Turtles and tortoises are among the most endangered vertebrates (Lovich et al. 2018; Rhodin et al. 2018) and face a variety of conservation threats including habitat loss and fragmentation (Mitchell and Klemens 2000), human overexploitation (Thorbjarnson et al. 2000), and climate change (Ihlow et al. 2012; Böhm et al. 2013; see Burke et al. 2000 and Todd et al. 2010 for general reviews). One particular threat to turtles is habitat degradation through contamination or alteration of aquatic ecosystems as well as impacts on nesting habitats (Linder et al. 2010; Todd et al. 2010; Carrizo et al. 2017). Such habitat degradation may potentially result in developmental or morphological problems for turtles.

In turtles, the frequency of shell anomalies (i.e., changes from the standard carapace or plastron morphology) might be an index of suboptimal habitat conditions. For example, the frequency of shell anomalies can change with variation in incubation conditions including temperature (Herlands et al. 2004; Davy and Murphy 2009; Telemeco et al. 2013; Zimm et al. 2017) and moisture or humidity (Lynn and Ullrich 1950; Zimm et al. 2017; but see Velo-Antón et al. 2011). In addition, turtles from less anthropogenically impacted sites tend to have fewer shell anomalies than do turtles from more anthropogenically impacted sites (Bishop et al. 1991; Bell et al. 2006; de Solla et al. 2008). In particular, sites with greater levels of contamination (i.e., chemical pollution) tend to have higher frequencies of shell anomalies (Bishop et al. 1991; Bell et al. 2006; de Solla et al. 2008; Nagle et al. 2018). However, Davy and Murphy (2009) found no correlation between shell anomalies and contamination, and Neuman-Lee and Janzen (2011) found no effect of atrazine exposure during development on morphological anomalies in hatchling map turtles (Graptemys ouachitensis and Graptemys pseudogeographica). In addition, an increased frequency of shell anomalies may be related to low genetic diversity or inbreeding (Velo-Antón et al. 2011).

One might expect that if increased habitat degradation, whatever the source, results in developmental or morphological abnormalities, then tracking the frequency of such abnormalities might give an indication of the change in the effects of habitat degradation on turtle populations. Examining temporal changes in the frequency of shell anomalies in a turtle population or community may thus give us a better understanding of temporal variation of stressors experienced by those turtle populations and communities. Here we use a 36-yr study of a turtle community in a northern Indiana lake (Dewart Lake, Kosciusko County; 41°22′N, 85°45′W) to examine the frequency of shell anomalies in 3 species of freshwater turtle (painted turtle, Chrysemys picta; northern map turtle, Graptemys geographica, and red-eared slider, Trachemys scripta elegans) and to identify temporal variation in their frequency. We have previously shown shifts in the composition of the turtle community of Dewart Lake, primarily due to declines in the numbers of C. picta (Smith et al. 2006). In addition, we have shown that there have been changes in the frequency of boat or propeller wounds over time in this community, especially in C. picta, which are likely associated with levels of boat traffic on the lake (Smith et al. 2006, 2018). These observations suggest that potential temporal changes in anthropogenic stressors are occurring in Dewart Lake, and we have observed changes in aquatic and terrestrial herbicide and pesticide use over the course of our study and an increase in human impacts along the shoreline from historical photography of Dewart Lake (J.B.I., G.R.S., and J.E.R., pers. obs.). We have also observed a reduction in water clarity over time, with turtles visible in the water column to depths > 4 m in the early years to depths < 1 m late in the study (J.B.I., pers. obs.). Unfortunately, we do not possess quantitative data on the temporal trends in potential anthropogenic stressors. However, based on the temporal changes in populations and propeller damage in the turtles in Dewart Lake (Smith et al. 2006, 2018), we expected there might be an increase in the frequency of shell anomalies over the course of our study due to potential increases in anthropogenic stressors in Dewart Lake. However, our main goal was to describe the frequency and types of shell anomalies and whether their frequency varies with turtle size or over the course of the study in the turtle community in Dewart Lake.

