Study Site

The Pátzcuaro Lake is an endorreic basin located in the Mexican Transvolcanic Belt. Water that flows from the mountain around the basin during the rainy season feeds the lake. The basin has been isolated from the Lerma River Basin for 540 kyrs owing to the dynamic volcanic activity of the area (Robles-Camacho et al. 2010). As the result of its isolation, the Pátzcuaro Basin holds an important number of endemic vertebrate taxa, such as various species of silverside (Chirostoma sp.), and other native fishes of Central México, such as Algansea lacustris, Alloophorus robustus, Goodea atripinnis, Neophorus diazi, Skiffia lermae, and Alloophorus dugesi (Zambrano et al. 2014). Pátzcuaro Lake also holds one microendemic species of amphibian (Ambystoma dumerilii; Huacuz-Elías 2005), a subspecies of garner snake (Thamnophis eques patzcuaroensis; Conant 2003), and a subspecies of mud turtle (K. h. tarascense; Iverson 1982), which was the only reported turtle in the basin until the present study.

We collected turtles in 2 cattle tanks and in an intermittent seasonal stream that flows down a hillside during rainy season to Pátzcuaro Lake. Our study site is located at San Jerónimo Purenchecuaro (Municipality of Quiroga) at the northern part of the lake. Cattle tanks are located at 19°41′44.02″N, 101°36′23.74″W and 19°41′37.87″N, 101°36′23.74″W in a hill beside the lake (Fig. 1). The cattle tanks are only 182 m apart from each other but are separated by a rustic stone wall. Both tanks hold water from late June through middle November, while the intermittent stream flows beside one of the tanks with a flow orientated at 180° southwest. This stream flows from the peak of the rainy season (late July) through the middle of September; during this time, the stream forms small pools (< 70-cm depth) on its way to the lake. An asphalt 2-way road interrupts the water flow, which eventually reaches the lake by an underpass and infiltration. The cattle tanks are mainly circular in shape. The diameter of the larger tank is near 17.19 m, with a variable depth of 50–80 cm. The diameter of the smaller tank is 10.19 m, but only 20–50 cm deep. The entire hillside area when the study was conducted was approximately 52.7 ha.

View Figure

• Download as Powerpoint
• Download Figure

Habitat type around the cattle tanks is a pine–oak (Pinus–Quercus sp.) remnant, with the typical secondary vegetation generated by historical cattle grazing around the area—such Acacia sp. trees, small bushes, agaves (Agave sp.)—and the substrate is covered by grass and bare soil. A system of rustic stone walls divides the private property patches from public land; nevertheless, cattle have access to most of the patches, which results in a homogenous structure because of grazing. The landscape resembles typical Tarasco country (Rzedowski et al. 2014): a matrix of cornfields, cattle patches, and managed forest patches from which local people extract timber for cooking and other nonwoody parts.

Methods

We captured turtles from August 2016 to November 2017 in both water tanks. We captured turtles using a 122 × 914-cm seine (Forestry Suppliers, Inc, Jackson, MS). We swiped cattle tanks 8 times in each wind-rose direction from the center to the shore of the tank. We also captured turtles by hand when possible. All turtles were marked using the shell-notch code modified from Cagle (1939). For each turtle, we measured the linear carapace length (CL), plastron length (PL), carapace width (CW), and carapace height (CH) with a dial caliper (Spi, Swiss Precision Instruments, Inc, Garden Grove, CA). Body mass was measured with 100-g and 1-kg spring scales (Pesola AG, Schindellegi, Switzerland). We sexed turtles by noting the presence of secondary sexual characters in males, which included elongated tail, concave plastron, and a big head (Iverson 1999), whereas females showed smaller tails, domed carapaces, and smaller heads.

Based on previous studies (Macip-Ríos et al. 2009, 2011; Vázquez-Gómez et al. 2016), we categorized turtles by body size as follows: 30–50 mm in CL as hatchlings or yearling, 51–70 mm in CL as juveniles, 71–90 mm in CL as immatures (although some individuals showed enough secondary sexual characteristics to be sexed), 91–150 mm in CL as adults, and > 150 mm in CL as “old” or asymptotic adults because they grow at a reduced rate.

