higher plants Springer New YorkPillar VD (2006) Multivariate exploratory analysis randomization testing and bootstrap resampling version
2320 Departamento de Ecologia UFRGS Porto AlegrePollock MM Naiman RJ Hanley TA (1998) Plant species richness in riparian wetlandsmdasha test of biodi-
versity theory Ecology 7994ndash105Robertson KM (2006) Distributions of tree species along point bars of 10 rivers in the south-eastern US
Coastal Plain J Biogeogr 33121ndash132 doi101111j1365-2699200501371xRosales J Petts G Knab-Vispo C (2001) Ecological gradients within the riparian forests of the lower Caura
river Venezuela Plant Ecol 152101ndash118 doi101023A1011411020040Stevens MHH Carson WP (2002) Resource quantity not resource heterogeneity maintains plant diversity
Ecol Lett 5420ndash426 doi101046j1461-0248200200333xStreck EV Kampf N Dalmolin RSD Klamt E Nascimento PC Schneider P (2002) Solos do Rio Grande do
Sul EMATERRS and UFRGS Porto AlegreSwaine MD Whitmore TC (1988) On the definition of ecological species groups in tropical rain forests
Vegetatio 7581ndash86Tabacchi E Correll DL Hauer R Pinay G Planty-Tabacchi AM Wissmar R (1998) Development
maintenance and role of riparian vegetation in the river landscape Freshw Biol 40497ndash516 doi101046j1365-2427199800381x
ter Braack CJF Smilauer P (1998) Canoco reference manual and userrsquos guide to Canoco for Windowssoftware for canonical community ordination (version 40) Microcomputer Power Ithaca
ter Braak CJF (1995) Ordination In Jongman RHG ter Braak CJF van Togeren OFR (eds) Data analysis incommunity and landscape ecology Cambridge University Press New York
Titeux N Dufrene M Jacob JP Paquay M Defourny P (2004) Multivariate analysis of fine-scale breedingbird atlas using a geographical information system and partial canonical correspondence analysisenvironmental and spatial effects J Biogeogr 311841ndash1856 doi101111j1365-2699200401125x
Turner MG Gergel SE Dixon MD Miller JR (2004) Distribution and abundance of trees in floodplainforests of the Wisconsin river environmental influences at different scales J Veg Sci 15729ndash738
Vreugdenhil SJ Kramer K Pelsma T (2006) Effects of flooding duration frequency and depth on thepresence of saplings of six woody species in north-west Europe For Ecol Manage 23647ndash55 doi101016jforeco200608329
Walker KF Sheldon F Puckridge JT (1995) A perspective on dryland river ecosystems Regul River 1185ndash104 doi101002rrr3450110108
Weiher E (2003) Species richness along multiple gradients testing a general multivariate model in oaksavannas Oikos 101311ndash316 doi101034j1600-0706200312216x
Wittmann F Junk WJ Piedade MTF (2004) The varzea forests in Amazonia flooding and the highlydynamic geomorphology interact with natural forest succession For Ecol Manage 196199ndash212 doi101016jforeco200402060
Worbes M Klinge H Revilla JD Martius C (1992) On the dynamics floristic subdivision and geographicaldistribution of Varzea forests in Central Amazonia J Veg Sci 3553ndash564
Zar JH (1996) Biostatistical analysis Prentice-Hall New Jersey
Biodivers Conserv (2010) 192371ndash2387 2387
123
Authors personal copy
regime The inundation regime also varies according to the topographical position fol-
lowing to frequently flooded forests to well drained non-flooded forests Furthermore we
investigated the relationships among tree component structure species and functional
groups distribution and spatialndashenvironmental variables We hypothesized that (1) as
flooding may gradually affect environmental heterogeneity richness and diversity will be
higher at intermediate elevations and directly associated with increased environmental
heterogeneity and (2) both species and functional groups will reflect variations in elevation
andor soil texture and chemistry
Methods
Study area
The study area is a forest remnant of ca 20 ha situated in the riparian fringes of the
Botucaraı river near its confluence with the Jacuı river (Lat 30010S Long 52470W)
(Budke et al 2007) The headwaters of the river lie in the southernmost extent (ca 650 m
asl) of the high planes region locally known as Planalto Meridional which geologically
is part of the Serra Geral formation made up of Cretaceous basalts originated from giant
lava flows that covered the sedimentary lowlands of the Parana Basin (Leinz 1949)
Downstream at its mid-course the Botucaraı river reaches the lowlands (ca 100 m asl)
and the topography is dominated by recently flood-deposited sedimentsmdashmeanders and
point bars At its lower course near the study area flooding events are enhanced by the
confluence with the stronger adjoining stream flow of the Jacuı river therefore promoting
lateral overflow According to Budke et al (2007) soils in such areas reflect not only the
geomorphic features from the basin but they also reflect flooding dynamics which fre-
quently produces non-stratified layers of fine gravel wood debris litter and sediment As a
consequence different soil profiles occur from well structured planosols in the riverside
slopes to recent deposited layers of sediment in the lowlands
The regional climate is moist subtropical without a regular dry season mean tem-
peratures ranges from 249C (hottest month) to 142C (coldest month) with high tem-
perature variation (absolutes values ranges from 42C in the summer to -3C in the
winter) mean annual rainfall is 1594 mm year-1 respectively (IPAGRO 1982) The
predominant soil is a Hydromorphic Planosol with typical stratified layers of depositional
sediments (Streck et al 2002)
Floods in the area are highly unpredictable because there is no marked seasonal rainy
period and rainfall is relatively well distributed throughout the year As a consequence
floods occur at any time of the year with duration of overflow periods varying from some
days to a few weeks (Budke et al 2008)
Regional vegetation is an extent of the Atlantic Forest Domain (Oliveira-Filho et al
2006) and includes overlapping patches of Seasonal Semideciduous Forests and Araucaria
Rain Forests at the river headwaters at Serra Geral formation Seasonal Semideciduous
Forests shows several genera of deciduous Fabaceae trees as Apuleia leiocarpa Myro-carpus frondosus Enterolobium contortisiliquun Parapiptadenia rigida and Erythrinafalcata as well as perennial ones which include Myrtaceae Lauraceae Sapotaceae and
Rubiaceae among others Canopy and emergent tree species can reach 25 m high
although mean vegetation stature is near 12ndash15 m In the lowlands of the river basin
Seasonal Semideciduous Forests is gradually changed by grasslands of the Pampa Domain
Biodivers Conserv (2010) 192371ndash2387 2373
123
Authors personal copy
(Oliveira-Filho et al 2006) and the river basin play a typical role of forest corridor toward
south reaching the Uruguay pampas as forest enclaves or galleries (Budke et al 2006)
Data collection
We carried out a tree survey in a 1 ha plot installed in a toposequence in the lowland areas
from the river margin to the lateral slope and therefore liable to different flooding regimes
The plot was divided in five 10 9 200 m transects and each transect was subdivided in
sampling units of 10 9 10 m All individual living trees having at least one stem and with
perimeter at breast height (pbh) C15 cm were sampled Voucher specimens of the different
species were collected prepared and lodged in the Herbarium ICN of the Universidade
Federal do Rio Grande do Sul (UFRGS)
A detailed topographic survey of the transects was carried out using a 10 m long water-
filled levelling hose 38 in a tape measure and a compass according to Cardoso and
Schiavini (2002) The resulting grid of vertical transects was used to produce contour maps
and to obtain the relative elevation of each sampling unit rather to the river To estimate
flooding frequency in each sampling unit we overlap their relative elevation to the
hydrometer records of the Jacuı river station (data calibrated according to topography)
Through Pulse 111 software (Neiff and Neiff 2003) we estimated the mean number of
floods per year from 1981 to 2004 and we used this variable as a pulse disturbance estimate
to sampling units (hereafter named flooding)
We collected samples of the topsoil (0ndash20 cm depth) from 15 sites distributed in dif-
ferent positions in such a way that its overall topographic variation was encompassed The
soil samples were kept in polyethylene bags and taken to the UFRGS Soil Laboratory for
chemical and textural analyses The variables were pH in water suspension levels of
potential acidity (Al H) bases saturation (V) sum of bases (S) cation exchange
capacity (CEC) organic matter (OM) and levels of clay sand and silt All procedures
followed EMBRAPA (1997) protocol In those plots without a soil subsample we
extrapolated real values by distance-proportional mean of the closest plots (ter Braak
1995) We compared the means of each soil property among transects by using one-way
ANOVA (Zar 1996)
Data analysis
Phytosociological parameters of density frequency and dominance (derived from tree
basal area) were calculated to describe tree community structure (Mueller-Dombois and
Ellenberg 1974) Frequency distributions into classes of diameter for each transect were
prepared and one-way ANOVA was used to compare transects Classes of exponentially
increasing range were used for diameters to make up for the accentuated decline in tree
frequency towards larger diameters (Oliveira-Filho et al 2001)
We applied rarefaction curves for each transect in order to analyse the range of species
richness within the toposequence The rarefaction curve technique generates expected
number of species based on the individualsrsquo density and then providing statistical
assumptions to this comparison (Gotelli and Colwell 2001) We also compared Shannon
diversity indices (H0) of each transect by bootstrap resampling tests with the software
Multiv (Pillar 2006) and depicted diversity and topography in a regression model
To verify topographical ranges of species we used an Indicator Species AnalysismdashISA
(Dufrene and Legendre 1997) which is a direct analysis of association between flooding
2374 Biodivers Conserv (2010) 192371ndash2387
123
Authors personal copy
and species distribution As the aim of this analysis was to assess the association between
species and topographyflooding it was used a non-hierarchical clustering procedure kmeans to produce k groups from the mean elevation of the original sampling units and by
using the resulting groups as the clustering factor required in the ISA (Dufrene and
Legendre 1997 Budke et al 2008) The analysis was performed in the PC-Ord program
(McCune and Mefford 1997)
We partitioned the variance of species distribution over the toposequence accounted by
spatial and environmental variables by successive partial Correspondence Canonical
Analysis (Borcard et al 1992) This approach combines three different matrices to
decompose all species variation in four components pure effect of environment pure
effect of spatial pattern combined variation of environment and spatial pattern and finally
unexplained variation Species assemblages from a determined position are affected by
surrounding sites because of contagious biotic process and environmental variables used to
describe biological processes are also neither randomly or uniformly spatially distributed
(Legendre 1993) In such case it is necessary to incorporate the spatial structure in the
modelling because the independence of observations is not respected (Legendre 1993) The
first matrix or species matrix included the abundances of all species with density C10
individuals The environmental matrix included initially all chemical and granulometric
figures the topographic variable (average elevation) and an ordinal (ranking) variable
labeled lsquolsquoflooding frequencyrsquorsquo We obtained the last variable directly from the topographic
survey summarizing flood occurrences and their intensity in each plot (Budke et al 2008)
The third matrix or spatial matrix included all terms of a polynomial function of geo-
graphical coordinates ie centers of each sampling unit and it was made by adding all
terms of a cubic trend surface regression
f x yeth THORN frac14 x y xy x2 y2 x2y xy2 x3 y3
According to Borcard et al (1992) this ensures the detection of more complex spatial
features as gaps or patches which require the quadratic and cubic terms of the coordinates
and their interactions
The variance partitioning proceeded in two steps First we extracted from each
explanatory matrix (environmental variables and spatial variables) all non-significant
variables by forward stepwise regression using Monte Carlo permutations (999 permuta-
tions P 005) with CANOCO 40 (ter Braack and Smilauer 1998) and performed two
canonical ordinations that are redundant in terms of explained variation over the species
data due to spatial structuring (Borcard et al 1992) Then two partial canonical analyses
were carried out (lsquoenvironmentalrsquo and lsquospatialrsquo) each of them constrained by one of the
sets of explanatory variables to determine the relative contribution of environmental and
spatial variables in accounting for species variation Final partition is possible by using the
sum of all canonical eigenvalues of two canonical ordinations constrained by one set of
explanatory variables and of two partial canonical ordinations each of them constrained
by one set of explanatory variables while controlling for the effect of the others (covari-
ables) (Borcard et al 1992 Titeux et al 2004)
To search for ecological differences in the toposequence we classified the species in
ecological groups of regeneration vertical distribution and dispersal We defined regen-
eration based on the categories proposed by Swaine and Whitmore (1988) The two main
levels are (a) lsquopioneerrsquo which includes the species showing an entirely heliophilous life
cycle a seed bank but no bank of juveniles and (b) lsquolate successional speciesrsquo which are
those able to germinate and establish under some degree of shade to form a bank of
Biodivers Conserv (2010) 192371ndash2387 2375
123
Authors personal copy
juveniles The later was divided into (b1) lsquoshade-tolerantrsquo and (b2) lsquolight-demanding late
successional speciesrsquo which are better seen as the two sides of a continuum of solar
radiation required by the trees to lsquoreleasersquo the bank of juveniles (Oliveira-Filho et al 1994)
We defined the vertical distribution based on the strata commonly reached by the adult
individuals (a) small tree species (b) medium tree species and (c) tall tree species (see
Oliveira-Filho et al 1994) The dispersal was (a) zoochorous species with animal-med-
iated dispersal syndrome (b) anemochorous and hydrochorous those with mechanisms to
facilitate wind-dispersal or flotation and (c) autochorous those dispersed by free fall or
ballistic mechanisms (Pijl 1982) The classification of each species into the ecological
groups was based on observations during fieldwork from 2004 to 2005 and on scientific
literature (Barroso et al 1999 Budke et al 2005 2008) We tested the distribution of trees
into frequency classes according to the ecological group by KruskalndashWallis tests (Zar
1996)
Results
River corridor along the studied area has a typical meandering system with well-defined
geomorphic features The lowest sector encompasses the levee and depression which
interacts directly with river floods Next to these sites we identified the lower-slope the
middle-slope and the ridge according to the relative elevation to the river channel
(Table 1) and these sectors corresponded to our installed transects The lower slope veg-
etation is a sharp transition between lowland and upland forests and only large inundation
floods this sector whereas upland sites present slight differences in vegetation structure
due to absence of flooding and allied effects Nevertheless there is a distinct gradient of
organic matter (OM) clay and cation exchange capacity (CEC) being higher toward upper
sites as also showed by potential acidity (Al H) (Table 1) By other hand sum of bases
(S) and phosphorus contents (P) showed a tendency of decreasing toward upper sites
(Table 1) Furthermore the variance of some soil variables was quite high and demon-
strated the high heterogeneity across transects
The field inventory yielded a total of 1229 individuals belonging to 72 species and 35
families from which Myrtaceae and Fabaceae were the richest families with 11 species
followed by Rubiaceae and Sapotaceae with four species (Table S1) Although Myrtaceae
and Fabaceae presented the highest richness both families appeared generally with low
density or basal area The stand showed a forest of low stature with most individuals
between 5 and 7 m tall and few emergent trees reaching up 15 m The diameter-class
distribution of trees revealed typical inverted-J distribution with most individuals situated
in the first two classes (Fig 1) Across the toposequence higher density was found near the
river (Levee) followed by lower density values in the depression and again an increased
density through lower and middle slope On the other hand the ridgetop transect presented
the lowest density but an increased basal area (Table 2) and several trees with diameter
[40 cm Vertical distribution of trees also showed the predominance of medium-sized
individuals followed by a decreased proportion of small and emergent trees (Fig 2A)
The proportion of light-demanding trees was higher towards the upper sites (Fig 2B)
Pioneer trees presented an opposite pattern being more abundant in low sites Shade-
tolerant trees also showed an increased density at upper sites where flooding is restrict or
absent Within the dispersal groups zoochorous trees presented higher proportion in all
transects Autochorous and hydrochorous trees decreased toward the ridgetop whereas
anemochorous trees followed the inverse pattern (Fig 2C) These structural patterns
2376 Biodivers Conserv (2010) 192371ndash2387
123
Authors personal copy
Tab
le1
So
ilv
aria
ble
so
ffi
ve
tran
sect
so
fri
ver
ine
fore
sto
nth
eB
otu
cara
ıri
ver
so
uth
ern
Bra
zil
Soil
var
iable
sL
evee
Dep
ress
ion
L-s
lope
M-s
lope
Rid
ge
FP
Rel
ativ
eel
evat
ion
(m)
38
plusmn0
5a
54
plusmn0
7a
85
plusmn2
3b
11
8plusmn
35
bc
13
6plusmn
37
c8
03
0
00
1
pH
(H2O
)4
8plusmn
04
47
plusmn0
44
7plusmn
04
47
plusmn0
64
7plusmn
06
08
50
93
ns
Pmdash
Meh
lich
(mg
dm
-3)
71
plusmn2
17
1plusmn
23
63
plusmn1
66
plusmn1
75
7plusmn
14
22
30
07
ns
K(m
gd
m-
3)
76
1plusmn
12
57
96
plusmn1
64
89
plusmn2
62
94
3plusmn
34
87
8plusmn
24
14
36
03
5n
s
Ca
(cm
olc
dm
-3)
62
plusmn4
36
8plusmn
44
7plusmn
47
61
plusmn4
57
plusmn3
73
19
05
2n
s
Mg
(cm
olc
dm
-3)
15
plusmn0
71
6plusmn
07
15
plusmn0
71
5plusmn
06
14
plusmn0
50
36
09
8n
s
Al
H
(cm
olc
dm
-3)
66
plusmn3
1a
71
plusmn3
ab8
5plusmn
4ab
96
plusmn5
5ab
10
plusmn4
2b
97
60
04
S(c
mo
lcd
m-
3)
8plusmn
48
87
plusmn5
18
1plusmn
46
84
plusmn5
17
4plusmn
41
27
70
59
ns
CE
C(c
mo
lcd
m-
3)
15
2plusmn
48
16
2plusmn
38
17
1plusmn
33
17
9plusmn
43
17
3plusmn
34
14
50
22
ns
V(
)5
57
plusmn1
71
49
7plusmn
20
24
59
plusmn2
46
45
5plusmn
23
64
18
plusmn2
07
40
60
39
ns
OM
()
26
plusmn1
1a
28
plusmn1
a3
2plusmn
1ab
37
plusmn1
1b
38
plusmn0
8b
22
9
00
01
Cla
y(
)1
37
plusmn2
4a
15
4plusmn
33
ab1
58
plusmn2
8ab
16
plusmn2
1b
15
5plusmn
17
ab2
49
00
4
San
d(
)2
0plusmn
74
23
1plusmn
22
24
2plusmn
10
12
23
plusmn7
52
23
plusmn6
28
91
00
6n
s
Sil
t(
)6
42
plusmn1
09
60
4plusmn
14
59
8plusmn
12
96
2plusmn
91
62
plusmn7
24
91
02
9n
s
Val
ues
are
mea
ns
plusmnst
and
ard
dev
iati
on
sfr
om
0to
20
cmd
epth
top
soil
sam
ple
s(N
=2
0fo
rea
chtr
anse
ct)
Dif
fere
nt
lett
ers
afte
rv
alu
esin
dic
ate
sign
ifica
nt
dif
fere
nce
sin
AN
OV
Ate
sts
(ns
=n
on
-sig
nifi
can
t)
Biodivers Conserv (2010) 192371ndash2387 2377
123
Authors personal copy
shaped the physiognomic features of different sectors that varied according to the topo-
sequence and consequently in flooding regime The depression sector presented lower
density basal area and also low tree diameters whereas the levee portion presented high
density and basal area
Species distribution across the topographic gradient is presented in Table 3 according
to the Indicator Species Analysis Some species were clearly distributed from lowland to
medium sites as Eugenia uniflora Myrciaria tenella Eugenia uruguayensis whereas
others were restricted to upland areas as Chomelia obtusa and Cordia americana Many
species did not show a specific site distribution and occurred over a wide distribution range
as Gymnanthes concolor and Casearia sylvestris
The relative elevation of each transect reflects the pattern of flooding frequency and
duration in each site then spatial aggregation of trees may indicate preferences or
restriction on the establishment of some species Typical riverine species appeared near the
river margin as Pouteria gardneriana Guettarda uruguensis and others (Table 3) whereas
typical species of well-drained forests as Sorocea bonplandii Parapiptadenia rigida and
Cupania vernalis occurred frequently in the ridgetop transect Furthermore 13 species did
Fig 1 Diameter-class distributions of trees with pbh C 15 cm surveyed in five transects of riverine foreston the Botucaraı river southern Brazil Diameter-classes are used for increasing intervals (see lsquolsquoMethodsrsquorsquosection) Bars and ranges are means and 95 confidence intervals of 100 sampling units respectively
Table 2 Density (ind ha-1) dominance (m2 ha-1) mean height (m) and mean diameter (cm) for differenttransects of the riverside forest of the Botucaraı river southern Brazil
Transect AD ADo Height Diameter
Levee 1655 plusmn 467 a 3927 plusmn 268 a 62 plusmn 25 a 1761 plusmn 1697
Depression 1005 plusmn 369 b 198 plusmn 137 b 67 plusmn 21 b 161 plusmn 1194
L-slope 1120 plusmn 443 ab 229 plusmn 165 b 71 plusmn 26 b 1547 plusmn 1302
M-slope 1415 plusmn 438 a 274 plusmn 165 b 69 plusmn 25 b 1453 plusmn 98
Ridge 950 plusmn 294 b 282 plusmn 194 b 7 plusmn 24 b 1727 plusmn 1555
ANOVA F = 107 F = 129 F = 305 F = 35
Different letters after values indicate significant differences in t tests ( P 005 P 0001)
AD density ADo dominance
2378 Biodivers Conserv (2010) 192371ndash2387
123
Authors personal copy
not present a topographic association due to wide distribution through the gradient On the
other hand the distinction among environmental and spatial effects showed that space
contributes significantly to the distribution of tree species (Fig 3) Environmental variables
selected by forward selection procedure (P 005) are summarized in Table 4 All geo-
graphical terms of the polynomial function were significant (P 005) during spatial CCA
and were add to the model The four CCA analyses provided the following results
1 CCA of the species matrix constrained by the environmental matrix sum of all
canonical eigenvalues = 0944 Monte Carlo tests for overall analysis F = 286
P 0001
2 CCA of the species matrix constrained by spatial matrix sum of all canonical
eigenvalues = 1017 Monte Carlo tests for overall analysis F = 359 P 0001
Fig 2 Ecological groups of vertical distribution (A) regeneration (B) and dispersal (C) in five transects ofriverine forest of Botucaraı river southern Brazil Pi pioneer Ld light-demanding St shade-tolerant Zoozoochorous Auto autochorous Ane anemochorous Hydro hydrochorous
Biodivers Conserv (2010) 192371ndash2387 2379
123
Authors personal copy
3 Environmental partial CCA (after removing the effect of geographical matrix) sum of
all canonical eigenvalues = 0416 Monte Carlo tests for overall analysis F = 135
P 0001
Table 3 Indicator species analysis (ISA) performed for species with density C10 individuals sampled infive transects with 20 sampling units each Botucaraı river southern Brazil
Biplots of environmental variables and species or environmental variables and sampling
units were depicted with the environmental partial analyses results (Fig 4) In this step
species distributions are purely based on speciesndashenvironment relationships The first two
axes accounted respectively for 268 (eigenvalue = 0112) and 181 (eigen-
value = 0075) of the speciesndashenvironment relationships and speciesndashenvironment corre-
lations for these axes were 0742 and 0649 (P 005) respectively The first four axes
accounted for 691 of speciesndashenvironment relationships Table 4 shows the intraset
correlations among variables and canonical coefficients with the first two axes The first
canonical axis was positively correlated with topography and several soil variables that are
Fig 3 Variation partitioning ofthe tree species matrix
Table 4 Intraset correlations among environmental variables selected for the model during lsquolsquoenviron-mentalrsquorsquo partial CCA of the species matrix and canonical coefficients of the first two axes
Environmental variables were selected by forward stepwise selection and included on the model if sig-nificant in Monte Carlo tests (P 005)
Biodivers Conserv (2010) 192371ndash2387 2381
123
Authors personal copy
influenced by flooding events Percentages of organic matter saturation of bases and cation
exchange capacity where higher through upper sites whereas aluminium contents were
higher in lower areas Thus this environmental gradient may affect tree distribution pat-
terns by restricting or facilitating species establishment As related in the Indicator
Analysis (although not accounting for spatial patterns directly) typical species of flooding
areas occurred near the levee and depression and the zonation was sharp once frequent
species of well-drained areas occurred only in the middle slope and ridgetop Sampling
units (Fig 5) also appeared distributed according to the toposequence with some over-
lapping due to species distribution
Rarefaction curves of species revealed significant differences on the expected total
number of species in each transect with higher richness within the middle slope (Fig 6)
A regression model (Fig 6) fitted diversity in a second-order polynomial regression
(y = -00586x2 03594x 291 R2 = 077) that showed the same pattern
Discussion
Environmental and spatial patterns
Tree species distribution throughout the topographical gradient indicated that both envi-
ronmental and spatial features were particularly important in predicting species and
community patterns This agrees with the well-know influence of geomorphic features and
hydrological regimes on riparian forests over different temporal and spatial scales
(Tabacchi et al 1998 Turner et al 2004 Desilets and Houle 2005) even though dis-
tinctions between environmental and spatial effects remain poorly studied (Titeux et al
Fig 4 Ordination biplotdepicting the two axes of theenvironmental partial CCA ofsampling units in a riverine forestin southern Brazil Eachsampling unit was identified bydifferent symbols according tothe respective transectEnvironmental variables arerepresented by their acronyms(see Table S1)
2382 Biodivers Conserv (2010) 192371ndash2387
123
Authors personal copy
2004) Here we employed a routine to partialling out the spatial effects on the analysis of
speciesndashenvironment relationships that also highlight the spatial component embedded in
such analysis (Legendre 1993) Once several biotic processes as growth mortality dis-
persal and predation influence the observed distributions of organisms (resulting in spatial
correlation) or if their distributions are dependent on explanatory variables which are
Fig 5 Ordination biplot depicting the two axes of the environmental partial CCA of species of a riverineforest in southern Brazil Species and environmental variables are represented by their acronyms (seeTable S1)
Fig 6 Rarefaction curves of tree species and Shannon diversity indices from five transects of riverine foreston the Botucaraı river southern Brazil Sampling units are representing different transects
Biodivers Conserv (2010) 192371ndash2387 2383
123
Authors personal copy
themselves spatially structured (Legendre 1993 Titeux et al 2004) spatial structuring is
an intrinsic component of ecosystems In our analysis lsquopurersquo spatial trends were more
attributed to species distribution than to lsquopurersquo environmental effects which link conta-
gious biological processes as important to the tree species distribution Furthermore
species and environmental data have a reasonable proportion of similar spatial structuring
identified by the largest proportion on the species variation due to spatially structured
environmental variation (1246) According to Borcard et al (1992) species and envi-
ronmental variables have in this case the same response to some common underlying
causes as the topographicndashflooding gradient In fact several studies have demonstrated the
direct effect of wetting and desiccation processes on both mineralogy and microbial
ecology of the sediment including nutrient dynamics (Baldwin and Mitchell 2000)
Once sediment or soils are submerged the inundation leads to a decrease in oxygen
contents and then resulting in progressive anaerobic conditions Rapid cycling of litter may
occur due to an increase on microbial activity which generates pulses on nutrient contents
and finally result in highly productive systems (Baldwin and Mitchell 2000) However a
negative effect is the rapid oxygen consumption which quickly leads to soil hypoxia or
anoxia When flood ends the anaerobic zones of sediments are newly oxygenated and
microbiota is replaced gradually to a new phase
Other spatially structured variables are sediment deposition and litter displacement
which are not covered in our study As demonstrated in lsquovarzearsquo forests of Amazonia
(Wittmann et al 2004) sediment deposition decrease toward upper sites and species
colonizing such lower areas show specific adaptations to the new site conditions regarding
to adventitious roots that probably offer mechanical support (Parolin et al 2004 Wittmann
et al 2004) However litter removal or deposition including seed bank may affect
directly species distribution once flooding and allied effects reallocate litter and seeds
among sites (Johansson et al 1996) Moreover studies have showed that flooding timing
frequency and magnitude can be used as indicators of sapling zonation on floodplain
forests (Vreugdenhil et al 2006)
All these processes are included in the unmeasured variables or spatially structuring
processes that have been missed by the geographical terms (Titeux et al 2004) and
accounted to the far unexplained variation (662) As also stressed by these authors the
stochastic spacendashtime fluctuations of each population the lsquounsaturationrsquo pattern (some
species do not use all suitable habitats) and species recording in not appropriated spatial
scales contribute to this unexplained variation too Notwithstanding occurrence data or
species abundances are often noisy (ter Braak 1995) and widespread in ecological studies
(Borcard et al 1992 Titeux et al 2004)
Richness and diversity patterns
Significant transitions occurred from the levee and depression to the following lower slope
transect regarding to stand structure and ecological groups Inversions on the proportion of
pioneershade-tolerant trees and auto-hydrochorous to anemochorous trees occurred in that
small transition and affected not only ecological groups but also species occurrences As a
consequence this zonation transect may consist spatially as a boundary for tolerant and
intolerant trees with regarding to flooding In fact few species occurred over the entire
flooding gradient and the lower slope also appears as an edge for several species
In a temporal scale the lower slope area will probably present more heterogeneous
spans in flooding events and it may consist in the most heterogeneous temporalndashspatial
sector across the topographic gradient which agree with the findings of Pollock et al
2384 Biodivers Conserv (2010) 192371ndash2387
123
Authors personal copy
(1998) from wetlands with different flooding regimes In the structured view of the
dynamic-equilibrium model (Huston 1994) different patches from different seral stages
result from spatial variation of disturbance frequencies If disturbance frequencies vary
over time a landscape could also contain patches of different seral stages (Pollock et al
1998) These authors used that assumption in a model with temporally synchronous dis-
turbance and found that at the community-scale level the results supported many pre-
dictions of the dynamic-equilibrium model especially regarding to species richness
In our study there are two major factors related to disturbance frequency First and
foremost is that once river floods are unpredictable due to a hydrological regime that varies
with occasional long-rainy periods temporal heterogeneity should be higher than in sea-
sonal predicted flood areas and second microtopography must create spatial heterogeneity
during floods events in the local area However well-drained upper sites are probably more
affected by inherent community processes as gap-phase dynamics and direct supply rates
of light (Stevens and Carson 2002) As related by Worbes et al (1992) and Parolin (2001)
hundreds of tree species with different phenological and other ecological traits grow in
seasonal flooded forests In such cases the cyclic alternation on floods and droughts drove
species to life history behavioral and morphological adaptations (Lytle and Poff 2004)
But in the case of unpredictable floods and droughts as assigned in our study bet-hedging
strategies might be evolved for example by persistent seed bank or asynchronous
reproductive phenologies (Brock 2003) although there are no conclusive studies related to
this theory (Lytle and Poff 2004)
Allowed by a transition in the ecological groups from the lower sites to the upper ones
species richness has a maximum at the lower slope transect probably due to higher het-
erogeneity in disturbance events (space and time) and correlated variables As reported by
Desilets and Houle (2005) the spatial gradient provides some evidences for stress toler-
ance and competition as factors structuring species distribution across the topographicndash
flooding gradient also boosted by an unpredictable pattern of floods that vary in frequency
timing and magnitude Lower sectors showed predicted ecological groups already
described for these areas (Budke et al 2007 2008 Junk et al 1989 Lytle and Poff 2004)
and expected structuring changes also occurred toward upper sites where the proportion of
shade-tolerant and small trees increased as well as anemochorous trees
In summary spatialndashtemporal and environmental variables are arranging tree species
distribution across the toposequence of our study site Furthermore predicted ecological
groups reflected the dynamics of disturbance in the topographicndashhydrological gradient
Species richness and diversity also reflected such pattern and were higher in the mid-sector
where occasional floods should prevent competitive exclusion and generate high envi-
ronmental heterogeneity
Acknowledgements We are grateful to the Programa de Pos-Graduacao em Botanica of the UniversidadeFederal do Rio Grande do SulmdashUFRGS for the opportunity to undertake this study and to CAPES Agencyfor the scholarship granted to the first author Our special thanks to Diogo lsquolsquoBagualrsquorsquo Lindenmaier forfieldwork assistance and to Ricardo Braga Eduardo Rossi and colleagues of the Laboratorio de Fitoeco-logiamdashUFRGS for critiques and suggestions We also appreciated the reviewing efforts of anonymouscontributors for providing useful comments to the manuscript
References
Baldwin DS Mitchell AM (2000) The effects of drying and re-flooding on the sediment and soil nutrientdynamics of lowland river-floodplain systems a synthesis Regul River 16457ndash467 doi1010021099-1646
Biodivers Conserv (2010) 192371ndash2387 2385
123
Authors personal copy
Barroso GM Morim MP Peixoto AL Ichaso CLF (1999) Frutos e sementes morfologia aplicada a si-stematica de dicotiledoneas Editora UFV Vicosa
Borcard D Legendre P Drapeau P (1992) Partialling out the spatial component of ecological variationEcology 731045ndash1055 doi1010292006WR005044
Brock MA (2003) Drought and aquatic community resilience the role of eggs and seeds in sediments oftemporary wetlands Freshw Biol 481207ndash1218 doi101046j1365-2427200301083x
Budke JC Athayde EA Giehl ELH Zachia RA Eisinger SM (2005) Composicao florıstica e estrategias dedispersao de especies lenhosas em uma floresta ribeirinha arroio Passo das Tropas Santa Maria RSBrasil Iheringia Bot 6017ndash24
Budke JC Jarenkow JA Oliveira-Filho AT Lindenmaier DS (2006) Padroes de riqueza e diversidade emrios de pequeno porte In Mariath JEA Santos RP (eds) Os avancos da botanica no inıcio do seculoXXI SBB Porto Alegre
Budke JC Jarenkow JA Oliveira-Filho AT (2007) Relationships between tree component structuretopography and soils of a riverine forest Rio Botucaraı southern Brazil Plant Ecol 189187ndash200 doi101007s11258-006-9174-8
Budke JC Jarenkow JA Oliveira-Filho AT (2008) Tree community features of two stands of riverine forestunder different flooding regimes in southern Brazil Flora 203162ndash174 doi101016jflora200703001
Cardoso E Schiavini I (2002) Relacao entre distribuicao de especies arboreas e topografia em um gradienteflorestal na Estacao Ecologica do Panga (Uberlandia MG) Rev Bras Bot 25277ndash289
Connell JH (1978) Diversity in tropical rain forests and coral reefs Science 1991302ndash1310 doi101126science19943351302
Damasceno-Junior GA Semir J Santos FAM Leitao-Filho HF (2005) Structure distribution of species andinundation in a riparian forest of Rio Paraguai Pantanal Brazil Flora 200119ndash135 doi101016jflora200409002
Desilets P Houle G (2005) Effects of resource availability and heterogeneity on the slope of the species-areacurve along a floodplain-upland gradient J Veg Sci 16487ndash496 doi1016581100-9233
Dufrene M Legendre P (1997) Species assemblages and indicator species the need for a flexible asym-metrical approach Ecol Monogr 67345ndash366
EMBRAPA (1997) Manual de metodos de analises de solo Empresa Brasileira de Pesquisa Agropecuariaand Centro Nacional de Pesquisas de Solos Rio de Janeiro
Ferreira LV (2000) Effects of flooding duration on species richness floristic composition and forest structurein river margin habitat in Amazonian blackwater floodplain forests implications for future design ofprotected areas Biodivers Conserv 91ndash14 doi101023A1008989811637
Ferreira LV Stohlgren TJ (1999) Effects of river level fluctuation on plant species richness diversity anddistribution in a floodplain forest in Central Amazonia Oecologia 120582ndash587 doi101007s004420050893
Ferreira C Piedade MTF Franco AC Goncalves JFC Junk WJ (2009) Adaptive strategies to tolerateprolonged flooding in seedlings of floodplain and upland populations of Himatanthus sucuuba aCentral Amazon tree Aquat Bot 90246ndash252 doi101016jaquabot200810006
Gotelli NJ Colwell RK (2001) Quantifying biodiversity procedures and pitfalls in the measurement andcomparison of species richness Ecol Lett 4379ndash391 doi101046j1461-0248200100230x
Guilherme FAG Oliveira-Filho AT Appolinario V Bearzoti E (2004) Effects of flooding regime andwoody bamboos on tree community dynamics in a section of tropical semideciduous forest in south-eastern Brazil Plant Ecol 17419ndash36 doi101023BVEGE000004605197752cd
Huston M (1994) Biological diversity the coexistence of species in changing landscapes CambridgeUniversity Press Cambridge
IPAGRO (1982) Atlas agroclimatico do Rio Grande do Sul Pallotti Porto AlegreJohansson ME Nilsson C Nilsson E (1996) Do rivers function as corridors for plant dispersal J Veg Sci
7593ndash598Junk WJ Bayley PB Sparks RE (1989) The flood pulse concept in river-floodplain systems Can J Fish
Aquat Sci 106110ndash127Legendre P (1993) Spatial autocorrelationmdashtrouble or new paradigm Ecology 741659ndash1673Leinz V (1949) Contribuicao a geologia dos derrames basalticos do Rio Grande do Sul Bol Fac Filos Let
581ndash55Lytle DA Poff NL (2004) Adaptation to natural flow regimes Trends Ecol Evol 1994ndash100 doi
101016jtree200310002McCune B Mefford MJ (1997) PCndashORD Multivariate analysis of ecological data version 436 MjM
Software Design Glaneden BeachMueller-Dombois D Ellenberg H (1974) Aims and methods of vegetation ecology John Wiley New York
2386 Biodivers Conserv (2010) 192371ndash2387
123
Authors personal copy
Neiff JJ (1990) Ideas para la interpretacion ecologica del Parana Interciencia 15424ndash441Neiff JJ Neiff M (2003) Pulso software para analisis de fenomenos recurrentes httpwwwneiffcom
Cited 25 May 2009Oliveira-Filho AT Vilela EA Gavilanes ML Carvalho DA (1994) Effect of flooding regime and understory
bamboos on the physiognomy and tree species composition of a tropical semideciduous forest in south-eastern Brazil Vegetatio 11399ndash124
Oliveira-Filho AT Curi N Vilela EA Carvalho DA (2001) Variation in tree community composition andstructure with changes in soil properties within a fragment of semideciduous forest in south-easternBrazil Edinb J Bot 58139ndash158 doi101017S0960428601000506
Oliveira-Filho AT Jarenkow JA Rodal MJN (2006) Floristic relationships of seasonally dry forests ofeastern South America based on tree species distribution patterns In Pennington RT Ratter JA LewisGP (eds) Neotropical savannas and dry forests plant diversity biogeography and conservation CRCPress Boca Raton
Parolin P (2001) Morphological and physiological adjustments to waterlogging and drought in seedlings ofAmazonian floodplain trees Oecologia 128326ndash335 doi101007s004420100660
Parolin P de Simone O Haase K Waldhoff D Rottenberger S Kuhn U Kesselmeier J Kleiss B SchmidtW Piedade MTF Junk WJ (2004) Central Amazonian floodplain forests tree adaptations in a pulsingsystem Bot Rev 70357ndash380 doi1016630006-8101(2004)070[0357CAFFTA]20CO2
Pijl L (1982) Principles of dispersal in higher plants Springer New YorkPillar VD (2006) Multivariate exploratory analysis randomization testing and bootstrap resampling version
2320 Departamento de Ecologia UFRGS Porto AlegrePollock MM Naiman RJ Hanley TA (1998) Plant species richness in riparian wetlandsmdasha test of biodi-
versity theory Ecology 7994ndash105Robertson KM (2006) Distributions of tree species along point bars of 10 rivers in the south-eastern US
Coastal Plain J Biogeogr 33121ndash132 doi101111j1365-2699200501371xRosales J Petts G Knab-Vispo C (2001) Ecological gradients within the riparian forests of the lower Caura
river Venezuela Plant Ecol 152101ndash118 doi101023A1011411020040Stevens MHH Carson WP (2002) Resource quantity not resource heterogeneity maintains plant diversity
Ecol Lett 5420ndash426 doi101046j1461-0248200200333xStreck EV Kampf N Dalmolin RSD Klamt E Nascimento PC Schneider P (2002) Solos do Rio Grande do
Sul EMATERRS and UFRGS Porto AlegreSwaine MD Whitmore TC (1988) On the definition of ecological species groups in tropical rain forests
Vegetatio 7581ndash86Tabacchi E Correll DL Hauer R Pinay G Planty-Tabacchi AM Wissmar R (1998) Development
maintenance and role of riparian vegetation in the river landscape Freshw Biol 40497ndash516 doi101046j1365-2427199800381x
ter Braack CJF Smilauer P (1998) Canoco reference manual and userrsquos guide to Canoco for Windowssoftware for canonical community ordination (version 40) Microcomputer Power Ithaca
ter Braak CJF (1995) Ordination In Jongman RHG ter Braak CJF van Togeren OFR (eds) Data analysis incommunity and landscape ecology Cambridge University Press New York
Titeux N Dufrene M Jacob JP Paquay M Defourny P (2004) Multivariate analysis of fine-scale breedingbird atlas using a geographical information system and partial canonical correspondence analysisenvironmental and spatial effects J Biogeogr 311841ndash1856 doi101111j1365-2699200401125x
