Impact of Human Disturbance on the Thermoregulatory Behaviour of the Endangered Ringed Sawback Turtle (Graptemys Oculifera)

Home , Northern map turtle, Ringed map turtle

Received: 14 May 2018 Revised: 8 January 2019 Accepted: 27 February 2019 DOI: 10.1002/aqc.3106

RESEARCH ARTICLE

Impact of human disturbance on the thermoregulatory behaviour of the endangered ringed sawback turtle (Graptemys oculifera)

Jessica M. Heppard1 | Richard Buchholz1,2

1 Department of Biology, University of Mississippi, Oxford, Mississippi Abstract 2 Center for Biodiversity and Conservation 1. Human recreational disturbance of riverine environments may be contributing to Research, University of Mississippi, Oxford, population declines of aquatic organisms worldwide. Aquatic turtles that are heavily Mississippi dependent on basking in sunlight to maintain an optimal body temperature may Correspondence abandon basking perches because they perceive humans as potential predators. Richard Buchholz, University of Mississippi – Biology, Oxford, MS 38677‐1848, USA. 2. The impact of boat traffic, the availability of basking structures, and various envi- Email: [email protected] ronmental factors on the basking behaviour of the ringed sawback turtle (Graptemys oculifera) was assessed at two sites on the Pearl River, Mississippi, USA. This information was used to model the likely effects of human disturbance on the body temperature of large female turtles. 3. Both the duration of basking and the percentage of turtles in a basking group that abandoned their basking perch were influenced by the availability of alternative basking sites, frequency of boat traffic, boat speed enforcement zones (‘wake’ or ‘no wake’), boat type, air temperature, weather conditions, and Julian day. A simulation model of anthropogenically disturbed and undisturbed adult female ringed sawbacks showed that disturbance probably decreases the mean daily body temperature, especially at higher disturbance probabilities and in May, when the air temperature is low. 4. Management recommendations are provided that minimize the adverse effects of unintentional human disturbance of the ringed sawback to enable more effective conservation of this endangered, endemic species. Such methods are likely to be useful elsewhere as recreational disturbance becomes a greater conservation issue worldwide.

KEYWORDS

behaviour, habitat management, physiology, recreation, reptiles, river

1 | INTRODUCTION (Şekercioğlu et al., 2011) on animal populations are well known, but a consensus on the population‐level effects of indirect forms of The direct adverse effects of human activities such as hunting (Rosser human disturbance has yet to be reached (Tablado & Jenni, 2017). & Mainka, 2002), pollution (Trevors & Saier, 2009), and habitat loss Indirect disturbance is typically unintentional (Buchholz & Hanlon,

