Predicting effects of habitat destruction on plant communities: A test of a model using Amazonian trees

Abstract

We devised a 'random-clearing model' to predict effects of habitat loss on plant communities and populations. The model yields the probability that a species will be extirpated by land clearing, based on only two parameters: its density and the percentage of the landscape that has been cleared. It can also be used to predict species richness of plant communities following clearing, so long as densities of individual species are known. We tested the model using data on the distributions of 200 tree species (≥ 10 cm dbh) within two 9-ha experimental landscapes in central Amazonia. Deforestation levels ranging from 20% to 99% of the landscape were simulated randomly, with the actual persistence of each species being the number of times it remained in the landscape after 1000 runs. The model was effective in all cases, explaining 83-99% of the total variability in species persistence on each plot. Species' distribution patterns explained some residual variation in persistence but were of negligible importance compared to the predictions of the model. We also used the model to predict species richness, simulating both random and realistically contagious patterns of deforestation. Again, the model was highly effective, explaining 96% to 98% of the total variation in richness. Surprisingly, there was little difference in richness between random and contagious clearing patterns. These results suggest that, at least at the limited spatial scale of our analysis, the effects of deforestation on plant persistence and richness can be predicted using a simple model that assumes random species distributions and deforestation patterns. The model makes four predictions: (1) Density has an overriding influence on the susceptibility of species to clearing, while distribution patterns are usually of much lesser importance. (2) The relationship between density and persistence is nonlinear. (3) Rare species (≤1 tree/ha) become exceptionally vulnerable in heavily degraded landscapes and in small forest remnants. (4) The distinction between high (e.g., 95%) and very high (e.g., 99%) levels of habitat clearing in terms of species persistence is often dramatic.