Turner MG Gergel SE Dixon MD Miller JR (2004) Distribution and abundance of trees in floodplainforests of the Wisconsin river environmental influences at different scales J Veg Sci 15729ndash738
Vreugdenhil SJ Kramer K Pelsma T (2006) Effects of flooding duration frequency and depth on thepresence of saplings of six woody species in north-west Europe For Ecol Manage 23647ndash55 doi101016jforeco200608329
Walker KF Sheldon F Puckridge JT (1995) A perspective on dryland river ecosystems Regul River 1185ndash104 doi101002rrr3450110108
Weiher E (2003) Species richness along multiple gradients testing a general multivariate model in oaksavannas Oikos 101311ndash316 doi101034j1600-0706200312216x
Wittmann F Junk WJ Piedade MTF (2004) The varzea forests in Amazonia flooding and the highlydynamic geomorphology interact with natural forest succession For Ecol Manage 196199ndash212 doi101016jforeco200402060
Worbes M Klinge H Revilla JD Martius C (1992) On the dynamics floristic subdivision and geographicaldistribution of Varzea forests in Central Amazonia J Veg Sci 3553ndash564
Zar JH (1996) Biostatistical analysis Prentice-Hall New Jersey
Biodivers Conserv (2010) 192371ndash2387 2387
123
Authors personal copy
(Oliveira-Filho et al 2006) and the river basin play a typical role of forest corridor toward
south reaching the Uruguay pampas as forest enclaves or galleries (Budke et al 2006)
Data collection
We carried out a tree survey in a 1 ha plot installed in a toposequence in the lowland areas
from the river margin to the lateral slope and therefore liable to different flooding regimes
The plot was divided in five 10 9 200 m transects and each transect was subdivided in
sampling units of 10 9 10 m All individual living trees having at least one stem and with
perimeter at breast height (pbh) C15 cm were sampled Voucher specimens of the different
species were collected prepared and lodged in the Herbarium ICN of the Universidade
Federal do Rio Grande do Sul (UFRGS)
A detailed topographic survey of the transects was carried out using a 10 m long water-
filled levelling hose 38 in a tape measure and a compass according to Cardoso and
Schiavini (2002) The resulting grid of vertical transects was used to produce contour maps
and to obtain the relative elevation of each sampling unit rather to the river To estimate
flooding frequency in each sampling unit we overlap their relative elevation to the
hydrometer records of the Jacuı river station (data calibrated according to topography)
Through Pulse 111 software (Neiff and Neiff 2003) we estimated the mean number of
floods per year from 1981 to 2004 and we used this variable as a pulse disturbance estimate
to sampling units (hereafter named flooding)
We collected samples of the topsoil (0ndash20 cm depth) from 15 sites distributed in dif-
ferent positions in such a way that its overall topographic variation was encompassed The
soil samples were kept in polyethylene bags and taken to the UFRGS Soil Laboratory for
chemical and textural analyses The variables were pH in water suspension levels of
potential acidity (Al H) bases saturation (V) sum of bases (S) cation exchange
capacity (CEC) organic matter (OM) and levels of clay sand and silt All procedures
followed EMBRAPA (1997) protocol In those plots without a soil subsample we
extrapolated real values by distance-proportional mean of the closest plots (ter Braak
1995) We compared the means of each soil property among transects by using one-way
ANOVA (Zar 1996)
Data analysis
Phytosociological parameters of density frequency and dominance (derived from tree
basal area) were calculated to describe tree community structure (Mueller-Dombois and
Ellenberg 1974) Frequency distributions into classes of diameter for each transect were
prepared and one-way ANOVA was used to compare transects Classes of exponentially
increasing range were used for diameters to make up for the accentuated decline in tree
frequency towards larger diameters (Oliveira-Filho et al 2001)
We applied rarefaction curves for each transect in order to analyse the range of species
richness within the toposequence The rarefaction curve technique generates expected
number of species based on the individualsrsquo density and then providing statistical
assumptions to this comparison (Gotelli and Colwell 2001) We also compared Shannon
diversity indices (H0) of each transect by bootstrap resampling tests with the software
Multiv (Pillar 2006) and depicted diversity and topography in a regression model
To verify topographical ranges of species we used an Indicator Species AnalysismdashISA
(Dufrene and Legendre 1997) which is a direct analysis of association between flooding
2374 Biodivers Conserv (2010) 192371ndash2387
123
Authors personal copy
and species distribution As the aim of this analysis was to assess the association between
species and topographyflooding it was used a non-hierarchical clustering procedure kmeans to produce k groups from the mean elevation of the original sampling units and by
using the resulting groups as the clustering factor required in the ISA (Dufrene and
Legendre 1997 Budke et al 2008) The analysis was performed in the PC-Ord program
(McCune and Mefford 1997)
We partitioned the variance of species distribution over the toposequence accounted by
spatial and environmental variables by successive partial Correspondence Canonical
Analysis (Borcard et al 1992) This approach combines three different matrices to
decompose all species variation in four components pure effect of environment pure
effect of spatial pattern combined variation of environment and spatial pattern and finally
unexplained variation Species assemblages from a determined position are affected by
surrounding sites because of contagious biotic process and environmental variables used to
describe biological processes are also neither randomly or uniformly spatially distributed
(Legendre 1993) In such case it is necessary to incorporate the spatial structure in the
modelling because the independence of observations is not respected (Legendre 1993) The
first matrix or species matrix included the abundances of all species with density C10
individuals The environmental matrix included initially all chemical and granulometric
figures the topographic variable (average elevation) and an ordinal (ranking) variable
labeled lsquolsquoflooding frequencyrsquorsquo We obtained the last variable directly from the topographic
survey summarizing flood occurrences and their intensity in each plot (Budke et al 2008)
The third matrix or spatial matrix included all terms of a polynomial function of geo-
graphical coordinates ie centers of each sampling unit and it was made by adding all
terms of a cubic trend surface regression
f x yeth THORN frac14 x y xy x2 y2 x2y xy2 x3 y3
According to Borcard et al (1992) this ensures the detection of more complex spatial
features as gaps or patches which require the quadratic and cubic terms of the coordinates
and their interactions
The variance partitioning proceeded in two steps First we extracted from each
explanatory matrix (environmental variables and spatial variables) all non-significant
variables by forward stepwise regression using Monte Carlo permutations (999 permuta-
tions P 005) with CANOCO 40 (ter Braack and Smilauer 1998) and performed two
canonical ordinations that are redundant in terms of explained variation over the species
data due to spatial structuring (Borcard et al 1992) Then two partial canonical analyses
were carried out (lsquoenvironmentalrsquo and lsquospatialrsquo) each of them constrained by one of the
sets of explanatory variables to determine the relative contribution of environmental and
spatial variables in accounting for species variation Final partition is possible by using the
sum of all canonical eigenvalues of two canonical ordinations constrained by one set of
explanatory variables and of two partial canonical ordinations each of them constrained
by one set of explanatory variables while controlling for the effect of the others (covari-
ables) (Borcard et al 1992 Titeux et al 2004)
To search for ecological differences in the toposequence we classified the species in
ecological groups of regeneration vertical distribution and dispersal We defined regen-
eration based on the categories proposed by Swaine and Whitmore (1988) The two main
levels are (a) lsquopioneerrsquo which includes the species showing an entirely heliophilous life
cycle a seed bank but no bank of juveniles and (b) lsquolate successional speciesrsquo which are
those able to germinate and establish under some degree of shade to form a bank of
Biodivers Conserv (2010) 192371ndash2387 2375
123
Authors personal copy
juveniles The later was divided into (b1) lsquoshade-tolerantrsquo and (b2) lsquolight-demanding late
successional speciesrsquo which are better seen as the two sides of a continuum of solar
radiation required by the trees to lsquoreleasersquo the bank of juveniles (Oliveira-Filho et al 1994)
We defined the vertical distribution based on the strata commonly reached by the adult
individuals (a) small tree species (b) medium tree species and (c) tall tree species (see
Oliveira-Filho et al 1994) The dispersal was (a) zoochorous species with animal-med-
iated dispersal syndrome (b) anemochorous and hydrochorous those with mechanisms to
facilitate wind-dispersal or flotation and (c) autochorous those dispersed by free fall or
ballistic mechanisms (Pijl 1982) The classification of each species into the ecological
groups was based on observations during fieldwork from 2004 to 2005 and on scientific
literature (Barroso et al 1999 Budke et al 2005 2008) We tested the distribution of trees
into frequency classes according to the ecological group by KruskalndashWallis tests (Zar
1996)
Results
River corridor along the studied area has a typical meandering system with well-defined
geomorphic features The lowest sector encompasses the levee and depression which
interacts directly with river floods Next to these sites we identified the lower-slope the
middle-slope and the ridge according to the relative elevation to the river channel
(Table 1) and these sectors corresponded to our installed transects The lower slope veg-
etation is a sharp transition between lowland and upland forests and only large inundation
floods this sector whereas upland sites present slight differences in vegetation structure
due to absence of flooding and allied effects Nevertheless there is a distinct gradient of
organic matter (OM) clay and cation exchange capacity (CEC) being higher toward upper
sites as also showed by potential acidity (Al H) (Table 1) By other hand sum of bases
(S) and phosphorus contents (P) showed a tendency of decreasing toward upper sites
(Table 1) Furthermore the variance of some soil variables was quite high and demon-
strated the high heterogeneity across transects
The field inventory yielded a total of 1229 individuals belonging to 72 species and 35
families from which Myrtaceae and Fabaceae were the richest families with 11 species
followed by Rubiaceae and Sapotaceae with four species (Table S1) Although Myrtaceae
and Fabaceae presented the highest richness both families appeared generally with low
density or basal area The stand showed a forest of low stature with most individuals
between 5 and 7 m tall and few emergent trees reaching up 15 m The diameter-class
distribution of trees revealed typical inverted-J distribution with most individuals situated
in the first two classes (Fig 1) Across the toposequence higher density was found near the
river (Levee) followed by lower density values in the depression and again an increased
density through lower and middle slope On the other hand the ridgetop transect presented
the lowest density but an increased basal area (Table 2) and several trees with diameter
[40 cm Vertical distribution of trees also showed the predominance of medium-sized
individuals followed by a decreased proportion of small and emergent trees (Fig 2A)
The proportion of light-demanding trees was higher towards the upper sites (Fig 2B)
Pioneer trees presented an opposite pattern being more abundant in low sites Shade-
tolerant trees also showed an increased density at upper sites where flooding is restrict or
absent Within the dispersal groups zoochorous trees presented higher proportion in all
transects Autochorous and hydrochorous trees decreased toward the ridgetop whereas
anemochorous trees followed the inverse pattern (Fig 2C) These structural patterns
2376 Biodivers Conserv (2010) 192371ndash2387
123
Authors personal copy
Tab
le1
So
ilv
aria
ble
so
ffi
ve
tran
sect
so
fri
ver
ine
fore
sto
nth
eB
otu
cara
ıri
ver
so
uth
ern
Bra
zil
Soil
var
iable
sL
evee
Dep
ress
ion
L-s
lope
M-s
lope
Rid
ge
FP
Rel
ativ
eel
evat
ion
(m)
38
plusmn0
5a
54
plusmn0
7a
85
plusmn2
3b
11
8plusmn
35
bc
13
6plusmn
37
c8
03
0
00
1
pH
(H2O
)4
8plusmn
04
47
plusmn0
44
7plusmn
04
47
plusmn0
64
7plusmn
06
08
50
93
ns
Pmdash
Meh
lich
(mg
dm
-3)
71
plusmn2
17
1plusmn
23
63
plusmn1
66
plusmn1
75
7plusmn
14
22
30
07
ns
K(m
gd
m-
3)
76
1plusmn
12
57
96
plusmn1
64
89
plusmn2
62
94
3plusmn
34
87
8plusmn
24
14
36
03
5n
s
Ca
(cm
olc
dm
-3)
62
plusmn4
36
8plusmn
44
7plusmn
47
61
plusmn4
57
plusmn3
73
19
05
2n
s
Mg
(cm
olc
dm
-3)
15
plusmn0
71
6plusmn
07
15
plusmn0
71
5plusmn
06
14
plusmn0
50
36
09
8n
s
Al
H
(cm
olc
dm
-3)
66
plusmn3
1a
71
plusmn3
ab8
5plusmn
4ab
96
plusmn5
5ab
10
plusmn4
2b
97
60
04
S(c
mo
lcd
m-
3)
8plusmn
48
87
plusmn5
18
1plusmn
46
84
plusmn5
17
4plusmn
41
27
70
59
ns
CE
C(c
mo
lcd
m-
3)
15
2plusmn
48
16
2plusmn
38
17
1plusmn
33
17
9plusmn
43
17
3plusmn
34
14
50
22
ns
V(
)5
57
plusmn1
71
49
7plusmn
20
24
59
plusmn2
46
45
5plusmn
23
64
18
plusmn2
07
40
60
39
ns
OM
()
26
plusmn1
1a
28
plusmn1
a3
2plusmn
1ab
37
plusmn1
1b
38
plusmn0
8b
22
9
00
01
Cla
y(
)1
37
plusmn2
4a
15
4plusmn
33
ab1
58
plusmn2
8ab
16
plusmn2
1b
15
5plusmn
17
ab2
49
00
4
San
d(
)2
0plusmn
74
23
1plusmn
22
24
2plusmn
10
12
23
plusmn7
52
23
plusmn6
28
91
00
6n
s
Sil
t(
)6
42
plusmn1
09
60
4plusmn
14
59
8plusmn
12
96
2plusmn
91
62
plusmn7
24
91
02
9n
s
Val
ues
are
mea
ns
plusmnst
and
ard
dev
iati
on
sfr
om
0to
20
cmd
epth
top
soil
sam
ple
s(N
=2
0fo
rea
chtr
anse
ct)
Dif
fere
nt
lett
ers
afte
rv
alu
esin
dic
ate
sign
ifica
nt
dif
fere
nce
sin
AN
OV
Ate
sts
(ns
=n
on
-sig
nifi
can
t)
Biodivers Conserv (2010) 192371ndash2387 2377
123
Authors personal copy
shaped the physiognomic features of different sectors that varied according to the topo-
sequence and consequently in flooding regime The depression sector presented lower
density basal area and also low tree diameters whereas the levee portion presented high
density and basal area
Species distribution across the topographic gradient is presented in Table 3 according
to the Indicator Species Analysis Some species were clearly distributed from lowland to
medium sites as Eugenia uniflora Myrciaria tenella Eugenia uruguayensis whereas
others were restricted to upland areas as Chomelia obtusa and Cordia americana Many
species did not show a specific site distribution and occurred over a wide distribution range
as Gymnanthes concolor and Casearia sylvestris
The relative elevation of each transect reflects the pattern of flooding frequency and
duration in each site then spatial aggregation of trees may indicate preferences or
restriction on the establishment of some species Typical riverine species appeared near the
river margin as Pouteria gardneriana Guettarda uruguensis and others (Table 3) whereas
typical species of well-drained forests as Sorocea bonplandii Parapiptadenia rigida and
Cupania vernalis occurred frequently in the ridgetop transect Furthermore 13 species did
Fig 1 Diameter-class distributions of trees with pbh C 15 cm surveyed in five transects of riverine foreston the Botucaraı river southern Brazil Diameter-classes are used for increasing intervals (see lsquolsquoMethodsrsquorsquosection) Bars and ranges are means and 95 confidence intervals of 100 sampling units respectively
Table 2 Density (ind ha-1) dominance (m2 ha-1) mean height (m) and mean diameter (cm) for differenttransects of the riverside forest of the Botucaraı river southern Brazil
Transect AD ADo Height Diameter
Levee 1655 plusmn 467 a 3927 plusmn 268 a 62 plusmn 25 a 1761 plusmn 1697
Depression 1005 plusmn 369 b 198 plusmn 137 b 67 plusmn 21 b 161 plusmn 1194
L-slope 1120 plusmn 443 ab 229 plusmn 165 b 71 plusmn 26 b 1547 plusmn 1302
M-slope 1415 plusmn 438 a 274 plusmn 165 b 69 plusmn 25 b 1453 plusmn 98
Ridge 950 plusmn 294 b 282 plusmn 194 b 7 plusmn 24 b 1727 plusmn 1555
ANOVA F = 107 F = 129 F = 305 F = 35
Different letters after values indicate significant differences in t tests ( P 005 P 0001)
AD density ADo dominance
2378 Biodivers Conserv (2010) 192371ndash2387
123
Authors personal copy
not present a topographic association due to wide distribution through the gradient On the
other hand the distinction among environmental and spatial effects showed that space
contributes significantly to the distribution of tree species (Fig 3) Environmental variables
selected by forward selection procedure (P 005) are summarized in Table 4 All geo-
graphical terms of the polynomial function were significant (P 005) during spatial CCA
and were add to the model The four CCA analyses provided the following results
1 CCA of the species matrix constrained by the environmental matrix sum of all
canonical eigenvalues = 0944 Monte Carlo tests for overall analysis F = 286
P 0001
2 CCA of the species matrix constrained by spatial matrix sum of all canonical
eigenvalues = 1017 Monte Carlo tests for overall analysis F = 359 P 0001
Fig 2 Ecological groups of vertical distribution (A) regeneration (B) and dispersal (C) in five transects ofriverine forest of Botucaraı river southern Brazil Pi pioneer Ld light-demanding St shade-tolerant Zoozoochorous Auto autochorous Ane anemochorous Hydro hydrochorous
Biodivers Conserv (2010) 192371ndash2387 2379
123
Authors personal copy
3 Environmental partial CCA (after removing the effect of geographical matrix) sum of
all canonical eigenvalues = 0416 Monte Carlo tests for overall analysis F = 135
P 0001
Table 3 Indicator species analysis (ISA) performed for species with density C10 individuals sampled infive transects with 20 sampling units each Botucaraı river southern Brazil
Biplots of environmental variables and species or environmental variables and sampling
units were depicted with the environmental partial analyses results (Fig 4) In this step
species distributions are purely based on speciesndashenvironment relationships The first two
axes accounted respectively for 268 (eigenvalue = 0112) and 181 (eigen-
value = 0075) of the speciesndashenvironment relationships and speciesndashenvironment corre-
lations for these axes were 0742 and 0649 (P 005) respectively The first four axes
accounted for 691 of speciesndashenvironment relationships Table 4 shows the intraset
correlations among variables and canonical coefficients with the first two axes The first
canonical axis was positively correlated with topography and several soil variables that are
Fig 3 Variation partitioning ofthe tree species matrix
Table 4 Intraset correlations among environmental variables selected for the model during lsquolsquoenviron-mentalrsquorsquo partial CCA of the species matrix and canonical coefficients of the first two axes
Environmental variables were selected by forward stepwise selection and included on the model if sig-nificant in Monte Carlo tests (P 005)
Biodivers Conserv (2010) 192371ndash2387 2381
123
Authors personal copy
influenced by flooding events Percentages of organic matter saturation of bases and cation
exchange capacity where higher through upper sites whereas aluminium contents were
higher in lower areas Thus this environmental gradient may affect tree distribution pat-
terns by restricting or facilitating species establishment As related in the Indicator
Analysis (although not accounting for spatial patterns directly) typical species of flooding
areas occurred near the levee and depression and the zonation was sharp once frequent
species of well-drained areas occurred only in the middle slope and ridgetop Sampling
units (Fig 5) also appeared distributed according to the toposequence with some over-
lapping due to species distribution
Rarefaction curves of species revealed significant differences on the expected total
number of species in each transect with higher richness within the middle slope (Fig 6)
A regression model (Fig 6) fitted diversity in a second-order polynomial regression
(y = -00586x2 03594x 291 R2 = 077) that showed the same pattern
Discussion
Environmental and spatial patterns
Tree species distribution throughout the topographical gradient indicated that both envi-
ronmental and spatial features were particularly important in predicting species and
community patterns This agrees with the well-know influence of geomorphic features and
hydrological regimes on riparian forests over different temporal and spatial scales
(Tabacchi et al 1998 Turner et al 2004 Desilets and Houle 2005) even though dis-
tinctions between environmental and spatial effects remain poorly studied (Titeux et al
Fig 4 Ordination biplotdepicting the two axes of theenvironmental partial CCA ofsampling units in a riverine forestin southern Brazil Eachsampling unit was identified bydifferent symbols according tothe respective transectEnvironmental variables arerepresented by their acronyms(see Table S1)
2382 Biodivers Conserv (2010) 192371ndash2387
123
Authors personal copy
2004) Here we employed a routine to partialling out the spatial effects on the analysis of
speciesndashenvironment relationships that also highlight the spatial component embedded in
such analysis (Legendre 1993) Once several biotic processes as growth mortality dis-
persal and predation influence the observed distributions of organisms (resulting in spatial
correlation) or if their distributions are dependent on explanatory variables which are
Fig 5 Ordination biplot depicting the two axes of the environmental partial CCA of species of a riverineforest in southern Brazil Species and environmental variables are represented by their acronyms (seeTable S1)
Fig 6 Rarefaction curves of tree species and Shannon diversity indices from five transects of riverine foreston the Botucaraı river southern Brazil Sampling units are representing different transects
Biodivers Conserv (2010) 192371ndash2387 2383
123
Authors personal copy
themselves spatially structured (Legendre 1993 Titeux et al 2004) spatial structuring is
an intrinsic component of ecosystems In our analysis lsquopurersquo spatial trends were more
attributed to species distribution than to lsquopurersquo environmental effects which link conta-
gious biological processes as important to the tree species distribution Furthermore
species and environmental data have a reasonable proportion of similar spatial structuring
identified by the largest proportion on the species variation due to spatially structured
environmental variation (1246) According to Borcard et al (1992) species and envi-
ronmental variables have in this case the same response to some common underlying
causes as the topographicndashflooding gradient In fact several studies have demonstrated the
direct effect of wetting and desiccation processes on both mineralogy and microbial
ecology of the sediment including nutrient dynamics (Baldwin and Mitchell 2000)
Once sediment or soils are submerged the inundation leads to a decrease in oxygen
contents and then resulting in progressive anaerobic conditions Rapid cycling of litter may
occur due to an increase on microbial activity which generates pulses on nutrient contents
and finally result in highly productive systems (Baldwin and Mitchell 2000) However a
negative effect is the rapid oxygen consumption which quickly leads to soil hypoxia or
anoxia When flood ends the anaerobic zones of sediments are newly oxygenated and
microbiota is replaced gradually to a new phase
Other spatially structured variables are sediment deposition and litter displacement
which are not covered in our study As demonstrated in lsquovarzearsquo forests of Amazonia
(Wittmann et al 2004) sediment deposition decrease toward upper sites and species
colonizing such lower areas show specific adaptations to the new site conditions regarding
to adventitious roots that probably offer mechanical support (Parolin et al 2004 Wittmann
et al 2004) However litter removal or deposition including seed bank may affect
directly species distribution once flooding and allied effects reallocate litter and seeds
among sites (Johansson et al 1996) Moreover studies have showed that flooding timing
frequency and magnitude can be used as indicators of sapling zonation on floodplain
forests (Vreugdenhil et al 2006)
All these processes are included in the unmeasured variables or spatially structuring
processes that have been missed by the geographical terms (Titeux et al 2004) and
accounted to the far unexplained variation (662) As also stressed by these authors the
stochastic spacendashtime fluctuations of each population the lsquounsaturationrsquo pattern (some
species do not use all suitable habitats) and species recording in not appropriated spatial
scales contribute to this unexplained variation too Notwithstanding occurrence data or
species abundances are often noisy (ter Braak 1995) and widespread in ecological studies
(Borcard et al 1992 Titeux et al 2004)
Richness and diversity patterns
Significant transitions occurred from the levee and depression to the following lower slope
transect regarding to stand structure and ecological groups Inversions on the proportion of
pioneershade-tolerant trees and auto-hydrochorous to anemochorous trees occurred in that
small transition and affected not only ecological groups but also species occurrences As a
consequence this zonation transect may consist spatially as a boundary for tolerant and
intolerant trees with regarding to flooding In fact few species occurred over the entire
flooding gradient and the lower slope also appears as an edge for several species
In a temporal scale the lower slope area will probably present more heterogeneous
spans in flooding events and it may consist in the most heterogeneous temporalndashspatial
sector across the topographic gradient which agree with the findings of Pollock et al
2384 Biodivers Conserv (2010) 192371ndash2387
123
Authors personal copy
(1998) from wetlands with different flooding regimes In the structured view of the
dynamic-equilibrium model (Huston 1994) different patches from different seral stages
result from spatial variation of disturbance frequencies If disturbance frequencies vary
over time a landscape could also contain patches of different seral stages (Pollock et al
1998) These authors used that assumption in a model with temporally synchronous dis-
turbance and found that at the community-scale level the results supported many pre-
dictions of the dynamic-equilibrium model especially regarding to species richness
In our study there are two major factors related to disturbance frequency First and
foremost is that once river floods are unpredictable due to a hydrological regime that varies
with occasional long-rainy periods temporal heterogeneity should be higher than in sea-
sonal predicted flood areas and second microtopography must create spatial heterogeneity
during floods events in the local area However well-drained upper sites are probably more
affected by inherent community processes as gap-phase dynamics and direct supply rates
of light (Stevens and Carson 2002) As related by Worbes et al (1992) and Parolin (2001)
hundreds of tree species with different phenological and other ecological traits grow in
seasonal flooded forests In such cases the cyclic alternation on floods and droughts drove
species to life history behavioral and morphological adaptations (Lytle and Poff 2004)
But in the case of unpredictable floods and droughts as assigned in our study bet-hedging
strategies might be evolved for example by persistent seed bank or asynchronous
reproductive phenologies (Brock 2003) although there are no conclusive studies related to
this theory (Lytle and Poff 2004)
Allowed by a transition in the ecological groups from the lower sites to the upper ones
species richness has a maximum at the lower slope transect probably due to higher het-
erogeneity in disturbance events (space and time) and correlated variables As reported by
Desilets and Houle (2005) the spatial gradient provides some evidences for stress toler-
ance and competition as factors structuring species distribution across the topographicndash
flooding gradient also boosted by an unpredictable pattern of floods that vary in frequency
timing and magnitude Lower sectors showed predicted ecological groups already
described for these areas (Budke et al 2007 2008 Junk et al 1989 Lytle and Poff 2004)
and expected structuring changes also occurred toward upper sites where the proportion of
shade-tolerant and small trees increased as well as anemochorous trees
In summary spatialndashtemporal and environmental variables are arranging tree species
distribution across the toposequence of our study site Furthermore predicted ecological
groups reflected the dynamics of disturbance in the topographicndashhydrological gradient
Species richness and diversity also reflected such pattern and were higher in the mid-sector
where occasional floods should prevent competitive exclusion and generate high envi-
ronmental heterogeneity
Acknowledgements We are grateful to the Programa de Pos-Graduacao em Botanica of the UniversidadeFederal do Rio Grande do SulmdashUFRGS for the opportunity to undertake this study and to CAPES Agencyfor the scholarship granted to the first author Our special thanks to Diogo lsquolsquoBagualrsquorsquo Lindenmaier forfieldwork assistance and to Ricardo Braga Eduardo Rossi and colleagues of the Laboratorio de Fitoeco-logiamdashUFRGS for critiques and suggestions We also appreciated the reviewing efforts of anonymouscontributors for providing useful comments to the manuscript
References
Baldwin DS Mitchell AM (2000) The effects of drying and re-flooding on the sediment and soil nutrientdynamics of lowland river-floodplain systems a synthesis Regul River 16457ndash467 doi1010021099-1646
Biodivers Conserv (2010) 192371ndash2387 2385
123
Authors personal copy
Barroso GM Morim MP Peixoto AL Ichaso CLF (1999) Frutos e sementes morfologia aplicada a si-stematica de dicotiledoneas Editora UFV Vicosa
Borcard D Legendre P Drapeau P (1992) Partialling out the spatial component of ecological variationEcology 731045ndash1055 doi1010292006WR005044
Brock MA (2003) Drought and aquatic community resilience the role of eggs and seeds in sediments oftemporary wetlands Freshw Biol 481207ndash1218 doi101046j1365-2427200301083x
Budke JC Athayde EA Giehl ELH Zachia RA Eisinger SM (2005) Composicao florıstica e estrategias dedispersao de especies lenhosas em uma floresta ribeirinha arroio Passo das Tropas Santa Maria RSBrasil Iheringia Bot 6017ndash24
Budke JC Jarenkow JA Oliveira-Filho AT Lindenmaier DS (2006) Padroes de riqueza e diversidade emrios de pequeno porte In Mariath JEA Santos RP (eds) Os avancos da botanica no inıcio do seculoXXI SBB Porto Alegre
Budke JC Jarenkow JA Oliveira-Filho AT (2007) Relationships between tree component structuretopography and soils of a riverine forest Rio Botucaraı southern Brazil Plant Ecol 189187ndash200 doi101007s11258-006-9174-8
Budke JC Jarenkow JA Oliveira-Filho AT (2008) Tree community features of two stands of riverine forestunder different flooding regimes in southern Brazil Flora 203162ndash174 doi101016jflora200703001
Cardoso E Schiavini I (2002) Relacao entre distribuicao de especies arboreas e topografia em um gradienteflorestal na Estacao Ecologica do Panga (Uberlandia MG) Rev Bras Bot 25277ndash289
Connell JH (1978) Diversity in tropical rain forests and coral reefs Science 1991302ndash1310 doi101126science19943351302
Damasceno-Junior GA Semir J Santos FAM Leitao-Filho HF (2005) Structure distribution of species andinundation in a riparian forest of Rio Paraguai Pantanal Brazil Flora 200119ndash135 doi101016jflora200409002
Desilets P Houle G (2005) Effects of resource availability and heterogeneity on the slope of the species-areacurve along a floodplain-upland gradient J Veg Sci 16487ndash496 doi1016581100-9233
Dufrene M Legendre P (1997) Species assemblages and indicator species the need for a flexible asym-metrical approach Ecol Monogr 67345ndash366
EMBRAPA (1997) Manual de metodos de analises de solo Empresa Brasileira de Pesquisa Agropecuariaand Centro Nacional de Pesquisas de Solos Rio de Janeiro
Ferreira LV (2000) Effects of flooding duration on species richness floristic composition and forest structurein river margin habitat in Amazonian blackwater floodplain forests implications for future design ofprotected areas Biodivers Conserv 91ndash14 doi101023A1008989811637
Ferreira LV Stohlgren TJ (1999) Effects of river level fluctuation on plant species richness diversity anddistribution in a floodplain forest in Central Amazonia Oecologia 120582ndash587 doi101007s004420050893
Ferreira C Piedade MTF Franco AC Goncalves JFC Junk WJ (2009) Adaptive strategies to tolerateprolonged flooding in seedlings of floodplain and upland populations of Himatanthus sucuuba aCentral Amazon tree Aquat Bot 90246ndash252 doi101016jaquabot200810006
Gotelli NJ Colwell RK (2001) Quantifying biodiversity procedures and pitfalls in the measurement andcomparison of species richness Ecol Lett 4379ndash391 doi101046j1461-0248200100230x
Guilherme FAG Oliveira-Filho AT Appolinario V Bearzoti E (2004) Effects of flooding regime andwoody bamboos on tree community dynamics in a section of tropical semideciduous forest in south-eastern Brazil Plant Ecol 17419ndash36 doi101023BVEGE000004605197752cd
Huston M (1994) Biological diversity the coexistence of species in changing landscapes CambridgeUniversity Press Cambridge
IPAGRO (1982) Atlas agroclimatico do Rio Grande do Sul Pallotti Porto AlegreJohansson ME Nilsson C Nilsson E (1996) Do rivers function as corridors for plant dispersal J Veg Sci
7593ndash598Junk WJ Bayley PB Sparks RE (1989) The flood pulse concept in river-floodplain systems Can J Fish
Aquat Sci 106110ndash127Legendre P (1993) Spatial autocorrelationmdashtrouble or new paradigm Ecology 741659ndash1673Leinz V (1949) Contribuicao a geologia dos derrames basalticos do Rio Grande do Sul Bol Fac Filos Let
581ndash55Lytle DA Poff NL (2004) Adaptation to natural flow regimes Trends Ecol Evol 1994ndash100 doi
101016jtree200310002McCune B Mefford MJ (1997) PCndashORD Multivariate analysis of ecological data version 436 MjM
Software Design Glaneden BeachMueller-Dombois D Ellenberg H (1974) Aims and methods of vegetation ecology John Wiley New York
2386 Biodivers Conserv (2010) 192371ndash2387
123
Authors personal copy
Neiff JJ (1990) Ideas para la interpretacion ecologica del Parana Interciencia 15424ndash441Neiff JJ Neiff M (2003) Pulso software para analisis de fenomenos recurrentes httpwwwneiffcom
Cited 25 May 2009Oliveira-Filho AT Vilela EA Gavilanes ML Carvalho DA (1994) Effect of flooding regime and understory
bamboos on the physiognomy and tree species composition of a tropical semideciduous forest in south-eastern Brazil Vegetatio 11399ndash124
Oliveira-Filho AT Curi N Vilela EA Carvalho DA (2001) Variation in tree community composition andstructure with changes in soil properties within a fragment of semideciduous forest in south-easternBrazil Edinb J Bot 58139ndash158 doi101017S0960428601000506
Oliveira-Filho AT Jarenkow JA Rodal MJN (2006) Floristic relationships of seasonally dry forests ofeastern South America based on tree species distribution patterns In Pennington RT Ratter JA LewisGP (eds) Neotropical savannas and dry forests plant diversity biogeography and conservation CRCPress Boca Raton
Parolin P (2001) Morphological and physiological adjustments to waterlogging and drought in seedlings ofAmazonian floodplain trees Oecologia 128326ndash335 doi101007s004420100660
Parolin P de Simone O Haase K Waldhoff D Rottenberger S Kuhn U Kesselmeier J Kleiss B SchmidtW Piedade MTF Junk WJ (2004) Central Amazonian floodplain forests tree adaptations in a pulsingsystem Bot Rev 70357ndash380 doi1016630006-8101(2004)070[0357CAFFTA]20CO2
Pijl L (1982) Principles of dispersal in higher plants Springer New YorkPillar VD (2006) Multivariate exploratory analysis randomization testing and bootstrap resampling version
2320 Departamento de Ecologia UFRGS Porto AlegrePollock MM Naiman RJ Hanley TA (1998) Plant species richness in riparian wetlandsmdasha test of biodi-
versity theory Ecology 7994ndash105Robertson KM (2006) Distributions of tree species along point bars of 10 rivers in the south-eastern US
Coastal Plain J Biogeogr 33121ndash132 doi101111j1365-2699200501371xRosales J Petts G Knab-Vispo C (2001) Ecological gradients within the riparian forests of the lower Caura
river Venezuela Plant Ecol 152101ndash118 doi101023A1011411020040Stevens MHH Carson WP (2002) Resource quantity not resource heterogeneity maintains plant diversity
Ecol Lett 5420ndash426 doi101046j1461-0248200200333xStreck EV Kampf N Dalmolin RSD Klamt E Nascimento PC Schneider P (2002) Solos do Rio Grande do
Sul EMATERRS and UFRGS Porto AlegreSwaine MD Whitmore TC (1988) On the definition of ecological species groups in tropical rain forests
Vegetatio 7581ndash86Tabacchi E Correll DL Hauer R Pinay G Planty-Tabacchi AM Wissmar R (1998) Development
maintenance and role of riparian vegetation in the river landscape Freshw Biol 40497ndash516 doi101046j1365-2427199800381x
ter Braack CJF Smilauer P (1998) Canoco reference manual and userrsquos guide to Canoco for Windowssoftware for canonical community ordination (version 40) Microcomputer Power Ithaca
ter Braak CJF (1995) Ordination In Jongman RHG ter Braak CJF van Togeren OFR (eds) Data analysis incommunity and landscape ecology Cambridge University Press New York
Titeux N Dufrene M Jacob JP Paquay M Defourny P (2004) Multivariate analysis of fine-scale breedingbird atlas using a geographical information system and partial canonical correspondence analysisenvironmental and spatial effects J Biogeogr 311841ndash1856 doi101111j1365-2699200401125x
Turner MG Gergel SE Dixon MD Miller JR (2004) Distribution and abundance of trees in floodplainforests of the Wisconsin river environmental influences at different scales J Veg Sci 15729ndash738
Vreugdenhil SJ Kramer K Pelsma T (2006) Effects of flooding duration frequency and depth on thepresence of saplings of six woody species in north-west Europe For Ecol Manage 23647ndash55 doi101016jforeco200608329
Walker KF Sheldon F Puckridge JT (1995) A perspective on dryland river ecosystems Regul River 1185ndash104 doi101002rrr3450110108
Weiher E (2003) Species richness along multiple gradients testing a general multivariate model in oaksavannas Oikos 101311ndash316 doi101034j1600-0706200312216x
Wittmann F Junk WJ Piedade MTF (2004) The varzea forests in Amazonia flooding and the highlydynamic geomorphology interact with natural forest succession For Ecol Manage 196199ndash212 doi101016jforeco200402060
Worbes M Klinge H Revilla JD Martius C (1992) On the dynamics floristic subdivision and geographicaldistribution of Varzea forests in Central Amazonia J Veg Sci 3553ndash564
Zar JH (1996) Biostatistical analysis Prentice-Hall New Jersey
Biodivers Conserv (2010) 192371ndash2387 2387
123
Authors personal copy
and species distribution As the aim of this analysis was to assess the association between
species and topographyflooding it was used a non-hierarchical clustering procedure kmeans to produce k groups from the mean elevation of the original sampling units and by
using the resulting groups as the clustering factor required in the ISA (Dufrene and
Legendre 1997 Budke et al 2008) The analysis was performed in the PC-Ord program
(McCune and Mefford 1997)
We partitioned the variance of species distribution over the toposequence accounted by
spatial and environmental variables by successive partial Correspondence Canonical
Analysis (Borcard et al 1992) This approach combines three different matrices to
decompose all species variation in four components pure effect of environment pure
effect of spatial pattern combined variation of environment and spatial pattern and finally
unexplained variation Species assemblages from a determined position are affected by
surrounding sites because of contagious biotic process and environmental variables used to
describe biological processes are also neither randomly or uniformly spatially distributed
(Legendre 1993) In such case it is necessary to incorporate the spatial structure in the
modelling because the independence of observations is not respected (Legendre 1993) The
first matrix or species matrix included the abundances of all species with density C10
individuals The environmental matrix included initially all chemical and granulometric
figures the topographic variable (average elevation) and an ordinal (ranking) variable
labeled lsquolsquoflooding frequencyrsquorsquo We obtained the last variable directly from the topographic
survey summarizing flood occurrences and their intensity in each plot (Budke et al 2008)
The third matrix or spatial matrix included all terms of a polynomial function of geo-
graphical coordinates ie centers of each sampling unit and it was made by adding all
terms of a cubic trend surface regression
f x yeth THORN frac14 x y xy x2 y2 x2y xy2 x3 y3
According to Borcard et al (1992) this ensures the detection of more complex spatial
features as gaps or patches which require the quadratic and cubic terms of the coordinates
and their interactions
The variance partitioning proceeded in two steps First we extracted from each
explanatory matrix (environmental variables and spatial variables) all non-significant
variables by forward stepwise regression using Monte Carlo permutations (999 permuta-
tions P 005) with CANOCO 40 (ter Braack and Smilauer 1998) and performed two
canonical ordinations that are redundant in terms of explained variation over the species
data due to spatial structuring (Borcard et al 1992) Then two partial canonical analyses
were carried out (lsquoenvironmentalrsquo and lsquospatialrsquo) each of them constrained by one of the
sets of explanatory variables to determine the relative contribution of environmental and
spatial variables in accounting for species variation Final partition is possible by using the
sum of all canonical eigenvalues of two canonical ordinations constrained by one set of
explanatory variables and of two partial canonical ordinations each of them constrained
by one set of explanatory variables while controlling for the effect of the others (covari-
ables) (Borcard et al 1992 Titeux et al 2004)
To search for ecological differences in the toposequence we classified the species in
ecological groups of regeneration vertical distribution and dispersal We defined regen-
eration based on the categories proposed by Swaine and Whitmore (1988) The two main
levels are (a) lsquopioneerrsquo which includes the species showing an entirely heliophilous life
cycle a seed bank but no bank of juveniles and (b) lsquolate successional speciesrsquo which are
those able to germinate and establish under some degree of shade to form a bank of
Biodivers Conserv (2010) 192371ndash2387 2375
123
Authors personal copy
