Use of time and energy by any organism is affected by biotic and abiotic factors (McNab 2002). Predation risk and food availability are biotic factors that could affect activity times (Pianka and Pianka 1970) while environmental temperature is the most notable abiotic factor affecting time and energy budgets in ectotherms (Huey 1991; Angilletta 2009). Freshwater turtles usually thermoregulate behaviorally by aerial basking to increase body temperature (T_b; Lefevre and Brooks 1995; Krawchuk and Brooks 1998), with physiological processes (i.e., metabolic and digestive rates) increasing in association with the increase in T_b (Jackson 1971; Parmenter 1981; Bennett 1982). Moreover, some authors have suggested that aerial basking may have other benefits besides thermoregulation, for example parasite removal or vitamin D synthesis (Chessman 1987; Richard 1999). Freshwater turtles thermoregulate in the water when the preferred temperature is available and basking should be more frequent during early and late season when water temperatures (T_w) are lower (Edwards and Blouin-Demers 2007). Additionally, it has been suggested that females may increase basking frequency during the nesting season due to energetic demands when egg development and oviposition are taking place (Schwarzkopf and Brooks 1985; Krawchuk and Brooks 1998).

Despite the benefits associated with basking, behavioral thermoregulation imposes energy expenditure, mortality risk, and missed opportunities. For example, an organism shuttling between microclimates will expend more energy, attract more predators, and startle more prey than one remaining still (Huey 1974; Huey and Slatkin 1976). In freshwater turtles, time budgets related to basking should be constrained because they perform all their vital activities, such as food consumption and reproduction, in the water, except for nesting (Pritchard 1967). When individuals are basking they cannot engage in any other activity, which leads to missed opportunities such as reproducing and feeding. Given that no other activity can occur while basking, individuals should spend as little time as possible reaching their optimal body temperature (Angilletta 2009). Besides basking frequently, when water temperature is low, turtles should maximize net thermoregulation benefits, obtaining more energy per unit time (Huey and Slatkin 1976) by choosing to bask at midday when solar radiation and air temperature are higher. Nevertheless, during summer in a northern emydid turtle species, basking behavior showed a bimodal pattern, with more individuals basking in the morning and afternoon (Selman and Qualls 2011). The high temperatures and intense solar radiation of summer could explain this pattern, although other factors such as scute ecdysis may also be relevant (Selman and Qualls 2011).

Trachemys dorbigni (black-bellied sliders) and Phrynops hilarii (Hilaire's side-necked turtle) are freshwater turtles commonly found in rivers, lagoons, and ponds of Argentina, Uruguay, and southern Brazil and which have a predominantly carnivorous diet (Cabrera 1998; Achaval and Olmos 2007). Both species show sexual dimorphism, with females larger than males, but P. hilarii are larger than T. dorbigni, reaching up to 40 cm in carapace length while T. dorbigni reach 30 cm in carapace length (Achaval and Olmos 2007). Nesting in T. dorbigni occurs mainly in November and December (Bager et al. 2009), so we expected to observe more basking female T. dorbigni during these months. For P. hilarii, nesting occurs during 2 periods, between September and December and between February and May (Bager 1997), thus we expected to observe more basking female P. hilarii in these months. As females bask more during the nesting period to increase body temperature before leaving the water to nest (Schwarzkopf and Brooks 1985; Krawchuk and Brooks 1998), we expected to find differences in basking pattern between the 2 species.

Basking ecology, habitat use, and thermal regimes of most freshwater turtles are poorly known, especially for species occurring at southern latitudes. Thus, we analyzed activity patterns by quantifying the basking behavior of T. dorbigni and P. hilarii during austral summer and early fall. The goal of this study was to study the effects of season (D), water temperature, and time of day (H) on basking.

METHODS

Study Area

The study area consisted of 2 connected ponds of 26 × 72 m and 190 × 168 m, located in Lecocq Park (34°47′S, 56°19′W) 18 km northwest of Montevideo, Uruguay. Three species of turtles live in these ponds (T. dorbigni, P. hilarii, and the South American snake-necked turtle, Hydromedusa tectifera), with individuals not being fed, not having contact with park employees, and having few visitors (S.C.-B., pers. obs., December 2002). Basking turtles were observed over emergent deadwood in the water, on the shorelines of both ponds, and on 2, 2-m-long artificial cylindrical structures made of cement that were present in the smaller pond.

