Northeastern Naturalist NENHC-14 L. Nolfo-Clements and M. Clements 22001155 NORTHEASTERN NATURALIST 22(1):NENHC-14V–oNl.E 2N2H, NCo-2. 51 Small-Mammal Population Dynamics and Habitat Use on Bumpkin Island in the Boston Harbor Lauren Nolfo-Clements1,* and Mark Clements2 Abstract - We performed short-interval mark–recapture trapping on small mammals on Bumpkin Island in Boston Harbor in 2008, 2009, and 2011 in an attempt to record patterns of species distribution, population dynamics, and habitat use. The only species captured during these intervals were native Peromyscus leucopus (White-footed Mouse) and Microtus pennsylvanicus (Meadow Vole). Both mice and voles were trapped in 2008 and 2009, while only mice were trapped in 2011. Animal densities varied by vegetation type and by year. The variation in the densities between years may be attributed to a number of factors including food availability and the sporadic presence of predators, a unique characteristic of the some of the harbor islands. Introduction Mammal populations on islands, both actual and virtual (habitat patches), have been the focus of numerous studies for decades (Foster 1964, Lawlor 1986, Lomolino et al. 2013). Rodent populations have been of particular interest due to the ease of isolation for these small animals and their presence on many islands in marine, freshwater, and terrestrial environments. Of rodent populations studied, Mus musculus L. (House Mouse) and those in the genus Rattus have received the most attention due to their status as introduced species and their cosmopolitan distribution as a result of inadvertent human transport (Howald et al. 2007) . Rodents of the genera Peromyscus and Microtus have been isolated on all types of islands (Crowell 1983, Forsman et al. 2011, Munshi-South and Kharchenko 2010). In Massachusetts, the status of Peromyscus leucopus Rafinesque (White- Footed Mouse) and Microtus breweri Baird (Beach Vole; endemic to Muskeget Island off Cape Cod) have been examined and their island population dynamics compared with those of mainland populations. For M. breweri, population cycling appeared to be absent on Muskeget Island when compared to mainland populations of Microtus pennsylvanicus Ord (Meadow Vole), though peak densities were similar between the 2 species (Tamarin 1977). For White-footed Mice, population dynamics were similar in both locations, but average densities were higher on the mainland (Adler and Tamarin 1984). With this information in mind, we initiated a small-mammal trapping project on Bumpkin Island in Boston Harbor in 2008, as part of a larger effort assisting the National Park Service (NPS) and the Massachusetts Department of Conservation and Recreation (DCR) in the inventory and monitoring of mammal species of the Boston 1Suffolk University, Biology Department, Boston, MA 02108. 2Northern Essex Community College, Natural Sciences Department, Haverhill, MA 01830. *Corresponding author - lnolfoclements@suffolk.edu. Manuscript Editor: Thomas J. Maier Northeastern Naturalist Vol. 22, No. 1 L. Nolfo-Clements and M. Clements 2015 NENHC-15 Harbor Islands National Recreation Area (BOHA). Through our island trapping efforts, we hoped to assess small-mammal species diversity, evaluate small-mammal population distribution and dynamics, and note any habitat preferences. Field-Site Description Bumpkin Island is one of BOHA’s 34 islands and peninsulas east of Boston, MA (Fig. 1). The climate is temperate with well-defined seasons. The average temperature is ~10 °C annually, ~21 °C in the summer ,and about -1 °C in the winter (National Weather Service 2014). Average annual precipitation is about 1300 mm; there is no distinct wet or dry season, although precipitation averages are highest during December and March (National Weather Service 2014). Figure 1. Location of Bumpkin Island in the Boston Harbor. Note the close proximity to the mainland at Hull, MA. (Credit: National Park Service) Northeastern Naturalist NENHC-16 L. Nolfo-Clements and M. Clements 2015 Vol. 22, No. 1 Bumpkin Island (24.9 ha) includes shoreline and intertidal-zone areas (42°16'51.18"N, 70°53'58.17"W). The island is a vegetated drumlin surrounded by rocky