Edaphic and Human Effects on Landscape-Scale Distributions of Tropical Rain Forest Palms Author(s): Deborah A. Clark, David B.

Clark, Rosa Sandoval M., Marco Vinicio Castro C. Source: Ecology, Vol. 76, No. 8 (Dec., 1995), pp. 2581-2594 Published by: Ecological Society of America Stable URL: http://www.jstor.org/stable/2265829 Accessed: 23/09/2009 17:03
Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the JSTOR archive only for your personal, non-commercial use. Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at http://www.jstor.org/action/showPublisher?publisherCode=esa. Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed page of such transmission. JSTOR is a not-for-profit organization founded in 1995 to build trusted digital archives for scholarship. We work with the scholarly community to preserve their work and the materials they rely upon, and to build a common research platform that promotes the discovery and use of these resources. For more information about JSTOR, please contact support@jstor.org.

Ecological Society of America is collaborating with JSTOR to digitize, preserve and extend access to Ecology.

http://www.jstor.org

Ecology, 76(8), 1995, pp. 2581-2594 ? 1995 by the Ecological Society of America

EDAPHIC AND HUMAN EFFECTS ON LANDSCAPE-SCALE DISTRIBUTIONS OF TROPICAL RAIN FOREST PALMS1
DEBORAH

A.

CLARK,2 DAVID

B.

CLARK,2 ROSA SANDOVAL

M., AND

MARCO VINICIO CASTRO C.3

Organization for Tropical Studies, La Selva Biological Station, INTERLINK-341, P.O. Box 02-5635, Miami, Florida 33152 USA

Abstract. We studied the landscape-level spatial variation in distribution and abundance of seven species of canopy and subcanopy palms in a neotropical rain forest. Within 568 ha of nonswamp old-growth lowland forest at the La Selva Biological Station, Costa Rica, we sampled at 516 intersections of a reserve-wide grid to systematically assess the species' distributions across the major edaphic gradients. We found mosaics of community structure related to marked within-forest variability in both soil and topography. Total stem density of this guild varied with both edaphic factors. Steep sites had twice as many large palms (> 10 m tall) per hectare than those on gentler topography or at lower slope positions. The combined density of subcanopy and canopy palms also varied significantly among soil types in the smallest size class (1-5 m tall), but not for larger individuals. Local (point) species richness of the subcanopy and canopy palms varied among soil types. Of the seven species, two (Astrocaryum alatum and A. confertum) were rare. All others showed significant edaphic variation in their distribution and/or estimated density. Two closely related species had strong and opposite edaphic associations. Euterpe macrospadix was biased toward steep topography and less fertile sites. Prestoea decurrens was nearly ommipresent on soil types with gentle topography while absent from half the points on soils with steep slopes. Two other closely related species, Iriartea deltoidea and Socratea exorrhiza, while virtually omnipresent across soil types and topographic positions, showed marked reciprocal variation in density between related soils. Iriartea's spatial distribution further indicates local removal of this species from one sector of the old growth by human harvesting (with subsequent apparent "release" of Socratea in this site). The most abundant species, Welfia georgii, while present at all sample points, showed significant among-soil variation in density. This substructuring of the arborescent palm guild results from the interplay of edaphic variation and past human activity. These findings and evidence from other sites suggest that marked spatial heterogeneity in community structure, at small to large scales (0.5-103 ha), may be general among tropical wet forests. Study of the spatial scales and causes of this variability will produce a more robust understanding of these complex ecosystems.
Key words: Astrocaryum; Costa Rica; edaphic variation; Euterpe; Iriartea; landscape ecology; palms; Prestoea; Socratea; spatial heterogeneity; tropical rain forest; Welfia.
INTRODUCTION

To understand many aspects of populations, communities, and ecosystems, a first need is to determine how the distributions and abundances of the organisms vary across the landscape. To develop predictive power further requires understanding the processes, past and present, underlying the spatial heterogeneity thus identified (Levin 1992). Although these ideas are generally recognized by most ecologists, current ecological studI Manuscript received 12 September 1994; revised 21 February 1995; accepted 4 March 1995; final version received 27 March 1995. 2 Present institutional address: Department of Biology, University of Missouri-St. Louis, 8001 Natural Bridge Road, St. Louis, Missouri 63121 USA (mailing address remains as above). I Present address: Instituto Nacional de Biodiversidad, Santo Domingo de Heredia, Costa Rica.

ies often lack the underpinnings of a clear understanding of the important scales of spatial variation within study sites. In the wet tropics a growing number of studies at small-to-large scales within old-growth forest have demonstrated strong effects of edaphic heterogeneity on floristic composition (Ashton 1969, Newbery and Proctor 1984 and cited references, Lescure and Boulet 1985, Lieberman et al. 1985, Basnet 1992, Johnston 1992, Ruokolainen and Tuomisto 1993, Gentry and Ortiz 1993, Oliveira-Filho et al. 1994). Much of the Amazon Basin is a highly interdigitated mosaic of forest types produced by the erosion/deposition cycles of the major rivers (Salo and Rasanen 1989, Foster 1990) and marked variation in terra firme parent material and geochemistry (Jordan 1985, Chauvel et al. 1987, Guillaumet 1987). Another potential source of small-to-medium scale spatial heterogeneity within tropical wet

2582
TABLE

DEBORAH A. CLARK ET AL.

Ecology, Vol. 76, No. 8

1. The seven species of canopy and subcanopy palms (Arecaceae, Subfamily Arecoideae) in old-growth forest at La Selva and their local uses. Tribe designations are from Uhl and Dransfield (1987). Voucher specimens are from the Costa Rican National Herbarium. Local uses were determined by interviewing eight long-term residents of the surrounding region (Cant6n de Sarapiquf) about recent historical uses of each palm species, which they identified from line drawings. Tribe Iriarteeae Species (voucher no.) Iriartea deltoidea Ruiz & Pav6n [Chac6n, Chac6n, Mora 1968] Socratea exorrhiza (Martius) Wendl. [W. D. Stevens 24559] Prestoea decurrens (Wendl.) H. E. Moore [M. H. Grayum, C. Jermy 6783] Euterpe macrospadix Oersted [M. H. Grayum 7813] Astrocaryum confertum Wendl. ex Burret [G. de Nevers, B. E. Hammel 7820] Astrocaryum alatum Loomis [W. D. Stevens 24625] Welfia georgii Wendl. ex Burret [M. Weimann & P. M. Rich 137] Local uses heart-of-palm, construction wood construction wood (less used than Iriartea), rarely heart-of-palm none heart-of-palm (less used than Iriartea) none none thatching

Areceae

Cocoeae

Geonomeae

forests is the historical or current impact of local human activity, from silviculture to selective harvesting, even within stands considered to be old-growth (Gordon 1982, Gomez-Pompa and Kaus 1990, Anderson 1990, Bush and Colinvaux 1994). Understanding the patterns and causes of spatial variability in the community structure of tropical wet forests can contribute to the resolution of important questions about these ecosystems. Is small-to-medium scale edaphic heterogeneity more characteristic of these forests than of temperate stands? Could this be a factor in the maintenance of high species richness in tropical rain forests? Are wet tropical forests characterized by major internal spatial heterogeneity in ecosystem-level processes such as primary productivity? Is variability in site conditions and forest communities likely to have major impacts for efforts in tropical forest conservation and restoration? Our goal in this study was to assess landscape-scale spatial heterogeneity within an old-growth neotropical rain forest, as reflected in the distribution and abundance patterns of the large palms. An unprecedented set of research tools for landscape-level studies of tropical rain forest (a reserve-wide grid system, soil map, and Geographical Information System [GIS]) is newly available at the study site. We used them to ask the following questions, at the scale of 500 ha of contiguous forest: (1) How do the abundance, size distribution, and local species diversity of the guild of subcanopy and canopy palms vary with respect to the landscape-scale variation in soil and topography? (2) Do the distribution and abundance of the individual species vary with edaphic conditions, and do these patterns differ among species? (3) Is there evidence of human impact on the distributions of large palms in the oldgrowth forest?
STUDY SITE AND SPECIES

