EVALUATING SMALL MAMMAL RESPONSE TO NATURAL DISTURBANCE AND RESTORATION IN OAK ECOSYSTEMS IN THE MISSISSIPPI ALLUVIAL VALLEY

oak species form a conspicuous and often dominant component of bottomland forests of the mississippi alluvial valley. the extent of these forests has been drastically reduced as a result of clearing for agri culture in the past two centuries. patterns of clearing have reduced the distribution of remaining forest patches to a much more flood-prone subset of the landscape than was historically the case, reducing the diversity of oak species currently present on the landscape. intensive harvesting has further changed the composition of the remaining stands. small remnant patches of primary forest continue to exist as research natural areas on the delta national forest in sharkey county, mississippi. in particular, the overcup oak (quercus lyrata) and redgum (liquidambar styraciflua) research natural areas pres ent substantial components of the trees for which the areas were named, as well as quercus nuttallii and smaller components of other species. recent interest in afforestation has produced a resurgence of interest in restoration of oak forest to abandoned farmland in the region. we have studied small mammal response to restoration on an extensive experiment near the delta national forest since 1995. we have also ex amined small mammal response to a tornado that disturbed approximately half of the overcup oak research natural area in 2008. we use these studies to demonstrate how population estimates of small mammals can be obtained from capture-recapture studies, employing different designs, and utilizing program capture for population estimation. small mammal communities in these stands are more species-rich in early succession than in primary forest. the study of response to tornado damage to the overcup oak research natural area is complicated by the fact that this particular forest type is very flood-prone, creating obstacles to colonization by small mammals. analysis of capture-recapture data with robust methods illustrated in this study permits extraction of