juveniles The later was divided into (b1) lsquoshade-tolerantrsquo and (b2) lsquolight-demanding late
successional speciesrsquo which are better seen as the two sides of a continuum of solar
radiation required by the trees to lsquoreleasersquo the bank of juveniles (Oliveira-Filho et al 1994)
We defined the vertical distribution based on the strata commonly reached by the adult
individuals (a) small tree species (b) medium tree species and (c) tall tree species (see
Oliveira-Filho et al 1994) The dispersal was (a) zoochorous species with animal-med-
iated dispersal syndrome (b) anemochorous and hydrochorous those with mechanisms to
facilitate wind-dispersal or flotation and (c) autochorous those dispersed by free fall or
ballistic mechanisms (Pijl 1982) The classification of each species into the ecological
groups was based on observations during fieldwork from 2004 to 2005 and on scientific
literature (Barroso et al 1999 Budke et al 2005 2008) We tested the distribution of trees
into frequency classes according to the ecological group by KruskalndashWallis tests (Zar
1996)
Results
River corridor along the studied area has a typical meandering system with well-defined
geomorphic features The lowest sector encompasses the levee and depression which
interacts directly with river floods Next to these sites we identified the lower-slope the
middle-slope and the ridge according to the relative elevation to the river channel
(Table 1) and these sectors corresponded to our installed transects The lower slope veg-
etation is a sharp transition between lowland and upland forests and only large inundation
floods this sector whereas upland sites present slight differences in vegetation structure
due to absence of flooding and allied effects Nevertheless there is a distinct gradient of
organic matter (OM) clay and cation exchange capacity (CEC) being higher toward upper
sites as also showed by potential acidity (Al H) (Table 1) By other hand sum of bases
(S) and phosphorus contents (P) showed a tendency of decreasing toward upper sites
(Table 1) Furthermore the variance of some soil variables was quite high and demon-
strated the high heterogeneity across transects
The field inventory yielded a total of 1229 individuals belonging to 72 species and 35
families from which Myrtaceae and Fabaceae were the richest families with 11 species
followed by Rubiaceae and Sapotaceae with four species (Table S1) Although Myrtaceae
and Fabaceae presented the highest richness both families appeared generally with low
density or basal area The stand showed a forest of low stature with most individuals
between 5 and 7 m tall and few emergent trees reaching up 15 m The diameter-class
distribution of trees revealed typical inverted-J distribution with most individuals situated
in the first two classes (Fig 1) Across the toposequence higher density was found near the
river (Levee) followed by lower density values in the depression and again an increased
density through lower and middle slope On the other hand the ridgetop transect presented
the lowest density but an increased basal area (Table 2) and several trees with diameter
[40 cm Vertical distribution of trees also showed the predominance of medium-sized
individuals followed by a decreased proportion of small and emergent trees (Fig 2A)
The proportion of light-demanding trees was higher towards the upper sites (Fig 2B)
Pioneer trees presented an opposite pattern being more abundant in low sites Shade-
tolerant trees also showed an increased density at upper sites where flooding is restrict or
absent Within the dispersal groups zoochorous trees presented higher proportion in all
transects Autochorous and hydrochorous trees decreased toward the ridgetop whereas
anemochorous trees followed the inverse pattern (Fig 2C) These structural patterns
2376 Biodivers Conserv (2010) 192371ndash2387
123
Authors personal copy
Tab
le1
So
ilv
aria
ble
so
ffi
ve
tran
sect
so
fri
ver
ine
fore
sto
nth
eB
otu
cara
ıri
ver
so
uth
ern
Bra
zil
Soil
var
iable
sL
evee
Dep
ress
ion
L-s
lope
M-s
lope
Rid
ge
FP
Rel
ativ
eel
evat
ion
(m)
38
plusmn0
5a
54
plusmn0
7a
85
plusmn2
3b
11
8plusmn
35
bc
13
6plusmn
37
c8
03
0
00
1
pH
(H2O
)4
8plusmn
04
47
plusmn0
44
7plusmn
04
47
plusmn0
64
7plusmn
06
08
50
93
ns
Pmdash
Meh
lich
(mg
dm
-3)
71
plusmn2
17
1plusmn
23
63
plusmn1
66
plusmn1
75
7plusmn
14
22
30
07
ns
K(m
gd
m-
3)
76
1plusmn
12
57
96
plusmn1
64
89
plusmn2
62
94
3plusmn
34
87
8plusmn
24
14
36
03
5n
s
Ca
(cm
olc
dm
-3)
62
plusmn4
36
8plusmn
44
7plusmn
47
61
plusmn4
57
plusmn3
73
19
05
2n
s
Mg
(cm
olc
dm
-3)
15
plusmn0
71
6plusmn
07
15
plusmn0
71
5plusmn
06
14
plusmn0
50
36
09
8n
s
Al
H
(cm
olc
dm
-3)
66
plusmn3
1a
71
plusmn3
ab8
5plusmn
4ab
96
plusmn5
5ab
10
plusmn4
2b
97
60
04
S(c
mo
lcd
m-
3)
8plusmn
48
87
plusmn5
18
1plusmn
46
84
plusmn5
17
4plusmn
41
27
70
59
ns
CE
C(c
mo
lcd
m-
3)
15
2plusmn
48
16
2plusmn
38
17
1plusmn
33
17
9plusmn
43
17
3plusmn
34
14
50
22
ns
V(
)5
57
plusmn1
71
49
7plusmn
20
24
59
plusmn2
46
45
5plusmn
23
64
18
plusmn2
07
40
60
39
ns
OM
()
26
plusmn1
1a
28
plusmn1
a3
2plusmn
1ab
37
plusmn1
1b
38
plusmn0
8b
22
9
00
01
Cla
y(
)1
37
plusmn2
4a
15
4plusmn
33
ab1
58
plusmn2
8ab
16
plusmn2
1b
15
5plusmn
17
ab2
49
00
4
San
d(
)2
0plusmn
74
23
1plusmn
22
24
2plusmn
10
12
23
plusmn7
52
23
plusmn6
28
91
00
6n
s
Sil
t(
)6
42
plusmn1
09
60
4plusmn
14
59
8plusmn
12
96
2plusmn
91
62
plusmn7
24
91
02
9n
s
Val
ues
are
mea
ns
plusmnst
and
ard
dev
iati
on
sfr
om
0to
20
cmd
epth
top
soil
sam
ple
s(N
=2
0fo
rea
chtr
anse
ct)
Dif
fere
nt
lett
ers
afte
rv
alu
esin
dic
ate
sign
ifica
nt
dif
fere
nce
sin
AN
OV
Ate
sts
(ns
=n
on
-sig
nifi
can
t)
Biodivers Conserv (2010) 192371ndash2387 2377
123
Authors personal copy
shaped the physiognomic features of different sectors that varied according to the topo-
sequence and consequently in flooding regime The depression sector presented lower
density basal area and also low tree diameters whereas the levee portion presented high
density and basal area
Species distribution across the topographic gradient is presented in Table 3 according
to the Indicator Species Analysis Some species were clearly distributed from lowland to
medium sites as Eugenia uniflora Myrciaria tenella Eugenia uruguayensis whereas
others were restricted to upland areas as Chomelia obtusa and Cordia americana Many
species did not show a specific site distribution and occurred over a wide distribution range
as Gymnanthes concolor and Casearia sylvestris
The relative elevation of each transect reflects the pattern of flooding frequency and
duration in each site then spatial aggregation of trees may indicate preferences or
restriction on the establishment of some species Typical riverine species appeared near the
river margin as Pouteria gardneriana Guettarda uruguensis and others (Table 3) whereas
typical species of well-drained forests as Sorocea bonplandii Parapiptadenia rigida and
Cupania vernalis occurred frequently in the ridgetop transect Furthermore 13 species did
Fig 1 Diameter-class distributions of trees with pbh C 15 cm surveyed in five transects of riverine foreston the Botucaraı river southern Brazil Diameter-classes are used for increasing intervals (see lsquolsquoMethodsrsquorsquosection) Bars and ranges are means and 95 confidence intervals of 100 sampling units respectively
Table 2 Density (ind ha-1) dominance (m2 ha-1) mean height (m) and mean diameter (cm) for differenttransects of the riverside forest of the Botucaraı river southern Brazil
Transect AD ADo Height Diameter
Levee 1655 plusmn 467 a 3927 plusmn 268 a 62 plusmn 25 a 1761 plusmn 1697
Depression 1005 plusmn 369 b 198 plusmn 137 b 67 plusmn 21 b 161 plusmn 1194
L-slope 1120 plusmn 443 ab 229 plusmn 165 b 71 plusmn 26 b 1547 plusmn 1302
M-slope 1415 plusmn 438 a 274 plusmn 165 b 69 plusmn 25 b 1453 plusmn 98
Ridge 950 plusmn 294 b 282 plusmn 194 b 7 plusmn 24 b 1727 plusmn 1555
ANOVA F = 107 F = 129 F = 305 F = 35
Different letters after values indicate significant differences in t tests ( P 005 P 0001)
AD density ADo dominance
2378 Biodivers Conserv (2010) 192371ndash2387
123
Authors personal copy
not present a topographic association due to wide distribution through the gradient On the
other hand the distinction among environmental and spatial effects showed that space
contributes significantly to the distribution of tree species (Fig 3) Environmental variables
selected by forward selection procedure (P 005) are summarized in Table 4 All geo-
graphical terms of the polynomial function were significant (P 005) during spatial CCA
and were add to the model The four CCA analyses provided the following results
1 CCA of the species matrix constrained by the environmental matrix sum of all
canonical eigenvalues = 0944 Monte Carlo tests for overall analysis F = 286
P 0001
2 CCA of the species matrix constrained by spatial matrix sum of all canonical
eigenvalues = 1017 Monte Carlo tests for overall analysis F = 359 P 0001
Fig 2 Ecological groups of vertical distribution (A) regeneration (B) and dispersal (C) in five transects ofriverine forest of Botucaraı river southern Brazil Pi pioneer Ld light-demanding St shade-tolerant Zoozoochorous Auto autochorous Ane anemochorous Hydro hydrochorous
Biodivers Conserv (2010) 192371ndash2387 2379
123
Authors personal copy
3 Environmental partial CCA (after removing the effect of geographical matrix) sum of
all canonical eigenvalues = 0416 Monte Carlo tests for overall analysis F = 135
P 0001
Table 3 Indicator species analysis (ISA) performed for species with density C10 individuals sampled infive transects with 20 sampling units each Botucaraı river southern Brazil
Biplots of environmental variables and species or environmental variables and sampling
units were depicted with the environmental partial analyses results (Fig 4) In this step
species distributions are purely based on speciesndashenvironment relationships The first two
axes accounted respectively for 268 (eigenvalue = 0112) and 181 (eigen-
value = 0075) of the speciesndashenvironment relationships and speciesndashenvironment corre-
lations for these axes were 0742 and 0649 (P 005) respectively The first four axes
accounted for 691 of speciesndashenvironment relationships Table 4 shows the intraset
correlations among variables and canonical coefficients with the first two axes The first
canonical axis was positively correlated with topography and several soil variables that are
Fig 3 Variation partitioning ofthe tree species matrix
Table 4 Intraset correlations among environmental variables selected for the model during lsquolsquoenviron-mentalrsquorsquo partial CCA of the species matrix and canonical coefficients of the first two axes
Environmental variables were selected by forward stepwise selection and included on the model if sig-nificant in Monte Carlo tests (P 005)
Biodivers Conserv (2010) 192371ndash2387 2381
123
Authors personal copy
influenced by flooding events Percentages of organic matter saturation of bases and cation
exchange capacity where higher through upper sites whereas aluminium contents were
higher in lower areas Thus this environmental gradient may affect tree distribution pat-
terns by restricting or facilitating species establishment As related in the Indicator
Analysis (although not accounting for spatial patterns directly) typical species of flooding
areas occurred near the levee and depression and the zonation was sharp once frequent
species of well-drained areas occurred only in the middle slope and ridgetop Sampling
units (Fig 5) also appeared distributed according to the toposequence with some over-
lapping due to species distribution
Rarefaction curves of species revealed significant differences on the expected total
number of species in each transect with higher richness within the middle slope (Fig 6)
A regression model (Fig 6) fitted diversity in a second-order polynomial regression
(y = -00586x2 03594x 291 R2 = 077) that showed the same pattern
Discussion
Environmental and spatial patterns
Tree species distribution throughout the topographical gradient indicated that both envi-
ronmental and spatial features were particularly important in predicting species and
community patterns This agrees with the well-know influence of geomorphic features and
hydrological regimes on riparian forests over different temporal and spatial scales
(Tabacchi et al 1998 Turner et al 2004 Desilets and Houle 2005) even though dis-
tinctions between environmental and spatial effects remain poorly studied (Titeux et al
Fig 4 Ordination biplotdepicting the two axes of theenvironmental partial CCA ofsampling units in a riverine forestin southern Brazil Eachsampling unit was identified bydifferent symbols according tothe respective transectEnvironmental variables arerepresented by their acronyms(see Table S1)
2382 Biodivers Conserv (2010) 192371ndash2387
123
Authors personal copy
2004) Here we employed a routine to partialling out the spatial effects on the analysis of
speciesndashenvironment relationships that also highlight the spatial component embedded in
such analysis (Legendre 1993) Once several biotic processes as growth mortality dis-
persal and predation influence the observed distributions of organisms (resulting in spatial
correlation) or if their distributions are dependent on explanatory variables which are
Fig 5 Ordination biplot depicting the two axes of the environmental partial CCA of species of a riverineforest in southern Brazil Species and environmental variables are represented by their acronyms (seeTable S1)
Fig 6 Rarefaction curves of tree species and Shannon diversity indices from five transects of riverine foreston the Botucaraı river southern Brazil Sampling units are representing different transects
Biodivers Conserv (2010) 192371ndash2387 2383
123
Authors personal copy
themselves spatially structured (Legendre 1993 Titeux et al 2004) spatial structuring is
an intrinsic component of ecosystems In our analysis lsquopurersquo spatial trends were more
attributed to species distribution than to lsquopurersquo environmental effects which link conta-
gious biological processes as important to the tree species distribution Furthermore
species and environmental data have a reasonable proportion of similar spatial structuring
identified by the largest proportion on the species variation due to spatially structured
environmental variation (1246) According to Borcard et al (1992) species and envi-
ronmental variables have in this case the same response to some common underlying
causes as the topographicndashflooding gradient In fact several studies have demonstrated the
direct effect of wetting and desiccation processes on both mineralogy and microbial
ecology of the sediment including nutrient dynamics (Baldwin and Mitchell 2000)
Once sediment or soils are submerged the inundation leads to a decrease in oxygen
contents and then resulting in progressive anaerobic conditions Rapid cycling of litter may
occur due to an increase on microbial activity which generates pulses on nutrient contents
and finally result in highly productive systems (Baldwin and Mitchell 2000) However a
negative effect is the rapid oxygen consumption which quickly leads to soil hypoxia or
anoxia When flood ends the anaerobic zones of sediments are newly oxygenated and
microbiota is replaced gradually to a new phase
Other spatially structured variables are sediment deposition and litter displacement
which are not covered in our study As demonstrated in lsquovarzearsquo forests of Amazonia
(Wittmann et al 2004) sediment deposition decrease toward upper sites and species
colonizing such lower areas show specific adaptations to the new site conditions regarding
to adventitious roots that probably offer mechanical support (Parolin et al 2004 Wittmann
et al 2004) However litter removal or deposition including seed bank may affect
directly species distribution once flooding and allied effects reallocate litter and seeds
among sites (Johansson et al 1996) Moreover studies have showed that flooding timing
frequency and magnitude can be used as indicators of sapling zonation on floodplain
forests (Vreugdenhil et al 2006)
All these processes are included in the unmeasured variables or spatially structuring
processes that have been missed by the geographical terms (Titeux et al 2004) and
accounted to the far unexplained variation (662) As also stressed by these authors the
stochastic spacendashtime fluctuations of each population the lsquounsaturationrsquo pattern (some
species do not use all suitable habitats) and species recording in not appropriated spatial
scales contribute to this unexplained variation too Notwithstanding occurrence data or
species abundances are often noisy (ter Braak 1995) and widespread in ecological studies
(Borcard et al 1992 Titeux et al 2004)
Richness and diversity patterns
Significant transitions occurred from the levee and depression to the following lower slope
transect regarding to stand structure and ecological groups Inversions on the proportion of
pioneershade-tolerant trees and auto-hydrochorous to anemochorous trees occurred in that
small transition and affected not only ecological groups but also species occurrences As a
consequence this zonation transect may consist spatially as a boundary for tolerant and
intolerant trees with regarding to flooding In fact few species occurred over the entire
flooding gradient and the lower slope also appears as an edge for several species
In a temporal scale the lower slope area will probably present more heterogeneous
spans in flooding events and it may consist in the most heterogeneous temporalndashspatial
sector across the topographic gradient which agree with the findings of Pollock et al
2384 Biodivers Conserv (2010) 192371ndash2387
123
Authors personal copy
(1998) from wetlands with different flooding regimes In the structured view of the
dynamic-equilibrium model (Huston 1994) different patches from different seral stages
result from spatial variation of disturbance frequencies If disturbance frequencies vary
over time a landscape could also contain patches of different seral stages (Pollock et al
1998) These authors used that assumption in a model with temporally synchronous dis-
turbance and found that at the community-scale level the results supported many pre-
dictions of the dynamic-equilibrium model especially regarding to species richness
In our study there are two major factors related to disturbance frequency First and
foremost is that once river floods are unpredictable due to a hydrological regime that varies
with occasional long-rainy periods temporal heterogeneity should be higher than in sea-
sonal predicted flood areas and second microtopography must create spatial heterogeneity
during floods events in the local area However well-drained upper sites are probably more
affected by inherent community processes as gap-phase dynamics and direct supply rates
of light (Stevens and Carson 2002) As related by Worbes et al (1992) and Parolin (2001)
hundreds of tree species with different phenological and other ecological traits grow in
seasonal flooded forests In such cases the cyclic alternation on floods and droughts drove
species to life history behavioral and morphological adaptations (Lytle and Poff 2004)
But in the case of unpredictable floods and droughts as assigned in our study bet-hedging
strategies might be evolved for example by persistent seed bank or asynchronous
reproductive phenologies (Brock 2003) although there are no conclusive studies related to
this theory (Lytle and Poff 2004)
Allowed by a transition in the ecological groups from the lower sites to the upper ones
species richness has a maximum at the lower slope transect probably due to higher het-
erogeneity in disturbance events (space and time) and correlated variables As reported by
Desilets and Houle (2005) the spatial gradient provides some evidences for stress toler-
ance and competition as factors structuring species distribution across the topographicndash
flooding gradient also boosted by an unpredictable pattern of floods that vary in frequency
timing and magnitude Lower sectors showed predicted ecological groups already
described for these areas (Budke et al 2007 2008 Junk et al 1989 Lytle and Poff 2004)
and expected structuring changes also occurred toward upper sites where the proportion of
shade-tolerant and small trees increased as well as anemochorous trees
In summary spatialndashtemporal and environmental variables are arranging tree species
distribution across the toposequence of our study site Furthermore predicted ecological
groups reflected the dynamics of disturbance in the topographicndashhydrological gradient
Species richness and diversity also reflected such pattern and were higher in the mid-sector
where occasional floods should prevent competitive exclusion and generate high envi-
ronmental heterogeneity
Acknowledgements We are grateful to the Programa de Pos-Graduacao em Botanica of the UniversidadeFederal do Rio Grande do SulmdashUFRGS for the opportunity to undertake this study and to CAPES Agencyfor the scholarship granted to the first author Our special thanks to Diogo lsquolsquoBagualrsquorsquo Lindenmaier forfieldwork assistance and to Ricardo Braga Eduardo Rossi and colleagues of the Laboratorio de Fitoeco-logiamdashUFRGS for critiques and suggestions We also appreciated the reviewing efforts of anonymouscontributors for providing useful comments to the manuscript
References
Baldwin DS Mitchell AM (2000) The effects of drying and re-flooding on the sediment and soil nutrientdynamics of lowland river-floodplain systems a synthesis Regul River 16457ndash467 doi1010021099-1646
Biodivers Conserv (2010) 192371ndash2387 2385
123
Authors personal copy
Barroso GM Morim MP Peixoto AL Ichaso CLF (1999) Frutos e sementes morfologia aplicada a si-stematica de dicotiledoneas Editora UFV Vicosa
Borcard D Legendre P Drapeau P (1992) Partialling out the spatial component of ecological variationEcology 731045ndash1055 doi1010292006WR005044
Brock MA (2003) Drought and aquatic community resilience the role of eggs and seeds in sediments oftemporary wetlands Freshw Biol 481207ndash1218 doi101046j1365-2427200301083x
Budke JC Athayde EA Giehl ELH Zachia RA Eisinger SM (2005) Composicao florıstica e estrategias dedispersao de especies lenhosas em uma floresta ribeirinha arroio Passo das Tropas Santa Maria RSBrasil Iheringia Bot 6017ndash24
Budke JC Jarenkow JA Oliveira-Filho AT Lindenmaier DS (2006) Padroes de riqueza e diversidade emrios de pequeno porte In Mariath JEA Santos RP (eds) Os avancos da botanica no inıcio do seculoXXI SBB Porto Alegre
Budke JC Jarenkow JA Oliveira-Filho AT (2007) Relationships between tree component structuretopography and soils of a riverine forest Rio Botucaraı southern Brazil Plant Ecol 189187ndash200 doi101007s11258-006-9174-8
Budke JC Jarenkow JA Oliveira-Filho AT (2008) Tree community features of two stands of riverine forestunder different flooding regimes in southern Brazil Flora 203162ndash174 doi101016jflora200703001
Cardoso E Schiavini I (2002) Relacao entre distribuicao de especies arboreas e topografia em um gradienteflorestal na Estacao Ecologica do Panga (Uberlandia MG) Rev Bras Bot 25277ndash289
Connell JH (1978) Diversity in tropical rain forests and coral reefs Science 1991302ndash1310 doi101126science19943351302
Damasceno-Junior GA Semir J Santos FAM Leitao-Filho HF (2005) Structure distribution of species andinundation in a riparian forest of Rio Paraguai Pantanal Brazil Flora 200119ndash135 doi101016jflora200409002
Desilets P Houle G (2005) Effects of resource availability and heterogeneity on the slope of the species-areacurve along a floodplain-upland gradient J Veg Sci 16487ndash496 doi1016581100-9233
Dufrene M Legendre P (1997) Species assemblages and indicator species the need for a flexible asym-metrical approach Ecol Monogr 67345ndash366
EMBRAPA (1997) Manual de metodos de analises de solo Empresa Brasileira de Pesquisa Agropecuariaand Centro Nacional de Pesquisas de Solos Rio de Janeiro
Ferreira LV (2000) Effects of flooding duration on species richness floristic composition and forest structurein river margin habitat in Amazonian blackwater floodplain forests implications for future design ofprotected areas Biodivers Conserv 91ndash14 doi101023A1008989811637
Ferreira LV Stohlgren TJ (1999) Effects of river level fluctuation on plant species richness diversity anddistribution in a floodplain forest in Central Amazonia Oecologia 120582ndash587 doi101007s004420050893
Ferreira C Piedade MTF Franco AC Goncalves JFC Junk WJ (2009) Adaptive strategies to tolerateprolonged flooding in seedlings of floodplain and upland populations of Himatanthus sucuuba aCentral Amazon tree Aquat Bot 90246ndash252 doi101016jaquabot200810006
Gotelli NJ Colwell RK (2001) Quantifying biodiversity procedures and pitfalls in the measurement andcomparison of species richness Ecol Lett 4379ndash391 doi101046j1461-0248200100230x
Guilherme FAG Oliveira-Filho AT Appolinario V Bearzoti E (2004) Effects of flooding regime andwoody bamboos on tree community dynamics in a section of tropical semideciduous forest in south-eastern Brazil Plant Ecol 17419ndash36 doi101023BVEGE000004605197752cd
Huston M (1994) Biological diversity the coexistence of species in changing landscapes CambridgeUniversity Press Cambridge
IPAGRO (1982) Atlas agroclimatico do Rio Grande do Sul Pallotti Porto AlegreJohansson ME Nilsson C Nilsson E (1996) Do rivers function as corridors for plant dispersal J Veg Sci
7593ndash598Junk WJ Bayley PB Sparks RE (1989) The flood pulse concept in river-floodplain systems Can J Fish
Aquat Sci 106110ndash127Legendre P (1993) Spatial autocorrelationmdashtrouble or new paradigm Ecology 741659ndash1673Leinz V (1949) Contribuicao a geologia dos derrames basalticos do Rio Grande do Sul Bol Fac Filos Let
581ndash55Lytle DA Poff NL (2004) Adaptation to natural flow regimes Trends Ecol Evol 1994ndash100 doi
101016jtree200310002McCune B Mefford MJ (1997) PCndashORD Multivariate analysis of ecological data version 436 MjM
Software Design Glaneden BeachMueller-Dombois D Ellenberg H (1974) Aims and methods of vegetation ecology John Wiley New York
2386 Biodivers Conserv (2010) 192371ndash2387
123
Authors personal copy
Neiff JJ (1990) Ideas para la interpretacion ecologica del Parana Interciencia 15424ndash441Neiff JJ Neiff M (2003) Pulso software para analisis de fenomenos recurrentes httpwwwneiffcom
Cited 25 May 2009Oliveira-Filho AT Vilela EA Gavilanes ML Carvalho DA (1994) Effect of flooding regime and understory
bamboos on the physiognomy and tree species composition of a tropical semideciduous forest in south-eastern Brazil Vegetatio 11399ndash124
Oliveira-Filho AT Curi N Vilela EA Carvalho DA (2001) Variation in tree community composition andstructure with changes in soil properties within a fragment of semideciduous forest in south-easternBrazil Edinb J Bot 58139ndash158 doi101017S0960428601000506
Oliveira-Filho AT Jarenkow JA Rodal MJN (2006) Floristic relationships of seasonally dry forests ofeastern South America based on tree species distribution patterns In Pennington RT Ratter JA LewisGP (eds) Neotropical savannas and dry forests plant diversity biogeography and conservation CRCPress Boca Raton
Parolin P (2001) Morphological and physiological adjustments to waterlogging and drought in seedlings ofAmazonian floodplain trees Oecologia 128326ndash335 doi101007s004420100660
Parolin P de Simone O Haase K Waldhoff D Rottenberger S Kuhn U Kesselmeier J Kleiss B SchmidtW Piedade MTF Junk WJ (2004) Central Amazonian floodplain forests tree adaptations in a pulsingsystem Bot Rev 70357ndash380 doi1016630006-8101(2004)070[0357CAFFTA]20CO2
Pijl L (1982) Principles of dispersal in higher plants Springer New YorkPillar VD (2006) Multivariate exploratory analysis randomization testing and bootstrap resampling version
2320 Departamento de Ecologia UFRGS Porto AlegrePollock MM Naiman RJ Hanley TA (1998) Plant species richness in riparian wetlandsmdasha test of biodi-
versity theory Ecology 7994ndash105Robertson KM (2006) Distributions of tree species along point bars of 10 rivers in the south-eastern US
Coastal Plain J Biogeogr 33121ndash132 doi101111j1365-2699200501371xRosales J Petts G Knab-Vispo C (2001) Ecological gradients within the riparian forests of the lower Caura
river Venezuela Plant Ecol 152101ndash118 doi101023A1011411020040Stevens MHH Carson WP (2002) Resource quantity not resource heterogeneity maintains plant diversity
Ecol Lett 5420ndash426 doi101046j1461-0248200200333xStreck EV Kampf N Dalmolin RSD Klamt E Nascimento PC Schneider P (2002) Solos do Rio Grande do
Sul EMATERRS and UFRGS Porto AlegreSwaine MD Whitmore TC (1988) On the definition of ecological species groups in tropical rain forests
Vegetatio 7581ndash86Tabacchi E Correll DL Hauer R Pinay G Planty-Tabacchi AM Wissmar R (1998) Development
maintenance and role of riparian vegetation in the river landscape Freshw Biol 40497ndash516 doi101046j1365-2427199800381x
ter Braack CJF Smilauer P (1998) Canoco reference manual and userrsquos guide to Canoco for Windowssoftware for canonical community ordination (version 40) Microcomputer Power Ithaca
ter Braak CJF (1995) Ordination In Jongman RHG ter Braak CJF van Togeren OFR (eds) Data analysis incommunity and landscape ecology Cambridge University Press New York
Titeux N Dufrene M Jacob JP Paquay M Defourny P (2004) Multivariate analysis of fine-scale breedingbird atlas using a geographical information system and partial canonical correspondence analysisenvironmental and spatial effects J Biogeogr 311841ndash1856 doi101111j1365-2699200401125x
Turner MG Gergel SE Dixon MD Miller JR (2004) Distribution and abundance of trees in floodplainforests of the Wisconsin river environmental influences at different scales J Veg Sci 15729ndash738
Vreugdenhil SJ Kramer K Pelsma T (2006) Effects of flooding duration frequency and depth on thepresence of saplings of six woody species in north-west Europe For Ecol Manage 23647ndash55 doi101016jforeco200608329
Walker KF Sheldon F Puckridge JT (1995) A perspective on dryland river ecosystems Regul River 1185ndash104 doi101002rrr3450110108
Weiher E (2003) Species richness along multiple gradients testing a general multivariate model in oaksavannas Oikos 101311ndash316 doi101034j1600-0706200312216x
Wittmann F Junk WJ Piedade MTF (2004) The varzea forests in Amazonia flooding and the highlydynamic geomorphology interact with natural forest succession For Ecol Manage 196199ndash212 doi101016jforeco200402060
Worbes M Klinge H Revilla JD Martius C (1992) On the dynamics floristic subdivision and geographicaldistribution of Varzea forests in Central Amazonia J Veg Sci 3553ndash564
Zar JH (1996) Biostatistical analysis Prentice-Hall New Jersey
Biodivers Conserv (2010) 192371ndash2387 2387
123
Authors personal copy
juveniles The later was divided into (b1) lsquoshade-tolerantrsquo and (b2) lsquolight-demanding late
successional speciesrsquo which are better seen as the two sides of a continuum of solar
radiation required by the trees to lsquoreleasersquo the bank of juveniles (Oliveira-Filho et al 1994)
We defined the vertical distribution based on the strata commonly reached by the adult
individuals (a) small tree species (b) medium tree species and (c) tall tree species (see
Oliveira-Filho et al 1994) The dispersal was (a) zoochorous species with animal-med-
iated dispersal syndrome (b) anemochorous and hydrochorous those with mechanisms to
facilitate wind-dispersal or flotation and (c) autochorous those dispersed by free fall or
ballistic mechanisms (Pijl 1982) The classification of each species into the ecological
groups was based on observations during fieldwork from 2004 to 2005 and on scientific
literature (Barroso et al 1999 Budke et al 2005 2008) We tested the distribution of trees
into frequency classes according to the ecological group by KruskalndashWallis tests (Zar
1996)
Results
River corridor along the studied area has a typical meandering system with well-defined
geomorphic features The lowest sector encompasses the levee and depression which
interacts directly with river floods Next to these sites we identified the lower-slope the
middle-slope and the ridge according to the relative elevation to the river channel
(Table 1) and these sectors corresponded to our installed transects The lower slope veg-
etation is a sharp transition between lowland and upland forests and only large inundation
floods this sector whereas upland sites present slight differences in vegetation structure
due to absence of flooding and allied effects Nevertheless there is a distinct gradient of
organic matter (OM) clay and cation exchange capacity (CEC) being higher toward upper
sites as also showed by potential acidity (Al H) (Table 1) By other hand sum of bases
(S) and phosphorus contents (P) showed a tendency of decreasing toward upper sites
(Table 1) Furthermore the variance of some soil variables was quite high and demon-
strated the high heterogeneity across transects
The field inventory yielded a total of 1229 individuals belonging to 72 species and 35
families from which Myrtaceae and Fabaceae were the richest families with 11 species
followed by Rubiaceae and Sapotaceae with four species (Table S1) Although Myrtaceae
and Fabaceae presented the highest richness both families appeared generally with low
density or basal area The stand showed a forest of low stature with most individuals
between 5 and 7 m tall and few emergent trees reaching up 15 m The diameter-class
distribution of trees revealed typical inverted-J distribution with most individuals situated
in the first two classes (Fig 1) Across the toposequence higher density was found near the
river (Levee) followed by lower density values in the depression and again an increased
density through lower and middle slope On the other hand the ridgetop transect presented
the lowest density but an increased basal area (Table 2) and several trees with diameter
[40 cm Vertical distribution of trees also showed the predominance of medium-sized
individuals followed by a decreased proportion of small and emergent trees (Fig 2A)
The proportion of light-demanding trees was higher towards the upper sites (Fig 2B)
Pioneer trees presented an opposite pattern being more abundant in low sites Shade-
tolerant trees also showed an increased density at upper sites where flooding is restrict or
absent Within the dispersal groups zoochorous trees presented higher proportion in all
transects Autochorous and hydrochorous trees decreased toward the ridgetop whereas
anemochorous trees followed the inverse pattern (Fig 2C) These structural patterns
2376 Biodivers Conserv (2010) 192371ndash2387
123
Authors personal copy
Tab
le1
So
ilv
aria
ble
so
ffi
ve
tran
sect
so
fri
ver
ine
fore
sto
nth
eB
otu
cara
ıri
ver
so
uth
ern
Bra
zil
Soil
var
iable
sL
evee
Dep
ress
ion
L-s
lope
M-s
lope
Rid
ge
FP
Rel
ativ
eel
evat
ion
(m)
38
plusmn0
5a
54
plusmn0
7a
85
plusmn2
3b
11
8plusmn
35
bc
13
6plusmn
37
c8
03
0
00
1
pH
(H2O
)4
8plusmn
04
47
plusmn0
44
7plusmn
04
47
plusmn0
64
7plusmn
06
08
50
93
ns
Pmdash
Meh
lich
(mg
dm
-3)
71
plusmn2
17
1plusmn
23
63
plusmn1
66
plusmn1
75
7plusmn
14
22
30
07
ns
K(m
gd
m-
3)
76
1plusmn
12
57
96
plusmn1
64
89
plusmn2
62
94
3plusmn
34
87
8plusmn
24
14
36
03
5n
s
Ca
(cm
olc
dm
-3)
62
plusmn4
36
8plusmn
44
7plusmn
47
61
plusmn4
57
plusmn3
73
19
05
2n
s
Mg
(cm
olc
dm
-3)
15
plusmn0
71
6plusmn
07
15
plusmn0
71
5plusmn
06
14
plusmn0
50
36
09
8n
s
Al
H
(cm
olc
dm
-3)
66
plusmn3
1a
71
plusmn3
ab8
5plusmn
4ab
96
plusmn5
5ab
10
plusmn4
2b
97
60
04
S(c
mo
lcd
m-
3)
8plusmn
48
87
plusmn5
18
1plusmn
46
84
plusmn5
17
4plusmn
41
27
70
59
ns
CE
C(c
mo
lcd
m-
3)
15
2plusmn
48
16
2plusmn
38
17
1plusmn
33
17
9plusmn
43
17
3plusmn
34
14
50
22
ns
V(
)5
57
plusmn1
71
49
7plusmn
20
24
59
plusmn2
46
45
5plusmn
23
64
18
plusmn2
07
40
60
39
ns
OM
()
26
plusmn1
1a
28
plusmn1
a3
2plusmn
1ab
37
plusmn1
1b
38
plusmn0
8b
22
9
00
01
Cla
y(
)1
37
plusmn2
4a
15
4plusmn
33
ab1
58
plusmn2
8ab
16
plusmn2
1b
15
5plusmn
17
ab2
49
00
4
San
d(
)2
0plusmn
74
23
1plusmn
22
24
2plusmn
10
12
23
plusmn7
52
23
plusmn6
28
91
00
6n
s
Sil
t(
)6
42
plusmn1
09
60
4plusmn
14
59
8plusmn
12
96
2plusmn
91
62
plusmn7
24
91
02
9n
s
Val
ues
are
mea
ns
plusmnst
and
ard
dev
iati
on
sfr
om
0to
20
cmd
epth
top
soil
sam
ple
s(N
=2
0fo
rea
chtr
anse
ct)
Dif
fere
nt
lett
ers
afte
rv
alu
esin
dic
ate
sign
ifica
nt
dif
fere
nce
sin
AN
OV
Ate
sts
(ns
=n
on
-sig
nifi
can
t)
Biodivers Conserv (2010) 192371ndash2387 2377
123
Authors personal copy
shaped the physiognomic features of different sectors that varied according to the topo-
sequence and consequently in flooding regime The depression sector presented lower
density basal area and also low tree diameters whereas the levee portion presented high
density and basal area
Species distribution across the topographic gradient is presented in Table 3 according
to the Indicator Species Analysis Some species were clearly distributed from lowland to
medium sites as Eugenia uniflora Myrciaria tenella Eugenia uruguayensis whereas
others were restricted to upland areas as Chomelia obtusa and Cordia americana Many
species did not show a specific site distribution and occurred over a wide distribution range
as Gymnanthes concolor and Casearia sylvestris
The relative elevation of each transect reflects the pattern of flooding frequency and
duration in each site then spatial aggregation of trees may indicate preferences or
restriction on the establishment of some species Typical riverine species appeared near the
river margin as Pouteria gardneriana Guettarda uruguensis and others (Table 3) whereas
typical species of well-drained forests as Sorocea bonplandii Parapiptadenia rigida and
Cupania vernalis occurred frequently in the ridgetop transect Furthermore 13 species did
Fig 1 Diameter-class distributions of trees with pbh C 15 cm surveyed in five transects of riverine foreston the Botucaraı river southern Brazil Diameter-classes are used for increasing intervals (see lsquolsquoMethodsrsquorsquosection) Bars and ranges are means and 95 confidence intervals of 100 sampling units respectively
Table 2 Density (ind ha-1) dominance (m2 ha-1) mean height (m) and mean diameter (cm) for differenttransects of the riverside forest of the Botucaraı river southern Brazil
Transect AD ADo Height Diameter
Levee 1655 plusmn 467 a 3927 plusmn 268 a 62 plusmn 25 a 1761 plusmn 1697
Depression 1005 plusmn 369 b 198 plusmn 137 b 67 plusmn 21 b 161 plusmn 1194
L-slope 1120 plusmn 443 ab 229 plusmn 165 b 71 plusmn 26 b 1547 plusmn 1302
M-slope 1415 plusmn 438 a 274 plusmn 165 b 69 plusmn 25 b 1453 plusmn 98
Ridge 950 plusmn 294 b 282 plusmn 194 b 7 plusmn 24 b 1727 plusmn 1555
ANOVA F = 107 F = 129 F = 305 F = 35
Different letters after values indicate significant differences in t tests ( P 005 P 0001)
AD density ADo dominance
2378 Biodivers Conserv (2010) 192371ndash2387
123
Authors personal copy
not present a topographic association due to wide distribution through the gradient On the
other hand the distinction among environmental and spatial effects showed that space
contributes significantly to the distribution of tree species (Fig 3) Environmental variables
selected by forward selection procedure (P 005) are summarized in Table 4 All geo-
graphical terms of the polynomial function were significant (P 005) during spatial CCA
and were add to the model The four CCA analyses provided the following results
1 CCA of the species matrix constrained by the environmental matrix sum of all
canonical eigenvalues = 0944 Monte Carlo tests for overall analysis F = 286
P 0001
2 CCA of the species matrix constrained by spatial matrix sum of all canonical
eigenvalues = 1017 Monte Carlo tests for overall analysis F = 359 P 0001
Fig 2 Ecological groups of vertical distribution (A) regeneration (B) and dispersal (C) in five transects ofriverine forest of Botucaraı river southern Brazil Pi pioneer Ld light-demanding St shade-tolerant Zoozoochorous Auto autochorous Ane anemochorous Hydro hydrochorous
Biodivers Conserv (2010) 192371ndash2387 2379
123
Authors personal copy
3 Environmental partial CCA (after removing the effect of geographical matrix) sum of
all canonical eigenvalues = 0416 Monte Carlo tests for overall analysis F = 135
P 0001
Table 3 Indicator species analysis (ISA) performed for species with density C10 individuals sampled infive transects with 20 sampling units each Botucaraı river southern Brazil
Biplots of environmental variables and species or environmental variables and sampling
units were depicted with the environmental partial analyses results (Fig 4) In this step
species distributions are purely based on speciesndashenvironment relationships The first two
axes accounted respectively for 268 (eigenvalue = 0112) and 181 (eigen-
value = 0075) of the speciesndashenvironment relationships and speciesndashenvironment corre-
lations for these axes were 0742 and 0649 (P 005) respectively The first four axes
accounted for 691 of speciesndashenvironment relationships Table 4 shows the intraset
correlations among variables and canonical coefficients with the first two axes The first
canonical axis was positively correlated with topography and several soil variables that are
Fig 3 Variation partitioning ofthe tree species matrix
Table 4 Intraset correlations among environmental variables selected for the model during lsquolsquoenviron-mentalrsquorsquo partial CCA of the species matrix and canonical coefficients of the first two axes
Environmental variables were selected by forward stepwise selection and included on the model if sig-nificant in Monte Carlo tests (P 005)
Biodivers Conserv (2010) 192371ndash2387 2381
123
Authors personal copy
influenced by flooding events Percentages of organic matter saturation of bases and cation
exchange capacity where higher through upper sites whereas aluminium contents were
higher in lower areas Thus this environmental gradient may affect tree distribution pat-
terns by restricting or facilitating species establishment As related in the Indicator
Analysis (although not accounting for spatial patterns directly) typical species of flooding
areas occurred near the levee and depression and the zonation was sharp once frequent
species of well-drained areas occurred only in the middle slope and ridgetop Sampling
units (Fig 5) also appeared distributed according to the toposequence with some over-
lapping due to species distribution
Rarefaction curves of species revealed significant differences on the expected total
number of species in each transect with higher richness within the middle slope (Fig 6)
A regression model (Fig 6) fitted diversity in a second-order polynomial regression
(y = -00586x2 03594x 291 R2 = 077) that showed the same pattern
Discussion
Environmental and spatial patterns
Tree species distribution throughout the topographical gradient indicated that both envi-
ronmental and spatial features were particularly important in predicting species and
community patterns This agrees with the well-know influence of geomorphic features and
hydrological regimes on riparian forests over different temporal and spatial scales
(Tabacchi et al 1998 Turner et al 2004 Desilets and Houle 2005) even though dis-
tinctions between environmental and spatial effects remain poorly studied (Titeux et al
Fig 4 Ordination biplotdepicting the two axes of theenvironmental partial CCA ofsampling units in a riverine forestin southern Brazil Eachsampling unit was identified bydifferent symbols according tothe respective transectEnvironmental variables arerepresented by their acronyms(see Table S1)
