North American box turtles (Terrapene spp.) occur in a variety of habitats throughout much of the United States (Ernst and Lovich 2009). This range is inclusive of several recognized species and subspecies that occupy a variety of different climactic regions (Martin et al. 2013). This expansive range likely impacts reproductive phenology, frequency, and output. For example, most known T. carolina populations in northern latitudes show distinct nesting seasons beginning in the late boreal spring and ceasing at the end of the summer months (Dodd 2001). Courtship and mating in these populations occurs primarily in the fall and to a lesser extent the spring (Ernst and Lovich 2009). However, reproductive phenology is more protracted in many species and populations of turtles in warmer climates compared with those in cooler climates (Iverson 1977; Jackson 1988).

The Florida box turtle (Terrapene bauri, considered as a subspecies of T. carolina by many) ranges from the Florida Keys to southern Georgia (Martin et al. 2020; Rhodin et al. 2021). Given this latitudinal range, T. bauri is exposed to a variety of different climactic regimes that likely impact its reproductive phenology. For example, on Egmont Key (Hillsborough County, Florida), populations show a similar but extended reproductive period compared with more northern congeners, with gravid females detected from March into August (Dodd 1997). Further, mating has been observed in essentially any period in which active turtles could be found at this site, likely influenced by the milder coastal climate of the region (Dodd 2001). Near the southern extent of their range in Everglades National Park, Meshaka and Layne (2015) detected oviducal eggs a month earlier in February, indicating further latitudinal extension of the nesting season.

In southwestern Florida, very little is known of the ecology of T. bauri other than information on habitat use and population size (Jones et al. 2016). The mild winters and higher annual temperatures of the southern Florida region may further influence the activity patterns and phenology of T. bauri (Kuchling 1999a). Thus, a better understanding of the variation in life history traits and reproductive timing of this species in this part of its range is imperative not only for modeling reproductive biology and influencing conservation action but also to serve as baseline data for comparisons as anthropogenic climate change continues to impact organisms and ecosystems. In the present study, we investigated potential extended reproduction via observations of courtship and mating, the presence of oviducal eggs, clutch frequency, clutch size, and egg size over a calendar year in a wild population from southwestern Florida.

METHODS

Field Site. — To protect the study population from potential illegal take for the wild animal trade, we suppress detailed descriptions of location and habitat. The site is a former recipient site for gopher tortoises (Gopherus polyphemus) in inland Collier County, Florida. The site is approximately 5 ha in size and consists of 3 major habitats: oak scrub (Quercus virginiana and Ceratiola ericoides), slash pine flatwoods (Pinus elliottii and Serenoa repens), and dry prairie (Andropogon sp. and Opuntia sp.). A variety of high-value forage plants exist at the site, including cocoplum (Chrysobalanus icaco), American beautyberry (Callicarpa americana), Mexican clover (Richardia brasiliensis), gopher apple (Licania michauxii), and various sources of invertebrate prey.

Data Collection. — Beginning in March 2019, a mark–recapture project on Florida box turtles was initiated at the site and supplemented by a radiotelemetry study in November 2019. Five males and 5 females were tagged and monitored for ∼ 13 mo. While tracking and surveying for these animals, we also documented reproductive behaviors such as courting, mating, and nesting. Starting in July 2020, all 5 radio-tagged female turtles were typically collected for monthly radiographs within the last 7–10 d of each month to determine presence of eggs. Average time between sampling periods was 30.7 ± 5.2 d (Table 1). This time frame fits the documented internesting period in the related T. carolina triunguis (19.1 ± 0.3 d; Messinger and Patton 1995). Any additional nontagged adult females that were encountered were collected for radiographs opportunistically. In December 2020, tags were removed from all males, and tags on females were replaced. A sixth female was also radio-tagged at this time to increase monthly sample size. Both radiographs and behavioral observations continued until August 2021, when tags were permanently removed.