Nine turtles (only females) were equipped with radiotransmitters attached with epoxy glue. We focused our telemetry study on females because of our initial, but uncompleted, objective to locate nests. Two models of transmitters were used: TXE-315G (Telenax, Ciudad del Carmen, Quitana Roo) and R1900 series (Advanced Telemetry Systems, Isanti, MN) because of budget issues; some transmitters were purchased directly and others were provided by one of our sponsors. The R1900 transmitters were programed in the frequency range of 164.000–164.999 and TXE-315G transmitters were programed in the frequency range of 216.000–216.999. Therefore, we also used 2 different receivers, a Telenax R1000 and an ATS R2000 (Advanced Telemetry Systems), and two Yagi antennas to locate and relocate the turtles in the field. A Garmin eTrex^® 10 Global Positioning System (Garmin, Olathe, KS) was used to record the locations with the higher precision available (3 m). We obtained locations at least twice per month; nevertheless, we missed some of the tracked turtles for several location events. During dry-season aestivation, we described microhabitats qualitatively in order to understand the number and frequency of sites used by turtles.

Statistical Analysis

We estimated population size using a log-linear Jolly-Seber model for closed populations in the Rcapture package (Baillargeon and Rivest 2012) in Program R (R Development Core Team 2015). We estimated capture probabilities (P) per period or capture event by a log-linear model (Baillargeon and Rivest 2012). To test for a sex ratio of 1:1, we used a χ² test (Zar 1999). We compared body size between males and females using a Wilcoxon test. We performed statistical analyses in JMP version 5.0.1 (SAS Institute 2002) or as otherwise indicated. We originally collected locations in decimal degree but transformed them to the Universal Transverse Mercator (UTM) coordinate system using a batch from the Earth Point Web site (Clark 2018). We estimated home range two ways, using the minimum convex polygon (MCP) and by the 50% kernel technique. We ran the home range analysis in the package adehabitat (Calenge 2006) in R (R Development Core Team 2008). We calculated overall movement between locations and estimated daily movements by hand using the UTM coordinates and tabulating data in a location-sequenced worksheet with a Microsoft Excel (version 11; Microsoft Corporation, Redmond, WA) macro. For all analyses, we considered an α = 0.05 to be significant. Because of the importance of temporal autocorrelation in radiotelemetry studies (Boyce et al. 2010), we used the Program R basic statistical package (R Development Core Team 2008) for testing, whereas to test for spatial autocorrelation, we ran Moran's I (Dray et al. 2010; Fieberg et al. 2010) with the package ape (Paradis et al. 2004) in R (R Development Core Team 2008).

Results

We conducted 11 sampling events. We marked 46 turtles and recaptured 25 individuals during the study. One individual was recaptured 6 times, another turtle was recaptured 5 times, 6 turtles were recaptured 4 times, 2 were recaptured 3 times, 15 were recaptured 2 times, and 21 were captured only once. Estimated population size using the model Mth Chao (LB) was 60 (± 6.6 SE) individuals for the 2 tanks sampled, with an Akaike Information Criterion = 239.14 (163.24 in deviance; CI = 49.6–76.2 individuals), suggesting a turtle density of 0.22 turtles/m² of surface water and 1.13 turtles/ha. Capture probabilities between sampling events averaged P = 0.44 (± 0.15) and ranged from 0.64 to 0.20.

Of the 46 turtles captured, only 4 were differentiated as males (2 old adults and 2 adults), 21 as females (19 adults and 2 clearly distinguished as females, but with a body size in the immature category), 3 as undetermined immatures, 12 as juveniles, and 11 as hatchlings or yearlings. Sex ratio was significantly biased to females (1:4.75 (χ²₁ = 11.56, p = 0.0007), while the population was composed mostly of adult females, followed by juveniles, hatchlings or yearlings, immatures, and very few old adults (Fig. 2).

