SEARCH

SEARCH BY CITATION

Keywords:

  • community ecology;
  • dispersal assembly;
  • Panama;
  • seed addition experiment;
  • seed limitation;
  • seed size;
  • seedling ecology;
  • seedling recruitment;
  • tropical ecology;
  • tropical forest

Summary

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  • 1
    The role of seed limitation in tropical forests remains uncertain owing to the scarcity of experimental evidence. We performed seed addition experiments to assess seed limitation for 32 shade-tolerant tropical forest species and monitored the natural seed rain of 25 of these species for 17 years.
  • 2
    One, two or five seeds were sown into 0.0079-m2 plots for large- (n = 5 species), medium- (n = 5) and small-seeded species (n = 22), respectively. The experiment was replicated at 69 sites, placed in groups of three at 23 locations. Seedling establishment was evaluated after 1 and 2 years in paired seed addition and control plots. Natural seedling emergence and understorey plant density were also measured.
  • 3
    Median natural seed rain was 0.31 seeds m−2 year−1 per focal species.
  • 4
    Seed addition enhanced seedling establishment in 31 and 26 of the 32 species after 1 and 2 years, respectively. Mean number of focal species’ seedlings after 2 years was 0.002 seedlings in control plots and 0.12, 0.37 and 0.60 seedlings in seed addition plots for large-, medium- and small-seeded species, respectively.
  • 5
    A 25 seeds added treatment increased seedling establishment by ≥ 2.0-fold over the five seeds added treatment after 2 years.
  • 6
    Community-wide recruitment and understorey plant density were strongly seed-limited. The natural density of understorey plants averaged 12 plants m−2 and was significantly less than for seedlings of the single focal species in plots with ≥ 2 seeds added 2 years earlier.
  • 7
    The number of established seedlings per seed added was independent of seed size.
  • 8
    Treatment (adding zero or five seeds), species identity and location all affected seedling establishment for the 11 small-seeded species represented at all sites, with treatment and its interactions accounting for 86% of the explained variation.
  • 9
    Our results suggest that seed limitation plays a dominant role in seedling recruitment and understorey plant community assembly in tropical forests. Although strong seed limitation may set the stage for species-neutral community assembly, the species differences in seedling establishment rate and its spatial variation demonstrate an important role for species-specific processes.

Introduction

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Seed limitation refers to the failure of seeds of a given plant species to arrive at suitable sites in sufficient numbers to establish as seedlings (Nathan & Muller-Landau 2000; Levin et al. 2003). Seed addition experiments have demonstrated that the failure of seeds to arrive limits recruitment in temperate herb communities (Tilman 1997; Ehrlén & Eriksson 2000; Rand 2000; Turnbull et al. 2000; Foster & Tilman 2003; Foster et al. 2004). Seed limitation may limit the scope for pairwise competitive interactions between species because direct competition between sessile plants is restricted to near neighbours whose zones of resource depletion overlap. In extreme cases, seed limitation may also limit competition among all individuals by keeping total plant density at low levels (Wright 2002). Thus, seed limitation may contribute to the maintenance of species richness by slowing competitive dynamics (Hurtt & Pacala 1995; Hubbell et al. 1999; Hubbell 2001; Schupp et al. 2002).

In contrast to temperate herbaceous communities, the role of local-scale seed limitation is unclear in tropical forests. Makana & Thomas (2004) performed seed addition experiments for two light-demanding African tree species and found that both were strongly seed-limited. Indirect evidence for seed limitation comes from seed trap studies, which suggest strong seed limitation in many tropical forest plant species (Hubbell et al. 1999; Dalling et al. 2002; Murray & Garcia-C. 2002; Terborgh et al. 2002). Additional evidence for seed limitation is provided by autocorrelation in the spatial distribution of tropical plants that occurs independently of environmental variation (Dalling et al. 1998; Valencia et al. 2004) and by correlations between seedling distributions and seed rain density predicted from seed trap studies (Dalling et al. 2002). These indirect measures cannot provide conclusive evidence for seed limitation, however, because it is unclear which sites are suitable for germination and seedling establishment (Levine & Murrell 2003). Furthermore, community-level seed production can be very high in tropical forests. For example, seed rain density averages 965 seeds m−2 year−1 for seeds with minimum dimension larger than 1.0 mm on Barro Colorado Island, Panama (Harms et al. 2000) and ≥ 500 seeds m−2 year−1 for seeds with minimum dimension larger than 1.5 mm at Cocha Cashu, Peru (Terborgh et al. 2002). Hence, it is far from clear whether seed input limits total seedling recruitment in tropical forests (Terborgh et al. 2002; Wright 2002). In fact, Webb & Peart (2001) argued that community-wide seed dispersal in an Indonesian forest approached the level at which seed dispersal limitation is unimportant in curtailing competitive exclusion.

Here, we test directly for local-scale seed limitation for 32 shade-tolerant species in old-growth tropical moist forest in Panama using seed sowing experiments (Turnbull et al. 2000). We asked the following questions:

  • 1
    Is seedling recruitment of individual species seed-limited?
  • 2
    Is seedling recruitment of the plant community as a whole seed-limited?
  • 3
    What is the relative importance of species identity, location, seed limitation and their interactions as controls of seedling recruitment? An important role for location or a location-by-species interaction would provide evidence that not all sites are equally suitable for germination and seedling establishment of all shade-tolerant species.

Materials and methods

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

study site and species

Barro Colorado Island (BCI; 9°09′ N, 79°51′ W) supports tropical semi-deciduous forest (Croat 1978; Foster & Brokaw 1996; Leigh 1999). Annual precipitation averages c. 2600 mm, and the climate is seasonally dry. Elevation ranges between 27 and 160 m a.s.l. About half the forest is old-growth, at least parts of which have escaped agriculture for 2000 years or more (Leigh 1999). Detailed descriptions of the climate, geology and biota of BCI can be found in Croat (1978), Gentry (1990), Leigh et al. (1996) and Leigh (1999). The 32 study species include 18 families and a wide range of growth forms and all 32 are relatively shade-tolerant (Table 1). Nomenclature follows Croat (1978) as updated by Condit et al. (1995) with two additional changes: Cephaelis ipecacuanha and Virola surinamensis have been renamed as Psychotria ipecacuanha (Brot.) Stokes and V. nobilis A. C. Sm., respectively.

