COS 231-4 - Anthropogenic footfall is stronger than footprint in predicting fisher (Pekania pennanti) occurrence across landscapes

Background/Question/Methods

Species distribution models (SDMs) are one method to explain and then predict animal occurrences more broadly. Occurrence, however, is influenced by multiple ecological processes that vary across landscapes and may not be fully captured in models. This can lead to misrepresentation of the importance and strength of the specific habitat features driving distributions. Fishers (Pekania pennanti) are a mesocarnivore known for their requirement of mature forest structure, yet the influence of other abiotic and biotic factors in driving their distribution is largely inconclusive. Therefore, the aim of this study was to determine the strongest suite of predictors for fisher occurrence across habitats that vary in abiotic and biotic conditions, and anthropogenic disturbance densities. We used camera traps deployed systematically across three landscapes in Alberta, Canada to sample fisher occurrence, and then used generalized linear mixed-effects models to weigh evidence for our competing hypotheses: that fisher occurrence is best explained by (1) natural land cover; (2) anthropogenic landscape disturbance; (3) snow and topography; and (4) co-occurrence of heterospecifics.

Results/Conclusions

Anthropogenic disturbance features generated by extensive resource extraction, roading, and cultivation best predicted fisher occurrence across the landscapes (∆AIC = 0.00, AIC_w ~ 1.00). Occurrence was differentially explained by “footfall” – actively used anthropogenic features – than “footprint” – unused anthropogenic features left from past activity. In contrast, natural landscape features, including dominant forest cover, were not the most significant predictor of fisher occurrence (∆AIC = 26.25, AIC_w ~ 0.00). These findings demonstrate the need for species distribution models (SDMs) to consider landscape scale ecological processes through including multi-habitat observations across a species range. Furthermore, specific features and conditions must be parsed apart from one another to ensure differential occurrence patterns do not negate one another. These considerations can extend to species management plans, where SDMs could provide a comprehensive statistical representation of species-landscape feature associations to increase the effectiveness of proposed conservation strategies.