Atlantic lowlands of Costa Rica (10?26' N, 84?00' W; elevation 37-150 m). It is classified in the Holdridge life zone system as tropical wet forest (Hartshorn and Hammel 1994). Mean annual rainfall is 3962 mm, with every month averaging at least 100 mm of rain (Sanford et al. 1994). The flora includes 323 tree species (Hartshorn and Hammel 1994) and is rich in palms (31 native species, Chazdon 1985; 25% of all woody stems ?10 cm in diameter, Lieberman et al. 1985). Although not subject to large-scale disturbances such as hurricanes, the forest is very dynamic. Stem turnover is high (2.0-2.3%/ yr for trees ?10 cm dbh; Lieberman et al. 1990), as is the frequency of gap formation (Hartshorn 1978, Sanford et al. 1986). La Selva's soils range from relatively fertile entisols and inceptisols to infertile ultisols (Sollins et al. 1994). The 1:10 000 soils map of La Selva (Sancho and Mata 1987) demarcates 23 consociations (mapping units within which ?75% of the area is the described soil type) and one complex. Our study species are the seven subcanopy to canopy palms found in old-growth at La Selva (Table 1). Henceforth all except the Astrocaryum species are designated by genus name only. Four species have been locally harvested (for heart-of-palm, construction wood, thatching) in recent history (Table 1).
METHODS

The La Selva Biological Station of the Organization for Tropical Studies (OTS) is a 1550-ha reserve in the

To evaluate landscape-scale distributions of the large palms in upland old-growth forest, we used the original La Selva reserve ("Original La Selva"; McDade and Hartshorn 1994), after excluding swamp, secondary forest, disturbed habitats, and the 12.4 ha of restricted access plots. Within the resulting study area (568 ha; Fig. 1) we sampled at intersections of La Selva's reserve-wide grid (posts every 50 m along lines at 100m intervals; accuracy ?20 cm). In the limited areas of alluvial soils, we sampled at all grid intersections (N = 78). On the nonalluvial soils we sampled at the in-

December1995

TROPICALPALM COMMUNITYSTRUCTURE

2583

tersections along every other grid line (thus along lines separated by 200 m; N = 438 grid points). We excluded points that fell within streams or floodable microsites m as well as those 30() from secondary forest or altered habitat. Pailnm samnplinig.-From data at each sample point (grid post) we evaluated distributions of the seven palm species in two ways: in terms of estimated density and presence/absence. 1. Estimnated densitv.-For each of four size classes, we measured the distance from each sample point to the nearest individual of any of the species and noted the species of that palm. This "Closest Individual Method" (Cottam and Curtis 1956) enables estimation of stem density [Densitv (N/ha) = 10 000/(4D2)J, where D is the mean distance in metres to the closest individual (the mean of the distances of Closest Individuals from all sample points). At each point we used this technique tor each of four size classes: 1-5, >5-10, > 10(-15, and > 15 m tall. A telescoping 15--mmeasuring pole was used to evaluate palm heights. For each size class, density of each species in a given area (i.e., in a given soil type or at a given topographic position) was calculated by multiplying the total stem density of canopy and subcanopy palm species (see formula above), by the proportion of that species among the Closest Individuals sampled in the area. 2. Preseele/Absence.-For each of the seven study species we noted whether any individual > 1 m tall was visible (at any distance) in any direction from the grid post or while measuring the Closest Individuals. Edaiphic factors: Soils.-The La Selva soils map is in the station's GIS (Arc/Info 6.0.1, on Sun Sparc II workstations). We used our field notes regarding streams, swamps, and topography to refine this map, and then grouped related soil map units into four categories. The resulting GIS coverage was used to classify each sample point by major soil type. Alluvium is an Andic Humitropept (Inceptisol) on gently undulating terrain. Although divided into the Experinental and Holdridge consociations, which were previously considered distinct terraces (Sollins et al. 1994), the grid survey data (OTS records) show these sectors to be at equivalent elevations (Experimental:
mode = 42-44 m, N = 132; Holdridge: mode = 42-

S @0
N 4.

4 '4

1~~~~~0p
1

0 100 300

800 metres
MVCC-JJP/Jan95

Ftc;. 1. The study area in old-growth(nonswamnp) forest (the thick blackborderindicatesthe boundaryof the sampled area). Circles denote the 516 points sampledalong the 50 x 200 m grid covering La Selva. Differentgrey tones indicate the four soil types (from lightest to darkest grey tone: Alluvium [Holdridge and ExperimentalConsociations of the Alluvium are labeled as "Ho" and "Ex' l; Arboleda soil, soil Residualsoil; Streamis [swampsare indicatedby the figuredhatchingpattern]). Trailsare shown as dashedlines. The is distributionof Iri(artea dleltoidleat shown by: present (0); absent (0).

44 m, N = 398). While low in pH and exchangeable bases, they have higher pH, extractable phosphorus (P), and/or exchangeable bases than the other three soil types (Sollins et al. 1994; J. S. Denslow and G. Chaverri, unpublished data). Arboleda is a consociation thought to represent old, highly weathered alluvial deposits that overlie very broken terrain (Sollins et al. 1994). This soil appears intermediate between the Alluvium (Holdridge consociation) and Residual soils in extractable P, but equivalent in pH to the Residual soils (J. S. Denslow and G. Chaverri. unlpublished data) The Residual consociations (Jagutar, Matabttey, and

Esquina) are highly weathered soils derived from old lava flows and overlie broken ridge-valley topography. They are acidic and low in exchangeable bases and extractable P (Sollins et al. 1994; J. S. Denslow and G. Chaverri, un;ipublishled daita). The Jaguar and Maitabuievconsociations, classified as Ultisols (Typic Tropohumults), cover 45% of La Selva (Sollins et al. 1994). The Stream consociations occur in the valley bottoms of principal streams: El Scalto (Lithic Humitropept), El Surd-Saltito (Typic Humitropept), and El Taconazo (Typic Tropaquept). All are strongly acidic, very low in exchangeable bases, and with poor-to-moderate drainage (Sollins et al. 1994). Edap)hic factors: To)pograiphvy.-In the field we categorized each sampling point (grid post) by topographic position: Slope Crest (just above to just below the upper breakpoint of a slope); Steep Slope, Moderate Slope, Gentle Slope (all the slope positions between Crest and Base, with slopes classed by relative steep-

2584 550 500cu 450o4001350 300z

ii

DEBORAH A. CLARK ET AL.

Ecology, Vol. 76, No. 8

sZZ l 15

IEI RESIDUAL 111ARBOLEDA STREAMS ALLUVIUM

250-

200C<

15010050 1m > 5-lOin >10-15m >15m

HEIGHTSIZE CLASS 2. Variation in total palm density (all seven species combined) by size class and soil type. Sample sizes are given in Table 2. Probabilities are for a Kruskal-Wallis test of the distances to the Closest Individual (see Methods), soil type. by * P ' 0.05.
FIG.

total estimated density across topographic and soil gradients produce marked spatial heterogeneity in their local abundance. Density and palm size.-By far the highest estimated densities (203-509 palms/ha, the seven species combined) occurred in the smallest size class (1-5 m tall; Fig. 2). Abundance dropped sharply at larger sizes. Topographic effects.-Estimated stem density varied significantly in relation to topography. In the two larger size classes (>10-15 m tall, >15 m tall), total density decreased continuously from slope crests, to slopes of decreasing steepness, to slope bases and flat terrain (Fig. 3B). In both size classes palm density varied twofold among topographic positions. The number of smaller palms per hectare, however, did not vary significantly with topography (Fig. 3A).

