Northeastern Naturalist Vol. 24, No. 3 M.C. Henriquez, S.K. Macey, E.E. Baker, L.B. Kelly, R.L. Betts, M.J. Rubbo, and J.A. Clark 2017 249 2017 NORTHEASTERN NATURALIST 24(3):249–266 Translocated and Resident Eastern Box Turtles (Terrapene c. carolina) in New York: Movement Patterns and Habitat Use Megan C. Henriquez1,2, Suzanne K. Macey1,3, Erin E. Baker4,5, Lisa B. Kelly4, Rachel L. Betts1,6, Michael J. Rubbo4,7, and J. Alan Clark1,* Abstract - Translocation of animals to new habitats is a common conservation management strategy but is of uncertain effectiveness. Terrapene c. carolina (Eastern Box Turtle) are often the subject of translocation efforts. To understand the effectiveness of this strategy, we radio-tracked 19 translocated and 7 resident Eastern Box Turtles to assess movement patterns and habitat use, including hibernacula selection. Using data collected over 4 years from a nature reserve in New York, we compared home range, maximum distance traveled, and total distance traveled for both translocated and resident turtles. We found no difference between translocated or resident turtles or between sexes for any of these measures. These results suggest that translocated turtles at this site adapted well to their new habitat. Introduction Many turtle species in North America need to be actively managed because of increased urbanization (Dodd and Seigel 1991, Ferronato et al. 2015, Gibbs and Shriver 2002). Urbanization not only decreases the amount of suitable habitat available for these species, but it also fragments existing habitat (Belzer and Steisslinger 1999) and could lead to inhibition of turtle population growth (Ferronato et al. 2015, Gibbs and Shriver 2002, Steen and Gibbs 2004). Additionally, turtles in urban and suburban areas are threatened by collisions with vehicles, mowing activity, and people removing them from the wild either out of a concern for the turtles’ safety or to keep them as pets (Belzer and Steisslinger 1999, Budischak et al. 2006, Greenspan et al. 2015, Hester et al. 2008). Because turtles have delayed sexual maturity (Congdon and Gibbons 1990), the loss of individual adult turtles by these threats may decrease overall local population viability (Belzer and Steisslinger 1999, Budischak et al. 2006, Hall et al. 1999, Kipp 2003). To mitigate some of urbanization’s negative effects on turtle populations, researchers and conservation managers often use techniques known as “RRT”, i.e., the “repatriation, relocation, and translocation” of individuals, as a conservation management strategy (Dodd and Seigel 1991, Ferronato et al. 2015, Reinert 1991, Sosa and Perry 2015). The use of each of these terms by researchers and conservation managers varies throughout the literature. The International Union for the 1Department of Biological Sciences, Fordham University, Bronx, NY 10458. 2Current address - Department of Anthropology, The Graduate Center, City University of New York, New York, NY 10016. 3Center for Biodiversity and Conservation, American Museum of Natural History, New York, NY 10024. 4Teatown Lake Reservation, Ossining, NY 10562. 5Ramapo Ridge Middle School, Mahwah, NJ 07430. 6Centre College, Danville, KY 40422. 7Woodcock Nature Center, Wilton, CT 06897. *Corresponding author - jaclark@fordham.edu. Manuscript Editor: Peter K. Ducey Northeastern Naturalist 250 M.C. Henriquez, S.K. Macey, E.E. Baker, L.B. Kelly, R.L. Betts, M.J. Rubbo, and J.A. Clark 2017 Vol. 24, No. 3 Conservation of Nature defines the term “conservation translocation” as the act of moving organisms from one area and releasing them in another for conservation purposes (IUCN SSC 2013). For this study, we use the term “translocation”. To better understand the effectiveness of translocation efforts, researchers and conservation managers often track the movement patterns and survivorship of translocated individuals to investigate the relative success of translocation at an individual level (Ashton and Burke 2007, Cook 2004, Sosa and Perry 2015). Previous studies suggest that translocated individuals may have more difficulty acquiring resources, and, consequently, may have larger home ranges and travel greater distances than resident turtles in the same area (Attum and Cutshall 2015, Rittenhouse et al. 2007). The study of movement patterns and home ranges for translocated turtles can provide researchers with important information not only about dispersal and survival but also about how translocated individuals use their new habitat and the quality of that habitat (Greenspan et al. 2015, Kapfer et al. 2013, Stickel 1950). Terrapene spp. (box turtles) are commonly managed through RRT. In most published box turtle translocation studies, individuals are drawn from known populations and released into new habitats (introduction) or habitats supporting an existing population (reinforcement) (e.g., Cook 2004, Farnsworth and Seigel 2013, Samuelson 2012). Our reinforcement translocation study of Terrapene c. carolina (L.) (Eastern Box Turtle), listed as “vulnerable” by the IUCN (van Dijk 2011) and of “special concern” by New York State (Breisch and Behler 2002, NYSDEC 2017), is distinctive as it evaluated the effects of translocation on individuals for which little, if any, information was available on their provenance. We tracked turtles’ movement patterns and habitat use to better understand the potential for translocation as a conservation tool for nature reserves and wildlife rehabilitation centers receiving Eastern Box Turtles whose provenance was unknown—a common situation for this species. Because translocated turtles are unlikely to be familiar with their new habitat, we predicted that translocated turtles would have larger home ranges and travel farther than resident turtles. In addition, we explored whether translocated turtles spent more time in habitat types different than those used by resident turtles. We also predicted that differences in movement patterns between translocated turtles and resident turtles might decrease over time as translocated turtles gained familiarity with their new habitat. Finally, based on results of previous studies (e.g., Cook 2004, Kapfer et al. 2013, Stickel 1989), we did not expect to find a difference in movement patterns and habitat use between males and females. Field-site Description This study was conducted at a private nature reserve in suburban Westchester County, NY, ~55 km north of New York City. The site was comprised of 338 ha of mixed habitat, including lakes, deciduous Quercus (oak)–Carya (hickory) forests, hardwood