Effects of isolation on ant assemblages depend on microhabitat

How isolation affects biological communities is a fundamental question in ecology and conservation biology. Local diversity (α) and regional diversity (γ) are consistently lower in insular areas. The pattern of species turnover (β diversity) and the influence of isolation on competitive interactions are less predictable. Differences in communities across microhabitats within an isolated patch could contribute to the variability in patterns related to isolation. Trees form characteristically dense and sparse patches (low vs. high isolation) in floating marshes in coastal Louisiana, and canopy and root areas around these trees could support distinct ant communities. Consequently, trees in floating marshes provide an ideal environment to study the effects of isolation on community assemblages in different microhabitats. We sampled ant communities in 120 trees during the summer of 2016. We found ant α diversity was not different between the canopy and roots, and the magnitude and directional effects of isolation on ants were inconsistent between the canopy and root areas. In the roots of sparse sites, ant diversity (α, β, and γ) was lower, species composition was changed, and the signature of interspecific competition was more prominent compared to dense sites. In the canopy, however, significant differences between dense and sparse sites were only detected in α and γ diversity, and ant species co‐occurrence was not significantly different from a random distribution. The inconsistent responses of ants in canopy and root areas to isolation may be due to the differences of species pool size, environmental harshness, and species interactions between strata. In addition, these findings indicate that communities in distinct microenvironments can respond differentially to habitat isolation. We suggest incorporating organisms from different microhabitats into future research to better understand the influence of isolation on the assembly of biological communities.