The Phenology of Brumation Emergence in the Yellow Mud Turtle (Kinosternon flavescens) in Nebraska
Abstract
Spring emergence from brumation has been studied in several terrestrial turtles, particularly box turtles (Terrapene spp.) and tortoises (Gopherus spp.). However, most research involved small samples (often captive), excluded juveniles, lasted only 1 or a few years, and/or used only estimates of emergence dates. We quantified brumation emergence dates for 14,668 Yellow Mud Turtles (Kinosternon flavescens) captured from a population in the Sandhills of Nebraska between 1983 and 2018. Average spring emergence date over 9 yr for 1961 adult (≥ 85 mm carapace length) males was 1 May and for 2882 females was 5 May. For 5985 juveniles and 3840 hatchlings it was each 21 May over 12 yr. However, these means varied by about 3 wk among years, and emergence seasons for each cohort were typically about 40 d in length. Average emergence date across years for adult males and females was related primarily to spring temperatures. Predicted emergence for adults began when mean minimum and maximum temperatures over the past 2 d reached approximately –6.0°C and 7.0°C, respectively. Average dates for juveniles and hatchlings were not correlated with general patterns of spring weather; however, the combination of warm temperatures for several days with rainfall stimulated the most emergence. Emergence date was inversely related to body size and age. Females that nested in a respective year of emergence had better body condition, but did not emerge earlier than those that did not. Although spring emergence was influenced by temperature and moisture, precise timing varied primarily by individual and year.
In temperate climates, winter conditions preclude general activity in ectotherms such as turtles (Ultsch 2006). As winter approaches, chelonians in these regions seek microhabitats that minimize physiological stress, reduce access by predators, and allow their body temperatures to fall to match ambient temperatures. This overwintering behavior (or winter dormancy) is herein termed brumation (after Mayhew 1965; Holden et al. 2021) to distinguish it from the more inclusive term hibernation, which is commonly applied to several temperate mammals that reduce activity in the winter, but still often maintain body temperatures above ambient (Dubiner et al. 2023).
Although the physiology of brumating turtles has been well studied (e.g., reviews in Ultsch 1989, 2006), the cues used by them to emerge in the spring have been much less well studied. For example, our understanding of the timing of spring emergence from brumation by turtles in temperate regions has been based primarily on anecdotal observations (e.g., Wetmore 1920), small samples of short duration (e.g., Doroff and Keith 1990; Harris et al. 2015), captive individuals (e.g., Grobman 1990), or only for adults (most studies), and were focused mainly on terrestrial box turtles of the genus Terrapene (reviews in DeGregorio et al. 2017 and Woodley 2013; also Roe and Bayles 2021 and Roe et al. 2023) or tortoises of the genus Gopherus (e.g., Nussear et al. 2007; Harris et al 2015; Averill-Murray et al. 2018) (Table S1). This pattern is not surprising given that it is extremely difficult to precisely monitor brumation emergence in most aquatic turtles, and even harder to do that simultaneously for a large sample (apart from hatchlings that overwintered in the nest; e.g., see Baker et al. 2010, 2013). This has impeded our ability to isolate the precise cues used by turtles to time their emergence from winter dormancy.
There has been much speculation about these cues, although most studies have simply concluded that temperature and/or moisture are involved. Hence, precise cues, based on large samples and over multiple years, are rarely quantified (but see DeGregorio et al. 2017 or Averill-Murray et al. 2018). Some researchers have claimed that the primary cue for terrestrial brumators is temperature, generally) when some threshold is reached (e.g., Grobman 1990 for Terrapene) or when temperatures in the shallow soil become inverted relative to deeper temperatures as spring progresses (e.g., Currylow et al. 2013 for Terrapene carolina). Others have argued that the primary cue is rainfall (e.g., LaDuc and Christiansen 2012) or soil moisture (Legler 1960). Still others believed it is a combination of these factors (e.g., Fitch 1956; Long 1986).
Furthermore, although adults of several species have been reasonably well studied (Table S1), data are rarely available by sex or include juveniles. Only a single paper has examined the effect of body size on emergence timing in turtles (Roe and Bayles 2021), and only 1 paper has examined the effects of reproductive condition on timing (Averill-Murray et al. 2018). Only 2 have examined emergence in individual turtles over time (DeGregorio et al. 2017; Roe and Bayles 2021). In contrast, our study quantified all these factors for a very large sample of turtles in Nebraska under study for 39 yr.
The Yellow Mud Turtle (YMT; Kinosternon flavescens) is a small (generally < 15 cm), diurnal, semiaquatic kinosternid species that spends most of its short activity season in lentic wetlands (often ephemeral), but passes most of the year terrestrially in fossorial activities (nesting, estivation, and brumation; Christiansen et al. 1985; Iverson 1990; Tuma 2006; Iverson and Christiansen 2023). An isolated population of YMTs in the Sandhills of western Nebraska was the subject of our research from 1981 to 2019. Intensive sampling using drift fences produced 28,283 total captures of YMTs, reproductive data from over 4984 females, and 14,668 captures of turtles as they emerged from brumation over 12 of our 39 study years. The latter data set forms the basis of this study, and our goals were to quantify emergence timing by sex, size, reproductive state, body condition, and year, and to examine individual variability and the climatic correlates of those emergence patterns.
Specifically, we predicted that 1) emergence would be earlier in warmer years (after Janzen et al. 2018); 2) males would emerge earlier than females (to ensure their access to as many females as possible for mating) (Ultsch 2006, p. 347); 3) reproductive adult females would emerge before nonreproductive adult females (to ensure mating and the final development of the clutch) and have better body condition (Averill-Murray et al. 2018); 4) individual adults would be consistent in their timing of emergence (e.g., usually early or late; DeGregorio et al 2017; Roe and Bayles 2021); and 5) adults would emerge before juveniles, and juveniles before overwintered hatchlings (i.e., emergence related to body size and age; Blouin-Demers et al. 2000).
METHODS
Field Site. —
We monitored YMTs as they emerged from brumation and moved to the Gimlet Lake wetland complex (lat 41°45.24’N, long 102°26.12’W; Fig. 1) on the Crescent Lake National Wildlife Refuge, Garden County, Nebraska, between 1983 and 2019 (see Gunderson 1973; Weaver 1965; Iverson 1991; and Iverson and Smith 1993 for study site descriptions). The Gimlet Lake complex supports 3 subpopulations of YMTs (Iverson 1991). These 3 subpopulations are associated with separate, upland, generally south-facing sandhills in proximity to the wetlands (Fig. S1; all supplemental material is available at http://dx.doi.org/10.2744/CCB-1622.1.s1).
