Methods. —

Our study took place at 2 ponds in Plymouth County, Massachusetts. We do not report the specific locations due to concerns of illegal collection and instead use the pseudonyms Pond A (4.5 ha) and Pond B (6.4 ha). Both sites are coastal plain ponds and are 1.3 km apart and surrounded by managed stands of pitch pine (Pinus rigida) and scrub oak (Quercus ilicifolia).

Under a federal recovery permit and working with private landowners in Plymouth County, nests were located in May, June, and July 2022 and protected from depredation using wire mesh cages. We recorded the date each nest was oviposited (nesting date) and recorded the date that hatchlings were first observed pipping the eggs or emerging from the eggs (emergence date). The nests were carefully excavated ≥ 55 d after oviposition and were incubated ex situ until hatching occurred.

After hatching, the headstarted turtles were raised in captivity for 9 mo (September–May) at volunteer facilities that included schools, museums, and nongovernment organizations, following a standardized protocol. In May, we weighed and PIT-tagged (Mini HTP8 Pre-Load Tray; Biomark, Boise, Idaho) each individual prior to release and adhered radio transmitters (R1680 radio transmitters; Advanced Telemetry Systems, Isanti, Minnesota) to the right posterior carapace of 10 headstarted turtles using epoxy (8277EC WaterWeld; J-B Weld, Lawrenceville, Georgia). We refer to each individual as the last 4 alphanumerics of its PIT tag identification number in this paper. Each radio transmitter weighed 10 g including the epoxy. One day prior to release, we recorded body mass (BM) to the nearest g and straight carapace length (SCL) to the nearest mm for each individual using a digital scale (Pronto Digital Kitchen Scale; Ozeri Corporation, San Diego, California) and 150 mm digital calipers (Erebus Company, Zhejiang, China). We randomly selected ten individuals (≥ 100 g) from the 2023 cohort (Table 1). The weight of the transmitter and epoxy was < 5% of the individual's body weight, to ensure the transmitter did not disrupt the individual's movement or behavior (Graeter et al. 2013). On 18 May 2023, we released 34 headstarted turtles into Pond A and 36 into Pond B. At each pond, 5 turtles were equipped with radio transmitters.

Table 1. For each individual, we report the body mass (BM) prior to release, the straight carapace length (SCL) prior to release, home range estimates that were calculated as 95% minimum convex polygons (MCP), the name of the pond, the size of the pond, and the proportion of the pond used for ten headstarted northern red-bellied cooters (Pseudemys rubriventris) in Plymouth County, Massachusetts, May–October 2022.

View Fullscreen

After release, we located individuals using radiotelemetry consisting of a radio receiver (ATS R410 Scanning Receiver; Advanced Telemetry Systems) and a rubber “H” antenna (RA-23K; VHF Telonics, Inc, Mesa, Arizona). We conducted radiotelemetry surveys every 7 d from 18 May to 28 September, at which point the turtles were located biweekly until 26 October (161 d) when turtle activity became limited. We conducted radiotelemetry surveys from the shore during the first 2 wk of the study to determine whether the headstarted turtles were still occupying the pond and began conducting radio telemetry via kayak 3 wk after release (starting 8 June) to collect more precise location data. We recorded the exact location of headstarted turtles if we were able to visually observe them and approximate locations if the headstarted turtles were submerged underwater. Approximate locations were recorded where the signal from the receiver was at the lowest gain possible and direction was still discernible. We recorded locations with a handheld GPS (Garmin GPSMap70; Garmin Ltd., Olathe, Kansas) and Google Earth Mobile (Google 10.41.0.6; Google, Mountain View, California). We used binoculars to view the headstarted turtles from afar and recaptured them using a handnet. Each time we located an individual, we recorded whether it was visually observed, its behavior (submerged in water but not swimming, swimming, unknown, basking, quiescent on bottom, and submerged in vegetation), and whether it was found in aquatic or terrestrial habitat.

We conducted all statistical analyses using R (R Core Team, version 4.3.2). We calculated home ranges for each individual that was observed > 5 times using 95% minimum convex polygons (MCPs) using the R package “AdehabitatHR” (Calenge 2006). We excluded the data collected from the shore during the first 2 wk of the study to improve the precision of MCPs. We measured the surface area of both ponds using the measurement tool in Google Earth (Google Earth Pro version 7.3; Google) and delineated the edge of the pond where the wetland vegetation met the uplands. We calculated a space use ratio by dividing the home range estimates for each individual by the surface area of the pond(s) in which they occurred.

