Journal of Ecology 2006 94, 157-170 Soil nutrients and beta diversity in the Bornean Dipterocarpaceae: evidence for niche partitioning by tropical rain forest trees GARY D. PAOLI1*, LISA M. CURRAN| and DONALD R. ZAK*| ^Department of Ecology and Evolutionary Biology, University of Michigan, 830 N. University Avenue, Ann Arbor, Ml 48109, USA, j" Yale School ofForestry and Environmental Studies, 205 Prospect Street, New Haven, CT06511, USA, and %School of Natural Resources and Environment, University of Michigan, Ann Arbor, MI 48109, USA Summary 1 The relative importance of niche- and dispersal-mediated processes in structuring diverse tropical plant communities remains poorly understood. Here, we link mesoscale beta diversity to soil variation throughout a lowland Bornean watershed underlain by alluvium, sedimentary and granite parent materials (c. 340 ha, 8-200 m a.s.L). We test the hypothesis that species turnover across the habitat gradient reflects interspecific partitioning of soil resources. 2 Floristic inventories (> 1 cm d.b.h.) of the Dipterocarpaceae, the dominant Bornean canopy tree family, were combined with extensive soil analyses in 30 (0.16 ha) plots. Six samples per plot were analysed for total C, N, P, K, Ca and Mg, exchangeable K, Ca and Mg, extractable P, texture, and pH. 3 Extractable P, exchangeable K, and total C, N and P varied significantly among substrates and were highest on alluvium. Thirty-one dipterocarp species (« = 2634 individuals, five genera) were recorded. Dipterocarp density was similar across substrates, but richness and diversity were highest on nutrient-poor granite and lowest on nutrient-rich alluvium. 4 Eighteen of 22 species were positively or negatively associated with parent material. In 8 of 16 abundant species, tree distribution (> 10 cm d.b.h.) was more strongly non-random than juveniles (1-10 cm d.b.h.), suggesting higher juvenile mortality in unsuitable habitats. The dominant species Dipterocarpus sublamellatus (> 50% of stems) was indifferent to substrate, but nine of 11 'subdominant' species (> 8 individuals ha-1) were substrate specialists. 5 Eighteen of 22 species were significantly associated with soil nutrients, especially P, Mg and Ca. Floristic variation was significantly correlated with edaphic and geographical distance for all stems > 1 cm d.b.h. in Mantel analyses. However, juvenile variation (1-10 cm d.b.h.) was more strongly related to geographical distance than edaphic factors, while the converse held for established trees (> 10 cm d.b.h.), suggesting increased importance of niche processes with size class. 6 Pervasive dipterocarp associations with soil factors suggest that niche partitioning structures dipterocarp tree communities. Yet, much floristic variation unrelated to soil was correlated with geographical distance between plots, suggesting that dispersal and niche processes jointly determine mesoscale beta diversity in the Bornean Dipterocarpaceae. Key-words: diversity, dominance, magnesium, mass effects, mesoscale, neutral, niche, phosphorus, Shorea Journal of Ecology (2006) 94, 157-170 doi: 10.HH/j.1365-2745.2005.01077.x © 2005 The Authors Journal compilation © 2006 British Ecological Society Correspondence: Gary D. Paoli (e-mail hancur_lagi@yahoo.com or gpaoli@umich.edu). 158 G. D. Paoli, L. M. Curran & D. R. Zak © 2005 The Authors Journal compilation © 2006 British Ecological Society. Journal of Ecology, 94, 157-170 Introduction Discontinuities in soil drainage cause major changes in the species composition of lowland tropical tree communities and promote the coexistence of species adapted to contrasting edaphic environments (Lieberman et al. 1985; Newbery, Gartlan, McKey & Waterman 1986; Clark et al. 1999). The effects of discontinuities in geological substrate on community structure (Clark et al. 1998, 1999; Phillips et al. 2003) are, however, less well understood. Plant-soil surveys across substrate gradients at a regional scale (up to 75 000 km2) have documented distinct plant communities on contrasting substrates, and linked these differences to variation in soil phosphorus (P) or magnesium (Mg) (Austin etal. 1972; Gartlan et al. 1986; Baillie et al. 1987; Swaine 1996; Potts et al. 2002). This pattern suggests that substrate-mediated nutrient variation promotes regional beta diversity and, thus, coexistence of tropical trees (Sollins 1998). However, interpreting regional correlations is complex, because regional-scale studies potentially confound the effects of habitat partitioning with dispersal limitation and historical biogeographical processes (but see Phillips et al. 2003). Current understanding of substrate influences on beta diversity of tropical trees is further complicated by field sampling problems, including inadequate attention to the spatial design of soil sampling (e.g. Gartlan etal. 1986; Newbery, Renshaw & Brunig, 1986; Ola-Adams & Hall 1987) and insufficient replication of spatially independent floristic plots within substrates (Proctor et al. 1983; Newbery & Proctor 1984). Here, we address these limitations by combining intensive soil sampling with replicated tree surveys throughout a lowland Bornean watershed on contrasting geological substrates. We test the hypothesis that canopy tree beta diversity reflects interspecific partition of geological substrates and identify candidate soil nutrients underlying this pattern. Mesoscale beta diversity (sensu Clark etal. 1998) measures the extent of species turnover at spatial scales that encompass repeated sampling of landscape units, such as topographic catenas or geological substrates. Mesoscale sampling (1-100 km2) is broader than local studies (< 100 ha), but smaller than regional studies (102 to c. 105 km2), and offers three advantages for studying edaphic influences on community structure. First, the mesoscale captures a broader range of soil conditions than local scales and increases power for detecting plant-soil associations. Secondly, because dispersal limitation is less severe across mesoscale landscapes, and biogeographical and climatic histories are similar, plant-soil correlations at the mesoscale more strongly suggest niche partitioning than regional studies. Finally, the mesoscale enables convenient study of the combined influences of deterministic and neutral processes: niche partitioning, for instance, may cause species segregation across substrates, whilst dispersal-mediated mass effects (Shmida & Wilson 1985) may cause a degree of floristic mixing among them. Despite advantages of the mesoscale, and continued interest in tropical tree beta diversity (Hubbell 2001; Condit et al. 2002; Chave & Leigh 2002), relatively few studies have been conducted at this scale (Ashton 1976; ter Steege et al. 1993; Clark et al. 1998, 1999; Cannon & Leighton 2004; Hall et al. 2004). All of these studies found a significant effect of geological substrate, soil type or topography, but only Hall et al. (2004) quantified soil nutrients correlated with floristic change. Whether mesoscale beta diversity reflects niche partitioning, and, if so, the role of specific soil nutrients, thus remains poorly understood. Here, we combine a floristic inventory of rain forest trees in the Dipterocarpaceae, the dominant family of canopy trees in Borneo (Curran & Leighton 2000; Slik et al. 2003), with extensive soil analyses throughout a forested watershed (c. 340 ha) underlain by three geological substrates. We quantify soil nutrient variation and its association with species distributions. To evaluate the degree to which species-site correlations reflect niche partitioning vs. dispersal limitation, we conducted Monte-Carlo randomizations and Mantel analyses on specific size classes. Finally, we investigate dominance-distribution relationships throughout the watershed to test for an effect of generalist vs. specialist strategies on rank abundance. Dominant tree species at regional scales in western Amazonia are widespread ecological generalists (Ruokolainen & Vormisto 2000; Pitman et al. 2001; Vormisto et al. 2004), but dominance-distribution relationships have not been studied at smaller