2012) and results from activities such as ecotourism (Müllner, Ringed sawbacks are slow to mature (2.5 to 4.5 years in males, 10 Linsenmair, & Wikelski, 2004), outdoor sports (Arlettaz et al., 2007), to 16 years in females) and lay few eggs per clutch (X = 3.7) (Jones, hiking (Taylor & Knight, 2003), and recreational boating (Bulté, 2006). Because they typically produce only one clutch per year Carriere, & Blouin‐Demers, 2010; Moore & Seigel, 2006). (Jones, 2006), and basking has been theorized to play a role in clutch Unfortunately, even when not focused on the wildlife itself, human development (Moore & Seigel, 2006), it may be that the loss of disturbance may be perceived as a predation threat that elicits an basking substrate coupled with basking interruptions during the sum- escape response (Blumstein, 2006; Blumstein, 2016). Repeated mer months may be adversely affecting egg formation (Hammond unnecessary escape actions may reduce population viability through et al., 1988). We hypothesize that human disturbance of basking direct costs such as energy expenditure and through missed opportu- may be contributing to the population decline of this species at mul- nities for feeding, reproduction and thermoregulation (Frid & Dill, tiple sites (Jones, 2017). 2002). The objectives of this study were to characterize the indirect Human disturbance of rivers can be particularly problematic for human disturbances occurring in the habitat of the endangered aquatic species because the linear nature of this habitat limits the ringed sawback, to investigate how human and environmental factors spatial escape options available to them (Bodie & Semlitsch, 2000). interact to affect basking, and to model how repeated human distur- This problem is especially acute for air‐breathing animals that bance of basking is likely to affect the mean body temperature of the cannot remain below the water surface indefinitely (Kramer, 1988). largest females. The largest females are probably the only age class If the species is also ectothermic, exposure to solar radiation by that lays more than one clutch of eggs per breeding season (Jones, aerial basking may be essential for maintaining a body temperature 2006), and thus are crucial to population recovery. We predict that that promotes energy assimilation through digestion (Hammond, frequent boat traffic and the rarity of basking structures on some Spotila, & Standora, 1988), enables egg development (Lindeman, parts of the river will result in fewer basking turtles and reduced 1999; Moore & Seigel, 2006), facilitates the growth of juveniles basking duration, with adverse consequences for maintaining an opti- (Bulté & Blouin‐Demers, 2010), and fosters other metabolic activities mal body temperature. (Lindeman, 2013). Thus, disturbances by passing watercraft (Moore & Seigel, 2006; Selman, Qualls, & Owen, 2013), and de‐snagging (i.e. the removal of logs and branches that serve as aerial basking structures for turtles) may be contributing to the widespread decline and endangerment of turtles, by reducing their metabolic rate (Jain‐Schlaepfer, 2 | METHODS Blouin‐Demers, Cooke, & Bulté, 2017), elevating stress (Selman et al., 2013), and impairing health (Galois & Ouellet, 2007; Selman 2.1 | Study site et al., 2013; Selman & Qualls, 2009). Although non‐destructive disturbance is not among the top causes of endangerment of chelo- Two study sites were chosen on the Pearl River near Jackson, nian taxa (Turtle Conservation Coalition, 2018), increasingly it will Mississippi, USA based on apparent differences in human disturbance be a contributing factor to population decline as ecotourism and the (Jones, 2017; Jones & Hartfield, 1995; Figure 1). Ratliff Ferry (RF) recreational use of protected areas continues to expand globally (Latitude −89.86°, Longitude 32.62°) is upstream of the Ross Barnett (Hogenson, 2017). Reservoir. This site comprises the 4.83 km of river between the Ratliff Turtle species in the genus Graptemys, the majority of which are Ferry Trading Post and a large sandbar colloquially known as ‘Flag threatened or endangered (Lindeman, 2013), bask more than other riv- Island’. RF is within the Pearl River Wildlife Management Area erine turtles (Ryan & Lambert, 2005). Thus, despite their primarily (PRWMA) where de‐snagging is prohibited, and the river is not chan- southern distribution in the USA, Graptemys species are probably nelized. However, the naturalistic setting and high numbers of sand- more susceptible to the adverse effects of disturbed basking than bars make it attractive for boating and other recreational activities. other riverine turtles. Boat traffic is high, particularly at weekends and holidays (Jones, The federally threatened ringed sawback (Graptemys oculifera), 2017), presenting a significant potential disturbance to turtles. All also known as the ringed map turtle, is found solely in the Pearl River three of the ‘no wake’ zones at RF, where watercraft must operate and its tributaries in Mississippi and Louisiana, USA (Jones & Selman, at the minimum speed that maintains steering and headway so as to 2009). The Pearl River is a narrow, 677‐km‐long forested river that minimize waves that cause shore erosion, and the two intervening flows southwards from central Mississippi through Louisiana and ‘wake’ zones, were included in the study. drains into the Gulf of Mexico through Lake Borgne and the The second study site is downstream of the Ross Barnett Mississippi Sound (Stolpe, Guo, Shiller, & Hassellöv, 2010). Despite Reservoir (Latitude 32.33°, Longitude 90.15°). The Lakeland site the species being classified as endangered by its home states, the (LL) (Jones, 2017) comprises the 4.83 km upstream of the Pearl proximity of the Pearl River to urban areas has meant that its habitat River access boat ramp within LeFleur's Bluff State Park and lacks is repeatedly targeted for river modifications that seem incongruous ‘no wake’ zones. The highly variable water levels at LL discourages with species recovery (Lynch, 2008; Lynch, 2009; The Associated boating use, and may limit the availability of basking structures for Press, 2015). turtles. HEPPARD AND BUCHHOLZ 3

FIGURE 1 Location of two study sites on the Pearl River outside Jackson, MS (image a). Ratliff Ferry Trading Post (RF, image b) is upstream of the Ross Barnett Reservoir in Canton, MS while Lakeland (LL, image c) is in Jackson, MS. Basking structures are indicated by black circle points in images (b) and (c) 4 HEPPARD AND BUCHHOLZ

