Reproduction in the Red-Cheeked Mud Turtle (Kinosternon scorpioides cruentatum) in Southeastern Mexico and Belize, with Comparisons Across the Species Range
Abstract
Reproduction in the red-cheeked mud turtle (Kinosternon scorpioides cruentatum) was studied in southeastern Mexico and Belize (Yucatan Peninsula) on the basis of museum and field-collected specimens. Adult females (mean carapace length [CL] = 110 mm) were not significantly larger than adult males (mean CL = 108 mm). Females matured at 100–105 mm CL and an estimated age of 9–10 years. Annual reproduction by females was apparently continuous for at least 10 months of the year (August–June), with the production of multiple annual clutches (possibly as many as 5) being typical and the number of clutches per year increasing with female size. Eggs exhibit diapause and embryonic estivation and apparently hatch in nature during the wet season (June–August) after up to 9 months in the nest. Modal clutch size was only 2 eggs (mean 2.2; range 1–4), and clutch size increased with female body size. Egg size averaged 31.5 × 16.6 mm and 5.5 g and did not vary with female body size. Relative egg mass (REM = mean egg mass × 100/[body mass − clutch mass]) averaged 2.3, was lower in larger clutches, and decreased with increased body size. Relative clutch mass (RCM = REM × clutch size) averaged 4.85, similar to that for the sympatric K. creaseri (4.5), the 2 lowest values reported for any studied Kinosternon population. RCM did not vary with female body size or clutch size. Females devote a relatively constant proportion of body mass to each clutch, and increases in reproductive output with size and age are apparently accomplished by increases in clutch size (but not egg size) and clutch frequency. This unusual suite of reproductive traits (small body size, small clutch size, production of up to 5 clutches per year, reduced relative clutch mass, and embryonic diapause and estivation) may have been instrumental in the success of this species at colonizing more of South America than any other Mesoamerican turtle genus or species and in less than 4 million years.
Knowledge of the reproductive ecology of a species is an essential component of understanding its life history strategy (e.g., Stearns 1992) and also for its management and conservation (Hamann et al. 2010). For freshwater turtles, detailed reproductive studies have overwhelmingly been focused on temperate North American species (e.g., Gibbons 1990; Steyermark et al. 2008; Greaves and Litzgus 2009). Although this pattern is changing rapidly for tropical South America (e.g., Bager et al. 2007; Daza and Páez 2007; Batistella and Vogt 2008; Correa-H. et al. 2010; Ferreira Junior and Castro 2010), Mesoamerica (spanning 8 countries and over 2000 km) has been largely ignored since Moll and Legler's (1971) seminal study of the slider turtle (now Trachemys venusta) 40 years ago. This study was undertaken to begin to correct this deficiency.
With a geographic range extending from Tamaulipas in northeastern Mexico to northern Argentina (Berry 1978; Berry and Iverson 2001), the scorpion mud turtle (Kinosternon scorpioides), as currently recognized, has the greatest latitudinal distribution of any Neotropical freshwater or terrestrial turtle. Despite this, few data exist to document its reproductive biology. All published references on reproduction to date report anecdotal field data based on few specimens (i.e., Fretey 1976, 1977; Alvarez del Toro 1983; Métrailler and Le Gratiet 1996; Campbell 1998; Iverson 2008; Barreto et al. 2009) or captive animals, often of uncertain provenance (i.e., Sexton 1960; Lardie 1983; Rudloff 1986; Rocha and Molina 1990; Goode 1994; Hofer 1999; Schilde 2001; Nickl 2009). In preparation for a forthcoming review of the evolution of life history characteristics of all kinosternid turtles, I compiled reproductive data based on field and museum specimen dissections of red-cheeked mud turtles (K. s. cruentatum) from the Yucatan Peninsula in Mexico and adjacent Belize. Those data are summarized here and compared to reproductive data from across the extensive range of the species.
METHODS
This study was based primarily on dissections of specimens of K. s. cruentatum in the collection of the Florida Museum of Natural History (UF), including specimens collected on the Yucatan Peninsula during my fieldwork on 26–28 April 1981, 18–28 August 1985, and 14–21 August 1987 (see Iverson 1988). I also examined but did not dissect over 300 additional museum specimens (see Acknowledgments) of K. scorpioides from across the species' range (including 194 of K. s. cruentatum).
