To provide estimates of Blanding's Turtle (Emydoidae blandingii) hatchling survival and to better understand the utility of alternative management tactics targeting this age class, we monitored survival and movements after natural (caged) and artificial incubation by using radio telemetry. We found that survival was similarly high (ca. 80% over 88 days) across treatments and study locations. Movement distances were similar among treatments but differed among study locations, perhaps because of differences in release site habitat variables. Our results suggest that nest cages and artificial incubation are equally effective methods for increasing survival to hatching. Extrapolating from the 88 days of our study, until resumption of activity following hibernation, we found survival estimates of 40%–78%, depending on the survival function used. When coupled with published rates of nest survival (6%–41%) and hatch success (47%–87%), anticipated age 0 survival, from egg deposition to emergence from hibernation, ranged from 1%–28%. Although our analysis fills a knowledge gap in Blanding's Turtle demography, further study is needed to improve the precision of survival estimates.
Managers of small populations of threatened and endangered species often implement actions to enhance survival during vulnerable life history stages. A frequent concern regarding such actions is that follow-up monitoring is insufficient to meaningfully evaluate alternative tactics (Stem et al., 2005). This is especially true of species with cryptic life stages for which monitoring effort may be prohibitive and technological solutions are lacking (Pike et al. 2008). Hatchling turtles represent one such example. One strategy in turtle conservation is to reduce nest depredation by caging nests in situ or by inducing oviposition in wild-caught females, artificially incubating eggs, and releasing hatchlings (Burke, 2015; O'Connor et al., 2017). These tactics require a significant allocation of resources to track and monitor nesting females and their nests or to track and capture gravid females for induction and egg incubation. In addition, environmental conditions during incubation (temperature, humidity, and their variability) differ between nest-caged and artificially incubated hatchlings, and these differences may influence performance (survival and behavior) posthatching (Usategui-Martin et al., 2019). Consequently, evaluating the outcome of these alternatives is needed to better achieve conservation goals.
The Blanding's Turtle Emydoidea blandingii is a long-lived, late-maturing turtle that uses wetland and adjacent upland habitat (Congdon et al., 2008; Congdon et al., 2011; Reid et al., 2016). Threats to this species include habitat loss, road mortality, and elevated rates of nest predation by subsidized predators, resulting in a lack of recruitment (Congdon et al., 2008). The Blanding's Turtle is ranked as endangered by the International Union for Conservation of Nature (IUCN, 2020), is recognized as being in need of conservation or is listed as threatened or endangered in each U.S. State and Canadian Province in which it occurs (Congdon et al., 2008; COSEWIC, 2016), and is scheduled for candidate status review under the U.S. Endangered Species Act (U.S. Fish and Wildlife Service, 2015).
Posthatch survival of juvenile Blanding's Turtles is poorly known. Nest failure rates, mostly because of nest predators, can be high, ranging from 59%–94% (survival rate = 6%–41%; Butler and Graham, 1995; Congdon et al., 2000; Standing et al., 2000; Reid et al., 2016; Urbanek et al., 2016). Hatching failure, because of infertility or abnormal development, is less frequent, ranging from 13%–53% (survival rate = 47%–87%; Emrich, 1991; Butler and Graham, 1995; Congdon et al., 2000; Joyal et al., 2000; Standing et al., 2000). Posthatch survival estimates from tracking studies using telemetry or fluorescent powder range from 20%–82% (Camaclang, 2007; Arsenault, 2011; Jones and Sievert, 2012; Paterson et al., 2012). Unfortunately, these studies of hatchling survival span variable (and sometimes unreported) time intervals and fail to account for unknown outcomes (e.g., transmitter loss or failure) or removal from study as transmitters reach the anticipated end of battery life or fluorescent powder trails become undetectable.
Management practices aimed at increasing early life stage survival of Blanding's Turtles include nest caging, hatchling releases after artificial incubation, headstarting, and mesopredator control (Standing et al., 2000; Urbanek et al., 2016; Starking-Symanski et al., 2018; Carstairs et al., 2019; Thompson et al., 2020). Here, we compare the effectiveness of nest caging and artificial incubation by examining Blanding's Turtle survival and movement distances after hatchling release. We focus on these two methods because they are relatively simple, requiring only the monitoring of adult females, and thus might be undertaken by individuals or agencies lacking the resources for headstarting or mesopredator control. By providing estimates of posthatch survival, we fill a knowledge gap in Blanding's Turtle demography. We make use of formal survival analysis, in which outcomes other than mortality are treated as “censored,” to estimate survival and associated confidence limits over the duration of our study and interpolate survival over longer time periods (Collet, 2003), e.g., from hatching through spring emergence after first hibernation. Our analyses, together with recent estimates of survival among older (≥1 yr) juveniles (Golba, 2019) and adults (Congdon et al., 1993; Rubin et al., 2004; Ruane et al., 2008; Reid et al., 2016; Golba, 2019), enhance knowledge of Blanding's Turtle survival more generally.
