Linear Activity Areas and Basking-Site Fidelity of Two Imperiled River Turtle Species (Graptemys oculifera and Graptemys flavimaculata) of Mississippi
Abstract
Many freshwater turtles in the United States are considered species of greatest conservation need in the states in which they occur, including turtles in the genus Graptemys (map turtles and sawbacks). However, basic life-history information is still lacking for many Graptemys species, particularly details about their movements, home range, and habitat use. The objective of this study was to describe linear activity area (LAA) and basking-site fidelity for 2 federally threatened Graptemys species in Mississippi (Graptemys flavimaculata, Graptemys oculifera) via capture–mark–recapture events. For G. flavimaculata, 97 individuals (30 female, 67 male) were recaptured during the study at 3 sites. Mean LAA for females (138 ± 293 m; 0–1489 m) and males (308 ± 637 m; 0–3773 m) was statistically similar, but turtle LAAs from 2 upstream sites were longer than LAAs from a downstream site. The maximum single movement was 2877 m by a male; alternatively, many males and females were recaptured at the same basking log across multiple years. For G. oculifera, 28 individuals (14 female, 14 male) were recaptured during the study at 1 site and mean LAAs were statistically similar for females (130 ± 206 m; 4.6–747 m) and males (231 ± 357 m; 0–1310 m). Differences in LAAs among sites for G. flavimaculata may be associated with habitat differences observed among the sites (i.e., river size, hydrology, position on river continuum) or associated with different turtle densities. Basking-site fidelity appears to be strong in both species and is likely driven by deadwood persistence time and/or sexually divergent preferences in basking sites. Additional movement ecology studies are needed for Graptemys species to determine how differences in habitat, turtle densities, and deadwood interact to influence linear home ranges.
Riverine turtles face significant conservation threats in the southeastern United States. Specifically, many southeastern US rivers have been modified via impoundments, stream channelization, and desnagging (Dynesius and Nilsson 1994; Moll and Moll 2004). Further, water quality in many of these same rivers has decreased, which can be attributed to increased nutrients and sedimentation due to land-use changes (i.e., development, agriculture, silviculture; Stets et al. 2020); a host of other types of contaminants are also of emerging concern (e.g., endocrine-disrupting chemicals; Noguera-Oviedo and Aga 2016). However, the southeastern United States is also considered a turtle biodiversity hotspot (Buhlmann et al. 2009), primarily driven by speciation of river turtles along the coastal margin of the Gulf of Mexico (Thomson et al. 2018). Over 60% of the turtle species in the southeastern United States are considered imperiled and at risk for declines, and 3 major reasons for decline include collection for harvest/pet trade; habitat destruction, degradation, and fragmentation; and incidental drowning through commercial or recreational fishing activities (Buhlmann and Gibbons 1997; Larocque et al. 2012; Shook et al. 2023).
Many turtles of the genus Graptemys (map turtles and sawbacks) are associated with southeastern US streams and rivers as well as their associated habitats (e.g., bayous and oxbow ponds; Lindeman 1999a). Nine of the 14 Graptemys species are endemic to single river systems along the Gulf of Mexico (Lindeman 2013), and many Graptemys are species of greatest conservation need (SGCN) in states within their respective ranges. For example, of the 9 Graptemys species found in Mississippi, 6 are considered SGCN: Graptemys flavimaculata, Graptemys gibbonsi, Graptemys nigrinoda, Graptemys oculifera, Graptemys pearlensis, and Graptemys pulchra (Mississippi Museum of Natural Science 2015); a seventh species, Graptemys geographica, was recently discovered in Mississippi (Brown et al. 2020) and will be included as an SGCN in the next State Wildlife Action Plan due to its limited range in the state (E. Field, pers. comm. October 2023). Many Graptemys have high conservation threat levels, yet they are also the most poorly studied species in the United States, particularly in comparison to species that more commonly occur in pond settings (e.g., Trachemys scripta, Sternotherus odoratus; Lovich and Ennen 2013). One of the reasons that information for Graptemys species is limited is that riverine habitats provide significant challenges for investigators conducting field research. For example, many Graptemys species rarely enter baited hoop traps, thus capture methods must include active sampling methods such as basking traps, snorkeling, or dip-netting (Sterrett et al. 2010; Selman et al. 2012). Second, many Graptemys species have small geographic ranges and are highly localized in their distribution, and their ranges may be distant from research stations or colleges/universities. Third, because of the distance and sampling methods needed, field work for Graptemys is usually more expensive due to the fuel and maintenance expenses associated with boats and vehicles to access distant locations.