shoreline. During especially low tides, the island is attached to the mainland at Hull, MA, by a thin sand spit for limited periods of time (Fig. 1). About 51% of the plant species on this island are non-native (Elliman 2005). As with most of the islands in Boston Harbor, Bumpkin Island has a long history of human use. The islands were used by native North Americans as fishing areas and for durable resources, and then later by early European settlers for similar uses and some farming, although the Europeans clearcut many of the forested islands to assist in the building and heating of Boston (Richburg and Patterson 2005). As a result of this intensive use and historical clearcutting, Bumpkin and a few other islands are currently dominated by shrubby and brushy vegetation and not the stands of trees that were present before European colonization. Currently, Bumpkin Island is not inhabited year-round. Primitive camping is permitted on the island during the open season (3rd week of June–Labor Day). There are both grassy and paved trails that bisect the island. The grassy trails, camping areas, and scenic overlooks are maintained by periodic mowing. There is one intact permanent structure on the island where a caretaker seasonally resides, in addition to a series of ruins. We selected our trapping areas so that they did not contain campgrounds or paved trails (Fig. 2). The plant species composition and vegetative structure appeared relatively uniform across the entirety of Bumpkin Island away from the shoreline. While the island did contain a few small stands of trees, most of the vegetation was shrubs, vines, and herbaceous vegetation. Common plant species included Rhus typhina L. (Staghorn Sumac), Celestrus orbiculatus Thunb. (Asian Bitterweet), Rubus idaeus L. (Common Red Raspberry), Rubus fruticosus L. (Blackberry), Rosa rugosa Thunb. (Japanese Rose), Morella pensylvanica (Mirb.) Kartesz (Northern Bayberry), Rhamnus sp. (buckthorn), Acer platanoides L. (Norway Maple), Solidago spp. (goldenrods), and Toxicodendron radicans (L.) Kuntze (Poison Ivy). Methods Field methods All trapping occurred in late June and July. We set traps in a grid pattern that covered roughly the same area across years (Fig. 2). In 2009, we set additional trapping grids on the northwestern side of the island. Since the island receives many Figure 2. Map of Bumpkin Island showing trapping-grid locations. A: 2008, B: 2009, and C: 2011 Hiking trails are shown with dotted lines. Northeastern Naturalist Vol. 22, No. 1 L. Nolfo-Clements and M. Clements 2015 NENHC-17 human and wildlife visitors year-round, we were not able to establish permanent trapping grids due to the high probability of disturbance. We set traps about 7 m apart with a single model LFATDG Sherman live trap (9 in [L] x 3 in [W] x 3.5 in [H]) at each station. We set a single 10 x 10 trap grid in 2008, one 10 x 5 and two 5 x 5 grids in 2009, and one 10 x 9 grid in 2011. We flagged and noted the GPS location of each trap. We baited traps with a mixture of peanut butter and oats rolled into quarter-sized balls and stuck to the back wall of the traps. We checked traps once a day, in the morning around 9 am. Checking the traps twice a day was considered, but a similar small-mammal trapping survey on Cape Cod revealed that checking traps twice a day did not reduce trapping mortality nor did it result in a significant increase in captures (Cook et al. 2006). Additionally, trapping survivorship was >98% for all 3 years of this study, with the 5 mortalities that did occur primarily due to prior injury and/or aggressive trap disturbance by Canis latrans Say (Coyote). Upon capture, we transferred small mammals, individually, into an unsealed large plastic Ziploc bag to allow for species identification, sexing, maturity evaluation (adult or juvenile), and weighing with a spring scale. If animals weighed less than 12 grams, they were not implanted with a passive integrated transponder (PIT) tag for the mark–recapture portion of the study. We used 8.5 mm x 2.12 mm PIT tags (Biomark model TXP1485B), implanted with a 12-gauge needle at the “scruff” of the neck right in front of the shoulders. PIT tags have been used on a vast