2.2 | Basking duration observations and human 2.3 | Basking group‐size surveys disturbance Fourteen morning and 12 afternoon basking surveys at RF and six During the summer of 2017, assessment of behaviour and distur- morning and two afternoon surveys at LL were conducted. These bance alternated between sites such that they took place at RF observations were taken via a passing jonboat at a speed (3.2 km between 21 May – 2 June and 30 June – 15 July, and at LL between h‐1) and distance (75–100 m) to minimize turtle disturbance. Air and 7 June – 14 June and 16 July – 22 July. Each site was divided into water temperatures at a depth of 0.6 m were taken at three points three consecutive 1.61 km sections, and human disturbance was during the survey, at the beginning, halfway point, and end, and aver- assessed in each section. A study of a related species, the yellow‐ aged. Surveys began at one end of the site and continued until the last blotched map turtle (Graptemys flavimaculata), showed mean home basking structure at the opposite end to minimize potential distur- range lengths of 1.8 km for males and 1.5 km for females (Jones, bance events due to the research vessel. It was assumed that the pop- 1996). Therefore, if the ringed sawback has similar home range ulations in our study sites were closed to significant immigration, movements, the river section where a turtle was observed probably emigration and mortality during the study period. Population survey represents the amount of disturbance experienced by that individual results are reported as mean number (± SD) of turtles per structure. over recent time. At the start of the study, basking structures with two or more 2.4 | Modelling body temperature turtles present during three days of assessment were chosen for continued observation – this eliminated any smaller structures that Thermal modelling gives important insight into the thermal ecology of were potentially only accessible to juveniles and small males. Summer organisms when it is difficult or impossible to extract the same informa- precipitation was the fourth highest on record (National Weather tion from live organisms (Dubois, Blouin‐Demers, Shipley, & Thomas, Service, 2018), resulting in high river levels early in the study period. 2009; Dzialowski, 2005; Jain‐Schlaepfer et al., 2017; Yagi & Litzgus, Additional, previously submerged, basking structures, were added as 2013). Heat loss or gain, in water or the air, are each a function of the the field season progressed and the river level dropped. The maximum difference between body temperature (Tb) and the surrounding envi- density of basking structures per river section was calculated from ronment (Dzialowski & O'Connor, 2001; Jain‐Schlaepfer et al., 2017). GPS location points (Garmin eTrex 10) imported into ArcGIS (ESRI, For a particular body mass, these proportions serve as thermal 2014 vs. 10.2.2). constants (Dzialowski & O'Connor, 2001; Jain‐Schlaepfer et al., 2017) Observations of basking behaviour and boat traffic were made that can be used to predict Tb when air temperature (Ta), and water with a spotting scope from a sandbar or jonboat located on the temperature (Tw) are known. opposite bank of the river from a basking structure, approximately 100–250 m away. Basking turtles were observed for a total of 6 h between 08:30 and 16:30, with results calculated over 3 h morning 2.5 | Simulation of disturbance and 3 h afternoon time periods. Only one focal structure was under observation per observation period. Although the individual turtles The method of Jain‐Schlaepfer et al. (2017) was used to simulate the were not marked, duration was calculated for a single individual by effect of disturbance on the Tb of large female ringed sawbacks. The opportunistically observing an animal from the point it emerged from simulation uses heating and cooling constants, Ta and Tw, optimal the water until re‐entry, while noting whether basking ended prema- body temperature (Topt), and duration of basking and of post‐basking turely owing to an obvious disturbance. If turtles terminated basking submergence in water to estimate the mean daily Tb of a 1500 g turtle by entering the water because of passing watercraft, the type of across a range of disturbance probabilities. The thermal constants boat was recorded. Boat types observed included motorboats, empirically determined by Jain‐Schlaepfer et al. (2017) from experi- jonboats, anglers, kayaks, personal watercraft (pwc), and airboats. mentation with G. geographica were used with a mean mass equivalent Anglers were defined as any slow moving boat containing a person to the largest females of the study species. The body temperatures of who is actively fishing. Boat traffic was quantified at each site by undisturbed and anthropogenically disturbed turtles were compared recording the number and type of passing boats. The percentage of under nine disturbance rates: 0, 0.01, 0.05, 0.1, 0.2, 0.3, 0.5, 0.7, and turtles on a focal structure that were disturbed due to each passing 1 per hour. Four time periods were chosen for simulations based on watercraft was recorded as well. Boat traffic (number of boats per activity periods from Jones (2006): 27–31 May (early nesting, primar- hour) was calculated separately for each 3 h morning or afternoon ily for large females), 13–17 June (peak nesting), 13–17 July (late window. Because turtles were not individually marked, not all data nesting, primarily for large females with a second clutch), and 1–5 points are independent, although assumptions of independence were August (post‐nesting). It was assumed that the turtles begin the day made for analyses as is common in studies of disturbance (e.g. with Tb =Tw, and then bask without returning to the water until either