Two Graptemys species are considered federally Threatened and occur in Mississippi: Graptemys flavimaculata (yellow-blotched sawback; US Fish and Wildlife Service 1991) from the Pascagoula River system of southeastern Mississippi and Graptemys oculifera (ringed sawback; US Fish and Wildlife Service 1986) from the Pearl River of central Mississippi and southeastern Louisiana. For the aforementioned logistical reasons and because most of their daily activities occur underwater (i.e., outside of aerial basking and nesting forays), relatively little is known about their life history and ecology. Further, there is little information available on the activity patterns and movements of southeastern Graptemys species (see Jones 1996; Sterrett et al. 2015), particularly when compared to their pond turtle equivalents (e.g., Chrysemys picta, Trachemys sp.). In the absence of a more detailed radiotelemetry data to define linear home ranges for riverine turtles, linear activity area (LAA) data from recapture points can be informative and can provide an estimate of minimal activity area while also allowing an estimation of basking fidelity to particular basking structures. Therefore, the primary objective of this study was to describe linear activity areas of male and female G. flavimaculata and G. oculifera based on long-term capture and recapture points. Along with this objective, I present information regarding site fidelity of individual turtles to both small reaches of the river and singular basking structures.
METHODS
Study Sites. —
Trapping for G. flavimaculata was done at 3 sites in the Pascagoula River system. These sites were the Leaf River (LR, Forrest County; trapping site length: ∼ 3.1 river kilometers [rkm]), lower Pascagoula River (PaR, Jackson County; ∼ 5.4 rkm), and lower Chickasawhay River (LCR, Greene County; ∼5 .0 rkm). Habitat characteristics for these sites are detailed in Selman (2012) and Selman and Lindeman (2015).
Graptemys oculifera were sampled by trapping at 1 site on the Pearl River (PeR, Marion County; ∼ 4.4 rkm). The PeR site is characteristic of a medium-sized (∼ 75–175 m wide) Gulf Coastal Plain river, with alternating gravel/sand bars and cutbank sections, abundant submergent and emergent deadwood snags, and a sand and gravel substrate. The flow of the Pearl River (discharge: 24–1700 m3/sec) is primarily regulated by the Ross Barnett Reservoir spillway, approximately 240 rkm upstream from the site. The site is surrounded by riparian bottomland forest (i.e., water oak [Quercus nigra], baldcypress [Taxodium distichum], sycamore [Platanus occidentalis], spruce pine [Pinus glabra]), as well as a small amount of pasture land and a small number of fishing camps.
Capture Methods and Measurements. —
From April through October of 2005 and 2006, sampling was conducted once per month at all 4 sites for 3 to 5 d each month. During the 2007 and 2008 field seasons, I continued to sample turtles using similar methods, but only at the LR and PaR sites. At all sites, turtles were captured by attaching open-topped basking traps (made of 1.9-cm polyvinyl chloride–coated crawfish wire; The Fish Net Company, Jonesville, LA) to known turtle basking structures as described by Selman et al. (2012). A maximum of 17 traps were used in a single trap day, but some traps were moved throughout the day, especially if turtles avoided the trap log or there was a change in river water levels. Traps were placed on different basking snags (e.g., logs, branches, tree crowns, tangles) to sample structures preferred by different size classes and sexes of turtles. I also captured individuals opportunistically by hand or by dip net at all sites.