2382 Biodivers Conserv (2010) 192371ndash2387
123
Authors personal copy
2004) Here we employed a routine to partialling out the spatial effects on the analysis of
speciesndashenvironment relationships that also highlight the spatial component embedded in
such analysis (Legendre 1993) Once several biotic processes as growth mortality dis-
persal and predation influence the observed distributions of organisms (resulting in spatial
correlation) or if their distributions are dependent on explanatory variables which are
Fig 5 Ordination biplot depicting the two axes of the environmental partial CCA of species of a riverineforest in southern Brazil Species and environmental variables are represented by their acronyms (seeTable S1)
Fig 6 Rarefaction curves of tree species and Shannon diversity indices from five transects of riverine foreston the Botucaraı river southern Brazil Sampling units are representing different transects
Biodivers Conserv (2010) 192371ndash2387 2383
123
Authors personal copy
themselves spatially structured (Legendre 1993 Titeux et al 2004) spatial structuring is
an intrinsic component of ecosystems In our analysis lsquopurersquo spatial trends were more
attributed to species distribution than to lsquopurersquo environmental effects which link conta-
gious biological processes as important to the tree species distribution Furthermore
species and environmental data have a reasonable proportion of similar spatial structuring
identified by the largest proportion on the species variation due to spatially structured
environmental variation (1246) According to Borcard et al (1992) species and envi-
ronmental variables have in this case the same response to some common underlying
causes as the topographicndashflooding gradient In fact several studies have demonstrated the
direct effect of wetting and desiccation processes on both mineralogy and microbial
ecology of the sediment including nutrient dynamics (Baldwin and Mitchell 2000)
Once sediment or soils are submerged the inundation leads to a decrease in oxygen
contents and then resulting in progressive anaerobic conditions Rapid cycling of litter may
occur due to an increase on microbial activity which generates pulses on nutrient contents
and finally result in highly productive systems (Baldwin and Mitchell 2000) However a
negative effect is the rapid oxygen consumption which quickly leads to soil hypoxia or
anoxia When flood ends the anaerobic zones of sediments are newly oxygenated and
microbiota is replaced gradually to a new phase
Other spatially structured variables are sediment deposition and litter displacement
which are not covered in our study As demonstrated in lsquovarzearsquo forests of Amazonia
(Wittmann et al 2004) sediment deposition decrease toward upper sites and species
colonizing such lower areas show specific adaptations to the new site conditions regarding
to adventitious roots that probably offer mechanical support (Parolin et al 2004 Wittmann
et al 2004) However litter removal or deposition including seed bank may affect
directly species distribution once flooding and allied effects reallocate litter and seeds
among sites (Johansson et al 1996) Moreover studies have showed that flooding timing
frequency and magnitude can be used as indicators of sapling zonation on floodplain
forests (Vreugdenhil et al 2006)
All these processes are included in the unmeasured variables or spatially structuring
processes that have been missed by the geographical terms (Titeux et al 2004) and
accounted to the far unexplained variation (662) As also stressed by these authors the
stochastic spacendashtime fluctuations of each population the lsquounsaturationrsquo pattern (some
species do not use all suitable habitats) and species recording in not appropriated spatial
scales contribute to this unexplained variation too Notwithstanding occurrence data or
species abundances are often noisy (ter Braak 1995) and widespread in ecological studies
(Borcard et al 1992 Titeux et al 2004)
Richness and diversity patterns
Significant transitions occurred from the levee and depression to the following lower slope
transect regarding to stand structure and ecological groups Inversions on the proportion of
pioneershade-tolerant trees and auto-hydrochorous to anemochorous trees occurred in that
small transition and affected not only ecological groups but also species occurrences As a
consequence this zonation transect may consist spatially as a boundary for tolerant and
intolerant trees with regarding to flooding In fact few species occurred over the entire
flooding gradient and the lower slope also appears as an edge for several species
In a temporal scale the lower slope area will probably present more heterogeneous
spans in flooding events and it may consist in the most heterogeneous temporalndashspatial
sector across the topographic gradient which agree with the findings of Pollock et al
2384 Biodivers Conserv (2010) 192371ndash2387
123
Authors personal copy
(1998) from wetlands with different flooding regimes In the structured view of the
dynamic-equilibrium model (Huston 1994) different patches from different seral stages
result from spatial variation of disturbance frequencies If disturbance frequencies vary
over time a landscape could also contain patches of different seral stages (Pollock et al
1998) These authors used that assumption in a model with temporally synchronous dis-
turbance and found that at the community-scale level the results supported many pre-
dictions of the dynamic-equilibrium model especially regarding to species richness
In our study there are two major factors related to disturbance frequency First and
foremost is that once river floods are unpredictable due to a hydrological regime that varies
with occasional long-rainy periods temporal heterogeneity should be higher than in sea-
sonal predicted flood areas and second microtopography must create spatial heterogeneity
during floods events in the local area However well-drained upper sites are probably more
affected by inherent community processes as gap-phase dynamics and direct supply rates
of light (Stevens and Carson 2002) As related by Worbes et al (1992) and Parolin (2001)
hundreds of tree species with different phenological and other ecological traits grow in
seasonal flooded forests In such cases the cyclic alternation on floods and droughts drove
species to life history behavioral and morphological adaptations (Lytle and Poff 2004)
But in the case of unpredictable floods and droughts as assigned in our study bet-hedging
strategies might be evolved for example by persistent seed bank or asynchronous
reproductive phenologies (Brock 2003) although there are no conclusive studies related to
this theory (Lytle and Poff 2004)
Allowed by a transition in the ecological groups from the lower sites to the upper ones
species richness has a maximum at the lower slope transect probably due to higher het-
erogeneity in disturbance events (space and time) and correlated variables As reported by
Desilets and Houle (2005) the spatial gradient provides some evidences for stress toler-
ance and competition as factors structuring species distribution across the topographicndash
flooding gradient also boosted by an unpredictable pattern of floods that vary in frequency
timing and magnitude Lower sectors showed predicted ecological groups already
described for these areas (Budke et al 2007 2008 Junk et al 1989 Lytle and Poff 2004)
and expected structuring changes also occurred toward upper sites where the proportion of
shade-tolerant and small trees increased as well as anemochorous trees
In summary spatialndashtemporal and environmental variables are arranging tree species
distribution across the toposequence of our study site Furthermore predicted ecological
groups reflected the dynamics of disturbance in the topographicndashhydrological gradient
Species richness and diversity also reflected such pattern and were higher in the mid-sector
where occasional floods should prevent competitive exclusion and generate high envi-
ronmental heterogeneity
Acknowledgements We are grateful to the Programa de Pos-Graduacao em Botanica of the UniversidadeFederal do Rio Grande do SulmdashUFRGS for the opportunity to undertake this study and to CAPES Agencyfor the scholarship granted to the first author Our special thanks to Diogo lsquolsquoBagualrsquorsquo Lindenmaier forfieldwork assistance and to Ricardo Braga Eduardo Rossi and colleagues of the Laboratorio de Fitoeco-logiamdashUFRGS for critiques and suggestions We also appreciated the reviewing efforts of anonymouscontributors for providing useful comments to the manuscript
References
Baldwin DS Mitchell AM (2000) The effects of drying and re-flooding on the sediment and soil nutrientdynamics of lowland river-floodplain systems a synthesis Regul River 16457ndash467 doi1010021099-1646
Biodivers Conserv (2010) 192371ndash2387 2385
123
Authors personal copy
Barroso GM Morim MP Peixoto AL Ichaso CLF (1999) Frutos e sementes morfologia aplicada a si-stematica de dicotiledoneas Editora UFV Vicosa
Borcard D Legendre P Drapeau P (1992) Partialling out the spatial component of ecological variationEcology 731045ndash1055 doi1010292006WR005044
Brock MA (2003) Drought and aquatic community resilience the role of eggs and seeds in sediments oftemporary wetlands Freshw Biol 481207ndash1218 doi101046j1365-2427200301083x
Budke JC Athayde EA Giehl ELH Zachia RA Eisinger SM (2005) Composicao florıstica e estrategias dedispersao de especies lenhosas em uma floresta ribeirinha arroio Passo das Tropas Santa Maria RSBrasil Iheringia Bot 6017ndash24
Budke JC Jarenkow JA Oliveira-Filho AT Lindenmaier DS (2006) Padroes de riqueza e diversidade emrios de pequeno porte In Mariath JEA Santos RP (eds) Os avancos da botanica no inıcio do seculoXXI SBB Porto Alegre
Budke JC Jarenkow JA Oliveira-Filho AT (2007) Relationships between tree component structuretopography and soils of a riverine forest Rio Botucaraı southern Brazil Plant Ecol 189187ndash200 doi101007s11258-006-9174-8
Budke JC Jarenkow JA Oliveira-Filho AT (2008) Tree community features of two stands of riverine forestunder different flooding regimes in southern Brazil Flora 203162ndash174 doi101016jflora200703001
Cardoso E Schiavini I (2002) Relacao entre distribuicao de especies arboreas e topografia em um gradienteflorestal na Estacao Ecologica do Panga (Uberlandia MG) Rev Bras Bot 25277ndash289
Connell JH (1978) Diversity in tropical rain forests and coral reefs Science 1991302ndash1310 doi101126science19943351302
Damasceno-Junior GA Semir J Santos FAM Leitao-Filho HF (2005) Structure distribution of species andinundation in a riparian forest of Rio Paraguai Pantanal Brazil Flora 200119ndash135 doi101016jflora200409002
Desilets P Houle G (2005) Effects of resource availability and heterogeneity on the slope of the species-areacurve along a floodplain-upland gradient J Veg Sci 16487ndash496 doi1016581100-9233
Dufrene M Legendre P (1997) Species assemblages and indicator species the need for a flexible asym-metrical approach Ecol Monogr 67345ndash366
EMBRAPA (1997) Manual de metodos de analises de solo Empresa Brasileira de Pesquisa Agropecuariaand Centro Nacional de Pesquisas de Solos Rio de Janeiro
Ferreira LV (2000) Effects of flooding duration on species richness floristic composition and forest structurein river margin habitat in Amazonian blackwater floodplain forests implications for future design ofprotected areas Biodivers Conserv 91ndash14 doi101023A1008989811637
Ferreira LV Stohlgren TJ (1999) Effects of river level fluctuation on plant species richness diversity anddistribution in a floodplain forest in Central Amazonia Oecologia 120582ndash587 doi101007s004420050893
Ferreira C Piedade MTF Franco AC Goncalves JFC Junk WJ (2009) Adaptive strategies to tolerateprolonged flooding in seedlings of floodplain and upland populations of Himatanthus sucuuba aCentral Amazon tree Aquat Bot 90246ndash252 doi101016jaquabot200810006
Gotelli NJ Colwell RK (2001) Quantifying biodiversity procedures and pitfalls in the measurement andcomparison of species richness Ecol Lett 4379ndash391 doi101046j1461-0248200100230x
Guilherme FAG Oliveira-Filho AT Appolinario V Bearzoti E (2004) Effects of flooding regime andwoody bamboos on tree community dynamics in a section of tropical semideciduous forest in south-eastern Brazil Plant Ecol 17419ndash36 doi101023BVEGE000004605197752cd
Huston M (1994) Biological diversity the coexistence of species in changing landscapes CambridgeUniversity Press Cambridge
IPAGRO (1982) Atlas agroclimatico do Rio Grande do Sul Pallotti Porto AlegreJohansson ME Nilsson C Nilsson E (1996) Do rivers function as corridors for plant dispersal J Veg Sci
7593ndash598Junk WJ Bayley PB Sparks RE (1989) The flood pulse concept in river-floodplain systems Can J Fish
Aquat Sci 106110ndash127Legendre P (1993) Spatial autocorrelationmdashtrouble or new paradigm Ecology 741659ndash1673Leinz V (1949) Contribuicao a geologia dos derrames basalticos do Rio Grande do Sul Bol Fac Filos Let
581ndash55Lytle DA Poff NL (2004) Adaptation to natural flow regimes Trends Ecol Evol 1994ndash100 doi
101016jtree200310002McCune B Mefford MJ (1997) PCndashORD Multivariate analysis of ecological data version 436 MjM
Software Design Glaneden BeachMueller-Dombois D Ellenberg H (1974) Aims and methods of vegetation ecology John Wiley New York
2386 Biodivers Conserv (2010) 192371ndash2387
123
Authors personal copy
Neiff JJ (1990) Ideas para la interpretacion ecologica del Parana Interciencia 15424ndash441Neiff JJ Neiff M (2003) Pulso software para analisis de fenomenos recurrentes httpwwwneiffcom
Cited 25 May 2009Oliveira-Filho AT Vilela EA Gavilanes ML Carvalho DA (1994) Effect of flooding regime and understory
bamboos on the physiognomy and tree species composition of a tropical semideciduous forest in south-eastern Brazil Vegetatio 11399ndash124
Oliveira-Filho AT Curi N Vilela EA Carvalho DA (2001) Variation in tree community composition andstructure with changes in soil properties within a fragment of semideciduous forest in south-easternBrazil Edinb J Bot 58139ndash158 doi101017S0960428601000506
Oliveira-Filho AT Jarenkow JA Rodal MJN (2006) Floristic relationships of seasonally dry forests ofeastern South America based on tree species distribution patterns In Pennington RT Ratter JA LewisGP (eds) Neotropical savannas and dry forests plant diversity biogeography and conservation CRCPress Boca Raton
Parolin P (2001) Morphological and physiological adjustments to waterlogging and drought in seedlings ofAmazonian floodplain trees Oecologia 128326ndash335 doi101007s004420100660
Parolin P de Simone O Haase K Waldhoff D Rottenberger S Kuhn U Kesselmeier J Kleiss B SchmidtW Piedade MTF Junk WJ (2004) Central Amazonian floodplain forests tree adaptations in a pulsingsystem Bot Rev 70357ndash380 doi1016630006-8101(2004)070[0357CAFFTA]20CO2
Pijl L (1982) Principles of dispersal in higher plants Springer New YorkPillar VD (2006) Multivariate exploratory analysis randomization testing and bootstrap resampling version
2320 Departamento de Ecologia UFRGS Porto AlegrePollock MM Naiman RJ Hanley TA (1998) Plant species richness in riparian wetlandsmdasha test of biodi-
versity theory Ecology 7994ndash105Robertson KM (2006) Distributions of tree species along point bars of 10 rivers in the south-eastern US
Coastal Plain J Biogeogr 33121ndash132 doi101111j1365-2699200501371xRosales J Petts G Knab-Vispo C (2001) Ecological gradients within the riparian forests of the lower Caura
river Venezuela Plant Ecol 152101ndash118 doi101023A1011411020040Stevens MHH Carson WP (2002) Resource quantity not resource heterogeneity maintains plant diversity
Ecol Lett 5420ndash426 doi101046j1461-0248200200333xStreck EV Kampf N Dalmolin RSD Klamt E Nascimento PC Schneider P (2002) Solos do Rio Grande do
Sul EMATERRS and UFRGS Porto AlegreSwaine MD Whitmore TC (1988) On the definition of ecological species groups in tropical rain forests
Vegetatio 7581ndash86Tabacchi E Correll DL Hauer R Pinay G Planty-Tabacchi AM Wissmar R (1998) Development
maintenance and role of riparian vegetation in the river landscape Freshw Biol 40497ndash516 doi101046j1365-2427199800381x
ter Braack CJF Smilauer P (1998) Canoco reference manual and userrsquos guide to Canoco for Windowssoftware for canonical community ordination (version 40) Microcomputer Power Ithaca
ter Braak CJF (1995) Ordination In Jongman RHG ter Braak CJF van Togeren OFR (eds) Data analysis incommunity and landscape ecology Cambridge University Press New York
Titeux N Dufrene M Jacob JP Paquay M Defourny P (2004) Multivariate analysis of fine-scale breedingbird atlas using a geographical information system and partial canonical correspondence analysisenvironmental and spatial effects J Biogeogr 311841ndash1856 doi101111j1365-2699200401125x
Turner MG Gergel SE Dixon MD Miller JR (2004) Distribution and abundance of trees in floodplainforests of the Wisconsin river environmental influences at different scales J Veg Sci 15729ndash738
Vreugdenhil SJ Kramer K Pelsma T (2006) Effects of flooding duration frequency and depth on thepresence of saplings of six woody species in north-west Europe For Ecol Manage 23647ndash55 doi101016jforeco200608329
Walker KF Sheldon F Puckridge JT (1995) A perspective on dryland river ecosystems Regul River 1185ndash104 doi101002rrr3450110108
Weiher E (2003) Species richness along multiple gradients testing a general multivariate model in oaksavannas Oikos 101311ndash316 doi101034j1600-0706200312216x
Wittmann F Junk WJ Piedade MTF (2004) The varzea forests in Amazonia flooding and the highlydynamic geomorphology interact with natural forest succession For Ecol Manage 196199ndash212 doi101016jforeco200402060
Worbes M Klinge H Revilla JD Martius C (1992) On the dynamics floristic subdivision and geographicaldistribution of Varzea forests in Central Amazonia J Veg Sci 3553ndash564
Zar JH (1996) Biostatistical analysis Prentice-Hall New Jersey
Biodivers Conserv (2010) 192371ndash2387 2387
123
Authors personal copy
Tab
le1
So
ilv
aria
ble
so
ffi
ve
tran
sect
so
fri
ver
ine
fore
sto
nth
eB
otu
cara
ıri
ver
so
uth
ern
Bra
zil
Soil
var
iable
sL
evee
Dep
ress
ion
L-s
lope
M-s
lope
Rid
ge
FP
Rel
ativ
eel
evat
ion
(m)
38
plusmn0
5a
54
plusmn0
7a
85
plusmn2
3b
11
8plusmn
35
bc
13
6plusmn
37
c8
03
0
00
1
pH
(H2O
)4
8plusmn
04
47
plusmn0
44
7plusmn
04
47
plusmn0
64
7plusmn
06
08
50
93
ns
Pmdash
Meh
lich
(mg
dm
-3)
71
plusmn2
17
1plusmn
23
63
plusmn1
66
plusmn1
75
7plusmn
14
22
30
07
ns
K(m
gd
m-
3)
76
1plusmn
12
57
96
plusmn1
64
89
plusmn2
62
94
3plusmn
34
87
8plusmn
24
14
36
03
5n
s
Ca
(cm
olc
dm
-3)
62
plusmn4
36
8plusmn
44
7plusmn
47
61
plusmn4
57
plusmn3
73
19
05
2n
s
Mg
(cm
olc
dm
-3)
15
plusmn0
71
6plusmn
07
15
plusmn0
71
5plusmn
06
14
plusmn0
50
36
09
8n
s
Al
H
(cm
olc
dm
-3)
66
plusmn3
1a
71
plusmn3
ab8
5plusmn
4ab
96
plusmn5
5ab
10
plusmn4
2b
97
60
04
S(c
mo
lcd
m-
3)
8plusmn
48
87
plusmn5
18
1plusmn
46
84
plusmn5
17
4plusmn
41
27
70
59
ns
CE
C(c
mo
lcd
m-
3)
15
2plusmn
48
16
2plusmn
38
17
1plusmn
33
17
9plusmn
43
17
3plusmn
34
14
50
22
ns
V(
)5
57
plusmn1
71
49
7plusmn
20
24
59
plusmn2
46
45
5plusmn
23
64
18
plusmn2
07
40
60
39
ns
OM
()
26
plusmn1
1a
28
plusmn1
a3
2plusmn
1ab
37
plusmn1
1b
38
plusmn0
8b
22
9
00
01
Cla
y(
)1
37
plusmn2
4a
15
4plusmn
33
ab1
58
plusmn2
8ab
16
plusmn2
1b
15
5plusmn
17
ab2
49
00
4
San
d(
)2
0plusmn
74
23
1plusmn
22
24
2plusmn
10
12
23
plusmn7
52
23
plusmn6
28
91
00
6n
s
Sil
t(
)6
42
plusmn1
09
60
4plusmn
14
59
8plusmn
12
96
2plusmn
91
62
plusmn7
24
91
02
9n
s
Val
ues
are
mea
ns
plusmnst
and
ard
dev
iati
on
sfr
om
0to
20
cmd
epth
top
soil
sam
ple
s(N
=2
0fo
rea
chtr
anse
ct)
Dif
fere
nt
lett
ers
afte
rv
alu
esin
dic
ate
sign
ifica
nt
dif
fere
nce
sin
AN
OV
Ate
sts
(ns
=n
on
-sig
nifi
can
t)
Biodivers Conserv (2010) 192371ndash2387 2377
123
Authors personal copy
shaped the physiognomic features of different sectors that varied according to the topo-
sequence and consequently in flooding regime The depression sector presented lower
density basal area and also low tree diameters whereas the levee portion presented high
density and basal area
Species distribution across the topographic gradient is presented in Table 3 according
to the Indicator Species Analysis Some species were clearly distributed from lowland to
medium sites as Eugenia uniflora Myrciaria tenella Eugenia uruguayensis whereas
others were restricted to upland areas as Chomelia obtusa and Cordia americana Many
species did not show a specific site distribution and occurred over a wide distribution range
as Gymnanthes concolor and Casearia sylvestris
The relative elevation of each transect reflects the pattern of flooding frequency and
duration in each site then spatial aggregation of trees may indicate preferences or
restriction on the establishment of some species Typical riverine species appeared near the
river margin as Pouteria gardneriana Guettarda uruguensis and others (Table 3) whereas
typical species of well-drained forests as Sorocea bonplandii Parapiptadenia rigida and
Cupania vernalis occurred frequently in the ridgetop transect Furthermore 13 species did
Fig 1 Diameter-class distributions of trees with pbh C 15 cm surveyed in five transects of riverine foreston the Botucaraı river southern Brazil Diameter-classes are used for increasing intervals (see lsquolsquoMethodsrsquorsquosection) Bars and ranges are means and 95 confidence intervals of 100 sampling units respectively
Table 2 Density (ind ha-1) dominance (m2 ha-1) mean height (m) and mean diameter (cm) for differenttransects of the riverside forest of the Botucaraı river southern Brazil
Transect AD ADo Height Diameter
Levee 1655 plusmn 467 a 3927 plusmn 268 a 62 plusmn 25 a 1761 plusmn 1697
Depression 1005 plusmn 369 b 198 plusmn 137 b 67 plusmn 21 b 161 plusmn 1194
L-slope 1120 plusmn 443 ab 229 plusmn 165 b 71 plusmn 26 b 1547 plusmn 1302
M-slope 1415 plusmn 438 a 274 plusmn 165 b 69 plusmn 25 b 1453 plusmn 98
Ridge 950 plusmn 294 b 282 plusmn 194 b 7 plusmn 24 b 1727 plusmn 1555
ANOVA F = 107 F = 129 F = 305 F = 35
Different letters after values indicate significant differences in t tests ( P 005 P 0001)
AD density ADo dominance
2378 Biodivers Conserv (2010) 192371ndash2387
123
Authors personal copy
not present a topographic association due to wide distribution through the gradient On the
other hand the distinction among environmental and spatial effects showed that space
contributes significantly to the distribution of tree species (Fig 3) Environmental variables
selected by forward selection procedure (P 005) are summarized in Table 4 All geo-
graphical terms of the polynomial function were significant (P 005) during spatial CCA
and were add to the model The four CCA analyses provided the following results
1 CCA of the species matrix constrained by the environmental matrix sum of all
canonical eigenvalues = 0944 Monte Carlo tests for overall analysis F = 286
P 0001
2 CCA of the species matrix constrained by spatial matrix sum of all canonical
eigenvalues = 1017 Monte Carlo tests for overall analysis F = 359 P 0001
Fig 2 Ecological groups of vertical distribution (A) regeneration (B) and dispersal (C) in five transects ofriverine forest of Botucaraı river southern Brazil Pi pioneer Ld light-demanding St shade-tolerant Zoozoochorous Auto autochorous Ane anemochorous Hydro hydrochorous
Biodivers Conserv (2010) 192371ndash2387 2379
123
Authors personal copy
3 Environmental partial CCA (after removing the effect of geographical matrix) sum of
all canonical eigenvalues = 0416 Monte Carlo tests for overall analysis F = 135
P 0001
Table 3 Indicator species analysis (ISA) performed for species with density C10 individuals sampled infive transects with 20 sampling units each Botucaraı river southern Brazil
Biplots of environmental variables and species or environmental variables and sampling
units were depicted with the environmental partial analyses results (Fig 4) In this step
species distributions are purely based on speciesndashenvironment relationships The first two
axes accounted respectively for 268 (eigenvalue = 0112) and 181 (eigen-
value = 0075) of the speciesndashenvironment relationships and speciesndashenvironment corre-
lations for these axes were 0742 and 0649 (P 005) respectively The first four axes
accounted for 691 of speciesndashenvironment relationships Table 4 shows the intraset
correlations among variables and canonical coefficients with the first two axes The first
canonical axis was positively correlated with topography and several soil variables that are
Fig 3 Variation partitioning ofthe tree species matrix
Table 4 Intraset correlations among environmental variables selected for the model during lsquolsquoenviron-mentalrsquorsquo partial CCA of the species matrix and canonical coefficients of the first two axes
Environmental variables were selected by forward stepwise selection and included on the model if sig-nificant in Monte Carlo tests (P 005)
Biodivers Conserv (2010) 192371ndash2387 2381
123
Authors personal copy
influenced by flooding events Percentages of organic matter saturation of bases and cation
exchange capacity where higher through upper sites whereas aluminium contents were
higher in lower areas Thus this environmental gradient may affect tree distribution pat-
terns by restricting or facilitating species establishment As related in the Indicator
Analysis (although not accounting for spatial patterns directly) typical species of flooding
areas occurred near the levee and depression and the zonation was sharp once frequent
species of well-drained areas occurred only in the middle slope and ridgetop Sampling
units (Fig 5) also appeared distributed according to the toposequence with some over-
lapping due to species distribution
Rarefaction curves of species revealed significant differences on the expected total
number of species in each transect with higher richness within the middle slope (Fig 6)
A regression model (Fig 6) fitted diversity in a second-order polynomial regression
(y = -00586x2 03594x 291 R2 = 077) that showed the same pattern
Discussion
Environmental and spatial patterns
Tree species distribution throughout the topographical gradient indicated that both envi-
ronmental and spatial features were particularly important in predicting species and
community patterns This agrees with the well-know influence of geomorphic features and
hydrological regimes on riparian forests over different temporal and spatial scales
(Tabacchi et al 1998 Turner et al 2004 Desilets and Houle 2005) even though dis-
tinctions between environmental and spatial effects remain poorly studied (Titeux et al
Fig 4 Ordination biplotdepicting the two axes of theenvironmental partial CCA ofsampling units in a riverine forestin southern Brazil Eachsampling unit was identified bydifferent symbols according tothe respective transectEnvironmental variables arerepresented by their acronyms(see Table S1)
2382 Biodivers Conserv (2010) 192371ndash2387
123
Authors personal copy
2004) Here we employed a routine to partialling out the spatial effects on the analysis of
speciesndashenvironment relationships that also highlight the spatial component embedded in
such analysis (Legendre 1993) Once several biotic processes as growth mortality dis-
persal and predation influence the observed distributions of organisms (resulting in spatial
correlation) or if their distributions are dependent on explanatory variables which are
Fig 5 Ordination biplot depicting the two axes of the environmental partial CCA of species of a riverineforest in southern Brazil Species and environmental variables are represented by their acronyms (seeTable S1)
Fig 6 Rarefaction curves of tree species and Shannon diversity indices from five transects of riverine foreston the Botucaraı river southern Brazil Sampling units are representing different transects
Biodivers Conserv (2010) 192371ndash2387 2383
123
Authors personal copy
themselves spatially structured (Legendre 1993 Titeux et al 2004) spatial structuring is
an intrinsic component of ecosystems In our analysis lsquopurersquo spatial trends were more
attributed to species distribution than to lsquopurersquo environmental effects which link conta-
gious biological processes as important to the tree species distribution Furthermore
species and environmental data have a reasonable proportion of similar spatial structuring
identified by the largest proportion on the species variation due to spatially structured
environmental variation (1246) According to Borcard et al (1992) species and envi-
ronmental variables have in this case the same response to some common underlying
causes as the topographicndashflooding gradient In fact several studies have demonstrated the
direct effect of wetting and desiccation processes on both mineralogy and microbial
ecology of the sediment including nutrient dynamics (Baldwin and Mitchell 2000)
Once sediment or soils are submerged the inundation leads to a decrease in oxygen
contents and then resulting in progressive anaerobic conditions Rapid cycling of litter may
occur due to an increase on microbial activity which generates pulses on nutrient contents
and finally result in highly productive systems (Baldwin and Mitchell 2000) However a
negative effect is the rapid oxygen consumption which quickly leads to soil hypoxia or
anoxia When flood ends the anaerobic zones of sediments are newly oxygenated and
microbiota is replaced gradually to a new phase
Other spatially structured variables are sediment deposition and litter displacement
which are not covered in our study As demonstrated in lsquovarzearsquo forests of Amazonia
(Wittmann et al 2004) sediment deposition decrease toward upper sites and species
colonizing such lower areas show specific adaptations to the new site conditions regarding
to adventitious roots that probably offer mechanical support (Parolin et al 2004 Wittmann
et al 2004) However litter removal or deposition including seed bank may affect
directly species distribution once flooding and allied effects reallocate litter and seeds
among sites (Johansson et al 1996) Moreover studies have showed that flooding timing
frequency and magnitude can be used as indicators of sapling zonation on floodplain
forests (Vreugdenhil et al 2006)
All these processes are included in the unmeasured variables or spatially structuring
processes that have been missed by the geographical terms (Titeux et al 2004) and
accounted to the far unexplained variation (662) As also stressed by these authors the
stochastic spacendashtime fluctuations of each population the lsquounsaturationrsquo pattern (some
species do not use all suitable habitats) and species recording in not appropriated spatial
scales contribute to this unexplained variation too Notwithstanding occurrence data or
species abundances are often noisy (ter Braak 1995) and widespread in ecological studies
(Borcard et al 1992 Titeux et al 2004)
Richness and diversity patterns
Significant transitions occurred from the levee and depression to the following lower slope
transect regarding to stand structure and ecological groups Inversions on the proportion of
pioneershade-tolerant trees and auto-hydrochorous to anemochorous trees occurred in that
small transition and affected not only ecological groups but also species occurrences As a
consequence this zonation transect may consist spatially as a boundary for tolerant and
intolerant trees with regarding to flooding In fact few species occurred over the entire
flooding gradient and the lower slope also appears as an edge for several species
In a temporal scale the lower slope area will probably present more heterogeneous
spans in flooding events and it may consist in the most heterogeneous temporalndashspatial
sector across the topographic gradient which agree with the findings of Pollock et al
2384 Biodivers Conserv (2010) 192371ndash2387
123
Authors personal copy
(1998) from wetlands with different flooding regimes In the structured view of the
dynamic-equilibrium model (Huston 1994) different patches from different seral stages
result from spatial variation of disturbance frequencies If disturbance frequencies vary
over time a landscape could also contain patches of different seral stages (Pollock et al
1998) These authors used that assumption in a model with temporally synchronous dis-
turbance and found that at the community-scale level the results supported many pre-
dictions of the dynamic-equilibrium model especially regarding to species richness
In our study there are two major factors related to disturbance frequency First and
foremost is that once river floods are unpredictable due to a hydrological regime that varies
with occasional long-rainy periods temporal heterogeneity should be higher than in sea-
sonal predicted flood areas and second microtopography must create spatial heterogeneity
during floods events in the local area However well-drained upper sites are probably more
affected by inherent community processes as gap-phase dynamics and direct supply rates
of light (Stevens and Carson 2002) As related by Worbes et al (1992) and Parolin (2001)
hundreds of tree species with different phenological and other ecological traits grow in
seasonal flooded forests In such cases the cyclic alternation on floods and droughts drove
species to life history behavioral and morphological adaptations (Lytle and Poff 2004)
But in the case of unpredictable floods and droughts as assigned in our study bet-hedging
strategies might be evolved for example by persistent seed bank or asynchronous
reproductive phenologies (Brock 2003) although there are no conclusive studies related to
this theory (Lytle and Poff 2004)
Allowed by a transition in the ecological groups from the lower sites to the upper ones
species richness has a maximum at the lower slope transect probably due to higher het-
erogeneity in disturbance events (space and time) and correlated variables As reported by
Desilets and Houle (2005) the spatial gradient provides some evidences for stress toler-
ance and competition as factors structuring species distribution across the topographicndash
flooding gradient also boosted by an unpredictable pattern of floods that vary in frequency
timing and magnitude Lower sectors showed predicted ecological groups already
described for these areas (Budke et al 2007 2008 Junk et al 1989 Lytle and Poff 2004)
and expected structuring changes also occurred toward upper sites where the proportion of
shade-tolerant and small trees increased as well as anemochorous trees
In summary spatialndashtemporal and environmental variables are arranging tree species
distribution across the toposequence of our study site Furthermore predicted ecological
groups reflected the dynamics of disturbance in the topographicndashhydrological gradient
Species richness and diversity also reflected such pattern and were higher in the mid-sector
where occasional floods should prevent competitive exclusion and generate high envi-
ronmental heterogeneity
Acknowledgements We are grateful to the Programa de Pos-Graduacao em Botanica of the UniversidadeFederal do Rio Grande do SulmdashUFRGS for the opportunity to undertake this study and to CAPES Agencyfor the scholarship granted to the first author Our special thanks to Diogo lsquolsquoBagualrsquorsquo Lindenmaier forfieldwork assistance and to Ricardo Braga Eduardo Rossi and colleagues of the Laboratorio de Fitoeco-logiamdashUFRGS for critiques and suggestions We also appreciated the reviewing efforts of anonymouscontributors for providing useful comments to the manuscript
References
Baldwin DS Mitchell AM (2000) The effects of drying and re-flooding on the sediment and soil nutrientdynamics of lowland river-floodplain systems a synthesis Regul River 16457ndash467 doi1010021099-1646
Biodivers Conserv (2010) 192371ndash2387 2385
123
Authors personal copy
Barroso GM Morim MP Peixoto AL Ichaso CLF (1999) Frutos e sementes morfologia aplicada a si-stematica de dicotiledoneas Editora UFV Vicosa
Borcard D Legendre P Drapeau P (1992) Partialling out the spatial component of ecological variationEcology 731045ndash1055 doi1010292006WR005044
Brock MA (2003) Drought and aquatic community resilience the role of eggs and seeds in sediments oftemporary wetlands Freshw Biol 481207ndash1218 doi101046j1365-2427200301083x
Budke JC Athayde EA Giehl ELH Zachia RA Eisinger SM (2005) Composicao florıstica e estrategias dedispersao de especies lenhosas em uma floresta ribeirinha arroio Passo das Tropas Santa Maria RSBrasil Iheringia Bot 6017ndash24
Budke JC Jarenkow JA Oliveira-Filho AT Lindenmaier DS (2006) Padroes de riqueza e diversidade emrios de pequeno porte In Mariath JEA Santos RP (eds) Os avancos da botanica no inıcio do seculoXXI SBB Porto Alegre
Budke JC Jarenkow JA Oliveira-Filho AT (2007) Relationships between tree component structuretopography and soils of a riverine forest Rio Botucaraı southern Brazil Plant Ecol 189187ndash200 doi101007s11258-006-9174-8
Budke JC Jarenkow JA Oliveira-Filho AT (2008) Tree community features of two stands of riverine forestunder different flooding regimes in southern Brazil Flora 203162ndash174 doi101016jflora200703001
Cardoso E Schiavini I (2002) Relacao entre distribuicao de especies arboreas e topografia em um gradienteflorestal na Estacao Ecologica do Panga (Uberlandia MG) Rev Bras Bot 25277ndash289
Connell JH (1978) Diversity in tropical rain forests and coral reefs Science 1991302ndash1310 doi101126science19943351302
Damasceno-Junior GA Semir J Santos FAM Leitao-Filho HF (2005) Structure distribution of species andinundation in a riparian forest of Rio Paraguai Pantanal Brazil Flora 200119ndash135 doi101016jflora200409002
Desilets P Houle G (2005) Effects of resource availability and heterogeneity on the slope of the species-areacurve along a floodplain-upland gradient J Veg Sci 16487ndash496 doi1016581100-9233
Dufrene M Legendre P (1997) Species assemblages and indicator species the need for a flexible asym-metrical approach Ecol Monogr 67345ndash366
EMBRAPA (1997) Manual de metodos de analises de solo Empresa Brasileira de Pesquisa Agropecuariaand Centro Nacional de Pesquisas de Solos Rio de Janeiro
Ferreira LV (2000) Effects of flooding duration on species richness floristic composition and forest structurein river margin habitat in Amazonian blackwater floodplain forests implications for future design ofprotected areas Biodivers Conserv 91ndash14 doi101023A1008989811637
Ferreira LV Stohlgren TJ (1999) Effects of river level fluctuation on plant species richness diversity anddistribution in a floodplain forest in Central Amazonia Oecologia 120582ndash587 doi101007s004420050893
Ferreira C Piedade MTF Franco AC Goncalves JFC Junk WJ (2009) Adaptive strategies to tolerateprolonged flooding in seedlings of floodplain and upland populations of Himatanthus sucuuba aCentral Amazon tree Aquat Bot 90246ndash252 doi101016jaquabot200810006
Gotelli NJ Colwell RK (2001) Quantifying biodiversity procedures and pitfalls in the measurement andcomparison of species richness Ecol Lett 4379ndash391 doi101046j1461-0248200100230x
Guilherme FAG Oliveira-Filho AT Appolinario V Bearzoti E (2004) Effects of flooding regime andwoody bamboos on tree community dynamics in a section of tropical semideciduous forest in south-eastern Brazil Plant Ecol 17419ndash36 doi101023BVEGE000004605197752cd
Huston M (1994) Biological diversity the coexistence of species in changing landscapes CambridgeUniversity Press Cambridge
IPAGRO (1982) Atlas agroclimatico do Rio Grande do Sul Pallotti Porto AlegreJohansson ME Nilsson C Nilsson E (1996) Do rivers function as corridors for plant dispersal J Veg Sci
7593ndash598Junk WJ Bayley PB Sparks RE (1989) The flood pulse concept in river-floodplain systems Can J Fish
Aquat Sci 106110ndash127Legendre P (1993) Spatial autocorrelationmdashtrouble or new paradigm Ecology 741659ndash1673Leinz V (1949) Contribuicao a geologia dos derrames basalticos do Rio Grande do Sul Bol Fac Filos Let
581ndash55Lytle DA Poff NL (2004) Adaptation to natural flow regimes Trends Ecol Evol 1994ndash100 doi
101016jtree200310002McCune B Mefford MJ (1997) PCndashORD Multivariate analysis of ecological data version 436 MjM
Software Design Glaneden BeachMueller-Dombois D Ellenberg H (1974) Aims and methods of vegetation ecology John Wiley New York
2386 Biodivers Conserv (2010) 192371ndash2387
123
Authors personal copy
Neiff JJ (1990) Ideas para la interpretacion ecologica del Parana Interciencia 15424ndash441Neiff JJ Neiff M (2003) Pulso software para analisis de fenomenos recurrentes httpwwwneiffcom
Cited 25 May 2009Oliveira-Filho AT Vilela EA Gavilanes ML Carvalho DA (1994) Effect of flooding regime and understory
bamboos on the physiognomy and tree species composition of a tropical semideciduous forest in south-eastern Brazil Vegetatio 11399ndash124
Oliveira-Filho AT Curi N Vilela EA Carvalho DA (2001) Variation in tree community composition andstructure with changes in soil properties within a fragment of semideciduous forest in south-easternBrazil Edinb J Bot 58139ndash158 doi101017S0960428601000506
Oliveira-Filho AT Jarenkow JA Rodal MJN (2006) Floristic relationships of seasonally dry forests ofeastern South America based on tree species distribution patterns In Pennington RT Ratter JA LewisGP (eds) Neotropical savannas and dry forests plant diversity biogeography and conservation CRCPress Boca Raton
Parolin P (2001) Morphological and physiological adjustments to waterlogging and drought in seedlings ofAmazonian floodplain trees Oecologia 128326ndash335 doi101007s004420100660
Parolin P de Simone O Haase K Waldhoff D Rottenberger S Kuhn U Kesselmeier J Kleiss B SchmidtW Piedade MTF Junk WJ (2004) Central Amazonian floodplain forests tree adaptations in a pulsingsystem Bot Rev 70357ndash380 doi1016630006-8101(2004)070[0357CAFFTA]20CO2
Pijl L (1982) Principles of dispersal in higher plants Springer New YorkPillar VD (2006) Multivariate exploratory analysis randomization testing and bootstrap resampling version
2320 Departamento de Ecologia UFRGS Porto AlegrePollock MM Naiman RJ Hanley TA (1998) Plant species richness in riparian wetlandsmdasha test of biodi-
versity theory Ecology 7994ndash105Robertson KM (2006) Distributions of tree species along point bars of 10 rivers in the south-eastern US
Coastal Plain J Biogeogr 33121ndash132 doi101111j1365-2699200501371xRosales J Petts G Knab-Vispo C (2001) Ecological gradients within the riparian forests of the lower Caura
river Venezuela Plant Ecol 152101ndash118 doi101023A1011411020040Stevens MHH Carson WP (2002) Resource quantity not resource heterogeneity maintains plant diversity
Ecol Lett 5420ndash426 doi101046j1461-0248200200333xStreck EV Kampf N Dalmolin RSD Klamt E Nascimento PC Schneider P (2002) Solos do Rio Grande do
Sul EMATERRS and UFRGS Porto AlegreSwaine MD Whitmore TC (1988) On the definition of ecological species groups in tropical rain forests
Vegetatio 7581ndash86Tabacchi E Correll DL Hauer R Pinay G Planty-Tabacchi AM Wissmar R (1998) Development
maintenance and role of riparian vegetation in the river landscape Freshw Biol 40497ndash516 doi101046j1365-2427199800381x
ter Braack CJF Smilauer P (1998) Canoco reference manual and userrsquos guide to Canoco for Windowssoftware for canonical community ordination (version 40) Microcomputer Power Ithaca
ter Braak CJF (1995) Ordination In Jongman RHG ter Braak CJF van Togeren OFR (eds) Data analysis incommunity and landscape ecology Cambridge University Press New York
