Date of Award

Fall 12-2021

Level of Access Assigned by Author

Open-Access Thesis

Degree Name

Master of Science (MS)


Wildlife Ecology and Wildlife Conservation


Erin M. Simons-Legaard

Second Committee Member

Cynthia S. Loftin

Third Committee Member

Erik J. Blomberg


Wildlife habitat conservation in landscapes where human activities cause chronic habitat disturbance is contingent upon developing land management strategies that minimize the effects of future habitat changes on wildlife populations. Long term studies can provide unique opportunities to understand how species respond to progressive habitat change, and such an understanding can reveal ways in which the often conflicting objectives of wildlife habitat conservation and human land use can be reconciled. Characterizing how animals respond behaviorally to habitat conditions may be a useful tool for identifying potential negative effects of disturbance before such effects impact rates of species occurrence, population demography, or other metrics indicative of population viability. The composition and configuration of Maine’s forests have been progressively and drastically altered by forest harvesting, and behavioral responses displayed by forest-associated species to the temporally cumulative effects of forest harvesting can inform forest management strategies for the conservation of Maine’s forest-associated wildlife. I evaluated the effects of forest composition and patch configuration on patterns of patch-scale habitat selection displayed by American marten (Martes americana) to identify marten responses to the cumulative and multidimensional effects of forest harvesting across a 30-year period (1989-2019). I used a spatially-explicit time series of forest harvesting and forest inventory data to classify forest types that were ecologically distinct to marten and were easily interpretable from a forest management perspective. I coupled marten habitat maps with a telemetry dataset collected from resident, non-juvenile marten to estimate patterns of marten habitat selection as they related to forest height and harvest history, the availability of different forest types within marten home ranges, and forest patch configuration using resource selection functions within an information-theoretic framework. Tall well-stocked forests (TWF) >12m in height, whether mature, uncut forest or originating from previously clearcut forest, received similar selection, and selection by marten for TWF increased relative to other forest types as TWF availability decreased as a result of forest harvesting. Decreased TWF availability within marten home ranges was also associated with an increase in the relative preference marten displayed for less isolated and smaller patches of TWF. The negative effect of patch area suggests that marten increasingly utilized all available TWF patches, regardless of patch size, as this forest type became less abundant. These results indicate that marten require extensive use of TWF, which is limited when forest harvesting results in a scarcity of sparsely distributed TWF patches, and that regenerating clearcuts regain the structural characteristics selected by marten when trees reach 12m. Relative avoidance of scrub and early-successional clearcuts (height) was stronger among marten occupying home ranges where these forest types were abundant, indicating a risk or cost associated with individuals increasing their use of early-successional forest in proportion to increased availability. Partial harvests and mid-successional forest types (9m-12m tree height) received similar selection as well as extensive use by marten. These forest types were preferred relative to unforested areas, forested road edges, and scrub and early-successional clearcuts, suggesting that some harvested forest types constitute a hospitable matrix for marten. Marten responses in selection to the shape and area of hospitable matrix patches were consistent with marten use of hospitable matrix being primarily associated with movement between TWF patches, which indicates that marten selection of partial harvests and mid-successional forest is dependent on the selection of adjacent TWF patches. Patterns of marten habitat selection I documented suggest that the effects of forest harvesting on marten may be reduced if harvests are 1) positioned adjacent to multiple TWF patches, 2) do not separate TWF patches by more than 300m, and 3) are located within female marten home range sized areas (~2.5km2) with high proportions of TWF relative to the surrounding landscape. Small patches of TWF within occupied areas should not be preferentially harvested unless they are highly isolated, especially in areas with low relative TWF abundance. Additionally, minimizing road construction within areas occupied by marten, decommissioning established roads wherever future access is not an immediate management priority, and reducing the intensity of road edge maintenance will diminish the impacts of logging roads on marten. This research illustrates how animal location data collected across periods of habitat change can be used to precisely characterize species’ responses to the interrelated components of that change. Further, this study demonstrates how the components of habitat change can be quantified in ecologically meaningful ways that are also easily interpretable from a land management perspective, ensuring the translatability of results to readily implementable recommendations for habitat conservation.