scales across well-defined nutrient gradients. We focus on the Dipterocarpaceae, a species-rich (c. 520 species) monophyletic family of tropical trees, for several reasons. First, Bornean dipterocarps reproduce synchronously every 3 -7 years (Curran et al. 1999) and share wide-ranging generalist seed predators (Curran & Leighton 2000; Curran & Webb 2000; Lyal & Curran 2000,2003). This controls for climatic variation during seedling establishment, as well as habitat-specificity in seed predators, two potentially confounding influences on species distributions. Secondly, dipterocarp seed is wind or gravity dispersed, minimizing the effect of divergent dispersal syndromes on species distributions (Ashton 1982). Thirdly, all dipterocarps form ectomycorrhizal associations. Fourthly, 60% of the 267 species in Borneo are endemic (Ashton 1982), implying a shared biogeographical and evolutionary history for much of the flora. Finally, a robust quantification of individual dipterocarp species associations with soil nutrients has been performed only at regional scales (Austin et al. 1972; Baillie et al. 1987). Materials and methods study site The study was conducted at the Cabang Panti Research Station (15 km2) in Gunung Palung National Park 159 Beta diversity of Bornean dip tero carps © 2005 The Authors Journal compilation © 2006 British Ecological Society. Journal of Ecology, 94,157-170 (GPNP, 90 000 ha) in West Kalimantan, Indonesia (1°00'-1°20' S, 109°00' 110°25' E). Mean annual rainfall is 4125 + 950 mm (mean + SD 1985-2002), with marked interannual variation in dry season intensity corresponding to El Nino Southern Oscillation (ENSO) events. The western slopes of two interconnected peaks in the centre of GPNP, Mt Palung (1070 m) and Mt Panti (1130 m), form the watershed of the research station. A range of Bornean lowland forest types occurs at GPNP, including peat swamp, freshwater swamp and lowland forests on well-drained mineral soils up to c. 300 m a.s.l. Upland forest is underlain by a variety of parent materials. Alluvial deposits occur along banks of the Air Putih River from 5 to 15 m a.s.l., composing a contiguous area of c. 100 ha in the study site. Upslope (40-140 m a.s.l.) is a texturally complex zone of sedimentary rock, in which localized patches of quartzite occur. Beyond sedimentary rock, forest is underlain by granite to the summit of both peaks. The Air Putih River separates the sedimentary and alluvium substrates by only 8 -15 m. The species composition of woody plants varies among soil types (Cannon & Leighton 2004), but edaphic correlates of floristic variation have not been studied. Throughout Kalimantan, the alluvium, sedimentary and granite substrates present at GPNP underlie 1%, 53% and 7%, respectively, of land area < 500 m a.s.l. (L. M. Curran et al, unpublished data). floristic and edaphic variation Thirty 40x40 m plots (0.16 ha) were established between 8 and 180 m a.s.l. throughout the watershed (c. 340 ha). This plot size was chosen to ensure plots could be treated as 'homogeneous' edaphic units on all substrates. Plots were stratified throughout the watershed on alluvium (n — 10), sedimentary (n — 8) and granite substrates (n — 12). Plot establishment was aided by a network of permanent trails traversing alluvial plain, ridge, valley and slope microhabitats. Plots were positioned at a randomly determined distance (30-150 m) and angle (30-150°) from randomly determined points along trails. Areas with recent gaps (canopy < 5 m) representing > 30% of total plot area were rejected. Distances between plots ranged from 51 to 1747 m; plots were less dispersed on alluvium (median = 362 m) than sedimentary (median — 773 m) or granite habitats (median = 661 m), reflecting the smaller extent of alluvium. Rainfall and maximum-minimum temperature were monitored over 12 months at 10, 40, 60, 180 and 210 m a.s.l. and showed no significant relationship with elevation throughout the study area (G. D. Paoli, unpublished data). soil sampling We quantified texture and nutrient content of surface soils (0-20 cm) in all 30 plots. Six sampling points were randomly stratified in each plot (5-24 m between points), and five soil cores (2-cm diameter) were collected at each point, one at the centre and four at 2 m in cardinal directions. The five cores per point were combined in the field in — 6 samples per plot), air dried, lightly ground and sieved to remove coarse particles. Mineral fractions (< 2 mm) were subsampled (c. 150 g) and transported to the USA for analyses. Total nutrient assays indicate potential long-term nutrient supply, but overestimate short-term availability, whereas labile nutrient assays measure readily available forms that may be poor indicators of long-term availability. Both pools were therefore measured to maximize the power of tests for plant-soil associations. Total C and N were measured by combustion using c. 40 mg samples of finely ground soil oxidized at 700 °C (NC2500, CE Instruments, Milan, Italy). Total P, K, Ca and Mg concentrations were determined by digesting 500 mg of finely ground soil in sequential additions of concentrated HF, HC1 and H202 (Bowman 1990). Total P was measured using the orthophosphate procedure on the Alpkem Rapid Flow AutoAnalyser (RFA 3550; OI Analytical, Austin, Texas, USA); cations were measured using ICP Optical Emission Spectrometry (Optima 3000; Perkin Elmer, Shelton, Connecticut, USA). Exchangeable cations were extracted from 10 g samples using 100 mL of M ammonium acetate (without pH adjustment). Extractable P was quantified using the Olsen method (Olsen et al. 1954) on 3-g samples. Texture was quantified for c. 50 g samples using the hydrometer method (Sheldrick & Wang 1993) and pH on two 10-g samples, using 20 mL of distilled water or 0.01 m CaCl2 after equilibrating for 1 hour. Soil nutrient content was expressed in kg or Mg (megagram) ha~' by multiplying nutrient concentration (ng g-1) by soil bulk density (g cm4) and sampling depth (20 cm). Soil bulk density was estimated using Rawl's (1983) regression method. All analyses used nutrient content, not concentration, because the former better describes availability by incorporating bulk density differences. Nutrient concentrations are presented in Table 1 for comparison with other studies. dipterocarpaceae surveys The Dipterocarpaceae were enumerated in all 30 plots. Each plot was separated into 32 quadrats (5 x 10 m) surveyed for all stems > 1 cm d.b.h. Dipterocarp individuals were measured (d.b.h.) and identified to species [n — 2602 stems) or morphospecies (n — 30 stems). Individuals were identified using published field guides and taxonomic treatments (Ashton 1982; Newman et al. 1996, 1998; Webb & Curran 1996), as well as a field guide based on trees previously field-identified by P. S. Ashton. data analysis Univariate and multivariate procedures were used to analyse soil variation throughout the watershed. 