Wood turtles (Glyptemys insculpta, formerly Clemmys insculpta, Holman and Fritz 2001) are among many freshwater turtle species with declining populations in eastern North America (DeGraaf and Rudis 1987; Doroff and Keith 1990; Stearns et al. 1990; Heppell 1998; Lewis and Faulhaber 1999). The IUCN Red List classifies the wood turtle as a vulnerable species (Hilton-Taylor 2000). The Convention on International Trade in Endangered Species of Wild Fauna and Flora regulates import and export of wood turtles internationally (species listed June 1992, Appendix II), and the Michigan Department of Natural Resources lists the wood turtle as a “Species of Special Concern,” which prohibits collection without a research permit. The US Forest Service develops and implements management practices to ensure that sensitive species are not adversely affected because of Forest Service actions. Effective management recommendations for the remaining populations of wood turtles require defining their home ranges and monitoring populations.

Wood turtles are especially vulnerable to disturbance because of a high natural mortality of eggs and juveniles combined with relatively slow reproductive and growth rates (Harding 1990; Buech and Nelson 1997). Heppell (1998) noted that, unlike desert tortoises and sea turtles with relatively high juvenile elasticities, most freshwater turtles share high adult survivorship but low fecundity. These population patterns suggest that conservation efforts to reduce adult mortality will most effectively serve to stabilize declining populations (Garber 1989; Harding 1990; Congdon et al. 1993; Heppell 1998). Many individual wood turtles now live for only a fraction of their potential reproductive life spans (Harding 1990; Ross et al. 1991). Garber and Burger (1995) discussed loss of adults and failure of recruitment as causes for a rapid decline of a population in Connecticut. The present study provides information on home ranges and movements that conservationists can apply to management plans to increase adult survival.

Once common and inhabiting much of the north-central and eastern United States and adjacent Canadian regions, wood turtles now exhibit a less continuous distribution in southeastern Canada, the Great Lakes area, New England, and northern Virginia (Harding and Bloomer 1979; Ernst et al. 1994). Studies documenting recent declines in G. insculpta attribute losses primarily to human activities such as habitat alteration, road mortality, and commercial collection of turtles for the pet trade and biological supply (Harding and Bloomer 1979; Harding 1990; Brooks et al. 1992; Garber and Burger 1995). For example, when municipal watershed managers opened a formerly restricted forest in Connecticut to hiking and fishing, a thriving population of wood turtles declined to apparent extirpation within 10 years (Garber and Burger 1995). Human activity in natural areas results in removal of turtles by visitors, disturbance by their dogs, road kills from increased traffic, and increased predation by raccoons (Procyon lotor) and other mammalian predators attracted to garbage. Our study site provides a critical study area because it is a river corridor receiving heavy recreational use by campers, fishers, hikers, off-road vehicle recreationists, canoeists, and rafters.

The results of previous studies on the ecology of G. insculpta vary considerably, suggesting that management decisions require studies of local populations. Harding and Bloomer (1979) and Ernst et al. (1994) provided general information on the life history and habits of wood turtles. The animals utilize openings in the streamside forest for feeding and basking (Harding 1990). Mating and hibernation generally occur in water (Bishop and Schoonmacher 1921; Harding and Bloomer 1979; Ernst 1986; Harding 1990; Farrell and Graham 1991), whereas the majority of wood turtle terrestrial activity occurs away from the river during the summer. Based on a study of intermediate-range homing, Carroll and Ehrenfeld (1978) proposed that wood turtle orientation may be based on olfactory cues or a combination of olfactory and magnetic inputs. Studies of wood turtle home ranges and habitat use have been conducted in Pennsylvania (Ernst 1968a, 2001; Strang 1983; Kaufmann 1995), West Virginia (Niederberger and Seidel 1999), New Hampshire (Tuttle and Carroll 1997), Canada (Quinn and Tate 1991; Arvisais et al. 2002), Wisconsin (Ross et al. 1991), and the upper peninsula of Michigan (Harding and Bloomer 1979; Harding 1985, 1990). In addition to the present study, further research is needed on the wood turtle population at our study site in northern Michigan. Our purpose in conducting this study was to determine home ranges of wood turtles in a population within a heavily used river corridor and to consider potential variables affecting home range.