After capture, I determined the sex of individuals when possible based on the presumption that males were smaller, had longer foreclaws and thicker and longer tails, with the cloaca posterior to the carapace rim (Jones and Selman 2009, Selman and Jones 2011). I measured midline plastron length (PL) to the nearest 1 mm with tree calipers and permanently marked the marginal scutes with holes from an electric drill according to the method by Cagle (1939) in order to later identify recaptured individuals. I also recorded capture locations using a handheld global positioning system (GPS) accurate to within 6 m. For basking structures where traps were located, a name/number was given to the structure, which permitted me to determine if individuals were captured on the same structure in subsequent capture events. Following marking and measuring of turtles, I released all individuals at their points of capture.
Linear Activity Areas. —
LAAs have been calculated for riverine turtles by calculating the distances between the 2 most distant capture points (Sexton 1959; Kornilev et al. 2010). LAAs for G. flavimaculata and G. oculifera were inferred from marked individuals that were thereafter recaptured. For capture and recapture points, I plotted the GPS coordinates within GoogleEarth (v. 6.1; Google Inc., Mountain View, CA), and then the distance was measured between capture and recapture GPS points, with longer distances traced along the midriver channel using the measure feature in GoogleEarth (Kornilev et al. 2010). I also calculated the number of days between captures. I acknowledge that more points are needed to accurately measure home ranges for riverine Graptemys species (Jones 1996). Thus, that is why I make the distinction to call these LAAs, as these calculations are not representative of true home range lengths.
For G. flavimaculata, a nonparametric Wilcoxon rank sums test was used to determine if the total time between captures was similar by sex. I also used Wilcoxon rank sums tests to determine if LAAs were similar by sex, by site (LCR, LR, PaR), and by the number of recaptures. For G. oculifera, Wilcoxon rank sums tests were used to determine if the total time between captures was similar by sex and if LAA was similar by sex. If any of the above tests were significantly different, I used a Wilcoxon nonparametric multiple comparisons test to delineate differences between groups. For each species, a linear regression was used to see if LAA was correlated with the total time between the first and last capture. For all statistical analyses, I used the software JMP 12.2.0 (SAS Institute, Cary, NC), with significance of tests at α = 0.05.
RESULTS
Graptemys flavimaculata Movements and LAA. —
Of the 97 individuals that were recaptured in the study, 79 were recaptured once (45 LR, 5 LCR, 29 PaR), 13 were recaptured twice (10 LR, 1 LCR, 2 PaR), and 5 were recaptured 3 times (3 LR, 1 LCR, 1 PaR). The mean total time between the first capture and last capture for females was 356 ± 274 d (25–976 d; n = 30) and for males was 335 ± 298 d (3–1194 d; n = 67). There was no difference between sexes for the total time between first and last capture (χ2 = 0.23, p = 0.63).
The mean LAA for female G. flavimaculata was 138 ± 293 m (0–1489 m; n = 30); the longest female LAA was a PaR female (17.0 cm PL) captured twice between 30 October 2006 and 29 April 2008. The mean LAA for males was 308 ± 636 m (0–3773 m; n = 67); the longest male LAA was an LCR male (7.6-cm PL) captured 4 times between 16 June 2005 and 30 September 2006. For LAA comparisons, there was no difference by sex (χ2 = 0.76, p = 0.38), but LAAs were different by site (χ2 = 12.7, p = 0.002; Table 1). Turtles from the LCR (Fig. 1) had longer LAAs compared to LR (Fig. 2) and PaR (Fig. 3) turtles, while LR turtles had longer LAAs than PaR turtles. There was no significant relationship between the LAA and number of captures (F1,97 = 30.4, p = 0.08) or LAA and the total time between first and last capture (F1,97 = 0.82, p = 0.37).