Buchholz & Hanlon, 2012; Moore & Seigel, 2006; Selman et al., they have reached Topt, they are disturbed, or at 18:00, the time of day

2013). when most basking ceases (R. Jones, pers. comm). Hourly Ta and Tw HEPPARD AND BUCHHOLZ 5 from each of the four given activity periods were averaged over five (the dependent variable) and eight independent variables: Julian day, days in 2016 and 2017, as recorded by the United States Geological weekday, weather, air temperature, water temperature, zone type Survey (USGS) NSTL Station on the Pearl River, located in Hancock (‘no wake’ or ‘wake’), density of basking structures and average boat County, MS (Lat 30°21′08“, Long 89°38’45” NAD27). Paired compar- traffic of the river mile of the basking structure. This model included isons (matched for date and time) with the infrequent (n = 68) local observations when no turtles were seen on structures. The best statis- measurements collected during basking surveys, showed no significant tical model for explaining basking group size contained the effects of difference between temperature values from Hancock County and weekday, Julian day, weather, air temperature, boat traffic, and the this study site (Ta, t = 0.5120, P = 0.61; Tw, t = −0.8012, P = 0.42). interaction between weekday and Julian day on the number of basking

To determine Topt under conditions where the species was completely turtles. Descriptive statistics on group size only include observations habituated to humans, four captive adult ringed sawbacks (two males, of structures where turtles were present, as by definition a group two females) were observed voluntarily basking under infrared heat cannot be zero. lamps in a climate controlled (Tw ≈ 29° C; Ta ≈ 26° C) naturalistic To assess the effect of environmental factors and human distur- exhibit at the Mississippi Museum of Natural Science, (Jackson, MS) bance on the duration of basking (the dependent variable), a multiple during daytime over 4 days, while their carapace temperatures were regression was carried out using seven independent variables: Julian simultaneously recorded every 10 min by Thermochron iButtons day, weekday, air temperature, water temperature, and the zone type (Model # DS1921H, Maxim Integrated San Jose, CA) attached dorsally (‘wake’ or no ‘wake’), number of basking structures within the river with marine epoxy. The Topt value used in the model is the mean of mile of observation, and boat traffic from the morning or afternoon carapace temperatures (31.6° C ± 2.8, range 26.0–38.5° C) of the period when the turtle was observed. The best model of independent two turtles that basked while under observation. One turtle basked variables explaining the duration of basking contained the effects of four times for a total of 290 min and the other basked three times Julian day, weekday, air temperature, water temperature, zone, num- for a total of 440 min. The simulation assumed that the preferred car- ber of basking structures, boat traffic, and interaction of number of apace temperature under these conditions represents the Topt of wild, basking structures and boat traffic. free‐living turtles in the absence of predation risk. Focal observations The odds of turtles ceasing basking through disturbance was calcu- in the field, as described earlier, provided a data set of cooling times lated using a logistic regression with the same initial predictors as the (duration in water) after a disturbance that was sampled at random previous multiple regression. The model with the lowest AIC value for for the simulation. For the disturbance simulation, time to return from calculating the odds of a turtle being disturbed contained weekday, basking included only cases where all the turtles on a basking perch Julian day, and the density of basking structures (Cox & Snell plunged into the water owing to passing watercraft, and was calcu- R2 = 0.24). Monday was chosen as the reference day with which other lated as the latency after disturbance to re‐emergence of a turtle onto days were compared when calculating the odds of disturbance. the basking structure. This was assumed to be the submerged time of A Chi‐square analysis was used to test whether type of disturbance, one of the same individuals disturbed from the perch immediately pre- including non‐watercraft, was related to whether a turtle was dis- viously. For undisturbed simulations, caused by difficulties in tracking turbed. A two‐way ANOVA tested data from all passing boats to assess unmarked individuals, the latency to begin basking after termination whether zone (‘wake’ or ‘no wake’) or watercraft type (independent was calculated as the time between a disturbance incidence by a variables) affected the percentage of basking turtles (the dependent natural disturbance (another turtle) to emergence of the next basking variable) disturbed per structure. As there was no ‘no wake’ zone at turtle as this provides a contrast to the effect of human disturbance. LL to compare with RF, and few boats at LL (only 11 out of 676 total This data set included observations from a previous exploratory field boats), the effect of location on boat type effects was not examined. season (27 June and 6 August 2016). The time to return from A post hocTukey test was used to assess the significance of differences human‐disturbed basking was 15.91 ± 17.29 min (n = 22) and non‐ in percentages of disturbance between boat types. The results are human disturbance was 6.73 ± 6.81 min (n = 46). given as means ± standard deviation unless otherwise noted. The results of the simulations of disturbance effects on modelled body temperatures were subjected to a multiple regression, to test 2.6 | Statistical tests whether differences in body temperatures (the dependent variable) varied between human‐disturbed and undisturbed simulations across All statistical tests were carried out using the statistical program R probabilities of disturbance (the independent variables). Studio (version 3.4.2; R Core Team, 2017) with an α of 0.05. All facto- rial ANOVAs used a type II Sum of Squares model. Before performing | the multiple regression tests described below, model residuals were 3 RESULTS confirmed as normally distributed and did not exhibit multicollinearity. Statistical testing was of the multivariate model with the lowest AIC 3.1 | Basking structures and basking value but excluding predictors with RVI scores less than 0.5. A multiple regression was carried out to assess the relationship Ratliff Ferry had more basking structures (X= 13.0 ± 6.6) available per between the number of basking turtles seen per structure on surveys river section than LL (X= 8.3 ± 1.7). Observations on the number of 6 HEPPARD AND BUCHHOLZ