Table 1.  Descriptions of the 32 study species, including family (names lack the -aceae suffix), growth form (GF: H = herb, L = liana, S = shrub, U = treelet, M = medium-sized tree, T = large tree), seed dry mass and natural seed rain quantified for 17 years and 200 0.5-m2 traps as population-level seed production (the mean ± SD number of seeds captured standardized to a square metre basis) and seed arrival (the number of traps that captured one or more seeds). NA = data not available
SpeciesFamilyGFSeed mass (g dry weight)Seed production (m−2 year−1)Seed arrival (no. of traps)
  • *

    Seeds were never captured for the seven study species that are always shorter than the 80-cm-high traps as adults.

Beilschmiedia pendulaLaur-T 3.8761 0.78 ± 0.74179
Brosimum alicastrumMor-T 0.6472 5.20 ± 3.99172
Calophyllum longifoliumClusi-T 1.9501 0.07 ± 0.07 87
Chrysochlamys eclipesClusi-S 0.0962 0.00 ± 0.00  4
Cupania sylvaticaSapind-U 0.1187 0.02 ± 0.03 51
Eugenia nesioticaMyrt-M 0.3465 0.01 ± 0.02 36
Faramea occidentalisRubi-U 0.1962 8.89 ± 12.32200
Geophila repensRubi-HNA**
Gustavia superbaLecythid-M 4.8342 0.62 ± 0.60144
Herrania purpureaSterculi-U 0.1847 0.01 ± 0.04  2
Hybanthus prunifoliusViol-S 0.018211.63 ± 8.07200
Lacistema aggregatumLacistem-U 0.0107 0.01 ± 0.03 12
Mouriri myrtilloidesMelastomat-S 0.0447 0.42 ± 0.27184
Ocotea whiteiLaur-TNA 0.08 ± 0.13 59
Oenocarpus maporaArec-M 1.3327 0.91 ± 0.70199
Paullinia turbacensisSapind-L 0.0925 0.10 ± 0.11115
Pharus latifoliusPo-HNA**
Pouteria reticulataSapot-T 0.4310 0.31 ± 0.41166
Prioria copaiferaFab-T22.6387 0.17 ± 0.11152
Psychotria deflexaRub-S 0.0026 0.01 ± 0.02  3
Psychotria emeticaRub-S 0.0030**
Psychotria horizontalisRub-S 0.0040 1.54 ± 1.06185
Psychotria ipecacuanhaRub-S 0.0060**
Psychotria limonensisRub-S 0.0066 0.01 ± 0.03  6
Psychotria marginataRub-S 0.0075 0.01 ± 0.01 19
Sorocea affinisMor-S 0.0777 0.40 ± 0.41153
Streptochaeta sodiroanaPo-HNA**
Streptochaeta spicataPo-HNA**
Streptogyne americanaPo-HNA**
Tabernaemontana arboreaApocyn-T 0.0694 0.86 ± 0.78189
Trichilia tuberculataMeli-T 0.119622.75 ± 22.66200
Virola nobilisMyristic-T 1.6382 0.45 ± 0.16153

experimental design

The rain of seeds and flowers was censused weekly from 1 January 1987 to 21 May 2003, using 200 seed traps set along 2.7 km of trails within a 50-ha forest dynamics plot (Wright & Calderón 1995; Wright et al. 2005). Each seed trap consisted of a square, 0.5-m2 PVC frame supporting a shallow, open-topped, 1-mm mesh bag, and suspended 0.8 m above the ground on four PVC posts. Traps were located at 13.5-m intervals on alternating sides of the trail and randomly between 4 m and 10 m from the trail so that distances between nearest traps averaged 18.9 ± 3.6 m (± 1 SD). All flowers, seeds, fruits, capsules and other reproductive parts of plants that fell into the traps were identified to species level and counted (only presence was recorded for flowers). Fruits and seeds were further categorized as aborted, immature, mature (endosperm filled), or damaged by insects or vertebrates. The number of undamaged, mature fruit was multiplied by the species-specific average seed : fruit ratio and added to the number of undamaged seeds to estimate the total number of viable seeds falling into each trap for each species.

Seeds were sown into 0.0079-m2 plots between March and October 2000 as each species matured fruit. Just one species was sown into each plot. Seedling establishment was evaluated near the end of the next two wet seasons in October–November 2001 and again in November–December 2002. Number of leaves and stem height (measured from the ground to the tip of stem) were recorded for the tallest seedling of the focal species in each plot during the 2001 and 2002 censuses. We refer to the period between sowing and the 2001 census as the first year and the period between the 2001 and 2002 censuses as the second year.

The experiment was replicated at 69 closed-canopy sites in old-growth forest. The 69 sites occurred in groups of three at 23 locations, sited every 100 m along the trails in old-growth forest in the south-western part of BCI. At each location three replicate sites were chosen, essentially at random, except that all sites were at least 3 m from the nearest trail, sites were separated from one another by 2–5 m, and canopy gaps sensuBrokaw (1982) and large palms were avoided. Although much topographic and edaphic variation was sampled, including the two major soil types on BCI (Leigh 1999), using the trail system introduced some bias away from steep slopes and swamps (Svenning et al. 2004). At each site, paired seed addition and control plots separated by 15 cm were established for each species. Each seed addition and control plot was delimited by a 10-cm-diameter circle of 5-cm-high aluminium mesh to prevent seeds from being washed away by rain. We found no indications that the mesh reduced seed predation. Unknown vertebrates removed a large fraction of the seeds weighing ≥ 0.3 g added to our experimental plots within a few days (J.-C.S., personal observation). Furthermore, in a pilot study ants were observed climbing the mesh to retrieve aril-coated seeds of Paullinia turbacensis placed inside plots. Finally, total annual seedling emergence rates in the control plots (see Results) were similar to those observed in other studies on BCI (c. 10–40 seedlings m−2 in Daws et al. in press). Several species were seeded into fewer than 69 sites (Table 2; range = 12–69 sites, average = 54.6). One, two and five seeds were added per plot for large- (2.0–22.6 g dry weight), medium- (0.3–1.6 g) and small- (0.03–0.2 g) seeded species, respectively. The few species with no available seed mass data (Table 1) were placed in a seed mass class based our field experience. For all study species seed dimensions are given directly or can be inferred from the species descriptions in Croat (1978). An additional plot with 25 seeds added was established for seven of the small-seeded species at 29–60 sites (average = 40). Seeds were collected distant from the experiment and were dropped into the experimental plots within 1–3 days. Plots were placed to avoid all naturally occurring plants and were maintained by removing naturally emerging seedlings of non-focal species and loosening litter suspended on the aluminium mesh during the first year. Naturally emerging seedlings were identified, counted and removed approximately monthly during the first year, except during the dry season (January–April 2001) when seedling emergence was negligible. The object of this weeding was to be able to evaluate the limitation of recruitment by seed arrival without confounding competitive interactions. Weeding turned out to have little or no impact on focal species’ recruitment (see Results).

