Introduction

Scientists have observed that insular animals may exhibit physical and behavioral characteristics that distinguish them from their mainland counterparts. The “island rule” states that larger animals will exhibit dwarfism and smaller animals will exhibit gigantism on islands (Lomolino 1985). This rule was first set forth with more specific parameters by Foster (1964) who noted that insular rodents and some marsupials tend towards gigantism while carnivores, artiodactyls, and lagomorphs tended towards dwarfism. Meiri et al. (2008) suggested that the island rule is not a general pattern shared among all animals or within clades, but similar to Foster (1964), showed that rodents exhibit a tendency for gigantism on islands.

The reasons behind these size trends are multifaceted. Food availability, predation pressure, and/or competition are the three main factors that are thought to influence the anomalous size trends of insular animals (Case 1978). Detailed studies have been made on various insular animal species and these factors and the size trends have remained consistent for the originally specified groups through decades of research worldwide (Lomolino 2005; Lomolino 2013).

Since some of the first observations that led to the formulation of the island rule were made on rodents, it is no surprise that numerous studies have focused on the various physical and behavioral aspects of these island giants. This suite of differing characteristics is pronounced in rodents and is often referred to as the “island syndrome” or “insular rodent syndrome” (Millien and Damuth 2004). This syndrome is typified by animals that have an overall larger body size, more stable population numbers (both seasonally and annually), lower fecundity, higher survivorship, and higher population density accompanied by a reduction in territoriality and other competitive behaviors (Adler and Levins 1994). Studies have also shown that island populations exhibit accelerated rates of morphological evolution (Pergams and Ashley 2001; Millien 2006).

Insular house mice (Mus musculus) have been noted as exhibiting the island syndrome in a variety of locations. The most notable of these is on Gough Island, in the South Atlantic Ocean. Mice from this population are the largest non-laboratory examples of this species ever recorded, they occur in high population densities, and are longer lived than typical house mice (Jones et al. 2003). House mice from the small Mediterranean islet of Piana have a significantly larger body size than mice from both the European mainland and the larger nearby islands of Sardinia and Corsica (Renaud and Auffray 2009).

Peromyscus mice on islands have been noted as exhibiting increased body size by some researchers but not by others. Foster (1964) noted an increase in insular Peromyscus body size off the coast of British Columbia, Canada. Insular populations of P. keeni from northwestern North America are over twice the mass of their mainland counterparts (Lomolino et al. 2012). However, Adler and Tamarin (1984) did not note a difference in mean body mass of adult Peromyscus leucopus in mainland versus island populations off the coast of Massachusetts, USA.

While P. leucopus has been studied throughout its range, including in major urban centers (Munshi-South 2012), studies on island populations of this species are limited. Researchers previously conducted studies on island and mainland populations of P. leucopus in Massachusetts, USA but did not note any unusual size or morphological characteristics in their study animals (Adler and Tamarin 1984; Adler and Wilson 1985).

Island rodents were also the focus of some of the earliest observations of microevolution (Pergams and Ashley 2001). While other island species have been observed undergoing microevolutionary changes, most notable Darwin’s finches on the Galapagos Islands (Grant and Grant 2002), the case of the white-footed mouse is somewhat unique in that individuals rarely live for a year in the wild so microevolutionary changes may be more readily observed in a short timeframe. Generally, evolutionary changes on islands are thought to occur rapidly initially and then stabilize over time (Raia and Meiri 2011).

Insular Microtus arvalis are twice the size of their mainland counterparts and have undergone significant microevolutionary changes over the last 5000 years since their island introductions. Some of these changes appear to have occurred over the span of a few decades (Cucchi et al. 2014). Microevolutionary changes have also been observed in three subspecies of Peromyscus maniculatus on the California Channel Islands. Pergams and Ashley (1999) studied morphological changes in P. maniculatus specimens collected from Santa Barbara, Anacapa, and Santa Cruz over a 90-year period. They found evidence for rapid, significant, changes in morphology of P. maniculatus from three different islands.

P. leucopus inhabits wooded and suburban areas as well as urban parks in the central and eastern USA, southern Canada, and eastern Mexico (Kays and Wilson 2009). In Massachusetts, USA, P. leucopus is found statewide while the more widespread P. maniculatus is only found in the central and western portions of the state. The average mass of an adult P. leucopus is typically about 20 g (Kays and Wilson 2009).

