«Urban Ecosystems, 8: 215–225, 2005 c 2005 Springer Science + Business Media, Inc. Manufactured in The Netherlands. Variation in winter microclimate ...»
Urban Ecosystems, 8: 215–225, 2005
c 2005 Springer Science + Business Media, Inc. Manufactured in The Netherlands.
Variation in winter microclimate and its potential
inﬂuence on Virginia opossum (Didelphis virginiana)
survival in Amherst, Massachusetts
L. LEANN KANDA∗ email@example.com
Organismic and Evolutionary Biology Graduate Program, 319 Morrill Science Center South,
University of Massachusetts, Amherst, MA 01003 TODD K. FULLER Organismic and Evolutionary Biology Graduate Program, 319 Morrill Science Center South, University of Massachusetts, Amherst, MA 01003; Department of Natural Resources Conservation, 160 Holdsworth Way, University of Massachusetts, Amherst, MA 01003 PAUL R. SIEVERT U.S. Geological Survey, Massachusetts Cooperative Fish & Wildlife Research Unit, Department of Natural Resources Conservation, University of Massachusetts, Amherst, MA 01003 KEVIN D. FRIEDLAND UMass/NOAA CMER Program, Blaisdell House, University of Massachusetts, Amherst, MA 01003 Abstract. If climate limits the geographical distribution of a species, local variation in microclimate may affect the species’ local distribution at the edge of its range. We hypothesized that warm urban microclimates may explain the distribution of the Virginia opossum (Didelphis virginiana) in central Massachusetts. We recorded winter temperatures with data-logging sensors in urban, coniferous, deciduous, and open habitats in the humandominated landscape of the Connecticut River Valley of Massachusetts. Overall, temperatures decreased with elevation. Daily maximum temperatures, a variable used in models of opossum biophysical constraints, were lowest at forested sites, intermediate at urban sites, and highest at open sites; however these were a poor indicator of evening temperatures, which are important to the nocturnal opossum. Open sites had the highest daily temperature ﬂuctuations, and were the coldest at night. Urban and coniferous sites had the least pronounced daily ﬂuctuations in temperature, and urban sites had the warmest nights. Habitat-speciﬁc winter temperatures in the Connecticut River Valley indicated that urban sites were most conducive to opossum persistence, but even they were unlikely to sustain populations. Other factors likely help mitigate the inﬂuence of harsh climatic conditions on persistence of opossum populations in western Massachusetts.
Keywords: microclimate, species distribution, temperature selection, urban, Virginia opossum Introduction Where temperature-induced energetic constraints are important, animals should select habitats with favorable thermal microclimates (Humphries et al., 2002). On a landscape scale, ∗ Authorto whom all correspondence should be addressed.
this may be reﬂected in patchy local distributions. At climatically constrained distributional edges, local variations in habitat and microclimate will deﬁne the exact geographical boundaries of the species’ range. Although human alteration of the landscape clearly alters local microclimate (notably the “urban heat island” effect; e.g., Brazel et al., 2000), the biological signiﬁcance of the climatic difference for local fauna is rarely noted. Perhaps this is because urbanization effects such as gross habitat loss, replacement by non-native species, and availability of urban food resources are far more dramatic (McKinney, 2002). For species at a climatic distributional limit, however, an altered urban climate may change the species’ range, particularly where warming from human development ameliorates restrictively cold climatic regimes.
Virginia opossums (Didelphis virginiana) are generalist medium-sized mammals that are thought to be highly sensitive to local winter temperature regimes at their northern distributional limit in northern New England and southern Ontario (Brocke, 1970). However, their current distribution (Gardner and Sunquist, 2003) extends farther north than these putative climatic limits (Brocke, 1970; Kanda, in press). The opossum is also an adaptable mesopredator able to coexist with humans throughout the urbanized landscape (Crooks, 2002). We investigated whether the presence of human development may be extending the opossum range by providing tolerable winter microclimates.