160 G. D. Paoli, L. M. Curran & D. R. Zak j3 O c © 2005 The Authors Journal compilation © 2006 British Ecological Society. Journal of Ecology, 94, 157-170 d is '3 u oo S « £ 1 i= t3 +1 c & 3 o o c c c c vi 0 3c vi a. = a s 2 rr! +r lis: "5" °" ^ o .2 2f o 3 S 'I « U aj U ^ ° 3 a c ■- c o p- 2 m cn o cn o rr rr rr d d d +1 +1 +1 +1 +1 +1 ,— DC DC in cn cn DC c: O d d d # cn in cn cn ^_ o <— •— i— d d d +1 +1 +1 +1 +1 +1 DC in rr cn o rr cn ^_ cn cn DC DC d d d cn # # C o i/S d d d +1 +1 +1 +1 +1 +1 DC cn DC cn cn cn d d r- O 00 cn rr cn d d +1 +1 +1 +1 +1 +1 <— m in cn in in cn cn # in cn rr d i/S d _ m cn i/~: cn <— i— o> +1 +1 +1 +1 +1 +1 in cn DC i— in- DC rr in d in o cn cn cn # in cn rr cn in <— '— <—i +1 +1 +1 +1 +1 +1 cn m d cn oc cn cn in- DC cn +1 +1 +1 DO '—l '— o od ts S a o DO m CN CN CN +1 +1 +1 OC IN IN- m CN IN- m CN * (N <— rr <— ir! in +1 +1 +1 +1 +1 +1 DC IN <— rr DC CN CN d rr IN CN # # ^ rr i— i— i— o d d d d d d +1 +1 +1 +1 +1 +1 DC IN rr m CN CN d d d # # DC tc \c to CN CN d d d +1 +1 +1 +1 +1 +1 o> 00 ^ 1- U r* -2 S 9 C OJ si -3 S Z S 3 c Mixed-model nested two-factor anova (sample nested within plot) was used to compare soil parameters among parent materials. Data were log transformed for total C and P, extractable P, and percentage clay, and In transformed for exchangeable Mg and total K, Ca and Mg. Principal Components Analysis (PCA) was performed on standardized data (z scores) to identify leading dimensions of nutrient variation among substrates. Dipterocarp abundance, richness and diversity (Fisher's a) were compared among substrates using one-way anova or Kruskal-Wallis test if required by the data. Analyses were performed using SPSS (version 11.5). To test species associations with substrate, we used a Monte-Carlo randomization procedure that accounts for non-independence of individuals within a plot due to restricted seed dispersal (Webb & Peart 2000). We permuted the substrates on which all 30 plots occurred and for each species calculated a chi-square deviation statistic for the simulated data set [xSim = 2(N0bs -Nexpec)2/Nexpec]) assuming a distribution of individuals across substrates proportional to total stem number (not number of plots) on each substrate. The procedure was repeated 1000 times to produce a simulated distribution of deviation statistics. Species were deemed significantly positively or negatively associated with a substrate if the observed deviation fell within the highest or lowest 5th percentile of the simulated distribution. This procedure was performed for all individuals (> 1 cm d.b.h.) and separately for juveniles (1-10 cm d.b.h.) and trees (> 10 dm d.b.h.). Stronger or more pervasive non-random distribution in larger size classes is a pattern consistent with juveniles experiencing higher mortality in unsuitable edaphic environments, a probable consequence of niche partitioning. Species associations with soil parameters were tested using Monte-Carlo randomization. First, for all species and soil parameters, the mean observed value of a parameter was calculated as the average across all plots in which the species occurred. Observed values were determined on the basis of presence-absence, rather than weighted average, to avoid biases caused by non-independence of individuals in a plot. Then, from the pool of 30 plots, random draws were made equal to the number of plots in which a species was present, and a simulated value for the parameter was computed as the average of these draws. This process was repeated 1000 times to produce a simulated distribution with which to compare observed values of each species. Associations with a parameter were deemed significant if observed values fell with the lowest or highest 5% of the simulated distribution. Finally, Mantel analyses were performed on distance matrices to compare the strength of correlations between floristic composition and geographical vs. environmental distance among plots (Legendre & Legendre 1998). Floristic dissimilarity was quantified using the Stein-haus dissimilarity index. Canonical correspondence analysis (CCA, ter Braak & Smilauer 2002) was used to identify soil parameters for inclusion in Mantel analyses 161 based on significant association with floristic patterns. Beta diversity of All soil parameters were included in a forward-selection Bornean model that retained extractable P, exchangeable Mg dipterocarps and total K and explained 28% of floristic variation (P < 0.0001; results available upon request from GDP). We thus computed four matrices to describe edaphic distance: one for each nutrient and a Euclidean combination using standardized values (z scores). We performed full Mantel analyses to quantify overall correlations among matrices and partial Mantel to control for cross-correlation between geographical and edaphic distance. Version 1.0 of zt software was used to compute Mantel correlations and test significance of the r statistic using permutation (Bonnet & Van de Peer 2002). Mantel analyses were first performed on all individuals (> 1 cm d.b.h.). If both niche partitioning and dispersal limitation influence dipterocarp community structure, then floristic divergence should be significantly related to both edaphic and geographical distances. We then performed separate analyses for juveniles (1-10 cm d.b.h.) and trees (> 10 cm d.b.h.) to test for changes with size class in the relative strength of floristic correlations with geographical and edaphic distance. If seed dispersal influences community structure by restricting local range expansions from one generation to the next, then differences among juveniles should be more strongly related to geographical distance than to edaphic factors. If, consistent with niche theory, juveniles of species poorly adapted to local conditions experience relatively high mortality, then floristic correlations with edaphic factors should be stronger for trees than for juveniles. Results soil variation Soil nutrients varied among substrates, and were generally highest in alluvium, intermediate in sedimentary soils and lowest in granitic soils. Total C, N and P, extractable P and exchangeable K were significantly higher in the alluvium than other substrates; total K and Ca, and exchangeable Ca, showed similar trends (Table 1). Averaging across all nutrients, the nutrient content of alluvium soils was nearly twice that of granite soils (Table 1). In contrast to soil nutrients, soil texture was similar across substrates. Only percentage silt varied significantly, and was highest on sedimentary soils (Table 2). Soil pH was significantly lower in the alluvium than other substrates (Table 2). In PCA of all data combined, axes 1 and 2 described 47% of variation throughout the watershed (Fig. 1). Alluvium and granite plots segregated along axis 1, representing a gradient of increasing P, N and exchangeable cations, and of decreasing pH; sedimentary plots were widely distributed across this axis (Fig. la). abundance and diversity of the dipterocarpaceae A total of 2632 dipterocarp individuals (> 1 cm d.b.h.) were enumerated, representing 31 species in five genera: Shorea, Dipterocarpus, Hopea, Vatica and Anisoptera (see Table S1 in Supplementary Material). Twenty-four taxa were identified to species; seven remain unidentified morphospecies, of which five were singletons. The genus Shorea was the most species-rich and dominant numerically, with 23 species and 1389 stems (52.8% of individuals). The genus Anisoptera was represented by two individuals of a single species (Table SI). The dipterocarp community (> 1 cm d.b.h.) was dominated numerically by Dipterocarpus sublamellatus (Fig. 2a), which comprised > 40% of all dipterocarp stems and was present in 20 of 30 plots (Table SI). Dominance by D. sublamellatus was pervasive in all size classes (Table SI). In contrast, most other dipterocarp species were rare and sparsely distributed, 15 of 31 species were present in < 3 plots. Across species, occurrence was positively correlated with log mean density throughout the watershed for stems > 1, > 10 or > 30 cm d.b.h. (for all categories Pearson correlation R > 0.88, P < 0.001; Fig. 2b). Dipterocarp density was similar across substrates, but species richness and diversity were significantly lower in the alluvium (Table 3). Summing across plots within a substrate, only 12 species were present on alluvium, 22 on sedimentary soils and 28 on the granite; lower richness in the alluvium was consistent across size categories (Table 3). Fisher's a index of diversity showed a similar pattern (Table 3). habitat distribution across substrates Low dipterocarp species richness in the alluvium reflected a nested distribution of species across substrates Table 2 Texture and pH properties of surface soils (0-20 cm) in lowland rain forest on three parent materials at Gunung Palung National Park, Indonesia. Superscripts indicate significant differences between substrates using SchefFe's test following two-factor nested anova. Data are mean ± 1 SE. *P < 0.05 Soil texture Soil pH © 2005 The Authors Journal compilation Parent material % Sand % Silt % Clay Water pH CaCl2 pH © 2006 British Ecological Society. Journal of Ecology, 94,157-170 Alluvium 65.1 ±0.8 22.7±0.8a* 12.2 ±0.7 4.20 ± 0.02a* 3.54±0.02a* Sedimentary 61.3 ±0.9 29.6±0.8b 9.2 ±0.8 4.49 ± 0.03b 3.70±0.02b Granite 