Ecological communities can be structured by processes that operate at local scales, such as food quality, and by processes that operate at larger spatial scales, such as climate or species pools. In addition, because different species may perform similar functions at different sites, community structure may manifest differently if species or guilds are the unit of study. We studied the arthropod food web based on a single species of high marsh plant (Iva frutescens) that occurs across a latitudinal gradient on the East Coast of the United States. We sampled eight Iva plants at each of fifteen sites across ~12 degree of latitude, three times during the summer of 2007. We calculated two metrics of beta-diversity (nestedness and turnover) over two organizational levels (taxon and guild) at two spatial scales (plant and site). We tested three hypotheses: (i) Community structure is more deterministic at the larger (site) than the smaller (plant) spatial scale. (ii) Variation in guild composition among sites is more correlated with environmental variables, whereas variation in taxon composition is more correlated with distance among sites. (iii) Beta-diversity calculated based on taxon will be high due to turnover, reflecting within-guild species replacement, while beta-diversity calculated based on guilds will be characterized by greater nestedness, reflecting environmental filtering.
Results/Conclusions
More of the variation in community structure was correlated with environment, latitude or time at the large (site) than the small (plant) scale, consistent with hypothesis (i). This outcome is likely explained by stronger demographic stochasticity and species interactions at the plant scale. At the site scale, differences in guild composition were better explained by spatial effects than by measured environmental variables, whereas differences in taxon composition were equally explained by spatial effects and environmental variables. This result is inconsistent with hypothesis (ii) and suggests that the measured environmental variables did not strongly select for particular functional groups. The taxon-based beta-diversity had a larger turnover component, while the guild-based beta-diversity had a larger nestedness component, consistent with hypothesis (iii). However, the nestedness in the guild-based beta-diversity index was not significantly larger than that in a null model where guild identity of each taxon was randomly assigned. Instead of plant and habitat qualities, latitudinal gradients in climate and the species pool might have affected the food web structure, which then dictated the arthropod communities associated with Iva. Our results suggest that food web communities might have different assembly rules than horizontal communities.