Table 2.  The outcome of the seed addition experiments 1 and 2 years after sowing (plain and bold type face, respectively). n refers to the number of paired seed addition (SA) and control (C) plots. Seeds refer to the number of seeds added to each SA plot. Seedling presence refers to the number of SA plots with focal species present as a seedling. Stem height and number of leaves are given as means (± 1 SD) for the largest focal species’ seedling in each SA plot. Number of seedlings is the mean (± 1 SD) number of seedlings present for the SA and C plots. P is the significance level for a one-tailed sign test comparing number of seedlings for paired SA and C plots, except for the 25-seed addition treatments where the sign test compares paired 25-seed and five-seed SA plots
SpeciesnSeedsSeedling presence (no. SA plots)Stem height (cm)No. of leavesNo. of seedlings per SA plotNo. of seedlings per C plotP
Beilschmiedia pendula69 11423.8 ± 9.5 3.6 ± 3.10.20 ± 0.410.01 ± 0.120.00049
 934.5 ± 12 6.6 ± 3.90.13 ± 0.340.00 ± 0.000.0020
Brosimum alicastrum69 2 715.0 ± 7.1 2.9 ± 2.20.13 ± 0.420.00 ± 0.000.0078
 617.5 ± 6.9 2.7 ± 1.20.10 ± 0.350.00 ± 0.000.016
Calophyllum longifolium69 11028.4 ± 10.2 5.3 ± 3.10.14 ± 0.350.00 ± 0.000.00098
 645.1 ± 23.010.3 ± 6.70.09 ± 0.280.00 ± 0.000.016
Chrysochlamys eclipes20 5 1 4.3 40.05 ± 0.220.00 ± 0.000.50
 1 6.2 60.05 ± 0.220.00 ± 0.000.50
Cupania sylvatica29 521 8.3 ± 1.9 3.5 ± 2.40.90 ± 1.140.00 ± 0.000.00003
16 8.9 ± 3.0 4.8 ± 1.90.62 ± 0.940.00 ± 0.000.00024
292521 8.7 ± 1.9 3.0 ± 1.92.76 ± 2.680.00 ± 0.000.00002
16 9.9 ± 2.1 5.0 ± 2.01.76 ± 2.100.00 ± 0.000.0037
Eugenia nesiotica23 21419.4 ± 4.015.6 ± 6.61.00 ± 0.900.00 ± 0.000.00005
1324.8 ± 4.020.8 ± 11.40.91 ± 0.900.00 ± 0.000.00012
Faramea occidentalis69 542 8.9 ± 3.0 2.0 ± 1.11.48 ± 1.580.04 ± 0.219 × 10−13
2310.8 ± 3.6 5.3 ± 2.30.90 ± 1.540.03 ± 0.175 × 10−7
Geophila repens69 537 3.6 ± 4.9 4.4 ± 2.80.84 ± 0.950.00 ± 0.007 × 10−12
2014.5 ± 30.516.4 ± 43.40.39 ± 0.670.00 ± 0.009 × 10−7
Gustavia superba69 11119.6 ± 4.5 5.5 ± 2.80.16 ± 0.370.00 ± 0.000.00049
 729.1 ± 5.3 9.7 ± 5.20.10 ± 0.300.00 ± 0.000.0078
Herrania purpurea55 53413.3 ± 3.1 4.1 ± 1.91.56 ± 1.620.00 ± 0.006 × 10−11
2815.5 ± 4.3 4.1 ± 1.51.05 ± 1.280.00 ± 0.004 × 10−9
Hybanthus prunifolius60 533 6.1 ± 2.1 4.4 ± 1.90.92 ± 1.060.00 ± 0.001 × 10−10
20 9.2 ± 4.2 5.6 ± 2.60.52 ± 0.850.00 ± 0.001 × 10−6
602535 7.5 ± 2.8 5.6 ± 2.62.78 ± 3.450.00 ± 0.000.0069
2611.3 ± 4.6 5.1 ± 2.91.42 ± 2.250.00 ± 0.000.012
Lacistema aggregatum39 531 8.7 ± 5.7 6.6 ± 3.82.21 ± 1.660.00 ± 0.005 × 10−10
2715.2 ± 13.0 9.9 ± 7.61.74 ± 1.570.00 ± 0.007 × 10−9
302525 8.7 ± 3.4 6.4 ± 2.15.83 ± 4.960.00 ± 0.000.00077
2113.3 ± 10.3 8.2 ± 3.93.13 ± 2.830.00 ± 0.000.0085
Mouriri myrtilloides69 51910.9 ± 2.510.7 ± 6.70.67 ± 1.290.00 ± 0.002 × 10−6
1615.0 ± 3.528.5 ± 18.40.48 ± 0.990.00 ± 0.000.00002
Ocotea whitei69 12142.7 ± 17.2 8.6 ± 11.60.30 ± 0.460.00 ± 0.005 × 10−7
1455.8 ± 28.730.9 ± 62.60.20 ± 0.410.00 ± 0.000.00006
Oenocarpus mapora69 218 5.9 ± 0.91 1.9 ± 0.20.43 ± 0.780.00 ± 0.004 × 10−6
18 6.8 ± 1.5 3.1 ± 1.10.42 ± 0.760.00 ± 0.004× 10−6
Paullinia turbacensis12 5 911.2 ± 3.6 5.3 ± 2.51.50 ± 1.310.00 ± 0.000.0020
 913.4 ± 4.9 7.0 ± 2.21.50 ± 1.310.00 ± 0.000.0020
Pharus latifolius69 533 5.4 ± 2.8 5.1 ± 2.10.99 ± 1.280.01 ± 0.122 × 10−9
19 9.1 ± 6.4 6.6 ± 3.50.41 ± 0.750.00 ± 0.002 × 10−6
Pouteria reticulata69 23314.3 ± 2.8 6.1 ± 3.00.65 ± 0.760.00 ± 0.001 × 10−10
2516.3 ± 4.0 8.3 ± 5.40.42 ± 0.600.00 ± 0.003 × 10−8
Prioria copaifera68 1 853.2 ± 24.9 2.8 ± 2.00.12 ± 0.320.00 ± 0.000.0039
 463.6 ± 33.4 7.3 ± 3.40.06 ± 0.240.00 ± 0.000.063
Psychotria deflexa69 519 3.5 ± 1.1 4.5 ± 1.60.48 ± 0.880.00 ± 0.002 × 10−6
 4 7.7 ± 2.2 7.5 ± 3.50.09 ± 0.370.00 ± 0.000.063
Psychotria emetica69 523 2.4 ± 0.8 2.7 ± 1.40.42 ± 0.720.00 ± 0.001 × 10−7
 00.00 ± 0.000.00 ± 0.00
Psychotria horizontalis69 535 2.9 ± 7.9 3.3 ± 1.30.94 ± 1.170.01 ± 0.123 × 10−11
 7 5.1 ± 1.3 6.9 ± 1.90.17 ± 0.590.01 ± 0.120.035
Psychotria ipecacuanha69 516 2.7 ± 0.9 2.9 ± 1.80.30 ± 0.600.00 ± 0.000.00002
 3 7.3 ± 0.8 9.0 ± 2.00.04 ± 0.210.00 ± 0.000.13
Psychotria limonensis69 520 5.6 ± 2.9 6.1 ± 3.20.52 ± 0.950.01 ± 0.121 × 10−6
1311.8 ± 7.2 7.8 ± 2.80.33 ± 0.820.01 ± 0.120.00012
542523 5.9 ± 2.8 5.8 ± 2.61.65 ± 3.190.02 ± 0.140.061
1810.1 ± 4.9 6.9 ± 2.90.72 ± 1.280.02 ± 0.140.032
Psychotria marginata41 511 4.6 ± 1.7 6.7 ± 4.10.51 ± 0.980.00 ± 0.000.00049
 7 7.5 ± 3.5 7.6 ± 2.80.24 ± 0.580.00 ± 0.000.0078
352516 4.0 ± 1.6 4.9 ± 3.11.54 ± 3.110.00 ± 0.000.032
 6 6.1 ± 1.9 7.7 ± 3.90.63 ± 1.930.00 ± 0.000.62
Sorocea affinis69 542 7.6 ± 2.2 2.5 ± 1.31.39 ± 1.430.00 ± 0.002 × 10−13
28 9.6 ± 2.8 4.0 ± 1.80.80 ± 1.110.00 ± 0.004 × 10−9
Streptochaeta sodiroana54 529 4.6 ± 2.0 4.8 ± 2.21.28 ± 1.460.00 ± 0.002 × 10−9
19 7.2 ± 3.0 5.7 ± 2.90.63 ± 0.980.00 ± 0.002 × 10−6
302521 6.1 ± 2.1 3.9 ± 1.65.40 ± 5.360.00 ± 0.000.0022
17 7.4 ± 2.2 5.9 ± 2.72.00 ± 2.550.00 ± 0.000.011
Streptochaeta spicata30 518 7.8 ± 3.6 7.7 ± 2.91.33 ± 1.450.00 ± 0.004 × 10−6
1611.9 ± 6.313.8 ± 7.11.07 ± 1.260.00 ± 0.000.00002
Streptogyne americana60 536 4.1 ± 1.7 3.7 ± 1.61.27 ± 1.330.02 ± 0.131 × 10−11
23 7.8 ± 6.9 5.9 ± 3.10.77 ± 1.180.02 ± 0.132 × 10−7
422529 4.1 ± 1.5 3.3 ± 1.44.14 ± 5.280.02 ± 0.150.00009
18 8.0 ± 5.2 6.0 ± 4.52.17 ± 3.810.02 ± 0.150.0059
Tabernaemontana arborea48 51814.4 ± 5.4 5.8 ± 3.30.75 ± 1.160.00 ± 0.004 × 10−6
1417.6 ± 6.2 7.0 ± 3.70.54 ± 0.990.00 ± 0.000.00006
Trichilia tuberculata69 537 9.2 ± 1.5 4.4 ± 1.71.39 ± 1.650.00 ± 0.007 × 10−12
3210.3 ± 2.1 7.2 ± 2.50.96 ± 1.290.00 ± 0.002 × 10−10
Virola nobilis69 21415.2 ± 5.8 2.1 ± 2.00.23 ± 0.490.00 ± 0.000.00006
 130.0 0.00.01 ± 0.120.00 ± 0.000.50

