Influence of Microhabitat on the Local Distribution of White Footed Mice (Peromyscus leucopus) and Red-Backed Voles (Clethrionomys gapperi) in Red Maple (Acer rubrum) Swamps and Contiguous Uplands in Rhode Island

Microhabitat choice by white-footed mice (Peromyscus leucopus) and red-backed voles (Clethrionomys gapperi) was studied in three Rhode Island red maple (Acer rubrum) swamps and their contiguous transition and upland zones. Significant differences were found in microhabitat use between the two species in each zone. High woody stem density (stems m2) and low herbaceous plant species richness were important factors determining white-footed mouse occurrence. Red-backed voles were found in areas of high density of shrub cover, as well as high herbaceous plant density (stems m2) and richness.

--Designation, descriptions and sampling methods for 29 characteristics of microhabitat structure in red maple (Acer rubrum) swamps, and contiguous transition and upland zones at three Rhode Island sites.                    Red maple swamps reach their greatest abundance in southern New England and northern New Jersey where they comprise 60-70% of the total area of palustrine wetlands (Golet, et al., in press). Because of their abundance, red maple swamps are among the most frequently altered wetland types.
Despite their abundance there is very little published information on the fauna of red maple swamps.
The distribution and abundance of small mammals are probably a function of habitat suitability (Vaughan, 1972:250-256), or more specifically, microhabitat suitability (Baker, 1968). Some small mammal species are limited in spatial distribution by specific habitat needs; others occupy a wide range of habitats (Kaufman and Fleharty, 1974;Kirkland and Griffin, 1974;Briese and Smith, 1974;Miller and Getz, 1977;Geier and Best, 1980).
Microhabitat use by small mammals has been documented for upland deciduous and coniferous forests in the East (Dueser andShugart, 1978, 1979;Kitchings and Levy, 1981;Adler, 1985;Parren and Capen, 1985;Seagle, 1985), but similar knowledge of small mammals in red maple swamps and bordering upland forests in the northeastern United States is limited. Dowler et al. (1985)  shrub cover had relatively high mammalian richness (9 species) and diversity (Shannon's H'2 = 2.21) (Miller and Getz, 1977). Differences in water content of food appeared to explain differences in local distributional patterns of short-tailed shrews (Blarina brevicauda) and red-backed voles (Clethrionomys gapperi), both of which have high water requirements (Getz, 1968).
The purpose of my study was to provide baseline information on: 1) small-mammal communities in Rhode Island red maple swamps and their contiguous upland forests as well as the transitional zone (ecotone) between them, and 2) the relationships between small mammals and their habitats. The ecotone is a transition between two or more diverse communities; each commonly containing many of the organisms of each of the overlapping communities (Odum, 1971 Criteria for selection study sites were: 1) size ~30 ha; 2) inclusion of a red maple swamp bordered by upland forest; 3) predominance of very poorly drained soils in the swamp (Rector, 1981); 4) low to moderate (~15%) slopes in surrounding upland (Rector, 1981); 5) presence of homogeneous forest cover ~85% deciduous; 6) canopy cover ~75%; 7) mature forest (minimum stand height of 10 m); 8) lack of recent human disturbance (~40 yrs); and 9) presence of a 100-m buffer around sampled areas.
At each site three parallel transects were located 90 m apart and perpendicular to the wetland boundary. A 50by 50-m grid was then located in the upland, transition and 3 wetland of each site (Appendix E).
The grids were positioned such that they were bisected by the transect line, and in the transition zone they were also bisected by the very poorly drained soil edge (Appendix F In wooded swamps, trees and shrubs grow primarily on mounds which are raised above the swamp's seasonal high water level (Golet et al., in press (Melchior and Iwen, 1965) and released at the trap site after species, weight (to nearest gram), sex, age (juvenile, subadult, adult), and reproductive condition (scrotal, abdominal, pregnant, lactating, none, unknown) had been recorded.
