Snowshoe Hare Pellet Count Surveys
Red Squirrel Surveys
Estimating Snowshoe Hare and Red Squirrel Relative Abundance
The snowshoe hare comprises between 76-94% of the lynx diet during periods of hare abundance (Nellis et al. 1972, Brand et al. 1976, O'Donoghue et al. 1998). Red squirrels may have an important secondary role in the lynx diet, particularly when hares are scarce or during snow-free months (O'Donoghue et al. 1998, Aubry et al. 2000). Grouse, small mammals and carrion seem to be less important components of the lynx diet (Aubry et al. 2000). This high reliance on one or two prey species indicates that lynx research should concurrently consider the ecologies of these herbivores.
Snowshoe hares and red squirrels have been studied successfully with indirect, noninvasive methodologies. Indirect methods of studying these species include various protocols designed to provide a population index of abundance without the intensive commitment required with mark-recapture trapping or radiotelemetry. For hares, fecal pellet counts have been commonly used as indirect survey methods (Krebs et al. 1987, 2001a, Murray et al. 2002). Due to the vocal nature of red squirrels, point count transects similar to those performed on birds can be used as an indirect means of obtaining abundance estimates for this species (Mattson and Reinhart 1996, Bayne and Hobson 2000). Densities and distributions of hare obtained from snow-tracking (and pellet counts, see below) will be compared with historical records from track counts conducted by MNDNR.
![]() |
| Figure 4. Snowshoe hare track indices and spring pellet count indices in Minnesota from 1974 to 2002. Data from an unpublished report by J. Erb, MNDNR. |
The goal of these prey species surveys is to obtain indices
of relative abundance for each species. These indices will
be analyzed across various habitats, spatial scales, and temporal
scales. Although the data collected on these prey species can
be analyzed independently of lynx data, information on relative
prey abundance will have a central role in lynx habitat analyses.
All hare fecal pellet transects, whether randomly placed
in habitats according to availability, systematically placed
in lynx home ranges, or stratified in high quality hare habitats,
will consist of five 1 m2 circular plots placed at 20 m intervals.
Large circular plots have recently been suggested as the preferred
method for inferring hare density in southern portions of snowshoe
hare range (McKelvey et al. 2002, Murray et al. 2002). All
plots will be permanently marked with a reinforcing bar (rebar)
stake at 3/8” diameter and revisited each May-June for
the duration of the project. During counts all fecal pellets
within the plot boundary will be counted and removed. To avoid
an inclusion bias among technicians, only 50% of the pellets
found directly on the plot boundary will be counted (McKelvey
et al. 2002). Vegetation obscuring pellets will be moved as
needed but pellets deeply incorporated into the organic layer
of the forest floor will not be counted. Over 180 permanent
transects were established in the spring of 2003 according
to the methods described above. The data sheet is in Appendix
4.
The fecal pellet data will permit the development of an index of relative density
across various habitat types that can also be analyzed at a variety of spatial
scales, from microhabitat to landscape-scale. Microhabitat analyses will be
based on stand-level vegetation measures collected at each pellet plot. Landscape-scale
analyses will be performed on a GIS. These multiple-scale analyses of the hare
data will allow us to estimate which spatial scale has the greatest influence
on hare populations, vital information for forest managers. For statistical
analysis, the transect will be made the experimental unit by summing counts
across transects. Counts will be normalized with a log-transformation and differences
between years and among habitat types will be tested with an analysis of covariance
(ANCOVA).
The fecal pellet data will permit the development of an index of relative density
across various habitat types that can also be analyzed at a variety of spatial
scales, from microhabitat to landscape-scale. Microhabitat analyses will be
based on stand-level vegetation measures collected at each pellet plot. Landscape-scale
analyses will be performed on a GIS. These multiple-scale analyses of the hare
data will allow us to estimate which spatial scale has the greatest influence
on hare populations, vital information for forest managers. For statistical
analysis, the transect will be made the observational unit by summing counts
across transects. Due to the anticipation of large numbers of zero counts,
it may be nearly impossible to transform the data to make them normally distributed.
Since count data often follows a Poisson distribution, various Poisson methods,
such as a Poisson ANOVA or regression, may be used. However, in a general sense,
the differences-between-habitats question will be answered with ANOVA methods.
Multiple comparisons within the overall ANOVA will be used to test between
pairs of vegetation types.
Another question of interest particularly applicable to a cyclic species such
as lynx are trends among years. Because future funding may not be available
for long-term monitoring of these pellet transects, we do not plan long-term
analyses involving trend analysis at this time. Therefore, to address between
year variability we will utilize a 2-way ANOVA in the habitat analysis and
include year to test whether the differences between habitats were different
between years.
A point count methodology can be a cost-effective method of acquiring baseline information on the habitat use and population dynamics of red squirrels. Red squirrel counts will be conducted in the same randomly selected stands delineated among available habitats that were used for the snowshoe hare pellet counts. NRRI has an existing source of quantitative data on red squirrels collected during forest bird point-counts across the Superior National Forest and other areas in the upper Great Lakes states (Howe et al. 1998). We have about five years of existing data that is added to each year. Similar statistical analyses described above for the snowshoe hare fecal pellet plots will be used for the red squirrel data to address differences between vegetation types and years.
Combining the use of pellet-surveys and point-counts with a track-based methodology will provide additional support to strengthen our evaluations of the data as well as help determine the utility and effectiveness of the various approaches. During the winter, line-transect sampling has proven to be useful for hares and squirrels (Thompson et al. 1989). To simplify data collection, all lynx prey species tracks encountered during the lynx/mesocarnivore track surveys will be counted. Indices of relative abundance will be developed for these lynx prey species and analyzed in relation to the landscape-scale composition of forest-types in the SNF. Compositional analysis (Aebischer et al. 1993) or similar distance-based classification schemes (Conner et al. 2003) will be used in the analysis.