To estimate natural densities of plants in the understorey (defined as plants with d.b.h. < 5 cm), we censused all plants in 10-cm-diameter plots located randomly c. 1.5 m away from each Hybanthus prunifolius and Faramea occidentalis experiment (n = 138 ‘natural density’ plots) during the 2001 census. A small flag was thrown backwards over one shoulder to locate the centre of each plot randomly with the caveat that plots were not allowed to contain large trees.

data analyses

We used sign tests to assess species-level seed limitation by comparing the number of seedlings established in paired control and seed addition plots for each focal species (Zar 1996). Sign tests were used because observed values were highly skewed and covered a limited range. We also used sign tests to assess community-wide seed limitation by comparing the total number of seedlings (focal species’ seedlings established after 1 year plus all other non-focal species’ seedlings emerging during the first year) in paired control and seed addition plots for each of the four seed addition levels (i.e. one, two, five or 25 seeds added). The sign tests were one-tailed given the expectation that seed addition could raise, but not lower, seedling recruitment. Probabilities were computed using binomial coefficients for the species-level tests and a normal approximation for the community-wide tests because of the larger sample size (Zar 1996). We also evaluated community-wide seed limitation by comparing the natural densities of all herbaceous and woody plants in the 138 ‘natural density’ plots with the densities of focal species’ seedlings for the four seed addition levels using Wilcoxon two-sample tests.

We evaluated the effect of weeding by comparing the number of focal species’ seedlings establishing in seed addition plots with and without naturally emerging seedlings using a one-way anova; by definition only seed addition plots with naturally emerging seedlings were weeded. We assessed the effect of seed mass on the mean number of seedlings produced per seed using a second anova to compare species from the three seed size categories described previously (see Experimental design above). The 32 species were replicates for the analysis of seed mass.

We determined the relative effects of seed addition (zero or five seeds added), species identity (n = the 11 species sown at all 69 sites), location (n = 23) and their interactions on the number of focal species’ seedlings established per plot after 1 year using a three-way anova. Species and treatment were treated as fixed effects and location as a random effect. We computed F-ratios following Quinn & Keough (2002) and variance components using the estimated mean square approach (Quinn & Keough 2002). We note that variance components for fixed and random factors have different interpretations, namely as the estimated variance between the group means for the specific populations sampled and between all possible group means, respectively (Quinn & Keough 2002). To enhance comparability between the variance components one could also treat location as a fixed factor. However, this changes the individual variance components by < 0.3% (data not shown). The limited numbers of seedlings in the one- and two-seed addition plots prohibited analyses of variation in seedling numbers per plot for the medium- and large-seeded species. For anova, the response variables were square-root transformed to improve normality and residual plots were investigated as a check thereof. Analyses were performed in Microsoft® Office Excel 2003 (Microsoft Corporation, Washington, USA) and JMP 5.0 (SAS Institute Inc. 2000).