67.7 ±0.8 19.8 ± 0.7C 12.5 ±0.6 4.51 ±0.02b 3.71 ±0.02b 162 G. D. Paoli, L. M. Curran & D. R. Zak (a) Factor scores (b) Factor loadings š o o Alluvium • Sedimentary • Granite -2-10 1 2 Axisl (26.5%) Fig. 1 Principal components analysis of soil nutrients in lowland rain forest on three parent materials at Gunung Palung National Park, Indonesia. PCA was performed on the mean of six surface soil samples (0-20 cm) in each of 30 (0.16 ha) plots analysed separately for 12 parameters including pH, and total and exchangeable nutrients. Factor scores for each plot (a) and axis loadings (b) are presented for PCA axes 1 and 2. Percentage of total variation explained by each axis is shown in parentheses. Total nutrients are denoted by subscript 'tot'. (b) (c) Rank Number of plots Stems ha-1 in alluvium Fig. 2 Demographic attributes of the lowland dipterocarp community at Gunung Palung National Park, Indonesia, (a) Rank vs. log density, (b) Plot occupancy vs. log density plotted separately for nested size classes; curves were fit using least squares linear regression and illustrate the positive correlation between occupancy and log density for all size categories (P < 0.001 and R > 0.885 for all categories), (c) Density on alluvium vs. density on granite substrates (subdominants > 8 individuals ha~' overall); dashed line indicates equal density on both substrates. The widespread dominant Dipterocarpus sublamellatus is labelled 'D sub'. © 2005 The Authors Journal compilation © 2006 British Ecological Society, Journal of Ecology, 94, 157-170 (Table S2). Of the 28 species present on granite, 20 also occurred on sedimentary soils, and these represented > 90% of the sedimentary flora (i.e. 20 of 22 species). Ten species were common to sedimentary and alluvial sites, comprising 83% of the alluvium flora (10 of 12 species). Despite the close proximity and spatial contiguity of these habitats, 11 species widespread on sedimentary and granite substrates were absent from the alluvium (Table S2). Eighteen of 22 dipterocarp species (> 1 cm d.b.h.) tested were significantly associated with at least one substrate (Fig. 3). The taxa Shorea quadrinervis, S. pinanga, Hopea dyeri, Dipterocarpus stellatus and Vatica sp. 1 were consistently encountered on granite soils, but absent from the alluvium, whereas S. johorensis, S. macrophylla and S. parvifolia were frequent on alluvium soils, but rare on the granite (Table S2). These associations were confirmed in Monte-Carlo randomization: the former group was associated positively with granite and negatively with alluvium, whereas the latter was associated positively with alluvium and negatively with granite (Fig. 3). The taxa S. hopeifolia and S. pau-ciflora were positively associated with sedimentary soils (Fig. 3). A final species group was negatively associated with one substrate, but indifferent to others: S. gibbosa, S. parvistipulata and Vatica sp. 3 were negatively associated with granite, S. longisperma was negatively associated with sedimentary soil, and S. crassa, S. laevis, S. gratissima and S. lamellata were negatively associated with the alluvium (Fig. 3). In contrast to this 163 Beta diversity of Bornean dip tero carps © 2005 The Authors Journal compilation © 2006 British Ecological Society. Journal of Ecology, 94,157-170 o. •- Fl & s II c c c £ ll Z 3 ■3 2 SO t< 6. « 5 u s3 3 5 S3 ^ o Al 12 3 O £ t3 — a ai -d — d d d +1 +1 +1 DC m d d cn o cn c i/~: m d d d +1 +1 +1 cn o m cn <—i cn cn cn cn DO cn cn # C d d +1 +1 +1 DC l> c-^ DC cn >— cn DO +1 +1 +1 DC in- i- i i/~: i/~: in 5 3 = < «8 O general pattern of non-random distribution, the dominant species D. sublamellatus was indifferent to substrate. We tested for stronger patterns of habitat association among larger size classes by comparing habitat associations of juveniles (1-10 cm d.b.h.) and established trees(> 10 cm d.b.h.). Of the 16 taxa sufficiently abundant in both size classes, five were randomly distributed as juveniles, but significantly associated with substrate as established trees (P < 0.05), and three other taxa displayed the same trend (P < 0.1, Table 4). Two species were significantly associated with a single substrate at both size classes and five species were randomly distributed at both stages. Only one species (Shorea mac-rophylla) was distributed non-randomly as juveniles but randomly as trees. This finding may reflect reduced power due to smaller sample size of established trees, rather than a broadening of habitat distribution with size class, because, for this species, the proportion of individuals across substrates did not differ between juveniles and trees (proportions on alluvium, sedimentary and granite substrates were 0.92, 0.06 and 0.02 for juveniles and 0.75, 0.07 and 0.18 for trees). Excluding the widespread dominant D. sublamellatus, species with high mesoscale population density (> 8 individuals ha~') showed a clear bias towards nutrient-poor granite or nutrient-rich alluvium soils (Fig. 2c). species distribution in relation to soil nutrients Eighteen of 22 species tested were significantly associated with one or more nutrient parameters (Table 5). Thirteen species were either positively (three species) or negatively (10 species) associated with extractable soil P (Table 5). Seven species were significantly associated with extractable Mg, six species with extractable Ca and four species with total K (Table 5). Most species were jointly associated with soil P and one or more cations. For example, S. quadrinervis and S. pauciflora were associated with soils low in P, K, Mg and Ca, while S. johorensis and S. gibbosa were associated with soils high in P, Ca and Mg (Table 5). Exceptions include Shorea gratissima and S. pinanga, which were negatively associated with P, but indifferent to base cations, and S. parvistipulata and S. longisperma, which were positively associated with base cations, but indifferent to P (Table 5). Most species associated with soil nutrients (Table 5) were also associated with one or more substrates (Fig. 3). However, the widespread dominant D. sublamellatus and the subdominant S. faguetiana were randomly distributed across substrates, but significantly associated with soils low in P, Mg and total C. beta diversity and geographical vs. environmental distance In Mantel analyses of all stems (> 1 cm d.b.h.), floristic divergence among plots was significantly related to Sedimentary Fig. 3 Venn diagram depicting substrate associations for 22 tree species in the Dipterocarpaceae in lowland tropical forest at Gunung Palung National Park, Indonesia. Significance of associations was assessed using Monte-Carlo randomization. Positive associations with a substrate are shown by species name in non-overlapping regions of the diagram, and negative associations at the intersection of the other two substrates. Taxa in the diagram centre were randomly distributed across substrates. Taxa positively associated with one substrate and negatively associated with others are denoted by superscript (alluvium (#), sedimentary (§) or granite (f)). Table 4 Comparison of substrate association between juveniles (1-10 cm d.b.h.) and established trees (> 10 cm d.b.h.) in the lowland Dipterocarpaceae at Gunung Palung National Park, Indonesia. Significant association (P < 0.05) was tested using Monte-Carlo randomization. Species present in at least three plots as both juveniles and trees (n = 16) were included in the analysis Juveniles (< 10 cm d.b.h.) Established trees (> 10 cm d.b.h.) Random Non-random Random Non-random Dipterocarpus stellatus Shorea crassa Shorea leprosula Shorea laevis Shorea gratissima Shorea macrophylla Dipterocarpus sublamellatus Shorea parvifolia Shorea johorensis Shorea faguetiana Shorea hopeifolia Shorea quadrinervis Shorea parvistipulata\ Shorea pinanga\ Hopea ferruginea\ Shorea pauciflora fNon-randomly distributed at a = 0.1. © 2005 The Authors Journal compilation © 2006 British Ecological Society, Journal of Ecology, 94, 157-170 geographical distance and to all edaphic parameters except total K (Table 6). However, separate analyses of juveniles (1-10 cm) and trees (> 10 cm