Divergent Size-Class Distributions of Gopher Tortoise Burrows in Scrub and Flatwoods Habitats of Peninsular Florida

The gopher tortoise (Gopherus polyphemus) is in decline throughout its range in the southeastern United States, due primarily to habitat loss and degradation (Auffenberg and Franz 1982; Van Lear et al. 2005; Mushinsky et al. 2006; US Fish and Wildlife Service 2011). Although tortoises occupy a variety of habitats, most studies of their ecology have been conducted in sandhill communities characterized by longleaf pine (Pinus palustris) and wiregrass (Aristida stricta) (Auffenberg and Franz 1982; Diemer 1986; MacDonald and Mushinsky 1988; Tuberville and Dorcas 2001). This community type, considered the archetypal habitat of the species (Franz and Quitmyer 2005), is largely absent from the southern half of peninsular Florida (Myers 1990). In this part of the range, gopher tortoises primarily occupy mesic flatwoods and human-modified habitats on flatwoods soils (e.g., improved pasture), which dominate southern Florida outside the Everglades. They also occupy more-restricted areas of Florida scrub habitats that largely replace sandhill communities on xeric sand ridges in central and south-central Florida (Abrahamson and Hartnett 1990; Myers 1990; Florida Fish and Wildlife Conservation Commission, unpubl. database, 2022). The humid, subtropical climate in the region also differs from other parts of the species' range, with year-round warm temperatures and a marked summer wet season (Chen and Gerber 1990). Together these soil, vegetation, and climatic characteristics create ecological conditions in the region that differ substantially from other parts of the gopher tortoise range.

Gopher tortoise habitat requirements include open-canopy conditions with ground-level sunlight for thermoregulation and nest incubation, herbaceous groundcover for forage, and friable soils appropriate for burrow excavation and nesting (Auffenberg and Franz 1982; Diemer 1986). Although fire-maintained scrub and flatwoods generally have suitable open-canopy conditions, other relevant features appear suboptimal (Castellón et al. 2012). In mesic flatwoods, high-quality forage is abundant (Abrahamson and Hartnett 1990; Castellón et al. 2012), but the poorly drained soils are ill-suited for burrowing and nesting. In Florida scrub, the xeric soils are suitable but herbaceous forage is extremely sparse (Myers 1990; Whitney et al. 2004; Castellón et al. 2012). Nonetheless, these habitats support populations of tortoises in southern Florida, albeit at lower densities than are typical of sandhill habitats (Castellón et al. 2012; Smith and Howze 2016; Florida Fish and Wildlife Conservation Commission, unpubl. database, 2022). In our study area in southcentral Florida, densities of noncollapsed burrows were relatively high within local populations (8.64–10.34 burrows/ha in flatwoods and 7.04–8.78/ha in scrub; Castellón et al. 2020). However, densities were lower at the landscape scale, especially in flatwoods (2.18/ha) relative to scrub (4.12/ha; Castellón et al. 2012; see also Goessling et al. 2021). For comparison, burrow densities in Florida sandhills reportedly range from 9.85/ha to 12.63/ha (Ashton et al. 2008; Nomani et al. 2008).

Although peninsular Florida is arguably the range-wide stronghold for the species (Smith et al. 2006), their ecology is poorly studied in the scrub and flatwoods habitats they occupy in the southern peninsula. This lack of knowledge hinders conservation decision making. For example, it is unclear whether these habitats constitute suitable recipient sites for gopher tortoise translocation (required under Florida state law when tortoise burrows are disturbed by development) and, if so, what stocking densities are appropriate. Thus, there is a need for better demographic data, especially recruitment rates, given that adult survival is typically high and demographic models suggest sensitivity to changes in survivorship of early life stages (Tuberville et al. 2014; Folt et al. 2021). However, juvenile recruitment is difficult to measure directly. Here we used burrow size-class distributions to assess population age structure within flatwoods and scrub habitats in south-central Florida. We also monitored tortoise burrows and nests with automated cameras to document nest fate and visitation of burrows by predators.

METHODS

Study Area. — Our research was conducted at Avon Park Air Force Range (APAFR), a 42,000-ha military training installation in south-central Florida (Fig. 1). Dominant upland habitats at APAFR are mesic and dry-mesic flatwoods, including slash pine (Pinus elliottii) plantations on flatwoods soils, with Florida scrub predominant on sand ridges. Management of flatwoods and plantations at APAFR at the time of our study included prescribed fire on a 2–3-yr rotation (US Air Force [USAF] 2000). Flatwoods had open canopies and dense groundcover (Abrahamson and Hartnett 1990; Castellón et al. 2012), and plantations were managed as even-aged stands, with a typical harvest age of 40–50 yrs. The plantations were also thinned periodically to maintain a basal area between 9 and 16 m²/ha (T. Meade, APAFR, pers. comm., 2011).