Materials and Methods
We combined data from three separate studies spearheaded by authors MK, ARK, and JK and located within 100 km of each other in northern Illinois and southern Wisconsin (Kane and Lee County, Illinois = Kane/Lee; Lake County, Illinois and Kenosha County, Wisconsin = Lake/Kenosha; Rock County, Wisconsin = Rock; Table 1). All locations consisted of shallow wetlands with abundant emergent aquatic vegetation and were the focus of ongoing Blanding's Turtle monitoring and telemetry (exact localities withheld at the request of land managers), providing us with the opportunity to observe nesting females and cage their nests or collect gravid females to obtain eggs for artificial incubation. To cage nests, we checked telemetered animals in late May and early June for afternoon or evening overland movements typical of nesting females. We discretely followed these females to their nesting sites. After a turtle oviposited, we caged nests by using closed cylinders made of 1.27-cm mesh hardware cloth measuring 30 cm in diameter and 25 cm high. Cages were buried to a depth of about 10 cm and held in place with rebar and zip ties or weights placed on flanges at their base. Caged nests were checked daily for hatchlings from late August through mid-September. To obtain artificially incubated hatchlings, we palpated females for shelled eggs, induced oviposition, and incubated eggs as described in Thompson et al. (2020).
Hatchlings from both caged and artificially incubated nests were measured and weighed. Hatchlings greater than 8 g were outfitted with a small radio transmitter (Advanced Telemetry Systems, 0.5–0.6 g) glued to their carapace by using quick-setting epoxy. Hatchlings were released into the shallows of their home preserve wetlands (Kane/Lee) or on land at known nest sites 20–150 m from wetland margins (Lake/Kenosha, Rock). Telemetered hatchlings were located every 1 to 3 days (using Advanced Communications R-1000, ATS R-2000, or Lotek Biotracker receiver) until the onset of cool weather, after which they were located approximately weekly. Hatchlings were monitored until death, disappearance, or transmitter detachment. Transmitters reaching the end of their expected battery life (30–40 days) were sometimes replaced to extend tracking duration. Remaining transmitters were removed before anticipated battery failure. The distance moved between sequential locations was measured using a flexible tape or from global positioning system (GPS) coordinates.
We used Kaplan-Meier survival analysis 1) to compare survival between incubation methods by using data on 19 hatchlings from caged nests and 27 artificially incubated hatchlings from the Kane/Lee study; 2) to compare survival among study locations by using data on 47 Kane/Lee hatchlings, 18 Lake/Kenosha hatchlings, and 17 Rock hatchlings; and 3) to estimate survival across methods and locations by using data on all 82 hatchlings combined. Wild-caught hatchlings were excluded from comparisons of hatchlings from caged nests to artificially incubated hatchlings (n = 1) but were included in comparisons among studies and combined estimates (n = 2). Analyses were carried out using SPSS 25.0, IBM. Documented deaths and disappearances occurring within the expected transmitter battery life were scored as mortalities. Disappearances were confirmed by extending the search area and repeating searches on subsequent dates. Transmitter detachments and removals were treated as censored data as was a single mortality attributed to antenna entanglement. For our combined analysis, we fit exponential, Weibull, and Gompertz survival functions to our Kaplan-Meier results by using the curve fitting function of SPSS. These functions differ in whether they treat the risk of mortality as constant (exponential function) or monotonically changing over time (Weibull and Gompertz functions; Collett, 2003). A useful feature of these functions is that they allow extrapolation of survival beyond the end of our telemetry study.
We restricted our analysis of movement distance to movements measured over intervals of 1 to 5 days. Because movement distances were skewed right, we transformed data by using natural logarithms after adding 1 to achieve normality. We computed mean movement distance for each turtle for which five or more distances were recorded. We used a t-test to compare mean movement distance between hatchlings from caged nests (n = 15) and hatchlings from artificially incubated eggs (n = 20) from the Kane/Lee study. We used analysis of variance with Tukey post-hoc tests to compare mean movement distances among hatchlings from the Kane/Lee (n = 36), Lake/Kenosha (n = 15), and Rock studies (n = 16). Equal variances were confirmed using Leven's test before analysis was conducted.
