Beta Diversity in Tropical Forests

See allHide authors and affiliations

Science  25 Jan 2002:
Vol. 295, Issue 5555, pp. 636-637
DOI: 10.1126/science.295.5555.636

The tropics support more than 200,000 species of flowering plants including many tree species (1). Yet even between different geographical areas, species composition may vary dramatically. For the tropical forests of Africa, Asia, and the Americas greater than 106 km2 in size, overall or gamma diversity varies from perhaps 30,000 to 120,000 species of flowering plants (2). It is well established that smaller forest plots ranging from 0.001 to 0.01 km2 in area contain from 30 to 300 tree species (alpha diversity) (3). Less information is available for beta diversity, which describes how species composition varies from one area to another. On page 666 of this issue, Condit et al. (4) present a new analysis of beta diversity in which they compare the species composition of forest plots that are located at distances of 10−1 to 103 kmapart in the neotropics of Panama (southern Mesoamerica) and in Ecuador and Peru (western Amazon).

Condit et al. explore how similarity in tree species composition from plot to plot declines as the distance between the plots increases. Regions in Panama and the western Amazon that are 104 km2 in area support 3500 to 5000 tree and shrub species (5). Yet at smaller scales (10−2 km2), the western Amazonian forests support 2 to 10 times as many species as do the Panamanian forests (6). It is possible to obtain rough values for beta diversity from the quotient of gamma and alpha diversity. This method predicts a relatively low beta diversity for the western Amazon, which Condit et al. confirm. However, this prediction is not in line with earlier views of strong beta diversity in western Amazonian forests (6). The higher beta diversity in Panama presumably reflects greater spatial variation in geology and climate and a lag in forest recovery after the marked temporal variations in climate during the last glacial cycle.

The investigators (4) compared their observations to predictions derived from a neutral model that takes into account dispersal capacity but ignores environmental and historical events controlling species distribution. Their data compare well with the neutral model at intermediate distances (0.2 to 50 km) between plots, underscoring the potential importance of dispersal as a key process in the structuring of tropical forest diversity (7). At smaller distances, they observed much greater similarity in species composition between plots than that predicted from the model (8). When the distances between plots were much larger as found in the western Amazon, the model again underestimated the similarity in species composition. Condit et al. conclude that dispersal is not the principal event that determines the diversity of tree species in western Amazonia (9).

To quantify the relative importance of spatial and environmental events in determining species similarity between plots, we have reanalyzed Condit et al.'s Panama plot data. Dispersal is a purely spatial process: Progeny grow close to their parents when dispersal capacity is low. We used straight-line distances between plots to represent the dispersal process. Condit et al. provided four crucial environmental variables: elevation, precipitation, age of the forest stand, and the type of bedrock. We used similarities converted from normalized differences between plots to quantify each environmental variable, and the Steinhaus coefficient to quantify species similarity. Distance and the four environmental variables were all significant predictors of species similarity between plots in permutation-based multiple regressions. We then partitioned the variance in species similarity by computing multiple regressions of species similarity against distance only, environmental variables only, and both distance and environmental variables (10). Distance alone and environmental variables alone explained minor portions of the variation in species similarity (see the figure). Distance and environment together, however, explained 24% of the variation. The inability to separate distance and environment reflects a strong gradient in rainfall that is highly correlated with distance between Panamanian plots (11). Perhaps most important, 59% of the variation in species similarity remained unexplained by either distance or environment. In an analogous study, distance and environment explained just 16% of the variation in upland tree species composition between Colombian forest plots (12). This unexplained variance is typical for studies of tree species similarity in tropical forests.

Tropical forest diversity.

Variance in tree species similarity among plots in Panamanian tropical forests. Distance and environment explain minor portions of the variation in species similarity. The bulk of the variation remains unexplained.

Condit et al.'s approach is an important step toward predicting the effects of plant dispersal on species composition in the tropics. However, given that most of the variation in species similarity in tropical forests cannot be explained, there is a clear need for additional data and analyses before we fully understand the events that determine tropical forest diversity.

References and Notes

Navigate This Article