Maximum carapace length (CL; not necessarily measured on the midline) and maximum plastron length (PL) were recorded to the nearest millimeter for each specimen. For juveniles with distinct growth annuli (rings) on the plastral scutes, right interabdominal seam length (IAB) was also measured, as was the medial length of each annulus on the right abdominal scute (representing former measures of IAB length at previous times of growth cessation). Based on the assumption that only a single annulus was produced each year during the winter (December–January; the dry and coldest season), these data were used to reconstruct estimated growth trajectories for those juveniles following the method of Ernst et al. (1973).
Standard reproductive parameters (Moll and Legler 1971) were collected for all possible museum and field-collected specimens. For each dissected female (n = 21), I recorded the diameters of all ovarian follicles > 3 mm in each ovary, the number and diameter of any identifiable corpora lutea, and the length and width of all oviductal shelled eggs. Females with no corpora lutea, oviductal eggs, or enlarged yolked follicles > 3 mm in diameter were considered to be sexually immature (Iverson 1999; Hamann et al. 2003). Because yolk diameters from dissected eggs were ~ 16 mm and the largest ovarian follicles were 16 mm in diameter, follicles > 13 mm were considered preovulatory. Hence, females that exhibited follicles > 13 mm in diameter were considered mature even if they bore no corpora lutea or oviductal eggs. Three field-collected females were retained in captivity until they oviposited, and the length, width, and mass of each egg were measured. All measures of mass are in grams.
Male and female samples were compared with t-tests; relationships between variables were examined with least-squares regression analysis when both were continuous and Spearman rank correlation analysis when the dependent variable was discrete; departures of sex ratio from expected (1:1) was tested with chi-square analysis. Relative egg mass (REM = mean egg mass × 100/[body mass − clutch mass]) and relative clutch mass (RCM = clutch mass × 100/[body mass − clutch mass] = REM × clutch size) were calculated for each gravid female as estimates of reproductive effort. Means are followed by ± 1 standard deviation. All analyses were performed with Statview™ software.
RESULTS
Sex Ratio
Every geographic sample of K. s. cruentatum in Table 1 was female biased, although only that for the Mexican Pacific was significantly skewed. The combined sample of 207 adults also exhibited a female-biased ratio (1:41:1; χ2 = 5.92; p = 0.015).
Body Size and Growth
During fieldwork in August 1985, I collected 2 neonatal K. scorpioides (with no posthatching shell growth) on roads during rainstorms (along with many K. creaseri; Iverson 1988). Those hatchlings measured 25.7 mm CL and 22.6 mm PL and 27.8 mm CL and 23.9 mm PL, respectively. Five neonates collected on 18 June 1961 in Yucatan State (CU 16136, 16138–41) averaged 27.7 ± 4.1 mm CL (23.0–34.0) and 23.1 ± 3.0 mm PL (19.8–27.6 mm). Another neonate collected 3 June 1949 in Quintana Roo (MCZ 53117) measured 26.5 mm CL. Mean CL for all 8 neonates was 27.3 ± 3.2 mm. Plastron length averaged 84.4 ± 4.1% of CL for 7 of those neonates; however, for 18 specimens less than 90 mm CL, this proportion increased with body size (PL/CL = 0.002CL + 0.793; r = 0.87; p < 0.0001) to a fairly uniform value among adults and subadults. For 40 male specimens > 80 mm CL from the Yucatan Peninsula in museum or field collections, PL/CL averaged 0.95 ± 0.02 (0.91–1.01), and for 42 female specimens > 80 mm CL, PL/CL averaged 0.98 ± 0.02 (0.94–1.04).
The largest specimen of K. s. cruentatum examined from the Yucatan Peninsula was a female (141 mm CL); however, females were not significantly larger than males (2-tailed t = 0.99; p = 0.33; Table 1; see also Iverson 1988). Body mass (BM) for 7 turtles that I weighed on field capture plus 5 juveniles reported by Hofer (1999) was related to CL by the regression, BM = 0.000102CL3.108 (r = 0.996; p < 0.0001).
The 4 largest immature females were 98 mm CL (Yucatan; 97 mm PL), 98 mm CL (Campeche; 97 mm PL), 88 mm CL (Yucatan; 87 mm PL), and 86 mm CL (Quintana Roo; 85 mm PL). The 4 smallest gravid females were 104–109 mm CL (Table 2), and 2 other females 104–106 mm CL bore enlarged (13–14 mm) preovulatory follicles but no eggs or corpora lutea. An additional female (97 mm CL; Table 2) had a single enlarged 10-mm yolked follicle (and 2 others at 7 mm), suggesting that she might have reproduced later during the year of capture. Together these data suggest that female maturity in this region is reached at ~ 100–105 mm CL (~ 99–104 mm PL; 165–200 g BM). Plastral growth histories were estimated for 8 juveniles (Fig. 1) and suggest that sexual maturity is delayed to 9 or 10 years.