-:

400 630-

ness); Base of Slope (just above to just below the lower slope breakpoint); and Flat (flat terrain, whether on broad ridges or terraces or in swales). When this field index was combined with slope angle measurement at the same point in another La Selva study (D. B. Clark et al., unpublished data), mean slope angles were (N,
points): Steep Slope, 24.70 (91); Moderate Slope, 15.70 (124); Gentle Slope, 6.60 (150); and Flat, 1.60 (60).

LU 200
-J

Statistical analysis.-To test for species associations with edaphic conditions, the presence-absence data were tested with chi-square (number of sample points with a given species present and absent vs. soil types or topographic positions). To test whether total palm density in a given size class differed among soil types (or topographic positions), a Kruskal-Wallis test was performed on the Closest Individual distances for that size class in each soil type (or topographic position); significantly different distances to Closest Individuals indicate significant density differences among the edaphic conditions (greater distances indicate lower density). When total palm densities did not differ across soils, species could be individually tested for an association between their density and soil. A chi-square test was used to compare the soil types in terms of the Closest Individuals that were and were not the species in question. Significant among-soil variation in a species' representation among Closest Individuals indicates that the species' density varies significantly among soils. Throughout, the probability used for significance is 0.05.
RESULTS

0-

iS5m
60 B.

>5-lOinm
*

50o 40

CO 30 z 20-J-

0 >10-15 m >15 m

HEIGHTSIZE CLASS 3. The relation of total palm density (all seven species combined) to topographic position (all soils combined). (A) Individuals in the two smaller size classes. (B) The two larger size classes (note different y axis). Probabilities are for a Kruskal-Wallis test of the distances to the Closest Individual (see Methods),by topographic position. Fill patterns, from left to right (with N, points sampled): slope crest-138; steep moderate slope-42; slope-105; gentle slope-91; slope base-57; flat-83. ** P ' 0.01; *** P ' 0.001.
FIG.

Total palm density The abundance of subcanopy and canopy palms varies greatly across the La Selva landscape. Changes in

December 1995

TROPICAL PALM COMMUNITY STRUCTURE

2585

100RESIDUAL 900
H

l_ARBOLEDA
w
E
-

V/A STREAMS
ALLUVIUM

w
W

80

C/)

CE
W

70 6050

0~~~~~~~~~~~~~~~~~~~~~~~~~~R
1

00 20 10 ;'30l
l

WE

SO

IR

EU SPECIES

PR

AL

CO

The distribution of the seven canopy and subcanopy palm species with respect to the four major soil classes. Data for each species are the percent of sample points in each soil type at which any individual >1 m tall was observed. Sample sizes as in Table 2. Species abbreviations: WE-Welfia, SO-Socratea, IR-Iriartea, EU-Euterpe, PR-Prestoea, AL-Astrocaryum alatum, CO-Astrocaryum confertum.
FIG. 4.

Soil effects.-The relation between estimated palm density (all seven species combined) and soil type (Fig. 2) contrasted with the topographic patterns. For the three larger size classes, palm densities were equivalent across the four soil types. Abundance of the smallest palms, however, did vary significantly among soils (Kruskal-Wallis, P < 0.03). Stem density at this size was much lower on the Streams soils than on the other three soil types (Tukey a posteriori pairwise comparison, P < 0.05). Species distributions The seven palm species have different distribution patterns across La Selva. The combined data on presence/absence and estimated density reveal: (1) general rarity of the two Astrocaryum species; (2) strong and opposite associations with soil type and topography for the species Euterpe and Prestoea; (3) evidence of local removal of one species (Iriartea) through human harvesting; (4) reciprocal density differences between Iriartea and Socratea; and (5) significant intersoil variation in the density of one ubiquitous species (Welfia). Astrocaryum.-Both Astrocaryum species were rare in the sampled area (Fig. 4). In all four soil types, they occurred at few sample points (3-5%, A. confertum; 2-i 1%, A. alatum). The estimated density of A. confertum (from Closest Individual data) was 0.0 stems/ ha in all size classes on all soils, except for the 1-5 m height class on Residual soils (1.2 stems/ha). Similarly, A. alatum estimated density was 0.0-3.0 stems/ha across sizes and soils. This species, however, is abundant in swamps at La Selva; Hartshorn (1983) reported

30.5 stems/ha (?10 cm diameter) in a 2-ha plot in swamp. Euterpe.-The presence/absence data (Fig. 4) revealed strong soil-related heterogeneity in Euterpe's distribution (chi-square, df = 3, P < 0.0001). Although present at 91% of sample points in the Residual soils, this species occurred at only 10% of the points in Alluvium, with intermediate frequencies in the other two soil types. A parallel pattern is seen with the Closest Individual data (percent of sample points with Euterpe as ?1 of the four Closest Individuals: Residual, 38%; Arboleda, 21%; Streams, 25%; Alluvium, 1%; chisquare, df = 3, P < 0.0001). Euterpe's abundance also significantly varied with soil type at all sizes (Appendix). Estimated densities were markedly higher on the Residual soils than on the other three soils. A single individual occurred in the density sampling on the Alluvium. Because topography varies with soil type (Table 2), a species' association with soil type could reflect responses either to topography and/or to substrate per se. To separate these factors, we carried out two additional analyses of Euterpe's distribution patterns. The many sample points on the Residual soils allow assessment of species' topographic associations within a single soil type. Euterpe's presence/absence patterns within the Residual soils (Fig. SA) strongly varied with topographic position. While present at >90% of points associated with steeper topography (Slope Crest, Steep Slope, Moderate Slope), Euterpe was absent from half of the Flat sites. Its distribution was also strongly associated with soil type per se. Comparing presence/

2586
TABLE

DEBORAH A. CLARK ET AL.

Ecology, Vol. 76, No. 8

2. Topographic variation on the four major soil types (see Methodsfor descriptions of soil classes and topographic positions). Data are the proportion of sample points in each topographic position, within each soil type. Proportion within each soil type Residual Arboleda Stream Alluvial

Topographic position

Slope crest 0.30 0.36 0.00 0.31 Steep slope 0.21 0.25 0.00 0.03 Moderate slope 0.27 0.11 0.00 0.13 Gentle slope 0.09 0.05 0.15 0.19 Base of slope 0.10 0.21 0.69 0.06 Flat 0.03 0.02 0.15 0.28 Total 1.00 1.00 1.00 1.00 N 44 329 65 78 Area sampled (ha)* 340 43 65 64 * Additional area was sampled at upland sites embedded within the 55 ha of swamp in the total study area; such withinswamp points were classified to soil type based on the local topography and proximity to soil map units.

density on the Streams soils and Arboleda soils, and its density on the Residual soils (Density ratios, 1-5 m, >5-10 m, and >10-15 m height classes, respectively: Streams/Residual-2.3, 3.8, 7.7; Arboleda/Residual-0.9, 2.1, 5.0). These patterns suggest that in both the 1-5 m and >5-10 m size classes, Prestoea suffers higher mortality on the Residual soils than on the Streams or Arboleda soils.
As with Euterpe, Prestoea's occurrence within the

