Spatial Ecology Of Eastern Box Turtles (Terrapene C. Carolina) In Central Massachusetts

Eastern box turtles are declining throughout their range as a result of habitat loss and fragmentation, associated road mortality, collection for the pet trade, and other localized threats such as incidental collection, disease, mowing, and fire. We evaluated the geomorphic habitat suitability for Eastern box turtles in Massachusetts using a Mahalanobis distance modeling approach. To evaluate habitat use, home range size, and box turtle population status in the Connecticut River Valley, Massachusetts, we conducted a four year mark-recapture and radio-telemetry effort. We surveyed over 30 sites, captured 191 animals and radio-tracked 91 animals at eight sites for 1-4 years for a total of 217 turtle-years. Estimated population sizes at these sites ranged from a summer high of 3.8 to a winter low of 0.3 turtles / ha. Home range size ranged from 140 - 2145 m straight line and 0.5 - 136 ha. These densities are much lower and home range sizes much larger than reported throughout the rest of the species range. Conservation efforts based on data from previous studies elsewhere in the species range would have underestimated habitat requirements and space needs, highlighting the need for local and regional information when planning for rare species conservation. We evaluated over-wintering habitat at multiple levels of habitat selection and spatial scales using logistic regression and classification trees in conjunction with remotely sensed geomorphic and land-use variables. We also conducted a micro-habitat assessment of over-wintering sites using variables measured in the field. Microtopography proved important, with 59% of transmitting turtles over-wintering in depressions (typically behind tip-up mounds of fallen trees). Other variables distinguishable between turtle and random locations were high basal areas of hardwood trees and abundant downed wood and mountain laurel. These results can be used to inform management practices and focus future survey efforts. We evaluated the influence of landscape composition and structure on the movements of box turtles using mixed effects linear regression models. Eastern box turtles have smaller home ranges in more urbanized landscapes and move further in more forested, less fragmented areas. The relationship varies significantly by site, underscoring the importance of evaluating effects across multiple sites, and suggesting that single-site studies may not be generalizable throughout regions. Population density estimates from five of the sites suggest a unimodal rather than linear relationship with urbanization. The highest-density site had a moderate level of fragmentation, suggesting that low levels of urbanization may be compatible with Eastern box turtle conservation, but that high levels of fragmentation may be detrimental to box turtle populations. Early successional habitat comprised more than 50% of June locations, suggesting these habitat features are important and may be limited on the landscape. We evaluated whether habitat management to create such areas could be effective by using utility distribution volume and a randomization approach to assess habitat use before and after forest was cleared at two sites. Use of the treatment areas increased significantly after clearing. Our results suggest that Eastern box turtles are willing to incorporate newly cleared areas into their home range within one year, and that it may be possible to improve habitat and minimize movement distance through active management. We also characterized seasonal movement and activity patterns, nesting habitat and reproductive parameters, and adult mortality rates. We suspect that mortality rates are much higher statewide and that habitat loss is likely the largest cause of mortality. Using our observed clutch size, nest success rate, and adult mortality rate and a deterministic population model, we estimated that hatchling and juvenile survival rates must exceed 77.6% annually in order to maintain a stable population in the absence of stochastic events. This survival rate is unrealistically high, particularly since the model does not account for variability and we suspect that our adult survival rate was artificially inflated, suggesting that populations may be declining in the Connecticut River Valley, even at protected sites. (Abstract shortened by UMI.