4 Discussion
Our surveys demonstrate substantial declines in muskrat populations at two large coastal wetlands in Ontario based on direct field observations spanning 40-50 years. The data we present are not confounded by trapper effort like harvest data often are, but they correspond to reported trends in provincial harvest data and are consistent with anecdotal evidence of muskrat decline. Ahlers and Heske (2017) called for direct evidence of muskrat population trends to avoid confounds with harvest effort, and we provide such evidence here.
It has been suggested that most survey techniques for muskrats are only feasible across small geographic extents and may not be useful for examining large-scale population trends (Roberts and Crimmins, 2010). While this may be true, we have two independent data sets from widely separated wetlands illustrating declines of similar magnitude over a similar time frame, which are also coincident with declines in our provincial harvest data over the same approximate time period. Collectively this suggests a broader trend of decline in muskrat populations across Ontario.
The greater than 90% decline in muskrat house abundance at each of our study sites is startling, yet we are confident the survey data indicate a true decline of this magnitude. In many wildlife surveys there is potential for over-estimating or under-estimating abundance as a result of observer error. In the case of muskrat house surveys, some houses may be missed, and some structures may be falsely labelled as active houses when they are in fact only feeding structures or inactive houses, or vice versa. However, we assume that in our study, on average, detection probabilities were similar from year to year (except the survey years that were deliberately eliminated from the analyses because the surveyors reported poor conditions that hindered their ability to detect houses), and that in most cases muskrat structures were correctly identified. We attempted to replicate the historic survey methods as closely as possible so that the results we obtained could be largely attributed to actual population change and not simply methodology differences. Furthermore, considering that our final house counts for each recent survey year at each site were adjusted based on imagery data that were used to supplement our field counts, and it is more difficult to distinguish new houses from old ones on air photos, it is more likely that we have over-estimated contemporary muskrat abundance by including houses in our tally that are no longer in use. Therefore, our estimates of the magnitude of muskrat population change at each site are likely conservative.
Muskrats have been shown to fluctuate periodically, thus it is possible that low house counts may be reflecting a low phase in a normal population cycle. Erb et al. (2000) examined ninety-one historic time series of muskrat harvest data from boreal, taiga and southern Arctic regions of Canada and found overall that mean period length ranged from 3.7 to 8.6 years. If the same pattern holds true in southern Ontario and the start of our surveys corresponded to a population low phase, we would expect to have seen a greater increase in our recent house count numbers over the five to six years we conducted the surveys. We observed no such increasing trend, and similarly, there is no increasing trend evident in muskrat harvest data collected over the same time period. Instead, it appears that muskrat populations at Point Pelee and Matchedash Bay-Gray Marsh are currently persisting at very low levels and are much reduced from the levels seen 30-50 years ago. Furthermore, harvest data suggest that these declines are widespread.