Titeux N Dufrene M Jacob JP Paquay M Defourny P (2004) Multivariate analysis of fine-scale breedingbird atlas using a geographical information system and partial canonical correspondence analysisenvironmental and spatial effects J Biogeogr 311841ndash1856 doi101111j1365-2699200401125x
Turner MG Gergel SE Dixon MD Miller JR (2004) Distribution and abundance of trees in floodplainforests of the Wisconsin river environmental influences at different scales J Veg Sci 15729ndash738
Vreugdenhil SJ Kramer K Pelsma T (2006) Effects of flooding duration frequency and depth on thepresence of saplings of six woody species in north-west Europe For Ecol Manage 23647ndash55 doi101016jforeco200608329
Walker KF Sheldon F Puckridge JT (1995) A perspective on dryland river ecosystems Regul River 1185ndash104 doi101002rrr3450110108
Weiher E (2003) Species richness along multiple gradients testing a general multivariate model in oaksavannas Oikos 101311ndash316 doi101034j1600-0706200312216x
Wittmann F Junk WJ Piedade MTF (2004) The varzea forests in Amazonia flooding and the highlydynamic geomorphology interact with natural forest succession For Ecol Manage 196199ndash212 doi101016jforeco200402060
Worbes M Klinge H Revilla JD Martius C (1992) On the dynamics floristic subdivision and geographicaldistribution of Varzea forests in Central Amazonia J Veg Sci 3553ndash564
Zar JH (1996) Biostatistical analysis Prentice-Hall New Jersey
Biodivers Conserv (2010) 192371ndash2387 2387
123
Authors personal copy
shaped the physiognomic features of different sectors that varied according to the topo-
sequence and consequently in flooding regime The depression sector presented lower
density basal area and also low tree diameters whereas the levee portion presented high
density and basal area
Species distribution across the topographic gradient is presented in Table 3 according
to the Indicator Species Analysis Some species were clearly distributed from lowland to
medium sites as Eugenia uniflora Myrciaria tenella Eugenia uruguayensis whereas
others were restricted to upland areas as Chomelia obtusa and Cordia americana Many
species did not show a specific site distribution and occurred over a wide distribution range
as Gymnanthes concolor and Casearia sylvestris
The relative elevation of each transect reflects the pattern of flooding frequency and
duration in each site then spatial aggregation of trees may indicate preferences or
restriction on the establishment of some species Typical riverine species appeared near the
river margin as Pouteria gardneriana Guettarda uruguensis and others (Table 3) whereas
typical species of well-drained forests as Sorocea bonplandii Parapiptadenia rigida and
Cupania vernalis occurred frequently in the ridgetop transect Furthermore 13 species did
Fig 1 Diameter-class distributions of trees with pbh C 15 cm surveyed in five transects of riverine foreston the Botucaraı river southern Brazil Diameter-classes are used for increasing intervals (see lsquolsquoMethodsrsquorsquosection) Bars and ranges are means and 95 confidence intervals of 100 sampling units respectively
Table 2 Density (ind ha-1) dominance (m2 ha-1) mean height (m) and mean diameter (cm) for differenttransects of the riverside forest of the Botucaraı river southern Brazil
Transect AD ADo Height Diameter
Levee 1655 plusmn 467 a 3927 plusmn 268 a 62 plusmn 25 a 1761 plusmn 1697
Depression 1005 plusmn 369 b 198 plusmn 137 b 67 plusmn 21 b 161 plusmn 1194
L-slope 1120 plusmn 443 ab 229 plusmn 165 b 71 plusmn 26 b 1547 plusmn 1302
M-slope 1415 plusmn 438 a 274 plusmn 165 b 69 plusmn 25 b 1453 plusmn 98
Ridge 950 plusmn 294 b 282 plusmn 194 b 7 plusmn 24 b 1727 plusmn 1555
ANOVA F = 107 F = 129 F = 305 F = 35
Different letters after values indicate significant differences in t tests ( P 005 P 0001)
AD density ADo dominance
2378 Biodivers Conserv (2010) 192371ndash2387
123
Authors personal copy
not present a topographic association due to wide distribution through the gradient On the
other hand the distinction among environmental and spatial effects showed that space
contributes significantly to the distribution of tree species (Fig 3) Environmental variables
selected by forward selection procedure (P 005) are summarized in Table 4 All geo-
graphical terms of the polynomial function were significant (P 005) during spatial CCA
and were add to the model The four CCA analyses provided the following results
1 CCA of the species matrix constrained by the environmental matrix sum of all
canonical eigenvalues = 0944 Monte Carlo tests for overall analysis F = 286
P 0001
2 CCA of the species matrix constrained by spatial matrix sum of all canonical
eigenvalues = 1017 Monte Carlo tests for overall analysis F = 359 P 0001
Fig 2 Ecological groups of vertical distribution (A) regeneration (B) and dispersal (C) in five transects ofriverine forest of Botucaraı river southern Brazil Pi pioneer Ld light-demanding St shade-tolerant Zoozoochorous Auto autochorous Ane anemochorous Hydro hydrochorous
Biodivers Conserv (2010) 192371ndash2387 2379
123
Authors personal copy
3 Environmental partial CCA (after removing the effect of geographical matrix) sum of
all canonical eigenvalues = 0416 Monte Carlo tests for overall analysis F = 135
P 0001
Table 3 Indicator species analysis (ISA) performed for species with density C10 individuals sampled infive transects with 20 sampling units each Botucaraı river southern Brazil
Biplots of environmental variables and species or environmental variables and sampling
units were depicted with the environmental partial analyses results (Fig 4) In this step
species distributions are purely based on speciesndashenvironment relationships The first two
axes accounted respectively for 268 (eigenvalue = 0112) and 181 (eigen-
value = 0075) of the speciesndashenvironment relationships and speciesndashenvironment corre-
lations for these axes were 0742 and 0649 (P 005) respectively The first four axes
accounted for 691 of speciesndashenvironment relationships Table 4 shows the intraset
correlations among variables and canonical coefficients with the first two axes The first
canonical axis was positively correlated with topography and several soil variables that are
Fig 3 Variation partitioning ofthe tree species matrix
Table 4 Intraset correlations among environmental variables selected for the model during lsquolsquoenviron-mentalrsquorsquo partial CCA of the species matrix and canonical coefficients of the first two axes
Environmental variables were selected by forward stepwise selection and included on the model if sig-nificant in Monte Carlo tests (P 005)
Biodivers Conserv (2010) 192371ndash2387 2381
123
Authors personal copy
influenced by flooding events Percentages of organic matter saturation of bases and cation
exchange capacity where higher through upper sites whereas aluminium contents were
higher in lower areas Thus this environmental gradient may affect tree distribution pat-
terns by restricting or facilitating species establishment As related in the Indicator
Analysis (although not accounting for spatial patterns directly) typical species of flooding
areas occurred near the levee and depression and the zonation was sharp once frequent
species of well-drained areas occurred only in the middle slope and ridgetop Sampling
units (Fig 5) also appeared distributed according to the toposequence with some over-
lapping due to species distribution
Rarefaction curves of species revealed significant differences on the expected total
number of species in each transect with higher richness within the middle slope (Fig 6)
A regression model (Fig 6) fitted diversity in a second-order polynomial regression
(y = -00586x2 03594x 291 R2 = 077) that showed the same pattern
Discussion
Environmental and spatial patterns
Tree species distribution throughout the topographical gradient indicated that both envi-
ronmental and spatial features were particularly important in predicting species and
community patterns This agrees with the well-know influence of geomorphic features and
hydrological regimes on riparian forests over different temporal and spatial scales
(Tabacchi et al 1998 Turner et al 2004 Desilets and Houle 2005) even though dis-
tinctions between environmental and spatial effects remain poorly studied (Titeux et al
Fig 4 Ordination biplotdepicting the two axes of theenvironmental partial CCA ofsampling units in a riverine forestin southern Brazil Eachsampling unit was identified bydifferent symbols according tothe respective transectEnvironmental variables arerepresented by their acronyms(see Table S1)
2382 Biodivers Conserv (2010) 192371ndash2387
123
Authors personal copy
2004) Here we employed a routine to partialling out the spatial effects on the analysis of
speciesndashenvironment relationships that also highlight the spatial component embedded in
such analysis (Legendre 1993) Once several biotic processes as growth mortality dis-
persal and predation influence the observed distributions of organisms (resulting in spatial
correlation) or if their distributions are dependent on explanatory variables which are
Fig 5 Ordination biplot depicting the two axes of the environmental partial CCA of species of a riverineforest in southern Brazil Species and environmental variables are represented by their acronyms (seeTable S1)
Fig 6 Rarefaction curves of tree species and Shannon diversity indices from five transects of riverine foreston the Botucaraı river southern Brazil Sampling units are representing different transects
Biodivers Conserv (2010) 192371ndash2387 2383
123
Authors personal copy
themselves spatially structured (Legendre 1993 Titeux et al 2004) spatial structuring is
an intrinsic component of ecosystems In our analysis lsquopurersquo spatial trends were more
attributed to species distribution than to lsquopurersquo environmental effects which link conta-
gious biological processes as important to the tree species distribution Furthermore
species and environmental data have a reasonable proportion of similar spatial structuring
identified by the largest proportion on the species variation due to spatially structured
environmental variation (1246) According to Borcard et al (1992) species and envi-
ronmental variables have in this case the same response to some common underlying
causes as the topographicndashflooding gradient In fact several studies have demonstrated the
direct effect of wetting and desiccation processes on both mineralogy and microbial
ecology of the sediment including nutrient dynamics (Baldwin and Mitchell 2000)
Once sediment or soils are submerged the inundation leads to a decrease in oxygen
contents and then resulting in progressive anaerobic conditions Rapid cycling of litter may
occur due to an increase on microbial activity which generates pulses on nutrient contents
and finally result in highly productive systems (Baldwin and Mitchell 2000) However a
negative effect is the rapid oxygen consumption which quickly leads to soil hypoxia or
anoxia When flood ends the anaerobic zones of sediments are newly oxygenated and
microbiota is replaced gradually to a new phase
Other spatially structured variables are sediment deposition and litter displacement
which are not covered in our study As demonstrated in lsquovarzearsquo forests of Amazonia
(Wittmann et al 2004) sediment deposition decrease toward upper sites and species
colonizing such lower areas show specific adaptations to the new site conditions regarding
to adventitious roots that probably offer mechanical support (Parolin et al 2004 Wittmann
et al 2004) However litter removal or deposition including seed bank may affect
directly species distribution once flooding and allied effects reallocate litter and seeds
among sites (Johansson et al 1996) Moreover studies have showed that flooding timing
frequency and magnitude can be used as indicators of sapling zonation on floodplain
forests (Vreugdenhil et al 2006)
All these processes are included in the unmeasured variables or spatially structuring
processes that have been missed by the geographical terms (Titeux et al 2004) and
accounted to the far unexplained variation (662) As also stressed by these authors the
stochastic spacendashtime fluctuations of each population the lsquounsaturationrsquo pattern (some
species do not use all suitable habitats) and species recording in not appropriated spatial
scales contribute to this unexplained variation too Notwithstanding occurrence data or
species abundances are often noisy (ter Braak 1995) and widespread in ecological studies
(Borcard et al 1992 Titeux et al 2004)
Richness and diversity patterns
Significant transitions occurred from the levee and depression to the following lower slope
transect regarding to stand structure and ecological groups Inversions on the proportion of
pioneershade-tolerant trees and auto-hydrochorous to anemochorous trees occurred in that
small transition and affected not only ecological groups but also species occurrences As a
consequence this zonation transect may consist spatially as a boundary for tolerant and
intolerant trees with regarding to flooding In fact few species occurred over the entire
flooding gradient and the lower slope also appears as an edge for several species
In a temporal scale the lower slope area will probably present more heterogeneous
spans in flooding events and it may consist in the most heterogeneous temporalndashspatial
sector across the topographic gradient which agree with the findings of Pollock et al
2384 Biodivers Conserv (2010) 192371ndash2387
123
Authors personal copy
(1998) from wetlands with different flooding regimes In the structured view of the
dynamic-equilibrium model (Huston 1994) different patches from different seral stages
result from spatial variation of disturbance frequencies If disturbance frequencies vary
over time a landscape could also contain patches of different seral stages (Pollock et al
1998) These authors used that assumption in a model with temporally synchronous dis-
turbance and found that at the community-scale level the results supported many pre-
dictions of the dynamic-equilibrium model especially regarding to species richness
In our study there are two major factors related to disturbance frequency First and
foremost is that once river floods are unpredictable due to a hydrological regime that varies
with occasional long-rainy periods temporal heterogeneity should be higher than in sea-
sonal predicted flood areas and second microtopography must create spatial heterogeneity
during floods events in the local area However well-drained upper sites are probably more
affected by inherent community processes as gap-phase dynamics and direct supply rates
of light (Stevens and Carson 2002) As related by Worbes et al (1992) and Parolin (2001)
hundreds of tree species with different phenological and other ecological traits grow in
seasonal flooded forests In such cases the cyclic alternation on floods and droughts drove
species to life history behavioral and morphological adaptations (Lytle and Poff 2004)
But in the case of unpredictable floods and droughts as assigned in our study bet-hedging
strategies might be evolved for example by persistent seed bank or asynchronous
reproductive phenologies (Brock 2003) although there are no conclusive studies related to
this theory (Lytle and Poff 2004)
Allowed by a transition in the ecological groups from the lower sites to the upper ones
species richness has a maximum at the lower slope transect probably due to higher het-
erogeneity in disturbance events (space and time) and correlated variables As reported by
Desilets and Houle (2005) the spatial gradient provides some evidences for stress toler-
ance and competition as factors structuring species distribution across the topographicndash
flooding gradient also boosted by an unpredictable pattern of floods that vary in frequency
timing and magnitude Lower sectors showed predicted ecological groups already
described for these areas (Budke et al 2007 2008 Junk et al 1989 Lytle and Poff 2004)
and expected structuring changes also occurred toward upper sites where the proportion of
shade-tolerant and small trees increased as well as anemochorous trees
In summary spatialndashtemporal and environmental variables are arranging tree species
distribution across the toposequence of our study site Furthermore predicted ecological
groups reflected the dynamics of disturbance in the topographicndashhydrological gradient
Species richness and diversity also reflected such pattern and were higher in the mid-sector
where occasional floods should prevent competitive exclusion and generate high envi-
ronmental heterogeneity
Acknowledgements We are grateful to the Programa de Pos-Graduacao em Botanica of the UniversidadeFederal do Rio Grande do SulmdashUFRGS for the opportunity to undertake this study and to CAPES Agencyfor the scholarship granted to the first author Our special thanks to Diogo lsquolsquoBagualrsquorsquo Lindenmaier forfieldwork assistance and to Ricardo Braga Eduardo Rossi and colleagues of the Laboratorio de Fitoeco-logiamdashUFRGS for critiques and suggestions We also appreciated the reviewing efforts of anonymouscontributors for providing useful comments to the manuscript
References
Baldwin DS Mitchell AM (2000) The effects of drying and re-flooding on the sediment and soil nutrientdynamics of lowland river-floodplain systems a synthesis Regul River 16457ndash467 doi1010021099-1646
Biodivers Conserv (2010) 192371ndash2387 2385
123
Authors personal copy
Barroso GM Morim MP Peixoto AL Ichaso CLF (1999) Frutos e sementes morfologia aplicada a si-stematica de dicotiledoneas Editora UFV Vicosa
Borcard D Legendre P Drapeau P (1992) Partialling out the spatial component of ecological variationEcology 731045ndash1055 doi1010292006WR005044
Brock MA (2003) Drought and aquatic community resilience the role of eggs and seeds in sediments oftemporary wetlands Freshw Biol 481207ndash1218 doi101046j1365-2427200301083x
Budke JC Athayde EA Giehl ELH Zachia RA Eisinger SM (2005) Composicao florıstica e estrategias dedispersao de especies lenhosas em uma floresta ribeirinha arroio Passo das Tropas Santa Maria RSBrasil Iheringia Bot 6017ndash24
Budke JC Jarenkow JA Oliveira-Filho AT Lindenmaier DS (2006) Padroes de riqueza e diversidade emrios de pequeno porte In Mariath JEA Santos RP (eds) Os avancos da botanica no inıcio do seculoXXI SBB Porto Alegre
Budke JC Jarenkow JA Oliveira-Filho AT (2007) Relationships between tree component structuretopography and soils of a riverine forest Rio Botucaraı southern Brazil Plant Ecol 189187ndash200 doi101007s11258-006-9174-8
Budke JC Jarenkow JA Oliveira-Filho AT (2008) Tree community features of two stands of riverine forestunder different flooding regimes in southern Brazil Flora 203162ndash174 doi101016jflora200703001
Cardoso E Schiavini I (2002) Relacao entre distribuicao de especies arboreas e topografia em um gradienteflorestal na Estacao Ecologica do Panga (Uberlandia MG) Rev Bras Bot 25277ndash289
Connell JH (1978) Diversity in tropical rain forests and coral reefs Science 1991302ndash1310 doi101126science19943351302
Damasceno-Junior GA Semir J Santos FAM Leitao-Filho HF (2005) Structure distribution of species andinundation in a riparian forest of Rio Paraguai Pantanal Brazil Flora 200119ndash135 doi101016jflora200409002
Desilets P Houle G (2005) Effects of resource availability and heterogeneity on the slope of the species-areacurve along a floodplain-upland gradient J Veg Sci 16487ndash496 doi1016581100-9233
Dufrene M Legendre P (1997) Species assemblages and indicator species the need for a flexible asym-metrical approach Ecol Monogr 67345ndash366
EMBRAPA (1997) Manual de metodos de analises de solo Empresa Brasileira de Pesquisa Agropecuariaand Centro Nacional de Pesquisas de Solos Rio de Janeiro
Ferreira LV (2000) Effects of flooding duration on species richness floristic composition and forest structurein river margin habitat in Amazonian blackwater floodplain forests implications for future design ofprotected areas Biodivers Conserv 91ndash14 doi101023A1008989811637
Ferreira LV Stohlgren TJ (1999) Effects of river level fluctuation on plant species richness diversity anddistribution in a floodplain forest in Central Amazonia Oecologia 120582ndash587 doi101007s004420050893
Ferreira C Piedade MTF Franco AC Goncalves JFC Junk WJ (2009) Adaptive strategies to tolerateprolonged flooding in seedlings of floodplain and upland populations of Himatanthus sucuuba aCentral Amazon tree Aquat Bot 90246ndash252 doi101016jaquabot200810006
Gotelli NJ Colwell RK (2001) Quantifying biodiversity procedures and pitfalls in the measurement andcomparison of species richness Ecol Lett 4379ndash391 doi101046j1461-0248200100230x
Guilherme FAG Oliveira-Filho AT Appolinario V Bearzoti E (2004) Effects of flooding regime andwoody bamboos on tree community dynamics in a section of tropical semideciduous forest in south-eastern Brazil Plant Ecol 17419ndash36 doi101023BVEGE000004605197752cd
Huston M (1994) Biological diversity the coexistence of species in changing landscapes CambridgeUniversity Press Cambridge
IPAGRO (1982) Atlas agroclimatico do Rio Grande do Sul Pallotti Porto AlegreJohansson ME Nilsson C Nilsson E (1996) Do rivers function as corridors for plant dispersal J Veg Sci
7593ndash598Junk WJ Bayley PB Sparks RE (1989) The flood pulse concept in river-floodplain systems Can J Fish
Aquat Sci 106110ndash127Legendre P (1993) Spatial autocorrelationmdashtrouble or new paradigm Ecology 741659ndash1673Leinz V (1949) Contribuicao a geologia dos derrames basalticos do Rio Grande do Sul Bol Fac Filos Let
581ndash55Lytle DA Poff NL (2004) Adaptation to natural flow regimes Trends Ecol Evol 1994ndash100 doi
101016jtree200310002McCune B Mefford MJ (1997) PCndashORD Multivariate analysis of ecological data version 436 MjM
Software Design Glaneden BeachMueller-Dombois D Ellenberg H (1974) Aims and methods of vegetation ecology John Wiley New York
2386 Biodivers Conserv (2010) 192371ndash2387
123
Authors personal copy
Neiff JJ (1990) Ideas para la interpretacion ecologica del Parana Interciencia 15424ndash441Neiff JJ Neiff M (2003) Pulso software para analisis de fenomenos recurrentes httpwwwneiffcom
Cited 25 May 2009Oliveira-Filho AT Vilela EA Gavilanes ML Carvalho DA (1994) Effect of flooding regime and understory
bamboos on the physiognomy and tree species composition of a tropical semideciduous forest in south-eastern Brazil Vegetatio 11399ndash124
Oliveira-Filho AT Curi N Vilela EA Carvalho DA (2001) Variation in tree community composition andstructure with changes in soil properties within a fragment of semideciduous forest in south-easternBrazil Edinb J Bot 58139ndash158 doi101017S0960428601000506
Oliveira-Filho AT Jarenkow JA Rodal MJN (2006) Floristic relationships of seasonally dry forests ofeastern South America based on tree species distribution patterns In Pennington RT Ratter JA LewisGP (eds) Neotropical savannas and dry forests plant diversity biogeography and conservation CRCPress Boca Raton
Parolin P (2001) Morphological and physiological adjustments to waterlogging and drought in seedlings ofAmazonian floodplain trees Oecologia 128326ndash335 doi101007s004420100660
Parolin P de Simone O Haase K Waldhoff D Rottenberger S Kuhn U Kesselmeier J Kleiss B SchmidtW Piedade MTF Junk WJ (2004) Central Amazonian floodplain forests tree adaptations in a pulsingsystem Bot Rev 70357ndash380 doi1016630006-8101(2004)070[0357CAFFTA]20CO2
Pijl L (1982) Principles of dispersal in higher plants Springer New YorkPillar VD (2006) Multivariate exploratory analysis randomization testing and bootstrap resampling version
2320 Departamento de Ecologia UFRGS Porto AlegrePollock MM Naiman RJ Hanley TA (1998) Plant species richness in riparian wetlandsmdasha test of biodi-
versity theory Ecology 7994ndash105Robertson KM (2006) Distributions of tree species along point bars of 10 rivers in the south-eastern US
Coastal Plain J Biogeogr 33121ndash132 doi101111j1365-2699200501371xRosales J Petts G Knab-Vispo C (2001) Ecological gradients within the riparian forests of the lower Caura
river Venezuela Plant Ecol 152101ndash118 doi101023A1011411020040Stevens MHH Carson WP (2002) Resource quantity not resource heterogeneity maintains plant diversity
Ecol Lett 5420ndash426 doi101046j1461-0248200200333xStreck EV Kampf N Dalmolin RSD Klamt E Nascimento PC Schneider P (2002) Solos do Rio Grande do
Sul EMATERRS and UFRGS Porto AlegreSwaine MD Whitmore TC (1988) On the definition of ecological species groups in tropical rain forests
Vegetatio 7581ndash86Tabacchi E Correll DL Hauer R Pinay G Planty-Tabacchi AM Wissmar R (1998) Development
maintenance and role of riparian vegetation in the river landscape Freshw Biol 40497ndash516 doi101046j1365-2427199800381x
ter Braack CJF Smilauer P (1998) Canoco reference manual and userrsquos guide to Canoco for Windowssoftware for canonical community ordination (version 40) Microcomputer Power Ithaca
ter Braak CJF (1995) Ordination In Jongman RHG ter Braak CJF van Togeren OFR (eds) Data analysis incommunity and landscape ecology Cambridge University Press New York
Titeux N Dufrene M Jacob JP Paquay M Defourny P (2004) Multivariate analysis of fine-scale breedingbird atlas using a geographical information system and partial canonical correspondence analysisenvironmental and spatial effects J Biogeogr 311841ndash1856 doi101111j1365-2699200401125x
Turner MG Gergel SE Dixon MD Miller JR (2004) Distribution and abundance of trees in floodplainforests of the Wisconsin river environmental influences at different scales J Veg Sci 15729ndash738
Vreugdenhil SJ Kramer K Pelsma T (2006) Effects of flooding duration frequency and depth on thepresence of saplings of six woody species in north-west Europe For Ecol Manage 23647ndash55 doi101016jforeco200608329
Walker KF Sheldon F Puckridge JT (1995) A perspective on dryland river ecosystems Regul River 1185ndash104 doi101002rrr3450110108
Weiher E (2003) Species richness along multiple gradients testing a general multivariate model in oaksavannas Oikos 101311ndash316 doi101034j1600-0706200312216x
Wittmann F Junk WJ Piedade MTF (2004) The varzea forests in Amazonia flooding and the highlydynamic geomorphology interact with natural forest succession For Ecol Manage 196199ndash212 doi101016jforeco200402060
Worbes M Klinge H Revilla JD Martius C (1992) On the dynamics floristic subdivision and geographicaldistribution of Varzea forests in Central Amazonia J Veg Sci 3553ndash564
Zar JH (1996) Biostatistical analysis Prentice-Hall New Jersey
Biodivers Conserv (2010) 192371ndash2387 2387
123
Authors personal copy
not present a topographic association due to wide distribution through the gradient On the
other hand the distinction among environmental and spatial effects showed that space
contributes significantly to the distribution of tree species (Fig 3) Environmental variables
selected by forward selection procedure (P 005) are summarized in Table 4 All geo-
graphical terms of the polynomial function were significant (P 005) during spatial CCA
and were add to the model The four CCA analyses provided the following results
1 CCA of the species matrix constrained by the environmental matrix sum of all
canonical eigenvalues = 0944 Monte Carlo tests for overall analysis F = 286
P 0001
2 CCA of the species matrix constrained by spatial matrix sum of all canonical
eigenvalues = 1017 Monte Carlo tests for overall analysis F = 359 P 0001
Fig 2 Ecological groups of vertical distribution (A) regeneration (B) and dispersal (C) in five transects ofriverine forest of Botucaraı river southern Brazil Pi pioneer Ld light-demanding St shade-tolerant Zoozoochorous Auto autochorous Ane anemochorous Hydro hydrochorous
Biodivers Conserv (2010) 192371ndash2387 2379
123
Authors personal copy
3 Environmental partial CCA (after removing the effect of geographical matrix) sum of
all canonical eigenvalues = 0416 Monte Carlo tests for overall analysis F = 135
P 0001
Table 3 Indicator species analysis (ISA) performed for species with density C10 individuals sampled infive transects with 20 sampling units each Botucaraı river southern Brazil
Biplots of environmental variables and species or environmental variables and sampling
units were depicted with the environmental partial analyses results (Fig 4) In this step
species distributions are purely based on speciesndashenvironment relationships The first two
axes accounted respectively for 268 (eigenvalue = 0112) and 181 (eigen-
value = 0075) of the speciesndashenvironment relationships and speciesndashenvironment corre-
lations for these axes were 0742 and 0649 (P 005) respectively The first four axes
accounted for 691 of speciesndashenvironment relationships Table 4 shows the intraset
correlations among variables and canonical coefficients with the first two axes The first
canonical axis was positively correlated with topography and several soil variables that are
Fig 3 Variation partitioning ofthe tree species matrix
Table 4 Intraset correlations among environmental variables selected for the model during lsquolsquoenviron-mentalrsquorsquo partial CCA of the species matrix and canonical coefficients of the first two axes
Environmental variables were selected by forward stepwise selection and included on the model if sig-nificant in Monte Carlo tests (P 005)
Biodivers Conserv (2010) 192371ndash2387 2381
123
Authors personal copy
influenced by flooding events Percentages of organic matter saturation of bases and cation
exchange capacity where higher through upper sites whereas aluminium contents were
higher in lower areas Thus this environmental gradient may affect tree distribution pat-
terns by restricting or facilitating species establishment As related in the Indicator
Analysis (although not accounting for spatial patterns directly) typical species of flooding
areas occurred near the levee and depression and the zonation was sharp once frequent
species of well-drained areas occurred only in the middle slope and ridgetop Sampling
units (Fig 5) also appeared distributed according to the toposequence with some over-
lapping due to species distribution
Rarefaction curves of species revealed significant differences on the expected total
number of species in each transect with higher richness within the middle slope (Fig 6)
A regression model (Fig 6) fitted diversity in a second-order polynomial regression
(y = -00586x2 03594x 291 R2 = 077) that showed the same pattern
Discussion
Environmental and spatial patterns
Tree species distribution throughout the topographical gradient indicated that both envi-
ronmental and spatial features were particularly important in predicting species and
community patterns This agrees with the well-know influence of geomorphic features and
hydrological regimes on riparian forests over different temporal and spatial scales
(Tabacchi et al 1998 Turner et al 2004 Desilets and Houle 2005) even though dis-
tinctions between environmental and spatial effects remain poorly studied (Titeux et al
Fig 4 Ordination biplotdepicting the two axes of theenvironmental partial CCA ofsampling units in a riverine forestin southern Brazil Eachsampling unit was identified bydifferent symbols according tothe respective transectEnvironmental variables arerepresented by their acronyms(see Table S1)
2382 Biodivers Conserv (2010) 192371ndash2387
123
Authors personal copy
2004) Here we employed a routine to partialling out the spatial effects on the analysis of
speciesndashenvironment relationships that also highlight the spatial component embedded in
such analysis (Legendre 1993) Once several biotic processes as growth mortality dis-
persal and predation influence the observed distributions of organisms (resulting in spatial
correlation) or if their distributions are dependent on explanatory variables which are
Fig 5 Ordination biplot depicting the two axes of the environmental partial CCA of species of a riverineforest in southern Brazil Species and environmental variables are represented by their acronyms (seeTable S1)
Fig 6 Rarefaction curves of tree species and Shannon diversity indices from five transects of riverine foreston the Botucaraı river southern Brazil Sampling units are representing different transects
Biodivers Conserv (2010) 192371ndash2387 2383
123
Authors personal copy
themselves spatially structured (Legendre 1993 Titeux et al 2004) spatial structuring is
an intrinsic component of ecosystems In our analysis lsquopurersquo spatial trends were more
attributed to species distribution than to lsquopurersquo environmental effects which link conta-
gious biological processes as important to the tree species distribution Furthermore
species and environmental data have a reasonable proportion of similar spatial structuring
identified by the largest proportion on the species variation due to spatially structured
environmental variation (1246) According to Borcard et al (1992) species and envi-
ronmental variables have in this case the same response to some common underlying
causes as the topographicndashflooding gradient In fact several studies have demonstrated the
direct effect of wetting and desiccation processes on both mineralogy and microbial
ecology of the sediment including nutrient dynamics (Baldwin and Mitchell 2000)
Once sediment or soils are submerged the inundation leads to a decrease in oxygen
contents and then resulting in progressive anaerobic conditions Rapid cycling of litter may
occur due to an increase on microbial activity which generates pulses on nutrient contents
and finally result in highly productive systems (Baldwin and Mitchell 2000) However a
negative effect is the rapid oxygen consumption which quickly leads to soil hypoxia or
anoxia When flood ends the anaerobic zones of sediments are newly oxygenated and
microbiota is replaced gradually to a new phase
Other spatially structured variables are sediment deposition and litter displacement
which are not covered in our study As demonstrated in lsquovarzearsquo forests of Amazonia
(Wittmann et al 2004) sediment deposition decrease toward upper sites and species
colonizing such lower areas show specific adaptations to the new site conditions regarding
to adventitious roots that probably offer mechanical support (Parolin et al 2004 Wittmann
et al 2004) However litter removal or deposition including seed bank may affect
directly species distribution once flooding and allied effects reallocate litter and seeds
among sites (Johansson et al 1996) Moreover studies have showed that flooding timing
frequency and magnitude can be used as indicators of sapling zonation on floodplain
forests (Vreugdenhil et al 2006)
All these processes are included in the unmeasured variables or spatially structuring
processes that have been missed by the geographical terms (Titeux et al 2004) and
accounted to the far unexplained variation (662) As also stressed by these authors the
stochastic spacendashtime fluctuations of each population the lsquounsaturationrsquo pattern (some
species do not use all suitable habitats) and species recording in not appropriated spatial
scales contribute to this unexplained variation too Notwithstanding occurrence data or
species abundances are often noisy (ter Braak 1995) and widespread in ecological studies
(Borcard et al 1992 Titeux et al 2004)
Richness and diversity patterns
Significant transitions occurred from the levee and depression to the following lower slope
transect regarding to stand structure and ecological groups Inversions on the proportion of
pioneershade-tolerant trees and auto-hydrochorous to anemochorous trees occurred in that
small transition and affected not only ecological groups but also species occurrences As a
consequence this zonation transect may consist spatially as a boundary for tolerant and
intolerant trees with regarding to flooding In fact few species occurred over the entire
flooding gradient and the lower slope also appears as an edge for several species
In a temporal scale the lower slope area will probably present more heterogeneous
spans in flooding events and it may consist in the most heterogeneous temporalndashspatial
sector across the topographic gradient which agree with the findings of Pollock et al
2384 Biodivers Conserv (2010) 192371ndash2387
123
Authors personal copy
(1998) from wetlands with different flooding regimes In the structured view of the
dynamic-equilibrium model (Huston 1994) different patches from different seral stages
result from spatial variation of disturbance frequencies If disturbance frequencies vary
over time a landscape could also contain patches of different seral stages (Pollock et al
1998) These authors used that assumption in a model with temporally synchronous dis-
turbance and found that at the community-scale level the results supported many pre-
dictions of the dynamic-equilibrium model especially regarding to species richness
In our study there are two major factors related to disturbance frequency First and
foremost is that once river floods are unpredictable due to a hydrological regime that varies
with occasional long-rainy periods temporal heterogeneity should be higher than in sea-
sonal predicted flood areas and second microtopography must create spatial heterogeneity
during floods events in the local area However well-drained upper sites are probably more
affected by inherent community processes as gap-phase dynamics and direct supply rates
of light (Stevens and Carson 2002) As related by Worbes et al (1992) and Parolin (2001)
hundreds of tree species with different phenological and other ecological traits grow in
seasonal flooded forests In such cases the cyclic alternation on floods and droughts drove
species to life history behavioral and morphological adaptations (Lytle and Poff 2004)
But in the case of unpredictable floods and droughts as assigned in our study bet-hedging
strategies might be evolved for example by persistent seed bank or asynchronous
reproductive phenologies (Brock 2003) although there are no conclusive studies related to
this theory (Lytle and Poff 2004)
Allowed by a transition in the ecological groups from the lower sites to the upper ones
species richness has a maximum at the lower slope transect probably due to higher het-
erogeneity in disturbance events (space and time) and correlated variables As reported by
Desilets and Houle (2005) the spatial gradient provides some evidences for stress toler-
ance and competition as factors structuring species distribution across the topographicndash
flooding gradient also boosted by an unpredictable pattern of floods that vary in frequency
timing and magnitude Lower sectors showed predicted ecological groups already
described for these areas (Budke et al 2007 2008 Junk et al 1989 Lytle and Poff 2004)
and expected structuring changes also occurred toward upper sites where the proportion of
shade-tolerant and small trees increased as well as anemochorous trees
In summary spatialndashtemporal and environmental variables are arranging tree species
distribution across the toposequence of our study site Furthermore predicted ecological
groups reflected the dynamics of disturbance in the topographicndashhydrological gradient
Species richness and diversity also reflected such pattern and were higher in the mid-sector
where occasional floods should prevent competitive exclusion and generate high envi-
ronmental heterogeneity
Acknowledgements We are grateful to the Programa de Pos-Graduacao em Botanica of the UniversidadeFederal do Rio Grande do SulmdashUFRGS for the opportunity to undertake this study and to CAPES Agencyfor the scholarship granted to the first author Our special thanks to Diogo lsquolsquoBagualrsquorsquo Lindenmaier forfieldwork assistance and to Ricardo Braga Eduardo Rossi and colleagues of the Laboratorio de Fitoeco-logiamdashUFRGS for critiques and suggestions We also appreciated the reviewing efforts of anonymouscontributors for providing useful comments to the manuscript
References
Baldwin DS Mitchell AM (2000) The effects of drying and re-flooding on the sediment and soil nutrientdynamics of lowland river-floodplain systems a synthesis Regul River 16457ndash467 doi1010021099-1646
Biodivers Conserv (2010) 192371ndash2387 2385
123
Authors personal copy
Barroso GM Morim MP Peixoto AL Ichaso CLF (1999) Frutos e sementes morfologia aplicada a si-stematica de dicotiledoneas Editora UFV Vicosa
Borcard D Legendre P Drapeau P (1992) Partialling out the spatial component of ecological variationEcology 731045ndash1055 doi1010292006WR005044
Brock MA (2003) Drought and aquatic community resilience the role of eggs and seeds in sediments oftemporary wetlands Freshw Biol 481207ndash1218 doi101046j1365-2427200301083x
Budke JC Athayde EA Giehl ELH Zachia RA Eisinger SM (2005) Composicao florıstica e estrategias dedispersao de especies lenhosas em uma floresta ribeirinha arroio Passo das Tropas Santa Maria RSBrasil Iheringia Bot 6017ndash24
Budke JC Jarenkow JA Oliveira-Filho AT Lindenmaier DS (2006) Padroes de riqueza e diversidade emrios de pequeno porte In Mariath JEA Santos RP (eds) Os avancos da botanica no inıcio do seculoXXI SBB Porto Alegre
Budke JC Jarenkow JA Oliveira-Filho AT (2007) Relationships between tree component structuretopography and soils of a riverine forest Rio Botucaraı southern Brazil Plant Ecol 189187ndash200 doi101007s11258-006-9174-8
Budke JC Jarenkow JA Oliveira-Filho AT (2008) Tree community features of two stands of riverine forestunder different flooding regimes in southern Brazil Flora 203162ndash174 doi101016jflora200703001
Cardoso E Schiavini I (2002) Relacao entre distribuicao de especies arboreas e topografia em um gradienteflorestal na Estacao Ecologica do Panga (Uberlandia MG) Rev Bras Bot 25277ndash289
Connell JH (1978) Diversity in tropical rain forests and coral reefs Science 1991302ndash1310 doi101126science19943351302
Damasceno-Junior GA Semir J Santos FAM Leitao-Filho HF (2005) Structure distribution of species andinundation in a riparian forest of Rio Paraguai Pantanal Brazil Flora 200119ndash135 doi101016jflora200409002
Desilets P Houle G (2005) Effects of resource availability and heterogeneity on the slope of the species-areacurve along a floodplain-upland gradient J Veg Sci 16487ndash496 doi1016581100-9233
Dufrene M Legendre P (1997) Species assemblages and indicator species the need for a flexible asym-metrical approach Ecol Monogr 67345ndash366
EMBRAPA (1997) Manual de metodos de analises de solo Empresa Brasileira de Pesquisa Agropecuariaand Centro Nacional de Pesquisas de Solos Rio de Janeiro
Ferreira LV (2000) Effects of flooding duration on species richness floristic composition and forest structurein river margin habitat in Amazonian blackwater floodplain forests implications for future design ofprotected areas Biodivers Conserv 91ndash14 doi101023A1008989811637
Ferreira LV Stohlgren TJ (1999) Effects of river level fluctuation on plant species richness diversity anddistribution in a floodplain forest in Central Amazonia Oecologia 120582ndash587 doi101007s004420050893
Ferreira C Piedade MTF Franco AC Goncalves JFC Junk WJ (2009) Adaptive strategies to tolerateprolonged flooding in seedlings of floodplain and upland populations of Himatanthus sucuuba aCentral Amazon tree Aquat Bot 90246ndash252 doi101016jaquabot200810006
Gotelli NJ Colwell RK (2001) Quantifying biodiversity procedures and pitfalls in the measurement andcomparison of species richness Ecol Lett 4379ndash391 doi101046j1461-0248200100230x
Guilherme FAG Oliveira-Filho AT Appolinario V Bearzoti E (2004) Effects of flooding regime andwoody bamboos on tree community dynamics in a section of tropical semideciduous forest in south-eastern Brazil Plant Ecol 17419ndash36 doi101023BVEGE000004605197752cd