Transmitters were placed on 82 hatchling turtles (Table 1), and survival was monitored for up to 88 days (Appendix 1). Thirteen turtles were confirmed or inferred to have died during our study. Causes of mortality included predation (n = 5 carcasses showing evidence of trauma), entanglement (n = 1), and unknown causes (2 carcasses without trauma and 5 animals that disappeared within expected battery life). The fates of the remaining 69 turtles were treated as censored because of transmitter loss (n = 13), battery failure (n= 3), or conclusion of the study (n = 53).
We analyzed 1,049 movements by 67 hatchling turtles (5–45 movements per turtle) (Appendix 1). Median distance moved between telemetry locations was 1.5 m (range = 0–294 m). Distance moved was positively correlated with the number of elapsed days between locations but only weakly so (r2 = 0.017, n = 1,049, P < 0.001), and thus no correction was made for elapsed days. We found no difference in distance moved between hatchlings from nest cages and hatchlings from artificially incubated eggs (back-transformed mean = 2.1 vs. 2.0 m; t = 3.323, df = 33, P = 0.749) in the Kane/Lee study. We found significant differences in movement distance among study locations (back-transformed mean = 2.1 m in Kane/Lee, 2.4 m in Lake/Kenosha, and 8.6 m in Rock; F2,64 = 41.943, P < 0.001). Tukey post-hoc comparisons revealed Kane/Lee = Lake/Kenosha < Rock (Fig. 3).
We used radiotelemetry to test for differences in survival and movement distances between Blanding's Turtle incubation methods and among study locations and to estimate survival across methods and locations. We found no significant difference in survival between Blanding's Turtle hatchlings from natural (caged) nests and those produced by artificial incubation. Likewise, we found no significant difference in survival among studies despite the fact that Lake/Kenosha and Rock hatchlings were released at nest sites and had to make overland movements to reach wetlands, whereas Kane/Lee turtles were released in wetland shallows. Admittedly, our ability to detect differences in survival is limited by sample size and study duration. For example, given our sample size and observed survival rates, our power to detect a difference among study locations is approximately 0.66 (https://www.statstodo.com/SSizSurvival_Pgm.php; Machin et al., 2009). We found no difference in movement distance between hatchlings from natural nests and those produced by artificial incubation. The similarity in the survival and movement distance of hatchlings from caged nests and those produced by artificial incubation suggests that both tactics may be effective for reducing mortality caused by nest predators. We recognize that hatchlings from caged nests and those produced by artificial incubation may differ in ways not detected in our study or that differences may not become apparent until later in life (e.g., effect on growth of constant vs. variable incubation temperature; Booth 2006). Thus, further study is warranted. Regardless, both tactics are labor intensive and should be evaluated against other options (no intervention, nest site restoration, headstarting, and mesopredator control) when designing management plans.
We did find differences in movement distance among studies, with Rock hatchlings moving significantly farther than Lake/Kenosha and Kane/Lee hatchlings. Possibly, the shorter movement distances exhibited by Kane/Lee hatchlings were a consequence of being released at wetland margins and not having to make overland movements. However, movement distances also differed between Rock and Lake/Kenosha hatchlings, of which all were released on land at known nest sites, suggesting that local habitat features may also be important.
Overall, during our investigation, we found hatchling Blanding's Turtle survival to be relatively high, ca. 80% over 88 days, exceeding that observed in Nova Scotia (20%–50%; n = 29, 36, 16, and 18 turtles tracked from hatching to hibernation in 2006, 2007, 2008, and 2010; Arsenault, 2011) and Ontario (42%, n = 48 turtles tracked for 53 days posthatching; Paterson et al., 2012). However, in the Nova Scotia and Ontario studies, 31%–75% of hatchlings were lost because of transmitter detachment, transmitter failure, or unknown causes, potentially biasing survival estimates downward. A similar rate of survival to our study was observed in Massachusetts by using fluorescent powder to track 72 turtles, but it occurred over a much shorter time interval (range = 1–48 days, mean = 4.5 days posthatching; Jones and Sievert, 2012). Our estimates also exceed those observed in hatchling wood turtles (11%, n = 42 turtles tracked for up to 58 days; Paterson et al. 2012) and hatchling gopher tortoises at 2 of 3 sites (88-day survival = ca. 30%, 50%, and 100%; n = 45, 20, and 20, estimated from Fig. 2 in Pike and Seigel 2006; gopher tortoise survival decreased to 0% at all 3 sites by 365–765 days).