FIGURE 2 Fewer turtles are seen basking as boat traffic increases (n = 1791 turtles). Each point represents an observation of group size on a basking structure at either Lakeland (LL) or Ratliff Ferry (RF). Raw data are shown. See text for results of multiple regression model results

turtles basking on 1036 structures were recorded, summing to 1791 as the study period progressed (multiple regression, F(1,1019) = 62.25, turtles over the field season. Basking group size ranged from 1 to slope estimate = −0.01, P < 0.001), and when there was higher boat

19 (X= 3.26 ± 2.88 turtles) at RF, and from 1 to 7 (X=1.84 ± 1.56 traffic (multiple regression, F(1,1019) = 8.26, slope estimate = −0.05, turtles) at LL. P = 0.004; Figure 2). Group sizes of turtles on structures were larger Between 23 May and 22 July 2017, 123 h (RF, 75 h; LL, 48 h) on Tuesday (5.23 ± 3.47 turtles), Wednesday (3.04 ± 2.21 turtles), of focal basking observations were made. During this time there and Thursday (3.06 ± 3.13 turtles), compared with Friday were 462 observations of start‐to‐finish basking by turtles: 96 turtles (2.56 ± 2.52 turtles), Saturday (2.67 ± 2.62 turtles) and Monday at LL and 366 turtles at RF. Duration of basking ranged (2.04 ± 1.32 turtles) (multiple regression, F(5,1019) = 3.81, P = 0.007); from <1 to 566 min (X=34.36 ± 50.11 min) at RF and from <1 to the interaction of Julian day and weekday resulted in fewer basking 297 (X=39.68 ± 47.55 min) at LL. turtles (multiple regression, F(5,1019) = 3.03). Weather had a significant effect on the number of basking turtles (multiple regression,

F(3,1019) = 5.35, P = 0.001), with the fewest turtles seen on rainy | 3.2 Boat traffic days (Figure 3).

