Scale-specific determinants of a mixed beech and oak seedling–sapling bank under different environmental and biotic conditions

Scale-specific determinants of a mixed beech and oakseedling–sapling bank under different environmentaland biotic conditions

Antonio Gazol • Ricardo Ibanez

Received: 22 September 2009 / Accepted: 6 April 2010 / Published online: 16 April 2010

� Springer Science+Business Media B.V. 2010

Abstract The persistence of seedlings in the forest

understorey is of major importance for the mainte-

nance and regeneration of canopy trees in several

forested ecosystems. In the present study, we exam-

ine the small-scale spatial pattern of a mixed beech

and oak seedling–sapling bank in two areas of an

unmanaged temperate deciduous forest with different

environmental conditions. We used environmental,

biotic and spatial variables to establish the main

factors that explain the spatial pattern of these

seedling–sapling banks at different scales. The stand

structure in both areas was similar, but while in plot

A beech dominated the canopy, plot B was dominated

by oaks. In both areas, established beech individuals

showed a clear reverse J-shaped distribution, whereas

established oaks showed a unimodal distribution with

only a few young individuals. Seedlings of beech and

oak were distributed in aggregates, whereas beech

saplings had a random distribution. At broader scales,

the abundance of seedlings and saplings is affected by

the environment as well as by inter-species compe-

tition, while at finer scales the spatial pattern is

mainly influenced by stochastic processes, probably

related to seed predation and establishment. The

structure of the seedling–sapling bank indicates an

advantage of beech over oak as far as regeneration is

concerned. Beech seedlings and saplings tolerate the

stress induced by the canopy and the understorey and

persist for many years, while oak seedlings decline in

a few years. Therefore, if current conditions persist,

after canopy opening beech seedlings and saplings

can grow rapidly into the canopy and the stands will

move towards beech dominance.

Keywords Stand structure � Variation partitioning �Environmental dependence � Spatial autocorrelation �Scale

Introduction

In many forest types, canopy tree species initially

establish abundant new individuals in the understo-

rey. Their growth is suppressed because of the strong

limiting resources of the forest understorey but they

can persist for many years waiting for improved

growing conditions. This ‘seedling bank’ (sensu

Grime et al. 1988) is an important ecological factor

in the regeneration of numerous tree species that

compose forest canopies (Lorimer et al. 1994; George

and Bazzaz 2003). After major disturbances such as

gap formation they can rapidly become new canopy

trees, and play a key role in the dynamics of many

forests.

In temperate Europe, beech (Fagus sylvatica) and

pedunculate oak (Quercus robur) are common spe-

cies in forested landscapes. Both species have

A. Gazol (&) � R. Ibanez

Department of Plant Biology, University of Navarra,

Irunlarrea s/n, 31008 Pamplona, Spain

e-mail: [email protected]

123

Plant Ecol (2010) 211:37–48

DOI 10.1007/s11258-010-9770-5

transient seed banks (Grime et al. 1988) and their

regeneration depends on the establishment and sur-

vival of seedlings. Some authors have stressed that

both species could form a seedling bank under a

dense canopy, although their duration in time may

vary (Ziegenhagen and Kaush 1995; Welander and

Ottosson 1998). While beech seedlings can survive in

the understorey for many years, oak seedlings can

only persist for a short period of time. Moreover,

pedunculate oak is considered by many authors to be

a species which requires gaps for its regeneration

(e.g. Rozas 2003), although in pure oak stands

establishment and regeneration of seedlings could

be possible (Dobrowolska 2008).

In general, beech seedlings can survive under low

levels of light and respond rapidly to canopy opening

(Collet and Chenost 2006), although they tolerate

drought less than oak seedlings (van Hees 1997).

Moreover, both species could have problems when

competing for water, light and other resources with the

understorey vegetation (Collet et al. 1998; Lof et al.

1998; Coll et al. 2003; Lof and Welander 2004;

Provendier and Balandier 2008). In acidophilous-

mixed forests, the light conditions and the understorey

composition are different depending on whether the

canopy is dominated by beech or oak (Hardtle et al.

2003), and, therefore, beech and oak seedlings could be

differently affected. In this sense, although seedlings of

the two species could become established, their ability

to survive and grow for several years could vary, taking

into account the different ability to tolerate the biotic

stress (sensu Antos et al. 2005) of the two species.

Despite the vast literature concerning the seedling

ecology of both species, few studies have focused on

the spatial pattern of their seedling banks in natural

conditions (e.g. Rozas 2003; Kunstler et al. 2004),

although studies in other forested ecosystems have

increased our knowledge of seedling requirements

(Garcıa and Houle 2005; Messaoud and Houle 2006;

Maltez-Mouro et al. 2007). The spatial pattern of a

species may be useful as a diagnostic tool in forest

management because it may reflect the ecology of the

species under study (Miller et al. 2002). When seed

rain is not limiting, these patterns depend on the

environmental factors (biotic and abiotic) that influ-

ence their establishment (Nathan and Muller-Landau

2000). Study of the location and the abundance of adult

canopy trees help to identify the potential source of

seeds. The quantification of the seedling bank spatial

patterns serves to generate hypotheses about the

processes that influence these patterns (Perry 1998).

Moreover, the partition of the factors that explain these

patterns helps to identify the scale-specific contribu-

tion of environmental heterogeneity as well as other

biotic relationships (Borcard and Legendre 2002; Dray

et al. 2006; Laliberte et al. 2009).