array of vertebrate species, both aquatic and terrestrial (Gibbons and Andrews 2004). Studies have shown that animals do not exhibit behavioral changes as result of such implantation and there is no outward marking of the animal that could impact predation rates, fitness, or social interactions (Gibbons and Andrews 2004, Harper and Batzli 1996, Schooley et al. 1993). We recorded the plant species present at each trapping location and noted proximity to mowed grass and paved trails. We identified plants to species whenever possible. Statistical analyses We estimated rodent population densities using spatially explicit capture-recapture (SECR) models (Borchers and Efford 2008, Efford 2004). Unlike other models (e.g., Jolley-Seber; see Leberton et al. 1992) that estimate survival, abundance, and/ or densities using mark–recapture data, SECR models use spatial information of mark–recapture data. This spatial information allows estimates of the distribution and density of animals in space. SECR analyses assume that animals have fixed, approximately circular, home ranges. Under this assumption, the probability of capturing an animal (i.e., the encounter rate) declines as the distance between the trap and the center of the home range increases. A variety of detection functions may be used to describe the relationship between encounter rate and distance from an animal’s home-range center. We selected the half-normal detection function for all analyses because it is the most common function used and the default in SECR 2.8.1 (Efford 2014). The half-normal Northeastern Naturalist NENHC-18 L. Nolfo-Clements and M. Clements 2015 Vol. 22, No. 1 detection function is a probability distribution with two parameters that determine its shape: 1) g0 is the encounter rate assuming that the trap and home-range center of an animal coincide in space; and 2) σ is a scale parameter that determines how the encounter rate declines with distance from the home-range center. The scale parameter (σ) can be interpreted as an estimate of the home range (radius only) of animals. The encounter rate (g0) can be interpreted as the probability of capture, but it is different from the detection probabilities reported by non-spatial methods, since they use only the frequency of detection to estimate probabilities; thus, an animal is either captured or not captured when it is near a trap. SECR probabilities are described by continuous functions that depend upon spatial information; thus, the further an animal is from a trap, the less likely it is to be captured. The probabilities of capture reported herein will be much lower than those reported for non-spatial methods, because g0 is not the total probability of capture. SECR models and their parameters were estimated by maximizing the full likelihood. Initial tests confirmed that model estimates were not measurably affected by variation in habitat-mask density (pixel size) or buffer width, so default SECR pixel sizes were used, and buffer widths were set to extend 30 m beyond trapping areas. Null SECR models assume homogenous animal density across trapping areas, and uniformity of home-range radius (σ) and probability of detection (g0) over individuals, traps, time, and habitat. Alternative SECR models that relaxed these assumptions were compared to null models using the Akaike information criterion with small-sample-size adjustment (AICc) to select preferred models (Burnham and Anderson 2002). We explored the effect of vegetation cover on populationdensity estimates (i.e., a measure of habitat preferences) by assigning trap sites to a vegetation category using fuzzy clustering of a distance matrix of vegetation presence– absence data (Kaufman and Rousseauw 1990). The vegetation category of each pixel in the SECR habitat mask was then assigned to the vegetation category of nearest trap site (Efford 2014, Efford and Fewster 2013). Results Population dynamics and density Across all 3 years, we only captured White-footed Mice and Meadow Voles on Bumpkin Island. We detected a total of 453 small mammals over the course of the study. We used a total of 159 individual mice in the analyses (77 with 167 detections in 2008, 67 with 152 