We used the nonparametric Kaplan-Meier survival analysis (Kaplan and Meier 1958) using the R package “survival” (Therneau 2022) to calculate the probabilities of survival and nondispersal using our radio-telemetry data (Altobelli et al. 2021). We censored headstarted turtles from our analysis once their radio transmitters detached or the individual died, or when dispersal occurred. To quantify all the possible outcomes for survival and nondispersal, we modeled 3 scenarios using the Kaplan-Meier analysis. In scenario 1, we calculated the probability of survival for all headstarted turtles and considered turtles that were confirmed to be alive as alive, considered any that were confirmed to be dead as dead, and considered any turtles whose fate was unknown as alive. In scenario 2, we calculated the probability of survival for all headstarted turtles and considered turtles that were confirmed to be alive as alive, considered any that were confirmed to be dead as dead, and considered any turtles whose fate was unknown as dead. In scenario 3, we calculated the probability of nondispersal and considered any headstarted turtle whose fate was unknown to have dispersed.

Results. —

We caged 25 northern red-bellied cooter nests throughout May, June, and July 2022. The nesting dates ranged from 31 May to 14 July (median 21 June). The nest emergence dates ranged from 20 August to 9 September (median 30 August). The BM of the headstarted turtles that had radio transmitters adhered to them ranged from 197 to 402 g (mean 310; SD 76.40), and the SCL ranged from 105 to 140 (mean 122.60; SD 11.19; Table 1).

We recorded a total of 181 locations of headstarted turtles, ranging from 2 to 20 (mean 15.7; SD 6.5) locations per individual. Overall, we visually located the headstarted turtles in 39% (n = 70) of observations and were unable to visually locate headstarted turtles in 61% (n = 111) of observations. The most frequently observed behavior was submerged in water (39%, n = 71), followed by swimming (35%, n = 64), unknown (17%, n = 30), basking (7%, n = 12), quiescent on bottom (2%, n = 3), and submerged in vegetation (0.6%, n = 1). We did not record any instances of terrestrial habitat use.

We calculated home range size estimates (MCPs) for 9 headstarted turtles: 5 in Pond A and 4 in Pond B. Home range size estimations ranged from 0.54 to 3.35 ha (Table 1), with an average home range size of 1.98 ha (SD 0.82). The pond space use ratio ranged from 0.12 to 0.52 (mean 0.37, SD 0.14; Table 1).

On 2 June (15 d after release), headstarted turtle no. 553E was found floating dead due to unknown causes along the edge of Pond B. On 29 June (42 d after release), the radio transmitter detached from no. 5535 in Pond A. On 24 August (98 d after release), the radio transmitter detached from headstarted turtle no. 5536 in Pond A. On 7 September (112 d after release), we lost the signal from the radio transmitter attached to headstarted turtle no. 552E in Pond A, and it is unclear whether it dispersed or died, or if the radio transmitter malfunctioned. The 6 remaining headstarted turtles survived to the beginning of the brumation period on 26 October. During the spring of 2024, we radio tracked the 6 remaining headstarted turtles, recording approximate locations, and they had moved from their overwintering locations and were assumed to be alive, though we have not recaptured them.

Overall, 8 of 10 headstarted turtles survived into October 2023, and 6 of 10 likely survived until May 2024. Using our survival data for Kaplan-Meier Analysis, in scenario 1, we estimate the probability of survival in Pond A to be 1.00 throughout the activity season (18 May–26 October), and in Pond B we estimate the probability of survival to be 1.00 from 1 to 15 d, then it drops to 0.987 and remains at 0.987 until the end of the study. In scenario 2, the probability of survival in Pond A is 1.00 until 112 d, then it drops to 0.929 and remains at 0.929 until the end of the study. In scenario 2, the probability of survival in Pond B is 1.00 from 1 to 15 d and then it drops to 0.987 and remains at 0.987 until the end of the study. In scenario 3, the probability of nondispersal in Pond A is 1.00 until 112 d, when it drops to 0.929 and remains at 0.929 until the end of the study. In scenario 3, the probability of nondispersal in Pond B is 1.00 throughout the study.

