Small mammals were studied from September 1978 to November 1980 in 5 Minnesota farmstead shelterbelts. Based on correlations between habitat variables characterizing vegetative features of shelterbelts and numbers of individuals captured in each season, Peromyscus leucopus and Clethrionomys gapperi were considered as woodland species, Microtus pennsylvanicus as a grassland species, and Sorex cinereus and Blarina brevicauda as intermediate in terms of their dependency on shelterbelts. Area and perimeter of shelterbelts were associated with numbers of all species except C. gapperi. Numbers of P. leucopus and C. gapperi tended to be higher in shelterbelts that were isolated from other wooded habitat, whereas numbers of S. cinereus were lower in isolated shelterbelts. Species richness was greater in larger shelterbelts with complex vegetative structure. Species of small mammals residing in shelterbelts were those that typically are not considered as agricultural pests. Management recommendations include maintenance practices that do not reduce stratification of vegetation, leaving woody and man-made debris within shelterbelts, and establishing shelterbelts that are as large as possible within the economic constraints of farming. These recommendations simultaneously would benefit other species of mammals and birds in the intensively farmed regions of the Midwest. J. WILDL. MANAGE. 47(1):74-84 Vegetative complexity and habitat size are major determinants of the abundance of individual mammalian species and the structure of mammalian communities in natural habitats (Dueser and Shugart 1978, Dueser and Brown 1980, Geier and Best 1980). Farmstead shelterbelts are small (generally 1.5 m in depth) that prevented access to traps. Traps were opened for 2 consecutive days each session and were closed between sessions. Rolled oats mixed with small amounts of peanut butter were used as bait. I marked each captured animal with ear tags (#1, National Band and Tag Co., Newport, Ky.) or toe clips for individual recognition. Sex, age, weight, reproductive condition, and trap location were recorded for each capture. Species captured >20 times during the study were included in the analyses; these were Sorex cinereus, Blarina brevicauda, Clethrionomys gapperi, Microtus pennsylvanicus. and Peromyscus leucopus. Fifty-four habitat variables were measured for each shelterbelt during August 1979 and 1980. These included 2 dimensional, 14 proximal land-use, 27 vegetative, and 11 physical variables (Appendix). Vegetative and physical features were based on sampling methods and procedures modified from Dueser and Shugart (1978) using 3 sampling units centered on each trap: a 10-m radius plot, a 1-m2 ring, and 2 perpendicular 20-m2 transects. Species variables were derived in each of the 9 seasons (autumn 1978, winter 1979, etc.) for the 5 species of small mammals. These variables consisted of the total number of individuals (all age-sex classes combined) and the total number of indiJ. Wildl. Manage. 47(1):1983 This content downloaded from 207.46.13.113 on Thu, 06 Oct 2016 04:06:29 UTC All use subject to http://about.jstor.org/terms 76 SMALL MAMMALS IN FARMSTEAD SHELTERBELTS * Yahner viduals per age-sex class (except for Sorex and Blarina) per shelterbelt in each season. In addition, species richness (S), or the total number of different species of small mammals captured, was determined per shelterbelt for the entire study period and for each season. Relationships between the 54 habitat variables and the species variables were examined. As a result of small sample size (N = 5 shelterbelts) and multicollinearity (Chatterjee and Price 1977) among habitat variables, simple correlation analyses were used (Sokal and Rohlf 1969). Both untransformed and log-transformed data were analyzed (after Dueser and Brown 1980). Throughout the text, significant correlations (P < 0.05) were based on product-moment correlation coefficients (r) ?0.878 and df = 3. RESULTS AND DISCUSSION