Colonization and extinction of ant communities in a fragmented landscape

Abstract In this paper we tested the assumption that smaller and more isolated remnants receive fewer ant colonizers and lose more species. We also tested hypotheses to explain such a pattern. We sampled ants in Brazil for 3 years in 18 forest remnants and in 10 grasslands between them. We tested the influence of remnant area and isolation on colonization rate, as well as the effect of remnant area on extinction rate. We tested the correlation between remnant area and isolation to verify the landscape design. Colonization rate was not affected by remnant area or isolation. Extinction rate, however, was smaller in larger remnants. Remnant area and isolation were negatively correlated. We tested two hypotheses related to the decrease in ant species extinction rate with increased remnant area: (i) small remnants support smaller and more extinction‐prone populations; and (ii) small remnants are more often invaded by generalist species, which suffer higher extinction inside remnants. The density of ant populations significantly increased with area. Generalist species presented a lower colonization rate in larger remnants, contrary to the pattern observed in forest species. Generalist species suffered more extinction than expected inside remnants. The lack of response of colonization rate to remnant area can be explained by the differential colonization by generalist and forest species. The decrease of ant population density in smaller remnants could be related to loss of habitat quality or quantity. The higher colonization by generalist ant species in the smaller remnants could be related to landscape design, because smaller remnants are more similar to the matrix than larger ones. Our results have important implications for conservation strategies because small remnants seem to be more affected by secondary effects of fragmentation, losing more forest species and being invaded more often by generalist species. Studies that compare only species richness between remnants cannot detect such patterns in species composition.