Citation: Chelonian Conservation and Biology: Celebrating 25 Years as the World's Turtle and Tortoise Journal 24, 1; 10.2744/CCB-1622
In 1981 a continuous ca. 750-m drift fence (25-cm aluminum coil stock) was constructed parallel to the northeastern shore of Gimlet Lake (“Main Fence”; Figs. S1–S3), but that fence was extended another 150 m in 1986 to census the subpopulation of turtles in an adjacent ephemeral pond and seasonal marsh southeast of Gimlet Lake (“Mallard Arm subpopulation”; Fig. S1). These fences were positioned to intercept turtles using the sandhill ridge to the northeast (Figs. S1, S3). Finally, in 1990, an additional 300-m fence (“Dike fence”) was constructed along the northwestern shore of Gimlet Lake parallel to the wetland shoreline to intercept the subpopulation of mud turtles that used the adjacent short sandhill ridge for nesting, estivation, and brumation (Fig. S1). Perennial cold (ca. 12°C) artesian springs emerging from the sandhill along this shore kept at least the shoreline habitat there cooler than for the other 2 subpopulations.
Because YMTs exhibit migration path fidelity at this site (Iverson 1991; Iverson et al. 2009; see also Tuma 2006), and require upland sites for estivation and brumation, especially south-facing (see Fig. S3), relatively few turtles at the Gimlet complex had migration paths beyond our fences. In addition, no YMT was ever found to shift among subpopulations, except in the area of an overflow berm where some turtles could move between the southeast end of Gimlet Lake and Mallard Arm. In addition, when in Gimlet Lake, turtles from either of the 2 subpopulations therein could occur in microsympatry. Hence, we have no reason to believe that there is a lack of gene flow among the subpopulations, despite some differences in life-history traits (unpublished) and differences in the migratory paths.
Although fieldwork at this site was ongoing from 1981 through 2019, we only monitored spring emergence at the fences in 1983, 1986, 1988, 1990, 1993, 1994, 1998, 1999, 2000, 2006, 2007, and 2018. Monitoring began in late March or early April prior to the first mud turtle emergence and continued at least until early June (to early July if nesting season was also monitored). In addition, because our arrival in 1998, 1999, and 2006 was delayed until late April, prior to the emergence of juveniles and hatchlings (but during adult emergence), emergence data for adults in those years were incomplete and not reported herein.
Capture Methods. —
Bucket traps (3 to 19 l) were buried along both sides of the fences at approximately 15-m intervals, and shaded, 12-mm wire mesh funnel traps were placed along both sides of the fences at ca. 8-m intervals (Fig. S2). This combination was necessary because adult turtles quickly learned to walk around sunken buckets without falling in. Once erected, 1 to 4 persons walked the fences continuously during all periods of potential turtle activity (generally 0800 to 1200 hrs and 1600 to 2100 hrs). All fences were removed (and buckets covered) when the field season ended, and then reconstructed each year they were used.
Because emerging turtles move directly to the wetland in the spring (generally less than 100 m from brumation sites), they were usually captured the day they emerge from brumation. However, some smaller juveniles (including hatchlings) may have taken more than a day to reach the fence during inclement weather or in areas where dense meadow vegetation was present. Nevertheless, we assumed that each capture in the spring on the hill side of a fence was a turtle that emerged that day.
For each capture we measured the maximum carapace length (CL in mm) and maximum plastron length (PL in mm) of the turtle (± 0.5 mm) with dial calipers in a plane parallel to the plastron (following Cagle 1946; see Iverson and Lewis 2018) and weighed each turtle (± 1 g) with PesolaTM scales. Carapace length for a few turtles with a malformed carapace was estimated from the ratio of PL/CL for adult males (mean = 0.932 ± 0.029, n = 2341) and adult females (mean = 1.000 ± 0.023, n = 3852). Turtles were individually marked with marginal scute notches following Cagle (1939) and/or notches along the margin of the plastral scutes. When possible, turtles were sexed based on external dimorphic characters (elongate tails and clasping organs in males; Ernst and Lovich 2009). All turtles ≥ 85 mm CL were easily sexed as male or female and were classified as adults for analyses reported herein. Turtles were classified as juveniles if they were larger than hatchlings and < 85 mm CL, the approximate size of the smallest of 4984 gravid females in this study (84-, 84.5-, and 85-mm CL).
Age at first capture (in winters posthatching) was estimated by counts of abdominal scute annuli (Iverson 2022). Annuli are very obvious for the first 10 yrs or so of life, and estimable in some cases up to about 20 yrs. Turtles having at least 20 annuli at first capture were considered to be at least 21 winters old. Age at subsequent recapture was based on this initial age assessment. Since most turtles in this study were marked as juveniles or subadults, we are confident of the accuracy of these ages ± 2 yrs.
Weather Data. —
Weather data were recorded at a NOAA weather station just north of Gimlet Lake (100 m north of bucket M-1 in Fig. S1), between 120 and 1000 m from the ends of the drift fences, and were available since 1970. Due to equipment failure in this station from 1 to 4 July 1988, daily maximum and minimum temperatures for those days were calculated from the Scottsbluff NOAA station (ca. 100 km due west). Degree-days (DD) were calculated as the difference in actual daily maximum temperature and 15.6°C (60° F), but only for days that reached above 15.6°C. Hence, a day with a maximum temperature of 25.6°C would have accumulated 10.0 DD. This threshold reflects the general activity of mud turtles at this site (Iverson, unpublished data).
Mean Emergence Timing. —
For analyses of mean emergence timing across years we examined the following spring climate variables for correlation (Table S2): April rain, March DD, April DD, 15 March–30 April DD, 15 March–15 May DD, mean April maximum temperature, and May DD and May rain (the last 2 for juveniles and hatchlings only). These relationships were examined separately for males, females, juveniles, and hatchlings using least squares regression analyses. Sample comparisons were made by 1-way ANOVA or Student t-test using Statview software (Abacus Concepts). Means are followed by ± 1 SD. Means of means are reported without variance measures (except range). Although data from many turtles were included in multiple samples across years (i.e., autocorrelated), our robust sample size minimizes that potential bias. In addition, analyses were repeated within a year to avoid this problem.
Data Selection.