In total, 675 boats were recorded during focal observations (motorboats, n = 502; jonboats, n = 77; anglers, n = 2; pwc, n = 90; kayaks, 3.4 | Duration of basking n = 2; airboats, n = 2), and 571 of these boats passed by structures that had basking turtles. Only three boat types ‐ jonboats, motorboats, Duration of basking increased across the field season (multiple regres- and kayaks (n = 11 boats) ‐ were seen at LL while jonboats, motor- sion, F(1, 449) = 29.63, slope estimate = 2.41, P < 0.001), but turtles boats, anglers, pwc, airboats, and kayaks were seen at RF (n = 665 boats). The highest number of boats was seen on the Saturday of Memorial day weekend at RF ( X=64.67 ± 19.33 boats per hour, n = 388). No boat traffic was observed at LL Monday through Thursday, and was low at the weekend (Friday, 0.33 ± 0.47; Saturday, 0.50 ± 0.71) relative to the much greater daily traffic at RF (Monday, 2.00 ± 0.47; Tuesday, 1.25 ± 1.00; Wednesday, 2.21 ± 1.15; Thursday, 3.53 ± 2.22; Friday, 2.67 ± 1.23; Saturday, 21.17 ± 12.96). Boat traffic was not associated with Julian day (Kendall's tau, tau = −0.12, P = 0.32).

3.3 | Survey counts of basking group size FIGURE 3 Weather had a significant effect on the number of turtles seen basking per structure during surveys (n = 1791 turtles). Overcast Air temperature had a positive association with the number of basking days had complete cloud cover, partly cloudy days were mostly cloudy turtles observed on structures (multiple regression, F(1, 1019) = 59.04, with some visibility of the sun, and sunny days had very few clouds. slope estimate = 0.36, P < 0.001), but fewer basking turtles were seen Bars represent standard error HEPPARD AND BUCHHOLZ 7 basked for shorter durations when air temperature was higher boat traffic (64.67 ± 19.33 boats per hour) which was the Saturday

(multiple regression, F(1,449) = 35.24, slope estimate = −22.45, of Memorial Day weekend (27 May in 2017). P < 0.001). Turtles basked for longer in ‘no wake’ zones (n = 192 turtles, Compared with Monday, turtles were 1.4x less likely to be X=41.69 ± 64.41 min) than ‘wake’ zones (n = 270 turtles, disturbed on a Tuesday (z = −3.3, P = 0.001), 1.1x less likely to be disturbed on a Wednesday (z = −2.7, P = 0.006), and 1.2x less likely to be X=31.03 ± 34.95 min) (multiple regression, F(1,449) = 5.58, ‘wake’ zone slope estimate = −21.72, P = 0.02; Figure 4a), and when there were disturbed on a Thursday (z = −3.1, P = 0.002). Other comparisons between days were not statistically significantly different. Basking tur- fewer basking structures (multiple regression, F(1,449) = 24.16, slope estimate = −2.29, P < 0.001). Turtles basked for longer when high boat tles were less likely to be disturbed as the field season progressed traffic was associated with high basking structure density (multiple (logistic regression, odds = −0.05, z‐‐5.5, P < 0.001) and were more likely to be disturbed when more basking structures were available regression, F(1, 449) = 6.68, slope estimate = 0.55, P = 0.01), despite boat traffic and basking structure density having independent negative (logistic regression, odds = 0.09, z = 5.2, P < 0.001). effects on duration of basking (slopes = −2.29 and − 14.82 respectively). 3.6 | Disturbance type

Watercraft type significantly affected the percentage of disturbed tur- 3.5 | Odds of disturbance tles (ANOVA, F(5,562) = 8.4806, P < 0.0001). Higher percentages of basking turtles were disturbed by kayaks and anglers than other water- Half (50.4%) of the turtles observed during focal observations were craft. The type of disturbance significantly affected not just the disturbed from their basking perch, 65% of those by human causes. percentage of turtles disrupted from each perch but also the total num- A higher percentage (59%) of basking turtles were disturbed at RF ber of individuals disturbed (n = 233 turtles disturbed, χ2 goodness of compared with LL (17%). Boats disturbed a higher percentage of tur- fit, χ2 = 522.76, df = 11, P < 0.001), with motorboats (39.9% of total), tles when passing in a ‘wake’ zone (n = 154 boats, other turtles (30.5% of total), and jonboats (13.7% of total) causing ‘ ’ X=19.78 ± 34.20%) than in a no wake zone (n = 417 boats, the greatest total number of turtles to abandon basking. A detailed dis-

X=2.28 ± 10.01%, F(1, 562) = 67.13, P < 0.001; Figure 4b). The observed tribution of disturbance types is provided by Heppard (2018). probability of disturbance per hour was 0.02 in May, 0.01 in June, 0.004 in July, and 0.02 in August. The observed probability of distur- 3.7 | Simulation of disturbance effects on Tb bance each hour was 0.16 during the day of highest observed hourly

The interaction between activity period and probability of hourly disturbance significantly affected the difference in simulated body temperature between human‐disturbed and undisturbed turtles (multiple

regression, F(3,35992) = 2990.9, P < 0.001). The difference between Tb of human‐disturbed and undisturbed turtles increased as the probability of disturbance increased, and was highest in May, followed by June, August, and July (Figure 5).

