Spatio-ecological niche segregation of two sympatric species ofClidemia (Melastomataceae) in Western Amazonian non-flooded rainforests

Folia Geobotanica 39: 143-160, 2004

SPATIO-ECOLOGICAL NICHE SEGREGATION OF TWO SYMPATRIC SPECIES OF CLIDEMIA (MELASTOMATACEAE) IN WESTERN AMAZONIAN NON-FLOODED RAINFORESTS

Leif Schu lman ~'2), Hanne le Ko ivunen ~) & Kalle Ruoko la inen 1)

1) Department of Biology, Section for Biodiversity and Environmental Studies, FI-20014 University of Turku, Finland 2) Address for correspondence: Botanic Garden, P.O. Box 44, FI-00014 University of Hetsinki, Finland, fax +358 9 191 50033; e-mail [email protected]

Abstract: The frequent occurrence of sympatric series of closely related plant species in tropical rainforests has evoked claims for and against the application of the competitive exclusion principle in these ecosystems. Narrow niche limits defined by biotic as well as abiotic specialization have been reported for sympatric species of the same genus or family. In Amazonian lowland rainforests this question deserves renewed attention because: (1) the existence of edaphically defined community types has recently been well established, and (2) spatio-ecological niche segregation of congeneric species may help explain not only the maintenance of the high Amazonian alpha-diversity, but also its origin through sympatric ecological speciation. In this study, the morphology, ecology, and distribution patterns of two species, Clidemia epiphytica and C. longifolia (Melastomataceae), from western Amazonia, were analyzed. The aims were to find out whether they really are two distinct taxonomic species and if so, whether they also can be considered biological species; if the species are sympatric; and if they are ecologically specialized. The results showed that the morphological variation of the species seems continuous, but that they exhibit opposite morphological responses to variation in soil cation concentration, which suggests that they also are separate biological species. Furthermore, the species occur sympatrically but in different habitats. It is suggested that a part of the enigma of sympatric congeners in rainforests may be explainable by spatial segregation stemming from ecological specialization in relation to subtle environmental variation. It is hypothesized that the studied species are a good candidate case of sympatric speciation driven by ecological specialization.

Keywords: Cations, Ecological specialization, Habitat partitioning, Morphological analysis, Neotropics, Sympatric speciation, Tierra firme

INTRODUCTION

The high alpha-diversity o f tropical rainforests and, in particular, the frequent occurrence o f sympatric series o f closely related plant species has puzzled botanists for more than half a century. Traditionally the discussion has centred on whether the concept o f the niche and Gause 's principle o f competitive exclusion apply in communities where such series coexist (FEDOROV 1966, ASHTON 1969, PdCHARDS 1969). Several theoretical solutions for the problem of coexistence o f species have been presented (see review by TILMAN & PACALA I993; also TILMAN 1994, HUBBELL 2001). Many observational studies have at least narrowed the problem by documenting fine-scale niche delimitation in rainforest plants. Niches o f sympatric species o f the same genus or family have been found to be delimited by biotic factors, such as differences in pollination and phenology (SNOW 1965, GENTRY 1976,

144 L. Schulman et al.

1990, HILTY 1980, BORSCHENIUS 2002), or by abiotic variables, such as specialization on different soils or topographic positions (GENTRY 1986, ROGSTAD 1990, SVENNING 2001 ). In the case of Amazonian lowland non-inundated (tierra firme) forests, two recent scientific developments have drawn renewed attention to these findings.

First, it has been shown that Amazonian tierra firme forests consist of numerous different forest types, as defined by plant community composition, and that a significant part of this variation can be attributed to edaphic heterogeneity (WERFF 1992, TUOMISTO et al. 1995, 2003, RUOKOLAINEN et al. 1997, RUOKOLAINEN & TUOMISTO 1998, TER STEEGE 1998, VORMISTO et al. 2000). This means that a considerable part of plant species in these forests have different niches at least at a relatively coarse resolution. It remains unclear, however, to what extent variation in community structure translates into spatial niche segregation among congeneric, generally sympatric species. In other words, the seeming co-occurrence of sympatric congeners in Amazonian tierra firme forests may be an artefact caused by observation at too coarse spatio-ecological resolution.

Second, the mainstream explanation for the origin of the Amazonian species richness, based on allopatric speciation in isolated forest refuges during periods of drier climate (HAFFER 1969, 1997, PRANCE 1973, HAFFER & PRANCE 2001), has met with strong opposition (BEVEN et al. 1984, SALO 1987, NELSON et al. 1990, BUSH 1994, 1996, COLINVAUX et al. 2000, 2001). Earlier suggestions that non-allopatric speciation might also play a role (e.g., BROWN 1982, ENDLER 1982, GENTRY 1982, 1986, 1989) did not gain much support, apparently due to the long-lasting singular orthodoxy of the allopatric speciation model (see, e.g., MAYR 1963, BUSH 1998). This latter view has, however, changed drastically in recent years. Both theoretical and empirical studies now show that sympatric speciation (with its parapatric variant) is a feasible alternative to the allopatric mode (see reviews in SCHLUTER 2001, TURELLI et al. 2001, and VIA 2001; also JOHANNESSON 2001, LEV1N 2001, BERLOCHER & FEDER 2002, GAVRILETS & WAXMAN 2002, NOSIL et al. 2002, OGDEN & THORPE 2002, ALMEIDA & VISTULO DE ABREU 2003, MIZERA & MESZI~NA 2003, MARTIN & HOSKEN 2003, SORENSON et al. 2003). In their review of more than 200 studies dealing with local genetic differentiation of plant populations, LINHART & GRANT (1996) concluded that natural selection operating on sympatric populations not only dominates the genetic structuring of plant populations, but also plays a central role in plant speciation.

If examples were found in Amazonia, where sympatric sibling species differ in their ecology, it would point to the possibility that part of the Amazonian species have originated through ecological specialization in sympatry. This would open a completely new view on the origin and maintenance of Amazonian alpha-diversity. One reason why such cases are not yet known is that the taxonomy of practically any Amazonian plant group is still woefully incomplete. The taxonomic understanding is mostly based on scanty herbarium material alone. Often no information exists on the relationship between environmental and phenotypic variation. This is problematic, because ignorance of this relationship may lead to errors in species circumscriptions and, hence, to misconceptions as regards distribution patterns and the autecology of species, as elucidated by ROGSTAD (1990) for some Southeast Asian Annonaceae. The problem is two-fold. First, if we study taxonomy only on the basis of herbarium specimens we tend to lump a set of specimens collected in the same geographic

Ecological segregation of Clidemia species 145

la ~ \

10 cm

lb

Fig. 1. Target taxa of this study. (a) Clidemia epiphytica (TRIANA) COGN. (drawn from KOIVUNEN 182, Peru: Loreto, INIA). (b) C. longifolia GLEASON (KoIVtrNEN 110, Peru: Loreto, Puerto Almendras). Drawn to same scale (scale bar = 10 cm) by H. Koivunen. See Table 1 for coordinates of collecting localities. Specimens are located at AMAZ with duplicates at TUR.

area into a single taxon despite some phenotypic variation, because there is no ecological information that could suggest narrower taxon circumscriptions. Second, if we study ecology and distribution patterns on the basis of this kind of prevailing taxonomy we tend not to detect spatio-ecological separation of taxa even where such patterns might exist. Because of this, it is essential that ecological, biogeographical, and taxonomic analyses of Amazonian plants proceed hand in hand.