Citation: Chelonian Conservation and Biology 9, 2; 10.2744/CCB-0827.1
Ovarian Cycle
Although adult females in August bore oviductal eggs and/or preovulatory and other sets of enlarged follicles (Table 2), none bore corpora lutea from previous clutches. Since corpora lutea regress completely (macroscopically) within a few months in other kinosternids (e.g., Iverson 1978, 1979), this suggests that ovulation may cease for some time prior to August but that reproduction continues in subsequent months. November females (n = 2) were both gravid and had enlarged follicles, suggesting that reproduction continues through the winter; however, neither had identifiable sets of old corpora lutea from previous clutches. Both were specimens preserved in 1966, and hence I might have been unable to detect small regressing corpora lutea from earlier autumn clutches. Adult females in January typically bore eggs, corpora lutea from up to 2 previous clutches, and enlarged follicles indicative of 1 to 2 future clutches, further suggesting that reproduction continues through the fall and winter. Only a single adult female was available from the spring (May), and, although gravid, she had only small yolked follicles in her ovaries. All kinosternids studied to date exhibit a postreproductive season of ovarian recrudesence, even those with nearly yearlong nesting (Iverson 1978, 1979), and hence ovarian recrudesence in K. s. cruentatum probably occurs during June and July.
Although females for dissections were not available for every month and females were collected over a broad region on the Yucatan Peninsula (including Belize), these data suggest that ovulation may begin in August and continue until at least May and that the production of at least 2 to 4 clutches per year may be typical (Table 2). Indeed, the ovaries of 1 female from Quintana Roo (113 mm CL) had evidence of 3 previous clutches and 2 potential future clutches (Table 2). Based on the number of observed sets of corpora lutea/eggs and the number of sets of enlarged follicles (≥ 8 mm) reported in Table 2, annual clutch frequency was estimated to average 2.8 ± 1.3 clutches per year (range 1–5) and to increase with body size (Z = 1.98; p = 0.05; n = 14). Despite the estimated length of the nesting season, neonates have been collected in the field only during the summer wet season (June and August; see above).
Clutch Size
Clutch size (CS) based on the number of fresh corpora lutea and/or oviductal eggs averaged 2.22 ± 0.44 (range 2–3; n = 9), CS based on older sets of corpora lutea averaged 1.80 ± 0.84 (range 1–3; n = 5), and CS estimated from size classes of enlarged follicles > 9 mm diameter averaged 2.04 ± 0.88 (range 1–4; n = 23). Modal CS was 2 eggs (i.e., for 21 of all 37 clutch size estimates: 57%).
Despite little variation, CS was correlated with CL when based on fresh corpora lutea and/or oviductal eggs (Z = 2.1; p = 0.04; Fig. 2) or sets of enlarged follicles (Z = 2.0; p = 0.05). However, the relationship was not significant when based on older sets of corpora lutea (Z = −1.1; p = 0.27), presumably because of the small sample and because some regressing corpora lutea may have been overlooked.
Citation: Chelonian Conservation and Biology 9, 2; 10.2744/CCB-0827.1
Egg Size
Nineteen eggs dissected from or laid by females from the Yucatan Peninsula averaged 31.4 ± 1.8 mm in length (29.6–34.8 mm) and 16.9 ± 0.6 mm in width (15.8–17.5 mm); 8 of those weighed 5.41 ± 0.21 g (5.19–5.86 g). Based on a compilation of egg size data from this study and those of Castillo (1986) and M. Ewert (pers. comm.), representing females from southeastern Mexico to Costa Rica (subspecies cruentatum and albogulare), egg length (EL) was related to egg mass (EM) by the regression equation EM = 0.345EL − 4.863 (r = 0.79; p < 0.0001; n = 345), and egg width (EW) was related to EM by the equation EM = 0.932EW − 9.984 (r = 0.87; p < 0.0001; n = 342). Finally, EM was related to EL and EW by the regression equation EM = 0.220EL + 0.696EW − 13.141 (r = 0.98; p < 0.0001; n = 342). Based on the last equation, mean estimated egg mass for all 19 Yucatan eggs I obtained by dissection (but did not weigh) averaged 5.47 ± 0.55 (4.37–6.55; n = 19). Although there was a trend for mean egg width in a clutch to increase with CL (r = 0.59; p = 0.12), no measure of egg size (whether EL, EW, or estimated EM) was significantly correlated with CL. Insufficient data were available to test whether there was monthly variation in absolute or relative egg size or clutch size.