Results

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

natural seed rain

The natural seed rain ranged from 0.00 m−2 year−1 to 22.8 m−2 year−1 (median = 0.31) for the 25 focal species that, as adults, are taller than the 80-cm-high traps (Table 1). This contrasts strongly with experimental seed densities of 1–25 seeds per 0.0079 m2. The mean number of traps where seeds of a species arrived was closely related to mean seedfall density (Fig. 1). After 17 years, the cumulative percentage of the 200 possible traps reached by seeds varied from 0% to 100% among the 25 focal species (median = 152; Table 1).

image

Figure 1. Seed arrival increases with population-level seed production. Seed arrival on the vertical axis is the mean number of 0.5-m2 traps (out of 200 in total) that captured a seed of the species. Seed production on the horizontal axis is the mean number of seeds captured standardized to a hectare basis. Means are for 17 years. Symbols represent the 25 study species that are taller than the 80-cm tall traps. The seven remaining study species are always shorter than the traps and were never captured as seeds. Two outliers, Psychotria limonensis and P. deflexa, bear many fruit below 80 cm. The final outlier, Herrania purpurea, is a small cauliflorous treelet that often displays its fruit below 80 cm.

Download figure to PowerPoint

species-level seed limitation

Seed addition enhanced seedling establishment significantly in 31 of 32 species after 1 year and in 26 of 32 species after 2 years (Table 2). The one species that showed no evidence of seed limitation, Chrysochlamys eclipes, produces many sterile seeds (S.J.W., personal observation), and these would not have been distinguished before sowing. The mean number of focal species’ seedlings (± 1 SD) per 0.0079 m2 seed addition plots was 0.19 ± 0.07 (n = 5 species), 0.49 ± 0.35 (n = 5 species) and 0.99 ± 0.51 (n = 22 species) seedlings for large-, medium- and small-seeded (five-seed treatment) species after 1 year and 0.12 ± 0.05, 0.37 ± 0.35 and 0.60 ± 0.47 seedlings after 2 years, respectively. In contrast, mean seedling number of single focal species was only 0.003 ± 0.009 and 0.002 ± 0.007 seedlings in the control plots (n = 32 species) after 1 and 2 years, respectively. The 25-seed addition treatment increased seedling establishment over the five-seed addition by 2.7-fold to 3.7-fold (mean increase 3.6-fold ± 0.38 SD) after 1 year and by 2.0-fold to 3.4-fold (mean increase 2.5-fold ± 0.51 SD) after 2 years (Fig. 2), with the one-tailed sign test significant at P < 0.05 for six of seven species for both censuses (Table 2).

image

Figure 2. Mean focal species’ seedling densities in control (white bars), five-seed addition (light grey) and 25-seed addition (dark grey) plots for Cupania sylvatica (n = 29 plots per treatment), Hybanthus prunifolius (n = 60), Lacistema aggregatum (n = 30), Psychotria limonensis (n = 54), P. marginata (n = 35), Streptochaeta sodiroana (n = 30) and Streptogyne americana (n = 42). For each species, seedling densities after 1 or 2 years are given by three bars to the left or right, respectively. Note that the white bars are absent or very small owing to the zero or low densities in the control plots. Only sites with all three treatments represented were included. Error bars give the standard error of the mean.

Download figure to PowerPoint

Most seedlings that established in the seed addition plots survived and grew during the second year (Table 2). Many achieved impressive stem lengths, the largest being 135 cm in Ocotea whitei, 100 cm in Prioria copaifera, 88 cm in Calophyllum longifolium and 64 cm in Beilschmiedia pendula (all large-seeded canopy species), 57 cm in the treelet Lacistema aggregatum and 140 cm in the trailing herb Geophila repens (both small-seeded).

Weeding had little impact on focal species’ recruitment. No wildings emerged during the first year in 70% of the one-seed, 74% of the two-seed, 77% of the five-seed and 67% of the 25-seed addition plots where focal species’ seedlings established (n = 64, 86, 558 and 170 plots, respectively). Furthermore, the number of focal species’ seedlings in these plots did not differ significantly from the number of focal species’ seedlings in plots where non-focal species’ seedlings did emerge and were subsequently removed by weeding (anova, r2 = 0.005, P = 0.08 and r2 = 0.003, P = 0.46 for the five- and 25-seed addition plots, respectively). The number of naturally emerging seedlings was too limited to allow similar analyses for the one- and two-seed addition plots.

community-wide seed limitation

The total number of seedlings (focal species’ seedlings established after 1 year plus all other non-focal species’ seedlings emerging during the first year) was greater in the seed addition plots than in the control plots for all treatments [proportion of non-zero differences with greater number of seedlings in the seed addition plot: 59% (one-tailed sign test P < 0.05), 77% (P < 0.0005), 82% (P < 0.0005), 87% (P < 0.0005) for the one-, two-, five- and 25-seed addition treatments, n = 165, 149, 761 and 218 non-zero differences, respectively; Table 3].

Table 3.  Evidence for community-wide seed limitation. Entries are the mean number of seedlings (1 SD) that recruited and survived until the first census (2001) for the focal species in the seed addition (SA) plots, that emerged during that same year for all non-focal species in the SA plots and that emerged during that same year for all species in the control plots. Means and SDs were calculated over all species in each seed size category
Plot typeSeedling typeSeed size (no. of seeds added)
Small (5)Small (25)Medium (2)Large (1)
SAFocal species   0.96 (1.30)  3.22 (4.27)  0.41 (0.70)  0.19 (0.39)
SAOther species   0.36 (0.80)  0.52 (0.90)  0.35 (0.90)  0.38 (1.22)
ControlAll species   0.36 (0.77)  0.48 (0.96)  0.30 (0.73)  0.38 (0.62)
No. of seeds added    5 25  2  1
No. of SA plots of each type 1207280299344

Further evidence for community-wide seed limitation comes from comparisons of natural densities of all herbaceous and woody plants (0.094 ± 0.399 plants per 0.0079-m2 plot, mean ± SD calculated for 138 plots) with the densities of focal species’ seedlings present in seed addition plots. Even after 2 years, focal species’ seedling densities were significantly greater than natural densities of all understorey plants for the two-seed (mean 0.291 ± 0.607 seedlings per 0.0079-m2 plot, Wilcoxon's two-sample test, P = 0.0001, n = 299 plots), five-seed (mean 0.544 ± 1.030, P < 0.0001, n = 1207 plots) and 25-seed (mean 1.579 ± 2.551; P < 0.0001, n = 280 plots) addition, but not for the one-seed (mean 0.116 ± 0.321, P = 0.11, n = 344 plots) addition treatment. Natural densities are lower than natural emergence rates owing to the death of many seedlings during their first year post-germination.

relative importance of seed limitation, species identity and location

The mean number of seedlings produced per seed added did not differ among the three seed size categories [Table 4; F2,29 = 0.22, P = 0.81 for 2001 and F2,29 = 0.36, P = 0.70 for 2002 for one-way anovas comparing the mean number of seedlings per seed among the five large-, five medium- and 22 small-seeded species (five-seed treatment)].