swamps, and successional shrubland (Fig. 1). The successional shrubland portion of the site consisted of an open area located under 2 sets of power lines. The vegetation under the power lines was actively maintained and mowed or Northeastern Naturalist Vol. 24, No. 3 M.C. Henriquez, S.K. Macey, E.E. Baker, L.B. Kelly, R.L. Betts, M.J. Rubbo, and J.A. Clark 2017 251 cut down bi-annually, especially at the bases of the main power line stanchions. The area surrounding the reserve consisted mostly of private property and residences, which included manicured lawns, continuation of the oak–hickory forest, and orchards. For this study, any habitat not on the reserve’s property was considered “off-property” and not further categorized into specific habitat ty pes. Methods Field methods Resident turtles were found on the reserve prior to the beginning of the study. Whether resident turtles hatched from nests at the reserve, migrated naturally to the reserve, or were translocated to the reserve by local citizens at an earlier time was unknown, and, therefore, we labeled them as resident rather than native. The nature center at our study site in New York State regularly receives Eastern Box Turtles from concerned citizens who remove turtles from roads or private property, from people who no longer want them as pets, and from the New York State Department of Environmental Conservation (NYSDEC), which confiscates turtles from the pet trade (box turtles cannot be kept as pets in New York State, pursuant to NY Environmental Conservation Law §§11-0512). Upon receipt of these turtles by the Figure 1. Map of study site including habitat types used by Eastern Box Turtles and a subset of home ranges from one year (selected 2013 individuals, including, but not distinguishing between, males and females or translocated and resident turtles). White areas represent “off-property” areas with no habitat data. Shrubland was found only in the power line rights-of-way. Northeastern Naturalist 252 M.C. Henriquez, S.K. Macey, E.E. Baker, L.B. Kelly, R.L. Betts, M.J. Rubbo, and J.A. Clark 2017 Vol. 24, No. 3 nature center, every effort was made to obtain information about the individual's background and to return it to its habitat of origin. If, however, a turtle’s provenance could not be determined, we considered it for inclusion in this study. Translocated turtles were released at different times throughout the 4 years of this study in small groups (5–7 individuals) at a single designated release point within successional shrubland habitat on the reserve. We marked all study turtles, both translocated and resident, using a triangular metal file to notch the marginal scutes with unique notch codes (modified from Cagle 1939) and attached small radio transmitters (ATS, R1850, Isanti, MN) to the back-right marginal and costal scutes using waterproof epoxy (Oatey, Cleveland, OH) which was then rubbed with soil to obtain a more natural color and smell. The total weight of the transmitter and epoxy was less than 5% of the turtles’ weight. The transmitters’ antennae were not glued down, were 12.5 cm long, and came off the shell at a 15-degree angle facing the tail. Transmitter attachment was completed within 30 minutes. For resident turtles, we attached transmitters in the field and released the individuals at their capture location. For translocated turtles, we attached transmitters at the reserve’s on-site wildlife center prior to their release onto the reserve. All capture and study procedures were conducted under a NYSDEC Scientific License to Collect or Possess. Tracking To evaluate Eastern Box Turtle movement patterns and habitat use, we tracked 26 turtles for up to 4 years (2010–2013). Nineteen turtles were translocated (9 males and 10 females), and 7 turtles were residents (3 males and 4 females). We tracked turtles 2–3 times per week during their yearly active period, which we defined as the time they emerged from their hibernacula (~April), to the time they buried themselves underground for the winter (~November). We used a handheld R-1000 receiver and an RA-150 Yagi antenna (Communication Specialist, Inc., Beacon, NY) to locate individuals and a handheld GPS unit (Garmin GPS map60, accuracy less than 15 m) to record their location. If we were unable to track turtles multiple times per week, we checked their radio signal weekly to ensure that the transmitter was still working and the individual was still within range. During tracking events, any turtle that showed signs of disease or injury was brought to an on-site health facility and rehabilitated until deemed healthy by a certified wildlife rehabilitator. After rehabilitation, turtles were released at their last point of capture or another location within their established home range. When a turtle was found in an area considered unsafe (e.g., a road), we moved the turtle to its nearest previous tracking location. We tracked each turtle for the entire winter, albeit less frequently once an individual went below the soil surface, and considered the last location recorded for that calendar year to be the hibernaculum point for that year. Movement patterns Because of seasonal variation in Eastern Box Turtle activity (Dolbeer 1969, Stickel 1950) and the variation in our yearly tracking of study individuals, we standardized Northeastern Naturalist Vol. 24, No. 3 M.C. Henriquez, S.K. Macey, E.E. Baker, L.B. Kelly, R.L. Betts, M.J. Rubbo, and J.A. Clark 2017 253 the tracking data by using the same 20-week period (21 May–22 October) in all years of this study for the movement-pattern analysis. We define “tracking season” as this standardized 20-week period. Any turtle that was not located for 3 or more consecutive weeks at any time during the tracking season was excluded from all analyses for that year. We examined home range, total distance traveled, and maximum distance traveled for each tracking season for each individual and also averaged these measurements across all tracking seasons an individual was observed. We used global positioning system (GPS) coordinates to map and calculate each turtle’s home range using the Minimum Convex Polygon (MCP) function in ArcGIS 10.0 (ESRI 2011). Although some studies used fixed kernel density and harmonic means to estimate home range (e.g., Cook 2004, Kapfer et al. 2013, Refsnider et al. 2012), we used the MCP method because it is more commonly used in box turtle movement studies and so allowed us to more easily compare our results to these studies (see Appendix 2). To calculate the total distance