View Figure

• Download as Powerpoint
• Download Figure

The dominant flatwoods soil in the study area is Myakka, followed distantly by Immokalee and Basinger Sands. Soils in our burrow-mapping sites were Myakka and Smyrna. These soils are limited to peninsular Florida and are poorly to very poorly drained. The water table depth in these series is < 46 cm for several months during the rainy season in most years, with the water occasionally rising above the surface for the Smyrna series. During our study, standing water or sheet flow was often present in flatwoods sites during periods of heavy rain.

Florida scrub assemblages at APAFR included sand pine (P. clausa) scrub, oak (Quercus spp.) scrub, mixed scrub, and scrubby flatwoods (Hokit et al. 1999), which were managed with prescribed fire on a 7–20-yr rotation (USAF 2000). These assemblages occur on deep, xeric, infertile soils and are characterized by stunted, shrubby vegetation and sparse groundcover (Myers 1990). Dominant soils were Satellite, Narcoossee, and Duette. These are very deep soils, but water table depths are 25–102 cm from the surface during the rainy season for the Satellite series (National Resource Conservation Service 2019).

The study area is in the humid subtropical climate zone, with mild, dry winters and hot, wet summers. Summer rainfall is characterized by frequent afternoon thunderstorms and occasional tropical weather systems that bring heavy rain. The 30-yr mean daily temperatures were 16.7°C (low = 9.6°C) in winter, and 27.6°C (high = 33.6°C) in summer. Mean rainfall was 168.9 mm in winter and 595.6 mm in summer (National Centers for Environmental Information 2017). The first year of our burrow mapping surveys, 2011, was wetter than the subsequent survey year in 2015, with rainfall in 2011 exceeding that in 2015 during the 6 mo (+ 227 mm) and 12 mo (+ 55 mm) preceding our surveys (Station GHCND:USC00080369; http://www.ncdc.noaa.gov/cdoweb/search).

Line-Transect Distance Sampling. — In 2009, from April through October, we implemented a landscape-level line-transect distance sampling (LTDS) survey (Buckland et al. 2001) for tortoise burrows, stratified by habitat in scrub and mesic flatwoods/plantations (Fig. 1). We used the survey method of Smith et al. (2009), wherein one observer searched for burrows while walking slowly along the transect, navigating using a global positioning system with real-time submeter accuracy (Trimble GeoXT). Two other members of the survey team searched adjacent to the transect (one on each side), meandering in sigmoid paths to cover the area within 20 m on each side of the transect.

During this survey, designed to estimate burrow detection probabilities and abundances (see Castellón et al. 2012 for detail), we also collected data on burrow sizes by measuring the diameter of noncollapsed burrows at a depth of 50 cm from the mouths using burrow calipers. Burrows that were too small to measure this way were measured at the entrance with a measuring tape. We then categorized burrows as belonging to juvenile < 14 cm), subadult (14–23 cm), or adult (> 23 cm) size classes (Smith 1995). Because carapace length (CL) is related to age in gopher tortoises (Landers et al. 1982), and burrow width is strongly correlated with CL (Alford 1980; Martin and Layne 1987), we were able to generate approximate demographic profiles for each habitat based on burrow widths (Landers et al. 1982; Smith 1995).

Burrow Mapping. — In addition to our LTDS surveys, as part of a broader research effort (Rothermel and Castellón 2014; Castellón et al. 2018, 2020), we also surveyed, mapped, and measured all detected burrows in 6 study sites during summer months (Fig. 1). We initially surveyed 1 flatwoods site (FW1) and 1 scrub site (SC1) in 2011. In 2015, we resurveyed these 2 sites and conducted first-time surveys in 1 flatwoods site (FW2), 1 scrub site (SC2), and 1 mixed-habitat site (approximately half flatwoods and half scrub; MX1); then in 2016, we surveyed 1 additional mixed habitat site (MX2). Site selection was limited to areas of APAFR where Air Force training activities would not interrupt access.