4 | DISCUSSION

4.1 | Metabolic consequences of disturbance

These results demonstrate that ringed sawbacks at the highest observed disturbance rate (0.20) could have a 1–2°C lower body temperature than undisturbed individuals, depending on the month. In other turtles this range of temperature difference has been shown to be relevant to post‐ingestive habitat preference (Gatten, 1974) and temperature preference after injection with bacterial antigens (do Amaral, Marvin, & Hutchison, 2002). Standard metabolic rate has not been measured in ringed sawbacks, but if a temperature‐dependent rate similar to the common map turtle is assumed (Jain‐Schlaepfer FIGURE 4 a) Duration of basking by turtles (n = 462 turtles) was et al., 2017), even minimal disturbance (0.01) in May would reduce lower in ‘wake’ zones (P = 0.02); b) Percentage of basking turtles metabolism of the ringed sawback by 2.7%. Lowered metabolism has disturbed by a passing watercraft (n = 571 boats) was greater in ‘wake’ zones (P = 0.001). Error bars represent standard error. Raw data are several adverse fitness outcomes for turtles. The metabolic rate deter- shown. See text for results of multiple regression model results mines growth rate (Bulté & Blouin‐Demers, 2010; Litzgus & Hopkins, 8 HEPPARD AND BUCHHOLZ

FIGURE 5 Differences in simulated body temperatures of undisturbed and human‐ disturbed adult female ringed sawbacks in the Pearl River over different probabilities of disturbance during four activity periods: 27– 31 May (early nesting, primarily for large females), 13–17 June (peak nesting), 13–17 July (late nesting, primarily for large females with second clutch), and 1–5 August (nesting ceased)

