Seasonal semi-deciduous forest is represented by only 3,5% of the area of the state of Mato Grosso. Heavy deforestation has reduced it mostly, to isolated fragments. Studies of mammals in these forests have been rare, with just a few species examined. Between November 2002 and August 2004, we studied the mammal populations of 18 semi-deciduous forest fragments and four control areas, with a total effort of 24,200 trapsnights. We captured 708 individuals, with a 2.8 % capture success rate. We captured 23 species of small mammals from 18 genera, including, 13 species were rodents and 10 marsupials. To evaluate the effect of seasonality on the community, we analyzed eight of the 22 areas during the two seasons (dry and wet) of the same year. Although the differences were not significant, we obtained a slightly higher abundance and species richness of small mammals during the dry season in total numbers. Independent of season, at two of our fragment study sites, F1 and F2, we captured more than twice the number of rodents when compared to the other six areas studied in that period. While the effect of seasonality was not significant for the community considered as a whole or for the rodents, the marsupials responded significantly, with more than twice the number captured during the dry season as compared to the wet season. Marmosops noctivagus and Micoureus demerarae had by far the higher capture rates over all study areas. Although not significant, we captured more than twice the number of Marmosops noctivagus individuals during the dry season than during the wet season. The high abundance of marsupials during the dry season could be related to food availability, bait search and success rates, or recruitment of young born in the area. Only two species, Bolomys lasiurus and Calomys sp., exhibited similar abundances in both forest and matrix (pasture) habitat. For the majority of species, the pasture matrix represented a barrier to dispersal, suggesting the possible extinction of small mammals in the long term, especially in the smaller fragments. In this study the size, the form and the age of the fragments, did not significantly affect the abundance or species richness of small mammals on our study sites. Fragment shape and time since isolation also had no effect significative on small mammal abundance and species richness. For marsupials, only the most irregularly-shaped fragments had significantly higher marsupial abundance; size and time since isolation had no effect significative on marsupial abundance or species richness. Since the physical variables of size, shape, and time since isolation of the fragments did not influence significantly rodent abundance, other variables, such as historical factors not analyzed in this study, may better explain the variation in the data. Vegetation structure affected the relative abundance of some species across study sites. Canopy cover, number of trees, and litter volume did not affect significantly the overall abundance or species richness of small mammals. However, we captured more rodents than marsupials in study areas with higher litter volume. Each study area had a different post-fragmentation history. Understanding the factors that affect the small mammal community of each area is fundamental for creating public policy and management strategies for conservation areas presently in existence and those that will be created.