Table 1. Summary of female body size, clutch frequency, and clutch size in 6 free-ranging Florida box turtles from southwestern Florida from July 2020 to August 2021. Consecutively observed clutches are shown in bold. ND = no eggs detected; — = lack of data.

View Fullscreen

Animals were transported to the Humane Society Naples Veterinary Clinic and radiographed in dorsoventral recumbency at settings of 77 kVp and 5.6 mAs. A US quarter (24-mm diameter) was placed on each radiograph for scale to allow for minimum egg length measurements. All measurements were performed post hoc in ImageJ (Schneider et al. 2012) and estimated to the nearest 0.1 mm as seen in other wildlife morphometric studies (Erdmann et al. 2018; Carrasco-Harris et al. 2020). All data summaries are reported as mean ± standard deviation. Animals were captured and held from 2 to 3 hrs for processing before being released at their initial capture locations.

RESULTS

We observed 20 courtship or mating events from April 2019 to August 2021. Behaviors ranged from mounting attempts to copulation. Most observations occurred from October to November (n = 7), although courtship behaviors or mating attempts were detected in all months except January and May (Table 2). The sole nesting event was observed on 11 April 2019. This female (#1) was observed depositing a clutch of 5 eggs into a nest chamber and covering them. These eggs were carefully excavated, measured, and weighed by hand before being returned to the nest chamber and backfilled. Egg weights ranged from 6 to 10 g, and egg lengths (measured by handheld vernier calipers) ranged from 30 to 34 mm. For consistency, these measurements were not included in the other data summaries where eggs were measured by radiographs.

Table 2. Observed reproductive behaviors and status in Florida box turtles from southwestern Florida. Y = yes; N = no.

View Fullscreen

Eight turtles were radiographed 84 times from July 2020 to August 2021. Radiographed turtles averaged 140.7 ± 5.7 mm straight carapace length. Twenty-eight observations of gravidity in 7 individuals occurred over the 14-mo examination period. Eggs were detected in July and August 2020, absent from September to December 2020, and again detected from January to August 2021, for a total of 8 consecutive months (Table 1). One female (#1) had a clutch of 2 observable eggs in June 2021; because 1 egg was abnormally small (< 10 mm) and may have been a mineralized follicle (Edmonds et al. 2020), it was excluded from summaries. Another female (#5) was captured only opportunistically throughout the study (March, July, and August 2021) and was found to be gravid a single time (March 2021) with a clutch of 4 eggs. The data from this female were included in overall population data for mean egg length and clutch size but were omitted from other analyses in which females were radiographed regularly.

Clutch frequency varied by individual, with some turtles appearing to produce distinct clutches in consecutive months, while others delayed additional clutches for multiple months (Table 1). Annual clutch number per calendar year ranged from 3 to 6; average clutch number for the 2021 season (January–August) was 3.5 ± 1.1 (n = 2–4). Average clutch size per turtle from 2020 to 2021 was 3.0 ± 0.9 eggs (n = 1–4). Average egg length was 33.6 ± 3.6 mm with a range of 24.8–42.1 mm.

DISCUSSION

Our study revealed an extended period of reproduction in T. bauri in southwestern Florida. Specifically, we found earlier timing of female reproduction and a more protracted reproductive season in females than found in other populations of T. bauri and in the closely related T. carolina. In addition, regular sampling allowed us to estimate clutch frequency and internesting intervals in the population. Our study contributes to a more complete understanding of the reproductive ecology of box turtles.

Courtship and mating attempts were documented in 10 mo of the year (all except May and January), with peak observations occurring in October–November. This corresponds to similar observed mating periods in the T. carolina clade across its range (Dolbeer 1969; Currylow et al. 2013b). Observations from November, December, and February most closely match Dodd's (1997) coastal central Florida observations of opportunistic breeding attempts during any active period throughout the year. We may simply have not observed breeding attempts that occurred in May and January, although they may have been reduced by climatic conditions. For example, the average temperature in January 2021 for the region was 18.2°C, with some mornings dropping lower than 10°C, while average temperature in May 2021 was 26.7°C with an accompanying drought period (25.7 mm of rainfall; US National Weather Service 1995). Cooler mornings may have limited turtle activity during most surveys in January, as this species appears to be most active when temperatures exceed 17°C (Dodd et al. 1994), while dry conditions in May potentially decreased the activity of animals, as Florida box turtles may be particularly sensitive to lower humidity levels (Bogert and Cowles 1947; Dodd et al. 1994).

