Beta-Diversity in Tropical Forest Trees
Tóm tắt
Từ khóa
Tài liệu tham khảo
S. P. Hubbell The Unified Neutral Theory of Biodiversity and Biogeography (Princeton Univ. Press Princeton NJ 2001).
C. R. Pyke R. Condit S. Aguilar and S. Lao [ J. Veg. Sci. 12 553 (2001)] described the network in Panama: 31 1-ha plots in lowland forest in Panama (10 smaller plots at higher elevation were omitted from the analysis) single 4- and 6-ha plots and a 50-ha plot on Barro Colorado Island (BCI). The large plots were divided into individual 1-ha plots for this analysis. The plots were scattered along the Panama Canal over a region of about 15 km by 50 km and included 513 morphospecies and 39 645 individuals ≥10 cm in diameter. A map showing the plot locations as well as a matrix of species abundance per plot is provided with the supplemental material (7).
K. Romoleroux et al. [in Estudios Sobre Diversidad y Ecologı́a de Plantas R. Valencia H. Balslev Eds. (Pontificia Universidad Católica del Ecuador Quito Ecuador 1997) pp. 189–215] described a 25-ha plot in Yasunı́ National Park in Ecuador. As in the larger Panamanian plots the Yasunı́ plot was treated as 25 separate 1-ha plots for this analysis and only trees ≥10 cm in diameter were included (820 species 17 546 individuals).
N. C. A. Pitman et al. [ Ecology 82 2101 (2001)] described one plot network in lowland forest of Yasunı́ National Park in eastern Ecuador where 15 1-ha plots held 1015 morphospecies and 9530 individuals ≥10 cm diameter and another in the Manu Biosphere Reserve in Amazonian Peru where 14 1-ha plots harbored 687 morphospecies and 8287 individuals (floodplain plots were omitted from the analysis). Within each site taxonomy was uniform. For intersite comparisons (including Panama) only fully identified species were included; this meant excluding 200 to 300 species per region that were recognized as morphospecies but not named. Maps of plot locations in both areas are provided with the supplemental material (7).
Supplementary Web material is available on Science Online at www.sciencemag.org/cgi/content/full/295/5555/666/DC1.
The Sørensen index is S 12 /[0.5( S 1 + S 2 )] where S 12 is the number of species common to both sites and S i is the total found at site i. Jaccard's index is S 12 /( S 1 + S 2 − S 12 ). For two plots the probability F can be calculated as F = ∑ f i 1 f i 2 where f ij is the relative abundance of species i at site j and the sum is over all species at both sites. More generally F ( r ) can be calculated by finding all pairs of trees separated by distances between r and r + Δ r and then determining what proportion of these pairs are the same species.
By overall decay we mean that all pairs of plots were considered together with similarity evaluated as a function of distance. The similarity-distance function predicts the slope of a power-law species-area curve (2) making it a powerful approach to beta-diversity. Other measures of beta-diversity based on species turnover (22) have not used the similarity-distance function but are nevertheless closely related theoretically. On the other hand we lose information by averaging all pairs of plots (at a given distance); this allows the data to be smoothed and provides theoretically relevant numbers but abrupt transitions due to habitat change would be missed.
Adjacent hectares in Panama are more similar because species richness is lower there—79 species/ha compared with 173 in Peru and 247 in Ecuador.
J. C. Nekola and P. S. White [ J. Biogeogr. 26 867 (1999)] linearized the similarity-distance curve by plotting log-Jaccard versus distance. Our data could not be linearized at all scales with this or any other logarithmic transformation. At distances less than 50 km the decline in similarity we observed was linear with the log of distance (opposite to the Nekola and White transformation) but Nekola and White lacked data at such short scales. At larger distances the log of similarity declined linearly with distance in our data as in boreal forest. We took estimates of the natural log of the Jaccard index from plots 10 to 20 km apart and from plots >1000 km apart and calculated a regression.
Single soil samples were taken from each of the 15 1-ha plots in Yasunı́. Soil pH nitrogen phosphorus sand silt content and nine other measures of soil chemistry showed no spatial autocorrelation; only copper content did.
G. Malécot The Mathematics of Heredity (Freeman San Francisco CA 1969).
Dispersal kernels where seeds have a high probability of dispersing long distances are called “fat-tailed” (13). These may have infinite mean or infinite higher moments meaning empirically that no matter how many distances have been measured the next might double the estimate of the moment concerned.
Seed dispersal distances were estimated from seedfall into 200 seed traps in the BCI 50-ha plot (23) with inverse modeling (24 25). Gaussian dispersal functions fit significantly better than a null model in 65 tree species (26).
For all comparisons of plots 18 to 20 km apart in Panama the mean ± SD of F was 0.0090 ± 0.0071 ( N = 77 plot pairs). At Yasunı́ at 17 to 22 km it was 0.0100 ± 0.0027 ( N = 21) and at Manu at 15 to 21 km it was 0.0129 ± 0.0062 ( N = 18). The same trend held for larger distances.
Typical wet-forest species occurred on an island of andesite toward the dry side of the isthmus (4). Consider six plots on BCI two plots on the andesite 9 to 12 km south of BCI and four plots on a sedimentary formation 10 to 13 km east of BCI. For BCI versus sedimentary the mean F was 0.0145 ± 0.0051 (24 comparisons ± SD); for BCI versus andesite the mean F was 0.0038 ± 0.0023 (12 comparisons). The latter is lower than the average F between all plots in Ecuador and all plots in Peru 1367 km apart ( F = 0.0092).
All three similarity indices (8) are sensitive to species' abundances even though Sørensen and Jaccard are based only on presence-absence data. In a sample as small as 1 ha of diverse forest many local species are absent but abundant species are nearly always present. Thus presence-absence indices are elevated when the same species are dominant at two sites relative to a situation where the dominant species at one site are rare at the other.
H. C. Muller-Landau dissertation Princeton University Princeton NJ (2001).
We thank the Smithsonian Tropical Research Institute for logistical and financial support; the U.S. Department of Defence Legacy Fund and the U.S. Agency for International Development for financial support of the plot network in Panama; and the Andrew W. Mellon Foundation the John D. and Catherine T. MacArthur Foundation and the NSF for supporting the plot networks in Peru and Ecuador.