traveled, we summed the distance traveled between each consecutive tracking event. To calculate the maximum distance traveled, we measured the greatest distance between any 2 (not necessarily consecutive) points in each turtle’s home range. We visually inspected and tested residual plots of the tracking season’s movement data for normality using the Shapiro-Wilkes test (shapiro.test function) in the stats package for R (R Code Team 2014). Because home range, total distance traveled, and maximum distance traveled were non-normal and residual variance increased with the mean, we log transformed movement data to meet the assumptions of normality. The final comparisons of movements were made using the linear mixed effect analysis (lmer function) in the lme4 package (Bates et al. 2014). In the linear mixed model, we set sex (i.e., male and female) and status (i.e., translocated or resident) as fixed effects. We set individual, calendar year, and number of tracking points as random effects. We compared each candidate model (i.e., sex and status interaction, sex and status, sex alone, status alone) to the null using a likelihood ratio test (anova function, stats package), and we compared candidate models to each other by creating AICc tables using the aictab function in the AICcmodavg package. For models significantly different than the null model, but not distinguishable from other candidate models, we performed model averaging to assess the effect of the parameters given model uncertainty (modavg function, AICcmodavg package). To detect trends in movement patterns over time for translocated turtles, we analyzed movement patterns considering the number of years (post-release) the turtle was present at the site (site year). Of the 19 translocated turtles tracked, 15 had adequate tracking data over multiple tracking seasons to be considered for this analysis, and we ran a separate set of linear mixed models on the movement data of these individuals. In these models, all of the above-mentioned random effects were maintained to account for differences in individuals, calendar year, and number of seasonal tracking points collected. The candidate models included sex and site year interaction, sex and site year, sex alone, and site year alone; a comparison of models was performed with the same approach as described above. Northeastern Naturalist 254 M.C. Henriquez, S.K. Macey, E.E. Baker, L.B. Kelly, R.L. Betts, M.J. Rubbo, and J.A. Clark 2017 Vol. 24, No. 3 Habitat use To assess relative habitat use for all study turtles, we used ArcGIS to superimpose GPS tracking locations onto a data layer of the habitat types for the reserve (i.e., oak–hickory forest, wet meadow, vernal pool, upland meadow, successional shrubland, hardwood swamp, shallow lake, and rocky slope). The habitat layer was constructed using aerial photography to delineate the different habitat types and was verified in the field. We conducted a compositional analysis to study habitat use at the study site, with the compana function using the parametric test in the adehabitatHS package for R (Calenge 2006). This analysis takes into consideration both the frequency of habitat use for each individual and the amount of habitat available within the reserve (Aarts et al. 2008, Aebischer et al. 1993). Any turtle position not on the reserve was considered “off-property” and not included in the compositional analysis as the description and proportion of off-property habitat was unknown. Compositional analyses cannot be used to compare differences between groups of individuals (e.g., males vs. females; Aebischer et al.1993). Therefore, for the 2 habitat types used most (based on the results of the compositional analysis), we performed generalized linear mixed models (glmer function; binomial family with logit link function; lme4 pacakage) to assess differences of habitat use among groups. The candidate models included sex and status interaction, sex and status, sex alone, and status alone; comparison of models was performed with the same approach as described above for the movement analysis. Overwintering behavior Box turtles often dig shallow cavities into the ground or use natural pits for hibernating during the winter (i.e., hibernacula) when food is scarce and temperatures are prohibitive to more active metabolic processes (Ultsch 2006). When turtles are translocated into new environments, finding appropriate hibernaculum sites may be challenging, as these turtles are not familiar with local habitat conditions (but see Cook 2004, Farnsworth and Seigel 2013). In New York, temperatures regularly drop below freezing, and securing a good hibernaculum is critical to a turtle’s survival (Carpenter 1957, Claussen et al. 1991). Therefore, we also tracked turtles’ overwintering behavior, including their selection of hibernacula. To evaluate overwintering behavior, we recorded the habitat type in which each hibernaculum site was located based on the habitat layer used in the habitat-use analysis. We also noted the frequency with which each hibernaculum habitat type was used, year to year, by each individual turtle. To assess hibernaculum fidelity, we defined fidelity to occur when a turtle buried itself less than 15 m from the previous year’s hibernaculum. The choice to use a measurement of less than 15 m was based on other studies (e.g., Refsnider et al. 2012, Stickel 1989) as well as our GPS unit’s accuracy, which was limited to less than 15 m. Northeastern Naturalist Vol. 24, No. 3 M.C. Henriquez, S.K. Macey, E.E. Baker, L.B. Kelly, R.L. Betts, M.J. Rubbo, and J.A. Clark 2017 255 Results Tracking The number of individual turtles whose tracking data were used for the study varied between years (2010 and 2011: n = 12; 2012: n = 20; 2013: n = 18) due to tracking inconsistency within the 20-week tracking season, transmitter failure/ availability, or mortality. Only 3 turtles (FRes1, FTrans4, and MRes2) in this study had data for all 4 tracking seasons (Appendix 1). However, an additional 3 turtles (FRes2, MRes3, MTrans4) were tracked over the entire 4-year study period, but only 3 tracking seasons within the 4-year period were included because of lack of sufficient data for 1 of the tracking seasons (Appendix 1). All measures and results are given ± S.D. Most individuals were tracked for 2 or more calendar years (mean = 2.4 ± 1.0 years). We collected 1721 tracking points during this study, and the number of points for an individual turtle during a tracking season varied from 16 to 61 points (mean = 28 ± 12.1 points) per tracking