Vegetation sampling.--Sampling followed the methods outlined by Dueser and Shugart (1978 (Getz, 1961~;Doucet and Bider, 1974); moisture (Chenoweth, 1917;Chew, 1951;Getz, 1961&, 1961£;Miller and Getz, 1977); temperature (Brower and Cade, 1966;Getz, 1961~, 1961£); and penetration resistance of the substrate (Dueser and Shugart, 1978;Jameson, 1949). Because these variables are dynamic, they were sampled each trap day throughout the 1991 field season. Air and soil temperatures were determined by an array of three maximum/minimum thermometers (Fig. 2), with the base of one thermometer each at 7 cm below the ground surface, and 25 cm and 100 cm above the ground surface (Getz, 1961£). One thermometer 9 was calculated using frequency of occurrence and basal area (Mueller-Dombois and Ellenberg, 1974 (Getz, 1961~;Doucet and Bider, 1974); moisture (Chenoweth, 1917;Chew, 1951;Getz, 1961Q, 1961Q;Miller and Getz, 1977); temperature (Brower and Cade, 1966;Getz, 1961~, 1961Q); and penetration resistance of the substrate (Dueser and Shugart, 1978;Jameson, 1949). Because these variables are dynamic, they were sampled each trap day throughout the 1991 field season. Air and soil temperatures were determined by an array of three maximum/minimum thermometers (Fig. 2), with the base of one thermometer each at 7 cm below the ground surface, and 25 cm and 100 cm above the ground surface (Getz, 1961Q feature some models were better at predicting a species absence than its presence at a zone. The average species richness per zone was calculated by the jackknife method (Heltshe and Forrester, 1983).
Four different regressions using the habitat zones at each study site as sample points at each were performed with small-mammal species richness as the dependent variable and overstory, understory, shrub, and herbaceous vegetation layer richness as single independent variables (n = 9, 7 d.f.). All means are reported with their respective standard deviations unless otherwise indicated.
Differences were considered significant at~ ~0.05. SAS programs (SAS Institute, Inc., 1985) were used to sort data and conduct all analyses.

RESULTS
Mammal Captures.
--Seven small mammal species were captured on the three sites (Table 2). White-footed mouse was the species most frequently captured. Captures in the upland were 5.4/100 trap nights (htn), 5.9/htn in the transition zone, and 5.7/htn in the wetland. Red-backed voles and masked shrews (Sorex cinereus) were the only other two species caught in all three zones. The site with the highest capture rate for white-footed mouse was Burlingame (9.0/htn), which also had the lowest capture rates of red-backed vole (1.0/htn). This is likely a reflection of an absence of red-backed vole captures in the Burlingame transition zone. The average species richness per zone was 5.7 ± 1.4 in the upland, 6.7 ± 1.4 species in the transition, and 3.0 + 1.4 in the wetland ( Table 3).
All of the eastern chipmunks (Tamias striatus), 33% of the white-footed mice, 30% of the red-backed voles, and 22% of masked shrews were captured in upland zones (Table 4).
White-footed mouse and red-backed vole accounted for most of the 101 captures in the upland zones (67% and 25%, respectively Of all the captures, 18% of white-footed mouse capture points overlapped with red-backed voles and 43% of 13 red-backed vole points overlapped with white-footed mice (Table 5). Thirty-three percent of masked shrew points overlapped with white-footed mice and 22% overlapped with red-backed voles. A common resource used to measure the niche overlap of two species is space or microhabitat use (Krebs, 1989). The percentage of a generalist's capture points that also are a specialist's capture points will be smaller than the percentage for the converse.
Of the 131 total capture points for the white-footed mouse, 23 points (18%) also had captures of red-backed voles.
Of the 54 total capture points for the red-backed vole, 23 points (43%) also had captures of white-footed mice.
Microhabitat.--Because of the low capture rates for most species, microhabitat was analyzed only for white-footed mice and red-backed voles.
Upland canopy coverage averaged about 81%, of which evergreens accounted for <5% (Table 6) plant species found in the overstory, 11 occurred in the upland, 16 in the transition, and 6 in the wetland (Table   7). I found 26 plant species in the understory with 18, 19, and 12 occurring in the upland, transition, and wetland, respectively (Table 8).
In the shrub stratum, 12 out of 21 plant species occurred in the upland, 15 in the transition, and 6 in the wetland (Table 9). In the herbaceous stratum, the upland contained 12 of the total 22 plant species , 15 in the transition, and 6 in the wetland (Table 10).