Huston M (1994) Biological diversity the coexistence of species in changing landscapes CambridgeUniversity Press Cambridge
IPAGRO (1982) Atlas agroclimatico do Rio Grande do Sul Pallotti Porto AlegreJohansson ME Nilsson C Nilsson E (1996) Do rivers function as corridors for plant dispersal J Veg Sci
7593ndash598Junk WJ Bayley PB Sparks RE (1989) The flood pulse concept in river-floodplain systems Can J Fish
Aquat Sci 106110ndash127Legendre P (1993) Spatial autocorrelationmdashtrouble or new paradigm Ecology 741659ndash1673Leinz V (1949) Contribuicao a geologia dos derrames basalticos do Rio Grande do Sul Bol Fac Filos Let
581ndash55Lytle DA Poff NL (2004) Adaptation to natural flow regimes Trends Ecol Evol 1994ndash100 doi
101016jtree200310002McCune B Mefford MJ (1997) PCndashORD Multivariate analysis of ecological data version 436 MjM
Software Design Glaneden BeachMueller-Dombois D Ellenberg H (1974) Aims and methods of vegetation ecology John Wiley New York
2386 Biodivers Conserv (2010) 192371ndash2387
123
Authors personal copy
Neiff JJ (1990) Ideas para la interpretacion ecologica del Parana Interciencia 15424ndash441Neiff JJ Neiff M (2003) Pulso software para analisis de fenomenos recurrentes httpwwwneiffcom
Cited 25 May 2009Oliveira-Filho AT Vilela EA Gavilanes ML Carvalho DA (1994) Effect of flooding regime and understory
bamboos on the physiognomy and tree species composition of a tropical semideciduous forest in south-eastern Brazil Vegetatio 11399ndash124
Oliveira-Filho AT Curi N Vilela EA Carvalho DA (2001) Variation in tree community composition andstructure with changes in soil properties within a fragment of semideciduous forest in south-easternBrazil Edinb J Bot 58139ndash158 doi101017S0960428601000506
Oliveira-Filho AT Jarenkow JA Rodal MJN (2006) Floristic relationships of seasonally dry forests ofeastern South America based on tree species distribution patterns In Pennington RT Ratter JA LewisGP (eds) Neotropical savannas and dry forests plant diversity biogeography and conservation CRCPress Boca Raton
Parolin P (2001) Morphological and physiological adjustments to waterlogging and drought in seedlings ofAmazonian floodplain trees Oecologia 128326ndash335 doi101007s004420100660
Parolin P de Simone O Haase K Waldhoff D Rottenberger S Kuhn U Kesselmeier J Kleiss B SchmidtW Piedade MTF Junk WJ (2004) Central Amazonian floodplain forests tree adaptations in a pulsingsystem Bot Rev 70357ndash380 doi1016630006-8101(2004)070[0357CAFFTA]20CO2
Pijl L (1982) Principles of dispersal in higher plants Springer New YorkPillar VD (2006) Multivariate exploratory analysis randomization testing and bootstrap resampling version
2320 Departamento de Ecologia UFRGS Porto AlegrePollock MM Naiman RJ Hanley TA (1998) Plant species richness in riparian wetlandsmdasha test of biodi-
versity theory Ecology 7994ndash105Robertson KM (2006) Distributions of tree species along point bars of 10 rivers in the south-eastern US
Coastal Plain J Biogeogr 33121ndash132 doi101111j1365-2699200501371xRosales J Petts G Knab-Vispo C (2001) Ecological gradients within the riparian forests of the lower Caura
river Venezuela Plant Ecol 152101ndash118 doi101023A1011411020040Stevens MHH Carson WP (2002) Resource quantity not resource heterogeneity maintains plant diversity
Ecol Lett 5420ndash426 doi101046j1461-0248200200333xStreck EV Kampf N Dalmolin RSD Klamt E Nascimento PC Schneider P (2002) Solos do Rio Grande do
Sul EMATERRS and UFRGS Porto AlegreSwaine MD Whitmore TC (1988) On the definition of ecological species groups in tropical rain forests
Vegetatio 7581ndash86Tabacchi E Correll DL Hauer R Pinay G Planty-Tabacchi AM Wissmar R (1998) Development
maintenance and role of riparian vegetation in the river landscape Freshw Biol 40497ndash516 doi101046j1365-2427199800381x
ter Braack CJF Smilauer P (1998) Canoco reference manual and userrsquos guide to Canoco for Windowssoftware for canonical community ordination (version 40) Microcomputer Power Ithaca
ter Braak CJF (1995) Ordination In Jongman RHG ter Braak CJF van Togeren OFR (eds) Data analysis incommunity and landscape ecology Cambridge University Press New York
Titeux N Dufrene M Jacob JP Paquay M Defourny P (2004) Multivariate analysis of fine-scale breedingbird atlas using a geographical information system and partial canonical correspondence analysisenvironmental and spatial effects J Biogeogr 311841ndash1856 doi101111j1365-2699200401125x
Turner MG Gergel SE Dixon MD Miller JR (2004) Distribution and abundance of trees in floodplainforests of the Wisconsin river environmental influences at different scales J Veg Sci 15729ndash738
Vreugdenhil SJ Kramer K Pelsma T (2006) Effects of flooding duration frequency and depth on thepresence of saplings of six woody species in north-west Europe For Ecol Manage 23647ndash55 doi101016jforeco200608329
Walker KF Sheldon F Puckridge JT (1995) A perspective on dryland river ecosystems Regul River 1185ndash104 doi101002rrr3450110108
Weiher E (2003) Species richness along multiple gradients testing a general multivariate model in oaksavannas Oikos 101311ndash316 doi101034j1600-0706200312216x
Wittmann F Junk WJ Piedade MTF (2004) The varzea forests in Amazonia flooding and the highlydynamic geomorphology interact with natural forest succession For Ecol Manage 196199ndash212 doi101016jforeco200402060
Worbes M Klinge H Revilla JD Martius C (1992) On the dynamics floristic subdivision and geographicaldistribution of Varzea forests in Central Amazonia J Veg Sci 3553ndash564
Zar JH (1996) Biostatistical analysis Prentice-Hall New Jersey
Biodivers Conserv (2010) 192371ndash2387 2387
123
Authors personal copy
3 Environmental partial CCA (after removing the effect of geographical matrix) sum of
all canonical eigenvalues = 0416 Monte Carlo tests for overall analysis F = 135
P 0001
Table 3 Indicator species analysis (ISA) performed for species with density C10 individuals sampled infive transects with 20 sampling units each Botucaraı river southern Brazil
Biplots of environmental variables and species or environmental variables and sampling
units were depicted with the environmental partial analyses results (Fig 4) In this step
species distributions are purely based on speciesndashenvironment relationships The first two
axes accounted respectively for 268 (eigenvalue = 0112) and 181 (eigen-
value = 0075) of the speciesndashenvironment relationships and speciesndashenvironment corre-
lations for these axes were 0742 and 0649 (P 005) respectively The first four axes
accounted for 691 of speciesndashenvironment relationships Table 4 shows the intraset
correlations among variables and canonical coefficients with the first two axes The first
canonical axis was positively correlated with topography and several soil variables that are
Fig 3 Variation partitioning ofthe tree species matrix
Table 4 Intraset correlations among environmental variables selected for the model during lsquolsquoenviron-mentalrsquorsquo partial CCA of the species matrix and canonical coefficients of the first two axes
Environmental variables were selected by forward stepwise selection and included on the model if sig-nificant in Monte Carlo tests (P 005)
Biodivers Conserv (2010) 192371ndash2387 2381
123
Authors personal copy
influenced by flooding events Percentages of organic matter saturation of bases and cation
exchange capacity where higher through upper sites whereas aluminium contents were
higher in lower areas Thus this environmental gradient may affect tree distribution pat-
terns by restricting or facilitating species establishment As related in the Indicator
Analysis (although not accounting for spatial patterns directly) typical species of flooding
areas occurred near the levee and depression and the zonation was sharp once frequent
species of well-drained areas occurred only in the middle slope and ridgetop Sampling
units (Fig 5) also appeared distributed according to the toposequence with some over-
lapping due to species distribution
Rarefaction curves of species revealed significant differences on the expected total
number of species in each transect with higher richness within the middle slope (Fig 6)
A regression model (Fig 6) fitted diversity in a second-order polynomial regression
(y = -00586x2 03594x 291 R2 = 077) that showed the same pattern
Discussion
Environmental and spatial patterns
Tree species distribution throughout the topographical gradient indicated that both envi-
ronmental and spatial features were particularly important in predicting species and
community patterns This agrees with the well-know influence of geomorphic features and
hydrological regimes on riparian forests over different temporal and spatial scales
(Tabacchi et al 1998 Turner et al 2004 Desilets and Houle 2005) even though dis-
tinctions between environmental and spatial effects remain poorly studied (Titeux et al
Fig 4 Ordination biplotdepicting the two axes of theenvironmental partial CCA ofsampling units in a riverine forestin southern Brazil Eachsampling unit was identified bydifferent symbols according tothe respective transectEnvironmental variables arerepresented by their acronyms(see Table S1)
2382 Biodivers Conserv (2010) 192371ndash2387
123
Authors personal copy
2004) Here we employed a routine to partialling out the spatial effects on the analysis of
speciesndashenvironment relationships that also highlight the spatial component embedded in
such analysis (Legendre 1993) Once several biotic processes as growth mortality dis-
persal and predation influence the observed distributions of organisms (resulting in spatial
correlation) or if their distributions are dependent on explanatory variables which are
Fig 5 Ordination biplot depicting the two axes of the environmental partial CCA of species of a riverineforest in southern Brazil Species and environmental variables are represented by their acronyms (seeTable S1)
Fig 6 Rarefaction curves of tree species and Shannon diversity indices from five transects of riverine foreston the Botucaraı river southern Brazil Sampling units are representing different transects
Biodivers Conserv (2010) 192371ndash2387 2383
123
Authors personal copy
themselves spatially structured (Legendre 1993 Titeux et al 2004) spatial structuring is
an intrinsic component of ecosystems In our analysis lsquopurersquo spatial trends were more
attributed to species distribution than to lsquopurersquo environmental effects which link conta-
gious biological processes as important to the tree species distribution Furthermore
species and environmental data have a reasonable proportion of similar spatial structuring
identified by the largest proportion on the species variation due to spatially structured
environmental variation (1246) According to Borcard et al (1992) species and envi-
ronmental variables have in this case the same response to some common underlying
causes as the topographicndashflooding gradient In fact several studies have demonstrated the
direct effect of wetting and desiccation processes on both mineralogy and microbial
ecology of the sediment including nutrient dynamics (Baldwin and Mitchell 2000)
Once sediment or soils are submerged the inundation leads to a decrease in oxygen
contents and then resulting in progressive anaerobic conditions Rapid cycling of litter may
occur due to an increase on microbial activity which generates pulses on nutrient contents
and finally result in highly productive systems (Baldwin and Mitchell 2000) However a
negative effect is the rapid oxygen consumption which quickly leads to soil hypoxia or
anoxia When flood ends the anaerobic zones of sediments are newly oxygenated and
microbiota is replaced gradually to a new phase
Other spatially structured variables are sediment deposition and litter displacement
which are not covered in our study As demonstrated in lsquovarzearsquo forests of Amazonia
(Wittmann et al 2004) sediment deposition decrease toward upper sites and species
colonizing such lower areas show specific adaptations to the new site conditions regarding
to adventitious roots that probably offer mechanical support (Parolin et al 2004 Wittmann
et al 2004) However litter removal or deposition including seed bank may affect
directly species distribution once flooding and allied effects reallocate litter and seeds
among sites (Johansson et al 1996) Moreover studies have showed that flooding timing
frequency and magnitude can be used as indicators of sapling zonation on floodplain
forests (Vreugdenhil et al 2006)
All these processes are included in the unmeasured variables or spatially structuring
processes that have been missed by the geographical terms (Titeux et al 2004) and
accounted to the far unexplained variation (662) As also stressed by these authors the
stochastic spacendashtime fluctuations of each population the lsquounsaturationrsquo pattern (some
species do not use all suitable habitats) and species recording in not appropriated spatial
scales contribute to this unexplained variation too Notwithstanding occurrence data or
species abundances are often noisy (ter Braak 1995) and widespread in ecological studies
(Borcard et al 1992 Titeux et al 2004)
Richness and diversity patterns
Significant transitions occurred from the levee and depression to the following lower slope
transect regarding to stand structure and ecological groups Inversions on the proportion of
pioneershade-tolerant trees and auto-hydrochorous to anemochorous trees occurred in that
small transition and affected not only ecological groups but also species occurrences As a
consequence this zonation transect may consist spatially as a boundary for tolerant and
intolerant trees with regarding to flooding In fact few species occurred over the entire
flooding gradient and the lower slope also appears as an edge for several species
In a temporal scale the lower slope area will probably present more heterogeneous
spans in flooding events and it may consist in the most heterogeneous temporalndashspatial
sector across the topographic gradient which agree with the findings of Pollock et al
2384 Biodivers Conserv (2010) 192371ndash2387
123
Authors personal copy
(1998) from wetlands with different flooding regimes In the structured view of the
dynamic-equilibrium model (Huston 1994) different patches from different seral stages
result from spatial variation of disturbance frequencies If disturbance frequencies vary
over time a landscape could also contain patches of different seral stages (Pollock et al
1998) These authors used that assumption in a model with temporally synchronous dis-
turbance and found that at the community-scale level the results supported many pre-
dictions of the dynamic-equilibrium model especially regarding to species richness
In our study there are two major factors related to disturbance frequency First and
foremost is that once river floods are unpredictable due to a hydrological regime that varies
with occasional long-rainy periods temporal heterogeneity should be higher than in sea-
sonal predicted flood areas and second microtopography must create spatial heterogeneity
during floods events in the local area However well-drained upper sites are probably more
affected by inherent community processes as gap-phase dynamics and direct supply rates
of light (Stevens and Carson 2002) As related by Worbes et al (1992) and Parolin (2001)
hundreds of tree species with different phenological and other ecological traits grow in
seasonal flooded forests In such cases the cyclic alternation on floods and droughts drove
species to life history behavioral and morphological adaptations (Lytle and Poff 2004)
But in the case of unpredictable floods and droughts as assigned in our study bet-hedging
strategies might be evolved for example by persistent seed bank or asynchronous
reproductive phenologies (Brock 2003) although there are no conclusive studies related to
this theory (Lytle and Poff 2004)
Allowed by a transition in the ecological groups from the lower sites to the upper ones
species richness has a maximum at the lower slope transect probably due to higher het-
erogeneity in disturbance events (space and time) and correlated variables As reported by
Desilets and Houle (2005) the spatial gradient provides some evidences for stress toler-
ance and competition as factors structuring species distribution across the topographicndash
flooding gradient also boosted by an unpredictable pattern of floods that vary in frequency
timing and magnitude Lower sectors showed predicted ecological groups already
described for these areas (Budke et al 2007 2008 Junk et al 1989 Lytle and Poff 2004)
and expected structuring changes also occurred toward upper sites where the proportion of
shade-tolerant and small trees increased as well as anemochorous trees
In summary spatialndashtemporal and environmental variables are arranging tree species
distribution across the toposequence of our study site Furthermore predicted ecological
groups reflected the dynamics of disturbance in the topographicndashhydrological gradient
Species richness and diversity also reflected such pattern and were higher in the mid-sector
where occasional floods should prevent competitive exclusion and generate high envi-
ronmental heterogeneity
Acknowledgements We are grateful to the Programa de Pos-Graduacao em Botanica of the UniversidadeFederal do Rio Grande do SulmdashUFRGS for the opportunity to undertake this study and to CAPES Agencyfor the scholarship granted to the first author Our special thanks to Diogo lsquolsquoBagualrsquorsquo Lindenmaier forfieldwork assistance and to Ricardo Braga Eduardo Rossi and colleagues of the Laboratorio de Fitoeco-logiamdashUFRGS for critiques and suggestions We also appreciated the reviewing efforts of anonymouscontributors for providing useful comments to the manuscript
References
Baldwin DS Mitchell AM (2000) The effects of drying and re-flooding on the sediment and soil nutrientdynamics of lowland river-floodplain systems a synthesis Regul River 16457ndash467 doi1010021099-1646
Biodivers Conserv (2010) 192371ndash2387 2385
123
Authors personal copy
Barroso GM Morim MP Peixoto AL Ichaso CLF (1999) Frutos e sementes morfologia aplicada a si-stematica de dicotiledoneas Editora UFV Vicosa
Borcard D Legendre P Drapeau P (1992) Partialling out the spatial component of ecological variationEcology 731045ndash1055 doi1010292006WR005044
Brock MA (2003) Drought and aquatic community resilience the role of eggs and seeds in sediments oftemporary wetlands Freshw Biol 481207ndash1218 doi101046j1365-2427200301083x
Budke JC Athayde EA Giehl ELH Zachia RA Eisinger SM (2005) Composicao florıstica e estrategias dedispersao de especies lenhosas em uma floresta ribeirinha arroio Passo das Tropas Santa Maria RSBrasil Iheringia Bot 6017ndash24
Budke JC Jarenkow JA Oliveira-Filho AT Lindenmaier DS (2006) Padroes de riqueza e diversidade emrios de pequeno porte In Mariath JEA Santos RP (eds) Os avancos da botanica no inıcio do seculoXXI SBB Porto Alegre
Budke JC Jarenkow JA Oliveira-Filho AT (2007) Relationships between tree component structuretopography and soils of a riverine forest Rio Botucaraı southern Brazil Plant Ecol 189187ndash200 doi101007s11258-006-9174-8
Budke JC Jarenkow JA Oliveira-Filho AT (2008) Tree community features of two stands of riverine forestunder different flooding regimes in southern Brazil Flora 203162ndash174 doi101016jflora200703001
Cardoso E Schiavini I (2002) Relacao entre distribuicao de especies arboreas e topografia em um gradienteflorestal na Estacao Ecologica do Panga (Uberlandia MG) Rev Bras Bot 25277ndash289
Connell JH (1978) Diversity in tropical rain forests and coral reefs Science 1991302ndash1310 doi101126science19943351302
Damasceno-Junior GA Semir J Santos FAM Leitao-Filho HF (2005) Structure distribution of species andinundation in a riparian forest of Rio Paraguai Pantanal Brazil Flora 200119ndash135 doi101016jflora200409002
Desilets P Houle G (2005) Effects of resource availability and heterogeneity on the slope of the species-areacurve along a floodplain-upland gradient J Veg Sci 16487ndash496 doi1016581100-9233
Dufrene M Legendre P (1997) Species assemblages and indicator species the need for a flexible asym-metrical approach Ecol Monogr 67345ndash366
EMBRAPA (1997) Manual de metodos de analises de solo Empresa Brasileira de Pesquisa Agropecuariaand Centro Nacional de Pesquisas de Solos Rio de Janeiro
Ferreira LV (2000) Effects of flooding duration on species richness floristic composition and forest structurein river margin habitat in Amazonian blackwater floodplain forests implications for future design ofprotected areas Biodivers Conserv 91ndash14 doi101023A1008989811637
Ferreira LV Stohlgren TJ (1999) Effects of river level fluctuation on plant species richness diversity anddistribution in a floodplain forest in Central Amazonia Oecologia 120582ndash587 doi101007s004420050893
Ferreira C Piedade MTF Franco AC Goncalves JFC Junk WJ (2009) Adaptive strategies to tolerateprolonged flooding in seedlings of floodplain and upland populations of Himatanthus sucuuba aCentral Amazon tree Aquat Bot 90246ndash252 doi101016jaquabot200810006
Gotelli NJ Colwell RK (2001) Quantifying biodiversity procedures and pitfalls in the measurement andcomparison of species richness Ecol Lett 4379ndash391 doi101046j1461-0248200100230x
Guilherme FAG Oliveira-Filho AT Appolinario V Bearzoti E (2004) Effects of flooding regime andwoody bamboos on tree community dynamics in a section of tropical semideciduous forest in south-eastern Brazil Plant Ecol 17419ndash36 doi101023BVEGE000004605197752cd
Huston M (1994) Biological diversity the coexistence of species in changing landscapes CambridgeUniversity Press Cambridge
IPAGRO (1982) Atlas agroclimatico do Rio Grande do Sul Pallotti Porto AlegreJohansson ME Nilsson C Nilsson E (1996) Do rivers function as corridors for plant dispersal J Veg Sci
7593ndash598Junk WJ Bayley PB Sparks RE (1989) The flood pulse concept in river-floodplain systems Can J Fish
Aquat Sci 106110ndash127Legendre P (1993) Spatial autocorrelationmdashtrouble or new paradigm Ecology 741659ndash1673Leinz V (1949) Contribuicao a geologia dos derrames basalticos do Rio Grande do Sul Bol Fac Filos Let
581ndash55Lytle DA Poff NL (2004) Adaptation to natural flow regimes Trends Ecol Evol 1994ndash100 doi
101016jtree200310002McCune B Mefford MJ (1997) PCndashORD Multivariate analysis of ecological data version 436 MjM
Software Design Glaneden BeachMueller-Dombois D Ellenberg H (1974) Aims and methods of vegetation ecology John Wiley New York
2386 Biodivers Conserv (2010) 192371ndash2387
123
Authors personal copy
Neiff JJ (1990) Ideas para la interpretacion ecologica del Parana Interciencia 15424ndash441Neiff JJ Neiff M (2003) Pulso software para analisis de fenomenos recurrentes httpwwwneiffcom
Cited 25 May 2009Oliveira-Filho AT Vilela EA Gavilanes ML Carvalho DA (1994) Effect of flooding regime and understory
bamboos on the physiognomy and tree species composition of a tropical semideciduous forest in south-eastern Brazil Vegetatio 11399ndash124
Oliveira-Filho AT Curi N Vilela EA Carvalho DA (2001) Variation in tree community composition andstructure with changes in soil properties within a fragment of semideciduous forest in south-easternBrazil Edinb J Bot 58139ndash158 doi101017S0960428601000506
Oliveira-Filho AT Jarenkow JA Rodal MJN (2006) Floristic relationships of seasonally dry forests ofeastern South America based on tree species distribution patterns In Pennington RT Ratter JA LewisGP (eds) Neotropical savannas and dry forests plant diversity biogeography and conservation CRCPress Boca Raton
Parolin P (2001) Morphological and physiological adjustments to waterlogging and drought in seedlings ofAmazonian floodplain trees Oecologia 128326ndash335 doi101007s004420100660
Parolin P de Simone O Haase K Waldhoff D Rottenberger S Kuhn U Kesselmeier J Kleiss B SchmidtW Piedade MTF Junk WJ (2004) Central Amazonian floodplain forests tree adaptations in a pulsingsystem Bot Rev 70357ndash380 doi1016630006-8101(2004)070[0357CAFFTA]20CO2
Pijl L (1982) Principles of dispersal in higher plants Springer New YorkPillar VD (2006) Multivariate exploratory analysis randomization testing and bootstrap resampling version
2320 Departamento de Ecologia UFRGS Porto AlegrePollock MM Naiman RJ Hanley TA (1998) Plant species richness in riparian wetlandsmdasha test of biodi-
versity theory Ecology 7994ndash105Robertson KM (2006) Distributions of tree species along point bars of 10 rivers in the south-eastern US
Coastal Plain J Biogeogr 33121ndash132 doi101111j1365-2699200501371xRosales J Petts G Knab-Vispo C (2001) Ecological gradients within the riparian forests of the lower Caura
river Venezuela Plant Ecol 152101ndash118 doi101023A1011411020040Stevens MHH Carson WP (2002) Resource quantity not resource heterogeneity maintains plant diversity
Ecol Lett 5420ndash426 doi101046j1461-0248200200333xStreck EV Kampf N Dalmolin RSD Klamt E Nascimento PC Schneider P (2002) Solos do Rio Grande do
Sul EMATERRS and UFRGS Porto AlegreSwaine MD Whitmore TC (1988) On the definition of ecological species groups in tropical rain forests
Vegetatio 7581ndash86Tabacchi E Correll DL Hauer R Pinay G Planty-Tabacchi AM Wissmar R (1998) Development
maintenance and role of riparian vegetation in the river landscape Freshw Biol 40497ndash516 doi101046j1365-2427199800381x
ter Braack CJF Smilauer P (1998) Canoco reference manual and userrsquos guide to Canoco for Windowssoftware for canonical community ordination (version 40) Microcomputer Power Ithaca
ter Braak CJF (1995) Ordination In Jongman RHG ter Braak CJF van Togeren OFR (eds) Data analysis incommunity and landscape ecology Cambridge University Press New York
Titeux N Dufrene M Jacob JP Paquay M Defourny P (2004) Multivariate analysis of fine-scale breedingbird atlas using a geographical information system and partial canonical correspondence analysisenvironmental and spatial effects J Biogeogr 311841ndash1856 doi101111j1365-2699200401125x
Turner MG Gergel SE Dixon MD Miller JR (2004) Distribution and abundance of trees in floodplainforests of the Wisconsin river environmental influences at different scales J Veg Sci 15729ndash738
Vreugdenhil SJ Kramer K Pelsma T (2006) Effects of flooding duration frequency and depth on thepresence of saplings of six woody species in north-west Europe For Ecol Manage 23647ndash55 doi101016jforeco200608329
Walker KF Sheldon F Puckridge JT (1995) A perspective on dryland river ecosystems Regul River 1185ndash104 doi101002rrr3450110108
Weiher E (2003) Species richness along multiple gradients testing a general multivariate model in oaksavannas Oikos 101311ndash316 doi101034j1600-0706200312216x
Wittmann F Junk WJ Piedade MTF (2004) The varzea forests in Amazonia flooding and the highlydynamic geomorphology interact with natural forest succession For Ecol Manage 196199ndash212 doi101016jforeco200402060
Worbes M Klinge H Revilla JD Martius C (1992) On the dynamics floristic subdivision and geographicaldistribution of Varzea forests in Central Amazonia J Veg Sci 3553ndash564
Zar JH (1996) Biostatistical analysis Prentice-Hall New Jersey
Biodivers Conserv (2010) 192371ndash2387 2387
123
Authors personal copy
4 Spatial CCA (after removing the effects of environmental matrix) sum of all
canonical eigenvalues = 0489 Monte Carlo tests for overall analysis F = 178
P 0001
The total variation in the species matrix (total inertia) was 4238 According to Borcard
et al (1992) the percentage of the total variation in the species matrix that accounted for
different steps (partition) is numbered as follows (a) non-spatial environmental variation
Biplots of environmental variables and species or environmental variables and sampling
units were depicted with the environmental partial analyses results (Fig 4) In this step
species distributions are purely based on speciesndashenvironment relationships The first two
axes accounted respectively for 268 (eigenvalue = 0112) and 181 (eigen-
value = 0075) of the speciesndashenvironment relationships and speciesndashenvironment corre-
lations for these axes were 0742 and 0649 (P 005) respectively The first four axes
accounted for 691 of speciesndashenvironment relationships Table 4 shows the intraset
correlations among variables and canonical coefficients with the first two axes The first
canonical axis was positively correlated with topography and several soil variables that are
Fig 3 Variation partitioning ofthe tree species matrix
Table 4 Intraset correlations among environmental variables selected for the model during lsquolsquoenviron-mentalrsquorsquo partial CCA of the species matrix and canonical coefficients of the first two axes
Environmental variables were selected by forward stepwise selection and included on the model if sig-nificant in Monte Carlo tests (P 005)
Biodivers Conserv (2010) 192371ndash2387 2381
123
Authors personal copy
influenced by flooding events Percentages of organic matter saturation of bases and cation
exchange capacity where higher through upper sites whereas aluminium contents were
higher in lower areas Thus this environmental gradient may affect tree distribution pat-
terns by restricting or facilitating species establishment As related in the Indicator
Analysis (although not accounting for spatial patterns directly) typical species of flooding
areas occurred near the levee and depression and the zonation was sharp once frequent
species of well-drained areas occurred only in the middle slope and ridgetop Sampling
units (Fig 5) also appeared distributed according to the toposequence with some over-
lapping due to species distribution
Rarefaction curves of species revealed significant differences on the expected total
number of species in each transect with higher richness within the middle slope (Fig 6)
A regression model (Fig 6) fitted diversity in a second-order polynomial regression
(y = -00586x2 03594x 291 R2 = 077) that showed the same pattern
Discussion
Environmental and spatial patterns
Tree species distribution throughout the topographical gradient indicated that both envi-
ronmental and spatial features were particularly important in predicting species and
community patterns This agrees with the well-know influence of geomorphic features and
hydrological regimes on riparian forests over different temporal and spatial scales
(Tabacchi et al 1998 Turner et al 2004 Desilets and Houle 2005) even though dis-
tinctions between environmental and spatial effects remain poorly studied (Titeux et al
Fig 4 Ordination biplotdepicting the two axes of theenvironmental partial CCA ofsampling units in a riverine forestin southern Brazil Eachsampling unit was identified bydifferent symbols according tothe respective transectEnvironmental variables arerepresented by their acronyms(see Table S1)
2382 Biodivers Conserv (2010) 192371ndash2387
123
Authors personal copy
2004) Here we employed a routine to partialling out the spatial effects on the analysis of
speciesndashenvironment relationships that also highlight the spatial component embedded in
such analysis (Legendre 1993) Once several biotic processes as growth mortality dis-
persal and predation influence the observed distributions of organisms (resulting in spatial
correlation) or if their distributions are dependent on explanatory variables which are
Fig 5 Ordination biplot depicting the two axes of the environmental partial CCA of species of a riverineforest in southern Brazil Species and environmental variables are represented by their acronyms (seeTable S1)
Fig 6 Rarefaction curves of tree species and Shannon diversity indices from five transects of riverine foreston the Botucaraı river southern Brazil Sampling units are representing different transects
Biodivers Conserv (2010) 192371ndash2387 2383
123
Authors personal copy
themselves spatially structured (Legendre 1993 Titeux et al 2004) spatial structuring is
an intrinsic component of ecosystems In our analysis lsquopurersquo spatial trends were more
attributed to species distribution than to lsquopurersquo environmental effects which link conta-
gious biological processes as important to the tree species distribution Furthermore
species and environmental data have a reasonable proportion of similar spatial structuring
identified by the largest proportion on the species variation due to spatially structured
environmental variation (1246) According to Borcard et al (1992) species and envi-
ronmental variables have in this case the same response to some common underlying
causes as the topographicndashflooding gradient In fact several studies have demonstrated the
direct effect of wetting and desiccation processes on both mineralogy and microbial
ecology of the sediment including nutrient dynamics (Baldwin and Mitchell 2000)
Once sediment or soils are submerged the inundation leads to a decrease in oxygen
contents and then resulting in progressive anaerobic conditions Rapid cycling of litter may
occur due to an increase on microbial activity which generates pulses on nutrient contents
and finally result in highly productive systems (Baldwin and Mitchell 2000) However a
negative effect is the rapid oxygen consumption which quickly leads to soil hypoxia or
anoxia When flood ends the anaerobic zones of sediments are newly oxygenated and
microbiota is replaced gradually to a new phase
Other spatially structured variables are sediment deposition and litter displacement
which are not covered in our study As demonstrated in lsquovarzearsquo forests of Amazonia
(Wittmann et al 2004) sediment deposition decrease toward upper sites and species
colonizing such lower areas show specific adaptations to the new site conditions regarding
to adventitious roots that probably offer mechanical support (Parolin et al 2004 Wittmann
et al 2004) However litter removal or deposition including seed bank may affect
directly species distribution once flooding and allied effects reallocate litter and seeds
among sites (Johansson et al 1996) Moreover studies have showed that flooding timing
frequency and magnitude can be used as indicators of sapling zonation on floodplain
forests (Vreugdenhil et al 2006)
All these processes are included in the unmeasured variables or spatially structuring
processes that have been missed by the geographical terms (Titeux et al 2004) and
accounted to the far unexplained variation (662) As also stressed by these authors the
stochastic spacendashtime fluctuations of each population the lsquounsaturationrsquo pattern (some
species do not use all suitable habitats) and species recording in not appropriated spatial
scales contribute to this unexplained variation too Notwithstanding occurrence data or
species abundances are often noisy (ter Braak 1995) and widespread in ecological studies
(Borcard et al 1992 Titeux et al 2004)
Richness and diversity patterns
Significant transitions occurred from the levee and depression to the following lower slope
transect regarding to stand structure and ecological groups Inversions on the proportion of
pioneershade-tolerant trees and auto-hydrochorous to anemochorous trees occurred in that
small transition and affected not only ecological groups but also species occurrences As a
consequence this zonation transect may consist spatially as a boundary for tolerant and
intolerant trees with regarding to flooding In fact few species occurred over the entire
flooding gradient and the lower slope also appears as an edge for several species
In a temporal scale the lower slope area will probably present more heterogeneous
spans in flooding events and it may consist in the most heterogeneous temporalndashspatial
sector across the topographic gradient which agree with the findings of Pollock et al
2384 Biodivers Conserv (2010) 192371ndash2387
123
Authors personal copy
(1998) from wetlands with different flooding regimes In the structured view of the
dynamic-equilibrium model (Huston 1994) different patches from different seral stages
result from spatial variation of disturbance frequencies If disturbance frequencies vary
over time a landscape could also contain patches of different seral stages (Pollock et al
1998) These authors used that assumption in a model with temporally synchronous dis-
turbance and found that at the community-scale level the results supported many pre-
dictions of the dynamic-equilibrium model especially regarding to species richness
In our study there are two major factors related to disturbance frequency First and
foremost is that once river floods are unpredictable due to a hydrological regime that varies
with occasional long-rainy periods temporal heterogeneity should be higher than in sea-
sonal predicted flood areas and second microtopography must create spatial heterogeneity
during floods events in the local area However well-drained upper sites are probably more
affected by inherent community processes as gap-phase dynamics and direct supply rates
of light (Stevens and Carson 2002) As related by Worbes et al (1992) and Parolin (2001)
hundreds of tree species with different phenological and other ecological traits grow in
seasonal flooded forests In such cases the cyclic alternation on floods and droughts drove
species to life history behavioral and morphological adaptations (Lytle and Poff 2004)
But in the case of unpredictable floods and droughts as assigned in our study bet-hedging
strategies might be evolved for example by persistent seed bank or asynchronous
reproductive phenologies (Brock 2003) although there are no conclusive studies related to
this theory (Lytle and Poff 2004)
Allowed by a transition in the ecological groups from the lower sites to the upper ones
species richness has a maximum at the lower slope transect probably due to higher het-
erogeneity in disturbance events (space and time) and correlated variables As reported by
Desilets and Houle (2005) the spatial gradient provides some evidences for stress toler-
ance and competition as factors structuring species distribution across the topographicndash
flooding gradient also boosted by an unpredictable pattern of floods that vary in frequency
timing and magnitude Lower sectors showed predicted ecological groups already
described for these areas (Budke et al 2007 2008 Junk et al 1989 Lytle and Poff 2004)
and expected structuring changes also occurred toward upper sites where the proportion of
shade-tolerant and small trees increased as well as anemochorous trees
In summary spatialndashtemporal and environmental variables are arranging tree species
distribution across the toposequence of our study site Furthermore predicted ecological
groups reflected the dynamics of disturbance in the topographicndashhydrological gradient
Species richness and diversity also reflected such pattern and were higher in the mid-sector
where occasional floods should prevent competitive exclusion and generate high envi-
ronmental heterogeneity
Acknowledgements We are grateful to the Programa de Pos-Graduacao em Botanica of the UniversidadeFederal do Rio Grande do SulmdashUFRGS for the opportunity to undertake this study and to CAPES Agencyfor the scholarship granted to the first author Our special thanks to Diogo lsquolsquoBagualrsquorsquo Lindenmaier forfieldwork assistance and to Ricardo Braga Eduardo Rossi and colleagues of the Laboratorio de Fitoeco-logiamdashUFRGS for critiques and suggestions We also appreciated the reviewing efforts of anonymouscontributors for providing useful comments to the manuscript
References
Baldwin DS Mitchell AM (2000) The effects of drying and re-flooding on the sediment and soil nutrientdynamics of lowland river-floodplain systems a synthesis Regul River 16457ndash467 doi1010021099-1646
Biodivers Conserv (2010) 192371ndash2387 2385
123
Authors personal copy
Barroso GM Morim MP Peixoto AL Ichaso CLF (1999) Frutos e sementes morfologia aplicada a si-stematica de dicotiledoneas Editora UFV Vicosa
Borcard D Legendre P Drapeau P (1992) Partialling out the spatial component of ecological variationEcology 731045ndash1055 doi1010292006WR005044
Brock MA (2003) Drought and aquatic community resilience the role of eggs and seeds in sediments oftemporary wetlands Freshw Biol 481207ndash1218 doi101046j1365-2427200301083x
Budke JC Athayde EA Giehl ELH Zachia RA Eisinger SM (2005) Composicao florıstica e estrategias dedispersao de especies lenhosas em uma floresta ribeirinha arroio Passo das Tropas Santa Maria RSBrasil Iheringia Bot 6017ndash24
Budke JC Jarenkow JA Oliveira-Filho AT Lindenmaier DS (2006) Padroes de riqueza e diversidade emrios de pequeno porte In Mariath JEA Santos RP (eds) Os avancos da botanica no inıcio do seculoXXI SBB Porto Alegre
Budke JC Jarenkow JA Oliveira-Filho AT (2007) Relationships between tree component structuretopography and soils of a riverine forest Rio Botucaraı southern Brazil Plant Ecol 189187ndash200 doi101007s11258-006-9174-8
Budke JC Jarenkow JA Oliveira-Filho AT (2008) Tree community features of two stands of riverine forestunder different flooding regimes in southern Brazil Flora 203162ndash174 doi101016jflora200703001
Cardoso E Schiavini I (2002) Relacao entre distribuicao de especies arboreas e topografia em um gradienteflorestal na Estacao Ecologica do Panga (Uberlandia MG) Rev Bras Bot 25277ndash289
Connell JH (1978) Diversity in tropical rain forests and coral reefs Science 1991302ndash1310 doi101126science19943351302
Damasceno-Junior GA Semir J Santos FAM Leitao-Filho HF (2005) Structure distribution of species andinundation in a riparian forest of Rio Paraguai Pantanal Brazil Flora 200119ndash135 doi101016jflora200409002
Desilets P Houle G (2005) Effects of resource availability and heterogeneity on the slope of the species-areacurve along a floodplain-upland gradient J Veg Sci 16487ndash496 doi1016581100-9233
Dufrene M Legendre P (1997) Species assemblages and indicator species the need for a flexible asym-metrical approach Ecol Monogr 67345ndash366
EMBRAPA (1997) Manual de metodos de analises de solo Empresa Brasileira de Pesquisa Agropecuariaand Centro Nacional de Pesquisas de Solos Rio de Janeiro
Ferreira LV (2000) Effects of flooding duration on species richness floristic composition and forest structurein river margin habitat in Amazonian blackwater floodplain forests implications for future design ofprotected areas Biodivers Conserv 91ndash14 doi101023A1008989811637
Ferreira LV Stohlgren TJ (1999) Effects of river level fluctuation on plant species richness diversity anddistribution in a floodplain forest in Central Amazonia Oecologia 120582ndash587 doi101007s004420050893
Ferreira C Piedade MTF Franco AC Goncalves JFC Junk WJ (2009) Adaptive strategies to tolerateprolonged flooding in seedlings of floodplain and upland populations of Himatanthus sucuuba aCentral Amazon tree Aquat Bot 90246ndash252 doi101016jaquabot200810006
Gotelli NJ Colwell RK (2001) Quantifying biodiversity procedures and pitfalls in the measurement andcomparison of species richness Ecol Lett 4379ndash391 doi101046j1461-0248200100230x
Guilherme FAG Oliveira-Filho AT Appolinario V Bearzoti E (2004) Effects of flooding regime andwoody bamboos on tree community dynamics in a section of tropical semideciduous forest in south-eastern Brazil Plant Ecol 17419ndash36 doi101023BVEGE000004605197752cd
Huston M (1994) Biological diversity the coexistence of species in changing landscapes CambridgeUniversity Press Cambridge
IPAGRO (1982) Atlas agroclimatico do Rio Grande do Sul Pallotti Porto AlegreJohansson ME Nilsson C Nilsson E (1996) Do rivers function as corridors for plant dispersal J Veg Sci