detections in 2009, and 15 with 39 detections in 2011) and a total of 68 individual voles in the analyses (27 with 35 detections in 2008, 41 with 60 detections in 2009, and 0 in 2011). A total of 83 mouse and 20 vole detections were dropped prior to SECR analysis due to small animal size, incomplete data, or our failure to successfully scan PIT tags. Models that included a learned trap response fit capture–recapture data best for both mice and voles. In all years, the probability of capture (g0) increased after previous capture. In 2008, the estimated probability of capture (g0) for mice was 0.03 (SE = 0.01) and recapture probability (RP) was 0.09 (SE = 0.01). Estimates in 2009 were similar (g0 = 0.014, SE = 0.003; RP = 0.01, SE = 0.01) and in 2011, Northeastern Naturalist Vol. 22, No. 1 L. Nolfo-Clements and M. Clements 2015 NENHC-19 the mice capture probability was similar to estimates in 2008 and 2009 (g0 = 0.01, SE = 0.01), but their recapture probability was lower (RP = 0.06, SE = 0.02). For voles, models that pooled estimates of g0 over years were preferred based on AICc. Pooled Estimates of g0 = 0.03 (SE = 0.01) and RP = 0.07 (SE = 0.02) for voles. There was evidence for individual heterogeneity in estimates of home-range radius (σ) among mice in 2008 because a finite-mixture model with 2 classes for σ was preferred based on AICc. The mixture model estimated that 27.6% of individuals had home ranges estimated at 13.36 m (SE = 1.02 m) while the rest (72.3%) had smaller home ranges estimated 6.20 m (SE = 0.89 m). There was no evidence for individual heterogeneity in σ for mice or voles during the remaining trapping sessions. Home-range radius of mice in 2009 and 2011 was estimated as 10.85 m (SE = 0.79 m) and 23.46 m (SE = 3.13 m), respectively. The pooled, across-year estimate of σ for voles was 6.05 m (SE = 0.68 m; Table 1). There was support for inhomogeneous densities across trapping areas for mice. Models that estimated a trend in density along a north–south axis were favored based on AICc in 2008 and 2011. Densities increased from north to south in 2008 (76.95–234.30 animals/ha) and increased from south to north in 2011 (1.11–39.14 animals/ha) (Table 2). Table 1. Home-range (σ) estimates for rodents captured on Bumpkin Island in 2008, 2009, and 2011. SE provided in parentheses. In 2008, Peromyscus leucopus (White-footed Mouse) populations showed heterogeneity in their home-range radius estimates. These two estimates are represented by σ1 and σ2. Home-range radius estimates were homogeneous in all other instances. Species σ1 (m) σ2 (m) Peromyscus leucopus 2008 13.36 (1.01) 6.20 (0.89) 2009 10.85 (0.79) NA 2011 23.46 (3.13) NA Microtus pennsylvanicus 2008, 2009 (combined) 6.05 (0.68) NA Table 2. Density estimates (animals/ha) for rodents captured on Bumpkin Island in 2008, 2009, and 2011. SE in parentheses. Microtus pennsylvanicus (Meadow Vole) populations had homogenous (hom.) densities across all habitat types while Peromyscus leucopus (White-footed Mouse) densities varied on both a north–south gradient and by habitat type. Prunus, Pinus, Rhus and Species Hom. North–south Rosa spp. Solidago spp. and Morella spp. Rubus spp. Peromyscus leucopus 2008 NA 76.95–234.30 162.17 (51.99) 147.11 (34.35) 133.45 (29.97) 121.06 (36.32) 2009 NA NA 0.00 137.29 (33.93) NA 48.57 (31.40) 2011 NA 39.14–1.11 6.00 (4.91) 1.63 (9.39) NA 22.11 (5.58) Microtus pennsylvanicus 2008 77.90 (21.89) NA NA NA NA NA 2009 84.50 (20.12) NA NA NA NA NA Northeastern Naturalist NENHC-20 L. Nolfo-Clements and M. Clements 2015 Vol. 22, No. 1 Habitat use Density estimates that used categorical vegetation cover variables (see methods) were also favored based on AICc, although these estimates varied widely between years (Table 2). In 2008, areas with roses had densities of 162.17 animals/ha (SE = 51.99 animals/ha), areas with goldenrods and maples had densities of 147.11 animals/ ha (SE = 34.35 animals/ha), areas with Prunus spp., pines, and bayberries had densities of 133.45 animals/ha (SE 29.97 animals/ha), and areas that were primarily sumacs and Rubus spp. (the most common species at all areas over all years) with few bayberry plants and maples, had densities of 121.06 