In the north of Spain, where temperate forests have

their southern distribution limit, there are few studies

about the spatial pattern of seedling banks (e.g. Rozas

2003). While some characteristics of these forests

could be similar to the rest of Europe, their sensitivity

to climate warming and drying, under climate change,

could be more severe due to their vicinity to the

Mediterranean region. Thus, our main objective is to

study the scale-specific environmental and biotic

determinants of a mixed beech and oak seedling–

sapling bank and to discover if they vary depending on

the proportion of beech and oak in the canopy. For this

purpose we: (1) described the stand structure of both

species in two different areas of a mixed forest; (2)

determined the small-scale spatial pattern of the beech

and oak seedling–sapling bank; and (3) studied their

scale-specific relationship with environmental vari-

ables and the surrounding vegetation.

Materials and methods

Study site

The study site is located in the ‘Suspiro’ watershed of

the Bertiz Natural Park (438100N, 18360W), in the

northwest of Navarra, Spain. The Bertiz Natural Park

is an old-growth unmanaged temperate forest that has

been preserved without any human intervention since

its donation to the government of Navarra in 1949.

The Suspiro watershed has 132 ha dominated by

beech (F. sylvatica), with areas of a mixed beech–oak

canopy (mainly with Q. robur). The climatic condi-

tions are characterised by mild winters and moder-

ately warm summers. Temperature ranges from a

monthly low of 7.2�C in January to a monthly high of

21.6�C in July, and the long-term annual precipitation

is 1525 mm (Gobierno de Navarra 2009). The

bedrock belongs to the Palaeozoic era and is mainly

composed of silicic shales and schists.

Based on a previous study (Gazol and Ibanez

2009), two study areas (hereafter areas A and B) were

38 Plant Ecol (2010) 211:37–48

123

selected because they presented different environ-

mental conditions and a mixed beech–oak canopy.

While area A had a higher proportion of beech in the

canopy and higher values of soil moisture, area B

showed a canopy dominated by oaks and higher

values of relative radiation. The two selected study

areas were located on southern slopes, their elevation

values were 405 and 324 m, respectively, and they

were separated by 425 m.

Stand structure

In each study area, an outer square of 1,600 m2

(40 9 40 m) was established. We annotated the

coordinates and recorded the diameter of each adult

tree with a Diameter at Breast Height (hereafter

DBH) greater than 10 cm. Additionally, established

beech and oak saplings and small trees with a height

greater than 50 cm and a diameter between 1 and

10 cm (measured 25 cm above ground level) were

counted and classified in three classes according to

their diameter: small saplings (1–2 cm); large sap-

lings (2–5 cm) and small trees (5–10 cm). Seedlings

were only spatially located in a 400 m2 (20 9 20 m)

area, located in the centre of the whole study area.

Spatial patterns of the seedling–sapling banks

These two 400 m2 (20 9 20 m) areas were divided

into a grid of 100 plots of 2 9 2 m. In each plot, beech

and oak individuals with a diameter of less than 10 cm

were classified and counted in four classes according

to their diameter measured 25 cm above ground level:

seedlings (0–1 cm) and the above-mentioned diame-

ter classes (small saplings, large saplings and small

trees). The distance of each plot to the nearest beech

and oak adult was calculated. Additionally, we

assessed the abundance of each vascular plant species

using a cover-percentage scale, and species were

classified into four groups: graminoids, forbs, ferns,

shrubs and sub-canopy trees.

Several environmental variables were measured.

Cover of leaf-litter, rocks and logs was visually

assessed and depth of leaf-litter was measured twice

in each plot. Slope was measured with a clinometer

(Silva Clino Master, Silva Sweden) and aspect was

obtained with a compass and then transformed into an

exposure index on a scale of -1.0 (northeast-facing

slopes) to 1.0 (southwest-facing slopes) (Beers et al.

1966). The slope microtopography of each plot was

visually assessed using a categorical scale: concave,

flat and convex. Elevation was obtained using a laser

pointer and a 2-m long centimetre rule, referring all

the measurements to the lowest point in each area.

Top soil water content was measured five times in

each plot with a soil moisture sensor SM200 (Delta-T

device, Cambridge UK) on one dry day. During a

sunny day, a fifteen second integration of the incident

radiation was taken in each plot with a pyranometric

sensor LiCor (Li-Cor, USA). Simultaneously, the

total radiation was measured outside the forest to

calculate the relative radiation.

We annotated the UTM coordinates of each plot.

The Principal Coordinates of Neighbour Matrices

method (hereafter PCNM; Borcard and Legendre

2002) was used to transform the UTM coordinates

into spatial variables. The PCNM variables are

eigenvectors that represent the spatial structure of

the studied site at several scales. For a complete

description of the method see Dray et al. (2006). 69

PCNM variables were created with the program

SpaceMaker (Borcard and Legendre 2004) and then

classified into scales according to the pattern they

represented: variables 1 to 10 were classified as

meso-scale and the rest were coded as fine-scale

variables. The terms meso- and fine-scales do not

have absolute meaning, although they are relative to

the study grain (plot) and extent (area of study).

The spatial patterns of the different classes of

beech and oak seedlings and saplings in the plots

were analysed using Spatial Analysis by Distance

Indices (hereafter SADIE; Perry 1998; Perry et al.

1999). We focused on the index of aggregation, Ia,

and the index of clustering, v, produced by SADIE.