d.b.h.) suggested that the effects of geographical and edaphic distance on floristic composition varied with size class (Table 6). In simple Mantel analyses, floristic divergence among juvenile dipterocarps was more strongly related to geographical distance than to edaphic factors, but the opposite was observed for established trees (Table 6). Differences with size class were even more pronounced in partial Mantel. Controlling for cross-correlation between geographical and edaphic distance, floristic differences among juveniles were weakly or unrelated to edaphic factors, but were strongly related to geographical distance. Among established trees, the opposite pattern was observed for the Euclidean index and for exchangeable Mg. Discussion High mesoscale beta diversity in the Dipterocarpaceae at GPNP appears to reflect the joint influences of interspecific edaphic partitioning and, to a lesser degree, dispersal limitation. The majority of common diptero-carp taxa were significantly associated with at least one parent material, species distributions were related to soil nutrients, especially P, and, for many species, habitat associations with substrate were stronger for established trees than juveniles. This suggests that substrate heterogeneity promotes the coexistence of a diverse assemblage of lowland dipterocarp species by allowing spatial partitioning of habitat throughout the watershed. These findings have important implications for understanding mechanisms underlying dipterocarp community structure in Borneo and highlight the Table 5 Species associations with surface soil (0-20 cm) nutrient content and soil texture in lowland forest at Gunung Palung National Park, Indonesia. Significance of association was tested using a randomization procedure, wherein species with observed means in the highest or lowest 5th percentile of the simulated distribution were deemed significantly positively (hi) or negatively (lo) associated with that parameter. *P< 0.05, **P < 0.01, ***p < 0.005. ****_• < 0.001 Species Total fraction N K Exchangeable fraction - Olsen - pH pH Ca Mg P K Ca Mg H20 CaCl2 % Sand % Silt % Clay D. sublamellatus iS. crassa S. johorensis iS. laevis S. gibbosa H. ferruginea S. faguetiana S. gratissima S. macrophylla S. parvistipulata S. pauciflora S. leprosula S. pinanga S. quadrinervis D. stellatus H. dyeri iS. hopeifolia iS. lamellata S. longisperma S. parvifolia Vatica sp. 1 Vatica sp. 3 lo* hi* lo* lo* lo* lo** lo* hi** lo** lo** lo* lo* hi* hi* lo* lo* lo* lo* lo* lo* lo* lo* lo* hi* lo! lo* lo* lo* hi** 1* hi** lo** lo* lo** lo** lo** lo** lo* lo** lo* hi* hi* lo* hi* lo** lo* lo** lo* lo* hi* lo* hi* © 2005 The Authors Journal compilation © 2006 British Ecological Society. Journal of Ecology, 94,157-170 Table 6 Mantel tests of correlation between floristic variation, soil nutrients and geographical distance among 30 (0.16 ha) plots in lowland tropical forest at Gunung Palung National Park, Indonesia. Separate analyses were performed on three size categories: > 1 cm d.b.h., 1-10 cm d.b.h. and > 10 cm d.b.h. *P < 0.05, **P < 0.01, ***P < 0.001, ****p < 0.0001 All stems - 1 cm d.b.h. Stems 1-10 cm d.b.h. Stems = 10 cm d.b.h. Soil factor Floristics Soil Distance Floristics Soil Distance Floristics Soil Distance (a) Euclidean index Floristics -Soil 0.175** Distance 0.275**** 0.280**** 0.348**** 0.366*** 0.113°s 0 309***** 0.231** 0.365**** 0.366**** 0.214** 0.178** 0.289*** 0.264**** 0.366**** (b) Extractable P Floristics -Soil 0.193* Distance 0.262*** 0.302*** 0.348**** 0 390**** 0.145* 0 294**** 0.267*** 0.365**** 0 390**** 0.131* 0.198** 0.302*** 0.264**** 0 390**** (c) Exchangeable Mg Floristics -Soil 0.134°s Distance 0.310**** 0.211* 0.348**** 0.260* 0.098"s 0 334***** 0.183* 0.365**** 0.260* 0.221** 0.207* 0.275** 0.264**** 0.260* (d) Total K Floristics Soil Distance 0.004 "s 0.348**** 0.017"s 0.348**** 0.039"s 0.000"s 0.364**** 0.014"s 0.365**** 0.039"s - 0.003™ 0.264*** 0.007 "s 0.264**** 0.039"s Mantel statistics above the diagonal are simple correlations; below the diagonal are partial Mantel statistics after controlling for cross-correlation with the third factor. Euclidean index is the sum of absolute differences in extractable P, and exchangeable Mg and K content between all possible pairs of plots after data were standardized using z scores; these nutrients were retained in canonical correspondence analysis of the association between nutrients and dipterocarp species abundances (available on request from lead author). effects of substrate heterogeneity on tropical woody plant diversity. species distributions in relation to soil factors Across size classes and substrates, the dipterocarp community was dominated by the emergent tree Dipterocarpus sublamellatus, which was indifferent to substrate. The broad habitat distribution of this species, however, was exceptional. Ten of the 22 dipterocarp species tested were both positively associated with at least one parent material and negatively associated with another. Eight additional taxa were negatively associated with one substrate, but indifferent to others, increasing the number of non-randomly distributed species to 82% of those tested. Substrate heterogeneity thus influences the distribution and abundance of most common dipterocarp species in upland soils at GPNP. Plant associations with parent material have been documented at spatial scales ranging from c. 50 ha to > 40 000 km2 in tropical palms, ferns, understorey and overstorey trees (Gartlan et al. 1986; Clark et al. 1999; Svenning 1999; Debski et al. 2002; Duque et al. 2002; Potts et al. 2002; Phillips et al. 2003; Tuomisto et al. 2003). However, the inability to infer process from pattern remains a limitation of the correlative approach. We addressed this problem by testing for indirect evidence of differential juvenile mortality in unsuitable habitats by comparing patterns of habitat association between juveniles and established trees > 10 cm d.b.h. Five of 14 species tested were randomly distributed as juveniles, but were non-random as established trees (P < 0.05), and three additional taxa displayed the same trend (P < 0.1), despite limited power resulting from small sample sizes for trees. It is possible for such a pattern to result from ecological processes unrelated to edaphic specialization: generalist herbivores distributed across a resource gradient could, for example, confer advantage to well-defended species on poor soils, but fast-growing species on rich soils, if a growth-defence trade-off exists (Fine et al. 2004). Nevertheless, we believe that stronger patterns of habitat association for established trees than juveniles suggest that dipterocarp beta diversity at our site is driven, at least in part, by interspecific edaphic partitioning. Webb & Peart (2000) reached a similar conclusion at a smaller spatial scale, based on stronger associations with topographic position for adult trees than for seedlings on the granite substrate at GPNP. Dipterocarp species distributions were unrelated to soil texture or pH, but strongly related to soil nutrients. Seventeen of 22 species tested were significantly associated with one or more nutrients, in particular extract-able P (13 species), total P (nine species), and total or exchangeable forms of Mg and Ca (eight species each; Table 5). If such correlation reflects causation, then a hierarchy of control appears to operate, with soil P having the strongest, most pervasive influence (14 species). A similar conclusion was reached by Hall et al. (2004) in Cameroon, but unlike in their study, we found little evidence for finer-scale nutrient partitioning within high or low P soils; most dipterocarp species were associated with soils high or low in both P and cations. Dipterocarp associations with nutrients at GPNP thus imply a generalized adaptation to rich or poor soils, consistent with findings across northern Borneo by Potts et al. (2002). Future work will focus on mechanisms of interspecific P partitioning (Palmiotto et al. 2004) that potentially explain this pattern, as well as the possibility that substrate associations may, for some