Survival beyond the 88 days of our study can be extrapolated using survival functions fit to the results of our Kaplan-Meier analysis. Given a typical hatch date for our study locations of 3 September, we computed survival until resumption of activity following hibernation (ca. 1 May), an interval of 240 days. The Gompertz survival function results in inferred survival of 78% over 240 days, with almost no mortality occurring beyond the end of our study (Fig. 2). The Weibull function results in inferred survival of 65% with modest overwinter mortality, and the exponential function results in inferred survival of 40% (Fig. 2). Statistical fit alone favors the Gompertz function (r2 = 0.93 vs. 0.86 and 0.52 for the Weibull and exponential functions, respectively), but it seems likely that at least modest rates of overwinter mortality occur among young Blanding's Turtles. Regardless, when coupled with published rates of nest survival (6%–41%; Butler and Graham, 1995; Congdon et al., 2000; Standing et al., 2000; Reid et al., 2016; Urbanek et al., 2016) and hatch success (47%–87%; Butler and Graham, 1995; Congdon et al., 2000; Emrich, 1991; Joyal et al., 2000; Standing et al., 1999), extrapolated hatchling survival rates (40%–78%) result in anticipated age 0 survival (defined here as extending from egg deposition to emergence from hibernation) of just 1%–28%.
Although estimates of age class 0 turtle survival are accumulating, most are limited to nest and egg survival and few encompass the entire age class. For example, of 54 estimates of age 0 survival tabulated by Iverson (1991), just 7 included hatchlings. Fifteen such estimates were tabulated by Heppell (1998), but no distinction was made between egg survival and survival over the entire age class. An alternative approach, advocated by Pike et al. (2008), is to infer juvenile survival rates from information on adult survival, age at maturity, and clutch size, assuming constant population size. The result is a mean annual survival rate, averaged over the entire juvenile stage (ca. 14 yr in Blanding's Turtles) that reveals little about survival during specific year classes (e.g., age 0 as in this study). The paucity of data on hatchling and juvenile turtle survival (compared, e.g., to adult survival; Rachmansah et al. 2020) reflects the technical challenges of tracking (e.g., via telemetry) and monitoring (via capture–mark–recapture) young turtles. Filling this knowledge gap for additional species will provide a better understanding of turtle demography and aid conservation planning. Regardless, evidence is growing that increasing the survival of this vulnerable age class (via nest protection, artificial incubation, headstarting, and mesopredator control) increases population persistence both in Blanding's Turtles (Urbanek et al., 2016; Carstairs et al., 2019; Golba 2019; Thompson et al., 2020) and in other turtle species (e.g., Vander Haegen et al. 2009; Munscher et al. 2012; Milinkovitch et al. 2013; Peñaloza et al. 2015; Shaver and Caillouet 2015; Engeman et al. 2016; Quinn et al. 2018).
Funding and logistical support was provided by The Nature Conservancy, Lake County Forest Preserve District, Forest Preserve District of Kane County, Forest Preserve District of DuPage County, Northern Illinois University, University of Illinois Urbana Champaign, U.S. Fish and Wildlife Service, University of Wisconsin–Whitewater, P. and G. Shackelford, S. Foster, R. Hay and Turtles for Tomorrow, C. Vogel, the Shearer Family, Mr. and Mrs. Hodge, C. Sweeny, M. Watrous, R. Conway, B. Parker, E. Sweeney, and J. Van Altena. Work was carried out under permits from the Illinois Department of Natural Resources (05-11s, 07-04s, and 16-045), Wisconsin Department of Natural Resources (SCPSRLN-19-26, 586, and 645), Illinois Nature Preserves Commission, and The Nature Conservancy. Institutional Animal Care and Use Committee approval was provided by Northern Illinois University (LA16-0015), University of Illinois (06129), and University of Wisconsin–Whitewater (K145011020Q). We gratefully acknowledge field assistance provided by J. Atkinson, K. Cassel, D. Fritz, C. Golba, K. Hausmann, B. House, J. Lorenz, P. Pieper, K. Rebman, K. Schmidt, Z. Welch, and S. Wyrick and the comments of two anonymous reviewers.
Present Address: Department of Forestry and Natural Resources, University of Kentucky, Lexington, Kentucky, 40546, USA
Present Address: Mountain Lake Biological Station, Department of Biology, University of Virginia, Charlottesville, Virginia, 22904, USA
Present Address: School of Medicine and Public Health. University of Wisconsin-Madison. Madison, Wisconsin, 53719, USA
Present Address: West Allis Central High School, West Allis, Wisconsin, 53227, USA