METHODS

Research took place along a 42-km stretch of a river in northeastern Michigan. The lands along the river fall within the boundaries of the 177,023-ha (437,247-acre) Huron National Forest. Dominant vegetative types are 23% hardwood, 21% aspen (Populus spp.) and birch (Betula spp.), 44% pine (Pinus spp.), 5% lowland conifer, 1% mixed swamp hardwood, and 6% open areas. The river averages 30 m wide and 1 m deep, with mean summer temperatures around 20°C (US Department of Agriculture 1986). The substrate of the river and surrounding lands is primarily sandy. The study area receives heavy use during the summer months for recreational purposes such as canoeing, tubing, fishing, and camping in certain designated areas along the river.

From May through June of 1998–2000, using a canoe, we actively searched for basking wood turtles along the banks of the river. Several incidental captures were also made while locating radiotelemetered turtles. All turtles were captured by hand and released at the point of capture within 1 hour. Data recorded for each turtle included age, sex, body size, and mass. We measured body mass with a hanging spring scale accurate to ± 10 g. We recorded carapace width, carapace length, and plastron length of the midline along the shell contour to the nearest 0.1 cm with a flexible ruler so we could compare them to previous measurements taken along the river before this study. Minimum age was estimated by counting annular rings on plastral scutes, although annuli become less reliable after 15–20 years of growth (Harding and Bloomer 1979; Garber 1989; Kaufmann 1992). Counting rings at least 3 times improved accuracy. For adult turtles (age > 9), plastron concavity and position of the cloaca relative to the edge of the carapace were used to determine the sex of individuals (Pope 1939; Ernst et al. 1994), although other data suggest that, in Michigan, a higher age (> 12) might be more appropriate to determine sex in this way (J. Harding, pers. comm., June 2003). However, annual monitoring of this population through 2003 did not lead to reclassification of turtles aged between 9 and 12 years during this study. We used a US Forest Service numbering scheme, similar to that of Cagle (1939), to mark each turtle. Using a triangular file, we notched the marginal scutes of the carapace.

Radio transmitters (Advanced Telemetry System, Isanti, MN) weighing < 10 g (< 2% of the average wood turtle mass) were affixed to 10 turtles per field season, with one transmitter failing too early in 1998 to provide data for this study. For nontelemetered turtles, we recorded the same morphological and environmental data. Using 5-minute epoxy (Devcon, Danvers, MA), we attached transmitters to the carapace on a rear costal or marginal scute so that the antenna extended horizontally behind the turtle. Observations of mounting behavior suggested that this placement would not interfere with mating. A 3-element yagi antenna and a 2-MHZ scanning receiver (Advanced Telemetry System, Isanti, MN) were used for radiotelemetry. The telemetered turtles were relocated a minimum of 21 (average = 31) times each between the end of May and the middle of August, the period during which we expected wood turtles to travel farthest from the river. Observations in May, October, and February supported this impression. Farrell and Graham (1991) stated that wood turtles in New Jersey remain active from late April or early May until October, when they begin hibernation. Most terrestrial movements occur between the periods of spring mating and hibernation, which are aquatic activities (Bishop and Schoonmacher 1921; Harding and Bloomer 1979; Ernst 1986; Harding 1990; Farrell and Graham 1991). We tracked turtles from the date of spring capture through the life expectancy of the transmitters, and we replaced 5 transmitters in the fall of 2000 for winter telemetry and further studies. We relocated turtles 3–6 times per week during the summer and once per month for the winter locations, with a minimum of 1 day between successive observations to minimize autocorrelation (Anderson 1982).