species, reflect adaptation to periodic variation in drought stress (G. D. Paoli and L. M. Curran, unpublished data). Baillie et al. (1987) studied nutrient correlates of regional beta diversity in northern Borneo and found that tree species distributions were more frequently related to total nutrient pools ('reserve contents', Baillie et al. 1987, p. 206) than to labile nutrient forms. They concluded that total nutrients were a better indicator of relative long-term nutrient availability to plants. Our findings suggest a more complicated scenario. We measured labile and total forms of P, K, Ca and Mg, and found that, for three out of four elements, labile forms were more frequently, and often more strongly, associated with species distributions than total forms. Thirteen species were associated with extractable P vs. nine for total P; six vs. two species with exchangeable and total Ca; seven vs. one species for exchangeable and total Mg; and zero vs. four species for exchangeable and total K (Table 5). Most importantly, in 28 of 34 significant associations with P, K, Mg or Ca, species were associated with total or labile forms, not both; only P showed concordance between forms (Table 5). Our results demonstrate the utility of well-replicated labile nutrient assays for detecting plant-soil associations, and the need to quantify both nutrient pools, because they provide complementary information. Plant-soil associations in this study also reveal a striking relationship between phylogeny and habitat distribution. The distributions of several species pairs within the same section or subsection of Shorea (i.e. putative close relatives) showed opposite relationships to soil factors. Thus (i) S. macrophylla and S. pinanga (section Pachycarpae), (ii) S. leprosula and S. quadrinervis (subsection Mutica), (iii) S. johorensis and S. pau-ciflora (subection Brachypterae) and (iv) S. gibbosa and S. faguetiana (subsection Richetioides) occupied different edaphic niches (Fig. 3, Table 5). Congeneric rain forest pairs have been shown to partition gradients of climate (Hoffman & Franco 2003), inundation (Rogstad 1990), elevation (Gunatilleke et al. 1997; Yamada et al. 2000) and white-sand vs. clay soils in western Amazonia (Fine et al. 2004), but to our knowledge have not been documented in such large numbers within a single lineage of plants across a well-defined nutrient gradient in a single watershed. This suite of related dipterocarp species with divergent habitat associations offers a unique opportunity to study ecological 167 Beta diversity of Bornean dipterocarps £> 2005 The Authors Journal compilation £> 2006 British Ecological Society. Journal of Ecology, 94,157-170 correlates of edaphic specialization and to assess its potential role in the evolution of a species-rich series of sympatric rain forest trees (Paoli 2004). dispersal vs. niche assembly The neutral model of Hubbell (2001) posits that beta diversity arises from random, spatially restricted seed dispersal and the local dynamics of extinction and speciation. Dipterocarp beta diversity at GPNP is not explained by local speciation, because most species present are widespread throughout Borneo (Ashton 1982), but dispersal limitation across substrates may be important, given that upland forests at GPNP may be recent post- or late-Pleistocene assemblages (Gathorne-Hardy et al. 2002). We used Mantel correlation analysis to test for evidence that seed dispersal limitation influences dipterocarp beta diversity by comparing floristic associations with edaphic vs. geographical distance for all stems > 1 cm d.b.h. and at different size stages. For all stems > 1 cm, floristic variation among plots was related more to geographical distance than to edaphic factors, even when cross-correlation between them was controlled (Table 6). However, interpreting this pattern is complex, because the relative strength of edaphic vs. geographical influences on floristic patterns appears to vary with size class. Floristic similarity of juveniles (1-10 cm d.b.h.) was more strongly related to geographical distance than to edaphic factors, whereas the reverse held for established trees (> 10 cm d.b.h.). This implies that the composition of juvenile dipterocarp assemblages is strongly affected by dispersal processes, but that the effects of edaphic partitioning increase through subsequent size classes. This pattern is consistent with the results of habitat association tests on different size classes (Table 4), and would be expected if spatial patterns of dipterocarp establishment were determined largely by seed dispersal and escape from predation (Curran & Webb 2000), but then juveniles experienced higher mortality in suboptimal habitats due to tradeoffs related to edaphic specialization. In this scenario, seed dispersal limits the spatial distribution and thus range of edaphic conditions encountered by new individuals, after which niche partitioning determines the subset of conditions suitable for long-term growth and survival. Future comparisons of intraspecific growth rate and mortality across the substrate gradient will provide a direct test of this hypothesis. patterns of species richness Dipterocarp stem densities were similar across substrates, but species richness and diversity were highest on the nutrient-poor granite (28 species), intermediate on sedimentary soils (20 species) and lowest on nutrient-rich alluvium (12 species). Lower dipterocarp richness on alluvium reflects a general pattern of low woody plant diversity on this substrate (Cannon & Leighton 2004). Low dipterocarp richness on alluvium reflected the local absence of species common on sedimentary and granite soils (hereafter 'hill substrates'). Only 11 of 30 species present on hill substrates (36.3%) occurred in the alluvium, whereas 11 of 12 species present on the alluvium (93.5%) were found on hill substrates (Table S2). The absence of hill species in the alluvium is more striking when just habitat specialists are compared. Of the 11 dipterocarp species positively associated with hill substrates, only one Hopea ferruginea individual (d.b.h. = 17 cm) was present on the alluvium, whereas all three species positively associated with alluvium were common on hill substrates (Table S2). The asymmetric distribution of habitat specialists across the gradient is surprising, given the close proximity of alluvium and hill substrates and the expectation that, because dipterocarps are dispersed by gravity or wind, seed dispersal downslope to the alluvium should be more frequent than the converse. The absence of hill dipterocarps from the alluvium is unlikely to be explained by disproportionate seed predation of hill species in the alluvium, because the most important dipterocarp seed predators at GPNP are mobile generalists that prey on seeds of all dipterocarp species, including the bearded pig Sus barbatus (Curran & Leighton 2000) and weevils in the Curculio-nidae (Lyal & Curran 2000, 2003). Nor does it seem likely to result from lower understorey light in the alluvium, because biomass, basal area and leaf area index are similar across substrates (Paoli et al. 2005) and rates of gap formation are higher in the alluvium than the granite (L. M. Curran, unpublished data). Rather, itmay be explained by competitive exclusion of hill species from the alluvium, where tree growth is twofold higher than in the granite (Paoli et al. 2005), and dominant species have leaf traits that confer higher intrinsic growth potential (Paoli 2004). More rapid competitive exclusion in the alluvium would lead to faster elimination of hill species and, thus, lower dipterocarp richness. This pattern is consistent with ecological theory (Huston 1979; Tilman 1982) and empirical studies documenting declines in species richness on productive sites (Huston 1993; Rajaniemi 2003), further suggesting a deterministic explanation. dominance-distribution relationships throughout the watershed Nearly 50% of all dipterocarp stems at GPNP are from a single species (Dipterocarpus sublamellatus). Yet, an assemblage of 11 subdominant dipterocarp species also attained high