How far do adult turtles move? Home range and dispersal of Kinosternon integrum

Abstract We describe the home range and movements of a population of Kinosternon integrum in Tonatico, Estado de México, México, over 3.5 years (during rainy and dry season months) using radiotelemetry in 37 adult turtles. The results showed that the home range of K. integrum was 0.151 ± 0.051 ha using 50% kernel density estimator (KDE), and 0.657 ± 0.214 ha using 95% KDE; the home range did not vary between sexes. Kinosternon integrum showed low distances traveled 51.44 ± 4.50 m, where 87.3% (n = 373) of movements were <100 m. The distance traveled differed by season, and movement category (aquatic and terrestrial movements). The shortest distance occurred during the dry season, during which some individuals move to estivation sites, and these movements were shorter than movements to artificial ponds (cattle ponds). In this population, home range and movement are similar to other species of the genus Kinosternon. Overall, the results indicate than K. integrum are highly dependent on aquatic habitats, but also utilize the terrestrial habitats for different biological activities, and to maintain viable populations. Therefore, the conservation of the entire inhabited area is fundamental. This study highlights the need to increase the studies, in Central México, concerning habitat use of freshwater turtles in order to increase the efficiency of conservation strategies.

The quality and quantity of resources varies temporally and spatially, and variation in the use of resources can influence the movement patterns and space use by animals (Roe & Georges, 2007). In habitats where extremely high temperatures or drought conditions occur during any season of the year, resources such as water or food may be scarce (Ligon & Stone, 2003;Litzgus & Mousseau, 2004;Pinder, Storey, & Ultsch, 1992;Storey, 2001). Therefore, individuals have developed different strategies to survive (Hall & Steidl, 2007) such as a) migration toward permanent water bodies, b) congregation in the local aquatic habitat (Ligon & Stone, 2003), or c) estivation until the water supply is restored (Iverson, 1990;Ligon & Stone, 2003;Wygoda, 1979); the use of these strategies manifest the necessity of identifying variations in home range, movements, and habitat requirements of freshwater turtles in order to effectively design and implement conservation strategies (Pittman & Dorcas, 2009;Rizkalla & Swihart, 2006), as well as improve management plans of endemic species, especially those inhabiting highly urbanized and polluted areas (Harden, Price, & Dorcas, 2009).
Kinosternon integrum is endemic to México (Lemos-Espinal & Smith, 2009;Figure 1), and, although considered least concern by the IUCN Red List, is under special protection by Mexican laws (SEMARNAT, 2010). Vulnerability was assessed using an Environmental Vulnerability Score (EVS) ranking K. integrum an 11, placing it as a medium vulnerability species (Wilson, Mata-Silva, & Johnson, 2013). Although it is the most widely distributed and the most commonly encountered freshwater turtle in México (Iverson, 1999), it is not excluded from the issues that affect freshwater turtles worldwide (e.g., water pollution, habitat degradation; Dudgeon et al., 2006). Currently, data concerning home range and movements in the genus Kinosternon in México are sparse and, to our knowledge, movement patterns have only been reported for one species in México (Kinosternon leucostomum; Morales-Verdeja & Vogt, 1997). Therefore, the purpose of this study was to evaluate the home range of K. integrum, as well as the movements and their variations according to sex, season, and category (aquatic and terrestrial movements), across years and estimate which of these factors affect the movements. Finally, the home range and movements were compared with other species of family Kinosternidae.

| Study site
This study was carried out in Zapote River, located in Tonatico, Estado de México, México (18°45′ N, 99°38′ W; 1,500 m.a.s.l. Figure 2). The area has a semiwarm humid climate with summer rains with average annual temperature ranging from 18 to 22°C, and annual precipitation ranging from 1,000 to 1,200 mm (INEGI, 2008;Luna, Morrone, & Espinosa, 2007). The rainy season typically occurs from mid-June to mid-September but sometimes is extended and includes the end of May and beginning of October, and the dry season occurs during the remaining months (Hernández-Gallegos & Domínguez-Vega, 2012; Figure 2). The Zapote River is fed by springs and runoffs, and the water level in the river changes considerably due to seasonal and annual variations in rainfall. Most of the river dries prior to the rainy season during April and May (the warmer and drier months of the year), and only a few permanent ponds remain ( Figure 2). When the summer rains begin, the amount of water increases, and the water flow becomes continuous. Sand and rocks dominate the substratum, and the dominant vegetation in the riverbank is gallery forest (trees and shrubs).