Preliminary analyses (our mixed model approach described below), with CL as a fixed effect to account for body size differences on emergence, and year and individual ID as random effects (except for hatchlings, which excluded CL and Individual ID) indicated slight differences in mean emergence dates among the 3 fences (Table S3). We evaluated our models with and without fence as a random effect, but the remaining residual variance did not have a meaningful change with fence included vs. without. Additionally, fence variance was accounted for in the variance of individual ID as a random effect due to the consistent migration path of each individual (i.e., each individual was nearly always captured along same section of fence year after year; Iverson et al. 2009). Therefore, we combined data across fences for all cohorts, but report timing differences between cohorts and fences.
One problem with simply scoring the number of emergences on a given date is that after each day there are fewer turtles left to emerge. Thus, a day in late May with perfect emergence conditions might result in a relatively small number of emergences but represent a large proportion of the turtles left to emerge (e.g., see Blouin-Demers et al. 2000). Therefore, we also calculated the number that emerged as a percentage of the number left to emerge (e.g., Table S4) to identify the peaks of emergence. However, even this approach is biased at the end of the season (e.g., the last emergence gets scored as 100%). Nevertheless, peaks identified by this method mirrored those based on raw daily numbers.
Climatic Effects on Emergence.
We evaluated climatic effects on the number of individuals that emerged each day using linear mixed effect models via maximum likelihood in R using the package “lme4” (Bates et al. 2015; R Core Team 2024). Preliminary analyses indicated no differences in emergence between male and female juveniles; therefore, we pooled sexes in our juvenile data set, and then conducted analyses separately for each cohort. For each year, we considered the first day of the season to be the first day an individual was encountered, and the last day to be the day the last individual was found on the hill side of a fence heading to the wetland. We included zeros for days when no turtles were captured to identify conditions that influenced emergence. Our predictor variables included rainfall (mm) during the previous 24 hrs, rainfall in the previous 48 hrs, maximum and minimum daily temperature (°C) on the day of capture, and mean daily maximum and minimum temperatures over the previous 48 hrs. To improve model convergence and determine relationships with emergence dates, we z-standardized continuous covariates.
First, we examined relationships among our temperature covariates (i.e., temperature variables over the last 24 hrs vs. mean temperature variables in the last 48 hrs) and dropped 1 of 2 variables if their Pearson’s correlation coefficient was > |0.70|, retaining the variable that had the lower AIC value in single-predictor models. Then we fit a global model with the remaining climatic predictor variables as fixed effects, and year as a random effect to account for unidentified variance shared among individuals. We included temperature variables as linear responses to account for the increasing temperatures as a season progressed and as quadratics to account for the combined effect of warmer temperatures and a decreasing number of individuals available to emerge towards the end of a nesting season. We used backward selection to remove nonsignificant variables until all variables were significant. We determined the covariate’s predictive importance by inspecting conditional beta coefficient (β) estimates and their 95% confidence intervals (CIs), with significance defined as CIs for a variable that did not include zero. We used unstandardized coefficients in linear mixed effect models (i.e., with random effects) to predict emergence using the “ggeffects” package (Lüdecke 2018) in R.
Individual Variation on Emergence. —
To test for emergence patterns in individuals (i.e., consistently early or late), we culled our data set to include only adults with emergence data for at least 6 (males) or 7 (females) yrs. For each individual, we calculated the mean emergence deviation (in days) across years of each emergence from the overall mean emergence date for the appropriate sex and year. Variation in these deviations was examined graphically (by sex) to determine whether it tended toward a narrow, normal distribution (i.e., no individual propensity for early or late emergence) or a broad, uniform distribution (suggesting individual “preference”).
In addition, because larger turtles tended to emerge earlier than smaller ones (see below), we expected an ontogenetic shift for individual adults from relatively later emergence to earlier emergence. To test this hypothesis, we regressed emergence deviation from mean emergence by year for males with 6 or more emergence dates (n = 49) and for females with 7 or more emergence dates (n = 72).
We also evaluated the effect of reproductive condition (nesting vs. nonnesting in the year of capture) on emergence timing in females, and body condition (the residual of the log CL vs. log body mass [BM] regression, calculated separately for each year’s data).
Temporal Changes. —
We submitted our individual level (i.e., from mark-recapture) data to linear mixed effect models separately for each cohort using the same process mentioned above. First, we evaluated whether the day of emergence from brumation (i.e., the nth day of the year [1–365], corrected for leap years; hereinafter “ordinal day”) had changed across the study. Here we used year as the fixed effect, carapace length as a fixed effect (CL was highly correlated |≥ 0.96| with other body size measurements) to account for an individual’s growth over time, and individual identification (all cohorts except hatchlings) as a random effect to account for individual variation and an individual’s expected earlier emergence across the study due to an increase in body size and age with time. Second, we predicted how age affected ordinal day of emergence, with individual identification and year as random effects. Third, we evaluated whether carapace length was changing across years using a linear model.
Spring emergence from nests of overwintered hatchlings has been well studied in a number of turtles (e.g., Baker et al. 2010, 2013; Lovich et al. 2014), and the climatic correlates have been studied in several cases (Crawford 1991; Geller et al. 2020; Murphy et al. 2020; among many others). That literature is not summarized here because hatchling YMTs do not overwinter in the nest, but rather bury individually in the vicinity of the nest site after hatching in the fall, and presumably emerge individually in the spring.
RESULTS
From 1981 through 2019, surveyors walked an estimated 30,000 km along our fences and made over 28,283 YMT total captures and individually marked 5231 (excluding 3817 cohort-marked hatchlings). We also recorded spring emergence dates for 14,668 turtles (Fig. S4).
Climate Variation. —
Between 1970 and 2018, there was no change over time in any of the spring climate variables (precipitation or temperature) that we examined (r2 < 0.04 and p > 0.15 in all cases). However, there was extreme variation in the progress of spring warmth across the years sampled (Fig. 2; Table S2). Of all the emergence sample years, 1983 had the coldest spring and 2006 had the warmest spring (Figs. 2 and 3; Table S2).
Citation: Chelonian Conservation and Biology: Celebrating 25 Years as the World's Turtle and Tortoise Journal 24, 1; 10.2744/CCB-1622
Citation: Chelonian Conservation and Biology: Celebrating 25 Years as the World's Turtle and Tortoise Journal 24, 1; 10.2744/CCB-1622
Among-Year Emergence Variation. —
Across all spring emergences 14,668 YMTs emerged from brumation from 1 April (an adult male) to 28 June (a juvenile and a hatchling; Fig. S4). On average males and females typically emerged at the end of April or early May, and hatchlings and juveniles typically emerged about 3 wks later in May (Table 1). However, average emergence dates for males, females, juveniles, and hatchlings varied by 3 wks across the years of study (Table 2; Figs. 3 and 4), and within years the emergence of each of these cohorts extended over about 40 d (Table 1).