The boundaries of these study sites were defined based on the natural limits of the burrow aggregations following McCoy and Mushinsky (2007). To do this, we expanded the survey area outward while searching for burrows until no more active or potentially active burrows were found within a 100-m radius of the outermost active or potentially active burrows (excluding those that were collapsed, structurally degraded, or with significant accumulations of debris at the mouth). Burrow searches were conducted by 3–5 people walking side-by-side along parallel transects spaced 5 m apart. A previous analysis of our LTDS data (using only burrows detected from the transect lines; i.e., excluding those detected by observers searching adjacent to the line) provided an effective strip width (ESW) estimate of 4.38 m, regardless of habitat (Castellón et al. 2020; Supplement A available at https://static1.squarespace.com/static/570d1ea37da24f381ca53c95/t/5e2deab94914ae3cb43dbd1b/1580067513880/CastellonSupplementA.pdf), which can be interpreted as perfect detection to this distance on either side of a transect. Therefore, our 5-m transect spacing provided double coverage over most of the study area, except for areas within 1 m of the transects where detection was near-perfect. Because the LTDS model used to estimate the ESW lacked power to discriminate detection probabilities based on burrow size, we cannot rule out the possibility that some small juvenile-sized burrows were missed in our surveys; however, the double-coverage of the survey area by the closely spaced observers should have ensured that most were detected.

We compared proportional distributions of burrow size classes (excluding collapsed burrows) among all 6 sites surveyed in 2015 and 2016 using a generalized linear model with a multinomial distribution (logit link function) and habitat type as a fixed factor. For the 2 sites that were surveyed in both 2011 and 2015 (FW1 and SC1), we compared size-class distributions between sites in both years and between years for individual sites using chi-square tests of independence.

Nest and Burrow Monitoring. — We recorded data on any nests we encountered opportunistically (e.g., through visual detection of exposed eggs) during burrow mapping and other monitoring activities in 2011. Then, from 13 May to 24 June 2015, we actively searched for nests in burrow aprons at our study sites by probing the sand with a metal wire (after Smith 1995). We avoided more-aggressive techniques (e.g., using spades) due to presence of unexploded ordnance at APAFR. We initially searched aprons of active burrows only but began searching all identifiable burrow aprons after we opportunistically found a nest at a collapsed burrow. Upon finding a nest, we deployed an automated camera (Bushnell NatureView, Model 119438) that was programmed to take 3 consecutive photographs each time it was triggered by motion. Each nest was monitored > 110 d to cover the entire incubation period. We recorded predation and visitation by potential predators as well as events such as flooding that could influence nest success. At the end of monitoring, we excavated the nests to check for unhatched eggs or hatchlings.

From 25 October to 29 November 2016, we monitored a subsample of burrows in each study site using automated cameras. Given the season, it was unlikely that nests were present, so the aim was to assess general habitat-specific predation risk by quantifying visitation by vertebrates known to depredate tortoise nests or juveniles. All monitored burrows appeared active or possibly active and were > 15 cm wide. The cameras took series of 3 photographs at 5-min intervals and each time they were triggered by motion. We defined independent observations of potential predators (per species) at each burrow as those separated by > 24 hrs. Therefore, a given predator species was recorded as detected no more than once per 24-hr period (camera-day) at each burrow. We used a t-test to compare mean numbers of predator detections per burrow, per day, between habitats. All statistical analyses were conducted using SPSS 22.0.0.0 (IBM SPSS, Armonk, NY).

RESULTS

Burrow size-class distributions derived from our 2009 LTDS survey were strongly skewed toward adult sizes, especially in flatwoods, where 82% of noncollapsed burrows were adult-sized. Subadult- and juvenile-sized burrows represented 15% and 3%, respectively (Fig. 2). In scrub communities, adult-, subadult-, and juvenile-sized burrows represented 74%, 15%, and 11% of burrow abundance, respectively. In the 6 sites where we mapped burrows in 2015 and 2016, size-class distributions differed significantly among habitats (ref: Juvenile/Flatwoods, β = –0.974, SE = 0.396, Wald χ² = 6.05, p = 0.014; Figs. 3 and 4). Again, in flatwoods sites the distributions were strongly skewed toward adult size classes, with very few juvenile-sized burrows (≤ 3%). Distributions were also skewed in scrub sites, but dominance by adult-sized burrows was less extreme. Distributions in mixed-habitat sites were more like those in flatwoods, although there was a higher percentage of juvenile-sized burrows in MX2. Comparisons between the 2 sites surveyed in 2011 (FW1 and SC1) showed an even stronger difference in burrow size-class distributions between habitats (χ² = 40.32, p < 0.001), with a much more strongly skewed distribution in FW1. In the flatwoods site, there was no significant difference in distributions between years (2011 and 2015; χ² = 2.07, p = 0.356), but there was a significant shift between years in the scrub site (χ² = 16.84, p < 0.001), with fewer juvenile- and more subadult-sized burrows in 2015 (Figs. 3 and 4).