| Sampling methods
Samplings were performed from December 2012 to May 2016, and turtles were captured by hand and snorkeling along 1.25 km of the Zapote River. All turtles were sexed and weighed, and their straight carapace length and curved carapace length were measured. All F I G U R E 1 Adult female of Kinosternon integrum from Tonatico, Estado de México, México (Photograph by Ailed Pérez-Pérez) turtles were marked using shell notching following Cagle (1939).
Thirty-seven adult turtles (18 females, 19 males; carapace length >120 mm; weight >300 g) were fitted with radiotransmitters (LL Electronics, Mahomet IL, model LF2 CR2477; weight 16.5 g, did not exceed 7% of body mass), which were attached to the posterior marginal scutes with silicon sealer. Turtles were held in the laboratory for 48 hr before being released at their exact capture locations. Turtles were monitored by a yagi antenna and a receiver (LL Electronics MN4000); the locations were recorded using a global positioning system unit (GPS, Garmin Oregon). Turtles were tracked at different periods throughout the sampling period (Table 1); the relocations were performed fortnightly; the sampling area of relocations included 1.12 km 2 (Figure 2), due the movements of the turtles to other habitats (estivation sites, spring, artificial ponds, and runoffs). Movements from the river to other habitats (estivation sites, artificial ponds, and marsh) were considered terrestrial movements, whereas movements within the river and within artificial ponds were considered aquatic movements.
In studies concerning movements and home range, independence is not biologically possible (Blundell, Maier, & Debevec, 2001); animals typically move in a nonrandom fashion and will return repeatedly to important areas (e.g., foraging sites, or locations that provide shelter or other important resources); this causes strong autocorrelation (Powell, 2000;de Solla, Bonduriansky, & Brooks, 1999;Swihart & Slade, 1985). de Solla et al. (1999) and Blundell et al. (2001) demonstrated that the kernel densities do not require serial independence of observations; the autocorrelation had no apparent effect on linear estimates of home ranges with any kernel method. Due to this, to estimate home range, we calculated 50% and 95% kernel density estimator (KDE) (only for turtles with more than 15 locations; 6 females and 6 males), using R software (R Development Core Team, R Foundation for Statistical Computing, Vienna, Austria). Most locations were highly clustered; thus, we used least-squares cross-validation for optimum bandwidth selection (Bowman, 1984). Kernel method is one of the least biased and most precise home estimator (Seaman, Griffith, & Powell, 1998;Worton, 1989); 50% KDE reflected core area of home range, where the high intensity of use of 95% indicated size of home range (Donaldson & Echternacht, 2005). We calculated the home ranges size of all animals with at least three captures or relocations, using 100% minimum convex polygon method (MCP), and estimate total distance of the movements between locations using the Xtool Pro extension in ArcGis 10.3 (Environmental Systems Research).

| Statistical analyses
The size of home range (50% and 95% KDE) was compared separately by two-way ANOVA using as a factor sex and mobility of the animals (two levels; animals that remain in the river all time and individuals that performed terrestrial movements) as cofactors. The correlations of the home range with straight carapace length and weight of turtles F I G U R E 2 Study site and climograph (2012-2013) of Tonatico, Estado de México, México were analyzed by linear regression. The correlation of the days between captures with distance of the movements was tested by simple regression; the correlation of the distance of movements with weight and carapace length was tested by means of the Spearman's rank test.
The best analysis to evaluate distance of the movements can be a linear mixed effect model (LME), which includes individuals as a random factor, as has been performed in other reptile studies, including freshwater turtles (Tiliqua rugosa, Kerr & Bull, 2006; Sternotherus odoratus, Rowe, Lehr, McCarthy, & Converse, 2009; Chrysemys picta marginata, Rowe & Dalgarn, 2010; Sternotherus minor peltifer, Ennen & Scott, 2013). According to von Ende (1993) and Schafer and Graham (2002), this form of analysis accommodated T A B L E 1 Characteristics of tracking, home range, and movements of 37 adult turtles of Kinosternon integrum in Tonatico, Estado de México, México. Females (H), Males (M). ❊ movements into the river, ✣ movements to estivation sites, ❖ movements to artificial ponds, and ♦ movements to spring All statistics were performed in SPSS version 20.0, means were presented ± 1 standard error (unless otherwise noted) and results were deemed significant if p < .05.

| Distance of the movements of Kinosternon integrum
The The distance traveled by turtles was mainly affected by season and by movement category (Tables 2 and 3). The movements during the rainy season were greater than the dry season (Table 3, Figure 3a), and terrestrial movements were greater than aquatic movements (Table 3, Figure 3b). When turtles moved from the river to artificial ponds (and return to the river) they traveled greater distances than those traveling to/from estivation sites (Table 3; Figure 3c). Only the interactions Year × Season, Season × Category, and Sex × Year × Season have an affect on the distance of the movements (Table 2). Differences in mean movement distance among years (Figure 4a) or category (Figure 4b) are not independent of the season. Moreover, differences in mean movement distance between sexes are not independent of the season or year ( Figure 5). range in the family Kinosternidae (considering 95% KDE; Table 4).