Citation: Chelonian Conservation and Biology: Celebrating 25 Years as the World's Turtle and Tortoise Journal 24, 1; 10.2744/CCB-1622
Across 9 sampled years annual average emergence date for males was 1 May (ranging from 22 April in warm 2007 to 12 May in cold 1983; Tables 1 and 2; see also Fig. 4). Individual males emerged from 1 April to 16 June across all years combined (n = 1961). Average commencement to termination of emergence was 15 April to 25 May across 9 yrs (Figs. 3 and 4). As predicted, across years, average emergence date for males was highly correlated with spring temperatures (Table 3; Fig. S5).
Average annual emergence date for females across 9 yrs was 5 May (ranging from 27 April in 2007 to 16 May in 1983; Tables 1 and 2; see also Fig. 4), on average about 4 d later than males. Individual females emerged from 2 April to 17 June across all years (n = 2882). Average commencement to termination of emergence was 17 April to 25 May across 9 yrs (Figs. 3 and 4). As predicted, across years, as for males, average emergence date for females was highly correlated with spring temperatures (Table 3; Fig. S5).
Average annual emergence date for juveniles across 12 yrs was 21 May (ranging from 10 May in 2007 to 1 June in 1983; Tables 1 and 2; see also Fig. 4). Individual juveniles emerged from 24 April to 28 June across all years (n = 5985). Average commencement to termination of emergence was 6 May to 18 June across 12 yrs (Figs. 3 and 4). Surprisingly, across years, none of the climate variables we examined were correlated with mean emergence date for juveniles (Table 3).
Average annual emergence date for hatchlings (following their first overwinter) across 12 yrs was 21 May (ranging from 7 May in 2007 to 4 June in 1983; excluding 1993 when only 3 hatchlings emerged; Tables 1 and 2; see also Fig. 4). Individual hatchlings emerged from 24 April to 28 June across all years (n = 3840). Average commencement to termination of emergence was 8 May to 16 June across 12 yrs (Fig. 3 and 4). Across years, none of the climate variables we examined were correlated with mean emergence date for hatchings (Table 3).
Cohort Emergence Variation. —
During the study, individuals emerged 1.5 d earlier at the Dike and 3.3 d earlier at Mallard Arm compared to the Main Fence, and was significantly different for all cohorts and fences, except hatchings at Main and Mallard Arm (Table S3). Hatchlings and juveniles exhibited the most evident peaks in emergence date in a given year (e.g., Figs. 3 and S4). For example, we captured 427 YMTs (including 296 hatchlings and 113 juveniles) on 20 May 1998 (maximum temperature 25°C; minimum temperature 12°C; maximum temperatures 28°–30.5°C on the previous 3 d; 19 mm rain in the previous 24 h and only 8 mm over the previous 10 d). In addition, we captured 463 YMTs (including 329 hatchlings and 70 juveniles) on 23 May 2006 (maximum temperature 25°C; minimum temperature 13°C; 30°–33°C over previous 5 d; 8.5 mm rain in last 24 h and no rain on previous 12 d).
All hatchling peaks (measured as the percentage of the hatchling cohort remaining to emerge) across years occurred on days with maximum temperatures between 19.4°C and 29.4°C, and 7 of the 9 peaks occurred after rainfall in the previous 24 hrs (Table S4). Over 9 yrs these peaks totaled 43% to 64% of the hatchlings remaining to emerge in 6 yrs and 19% to 27% in the other 3 yrs. Also, the magnitude of the greatest peak of hatchling emergence in a given year was greater in drier years (Fig. S6), illustrating the importance of both temperature and rainfall in the emergence of at least hatchlings.
Across all hatchling emergences in May, an average of 51% occurred on days with rain in the previous 24 hrs and 68% occurred after rainfall in the past 48 hrs (Table S5). This pattern continued into June when temperatures were higher, with 69% of emergences occurring within 48 hrs of rain (Table S6). From our mixed model analyses, 3 weather variables significantly influenced emergence (Table 4). We predicted that 7.9 hatchlings would emerge on a day without rain, and an additional 1.8 individuals for each 10 mm increase in rainfall (max = 18.8 individuals with 60 mm of rain; Fig. S7). Hatchlings were not predicted to emerge until the maximum temperature in the last 48 hrs reached 8.5°C (2.7 individuals), with a peak of 9.2 individuals at 22.4°C.
Peak daily movements for juveniles were less extreme than for hatchlings (Fig. S4). Those yearly peaks occurred on days with maximum temperatures between 19.4°C and 28.9°C, and 8 of 9 peaks occurred after rainfall in the previous 24 hrs (Table S7). Across all 12 yrs the proportion of juveniles that moved on days with rain during the previous 48 hrs averaged 56.7 ± 16.6% (range, 29%–79%) of the total juvenile cohort that year, even though the number of days with rain in May averaged only 11.8 ± 3.2 (range, 7–17). That percentage by year was not correlated with total May rainfall (r =−0.23; p = 0.47), but there was a trend for higher percentages in years with fewer rain days (r =−0.52; p = 0.08), suggesting that there are larger bursts of juvenile activity when rain is less uniformly spread across the month. However, in 2007, 299 juveniles (60% of the total that year) emerged between 8 and 14 May during a period without rain and mean daily high temperatures of 22°–32°C (mean 27°C).
Across all juvenile emergences in May, an average of only 37% occurred on days with rain within 24 hrs and 55% occurred after rain within the past 48 hrs (Table S5). However, in June, 68% of emergences occurred within 48 hrs of rain (Table S6). Not surprisingly, these data suggest that rainfall is a stronger cue for emergence of hatchlings than juveniles, and that the importance of rainfall as a cue increases later in the season. From our mixed model analyses, 5 variables were significant (Table 4). The predicted number of juveniles that emerged increased with total rainfall in the last 24 hrs, from 10 individuals with no rain, and an additional 2.8 individuals for every 10 mm of rain (Fig. S7). Juvenile emergence was also positively associated with increasing maximum temperatures on the day of emergence, with individuals emerging first at 8.5°C, and increasing quadratically until they peaked at 27.6°C with 12.6 individuals predicted to emerge (Fig. S7). No individuals were predicted to emerge until the mean minimum temperatures in the last 48 hrs reached −1.6°C and peaked at 12 individuals when temperatures reached 9.2°C (Fig. S7).