METHODS

We surveyed the turtle community in Station Bay in the southeastern corner of Dewart Lake nearly annually (in late July–early August) from 1979 to 2015 using a variety of trapping and capture methods. Prior to 1992, we used aquatic wire funnel traps (n = 5–15; see Iverson 1979 for design). After 1992, we used 2.5-cm mesh fyke nets (n = 2–12) deployed with 15-m (50-foot) leads between a pair of 90-cm (3-foot) hoop diameter funnel traps. We checked traps every 2–3 hrs from sunrise to 1–2 hrs post sunset. No turtles entered the traps during the night (Smith and Iverson 2004). We measured carapace length (CL) in millimeters, sexed (some individuals were too small to sex because they were smaller than the smallest males expressing secondary sexual characteristics), individually marked (using notches cut into marginals to create an individual number; Cagle 1939), retained, and subsequently released all captured turtles at the end of the sampling period (2–5 d). We examined each captured turtle for shell anomalies (i.e., deviations from the standard carapace and plastral scute morphology, excluding wounds or other damage) and recorded the nature of any such anomalies (Table 1). To allow us to calculate an estimated year of hatching, we aged turtles using counts of annuli on plastral scute when annuli were visible (Germano and Bury 1988; Wilson et al. 2003) or by comparing juvenile CL (e.g., in turtles < 10 yrs old) to that of known-aged turtles. The use of annuli to age G. geographica and T. s. elegans in Dewart Lake has been confirmed (Iverson 1988; Lewis et al. 2018; Iverson et al. 2019).

Table 1. Inventory of the types of carapacial and plastral anomalies found in the turtles from Dewart Lake in northcentral Indiana. We list the anomalies under the summary categories used in analyses of the frequencies of anomalies in different shell locations (see Table 2). Accessory scutes are smaller than normal extra scutes, whereas extra, double, or triple scutes are equal, nearly full-sized scutes.

View Fullscreen

Table 2. Frequency of shell anomalies for male, female, unsexed juvenile, and all Chrysemys picta, Graptemys geographica, and Trachemys scripta elegans from Station Bay, Dewart Lake, Indiana. Because some individuals had > 1 anomaly, the total number of individuals with anomalies is not necessarily the sum of the different types of anomalies. Percentages of turtles with an anomaly in a specific group are given in parentheses (i.e., percent of all individuals of a particular group with a given anomaly type).

View Fullscreen

We used logistic regression to determine whether there was a relationship between estimated year of hatching (back-calculated based on age estimate), year of first capture, and CL at first capture on the presence or absence of a shell anomaly in an individual. We ran these analyses for each species separately using all individuals, females only, and males only. To correct for multiple comparisons for each species, we used the Holm-Bonferroni correction to sequentially adjust alpha for multiple comparisons and determine significance. We used chi-square (χ²) analyses to compare frequencies of anomalies between carapace and plastron within each species, between the sexes within each species, and among species.

RESULTS

Chrysemys picta. — A total of 337 C. picta had at least 1 shell anomaly (17% of 1993 individuals). Fifty-two C. picta (3%) had multiple anomalies. No single type of anomaly appeared to be more frequent than others (Table 2). However, carapace anomalies were more frequent than plastral anomalies (χ²₁ = 11.77, p = 0.0006). Male and female C. picta did not differ in the frequency of anomalies (χ²₁ = 1.60, p = 0.21). Males did have a higher relative frequency of carapace anomalies vs. plastral anomalies than did females (χ²₁ = 5.36, p = 0.021).

There was a significant, but relatively weak, negative relationship between year of first capture and the presence of a shell anomaly for all C. picta (n = 1992, χ²₁ = 78.9, r² = 0.044, p < 0.0001; coefficients: slope = –0.06, intercept = 123.0). This negative relationship was also found when we analyzed females only (n = 778, χ²₁ = 11.9, r² = 0.15, p = 0.0006; coefficients: slope = –0.037, intercept = 75.3) and males only (n = 890, χ²₁ = 53.2, r² = 0.064, p < 0.0001; coefficients: slope = –0.072, intercept = 146.1).

There was a significant negative relationship between CL at first capture and the presence of a shell anomaly for all C. picta (n = 1993, χ²₁ = 28.35, r² = 0.016, p < 0.0001; coefficients: slope = –0.12, intercept = 2.97). In female C. picta, there was no relationship between CL at first capture and the presence of a shell anomaly (n = 778, χ²₁ = 0.22, r² = 0.0003, p = 0.64). However, in male C. picta there was a significant negative relationship between CL at first capture and the presence of a shell anomaly (n = 891, χ²₁ = 17.82, r² = 0.022, p < 0.0001; coefficients: slope = –0.024, intercept = 4.33).