animals/ha (SE = 36.32 animals/ha). In 2009, areas with goldenrods had densities of 137.29 animals/ha (SE = 33.93 animals/ha), areas with only sumacs and Rubus spp. had densities of 48.57 animals/ ha (SE = 31.40 animals/ha), and areas with roses had density estimates of 0. In 2011, areas with only sumacs and Rubus spp. had densities of 22.11 animals/ ha (SE = 5.58 animals/ha), areas with Rubus spp. and roses had densities of 6.00 animals/ha (SE = 4.91 animals/ha), and areas with sumacs, goldenrods, and few Rubus spp. had densities of 1.63 animals/ ha (SE = 9.39 animals/ha). Vole sample sizes were too small to allow for accurate estimates of variable densities across trapping areas. Vole densities were 77.90 animals/ha (SE = 21.89 animals/ha) in 2008 and 84.50 animals/ha (SE = 20.12 animals/ha) in 2009. Discussion Both mice and voles exhibited learned trap response behavior, i.e., once an animal was captured, the probability of capturing that animal again increased. For animals captured in baited traps, this is a well-known and extensively studied behavior (Pollock et al. 1990); thus, this result was not unexpected. This behavior was accounted for in our model and hence did not impact our results. Population dynamics and density When compared to other similar trapping studies on the islands and mainland of eastern Massachusetts, our population densities were high (Table 2). Adler and Tamarin (1984) found densities of 4 mice per 0.8 ha in both island and mainland populations in June–July. In a previous study by Tamarin (1977), voles on an island off the coast of eastern Massachusetts showed consistent densities of 35–40 animals per ha in June–July, while mainland vole populations exhibited cycling and hence occurred in highly variable densities that ranged from 6–80 individuals per ha in June–July, depending upon the year. However, mice population densities outside of Massachusetts can vary widely depending upon the size of the habitat patch and the quality of the habitat. Densities of over 250 individuals per hectare have been recorded in small woodland patches (less than 0.5 ha), although densities of 75–100/ha were more common in these small patches (Nupp and Swihart 1996). Our population-density estimates are well within these published ranges considering the abundance of dense ground cover and high-quality food plants in our study area, as discussed below. Northeastern Naturalist Vol. 22, No. 1 L. Nolfo-Clements and M. Clements 2015 NENHC-21 Vole densities outside of Massachusetts also vary widely. Since these voles are subject to population cycling in some habitats, their densities may range from undetectable to upwards of 150 animals per hectare (Christian 1971, Getz et al. 2001). Vole populations may be cycling on Bumpkin Island, although our study duration was not long enough to confirm this hypothesis. The well-known population cycling patterns of voles has been studied extensively, and evidence strongly suggests that predation is the driver behind these cycles (Andreassen et al. 2013). However, due to the short time interval over which our trapping occurred (3 seasons), other explanations, such as weather and variations in food availability, are also viable. A number of factors contribute to whether or not oceanic islands are home to predators including their size, distance from the mainland, and availability of prey (Gravel et al. 2011). However, most islands are either permanent homes for predators or lack predators completely. The islands of BOHA are unique in that they are home to ephemeral predators that use the island for variable periods on an annual or seasonal basis (L. Nolfo-Clements, unpubl. data). These predators are not limited by prey availability, as are most predators on islands (L. Nolfo-Clements, unpubl. data). On the islands of BOHA, if prey becomes scarce, predators may simply vacate the island for another island or the mainland by swimming or walking across the frozen Harbor or a temporary land bridge. In 2008, the presence of Coyotes was confirmed by direct animal sightings, scat, and prints (L. Nolfo-Clements, unpubl. data). In 2011, Mustela vison Schreber (American Mink) were also sighted on the island