Materials and methods

Study area

The Boston Harbor Islands are a group of 30 islands just east of the city of Boston, Massachusetts, USA (Fig. 1). The islands range in size from small rocky outcroppings to partially forested islands with over 80 ha of terrestrial habitat (Roman et al. 2005). Due to the significant tidal fluctuations of nearly 3 m in the harbor, many islands have large intertidal areas.

Fig. 1
figure 1

The study site is located in the USA in the state of Massachusetts (filled black inset map, a) near the city of Boston (inset map, b). The islands included in the study are located in the southern part of the inner Boston Harbor (gray box, inset map b). The islands where trapping occurred, Bumpkin and Peddocks, are labeled in map c. Maps were generated with R (R Core Team 2017) using GIS vector data downloaded from MassGIS and Natural Earth (naturalearthdata.com)

Full size image

The Boston Harbor Islands have a long history of human use. The islands were used by Native American Indians for food and supplies since before the islands lost their connection with the mainland about 9000 years ago when the sea level began to rise (Richburg and Patterson 2005). Upon the arrival of European colonists in the seventeenth century, the vegetation on the islands began a dramatic shift as the islands were clear cut for agriculture, fuel, and later for use as military installations. As a result of this disturbance, the vegetation on many of islands is composed of over 50% non-native species (Elliman 2005).

Our study took place on two of the inner Harbor Islands, Bumpkin and Peddocks Islands (Fig. 1c). White-tailed deer (Odocoileus virginianus), coyotes (Canis latrans), and wild turkeys (Meleagris gallopavo) are frequent visitors on both of these islands and move freely between the islands and the mainland on a seasonal and annual basis (Nolfo-Clements, unpublished data).

Bumpkin Island has a terrestrial area of 12.2 ha (42° 16′ 51.18″ N, 70° 53′ 58.17″ W). During especially low tides, the island is attached to the mainland at Hull, MA, by a thin sand spit for limited periods of time. The vegetation on Bumpkin Island is dominated by staghorn sumac (Rhus typhina), Asian bittersweet (Celastrus orbiculatus), common red raspberry (Rubus idaeus), blackberry (Rubus fruticosus), beach rose (Rosa rugosa), northern bayberry (Morella pensylvanica), buckthorn (Rhamnus sp), Norway maple (Acer platanoides), goldenrod (Solidago spp), and poison ivy (Toxicodendron radicans).

Peddocks is a 74.6 ha island that exhibits a surprising diversity of habitats including forests, marshes, and man-made structures both occupied and abandoned (42° 17′ 31.27″ N, 70° 56′ 14.52″ W). The vegetation on Peddocks Island is primarily Norway maple (A. platanoides), burning bush (Euonymus alatus), poison ivy (T. radicans), red barberry (Berberis thunbergii), greenbrier (Smilax spp), privet (Ligustrum spp), multiflora rose (Rosa multiflora), and Morrow’s honeysuckle (Lonicera morrowii).

Both Bumpkin and Peddocks Island allow primitive camping during the open season (3rd week of June–early September). Both islands have grassy and paved trails. The grassy trails, camping areas, and scenic overlooks are maintained by periodic mowing.

Small mammal trapping and handling

Trapping on Bumpkin Island occurred in late June and early July in 2008, 2009, 2011, 2014, and 2015. Trapping on Peddocks occurred in mid-June in 2012, 2014, and 2015. Each trapping interval took place over a 2-week period with traps checked 4 days a week, emptied and closed for weekends, and reset for the second week. We were not able to establish permanent trapping grids on the islands due to high probability of disturbance by humans and wildlife. We set single Sherman live traps (9 in [L] × 3 in [W] × 3.5 in [H]) in a grid at 7-m intervals. Grid sizes varied by year and location, but ranged from 5 × 5 (25 trap) to 10 × 10 (100 trap) grids.

All traps were baited with a mixture of peanut butter and oats and placed in vegetation or brush to minimize exposure to sun and rain. Sufficient leaf litter was placed in each trap to allow small mammals to insulate themselves from the metal walls of the trap.