Citation: Chelonian Conservation and Biology: Celebrating 25 Years as the World’s Turtle and Tortoise Journal 23, 1; 10.2744/CCB-1613.1
Citation: Chelonian Conservation and Biology: Celebrating 25 Years as the World’s Turtle and Tortoise Journal 23, 1; 10.2744/CCB-1613.1
Citation: Chelonian Conservation and Biology: Celebrating 25 Years as the World’s Turtle and Tortoise Journal 23, 1; 10.2744/CCB-1613.1
For all individuals, the mean distance moved between recapture locations was 121 ± 277 m (range: 0–1489 m; n = 30 individuals) for females and 266 ± 487 m (0–2877 m; n = 67) for males. The longest single movement observed was 2877 m by an LCR male (7.2-cm PL) between 23 October 2005 and 30 September 2006 (Fig. 4), while the longest female movement was 1489 m by a PaR female (17.0-cm PL) between 30 October 2006 and 29 April 2008. Conversely, 9 males (7 LR, 2 PaR) and 2 females (both PaR) were known to be recaptured on the same log at later recapture events, and 5 other individuals (3 LR males, 2 PaR females) were recaptured within GPS accuracy of their original capture location. One male from the LR was captured 3 times on the same branch with 1109 d between recaptures, and another LR male was recaptured at the same location 3 times over 932 d. Further, a PaR female was captured 4 times on the same log over 365 d.
Citation: Chelonian Conservation and Biology: Celebrating 25 Years as the World’s Turtle and Tortoise Journal 23, 1; 10.2744/CCB-1613.1
Graptemys oculifera Movements and LAA. —
Of the 28 individuals recaptured in the study, all were recaptured only once. The mean time between recaptures was 162 ± 150 d (21–427 d; n = 14) for females and 237 ± 109 d (65–371; n = 14) for males. There was no difference between the sexes in the total time between first and last capture (χ2 = 3.05, p = 0.08).
For G. oculifera, the mean LAA was 131 ± 206 m (0–747 m; Fig. 4A) for females and 231 ± 356 km (0–1310 m; Fig. 4B) for males. The longest female LAA (12.3-cm PL) occurred between 12 July 2005 and 10 April 2006, while the longest male LAA (7.9-cm PL) occurred between 5 October 2005 and 11 October 2006. There was no difference in mean male and female LAA (χ2 = 0.48, p = 0.49; Table 1) and there was no relationship between the LAA and the days between first and last capture (F1,28 = 3.91, p = 0.06). Similar to G. flavimaculata, 2 male G. oculifera were recaptured on the same log, both 184 d apart, while a single female was recaptured within the GPS accuracy after 29 d. Furthermore, many individuals during the 2 years of the study at the Pearl River site were found to be associated with a single, highly branched spruce pine (Pinus glabra) that had fallen into the river, with individuals either having their original capture location or final capture location at this single structure (Fig. 4C).
DISCUSSION
Very little information on LAAs or home range lengths is available in the literature for Graptemys species, particularly those that occur in the southeastern United States (Jones 1996; Sterrett et al. 2015; Selman and Lindeman 2015). In this study, I collected data on 2 Graptemys species, one of which has previously had a radiotelemetry study completed (G. flavimaculata; Jones 1996) and another for which no information is known about home range or movements (G. oculifera). Even though these data are imperfect and incomplete compared to a more detailed study using telemetry, these data contribute to a better understanding of the ecology for both species.
For G. flavimaculata, the mean LAAs in this study (female mean: 138 m; male mean: 308 m) were smaller than mean home range lengths previously described via radiotelemetry by Jones (1996) conducted at the same PaR site (female mean: 1550 m; male mean: 1861 m). Most of the G. flavimaculata LAAs in this study appear to fall within the lower to midrange of the linear home ranges described by Jones (1996), but this study did not detect the longer linear distances. I suspect the differences between the 2 studies could be attributed to multiple potential interacting factors. First, the LAAs I observed may have been overly influenced by the basking-site fidelity exhibited by some individuals (more below), as these basking structures were the only points where I could make relocations in this study. Using radiotelemetry, individuals can be detected even when they venture away from basking locations. Second, the lack of the “upper distribution” may have been due to the length of the site that was trapped during this study. All of the sites I trapped were < 5.4 rkm, and this is shorter than the maximum linear home range lengths previously described by Jones (1996). Therefore, it seems likely that some turtles may have left the study area and later returned to their previous location without my knowledge of their movements between captures. Third, home range sizes increase with increasing the number of relocation/recapture points (Girard et al. 2002). For example, the maximum number of locations for any G. flavimaculata individual in this study was 4, whereas all 16 G. flavimaculata individuals tracked by Jones (1996) had ≥ 26 relocation points (maximum: 133). Nonetheless, consistent findings between this study and Jones (1996) were that both sexes had relatively similar sized LAAs and linear home ranges. Of note, this study occurred over a longer time period (2 yrs: LCR, PeR; 4 yrs: LR, PaR), whereas the study by Jones (1996) was limited to the battery life of most transmitters (i.e., the maximum time for males was 15 mo, and 26 mo for females; Jones 1996).