season across years. Movement patterns During the 20-week tracking season, individual home ranges varied from 0.03 to 79.8 ha (mean = 8.4 ± 15.0 ha), while individual averages across tracking seasons ranged from 0.07 to 50.6 ha (mean = 8.2 ± 12.0 ha) (see 2013 home ranges in Fig. 1; all home ranges given in Appendix 1). Total distance traveled, or the sum of the distances between all tracking points collected per individual in 1 tracking season, varied from 280 to 13860 m (mean = 2247.6 ± 2529.8 m). Average total distances, or the average of all the tracking season total distances collected for each individual, ranged from 332 to 6787 m (mean = 2253.6 ± 1833.2 m; Appendix 1). Maximum distance, or the longest distance between any 2 points per individual in one tracking season, varied from 28 to1451 m (mean = 442.5 ± 342.5 m). Average maximum distances, or the average of all tracking season maximum distances collected per individual, ranged from 42 to 1441 m (mean = 590.3 ± 565.4 m) (Appendix 1). Based on the results of the linear mixed model comparisons, none of the candidate models in any movement category were better performing than the null model (Table 1). For translocated turtles, the amount of time (site year) after release did not appear to affect home ranges, although some individuals had large changes in home ranges from year to year (Table 2). For example, MTrans9 had an increase of 32.9 ha (284%) between site year 1 and site year 2, while MTrans2 had a 16.3-ha (80%) decrease in home range between site year 1 and site year 2. MTrans2’s home range then stabilized for the third year, keeping a ~3.9-ha home range for both the second and third year after release. This stabilization of home range over time was not observed in FTrans2, who had an 11.9-ha (75%) decrease in home range between site year 1 and 2 and then a 7.3-ha (187%) increase between site year 2 and 3. Not including these 3 individuals with relatively large home-range changes, the average change (either increasing or decreasing) of home range over 2 consecutive site years was 1.8 ± 1.9 ha. When comparing candidate models for movement patterns in translocated turtles, neither sex, site year, nor the interaction between sex and site year had a significant effect on movement patterns over time (Table 2). Northeastern Naturalist 256 M.C. Henriquez, S.K. Macey, E.E. Baker, L.B. Kelly, R.L. Betts, M.J. Rubbo, and J.A. Clark 2017 Vol. 24, No. 3 Table 1. Eastern Box Turtle movement model comparisons based on sex and status as translocated or resident. Models are listed in order of increasing AICc values. K is the number of estimable parameters (degrees of freedom), AIC is the Akaike Information Criteria, AICc is the AIC corrected for small sample size, ΔAICc is the difference in AICc from the “best” performing model, AICc Wt is the relative weight of each model, and P is the P-value from the likelihood ratio test comparing the candidate model to the null model (random effects only). No models were considered different than the null. Model K AIC AICc Δ AICc AICc Wt P Home range Status 6 186.8 188.30 0.00 0.43 0.624 Sex 6 186.9 188.41 0.11 0.19 0.718 Sex + status 7 188.6 190.71 2.41 0.13 0.828 Sex : status 8 190.2 192.94 6.85 0.02 0.850 Maximum distance Sex 6 160.0 161.55 0.00 0.43 0.563 Status 6 160.2 161.72 0.17 0.39 0.687 Sex + status 7 161.9 163.93 2.38 0.13 0.779 Sex : status 8 162.9 165.65 4.09 0.05 0.698 Total distance Status 6 160.9 162.38 0.00 0.45 0.429 Sex 6 161.4 162.89 0.52 0.34 0.742 Sex + status 7 163.2 164.80 2.42 0.13 0.688 Sex : status 8 163.2 165.89 3.51 0.08 0.512 Table 2. Translocated Eastern Box Turtle movement model comparisons based on sex and site year. Models are listed in order of increasing AICc values. K is the number of estimable parameters (degrees of freedom), AIC is the Akaike Information Criteria, AICc is the AIC corrected for small sample size, ΔAICc is the difference in AICc from the “best” performing model, AICc Wt is the relative weight of each model, and P is the P-value from the likelihood ratio test comparing the candidate model to the null model (random effects only). No models were considered different than the null. Model K AIC AICc Δ AICc AICc Wt P Home range Sex 6 122.4 124.64 0.00 0.92 0.936 Site Year 8 125.9 130.04 5.40 0.06 0.928 Sex + Site Year 9 127.9 133.21 8.56 0.01 0.976 Sex : Site Year 12 133.1 143.12 18.47 0.00 0.987 Maximum distance Sex 6 119.0 121.29 0.00 0.94 0.312 Site Year 8 123.5 127.58 6.29 0.04 0.902 Sex + Site Year 9 124.4 129.73 8.44 0.01 0.808 Sex : Site Year 12 125.9 135.99 14.70 0.00 0.526 Total distance Sex 6 113.3 115.57 0.00 0.77 0.747 Site Year 8 114.3 118.41 2.85 0.19 0.376 Sex + Site Year 9 116.1 121.42 5.85 0.04 0.512 Sex : Site Year 12 120.4 130.48 14.91 0.00 0.661 Northeastern Naturalist Vol. 24, No. 3 M.C. Henriquez, S.K. Macey, E.E. Baker, L.B. Kelly, R.L. Betts, M.J. Rubbo, and J.A. Clark 2017 257 Habitat use Considering the amount available of each different habitat type, the turtle location data collected during the tracking season indicated that turtles generally preferred certain habitats (i.e., successional shrubland and oak–hickory forest) to others ( λ = 6.9e-16, df = 9, P < 0.001; Table 3). Although the successional shrubland was found only under the power lines and was less available than the oak–hickory forest, successional shrubland was the most frequently used habitat by translocated turtles (Table 4). When investigating differences in successional shrubland habitat use considering sex and status, the model with status alone was significantly different than the null model but not different from the model with both sex and status (Table 5). When predictor variables were averaged across models, translocated turtles do not use successional shrubland more than resident turtles (95% unconditional CI of the model-averaged estimate of status includes zero; Fig. 2), but translocated turtles do use oak–hickory forest habitat significantly less than resident turtles (Table 5, Fig. 2). Table 4. Habitat type and percentage (± SD) use by Eastern Box Turtles organized by status as translocated or resident and sex (2010–2013). Any habitat types not used by study turtles are not included. Habitat types are listed by most to least used by all individuals pooled. Habitat All individuals Resident Translocated Female Male Shrubland 44.25 ± 0.25 33.44 ± 0.24 48.67 ± 0.25 40.37 ± 0.23 48.96 ± 0.28 Forest 37.20 ± 0.18 43.44 ± 0.18 34.65 ± 0.17 37.89 ± 0.17 37.89 ± 0.19 Off-property 8.89 ± 0.20 6.69 ± 0.12 9.79 ± 0.23 12.12 ± 