View Figure

• Download as Powerpoint
• Download Figure

View Figure

• Download as Powerpoint
• Download Figure

View Figure

• Download as Powerpoint
• Download Figure

Despite spending > 400 person-hours repeatedly searching 249 burrow aprons in the spring of 2015, we detected only 3 nests, 1 in flatwoods (FW1) and 2 in scrub (MX2). Camera monitoring of the flatwoods nest, from late May through mid-September, documented repeated visits by 1 or more eastern spotted skunks (Spilogale putorius) and 2 periods of flooding. Flooding of the nest first occurred on 29 August and lasted 1 wk. The second flooding occurrence on 5 September was 1 d after the water receded from the first flooding and lasted more than a month. At the end of monitoring, we excavated the intact nest and found 4 unhatched eggs. Upon opening 1 egg we found a relatively well-developed embryo. Over the course of fieldwork in 2011, we opportunistically encountered 2 nests in the FW1 site, 1 of which was already depredated by a mammal when it was found. Both nests were flooded repeatedly over the course of the summer and the nondepredated nest failed to hatch despite remaining intact.

The 2 nests in scrub habitat were monitored with automated cameras from early June through late September. One nest, found with 1 egg partially exposed, was visited repeatedly by 1 or more nine-banded armadillos (Dasypus novemcinctus) that sometimes dug around the nest and may have consumed 1 egg before the nest was completely depredated by a striped skunk (Mephitis mephitis) on the 44th day of monitoring. Continued monitoring of the depredated nest also documented visits by a feral hog (Sus scrofa) and a raccoon (Procyon lotor). The other nest was found buried at a depth of 4 cm in the apron of a collapsed burrow. This nest was visited by only 1 potential predator, a striped skunk, but failed to hatch despite remaining intact. When we excavated the nest, we found 2 dead but partially emerged hatchlings and 2 unhatched eggs. Upon opening an egg, we found a relatively well-developed embryo.

To further evaluate visitation by predators, we deployed automated cameras at 67 burrows in flatwoods for a mean duration of 11.78 camera-days per burrow (± 3.31 standard deviation [SD]) and at 44 burrows in scrub for a mean of 11.61 camera-days per burrow (± 2.45 SD). There was no significant difference between habitats in the daily rate of visitation by predators (t = 1.36, df = 109, p = 0.175). Burrows in flatwoods were visited at a rate of 0.17/d (± 0.21 SD) compared with 0.12/d (± 0.16 SD) in scrub. Potential predators detected at flatwoods burrows were eastern spotted skunks (n = 43), southern black racers (Coluber constrictor priapus; n = 28), nine-banded armadillos (n = 23), eastern coachwhips (Masticophis flagellum flagellum; n = 13), Virginia opossums (Didelphis virginiana; n = 6), bobcats (Lynx rufus; n = 2), 1 domestic dog (Canis lupus familiaris), 1 feral cat (Felis catus), and 1 striped skunk. Potential predators at scrub burrows were eastern spotted skunks (n = 28), southern black racers (n = 14), eastern coachwhips (n = 8), nine-banded armadillos (n = 7), eastern corn snakes (Pantherophis guttatus; n = 3), feral hogs (n = 2), and 1 eastern indigo snake (Drymarchon couperi).

DISCUSSION

The burrow size-class distributions we observed at APAFR were strongly skewed toward adult sizes (Figs. 2–4). Gopher tortoise life history, characterized by low fecundity and poor survival of early life stages, coupled with slowed growth following maturity and long adult lifespans, can lead to dominance of populations by adult tortoises (Tuberville et al. 2014). However, the aforementioned life history characteristics also make the species vulnerable to demographic perturbations (Iverson 1980; Brooks et al. 1991; Congdon et al. 1994; Tuberville et al. 2014), and highly skewed size-class distributions can indicate lack of recruitment (Alford 1980; McCoy et al. 2006; McCoy and Mushinsky 2007; Tuberville et al. 2014; Folt et al. 2021).