Previous examinations of Virginia opossum energetics have relied on generalizations based on the number of days with daily maxima ≤0◦ C, as recorded by local weather stations (Brocke, 1970; Kanda, in press). The energetic modeling relies on two inputs: opossum weight at the onset of winter and the winter temperature regime as represented by daily maxima, and the relationship between these based upon the temperature at which opossums cease foraging activity. The opossum, which is generally nocturnal, does not forage on days when the evening temperature is less than or equal to approximately −4◦ C (Brocke, 1970, Kanda et al., in press). Brocke (1970) hypothesized that in Lansing, Michigan (where he conducted his energetics research), −4◦ C evening temperatures occurred on days when the daily maximum was 0◦ C. Days at least this cold restrict opossum foraging, and thus opossum winter starvation hinges on how many cold, non-foraging days occur. Predictions of opossum winter survival in an area are therefore predicated upon (1) the relationship between evening temperatures and daily maximum temperatures, and (2) the ability of the daily maximum temperature set, usually from the local weather station, to predict landscape temperature regimes.
Opossums are not expected to exist in Amherst, Massachusetts, based on energetic modeling (Kanda, in press). Only a small proportion (e.g., 32% in 1999) of female juveniles, the most important breeding cohort yet the lightest weight class, would be expected to survive the winters, based upon daily maximum temperatures reported by the local weather station (Amherst, Massachusetts; National Oceanic and Atmospheric Administration, 2004). Such low over-winter survival would lead to population declines, and eventual local extinction (Kanda and Fuller, 2004). Nevertheless, opossums are known to occur patchily throughout the Amherst region, with a correlation of opossum occurrence and human development (Kanda, 2005).
We hypothesized that opossums use urban refugia with microclimates more favorable than those reported by the local weather station. We suspected that in Amherst the warming
MICROCLIMATE AND OPOSSUM SURVIVALeffect of human development in town would be sufﬁcient to increase expectations of opossum survival, because as little as a 1◦ C overall difference in the daily temperature could dramatically increase expected opossum survival rates (Kanda, in press). We therefore attempted to identify the magnitude of temperature differences among natural and urban habitat patches in the Amherst region, and whether daily maxima of ≤0◦ C are accurate predictors of evening temperatures ≤−4◦ C within these regimes. Also, because predictions of opossum survival in other areas will likely continue to rely on summary data from weather stations, we examine the relationship between weather station data and the evening temperatures measured across the landscape. By understanding microclimate variation due to time of day, habitat, and elevation, we hope to identify regions in central Massachusetts where opossum populations should be expected to persist.
Methods Sites Amherst, Massachusetts, lies in the Connecticut River Valley of central Massachusetts (42◦ 23’ N, 72◦ 32’ W). Anecdotal, road-transect survey, and camera-trap evidence suggest that opossums are less common in the Pelham Hills (elevation 335 m) that border the valley than in the valley itself (elevation 45 m) (Kanda, 2005). The hills also have less human development and more forested areas (coniferous, deciduous, and mixed), whereas the valley supports extensive urban, suburban, and agricultural development (cf. Danielson et al., 1997). Though Amherst, Massachusetts is not a large town (17,000 people in Amherst Town Center; U.S. Census Bureau, 2003), it is contiguous with the University of Massachusetts campus with an additional 11,000 resident students (University of Massachusetts Amherst Housing Services, 2004).
We deployed 1 20 HoBoTM data loggers that recorded air temperature every 30 minutes with an accuracy of ±0.5◦ C for 125 consecutive days during November 1999 to March
2000. 18 loggers were placed in a factorial design across three habitats (coniferous forest, deciduous forest, and open meadow) and three elevations (45, 165, and 335 m; each ±10 m), with two replicate sites for each habitat-elevation combination. These “natural” habitat sites were in habitat patches 1 ha to 100 ha in size imbedded in a matrix of human development at low elevation (urban, suburban, and agricultural) and a matrix of mixed natural habitats interspersed with suburban development at higher elevations (landuse types identiﬁed from Southern New England Gap Analysis Project; Slaymaker et al., 1996). Most patches available in the landscape were 50 ha, although deciduous forest formed blocks as large as 373 ha at the highest elevation. Uncultivated open land was rare and found only in patches smaller than 3 ha throughout the study area; however, cultivated open land was common at low elevation. In addition, two loggers were placed at urban sites on the University of Massachusetts Amherst campus, which lies at the low elevation (54 m). There is no comparable concentration of human development (a block of 188 ha) at the higher elevations in the study area. The total study area was 216 km2.
