2.2 Survey Methods
We aimed to replicate the historic survey methods as closely as possible to obtain comparable contemporary results. We undertook this task by following the detailed descriptions and survey maps provided in the old reports to re-locate the same survey areas for each site. At Point Pelee National Park, muskrat surveys were initiated in the 1950s by Parks Canada staff and standardized in 1963 when the marsh was divided into 14 survey zones, with a 15th zone added in 1971 (Reive, 1978). The zones remained consistent in subsequent years; however, a few boundary adjustments were made in 1979 to accommodate changes in the vegetation structure of the marsh (Reive, 1979). The surveys were conducted annually until 1980, after which time no further muskrat surveys were undertaken (Bremner and Reive, 1980). At Matchedash Bay-Gray Marsh, muskrat surveys were initiated in late fall 1978 by Ontario Ministry of Natural Resources Midhurst (formerly Huronia) District staff and continued annually until 1986, with the exception of 1982 when no surveys were conducted (LaFrance, 1986).
At both sites the basic survey method was a winter count of all active muskrat houses found within the respective search areas (survey zones). At Pelee the survey method covered only a portion of the entire marsh (the perimeter of all ponds along the open water-vegetation edge which is typically where muskrat house-building occurs) while at Matchedash the survey method aimed for a complete census of the entire marsh. We surveyed Pelee for muskrat houses during the winters of 2014 and 2015 and during the springs of 2017 and 2019 (four surveys spanning six years). We surveyed Matchedash for muskrat houses during the winter of 2014 and the springs of 2014-2018 (five consecutive survey years).
At Pelee we conducted surveys the first two years in winter (March 2014 and February 2015) on foot with two observers walking on the ice along the inner edge of all marsh areas (i.e., the perimeter of mapped ponds, as well as small channels and accessible pools) within each mapped survey zone, as depicted in the historic survey reports. Observers searched from the edge of the emergent vegetation to approximately 15 m inwards from the open water edge (or as far into the vegetation stands as we could detect muskrat houses) and counted all muskrat houses seen. This is the same survey method used in the historic surveys at Pelee, except that in some years the observers rode an ATV rather than walked. The latter two years of our surveys (2017 and 2019) were conducted by canoe, as ice conditions were not adequate in the winters of those years to allow surveyors to safely traverse the marsh. The canoe-based surveys followed the same routes covered on foot in the first two survey years, with the exception of one survey zone which was conducted on foot in chest waders as it did not contain enough open water to paddle a canoe through. When surveys were conducted by canoe, the lead observer was stationed in the bow in a kneeling position and frequently stood up to ensure a comparable search height to the surveys conducted on foot in winter. As well, spring surveys were conducted in early spring before new growth and green-up of the marsh vegetation so that visibility of muskrat houses was comparable to surveys conducted in winter.
At Matchedash we conducted the very first survey in late winter (March 2014) on a combination of foot and snowmobile with two observers covering the entire marsh (both perimeter and interior areas); however, the snow cover was very deep (over 5 feet) and we were concerned that muskrat houses might be buried and thus not detected. We verified this suspicion by returning to the site early the following spring (2 months later) and paddled the perimeter of the entire marsh by canoe, counting several muskrat houses that we missed due to snow depth in the winter. We know that these were not houses that were newly built in the interval between our winter and spring visits because we conducted the survey in May only a few weeks after ice-out which is not a time when muskrats conduct house-building in this region. Muskrats typically build houses in the fall before freeze-up and these remain in place over the winter until they start to deteriorate (if no longer in use) the following spring-summer (Dozier, 1948). Over the following four years (2015-2018) we conducted all surveys at this site in May by canoe as the Matchedash area typically receives heavy snow accumulation and thus we knew that detection of houses would continue to be hindered by snow cover in winter. The historic surveys, however, were conducted every year in winter by snowmobile (except for 1981 when it was conducted by boat in spring) and covered the entire marsh area, not just the perimeter of the vegetation stands which the canoe-based surveys were restricted to (even though this is where most muskrat houses typically occur). We compensated for this difference in survey coverage by using high-resolution aerial photographs of the study area (collected ourselves in March 2018 during low snow cover conditions) to count muskrat houses in the areas we could not access by canoe. We also used our aerial count of houses found in these inaccessible areas of the marsh, in combination with aerial verification of the houses found by ground survey, to calculate a percentage of the total count of houses missed by the spring ground survey (i.e. a field detection rate). We then applied that detection rate to the previous four survey years’ counts as a correction factor, yielding what we feel should represent the most accurate final house count possible for each year based on all available data. In other words, for each survey year we have a minimum count of muskrat houses observed from only the early spring canoe surveys and an estimated maximum count based on a combination of the canoe and aerial surveys (the latter presumably representing greater detection accuracy). We considered the estimated maximum count as our final house count for each year.
At Pelee we also used imagery-based house counts in 2017 and 2019 to supplement our field counts in the survey areas we could not access, similar to how we handled the Matchedash data. For this we used high-resolution imagery available from the County of Essex (Essex County, 2020) as well as imagery collected and shared by Point Pelee National Park. We used our final house counts (i.e. our estimated maximum counts) at both Pelee and Matchedash in all survey years to compute an average house count at each site for the contemporary survey period. Because we chose to use our estimated maximum house counts that were based on a combination of aerial and ground census methods, rather than our minimum ground-only observations (which were incomplete in some cases), we are confident that we have not underestimated the current muskrat population at either site by potential detection errors (i.e. missed houses) resulting from site conditions or incomplete survey coverage.
Historic house counts dating back to 1957 are available for Pelee, however a detailed breakdown of the count results by survey zone was not provided until 1967. Examining these data, we realized that not all areas of the marsh were surveyed consistently from year to year and that some years suffered from poor survey conditions (e.g., due to deep snow) which likely decreased house detection rates. Because we could not supplement the historic field data with imagery as we did for our recent surveys, we decided that a trend over time analysis of house counts would be more accurate if we omitted the survey years and zones where survey coverage was reportedly poor, incomplete or simply unknown due to a lack of information in historic reports. We therefore began our analysis of the Pelee muskrat population with the 1968 data and removed the years 1977-1978, as well the survey zones 14 and 15 in all years, from our final analyses to create a reduced but more consistent and comparable set of time series data. However, we also computed a mean house count for the 1968-1980 period at Pelee without excluding the two survey zones to compare with our final results. At Matchedash survey coverage was generally more consistent from year to year (with a couple minor exceptions) and thus we used all seven survey years and the annual results obtained for the entire survey area to calculate a mean house count for the historic period at that site.
As was done for the historic surveys, we used our annual counts of muskrat houses as indices of abundance for each site, a common method first described by Dozier (1948) and since undertaken by many researchers as a means to infer abundance and track annual change in muskrat populations (e.g., Greenhorn et al., 2017; Kroll and Meeks, 1985; Proulx and Gilbert, 1984; Toner et al., 2010).