Residual soils (Fig. SA) was strongly associated with topographic position, but in the opposite direction to that of Euterpe. While present at only 29 and 47% of points associated with steeper topography (Slope Crest and Steep Slopes, respectively), Prestoea was at 78 and 81% of the Slope Base and Flat sites. Its distribution

100 A. z w w w Tul) H

90 705040-

absence at the same topographic

position (the four po-

sitions with adequate samples) showed much more frequent occurrence on the Residual soils than on the Alluvium (Fig. 6A). Together these patterns suggest balanced avoidance of gentle topography (Table 2) and more fertile conditions (based on the limited nutrient data for La Selva soils; see Methods). Euterpe's distribution was: nearly omnipresent on Residual soils (steep topography, infertile); intermediate occurrence on both the Arboleda soils (steep topography, intermediate fertility) and the Streams soils (base of slopes/flat terrain, infertile); and nearly absent from Alluvial soils (gentle topography, fertile). Euterpe's significant density differences among soils (Appendix) paralleled these presence/absence patterns. Prestoea.-This close relative of Euterpe was also strongly associated with both topographic and soil variation, but in directions opposite to those found with Euterpe. Prestoea's presence/absence patterns (Fig. 4) varied markedly among soil types (chi-square, df = 3, P < 0.0001). While at only 48-53% of points in the Residual and Arboleda soils, it occurred at 91-92% of points in the Streams and Alluvial soils. A similar pattern exists in the percent of sample points with Prestoea as any of the four Closest Individuals (Residual, 14.3%; Arboleda, 20.5%; Streams, 52.3%; Alluvium, 46.2%; chi-square, df = 3, P < 0.0001). Prestoea's estimated density (stems per hectare; Appendix) varied significantly across soil types in all three size classes (smaller than the other six species, Prestoea only occurred once as a Closest Neighbor at >15 m height). Overall, Prestoea density was lowest on the Residual soils. The ratio between its density on the Alluvium and on the Residual soils increased with size class, from 3.8 (1-5 m tall palms) and 3.5 (5-10 m tall) to 5.0 (10-15 m tall). Stronger shifts with increasing size class occurred in the ratios between Prestoea's

z
0

30-

U_ 20-'0 100

EUTERPE
100 B w Ul) w

PRESTOEA

80-

cE cE

o~7060-

50-

ft

u)
Z

4030 q

0
u_

20
10S}
0

IRIARTEASOCRATEA WELFIA
FIG. 5. Topographic affinities of the abundant canopy/ subcanopy palm species, controlling for soil type. Data are the percent of sample points at which each species was present at each topographic position, within the Residual soils. Fill patterns, from left to right (with N, points): slope crest (98); steep slope (70); moderate slope (90); gentle slope (30); slope base (32); flat (9). Probability (**** P ' 0.0001) is from a chi-square test for association of species presence/absence with topographic position. (A) Euterpe and Prestoea (df = 3 for both chi-square tests; for each, two pairs of adjacent topographic positions were combined to eliminate expected values <5). (B) Iriartea, Socratea, and Welfia.

December 1995

TROPICAL PALM COMMUNITY STRUCTURE

A.

2587

was also significantly associated with soil type per se, in analyses controlling for topography (Fig. 6B). In contrast to Euterpe (Fig. 6A), Prestoea occurred more frequently on the Alluvium than on Residual soils at all four topographic positions that could be compared (chi-square, df = 1, P < 0.0001 and P = 0.01 for two of four comparisons). Taken together, these patterns indicate that Prestoea's distribution is associated with both soil and topography. A bias toward more fertile soil is suggested by the Alluvium/Residual comparison at given topographic positions (Fig. 6B) and by the size-related increase in relative density on the Arboleda soil, compared to the (less fertile, but topographically equivalent) Residual soils. The presence/absence data, however, suggest a stronger association with topography than with soil. Prestoea was nearly omnipresent on the Streams and Alluvium soils (both soils of low slope positions/gentle topography, while contrasting in fertility). Similarly, the species occurred at only about half the sample points on the two soils on highly dissected terrain (Residual, Arboleda; Fig. 4), in spite of these soils' apparent fertility difference. The significant among-soil differences in Prestoea's stem density in the two smaller size classes (Appendix) follow the same general trend (markedly lower densities in the Residual and Arboleda soils than in the Streams and Alluvium soils). In the largest size class, although the pattern is less clear, Prestoea's density is lowest on the steep Residual soils and highest on the flat Streams soils. Iriartea.-Unlike Euterpe and Prestoea, Iriartea did not show an association with topography (Fig. 5B). Its presence/absence patterns (Fig. 4), however, did differ significantly among soils (chi-square, df = 3, P = 0.0000). Iriartea was present at 91-98% of sample points on all soils but Alluvium, where it occurred at 50% of the points. Mapping (Fig. 1) revealed a strong spatial bias in Iriartea's distribution within the Alluvium. Although found at all 36 sample points in the Holdridge consociation of the Alluvium, the species was virtually absent from the Experimental consociation (present at only 2 of 42 points). The two consociations are both classified as Andic Humitropepts (Sollins et al. 1994), and have equivalent distributions of topographic positions (chi-square, df = 3, P = 0.89) and elevations (see Methods). An edaphic explanation for Iriartea's distribution within the Alluvium is unlikely given: (1) the species' lack of variation with topography; (2) its virtual omnipresence in all four soil types (within the Alluvium, omnipresent in the Holdridge consociation); (3) that both strongly edaphically varying species, Euterpe and Prestoea, had equal estimated densities on the two Alluvium consociations (Appendix); and (4) the similarity of the two consociations. A nonedaphic factor distinguishing the two soil map units, however, is location

10080-

***

***

Ul)

70m UJ

60
I50-

cl
Z

4030-

aO

20
10C

-0
SLOPE CREST MOD SLOPE GENTLE SLOPE FLAT

100F . Su

Ul)

80-

70w
LU

60I50-

U)
Z

403020-

aIL

10 0 SLOPE CREST MOD. SLOPE GENTLE SLOPE FLAT

FIG. 6. Soil affinities of (A) Euterpe and (B) Prestoea, controlling for topographicposition (Residual soils: white bars; Alluvium, black bars). Each probabilityis for a chisquare test for association of the species' presence/absence with soil type, at a fixed topographicposition (** P < 0.01;

*** P < 0.001; **** P < 0.0001). N, sample points (Residual

soils and Alluvium, respectively): slope crest, 98 and 24; moderateslope, 90 and 10; gentle slope, 30 and 15; flat, 9
and 22.

with respect to human activity and access. The Experimental consociation lies between a former commercial cacao plantation (operative until the 1980s) and the former housing for La Selva farm workers (abandoned in the early 1980s). In contrast, the old-growth areas of the Holdridge consociation, while not far from a neighboring cattle ranch, are relatively remote from modern human habitation. Taken together, these lines of evidence indicate that Iriartea's absence from the Experimental consociation is due to intense past harvesting of the species from this sector of the La Selva old-growth forest. Iriartea's absence from this consociation in all size classes >1 m tall suggests complete local removal of large stems by harvesting, with this process occurring over a protracted period. Such ac-

2588
TABLE

DEBORAH A. CLARK ET AL.