mesoscale densities throughout the watershed (> 8 individuals ha~'). Nine of these taxa were positively associated with one substrate and negatively associated with others, indicating that high mesoscale densities were attained through high local densities on one substrate (Figs 2 and 3). This suggests that subdominant dipterocarp species sacrifice broad habitat distribution for high local densities under restricted edaphic conditions. That dominant rain forest trees are habitat generalists, whereas subdominant taxa are habitat specialists, has been documented at local scales within communities (Valencia et al. 2004) and among communities at meso- and regional scales (see data in ter Steege et al. 1993; Pitman et al. 2001; Phillips et al. 2003). Such a dominance-distribution relationship may be a general property of lowland tropical communities, but more data are required. Dominance of the dipterocarp community by the habitat generalist Dipterocarpus sublamellatus is consistent with regional findings from western Amazonia, where separate communities are often dominated by a shared set of so-called 'oligarchs' (Pitman et al. 2001). In the present study, we identified only one dominant tree species, Dipterocarpus sublamellatus, but the non-dipterocarp Strombosia ceylanica (Olacaceae) is also locally abundant and widespread across the gradient (Cannon & Leighton 2004), suggesting dominance by an oligarchy of generalists throughout the watershed. The mechanisms that permit broad habitat distribution and high local densities are not well understood. Regeneration of D. sublamellatus is common throughout all gap phases, suggesting that light availability is not a prominent factor. It has been observed in Amazonia that dominant taxa, especially palms, tend to be tall trees with broad geographical ranges (Ruokolainen & Vormisto 2000; Pitman et al. 2001). Consistent with this pattern, Dipterocarpus sublamellatus is an emergent canopy tree with a broad geographical range spanning Peninsular Malaysia, Sumatra and Borneo (Ashton 1982), but many rare dipterocarps at GPNP (e.g. S. lamellata) reach larger maximum size and have broader geographical ranges in Asia. Thus, tree size and geographical range alone do not predict mesoscale distribution or abundance. Dipterocarpus sublamellatus may achieve widespread, locally dense populations through exceptional phenotypic plasticity, drought resistance, protection from herbivores and seedling pathogens, or some combination of these factors. Comparative studies of D. sublamellatus and edaphic specialists at GPNP will help identify plant traits underlying changes in relative abundance across local habitat gradients. Pervasive dipterocarp associations with soil nutrients, especially P, in this study support the hypothesis that niche processes structure upland tropical tree communities. Expanding the spatial scale of our study to include other watersheds within the park, or nearby mountain systems with similar substrates, will provide a robust test of generality for the habitat associations documented here. Substantial floristic variations unrelated to soil factors measured in this study, however, were significantly correlated with geographical distance between plots, suggesting a role for dispersal limitation as well. We hypothesize that effects of dispersal limitation are likely to operate more strongly within substrate types than between them; greater plot replication within each substrate will be required to test this. Under this scenario, interspecific trait differences cause habitat partitioning by dipterocarp species adapted to rich or poor soils, whereas seed dispersal influences the distribution and relative abundance of species within substrates, and promotes a degree of floristic mixing among them. Understanding the interaction between dispersal and niche processes, and their dependency on spatial scale and ecological conditions, represents an area of important future research at GPNP and other tropical sites. Acknowledgements We are grateful to the Indonesian Institute of Sciences (LIPI) and the Department of Forest Protection and Nature Conservation (PHKA) for granting permission to conduct research in Indonesia. We thank our Indonesian sponsor, the Center for Research and Development in Biology, for logistical support, and the students and faculty of Universitas Tanjungpura, for serving as our research counterparts. Farizal, Hon, Tang and Morni provided vital assistance in the field, and A. Budiman, Sugarjito, Ibu Ina, N. Paliama and M. Sinaga provided logistical support. We thank J. Harting for computing distances between plots. We thank D. Burslem, L. De Mattia, P. Fine, D. Goldberg, A. Gorog, P. Hall, J. Vandermeer, C. Webb, T. Young and anonymous reviewers for critical advice. GD.P received financial support from the Fulbright Indonesia Programme and the University of Michigan, and L.M.C. from the NASA Earth Science Programme (NAG 511335 & 511161), the University of Michigan and the Yale School of Forestry and Environmental Studies. References Ashton, PS. (1976) Mixed dipterocarp forest and its variation with habitat in the Malayan lowlands: a re-evaluation at Pasoh. Malaysian Forester, 39, 56-72. Ashton, PS. (1982) Dipterocarpaceae. Flora Malaesiana, I (9), 237-552. Austin, M.P, Ashton, PS. & Greig-Smith, P. (1972) The application of quantitative methods to survey vegetation. III. A re-examination of rain forest data from Brunei. Journal of Ecology, 60, 305-324. Baillie, I.C., Ashton, PS., Court, M.N., Anderson, J.A.R., Fitzpatrick, E.A. & Tinsley, J. (1987) Site characteristics and the distribution of tree species in mixed dipterocarp forests on tertiary sediments in Central Sarawak, Malaysia. Journal of Tropical Ecology, 3, 201-220. Bonnet, E. & Van de Peer, Y. (2002) zt: a software tool for simple and partial Mantel tests. Journal of Statistical Software, 7(10), 1-12. Bowman, R. (1990) A rapid method to determine total phosphorus in soils. Journal of the Soil Science Society of America, 52, 1301-1304. ter Braak, C.J.F. & Smilauer, P. (2002) canoco Reference Manual and CanoDraw for Windows User's Guide, Software for Canonical Community Ordination Version 4.5. Microcomputer Power, Ithaca. Cannon, C. & Leighton, M. (2004) Tree species distributions across five habitats in a Bornean rainforest. Journal of Vegetation Science, 15, 257-266. Chave, J. & Leigh, E.G. (2002) A spatially explicit neutral 169 Beta diversity of Bornean dip tero carps © 2005 The Authors Journal compilation © 2006 British Ecological Society. Journal of Ecology, 94,157-170 model of (3-diversity in tropical forests. Theoretical Population Biology, 62, 153-168. Clark, D.B., Clark, DA. & Read, J.M. (1998) Edaphic variation and the mesoscale distribution of tree species in a neotropical rain forest. Journal of Ecology, 86, 101-112. Clark, D., Clark, D. & Read, J.M. (1999) Edaphic factors and the landscape-scale distribution of rain forest trees. Ecology, 80, 2662-2675. Condit, R., Pitman, N., Leigh, E.G., Chave, J, Terborgh, J. Foster, R. et al. (2002) Beta-diversity in tropical forest trees. Science, 295, 666-669. Curran, L.M., Caniago, I., Paoli, G.D., Astiani, D. & Leighton, M. (1999) Impact of El Nino and logging on canopy tree recruitment in Borneo. Science, 286, 2182- 2188. Curran, L.M. & Leighton, M. (2000) Vertebrate responses to spatio-temporal variation in seed production by mast-fruiting Bornean Dipterocarpaceae. Ecological Monographs, 70. 