Patterns of habitat use by murid rodents were examined in lowland grasslands in Taiwan. Fifteen grassland sites were live-trapped in 1990, and 16 microhabitat variables were measured at each of 539 trap stations in the sampling sites. Mus caroli, Apodemus agrarius, Rattus losea, and Bandicota indica were captured regularly during the sampling. Rattus norvegicus and Nivivenrer coxingi were captured infrequently. The analysis focused on the four common species.
Mus caroli, R. losea, and B. indica have been considered to be commensals of humans and to be largely dependent upon agricultural crops. However, these species were captured in natural habitats, often far from agricultural areas. M. caroli and R. Iosea were by far the most abundant species.
Vertical stratification of the habitat did not increase the number of species or overall rodent abundance. Rather, sites with well-developed grass cover and little vertical development of vegetation above the grass layer had the highest overall densities of rodents.
Use of habitat was very similar among the four common species and was concentrated in areas with dense grass cover. Overlap in habitat use was particularly extensive in the core areas of greatest occupancy. Relative population abundances (number of individuals captured per trap station) of the four species were related weakly to the microhabitat variables. These variables were better predictors of capture probabilities of M. caroli and R. losea at individual trap stations, as revealed by multiple logistic regression. Abundances and capture probabilities did not appear to be affected by the presence or abundance of other species.
It is suggested that body size differences may facilitate coexistence of these four species of ecologically similar murids. Beginning with the smallest species (M. caroli), body masses of each subsequent species in the series differ by at least a factor of two. These grassland communities may be somewhat resistant to invasion by more similar-sized species of murids.