Ten females with CL, EL, and EW data were available to estimate relative egg mass (REM = mean EM × 100/[BM − CM]) for Yucatan turtles. Estimated REM averaged 2.28 ± 0.46 (range 1.57–2.96), was lower in 2 3-egg clutches than 8 2-egg clutches (1.64 ± 0.10 vs. 2.51 ± 0.25; t = 4.7; p = 0.002) and decreased with female body size (for CL, r = −0.81, p = 0.004; for BM, r = −0.83, p = 0.003).
Clutch Mass
Estimated clutch masses for 10 females (see above) averaged 11.85 ± 2.15 g (range 9.13–16.10 g), was greater in 3-egg clutches than 2-egg clutches (15.54 g vs. 11.08 g; t = 4.48; p = 0.002), and increased with body size (for both CL and BM, r = 0.90 and p = 0.0004). Relative clutch mass (RCM = REM × CS) was estimated for the same 10 males, averaged 4.85 ± 0.67 (range 4.42–5.91), did not differ between 2-egg and 3-egg clutches (t = 0.25; p = 0.81), and was not related to female body size (for CL, r = 0.19, p = 0.60; for BM, r = 0.20; p = 0.58).
DISCUSSION
Most authors have considered the taxonomic content of K. scorpioides to include material from Tamaulipas in northeastern Mexico to northern Argentina, including as many as 7 subspecies (e.g., Ernst and Barbour 1989; Iverson 1992). In a multivariate analysis of body-size standardized shell measurements, Berry (1978) identified 3 distinct phenetic groups within K. scorpioides. These corresponded to the recognized subspecies scorpioides (eastern Panama to Argentina), albogulare (western Panama to Honduras), and cruentatum plus abaxillare (Belize to Mexico), with geographic and phenetic intermediates between albogulare and cruentatum found along the Pacific Coast in Guatemala and El Salvador. More recent revisions (e.g., Cabrera and Colantonio 1997) have corroborated the recognition of only 4 subspecies (K. s. scorpioides, K. s. albogulare, K. s. cruentatum from Belize and Guatemala to Tamaulipas in Mexico, and K. s. abaxillare from the central valley of Chiapas in Mexico), and that taxonomy is followed here. However, without any formal analysis of geographic variation, some European turtle herpetoculturists have recently suggested that this taxon represents 2 species. For example, Schilde (2001) recognized K. cruentatum as a monotypic species (though he did allude to intraspecific variation) and K. scorpioides as a separate species, including abaxillare, albogulare, and scorpioides as subspecies. Given that the geographic range of abaxillare lies in the center of the range of cruentatum, this arrangement seems unfounded. In contrast, Artner (2003) recognized a monotypic K. scorpioides in South America and a polytypic K. cruentatum in Central America and Mexico (with 3 subspecies).
Of more immediate concern is the confusion in the literature (especially the husbandry literature) over the identification of K. s. cruentatum vs. K. s. albogulare. Although the former has long been referred to as the red-cheeked mud turtle (e.g., Iverson 1992; Artner 2003), many populations of the latter subspecies also include individuals with red markings on the head (Acuña-Mesén 1993; Leenders 2001; Savage 2002; Kohler 2003; McCranie et al. 2006). This has led to considerable confusion, particularly when specimens have originated in the pet trade (usually exported from “Honduras”) under the name “red-cheeked mud turtle.” For example, although the papers by Goode (1991, 1994) contain valuable captive reproductive data, they refer to the name K. s. cruentatum even though at least some specimens apparently originated in Honduras (Goode 1994). Because the data on at least body size, growth, clutch size, and egg size in those papers differ significantly from those for K. s. cruentatum from the Yucatan and Belize, I tentatively associate them with K. s. albogulare. Other possible examples of taxonomic confusion in the literature are noted below. Molecular phylogenetic work now in progress (M. Le et al., pers. comm.) is intended to address this taxonomic problem.