| DISCUSSION
However, it is important to note that the information concerning home range in Kinosternidae is still scarce, as there is only information of home range for nine of the 25 species. Furthermore, most studies have estimated home range using the MCP method, which has been criticized due to its sensitivity to extreme outlying locations, and generally can cause an overestimation of the home range (Powell, 2000). In this study, the results of MPC include habitats in which individuals of K. integrum were never observed (e.g., crops, uncovered soil).
In turtles, intra-and interspecific variation on home range and patterns of movements of turtles has been reported (Slavenko et al., 2016). Home range and movements can be more influenced by extrinsic factors (i.e., weather, season, size of wetlands, environmental variation, and distribution of resources and food) than intrinsic factors (i.e., sex, size, and sexual maturity.) (Slavenko et al., 2016). The movements of K. integrum were influenced by extrinsic factors such as season and distance between microhabitats essential for survival (estivation sites and artificial ponds), but the distance traveled is consistent among years. The intrinsic factors such as size or sex did not affect the movement distance, which has been reported in other species  (K. subrubrum, Bennett et al., 1970;K. baurii, Wygoda, 1979). Hence, identification of factors that promote the movements within a population could improve future conservations plans.
Most of the individuals of K. integrum in Tonatico presented low mobility and high fidelity for their aquatic habitat within the river (25.1% of the movements were <5 m). K. integrum exhibited both aquatic (within the river) and terrestrial movements (toward artificial ponds and estivation sites), which have been recorded in other species of Kinosternidae. According to Ennen and Scott (2008) (Tables 4   and 5). The high fidelity presented in turtles has been related to the availability of resources in the environment. The low mobility of K. integrum suggests that the Zapote River offers optimal environmental conditions to obtain resources locally that adequately satisfy the different energy demands of individuals. Interestingly, a similar situation has also been suggested by previous studies (Milam & Melvin, 2001).

Terrestrial movements have been recorded in other species
within the Kinosternon genus, and long-distance movements are rare.
Nevertheless, some organisms within populations can perform largescale movements within the same habitat or among habitats (Bowne, 2008;Gibbons, Greene, & Congdon, 1983;Hall & Steidl, 2007;Ligon & Stone, 2003;Pittman & Dorcas, 2009;Smar & Chambers, 2005;Stone, 2001). In environments where the water is scarce, the number of terrestrial movements by turtles should be relatively high as turtles may need to move to ephemeral pools when they become available and to more permanent pools or to estivation sites when ephemeral pools dry (Stone, 2001). Herein, we present a similar result, because during the

F I G U R E 4 Mean movement distance of Kinosternon integrum according to: a) interaction between Year and Season, and b) interaction between Category and Season
F I G U R E 5 Mean of distance of movements of Kinosternon integrum according to three-way interaction among Sex, Year, and Season samplings we found that turtles migrate temporally to other sites. The terrestrial movements performed by K. integrum were carried out between river and estivation sites, and between river and artificial ponds.
These strategies are present to avoid periods of droughts presented during the dry season (November-May) in the habitat of this population of K. integrum. The largest terrestrial movements carried out between river and artificial pond were recorded during the onset of the rainy season. This indicates that some individuals of K. integrum use permanent ponds of the river as a refuge during the dry season, but during the beginning of the rainy season migrate to temporal ponds (artificial ponds). Roe and Georges (2007) revealed that the movements between water bodies were not species-specific and some species of turtles use more than one wetland, normally traveling between two and three bodies of water. In the study site, it is common that some individuals of K. integrum perform seasonal migrations between river and artificial ponds; this ability to move between water bodies allows them to have a better survival condition and body condition, as the permanent pools can offer different resources and benefits, while others have dried (Roe & Georges, 2007. Because of this, the maintenance of landscape connectivity can be an important issue in the management of habitats adjacent to water bodies (ground safety zone). The conservation of the whole area is very important, as these environmental changes can be detrimental on the natural history of these organisms.
Anthropogenic barriers (i.e., crops, roads) have a profound effect upon the survival of the turtles, mainly on those that perform overland movements (Myfsud & Myfsud, 2008;Pittman & Dorcas, 2009). It has been documented that in areas where disturbances occur in their habitat, turtles move greater distances (Plummer & Mills, 2008). Hence, conservation plans should include actions to maintain natural characteristics of the rivers and promote land corridors, in order to ensure that turtle populations are not negatively impacted.

ACKNOWLEDGMENTS
We would like to thank Theodore Roosevelt Memorial Fund of the

CONFLICT OF INTEREST
None declared.
T A B L E 5 Mean of distance of the movements of some species of family Kinosternidae (including Kinosternon integrum). The estimation of the movements by ✢ linear home range, ◼ total daily distance movement, ★ mean distance, ✸ minimum movement distance, ◻ distance between captures Movements to nesting sites. Species are mentioned according to Iverson et al. (2013). Text in bold indicates the results of this paper.