One of our females exhibited shelled eggs on 27 January. To our knowledge, this is the earliest date for shelled eggs in wild box turtles; the previous earliest date was in February at a similar latitude (Everglades National Park; Meshaka and Layne 2015). A gravid female was documented with eggs in October in a Virginia population of T. carolina (Mitchell and de Sá 1994); presumably, this was a late or partially retained summer clutch as opposed to the beginning of a subsequent nesting season.

In the present study, only 4 mo (September–December 2020) without detectable eggs were documented. These observations are indicative of an extended reproductive period, similar to those observed by Iverson (1977) and Jackson (1988) in various other turtles in Florida. Dodd (1997) never observed gravid females or eggs prior to March on Egmont Key, while Meshaka and Layne (2015) observed gravid females from February to August (Everglades National Park) but suggested that further extension could be possible based on follicle size and presence. As latitude increases, the nesting period is generally contracted in T. bauri and other congeners (Jackson 1991; Wilson and Ernst 2005). At the same time that congeners may be minimally active during hibernation in more northern regions (e.g., Currylow et al. 2013a), the box turtles in our study population are regularly active and thus may be able to acquire energetic resources required to support reproduction.

Turtles at our site produced between 2 and 4 clutches in a single nesting season (Fig. 1). Multiple clutches in the T. carolina clade are well known (Stuart and Miller 1987; Jackson 1991; Buchman et al. 2010), and similar observations of clutch number in T. bauri specifically were documented by Meshaka and Layne (2015). However, frequency of clutches is not as well understood in wild populations of Terrapene. Repeated sampling afforded by radiotelemetry allowed us to estimate the amount of time between clutches. The period of time between subsequent clutches varied between individuals, with some individuals exhibiting potential clutches in 3 consecutive months (20–28 d apart), whereas others appeared to space clutches between 40 and 98 d apart (Table 1). Although turtles in the genus Terrapene (Nieuwolt-Dacanay 1997) are able to retain shelled eggs for periods of 8–50 d if environmental conditions are not optimal, we do not believe this to be the case in most instances in our study. Specifically, we suggest that egg number and general position in the oviduct may support distinct clutches in consecutive months. Outside of species in the softshell genus Lissemys, most turtles to our knowledge appear to simultaneously shell all ovulated eggs at the same time (Mahmoud and Licht 1997; Kuchling 1999b); thus, Florida box turtles documented in subsequent months with clutch sizes larger than in the preceding month are likely showing independent clutches. The desert tortoise species Gopherus morafkai exhibits egg retention between years due to extreme environmental conditions, leading to secondary clutches undergoing ovulation and shelling simultaneously while the earlier clutch is still retained (Lovich et al. 2017). Thus, it could be possible for multiple clutches to be present simultaneously in other species facing similar conditions. However, given the more benign conditions and time frame between clutches documented in our population, we find this to be an unlikely scenario in T. bauri. Some turtle species are known to deposit multiple partial clutches when disturbed (sea turtles, Miller 1997; Cuora mouhotii, Ji-Chao et al. 2011); thus, some of the T. bauri clutches that decreased in number in consecutive months could have been partially retained from a failed or abandoned nesting attempts. Hypothetically, the position of the eggs within the oviducts could change over days to weeks. However, in the present study, in some instances, varying numbers of eggs were on complete opposite sides of the animal on radiographic examination. For example, 1 turtle (#1) showed 1 egg in its left oviduct and 2 in its right oviduct on 10 July 2020, while a designated subsequent clutch on 20 August 2020 showed 2 eggs in the left oviduct and a single egg on the right oviduct (Fig. 2). At least some turtle species alternate the ovaries used for ovulation between consecutive clutches (Dobie 1971; Etches and Petitte 1990). This supports the notion that the eggs we observed across successive months were distinct clutches. It is also possible that some subsequent clutches were missed due to the monthly sample schedule, with Dodd (2001) indicating that underestimates of reproductive potential are likely across studies. Under-counting clutches is a possibility in this study, as more than 20 d occurred between samples, during which potential ovulation, shelling, and oviposition could occur (Messinger and Patton 1995). Specifically, Kuchling (1999b) indicates that ovulation may occur in as little as 12–48 hrs, with egg shell deposition occurring between 48 and 60 hrs postovulation in the Australian species Pseudemydura umbrina, while Mahmoud and Licht (1997) observed similar ovulation periods in snapping turtles (Chelydra seprentina). In the present study, we intentionally limited radiographs to a maximum of 1 per month to reduce stress on the animals.

