Do Species Sorting and Mass Effects Drive Assembly in Tropical Agroecological Landscape Mosaics?

Recent assessments of biodiversity in tropical agroecosystems have revealed surprisingly high functional and taxonomic diversity in systems with low management intensity. This biodiversity is the product of community assembly. Because agroecosystems are novel ecosystems and occur in landscape mosaics, the assembly processes generating communities in agroecosystems are poorly resolved. Broadly, two models have been proposed to explain landscape assembly: trade-offs in species performance across habitats (species sorting) and source-sink dynamics between habitats of differential quality (mass effects). These models are largely untested in tropical agroecosystems. We utilize an extensive data set on a tropical twig-nesting ant community from five microhabitat types in a shaded coffee agroecosystem to test for species sorting, mass effects, or a mixed model. To test among these models, we used community similarity and a variance decomposition on a focal microhabitat (a moderate-shade coffee farm) to partition community variance into spatial and environmental components. To identify the source habitat for mass effects and assess their strength, we measured dispersing alates (winged reproductives), artificial nests, and colony and nest size in shade trees and coffee. We found significant environmental and spatial signal and evidence for both species sorting and mass effects. We find sorting occurs among common species, but that mass effects are prevalent among rare species and likely originate in the shade trees. Our results indicate that both metacommunity models occur in tropical landscape mosaics, but they may not apply equally to all species in communities, habitat gradients, or timescales.