The Ia index quantifies the overall spatial pattern of

the variable under consideration. Values near to one

indicate a random distribution, whereas values higher

or lower indicate an aggregated or a regular distri-

bution, respectively. The v index, which indicates the

degree of clustering in the data, can be used to

represent the spatial pattern of the variable in a map.

In each study area, we performed a SADIE analysis

for beech and oak seedlings, small saplings, large

saplings and small tree class with a minimum number

of 10 individuals, as well as for each group of plant

species (cover values were previously transformed to

a decile scale). Another application of SADIE

measures the spatial association of counts (Perry

Plant Ecol (2010) 211:37–48 39

123

and Dixon 2002). Using the index of association, we

studied the relation between the spatial patterns of the

seedlings, small saplings, large saplings and small

trees. The geostatistical extension of ArcGis 9.1 (Esri

Redlands, USA) was used to create isoline maps of

the index of clustering.

Redundancy analyses with variation partitioning

(Borcard et al. 1992; Okland 2003) were used to

discover the fraction of variation in seedling–sapling

abundance explained by different explanatory matri-

ces. A separate analysis was performed in each study

area. The response matrix was composed by the

abundance of the above-mentioned beech and oak

seedlings, saplings and small trees and values were

previously log transformed (x0 = log (x ? 1)). Three

explanatory matrices were used to partition the

variation in abundance of beech and oak individuals:

(1) an environmental matrix created with the mea-

sured environmental factors; (2) a biotic matrix

composed by the cover values of the different groups

of vascular plants (graminoids, forbs, ferns, shrubs

and sub-canopy trees) as well as the distance to the

nearest beech and oak adult; and (3) a space matrix

comprising the X and Y coordinates and 69 PCNM

variables in each study area. Prior to the analyses, a

separate forward selection with 999 unrestricted

permutations was carried out to select the variables

that significantly explained the variations in seedling,

sapling and small tree abundance. The analyses were

performed using the program CANOCO 4.5 (ter

Braak and Smilauer 2002), and the CanoDraw

extension was used to create the ordination diagrams.

Finally, we tested the effect of the environmental

and biotic variables on four spatial models: the broad-

scale linear trend (x- and y-UTM coordinates),

PCNM scale (PCNM variables selected), and the

meso- (PCNM variables 1–10) and fine-scale models

(variables 11–69). The fitted values of beech and oak

abundance of each spatial model were used as

response matrices against the environmental and

biotic variables. Using this approach, we separated

the relative contribution of the environmental and

biotic factors on different scales. For a complete

description of the method see Laliberte et al. (2009).

These analyses were performed with the ‘vegan’

library (Oksanen et al. 2007) of the R statistical

language (R Development Core Team 2007).

Results

Stand structure

The overstorey of both study areas was co-dominated

by beech and oak adults in different proportions

(Table 1). In area A, the mean number of beech and

Table 1 Stand structure of beech and oak individuals in the A and B 40 9 40 m areas studied in the Bertiz forest

Area A Area B

N N hectare-1 DBH range (cm) N N hectare-1 DBH range (cm)

Fagus sylvatica

Adults (DBH [ 10 cm) 24 150 12.4–72.6 12 75 11.1–61.8

Seedlings (0–1cm)a 1,044 6,525 716 4,475

Small saplings (1–2 cm) 66 413 90 563

Large saplings (2–5 cm) 36 225 48 300

Small trees (5–10 cm) 7 44 15 94

Quercus robur

Adults (DBH [ 10 cm) 16 100 24.8–78.8 23 144 16.9–70.7

Seedlings (0–1 cm)a 3,088 19,300 3,052 19,075

Small saplings (1–2 cm) 0 0 0 0

Large saplings (2–5 cm) 0 0 2 13

Small trees (5–10 cm) 0 0 2 13

The total number of individuals (N), density (N hectare-1) and DBH range in centimetres (adult individuals only) are showna The total number of seedlings (N) and their density (N hectare-1) has been estimated using the number of seedlings counted in the

400 m2 area

40 Plant Ecol (2010) 211:37–48

123

oak adults (DBH [ 10 cm) was 150 and 100 per

hectare, respectively, and the total basal area was

36.3 m2 ha-1 (15.2 beech and 21.1 oak). In area B,

there were 75 beech and 144 oak adults per hectare and

the total basal area was 29.4 m2 ha-1 (8.6 beech and

20.8 oak). No oak individuals with a diameter between

1 and 10 cm were found in A, whereas only 4 were

found in B. In contrast, the mean number of beech

individuals with a diameter between 1 and 10 cm was

681 and 956 per hectare in areas A and B, respectively.

The number of beech individuals decreased from small

saplings to small trees (Table 1).

The distribution of individuals with a diameter

higher than 1 cm classified into diameter classes

(Fig. 1) showed similar results in both areas. Beech

presented a distribution where more than 80% of the

individuals had a diameter between 1 and 10 cm. In

contrast, all oaks in area A and 80% of them in area B

had a diameter greater than 20 cm, showing a more or

less unimodal distribution.