7593ndash598Junk WJ Bayley PB Sparks RE (1989) The flood pulse concept in river-floodplain systems Can J Fish
Aquat Sci 106110ndash127Legendre P (1993) Spatial autocorrelationmdashtrouble or new paradigm Ecology 741659ndash1673Leinz V (1949) Contribuicao a geologia dos derrames basalticos do Rio Grande do Sul Bol Fac Filos Let
581ndash55Lytle DA Poff NL (2004) Adaptation to natural flow regimes Trends Ecol Evol 1994ndash100 doi
101016jtree200310002McCune B Mefford MJ (1997) PCndashORD Multivariate analysis of ecological data version 436 MjM
Software Design Glaneden BeachMueller-Dombois D Ellenberg H (1974) Aims and methods of vegetation ecology John Wiley New York
2386 Biodivers Conserv (2010) 192371ndash2387
123
Authors personal copy
Neiff JJ (1990) Ideas para la interpretacion ecologica del Parana Interciencia 15424ndash441Neiff JJ Neiff M (2003) Pulso software para analisis de fenomenos recurrentes httpwwwneiffcom
Cited 25 May 2009Oliveira-Filho AT Vilela EA Gavilanes ML Carvalho DA (1994) Effect of flooding regime and understory
bamboos on the physiognomy and tree species composition of a tropical semideciduous forest in south-eastern Brazil Vegetatio 11399ndash124
Oliveira-Filho AT Curi N Vilela EA Carvalho DA (2001) Variation in tree community composition andstructure with changes in soil properties within a fragment of semideciduous forest in south-easternBrazil Edinb J Bot 58139ndash158 doi101017S0960428601000506
Oliveira-Filho AT Jarenkow JA Rodal MJN (2006) Floristic relationships of seasonally dry forests ofeastern South America based on tree species distribution patterns In Pennington RT Ratter JA LewisGP (eds) Neotropical savannas and dry forests plant diversity biogeography and conservation CRCPress Boca Raton
Parolin P (2001) Morphological and physiological adjustments to waterlogging and drought in seedlings ofAmazonian floodplain trees Oecologia 128326ndash335 doi101007s004420100660
Parolin P de Simone O Haase K Waldhoff D Rottenberger S Kuhn U Kesselmeier J Kleiss B SchmidtW Piedade MTF Junk WJ (2004) Central Amazonian floodplain forests tree adaptations in a pulsingsystem Bot Rev 70357ndash380 doi1016630006-8101(2004)070[0357CAFFTA]20CO2
Pijl L (1982) Principles of dispersal in higher plants Springer New YorkPillar VD (2006) Multivariate exploratory analysis randomization testing and bootstrap resampling version
2320 Departamento de Ecologia UFRGS Porto AlegrePollock MM Naiman RJ Hanley TA (1998) Plant species richness in riparian wetlandsmdasha test of biodi-
versity theory Ecology 7994ndash105Robertson KM (2006) Distributions of tree species along point bars of 10 rivers in the south-eastern US
Coastal Plain J Biogeogr 33121ndash132 doi101111j1365-2699200501371xRosales J Petts G Knab-Vispo C (2001) Ecological gradients within the riparian forests of the lower Caura
river Venezuela Plant Ecol 152101ndash118 doi101023A1011411020040Stevens MHH Carson WP (2002) Resource quantity not resource heterogeneity maintains plant diversity
Ecol Lett 5420ndash426 doi101046j1461-0248200200333xStreck EV Kampf N Dalmolin RSD Klamt E Nascimento PC Schneider P (2002) Solos do Rio Grande do
Sul EMATERRS and UFRGS Porto AlegreSwaine MD Whitmore TC (1988) On the definition of ecological species groups in tropical rain forests
Vegetatio 7581ndash86Tabacchi E Correll DL Hauer R Pinay G Planty-Tabacchi AM Wissmar R (1998) Development
maintenance and role of riparian vegetation in the river landscape Freshw Biol 40497ndash516 doi101046j1365-2427199800381x
ter Braack CJF Smilauer P (1998) Canoco reference manual and userrsquos guide to Canoco for Windowssoftware for canonical community ordination (version 40) Microcomputer Power Ithaca
ter Braak CJF (1995) Ordination In Jongman RHG ter Braak CJF van Togeren OFR (eds) Data analysis incommunity and landscape ecology Cambridge University Press New York
Titeux N Dufrene M Jacob JP Paquay M Defourny P (2004) Multivariate analysis of fine-scale breedingbird atlas using a geographical information system and partial canonical correspondence analysisenvironmental and spatial effects J Biogeogr 311841ndash1856 doi101111j1365-2699200401125x
Turner MG Gergel SE Dixon MD Miller JR (2004) Distribution and abundance of trees in floodplainforests of the Wisconsin river environmental influences at different scales J Veg Sci 15729ndash738
Vreugdenhil SJ Kramer K Pelsma T (2006) Effects of flooding duration frequency and depth on thepresence of saplings of six woody species in north-west Europe For Ecol Manage 23647ndash55 doi101016jforeco200608329
Walker KF Sheldon F Puckridge JT (1995) A perspective on dryland river ecosystems Regul River 1185ndash104 doi101002rrr3450110108
Weiher E (2003) Species richness along multiple gradients testing a general multivariate model in oaksavannas Oikos 101311ndash316 doi101034j1600-0706200312216x
Wittmann F Junk WJ Piedade MTF (2004) The varzea forests in Amazonia flooding and the highlydynamic geomorphology interact with natural forest succession For Ecol Manage 196199ndash212 doi101016jforeco200402060
Worbes M Klinge H Revilla JD Martius C (1992) On the dynamics floristic subdivision and geographicaldistribution of Varzea forests in Central Amazonia J Veg Sci 3553ndash564
Zar JH (1996) Biostatistical analysis Prentice-Hall New Jersey
Biodivers Conserv (2010) 192371ndash2387 2387
123
Authors personal copy
influenced by flooding events Percentages of organic matter saturation of bases and cation
exchange capacity where higher through upper sites whereas aluminium contents were
higher in lower areas Thus this environmental gradient may affect tree distribution pat-
terns by restricting or facilitating species establishment As related in the Indicator
Analysis (although not accounting for spatial patterns directly) typical species of flooding
areas occurred near the levee and depression and the zonation was sharp once frequent
species of well-drained areas occurred only in the middle slope and ridgetop Sampling
units (Fig 5) also appeared distributed according to the toposequence with some over-
lapping due to species distribution
Rarefaction curves of species revealed significant differences on the expected total
number of species in each transect with higher richness within the middle slope (Fig 6)
A regression model (Fig 6) fitted diversity in a second-order polynomial regression
(y = -00586x2 03594x 291 R2 = 077) that showed the same pattern
Discussion
Environmental and spatial patterns
Tree species distribution throughout the topographical gradient indicated that both envi-
ronmental and spatial features were particularly important in predicting species and
community patterns This agrees with the well-know influence of geomorphic features and
hydrological regimes on riparian forests over different temporal and spatial scales
(Tabacchi et al 1998 Turner et al 2004 Desilets and Houle 2005) even though dis-
tinctions between environmental and spatial effects remain poorly studied (Titeux et al
Fig 4 Ordination biplotdepicting the two axes of theenvironmental partial CCA ofsampling units in a riverine forestin southern Brazil Eachsampling unit was identified bydifferent symbols according tothe respective transectEnvironmental variables arerepresented by their acronyms(see Table S1)
2382 Biodivers Conserv (2010) 192371ndash2387
123
Authors personal copy
2004) Here we employed a routine to partialling out the spatial effects on the analysis of
speciesndashenvironment relationships that also highlight the spatial component embedded in
such analysis (Legendre 1993) Once several biotic processes as growth mortality dis-
persal and predation influence the observed distributions of organisms (resulting in spatial
correlation) or if their distributions are dependent on explanatory variables which are
Fig 5 Ordination biplot depicting the two axes of the environmental partial CCA of species of a riverineforest in southern Brazil Species and environmental variables are represented by their acronyms (seeTable S1)
Fig 6 Rarefaction curves of tree species and Shannon diversity indices from five transects of riverine foreston the Botucaraı river southern Brazil Sampling units are representing different transects
Biodivers Conserv (2010) 192371ndash2387 2383
123
Authors personal copy
themselves spatially structured (Legendre 1993 Titeux et al 2004) spatial structuring is
an intrinsic component of ecosystems In our analysis lsquopurersquo spatial trends were more
attributed to species distribution than to lsquopurersquo environmental effects which link conta-
gious biological processes as important to the tree species distribution Furthermore
species and environmental data have a reasonable proportion of similar spatial structuring
identified by the largest proportion on the species variation due to spatially structured
environmental variation (1246) According to Borcard et al (1992) species and envi-
ronmental variables have in this case the same response to some common underlying
causes as the topographicndashflooding gradient In fact several studies have demonstrated the
direct effect of wetting and desiccation processes on both mineralogy and microbial
ecology of the sediment including nutrient dynamics (Baldwin and Mitchell 2000)
Once sediment or soils are submerged the inundation leads to a decrease in oxygen
contents and then resulting in progressive anaerobic conditions Rapid cycling of litter may
occur due to an increase on microbial activity which generates pulses on nutrient contents
and finally result in highly productive systems (Baldwin and Mitchell 2000) However a
negative effect is the rapid oxygen consumption which quickly leads to soil hypoxia or
anoxia When flood ends the anaerobic zones of sediments are newly oxygenated and
microbiota is replaced gradually to a new phase
Other spatially structured variables are sediment deposition and litter displacement
which are not covered in our study As demonstrated in lsquovarzearsquo forests of Amazonia
(Wittmann et al 2004) sediment deposition decrease toward upper sites and species
colonizing such lower areas show specific adaptations to the new site conditions regarding
to adventitious roots that probably offer mechanical support (Parolin et al 2004 Wittmann
et al 2004) However litter removal or deposition including seed bank may affect
directly species distribution once flooding and allied effects reallocate litter and seeds
among sites (Johansson et al 1996) Moreover studies have showed that flooding timing
frequency and magnitude can be used as indicators of sapling zonation on floodplain
forests (Vreugdenhil et al 2006)
All these processes are included in the unmeasured variables or spatially structuring
processes that have been missed by the geographical terms (Titeux et al 2004) and
accounted to the far unexplained variation (662) As also stressed by these authors the
stochastic spacendashtime fluctuations of each population the lsquounsaturationrsquo pattern (some
species do not use all suitable habitats) and species recording in not appropriated spatial
scales contribute to this unexplained variation too Notwithstanding occurrence data or
species abundances are often noisy (ter Braak 1995) and widespread in ecological studies
(Borcard et al 1992 Titeux et al 2004)
Richness and diversity patterns
Significant transitions occurred from the levee and depression to the following lower slope
transect regarding to stand structure and ecological groups Inversions on the proportion of
pioneershade-tolerant trees and auto-hydrochorous to anemochorous trees occurred in that
small transition and affected not only ecological groups but also species occurrences As a
consequence this zonation transect may consist spatially as a boundary for tolerant and
intolerant trees with regarding to flooding In fact few species occurred over the entire
flooding gradient and the lower slope also appears as an edge for several species
In a temporal scale the lower slope area will probably present more heterogeneous
spans in flooding events and it may consist in the most heterogeneous temporalndashspatial
sector across the topographic gradient which agree with the findings of Pollock et al
2384 Biodivers Conserv (2010) 192371ndash2387
123
Authors personal copy
(1998) from wetlands with different flooding regimes In the structured view of the
dynamic-equilibrium model (Huston 1994) different patches from different seral stages
result from spatial variation of disturbance frequencies If disturbance frequencies vary
over time a landscape could also contain patches of different seral stages (Pollock et al
1998) These authors used that assumption in a model with temporally synchronous dis-
turbance and found that at the community-scale level the results supported many pre-
dictions of the dynamic-equilibrium model especially regarding to species richness
In our study there are two major factors related to disturbance frequency First and
foremost is that once river floods are unpredictable due to a hydrological regime that varies
with occasional long-rainy periods temporal heterogeneity should be higher than in sea-
sonal predicted flood areas and second microtopography must create spatial heterogeneity
during floods events in the local area However well-drained upper sites are probably more
affected by inherent community processes as gap-phase dynamics and direct supply rates
of light (Stevens and Carson 2002) As related by Worbes et al (1992) and Parolin (2001)
hundreds of tree species with different phenological and other ecological traits grow in
seasonal flooded forests In such cases the cyclic alternation on floods and droughts drove
species to life history behavioral and morphological adaptations (Lytle and Poff 2004)
But in the case of unpredictable floods and droughts as assigned in our study bet-hedging
strategies might be evolved for example by persistent seed bank or asynchronous
reproductive phenologies (Brock 2003) although there are no conclusive studies related to
this theory (Lytle and Poff 2004)
Allowed by a transition in the ecological groups from the lower sites to the upper ones
species richness has a maximum at the lower slope transect probably due to higher het-
erogeneity in disturbance events (space and time) and correlated variables As reported by
Desilets and Houle (2005) the spatial gradient provides some evidences for stress toler-
ance and competition as factors structuring species distribution across the topographicndash
flooding gradient also boosted by an unpredictable pattern of floods that vary in frequency
timing and magnitude Lower sectors showed predicted ecological groups already
described for these areas (Budke et al 2007 2008 Junk et al 1989 Lytle and Poff 2004)
and expected structuring changes also occurred toward upper sites where the proportion of
shade-tolerant and small trees increased as well as anemochorous trees
In summary spatialndashtemporal and environmental variables are arranging tree species
distribution across the toposequence of our study site Furthermore predicted ecological
groups reflected the dynamics of disturbance in the topographicndashhydrological gradient
Species richness and diversity also reflected such pattern and were higher in the mid-sector
where occasional floods should prevent competitive exclusion and generate high envi-
ronmental heterogeneity
Acknowledgements We are grateful to the Programa de Pos-Graduacao em Botanica of the UniversidadeFederal do Rio Grande do SulmdashUFRGS for the opportunity to undertake this study and to CAPES Agencyfor the scholarship granted to the first author Our special thanks to Diogo lsquolsquoBagualrsquorsquo Lindenmaier forfieldwork assistance and to Ricardo Braga Eduardo Rossi and colleagues of the Laboratorio de Fitoeco-logiamdashUFRGS for critiques and suggestions We also appreciated the reviewing efforts of anonymouscontributors for providing useful comments to the manuscript
References
Baldwin DS Mitchell AM (2000) The effects of drying and re-flooding on the sediment and soil nutrientdynamics of lowland river-floodplain systems a synthesis Regul River 16457ndash467 doi1010021099-1646
Biodivers Conserv (2010) 192371ndash2387 2385
123
Authors personal copy
Barroso GM Morim MP Peixoto AL Ichaso CLF (1999) Frutos e sementes morfologia aplicada a si-stematica de dicotiledoneas Editora UFV Vicosa
Borcard D Legendre P Drapeau P (1992) Partialling out the spatial component of ecological variationEcology 731045ndash1055 doi1010292006WR005044
Brock MA (2003) Drought and aquatic community resilience the role of eggs and seeds in sediments oftemporary wetlands Freshw Biol 481207ndash1218 doi101046j1365-2427200301083x
Budke JC Athayde EA Giehl ELH Zachia RA Eisinger SM (2005) Composicao florıstica e estrategias dedispersao de especies lenhosas em uma floresta ribeirinha arroio Passo das Tropas Santa Maria RSBrasil Iheringia Bot 6017ndash24
Budke JC Jarenkow JA Oliveira-Filho AT Lindenmaier DS (2006) Padroes de riqueza e diversidade emrios de pequeno porte In Mariath JEA Santos RP (eds) Os avancos da botanica no inıcio do seculoXXI SBB Porto Alegre
Budke JC Jarenkow JA Oliveira-Filho AT (2007) Relationships between tree component structuretopography and soils of a riverine forest Rio Botucaraı southern Brazil Plant Ecol 189187ndash200 doi101007s11258-006-9174-8
Budke JC Jarenkow JA Oliveira-Filho AT (2008) Tree community features of two stands of riverine forestunder different flooding regimes in southern Brazil Flora 203162ndash174 doi101016jflora200703001
Cardoso E Schiavini I (2002) Relacao entre distribuicao de especies arboreas e topografia em um gradienteflorestal na Estacao Ecologica do Panga (Uberlandia MG) Rev Bras Bot 25277ndash289
Connell JH (1978) Diversity in tropical rain forests and coral reefs Science 1991302ndash1310 doi101126science19943351302
Damasceno-Junior GA Semir J Santos FAM Leitao-Filho HF (2005) Structure distribution of species andinundation in a riparian forest of Rio Paraguai Pantanal Brazil Flora 200119ndash135 doi101016jflora200409002
Desilets P Houle G (2005) Effects of resource availability and heterogeneity on the slope of the species-areacurve along a floodplain-upland gradient J Veg Sci 16487ndash496 doi1016581100-9233
Dufrene M Legendre P (1997) Species assemblages and indicator species the need for a flexible asym-metrical approach Ecol Monogr 67345ndash366
EMBRAPA (1997) Manual de metodos de analises de solo Empresa Brasileira de Pesquisa Agropecuariaand Centro Nacional de Pesquisas de Solos Rio de Janeiro
Ferreira LV (2000) Effects of flooding duration on species richness floristic composition and forest structurein river margin habitat in Amazonian blackwater floodplain forests implications for future design ofprotected areas Biodivers Conserv 91ndash14 doi101023A1008989811637
Ferreira LV Stohlgren TJ (1999) Effects of river level fluctuation on plant species richness diversity anddistribution in a floodplain forest in Central Amazonia Oecologia 120582ndash587 doi101007s004420050893
Ferreira C Piedade MTF Franco AC Goncalves JFC Junk WJ (2009) Adaptive strategies to tolerateprolonged flooding in seedlings of floodplain and upland populations of Himatanthus sucuuba aCentral Amazon tree Aquat Bot 90246ndash252 doi101016jaquabot200810006
Gotelli NJ Colwell RK (2001) Quantifying biodiversity procedures and pitfalls in the measurement andcomparison of species richness Ecol Lett 4379ndash391 doi101046j1461-0248200100230x
Guilherme FAG Oliveira-Filho AT Appolinario V Bearzoti E (2004) Effects of flooding regime andwoody bamboos on tree community dynamics in a section of tropical semideciduous forest in south-eastern Brazil Plant Ecol 17419ndash36 doi101023BVEGE000004605197752cd
Huston M (1994) Biological diversity the coexistence of species in changing landscapes CambridgeUniversity Press Cambridge
IPAGRO (1982) Atlas agroclimatico do Rio Grande do Sul Pallotti Porto AlegreJohansson ME Nilsson C Nilsson E (1996) Do rivers function as corridors for plant dispersal J Veg Sci
7593ndash598Junk WJ Bayley PB Sparks RE (1989) The flood pulse concept in river-floodplain systems Can J Fish
Aquat Sci 106110ndash127Legendre P (1993) Spatial autocorrelationmdashtrouble or new paradigm Ecology 741659ndash1673Leinz V (1949) Contribuicao a geologia dos derrames basalticos do Rio Grande do Sul Bol Fac Filos Let
581ndash55Lytle DA Poff NL (2004) Adaptation to natural flow regimes Trends Ecol Evol 1994ndash100 doi
101016jtree200310002McCune B Mefford MJ (1997) PCndashORD Multivariate analysis of ecological data version 436 MjM
Software Design Glaneden BeachMueller-Dombois D Ellenberg H (1974) Aims and methods of vegetation ecology John Wiley New York
2386 Biodivers Conserv (2010) 192371ndash2387
123
Authors personal copy
Neiff JJ (1990) Ideas para la interpretacion ecologica del Parana Interciencia 15424ndash441Neiff JJ Neiff M (2003) Pulso software para analisis de fenomenos recurrentes httpwwwneiffcom
Cited 25 May 2009Oliveira-Filho AT Vilela EA Gavilanes ML Carvalho DA (1994) Effect of flooding regime and understory
bamboos on the physiognomy and tree species composition of a tropical semideciduous forest in south-eastern Brazil Vegetatio 11399ndash124
Oliveira-Filho AT Curi N Vilela EA Carvalho DA (2001) Variation in tree community composition andstructure with changes in soil properties within a fragment of semideciduous forest in south-easternBrazil Edinb J Bot 58139ndash158 doi101017S0960428601000506
Oliveira-Filho AT Jarenkow JA Rodal MJN (2006) Floristic relationships of seasonally dry forests ofeastern South America based on tree species distribution patterns In Pennington RT Ratter JA LewisGP (eds) Neotropical savannas and dry forests plant diversity biogeography and conservation CRCPress Boca Raton
Parolin P (2001) Morphological and physiological adjustments to waterlogging and drought in seedlings ofAmazonian floodplain trees Oecologia 128326ndash335 doi101007s004420100660
Parolin P de Simone O Haase K Waldhoff D Rottenberger S Kuhn U Kesselmeier J Kleiss B SchmidtW Piedade MTF Junk WJ (2004) Central Amazonian floodplain forests tree adaptations in a pulsingsystem Bot Rev 70357ndash380 doi1016630006-8101(2004)070[0357CAFFTA]20CO2
Pijl L (1982) Principles of dispersal in higher plants Springer New YorkPillar VD (2006) Multivariate exploratory analysis randomization testing and bootstrap resampling version
2320 Departamento de Ecologia UFRGS Porto AlegrePollock MM Naiman RJ Hanley TA (1998) Plant species richness in riparian wetlandsmdasha test of biodi-
versity theory Ecology 7994ndash105Robertson KM (2006) Distributions of tree species along point bars of 10 rivers in the south-eastern US
Coastal Plain J Biogeogr 33121ndash132 doi101111j1365-2699200501371xRosales J Petts G Knab-Vispo C (2001) Ecological gradients within the riparian forests of the lower Caura
river Venezuela Plant Ecol 152101ndash118 doi101023A1011411020040Stevens MHH Carson WP (2002) Resource quantity not resource heterogeneity maintains plant diversity
Ecol Lett 5420ndash426 doi101046j1461-0248200200333xStreck EV Kampf N Dalmolin RSD Klamt E Nascimento PC Schneider P (2002) Solos do Rio Grande do
Sul EMATERRS and UFRGS Porto AlegreSwaine MD Whitmore TC (1988) On the definition of ecological species groups in tropical rain forests
Vegetatio 7581ndash86Tabacchi E Correll DL Hauer R Pinay G Planty-Tabacchi AM Wissmar R (1998) Development
maintenance and role of riparian vegetation in the river landscape Freshw Biol 40497ndash516 doi101046j1365-2427199800381x
ter Braack CJF Smilauer P (1998) Canoco reference manual and userrsquos guide to Canoco for Windowssoftware for canonical community ordination (version 40) Microcomputer Power Ithaca
ter Braak CJF (1995) Ordination In Jongman RHG ter Braak CJF van Togeren OFR (eds) Data analysis incommunity and landscape ecology Cambridge University Press New York
Titeux N Dufrene M Jacob JP Paquay M Defourny P (2004) Multivariate analysis of fine-scale breedingbird atlas using a geographical information system and partial canonical correspondence analysisenvironmental and spatial effects J Biogeogr 311841ndash1856 doi101111j1365-2699200401125x
Turner MG Gergel SE Dixon MD Miller JR (2004) Distribution and abundance of trees in floodplainforests of the Wisconsin river environmental influences at different scales J Veg Sci 15729ndash738
Vreugdenhil SJ Kramer K Pelsma T (2006) Effects of flooding duration frequency and depth on thepresence of saplings of six woody species in north-west Europe For Ecol Manage 23647ndash55 doi101016jforeco200608329
Walker KF Sheldon F Puckridge JT (1995) A perspective on dryland river ecosystems Regul River 1185ndash104 doi101002rrr3450110108
Weiher E (2003) Species richness along multiple gradients testing a general multivariate model in oaksavannas Oikos 101311ndash316 doi101034j1600-0706200312216x
Wittmann F Junk WJ Piedade MTF (2004) The varzea forests in Amazonia flooding and the highlydynamic geomorphology interact with natural forest succession For Ecol Manage 196199ndash212 doi101016jforeco200402060
Worbes M Klinge H Revilla JD Martius C (1992) On the dynamics floristic subdivision and geographicaldistribution of Varzea forests in Central Amazonia J Veg Sci 3553ndash564
Zar JH (1996) Biostatistical analysis Prentice-Hall New Jersey
Biodivers Conserv (2010) 192371ndash2387 2387
123
Authors personal copy
2004) Here we employed a routine to partialling out the spatial effects on the analysis of
speciesndashenvironment relationships that also highlight the spatial component embedded in
such analysis (Legendre 1993) Once several biotic processes as growth mortality dis-
persal and predation influence the observed distributions of organisms (resulting in spatial
correlation) or if their distributions are dependent on explanatory variables which are
Fig 5 Ordination biplot depicting the two axes of the environmental partial CCA of species of a riverineforest in southern Brazil Species and environmental variables are represented by their acronyms (seeTable S1)
Fig 6 Rarefaction curves of tree species and Shannon diversity indices from five transects of riverine foreston the Botucaraı river southern Brazil Sampling units are representing different transects
Biodivers Conserv (2010) 192371ndash2387 2383
123
Authors personal copy
themselves spatially structured (Legendre 1993 Titeux et al 2004) spatial structuring is
an intrinsic component of ecosystems In our analysis lsquopurersquo spatial trends were more
attributed to species distribution than to lsquopurersquo environmental effects which link conta-
gious biological processes as important to the tree species distribution Furthermore
species and environmental data have a reasonable proportion of similar spatial structuring
identified by the largest proportion on the species variation due to spatially structured
environmental variation (1246) According to Borcard et al (1992) species and envi-
ronmental variables have in this case the same response to some common underlying
causes as the topographicndashflooding gradient In fact several studies have demonstrated the
direct effect of wetting and desiccation processes on both mineralogy and microbial
ecology of the sediment including nutrient dynamics (Baldwin and Mitchell 2000)
Once sediment or soils are submerged the inundation leads to a decrease in oxygen
contents and then resulting in progressive anaerobic conditions Rapid cycling of litter may
occur due to an increase on microbial activity which generates pulses on nutrient contents
and finally result in highly productive systems (Baldwin and Mitchell 2000) However a
negative effect is the rapid oxygen consumption which quickly leads to soil hypoxia or
anoxia When flood ends the anaerobic zones of sediments are newly oxygenated and
microbiota is replaced gradually to a new phase
Other spatially structured variables are sediment deposition and litter displacement
which are not covered in our study As demonstrated in lsquovarzearsquo forests of Amazonia
(Wittmann et al 2004) sediment deposition decrease toward upper sites and species
colonizing such lower areas show specific adaptations to the new site conditions regarding
to adventitious roots that probably offer mechanical support (Parolin et al 2004 Wittmann
et al 2004) However litter removal or deposition including seed bank may affect
directly species distribution once flooding and allied effects reallocate litter and seeds
among sites (Johansson et al 1996) Moreover studies have showed that flooding timing
frequency and magnitude can be used as indicators of sapling zonation on floodplain
forests (Vreugdenhil et al 2006)
All these processes are included in the unmeasured variables or spatially structuring
processes that have been missed by the geographical terms (Titeux et al 2004) and
accounted to the far unexplained variation (662) As also stressed by these authors the
stochastic spacendashtime fluctuations of each population the lsquounsaturationrsquo pattern (some
species do not use all suitable habitats) and species recording in not appropriated spatial
scales contribute to this unexplained variation too Notwithstanding occurrence data or
species abundances are often noisy (ter Braak 1995) and widespread in ecological studies
(Borcard et al 1992 Titeux et al 2004)
Richness and diversity patterns
Significant transitions occurred from the levee and depression to the following lower slope
transect regarding to stand structure and ecological groups Inversions on the proportion of
pioneershade-tolerant trees and auto-hydrochorous to anemochorous trees occurred in that
small transition and affected not only ecological groups but also species occurrences As a
consequence this zonation transect may consist spatially as a boundary for tolerant and
intolerant trees with regarding to flooding In fact few species occurred over the entire
flooding gradient and the lower slope also appears as an edge for several species
In a temporal scale the lower slope area will probably present more heterogeneous
spans in flooding events and it may consist in the most heterogeneous temporalndashspatial
sector across the topographic gradient which agree with the findings of Pollock et al
2384 Biodivers Conserv (2010) 192371ndash2387
123
Authors personal copy
(1998) from wetlands with different flooding regimes In the structured view of the
dynamic-equilibrium model (Huston 1994) different patches from different seral stages
result from spatial variation of disturbance frequencies If disturbance frequencies vary
over time a landscape could also contain patches of different seral stages (Pollock et al
1998) These authors used that assumption in a model with temporally synchronous dis-
turbance and found that at the community-scale level the results supported many pre-
dictions of the dynamic-equilibrium model especially regarding to species richness
In our study there are two major factors related to disturbance frequency First and
foremost is that once river floods are unpredictable due to a hydrological regime that varies
with occasional long-rainy periods temporal heterogeneity should be higher than in sea-
sonal predicted flood areas and second microtopography must create spatial heterogeneity
during floods events in the local area However well-drained upper sites are probably more
affected by inherent community processes as gap-phase dynamics and direct supply rates
of light (Stevens and Carson 2002) As related by Worbes et al (1992) and Parolin (2001)
hundreds of tree species with different phenological and other ecological traits grow in
seasonal flooded forests In such cases the cyclic alternation on floods and droughts drove
species to life history behavioral and morphological adaptations (Lytle and Poff 2004)
But in the case of unpredictable floods and droughts as assigned in our study bet-hedging
strategies might be evolved for example by persistent seed bank or asynchronous
reproductive phenologies (Brock 2003) although there are no conclusive studies related to
this theory (Lytle and Poff 2004)
Allowed by a transition in the ecological groups from the lower sites to the upper ones
species richness has a maximum at the lower slope transect probably due to higher het-
erogeneity in disturbance events (space and time) and correlated variables As reported by
Desilets and Houle (2005) the spatial gradient provides some evidences for stress toler-
ance and competition as factors structuring species distribution across the topographicndash
flooding gradient also boosted by an unpredictable pattern of floods that vary in frequency
timing and magnitude Lower sectors showed predicted ecological groups already
described for these areas (Budke et al 2007 2008 Junk et al 1989 Lytle and Poff 2004)
and expected structuring changes also occurred toward upper sites where the proportion of
shade-tolerant and small trees increased as well as anemochorous trees
In summary spatialndashtemporal and environmental variables are arranging tree species
distribution across the toposequence of our study site Furthermore predicted ecological
groups reflected the dynamics of disturbance in the topographicndashhydrological gradient
Species richness and diversity also reflected such pattern and were higher in the mid-sector
where occasional floods should prevent competitive exclusion and generate high envi-
ronmental heterogeneity
Acknowledgements We are grateful to the Programa de Pos-Graduacao em Botanica of the UniversidadeFederal do Rio Grande do SulmdashUFRGS for the opportunity to undertake this study and to CAPES Agencyfor the scholarship granted to the first author Our special thanks to Diogo lsquolsquoBagualrsquorsquo Lindenmaier forfieldwork assistance and to Ricardo Braga Eduardo Rossi and colleagues of the Laboratorio de Fitoeco-logiamdashUFRGS for critiques and suggestions We also appreciated the reviewing efforts of anonymouscontributors for providing useful comments to the manuscript
References
Baldwin DS Mitchell AM (2000) The effects of drying and re-flooding on the sediment and soil nutrientdynamics of lowland river-floodplain systems a synthesis Regul River 16457ndash467 doi1010021099-1646
Biodivers Conserv (2010) 192371ndash2387 2385
123
Authors personal copy
Barroso GM Morim MP Peixoto AL Ichaso CLF (1999) Frutos e sementes morfologia aplicada a si-stematica de dicotiledoneas Editora UFV Vicosa
Borcard D Legendre P Drapeau P (1992) Partialling out the spatial component of ecological variationEcology 731045ndash1055 doi1010292006WR005044
Brock MA (2003) Drought and aquatic community resilience the role of eggs and seeds in sediments oftemporary wetlands Freshw Biol 481207ndash1218 doi101046j1365-2427200301083x
Budke JC Athayde EA Giehl ELH Zachia RA Eisinger SM (2005) Composicao florıstica e estrategias dedispersao de especies lenhosas em uma floresta ribeirinha arroio Passo das Tropas Santa Maria RSBrasil Iheringia Bot 6017ndash24
Budke JC Jarenkow JA Oliveira-Filho AT Lindenmaier DS (2006) Padroes de riqueza e diversidade emrios de pequeno porte In Mariath JEA Santos RP (eds) Os avancos da botanica no inıcio do seculoXXI SBB Porto Alegre
Budke JC Jarenkow JA Oliveira-Filho AT (2007) Relationships between tree component structuretopography and soils of a riverine forest Rio Botucaraı southern Brazil Plant Ecol 189187ndash200 doi101007s11258-006-9174-8
Budke JC Jarenkow JA Oliveira-Filho AT (2008) Tree community features of two stands of riverine forestunder different flooding regimes in southern Brazil Flora 203162ndash174 doi101016jflora200703001
Cardoso E Schiavini I (2002) Relacao entre distribuicao de especies arboreas e topografia em um gradienteflorestal na Estacao Ecologica do Panga (Uberlandia MG) Rev Bras Bot 25277ndash289
Connell JH (1978) Diversity in tropical rain forests and coral reefs Science 1991302ndash1310 doi101126science19943351302
Damasceno-Junior GA Semir J Santos FAM Leitao-Filho HF (2005) Structure distribution of species andinundation in a riparian forest of Rio Paraguai Pantanal Brazil Flora 200119ndash135 doi101016jflora200409002
Desilets P Houle G (2005) Effects of resource availability and heterogeneity on the slope of the species-areacurve along a floodplain-upland gradient J Veg Sci 16487ndash496 doi1016581100-9233
Dufrene M Legendre P (1997) Species assemblages and indicator species the need for a flexible asym-metrical approach Ecol Monogr 67345ndash366
EMBRAPA (1997) Manual de metodos de analises de solo Empresa Brasileira de Pesquisa Agropecuariaand Centro Nacional de Pesquisas de Solos Rio de Janeiro
Ferreira LV (2000) Effects of flooding duration on species richness floristic composition and forest structurein river margin habitat in Amazonian blackwater floodplain forests implications for future design ofprotected areas Biodivers Conserv 91ndash14 doi101023A1008989811637
Ferreira LV Stohlgren TJ (1999) Effects of river level fluctuation on plant species richness diversity anddistribution in a floodplain forest in Central Amazonia Oecologia 120582ndash587 doi101007s004420050893
Ferreira C Piedade MTF Franco AC Goncalves JFC Junk WJ (2009) Adaptive strategies to tolerateprolonged flooding in seedlings of floodplain and upland populations of Himatanthus sucuuba aCentral Amazon tree Aquat Bot 90246ndash252 doi101016jaquabot200810006
Gotelli NJ Colwell RK (2001) Quantifying biodiversity procedures and pitfalls in the measurement andcomparison of species richness Ecol Lett 4379ndash391 doi101046j1461-0248200100230x
Guilherme FAG Oliveira-Filho AT Appolinario V Bearzoti E (2004) Effects of flooding regime andwoody bamboos on tree community dynamics in a section of tropical semideciduous forest in south-eastern Brazil Plant Ecol 17419ndash36 doi101023BVEGE000004605197752cd
Huston M (1994) Biological diversity the coexistence of species in changing landscapes CambridgeUniversity Press Cambridge
IPAGRO (1982) Atlas agroclimatico do Rio Grande do Sul Pallotti Porto AlegreJohansson ME Nilsson C Nilsson E (1996) Do rivers function as corridors for plant dispersal J Veg Sci
7593ndash598Junk WJ Bayley PB Sparks RE (1989) The flood pulse concept in river-floodplain systems Can J Fish
Aquat Sci 106110ndash127Legendre P (1993) Spatial autocorrelationmdashtrouble or new paradigm Ecology 741659ndash1673Leinz V (1949) Contribuicao a geologia dos derrames basalticos do Rio Grande do Sul Bol Fac Filos Let
581ndash55Lytle DA Poff NL (2004) Adaptation to natural flow regimes Trends Ecol Evol 1994ndash100 doi
101016jtree200310002McCune B Mefford MJ (1997) PCndashORD Multivariate analysis of ecological data version 436 MjM
Software Design Glaneden BeachMueller-Dombois D Ellenberg H (1974) Aims and methods of vegetation ecology John Wiley New York
2386 Biodivers Conserv (2010) 192371ndash2387
123
Authors personal copy
Neiff JJ (1990) Ideas para la interpretacion ecologica del Parana Interciencia 15424ndash441Neiff JJ Neiff M (2003) Pulso software para analisis de fenomenos recurrentes httpwwwneiffcom
Cited 25 May 2009Oliveira-Filho AT Vilela EA Gavilanes ML Carvalho DA (1994) Effect of flooding regime and understory
bamboos on the physiognomy and tree species composition of a tropical semideciduous forest in south-eastern Brazil Vegetatio 11399ndash124
Oliveira-Filho AT Curi N Vilela EA Carvalho DA (2001) Variation in tree community composition andstructure with changes in soil properties within a fragment of semideciduous forest in south-easternBrazil Edinb J Bot 58139ndash158 doi101017S0960428601000506
Oliveira-Filho AT Jarenkow JA Rodal MJN (2006) Floristic relationships of seasonally dry forests ofeastern South America based on tree species distribution patterns In Pennington RT Ratter JA LewisGP (eds) Neotropical savannas and dry forests plant diversity biogeography and conservation CRCPress Boca Raton
Parolin P (2001) Morphological and physiological adjustments to waterlogging and drought in seedlings ofAmazonian floodplain trees Oecologia 128326ndash335 doi101007s004420100660
Parolin P de Simone O Haase K Waldhoff D Rottenberger S Kuhn U Kesselmeier J Kleiss B SchmidtW Piedade MTF Junk WJ (2004) Central Amazonian floodplain forests tree adaptations in a pulsingsystem Bot Rev 70357ndash380 doi1016630006-8101(2004)070[0357CAFFTA]20CO2
Pijl L (1982) Principles of dispersal in higher plants Springer New YorkPillar VD (2006) Multivariate exploratory analysis randomization testing and bootstrap resampling version
2320 Departamento de Ecologia UFRGS Porto AlegrePollock MM Naiman RJ Hanley TA (1998) Plant species richness in riparian wetlandsmdasha test of biodi-
versity theory Ecology 7994ndash105Robertson KM (2006) Distributions of tree species along point bars of 10 rivers in the south-eastern US
Coastal Plain J Biogeogr 33121ndash132 doi101111j1365-2699200501371xRosales J Petts G Knab-Vispo C (2001) Ecological gradients within the riparian forests of the lower Caura
river Venezuela Plant Ecol 152101ndash118 doi101023A1011411020040Stevens MHH Carson WP (2002) Resource quantity not resource heterogeneity maintains plant diversity
Ecol Lett 5420ndash426 doi101046j1461-0248200200333xStreck EV Kampf N Dalmolin RSD Klamt E Nascimento PC Schneider P (2002) Solos do Rio Grande do
Sul EMATERRS and UFRGS Porto AlegreSwaine MD Whitmore TC (1988) On the definition of ecological species groups in tropical rain forests
Vegetatio 7581ndash86Tabacchi E Correll DL Hauer R Pinay G Planty-Tabacchi AM Wissmar R (1998) Development
maintenance and role of riparian vegetation in the river landscape Freshw Biol 40497ndash516 doi101046j1365-2427199800381x
ter Braack CJF Smilauer P (1998) Canoco reference manual and userrsquos guide to Canoco for Windowssoftware for canonical community ordination (version 40) Microcomputer Power Ithaca
ter Braak CJF (1995) Ordination In Jongman RHG ter Braak CJF van Togeren OFR (eds) Data analysis incommunity and landscape ecology Cambridge University Press New York
Titeux N Dufrene M Jacob JP Paquay M Defourny P (2004) Multivariate analysis of fine-scale breedingbird atlas using a geographical information system and partial canonical correspondence analysisenvironmental and spatial effects J Biogeogr 311841ndash1856 doi101111j1365-2699200401125x
Turner MG Gergel SE Dixon MD Miller JR (2004) Distribution and abundance of trees in floodplainforests of the Wisconsin river environmental influences at different scales J Veg Sci 15729ndash738
Vreugdenhil SJ Kramer K Pelsma T (2006) Effects of flooding duration frequency and depth on thepresence of saplings of six woody species in north-west Europe For Ecol Manage 23647ndash55 doi101016jforeco200608329
Walker KF Sheldon F Puckridge JT (1995) A perspective on dryland river ecosystems Regul River 1185ndash104 doi101002rrr3450110108
Weiher E (2003) Species richness along multiple gradients testing a general multivariate model in oaksavannas Oikos 101311ndash316 doi101034j1600-0706200312216x
Wittmann F Junk WJ Piedade MTF (2004) The varzea forests in Amazonia flooding and the highlydynamic geomorphology interact with natural forest succession For Ecol Manage 196199ndash212 doi101016jforeco200402060
Worbes M Klinge H Revilla JD Martius C (1992) On the dynamics floristic subdivision and geographicaldistribution of Varzea forests in Central Amazonia J Veg Sci 3553ndash564
Zar JH (1996) Biostatistical analysis Prentice-Hall New Jersey
Biodivers Conserv (2010) 192371ndash2387 2387
123
Authors personal copy
themselves spatially structured (Legendre 1993 Titeux et al 2004) spatial structuring is
an intrinsic component of ecosystems In our analysis lsquopurersquo spatial trends were more
attributed to species distribution than to lsquopurersquo environmental effects which link conta-
