Small mammals such as mice and voles have potential as indicators of sustainable forest management. They have an important functional role in forests, they are economically important as prey for furbearer populations, and they respond to disturbance in a characteristic manner. In Ontario, Canada, several small mammal species have been suggested as bioindicators. However, strong year-to-year variation in population levels independent of forest disturbance means that very long time frames would be required to detect trends. Models of habitat supply have been suggested as a method of monitoring small mammals. We explore the feasibility of monitoring structural measurements and habitat supply for small mammal species using an area near White River, Ontario, Canada, as a case study. Small mammals were surveyed in the region for 3 years, and associations with mapped and stand level habitat attributes examined. Thirteen species were recorded, but only five species were recorded in sufficient numbers for habitat associations to be examined. The deer mouse and red-backed vole were recorded from all mature forest habitats, although both were more prevalent in mixedwood stands. Red-backed vole abundance was linearly related to stand age and the volume of downed logs. Deer mice were most abundant in recently clearcut stands, with abundance declining sharply in 5-15-year-old stands. They were also abundant in mature forest, where they were significantly associated with downed wood volume. Vegetation complexity was also significant for both species. Habitat supply maps for both species could be readily developed, and structural attributes modified by forest practices were important. However, strong year-to-year variation in the abundance of both species in mature forest prevented carrying capacities from being reliably assigned to habitat supply maps. Thus, while relative changes in the availability of high, medium and low quality habitat are identifiable, expected changes in minimum population size cannot be inferred. The effect of cumulative disturbances on the quality of available habitat is also unknown. Without this information, change in habitat supply cannot be used to assess the sustainability of forest management actions. We suggest that dynamic landscape meta-population (DLMP) models may provide one solution, and require further exploration as a sustainability assessment tool. (c) 2004 Elsevier B.V. All rights reserved.