The physical, climatic, botanic, and herpetofaunal characteristics of the Yucatan Peninsula (including Belize and northern Guatemala) have been thoroughly reviewed by Lee (1980, 1996, 2000), Campbell (1998), and Stafford and Meyer (2000). This is a region characterized by a low, karst topography, dominated by tropical seasonal dry forest, and experiencing a distinct dry season lasting from December through April, followed by a June–November monsoon season during which about 72% of the annual 207 cm of rain falls (e.g., Cancun 1971–2000, World Meteorological Organization: http://www.wmo.int/pages/prog/wcp/cop16wmo.php). Seasonal shallow wetlands are abundant during the wet season, but many of these hold no water during part or all of the dry season, forcing K. scorpioides to seek more permanent aquatic sites (e.g., solution holes or perennial ponds and lakes; pers. obs.) or to estivate beneath soil or vegetation (Duellman 1965; Dean 1980; Legler 1990; Cabrera 1998; Stafford and Meyer 2000). For example, Dean (1980) reported that although activity in K. s. cruentatum in coastal Chiapas is reduced during the dry season, not all individuals estivate for its entirety. Given that this species has been collected on the Yucatan Peninsula during every month of the year but September (based on museum records), it is probable that activity coincides with the availability of standing water.
Sex Ratio
Sex ratios of K. s. cruentatum and K. s. abaxillare are consistently female biased in field and museum collections (Table 1), and the only well-sampled population of K. s. albogulare shows the same pattern (1.97 females per male; Forero-Medina et al. 2007). However, samples of K. s. scorpioides from South America reviewed in Table 1 show no such trend. It is known that this species exhibits temperature-dependent sex determination (Ewert and Nelson 1991), but interpreting the significance of these sex ratio patterns is not yet possible.
Size and Growth
Measurements of hatchling K. s. cruentatum produced in captivity (29.8 mm CL, 3.7 g, n = 3, Ewert 1979; 26.8, 28.0, and 28.5 mm CL, Lardie 1983; and 27.0 mm CL, 2.6 g, Rudloff 1986) are similar to those from field-collected specimens (mean, 27.3 mm CL). In contrast, 67 captive hatchlings (probably representing K. s. albogulare) produced by Goode (1994) averaged 33.0 mm CL (range 27–36 mm). Hatchlings of K. s. scorpioides from South America (27 mm CL, French Guiana, Métrailler and Le Gratiet 1996; 28.5 mm CL, 5.3 g, Brazil, Rocha and Molina 1990; and 30 mm CL, Argentina, Cei 1993) were similar to those of K. s. cruentatum. However, the report of 2 hatchlings of K. scorpioides from Costa Rica at 22 and 15 g (Acuña-Mesén 1993, citing Castillo-Centeno 1986) was clearly erroneous. In addition, Campbell's (1998) statement that hatchling K. s. cruentatum are 35–36 mm CL is suspect.
Populations of K. s. cruentatum along the Pacific Coast of Mexico are similar to those on the Yucatan Peninsula and Belize in adult body size, although they differ in exhibiting a slightly female-biased sexual size dimorphism (Table 1). However, populations in the Tampico embayment and the Veracruz lowlands appear to reach larger body sizes even though they apparently lack size dimorphism (Table 1). Kinosternon s. abaxillare exhibits similar body sizes to the latter populations and also lacks sexual size dimorphism. Both K. s. albogulare and K. s. scorpioides grow to larger body sizes than Yucatan populations and show a weak tendency for males to be larger than females (Table 1). However, mean male size relative to female size by population did not increase with mean overall body size (r = 0.05; p = 0.90; n = 12), although that positive correlation has been identified across species of kinosternid turtles (Iverson 1991b). Lee (1996) reported that adult K. s. cruentatum on the Yucatan Peninsula averaged between 155 and 175 mm CL, but he must have been referring to literature data for more southern subspecies; I could find no Yucatan adult larger than 141 mm CL in any collection (Table 1).
Despite differences in general shell shape (more highly domed in Central than South America; Berry 1978), BM of 8 K. s. scorpioides from French Guiana (Métrailler and Le Gratiet 1996) was very similar to that estimated for K. s. cruentatum based on the BM–CL regression. Inclusion of those data produced a highly significant regression for the species: BM = 0.000404CL2.783 (r = 0.99; p < 0.0001; n = 20).