Interpretation of burrow size-class distributions, as opposed to sizes of the tortoises themselves, can be complicated by some potential sources of bias. Most importantly, small burrows are more difficult to detect than larger ones (Diemer 1992a; Smith et al. 2009), although our double-coverage surveys in the burrow mapping sites should have ensured detection of most burrows, irrespective of size. Comparisons could also be biased if detection differs by habitat, although we observed no such difference among adult-sized burrows (Castellón et al. 2020; Supplement A). Furthermore, juveniles may use fewer burrows than do adults because they sometimes use “pallets” (shallow excavations that are typically well hidden) or hide aboveground in leaf litter (Douglass and Layne 1978; Diemer 1992b; Pike 2006). It is also possible that small burrows collapse and become unrecognizable more quickly than the larger burrows (Castellón, pers. obs., 2015; Guyer and Hermann 1997). To better address nondetection of juvenile-sized burrows, for future surveys we recommend Gaya's (2019) double-observer LTDS protocol, which increased detection of juvenile burrows and overall estimate precision in her trial of the method. Surveys conducted immediately following a controlled burn of the habitat may also increase detection (Gaya 2019), although we have concerns that standardizing surveys to a period after a controlled burn could bias data if the structure of the pre-burn vegetation, which may be somewhat overgrown, reduces nesting attempts by females that may move to more-open habitat (Diemer 1992b).

Despite the possibility that some small burrows were missed by our surveys, the differences in size-class distributions we observed between habitats were striking and broadly consistent across sites and survey methods. We found very few juvenile-sized burrows (≤ 3%; Figs. 2–4) in flatwoods habitats at local and landscape scales, which could imply low survival of early life stages. We suggest that frequent flooding of the poorly drained soils during the summer rainy season may severely limit nest success in flatwoods because eggs, which require gas exchange, cannot remain viable when submerged for extended periods (Kam 1994). Although we only monitored 3 nests in flatwoods directly, all 3 were flooded during the nesting season (1 for more than a month) and all 3 failed to hatch, with only 1 failure attributed to predation.

Additional data on egg survival in flatwoods are lacking because our search effort yielded few nests. However, we attribute this to our reliance on a minimally invasive search technique rather than a lack of nesting attempts, as Rothermel and Castellón (2014) found similar proportions of gravid females and similar clutch sizes in flatwoods (5.7 ± 0.9 SD) and scrub (6.2 ± 1.4 SD) at APAFR. Furthermore, although we initially hypothesized that gravid females might relocate or make brief excursions to nest in drier habitats, there was no evidence that this occurred based on telemetry data. In fact, there was little movement between mesic flatwoods and more-xeric habitats, as only 1 of the tortoises we monitored moved from mesic flatwoods into scrubby-flatwoods after a period of flooding, when she spent several days outside a burrow sheltered under woody debris (Castellón et al. 2018). Therefore, we attribute the apparent lack of recruitment in flatwoods to nest failure from flooding rather than low fecundity or nesting forays by females to drier habitats. Although it is possible that females selected nesting sites within the flatwoods that were slightly higher than the mean elevation (see Castellón et al. 2020), few locations would have escaped flooding during our study because the flatwoods sites were entirely inundated by shallow surface water during heavy rains from tropical storms.

Given the 80- to 90-d incubation period (Mushinsky et al. 2006), eggs laid in flatwoods from late April through May would not typically hatch before the onset of the wet season in June, when flooding risk increases. Although early nests might avoid flooding in some years, relatively dry weather for 3 yrs leading up to the 2015 survey did not produce a shift in the size-class distribution toward smaller/younger classes in our flatwoods sites (Figs. 3 and 4), as would be expected if recruitment increased during drier years. It was notable, however, that subadult-sized burrows represented relatively large percentages (8%–15%; Figs. 2–4) of the distributions in flatwoods, suggesting that recruitment had been successful in previous years.

We speculate that an extended multiyear drought with lower-than-average tropical moisture, as might be expected under strong La Niña conditions, could lead to higher rates of nest success in mesic flatwoods. Under these conditions, the water table in flatwoods might be low enough to prevent nest flooding, but forage would likely remain relatively abundant. Given the long reproductive lifespan of female gopher tortoises, occasional years of successful recruitment during droughts might offset years when nest success is lower, especially if juvenile survival is high and age at maturity is reduced due to abundant forage (Mushinsky et al. 2003). Long-term monitoring encompassing years with more-extreme drought conditions could reveal whether recruitment in flatwoods follows such an episodic pattern. It also bears noting that many gopher tortoise populations in southern Florida may benefit from extensive hydrologic manipulations (e.g., channelization for flood control) that have artificially drained many areas with previously flood-prone soils (Gentile et al. 2001; Ogden et al. 2005).

