Marsh interspersion and muskrat (Ondatra zibethicus) habitat use

Muskrat (Ondatra zibethicus) populations have been declining in North America for decades. The precise cause of these widespread declines has not yet been identified. Over a similar timeframe, wetlands across large regions of North America have been experiencing an invasion of hybrid cattail *Typha *x glauca. This invasion is associated with many negative consequences for wetlands, including a reduction in biodiversity, open water habitat, and interspersion of water and vegetation. Muskrats are strongly tied to wetlands, especially where there is a high degree of interspersion of water and emergent vegetation. Therefore, a widespread reduction in interspersion caused by *T. *x *glauca *invasions may be contributing to widespread muskrat population declines. We sought to understand the impact of reduced marsh interspersion on fine-scale muskrat habitat use which will shed more light on broad-scale population trends. We measured intensity of habitat use by muskrats in a large, Typha-dominated marsh in south-central Ontario using camera traps, stratifying camera placement along a gradient of marsh interspersion. We found no correlation between interspersion and intensity of use, suggesting that factors other than interspersion may drive intensity of use. The best predictor of intensity of use in our study was the presence of channelized water features. Our study site, like most marshes in the region, was highly dominated by *T. *x glauca. Further research is needed to determine the impact of *T. *x *glauca *invasions on muskrats, as well as the cause of widespread muskrat declines.   

Methods
We used camera traps to measure intensity of use by muskrats along a gradient of marsh interspersion. We used aerial imagery and land cover classifications in ArcGIS Pro to measure interspersion ("interspersion"). We used Pearson correlation to determine correlations between intensity of use and interspersion along with other predictor variables, as well as zero-inflated negative binomial models to model intensity of use using these predictor variables. Aside from interspersion, other predictor variables used in our statistical analyses included surrounding water area ("water area"), whether the surrounding habitat was channelized ("channelization"), season ("sampling period"),  an index of camera viewshed obstruction ("viewshed obstruction"), and an index of the extent of surface water visible within the viewshed ("sample area"). We used R to analyze the data using the following packages: dplyr, Hmisc, pscl, and PerformanceAnalytics.