We conducted such a combined study on two species, Clidemia epiphytica and C. longifolia (Melastomataceae), in western Amazonia. From our own field experience we have come to suspect that these two taxonomic species may actually represent extreme ends of a morphological continuum. To test our suspicion, we analyzed some aspects of their morphology and ecology, and their local to regional distribution patterns. Our aim was to answer the following questions: (1) Are there two morphologically distinct forms that can be regarded as good taxonomic as well as biological species? (2) If yes, are the species sympatric? And, (3) do the species show differing ecological specialization? Here we present the

answers we found and discuss them in an evolutionary context with the rationale that positive answers to the three questions would provide circumstantial evidence in support of sympatric speciation.

MATERIALS AND METHODS

Study species, taxonomy

Clidemia epiphytica (TRIANA) COGN. (Fig. la) and C. longifolia GLEASON (Fig. lb) are rainforest lianas that adhere to tree trunks with the help of adventitious roots. To our knowledge, the only comparative taxonomic treatment of these species is provided by WURDACK (1964). He places four species in what he calls "the node around C. epiphytica": C. discolor (TRIANA) COGN., C. serpens (TRIANA) COGN., C. longifolia, and C. epiphytica.


In the key he presents, the first forking separates the two former species on the basis of lack of setae on the hypanthium. The two species with setose hypanthia, C. longifolia and C. epiphytica, are separated from each other by the quantitative character "lamina length-to-width ratio": in the former the ratio is said to be 3.1-4.1, in the latter 1.5-2. WURDACK (1964) also states that leaf shape is the only consistent feature separating C. longifolia from C. epiphytica, but that "it is a constant and striking one". In the original description of C. longifolia, GLEASON (1931) also recognized stouter petioles as an attribute separating the new species from C. epiphytica. WURDACK (1964) considers a fifth published name, C. panamensis (BLAKE et STANDL.) GLEASON, a synonym ofC. epiphytica, and a sixth name, C. radicans PILG., a synonym of a seventh, C. trichocalyx (BLAKE) GLEASON. The latter he places under C. epiphytica as a variety, C. epiphytica var. trichocalyx (BLAKE) WURDACK, separating it from the type variety because it has persistently setulose tertiary veins and veinlets on the abaxial side of the lamina. The typical variety has only sparse setulae restricted to the primary and secondary veins. Clidemia discolor was later found to be a shrub, rather than a liana, and to differ from the three lianoid species also by a number of reproductive characters (WURDACK 1988).

Based on our own field experience, we agree with WURDACK (1964, 1988) that C. serpens is distinct from C. epiphytica. In addition to being esetose, C. serpens has strikingly red petioles, whereby it is readily distinguished in the field. However, the distinctiveness of C. longifolia is much less clear. The relative width of the lamina seems to vary rather continuously in specimens that otherwise fit published descriptions of C. epiphytica and C. longifolia (we have not noticed differences in the reproductive structures, which is in accordance with previous descriptions).

Study species, phylogenetic relationships According to WURDACK (1964) the climbing species of Clidemia sect. Sagraea (DC.)

COGN. (i.e., C. epiphytica, C. longifolia, and C. serpens) have 4-merous flowers, whereas in the climbing species of sect. Staphidium (NAUD.) COGN. (C. epibaterium DC., C. repens TRIANA, and C. sandwithii WURDACK) the flowers are 5-merous. Unfortunately, phylogenetic analyses clarifying the relationships between species of Clidemia D. DON have not been carried out. However, based on a cladistic analysis at the tribal level, JUDD (1989) recognized genus Sagraea DC. (with 4-merous flowers) as separate from the genus Clidemia (with 5-merous flowers). Hence, it can be inferred from the cladistic study of JUDD (1989) that C. epiphytica, C. longifolia, and C. serpens are phylogenetically distinct from other climbing species traditionally placed in the same genus. Because C. epiphytica and C. longifolia share the character of possessing setae on the hypanthia, and lack the red petiole of C. serpens, it seems likely that C. epiphytica and C. longifolia would come out as a sibling species pair in a cladistic analysis.

Study species, distribution Clidemia epiphytica ranges from Guatemala and Honduras in the north to Bolivia in the

south, and from central Peru in the west to Venezuela in the east (WURDACK 1964, 1988, BRAKO & ZARUCCHI 1993). The distributions of the two varieties differ only in that the


southernmost records ofvar, trichocalyx are from Peruvian Amazonia, rather than Bolivia. In addition, in French Guyana there is an apparently disjunct population of C. epiphytica, which has not been placed in either variety (WURDACK 1993). Clidemia longifolia occurs in northern Peru, where both species are Amazonian to low Andean in their elevational distribution (BRAKO & ZARUCCHI 1993), and in southern Colombia. In summary, the range of C. longifolia is nested within that of C. epiphytica.

Study area

The study material was collected in primary lowland non-flooded rainforests in Amazonian Colombia, Ecuador, and Peru (Fig. 2). The climate of the study area is aseasonally hot and wet. The Ecuadorean part is apparently slightly drier (annual precipitation approx. 2600 mm; ROMOLEROUX et al. 1997) than the Peruvian and Colombian areas (annual precipitation approx. 3000 nun; DUIVENVOORDEN & LIPS 1993, MARENGO 1998). The geological setting is dominated by large rivers and various fluvial processes that have modified the environment since the Tertiary and left behind a mosaic of present and fossil floodplains (so-called river terraces), and areas of Miocene semimarine deposits (SALO et al. 1986, RASANEN et al. 1987, 1992, 1995).

Field work and soil sample analyses

The plant specimens and soil samples used in this study were obtained within two collecting schemes. In both schemes sampling site selection was guided by satellite image (Landsat TM) interpretation, and sites were selected within landscape sections, which looked relatively homogeneous in the images. In the first scheme, a total of 90 specimens were gathered at seven sites (each collecting area approx. 2 km across) within approx. 50 km from the Peruvian city of Iquitos (Table 1; Fig. 2). These specimens were used for two purposes: their morphology was analyzed and compared with soil cation concentration estimates obtained from soil samples collected at the same sites (see below), and they provided insight to the distribution patterns of the taxa at a local scale.

The dataset was augmented with data obtained during a larger programme documenting the composition of the Melastomataceous flora in western Amazonian rainforests (see TUOMISTO et al. 1995, 2003, RUOKOLAINEN et al. 1997, RUOKOLAINEN & TUOMISTO 1998). In this second scheme, the data come from inventory plots of fixed size (25 x 25 m, 2 x 500 m, 5 × 500 m, or 5 x 1300 m; Table 1). In most of the plots all Melastomataceae individuals have been counted, in some only the presence of species has been recorded. In all plots each encountered morphotype has been documented by means of a voucher collection. Soil samples have been taken from almost all of these plots. In this scheme, a total of 65 specimens of the target taxa of the present study were collected from 43 sites (Table 1; Fig. 2). These specimens were analyzed together with the 90 collections of the first scheme, and they were also used to link the morphological variation of the studied taxa to geographical distribution at a regional scale. To detect possible correlation between plant morphology and size (a proxy of age), the height of each individual from which a specimen was collected was determined in meters.