Fieldwork was conducted from December to April in 2 consecutive austral summers between 2002 and 2004. During this period, individuals were periodically captured and marked with a painted code on their carapace. A total of 29 T. dorbigni (14 females, 7 males, and 8 juveniles) and 22 P. hilarii (2 females, 8 males, and 12 juveniles) were captured. Female and male carapace length and body mass distributions of both species are shown in Appendix 1. We also captured and marked 11 individuals of H. tectifera, but observed this species basking only twice during the period of study. Basking behavior was assessed through scans with 10 × 50 binoculars while walking along the pond shorelines. The basking individuals (marked and unmarked) where counted every hour from 0700 to 1800 hrs on 16 days, with a total of 106 observations (hourly counts) for each species. Water temperature was recorded hourly with a thermometer (0.01°C) at a depth of 15 cm.

Data Analysis

We tested for differences in basking frequency and water temperatures among months using contingency tables and analysis of variance (ANOVA), respectively. Post hoc comparisons were made with Fisher's Exact test and Tukey's HSD test, respectively. In addition, to evaluate seasonal effects on basking behavior, we transformed dates to Julian days (increasing along the summer) and incorporated date as a predictor variable in data analysis. We fitted generalized linear models (GLM) with a Poisson link (Faraway 2004; Zuur et al. 2009) including water temperature (T_w), hour (H), quadratic hour (H²), Julian day (D), and quadratic Julian day (D²) and with their 2-way interaction terms as predictor variables. We included quadratic factors for D and H to test a unimodal relationship with basking frequency to test the hypothesis that basking frequency has a minimum or maximum at intermediate values of D and H (Table 1). All analyses were performed with R software (R Development Core Team 2012). Model selection was made using Akaike Information Criterion (AIC) and a log-likelihood ratio test (LRT; Akaike 1971, 1974; Zuur et al. 2009). When AIC values could not distinguish between competing models (ΔAIC < 2), the log-likelihood ratio test was performed (Zuur et al. 2009). When the nested model has a significant p-value, the excluded variable was interpreted to be relevant to explaining the response variable (Zuur et al. 2009).

Table 1. Variables included in data analyses with biological hypotheses. T_w = water temperature in °C; H = hour; H² = quadratic term for hour; D = Julian day; and D² = quadratic term for Julian day.

View Fullscreen

RESULTS

We observed significant differences in the frequency of basking among months in T. dorbigni (χ²₃ = 24.034, p < 0.001; Fig. 1A). In particular, T. dorbigni showed higher basking frequency in both December and February than in January (Fisher's Exact test; p < 0.001 and p < 0.05, respectively) but no differences between December and April (p = 0.552; Fig. 1A). T_w also differed significantly among months (F_116,3 = 84.8, p < 0.001; Fig. 1C). Tukey's HSD post hoc comparisons showed that all months were significantly different from one another in T_w (p < 0.05; Fig. 1C) except for December and February (p = 0.99; Fig. 1C).

View Figure

• Download as Powerpoint
• Download Figure

The best model predicting the number of basking individuals in T. dorbigni included effects of T_w, H, and H² (model 5) and the interaction term T_w × H. Models with the lowest AIC values were models 4 and 5, but the LRT test showed that model 4 included a nonsignificant D effect to explain basking behavior (Table 2). Model 5 explained 34% of the deviance of the data, and the estimated parameters showed that H had a positive effect on basking behavior, whereas the main T_w effect was significant but negligible (Table 3). Indeed, the stronger effect of T_w was observed in its interaction with H (Table 3). We observed that the number of basking individuals decreased while T_w increased at noon and in the afternoon (Fig. 2). However, in the morning there was not a clear T_w effect on the number of basking individuals (Fig. 2). Regarding use of time, we observed a bell-shaped curve at lower T_w values for basking with several individuals basking at midday, few individuals basking during afternoon, and almost no individual basking in the morning (Fig. 2). However, this bell-shaped pattern diminished while the T_w increased, with fewer individuals basking at noon as T_w increased (Fig. 2).