Adults demonstrated no clear pattern of temperature or rainfall as cues for emergence in our initial analyses (e.g., Figs. 5, S8, and S9). However, from our mixed model analyses, mean temperatures over the last 48 hrs significantly influenced emergence for adults (Table 4; Fig. S10). For males, no individuals were predicted to emerge until mean maximum temperatures during the past 48 hrs reached 7.0°C, then steadily increased to 5.1 individuals at 22.0°C. Males were not predicted to emerge until mean minimum temperatures in the last 48 hrs reached −6°C, then steadily increased to 5.8 individuals at 4.2°C. For females, no individuals were predicted to emerge until mean daily maximum temperatures over the last 48 hrs reached 6.4°C and peaked at 6.9 individuals at 20.0°C. A similar relationship occurred for mean minimum temperatures over the last 48 hrs, with no individuals predicted to emerge until temperatures reached −5°C, then peaked at 8.6 individuals at 4.3°C (Fig. S10).
Citation: Chelonian Conservation and Biology: Celebrating 25 Years as the World's Turtle and Tortoise Journal 24, 1; 10.2744/CCB-1622
Life History Effects. —
Emergence date was inversely correlated with body size (carapace length) for all males (r =−0.70; n = 3604; p < 0.0001) and all females (r =−0.32; n = 6255; p < 0.0001) across all years, including those early unsexable captures that were sexed retrospectively upon recapture. The same relationship applied just to males > 90 mm CL (r =−0.58; n = 1961; p < 0.0001; Fig. 6), females > 90 mm CL (r =−0.28; n = 2882; p < 0.0001; Fig. 6), and juveniles (r =−0.185; n = 523; p < 0.0001) across all years. Finally, this relationship was supported for males, females, and juveniles within every sample year (Table 5). The range of variation in hatchling size was too narrow to relate to emergence date.
Citation: Chelonian Conservation and Biology: Celebrating 25 Years as the World's Turtle and Tortoise Journal 24, 1; 10.2744/CCB-1622
Emergence date was also inversely correlated with age for all males (r =−0.68; n = 2985; p < 0.0001) and all females (r =−0.57; n = 4956; p < 0.0001). Like body size, older turtles emerged earlier for all cohorts. From our linear model analyses, males were predicted to advance their emergence 1.1 d for each year of age, ranging from ordinal day 131.7 for age 6 to ordinal day 91.3 for age 44 (β =−1.066, SE = 0.038, 95% CI =−1.141 to −0.990). Females were predicted to advance their emergence 0.7 d each year of age, ranging from ordinal day 133.3 at age 5 to ordinal day 103 at age 50 (β =−0.673, SE = 0.031, 95% CI =−0.734 to −0.162). Juveniles were predicted to advance 1.0 day for every year of age, ranging from ordinal day 144.0 at age 2 to ordinal day 129.8 at age 17 (β =−0.948, SE = 0.043, 95% CI =−1.032 to −0.864).
Nonreproductive females in a given year (i.e., nonnesting) emerged later than reproductive females (Table 6). However, nonreproductive females were on average smaller than reproductive females (Table 6), and hence the delay was expected, since many small females (e.g., 85–95 mm CL) were still immature or more likely to skip reproduction in a given year (Iverson unpublished). When the mean emergence date for only those females ≥ 95 mm CL (i.e., certainly mature) was examined, there was no difference in emergence date between reproductive vs. nonreproductive females (Table 6).
Based on the data from 4 yrs when the entire emergence through nesting season was monitored, body condition (as the residual of log CL vs. log BM) was nearly or significantly higher (p < 0.07) for females that nested in a given year vs. those that did not (Table 7). In addition, analyses of body condition vs. emergence date within each year (Table S8) suggested a difference between males and females. For males over 9 study years, slopes were negative in 8 and significant in 5 yrs (all with negative slopes). Hence, males in better body condition tended to emerge earlier in a given year. In contrast, for females, slopes were positive for 7 of 9 sample years, and significantly positive in 4 yrs and nearly so in 1 other (p = 0.06). Thus, females in poorer body condition tended to emerge earlier.
Temporal Effects. —
Least squares analyses of mean emergence dates by year and cohort suggested no change over time during our study period (Table 2; males, r =−0.13, p = 0.73; females, r =−0.16, p = 0.69; juveniles, r =−0.16, p = 0.63; hatchlings, r =−0.46; p = 0.15), but sample sizes were only 9–12, depending on cohort, with no consideration of individuals. However, when considering individuals, including carapace length as a fixed effect and individual ID as a random effect to account for individual’s growth and survival (except hatchlings, which were a linear model), we found all cohorts advanced over time from 1983 to 2018; males (5.3 d; β =−0.191, SE = 0.025, 95% CI =−0.240 to −0.142); females (4.7 d; β =−0.117, SE = 0.019, 95% CI =−0.154 to −0.080); juveniles (3.6 d; β =−0.116, SE = 0.0135, 95% CI =−0.143 to −0.089); and hatchlings (6.4 d; β =−0.182, SE = 0.020, 95% CI =−0.221 to −0.143). However, we caution the interpretation of this advancement for several reasons (see “Discussion”).
Carapace length for 1958 adult male captures averaged 112.2 ± 11.3 mm; for 3851 adult female captures it averaged 98.6 ± 6.5 mm; and for 5229 juvenile captures it averaged 55.1 ± 15.5 mm (Table S9). Mean annual CL of males decreased by about 1.12 mm CL per decade between 1983 and 2018 (data in Table S9; r =−0.82; n = 9; p = 0.007), and that for females decreased by about 1.19 mm per decade (Table S9; r =−0.78; n = 9; p = 0.013); however, mean annual CL for juveniles did not vary across years (r =−0.30; n = 11; p = 0.37).
Individual Effects. —
Average deviations of emergence days from annual averages for individual turtles revealed considerable variation (Fig. 7). Although those deviations were normally distributed, some individuals clearly emerged consistently earlier or later (Table 8). However, we could identify no correlates of these patterns except individual tendency. Future work will examine reproductive output and other possible factors in early vs. late emerging turtles.
Citation: Chelonian Conservation and Biology: Celebrating 25 Years as the World's Turtle and Tortoise Journal 24, 1; 10.2744/CCB-1622
Regressions of year vs. emergence deviation for individual males and females revealed considerable variation and no clear patterns. For 50 males (with n = 6–10 yrs of emergence data), 16 regressions had negative slopes (i.e., emergence later with time/age), but only 2 were significant at 5%. Positive slopes were found for 33 regressions, and 9 were significant at 5%. For 72 females (with n = 7–11 yrs of emergence captures), 28 regressions had negative slopes (but only 7 significant) and 44 had positive slopes (only 5 significant). Considering experiment-wise error rate, no consistent relationship between year (age) and emergence deviance apparently exists, but may still reflect individual tendencies.