Species In general, areas of white-footed mouse occurrence were characterized by high woody stem density, tree stump dispersion, overstory tree size, soil surface exposure, and low evergreen coverage, sphagnum moss exposure, and evergreen herb stratum (Table   11). Significant differences between white-footed mouse capture and noncapture points were detected in 21% (5/24 ) of the microhabitat variables at AS and BG, and 25% (6/24) at GS. White-footed mouse occurrence in Arrow Swamp was in areas of relatively high tree stump size and evergreen closure , and low herbaceous foliage profile density, understory tree size, and tree stump dispersion.
In Burlingame, white-footed mouse microhabitat was characterized by high numbers of woody species, evergreen shrub cover, evergreenness of herbaceous stratum, and soil surface exposure, as well as by low sphagnum moss exposure.
Great Swamp microhabitat for white-footed mice was characterized by high overstory tree size, and soil surface exposure, as well as by low numbers of herbaceous species, percent total canopy coverage, evergreen coverage, and evergreen herbaceous stratum.
Significant differences in white-footed mouse capture and noncapture points also were detected in 33% (8/ England forests may be related primarily to the diversity of trees and shrubs, which directly or indirectly reflects a greater diversity of available food types (Miller and Getz, 1977). In this study, higher plant species richness values likely reflect a higher diversity of food items.
Greater plant species richness also may result in a more diverse phenology of food production (stem, leaf, or fruit ) (Martin et al., 1951, Graves, 1952Sutton and Sutton, 1985). A lower plant species diversity may result in food being produced at one or only a few points in the growing season.
Though masked shrews were caught in each grid, the most were captured in the Great Swamp wetland grid. Masked shrews were caught where there was a carpet of sphagnum moss.
As expected, white-footed mouse captures occurred in all grids and all sites. Densities of white-footed mice are higher in upland sites, compared to swamps, which contain seed, fruit, nut etc. producers that provide a reliable year-round food supply (Getz, 1961Q).
In this study, no statistical difference was found in the numbers of white-footed mouse captures between zones. The white-footed mouse appears to be a microhabitat generalist.
Unexpectedly, red-backed vole captures occurred in all grids except the transition zone at Burlingame. Captures of red-backed voles in the upland zones are in conflict with most of the existing literature, which indicates their dependence upon moisture (Chew, 1951;Brower and Cade, 1966;Getz, 1968;Kirkland and Griffin, 1974;Miller and Getz, 1977;Degraaf and Rudis, 1983). Red-backed voles have a water requirement 2.2 times that of white-footed mice (Getz 1968 white-footed mice prefers a high stern density. This preference for high stern density also was reported as high density of shrub-understory vegetation (Dueser and Shugart, 1978;M'Closkey and Lajoie, 1975;Seagle, 1985). My study found that the white-footed mouse preferred stone walls in the transition zone, confirming a preference for rocks noted in previous studies (Lackey et al., 1985). The red-backed vole was present in Arrow Swamp's thickets of mountain laurel (Fig. 6), which had a low density of sterns with many leafy lateral branches. Similarly, red-backed voles in Connecticut were more abundant at sites with >50% shrub cover (Miller and Getz, 1977). In this study, the average shrub coverage was 53% ± 6.7. In the transition zones, red-backed voles were primarily caught on the very poorly drained side of the transition grid. A high mound-pool ratio (drier) predicted the absence of red-backed voles in the transition zone in the logistic regression model. Red-backed voles are restricted to low, wet areas, often with standing water, or to moist areas where the water content of the vegetation was high as a result of high soil moisture (Getz, 1968). Mound/pool topography is an indicator of low wet areas. 24 Red-backed voles in New England occur in cool, moist forests with mossy rocks, logs, tree roots, or other cover; they also require water sources such as springs, brooks, or bogs, and debris cover such as fallen trees, stumps, rocks, or slash (Degraaf and Rudis, 1983).
In the wetlands, mound-pool topography and evergreen coverage were predictors of red-backed vole presence.
Burlingame's wetland, which had a high mound-pool-ratio and a low value for evergreen coverage, yielded the lowest capture of red-backed voles. Water appears to be a key factor in the local distribution of the red-backed vole.