~ 4¢j 37

ECUADOR

21" I " - "s { 31 ' ' ~ 4 3

PERU

~L COLOMBIA~'~

33

21

t4,

BRAZIL

Fig. 2. Study area with locations of study sites. Insert shows location of study area in South America. Coordinates of study sites are given in Table 1 (numbers denoting sites are the same). Symbols denoting sites mean: cross - only Clidemia epiphytica encountered, solid circle - only C. longifolia encountered, solid square - both species encountered.

In the first collecting scheme, one soil sample (top 5 cm of the mineral soil after removing organic material) was collected per plot. In the second scheme, two or more samples were collected per plot (save in one plot; Table 1). In order to catch the full local variation in the soil, at least one sample was taken from a hillock top and one from the bottom of a depression. In general, we collected more soil samples when the plot was bigger, and when depressions and hillocks were unevenly represented, there were always more samples from hillocks.

The soil samples were analyzed for average cation concentration. Here we report only the sum of exchangable cations (Ca, K, Mg, Na; Table 1), which has been shown to be one of the most important soil characteristics explaining distribution pattems of Amazonian Melastomataceae (TUOMISTO et al. 1995, 2003, RUOKOLA1NEN et al. 1997, RUOKOLAINEN & TUOMISTO 1998, VORMISTO et al. 2000). For a description of the laboratory analyses of the soil samples see RUOKOLAINEN & TUOMISTO (1998). In addition, we tried to designate the environment of each collecting site as either an old river terrace, or not a terrace, on the basis of field observations, Landsat TM satellite images, and geomorphological maps (Table 1).


Table 1. Study sites (numbers are the same as in Fig. 2), their locations, plot sizes (in 1,000 m2; n.d. - area not defined), number of soil samples analyzed per site (n; n.a. - soil samples not available), mean sum of cations in the soil samples from the corresponding site (s.cat.; g iven as cmol+/kg), number of Clidemia epiphytica and C. longifolia individuals observed and collected at each site, population density (no. o f observed individuals per 1,000 m2; n.d. - population density not definable because either number of individuals not recorded or plot size not determined), and geological history of site (a r iver terrace or not; see text). Number of Clidemia individuals marked as + means that only presence was recorded in the corresponding site; hence, totals of observed individuals are min imum numbers. Sites marked with * before the name are those referred to as "the first collecting scheme" in the text. Sites marked with c before the name are in Colombia, those marked with E in Ecuador; the remaining sites are in Peru. Estr. - Estrecho; Lib. Agr. - Libertad Agraria; Aim. - Almendras; Febr. Febrero.

Study site locality plot soil samp. C. epiphytica C. longifolia riv. lat. (S) long. (W) size n s.cat, obs. coll. dens. obs. coll. dens. ter.

1 Mishana a 3o53'23 " 73°29 '19" 0.625 1 0.06 11 2 16.9 Yes 2 M o m 6 n 2 3°39 '45 " 73°17'12 " 1.25 2 0.08 1 1 0.8 No 3 Jenaro Herrera 2 4o52 '14" 73038'59" 2.5 3 0.12 5 1 2.0 Yes 4 Nauta 4026'42" 73035'07" 6.5 16 0.14 7 1 1.1 No 5 Huanta 3030'00 " 72°00'00" 6.5 11 0.14 36 2 5.5 Yes 6 Panguana 1 3052'32 " 73°04'52" 2.5 7 0.15 13 2 5.2 Yes 7 c Estr. krn 15.55 l°47 '13" 71°02'48 " 2 2 0.16 + 1 n.d. Yes 8 c Pam~ikm 6.2 l °10 ' 20 " 71003'44 ' 2 2 0.17 + 1 n.d. No 9 SanPedro 4°21 '40" 73o00'59" 2.5 3 0.17 4 1 1.6 Yes 10 Maniti 3°36 '16 " 72o05'22" 6.5 8 0.20 128 2 19.7 Yes 11 Requena 1 5o37'28 " 73004'50" 2.5 3 0.24 4 1 1.6 No 12 NuevaEsperanza 1 3°20 '18 " 72°02'10 " 2.5 3 0.25 58 1 23.2 Yes 13 Lib. Agr. krn 3 4°09 '11" 73o00'48 " 2.5 3 0.26 + 1 n.d. Yes 14 San Gerardo b 3o54'43 " 73o02 ' 14" n.d. 2 0.29 + 1 n.d. Yes 15 Puerto Aim. 1 3050'40" 73°02'14" 2.5 3 0.29 I 1 0.4 Yes 16 Puerto Izango 5 3°18 '10" 72004'57 " 6.5 7 0.30 120 2 18.5 Yes 17 Mishanac 3°53'23" 73002'56 " 5.5 3 0.31 77 3 14.0 71 1 12.9 Yes 18 *San Gerardo a 3054'43 " 73002 ' 14" 2.5 1 0.46 6 2 2.4 Yes 19 Mishanab 3°53'23" 73°02'56 " 0.625 1 0.50 7 1 10.8 7 1 10.8 Yes 20 Supai km 3.9 3°17'49" 72004'58 " 2 2 0.52 + 1 n.d. Yes 21 S u p a i k m 6.2 3°17 '18" 72°00'54 " 2 2 0.62 + 1 n.d. Yes 22 *PuertoAlm. 2 3049'47" 73002 '15" n.d. 1 0.62 30 22 n.d. Yes 23 Sucusari 1 3°15'17" 72005'28 " 2.5 4 0.69 30 1 12.0 No 24 Gengen 3°36'41" 73o01'47 " 7.2 10 0.72 24 2 3.3 No 25 * I N I A b 3o56'53" 73002'27 " n.d. 1 0.77 5 5 n.d. No 26 *SantaAna 4055 '50 " 73o00'42 " 6.5 1 0.83 87 21 n.d. ? 27 *Palo Seco 2 3059'26" 73°02'30" 2.5 1 0.97 8 2 3.2 No 28 SanAntonio 1 4°32 '10 " 73°03'50" 6.5 16 1.33 18 10 2.8 No 29 13 de Febr. clay 4°15 '00 " 73o02'43 " 1.25 4 1.41 5 2 4.0 No 30 Carbaja lplot 2 4°17 '36 " 73o03'33" 2 2 1.67 + 1 n.d. No 31 ExPetroleros 4o39'50 " 73°02'46" 6.5 17 2.33 29 2 4.5 No 32 * I N I A a 3056'53 " 73°02'27" n.d. 1 2.34 153 27 n.d. No 33 Puer to lzango 1 3°15 '28 " 72000'49" 6.5 9 2.34 117 1 18.0 No 34 E Yastmi 15 0039'34 " 76002'8" 2.5 3 2.40 2 1 0.8 No 35 Allpahuayo 3056 ' 19" 73o02'36 " 2.5 1 2.47 7 1 2.8 No 36 E Yasuni 1 0°40'31" 76002'20" 2.5 3 3.23 2 1 0.8 No 37 E Dicaro 3 1°14'11" 76o01'7" 2.5 3 3.81 6 1 2.4 No 38 *Tarapoto a 3046'35 " 73°02'41 " n.d. 1 5.51 89 6 n.d. No 39 Lib. Agr. km 8 4009 '15" 73o05'20 " 2.5 3 5.75 + 1 n.d. 1 1 0.4 No 40 7 de Julio 4022'30 " 73°00'42 " 2.5 3 7.16 3 1 1.2 No 4 l E Yasunl 10 0032'38 " 76003'7 " 2.5 3 8.54 2 1 0.8 No 42 Tarapotob 3046'35 " 73°02'41 " 6.5 14 10.69 16 1 2.5 No 43 *Palo Seco 1 3059'43 " 73°02'26 " n.d. 1 18.43 30 5 n.d. No 44 Jenaro Herrera 1 4053 '13" 73°03'45" n . d . n . a , n.a. 25 1 10.0 Yes 45 Sucusari 3°16'00" 72°05'41 " 2.5 n.a. n.a. + 1 n.d. Yes 46 Sucusar iplot 10 3°15 '42 " 72005'27 " n . d . n . a , n.a. + 1 n.d. No 47 Sucusar iplot 106 3°15 '16 " 72°04'53 " 1 n.a. n.a. + 1 n.d. ? 48 Sucusar ip lo t 239 3°14'49" 72004 '16" 1 n.a. n.a. + 1 n.d. ? 49 Sucusari plot 77 3°15 '20 " 72°05'19" 1 n.a. n.a. ÷ 1 n.d. ? 50 Yarinal 1 3°57'17" 73002'33" n . d . n . a , n.a. + 2 n.d. No