Traps were checked once a day, in the morning around 9 a.m. Checking the traps twice a day was a consideration, however a similar small mammal trapping survey on Cape Cod, MA, USA 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). Our trapping mortality was <98% for P. leucopus in all years using this methodology.

Upon capture, the mouse was transferred into a large unsealed plastic Ziploc bag to allow for species identification, sexing, maturity evaluation (adult, young adult, or juvenile), and weighing with a spring scale. Animal with a mass <16 g were not implanted with a tag for the mark-recapture portion of the study and were immediately released. Animals >16 g were scanned for a PIT tag. If one was not found, the animal was removed from the bag and implanted with a tested PIT tag at the nape of its neck. The animal was then scanned to ensure that the tag was in place, its number will be recorded, and then it was released. If an animal had a permanent external marking that distinguished it (stumped tail, missing eye, etc.), that characteristic was used for identification in lieu of a PIT tag.

For small species like P. leucopus, 8.4 mm tags are implanted. This is the smallest size currently commercially available. Tags are implanted at the “scruff” of the neck, right in front of the shoulders with a small gauge implanter. No anesthesia or sutures are required and animals can be immediately released. Studies have shown that animals do not exhibit behavioral changes as result of implantation and there is no outward marking of the animal that could impact predation rates, fitness, or social interactions (Schooley et al. 1993; Harper and Batzli 1996; Gibbons and Andrews 2004).

Although young adult and adult animals were tagged during our trapping intervals, only adult mice were included in this study. We identified adult mice through the examination of pelage coloration and external indicators of reproductive readiness (external scrotum in males and enlarged nipples in females).

Masses from the literature

We searched for recent articles (since 2000) that reported the mass of adult P. leucopus for captive and wild populations. The reported mean and standard error for each population was recorded and used for comparison to our data.

Statistical analysis

All statistical analyses were performed in the R environment (R Core Team 2017). Linear models were used to explore patterns of body mass change over time for adult P. leucopus. Separate linear models were fit to body mass data for each sex and island. Fit of linear models was assessed using r-squared values and through visual inspection of residual value plots. We also assessed differences in body mass of adult P. leucopus by sex, by year, and interactions between sex and year using an ANOVA with Bonferroni correction for post hoc, pairwise comparisons. Pairwise comparisons were considered significant when p < 0.05. If an animal was captured more than once, an average of all of its capture masses was used for analysis. Masses for both sexes by year were pooled and plotted for comparison to masses obtained from the primary literature. Because we were unable to obtain raw data from the literature, we compared our results to others by comparing error bars.

Results

We had a total of 771 captures of P. leucopus over the course of the study. We had 633 captures on Bumpkin Island and 138 on Peddocks. Of those captures, 272 were individually identifiable adult animals that we used in this study, 187 from Bumpkin and 85 from Peddocks. We did not recapture any animals from previous years on either island, despite tagging young adult animals each year. More juvenile and young adult animals were caught on Bumpkin than Peddocks Island. The only other species captured during these trapping intervals was the meadow vole (Microtus pennsylvanicus) on Bumpkin Island in all years except 2011. For an overview of the population dynamics and movement patterns of P. leucopus on Bumpkin Island in 2008, 2009, and 2011, see Nolfo-Clements and Clements (2015).

The number of adult animals captured on the islands varied from year to year, as did the rate of captures (Table 1). Linear models (R 2 females = −0.02, R 2 males = −0.01) for Peddocks Island suggest that body mass of P. leucopus did not change considerably between 2012 and 2015 (Fig. 2). Although the linear model slope estimate for females was positive (0.34) and negative for males (−0.62) there was no evidence that either were different from 0 (p females = 0.61, p males = 0.19). Linear models for Bumpkin Island P. leucopus (R 2 females = 0.42, R 2 males = 0.12) suggest that body mass of males increased by an average of 1.2 g per year during the study period (slopemales = 1.23, p males = 1.09e−14) and female body mass increased at a slower, yet significant, average rate (slopefemales = 0.69, p females = 1.99e−3) (Fig. 2).