A common resource used to measure the niche overlap of two species is space or microhabitat use (Krebs, 1989 Average distance (m) from trap station to nearest overstory tree, in quarters (Cottam and Curtis 1956).
Average of 4 densiometer readings centered on trap station.
Average of 4 densiorneter readings, for presence of evergreen canopy, centered on trap station.

Understory
Average diameter (cm) of the nearest understory tree, in quarters around trap station (Cottam and Curtis 1956).
Average distance (m) from trap station to nearest understory tree, in quarters (Cottam and Curtis 1956).

Shrub Stratum
Woody species count within a l-m2 ring centered on the trap station.
Live woody stem count at ground level within a l-m2 ring centered on the trap station.
Live woody stem count within a l-m2 ring centered on the trap station (stems <0.40 meters in height) Average number of waist-height contacts (tree and shrub) along center lines of two perpendicular 10 m2 transects centered on trap station (James and Shugart 1971).
Average of 4 densiometer readings, for shrub-level vegetation, centered on trap station.
Average of 4 densiometer readings, for presence of evergreen shrub-level vegetation, centered on trap station.

Herbaceous Stratum
Herbaceous species count within a l-m2 ring centered on the trap station.
Live herbaceous stem count at ground level within a l-m2 ring centered on the trap station.
Live herbaceous stem count within a l-m2 ring centered on the trap station (stems ~0.40 meters in height) Average numbers of live herbaceous stem contacts with a 0.80-cm rod rotated 360 degrees, describing a l-m2 ring centered on the trap station and parallel to the ground, at heights of 0.05, 0.10, 0.20, 0. 40, 0. 60, ... , 2. 00 m above ground level.  (Dueser and Shugart 1978).
Average diameter (cm) of the nearest tree stump ~ 10 cm in diameter and~ 1.00 min height, in quarters around trap station.
Average distance (m) from trap station to nearest tree stump ~ 10 cm in diameter and~ 1.00 m in height, in quarters.
Average total length of fallen logs > 10 cm in diameter, per quarter.
Average distance (m) from trap station to nearest fallen log~ 10 cm in diameter, in quarters.
Average diameter (cm) of the nearest fallen log~ 10 cm in diameter, in quarters around trap station.
Percentage of points with exposed mineral soil or rock, from 21 step-point samples along center lines of 2 perpendicular 10 m2 transects centered on trap station (Dueser and Shugart 1978).
Percentage of points with sphagnum moss, from 21 step-point samples along center lines of 2 perpendicular 10 m2 transects centered on trap station (Dueser and Shugart 1978).
Ratio of the number of points with mound topography to those with pool topography.

Humus Layer
Average of four depth measurements taken within the l-m2 ring centered on the trap station. A 1.27 cm diameter wooden dowel was inserted into the duff until contact with mineral soil.
Average of four penetrometer readings of the duff layer taken within the 1-m2 ring centered on the trap station (Dueser and Shugart 1978).   Tami as striatus     1111----- Thelypteris noveboracensis x Symplocarpus foetidus Grass and sedges Mitchella repens x x (+) Average value of variable was significantly greater (f ~0.05) at Peromyscus !eucopus capture points than at non-capture points.
(-) Average value of variable was significantly lower (f ~0.05) at Peromyscus leucopus capture points than at non-capture points.      Of their five upland and wetland study sites at Great Swamp National Wildlife Refuge in New Jersey, the one dominated by red maple, sweet gum, and American hornbeam produced the highest numbers of small mammals (Dowler et al., 1985). No single forest type in Vermont had significantly higher mammalian diversity than any other, whereas in Connecticut, red maple swamps with 50-75% shrub cover had relatively high mammalian richness (9 species) and diversity (Shannon's H'2 = 2.21) (Miller and Getz, 1977).
Most species of small mammals in New England are primarily graniverous or insectivorous, while the most common sources of seeds are from trees and shrubs, especially mast from trees. The combined species richness of trees and shrubs at a site was the only variable correlated (+) with species diversity of small mammals; diversity and availability of food may affect the diversity of forest small mammals in New England (Miller and Getz, 1977).