TOTAL 133 187 >640 82 >619 73


F + 0.5

- 0.4

O0

0

- 0 , 2 "t+-tt-'~

+ +.1t.4t-

-~t.-+ %

% + + 0 ° -I- + O O

-0

-0.2 0 I I I

[]

[] . ~ . []

+ o= ~ ~a o o ~ l u

[]

0

0

n

[

- 0.5

- tO <5 lhd

+x.,

-m. o

o oo o ° clmO ~OD []

°°Q o

b h d -

0 . 2 0 . 4

I I

acJ o []

on

O

+ - ~ []

* , +*~-¢~Do ~ ~P ~ ~

Fig. 3. (a) An ordination of morphologically analyzed Clidemia specimens based on principal coordinates analysis (PCoA, Gower 's distance; n = 155). Specimens lacking minute reddish glands on the abaxial surface of the laminae are marked with a cross, those possessing such glands with a square. The encircled point represents the type specimen ofC. epiphytica vat. trichocalyx. The directions o f increase of the morphological variables are shown in the insert (lain - lamina length-to-width ratio, p e t - lamina-to-petiole length ratio, bhd - distribution of branched hairs, lhd - distribution o f long hairs). The axes account for 10.01% o f the total variation. (b) The same ordination as in (a) but with specimens from the sites "Mishana b" and "Mishana c" (n = 6) encircled. (c) The same ordination as in (a) but with specimens from the site "Libertad Agraria km 8" (n = 2) encircled. See Fig. 2 for locations o f study sites on the map, and Table 1 for coordinates.

Morphological analyses

We examined the morphology of the 155 specimens (less one with leaves too damaged) acquired within the two collecting schemes described above. For comparison, we also examined the type specimen of C. epiphytica var. trichocalyx (originating from Honduras, deposited at US). The specimens we collected are deposited at AMAZ, USM, QCA, QCNE, or COAH, with duplicates at TUR (herbarium acronyms after HOLMGREN et al. 1990).

We collected the morphological information from one leaf per specimen, because variation within specimens was deemed insignificant. We could not use floral or fruit characters, since the majority of the specimens were sterile. Measured characters were length and width of lamina, and length of petiole. We calculated a length-to-width ratio for the lamina, and a lamina-to-petiole length ratio for the whole leaf. In some specimens the abaxial surface of the lamina was sparsely to very densely covered with small, dark reddish protuberances or glands


(approx. 0.02-0.04 mm in diameter; much like those called "sessile glands with thin-walled heads" and shown in figure 41 in WURDACK (1986), although not verified with SEM). We coded the presence of these glands in a semi-quantitative manner: 0 - no glands, 1 - few glands, 2 - abundant glands.

We analyzed the distribution of laminar pubescence by first recording the presence of longish (generally between 1 and 3 mm) unbranched hairs, or setulae, separately on primary, secondary, and tertiary veins, and elsewhere on the lamina surface. The presence of setulae in these areas yielded one point per area, while absence was coded as 0. A single value describing the distribution of setulae on the lamina was then obtained by adding these values together. In the same way we recorded the presence of 0.1~).3 mm long branched hairs (corresponding to the type "dendritic hairs with well-developed axis and moderate number of terete short to moderately long arms" and figures 159-162 in WURDACK 1986). Hence, the distribution of hairs was described by two semi-quantitative values between 0 and 4. The distribution of both hair types followed the rank order of the veins so that presence on lower ranking veins was always accompanied by presence on all higher ranking veins.

We used the results of the morphological studies to construct a distance matrix of the specimens giving the value of Gower's distance for every possible comparison between two specimens. Gower's distance was selected as the distance measure because it can combine information from binary, ordinal, and continuous variables into a single meaningful index (LEGENDRE & LEGENDRE 1998). It is the sum of the squared differences between values in the cophenetic similarity matrix and the original similarity matrix:

DGOZeER = E i , j (original Sg - cophenetic Slj) 2

We then used the distance matrix for a principal coordinates ordination (PCoA), which illustrated the general pattern of morphological similarities and differences among the specimens. Gower's distances were calculated and the PCoA carried out with the software R-Package (LEGENDRE & VAUDOR 1991). For other statistics we used the program JMP (SAS INSTITUTE 1995).

RESULTS

Morphological variation and distinction of taxa

There is a clear morphological gradient among the analyzed specimens, which is illustrated by the correlation between the main axis of variation of the PCoA, and the variables "lamina length-to-width ratio", "lamina-to-petiole length ratio", and "distribution of branched hairs" (Fig. 3a). On the left side of the ordination are the specimens with broad laminae and long petioles, corresponding to published descriptions of C. epiphytica. The type specimen of C. epiphytica var. trichocalyx is also situated there. Specimens with narrow laminae and short petioles, recognizable as C. longifolia, are grouped to the right. However, in the middle there are a number of specimens that are intermediate in all analyzed variables, and a few that are intermediate in leaf shape but differ in their position along the second axis of variation because they are extreme in their score for distribution of long hairs.

152 L. Schulman etal.

Table 2. Summary of results of morphological analyses of 155 Clidemia specimens. Shown are means of scores recorded, with minimum and maximum values in parentheses, and standard deviations (s.d.). Specimens possessing minute reddish glands on the abaxial side of the laminae are here listed under the name U. longifolia, those lacking such glands under C. epiphytica. Specimens with the scores 0 or 1 in the variable "distribution of long hairs" are listed as C. epiphytica var. epiphytica, those with the score 4 as var. trichocalyx (see text).