The location of each observation was determined using a hand-held Trimble GPS and Pathfinder 2.10 postdifferential correction software (Trimble Navigation, Sunnydale, CA), which corrected all positions to < 1 m (Rempel and Rodgers 1997). TRACKER software (Gallerani Lawson, and Rodgers 1997) was used to calculate home ranges from the turtle location data. Home ranges were estimated using 95% and 50% adaptive kernel method (AK) (Whorton 1989), as well as 100% and 50% minimum convex polygon method (MCP) (Mohr 1947) (Fig. 1). The AK analysis computed probabilities of locations based on the density of locations where more weight is given to areas used most heavily, whereas MCP came from the area of a convex polygon formed by connecting the outermost animal locations (Anderson 1982). We calculated turtle distances from the river edge using Arc Map GIS 3.1 software (Environmental Systems Research Institute, Inc., Redlands, CA).

View Figure

• Download as Powerpoint
• Download Figure

With SigmaStat 2.0 software (SPSS Science, Chicago, IL), we checked for significant differences in home ranges between males and females of the same year using t-tests, between years of study with sexes combined using ANOVA tests with Dunn's method for multiple comparisons, between age cohorts of 5 years with Pearson correlations, and between carapace lengths and weights using linear regression tests. We also tested for significant differences in habitat wetness among the 3 study years using drought index data for the national forest, where the drought index indicates the net effect of evapotranspiration and precipitation on the upper soil (Keetch and Byram 1988; Bert Starr, U.S. Forest Service, pers. comm., May 2001).

RESULTS

We captured 68 different wood turtles, including 38 females, 20 males, and 10 juveniles. Of these, 6 individuals were recaptured in more than 1 year. We calculated home ranges for 29 turtles (20 females, 8 males, and 1 juvenile). Two individual turtles were tracked in 1998 and again in 2000. Home ranges spanned from 0.4 to 354 ha (95% adaptive kernel method). Although we found no statistical differences for home ranges associated with turtle age, carapace length, or sex (all p > 0.05), differences among home ranges between the 3 study years were statistically significant (p < 0.005). Summer home ranges were larger in 2000 than in the previous years (p < 0.05), although for 1 of 2 turtles tracked in both 1998 and 2000, the opposite trend occurred (from 21.6 to 7.8 ha in 95% adaptive kernal and from 19.3 to 4.6 ha in 100% MCP, although both measures of 50% core areas increased in 2000).

The summer of 2000 was wetter than either 1998 or 1999, although not in terms of simple median drought index value (p = 0.086). There were twice as many days with a drought index > 50 in 1998 and 1999 than in 2000. In addition, 2000 had no days with the drought index > 165, whereas 1998 had 3 days where the drought index was > 200, and 1999 had 2 days where the drought index was > 300.

Home ranges were spatially associated with the river corridor. Of 955 turtle locations in 1998–2000, 92.5% were within 200 m of the river. Ten of 29 telemetered turtles moved > 200 m from the river, 6 of these turtles remained in the same generally unforested areas away from the river for more than 2 days, typically for weeks. Only 2 turtles, composing less than 4% of turtle locations (n = 36), traveled more than 500 m from the river (both in 1999).

DISCUSSION

Comparisons among home range studies of the same species are difficult to make because even slight differences in methods can change estimates. Furthermore, comparing a density-weighted estimate with one that is not density-weighted can be misleading (Anderson 1982; White and Garrot 1990; Larkin and Halkin 1994; Gallerani Lawson and Rodgers 1997). Our average summer home range, 40.6 ha (using the 95% adaptive kernal) is larger than other reported values (Table 1), some of which include more time than just summers. Quinn and Tate (1991) reported average summer home ranges of 24.30 ha, whereas Ross et al. (1991) reported an average of 0.40 ha, both estimated using the MCP method. Our 100% MCP calculation, 6.7 ha, falls between these 2 sizes. Strang (1983) reported home range lengths averaging 447 m using a modified home range estimator.

Table 1. Previously reported average home ranges of wood turtles, including differences in reporting methods. Sex is indicated where studies differentiated. Minimum convex polygons (MCP) calculate the home range that encompasses the given percentage of all locations with no weighting given to areas of high numbers of locations whereas adaptive kernels (AK) give more weight to heavily used areas in estimating home range size.