Table 1 Captures per trap night and numbers of captures by sex for Peromyscus leucopus trapped on Bumpkin Island in 2008, 2009, 2011, 2014, and 2015 and Peddocks Island in 2012, 2014, and 2015. This table does not indicate the number of individuals trapped since some were trapped multiple times, but rather the total number of captures
Full size table
Fig. 2
figure 2

Male and female Peromyscus leucopus body masses from Bumpkin and Peddocks Island in Boston Harbor plotted against year of capture. To avoid overplotting and aid data visualization, body mass data points within years are “jittered” slightly (i.e., randomly offset along the x-axis). Linear models were fit independently for both sexes on each island. Bumpkin Island males (y-intercept = −2442.6 (p = 1.85e−14), slope = 1.22 (p = 1.01 e−14), R 2 = 0.42); Bumpkin Island females (y-intercept = −1352.3 (p = 2.44e−3), slope = 0.68 (p = 1.99 e−3), R 2 = 0.12); Peddocks Island males (y-intercept = −662.1 (p = 0.62), slope = 0.34 (p = 0.61), R 2 = 0.01); Peddocks Island males (y-intercept = 1289.5 (p = 0.18), slope = −0.62 (p = 0.19), R 2 = 0.03)

Full size image

The ANOVA to explore differences in body mass between sexes, over time, and interactions between sex and time confirm, at least partially, the analysis with linear models. Even though there was no overall difference in body mass between sexes (p sex = 0.12), the analysis did suggest, as did linear models, a strong effect of time on body mass (p year = 0.22e-16). The analysis further suggested an interaction between sex and time (p sex*year = 0.03). Pairwise comparisons (see Materials and methods) and inspection of an interaction plot (not shown) of average body mass and time revealed that mean mass of mice differed significantly between early (2008, 2009, and 2011) and later years (2014 and 2015) of the study. In particular, between 2008, 2009, and 2011 average male body mass increased steadily while average female body mass decreased. Both sexes show a large increase after 2011 (Fig. 2), but male body mass increased more than females so the average male was more massive than the average female in 2014 and 2015. Finally, males were significantly smaller than females in 2008 and 2009 by an average of 2.66 and 2.51 g, respectively.

Overall, masses of animals were significantly greater than those reported in the recent literature for both wild and captive mainland populations (Fig. 3). The average adult mass for P. leucopus from the recent literature was 19.9 g (Johnson et al. 2000; Young et al. 2000; Segre et al. 2002; Derting et al. 2003; Pyter et al. 2005; Cramer et al. 2006; Greenberg et al. 2006; Malo et al. 2010; Gibbes and Barrett 2011; Kaseloo et al. 2012; Thomason et al. 2013; Stephens et al. 2014). The masses were also significantly greater than those published by Adler and Tamarin (1984) for island and mainland populations of P. leucopus in eastern Massachusetts, USA which were also in the 19–20 g range.

Fig. 3
figure 3

Average body masses of Peromyscus leucopus captured on Bumpkin and Peddocks Islands in the Boston Harbor compared with masses reported for this species in the published literature since 2000. The error bars indicate one standard error. Multiple data bars for a citation indicate multiple study populations that were reported separately by the authors. White data bars indicate Bumpkin Island, black bars indicate Peddocks Island, light gray indicate wild populations, and dark gray bars indicate captive populations

Full size image

Discussion

One of the most notable aspects of our findings is that these Peromyscus mice exhibited the island syndrome despite their islands’ relatively small size and close proximity to the mainland. Studies have noted that rodents on smaller and more remote islands tend to be larger (Pergams and Ashley 2001). This is thought to be the case because the smaller and/or remote islands tend to have reduced trophic complexity when compared with mainland ecosystems (Russell et al. 2011). Reduced competition, predation, and food resources and/or shifts in life history strategy have all been noted as possible causes for morphological and behavioral changes observed in insular rodents (Palkovacs 2003).

Despite the similar mass results in the years for which we have data from both islands (2014 and 2015), the habitats on Bumpkin and Peddocks are notably different. The plant species composition varies between islands and Bumpkin Island is also home to a population of meadow voles while Peddocks Island only has P. leucopus in the habitats sampled. Although Peddocks Island is known to harbor Norway rats (Rattus novergicus) on other parts of the island. American mink (Neovison vison), a potential predator, have occasionally been reported on Bumpkin Island but not on Peddocks. Thus, the resource-based and competition-related environment probably varies for P. leucopus on these two islands. Both islands, however, are home to ephemeral populations of coyotes, white-tailed deer, and wild turkeys.