The probability of a C. picta having a shell anomaly decreased with estimated year of hatching (n = 1001, χ²₁ = 11.9, r² = 0.015, p = 0.0006; coefficients: slope = –0.042, intercept = 86.1). However, this relationship was not significant after applying the Holm-Bonferroni correction for females alone (n = 410, χ²₁ = 3.94, r² = 0.01, p = 0.047) or for males alone (n = 273, χ²₁ = 2.23, r² = 0.01, p = 0.14).

Graptemys geographica. — A total of 42 individuals had at least 1 shell anomaly (11% of 387 total individuals). Eight individuals (2%) had multiple anomalies. Anomalies of the marginals were apparently the most frequent followed by plastral anomalies and vertebral scute anomalies (Table 2). Carapace anomalies were more frequent than plastral anomalies (χ²₁ = 11.26, p = 0.0008). Male and female G. geographica did not differ in their frequency of shell anomaly (χ²₁ = 1.47, p = 0.23). Males and females also did not differ in the relative frequency of carapace vs. plastral anomalies (χ²₁ = 0.60, p = 0.44).

There was no relationship between year of first capture and presence of a shell anomaly (n = 387, χ²₁ = 2.11, r² = 0.008, p = 0.15). This relationship also held true when males (n = 169, χ²₁ = 0.23, r² = 0.0021, p = 0.63) and females (n = 137, χ²₁ = 2.64, r² = 0.0255, p = 0.10) were analyzed separately.

There was no significant relationship between CL at first capture and the presence of a shell anomaly after applying the Holm-Bonferroni correction (n = 387, χ²₁ = 6.94, r² = 0.024, p = 0.008). This relationship was also not significant after applying the Holm-Bonferroni correction when females were analyzed alone (n = 137, χ²₁ = 5.76, r² = 0.054, p = 0.016) or when males were analyzed alone (n = 169, χ²₁ = 0.01, r² = 0.0001, p = 0.92).

There was no relationship between the estimated year of hatching and the presence of an anomaly in a G. geographica (n = 319, χ²₁ = 0.57, r² = 0.003, p = 0.45). There was also no such relationship when females (n = 123, χ²₁ = 1.35, r² = 0.01, p = 0.24) or males (n = 117, χ²₁ = 0.08, r² = 0.001, p = 0.77) were analyzed separately.

Trachemys scripta elegans. — A total of 31 individuals had at least 1 shell anomaly (18% of 175 individuals). Seven individuals (4%) had multiple anomalies. Anomalies of the marginals and vertebrals were the more common anomalies, but were slightly less frequent than plastral anomalies (Table 2). There was no difference in the frequency of anomalies on the carapace and the plastron (χ²₁ = 0.12, p = 0.73). Male and female T. s. elegans had similar frequencies of shell anomalies (χ²₁ = 0.78, p = 0.38). Male and female T. s. elegans also did not differ in the frequency of carapace vs. plastral anomalies (χ²₁ = 0.27, p = 0.60).

There was no relationship between year at first capture and presence of a shell anomaly (n = 175, χ²₁ = 0.03, r² = 0.0002, p = 0.87). This finding also held true for males only (n = 54, χ²₁ = 1.40, r² = 0.027, p = 0.24) and females only (n = 62, χ²₁ = 0.28, r² = 0.0039, p = 0.60).

There was a significant but weak negative relationship between CL at first capture and shell anomaly presence (n = 175, χ²₁ = 8.59, r² = 0.05, p = 0.0034; coefficients: slope = –0.009, intercept = 2.93). However, the relationship between CL at first capture and shell anomaly was not significant for males alone (n = 54, χ²₁ = 0.45, r² = 0.008, p = 0.50) or for females alone (n = 62, χ²₁ = 0.78, r² = 0.0028, p = 0.38).

Estimated year of hatching had no effect on presence of shell anomalies in individual T. s. elegans (n = 138, χ²₁ = 1.82, r² = 0.016, p = 0.18). No relationship was also observed when we analyzed females (n = 48, χ²₁ = 0.28, r² = 0.005, p = 0.60) or males (n = 32, χ²₁ = 1.63, r² = 0.05, p = 0.23) separately.

Species Comparisons. — Graptemys geographica had a lower shell anomaly frequency than did C. picta or T. s. elegans (χ²₂ = 9.10, p = 0.01). The frequency of carapace vs. plastral anomalies did not significantly differ among the 3 species; however, there was a tendency for T. s. elegans to have a similar frequency of carapace and plastron anomalies, whereas C. picta and G. geographica both had higher frequencies of carapace anomalies than plastral anomalies (χ²₂ = 4.99, p = 0.08).