Ecology, Vol. 76, No. 8

3. The relationship between the densities of Socratea exorrhiza (S) and Iriartea deltoidea (I) on related soil types. Probabilities are for chi-square tests of Closest Individual data; these tests effectively compare densities (see Methods) of a given species on the two soils of each pair (or of the two species combined across the soil pair): *** P < 0.001; ** P < 0.01; * P < 0.05; NS, P > 0.05. Density (stems/ha)

Height class (m): Soil I Alluvial consociation Experimental 0

NS

1-5 S 58
NS

>5-10 (S + I) (58)
NS

>10-15 (S + I) (11)
NS *

>15 (S + I) (17)
NS *

I 0
**

S 11
**

I 1 13 5
NS

S 16
*

I 0 10 8
*

S 23
*

(S + I) (23)
NS

Holdridge Arboleda Residual

47 17
**

35 121
***

(82) (138)
NS

13 11
NS

0 19
NS

(13) (30)
NS

4 8
NS

(17) (13)
NS

5 17
NS

(15) (25)
NS

84

51

(135)

17

15

(32)

(16)

16

14

(30)

tivity could have been relatively recent, but is unlikely to have occurred after strict site protection was established in the early 1980s. Socratea.-As for Iriartea, the distribution of this closely related species was not associated with edaphic conditions. Within the Residual soils, Socratea's presence/absence patterns did not vary with topographic position (Fig. 5B). The species was present at 9 1-100% of the sample points in each of the four soil types (Fig. 4). Nevertheless, Socratea's estimated density did vary among soils (Appendix). In all four size classes, Socratea's lowest density occurred on the Streams soils. In two of the four size classes, the among-soil density differences were significant. Socratea and Iriartea have reciprocal density patterns on similar soils (Table 3). Between the Experimental and Holdridge consociations of the Alluvium, for example, both species showed significant density differences in three of the four size classes. The intersoil differences, however, are opposite for the two species. In spite of each species' large density differences between consociations, the summed density of the two palm species in each of these size classes is remarkably similar in the two consociations. None of the other abundant palm species (Euterpe, Prestoea, and Welfia) showed density differences between these two Alluvial consociations (Appendix). Iriartea and Socratea also showed reciprocal density patterns between the Residual and Arboleda soils (Table 3). Between these two soil types, which have similar topography (Table 2; chisquare, df = 4, P = 0.07), all seven palm species had similar frequencies of occurrence (Fig. 4). In the 1-5 m size class, however, Socratea and Iriartea had large significant differences in density between these soils, although the sum of the two species' densities in this size class was remarkably similar between soils. In all eight comparisons for these two soil pairs (four size classes, Experimental vs. Holdridge; four size classes, Arboleda vs. Residual), there is a reciprocal relationship between the densities of Iriartea and Socratea (one species' relatively higher density in the first soil compared to the second soil is complemented by the op-

posite pattern for the other species). The Binomial probability of such an outcome is <0.01 (two-tailed). Welfia.-The ubiquitousness of this palm was shown by the presence/absence survey (Fig. 4). Welfia was one or more of the four Closest Individuals at 79-90% of the sample points in all four soil types. Its estimated densities were overall the highest among the seven palm species, particularly in the two smallest size classes (Appendix). Welfia was also the only species accounting for 50% or more of the canopy and subcanopy palm stems per hectare in any size class on any soil (percent of stems that were Welfia: Arboleda, 52.3% [1-5 m tall] and 50.0% [10-15 m]; Alluvium, 63% [1-

m]).

Given Welfia's widespread distribution, it is not surprising that its occurrence within La Selva showed no association with topography (Fig. SB). Its density, however, did vary significantly among soil types (two of four size classes; Appendix). In all four size classes, Welfia was least abundant on the Streams soils. Local species richness and soil The local species richness of subcanopy/canopy palms varied significantly among soil types. We examined this relationship two ways. To compare all soils we excluded Iriartea because of the confounding effect of human harvesting of this species from the Experimental consociation. While at least four of the six palm species were present at 68% of sample points on the Streams soils and at 50% of points on the Residual soils, a much smaller proportion of points on the other soils had this many palm species present (Arboleda25%; Alluvium: Holdridge Consociation-19%, and Experimental Consociation-17%; Kruskal-Wallis, number of species/sample point on five soil types, P < 0.0001). We repeated this analysis with all seven species, but excluding the Experimental Consociation to remove the effect of harvesting of Iriartea. The patterns were unchanged (Kruskal-Wallis, number of species per sample point on four soils, P = 0.0000). The mean numbers of palm species present at a sample point were: Streams, 4.58; Residual, 4.51; Arboleda, 4.23;

December 1995

TROPICAL PALM COMMUNITY STRUCTURE

2589

Alluvium, 4.19 [Holdridge]. On no soil type were there any sample points with fewer than two of the palm species present, however. Conversely, at no sample points in any soil did all the species co-occur.
DISCUSSION

Edaphic factors and forest heterogeneity The variation in soil and topography within this oldgrowth forest has major impacts on the large palms: a mesoscale (hundreds of hectares) gradient in local species diversity; topographic changes in the group's overall abundance; one species' absence from a major soil type; and edaphic variation in density for many species size classes. All five common species showed marked distributional and/or density changes over the edaphic gradients within La Selva. Palm species richness.-Over the landscape we found highly significant variation in the local (point) species richness of subcanopy and canopy palms. The limited available soil data suggest that this diversity trend is inversely related to soil fertility. The lowest local richness of palm species occurs in the Alluvium, the most fertile soil type (Sollins et al. 1994; J. S. Denslow and G. Chaverri, unpublished data). The soils with highest local palm diversity, Streams and Residual, appear to be the two least fertile (Sollins et al. 1994; see Methods). Finally, the Arboleda soil, with intermediate local palm diversity, appears intermediate in fertility between the Alluvium and Residual soils (see Methods). This trend is not confounded by stem density differences, which could produce spurious differences in species richness (palm densities were equivalent across soils in three of the four size classes; in the 1-5 m size class, stem density was lowest in the Streams soil [Fig. 2], where local species richness was highest). Although the intersoil differences in local species diversity are clear, definitive analysis of their relation to soil fertility will await more systematic soil sampling. A negative correlation between tree species richness and soil nutrients was found by Huston (1980) in a data set from 46 forest plots distributed around Costa Rica. He attributed this pattern to lowered dominance by superior competitors in less productive sites. While very suggestive, his findings involved possibly confounding effects from wide differences in elevation (which affects the species pool; Gentry 1982) and rainfall patterns. The patterns we found for the large palms at La Selva, however, control for several factors that can vary among geographically separated sites (e.g., climate, elevation, species pool). They thus potentially provide interesting support for Huston's hypothesis (contingent on confirmation by more extensive soil nutrient data from La Selva). Total palm density.-The abundance patterns of canopy and subcanopy palms within the La Selva oldgrowth revealed marked topographic variation in this