Table 2. Model selection for basking in T. dorbigni by means of a log-likelihood ratio test (LRT). T_w = water temperature in °C; D = Julian day; D² = quadratic term for Julian day; H = hour; and H² = quadratic term for hour.

View Fullscreen

Table 3. The best fitting model (model 5) for basking in T. dorbigni. T_w = water temperature; H = hour; H² = quadratic term for hour.

View Fullscreen

View Figure

• Download as Powerpoint
• Download Figure

Phrynops hilarii also showed a seasonal pattern for activity with significant differences in basking frequency, which increased in fall as T_w decreased (χ²₃ = 18.125, p < 0.001; Fig. 1B–C). In this sense, the number of basking individuals was greater in April than in December (p < 0.05; Fisher's Exact test result) and January (p < 0.001), but basking frequency did not differ significantly between December and February (p = 0.063; Fig. 1B). In addition, there were fewer individuals basking in January than in February (p < 0.05) and April (p < 0.001).

The best model (model 5) predicting the number of basking individuals in P. hilarii included T_w, H, H², D, and the interaction term T_w × D (Table 4). This model had the lowest AIC value (Table 4) and explained 50% of the deviance. The estimated parameters showed a negative effect of T_w and a unimodal effect of H on basking behavior (Table 5). T_w and D interacted to determine the number of basking individuals; for example, T_w had an evident negative effect during December, January, and February. During these months, when T_w increased, the number of basking individuals decreased (Fig. 3). However, during fall as T_w increased, the estimated number of basking individuals also increased (Fig. 3).

Table 4. Model selection for basking in P. hilarii by means of a log likelihood ratio test (LRT). T_w =  water temperature in °C, D = Julian day; D² = quadratic term for Julian day; H =  hour; and H² = quadratic term for hour.

View Fullscreen

Table 5. The best fitting model (model 3) for basking in P. hilarii. T_w =  water temperature in °C; Day = Julian day; H = hour; and H² = quadratic term for hour.

View Fullscreen

View Figure

• Download as Powerpoint
• Download Figure

We observed that H had a bell-shaped effect on basking behavior with more individuals basking at noon than during the morning and the afternoon (Fig. 3). This bell-shaped pattern was conspicuous at lower T_w values in December, January, and February (Fig. 3), whereas in April, it became evident as T_w increased (Fig. 3).

Regarding seasonal effects on activity patterns, D showed a negative effect on the number of basking individuals of P. hilarii (Table 4), with decreasing numbers of basking individuals from December to January (Fig. 3). Additionally, this seasonal effect on basking was similar to that observed in the contingency tables (Fig. 1B). For example, February and April had different mean T_w values, yet similar basking frequencies of P. hilarii individuals were observed for both months (Fig. 1B–C).

DISCUSSION

Environmental temperature affects ectotherms more than other organisms (Paaijmans et al. 2013) because it determines all vital rates such as activity levels and reproduction (McNab 2002; Angilletta 2009). Here, we show that activity patterns of 2 South American freshwater turtles vary with water temperature, diminishing the time spent out of water as water temperature increases (Figs. 2 and 3). Additionally, basking behavior in P. hillarii changed with season (Fig. 3) while in T. dorbigni, basking occurred mainly after noon on those days that water temperature was lower during all parts of the season (Fig. 2).

We observed a seasonal activity pattern for Trachemys dorbigni, with significant differences among months for basking frequency. In particular, January showed the lowest basking frequency with no individuals basking (Fig. 1A). Moreover, December and April shared similar basking frequencies but showed different mean T_w values, making it difficult a priori to attribute basking only to T_w (Fig. 1). However, when we fit models for the number of basking individuals, the best model did not include D as a predictor variable (Tables 2 and 3), indicating the absence of a clear seasonal effect on basking in this species. This result could occur because the seasonal effect in basking is predicted only for females, which have different energetic requirements during nesting season (Schwarzkopf and Brooks 1985; Krawchuk and Brooks 1998). Moreover, the sex of basking individuals was not recorded, which could mask the seasonal effect (Edwards and Blouin-Demers 2007). Females and males could show different basking patterns due to reproductive needs (Selman and Qualls 2011). For example, females may bask longer and more frequently during spring to reach T_b optima for nesting. The differences in body size between the sexes could also produce different basking patterns due to greater thermal inertia in larger females (McNab 2002). In fact, sexual dimorphism in carapace length has been reported for T. dorbigni and P. hilarii (Astort 1984; Cabrera 1998). Carapace length differences between female and male individuals recorded in the present study were negligible (i.e., T. dorbigni 25%, P. hilarii 10%), but body size differences over 100% in T. dorbigni indeed may produce different basking patterns between the sexes (Appendix 1).