0.25 12.12 ± 0.10 Swamp 4.75 ± 0.13 13.46 ± 0.22 1.19 ± 0.04 6.79 ± 0.17 2.28 ± 0.06 Rocky slope 3.93 ± 0.11 2.11 ± 0.05 4.68 ± 0.13 2.27 ± 0.05 5.96 ± 0.15 Vernal pool 0.63 ± 0.02 0.66 ± 0.02 0.61 ± 0.02 0.23 ± 0.01 1.11 ± 0.02 Lake 0.29 ± 0.01 0.21 ± 0.21 0.32 ± 0.01 0.34 ± 0.01 0.23 ± 0.01 Wet meadow 0.06 ± 0.00 0.00 ± 0.00 0.08 ± 0.01 0.00 ± 0.00 0.13 ± 0.01 Table 3. Comparisons of habitat use for individual Eastern Box Turtles over all years sampled via compositional analysis. Table reads left to right, and sign relates to the comparative relationship between habitat types (i.e., - is used less, + is used more). Triple signs mean statistical difference (P < 0.05) in the sign direction. Area of habitat and percentage of that habitat type within the study site are listed under habitat type abbreviations. Habitat type SH F SW RS VP L WM Area (ha) 14.4 243.6 22.5 19.1 1.3 23.4 1.5 % 4.2 71.3 6.6 5.6 0.4 6.8 0.4 Shrubland (SH) +++ +++ +++ +++ +++ +++ Forest (F) +++ +++ +++ +++ +++ Swamp (SW) - - +++ + Rocky slope (RS) - +++ + Vernal pool (VP) +++ +++ Lake (L) --- Wet meadow (WM) Northeastern Naturalist 258 M.C. Henriquez, S.K. Macey, E.E. Baker, L.B. Kelly, R.L. Betts, M.J. Rubbo, and J.A. Clark 2017 Vol. 24, No. 3 Table 5. Eastern Box Turtle habitat-use model comparisons based on status as translocated or resident and sex for shrubland and forest habitat types. Models are listed in order of increasing AICc values. K is the number of estimable parameters (degrees of freedom), AIC is the Akaike Information Criteria, AICc is the AIC corrected for small sample size, ΔAICc is the difference in AICc from the “best” performing model, AICc Wt is the relative weight of each model, P is the P-value from the likelihood ratio test comparing the candidate model to the null model (random effects only). Model K AIC AICc Δ AICc AICc Wt P Shrubland Status** 5 412.4 413.44 0.00 0.49 0.049* Sex + status** 6 413.1 414.61 1.17 0.27 0.076 Sex 5 414.9 415.99 2.54 0.14 0.253 Sex : status 7 414.5 416.53 3.09 0.01 0.123 Forest Status** 5 391.4 392.44 0.00 0.45 0.003* Sex : status** 7 391.3 393.35 0.91 0.28 0.005* Sex + status** 6 392.0 393.55 1.10 0.26 0.006* Sex 5 399.0 400.06 7.61 0.01 0.295 *Models statistically different compared to null. **Models with less than 2 ΔAICc from the “best” model and therefore model averaging was p erformed. Figure 2. Comparisons of predictor variables’ model-averaged coefficients with 95% unconditional confidence intervals (CI) for successional shrubland and oak–hickory forest habitat use by Eastern Box Turtles. When 95% CI did not include zero (did not cross the dashed line), the predictor variable was considered to have an effect on habitat use. The direction of the effect is based on the comparison of translocated turtles to resident turtles for status and males to females for sex and on the sign of model-averaged coefficient; for example, translocated turtles use forest less than resident turtles (as represented by the negative model-averaged coefficient and the CI range not including zero). Northeastern Naturalist Vol. 24, No. 3 M.C. Henriquez, S.K. Macey, E.E. Baker, L.B. Kelly, R.L. Betts, M.J. Rubbo, and J.A. Clark 2017 259 Overwintering behavior We collected multiple years of hibernacula data for 22 of 26 turtles. Turtles overwintered in only 2 habitat types: oak–hickory forest and successional shrubland located under power lines. The oak–hickory forest was the preferred overwintering habitat, with 17 of 22 (77%) turtles hibernating there for at least 1 winter. Six of 7 (86%) resident turtles and 11 of 15 (73%) translocated turtles hibernated in oak– hickory forest at least once. Seventeen of 22 (77%) hibernating turtles exhibited fidelity to habitat type with 15 of those 17 (88%) selecting oak–hickory forest for each year observed. The average distance between any 2 hibernaculum locations across all individuals was 124.9 ± 125.7 m. Only 3 of 22 (14%) showed hibernaculum fidelity (within 15 m of a previous hibernaculum site), including 2 translocated turtles (FTrans6 and MTrans4) and 1 resident turtle (FRes2). Discussion Movement patterns Home range. Consistent with previous studies of Eastern Box Turtles (Cook 2004, Kapfer et al. 2013, Stickel 1989), we found no difference in home ranges between males and females. Unlike most previous studies comparing translocated and resident box turtles (Appendix 2), the home ranges of our translocated turtles were not larger than the home ranges of our resident turtles. Differences in home range in other studies may reflect differing analytical approaches, population densities, or habitat quality at different sites (Appendix 2; Stickel 1950, 1989). When comparing the results of other studies to ours, we note that our data were based on a 20-week tracking season that may not have captured all of a turtle’s movements during the entire active season or be directly comparable to other studies that used different tracking periods. In addition to differing tracking periods, the large variability of home ranges for translocated turtles in our study could reflect differences in the origins of the turtles themselves, as the provenance of our translocated turtles was unknown. Cook (2004), however, saw no differences between movement patterns for released pets versus translocated wild Eastern Box Turtles of known provenance. After translocated turtles acclimatize to their new environment, home ranges might be expected to decrease and stabilize over time (Hester et al. 2008); however, we found no pattern of change in home ranges over time. The large differences in home ranges between years exhibited by a few turtles may be due simply to individual variation in home range stability, as seen in Nichols (1939). Total distance traveled. Only a small number of turtles (both translocated and resident) in our study showed irregular movement patterns, such as long unidirectional movement. For example, a resident male showed strong linear movement (MRes1= 600 m) alternating between 2 smaller sub-home ranges during the course of each of tracking season. In both our study and Hester et al.’s (2008) study, the average total distance traveled by translocated and resident turtles did not differ. However, in both status categories, our turtles traveled ~1 km less than Hester et al.’s population, which may reflect the shorter time period we used for movement Northeastern Naturalist 260 M.C. Henriquez, S.K. Macey, E.E. Baker, L.B. Kelly, R.L. Betts, M.J. Rubbo, and J.A. Clark 2017 Vol. 24, No. 3 analysis within a year (20-weeks), differences