Although Yucatan females of K. s. cruentatum reach maturity at 100–105 mm CL (see also Campbell 1998) and perhaps 9–10 years of age, a female of 125 mm CL that I collected in April 1981 in Tamaulipas in northeastern Mexico had no ovarian follicles larger than 3 mm in diameter (and no corpora lutea) and appeared to be immature, whereas another from the same site (143 mm CL) had 3 near-ovulatory follicles (14–16 mm) and 3 others 9–10 mm in diameter. Though anecdotal, these data suggest that maturity may be at larger sizes in the Tampico embayment (> 125 mm CL) than on the Yucatan Peninsula (100 mm CL). Given that adults from the former region apparently reach larger body sizes (Table 1), this is to be expected. Furthermore, Goode (1994) reported maturity in his captives (presumably K. s. albogulare from Honduras) at 122–132 mm CL and 5.3 years of age. His adults also reached 154 mm CL, much larger than over 100 Yucatan adults I have examined.
The available data on juvenile growth in K. s. cruentatum are meager and based only on plastral annuli (and associated assumptions; Fig. 1). Captive growth rates for 3 hatchlings (Lardie 1983; 31 and 33 mm PL at age 0.5 years, 33 mm PL at 1 year) were even slower than my data suggested. On the contrary, growth rates reported by Goode (1991, 1994) and based on plastral annuli for captives and wild-caught specimens, presumably from Honduras, were much faster (estimated PL for ages 1, 2, 3, and 4 were 65, 94, 105, and 128 mm PL, respectively; compare Fig. 1) Similarly, growth rates reported for K. s. scorpioides from Venezuela by Pritchard and Trebbau (1984) were also apparently much faster; 5 unsexed specimens estimated to be 5 to 6.5 years old were 102–135 mm CL (estimated PL range, 98–130 mm).
Ovarian Cycle
A number of authors have noted the protracted nesting season of K. scorpioides (Table 3). For K. s. cruentatum and K. s. albogulare, nesting may begin as early as August and extend 9–10 months to May or June. A breeding colony of 5–10 adult females (presumably from Honduras) was studied by Goode (1994) over a 7-year period; 154 clutches were laid over a 312-day annual period from August to June, a period nearly identical to that suggested here for Yucatan K. s. cruentatum; 94% of those clutches were laid over a shorter 212-day period from October to May (Goode 1994). Corroboration of this long reproductive season is provided by a female (128 mm CL) collected 30 December 1976 in northwest Honduras that in January 1977 contained oviductal eggs, 1 set of corpora lutea, 4 preovulatory follicles (14–15 mm), and 3 other enlarged follicles (12, 11, and 6 mm). These data suggest that oviposition may span 7–10 months of the year in populations of K. scorpioides in at least northern Mesoamerica and that reproduction is suspended during the summer portion of the wet season when energy resources are presumably most plentiful.
Preliminary data suggest that K. scorpioides from northern South America may also have an extended nesting season (Table 3). For example, Barreto et al. (2009) found gravid females in August (early dry season) in northeastern Brazil, and Métrailler and Le Gratiet (1996) reported eggs laid in September (early dry season) in French Guiana. In addition, a female K. s. scorpioides (153 mm CL) collected in Venezuela (Edo. Portuguesa) in September (late wet season) 1974 contained 5 oviductal eggs (Table 3), 6 fresh corpora lutea, an atretic 11-mm follicle free in the coelom, 5 preovulatory follicles (13–15 mm), 5 enlarged follicles (8–10 mm), and 4 smaller follicles (6–7 mm). These data suggest an extended autumn reproductive season in northern South America (at least from the late wet season well into the dry season). The extent of the nesting season at the southern end of the species' range in southern South America also deserves attention given that the timing of the wet and dry seasons there is the reverse of that in Mesomerica. For example, Rocha and Molina (1990) reported that nesting in captivity in southeast Brazil occurred from March to August (i.e., late wet season into mid-dry season; autumn and winter there).
The production of multiple annual clutches by individual K. scorpioides in captivity has been previously reported by Lardie (1983), Goode (1994), Hofer (1999), Schilde (2001), and Nickl (2009), but my dissections are the first confirmation of this in the field (based on multiple sets of corpora lutea). Internesting intervals in captivity were generally 1–2 months, although Hofer (1999) reported a female that produced 2 clutches of 4 eggs only 16 days apart. Goode (1994) reported minimum internesting intervals of 32 and 35 days for presumably Honduran females, but he did not always monitor individual females and hence could not quantify typical intervals. Lardie (1983) recorded an internesting interval in captivity of 51 days for a female K. s. cruentatum of unknown origin, and Schilde (2001) suggested that intervals as short as 4 weeks are possible for captive K. s. albogulare from Nicaragua. Even if intervals averaged 2 months, the production of 5 annual clutches (as speculated above for the 113-mm CL Quintana Roo female) is possible. The preliminary data examined here provide no evidence that individual mature females forgo reproduction in a given year.