C. tongifolia C epiphytiea, all var. epiphytica var. trichocalyx (n = 73) (n = 82) (n = 47) (n = 29)

lamina length-width ratio mean (1.8-)4.1(8.2) (1.3-)2.1(-3.0) (1A-)2.1(-2.9) (1.3~2.0(-3.0) s.d. 1.5 0.4 0.4 0.4

petiole-lamina length ratio mean (1.4-)24.1(-101.0) (0.6-)4.2(-15.7) (1.6-)4.6(-15.7) (0.6-)3.6(-6.8) s.d. 19.3 2.4 2.8 1.6

petiolelength cm mean (0.1-)1.3G14.5) (0.2-)4.3(-18.5) (0.2-)3.7(-13.0) (1.0-)5.2(-18.5) s.d. 2.0 3.1 2.6 3.9

long hair distribution mean (0 - )1 (4) ( 0 ) 2 ( - 4 ) (0-)1(-1) (4-)4(-4) s.d. 1 2 0 0

branched hair distribution mean (0-)3(-4) (0-)1(-3) (0-)1(-3) (0-)1(-3) s.d. 1 1 1 1

Despite the relatively continuous variation in the quantitative and semiquantitative variables, the specimens can be divided into two reasonably well discernible groups, if they are marked in a qualitative manner as either possessing or not possessing small reddish glands on the abaxial surface of the laminae. The presence of such glands is a character state not previously reported for either taxon, but since it is congruent with other character states typical of C. longifolia, we shall henceforth apply the name C. longifolia to specimens with red glands, and the name C. epiphytica to those without red glands. (After having conducted the analyses described here, we studied a close-up photograph, kindly provided by R. Moran, showing the abaxial lamina surface of the type specimen of C. longifolia deposited at NY, and confirmed that it has plenty of reddish glands.)

Clidemia epiphytica presented a strongly bimodal pattern in the distribution of long hairs: most specimens had either a very restricted, or else a wide distribution of long hairs (scores for this variable 0-1 or 4, respectively; results not shown). The former group corresponds with t he description of the type variety, the latter with the description of var. trichocalyx (WURDACK 1964).

The results of the morphological analyses are summarized in Table 2. When presence or absence of red glands on the abaxial side of the lamina is used as the distinguishing criterion, the mean values of all variables except "distribution of long hairs" are clearly different in the two species. However, when minimum and maximum values are compared, there are various degrees of overlap in all variables.

Distribution patterns

The specimens of C. epiphytica and C. longifolia were collected from somewhat different but strongly overlapping areas (Fig. 2). In Ecuador we collected only C. epiphytica, and in Colombia only C. longifolia. In the Peruvian part of the study area both species were fotmd intermingled. On a more detailed scale, however, there are only two localities (three sites), where both species were collected (Table 1; Fig. 2).

Ecological segregation of Cfidemia species 153

8 4a I

3

2

1 0.01

,ooi4 1 80

60

40

20

0 0.01

7 e "

" 0

5 0

4 r - .

E c~

r =

-6

¢--

¢ -

¢E r -

E

0 []

[]

0

0 § o

o oB

I I I

0.1 1 10 100

ors o

0 n B

n o

[] n o ~ ,~1 .

0.1 1 10 100

sum of cations (cmol+/kg) Fig. 4. The relationship between soil cation concentration and lamina shape (a) and soil cation concentration and relative petiole length (b) among specimens of Clidemia epiphytica (crosses) and C. longifolia (squares). In C. epiphytica, the lamina becomes broader and the petiole relatively longer with increasing cation concentration. In C. longifolia, the lamina becomes narrower and the petiole relatively shorter with increasing cation concentration.

The two species were mostly encountered in soils with different cation concentrations (Table 1). Clidemia longifolia was restricted to substrates relatively poor in cations, apart from a single individual encountered in a site with a relatively rich soil (Libertad Agraria km 8). In the cation richest site where C. longifolia occurred in abundance (Santa Ana), C. epiphytica was not encountered. In comparison, the cation concentration gradient covered by C. epiphytica is much wider. However, on soils very poor in cations it occurred only in low densities and only in the absence of C. longifolia. The sole exception to this pattern were the adjacent sites Mishana b and c (same locality), where both taxa were found in high densities despite low cation concentrations.

The occurrences of the two species were divided between river terraces and non-terrace environments. Of the 26 sites where C. longifolia was encountered, 23 could be designated as either terrace or non-terrace. Only two of these were of the latter type. Of the 27

sites where C. epiphytica was encountered, 24 were non-terraces, two were terraces, and one was undefined.

Morphological variation and soil cation concentration

In Fig. 4, the values of the lamina length-to-width ratio and the lamina-to-petiole length ratio of all studied specimens are plotted against the cation content of their substrate (mean of values obtained from soil samples collected at respective study site). However, for the assessment of a possible correlation between variation in morphology and cation content we


could not use the individual specimens because our data did not allow us to connect a separate soil cation content to each specimen. Instead, we compared the population mean to the mean of cation content values obtained from soil samples collected at the corresponding study site. The procedure potentially hides fine-scale variation in the soil, and may not reveal the relationship between morphology and soil variation to full detail.

Analyzed in this way, lamina shape and relative petiole length of C. epiphytica correlated negatively with soil cation concentration so that the broadest laminae and longest petioles were encountered on the soils richest in cations (Box-Cox transformed variables; n = 26, r = -0.36, P = 0.0695 and r = -0.41, P = 0.0380, respectively). In contrast, C. longifolia showed a positive relationship between both morphological variables and cation concentration, i.e., the broadest laminae and longest petioles were encountered on the poorest soils, although the latter correlation was not significant (Box-Cox transformed variables; n = 22, r = 0.60, P = 0.0036 and r = 0.20, P = 0.3633, respectively).

We tested the statistical significance of the difference in responses between the two species by a permutation procedure. Assuming that our morphological definitions of C. longifolia and C. epiphytica would not bear any ecological significance, we should readily obtain an equal or a bigger difference between two correlation coefficients calculated for two randomly taken combinations of 26 (no. of sites with C. epiphytica) and 22 (no. of sites with C. longifolia) values of cation content and lamina shape, or cation content and relative petiole length. We performed this test with 1000 permutations. We found that the probability of randomly observing an equal or a bigger difference in correlation coefficients, which the two taxa have between lamina shape and soil cation content, is 0.6%. The corresponding probability for the difference in correlation coefficients between relative petiole length and cation content is 1.7%. Hence we contend that the two taxonomic species show different morphological responses to variation in soil cation content.

Lamina shape and relative petiole length correlated negatively with the size of the plant (the taller the specimen the wider the laminae and the longer the petioles) both in C. epiphytica (Box-Cox transformed variables; r = -0.24, P = 0.040 and r = -0.47, P < 0.001, respectively) and in C. longifolia (r = -0.28, P = 0.013 and r = -0.33, P = 0.010, respectively). However, these correlations cannot account for the correlations between soil cation concentration and lamina shape or relative petiole length, since the height of the plant and the cation concentration did not correlate with each other (C. epiphytica: r = -0.13, P = 0.290; C. longifolia: r = -0.13, P = 0.322). The effect of variation in plant height (=age) is one explanation of the within-population variation observable in Fig. 4.

The degree ofhairiness in specimens of C. epiphytica and C. longifolia was independent of the height of the plant and of cation concentration (none of the hair distribution classes had statistically different means of plant height or cation concentration at P < 0.05; Tukey-Kramer HSD). The two varieties of C. epiphytica did not show any statistical difference in their response to soil cation concentration (P > 0.05; Student's t-test), if long-hair distribution classes 0 and 1 are taken as the type variety, and class 4 as var. trichocalyx.

In one of the localities where both species were found together (sites Mishana b and c) the specimens are morphologically intermediate (Fig. 3b), whereas in the other such locality


(Libertad Agraria kin 8) the specimens are morphologically distinct (Fig. 3c). The two varieties of C. epiphytica were observed together in several sites.