on numerous occasions by both island rangers and visitors (L. Nolfo-Clements, unpubl. data). No predators were reported in 2009. Both American Mink and Coyotes are known predators of voles (Fey et al. 2010, Gese et al. 1996). Habitat use Extensive research has focused on the habitat use of both White-footed Mice and Meadow Voles, especially at the microhabitat scale (Jorgensen 2004). White-footed Mice, proficient climbers, prefer areas with vertical structure and may nest in standing hollow trees or even bird houses (Kaufman et al. 1983; L. Nolfo-Clements, pers. observ.). Meadow Voles prefer grasslands, but are also found at lower densities in woodlands (Reich 1981). There is also some evidence to suggest that Meadow Voles are adapted to living in disturbed patches and may prefer habitat edges (Bowers et al. 1996). Of the plant species recorded at our trapping locations, Rubus spp. and roses probably provided the most food and cover. While bayberries, sumacs, Prunus spp., maples, and pines may act as food sources for rodents, these plants do not provide dense, nearly continuous ground cover as do Rubus spp. and roses. The high density of mice found in goldenrods in 2009 was surprising, although these plants may occur in very high densities in certain areas of the island and hence potentially provide adequate cover, they are not a food source for this species. Additionally, habitat dominated by goldenrods was indicated as preferred mice habitat in at least one old-field study, although the mice in that study inhabited Northeastern Naturalist NENHC-22 L. Nolfo-Clements and M. Clements 2015 Vol. 22, No. 1 burrows in the habitat dominated by goldenrods (Pearson 1959). Previous studies have shown that mice prefer areas with large amounts of woody debris, a feature not present on Bumpkin Island (Kellner and Swihart 2014). Studies on Whitefooted Mice in non-forested habitat indicate that they prefer areas with dense, high shrub cover and occur at lower densities in grasslands (Adler et al. 1984, Clark et al. 1987, Kaufman et al. 1985). In most habitats, the primary food sources for White-Footed Mice include small seeds, insects, and their larvae, mast (acorns and nuts), fruit, and some vegetation (Whitaker 1966, Wolff et al. 1985). In contrast, Meadow Vole diets primarily consist of green vegetation during the growing season, and seeds and nuts during the winter months (Lindroth and Batzli 1984, Zimmerman 1965). Krebs et al. (2010) reported that vole and mice populations in the Yukon appeared to be influenced by fluctuations in berry crops that were primarily linked to weather conditions. Much of the understory in our trapping area consisted of Rubus plicatus Weihe & Nees (Bramble Blackberry), which is grown as a fruit crop. Additionally, Rosa rugosa Thunb. (Rugosa Rose) produces a large rosehip that could act as a potential food source for these species. Both of these berries occurred in high densities and persist in the environment well into the autumn and early winter in our study area. The seeds of these fruits could also provide winter and early spring forage for these rodents. The habitat on Bumpkin Island represents a “middle ground” between the habitat needs of White-Footed Mice and Meadow Voles. The dense shrubby overstory throughout our study area, coupled with low creeping brambles and grassy trails, provides the cover and food sources required by both of these species. While there are few masting trees on Bumpkin Island, fruit, small seeds, and insects are available in great abundance. Additionally, Meadow Voles are frequently spotted by visitors and rangers crossing grassy trails, and at the edges of campsites and fields. Conclusions Our results indicate that rodents on Bumpkin Island occurred at higher densities than expected for populations located in Massachusetts. However, our results are comparable to those recorded in small habitat patches and/or in high-quality habitat in other locales. These high densities arere most likely attributable to the vegetation on Bumpkin Island that provide dense ground cover and abundant food sources for both White-Footed Mice and Meadow Voles. While the sporadic presence of predators may impact