DISCUSSION
Spring emergence of YMTs from brumation varied significantly among years by as much as 3 wks, and the season of emergence for a given cohort (males, females, juveniles, or hatchlings) extended over about 40 d each year. Adult males emerged an average of 4 d before females (overall means, 1 May vs. 5 May), and juveniles and hatchlings emerged later (21 May, for both). Average emergence date across years for adult males and females was related primarily to spring air temperatures, specifically to temperatures during the 48 hrs prior to emergence (but not to rainfall). However, average emergence dates for juveniles and hatchlings were related primarily to rainfall in the previous 24 hrs as well as temperatures over the previous 48 hrs.
In addition, emergence date was inversely related to body size and age, and females that nested in a respective year of emergence had better body condition, but did not emerge earlier than those that did not nest. We also found evidence of individual effects on emergence timing across years. Each of these patterns is discussed below.
Life history effects. —
Males generally emerged and moved to water an average of 4 d before females, perhaps increasing their likelihood of intercepting females for mating as the latter reach the water. As predicted, larger and older turtles emerged earlier than smaller and younger individuals (regardless of sex; see Blouin-Demers et al. 2000). This pattern may simply be a result of larger turtles being able to dig up faster from brumation depths, but we have no data to test this hypothesis. The relationship between emergence date and body size in turtles has been examined in only 1 other study. In North Carolina, Roe and Bayles (2021; see also Roe et al. 2023) found no relationship between body size and emergence timing in Terrapene carolina. No previous study has investigated the effect of age on emergence.
Nesting females in a given year had better body condition when they emerged from brumation than those that did not nest, suggesting that the latter were unable to accumulate sufficient resources in the previous year to support reproduction in the current year. However, reproductive females did not emerge earlier than those that skipped reproduction that year, counter to our original hypothesis.
We found a surprising difference in timing of emergence with respect to body condition by sex, early males exhibited better body condition, but early females exhibited the opposite. Unfortunately, no data are available on the body condition of these turtles going into brumation the previous year. However, it is possible that males in the best condition can dig up and out from their overwintering sites more quickly (i.e., emerge earlier), and females with poor body condition are driven physiologically to move to the water as early as possible. Physiological studies of these turtles during the emergence process could be very enlightening and might explain emergence timing better than simply environmental factors.
Few turtle studies have quantified these emergence relationships, although DeGregorio et al. (2017) found that female Terrapene carolina emerged an average of 8 d before males, and Roe and Bayles (2021; also, Roe et al. 2023) found no difference in emergence timing for males and females of that same species. In addition, Bailey et al. (1995) and Averill-Murray et al. (2018) found that female Gopherus morafkai emerged before males; and Nussear et al. (2007) and Rautenstrauch et al. (1998) reported that juvenile Gopherus agassizii emerged before females, which emerged before males. These authors attributed this pattern to the shallower burrows of juveniles and females, which warmed sooner than the deeper burrows of males. Each of these reports differs from the pattern we observed, perhaps because YMTs require water in which to mate and feed.
We also found differences in emergence timing among fences, with those at the Dike emerging 1.5 d earlier than those at the Main Fence, and those at Mallard Arm emerging 3.3 d earlier than those at the Main Fence (hatchlings were not different). These differences may be related to the solar aspect of the respective brumation sandhills (Fig. S1). The Dike sandhill faces generally southeast and receives direct sun earliest in the morning. The Mallard Arm sandhill faces generally southward, and the Main fence faces southwestward and is the last to receive direct sun. These differences may produce slightly different soil temperature profiles that might explain the variation we observed among fences. It is not clear if these differences are biologically important; however, given that brumation emergence, nesting, and hatching are all delayed in cold years, sometimes to the point of complete recruitment failure (Iverson 1991, 2022), even small differences in spring emergence timing may have significant fitness impacts.
Temporal Effects. —
Juvenile and hatchling YMTs emerged later than adults, generally in mid-May, during the wettest month of the year and when temperatures were less limiting to activity, since even warm days offered windows of optimum temperatures for activity in the morning and evening. Hence, rainfall and benign temperatures appear to act synergistically to promote emergence in smaller turtles, with larger juveniles emerging before smaller turtles. Given their small size (especially hatchlings at only 2–3 g), and hence vulnerability to desiccation in this arid landscape with low humidity, this pattern is not surprising.
We found that emergence date advanced slightly with time (1983–2018) for all cohorts in our study. This finding was unexpected, particularly with the decrease in body size noted in adults over time. However, given the extensive annual variation in timing that we observed and our sporadic sampling (only 9–12 yrs over a 36-yr span), we remain cautious about concluding that emergence has advanced at this site. Only one other review has directly addressed possible changes in emergence dates with time in turtles. Janzen et al. (2018, Table 4) could find no change in emergence time over 25 yrs for Clemmys guttata, but there was a trend (p = 0.07) toward earlier emergence over 18 yrs for Chelydra serpentina. In addition, data presented by DeGregorio et al. (2017, Fig. 2) for Terrapene carolina showed no such change over their 19-yr study. However, those presented by Averill-Murray et al. (2018, Table 2) for Gopherus morafkai revealed that emergence averaged earlier over the course of their 9-yr study (r =−0.83; p < 0.01; from their Table 2). Unfortunately, too few long-term studies focus on the timing of emergence from brumation and its environmental correlates to suggest a general pattern for turtles, although earlier emergence is expected in the future with warming attributed to climate change.
We detected a slow decline in mean adult body size in both males and females between 1983 and 2018. This may reflect some long-term (> 40 yrs) demographic shift (e.g., changes in competition or predation; the subject of future analysis). It is also possible that the increase in average nighttime temperatures at our study site over the past 50 yrs (Hedrick et al. 2021) has resulted in a diversion of energy away from growth and reproduction, and toward maintenance, as a result of higher metabolic rates during the night. Ongoing analysis of long-term patterns of reproductive output in this population will test this hypothesis. If true, we expect a similar decline in the allocation of resources to reproduction.
Individual Effects. —
Our data suggest that emergence date in adults may reflect individual tendencies, with some consistently emerging early and others consistently later (Table 8), but most following the mean population emergence patterns. Whether these emergence patterns have a genetic basis (or simply reflect some environmental factor like brumation depth, which itself may have a genetic basis) remains to be determined. In contrast, we found no pattern in emergence deviation date over periods of 15–36 yrs for individual adults, despite the expectation that turtles would tend to emerge earlier in time over their lives as they increased in size and age (see above; e.g., Fig. 6).