In comparison to flatwoods, juvenile-sized burrows were relatively well represented in scrub sites, with percentages (11%–41%; Figs. 2–4) that were closer to the range reported for sandhill sites in northern Florida and southern Georgia (∼ 20%–47%; Diemer 1992a; Tuberville et al. 2014), suggesting that recruitment in scrub may be reasonably successful despite low abundance of forage. We also speculate that higher than average rainfall in 2011 may have contributed to the unusually large percentage of juvenile-sized burrows observed in SC1 that year due to a potential increase in forage abundance. Although gopher tortoise forage plants are generally sparse in scrub (Myers 1990; Whitney et al. 2004; Castellón et al. 2012), especially forbs that are considered critical for juvenile growth and survival (Garner and Landers 1981; Aresco and Guyer 1999; Mushinsky et al. 2003); they may be more abundant in years with higher-than-average rainfall as occurred in 2011 (Petrů and Menges 2003). A correspondingly large percentage of subadult-sized burrows in 2015 further suggested that juveniles recruited in 2011 may have survived to reach the subadult age class.

The surprisingly low percentage of subadult-sized burrows observed in SC2 during 2015 (Figs. 2 and 4), however, suggests that the pattern in SC1 may not have been typical across the study area. Nonetheless, subadult-sized burrows represented 15% of those encountered in scrub during the 2009 landscape-scale LTDS survey (Fig. 2), suggesting that recruitment across scrub habitats at APAFR was reasonably high in previous years (i.e., comparable to that in sandhill habitats elsewhere; Diemer 1992a; Tuberville et al. 2014). The lower percentages of juvenile-sized burrows detected in scrub by our LTDS surveys (11%), compared with burrow mapping results (13%–41%), may have been due to initiation of LTDS surveys in April, prior to emergence of hatchlings, whereas burrow mapping was conducted during late summer when hatchling abundance was at its peak. Also, compared with the double-coverage surveys in the burrow mapping sites, LTDS surveys may be more biased toward larger-sized burrows because they are visible at longer distances.

Predation on early life stages is another important determinant of recruitment. High rates of gopher tortoise nest predation have been documented elsewhere (Landers et al. 1980; Smith et al. 2013), and gopher tortoise survival during the first year is generally low (estimated 5%–10%), even in high-quality habitat (Perez-Heydrich et al. 2012). However, few data are available to assess predation rates in scrub and flatwoods. We found no significant difference in burrow visitation by predators in our flatwoods and scrub sites, but more direct monitoring of nest, hatchling, and juvenile survival would be required to rigorously assess predation rates.

As a counterbalance to potential constraints related to suboptimal habitat quality in flatwoods and scrub habitats, we suggest that persistence of tortoise populations in southern Florida could be fostered, in part, by compensatory mechanisms associated with the warmer climate. For instance, lack of winter dormancy and access to year-round foraging (Douglass and Layne 1978; Moore et al. 2009) may explain the faster growth rate and earlier age at first reproduction observed in this part of the range (Landers et al. 1982; Mushinsky et al. 1994; Aresco and Guyer 1999), which could provide a demographic boost to populations in southern Florida. Tortoises living in the southernmost extent of Florida may also benefit from an extended breeding season (Moore et al. 2009; Allman et al. 2019), increasing the chances that some nests in flatwoods might hatch prior to onset of the rainy season and thereby escape flooding. The longer breeding season also raises the possibility that females could produce more than 1 clutch per season, although additional research is needed to confirm this hypothesis (Allman et al. 2019).

These potential compensatory mechanisms, and the possibility of shifting source-sink dynamics depending on weather patterns (e.g., flatwoods and scrub alternately functioning as source habitats during dry and wet years, respectively), raise intriguing questions that warrant further investigation. Nonetheless, our research provides some of the first insights into the demography of gopher tortoises in this part of the range, where careful management may be critical if populations are unstable due to suboptimal conditions combined with increasing pressures from human encroachment and climate change that could cause shifts in demographic trajectories. This may be particularly important in flatwoods where recruitment appears limited and where longevity of adults could mask persistent lack of recruitment, leading to abrupt declines when the adults begin to die off (Lovich et al. 2018).