In comparison to a previous study with the syntopic G. gibbonsi (Pascagoula map turtle) at the same study sites, G. flavimaculata LAAs were smaller than G. gibbonsi LAAs (female mean: 600 m, range: 0–1690 m; male mean: 370 m, range: 0–940 m; Selman and Lindeman 2015). There was no difference in the environment because both studies occurred concurrently at the same site. Thus, it seems likely that the LAAs for G. flavimaculata were smaller than G. gibbonsi because of a biological difference between the 2 species. For example, G. flavimaculata are specialized to consume freshwater sponges (Selman and Lindeman 2018), and sponges are sessile and affixed to deadwood structures, the latter of which were ubiquitous at all of the river study sites. Conversely, G. gibbonsi primarily consume aquatic insect larvae and mollusks (Selman and Lindeman 2015; Vučenović and Lindeman 2021), and these prey species are mobile and more dispersed throughout the river channel. Therefore, this difference in prey availability and location may have necessitated fewer movements by G. flavimaculata to acquire food resources. Additionally, because G. gibbonsi are larger than G. flavimaculata, larger body sizes may confer larger home range sizes. Indeed, Suwanee cooters (Pseudemys suwanniensis), one of the largest emydid species in North America (> 43 cm carapace length), have been found to move > 100 km, homing from one location to another in Florida (Johnston et al. 2017).
One interesting aspect of this study is that LAAs for G. flavimaculata were different among the 3 study sites. It seems that there are many potential biotic or abiotic explanations for the LAA differences observed across sites. First, there were considerable differences in the habitats and river depths among the 3 sites. The upstream LR and LCR sites had narrower river widths and lower mean discharges, and both were shallower than the downstream PaR site (Selman 2012). Thus, the 3-dimensional volume of available habitat or space was considerably larger in the PaR; therefore, a relatively smaller distance of river may have provided all the necessary habitat requirements. Second, phytoplankton and zooplankton biomass varies along the stream continuum from low to high from the headwaters to river outlets (Vannote et al. 1980). Because sponges consume phytoplankton and zooplankton and they are the primary prey item of G. flavimaculata (Selman and Lindeman 2018), sponge densities could be higher and more concentrated at downstream sites. Thus, shorter movements may be needed by individual turtles downstream compared to upstream. Third, population estimates at these sites were also different during the study period and densities may influence movements. Populations were most dense at the PaR (281–602/rkm), while the LR (80–120/rkm) and the LCR (93/rkm) had lower densities (Selman and Qualls 2009). Because there were higher densities at the PaR, movements might have been shorter due to more individuals being available for conspecific interactions such as mating. Conversely, in areas with lower densities, these interactions might be less frequent and longer movements may be needed to interact with conspecifics. Fourth, the LAA differences among sites could also be due to differences in the river length that was sampled. The LR site was smaller than the PaR and LCR sites, as it was nonnavigable upstream and downstream due to extensive gravel bars at low water. However, while the LR site was shorter, it had longer LAAs than the PaR site. Therefore, the length of the site did not entirely explain the differences of LAAs observed between sites (at least within the river distances sampled). While it was not quantified in this study, basking structure density may influence river turtle home ranges and movements, which would be an interesting avenue to investigate in the future.
For G. oculifera, however, this is the first available data for LAAs because no detailed telemetry studies have been completed or linear home ranges defined. First, G. oculifera LAAs were relatively similar to LAAs for G. flavimaculata (Table 1). This finding should not be surprising because these species are sister taxa (Thomson et al. 2018), and they are ecological analogues to one another, having similar diets and habits yet occurring in separate but neighboring drainages (Lindeman et al. 2023). A future radio or satellite telemetry study of G. oculifera is likely to find larger home range lengths than the LAAs reported herein and possibly similar to those values found by Jones (1996) for G. flavimaculata. However, it seems possible that home range lengths could vary throughout the Pearl River system similar to the differences observed in G. flavimaculata LAAs by site as described above.