gious biological processes as important to the tree species distribution Furthermore
species and environmental data have a reasonable proportion of similar spatial structuring
identified by the largest proportion on the species variation due to spatially structured
environmental variation (1246) According to Borcard et al (1992) species and envi-
ronmental variables have in this case the same response to some common underlying
causes as the topographicndashflooding gradient In fact several studies have demonstrated the
direct effect of wetting and desiccation processes on both mineralogy and microbial
ecology of the sediment including nutrient dynamics (Baldwin and Mitchell 2000)
Once sediment or soils are submerged the inundation leads to a decrease in oxygen
contents and then resulting in progressive anaerobic conditions Rapid cycling of litter may
occur due to an increase on microbial activity which generates pulses on nutrient contents
and finally result in highly productive systems (Baldwin and Mitchell 2000) However a
negative effect is the rapid oxygen consumption which quickly leads to soil hypoxia or
anoxia When flood ends the anaerobic zones of sediments are newly oxygenated and
microbiota is replaced gradually to a new phase
Other spatially structured variables are sediment deposition and litter displacement
which are not covered in our study As demonstrated in lsquovarzearsquo forests of Amazonia
(Wittmann et al 2004) sediment deposition decrease toward upper sites and species
colonizing such lower areas show specific adaptations to the new site conditions regarding
to adventitious roots that probably offer mechanical support (Parolin et al 2004 Wittmann
et al 2004) However litter removal or deposition including seed bank may affect
directly species distribution once flooding and allied effects reallocate litter and seeds
among sites (Johansson et al 1996) Moreover studies have showed that flooding timing
frequency and magnitude can be used as indicators of sapling zonation on floodplain
forests (Vreugdenhil et al 2006)
All these processes are included in the unmeasured variables or spatially structuring
processes that have been missed by the geographical terms (Titeux et al 2004) and
accounted to the far unexplained variation (662) As also stressed by these authors the
stochastic spacendashtime fluctuations of each population the lsquounsaturationrsquo pattern (some
species do not use all suitable habitats) and species recording in not appropriated spatial
scales contribute to this unexplained variation too Notwithstanding occurrence data or
species abundances are often noisy (ter Braak 1995) and widespread in ecological studies
(Borcard et al 1992 Titeux et al 2004)
Richness and diversity patterns
Significant transitions occurred from the levee and depression to the following lower slope
transect regarding to stand structure and ecological groups Inversions on the proportion of
pioneershade-tolerant trees and auto-hydrochorous to anemochorous trees occurred in that
small transition and affected not only ecological groups but also species occurrences As a
consequence this zonation transect may consist spatially as a boundary for tolerant and
intolerant trees with regarding to flooding In fact few species occurred over the entire
flooding gradient and the lower slope also appears as an edge for several species
In a temporal scale the lower slope area will probably present more heterogeneous
spans in flooding events and it may consist in the most heterogeneous temporalndashspatial
sector across the topographic gradient which agree with the findings of Pollock et al
2384 Biodivers Conserv (2010) 192371ndash2387
123
Authors personal copy
(1998) from wetlands with different flooding regimes In the structured view of the
dynamic-equilibrium model (Huston 1994) different patches from different seral stages
result from spatial variation of disturbance frequencies If disturbance frequencies vary
over time a landscape could also contain patches of different seral stages (Pollock et al
1998) These authors used that assumption in a model with temporally synchronous dis-
turbance and found that at the community-scale level the results supported many pre-
dictions of the dynamic-equilibrium model especially regarding to species richness
In our study there are two major factors related to disturbance frequency First and
foremost is that once river floods are unpredictable due to a hydrological regime that varies
with occasional long-rainy periods temporal heterogeneity should be higher than in sea-
sonal predicted flood areas and second microtopography must create spatial heterogeneity
during floods events in the local area However well-drained upper sites are probably more
affected by inherent community processes as gap-phase dynamics and direct supply rates
of light (Stevens and Carson 2002) As related by Worbes et al (1992) and Parolin (2001)
hundreds of tree species with different phenological and other ecological traits grow in
seasonal flooded forests In such cases the cyclic alternation on floods and droughts drove
species to life history behavioral and morphological adaptations (Lytle and Poff 2004)
But in the case of unpredictable floods and droughts as assigned in our study bet-hedging
strategies might be evolved for example by persistent seed bank or asynchronous
reproductive phenologies (Brock 2003) although there are no conclusive studies related to
this theory (Lytle and Poff 2004)
Allowed by a transition in the ecological groups from the lower sites to the upper ones
species richness has a maximum at the lower slope transect probably due to higher het-
erogeneity in disturbance events (space and time) and correlated variables As reported by
Desilets and Houle (2005) the spatial gradient provides some evidences for stress toler-
ance and competition as factors structuring species distribution across the topographicndash
flooding gradient also boosted by an unpredictable pattern of floods that vary in frequency
timing and magnitude Lower sectors showed predicted ecological groups already
described for these areas (Budke et al 2007 2008 Junk et al 1989 Lytle and Poff 2004)
and expected structuring changes also occurred toward upper sites where the proportion of
shade-tolerant and small trees increased as well as anemochorous trees
In summary spatialndashtemporal and environmental variables are arranging tree species
distribution across the toposequence of our study site Furthermore predicted ecological
groups reflected the dynamics of disturbance in the topographicndashhydrological gradient
Species richness and diversity also reflected such pattern and were higher in the mid-sector
where occasional floods should prevent competitive exclusion and generate high envi-
ronmental heterogeneity
Acknowledgements We are grateful to the Programa de Pos-Graduacao em Botanica of the UniversidadeFederal do Rio Grande do SulmdashUFRGS for the opportunity to undertake this study and to CAPES Agencyfor the scholarship granted to the first author Our special thanks to Diogo lsquolsquoBagualrsquorsquo Lindenmaier forfieldwork assistance and to Ricardo Braga Eduardo Rossi and colleagues of the Laboratorio de Fitoeco-logiamdashUFRGS for critiques and suggestions We also appreciated the reviewing efforts of anonymouscontributors for providing useful comments to the manuscript
References
Baldwin DS Mitchell AM (2000) The effects of drying and re-flooding on the sediment and soil nutrientdynamics of lowland river-floodplain systems a synthesis Regul River 16457ndash467 doi1010021099-1646
Biodivers Conserv (2010) 192371ndash2387 2385
123
Authors personal copy
Barroso GM Morim MP Peixoto AL Ichaso CLF (1999) Frutos e sementes morfologia aplicada a si-stematica de dicotiledoneas Editora UFV Vicosa
Borcard D Legendre P Drapeau P (1992) Partialling out the spatial component of ecological variationEcology 731045ndash1055 doi1010292006WR005044
Brock MA (2003) Drought and aquatic community resilience the role of eggs and seeds in sediments oftemporary wetlands Freshw Biol 481207ndash1218 doi101046j1365-2427200301083x
Budke JC Athayde EA Giehl ELH Zachia RA Eisinger SM (2005) Composicao florıstica e estrategias dedispersao de especies lenhosas em uma floresta ribeirinha arroio Passo das Tropas Santa Maria RSBrasil Iheringia Bot 6017ndash24
Budke JC Jarenkow JA Oliveira-Filho AT Lindenmaier DS (2006) Padroes de riqueza e diversidade emrios de pequeno porte In Mariath JEA Santos RP (eds) Os avancos da botanica no inıcio do seculoXXI SBB Porto Alegre
Budke JC Jarenkow JA Oliveira-Filho AT (2007) Relationships between tree component structuretopography and soils of a riverine forest Rio Botucaraı southern Brazil Plant Ecol 189187ndash200 doi101007s11258-006-9174-8
Budke JC Jarenkow JA Oliveira-Filho AT (2008) Tree community features of two stands of riverine forestunder different flooding regimes in southern Brazil Flora 203162ndash174 doi101016jflora200703001
Cardoso E Schiavini I (2002) Relacao entre distribuicao de especies arboreas e topografia em um gradienteflorestal na Estacao Ecologica do Panga (Uberlandia MG) Rev Bras Bot 25277ndash289
Connell JH (1978) Diversity in tropical rain forests and coral reefs Science 1991302ndash1310 doi101126science19943351302
Damasceno-Junior GA Semir J Santos FAM Leitao-Filho HF (2005) Structure distribution of species andinundation in a riparian forest of Rio Paraguai Pantanal Brazil Flora 200119ndash135 doi101016jflora200409002
Desilets P Houle G (2005) Effects of resource availability and heterogeneity on the slope of the species-areacurve along a floodplain-upland gradient J Veg Sci 16487ndash496 doi1016581100-9233
Dufrene M Legendre P (1997) Species assemblages and indicator species the need for a flexible asym-metrical approach Ecol Monogr 67345ndash366
EMBRAPA (1997) Manual de metodos de analises de solo Empresa Brasileira de Pesquisa Agropecuariaand Centro Nacional de Pesquisas de Solos Rio de Janeiro
Ferreira LV (2000) Effects of flooding duration on species richness floristic composition and forest structurein river margin habitat in Amazonian blackwater floodplain forests implications for future design ofprotected areas Biodivers Conserv 91ndash14 doi101023A1008989811637
Ferreira LV Stohlgren TJ (1999) Effects of river level fluctuation on plant species richness diversity anddistribution in a floodplain forest in Central Amazonia Oecologia 120582ndash587 doi101007s004420050893
Ferreira C Piedade MTF Franco AC Goncalves JFC Junk WJ (2009) Adaptive strategies to tolerateprolonged flooding in seedlings of floodplain and upland populations of Himatanthus sucuuba aCentral Amazon tree Aquat Bot 90246ndash252 doi101016jaquabot200810006
Gotelli NJ Colwell RK (2001) Quantifying biodiversity procedures and pitfalls in the measurement andcomparison of species richness Ecol Lett 4379ndash391 doi101046j1461-0248200100230x
Guilherme FAG Oliveira-Filho AT Appolinario V Bearzoti E (2004) Effects of flooding regime andwoody bamboos on tree community dynamics in a section of tropical semideciduous forest in south-eastern Brazil Plant Ecol 17419ndash36 doi101023BVEGE000004605197752cd
Huston M (1994) Biological diversity the coexistence of species in changing landscapes CambridgeUniversity Press Cambridge
IPAGRO (1982) Atlas agroclimatico do Rio Grande do Sul Pallotti Porto AlegreJohansson ME Nilsson C Nilsson E (1996) Do rivers function as corridors for plant dispersal J Veg Sci
7593ndash598Junk WJ Bayley PB Sparks RE (1989) The flood pulse concept in river-floodplain systems Can J Fish
Aquat Sci 106110ndash127Legendre P (1993) Spatial autocorrelationmdashtrouble or new paradigm Ecology 741659ndash1673Leinz V (1949) Contribuicao a geologia dos derrames basalticos do Rio Grande do Sul Bol Fac Filos Let
581ndash55Lytle DA Poff NL (2004) Adaptation to natural flow regimes Trends Ecol Evol 1994ndash100 doi
101016jtree200310002McCune B Mefford MJ (1997) PCndashORD Multivariate analysis of ecological data version 436 MjM
Software Design Glaneden BeachMueller-Dombois D Ellenberg H (1974) Aims and methods of vegetation ecology John Wiley New York
2386 Biodivers Conserv (2010) 192371ndash2387
123
Authors personal copy
Neiff JJ (1990) Ideas para la interpretacion ecologica del Parana Interciencia 15424ndash441Neiff JJ Neiff M (2003) Pulso software para analisis de fenomenos recurrentes httpwwwneiffcom
Cited 25 May 2009Oliveira-Filho AT Vilela EA Gavilanes ML Carvalho DA (1994) Effect of flooding regime and understory
bamboos on the physiognomy and tree species composition of a tropical semideciduous forest in south-eastern Brazil Vegetatio 11399ndash124
Oliveira-Filho AT Curi N Vilela EA Carvalho DA (2001) Variation in tree community composition andstructure with changes in soil properties within a fragment of semideciduous forest in south-easternBrazil Edinb J Bot 58139ndash158 doi101017S0960428601000506
Oliveira-Filho AT Jarenkow JA Rodal MJN (2006) Floristic relationships of seasonally dry forests ofeastern South America based on tree species distribution patterns In Pennington RT Ratter JA LewisGP (eds) Neotropical savannas and dry forests plant diversity biogeography and conservation CRCPress Boca Raton
Parolin P (2001) Morphological and physiological adjustments to waterlogging and drought in seedlings ofAmazonian floodplain trees Oecologia 128326ndash335 doi101007s004420100660
Parolin P de Simone O Haase K Waldhoff D Rottenberger S Kuhn U Kesselmeier J Kleiss B SchmidtW Piedade MTF Junk WJ (2004) Central Amazonian floodplain forests tree adaptations in a pulsingsystem Bot Rev 70357ndash380 doi1016630006-8101(2004)070[0357CAFFTA]20CO2
Pijl L (1982) Principles of dispersal in higher plants Springer New YorkPillar VD (2006) Multivariate exploratory analysis randomization testing and bootstrap resampling version
2320 Departamento de Ecologia UFRGS Porto AlegrePollock MM Naiman RJ Hanley TA (1998) Plant species richness in riparian wetlandsmdasha test of biodi-
versity theory Ecology 7994ndash105Robertson KM (2006) Distributions of tree species along point bars of 10 rivers in the south-eastern US
Coastal Plain J Biogeogr 33121ndash132 doi101111j1365-2699200501371xRosales J Petts G Knab-Vispo C (2001) Ecological gradients within the riparian forests of the lower Caura
river Venezuela Plant Ecol 152101ndash118 doi101023A1011411020040Stevens MHH Carson WP (2002) Resource quantity not resource heterogeneity maintains plant diversity
Ecol Lett 5420ndash426 doi101046j1461-0248200200333xStreck EV Kampf N Dalmolin RSD Klamt E Nascimento PC Schneider P (2002) Solos do Rio Grande do
Sul EMATERRS and UFRGS Porto AlegreSwaine MD Whitmore TC (1988) On the definition of ecological species groups in tropical rain forests
Vegetatio 7581ndash86Tabacchi E Correll DL Hauer R Pinay G Planty-Tabacchi AM Wissmar R (1998) Development
maintenance and role of riparian vegetation in the river landscape Freshw Biol 40497ndash516 doi101046j1365-2427199800381x
ter Braack CJF Smilauer P (1998) Canoco reference manual and userrsquos guide to Canoco for Windowssoftware for canonical community ordination (version 40) Microcomputer Power Ithaca
ter Braak CJF (1995) Ordination In Jongman RHG ter Braak CJF van Togeren OFR (eds) Data analysis incommunity and landscape ecology Cambridge University Press New York
Titeux N Dufrene M Jacob JP Paquay M Defourny P (2004) Multivariate analysis of fine-scale breedingbird atlas using a geographical information system and partial canonical correspondence analysisenvironmental and spatial effects J Biogeogr 311841ndash1856 doi101111j1365-2699200401125x
Turner MG Gergel SE Dixon MD Miller JR (2004) Distribution and abundance of trees in floodplainforests of the Wisconsin river environmental influences at different scales J Veg Sci 15729ndash738
Vreugdenhil SJ Kramer K Pelsma T (2006) Effects of flooding duration frequency and depth on thepresence of saplings of six woody species in north-west Europe For Ecol Manage 23647ndash55 doi101016jforeco200608329
Walker KF Sheldon F Puckridge JT (1995) A perspective on dryland river ecosystems Regul River 1185ndash104 doi101002rrr3450110108
Weiher E (2003) Species richness along multiple gradients testing a general multivariate model in oaksavannas Oikos 101311ndash316 doi101034j1600-0706200312216x
Wittmann F Junk WJ Piedade MTF (2004) The varzea forests in Amazonia flooding and the highlydynamic geomorphology interact with natural forest succession For Ecol Manage 196199ndash212 doi101016jforeco200402060
Worbes M Klinge H Revilla JD Martius C (1992) On the dynamics floristic subdivision and geographicaldistribution of Varzea forests in Central Amazonia J Veg Sci 3553ndash564
Zar JH (1996) Biostatistical analysis Prentice-Hall New Jersey
Biodivers Conserv (2010) 192371ndash2387 2387
123
Authors personal copy
(1998) from wetlands with different flooding regimes In the structured view of the
dynamic-equilibrium model (Huston 1994) different patches from different seral stages
result from spatial variation of disturbance frequencies If disturbance frequencies vary
over time a landscape could also contain patches of different seral stages (Pollock et al
1998) These authors used that assumption in a model with temporally synchronous dis-
turbance and found that at the community-scale level the results supported many pre-
dictions of the dynamic-equilibrium model especially regarding to species richness
In our study there are two major factors related to disturbance frequency First and
foremost is that once river floods are unpredictable due to a hydrological regime that varies
with occasional long-rainy periods temporal heterogeneity should be higher than in sea-
sonal predicted flood areas and second microtopography must create spatial heterogeneity
during floods events in the local area However well-drained upper sites are probably more
affected by inherent community processes as gap-phase dynamics and direct supply rates
of light (Stevens and Carson 2002) As related by Worbes et al (1992) and Parolin (2001)
hundreds of tree species with different phenological and other ecological traits grow in
seasonal flooded forests In such cases the cyclic alternation on floods and droughts drove
species to life history behavioral and morphological adaptations (Lytle and Poff 2004)
But in the case of unpredictable floods and droughts as assigned in our study bet-hedging
strategies might be evolved for example by persistent seed bank or asynchronous
reproductive phenologies (Brock 2003) although there are no conclusive studies related to
this theory (Lytle and Poff 2004)
Allowed by a transition in the ecological groups from the lower sites to the upper ones
species richness has a maximum at the lower slope transect probably due to higher het-
erogeneity in disturbance events (space and time) and correlated variables As reported by
Desilets and Houle (2005) the spatial gradient provides some evidences for stress toler-
ance and competition as factors structuring species distribution across the topographicndash
flooding gradient also boosted by an unpredictable pattern of floods that vary in frequency
timing and magnitude Lower sectors showed predicted ecological groups already
described for these areas (Budke et al 2007 2008 Junk et al 1989 Lytle and Poff 2004)
and expected structuring changes also occurred toward upper sites where the proportion of
shade-tolerant and small trees increased as well as anemochorous trees
In summary spatialndashtemporal and environmental variables are arranging tree species
distribution across the toposequence of our study site Furthermore predicted ecological
groups reflected the dynamics of disturbance in the topographicndashhydrological gradient
Species richness and diversity also reflected such pattern and were higher in the mid-sector
where occasional floods should prevent competitive exclusion and generate high envi-
ronmental heterogeneity
Acknowledgements We are grateful to the Programa de Pos-Graduacao em Botanica of the UniversidadeFederal do Rio Grande do SulmdashUFRGS for the opportunity to undertake this study and to CAPES Agencyfor the scholarship granted to the first author Our special thanks to Diogo lsquolsquoBagualrsquorsquo Lindenmaier forfieldwork assistance and to Ricardo Braga Eduardo Rossi and colleagues of the Laboratorio de Fitoeco-logiamdashUFRGS for critiques and suggestions We also appreciated the reviewing efforts of anonymouscontributors for providing useful comments to the manuscript
References
Baldwin DS Mitchell AM (2000) The effects of drying and re-flooding on the sediment and soil nutrientdynamics of lowland river-floodplain systems a synthesis Regul River 16457ndash467 doi1010021099-1646
Biodivers Conserv (2010) 192371ndash2387 2385
123
Authors personal copy
Barroso GM Morim MP Peixoto AL Ichaso CLF (1999) Frutos e sementes morfologia aplicada a si-stematica de dicotiledoneas Editora UFV Vicosa
Borcard D Legendre P Drapeau P (1992) Partialling out the spatial component of ecological variationEcology 731045ndash1055 doi1010292006WR005044
Brock MA (2003) Drought and aquatic community resilience the role of eggs and seeds in sediments oftemporary wetlands Freshw Biol 481207ndash1218 doi101046j1365-2427200301083x
Budke JC Athayde EA Giehl ELH Zachia RA Eisinger SM (2005) Composicao florıstica e estrategias dedispersao de especies lenhosas em uma floresta ribeirinha arroio Passo das Tropas Santa Maria RSBrasil Iheringia Bot 6017ndash24
Budke JC Jarenkow JA Oliveira-Filho AT Lindenmaier DS (2006) Padroes de riqueza e diversidade emrios de pequeno porte In Mariath JEA Santos RP (eds) Os avancos da botanica no inıcio do seculoXXI SBB Porto Alegre
Budke JC Jarenkow JA Oliveira-Filho AT (2007) Relationships between tree component structuretopography and soils of a riverine forest Rio Botucaraı southern Brazil Plant Ecol 189187ndash200 doi101007s11258-006-9174-8
Budke JC Jarenkow JA Oliveira-Filho AT (2008) Tree community features of two stands of riverine forestunder different flooding regimes in southern Brazil Flora 203162ndash174 doi101016jflora200703001
Cardoso E Schiavini I (2002) Relacao entre distribuicao de especies arboreas e topografia em um gradienteflorestal na Estacao Ecologica do Panga (Uberlandia MG) Rev Bras Bot 25277ndash289
Connell JH (1978) Diversity in tropical rain forests and coral reefs Science 1991302ndash1310 doi101126science19943351302
Damasceno-Junior GA Semir J Santos FAM Leitao-Filho HF (2005) Structure distribution of species andinundation in a riparian forest of Rio Paraguai Pantanal Brazil Flora 200119ndash135 doi101016jflora200409002
Desilets P Houle G (2005) Effects of resource availability and heterogeneity on the slope of the species-areacurve along a floodplain-upland gradient J Veg Sci 16487ndash496 doi1016581100-9233
Dufrene M Legendre P (1997) Species assemblages and indicator species the need for a flexible asym-metrical approach Ecol Monogr 67345ndash366
EMBRAPA (1997) Manual de metodos de analises de solo Empresa Brasileira de Pesquisa Agropecuariaand Centro Nacional de Pesquisas de Solos Rio de Janeiro
Ferreira LV (2000) Effects of flooding duration on species richness floristic composition and forest structurein river margin habitat in Amazonian blackwater floodplain forests implications for future design ofprotected areas Biodivers Conserv 91ndash14 doi101023A1008989811637
Ferreira LV Stohlgren TJ (1999) Effects of river level fluctuation on plant species richness diversity anddistribution in a floodplain forest in Central Amazonia Oecologia 120582ndash587 doi101007s004420050893
Ferreira C Piedade MTF Franco AC Goncalves JFC Junk WJ (2009) Adaptive strategies to tolerateprolonged flooding in seedlings of floodplain and upland populations of Himatanthus sucuuba aCentral Amazon tree Aquat Bot 90246ndash252 doi101016jaquabot200810006
Gotelli NJ Colwell RK (2001) Quantifying biodiversity procedures and pitfalls in the measurement andcomparison of species richness Ecol Lett 4379ndash391 doi101046j1461-0248200100230x
Guilherme FAG Oliveira-Filho AT Appolinario V Bearzoti E (2004) Effects of flooding regime andwoody bamboos on tree community dynamics in a section of tropical semideciduous forest in south-eastern Brazil Plant Ecol 17419ndash36 doi101023BVEGE000004605197752cd
Huston M (1994) Biological diversity the coexistence of species in changing landscapes CambridgeUniversity Press Cambridge
IPAGRO (1982) Atlas agroclimatico do Rio Grande do Sul Pallotti Porto AlegreJohansson ME Nilsson C Nilsson E (1996) Do rivers function as corridors for plant dispersal J Veg Sci
7593ndash598Junk WJ Bayley PB Sparks RE (1989) The flood pulse concept in river-floodplain systems Can J Fish
Aquat Sci 106110ndash127Legendre P (1993) Spatial autocorrelationmdashtrouble or new paradigm Ecology 741659ndash1673Leinz V (1949) Contribuicao a geologia dos derrames basalticos do Rio Grande do Sul Bol Fac Filos Let
581ndash55Lytle DA Poff NL (2004) Adaptation to natural flow regimes Trends Ecol Evol 1994ndash100 doi
101016jtree200310002McCune B Mefford MJ (1997) PCndashORD Multivariate analysis of ecological data version 436 MjM
Software Design Glaneden BeachMueller-Dombois D Ellenberg H (1974) Aims and methods of vegetation ecology John Wiley New York
2386 Biodivers Conserv (2010) 192371ndash2387
123
Authors personal copy
Neiff JJ (1990) Ideas para la interpretacion ecologica del Parana Interciencia 15424ndash441Neiff JJ Neiff M (2003) Pulso software para analisis de fenomenos recurrentes httpwwwneiffcom
Cited 25 May 2009Oliveira-Filho AT Vilela EA Gavilanes ML Carvalho DA (1994) Effect of flooding regime and understory
bamboos on the physiognomy and tree species composition of a tropical semideciduous forest in south-eastern Brazil Vegetatio 11399ndash124
Oliveira-Filho AT Curi N Vilela EA Carvalho DA (2001) Variation in tree community composition andstructure with changes in soil properties within a fragment of semideciduous forest in south-easternBrazil Edinb J Bot 58139ndash158 doi101017S0960428601000506
Oliveira-Filho AT Jarenkow JA Rodal MJN (2006) Floristic relationships of seasonally dry forests ofeastern South America based on tree species distribution patterns In Pennington RT Ratter JA LewisGP (eds) Neotropical savannas and dry forests plant diversity biogeography and conservation CRCPress Boca Raton
Parolin P (2001) Morphological and physiological adjustments to waterlogging and drought in seedlings ofAmazonian floodplain trees Oecologia 128326ndash335 doi101007s004420100660
Parolin P de Simone O Haase K Waldhoff D Rottenberger S Kuhn U Kesselmeier J Kleiss B SchmidtW Piedade MTF Junk WJ (2004) Central Amazonian floodplain forests tree adaptations in a pulsingsystem Bot Rev 70357ndash380 doi1016630006-8101(2004)070[0357CAFFTA]20CO2
Pijl L (1982) Principles of dispersal in higher plants Springer New YorkPillar VD (2006) Multivariate exploratory analysis randomization testing and bootstrap resampling version
2320 Departamento de Ecologia UFRGS Porto AlegrePollock MM Naiman RJ Hanley TA (1998) Plant species richness in riparian wetlandsmdasha test of biodi-
versity theory Ecology 7994ndash105Robertson KM (2006) Distributions of tree species along point bars of 10 rivers in the south-eastern US
Coastal Plain J Biogeogr 33121ndash132 doi101111j1365-2699200501371xRosales J Petts G Knab-Vispo C (2001) Ecological gradients within the riparian forests of the lower Caura
river Venezuela Plant Ecol 152101ndash118 doi101023A1011411020040Stevens MHH Carson WP (2002) Resource quantity not resource heterogeneity maintains plant diversity
Ecol Lett 5420ndash426 doi101046j1461-0248200200333xStreck EV Kampf N Dalmolin RSD Klamt E Nascimento PC Schneider P (2002) Solos do Rio Grande do
Sul EMATERRS and UFRGS Porto AlegreSwaine MD Whitmore TC (1988) On the definition of ecological species groups in tropical rain forests
Vegetatio 7581ndash86Tabacchi E Correll DL Hauer R Pinay G Planty-Tabacchi AM Wissmar R (1998) Development
maintenance and role of riparian vegetation in the river landscape Freshw Biol 40497ndash516 doi101046j1365-2427199800381x
ter Braack CJF Smilauer P (1998) Canoco reference manual and userrsquos guide to Canoco for Windowssoftware for canonical community ordination (version 40) Microcomputer Power Ithaca
ter Braak CJF (1995) Ordination In Jongman RHG ter Braak CJF van Togeren OFR (eds) Data analysis incommunity and landscape ecology Cambridge University Press New York
Titeux N Dufrene M Jacob JP Paquay M Defourny P (2004) Multivariate analysis of fine-scale breedingbird atlas using a geographical information system and partial canonical correspondence analysisenvironmental and spatial effects J Biogeogr 311841ndash1856 doi101111j1365-2699200401125x
Turner MG Gergel SE Dixon MD Miller JR (2004) Distribution and abundance of trees in floodplainforests of the Wisconsin river environmental influences at different scales J Veg Sci 15729ndash738
Vreugdenhil SJ Kramer K Pelsma T (2006) Effects of flooding duration frequency and depth on thepresence of saplings of six woody species in north-west Europe For Ecol Manage 23647ndash55 doi101016jforeco200608329
Walker KF Sheldon F Puckridge JT (1995) A perspective on dryland river ecosystems Regul River 1185ndash104 doi101002rrr3450110108
Weiher E (2003) Species richness along multiple gradients testing a general multivariate model in oaksavannas Oikos 101311ndash316 doi101034j1600-0706200312216x
Wittmann F Junk WJ Piedade MTF (2004) The varzea forests in Amazonia flooding and the highlydynamic geomorphology interact with natural forest succession For Ecol Manage 196199ndash212 doi101016jforeco200402060
Worbes M Klinge H Revilla JD Martius C (1992) On the dynamics floristic subdivision and geographicaldistribution of Varzea forests in Central Amazonia J Veg Sci 3553ndash564
Zar JH (1996) Biostatistical analysis Prentice-Hall New Jersey
Biodivers Conserv (2010) 192371ndash2387 2387
123
Authors personal copy
Barroso GM Morim MP Peixoto AL Ichaso CLF (1999) Frutos e sementes morfologia aplicada a si-stematica de dicotiledoneas Editora UFV Vicosa
Borcard D Legendre P Drapeau P (1992) Partialling out the spatial component of ecological variationEcology 731045ndash1055 doi1010292006WR005044
Brock MA (2003) Drought and aquatic community resilience the role of eggs and seeds in sediments oftemporary wetlands Freshw Biol 481207ndash1218 doi101046j1365-2427200301083x
Budke JC Athayde EA Giehl ELH Zachia RA Eisinger SM (2005) Composicao florıstica e estrategias dedispersao de especies lenhosas em uma floresta ribeirinha arroio Passo das Tropas Santa Maria RSBrasil Iheringia Bot 6017ndash24
Budke JC Jarenkow JA Oliveira-Filho AT Lindenmaier DS (2006) Padroes de riqueza e diversidade emrios de pequeno porte In Mariath JEA Santos RP (eds) Os avancos da botanica no inıcio do seculoXXI SBB Porto Alegre
Budke JC Jarenkow JA Oliveira-Filho AT (2007) Relationships between tree component structuretopography and soils of a riverine forest Rio Botucaraı southern Brazil Plant Ecol 189187ndash200 doi101007s11258-006-9174-8
Budke JC Jarenkow JA Oliveira-Filho AT (2008) Tree community features of two stands of riverine forestunder different flooding regimes in southern Brazil Flora 203162ndash174 doi101016jflora200703001
Cardoso E Schiavini I (2002) Relacao entre distribuicao de especies arboreas e topografia em um gradienteflorestal na Estacao Ecologica do Panga (Uberlandia MG) Rev Bras Bot 25277ndash289
Connell JH (1978) Diversity in tropical rain forests and coral reefs Science 1991302ndash1310 doi101126science19943351302
Damasceno-Junior GA Semir J Santos FAM Leitao-Filho HF (2005) Structure distribution of species andinundation in a riparian forest of Rio Paraguai Pantanal Brazil Flora 200119ndash135 doi101016jflora200409002
Desilets P Houle G (2005) Effects of resource availability and heterogeneity on the slope of the species-areacurve along a floodplain-upland gradient J Veg Sci 16487ndash496 doi1016581100-9233
Dufrene M Legendre P (1997) Species assemblages and indicator species the need for a flexible asym-metrical approach Ecol Monogr 67345ndash366
EMBRAPA (1997) Manual de metodos de analises de solo Empresa Brasileira de Pesquisa Agropecuariaand Centro Nacional de Pesquisas de Solos Rio de Janeiro
Ferreira LV (2000) Effects of flooding duration on species richness floristic composition and forest structurein river margin habitat in Amazonian blackwater floodplain forests implications for future design ofprotected areas Biodivers Conserv 91ndash14 doi101023A1008989811637
Ferreira LV Stohlgren TJ (1999) Effects of river level fluctuation on plant species richness diversity anddistribution in a floodplain forest in Central Amazonia Oecologia 120582ndash587 doi101007s004420050893
Ferreira C Piedade MTF Franco AC Goncalves JFC Junk WJ (2009) Adaptive strategies to tolerateprolonged flooding in seedlings of floodplain and upland populations of Himatanthus sucuuba aCentral Amazon tree Aquat Bot 90246ndash252 doi101016jaquabot200810006
Gotelli NJ Colwell RK (2001) Quantifying biodiversity procedures and pitfalls in the measurement andcomparison of species richness Ecol Lett 4379ndash391 doi101046j1461-0248200100230x
Guilherme FAG Oliveira-Filho AT Appolinario V Bearzoti E (2004) Effects of flooding regime andwoody bamboos on tree community dynamics in a section of tropical semideciduous forest in south-eastern Brazil Plant Ecol 17419ndash36 doi101023BVEGE000004605197752cd
Huston M (1994) Biological diversity the coexistence of species in changing landscapes CambridgeUniversity Press Cambridge
IPAGRO (1982) Atlas agroclimatico do Rio Grande do Sul Pallotti Porto AlegreJohansson ME Nilsson C Nilsson E (1996) Do rivers function as corridors for plant dispersal J Veg Sci
7593ndash598Junk WJ Bayley PB Sparks RE (1989) The flood pulse concept in river-floodplain systems Can J Fish
Aquat Sci 106110ndash127Legendre P (1993) Spatial autocorrelationmdashtrouble or new paradigm Ecology 741659ndash1673Leinz V (1949) Contribuicao a geologia dos derrames basalticos do Rio Grande do Sul Bol Fac Filos Let
581ndash55Lytle DA Poff NL (2004) Adaptation to natural flow regimes Trends Ecol Evol 1994ndash100 doi
101016jtree200310002McCune B Mefford MJ (1997) PCndashORD Multivariate analysis of ecological data version 436 MjM
Software Design Glaneden BeachMueller-Dombois D Ellenberg H (1974) Aims and methods of vegetation ecology John Wiley New York
2386 Biodivers Conserv (2010) 192371ndash2387
123
Authors personal copy
Neiff JJ (1990) Ideas para la interpretacion ecologica del Parana Interciencia 15424ndash441Neiff JJ Neiff M (2003) Pulso software para analisis de fenomenos recurrentes httpwwwneiffcom
Cited 25 May 2009Oliveira-Filho AT Vilela EA Gavilanes ML Carvalho DA (1994) Effect of flooding regime and understory
bamboos on the physiognomy and tree species composition of a tropical semideciduous forest in south-eastern Brazil Vegetatio 11399ndash124
Oliveira-Filho AT Curi N Vilela EA Carvalho DA (2001) Variation in tree community composition andstructure with changes in soil properties within a fragment of semideciduous forest in south-easternBrazil Edinb J Bot 58139ndash158 doi101017S0960428601000506
Oliveira-Filho AT Jarenkow JA Rodal MJN (2006) Floristic relationships of seasonally dry forests ofeastern South America based on tree species distribution patterns In Pennington RT Ratter JA LewisGP (eds) Neotropical savannas and dry forests plant diversity biogeography and conservation CRCPress Boca Raton
Parolin P (2001) Morphological and physiological adjustments to waterlogging and drought in seedlings ofAmazonian floodplain trees Oecologia 128326ndash335 doi101007s004420100660
Parolin P de Simone O Haase K Waldhoff D Rottenberger S Kuhn U Kesselmeier J Kleiss B SchmidtW Piedade MTF Junk WJ (2004) Central Amazonian floodplain forests tree adaptations in a pulsingsystem Bot Rev 70357ndash380 doi1016630006-8101(2004)070[0357CAFFTA]20CO2
Pijl L (1982) Principles of dispersal in higher plants Springer New YorkPillar VD (2006) Multivariate exploratory analysis randomization testing and bootstrap resampling version
2320 Departamento de Ecologia UFRGS Porto AlegrePollock MM Naiman RJ Hanley TA (1998) Plant species richness in riparian wetlandsmdasha test of biodi-
versity theory Ecology 7994ndash105Robertson KM (2006) Distributions of tree species along point bars of 10 rivers in the south-eastern US
Coastal Plain J Biogeogr 33121ndash132 doi101111j1365-2699200501371xRosales J Petts G Knab-Vispo C (2001) Ecological gradients within the riparian forests of the lower Caura
river Venezuela Plant Ecol 152101ndash118 doi101023A1011411020040Stevens MHH Carson WP (2002) Resource quantity not resource heterogeneity maintains plant diversity
Ecol Lett 5420ndash426 doi101046j1461-0248200200333xStreck EV Kampf N Dalmolin RSD Klamt E Nascimento PC Schneider P (2002) Solos do Rio Grande do
Sul EMATERRS and UFRGS Porto AlegreSwaine MD Whitmore TC (1988) On the definition of ecological species groups in tropical rain forests
Vegetatio 7581ndash86Tabacchi E Correll DL Hauer R Pinay G Planty-Tabacchi AM Wissmar R (1998) Development
maintenance and role of riparian vegetation in the river landscape Freshw Biol 40497ndash516 doi101046j1365-2427199800381x
ter Braack CJF Smilauer P (1998) Canoco reference manual and userrsquos guide to Canoco for Windowssoftware for canonical community ordination (version 40) Microcomputer Power Ithaca
ter Braak CJF (1995) Ordination In Jongman RHG ter Braak CJF van Togeren OFR (eds) Data analysis incommunity and landscape ecology Cambridge University Press New York
Titeux N Dufrene M Jacob JP Paquay M Defourny P (2004) Multivariate analysis of fine-scale breedingbird atlas using a geographical information system and partial canonical correspondence analysisenvironmental and spatial effects J Biogeogr 311841ndash1856 doi101111j1365-2699200401125x
Turner MG Gergel SE Dixon MD Miller JR (2004) Distribution and abundance of trees in floodplainforests of the Wisconsin river environmental influences at different scales J Veg Sci 15729ndash738
Vreugdenhil SJ Kramer K Pelsma T (2006) Effects of flooding duration frequency and depth on thepresence of saplings of six woody species in north-west Europe For Ecol Manage 23647ndash55 doi101016jforeco200608329
Walker KF Sheldon F Puckridge JT (1995) A perspective on dryland river ecosystems Regul River 1185ndash104 doi101002rrr3450110108
Weiher E (2003) Species richness along multiple gradients testing a general multivariate model in oaksavannas Oikos 101311ndash316 doi101034j1600-0706200312216x
Wittmann F Junk WJ Piedade MTF (2004) The varzea forests in Amazonia flooding and the highlydynamic geomorphology interact with natural forest succession For Ecol Manage 196199ndash212 doi101016jforeco200402060
Worbes M Klinge H Revilla JD Martius C (1992) On the dynamics floristic subdivision and geographicaldistribution of Varzea forests in Central Amazonia J Veg Sci 3553ndash564
Zar JH (1996) Biostatistical analysis Prentice-Hall New Jersey
Biodivers Conserv (2010) 192371ndash2387 2387
123
Authors personal copy
Neiff JJ (1990) Ideas para la interpretacion ecologica del Parana Interciencia 15424ndash441Neiff JJ Neiff M (2003) Pulso software para analisis de fenomenos recurrentes httpwwwneiffcom
Cited 25 May 2009Oliveira-Filho AT Vilela EA Gavilanes ML Carvalho DA (1994) Effect of flooding regime and understory
bamboos on the physiognomy and tree species composition of a tropical semideciduous forest in south-eastern Brazil Vegetatio 11399ndash124
Oliveira-Filho AT Curi N Vilela EA Carvalho DA (2001) Variation in tree community composition andstructure with changes in soil properties within a fragment of semideciduous forest in south-easternBrazil Edinb J Bot 58139ndash158 doi101017S0960428601000506
Oliveira-Filho AT Jarenkow JA Rodal MJN (2006) Floristic relationships of seasonally dry forests ofeastern South America based on tree species distribution patterns In Pennington RT Ratter JA LewisGP (eds) Neotropical savannas and dry forests plant diversity biogeography and conservation CRCPress Boca Raton
Parolin P (2001) Morphological and physiological adjustments to waterlogging and drought in seedlings ofAmazonian floodplain trees Oecologia 128326ndash335 doi101007s004420100660
Parolin P de Simone O Haase K Waldhoff D Rottenberger S Kuhn U Kesselmeier J Kleiss B SchmidtW Piedade MTF Junk WJ (2004) Central Amazonian floodplain forests tree adaptations in a pulsingsystem Bot Rev 70357ndash380 doi1016630006-8101(2004)070[0357CAFFTA]20CO2
Pijl L (1982) Principles of dispersal in higher plants Springer New YorkPillar VD (2006) Multivariate exploratory analysis randomization testing and bootstrap resampling version
2320 Departamento de Ecologia UFRGS Porto AlegrePollock MM Naiman RJ Hanley TA (1998) Plant species richness in riparian wetlandsmdasha test of biodi-
versity theory Ecology 7994ndash105Robertson KM (2006) Distributions of tree species along point bars of 10 rivers in the south-eastern US
Coastal Plain J Biogeogr 33121ndash132 doi101111j1365-2699200501371xRosales J Petts G Knab-Vispo C (2001) Ecological gradients within the riparian forests of the lower Caura
river Venezuela Plant Ecol 152101ndash118 doi101023A1011411020040Stevens MHH Carson WP (2002) Resource quantity not resource heterogeneity maintains plant diversity
Ecol Lett 5420ndash426 doi101046j1461-0248200200333xStreck EV Kampf N Dalmolin RSD Klamt E Nascimento PC Schneider P (2002) Solos do Rio Grande do
Sul EMATERRS and UFRGS Porto AlegreSwaine MD Whitmore TC (1988) On the definition of ecological species groups in tropical rain forests
Vegetatio 7581ndash86Tabacchi E Correll DL Hauer R Pinay G Planty-Tabacchi AM Wissmar R (1998) Development
maintenance and role of riparian vegetation in the river landscape Freshw Biol 40497ndash516 doi101046j1365-2427199800381x
ter Braack CJF Smilauer P (1998) Canoco reference manual and userrsquos guide to Canoco for Windowssoftware for canonical community ordination (version 40) Microcomputer Power Ithaca
ter Braak CJF (1995) Ordination In Jongman RHG ter Braak CJF van Togeren OFR (eds) Data analysis incommunity and landscape ecology Cambridge University Press New York
Titeux N Dufrene M Jacob JP Paquay M Defourny P (2004) Multivariate analysis of fine-scale breedingbird atlas using a geographical information system and partial canonical correspondence analysisenvironmental and spatial effects J Biogeogr 311841ndash1856 doi101111j1365-2699200401125x
Turner MG Gergel SE Dixon MD Miller JR (2004) Distribution and abundance of trees in floodplainforests of the Wisconsin river environmental influences at different scales J Veg Sci 15729ndash738
Vreugdenhil SJ Kramer K Pelsma T (2006) Effects of flooding duration frequency and depth on thepresence of saplings of six woody species in north-west Europe For Ecol Manage 23647ndash55 doi101016jforeco200608329
Walker KF Sheldon F Puckridge JT (1995) A perspective on dryland river ecosystems Regul River 1185ndash104 doi101002rrr3450110108
Weiher E (2003) Species richness along multiple gradients testing a general multivariate model in oaksavannas Oikos 101311ndash316 doi101034j1600-0706200312216x
Wittmann F Junk WJ Piedade MTF (2004) The varzea forests in Amazonia flooding and the highlydynamic geomorphology interact with natural forest succession For Ecol Manage 196199ndash212 doi101016jforeco200402060
Worbes M Klinge H Revilla JD Martius C (1992) On the dynamics floristic subdivision and geographicaldistribution of Varzea forests in Central Amazonia J Veg Sci 3553ndash564
Zar JH (1996) Biostatistical analysis Prentice-Hall New Jersey