Spatial pattern of the seedling–sapling banks

The study of the 2 9 2 m plots showed that oak

seedlings with a diameter between 0 and 1 cm were the

most abundant, and that beech seedlings were more

abundant in area A than in B (Table 2). Seedlings

of beech and oak had a significantly aggregated

Fig. 1 Relative frequency (%) of established beech and oak individuals classified in diameter classes of 5 cm found in the A and B

40 9 40 m areas studied. Only individuals with a diameter greater than 1 cm were considered

Table 2 Spatial pattern of the beech and oak seedlings (0–1 cm), small saplings (1–2 cm), large saplings (2–5 cm) and small trees

(5–10 cm) studied in the 100 2 9 2 m plots of areas A and B

Area A Area B

N Ia Vj Vi N Ia Vj Vi

F. sylvatica

Seedlings 261 2.210* -2.171 1.742 179 2.467* -2.433 1.964

Small saplings 33 1.158 -1.166 1.207 11 1.699* -1.723 1.613

Large saplings 21 1.286 -1.287 1.255 12 1.393 -1.397 1.353

Small trees 2 - - - - - - -

Q. robur

Seedlings 772 2.201* -1.991 1.818 763 2.743* -2.436 2.682

Small saplings 0 – – – – – – –

Large saplings 0 – – – – – – –

Small trees 0 – – – – – – –

The total number of individuals (N), aggregation index (Ia), and negative and positive means of the index of clustering (Vj and Vi

respectively) are shown

* Significance of the index of aggregation (P \ 0.01) tested against 5,967 permutations

Plant Ecol (2010) 211:37–48 41

123

distribution, whereas beech saplings with higher

diameters tended to be randomly distributed, exclud-

ing small saplings (1–2 cm) in area B (Table 2). In

area A, oak seedlings formed significant patches on

southern zones whereas beech seedlings were aggre-

gated on the southeastern part of the plot (Fig. 2).

Large beech saplings (2–5 cm) were randomly

distributed but they were more abundant in elevated

zones of the study area. In area B, seedlings of both

species formed significant patches in the southern part

of the area (Fig. 2). Large beech saplings were

randomly distributed and small saplings were aggre-

gated in the southwestern part of the study area. Small

beech and oak trees (5–10 cm) were not studied due to

their low abundance (Table 2).

In the two areas, seedlings of both species tend to

be positively associated together and negatively

related to large beech saplings. In area A, small

beech saplings were positively associated with large

saplings, whereas their relationship in area B was

negative (Table 3).

Explanation of the spatial pattern

of the seedling–sapling banks

The main environmental differences between the two

areas were found in aspect, soil moisture and

radiation (Table 4). The species composition of the

Fig. 2 Spatial patterns of beech and oak seedlings in areas A

and B. The index of clustering (v) of each plot was interpolated

using kriging. Values of the index of clustering higher than 1.5

(dark colour) indicate zones where seedlings form significant

aggregates while values lower than -1.5 (light colour) indicate

zones where gaps of seedlings are significant. The beech

(black) and oak (white) adults present in the 40 9 40 m areas

are represented with a circle proportional to their DBH

Table 3 Index of association between the beech and oak

seedlings (0–1 cm), small saplings (1–2 cm) and large saplings

(2–5 cm) studied in the 100 2 9 2 m plots of areas A and B

Area A Area B

Fs1 Fs2 Fs3 Fs1 Fs2 Fs3

Qr1 0.119 -0.070 -0.259* 0.595* 0.263* -0.375*

Fs1 0.014 -0.158 0.468* -0.459*

Fs2 0.351* -0.387*

Qr1, Fs1 (oak and beech seedlings respectively), Fs2 (small

beech saplings), and Fs3 (large beech saplings)

* Significance of the positive (P \ 0.025) or negative

(P [ 0.975) association tested with the Dutilleul adjusted

probability level

42 Plant Ecol (2010) 211:37–48

123

understorey was similar in both study areas. The

graminoids Agrostis capillaris and Deschampsia

flexuosa, the forbs Hypericum pulchrum, Ranunculus

tuberosus and Veronica chamaedrys, the fern Pter-

idium aquilinum and the shrubs Calluna vulgaris,

Erica vagans and Hypericum androsaemum were the

most common species. The abundance and spatial

patterns of each group of species considered can be

seen in Table 5.

The ordination diagrams of the environmental and

biotic models showed that the first axes were mainly

created by the segregation between seedlings (0–1

cm) and large saplings (2–5 cm) (Fig. 3), although

the results were different in each study area. In area

A, three environmental (elevation, aspect and slope),

two biotic (forb and shrub cover) and 15 spatial

variables (x- and y-UTM coordinates and 13 PCNMs)

were selected, which explained together 60.6% of

variation in beech and oak seedling–sapling abun-

dance (Fig. 3). After removing their joint effects,

only the spatial fraction (31.7%) was significant. The

joint effect among the three sets of variables

accounted for 15.3% of the explained fraction.

11.4% was due to the joint effect of the environment

and space.

The division of the spatial-structured fraction

(58.8% of the variation) on different scales showed

that 26.5% of the spatial fraction was explained by

the UTM coordinates (linear trend at a broad scale)

and 38.5% by the PCNM variables (Fig. 3). 27.2% of

the PCNM fraction was due to the meso-scale

variables and 10.8% to the fine-scale variables. In

the linear trend fraction (x- and y-coordinates), 73%

of the variation was explained by the environmental

variables, 2% by the biotic factors and 19% by the

interaction between both sets of variables. Therefore,

Table 4 Quantitative environmental variables measured in the 100 2 9 2 m plots of areas A and B