View Figure

• Download as Powerpoint
• Download Figure

Male body size averaged (x̄ ± SD) 148.72 (± 26.49) mm in CL, whereas female body size averaged 111.01 (± 15.17) mm. Carapace length (Z = 6.14, p = 0.01) and PL (males = 130.12 ± 13.38 mm; females = 100.08 ± 14.79 mm; Z = 7.35, p = 0.006) differed significantly between sexes. Males and females did not differ in CH (males = 52.95 ± 7.99 mm; females = 44.70 ± 7.93 mm; Z = 2.90, p = 0.08); however, males (432.5 ± 199.56 g) were heavier (Z = 4.88, p = 0.02) than females (219 ± 91.75 g).

Home range size ranged from 0.08 to 2.76 ha when we used MCP and from 0.38 to 4.35 ha when the 50% kernel approximation was used (Table 1). The average home range of the females tracked (x̄ ± SD) was 0.80 (± 0.99) ha when we used MCP and 1.83 (± 1.44) ha when used 50% kernels. Tracked females moved on average 113.38 (± 142.10) m between location events, and their estimated daily movements averaged 5.78 (± 7.81) m/d. Table 2 shows movement of individual females tracked in the study site.

Table 1. Estimated home ranges and aestivation time (days out of the water) for Kinosternon integrum females tracked in San Jerónimo Purenchecuaro, Michoacán, México. Data were calculated for turtles with a minimum of 5 locations. MCP = minimum convex polygon.

View Fullscreen

Table 2. Movement of individual Kinosternon integrum females tracked in San Jerónimo Purenchecuaro, Michoacán, México.

View Fullscreen

Temporal autocorrelation values ranged from 0.5 to −0.23 and averaged 0.07. We only were able to compute the temporal autocorrelation for 6 individuals because of the limited number of relocations. We calculated Moran's I-values for spatial autocorrelation for 5 individuals only. Spatial autocorrelation values were also low, ranging from −0.25 to 0.011 and with an average of −0.10. No correlation values were statistically significant.

The tracked females used 10 recognizable aestivation microhabitats. The most used microhabitat was a rustic stone wall. We located turtles 12 times buried beside the wall, and located some turtles (≥ 6 times) that had slipped inside the stone wall. One turtle (no. 6) aestivated under an Agave sp. Even when the agave died during the first season (2016), the same turtle used the remains of the plant to aestivate during next dry season (2017). Turtle no. 12 also used the same hole in an old oak tree to aestivate during 2 sampling seasons. Other turtles used microhabitats under a dead Opuntia sp. tree, buried beside an Acacia sp. tree, buried in roots a few meters from one of the tanks, buried beside a small pile of rocks, and buried under fallen branches and leaf litter. No turtles were located in the lakeshore. The closest straight-line distance from a turtle location to the lakeshore was 560 m. Turtles remained in aestivation for 8–9 mo (229–279 d; Table 1), basically, from October or November (when cattle tanks hold minimum or no water) to July or August (when heavy rains fill the tanks and an intermittent stream runs downhill). Turtles spent an average of 254 (± 18.72) d out of water tanks.

Discussion

To our knowledge, this is the first record of the Mexican mud turtle in the Pátzcuaro Basin. Neither Martín del Campo (1940) nor Iverson (1982) reported the species as present in the lake or its surroundings. Here we report a small population, but one with an important demographic potential to increase owing to the following observations: we used X-rays of turtles to establish that the population is breeding (2 females with 4 eggs, and 1 female with 5 eggs, in oviduct), we observed recruitment of juvenile turtles into the population, and most of the adults are females. Both ponds are very small, so the carrying capacity of the habitat is unknown. However, in previous studies on the same species, slightly larger ponds (at lower altitude) could hold up to 200 individuals (Macip-Ríos et al. 2009, 2011).