Once his breeding colony was established, Goode (1994) found that mean clutch frequency varied across years (from 1.63 to 3.25) and generally increased with increasing size and age of the breeding females. Unfortunately, he did not follow clutch frequency in individual females, so maximum clutch frequency was not reported. However, given the length of the nesting season, annual clutch frequency for K. scorpioides in Mesoamerica may be the highest of all studied kinosternids (review in Iverson 1999).
It is not known whether females of K. scorpioides are capable of holding viable oviductal eggs for long periods, that is, until suitable nesting conditions prevail (e.g., after a rain event). Furthermore, copulation has never been observed in the field, although Sexton (1960) recorded mating in captivity in October and November, and R. Lardie (pers. comm.) witnessed repeated mating from 19 to 23 September. Whether sequential clutches in K. scorpioides are fertilized by stored sperm or repeated matings also deserves study.
Numerous reports of incubation times for the eggs of K. scorpioides in captivity are available (Table 4), ranging from 78 to 266 days, and do not exhibit the expected relationship with incubation temperature (Ewert 1979). This substantial variation is presumably the result of 2 specialized developmental mechanisms exhibited by the embryos of K. scorpioides (Ewert 1985, 1991). First, embryos undergo a period of diapause soon after oviposition, when development is arrested, later to be broken by a chilling event. Second, full-term embryos also enter a period of dormancy when metabolic rates fall, and the embryo estivates until sufficient moisture (and, to a lesser degree, cold or warm temperature pulses) breaks that dormancy. The turtle then pips the egg shell and emerges (Ewert 1991). Given the long reproductive season, during which conditions are likely suboptimal for hatchling turtles, these developmental mechanisms apparently function to synchonize hatching during the summer rainy season, no matter when during the previous 10 months the eggs were laid (e.g., Nickl 2009). These adaptations might explain why neonates have only been field collected on the Yucatan in June and August. Hatching in captivity has also been observed during July and August (Lardie 1983) and June–July (Nickl 2009) but also during November (Hofer 1999), presumably because of artificial incubation conditions that prematurely interrupted those developmental delays. The timing of hatching in K. scorpioides from South America is less clear. Métrailler and Le Gratiet (1996) reported a clutch (laid in French Guiana in September) that hatched in February during the rainy season, and Barreto et al. (2009) captured 3 posthatchlings early in the rainy season in northeastern Brazil in January and February.
A greatly extended nesting season spanning most of the dry season also characterizes the other 3 Mesoamerican kinosternids that have been studied (K. acutum, K. creaseri, and K. leucostomum; reviewed by Morales-Verdeja and Vogt 1997; Iverson 1999). Each of these (including K. scorpioides) is also able to nearly fully close the plastron to the carapace (Iverson 1991b). The other 2 Mesoamerican species (K. angustipons and K. dunni) have a reduced plastron and are apparently more aquatic (Medem 1962; Legler 1965) and perhaps less affected by seasonal rainfall patterns. Anecdotal evidence suggests that the latter 2 species may reproduce during the summer when K. scorpioides (in Central America) is nonreproductive; Legler (1960) reported a gravid female K. angustipons found in July, and Medem (1961, 1962) reported a gravid female K. dunni in May. These 2 species remain the most poorly known of all kinosternids, and reproductive studies are sorely needed.
Clutch Size
As the subspecies with the smallest adult body size, K. s. cruentatum also produces the smallest clutches, typically 1–3 eggs (Tables 2 and 3), with 2 eggs being modal. Given that maximum CS otherwise reported across the entire range of K. scorpioides is 8 eggs, the report by Alvarez del Toro (1983) that red-cheeked mud turtles lay 10 eggs is probably in error. It is possible that these records were based on counts of enlarged ovarian follicles and hence represented more than 1 clutch. The general pattern for K. scorpioides, both within and among populations, seems to be one of increasing clutch size with increases in body size (Goode 1994; Table 3). There is no consistent pattern of seasonal variation in clutch size (Goode 1994).