The increased body weight we observed is not likely due to natural geographic variation in body size for this species. The study data from wild populations presented in Fig. 3 comes from a variety of locations within the USA, including North Carolina, Ohio, Wisconsin, and Georgia (Cramer et al. 2006; Greenberg et al. 2006; Gibbes and Barrett 2011; Stephens et al. 2014). Despite this, the body masses recorded in those studies are very similar. Additionally, as previously mentioned, studies of P. leucopus from mainland Massachusetts report body weights in the same range as those from these disparate locations (Adler and Tamarin 1984).

As discussed, the size variation of Peromyscus mice on islands has been previously examined. Nonetheless, the genetic underpinnings of this variation have not been reported for this genus. However, there have been studies on the house mouse (Mus musculus) that examine the genetics of body size.

Gray et al. (2015) found that insular M. musculus on Gough Island in the Atlantic Ocean attain their unusually large body size through accelerated growth in the first 6 weeks of life. They identified 19 quantitative trait loci (QTLs) that were responsible for this increased growth rate (11 QTLs, 3–20% of variance) and overall larger body size (8 QTLs, 6–24% of variance) (Gray et al. 2015). Ishikawa et al. (2005) found that body size as related to growth and mass in wild M. musculus from the Philippines was in part controlled by 17 QTLs. Ishikawa and Okuno (2014) identified candidate genes that appear to significantly influence growth rate and body composition in M. musculus based upon 12 QTLs located on chromosome 2.

Divergence of insular house mice from mainland populations is frequently noted through microevolutionary changes in mandible shape which are attributed to dietary differences (Boell and Tautz 2011; Renaud et al. 2013). Babiker and Tautz (2015) studied the insular house mice on Heligoland, a small island in the North Sea off of the north-western coast of Germany. They found that, despite this island mouse population only being isolated from the mainland for about 400 years, its divergence was both morphologically and molecularly significant.

Although the genetics of island Peromyscus have not been explored, genetic variation and isolation in mainland populations has. Munshi-South and Kharchenko (2010) noted rapid genetic divergence in P. leucopus populations separated by urban development. Harris and Munshi-South (2016) uncovered genetic differentiation between populations of urban and rural P. leucopus.

Alternatively, the interannual variation in body size that we observed may be primarily due to phenotypic plasticity as a result of resource availability or climatic factors. Renaud et al. (2015) found that mandible shape in insular M. musculus on Guillou Island in the Kerguelen Archipelago can differ significantly between years most likely due to morphological plasticity in this genetically homogeneous island population. Seasonal and annual variations in copepod body size are often attributed to changes in temperature and animal density (Viitasalo et al. 1995), although comparing morphological changes in mammals to those in crustaceans should be done conservatively. We were not able locate any published studies that reported interannual variation in the body mass of wild mammals in general or rodents in particular.

Winters in Massachusetts, while relatively consistent in temperature range (averaging −3 to 5 °C), may vary widely in snowfall totals which typically range from 0.5 to 2.5 m or more. Summer rainfall totals are also highly variable, ranging from 0.1 to nearly 1 m. This variation has a direct impact not only on P. leucopus populations but also the plant and insect populations that form the resource base for this species. We have noted that the availability of raspberries and rose hips on Bumpkin Island varies widely from year to year (Nolfo-Clements, unpublished data). It is likely that the availability of other food sources also fluctuates due to this climatic variability.

In summary, P. leucopus on Bumpkin and Peddocks Island in the Boston Harbor, Massachusetts, USA appear to follow the island rule. This species’ unusually high body mass is significantly greater than masses reported in the literature for this species over a broad geographic range, even for an island population in Massachusetts. Furthermore, these mice have also undergone a significant increase in body mass over the course of our study. As body mass is known to be influenced at least in part by genetic inheritance, these changes may be an indication that this species is currently undergoing microevolutionary changes. These changes are readily observable on an annual basis due to this species’ yearly population turnover.