121-150. Curran, L.M. & Webb, CO. (2000) Spatio-temporal scale of seed predation in mast-fruiting Dipterocarpaceae: experimental studies of regional seed availability. Ecological Monographs, 70, 151-171. Debski, I., Burslem, D.F.R.P, Palmiotto, PA., Lafrankie, J.V., Lee, H.S. & Manokaran, N. (2002) Habitat preferences of Aporosa in two Malaysian forests: implications for abundance and coexistence. Ecology, 83, 2005-2018. Duque, A., Sanchez, M., Cavalier, J. & Duivenvoorden, J. (2002) Different floristic patterns of woody understorey and canopy plants in Colombian Amazonia. Journal of Tropical Ecology, 18, 499-525. Fine, P.V.A., Mesones, I. & Coley, PD. (2004) Herbivores promote habitat specialization by trees in Amazonian forests. Science, 305, 663-665. Gartlan, J.S., Newbery, D.M., Thomas, D.W & Waterman. PG (1986) The influence of topography and soil phosphorus on the vegetation of Korup Forest Reserve, Cameroun. Vegetatio, 65, 131-148. Gathorne-Hardy, F.J., Syaukani, Davies, R.G, Eggleton, F. & Jones, D.T. (2002) Quaternary rain forest refugia in south-east Asia: using termites (Isoptera) as indicators. Biology Journal of the Linnaean Society, 75, 453-466. Gunatilleke, C.V.S., Gunatilleke, I.A.U.N., Perera, G.A.D., Burslem, D.F.R.P, Ashton, P.M.S. & Ashton, PS. (1997) Response to nutrient addition among seedlings of eight closely related species of Shorea in Sri Lanka. Journal of Ecology, 85, 301-311. Hall, J.S., McKenna, J.J., Ashton, P.M. & Gregoire, T.G. (2004) Habitat characterizations underestimate the role of edaphic factors controlling the distribution of Entandro-phragma. Ecology, 85, 2171-2183. Hoffman, WA. & Franco, A.C (2003) Comparative growth analysis of tropical forest and savanna woody plants using phylogenetically independent contrasts. Journal of Ecology, 91, 475-484. Hubbell, S. (2001) The Unified Neutral Theory of Biodiversity and Biogeography. Princeton University Press, New Jersey. Huston, M.A. (1979) A general hypothesis of species diversity. American Naturalist, 113, 81-101. Huston, M.A. (1993) Biological diversity, soils and economics. Science, 262, 1676-1680. Legendre, P. & Legendre, L. (1998) Numerical Ecology. Elsevier, Amsterdam. Lieberman, D., Lieberman, M., Hartshorn, GS. &Peralta, R. (1985) Small-scale altitudinal variation in lowland tropical forest vegetation. Journal of Ecology, 73, 505-516. Lyal, C.H.C. & Curran, L.M. (2000) Seed-feeding beetles of the weevil tribe Mecysolobini (Insecta: Coleptera: Curcu-lionidae) developing in seeds of trees in the Dipterocarpaceae. Journal of Natural History, 34, 1743-1847. Lyal, C.H.C. & Curran, L.M. (2003) More than black and white: a new genus nanophyine seed predators of Dipterocarpaceae and a review of Meregallia Alonso-Zarazaga (Coleoptera: Curculionidae: Nanophyidae) in Southeast Asia. Journal of Natural History, 37, 57-105. Newbery, D.M., Gartlan, J.S., McKey, D.B. & Waterman, EG. (1986) The influence of drainage and soil phosphorus on the vegetation of Douala-Edea Forest Reserve, Cameroun. Vegetatio, 65, 149-162. Newbery, D.M. & Froctor, J. (1984) Ecological studies in four contrasting rain forests in Gunung Mulu National Park. Sarawak. IV. Associations between tree distribution and soil factors. Journal of Ecology, 12, 475-493. Newbery, D.M., Renshaw, E. & Brunig, E.F (1986) Spatial patterns of trees in kerangas forest, Sarawak. Vegetatio, 65. 77-89. Newman, M.F., Burgess, P.M. & Whitmore, T.C. (1996) Borneo Island Light Hardwoods. Royal Botanic Garden Edinburgh and Centre for International Forestry Research. Edinburgh. Newman, M.F, Burgess, P.M. & Whitmore, T.C. (1998) Borneo Island Medium and Heavy Hardwoods. Royal Botanic Garden Edinburgh and Centre for International Forestry Research, Edinburgh. Ola-Adams, B.A. & Hall, J.B. (1987) Plant-soil relations in a mature forest inviolate plot at Akure, Nigeria. Journal of Ecology, 3, 57-74. Olsen, S.R., Cole, C.U., Watanabe, FS. & Dean, L.A. (1954) Estimation of available phosphorus in soils by extraction with sodium bicarbonate. USD A Circular, 939, 19. Palmiotto, P. A., Davies, S. J, Vogt, K.A., Ashton, M.S., Vogt, D.J. & Ashton, PS. (2004) Soil-related habitat specialization in dipterocarp rain forest species in Borneo. Journal of Ecology, 92, 609-623. Paoli, GD. (2004) Community and ecosystem consequences of tree species turnover along soil nutrient gradients in lowland rain forest of Indonesian Borneo. PhD thesis. University of Michigan, Ann Arbor. Paoli, G.D., Curran, L.M. & Zak, DR. (2005) Phosphorus efficiency of aboveground productivity in Bornean rain forest: evidence against the unimodal response efficiency hypothesis. Ecology, 86, 1548-1561. Phillips, O.L., Vargas, P.N., Monteagudo, A.L., Cruz, A.P., Zans, M.C, Sanchez, WG. et al. (2003) Habitat association among Amazonian tree species: a landscape-scale approach. Journal of Ecology, 91, 757-775. Pitman, N.C., Terborgh, J., Silman, M.R., Nunez, P.V., Neill, DA. & Ceron, C.E. (2001) Dominance and distribution of tree species in upper Amazonian terra flrme forests. Ecology, 82,2101-2117. Potts, M.D., Ashton, PS., Kaufman, L.S. & Plotkin, J.B. (2002) Habitat patterns in tropical rain forests: a comparison of 105 plots in northwest Borneo. Ecology, 83, 2782-2797. Proctor, J, Anderson, J.M., Chai, P. & Vallack, WH. (1983) Ecological studies in four contrasting lowland rain forests in Gunung Mulu National Park, Sarawak. I. Forest environment, structure and floristics. Journal of Ecology, 71, 237-260. Rajaniemi, T. (2003) Explaining productivity-diversity gradients in plants. Oikos, 101, 449-457. Rawls, WJ. (1983) Estimating soil bulk density from particle size analysis and organic matter content. Soil Science, 135. 123-125. Rogstad, S.H. (1990) The biosystematics and evolution of the Polyalthia hypoleuca species complex (Annonaceae) of Malesia. II. Comparative distributional ecology. Journal of Tropical Ecology, 6, 387-408. Ruokolainen, K. & Vormisto, J. (2000) The most widespread Amazonian palms tend to be tall and habitat generalists. Basic and Applied Ecology, 1, 97-108. Sheldrick, B.H. & Wang, C. (1993) Particle size distribution. Vormisto, J., Svenning, J.C., Hall, P. & Balslev, H. (2004) Diversity and dominance in palm (Arecaceae) communities in terra firme forests in the western Amazon basin. Journal of Ecology, 92, 577-588. Webb, CO. & Curran, L.M. (1996) A field key to dipterocarp seedlings of the Gunung Palung National Park, West Kalimantan, Indonesia. Tropical Biodiversity, 3, 193-225. Webb, CO. & Peart, DR. (2000) Habitat associations of trees and seedlings in a Bornean rain forest. Journal of Ecology, 88, 464-478. Yamada, T., Itoh, A., Kanzaki, M., Yamakura, T., Suzuki, E. & Ashton, PS. (2000) Local and geographical distributions for a tropical tree genus, Scaphium (Sterculiaceae) in the Far East. Plant Ecology, 148, 23-30. Received 7 June 2005 revision accepted 13 September 2005 Handling Editor: David Burslem Supplementary material The following supplementary material is available online from www.Blackwell-Synergy.com: Table SI Population attributes of the Dipterocarpaceae. Table S2 Habitat associations of the Dipterocarpaceae. © 2005 The Authors Journal compilation © 2006 British Ecological Society, Journal of Ecology, 94, 157-170 D. R. Zak 170 Soil Sampling and Methods of Analysis (ed. M.R. Carter), G. D Paoli pp. 499-511. Lewis Publishers, Boca Raton. , „ „ Shmida, A. & Wilson, M.V. (1985) Biological determinants of L.M. Curran & .'. . ' , , , „„ species diversity. Journal oj Biogeograpny, 12, 1-20. Slik, J.W.F, Poulsen, A.D., Ashton, PS., Cannon, C.H., Eichhorn, K.A.O., Kartawinata, K. et al. (2003) A floristic analysis of the lowland dipterocarp forest of Borneo. Journal of Biogeography, 30, 1517-1531. Sollins, M. (1998) Factors influencing species composition in tropical lowland rain forests: does soil matter? Ecology, 79, 23-30. ter Steege, H, Tetten, V.G, Polak, A.M. & Werger, M.J.A. (1993) Tropical rain forest types and soil factors in a watershed area in Guyana. Journal of Vegetation Science, 4, 705-716. Svenning, J.C (1999) Microhabitat specialization in a species-rich palm community in Amazonian Ecuador. Journal of Ecology, 87, 55-65. Swaine, M.D. (1996) Rainfall and soil fertility as factors limiting forest species distributions in Ghana. Journal of Ecology, 84, 419-428. Tilman, D. (1982) Resource Competition and Community Structure. Princeton University Press, New Jersey. Tuomisto, H., Ruokolainen, K. & Yli-Halla, M. (2003) Dispersal, environment, and floristic variation of western Amazonian forests. Science, 299, 241-244. Valencia, R., Foster, R.B., Villa, G, Condit, R., Svenning, J, Hernandez, C. et al. (2004) Tree species distributions and local habitat variation in the Amazon: large forest plot in eastern Ecuador. Journal of Ecology, 92, 214-229.