One of the most intriguing findings of this study is that individuals of both Graptemys species exhibited long-term basking-site fidelity to small areas or singular basking structures over multiple years, with the longest basking-site fidelity (1109 d) exceeding other observations to date (932 d; Selman 2017). My observations herein confirm other studies that have also documented long-term basking-site fidelity. Jones (1996) documented a male G. flavimaculata using the same basking structure over 9 mo, while Selman and Lindeman (2015) reported a female G. gibbonsi basking on the same log with 397 d between captures.
There are several potential reasons for the basking-site fidelity observed in Graptemys species. First, there appear to be some deadwood snags that persist in the environment, particularly larger-trunked trees or tree species that have dense, resinous heartwood (e.g., Pinus) that are resistant to water penetration and slower to decay. Persistent deadwood snags may confer some familiarity for turtles, and individuals may return to desirable basking locations across years. For example, based on a chronology of images (via Google Earth), the large spruce pine (Pinus glabra) tree at the PeR site was at this location for at least 9 yrs (2004–2013) and possibly longer (i.e., high river levels in some photos obscured the snag and images were lacking for some years). Likewise, if these areas continue to be high-quality basking structures, researcher bias may also be introduced due to productivity of trapping and/or familiarity with the basking structure. Second, it seems likely that different-sized turtles show basking-site preferences due to their different body sizes. For example, larger females need larger, more supportive structures upon which to bask, while smaller males and juveniles may show fidelity and return to smaller deadwood snags (i.e., smaller limbs) due to encounters with larger turtles. Indeed, differential selection of basking snags likely allows smaller turtles to avoid competition and agonistic encounters with larger turtles (Lindeman 1999b).
Deadwood structures are an integral part of the biology of both species and serve as a basking platform (e.g., Selman and Qualls 2011), a substrate for prey attachment (Selman and Lindeman 2018), and a place for individuals to rest (Boyer 1965; Waters 1974). Further, Lindeman (1998) found that a higher density of basking deadwood correlated strongly to higher basking densities of turtles. Thus, conservation of riverine corridors and mature riparian forest buffers should be of utmost conservation concern for both of these Graptemys species, as well as other imperiled river turtles. Further studies are needed to quantify the persistence of deadwood over time in river environments, with particular attention given to understanding how tree species and sizes contribute to persistence. Further, additional movement ecology studies are needed in Gulf Coast Graptemys species, particularly studies that occur across multiple sites. Collecting data across multiple sites could elucidate how differences in habitat, turtle densities, and deadwood interact to influence linear home ranges and movements in riverine turtles.
Graptemys flavimaculata linear activity areas from the lower Chickasawhay River site for the lone recaptured female (A) and all males (B; n = 6). Four capture points are indicated for male L11, a male with the longest linear activity area in the study.
Graptemys flavimaculata linear activity area from the Leaf River for all females (A; n = 17) and a representative subset of males (B; n = 13 of 41). Asterisks indicate multiple captures of an individual at the same basking log/branch or within GPS accuracy.
Graptemys flavimaculata linear activity areas from the Pascagoula River site for females (A; n = 11) and males (B; n = 21). Female 2,8,9-9 was captured 4 times in the same location (marked with red arrow). Asterisks indicate individuals where recapture locations were from the same basking log/branch or within global positioning system accuracy.
Graptemys oculifera linear activity areas from the Pearl River site for females (A; n = 14) and males (B; n = 14). Asterisks indicate individuals where recapture locations were from the same basking log/branch or within global positioning system accuracy accuracy. An enlargement (C) of the red square in panel As and B of a spruce pine (Pinus glabra) basking structure that persisted for at least 9 yrs (see text).
Contributor Notes
Handling Editor: Peter V. Lindeman