DISCUSSION

Morphology and taxonomy

Our morphological analyses showed that there is a bimodal distribution in leaf shape and petiole length, which are the quantitative morphological characters used to separate C. epiphytica and C. longifolia (GLEASON 1931, WURDACK 1964). In addition, this pattern is congruent with that observed in one qualitative character: the presence of minute reddish glands on the abaxial surface of the lamina, which often gives the specimen an overall dark colour. This is the qualitative attribute best characterizing C. longifolia, which also usually has plenty of small dendritic hairs along the main veins on the abaxial surface of the lamina. The leaves of C. epiphytica have fewer dendritic hairs, and lack reddish glands (whitish glands are sometimes present). With these additions our results confirm and reinforce the taxonomic decisions of WURDACK (1964).

Distribution patterns and niche partitioning

The previous understanding that the geographical range of C. longifolia is nested within that ofC. epiphytica was not contradicted by our findings. However, the species seldom occur side by side. Rather, they grow on soils with different cation concentrations and in landscape sections with different geological history. Because general floristic composition of western Amazonian forests correlates well with cation concentration of the soil (TUOMISTO et al. 1995, 2003, RUOKOLAINEN et al. 1997, RUOKOLAINEN & TUOMISTO 1998, VORMISTO et al. 2000), it seems reasonable to regard forests on different soils as different habitats. Hence, C. longifolia is found in habitats characterized by soils poor in cations, and located on river terraces. The single individual of this species found in site Libertad Agraria km 8, which has a rich soil, can be attributed to mass effect (SHMIDA & WILSON 1985), since poor soils and dense populations of C. longifolia occur within a short distance (Table l, Fig. 2). Clidemia epiphytica dominates habitats with soils rich in cations, and located on non-terrace surfaces (we found no habitat differences between the two varieties ofC. epiphytica). Also the poorest sites in which C. epiphytica was found were located in non-terrace areas. It is noteworthy that when C. epiphytica does occur on soils with very few cations, its population density is low and it does not co-occur with C. longifolia.The explanation for this result may be that in these areas the environment has some feature that went undetected in our analyses, but still affects the success of plants growing there.

The only exceptions to the above-described patterns that we found, were the adjacent sites Mishana b and c. The explanation for the co-occurrence of the species in these sites may be their mixed geology. The sites are on a geologically very recent terrace of the black-water river Nanay (KALLIOLA et al. 1993), and the soils generally have a low cation content. However, the much more nutrient-rich Pebas Formation has been found to lie a few meters from the soil surface in the near-by village ofMishana (A. LINNA, unpubl, data). Hence, it is


possible that within the terrace these more nutrient-rich sediments can be found sufficiently close to the surface to affect the flora.

At least in the Peruvian part of our study area, the different soil types typically occur as distinct intermingled patches that are sometimes quite small, in the order of hundreds of meters across (TUOMISTO et al. 1995, KALLIOLA • FLORES PAIT,/~N 1998). Consequently, individuals of the apparently insect-pollinated and bird-dispersed C. epiphytica and C. longifolia growing in different patches, are certainly within "normal cruising range" of each other, i.e., physically capable of coming into reproductive contact with moderately high frequency (this is the definition of sympatry employed by MAYR (1963), and many subsequent authors). Hence we conclude that C. epiphytica and C. longifolia are sympatric within the range of the latter, but they are spatio-ecologically segregated into separate niches, which are defined by different habitat preferences.

Similar habitat partitioning in the surroundings of Iquitos, Peru, within the seemingly uniform tierra firme forests, has previously been reported for several angiosperm genera and some pteridophytes (GENTRY 1981, RUOKOLAINEN & TUOMISTO 1998). This suggests that part of the enigma of series of closely related plant species occurring sympatrically in tropical rainforests may be explainable by specialization in relation to subtle environmental variation that often is unknown in ill-studied regions, as predicted by RICHARDS (1969). Niches of sympatric congeners in rainforests may, thus, be definable spatio-ecologically, in addition to possible differences in phenology and other reproductive features (cf. SNOW 1965, HILTY 1980, GENTRY 1986, ROGSTAD 1990, BORSEHENIUS 2002).

Evolutionary implications We have shown that C. epiphytica and C. longifolia are morphologically separable, when

their phenotypic variation is understood, and can therefore be regarded as good taxonomic species. Furthermore, the two species occur in nature sympatrically but in different habitats. However, taxonomic distinctiveness and probable sister relationship in a cladogram (see above under "Study species, phylogenetic relationships") do not prove that C. epiphytica and C. longifolia are separate biological species. In evaluating the possibility that they are such, the few specimens showing intermediate morphology are of key interest. These could be interpreted as: (1) a sign of gradual differentiation due to distance, i.e., the two "species" actually being geographical races; (2) a sign of hybridization; or (3) phenotypic variation within one or among two species.

Geographical differentiation can be ruled out, because morphologically extreme forms of both kinds were collected in interspersed sites in the Peruvian part of our study area. That the intermediate specimens could be hybrids between two distinct biological species is possible, but two things speak against it. First, no intermediates were found in Libertad Agraria km 8, and only intermediates in Mishana b and c. This is only to be expected on the basis of the morphological response of the species to variation in cation concentration: when the soil is poor (Mishana b and c), their leaf shapes approach one another, whereas in rich soils (Libertad Agraria km 8) they are morphologically distinct. Also, if the intermediate specimens were the result of hybridization between two morphologically distinct species, one would expect all three sites to have a range of both extreme and intermediate forms. Second, despite the short


geographical distance between many o f the study sites, only in two did the population consist o f morphologically intermediate specimens. I f hybridization occurred frequently, more populations with signs o f this should have been found. Hence, we conclude that phenotypic variation among two separate gene pools, i.e., two separate biological species, is the most likely explanation for our results, despite the specimens with intermediate morphology.

This hypothesis should be tested: are the two morphologically and ecologically separated taxonomic species also genetically different, and are there mechanisms that prevent gene flow between them? This is an interesting question because the distribution patterns and autecological differences o f C. epiphytica and C. longifolia provide notable circumstantial evidence that their divergence is a good candidate case o f sympatric speciation.

GENTRY (1982) proposed that explosive adaptive speciation, essentially sympatric and mostly powered by Andean orogeny, accounts for the "excess plant diversity o f the Neotropics" compared to the Paleotropics. He estimated the "excess" to equal approximately 40,000 seed plant species! This number includes many extra-Amazonian taxa, but the whole scenario is very important also for the western, and to some extent the northern parts o f the Amazon basin. In conclusion, the prospects o f possibly documenting a case o f sympatric plant speciation driven by edaphic specialization in Amazonian tierra firme forests are exciting

indeed.

Acknowledgements: We thank all persons, too numerous to be listed here, who helped us in the field. We also want to acknowledge the kind help received from local inhabitants in all study areas. Special thanks go to H. Tuomisto, R. A~i, A. Alonen, A. Linna, E. Lusa, J. Prlkki, L. Rebata Hernani, I. E. S~i~iksj~rvi, and T. Toivonen for help in various phases of the study. Earlier versions of the manuscript were much improved by suggestions of B. Simpson and three anonymous reviewers. We are grateful to Universidad Nacional de la Amazonia Peruana and Instituto Nacional de la Investigacirn Agraria (Peru), to Pontificia Universidad Catrlica del Ecuador, and to Universidad de los Andes (Colombia) for co-operation, and to the herbarium of the University ofTurku (TUR) for excellent working facilities. The study formed part of work financially supported by the Ministry of Foreign affairs of Finland, the Academy of Finland, the STD-3 and INCO-DC programmes of the EC, TOP Foundation, Hilma and Heikki Honkanen Foundation, and the Finnish Biodiversity Research Programme FIBRE.