small-mammal populations, our data are not extensive enough to clarify this relationship. In the future, we hope to uncover whether or not Meadow Vole populations are cycling on Bumpkin Island and whether the presence of predators impacts those cycles. We also hope to uncover how genetically isolated small mammals on Bumpkin Island are from populations on other islands and the mainland to clarify if the island is truly an isolated habitat patch considering its sporadic connection to the mainland and the large number of boats that move between these islands and the mainland. Northeastern Naturalist Vol. 22, No. 1 L. Nolfo-Clements and M. Clements 2015 NENHC-23 Acknowledgments We would like to thank the National Park Service and the Massachusetts Department of Conservation and Recreation for their support. Thanks to C. Hogan and C. Surdyka who assisted with trapping and J. Demers for data entry and organization. This research was supported in part by funds from Suffolk University’s faculty summer stipend award. Transportation costs were covered by the NPS. This study protocol was reviewed and approved by the Institutional Animal Care and Use Committee at Loyola University in New Orleans. A special thanks to M. Albert who has supported and facilitated all of our research endeavors on the Boston Harbor Islands, as well as to 2 anonymous reviewers and T.J. Maier, the Manuscript Editor. Literature Cited Adler, G.H., and R.H. Tamarin. 1984. Demography and reproduction in island and mainland White-Footed Mice (Peromyscus leucopus) in southeastern Massachusetts. Canadian Journal of Zoology 62:58–64. Adler, G.H., L.M. Reich, and R.H. Tamarin. 1984. Characteristics of White-Footed Mice in woodland and grassland of Eastern Massachusetts. Acta Theriologica 29:57–62. Andreassen, H.P., P. Glorvigen, A. Remy, and R.A. Ims. 2013. New views on how population- intrinsic and community-extrinsic processes interact during vole population cycles. Oikos 122:507–515. Borchers, D.L., and M.G. Efford. 2008. Spatially explicit maximum likelihood methods for capture–recapture studies. Biometrics 64:377–385 Bowers, M.A., K. Gregario, C.J. Brame, S.F. Matter, and J.L.Dooley, Jr. 1996. Use of space and habitats by Meadow Voles at the home-range, patch, and landscape scales. Oecologia 105:107–115. Burnham, K.P., and D.R. Anderson. 2002. Model Selection and Multimodel Inference: A Practical Information-Theoretic Approach, 2nd Edition. Springer, New York, NY. 488 pp. Christian, J.J. 1971. Fighting, maturity, and population density in Microtus pennsylvanicus. Journal of Mammalogy 52:556–567. Clark, B.K., D.W. Kaufman, G.A. Kaufman, and E.J. Finck. 1987. Use of tallgrass prairie by Peromyscus leucopus. Journal of Mammology 68:158–160. Cook, R.P., K.M. Boland, and T. Dolbeare. 2006. Inventory of small mammals at Cape Cod National Seashore with recommendations for long-term monitoring. Technical Report NPS/NER/NRTR--2006/047. National Park Service. Boston, MA. Crowell, K.L. 1983. Islands: Insight or artifact? Population dynamics and habitat utilization in insular rodents. Oikos 41:442–454. Efford, M.G. 2004. Density estimation in live-trapping studies. Oikos 106:598–610 Efford, M.G. 2014. SECR: Spatially explicit capture–recapture models. R package version 2.8.1. Available online at http://CRAN.R-project.org/package=secr. Accessed 11 June 2014. Efford M.G., and R.M. Fewster. 2013. Estimating population size by spatially explicit capture– recapture. Oikos 122:918–928. Elliman, T. 2005. Vascular flora and plant communities of the Boston Harbor Islands. Northeastern Naturalist 12:49–74. Fey, K., P.B. Banks, and E. Korpimaki. 2010. Alien mink predation and colonisation processes of rodent prey on small islands of the Baltic Sea: Does prey naiveté matter? International Journal of Ecology 2010: 1–9. Northeastern Naturalist NENHC-24 L. Nolfo-Clements and M. Clements 2015 Vol. 22, No. 1 Forsman, A., J. Merila, and T. Ebenhard. 2011. Phenotypic evolution of dispersal enhancing traits in insular rodents. Proceedings of the Royal Society B 278:225–232. Foster, J.B. 1964. Evolution of mammals on islands. Nature 202:234–235. Gese, E.M., R.L. Ruff, and R.L. Crabtree. 