Area A Area B t-statistic

Mean ± SD Range Mean ± SD Range

Elevation (m)a 6.4 ± 3.0 0.0–12.2 6.3 ± 3.2 0.0–11.5 0.215

Aspectb 0.11 ± 0.51 -0.75–0.77 0.51 ± 0.13 0.13–0.78 -7.604*

Slope (degrees) 26 ± 6 10–45 24 ± 6 3–49 2.892*

Leaf-litter cover (%) 59 ± 20 15–95 49 ± 23 5–95 3.397*

Leaf-litter deep (cm) 3.4 ± 1.6 0.5–12.0 2.7 ± 1.1 0–8 3.352*

Rock cover (%) 3 ± 5 0–25 4 ± 6 0–30 -1.088

Log cover (%) 11 ± 8 0–40 10 ± 9 0–40 0.716

Radiationb 15.3 ± 5.5 8.2–40.1 26.9 ± 16.3 10.8–100 -6.742*

Soil moisture (mV) 219 ± 30 164–319 135 ± 19 101–193 23.590*

The means of the variables in areas A and B are compared using a permutation test of the t-statistic (Legendre 2005)a Based on the lowest point in each areab Relative scale

* Significance of the t-statistic (P \ 0.01) against 999 permutations

Table 5 Percentage cover and aggregation index of the groups of species studied in the 100 2 9 2 m plots of areas A and B

Area A Area B

Mean ± SD Range Ia Mean ± SD Range Ia

Graminoids 14 ± 15 0–74 2.53* 14 ± 12 0–66 2.13*

Forbs 8 ± 5 0–24 2.27* 7 ± 6 0–40 2.07*

Ferns 7 ± 8 0–50 2.14* 3 ± 9 0–50 2.57*

Shrubs 20 ± 19 0–81 2.46* 16 ± 14 0–70 2.13*

Sub-canopy trees 4 ± 8 0–50 1.37 1 ± 7 0–70 1.08

* Significance of the index of aggregation (P \ 0.01) tested against 5,967 permutations

Plant Ecol (2010) 211:37–48 43

123

6% of this fraction was purely spatial. In the PCNM

scale, the environmental variables were able to

explain 1% of the variation, 9% was explained by

the biotic factors and both sets of variables showed a

negative overlap. 93% of this fraction was due to the

pure effect of the spatial variables. The meso-scale

model was mainly due to the spatial variables (87%),

but 14% of it was explained by the biotic variables

and 2% by the environment. The fine-scale model

was completely created by the spatial variables.

In area B, the environmental (elevation, soil

moisture, flat microtopography, rock and log cover),

biotic (distance to nearest adult beech, graminoid,

forb, shrub and fern cover) and spatial (x- and y-UTM

coordinates and 15 PCNMs) variables selected

explained 72.3% of the variation in beech and oak

seedling–sapling abundance (Fig. 3). Only the pure

effect of the spatial variables (14.1%) was significant

after removing the joint effects among the three

groups. 32.3% of the explained variation was due to

the joint effect between the three sets of variables,

11.5% to the joint effect between environment and

space, and 10.7% to the overlap between space and

biotic variables.

The division of the shared spatial fraction into

scales showed that 45.7% was due to a linear trend

and 28.2% was explained by the PCNM variables

(Fig. 3). 12.5% of the PCNM fraction was explained

by the meso-scale variables and 15.2% of it by the

fine-scale variables. The environmental and biotic

Fig. 3 Ordination and venn diagrams of the variation parti-

tioning of seedling abundance of areas A (left) and B (right). At

the top of the Figure, for each study area, the ordination biplots

of the shared environmental (left) and biotic (right) fractions

are shown. Seedlings and saplings are coded as follows; Qr1oak seedlings (0–1 cm), Fs1 beech seedlings (0–1 cm), Fs2small beech saplings (1–2 cm), and Fs3 large beech saplings

(2–5 cm). In the centre of the Figure the venn diagrams

represent the general variation partitioning results showing the

environmental, biotic and spatial fractions, as well as their

shared fractions. At the bottom of the Figure, the circular

diagrams indicate the division of the shared spatial fraction of

the variation partitioning into different scales: the linear trend

and the PCNM scale, as well as the division of the PCNM scale

into meso- and fine-scale

44 Plant Ecol (2010) 211:37–48

123

variables explained 29 and 5% of the linear trend,

respectively, and 55% was due to their joint effect.

Only 11% of this model was due to the spatial

variables alone. In the PCNM model, 10% of the

variation was explained by the environmental vari-

ables, 11% by the biotic factors and 2% by their joint

effect. The dissection of this model showed that the

environmental and biotic variables explained 11 and

4% of the meso-scale fraction, respectively, and their

overlap was 5%. In the fine-scale model, the

environmental variables explained 2% and the biotic

factors 9%, with negative overlap. 80% of the meso-

scale model and 90% of the fine-scale model was of

purely spatial nature.

