Colloquium paper: how many tree species are there in the Amazon and how many of them will go extinct?

Abstract

New roads, agricultural projects, logging, and mining are claiming an ever greater area of once-pristine Amazonian forest. The Millennium Ecosystems Assessment (MA) forecasts the extinction of a large fraction of Amazonian tree species based on projected loss of forest cover over the next several decades. How accurate are these estimates of extinction rates? We use neutral theory to estimate the number, relative abundance, and range size of tree species in the Amazon metacommunity and estimate likely tree-species extinctions under published optimistic and nonoptimistic Amazon scenarios. We estimate that the Brazilian portion of the Amazon Basin has (or had) 11,210 tree species that reach sizes >10 cm DBH (stem diameter at breast height). Of these, 3,248 species have population sizes >1 million individuals, and, ignoring possible climate-change effects, almost all of these common species persist under both optimistic and nonoptimistic scenarios. At the rare end of the abundance spectrum, however, neutral theory predicts the existence of approximately 5,308 species with <10,000 individuals each that are expected to suffer nearly a 50% extinction rate under the nonoptimistic deforestation scenario and an approximately 37% loss rate even under the optimistic scenario. Most of these species have small range sizes and are highly vulnerable to local habitat loss. In ensembles of 100 stochastic simulations, we found mean total extinction rates of 20% and 33% of tree species in the Brazilian Amazon under the optimistic and nonoptimistic scenarios, respectively.

Fig. 1.

Fit of Fisher's logseries to the Amazonian relative tree species abundance data of Pires, Dobzhansky, and Black (1).

Fig. 2.

Fit of Fisher's logseries to the relative abundance data of Amazonian tree genera (data from ref. 18). Fisher'σ a (θ) is ≈71.

Fig. 3.

The predicted logseries rank abundance curve for tree species in the Brazilian Amazon. The size of the metacommunity for this calculation was taken as 4,648,400 km² times 60,000 stems >10 cm DBH km⁻². The value of Fisher's α (or θ) for this calculation was 500.

Fig. 4.

Log–log relationships between the mean distance to the nth nearest conspecific neighbor of a focal plant and the rank of nearest neighbor, for two arbitrarily chosen BCI tree species. (a) Alseis blackiana (Rubiaceae), stems >1 cm DBH; Y = 0.5589 + 0.5419X, R² = 0.999944. (b) A. blackiana (Rubiaceae), stems >20 cm DBH; Y = 1.2137 + 0.5514X, R² = 0.999376. (c) Beilschmiedia pendula (Lauraceae), stems >1 cm DBH; Y = 0.7472 + 0.5608X, R² = 0.99951. (d) B. pendula (Lauraceae), stems >20 cm DBH; Y = 1.3328 + 0.5706X, R² = 0.999638. (e) Distribution of slopes of the log–log relationship between the mean distance to the nth nearest conspecific neighbor of a focal plant and the rank of nearest neighbor for the 155 tree species on BCI with abundances >10². (f) Distribution of intercepts for the same relationship.

Fig. 5.

Landscape-level relationship between the mean distance to the nth nearest conspecific neighbor of a focal plant and the rank of nearest neighbor for the rare canopy tree Tabebuia guayacan (Bignoniaceae) over the 15.2 km² area of BCI, Panama. Power-law relationship for T. guayacan adults: log₁₀(distance to nth nearest neighbor = 1.7099 + 0.6586 log₁₀(rank nearest neighbor), R² = 0.998.

Fig. 6.

Scenarios of the future of the Brazilian Amazon [after Laurance et al. (3)]. (a) Optimistic scenario. (b) Pessimistic scenario. Black, heavy-impact areas; red, moderate-impact areas; yellow, light-impact areas; green, pristine areas. See text for description of these impacts. (c) Relative sizes of species ranges for some highly abundant species, on same scale as maps of the Brazilian Amazon.

Fig. 7.

Predicted extinction rates of tree species in the Brazilian Amazon under the optimistic and nonoptimistic scenarios of Laurance et al. (3) as a function of population size. The abundance bins (histogram bars) are labeled with the logarithm to the base 10 of the lower bound of abundance in the respective bin. (a) Extinction scenario 1: Species go extinct only if their entire geographic range is in heavy-impact areas (in black) on either the optimistic or the nonoptimistic scenarios of Laurance et al. (3). (b) Comparison of extinction scenarios 1 and 3. In extinction scenario 3, species manage to survive in heavy-impact areas if at least one occupied cell survives, with probability 0.05 per cell. Extinction scenario 2 (data not shown), in which species can go extinct if they have some portion of their range in areas other than heavy-impact, gave results very similar to extinction scenario 1.