tree guild. Estimated stem density of the two larger size classes varied twofold among topographic positions, with highest densities on the slope crests and steeper slopes. On steep slopes in La Selva old-growth forest height is lower, and recent gaps (with canopy ?5 m high) are three times more frequent than at other topographic positions (D. B. Clark et al., unpublished data). The density-topography variation of the larger size classes of palms may be due to the favoring of growth into these sizes by more frequent gaps on steeper terrain. Topographic variation in tree stem density may be frequent in tropical wet forests. In a Central Amazon forest, Kahn and de Castro (1985) found substantially higher palm densities on slope crests than at midslope, with lowest densities on the plateaus; they speculated that the higher densities at slope crests reflected greater gap formation on these more wind-prone sites. Similarly, in a Puerto Rican montane forest, Basnet (1992) found that the stem density of trees -10 cm dbh and of saplings 2.5-10 cm dbh decreased significantly from ridges to slopes to valley positions. In contrast, the soil variation at La Selva had little impact on the estimated density of canopy and subcanopy palms. Only for the 1-5 m height class did total density vary significantly among soils. For all other size classes, the total density of stems in this species group was remarkably similar in the four soil types (Fig. 2), in spite of major intersoil variation in the density of individual species. This constancy in total abundance of an internally varying species group suggests the operation of some sort of floristic "assembly rules" with respect to the trees in this forest (cf. Gentry and Ortiz 1993). Although we as yet can propose no mechanisms that could produce such patterning, this potentially important feature of tropical forests merits study. Species responses to edaphic variation The patterns of local species richness of large palms demonstrated a high level of species mixing in this guild. On all soils, the mean number of large palm species at each sample point was high (4.2-4.6 of seven species). Nevertheless, all five abundant species varied edaphically in density. Euterpe and Prestoea.-These closely related species showed marked associations with both soils and topography (Euterpe with steep topography and less fertile soils, Prestoea with gentle topography and more fertile soils). The limited additional data from La Selva support these patterns. Within three 4-ha forest inventory plots (one each on the Experimental and Holdridge consociations of the Alluvium and one on the Arboleda soil), Euterpe occurred at all the higher elevations (3872 m) but was absent from the lowest elevations (3236 m) (Lieberman et al. 1985). In a floristic study at La Selva (J. S. Denslow and G. Chaverri, unpublished data) Euterpe occurred in plots on both the Arboleda

2590

DEBORAH A. CLARK ET AL.

Ecology, Vol. 76, No. 8

and Residual soils, but not on the two consociations of the Alluvium; Prestoea showed the opposite pattern. Determining the proximal causes of these edaphic associations will require experimental work and intensive soil analysis. If Euterpe's distribution reflects lowered competitive ability on more productive sites, such competitive interactions must occur early in development. Even in the smallest (1-5 m) size class, Euterpe was virtually absent from the Alluvium. In contrast, the size-related increases in Prestoea's among-soil density differences suggest continuing mortality effects through regeneration. It is unlikely that direct competition between Euterpe and Prestoea causes their opposite edaphic associations, given these developmental differences and their relatively low densities. Iriartea, Socratea, and Welfia.-These palm species showed no distributional variation with either topography or soil. Similarly, Iriartea deltoidea was found by Phillips et al. (1994) in all seven forest types in a lowland Amazonian reserve in Peru, and L exorrhiza showed no edaphic associations within a Brazilian Amazonian forest (Kahn and de Castro 1985). Although Iriartea, Socratea, and Welfia were nearly omnipresent in the four soil types, their densities varied greatly within La Selva. Although their relation to the Alluvium is affected by human harvesting of Iriartea, among the other soils all three showed marked density differences (Appendix). Other sites.-There is increasing evidence of the importance of edaphically related spatial heterogeneity within old-growth tropical wet forests. Many cases are known from Southeast Asia, from numerous small plots over hundreds of square kilometres (cf. Brunei, Ashton 1969; Sarawak, Baillie et al. 1987), to small-scale within-forest variation (four 1-ha plots, Sarawak, Newbery and Proctor 1984; 0.4 ha, subtropical Queensland, Williams et al. 1969). Similar evidence has come from neotropical wet forests. As in Asia, edaphic variation in forest composition has been found at all scales: from very local effects (within a 60 x 120 m plot, Puerto Rico, Johnston 1992), to catenas or watersheds on scales of 0.5-500 ha (Central Amazon, Kahn and de Castro 1985, Peres 1994; French Guiana, Lescure and Boulet 1985; Puerto Rico, Basnet 1992; Guyana, ter Steege et al. 1993). In Panama, Hubbell and Foster (1986) found half of the 303 tree and shrub species in a 50-ha plot to be significantly associated with one of four broad topographic site types. Human impact in old-growth tropical rain forest In addition to the now well-documented impacts of edaphic variation, anthropogenic effects are a relatively little-studied but potentially important source of heterogeneity within old-growth tropical forests. In the present study, the distribution of Iriartea revealed a footprint of human activities (Hamburg and Sanford 1986) in forest previously considered "virgin."

Human activity is likely to have affected the composition of most neotropical forests (e.g., Anderson 1990, Gomez-Pompa and Kaus 1990, Garcia-Montiel and Scatena 1994). In Panama, for example, the Guaymi Indians practice selective silviculture in forests similar to La Selva (Gordon 1982). When clearing patches of forest they protect Iriartea and Socratea, and they often plant these species, as well as Welfia, into their "tree gardens." Archaeological work at La Selva (I. Quintanilla, unpublished data cited by Horn and Sanford 1992) revealed habitation sites and pottery shards from - BC 300 to AD 300; these pre-Columbians could have affected the composition of La Selva's current old growth. The distributional anomaly found for Iriartea, however, is likely to have resulted from more recent harvesting by local residents (otherwise, recolonization from neighboring populations should have occurred). Given the increasing evidence of anthropogenic impacts within neotropical old growth, researchers studying the structure and function of these forests should explicitly seek indications of human activities in their study sites (Hamburg and Sanford 1986). Interspecific interactions: the case of Iriartea and Socratea The local abundances of these palms (Table 3) suggest some type of interspecific interaction. On the Residual and Arboleda soils, these species showed reciprocal density variation at all sizes (significant at 1-5 m tall), and their summed density was remarkably similar between soils. On the Alluvium, the harvesting of Iriartea from the Experimental Consociation created a natural experiment. In all size classes, Socratea's current abundance is higher on the Experimental soil (Iriartea absent) than on the Holdridge (Iriartea present) (significant in the three larger size classes). No other species varied in density between these consociations. Socratea and Iriartea would seem to have a low probability of interspecific encounter. Neither accounts for >40-46% of the large palms per hectare in any size class on any soil (Appendix), and 75% of the stems -10 cm dbh in this forest are nonpalms (Lieberman et al. 1985). Possible forms of interaction between the two include competition for pollinators or indirect effects mediated by specific vesicular-arbuscular mycorrhizal fungi or natural enemies. Comparative field observations and experiments to assess the factor(s) underlying these species' reciprocal distributions would be very interesting. Such a study would be especially valuable given the widespread human use of Iriartea deltoidea in neotropical forests (Pinard 1993, Phillips et al. 1994). Methods for landscape-scale plant community studies The techniques in this study enabled a survey of palm distributions and densities over a large area, revealed unrecognized spatial heterogeneity in the guild, and led

December 1995

TROPICAL PALM COMMUNITY STRUCTURE

2591

to inferences about underlying factors. Both sampling methods, however, have drawbacks. Although the Closest Individual technique produces quantitative density data, intersite density comparisons for individual species require equivalent overall density of the total guild between sites. With Closest Individuals evaluated on a species group basis, one abundant species could mask variation in the others (taking data on each species at each sample point would have required seven times the field effort and would have involved data truncation at points where a species is absent or rare). The Presence/ Absence sampling provided data on each species at all sample points, but within a variable "plot" size (due to greater visibility in more open forest or steeper terrain). Additional noise in the data could result from edaphic variability within the search area. Combining methods helped resolve these issues and strengthened conclusions when the techniques produced equivalent patterns. Based on this experience, however, we recommend the following approach for studies of plant community variation over large landscapes. 1) Distributing many sample points over the major environmental gradients in the study area enables generalization to the landscape. Such unbiased inference is difficult to attain from few samples or plots, especially those not located with strict random sampling (see Botkin and Simpson 1990). 2) To assess community or guild structure, we recommend complete stem inventory and identification within a small, standard-radius circular plot (large enough to contain several stems of the target group) at each sample point. This would provide data at each point for each species of interest and would enable analysis of presence/absence patterns and local density. With stem measurement, population structure could be quantified for more abundant species and for the species group as a whole. If not provided by a GIS or other means, field data on the topographic position, slope, and soil characteristics should be taken in each plot, to directly link the plant distributions to local site conditions. This small-plot method would provide more quantitative and more site-controlled data than the techniques we used in the present study, at little additional field effort. Conclusions The findings from this study and from many other sites indicate that tropical wet forests are characterized by considerable internal heterogeneity in community structure. Much of this spatial variation is due to the frequent occurrence of edaphic mosaics (varying soil chemistry and texture, drainage, and topography) within small-to-large landscapes in these forests. Additional heterogeneity can derive from local human impacts, even in old-growth stands. Two challenging questions arise from this study. (1) What processes underly the nonuniform distributions of the large palms at La Selva? For example, what