Egg Size
Based on a compilation of egg size data from this study, Castillo (1986), and M. A. Ewert (pers. comm.), representing females from southeastern Mexico to Costa Rica (subspecies cruentatum and albogulare, though mostly the latter), average egg length (EL) was 33.58 ± 2.57 (24.2–39.7 mm; n = 348), average egg width (EW) was 17.93 ± 1.05 (14.8–20.1 mm; n = 345), and average egg mass (EM) was 6.72 ± 1.13 (3.82–9.45 mm; n = 345). Although most reports of egg size in K. scorpioides with known country of origin are based on small sample sizes (Table 3), a general pattern of increasing egg size from north to south seems evident. This pattern has previously been identified across turtle species as well as within many (but not all) species (Moll 1979; Iverson et al. 1993). However, the collection of many more data is needed to confirm this pattern for K. scorpioides. Further work will also be necessary to explain the variation in egg size across years observed in captivity by Goode (1994).
Relative egg mass (REM = [mean EM × 100]/[BM − CM]) could be estimated for only 5 populations of K. scorpioides (Table 5). Despite the very small sample sizes, REM (the relative investment a female makes in each egg) tends to decrease with decreasing latitude (though no data are available for south of the equator). Because this is the reverse of the general pattern previously reported for turtles (e.g., Moll 1979; Iverson et al. 1993), it deserves further study. Among all kinosternids for which REM data are available, typical values are 2.4 to 3.1 (Iverson 1991a, 1999; Iverson et al. 1991). Only K. integrum (1.1; Iverson 1999), K. arizonense (1.4; Iverson 1989), and K. flavescens (1.7; Iverson 1991a) exhibit smaller values than K. scorpioides (2.3).
Clutch Mass
Relative clutch mass (RCM = REM × CS) was estimated for 5 populations of K. scorpioides across part of its range (Table 5), but no geographic pattern was evident. However, among other kinosternids studied to date, mean RCM is 6.0 (Iverson 1991a, 1999; Iverson et al. 1991), and only that for Yucatan K. creaseri (4.5; Iverson 1988) is less than that for Yucatan K. s. cruentatum (4.85). This corroborates the general geographic pattern across kinosternid species of reduced RCM with decreased latitude (Iverson 1999).
In summary, in comparison to most other kinosternids, K. scorpioides produces very few, relatively small eggs, with low total investment per clutch, but may typically produce more clutches per year than most other kinosternids (Iverson 1999). This combination of life history traits appears to be an adaptation to the warm, seasonal tropics, presumably rich in predator diversity but where year-round activity may sometimes be possible (if only after rain events during the dry season). The exploitation of rain events during the dry season for nesting (and possibly some feeding), coupled with mechanisms of developmental delay that synchronize hatching to the rainy season, apparently allows this species to take maximum advantage of the bounty of the wet season. During the latter, food intake is presumably not limited by abdominal space occupied by shelled eggs, and hence females may maximize both growth and storage of energy for future reproduction. This pattern also allows hatchlings to emerge under optimal conditions for growth and survival, allowing them to double their length in their first year (Fig. 1). Furthermore, the production of many, relatively small (in number and mass) clutches likely counteracts the presumed high probability of clutch depredation in this tropical environment. This strategy differs significantly from that of the other widespread tropical kinosternid, K. leucostomum, which produces clutches of 1–2 relatively large eggs (Moll 1979:329 [note figure legends 8 and 9 are reversed]; Moll and Legler 1971; Török 2009). The complex reproductive and behavioral strategy of K. scorpioides may have been the key adaptive suite that allowed this single kinosternid species to colonize the majority of the Atlantic versant of South America after the closure of the Central American portal only about 4 million years ago (Iturralde-Vinent and MacPhee 1999).
Estimated plastral growth in mm for 7 juvenile Kinosternon scorpioides cruentatum on the Yucatan Peninsula based on plastron lengths reconstructed from plastral annuli (see text for method). Note that age is 1 year less than annulus number.
The relationship between maximum carapace length (CL) and clutch size (CS) as estimated from fresh corpora lutea and oviductal eggs, older sets of corpora lutea, and size class sets of enlarged follicles. Larger symbols indicate coincidence of data points. Note that some females are represented by more than 1 type of clutch size estimate. The least-squares regression line for counts of fresh corpora lutea is plotted: CS = 0.055CL − 3.99 (r = 0.93; p = 0.0003; see text).