View Figure

• Download as Powerpoint
• Download Figure

View Figure

• Download as Powerpoint
• Download Figure

As with many emydid species, most females in the present study exhibited annual reproduction. In the first 2 mo of the study (July–August 2020), 5 of 6 radiographed turtles had eggs in 1 or both months. Subsequently, all 5 produced 2 or more clutches in the 2021 nesting season from January to August 2021. Although many species may be expected to show annual clutching, the ability to do so is highly variable between box turtle populations (Dodd 2001) and is likely dependent on resource availability (Nieuwolt-Dacanay 1997). At least 1 individual (#1) nested in 3 consecutive years (2019–2021). At our field site, food sources are abundant throughout much of the year, with invertebrates, carrion, berries, mushrooms, leafy plants (Speer, unpubl. data), and rough forage plants (Rodriguez et al. 2021) all documented as part of turtle diets. In addition to plentiful food, warm annual temperatures (annual average of 25.3°C from 2019 to 2021; US National Weather Service 1995) allow for extended activity (Dodd et al. 1994; Currylow et al. 2012) and continued acquisition of resources, which could be allocated to repeated clutches within a season as well as in subsequent seasons.

Clutch sizes in this study (1–5 eggs) were consistent with observations by Dodd (1997), Farrell et al. (2006), and Meshaka and Layne (2015). Minimum egg length was similar but smaller on average compared with those reported by Meshaka and Layne, averaging 33.6 ± 3.6 mm compared with 35.8 ± 2.2 mm in their report. However, this is likely due to variations in measurement techniques (estimated radiographs vs. direct measurement of oviducal eggs). The largest egg observed (42.1 mm) may be one of the largest documented in a wild Florida box turtle, although direct comparisons may not be possible due to the aforementioned variation in measurement procedures across studies.

Although the extent may vary at the population level, our data, coupled with those of Meshaka and Layne (2015), suggest that extended reproduction is typical of Florida box turtles at southern latitudes. The current study describes the earliest documented wild gravid female Florida box turtle and also provides more information on clutch frequency and size in wild populations. Moreover, it is important to document basic life history characteristics in potentially temperature-sensitive organisms that may face an ever-warming world. Some species have already been shown to display reproductive output changes over multiple years of exposure to warming climates. Species like gopher tortoises exhibit increased reproductive output under these conditions (Hunter et al. 2021), while others, such as the common snapping turtles (Chelydra serpentina), have shown variable climate-induced impacts on both egg size and clutch size (Hedrick et al. 2018). Such basic life history data are imperative for assessing physiological tolerances in response to anthropogenic climate change, threat of habitat destruction, and overcollection for the illegal wildlife trade plaguing turtle species throughout their ranges (Stanford et al. 2020). Furthermore, these data can and will serve a critical role in assessing the continuing impact of anthropogenic threats on sensitive organisms, providing support for future conservation legislation.