factors lead to the opposite distributions of Euterpe and Prestoea? Possibilities include differential impacts of drainage conditions or nutrients on their ability to grow or survive in low light. More complex edaphic effects may produce the among-soil density changes in widespread species such as Welfia. What processes could produce the reciprocal densities of Socratea and Iriartea? Resolution of these issues will require field and shadehouse experiments (outplantings along soil/topographic gradients, soil trials with fertilization and watering; cf. Goldberg 1982, Burslem et al. 1994). (2) Does high internal spatial heterogeneity promote the exceptional species richness of tropical wet forests (cf. Ashton 1969)? Do these forests have relatively more plant species with strong edaphic associations and/or more pronounced edaphic variability than less speciesrich communities? Comparing these properties in temperate and tropical forests could be very revealing.
ACKNOWLEDGMENTS

We gratefullyacknowledgethe financialsupportprovided (BSR89-18185) and by the U.S. NationalScience Foundation the Andrew W. Mellon Foundation.Phil Sollins introduced us to La Selva's soil variability.Leonel Camposand William Brenes carriedout the field work and data input with skill and quality control. We also thank those interviewed about local uses of palms. We thank David Ackerly, Carol Augspurger,and Fred Scatena for their constructivereviews of the manuscript, JennyJuarezfor help with GIS mapping. and The La Selva GIS and reserve-widegrid were madepossible for by donationsto the Organization TropicalStudies (OTS) from the U.S. National Science Foundation,Environmental Systems Research Institute, the Andrew W. Mellon Foundation, and Sun Microsystems,Inc. We also thank OTS for on-going logistic support.
LITERATURE CITED

Anderson, A. B. 1990. Extraction and forest management by rural inhabitants in the Amazon estuary. Pages 65-85 in A. B. Anderson, editor. Alternatives to deforestation. Columbia University Press, New York, New York, USA. Ashton, P. S. 1969. Speciation among tropical forest trees: some deductions in the light of recent evidence. Biological Journal of the Linnean Society 1:155-196. Baillie, I. C., P. S. Ashton, M. N. Court, J. A. R. Anderson, E. A. Fitzpatrick, and J. Tinsley. 1987. Site characteristics and the distribution of tree species in Mixed Dipterocarp Forest on Tertiary sediments in central Sarawak, Malaysia. Journal of Tropical Ecology 3:201-220. Basnet, K. 1992. Effect of topography on the pattern of trees in Tabonuco (Dacryodes excelsa) dominated rain forest in Puerto Rico. Biotropica 24:31-42. Botkin, D. B., and L. G. Simpson. 1990. Biomass of the North American boreal forest. Biogeochemistry 9:161174. Burslem, D. F R. P., I. M. Turner, and P. J. Grubb. 1994. Mineral nutrient status of coastal hill dipterocarp forest and adinandra belukar in Singapore: bioassays of nutrient limitation. Journal of Tropical Ecology 10:579-599. Bush, M. B., and P. A. Colinvaux. 1994. Tropical forest disturbance: paleoecological records from Darien, Panama. Ecology 75:1761-1768. Chauvel, A., Y. Lucas, and R. Boulet. 1987. On the genesis of the soil mantle of the region of Manaus, Central Amazonia, Brazil. Experientia 43:234-241. Chazdon, R. L. 1985. The palm flora of Finca La Selva. Principes 29:74-78.

2592

DEBORAH A. CLARK ET AL.

Ecology, Vol. 76, No. 8

Cottam, G., and J. T. Curtis. 1956. The use of distance measures in phytosociological sampling. Ecology 37:451-460. Foster, R. B. 1990. The floristic composition of the Rio Manu floodplain forest. Pages 99-111 in A. H. Gentry, editor. Four neotropical rainforests. Yale University Press, New Haven, Connecticut, USA. Garcia-Montiel, D. C., and F N. Scatena. 1994. The effect of human activities on the structure and composition of a tropical forest in Puerto Rico. Forest Ecology and Management 63:57-78. Gentry, A. H. 1982. Patterns of neotropical plant species diversity. Evolutionary Biology 15:1-84. Gentry, A. H., and R. Ortiz S. 1993. Patrones de composicion floristica en la Amazonia Peruana. Pages 155-166 in R. Kalliola, M. Puhakka, and W. Danjoy, editors. Amazonia Peruana. PAUT and ONERN, Jyvaskyla, Finland. Goldberg, D. E. 1982. The distribution of evergreen and deciduous trees relative to soil type: an example from the Sierra Madre, Mexico, and a general model. Ecology 63: 942-951. Gomez-Pompa, A., and A. Kaus. 1990. Traditional management of tropical forests in Mexico. Pages 45-64 in A. B. Anderson, editor. Alternatives to deforestation. Columbia University Press, New York, New York, USA. Gordon, B. L. 1982. A Panama forest and shore. Boxwood Press, Pacific Grove, California, USA. Guillaumet, J.-L. 1987. Some structural and floristic aspects of the forest. Experientia 43:241-25 1. Hamburg, S. P., and R. L. Sanford, Jr. 1986. Disturbance, Homo sapiens, and ecology. Bulletin of the Ecological Society of America 67:169-17 1. Hartshorn, G. S. 1978. Tree falls and tropical forest dynamics. Pages 617-638 in P. B. Tomlinson and M. H. Zimmerman, editors. Tropical trees as living systems. Cambridge University Press, London, England. . 1983. Plants. Pages 118-157 in D. H. Janzen, editor. Costa Rican natural history. University of Chicago Press, Chicago, Illinois, USA. Hartshorn, G. S., and B. E. Hammel. 1994. Vegetation types and floristic patterns. Pages 73-89 in L. A. McDade, K. S. Bawa, H. A. Hespenheide, and G. S. Hartshorn, editors. La Selva: ecology and natural history of a neotropical rain forest. University of Chicago Press, Chicago, Illinois, USA. Horn, S. P., and R. L. Sanford, Jr. 1992. Holocene fires in Costa Rica. Biotropica 24:354-361. Hubbell, S. P., and R. B. Foster. 1986. Commonness and rarity in a neotropical forest: implications for tropical tree conservation. Pages 205-231 in M. E. Soule, editor. Conservation biology: the science of scarcity and diversity. Sinauer, Sunderland, Massachusetts, USA. Huston, M. A. 1980. Soil nutrients and tree species richness in Costa Rican forests. Journal of Biogeography 7:147157. Johnston, M. H. 1992. Soil-vegetation relationships in a tabonuco forest community in the Luquillo Mountains of Puerto Rico. Journal of Tropical Ecology 8:253-263. Jordan, C. E 1985. Soils of the Amazon forest. Pages 8394 in G. T. Prance and T. E. Lovejoy, editors. Amazonia. Pergamon, New York, New York, USA. Kahn, E, and A. de Castro. 1985. The palm community in a forest of Central Amazonia, Brazil. Biotropica 17:210216. Lescure, J. P., and R. Boulet. 1985. Relationships between soil and vegetation in a tropical rain forest in French Guiana. Biotropica 17:155-164. Levin, S. A. 1992. The problem of pattern and scale in ecology. Ecology 73:1943-1967. Lieberman, D., G. S. Hartshorn, M. Lieberman, and R. Per-