in habitat quality, or differences in resource distribution at each site. Maximum distance traveled. Translocation projects may be hindered by animals leaving the study site in their attempts to return to their habitat of origin, a process known as homing (Berry 1986, Rittenhouse et al. 2007) or because the target site for the translocation is too small or of too poor quality for the persistence of additional turtles (Cook 2004, Dodd and Seigel 1991). Translocation studies often attempt to capture dispersal events by tracking the maximum distance traveled and orientation of the dispersal (e.g., Cook 2004, Lemaku 1970, Rittenhouse et al. 2007). Based on the maximum dispersal distance recorded in his study (113–1295 m), Cook (2004) concluded that a translocation site should be at least 300 ha; our findings on the average maximum distance traveled for translocated turtles (42–1441 m) support this recommendation. Habitat use Preferred habitat type. Eastern Box Turtles are generally considered a forestdwelling, terrestrial species (Budischak et al. 2006, Dodd 2001, Williamson 2013). For example, Williams and Parker (1987) reported resident Eastern Box Turtles in Indiana strongly favored forest habitat and were rarely seen in power line rightsof- way. In our study, we also found that resident turtles used forest habitat most frequently but found that translocated turtles used the successional shrubland habitat under the power line rights-of-way most frequently. Our results suggest that translocated individuals may be using the reserve’s habitat differently than residents, but we acknowledge the reduced use of oak–hickory forest may be associated with the translocation release point being in the successional shrubland. Our choice for a release point possibly affected the translocated turtles’ habitat selection and home-range establishment (Brichieri-Colombi and Moehrenschlager 2016). Even so, both resident and translocated turtles preferred to hibernate in the oak–hickory forest which, for translocated turtles, was outside the habitat type of their release point. Power lines are ubiquitous in areas affected by urbanization, and our translocated turtles’ use of habitat beneath power lines raises interesting questions. While power lines could provide quality habitat for box turtles, the active management of powerline rights-of-way could also pose a threat. At our site, the local electric company mowed the areas below power lines bi-annually to avoid overgrowth of grasses, shrubs, and trees. Although none of our study turtles were injured or killed by this maintenance, previous studies have shown that mortality could result from mowing or harvesting grasses (Nazdrowicz et al. 2008). Consequently, if turtles preferentially use these actively managed areas, they may experience higher mortality. Off-property use. No turtles migrated away from the study site, though several individuals temporarily left the borders of the reserve. Turtles in unprotected offproperty habitat, such as yards and agriculture fields, may be more vulnerable to mortality events due to mowing, collisions with vehicles, and collection (Greenspan Northeastern Naturalist Vol. 24, No. 3 M.C. Henriquez, S.K. Macey, E.E. Baker, L.B. Kelly, R.L. Betts, M.J. Rubbo, and J.A. Clark 2017 261 et al. 2015). Although off-property was the third most frequently used habitat in this study, the unprotected off-property habitat at our site often included high quality continuations of oak–hickory forest or successional shrubland; as a result, the use of off-property habitat may not have conferred greater risks to individuals. Overwintering behavior Other Eastern Box Turtle studies showed variability in hibernaculum fidelity (e.g., Claussen et al. 1991, Cook 2004). For example, Stickel (1989) found that Eastern Box Turtles selected hibernacula sites from 8 to 24 m from previous years’ hibernacula sites. Cook (2004) reported distances between consecutive hibernacula averaged 97.6 ± 89.9 m, but he described the variability not only between individuals in a population, but also between years for a single individual. These results are comparable to our own findings of high levels of individual variability in hibernaculum selection. Conclusions Our study found no differences between home ranges and movement patterns of translocated and resident turtles. However, the translocated turtles in our study did not spend as much time in the oak–hickory forest habitat as resident turtles. This difference could be associated with the release point for our translocated turtles, which was in successional shrubland. Future translocation projects should carefully consider release locations. Translocation sites need to include enough space for relocated individuals to have sufficient access to food, reproductive opportunities, and other factors critical to the species’ survival at all life stages (Dodd and Seigel 1991). The size of our site and the variety of available habitat types may have allowed for translocated individuals to successfully acquire such resources. Nature reserves and wildlife rehabilitation centers often receive animals whose provenance was unknown. Our results suggest that Eastern Box Turtles can be successfully translocated to new habitats, even if their provenance is unknown. A limitation of our study is that we were not able to determine whether the resident turtles in our study were hatched on-site or whether they were previously released on-site by well-meaning private citizens. Perhaps through rigorous marking programs or genetic analyses, future studies could know with more certainty the origin of both resident and translocated turtles. Only 1 study turtle died during the 4 years of this study: a young translocated male (MTrans1) displaying evidence of winter exposure and dehydration (Appendix 1). Other studies of translocated box turtles had higher mortality rates (Farnsworth and Seigel 2013, Hester et al. 2008). Because the reserve had a wildlife rehabilitator on site, we were able to readily address health issues presented by either translocated or resident turtles. These healthcare measures might account for some of the difference between our study and others that had higher study turtle mortality. Even if consistent health monitoring is not possible for future translocation studies, we strongly recommend that turtles be quarantined and tested for disease prior to release, especially considering the increased concern Northeastern Naturalist 262 M.C. Henriquez, S.K. Macey, E.E. Baker, L.B. Kelly, R.L. Betts, M.J. Rubbo, and