DISCUSSION

The frequency of shell anomalies in C. picta at Dewart Lake (17%) was intermediate to other estimates for this species (John Heinz National Wildlife Refuge, Pennsylvania, 19%, Bell et al. 2006; E.S. George Reserve, Michigan, 14%, Bell et al. 2006; Pennsylvania, 13%, Ernst 1971; Ontario, Canada, 28%, Davy and Murphy 2009; Saskatchewan, Canada, 22%, MacCulloch 1981). Our observed frequencies of shell anomalies in G. geographica (11%) and T. s. elegans (18%) were low compared with other studies for these 2 species: 29% for a population of G. geographica from a contaminated site (Nagle et al. 2018) and 36% for T. scripta from the Chicago Museum of Natural History collections (Zangerl and Johnson 1957). We encourage the collection and publication of such information to allow a better understanding of the extent of variation in the frequency of shell anomalies among populations and to allow for clarity concerning the potential underlying causes of such variation.

In all 3 species we examined, the overall frequency of shell anomalies did not differ between males and females. This finding contrasts with results for other populations of C. picta in which females tended to have higher frequency of anomalies than do males. In Chrysemys picta bellii from southern Saskatchewan, females had a higher frequency of shell anomalies than did males (MacCulloch 1981). In addition, Davy and Murphy (2009) reported slightly higher frequencies of anomalies in females of both Chrysemys picta marginata and Chelydra serpentina. Our results for T. s. elegans are similar to the only other study on sex differences in shell anomalies in the genus Trachemys: in T. dorbigni, shell anomalies were found in 8% of males, 10% of females, and 14% of immatures (Bujes and Verrastro 2007).

In both C. picta and G. geographica, there were more anomalies on the carapace than the plastron, but in T. s. elegans there was no difference in the frequency of anomalies on the carapace and the plastron. Zangerl and Johnson (1957) found that in general, abnormalities in the plastron of C. picta and T. scripta were more common than those in the carapace, with slightly more carapacial anomalies occurring on the posterior half of the shell. Carapace anomalies were more prevalent in C. p. bellii from southern Saskatchewan than were plastron anomalies (MacCulloch 1981). No single type of anomaly was more frequent than any other in C. picta. In G. geographica, marginal scute anomalies were more frequent than the other types of anomalies. In T. s. elegans, plastral anomalies were most frequent, followed by anomalies of the marginals and vertebrals. Bell et al. (2006) found that the most frequent anomalies in C. picta were extra carapacial scutes or deformed shells. The most common shell anomaly in C. p. bellii from southern Saskatchewan was division of carapacial scutes or supernumary carapacial scutes (MacCulloch 1981).

The weak negative relationship between CL at first capture and the presence of a shell anomaly in C. picta and T. s. elegans might indicate weak selection against individuals with a shell anomaly (i.e., those turtles with an anomaly suffer earlier mortality and thus anomalies are less represented in larger turtles). For example, shell anomalies have been shown to decrease turtle survival (e.g., Mast and Carr 1989; Özdemir and Türkozan 2006; Telemeco et al. 2013; Maffucci et al. 2020), physiological performance and growth (Sim et al. 2014), or carapace strength (Rivera and Stayton 2013). However, the link or correlation between shell anomalies and reduced performance or individual fitness does not appear to be universal (e.g., Tucker 1997; Ergene et al. 2011; Moldowan et al. 2015). Given the weak nature of the relationship between CL at first capture and the presence of a shell anomaly in C. picta and T. s. elegans, it appears that the shell anomalies found in turtles in Dewart Lake do not appear to be very costly to an individual's survival, at least for those anomalies that permit successful hatching and early survival.

Whether or not an individual had a shell anomaly was affected by year of first capture and estimated year of hatching in C. picta, but not in G. geographica or T. s. elegans. The relationships in C. picta, while statistically significant, were all quite weak (r² ≤ 0.15 in all cases) despite the robust sample size (n = 1993). It thus appears that over the 36-yr span of our study, there was a subtle decrease in shell anomalies in C. picta, but this very weak shift does not appear to suggest any dramatic or systematic change in the frequency of shell anomalies over time. Indeed, the frequencies of shell anomalies we observed in all 3 species tended to be lower than many previous studies, especially those from contaminated sites (see above). Thus, while Dewart Lake has experienced an apparent increase in human impacts on its turtle populations (see Smith et al. 2006, 2018), these anthropogenic changes have apparently not manifested themselves in increases in turtle shell anomalies.