1996. Intrinsic and extrinsic factors influencing Coyote predation of small mammals in Yellowstone National Park. Canadian Journal of Zoology 74:784–797. Getz, L.L., J.E. Hofmann, B. McGuire, and T.W. Dolan III. 2001. Twenty-five years of population fluctations of Microtus orchrogaster and Microtus pennsylvanicus in three habitats in east-central Illinois. Journal of Mammalogy 82:22–34. Gibbons, J.W., and K.M. Andrews. 2004. PIT tagging: Simple technology at its best. Bioscience 54:447–454. Gravel, D., F. Massol, E. Canard, D. Mouillot, and N. Mouquet. 2011. Trophic theory of island biogeography. Ecology Letters 14:1010–1016. Harper, S.J., and G.O. Batzli. 1996. Monitoring use of runways by voles with passive integrated transponders. Journal of Mammalogy 77:364–369. Howald, G., C.J. Donlan, J.P. Galvan, J.C. Russell, J. Parkes, A. Samaniego, Y. Wang, D. Veitch, P. Genovesi, M. Pascal, A. Saunders, and B. Tershy. 2007. Invasive rodent eradications on islands. Conservation Biology 21:1258–1268. Jorgensen, E.E. 2004. Small-mammal use of microhabitat reviewed. Journal of Mammalogy 85:531–539. Kaufman, D.W., S.K. Peterson, R. Fristik, and G.A. Kaufman. 1983. Effects of microhabitat features on habitat use by Peromyscus leucopus. American Midland Naturalist 110:177–185. Kaufman, D.W., M.E. Peak, and G.A. Kaufman. 1985. Peromyscus leucopus in riparian woodlands: Use of trees and shrubs. Journal of Mammalogy 66:139–143. Kaufman, L., and P.J. Rousseauw. 1990. Finding Groups in Data: An Introduction to Cluster Analysis. John Wiley and Sons, New York, NY. 342 pp. Kellner, K.F. and R.K. Swihart. 2014. Changes in small-mammal microhabitat use following silvicultural disturbance. American Midland Naturalist 172:348–358. Krebs, C.J., K. Cowcill, R. Boonstra, and A.J. Kenney. 2010. Do changes in berry crops drive population fluctuations in small rodents in the southwestern Yukon? Journal of Mammalogy 91:500–509. Lawlor, T.E. 1986. Comparative biogeography of mammals on islands. Biological Journal of the Linnean Society 28:99–125. Lindroth, R.L., and G.O. Batzli. 1984. Food habits of the Meadow Vole (Microtus pennsylvanicus) in bluegrass and prairie habitats. Journal of Mammalogy 65:600–606. Lomolino, M.V., A.A. van der Geer, G.A. Lyras, M.R. Palombo, D.F. Sax, and R. Rozzi. 2013. Of mice and mammoths: Generality and antiquity of the island rule. Journal of Biogeography 40:1427–1439. Munshi-South, J., and K. Kharchenko. 2010. Rapid, pervasive genetic differentiation of urban White-Footed Mouse (Peromyscus leucopus) populations in New York City. Molecular Ecology 19:4242–4254. National Weather Service. 2014. Home page. Available online at http://www.weather.gov. Nupp, T.E., and R.K. Swihart. 1996. Effects of forest patch area on population attributes of White-Footed Mice (Peromyscus leucopus) in fragmented landscapes. Canadian Journal of Zoology 74:467–472. Pearson, P.G. 1959. Small-mammal and old-field succession on the Piedmont of New Jersey. Ecology 40:249–255. Northeastern Naturalist Vol. 22, No. 1 L. Nolfo-Clements and M. Clements 2015 NENHC-25 Pollock, K.H., J.D. Nichols, C. Brownie, and J.E. Hines. 1990. Statistical inference for capture–recapture experiments. Wildlife Monographs 107:3–97. Reich, L.M. 1981. Microtus pennsylvanicus. Mammalian Species 159:1–8. Richburg, J.A., and W.A. Patterson III. 2005. Historical description of the vegetation of the Boston Harbor Islands:1600–2000. Northeastern Naturalist 12:13–30. Schooley, R.L., B. van Horne, and K.P. Burnham. 1993. Passive integrated transponders for marking free-ranging Townsend's Ground Squirrels. Journal of Mammalogy 74:480–484. Tamarin, R.H. 1977. Demography of the Beach Vole (Microtus breweri) and the Meadow Vole (Microtus pennsylvanicus) in southeastern Massachusetts. Ecology 58:1310–1321. Whitaker, J.O., Jr. 1966. Food of Mus musculus, Peromyscus maniculatus bairdi, and Peromyscus leucopus in Vigo County, Indiana. Journal of Mammalogy 47:473–486. Wolff, J.O., R.D. Dueser, and K.S. Berry. 1985. Food habits of sympatric Peromyscus leucopus and Peromyscus maniculatus. Journal of Mammalogy 66:795–798. Zimmerman, E.G. 1965. A comparison of habitat and food of two species of Microtus. Journal of Mammalogy 46:605–612.