Phrynops hilarii showed fewer individuals basking in January than in February and April and similar basking frequencies in January and December despite differences in mean T_w (Fig. 1), suggesting a seasonal pattern not entirely associated with T_w. Instead, such increase in basking frequency in February and April could be due to the presence of nesting females in these months, given the 2 nesting periods described for this species: September to December and February to May (Bager 1997). However, the best model for P. hilarii included a negative relationship between D and the number of basking individuals, with basking frequency decreasing during the season (Table 5). Unfortunately, we do not have any information about nesting females in our study area or the reproductive status of basking individuals that could shed light on the cause of the seasonal effect on basking for P. hilarii.

As noted, T_w was a relevant factor affecting activity patterns in both species (Tables 3 and 5), with more individuals basking at lower T_w values (Figs. 2 and 3). This effect was more conspicuous in P. hilarii than in T. dorbigni (Tables 3 and 5); for the latter, the T_w effect was observed mainly with the interaction with H (Table 3; Fig. 2). This means that individuals of T. dorbigni decrease basking activity when T_w decreases but only during midday and afternoon (Fig. 2). Besides the main T_w negative effect on basking in P. hilarii, there was a significant interaction term between T_w and D (Table 5). We observed that the relationship between T_w and the number of basking individuals varied over the season, with a negative effect of T_w on basking in December, January, and February and a positive effect of T_w in April (Fig. 3). During April, T_w reached lower values and turtles probably were inactive while in the days with higher temperatures, individuals were active and more likely to bask, which produces the positive T_w effect on basking frequency (Fig. 3).

Regarding the use of time in basking activity, both species showed the same bell-shaped pattern, with more individuals basking at midday than during the morning and afternoon (Figs. 2 and 3). For T. dorbigni, this pattern was more conspicuous at lower T_w values. Individuals probably avoid costs of thermoregulation owing to missed opportunities and maximize benefits of thermoregulation by basking when solar radiation and the heating rate are greater, allowing the allocation of less time to basking (Huey and Slatkin 1976). In the case of P. hilarii, the bell-shaped pattern was conspicuous for low temperatures during summer and high temperatures in autumn (Fig. 3). However, whereas we expected the opposite pattern with more individuals basking in the morning and afternoon during midsummer (Bauwens et al. 1996), we observed that turtles did not bask during the days with the highest temperatures, perhaps to avoid the extremes of solar radiation and temperature.

We showed a strong relationship between basking activity and abiotic factors such as water temperature, time of day, and season. We presume that T. dorbigni and P. hilarii take advantage of aerial basking to obtain thermoregulatory benefits and that this activity is constrained by energy and time budgets. Individuals could adjust their energy and time budgets by changing basking frequency or duration depending on several factors. For example, with high radiation and temperature, individuals could bask more frequently for a shorter duration for each event of basking (Selman and Qualls 2011). However, we did not record or analyze basking duration and, under this scenario with a short duration of basking events, we could be underestimating basking frequency. In addition, species may compete for basking sites, which would produce the observed differential basking pattern among them (Cadi and Joly 2003). A better experimental design is needed to test both hypotheses with stronger evidence.

To our knowledge, H. tectifera does not aerial bask; we did not observe individuals of this species basking. Speculation on why this species does not bask is not possible because we cannot argue whether this absence of basking is owing to a thermoconformer strategy, the use of different microhabits inside the water, or a preferred body temperature substantially different from those of the other 2 freshwater turtle species.