Discussion. —

The low frequency of visual observations and high frequency of “submerged in water” behaviors are likely attributed to shyness toward humans, or conversely, the turtles are behaving normally. We observed many of the headstarted turtles basking on logs while standing hidden among the forested edge of the pond; however, the headstarted turtles would retreat from their basking locations when we were not hidden. We frequently observed the headstarted turtles swimming to the water’s surface, and upon encounter, they would swim down and conceal themselves. We made very few recaptures of the headstarted turtles using the handnet. Several studies have expressed concerns that headstarting may alter the natural behavior of the turtles and may result in human-habituated turtles (Seigel and Dodd 2000; Cann et al. 2021); however, habituation did not appear to occur in this population. Other studies have reported that human disturbance can cause freshwater turtles to bask less often and have documented these impacts to their physiology (Heppard and Buchholz 2019; Berberi et al. 2024); therefore this species may benefit from being released in areas of low human disturbance.

All of the headstarted turtles had home range estimates that were ≤ 52% of the available pond surface area, and the 95% MCP polygons were not all concentrated in the same area. These results might indicate that some areas of the pond had more resources for headstarted turtles to utilize (e.g., food, shelter) than other areas. However, we radio-tracked the headstarted turtles every 7 d, and they may have moved to areas outside of their estimated home ranges when we were not present. Future studies should investigate the home range of headstarted turtles for longer durations and record locations more frequently to allow for a better understanding of important habitat features (e.g., pond depths, patches of aquatic vegetation) and for comparison over time.

It is unknown what caused the death of headstart turtle no. 553E, and we did not visually observe this turtle until it was found dead. We suggest that future studies radio-track headstarted turtles frequently during the first month following release to allow for a better understanding of survival during this acclimation period. Two of our radio transmitters detached from headstarted turtles during this study, possibly due to water corrupting the bond of the epoxy to the carapace, the headstarted turtles knocking the radio off incidentally on submerged structures, or the epoxy not being mixed homogeneously. Future studies may benefit from using a different brand or type of epoxy for freshwater turtles than what we used. We lost the radio signal from headstarted turtle no. 552E, and the fate of this individual is unknown. There are several possibilities for the fate of this headstarted turtle: 1) it was preyed upon by an aerial predator such as bald eagle (Haliaeetus leucocephalus) or great blue heron (Ardea herodias) and was removed from the study location (Neimela and Bury 2012; Coleman 2023); 2) it was collected by a human (although unlikely due to how difficult it was for the authors to recapture individuals); 3) it dispersed over a long distance and moved beyond the range of our receiver (although we attempted to radio-track this headstart on the trails surrounding this pond and visited every nearby pond within a few km of Pond A and did not detect any signal); or 4) the radio transmitter’s battery died and it is still in Pond A and we were unable to receive a signal from it.

The actual probability of survival of the headstarted turtles was very high in both ponds, and modeled probability of survival in scenarios 1 and 2 was also high. This result indicates that the overall probability of survival is likely high among headstarted northern red-bellied cooters from the release date to the time of first overwintering. Previous studies have investigated survival rates in headstarted northern red-bellied cooters in Massachusetts and found that survival was lower during the first year following release and increased thereafter (Haskell 1993; Haskell et al. 1996). Our results may not be representative of all headstarted northern red-bellied cooters weight classes because we randomly selected individuals for radio telemetry monitoring only if they were ≥ 100 g. We suggest that future studies of northern red-bellied cooter survival in Massachusetts conduct radio telemetry for longer periods so that survival can be assessed in relation to a broader time scale.

We did not document any instances of headstarted turtles exiting the release pond and dispersing to other sites, although it is unclear whether no. 552E dispersed or not. Similarly, Haskell et al. (1996) did not document any movement of headstarted northern red-bellied cooters between ponds. The overall probability of nondispersal is very high, likely indicating that northern red-bellied cooters remain at the release site from the release date to the beginning of the first overwintering period.

In conclusion, we found that estimated home ranges for headstarted northern red-bellied cooters were ≤ 52% of the pond habitat available to them. The survival rate was high at both sites and likely indicates that the majority of headstarted turtles of this species survive to their first brumation period in Massachusetts. The probability of nondispersal was very high, which likely indicates that the majority of headstarted turtles of this species remain in the ponds in which they are released during the first year in the wild. We suggest that future radio-telemetry studies of headstarted northern red-bellied cooters in Massachusetts be conducted over a longer period to allow for a better understanding of how home range size, survival, and dispersal changes over time.