Table 4.  The number of focal species’ seedlings per seed added (mean ± 1 SD) 1 and 2 years after sowing for small- (five-seed treatment), medium- and large-seeded species. Means and standard deviations are calculated over species
 20012002
Small seeds (22 spp.)0.20 ± 0.100.12 ± 0.09
Medium seeds (five spp.)0.24 ± 0.170.19 ± 0.18
Large seeds (five spp.)0.19 ± 0.070.12 ± 0.05

Treatment (adding zero or five seeds), species identity, location and their interactions collectively explained a large proportion of the variation in the number of focal species’ seedlings established after 1 year for the 11 small-seeded species sown at all 69 sites (r2 = 62.0%). All main effects and interactions contributed significantly to the explained variance (Table 5). Treatment alone explained 60.6% of the explained variance; all interactions involving treatment explained another 25.3%; and the species, location and species–location interaction collectively explained just 14.2% of the explained variance (Table 5).

Table 5.  Three-way factorial mixed-effects analysis of variance of the number of focal species’ seedlings (square-root transformed) per 10-cm-diameter plot (n = 1518) in relation to treatment (fixed effect: zero or five seeds added), species identity (fixed effect: the 11 small-seeded species represented at all 69 sites, n = 11) and location (random effect; n = 23) 1 year after seed addition. Variance components were computed using the estimated mean square approach (Quinn & Keough 2002)
 d.f.Sum of squaresF ratioEstimated variance component (%)
  • ***

    P < 0.001;

  • ****

    P < 0.0001.

Species  10 20.34  6.88**** 2.7
Location  22 17.56  4.06**** 2.0
Treatment   1124.87156.60****35.0
Species × Location 220 65.06  1.50**** 3.5
Species × Treatment  10 18.42  6.61**** 4.9
Location × Treatment  22 17.54  4.05**** 3.9
Species × Location × Treatment 220 61.28  1.42*** 5.8
Error1012199.08 42.2

Discussion

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

This seed addition experiment demonstrated that seedling recruitment was strongly seed-limited for 31 of 32 shade-tolerant species in an old-growth tropical moist forest and that subsequent seedling survival was sufficiently high that significant effects of added seeds persisted for 2 years for 26 species. The experiment also produced strong evidence that the plant community as a whole was severely seed-limited in terms of both seedling recruitment and total understorey plant density. The total of number of focal species’ seedlings surviving 1 year plus all non-focal species’ seedlings emerging during the first year was significantly increased by all seed addition treatments. Furthermore, the average focal species’ seedling density when 2–25 seeds were added per 0.0079-m2 plot (≥ 37 seedlings m−2 after 2 years) greatly exceeded average natural understorey plant densities on BCI (12 plants m−2, this study) and in other lowland neotropical forests [Cocha Cashu, Peru: 20 plants m−2 (Terborgh et al. 2002); La Selva, Costa Rica: 9–15 seedlings m−2 (Marquis et al. 1986)]. Seed arrival clearly affects the species composition as well as the overall density of seedlings in the tropical forest understorey. Because seed arrival provides the spatial template for processes occurring in later ontogenetic stages, strong seed limitation of seedling recruitment is likely to have profound consequences for the adult plant community as well, although experimental demonstration of this will be difficult for long-lived species.

We included several herbaceous species, and after 2 years at least one (Geophila repens) achieved reproductive size (1.4 m stem length; cf. Croat 1978) in a seed addition plot. Furthermore, seedlings of several other species also grew to rather large size over the 2 years of the study and the surviving seedlings of all species generally experienced positive growth in both years (Table 2). Hence, there is no indication that seedlings emerging after seed addition in otherwise unoccupied understorey sites should have a particularly low chance of reaching later ontogenetic stages. The strong seed limitation documented here suggests that tropical forest understorey plant communities are strongly influenced by seed arrival patterns, in particular given the likely absence of strong competitive dynamics in this stratum of the forest. Acting together with seed limitation, competitive suppression by the overstorey and herbivory cause understorey plant density to be low and hence there is little scope for direct competitive interactions among understorey plants (Wright 2002).

Nathan & Muller-Landau (2000) emphasized the importance of adding seeds at variable densities to provide information on seedling recruitment at saturating seed arrival densities. The addition of 25 vs. five seeds per 0.0079-m2 plot elicited a strong increase in seedling recruitment for shade-tolerant species with seed mass < 0.2 g. Consequently, the saturating seed rain density must be > 5 seeds per 0.0079 m2 for these species. There is an indication that 25 seeds per 0.0079 m2 might be approaching saturation because the increase in recruitment (on average 3.6-fold) was less than the increase in the number of seeds added (5-fold). Regardless, following Nathan & Muller-Landau (2000) the high seedling densities achieved by the addition of 25 seeds per 0.0079 m2 provide the best estimate of the absolute degree of seed limitation. The observed rain of seeds with minimum dimension > 1 mm averages 7.6 seeds per 0.0079 m2 year−1, including a large proportion of seeds from light-demanding species in old-growth forest on BCI (Harms et al. 2000). Hence, even the five-seed addition rate is likely to exceed the annual seed rain for all shade-tolerant species combined, and certainly far exceeds the combined natural seed rain (0.44 seeds per 0.0079 m2 year−1) of the 32 shade-tolerant study species (Table 1).

Seed limitation may be caused by insufficient seed numbers (source limitation) or non-uniform seed distribution patterns (dispersal limitation) (Clark et al. 1998; Muller-Landau et al. 2002; Schupp et al. 2002). Muller-Landau et al. (2002) found both strong source and dispersal limitation in Beilschmiedia pendula and negligible source and moderate dispersal limitation in Trichilia tuberculata at our study site. Our experiment does not allow separation of these two components of limitation for individual species; however, the strong relationship between the proportion of sites where seeds arrive and population-level seed production suggests that source limitation may generally be most important at the species level (Fig. 1). Dispersal limitation can be discounted at the community level.

The estimated strength of seed limitation varies with temporal scale because the chance that a seed arrives can only increase with the passage of time (Muller-Landau et al. 2002). Summing seed rain over long periods may be misleading, however, because it ignores temporal variability in site favourability and hence the importance of being present when ‘windows of opportunity’ open (Schupp et al. 2002).