Two other studies have examined individual effects on emergence timing in turtles. DeGregorio et al. (2017) studied 11 Terrapene carolina in South Carolina over 4–17 winters and found a significant individual effect on emergence date. Some (but not all) turtles had high interannual consistency in emergence date. In addition, for the same species in North Carolina, Roe and Bayles (2021) found significant repeatability in nest date for 16 turtles each captured over 2 yrs. Furthermore, they subsequently found that “bolder” individuals emerged earlier than more “timid” individuals (Roe et al. 2023), with a maximum difference of 15 d. Taken together, these studies provide strong evidence for significant individual differences in emergence timing (i.e., responses to environmental emergence cues), which complicates our ability to pinpoint population-wide cues for emergence.
Fitness Impacts. —
Although temperatures ultimately limit the beginning of emergence in the spring, there are several selective advantages to early emergence, especially when the first and mean emergence dates each varied by 3 wks across years (Table 1). These advantages include 1) the increased likelihood of encountering receptive mates; 2) the increased time for foraging (i.e., a longer growing season); 3) the potential for earlier nesting and the coupled increased likelihood of successful hatching of the clutch before winter (Iverson 2022; see below); and 4) for small turtles, the reduced exposure to predators such as migratory birds (e.g., grackles, shrikes, kingbirds, and thrashers, that arrive in mid-May and were frequently observed depredating small YMTs) and Western Hognose Snakes (Heterodon nasicus, known to prey regularly on small YMTs; Platt et al. 1969; Iverson 1990 and which begin activity in mid-May just as hatchling mud turtle emergence is peaking; Fig. S11).
A major disadvantage of early emergence (particularly in adults) is the potential for exposure to lethal temperatures from subsequent cold fronts (see 22 April in Fig. S9). For example, the temperature range was −1.7°C to 3.3°C on 17 May 1983, −3.3°C to −1.1°C on 22 April 1998 (and −6.7°C the next morning), and −4.5°C to −2.7°C (with ca. 6 cm of snow) on 26 April 1994, after 146, 45, and 253 total adult turtle emergences, respectively! Furthermore, in 1994, high temperatures over the 5 d after 26 April 1994 averaged 1.4°C with a highest temperature of 6.4°C and a lowest of −5.7°C, and with a low that night of −11.7°C. In 2007, 16 adult turtles (14 male; 2 female) emerged 1–3 April, with no captures for the next 9 d during which the mean daily high temperature was 4.4°C and 2 consecutive daily highs were only 0°C and −1.7°C. YMTs are not freeze tolerant (Costanzo et al. 1995), which leaves them highly susceptible to death from cold events (e.g., Christiansen and Bickham 1989). Spring cold spells lethal to box turtles (Terrapene) were reported by Neill (1948) in Georgia and Schwartz and Schwartz (1974) in Missouri, and thus the risk of YMTs emerging too early is probably real. Indeed, Roe et al. (2023) found that bolder box turtles in North Carolina emerged earlier from brumation than those that were more shy, but had reduced survival.
Emergence timing clearly has important long-term fitness consequences for YMTs. The mean emergence date for females was 5 May (Table 1), and the mean time between emergence and departure from the wetlands on a nesting foray was 44.4 d (n = 1076; Iverson unpublished), suggesting an average nest date after 17 June. This estimate is in line with the median date of departure on a nesting foray over 18 yrs for 4984 gravid females, on 17.2 June (Iverson unpublished). If incubation length averages 103–105 d (Iverson 2022), then average hatching date would be at least 28 September. Given that the long-term mean daily maximum and minimum temperatures in October at the site were 17.6°C and 1.6°C, respectively, there is limited time that month (and potentially insufficient warmth) for hatchlings to dig down below the frost line before winter. Obviously, delays in any of these parameters (e.g., by cold weather) would put YMTs in jeopardy of reduced recruitment in any given year. Indeed, it is likely that hatching success of the nests of some late-nesting female is low even in a climatically normal year, and Iverson (2022) previously documented nearly complete recruitment failure (via unhatched nests) in very cold years.
Other Considerations. —
Although temperatures constrain the emergence of YMTs from the ground, their impacts on overwintering turtles buried at different depths (and with varying ground cover, slope, and solar aspect) are still unknown. Hence, the initial cue for turtles to begin digging upward in the spring is unknown, as is the time interval from initial upward movement to actual emergence. However, the temperature inversion that occurs below ground in the spring as air temperatures rise (Gregory 1982; Currylow et al. 2013) may be involved. Also, YMTs may dig to the surface and remain there for some period (as Eastern Box Turtles do; Woodley 2013) until other cues trigger their final emergence and movement to the wetlands. More research is clearly needed.
We lack data on maximum brumation depths of individual YMTs, although we found hatchlings at depths of > 70 cm (Costanzo et al. 1995), and Tuma (in Costanzo et al. 1995) found hatchlings in Iowa at 1.0–1.2 m deep. However, even if depths were all the same across sizes and sexes, given the inverse relationship in this species between size (and age) and emergence date, it is possible that emergence date simply reflects the time it takes for turtles of different body sizes to dig upward, i.e., after they experience the first warm spring pulse of temperature at brumation depth (as speculated by Carpenter 1953 for snakes). Monitoring underground movements in turtles prior to emergence will be a challenge, but increasing sophistication and miniaturization of telemetry devices should soon make this feasible (Zhou et al. 2020).
Most of the previous work on brumation emergence has focused on the genera Terrapene (reviewed in DeGregorio et al. 2017 and Milanovich et al. 2017; also Roe et al. 2023) and Gopherus (Nussear et al. 2007; Averill-Murray et al. 2018). Relatively little attention has been paid to other taxa or to subadult turtles (see Table S1), and most of those studies have invoked temperature as the primary cue for emergence, but usually only in very general terms (i.e., nonquantitative). Among kinosternid turtles other than K. flavescens, only a single emerging K. subrubrum has been reported (25 March, Virginia; Wetmore and Harper 1917). For K. flavescens, Ligon et al. (2011) and Sanders et al. (2012) reported the emergence of 7 presumed adults in Texas between 5 March and 21 April in a single year, and LaDuc and Christiansen (2012) reported emergence in May and June in Texas, apparently including all size classes (though samples sizes and precise dates were not reported), across 7 yrs, and primarily in association with rainfall. In Illinois, Tuma (2006) observed 9 adult YMTs emerging between 29 April and 24 May in a single year. Finally, in Iowa, Christiansen et al. (1985) observed 2 presumed adult YMTs emerging on 25 April and reported that emergence occurs during “the third or fourth week of April”; and Christiansen and Galloway (1984) captured 167 hatchlings in terrestrial drift fences (presumably having just emerged from brumation) between 29 April and 22 June across 5 yrs. Each of these reports is consistent with our more extensive emergence data (Table 2), although emergence timing in YMTs in Texas may be more variable than farther north.