J.A. Clark 2017 Vol. 24, No. 3 about and documentation of disease outbreaks such as Ranaviris in box turtles (Kimble et al. 2017). While the definition of a successful RRT project is unclear, many researchers and managers agree that successful reproduction and incorporation of translocated individuals into existing populations are some of the primary qualifiers (Burke 1991, Dodd and Seigel 1991, Brichieri-Colombi and Moehrenschlager 2016). In this study, we did not track reproductive patterns on either the individual or population level. Future translocation studies would benefit from following individuals for longer periods of time, while focusing on survivorship, mating patterns, reproduction, population stability, and the implications translocations have on the health and genetic structuring of resident populations. Nature centers and wildlife rehabilitation centers will likely continue to acquire individual Eastern Box Turtles whose provenance is unknown. The ability to successfully relocate such individuals may be important for this species’ conservation, and this study suggests that such translocations have the potential to be a useful conservation tool. Acknowledgements We thank the staff at the nature center at our study site for logistical support and the nature center’s wildlife center for receiving and rehabilitating turtles, while providing consistent support with field data collection. We are also grateful to Karina Polanco and James Herrera for assistance with data analysis. This project was supported by Fordham University, Fordham University’s Calder Summer Undergraduate Research program, and the National Science Foundation Research Experience for Undergraduates program. Literature Cited Aarts, G., M. MacKenzie, B. McConnell, M. Fedak, and J. Mattiopoulos. 2008. Estimating space-use and habitat preference from wildlife telemetry data. Ecography 31:140–160. Aebischer, N.J., P.A. Roberston, and R.E. Kenward. 1993. Compositional analysis of habitat use from animal radio-tracking data. Ecology 74:1313–1325. Ashton, K.G., and R.L. Burke. 2007. Long-term retention of a relocated population of Gopher Tortoises. The Journal of Wildlife Management 71:783–787. Attum, O., and C.D. Cutshall. 2015. Movement of translocated turtles according to translocation method and habitat structure. Restoration Ecology 23:588–594. Bates, D., M. Maechler, B. Bolker, and S. Walker. 2014. lme4: Linear mixed-effects models using Eigen and S4. R package version 1.1–7. Available online at http://CRAN.Rproject. org/package=lme4%3E. Accessed 3 January 2015. Belzer, B., and M.B. Steisslinger. 1999. The box turtle: Room with a view on species decline. The American Biology Teacher 61:510–513. Berry, K.H. 1986. Desert Tortoise (Gopherus agassizii) relocation: Implications of social behavior and movements. Herpotologica 42:113–142. Breisch, A.R., and J.L. Behler. 2002. Turtles of New York State. Conservationist 57:15–18. Brichieri-Colombi, T.A., and A. Moehrenschlager. 2016. Alignment of threat, effort, and perceived success in North American conservation translocations. Conservation Biology 30:1159–1172. Northeastern Naturalist Vol. 24, No. 3 M.C. Henriquez, S.K. Macey, E.E. Baker, L.B. Kelly, R.L. Betts, M.J. Rubbo, and J.A. Clark 2017 263 Budischak, S.A., J.M. Hester, S.J. Price, and M.E. Dorcas. 2006. Natural history of Terrapene carolina (Box Turtles) in an urbanized landscape. Southeastern Naturalist 5:191–204. Burke, R.L. 1991. Relocations, repatriations, and translocations of amphibians and reptiles: Taking a broader view. Herpetologica 47:350–357. Cagle, F.R. 1939. A system of marking turtles for future identification. Copeia 1939:170–173. Calenge, C. 2006. The package “adehabitat” for the R software: A tool for the analysis of space and habitat use by animals. Ecological Modelling 197:516–519. Carpenter, C.C. 1957. Hibernation, hibernacula, and associated behavior of the Three-toed Box Turtle (Terrapene carolina triunguis). Copeia 1957:278–282. Claussen, D.L., P.M. Daniel, S. Jiang, and N.A. Adams. 1991. Hibernation in the Eastern Box Turtle, Terrapene c. carolina. Journal of Herpetology 25:334–341. Congdon, J.D., and J.W. Gibbons. 1990. The evolution of turtle life histories. Pp. 45–54, In J.W. Gibbons (Ed.). Life History and Ecology of the Slider Turtle. Smithsonian Institution Press, Washington, DC. 317 pp. Cook, R.P. 2004. Dispersal, home-range establishment, survival, and reproduction of translocated Eastern Box Turtles, Terrapene c. carolina. Applied Herpetology 1:197–228. Dodd, Jr., C.K. 2001. North American Box Turtles: A Natural History. University of Oklahoma Press, Norman, OK. 231 pp. Dodd, C.K., and R.A. Seigel. 1991. Relocation, repatriation, and translocation of amphibians and reptiles: Are they conservation strategies that work? Herpetologica 47:336–350 Dolbeer, R.A. 1969. A study of population density, seasonal movements, and weight changes, and winter behavior of the Eastern Box Turtle, Terrapene c. carolina L., in Eastern Tennessee. Master's Thesis. University of Tennessee, Knoxville, TN. 53 pp. Environmental Systems Research Institute (ESRI). 2011. ArcGIS Desktop: Release 10. Redlands, CA. Farnsworth, S.D., and R.A. Seigel. 2013. The responses, movements, and survival of relocated box turtles during construction of the Intercounty Connector highway in Maryland. Transportation Research Record 2362:1–8. Ferronato, B.O., J.H. Roe, and A. Georges. 2015. Urban hazards: Spatial ecology and survivorship of a turtle in an expanding suburban environment. Urban Ecosystems 19:415–428. Gibbs, J.P., and W.G. Shriver. 2002. Estimating the effects of road mortality on turtle populations. Conservation Biology 16:1647–1652. Greenspan, S.E., E.P. Condon, and L.L. Smith. 2015. Home range and habitat selection in the Eastern Box Turtle (Terrapene carolina carolina) in a Longleaf Pine (Pinus palustris) reserve. Herpetological Conservation and Biology 10:99-111. Hall, R.J., P.F.P. Henry, and C.M. Bunck. 1999. Fifty-year trends in a box turtle population in Maryland. Biological Conservation 88:165–172. Hester, J.M., S.J. Price, and M.E. Dorcas. 2008. Effects of relocation on movements and home ranges of Eastern Box Turtles. The Journal of Wildlife Management 72:772–777. International Union for Conservation of Nature Species Survival Commission (IUCN SSC). 