REFERENCES

ALMEIDA C.R. & VISTULO DE ABREU F. (2003): Dynamical instabilities lead to sympatric speciation. Evol. Ecol. Res. 5: 739-757.

ASHTON P.S. (1969): Speciation among tropical forest trees: some deductions in the light of recent evidence. Biol. J. Linn. Sac. 1: 155-196.

BERLOCHER S.H. & FEDER J.L. (2002): Sympatric speciation in phytophagous insects: moving beyond controversy? Annual Rev. Entomol. 47: 773-815.

BEVEN S., CONNOR E.F. & BEVEN K. (1984): Avian biogeography in the Amazon basin and the biological model of diversification. J. Biogeogr. 1 l: 383-399.

BORSCHENIUS F. (2002): Staggered flowering in four sympatric varieties of Geonoma cuneata (Palmae). Biotropica 34: 603-606.

BRAKO L. & ZARUCCHI J.L. (1993): Catalogue of the flowering plants and gymnosperms of Peru, Cat~ilogo de las angiospermas y gimnospermas del Peril. Monogr. Syst. Bot. Missouri Bot. Gard. 45.

BROWN K.S., Jr. (1982): Historical and ecological factors in the biogeography of aposematic neotropical butterflies. Amer. Zool. 22: 453-471.

BUSH G.L. (1998): The conceptual radicalization of an evolutionary biologist. In: HOWARD D.J. & BERLOCHER S.H. (eds.), Endless forms: species and speciation, Oxford University Press, New York, pp. 425-438.


BUSH M.B. (1994): Amazonian speciation: a necessarily complex model. J. Biogeogr. 21: 5-17. BUSH M.B. (1996): Amazonian conservation in a changing world. Biol. Conservation 76: 219-228. COLINVAUX P.A., DE OLIVEIRA P.E. & BUSH M.B. (2000): Amazonian and Neottopical plant communities on

glacial time-scales: The failure of the aridity and refuge hypothesis. Quatern. Sci. Rev. 19: 141-169. COLINVAUX P.A., IPdON G., RASANEN M.E., BUSH M.B. & NUNES DE MELLO J.A.S. (2001): A paradigm to be

discarded: geological and paleoecological data falsify the Haffer & Prance refuge hypothesis of Amazonian speciation. Amazoniana 16: 609-646.

DUIVENVOORDEN J.F. & LIPS J.M. (1993): Landscape ecology of the middle Caquet6 basin. Explanatory notes to the maps. Studies on the Colombian Amazonia, III A. Tropenbos, Santaf6 de Bogot~i.

ENDLER J.A. (1982): Pleistocene forest refuges: fact or fancy? In PRANCE G.T. (ed.), Biological diversification in the tropics, Columbia University Press, New York, pp. 641-657.

FEDOROV An.A. (1966): The structure of the tropical rainforest and speciation in the humid tropics. J. Ecol. 54: 1 11.

GAVRILETS S. & WAXMAN D. (2002): Sympatric speciation by sexual conflict. Proc. Natl. Acad. Sci. USA 99: 10533-10538.

GENTRY A.H. (1976): Bignoniaceae of southem Central America: distribution and ecological specificity. Biotropica 8:117-131.

GENTRY A.H. (1981): Distributional patterns and an additional species of the Passiflora vitifolia complex: Amazonian species diversity due to edaphically differentiated communities. Pl. Syst. Evol. 137: 95-105.

GENTRY A.H. (1982): Neotropical floristic diversity: phytogeographical connections between Central and South America, Pleistocene climatic fluctuations, or an accident of the Andean orogeny? Ann. Missouri Bot. Gard. 69: 557-593.

GENTRY A.H. (1986): Endemism in tropical versus temperate plant communities. In: SOULI~ M.E. (ed.), Conservation biology, Sinauer, Sunderland, pp. 153-181.

GENTRY A.H. (1989): Speciation in tropical forests. In HOLM-NIELSEN L. B., NIELSEN I. C. & BALSLEV H. (eds.), Tropical forests: Botanical dynamics, speciation and diversity, Academic Press, London, pp. 113-134.

GENTRY A.H. (1990): Evolutionary patterns in neotropical Bignoniaceae. Mem. New York Bot. Gard. 55: 118-129.

GLEASON H.A. (1931): Studies on the flora of northern South America - XV. Recent collections of Melastomataceae from Peru and Amazonian Brazil. Bull. Torrey Bot. Club 58: 215-262.

HAFFER J. (1969): Speciation in Amazonian forest birds. Science 165: 131-137. HAFFER J. (1997): Alternative models of vertebrate speciation in Amazonia: an overview. Biodiv. Conservation

6: 451-476. HAFFER J. & PRANCE G.T. (2001): Climatic forcing of evolution in Amazonia during the Cenozoic: on the

refuge theory of biotic differentiation. Amazoniana 16: 579-607. HILTY S.L. (1980): Flowering and fruiting periodicity in a premontane rainforest in Pacific Colombia.

Biotropica 12: 292-306. HOLMGREN P.K., HOLMGREN N.H. & BARNETT L.C. (eds.) ( 1990): Index Herbariorum Part I: The Herbaria of

the World (8th edition). Regnum Veg. 120. HUBBELL S.P. (2001): The unified neutral theory of biodiversity and biogeography. Princeton University Press,

Princeton. JOHANNESSON K. (2001): Parallel speciation: a key to sympatric divergence. Trends Ecol. Evol. 16:148-153. JUDD W.S. (1989): Taxonomic studies in the Miconieae (Melastomataceae). III. Cladistic analysis of

axillary-flowered taxa. Ann. Missouri Bot. Gard. 76: 476-495. KALLIOLA R. & FLORES PAITAN S. (eds.) ( 1998): Geoecologiay desarrollo Amazrnico: estudio integrado en la

zona de Iquitos, Peril. Ann. Univ. Turku A II, 114. University of Turku, Turku. KALLIOLA R., LINNA A., PUHAKKA M., SALO J. & RASANEN M. (1993): Mineral nutrients in fluvial sediments

from the Peruvian Amazon. Catena 20: 333-349. LEGENDRE P. & LEGENDRE L. (1998): Numerical ecology (2nd English edition). Developments in

environmental modelling 20, Elsevier, Amsterdam. LEGENDRE P. & VAUDOR A. (1991 ): The R-package: Multidimensional analysis, spatial analysis. Drpartement

de sciences biologiques, Universit6 de Montr6al, Montreal. LEVIN D.A. (2001): 50 years of plant speciation. Taxon 50: 69-91.


LINHART Y.B. & GRANT M.C. (1996): Evolutionary significance of local genetic differentiation in plants. Annual Rev. EcoL Syst. 27: 237-277.

MARENGO J.A. (1998): Climatologia de la zona de Iquitos, Peril. In: KALLIOLA R. & FLORES PAITAN S. (eds.), Geoecologia y desarrollo Amaz6nico, estudio integrado en la zona de lquitos, Peril, Ann. Univ. Turku A II, 114, University of Turku, Turku, pp. 35-57.