Discussion

Stand structure

The distribution of beech individuals per diameter

class in both study areas shows a reverse J-shaped

diameter distribution which is in accordance with the

idealised well-balanced population structure for old-

growth forests as observed in other forests in the

same region (Rozas 2006). However, oak individuals

showed a unimodal distribution with only 4 out of 43

individuals with a diameter between 1 and 10 cm. As

has been indicated by other authors, the forest

understorey could be a habitat with high biotically

induced stress (Antos et al. 2005): while beech

seedlings and saplings suppress their growth and

survive for many years waiting for improved growing

conditions, oak seedlings and saplings cannot tolerate

these conditions. In another mixed forest in northern

Spain, oak establishment was strongly associated

with canopy openings (Rozas 2003). Additionally, in

a flood plain oak forest of Poland (Dobrowolska

2008), oak regeneration was possible with a stand

density of *150 individuals per hectare and a mean

basal area of *26 m2 ha-1. In our study, the two

large saplings and the two small oak trees were found

in area B, where the mean basal area was

*29.4 m2 ha-1 and the presence of oak adults was

twice the number of beeches. Probably the survival of

these young oaks was favoured by the higher values

of light enabled by the presence of a canopy

dominated by oaks (Hardtle et al. 2003). The

presence of oaks in the canopy could be explained

by historical factors: oaks would initially become

established under more open conditions, when the

area was still managed. The difference in the number

of young beech and oak trees indicates that, as has

been observed in other forests (Rozas 2003), natural

regeneration in the Bertiz forest could be sufficient to

maintain beech populations, but could lead to a

decline in oak populations. After canopy opening,

beech saplings can grow rapidly into the canopy and

the stands will move towards beech dominance.

Spatial patterns of the beech

and oak seedling–sapling banks

Oak seedlings seem to benefit from seed reserves

during the germination year (Welander and Ottosson

1998) and so they could have an advantage over

beech seedlings, giving as a result a higher number

of oak seedlings. However, the seed production of

both species is very variable over the years (Grime

et al. 1988), and this could, therefore, influence the

number of seedlings. Although we have no infor-

mation about seed rain in recent years, the large

difference in the number of seedlings between the

two species led us to think that oak seedlings may

have an advantage in germination. However, the lack

of oak saplings, with higher diameters, reinforces the

hypothesis that they can only tolerate the conditions

of the understorey for a few years, probably due to

the relatively low light availability in the understo-

rey (Antos et al. 2005).

The spatial pattern of beech and oak seedlings

departed significantly from random in the two studied

areas, as has been found for species recruitment in

other forests (Rozas 2003; Kunstler et al. 2004;

Maltez-Mouro et al. 2007). These clumped distribu-

tions could be caused by the environmental and biotic

factors that influence seedling establishment because

the presence of adult beech and oak individuals

ensures seed rain (Nathan and Muller-Landau 2000).

The positive association between seedlings of both

species probably indicates that they have similar

establishment requirements. Although the two species

perform differently during their development, beech

and oak seedlings could benefit from the same

microsites during their establishment. Therefore, at

earlier stages of development beech and oak seed-

lings seem to compete for the same resources

regardless of the dominant tree species in the canopy.

Plant Ecol (2010) 211:37–48 45

123

The random distribution of large beech saplings

may be due to the environmental history, intra-

species competition and scale limitations of the study.

Beech seedlings and saplings tend to compete for

light in natural conditions (Collet and Chenost 2006)

and, although they are able to survive under low light

levels for many years, canopy opening is necessary to

enable their growth into the canopy (Szwagrzyk et al.

2001). Periodic small gap creation by natural distur-

bances facilitates the growth of beech individuals

(Rozas 2003). In the two studied plots, the negative

association between large beech saplings and beech

and oak seedlings is maintained (Table 3). These

results, therefore, indicate that large beech saplings

could shade and prevent the establishment of beech

and oak seedlings (Welander and Ottosson 1998).

Otherwise, the random distribution of large beech

saplings can be a result of the high biotically induced

stress of the understorey. Therefore, many seedlings

and small saplings will die and only a few individuals

will persist for many years (Antos et al. 2005).

The variation partitioning analyses showed the

importance of the spatial component in the distribu-

tion of seedlings and saplings, as occurs in other

forested ecosystems (Maltez-Mouro et al. 2007). The

amount of variance explained by the environmental

and biotic variables was higher in area B than in A. In

this sense, area B had lower values of soil moisture

and higher levels of radiation than A. The higher

values of light could enable the presence of a more

diverse understorey (Hardtle et al. 2003) but the

lower values of soil moisture could cause a relative

water deficit. Moreover, the number of beech saplings

of large diameter (1–5 cm) was lower in this area,

indicating the influence of a possible water deficit

(van Hees 1997). Under the more severe environ-

mental conditions of area B, seedling and sapling

distribution could be influenced more by the under-

storey vegetation (Lof and Welander 2004), as is

indicated by the heightened joint effect between the

environmental and the biotic fractions. Thus, the

competition with the understorey vegetation could be

the reason for the lower number of small and large

beech saplings in area B than in A. These results

could indicate the importance that biotically induced

stress has on the growth and development of beech

saplings.

The environmental and the biotic variables were

spatially structured and induced a scale-specific

spatial dependence in the distribution of the seed-

ling–sapling bank (Borcard et al. 1992; Legendre

1993; Laliberte et al. 2009). At broader scales (linear

trend), the greater importance of the environmental

variables than the biotic factors indicates that niche

processes are the main factors determining seedling

and sapling distribution (Laliberte et al. 2009). The

large differences in elevation and the steep slopes

were the main factors determining the broad-scale

beech and oak distribution. Lower zones in both areas

had higher values of forb cover and lower of shrub

and fern. Therefore, the aggregation of seedlings in

these areas could be related to the competition for

light (Lorimer et al. 1994; Collet et al. 1998; Coll

et al. 2003; Provendier and Balandier 2008). Shrubs,

ferns and large seedlings create a tall understorey that

prevents the passing of light. The positive relation

between seedlings and forb cover indicated their

preference for the same microsite, as was found in

other forests (Miller et al. 2002).

