Food-web assembly during a classic biogeographic study: Species' "trophic breadth" corresponds to colonization order

Ecologists have found many patterns in food-web structure. Some, like the constant connectance hypothesis, lack definitive explanatory mechanisms. In response, we investigated whether community assembly mechanisms could explain why trophic complexity consistently scales with species richness among ecosystems. We analyzed how food-web structure developed during the community assembly recorded in Simberloff and Wilson's classic biogeography experiment. Using their arthropod surveys, we constructed six time series of food-webs from pre- and post-defaunation censuses of six experimental islands, and synthesized trophic information for 250 species from the literature and expert sources. We found that the fraction of specialist species increased and the fraction of generalists decreased during food-web assembly. Directed connectance initially declined over time, despite an increase in species richness, but eventually leveled off as predicted by the constant connectance hypothesis of diversity-complexity scaling. The initial decline was explained by later colonization by trophic specialists, probably due to limited resource availability during early colonization. Late-colonizing super-generalists maintained constant connectance at later dates. This relationship between colonization success and trophic breadth helps explain food-web patterns and corroborates assertions that community assembly is systematically influenced by species' trophic breadths.