We found that the seedling establishment rate per seed added did not differ between small-, medium- and large-seeded species, in contrast to several previous reports from BCI (e.g. Harms et al. 2000; Dalling & Hubbell 2002) and general meta-analyses (Moles & Westoby 2002, 2004). A relationship between seedling establishment and seed size may have emerged if the very smallest seeds, e.g. of Miconia or Ficus spp., had been included. However, in a study of 18 BCI tree species Augspurger (1984) likewise found no correlation between seed size and seedling survival rate under shade. Our impression from the field was that strong vertebrate predation on large seeds both before and after germination equalized overall establishment. A large proportion of the medium and large seeds added to our experimental plots were altogether removed, often within a few days of placement (J.-C.S., personal observation).

Seed addition, species identity, location and their interactions all had significant effects on seedling establishment for the 11 small-seeded, shade-tolerant species that were represented at all experimental sites in this study (Table 5). Differential establishment among locations and species with similar sized seeds could reflect many ecological differences, including attributes of microsites (e.g. light and moisture availability, proximity to pests) and attributes of species (e.g. shade and drought tolerance, pest defence). Niche differences with respect to edaphic conditions and light availability are well documented for the BCI flora (Kitajima 1994; Harms et al. 2001; Engelbrecht & Kursar 2003; Svenning et al. 2004). Specialized pests and pathogens could also cause seed and seedling survivorship to vary spatially, being low close to an adult conspecific or where conspecific density is high (Hyatt et al. 2003; Janzen 1970) as has indeed been shown for many of our study species on BCI (Condit et al. 1992, 1994; Schupp 1992; Harms et al. 2000; Wright et al. 2005). Nonetheless, the seed addition treatment and interactions involving the seed addition treatment collectively accounted for 86% of the explained variance in seedling recruitment in our experiment (Table 5).

We conclude that seed limitation strongly influences the local species composition of the understorey seedling community and limits the overall density of understorey plants on BCI and that site and species identity also have strong effects on establishment after seeds arrive. Strong seed limitation may set the stage for neutral dispersal-assembly of the shade-tolerant plant community on BCI (Hubbell 2001). However, the species differences in average seedling establishment per seed and its variation among sites demonstrate that niche differences among species and/or negative density dependence also affect community structure (Harms et al. 2000; Svenning et al. 2004; Wright et al. 2005).

Acknowledgements

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

We thank Sebastian Bernal for assisting with the field work. The Carlsberg Foundation (grants 990086/20 and 990576/20) and The Danish Natural Science Research Council (grants 9901835, 51-00-0138, and 21-01-0415) supported J.-C.S. The Andrew W. Mellon Foundation and the Environmental Sciences Program of the Smithsonian Institution supported S.J.W.