2013. Guidelines for reintroductions and other conservation translocations. IUCN Species Survival Commission, Gland, Switzerland. Kapfer, J.M., D.J. Muñoz, J.D. Groves, and R.W. Kirk. 2013. Home range and habitat preferences of Eastern Box Turtles (Terrapene carolina Linnaeus, 1758) in the Piedmont Ecological Province of North Carolina (USA). Herpetology Notes 6:251–260. Northeastern Naturalist 264 M.C. Henriquez, S.K. Macey, E.E. Baker, L.B. Kelly, R.L. Betts, M.J. Rubbo, and J.A. Clark 2017 Vol. 24, No. 3 Kimble, S.A.J., A.J. Johnson, R.N. Williams, and J.T. Hoverman. 2017. A severe Ranavirus outbreak in captive, wild-caught box turtles. EcoHealth doi:10.1007/s10393-017- 1263-8. Kipp, R.L. 2003. Nesting ecology of the Eastern Box Turtle (Terrapene carolina carolina) in a fragmented landscape. Master's Thesis. University of Delaware, Newark, DE. 78 pp. Lemaku, P.J. 1970. Movements of the Box Turtle, Terrapene c. carolina (Linnaeus) in unfamiliar territory. Copeia 1970:781–783. Nazdrowicz, N.H., J.L. Bowman, and R.R. Roth. 2008. Population ecology of the Eastern Box Turtle in a fragmented landscape. The Journal of Wildlife Management 72:745–753. New York State Department of Environmental Conservation (NYSDEC). 2017. List of Endangered, Threatened and Special Concern Fish and Wildlife Species of New York State. http://www.dec.ny.gov/animals/7494.html. Accessed 25 January 2017 Nichols, J.T. 1939. Data on size, growth, and age in the box turtle, Terrapene carolina. Copeia 1939:14–20. R Code Team. 2014. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. Refsnider, J.M., J. Strickland, and F.J. Janzen. 2012. Home range and site fidelity of imperiled Ornate Box Turtles (Terrapene ornata) in Northwestern Illinois. Chelonian Conservation and Biology 11:78–83. Reinert, H.K. 1991. Translocation as a conservation strategy for amphibians and reptiles: Some comments, concerns, and observations. Herpetologica 47:357–363. Rittenhouse, C.D., J.J. Millspaugh, M.W. Hubbard, and S.L. Sheriff. 2007. Movements of translocated and resident Three-toed Box Turtles. Journal of Herpetology 41:115–121. Samuelson, C.S. 2012. Movement patterns in resident and translocated Three-Toed Box Turtles (Terrapene carolina triunguis). Master’s Thesis. University of Texas, Tyler, TX. 54 pp. Sosa, J.A., and G. Perry. 2015. Site fidelity, movement, and visibility following translocation of Ornate Box Turtles (Terrapene ornata ornata) from a wildlife rehabilitation center in the high plains of Texas. Herpetalogical Conservation and Biology 10:255–262. Steen, D.A., and J.P. Gibbs. 2004. Effects of roads on the structure of freshwater turtle populations. Conservation Biology 18:1143–1148. Stickel, L.F. 1950. Populations and home range relationships of the Box Turtle, Terrapene c. carolina (Linnaeus). Ecological Monographs 20:351–378. Stickel, L.F. 1989. Home range behavior among Box Turtles (Terrapene c. carolina) of a bottomland forest in Maryland. Journal of Herpetology 23:40–44. Ultsch, G.R. 2006. The ecology of overwintering among turtles: Where turtles overwinter and its consequences. Biological Reviews 81:331–367. van Dijk, P.P. 2011. Terrapene carolina. (errata version published in 2016) The IUCN Red List of Threatened Species 2011:e.T21641A97428179. Accessed on 26 January 2017. Williams, E.C.J., and W.S. Parker. 1987. A long-term study of a Box Turtle (Terrapene carolina) population at Allee Memorial Woods, Indiana, with emphasis on survivorship. Herpetologica 1987:328–335. Williamson, B.A. 2013. Examining habitat selection and home-range behavior at multiple scales in a population of Eastern Box Turtles (Terrapene c. carolina), with notes on demographic changes after 17 years. Master’s Thesis. Marshall University, Huntington, WV. 88 pp. Northeastern Naturalist Vol. 24, No. 3 M.C. Henriquez, S.K. Macey, E.E. Baker, L.B. Kelly, R.L. Betts, M.J. Rubbo, and J.A. Clark 2017 265 Appendix 1. Sex, status, health notes, and average movement measurements over multiple tracking seasons taken for all Eastern Box Turtles in this study. Turtle ID corresponds to sex (F = female; M = male), status (Res = resident; Trans = translocated), and individual ID number. The lower portion of the table is data summarized based on categories of turtles. # of seasons = number of seasons tracked. In health notes, treated indicates individuals that were treated for upper respiratory tract infections, and the 1 individual who died likely succumbed to exposure during the overwintering period. Measurements are given ± SD. # of Average home Average total Average max. Turtle ID seasons Health notes range (ha) distance (m) distance (m) FRes1 4 14.44 ± 10.94 2701 ± 1550 613 ± 259 FRes2 3 1.18 ± 0.98 786 ± 369 215 ± 107 FRes3 1 5.16 1631 374 FRes4 1 0.61 813 173 FTrans1 3 2.90 ± 1.48 1515 ± 188 417 ± 235 FTrans2 3 10.31 ± 5.99 1874 ± 1015 581 ± 236 FTrans3 3 Treated 2012 0.61 ± 0.26 723 ± 539 131 ± 17 FTrans4 4 3.02 ± 1.21 1816 ± 1141 449 ± 238 FTrans5 3 3.05 ± 1.59 1848 ± 538 427 ± 178 FTrans6 3 2.01 + 0.64 2461 ± 1717 426 ± 315 FTrans7 2 3.90 ± 0.79 1523 ± 127 333 ± 92 FTrans8 2 4.42 ± 5.50 1389 ± 1388 293 ± 248 FTrans9 1 2.10 1404 370 FTrans10 1 48.04 4437 1441 MRes1 2 21.10 ± 25.57 6787 ± 7941 742 ± 589 MRes2 4 Treated 2012, 2013 28.78 ± 35.82 6301 ± 5959 831 ± 523 MRes3 3 1.03 ± 0.77 1137 ± 445 156 ± 53 MTrans1 2 Died 2011 0.84 ± 1.15 1041 ± 1077 151 ± 179 MTrans2 3 9.31 ± 9.36 3268 ± 1787 743 ± 51 MTrans3 3 Treated 2012 1.05 ± 0.61 1009 ± 524 223 ± 35 MTrans4 3 Treated 2010 0.57 ± 0.67 504 ± 279 119 ± 67 MTrans5 2 Treated 2013 7.23 ± 4.35 2225 ± 1068 456 ± 167 MTrans6 2 Treated 2013 1.08 ± 0.84 908 ± 408 213 ± 151 MTrans7 1 0.07 332 42 MTrans8 1 50.57 6134 1142 MTrans9 2 28.06 ± 23.29 4026 ± 1278 818 ± 448 Residents (n = 7) 1 treated 10.33 ± 11.29 2880 ± 2590 443 ± 283 Translocated (n = 19) 4 treated, 1 died 9.43 ± 15.42 2026 ± 1488 462 ± 359 Females (n = 14) 1 treated 7.27 ± 12.35 1784 ± 952 446 ± 318 Males (n = 12) 4 treated, 1 died 12.47 ± 16.22 2806 ± 2437 469 ± 368 Northeastern Naturalist 266 M.C. Henriquez, S.K. Macey, E.E. Baker, L.B. Kelly, R.L. Betts, M.J. Rubbo, and J.A. Clark 2017 Vol. 24, No. 3 Appendix 2. Comparison of home ranges in box turtle (Terrapene sp.) movement studies. Difference Average MCP (ha) between Species Translocated Resident groups? Analytical method Study Terrapene c. carolina (Eastern Box Turtle) 9.43 10.33 No MCP This study 9.80 - - 95% bivariate normal Cook 2004 4.80 - - 95% harmonic mean Cook 2004 14.71 4.31 Yes MCP Farnsworth and Seigel 2013 18.02 6.45 Yes MCP Hester et al. 2008 - 2.68 - MCP Kapfer et al. 2013 Terrapene c. triunguis (Agassiz) 14.22 7.89 No MCP Samuelson 2012 (Three-Toed Box Turtle) Terrapene ornate (Ornate Box Turtle) - 0.40 - 50% fixed kernel Refsnider et al. 2012