In general, the diversity of forest species of small mammals in New England is related primarily to the diversity of the trees and shrubs, as they directly or indirectly reflect a greater diversity of available food types (Miller and Getz, 1977). Differences in water content of food appeared to explain differences in local distributional patterns of short-tailed shrews and red-backed voles, both of which have high water requirements (Getz, 1968;Miller and Getz, 1977). Getz (196lh) described white-footed mouse habitat as areas in which the cover is in the form of shrub stratum or fallen trees as debris.
He found higher densities of white-footed mice in upland sites compared to swamps; he attributed this to lower densities of mast producers (primarily oak and hickory) which produce a reliable year-round food supply.
The white-footed mouse is nocturnal, and because the microclimates of swamps and uplands are similar at night, water balance is not a factor in the local distribution of this species (Getz, 1968). White-footed mice have a generalized distribution and reach their highest densities in brushy fields and in woodlots dominated by deciduous trees Hallet et al., 1983).
Microhabitat features that determine the distribution and abundance of white-footed mice are: deciduous canopy and low shrub evergreenness (Dueser and Shugart, 1978); high density of shrub-understory vegetation (Dueser and Shugart, 1978;M'Closkey and Lajoie, 1975;Seagle, 1985); and high plant species richness in herbaceous and shrub strata (Parren and Capen, 1985 1985). White-footed mouse densities increase with increasing shrub species richness, and generally are positively associated with woody microhabitat or negatively associated with herbaceous habitats (Adler, 1985).
In Southern New England, red-backed voles are restricted to low, wet areas, with standing water or an accessible water table (Getz, 1968). Red-backed voles aref ound in moist areas where the water content of the vegetation is directly influenced by the soil moisture (Miller and Getz, 1977). In Vermont, red-backed voles are less abundant in sites with less tree cover, while in Connecticut they are more abundant in sites with >50% shrub cover and in sites with >25% herbaceous cover. Red-backed voles also have been found in mesic forest habitats with an abundance of stumps and exposed roots (Merritt, 1981); a high density of fallen logs (Merritt, 1981;Belk et al., 1988); and dense canopy cover and dense woody vegetation (Belk et al., 1988 was the principle food of 75% of the red-backed voles sampled (Hamilton, 1941).
The eastern chipmunk occurs at sites with primarily deciduous canopy, a high density of trees, a low density of shrubs, and high shrub evergreenness (Dueser and Shugart, 1978). Eastern chipmunks also occur in bushy habitats as well as forest, especially where there was an abundance of crevices for refuge (Snyder, 1982). The eastern chipmunk is a woodland generalist, using a wide range of microhabitats within forests, and exhibits no particularly strong association with any one type (Kitchings and Levy, 1981).
The short-tailed shrew has more than double the water equilibrium of Peromyscus at a temperature of 610 C, and receives half of its water intake from the foods eaten (Chew, 1951). Even so, short-tailed shrews drink water, demonstrating that a need to drink may still exist even if a considerable amount of water is received from food.
Short-tailed shrews are subjected to less loss of body moisture from evaporation because of its subterranean 70 habitat (Chew, 1951).
Short-tailed shrews are found in moist habitats, but not in standing water (Getz, 1961Q).
They are scarce or absent in these habitats because the available food supply (consisting of large forms of invertebrates) are also low or absent. Short-tailed shrews are approximately twice as abundant in areas with > 50% herb cover than in areas with < 50% herb cover, and slightly more abundant in the more moist sites of Connecticut (Miller and Getz, 1977). Short-tailed shrew habitat is deciduous woodlands, but they rarely occur in areas with heavy undergrowth (Kitchings and Levy, 1981).
They consistently occupy areas of high stump and log density, hard ground, few shrubs, and dense overstory.
Masked shrews are less abundant in drier habitats than were short-tailed shrews and they did not avoid standing water (Getz, 1961Q). They may be able to better utilize smaller prey items (collembolans, ants, spiders) than do short-tailed shrews. Masked shrew activity increases or decreases with corresponding changes in the mean nighttime temperature from that of the previous night (Doucet and Bider, 1974). They also increase activity on cloudy nights; the highest activity was recorded with rainfall from 1800-2400h.
Smoky shrews are restricted to the cool forested regions of Pennsylvannia, New York, and New England in habitats with a ground cover of loose leaf mold (humus) and black friable soil (Hamilton, 1940).  Appendix J.