Tropical tree communities: A test of the nonequilibrium hypothesis

We have conducted a test of Hubbell's nonequilibrium model of tropical forest dynamics and found that it fails to account for high levels of compositional similarity in disjunct samples of floodplain forest in the Manu River basin of southeastern Peru. In describing how the composition of a tropical forest would change over many tree generations in the absence of density dependence, Hubbell's model, as presented, is untestable. However, by substituting space for time, it can be tested. Here we take advantage of the meandering of the Manu River to generate independently replicated, primary-successional chronosequences which culminate in tall, "mature-phase" forests possessing a characteristic complex vertical structure. By the logic of the nonequilibrium hypothesis, spatially uncorrelated forests should show uncorrelated species abundances. We tested this prediction by examining the compositional similarity of 1000-tree samples in four types of comparisons: (1) between nearby (potentially spatially correlated) and remote (presumably uncorrelated) plots of mature floodplain forest, (2) between real and synthetic forests generated by scrambling the relative abundances of the species, (3) between successional stages and the "mature phase" of floodplain forest, and (4) between mature floodplain forest and adjacent upland forest on a different soil type. In all comparisons, five mature floodplain-forest plots showed higher levels of similarity inter se than to any of the other real or synthetic forests. The 10 most abundant species in the three potentially spatially correlated plots appeared prominently in the 10 top ranks of two remote plots located 30 and 39 km to the east. The probability of obtaining the observed result under the expectation of the nonequilibrium hypothesis was ≈10-7. We conclude with a plea for the application of biological models to the problem of tropical forest tree diversity.

Duke Scholars

Author John W. Terborgh Environmental Sciences and Policy