MARTIN O.Y. & HOSKEN D.J. (2003): The evolution of reproductive isolation through sexual conflict. Nature 423: 979-982.

MAYR E. (1963): Animal species and evolution. Belknap, Cambridge. MIZERA F. & MESZI~NA G. (2003): Spatial niche packing, character displacement and adaptive speciation along

an environmental gradient. Evol. Ecol. Res. 5: 363-382. NELSON B.W., FERREIRA C.A., DA SILVA M.F. & KAWASAKI M.L. (1990): Refugia, endemism centers and

collecting density in Brazilian Amazonia. Nature 345: 714-716. NOSIL P., CRESPI B.J. & SANDOVAL C.P. (2002): Host-plant adaptation drives the parallel evolution of

reproductive isolation. Nature 417: 440-443. OGDEN R. & THORPE R.S. (2002): Molecular evidence for ecological speciation in tropical habitats. Proc. Natl.

Acad. Sci. USA 99: 13612-13615. PRANCE G.T. (1973): Phytogeographic support for the theory of Pleistocene forest refuges in the Amazon Basin,

based on evidence from distribution patterns in Caryocaraceae, Chrysobalanaceae, Dichapetalaceae and Lechytidaceae. Acta Amazon. 3: 5-25.

RAS~,NEN M.E., SALO J.S. & KALLIOLA R.J. (1987): Fluvial perturbance in the Western Amazon basin: Regulation by long-term sub-andean tectonics. Science 238:1398-1401.

RAS~.NEN M.E., NELLER R., SALO J.S. & JUNGNER H. ( 1992): Recent and ancient fluvial deposition systems in the Amazonian foreland basin, Peru. Geol. Mag. 129: 293-306.

RAS~.NEN M.E., L1NNA M., SANTOS J.C.R. & NEGRI F.R. (1995): Late Miocene tidal deposits in the Amazonian foreland basin. Science 269: 386-390.

RICHARDS P.W. (1969): Speciation in the tropical rainforest and the concept of the niche. Biol. ,1. Linn. Soc. 1: 149-153.

ROGSTAD S.H. (1990): The biosystematics and evolution of the Polyalthia hypoleuca species complex (Annonaceae) of Malesia. II. Comparative distributional ecology. J. Trop. EcoL 6: 387-408.

ROMOLEROUX K., FOSTER R., VALENCIA R., CONDIT R., BALSLEV H. & LOSOS E. (1997): Arboles y arbustos (dap > 1 cm) encontrados en dos hect~ireas de un bosque de la Amazonia ecuatoriana. In: VALENCIA R. & BALSLEV H. (eds.), Estudios sobre diversidad y ecologia de plantas, Memorias del H Congreso Ecuatoriano de Botdnica, Pontificia Universidad Cat61ica del Ecuador, Quito, pp. 189-215.

RUOKOLAINEN K., LINNA m. & TUOMISTO H. (1997): Use of Melastomataceae and pteridophytes for revealing phytogeographic patterns in Amazonian rainforests. J. Trop. Ecol. 13: 243-256.

RUOKOLAINEN K. & TUOMISTO H. (1998): Vegetaci6n natural de la zona de Iquitos. In: KALLIOLA R. & FLORES PAITAN S. (eds.), Geoecologia y desarrollo Amaz6nico, estudio integrado en la zona de lquitos, Peril, Ann. Univ. Turku A II, 114, University of Turku, Turku, pp. 253-365.

SALO J. (1987): Pleistocene forest refuges in the Amazon: evaluation of the biostratigraphical, lithostratigraphical and geomorphical data. Ann. Zool. Fenn. 24:203-211.

SALO J., KALLIOLA R., HAKKINEN I., M.~KINEN Y., NIEMELA. P., PUHAKKA M. & COLEY P.D. (1986): River dynamics and the diversity of Amazon lowland forest. Nature. 322: 254-258.

SAS INSTITUTE (1995): JMP Statistics and Graphics Guide: version 3.1 of JMP. SAS Institute, Cary. SCHLUTER D. (2001): Ecology and the origin of species. Trends Ecol. Evol. 16: 372-380. SHMIDA A. & WILSON M.W. (1985): Biological determinants of species diversity. J. Biogeogr. 12: 1-20. SNOW D.W. (1965): A possible selective factor in the evolution of fruiting seasons in tropical forest. Oikos 15:

274-281. SORENSON M.D., SEFC K.M. & PAYNE R.B. (2003): Speciation by host switch in brood parasitic indigobirds.

Nature 424: 928-931. SVENNING J.-C. (2001): On the role of microenvironmental heterogeneity in the ecology and diversification of

neotropical rain-forest palms (Arecaeeae). Bot. Rev. 67: 1-53. TILMAN D. (1994): Competition and biodiversity in spatially structured habitats. Ecology 75: 2-16. TILMAN D. & PACALA S. (1993): The maintenance of species richness in plant communities. In: RICKLEFS R.E.

& SCHLUTER D. (eds.), Species diversity in ecological communities: historical and geographical perspectives, The University of Chicago Press, Chicago, pp. 13-25.


TER STEEGE H. (1998): The use of forest inventory data for a national protected area strategy in Guyana. Biodiv. Conservation 7: 1457-1483.

TUOMISTO H., RUOKOLA1NEN K., KALLIOLA R., LINNA A., DANJOY W. & RODRJGUEZ Z. (1995): Dissecting Amazonian biodiversity. Science 269: 63-66.

TUOMISTO H., RUOKOLAINEN K. & YLI-HALLA M. (2003): Dispersal, environment, and floristic variation of western Amazonian forests. Science 299: 241-244.

TURELLI M., BARTON N.H. & COYNE J.A. (2001): Theory and speciation. Trends EcoL Evol. 16: 330-343. VIA S. (2001): Sympatric speciation in animals: the ugly duckling grows up. Trends Ecol. Evol. 16:381-390. VORMISTO J., PHILLIPS O., RUOKOLAINEN K., TUOMISTO H. & VA.SQUEZ R. (2000): A comparison of

small-scale distribution patterns of four plant groups in an Amazonian rainforest. Ecography 23: 349-359. WERFF H. VAN DER (1992): Substrate preference of Lauraceae and ferns in the Iquitos area, Peru. Candollea 47:

11-20. WURDACK J.J. (1964): Melastomataceas nuevas Venezolanas. Bol. Soc. Venez. Ci. Nat. 25:213-215. WURDACK J.J. (1986): Atlas of hairs for neotropical Melastomataceae. Smithsonian Contributions to Botany

63. Smithsonian Institution Press, Washington. WURDACK J.J. (1988): 138. Metastomataceae. In: HARLING G. & SPARRE B. (eds.), Flora of Ecuador, Opera

Botanica B 13, Swedish Natural Science Research Council, Editorial Service, Stockholm. WURDACK J.J. (1993): Melastomatoideae. In: GORTS-VAN RIJN A.R.A. (ed.), Flora of the Guianas, Series A:

Phanerogams, 13, Koeltz Scientific Books, Koenigstein, pp. 3-301.

Received 17 June 2003, revision received 21 January 2004, accepted 10 February 2004

Spatio-ecological niche segregation of two sympatric species ofClidemia (Melastomataceae) in Western Amazonian non-flooded rainforests

Documents