The main variations in the distribution of the

seedling–sapling bank at finer scales were explained

by the spatial variables. While we do not measure

important soil resources that could influence seedling

distribution (Lof and Welander 2004), such as

mineral nutrients, the fine-scale domain of this spatial

fraction could indicate the importance of stochastic

processes like dispersal and predation (Legendre

1993; Dray et al. 2006; Laliberte et al. 2009). The

presence of adult individuals of both species in the

canopy could minimise the influence of dispersal

processes (Nathan and Muller-Landau 2000) and thus

other processes such as variation in seed predation

and establishment (Garcıa and Houle 2005; Mess-

aoud and Houle 2006) or processes related to fine-

scale resource availability (Beatty 1984) could be the

reason for the unexplained spatial component.

Concluding remarks and implications

The combination of the SADIE analysis with the

variation partitioning method could be of high utility

for the interpretation of species patterns. Moreover,

the use of the PCNM variables allows the subsequent

division of the spatial component into scales facili-

tating the interpretation of the vegetation–environ-

ment relationships.

Regardless of the dominant tree species in the

canopy, seedlings of beech and oak form aggregates,

46 Plant Ecol (2010) 211:37–48

123

whereas small and large beech saplings tend to be

randomly distributed. The broad-scale seedling and

sapling abundance is mainly determined by the

environmental conditions and the competition with

the understorey vegetation. This determination was

higher under the influence of a canopy dominated by

oaks, where the lower water availability could

increase the competition with the understorey vege-

tation. The higher ability of beech seedlings to

tolerate the stress in the understorey could be the

reason for the presence of saplings and small trees of

beech. Conversely, oak seedlings can only tolerate

these conditions for a few years. At finer scales, the

seedling–sapling bank was mainly determined by

spatially structured processes probably related to seed

predation and establishment. The maintenance of the

natural conditions in the Bertiz forest could lead to a

decline in oak populations and a colonisation of the

canopy by beech individuals after canopy opening,

owing to their higher ability to tolerate the biotically

induced stress of the understorey. Therefore, we

could expect that the stands will move towards beech

dominance. The development of further studies with

a dynamic perspective could help to clarify the future

of these areas with a mixed beech–oak canopy.

Acknowledgements The authors thank Etienne Laliberte and

Daniel Borcard for technical advice in the variation

partitioning analysis performed. Helpful suggestions from

two anonymous reviewers further improved the manuscript.

We are also grateful to the Parque Natural ‘‘Senorıo de Bertiz’’

for giving permission to conduct this research within the

protected area. The research was partially supported by

Fundacion Caja Navarra (programa ‘‘Tu eliges, Tu decides’’)

as well as by Fundacion Universitaria de Navarra and by a

predoctoral grant from the Asociacion de Amigos de la

Universidad de Navarra to Antonio Gazol. We also thank

Jaime Urıa for his help in the field work.

References

Antos JA, Guest HJ, Parish R (2005) The tree seedling bank in

an ancient montane forest: stress tolerators in a productive

habitat. J Ecol 93:536–543

Beatty SW (1984) Influence of microtopography and canopy

species on spatial patterns of forest understory plants.

Ecology 65:1406–1419

Beers TW, Dress PE, Wensel LC (1966) Aspect transformation

in site productivity research. J For 64:691–692

Borcard D, Legendre P (2002) All-scale spatial analysis of

ecological data by means of principal coordinates of

neighbour matrices. Ecol Model 153:51–68

Borcard D, Legendre P (2004) SpaceMaker2. Version for Win-

dows. Departement des Sciences Biologiques, Universite de

Montreal. Available online http://www.bio.umontreal.ca/

casgrain/en/labo/spacemaker.html. Accessed 05 Sep 2009

Borcard D, Legendre P, Drapeau P (1992) Partialling out the

spatial component of ecological variation. Ecology 73:

1045–1055

Coll L, Balandier P, Picon-Cochard C, Prevosto B, Curt T

(2003) Competition for water between beech seedlings

and surrounding vegetation in different light and vegeta-

tion composition conditions. Ann For Sci 60:593–600

Collet C, Chenost C (2006) Using competition and light esti-

mates to predict diameter and height growth of naturally

regenerated beech seedlings growing under changing

canopy conditions. Forestry 79:489–502

Collet C, Ningre F, Frochot H (1998) Modifying the micro-

climate around young oaks through vegetation manipu-

lation: effects on seedling growth and branching. For Ecol

Manag 110:249–262

Dobrowolska D (2008) Effect of stand density on oak regen-

eration in flood plain forests in Lower Silesia, Poland.

Forestry 81:511–523

Dray S, Legendre P, Peres-Neto PR (2006) Spatial modelling: a

comprehensive framework for principal coordinate analysis

of neighbour matrices (PCNM). Ecol Model 196:483–493

Garcıa D, Houle G (2005) Fine-scale spatial patterns of

recruitment in red oak (Quercus rubra): what matters

most, abiotic or biotic factors? Ecoscience 12:223–235

Gazol A, Ibanez R (2009) Different response to environmental

factors and spatial variables of two attributes (cover and

diversity) of the understorey layers. For Ecol Manag

258:1267–1274

George LO, Bazzaz FA (2003) The herbaceous layer as a filter

determining spatial pattern in forest tree regeneration. In:

Gilliam FS, Roberts MR (eds) The herbaceous layer in

forest of eastern North America. Oxford University Press,

New York, pp 265–282

Gobierno de Navarra (2009) Meteorologıa y climatologıa de

Navarra. Available via DIALOG. http://meteo.navarra.es

of subordiante document. Accessed 05 Sep 2009

Grime JP, Hodgson JG, Hunt R (1988) Comparative plant

ecology: a functional approach to common British spe-

cies. Unwin Hyman, London

Hardtle W, von Oheimb G, Westphal C (2003) The effects of

light and soil conditions on the species richness of the

ground vegetation of deciduous forests in northern Ger-

many (Schleswig-Holstein). For Ecol Manag 182:327–338

Kunstler G, Curt T, Lepart J (2004) Spatial pattern of beech

(Fagus sylvatica L.) and oak (Quercus pubescens Mill.)

seedlings in natural pine (Pinus sylvestris L.) woodlands.