View Fullscreen

Larger home ranges for our study population may indicate poor habitat quality because turtles may need to travel farther to find suitable food and resources (Plummer and Shirer 1975; Gibbons et al. 1990; Manseau and Gauthier 1993; but see Brown and Brooks 1993). Sexton (1959) reported that painted turtles (Chrysemys picta) shift their feeding activity ranges in response to habitat conditions such as scarce vegetation. Accurate assessment of habitat quality requires studies of foraging behavior or forage resources because food quality can have important consequences on the fitness of an organism (Manseau and Gauthier 1993). Wood turtles are active foragers, feeding on numerous plant and animal taxa (Strang 1983; Ernst et al. 1994). This high degree of omnivory suggests opportunistic use of available habitats by wood turtles. In our study area, anthropogenic effects such as logging, stream bank stabilization, and fisheries habitat management or physical disturbances, including flooding or fire, could have reduced the number of favorable nesting or foraging sites, forcing individuals to search farther for acceptable habitats.

Male and female average home ranges in our study did not differ significantly. However, Ross et al. (1991) found that males travel greater daily distances than do females, despite similar home ranges. This may be attributed to males' active pursuit of mates (Morreale et al. 1984; Ernst 1986; Brown and Brooks 1993; Lewis and Faulhaber 1999). Several authors have reported seasonal long-distance movements of female freshwater turtles to ideal nesting sites later in the summer, a strategy that may enhance the individuals' genetic fitness (Morreale et al. 1984; Gibbons et al. 1990; Harding 1990; Quinn and Tate 1991; Lewis and Faulhaber 1999).

Home ranges were significantly larger in 2000 than in the previous 2 years. Drought index values were lower in 2000, which may have improved vegetation or turtles' ability to move along adjunct waterways. Farrell and Graham (1991) found that wood turtles tend to be highly aquatic all year, especially during a year with a dry, hot summer. Wood turtles lose water through evaporation faster than the more terrestrial box turtles (Terrapene carolina) (Ernst 1968b). A relatively moist season in 2000 may have enabled the turtles to travel farther in search of higher quality foods or more secure resting areas. However, this study was not designed to test this hypothesis, and the correlation is only noted here for future use.

Similar to wood turtles studied elsewhere, telemetered turtles in this study generally traveled less than 200 m from the river. Juvenile wood turtles (n = 10) in Wisconsin remained less than 40 m from the river (Brewster and Brewster 1991), whereas turtles in Pennsylvania (n = 205) wandered 100–400 m from the nearest water body (Ernst 1986). Kaufmann (1992) reported that more than 95% of observed movements occurred within 300 m of a Pennsylvania creek, with occasional movements of up to 600 m away from a creek. In a Maine population, 95% of wood turtle activity areas were within 304 m of rivers and streams (Compton et al. 2002). Harding and Bloomer's (1979) study in Michigan recorded most adults (n = 47) remaining within 150 m of a stream. Overall, telemetry data indicate heavy use of the riparian corridor by wood turtles.

Benefits of maintaining close proximity to the channel include protection from predation (Brewster and Brewster 1991) and a more varied microhabitat than in other areas (Kaufmann 1992). The densest wood turtle populations seem to occur where a variety of habitat types are available to provide diverse cover for thermoregulation and food sources (Kaufmann 1992; Ewert et al. 1998; Compton et al. 2002). Streams provide openings in a dense forest canopy, so turtles may have access to wide-ranging thermal and moisture gradients in their microenvironments by moving relatively short distances. When turtles were not in the river, we frequently observed them in forest clearings such as campsites or canopy gaps. Our observations of 6 turtles at the same nonriver locations (> 200 m from the river) throughout the summer suggest that these may be sites where the turtles found optimal food sources or safety.

The mobility we observed in individuals in this population may be a result of wood turtles' sensitivity to habitat disturbance. Relatively large home ranges reported here, which occasionally included nonriver sites, indicate that protection of the species' habitat in Michigan should include extensive riparian areas. The patterns of movements documented in this study identify known wood turtle habitats important in land use planning.