Climate change over the last 5 decades has produced warmer temperatures overall at our site, with the greatest impact being a noticeable increase in nighttime minimum temperatures (Hedrick et al. 2021). However, neither spring day-time temperatures nor spring rainfall at our site have changed over that period (e.g., Table S2). Nevertheless, we found some evidence of an advance in emergence timing over our 12 samples spanning a 35-yr period, though we are cautious about its interpretation. In any case, should spring day-time temperatures eventually warm at our site, earlier emergence of YMTs would be expected, if rainfall patterns do not change radically. However, increased uncertainty regarding rainfall and temperature patterns with climate change (Butler et al. 2016; Berriozabel-Islas et al. 2020) could negatively impact YMTs, although their extensive suite of morphological and physiological adaptations to arid climates may serve them well in some parts of the range (but see Christiansen et al. 2012 or Tuma 2006), assuming available habitat.
Regretfully, soil temperatures could not be measured during our study due to persistent rodent damage to data loggers. Soil temperatures might be expected to be important predictors of emergence dates. However, even those data would be complicated by possible spatial differences in temperature profiles (e.g., reflecting variation in slope, aspect, or soil composition and moisture) and the inability to know precisely when turtles break brumation and begin digging upward, and at what body temperatures and rates of upward movement. More detailed analyses of soil and turtle body temperatures coupled with telemetry designed to record underground movement (Zhou et al. 2020) might clarify the relationship of temperature and rainfall to emergence timing more precisely. However, the considerable variation in emergence dates that we observed, even among turtles of the same sex and size (within and between years), suggests that this will not be easy. Furthermore, physiological studies (e.g., blood chemistry; Walden 2017) of turtles before, during, and after they break brumation and eventually emerge could reveal significant biochemical cues for emergence that can potentially override air temperature and moisture, and perhaps explain the wide variation in emergence timing we report here, and might give us some insight into why 299 juveniles (60% of the annual total) emerged in May 2007 during a week with no rain and normal temperatures.
In addition, the role of changes in barometric pressure and associated humidity (and other meteorological) changes as cues for emergence have not been examined in turtles, although there is some correlative evidence of their influence on nesting in a few turtle species (Clay 1981; Pike 2008; Schofield et al. 2010; Palomino-González et al. 2020; Geller et al. 2022; Santoro et al. 2024), and pressure changes have been shown to be detectable by amniotes as diverse as sea snakes (Liu et al. 2010), birds (Breuner et al. 2013), and bats (e.g., Blomberg et al. 2021). The passages of “dry fronts” (i.e., low-pressure systems with increased humidity and cloud cover, but without precipitation; see Santoro et al. 2024) are common at our site and may also provide cues for the emergence of turtles. It was our subjective field observation that bursts of emergence sometimes accompanied these dry fronts, but these potential cues for emergence remain unexplored in turtles.
Finally, it is easy to underestimate the value of long-term individual-based studies (Clutton-Brock and Sheldon 2010; Nanglu et al. 2023). Prior to our work, only 2 long-term studies of brumation emergence in a turtle (Terrapene carolina) had been published (DeGregorio et al. 2017 for 17 yrs; and Roe and Bayles 2021 for 8 yrs), and we hope our work will stimulate more interest in such studies in the future.
Aerial view (looking NNW) of Gimlet Lake, Nebraska, study site in June 1999 at full water capacity following 18.6 cm of rain in April and May (long-term average 13.0 cm). Gimlet Lake is in the background south of Crescent Lake National Wildlife Refuge headquarters and housing area. Mallard Arm is in the foreground. Yellow mud turtles are associated with the sandhills in the upper left and upper right (identifiable here by the invasive red brome on the lower slopes). During dry years and in late summer, Mallard Arm dries down considerably (sometimes nearly dry), whereas Gimlet Lake is permanent (spring-fed). Photograph by Damon Hearne.
Accumulated degree-days from 1 April to 15 June for selected years, including the warmest (2006) and the coldest (1983). Degree-days are the sum of each day’s maximum temperature in °C over 15.6°C through time. Traces for 1986, 1990, 1998, and 1999 were very similar to that for 1993; those for 1988 and 2018 were similar to 1994; and that for 2007 was slightly cooler than 2000.
Timing of emergence of Kinosternon flavescens in western Nebraska from brumation in a cold year (1983; top) and a warm year (2000; bottom). Note that fences were closed on 13 June in 1983.
Box-and-whisker plots of ordinal day of emergence (y axis) by year for Kinosternon flavescens in western Nebraska for adult males (A), adult females (B), juveniles (C), and hatchlings (D). Boxes denote 25% and 75% data quartiles; median is horizontal line within each box; and whiskers represent earliest and latest dates. Note that only 3 hatchlings emerged in 1993 due to the effects of Mount Pinatubo on temperatures during the previous 2 yrs (Iverson 2022) and no hatchlings emerged in 1994.
Rainfall (in mm) over previous 24 hrs (top panel); maximum daily temperature (middle panel); and number of adult emergences (males = blue/lower bars; females = red; bottom panel) by date for Kinosternon flavescens in western Nebraska in 2000 (an average year; Fig. 2). Note the typical lower threshold for daily movement was 15.6°C.
Relationship between carapace length (CL ≥ 85 mm CL) and emergence date (ordinal day) for males (top; CL = 186.1–0.616*OD; n = 1961; r=−0.58; p < 0.0001) and females (bottom; CL = 121.63–0.183*OD; n = 2882; r=−0.28; p < 0.0001) across all years combined for Kinosternon flavescens in western Nebraska. See Table 4 for these relationships by year, all of which were statistically significant.
Frequency distribution of mean emergence deviance (in days) from the average yearly emergence date (by sex) of individual Kinosternon flavescens from western Nebraska for 71 females (top; n = 7–11 emergences per female; mean deviance −0.3 ± 4.1 d) and 50 males (bottom; n = 6–10 emergences per male; mean deviance 1.1 ± 3.6 d). Intervals are 0.5 d. Neither distribution differs from normality (males, Kolmogorov-Smirnov normality test D = 0.08, p = 0.89; females, D = 0.08, p = 0.65). Extreme cases are reviewed in Table 8.
Contributor Notes
Handling Editor: Jeffrey E. Lovich