alta. 1990. Forest dynamics at La Selva Biological Station, 1969-1985. Pages 509-521 in A. H. Gentry, editor. Four neotropical rainforests. Yale University Press, New Haven, Connecticut, USA. Lieberman, M., D. Lieberman, G. S. Hartshorn, and R. Peralta. 1985. Small-scale altitudinal variation in lowland wet tropical forest vegetation. Journal of Ecology 73:505-516. McDade, L. A., and G. S. Hartshorn. 1994. La Selva Biological Station. Pages 6-14 in L. A. McDade, K. S. Bawa, H. A. Hespenheide, and G. S. Hartshorn, editors. La Selva: ecology and natural history of a neotropical rain forest. University of Chicago Press, Chicago, Illinois, USA. Newbery, D. McC., and J. Proctor. 1984. Ecological studies in four contrasting lowland rain forests in Gunung Mulu National Park, Sarawak. Journal of Ecology 72:475-493. Oliveira-Filho, A. T., E. A. Vilela, D. A. Carvalho, and M. L. Gavilanes. 1994. Effects of soils and topography on the distribution of tree species in a tropical riverine forest in south-eastern Brazil. Journal of Tropical Ecology 10:483508. Peres, C. A. 1994. Composition, density, and fruiting phenology of arborescent palms in an Amazonian terra firme forest. Biotropica 26:285-294. Phillips, O., A. H. Gentry, C. Reynel, P. Wilkin, and B. Galvez-Durand. 1994. Quantitative ethnobotany and Amazonian conservation. Conservation Biology 8:225-248. Pinard, M. 1993. Impacts of stem harvesting on populations of Iriartea deltoidea (Palmae) in an extractive reserve in Acre, Brazil. Biotropica 25:2-14. Ruokolainen, K., and H. Tuomisto. 1993. La vegetacion de terrenos no inundables (tierra firme) en la selva baja de la Amazonia Peruana. Pages 139-153 in R. Kalliola, M. Puhakka, and W. Danjoy, editors. Amazonia Peruana. PAUT and ONERN, Jyvaskyla, Finland. Salo, J., and M. Rasanen. 1989. Hierarchy of landscape patterns in western Amazon. Pages 35-45 in L. B. HolmNielsen, I. C. Nielsen, and H. Balslev, editors. Tropical forests: botanical dynamics, speciation and diversity. Academic Press, New York, New York, USA. Sancho M., E, and R. Mata Ch. 1987. Estudio detallado de suelos, Estaci6n Biol6gica "La Selva." Organization for Tropical Studies, San Jos6, Costa Rica. Sanford, R. L., H. E. Braker, and G. S. Hartshorn. 1986. Canopy openings in a primary neotropical lowland forest. Journal of Tropical Ecology 2:277-282. Sanford, R. L., Jr., P. Paaby, J. C. Luvall, and E. Phillips. 1994. Climate, geomorphology, and aquatic systems. Pages 19-33 in L. A. McDade, K. S. Bawa, H. A. Hespenheide, and G. S. Hartshorn, editors. La Selva: ecology and natural history of a neotropical rain forest. University of Chicago Press, Chicago, Illinois, USA. Sollins, P., E Sancho M., R. Mata Ch., and R. L. Sanford, Jr. 1994. Soils and soil process research. Pages 34-53 in L. A. McDade, K. S. Bawa, H. Hespenheide, and G. S. Hartshorn, editors. La Selva: ecology and natural history of a neotropical rainforest. University of Chicago Press, Chicago, Illinois, USA. ter Steege, H., V. G. Jetten, A. M. Polak, and M. J. A. Werger. 1993. Tropical rain forest types and soil factors in a watershed area in Guyana. Journal of Vegetation Science 4: 705-716. Uhl, N. W., and J. Dransfield. 1987. Genera palmarum. Allen Press, Lawrence, Kansas, USA. Williams, W. T., G. N. Lance, L. J. Webb, J. G. Tracey, and J. H. Connell. 1969. Studies in the numerical analysis of complex rain-forest communities. IV. A method for the elucidation of small-scale forest pattern. Journal of Ecology 57:635-654.

December 1995

TROPICAL PALM COMMUNITY STRUCTURE APPENDIX

2593

Density of the five abundant palm species by size and soil type. For the Alluvium, data are presented both for the overall soil type and separately by consociation (Experimental, Holdridge). Probabilities are for chi-square tests of the numbers of Closest Individuals that were and were not a given species, across soils. When total palm density does not differ among soils, this tests for variation in a given species' density across soils (see Methods). Only in the 1-5 m height class did total density vary among soils (see Results); for this size, P* is for a chi-square test (as above) of a species' representation among Closest Individuals on the three soil types of equivalent palm density (Residual, Arboleda, and Alluvium). Density (stems/ha) Height class (m): Soil type Residual Arboleda Stream Alluvium P Consociation Experimental Holdridge P Soil type Residual Arboleda Stream Alluvium P Consociation Experimental Holdridge P Soil type Residual Arboleda Stream Alluvium
P

1-5

>5-10 Euterpe macrospadix

>10-15

>15

45.0 17.1 28.0 6.6 0.001* 14.6 0.0

NS

6.7 3.7 0.0 0.0 <0.001 0.0 0.0 Prestoea decurrens

5.1 1.5 2.8 0.0 0.002 0.0 0.0

...

6.3 4.0 2.4 0.0 0.003 0.0 0.0

...

29.5 25.9 68.6 110.8 0.0006* 129.9 93.5

NS

5.4 11.3 20.4 18.9 0.0000 14.9 25.0

NS

0.3 1.5 2.3 1.5 0.0003 2.0 0.9

NS

.. ... ... ... ... ...

...

Socratea exorrhiza 50.9 121.0 18.7 45.8

0.001*

15.0 18.8 6.2 7.0

NS

6.6 8.2 3.9 11.1 0.002 16.2 4.4

0.0001

14.2 16.8 6.3 14.0

NS

Consociation Experimental Holdridge

P

57.7 34.9
NS

11.4 0.0
<0.01

23.4 5.2
0.0001

Iriartea deltoidea Soil type Residual Arboleda Stream Alluvium P Consociation Experimental Holdridge
P

83.9 17.1 31.3 25.9 0.0002* 0.0 46.7

NS

16.6 11.3 8.0 4.9 0.02 0.0 12.5

<0.01

9.5 5.2 5.5 6.5

NS

15.8 7.9 14.6 5.4

0.0001

1.4 13.2
0.0001

0.0 10.4
0.0000

2594 APPENDIX. Continued.

DEBORAH A. CLARK ET AL.

Ecology, Vol. 76, No. 8

Density (stems/ha) Height class (m): Soil type Residual Arboleda Stream Alluvium P Consociation Experimental Holdridge P 1-5 >5-10 Welfia georgii 177.0 199.0 53.2 319.3 0.02* 404.8 245.5
NS

>10-15

>15

29.5 35.8 21.3 23.1

NS

14.8 16.3 10.6 10.7

NS

8.6 14.8 8.3 15.8

0.0000

21.7 25.0
NS

8.8 13.2
NS

15.9 15.6
NS