References

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  • Augspurger, C.K. (1984) Light requirements of neotropical tree seedlings: a comparative study of growth and survival. Journal of Ecology, 72, 777795.
  • Brokaw, N.V.L. (1982) The definition of treefall gap and its effect on measures of forest dynamics. Biotropica, 14, 158160.
  • Clark, J.S., Macklin, E. & Wood, L. (1998) Stages and spatial scales of recruitment limitation in southern Appalachian forests. Ecological Monographs, 68, 213235.
  • Condit, R., Hubbell, S.P. & Foster, R.B. (1992) Recruitment near conspecific adults and the maintenance of tree and shrub diversity in a neotropical forest. American Naturalist, 140, 261286.
  • Condit, R., Hubbell, S.P. & Foster, R.B. (1994) Density dependence in two understorey tree species in a neotropical forest. Ecology, 75, 671680.
  • Condit, R., Hubbell, S.P. & Foster, R.B. (1995) Mortality rates of 205 neotropical tree and shrub species and the impact of a severe drought. Ecological Monographs, 65, 419439.
  • Croat, T.B. (1978) Flora of Barro Colorado Island. Stanford University Press, Stanford.
  • Dalling, J.W. & Hubbell, S.P. (2002) Seed size, growth rate and gap microsite conditions as determinants of recruitment success for pioneer species. Journal of Ecology, 90, 557568.
  • Dalling, J.W., Hubbell, S.P. & Silvera, K. (1998) Seed dispersal, seedling establishment and gap partitioning among tropical pioneer trees. Journal of Ecology, 86, 674689.
  • Dalling, J.W., Muller-Landau, H.C., Wright, S.J. & Hubbell, S.P. (2002) Role of dispersal in the recruitment limitation of neotropical pioneer species. Journal of Ecology, 90, 714727.
  • Daws, M.I., Pearson, T.R.H., Burslem, D.F.R.P., Mullins, C.E. & Dalling, J.W. (in press) Effects of topographic position, leaf litter and seed size on seedling demography in a semi-deciduous tropical forest in Panama. Plant Ecology, in press.
  • Ehrlén, J. & Eriksson, O. (2000) Dispersal limitation and patch occupancy in forest herbs. Ecology, 81, 16671674.
  • Engelbrecht, B.M.J. & Kursar, T.A. (2003) Comparative drought-resistance of seedlings of 28 species of co-occurring tropical woody plants. Oecologia, 136, 383393.
  • Foster, R.B. & Brokaw, N.V.L. (1996) Structure and history of the vegetation of Barro Colorado Island. The Ecology of a Tropical Forest: Seasonal Rhythms and Long-Term Changes (eds E.G.Leigh, A.S.Rand & D.M.Windsor), pp. 6782. Smithsonian Institution Press, Washington, D.C.
  • Foster, B.L., Dickson, T.L., Murphy, C.A., Karel, I.S. & Smith, V.H. (2004) Propagule pools mediate community assembly and diversity-ecosystem regulation along a grassland productivity gradient. Journal of Ecology, 92, 435449.
  • Foster, B.L. & Tilman, D. (2003) Seed limitation and the regulation of community structure in oak savanna grassland. Journal of Ecology, 91, 9991007.
  • Gentry, A.H. (ed.) (1990) Four Neotropical Rainforests. Yale University Press, New Haven.
  • Harms, K.E., Condit, R., Hubbell, S.P. & Foster, R.B. (2001) Habitat associations of trees and shrubs in a 50-ha neotropical forest plot. Journal of Ecology, 89, 947959.
  • Harms, K.E., Wright, S.J., Calderón, O., Hernández, A. & Herre, E.A. (2000) Pervasive density-dependent recruitment enhances seedling diversity in a tropical forest. Nature, 404, 493495.
  • Hubbell, S.P. (2001) The Unified Neutral Theory of Biodiversity and Biogeography. Princeton University Press, Princeton.
  • Hubbell, S.P., Foster, R.B., O'Brien, S.T., Harms, K.E., Condit, R., Wechsler, B., Wright, S.J. & Loo de Lao, S. (1999) Light-gap disturbances, recruitment limitation, and tree diversity in a Neotropical forest. Science, 283, 554557.
  • Hurtt, G.C. & Pacala, S.W. (1995) The consequences of recruitment limitation: reconciling chance, history and competitive differences between plants. Journal of Theoretical Biology, 176, 112.
  • Hyatt, L.A., Rosenberg, M.S., Howard, T.G., Bole, G., Fang, W., Anastasia, J., Brown, K., Grella, R., Hinman, K., Kurdziel, J.P. & Gurevitch, J. (2003) The distance dependence prediction of the Janzen-Connell hypothesis: a meta-analysis. Oikos, 103, 590602.
  • Janzen, D.H. (1970) Herbivores and the number of tree species in tropical forests. American Naturalist, 104, 501528.
  • Kitajima, K. (1994) Relative importance of photosynthetic traits and allocation patterns as correlates of seedling shade tolerance of 13 tropical trees. Oecologia, 98, 419428.
  • Leigh, E.G. (1999) Tropical Forest Ecology: a View from Barro Colorado Island. Oxford University Press, New York.
  • Leigh, E.G., Rand, A.S. & Windsor, D.M. (eds) (1996) The Ecology of a Tropical Forest: Seasonal Rhythms and Long-Term Changes. Smithsonian Institution Press, Washington.
  • Levin, S.A., Muller-Landau, H.C., Nathan, R. & Chave, J. (2003) The ecology and evolution of seed dispersal: a theoretical perspective. Annual Review of Ecology, Evolution, and Systematics, 34, 575604.
  • Levine, J.M. & Murrell, D.J. (2003) The community-level consequences of seed dispersal patterns. Annual Review of Ecology, Evolution, and Systematics, 34, 549574.
  • Makana, J.R. & Thomas, C. (2004) Dispersal limits natural recruitment of African mahoganies. Oikos, 106, 6772.
  • Marquis, R.J., Young, H.J. & Braker, H.E. (1986) The influence of understorey vegetation cover on germination and seedling establishment in a tropical lowland wet forest. Biotropica, 18, 273278.
  • Moles, A.T. & Westoby, M. (2002) Seed addition experiments are more likely to increase recruitment in larger-seeded species. Oikos, 99, 241248.
  • Moles, A.T. & Westoby, M. (2004) Seedling survival and seed size: a synthesis of the literature. Journal of Ecology, 92, 372383.
  • Muller-Landau, H.C., Wright, S.J., Calderón, O., Hubbell, S.P. & Foster, R.B. (2002) Assessing recruitment limitation: concepts, methods and case-studies from a tropical forest. Seed Dispersal and Frugivory: Ecology, Evolution, and Conservation (eds D.J.Levey, W.R.Silva & M. Galetti), pp. 3553. CABI Publishing, Wallingford, UK.
  • Murray, K.G., Garcia-C., J.M. (2002) Contributions of seed dispersal and demography to recruitment limitation in a Costa Rican cloud forest. Seed Dispersal and Frugivory: Ecology, Evolution, and Conservation (eds D.J.Levey, W.R.Silva & M. Galetti), pp. 323338. CABI Publishing, Wallingford, UK.
  • Nathan, R. & Muller-Landau, H.C. (2000) Spatial patterns of seed dispersal, their determinants and consequences for recruitment. Trends in Ecology and Evolution, 15, 278285.
  • Quinn, G.P. & Keough, M.J. (2002) Experimental Design and Data Analysis for Biologists. Cambridge University Press, Cambridge.
  • Rand, T.A. (2000) Seed dispersal, habitat suitability and the distribution of halophytes across a salt marsh tidal gradient. Journal of Ecology, 88, 608621.
  • SAS Institute Inc. (2000) JMP® Statistics and Graphics Guide, Version 4. SAS Institute Inc, Cary, North Carolina.
  • Schupp, E.W. (1992) The Janzen–Connell model for tropical tree diversity: population implications and the importance of spatial scale. American Naturalist, 140, 526530.
  • Schupp, E.W., Milleron, T. & Russo, S.E. (2002) Dissemination limitation and the origin and maintenance of species-rich tropical forests. Seed Dispersal and Frugivory: Ecology, Evolution, and Conservation (eds D.J.Levey, W.R.Silva & M. Galetti), pp. 1933. CABI Publishing, Wallingford, UK.
  • Svenning, J.-C., Kinner, D.A., Stallard, R.F., Engelbrecht, B.M.J. & Wright, S.J. (2004) Ecological determinism in plant community structure across a tropical forest landscape. Ecology, 85, 25262538.
  • Terborgh, J., Pitman, N., Silman, M., Schichter, H. & Núñez, V. P. (2002) Maintenance of tree diversity in tropical forests. Seed Dispersal and Frugivory: Ecology, Evolution, and Conservation (eds D.J.Levey, W.R.Silva & M. Galetti), pp. 117. CABI Publishing, Wallingford, UK.
  • Tilman, D. (1997) Community invasibility, recruitment limitation, and grassland biodiversity. Ecology, 78, 8192.
  • Turnbull, L.A., Crawley, M.J. & Rees, M. (2000) Are plant populations seed-limited? A review of seed sowing experiments. Oikos, 88, 225238.
  • Valencia, R., Foster, R.B., Villa, G., Condit, R., Svenning, J.-C., Hernandéz, C., Romoleroux, K., Losos, E., Magård, E. & Balslev, H. (2004) Tree species distributions and local habitat variation in the Amazon: large forest plot in eastern Ecuador. Journal of Ecology, 92, 214229.
  • Webb, C.O. & Peart, D.R. (2001) High seed dispersal rates in faunally intact tropical rain forest: theoretical and conservation implications. Ecology Letters, 4, 491499.
  • Wright, S.J. (2002) Plant diversity in tropical forests: a review of mechanisms of species coexistence. Oecologia, 130, 114.
  • Wright, S.J. & Calderón, O. (1995) Phylogenetic constraints on tropical flowering phenologies. Journal of Ecology, 83, 937948.
  • Wright, S.J., Muller-Landau, H.C., Calderón, O. & Hernandéz, A. (2005) Annual and spatial variation in seedfall and seedling recruitment in a Neotropical forest. Ecology, 86, 848860.
  • Zar, J.H. (1996) Biostatistical Analysis. Prentice Hall Inc, Upper Saddle River, New Jersey.