Data-loggers were placed on level ground at least 20 m from the habitat edge and greater than 50 m from open water or marsh. The temperature loggers were placed in waterproof 218 KANDA ET AL.
clear plastic containers and hung in small three-sided covered shelters (20 × 20 × 30 cm) on stakes. The open face of the shelter was oriented north. Sensors were 60 cm above the ground and greater than 2 m from the nearest tree (10 cm dbh).
We obtained daily mean, minimum, and maximum temperatures for the winter from the Amherst Weather Station located at a sewage treatment plant at the west end of the University of Massachusetts Amherst campus (National Oceanic and Atmospheric Administration, 2004). Using the criteria we applied to other sites, the station is a low-elevation (45 m), open habitat site.
Our ﬁrst objective was to document temperature regime variation across the central Massachusetts landscape. To do this, we compared the daily maximum, minimum, and mean temperatures (1) among three elevations and (2) among four habitat types representative of the landscape. Because night temperatures are important to the opossum, we compared different habitat/elevation combinations by time of day to identify regional variation in this ecologically important variable.
For our examination of temperature differences by elevation, we restricted our analysis to the 18 natural habitat sites. Daily mean, minimum, and maximum temperatures were calculated for each site. Of course, these daily parameters are, at all sites, most strongly inﬂuenced day to day by the seasonal weather pattern. Because we wished to look for how sites differ from one another, we concentrated on the difference among the sites for each parameter each day by standardizing daily parameters to the deviation from the average across sites. The resulting 125 samples of each deviation parameter were found to be normally distributed at each site. For each parameter we used SAS v.8 (SAS Institute, 1999) to perform a mixed-model ANOVA considering habitat and elevation as ﬁxed effects, with individual site as a nested random effect. Least squares means with Bonferroni corrections were used in pairwise comparison among elevations.
In comparing temperature regimes across four habitat types (urban, coniferous, deciduous, and open), we used only low-elevation sites (the only elevation of urban sites). As described above, we calculated the difference from the mean for each parameter (in this case, the mean of 8 low-elevation sites) for each day. We again performed a mixed-model ANOVA on these deviation parameters, considering habitat as a ﬁxed effect and individual site as a random nested effect.
To examine the temperature variation among sites by time of day, we focused on the contrast between natural habitat sites and urban sites. The difference between each natural site and the urban average was calculated for every time point. For each time of day, a natural site’s deviations from the urban were averaged over all days. We selected four time points for statistical comparison: 19:00 (used to predict opossum foraging by Brocke, 1970) and three other times spaced at 6-hour intervals (01:00, 07:00, 13:00). Bonferroni corrections were used to adjust for family-wise Type I error rates when comparing the four time points.
Because our sensors were of different type and in different housing than the local weather station, direct comparison of temperature records between our sites and the local station was not appropriate. However, we wished to evaluate the temperature index
MICROCLIMATE AND OPOSSUM SURVIVALcurrently used for evaluating regimes for opossum survival, the count of daily maxima ≤0◦ C as recorded by the weather station, for its ability to act as proxy for evening temperature regimes across the landscape. We counted the records from the local weather station of days with daily maximum ≤0◦ C, the days with evening temperatures ≤−4◦ C at each site and the days with daily maximum ≤0◦ C at each site. Counts from sites and station were compared to identify which (if any) evening temperature regimes across the landscape were warmer than predicted by maximum daily temperatures at the weather station.
Results Temperature variation by elevation Elevation was a signiﬁcant effect on differences among natural sites in daily maxima, mean, and minima, but did not have high explanatory power. Maximum temperatures cooled as elevation increased ( p = 0.013, r 2 = 0.123), though only the difference between the low and high elevations was signiﬁcant ( p = 0.014). Daily mean temperatures also cooled with higher elevation ( p = 0.013, r 2 = 0.193). Pairwise comparisons could not distinguish between daily means at low-and mid-elevation sites, but found high-elevation sites to be cooler than both ( p = 0.016 and p = 0.026, respectively).