The population phenology seems to be strongly associated with the hydroperiod of the intermittent stream and the cattle tanks. Rainy season begins in late May or early June (Garcia 2004); however, it is not until middle July to early August, when rains fall almost every day, that the cattle tanks are filled to capacity. The same phenological pattern has been described in other K. integrum populations (Macip-Ríos et al. 2009, 2011; Pérez-Pérez et al. 2017) and other kinosternids (Lingon and Stone 2003; Forero-Medina et al. 2007; Vázquez-Gómez et al. 2016).

Home ranges of tracked individuals were limited to the periphery of the tanks, but aestivation sites were associated with highly protected microhabitats like rustic stone walls, agave plants, and fallen logs. Also, we detected fidelity of turtles to aestivation microhabitats. Some turtles used the same aestivation microhabitat every sampled dry season. Among freshwater turtles, kinosternids have small home ranges (Greenspan et al. 2015; McCoard et al. 2016; Slavenko et al. 2016). Compared with the study of Pérez-Pérez et al. (2017), home ranges (MCP and kernel) from our Pátzcuaro study site are larger, but the range size overlaps. Compared with other kinosternids (see Pérez-Pérez et al. 2017, table 4), our results fall within the reported home range sizes and resemble the home ranges of K. flavescens (Mahmoud 1969), K. subrubrum (Mahmoud 1969), and K. acutum (Iverson and Vogt 2011). Compared with other turtles, our data follow predictions by Slavenko et al. (2016) of a correlation of body size with home range.

Turtles moved short distances among tanks, and to temporal pools from the stream, during the rainy season. However, our results contrast with those found by Pérez-Pérez et al. (2017), when the same species moved an even shorter distance (51.44 m on average). Larger freshwater turtles tend to move larger distances (Ryan et al. 2008, 2014), whereas others could move just kilometers (Bower et al. 2012). The studied basin population moves between tanks during rainy season. When water dries out by middle November, turtles move variable distances looking for aestivation microhabitats. It is interesting that none of the tracked turtles aestivated in the dry cattle tanks or buried in the remaining mud as expected, similar to other mud turtles (Lingon and Stone 2003).

We recognize that our home range data are limited in sample size compared with other studies, but our data generally agree with other estimations for the genus (Pérez-Pérez et al. 2017, table 5). Actually, our results indicate that K. integrum are among the top 3 species of kinosternids in terms of distance of movement, which agrees with the prediction of Slavenko et al. (2016) of a correlation between movement and body size. Temporal and spatial autocorrelation seem to have little or no effect in our results; however, the small and statistically insignificant estimated values could be due to the relocation frequency (every 2 wks during rainy season, and monthly during dry season).

One important finding from our research is the capacity of K. integrum to exploit man-made structures to survive, which includes their main aquatic habitats and most of their aestivation microhabitats. It makes sense to regard K. integrum as a good colonizer species if it can successfully exploit, survive, and breed using highly altered and modified habitats. K. integrum has also been reported in other altered habitats (Iverson 1999), and its populations have been described as abundant in other highly modified landscapes (Iverson 1999; Macip-Ríos et al. 2009, 2011; Pérez-Pérez et al. 2017).

The Mexican mud turtle has been reported in new basins during at least the past 40 yrs (Iverson 1982; Reyes-Velasco et al. 2013; Vogt and Legler 2013). It could have been introduced accidentally in the Valley of México and as a result of hydraulic infrastructure changes, such as those in the Magdalena valley in Michoacán (Iverson 1982). Other K. integrum “invasions” or introductions could even be tracked to the beginning or before of the 20th century, like those in Laguna Zapotlán, Jalisco (Gadow 2011). Nonetheless, based on our observations of aestivation capacity, movements, and reproductive potential, we suggest that Pátzcuaro Basin colonization for K. integrum could be natural. However, it is unknown if its presence could potentially affect the endemic K. h. tarascense. Further research is needed to understand basic ecological parameters and conservation consequences of the interaction of these 2 species.