Eur J For Res 123:331–337

Laliberte E, Paquette A, Legendre P, Bouchard A (2009)

Assessing the scale-specific importance of niches and

other spatial processes on beta diversity: a case study from

a temperate forest. Oecologia 159:377–388

Legendre P (1993) Spatial autocorrelation: trouble or new

paradigm? Ecology 74:1659–1673

Legendre P (2005) t.perm.R: test t pour donnees independantes

avec test par permutations. Available online http://

www.bio.umontreal.ca/legendre/index.html#FonctionsR.Accessed 25 Mar 2010

Plant Ecol (2010) 211:37–48 47

123

Lof M, Welander T (2004) Influence of herbaceous competi-

tors on early growth in direct seeded Fagus sylvatica L.

and Quercus robur L. Ann For Sci 61:781–788

Lof M, Gemmel P, Nilsson U, Welander NT (1998) The

influence of site preparation on growth in Quercus roburL. seedlings in southern Sweden clear-cut and shelter-

wood. For Ecol Manag 109:241–249

Lorimer CG, Chapman JW, Lambert WD (1994) Tall under-

storey vegetation as a factor in the poor development of

oak seedling beneath mature stands. J Ecol 82:227–237

Maltez-Mouro S, Garcıa LV, Maranon T, Freitas H (2007)

Recruitment patterns in a Mediterranean oak forest: a case

study showing the importance of the spatial component.

For Sci 53:645–652

Messaoud Y, Houle G (2006) Spatial patterns of tree seedling

establishment and their relationship to environmental

variables in a cold-temperate deciduous forest of eastern

North America. Plant Ecol 185:319–331

Miller TF, Mladenoff DJ, Clayton MK (2002) Old-growth

northern hardwood forest: spatial autocorrelation and pat-

terns of understory vegetation. Ecol Monogr 72:487–503

Nathan R, Muller-Landau HC (2000) Spatial patterns of seed

dispersal, their determinants and consequences for

recruitment. Trends Ecol Evol 15:278–285

Okland RH (2003) Partitioning the variation in a plot-by-spe-

cies data matrix that is related to n sets of explanatory

variables. J Veg Sci 14:693–700

Oksanen J, Kindt R, Legendre P, O’Hara RB, Simpson GL,

Solymos P, Stevens MHH, Wagner H (2007) ‘‘vegan’’:

Community Ecology Package Version 1.15-1. Available

via DIALOG. http://www.R-project.org of subordinate

document. Accessed 05 Sep 2009

Perry JN (1998) Measures of spatial pattern for counts. Ecol-

ogy 79:1008–1017

Perry JN, Dixon P (2002) A new method to measure spatial

association for ecological count data. Ecoscience 9:133–141

Perry JN, Winder L, Holland JM, Alston RD (1999) Red-blue

plots for detecting clusters in count data. Ecol Lett 2:106–113

Provendier D, Balandier P (2008) Compared effects of com-

petition by grasses (Graminoids) and broom (Cytisusscoparius) on growth and functional traits of beech sap-

lings (Fagus sylvatica). Ann For Sci 65:510–518

R Development Core Team (2007) R: a language and envi-

ronment for statistical computing. R Foundation for Sta-

tistical Computing, Vienna, Austria. Available online

http://www.R-project.org Accessed 05 Sep 2009

Rozas V (2003) Regeneration patterns, dendroecology, and

forest-use history in an old-growth beech-oak lowland

forest in Northern Spain. For Ecol Manag 182:175–194

Rozas V (2006) Structural heterogeneity and tree spatial pat-

terns in an old-growth deciduous lowland forest in Can-

tabria, northern Spain. Plant Ecol 185:57–72

Szwagrzyk J, Szewczyk J, Bodziarczyk J (2001) Dynamics of

seedling banks in beech forest: results of a 10-year study

on germination, growth and survival. For Ecol Manag

141:237–250

ter Braak CJF, Smilauer P (2002) CANOCO reference manual

and CanoDraw for Windows user’s guide: software for

canonical community ordination (version 4.5). Micro-

computer Power, New York

van Hees AFM (1997) Growth and morphology of pedunculate

oak (Quercus robur L) and beech (Fagus sylvatica L)

seedlings in relation to shading and drought. Ann For Sci

54:9–18

Welander NT, Ottosson B (1998) The influence of shading on

growth and morphology in seedlings of Quercus robur L.

and Fagus sylvatica L. For Ecol Manag 107:117–126

Ziegenhagen B, Kaush W (1995) Productivity of young oak

shaded oaks (Quercus robur L.) as corresponding to shoot

morphology and leaf anatomy. For Ecol Manag 72:97–108

48 Plant Ecol (2010) 211:37–48

123