Control of Seedling Recruitment by Land Crabs in Rain Forest on a Remote Oceanic Island

Abstract

We tested the hypothesis that an omnivorous land crab determines the dynamics of seedling recruitment in rain forest on Christmas Island, Indian Ocean. The red crab (Gecarcoidea natalis) occurs at densities of ∼1 crab/m2 and consumes seeds, seedlings, and leaf litter. A crab-exclusion experiment below closed canopy and in light gaps showed that red crabs markedly reduced seedling recruitment. Overall, seedling emergence was 29-fold and 21-fold greater on the exclusion plots in the understory and in gaps, respectively. In control plots, mean seedling densities never exceeded 15 seedlings/25 m2 in the understory and 3 seedlings/25 m2 in gaps. Species richness of seedlings was significantly higher in the absence of crabs, averaging >10 and 12 species per plot in the understory and gaps, respectively. On control plots, mean species richness was less than one species per plot. Two-dimensional ordination (using nonmetric multidimensional scaling) showed that red crabs had a large impact on relative species abundances. Analyses of similarity confirmed that significant differences existed between both treatments and habitats. Seed removal and seedling transplant experiments suggested that most rain forest species are rapidly eaten by red crabs. Seedlings of many of these vulnerable species (e.g., Maclura cochinchinensis, Planchonella nitida, Schefflera elliptica) were abundant on exclusion plots. However, they were rare on control plots, and none persisted for >2 mo. Persistent seedlings were limited to two species (Inocarpus fagifer and Tristiropsis acutangula), both of which were resistant as seeds and seedlings to crab predation. Although red crabs are the major biotic determinants of seedling recruitment, this impact was not reflected in the overhead canopy. Some tree species unable to recruit in the presence of crabs were common as adults. Shifting mosaics in crab densities over space and time may offer the best explanation for recruitment of vulnerable species. Our results show that a dominant consumer has large effects on plant community structure on a remote oceanic island, consistent with the viewpoint that strong control by a single species is more likely in simple communities.