2003; Williamson, Spotila, & Standora, 1989), and slow growth may other turtles for access to basking substrate under conditions of prolong periods of size‐dependent predation risk of juveniles (Jones, greater snag availability. Indeed, this study showed that other turtles 2017). In adult females, low metabolism may result in reduced were the second most common reason for individual ringed sawbacks maternal investment in eggs (Rowe, Woodland, & Funck, 2017; to stop basking. Similarly, Polo‐Cavia, López, and Martín (2010) found Steyermark & Spotila, 2000). Ultimately this chronic reduction in that aggressive, invasive red‐eared sliders reduced the basking activity metabolism could impede individual survival and population growth and perch usage of native, endangered Spanish terrapins (Mauremys in disturbed populations. leprosa). Thus, the population‐level influence of human disturbance The simulation of ringed sawback metabolism is based on the best on threatened riverine turtles may depend on the combination of available evidence for this threatened species, but requires additional the rate of human disruption, basking perch density, the number of study of a key modelling variable in free‐living individuals. TheTopt used other turtles of the same species, and interspecific competitive inter- in the simulation was recorded from the carapace temperature of cap- actions, rather than simply human disturbance alone. tive individuals under an artificial heat source. Carapace temperature is known to be positively correlated with Tb, but it also reaches greater 4.2.2 | Boat wakes extremes of temperature than the body core where most metabolism is occurring (Edwards & Blouin‐Demers, 2007). Thus, carapace tempera- Considering that the present study showed that there are longer ture might over‐estimate Topt during basking, and show more rapid basking durations in ‘no wake’ zones, a higher percentage of basking cooling during immersion than the core. Also it is not known how body turtles that are disturbed by boats passing in ‘wake’ zones, and that condition, constant Tw, and the invariable, but low intensity, artificial there is lack of a statistical effect of the number of passing boats, it light source relative to natural sunlight may affect the temperature can be concluded that the wakes of boats appear to affect basking preferences of long‐term captive turtles compared with wild individ- more often than the mere presence of boats and boaters. On the Pearl uals. The field deployment of surgically implanted, temperature data River, wakes created by multiple passing boats compound each other loggers is the next step needed to better understand the thermal and sometimes completely submerge basking structures. In addition, preferences of ringed sawbacks and to assess the relevance of the large, slow‐moving boats that produce larger wakes (Selman et al., simulation to management needs under natural conditions. 2013) are known to cause more displacement of turtles than small, Although the simulation suggests that the large, female ringed fast‐moving boats (Moore & Seigel, 2006). Some authors have sug- sawbacks at this study site will find it more difficult to achieve Topt gested that wildlife will habituate to human presence over time when human disturbance is frequent and Ta is low, the results also (Blumstein, 2006), and thus learn to avoid the opportunity costs of indicate that the adverse impact of human disturbance on basking is responding to harmless humans as though they were predators. likely to depend on a number of additional biotic and abiotic factors Indeed Graptemys spp. sometimes habituate to human activity that have not been considered by previous researchers. (reviewed by Lindeman, 2013). In the case of riverine turtles, however, if habituation results in tolerance of passing boats, the turtles may still 4.2 | Mechanisms of physical disturbance from fail to reachTopt through being swept off the structure by the action of basking structures boat wakes. The surprising positive interaction between boat traffic and basking structure density in their effect on duration of basking, 4.2.1 | Competition is suspected to be an artefact of a single river section (#3) at RF where multiple ‘no wake’ zones happened to co‐occur with an abundance of Unexpectedly, turtles terminated basking sooner where there were basking structures (Figure 1B), and this area was also especially popu- more basking structures, possibly because of less competition with lar with holiday boaters (as high as 84 boats per hour) because of an HEPPARD AND BUCHHOLZ 9 attractive sandbar there. Uniquely this finding suggests that increases Huey, & Nevo, 1982). Even though the motor activity that allows in boat traffic can be mitigated to some extent by providing greater abandonment of aerial basking (i.e. walking off the log) does not basking perch abundance and by reducing boat speed and the inter‐ appear to be difficult, anecdotal evidence suggests that turtles may wake interval of passing boats. tradeoff basking duration with predation risk. For example, Boyer (1965) remarked that individuals that recently climbed out of the 4.3 | Mechanisms of behavioural disturbance from water onto a basking perch, and presumably have a Tb similar to basking structures Tw, are more likely to abandon basking upon disturbance than other individuals on the same structure who had been basking for some

4.3.1 | Approach speed time and thus have a higher Tb. Although Northern map turtles (Graptemys geographica) do not swim faster when warmer, Ben‐Ezra, There were three primary human sources of disturbance that Bulté, and Blouin‐Demers (2008) found that they take longer to right disrupted the greatest total number of turtles from their basking themselves at cooler temperatures in their preferred temperature perches: motorboats, anglers, and pwc. Similarly, kayaks and anglers range, a behaviour perhaps more relevant to escaping a terrestrial disturbed the greatest percentage of turtles from focal basking groups. predator than swimming. Optimization of basking could also be Moore and Seigel (2006) found that slow‐moving watercraft typically indicated by behaviour that we did not measure in our study, such remain out of the main river channel, and their closer approach to as flight initiation distance (Polich & Barazowski, 2016). Cooler basking structures may be perceived as a greater threat by the turtles. ectothermic animals may prepare for their slower escape by starting Speed of predator movement may also be used as a cue by potential to escape at a greater distance from the approaching threat, as prey to assess the risk of predator attack. Burger (2001) reported that occurs in northern water snakes (Nerodia sipedon) (Weatherhead & pedestrians who stop to watch basking snakes cause more snakes to Robertson, 1992). flee than equidistant pedestrians who walk past.

4.3.2 | Temperature and perceived risk 4.4 | Recommendations for further research and management Ringed turtles might not bask as frequently earlier in the season in order to conserve heat that would have been lost during basking Given projected global increases in recreational boating, disturbance due to heat transfer to low Ta and evaporation of the river water on of basking should receive additional research attention from conserva- their carapace (Boyer, 1965; Brattstrom, 1965). Indeed Ta did not con- tion scientists. A priority for research is to test the assumption that sistently remain several degrees higher than water temperatures until promoting aerial basking will result in increased population recruit- July, when basking censuses found more turtles basking, and focal ment. Meanwhile, for already threatened riverine turtle species and observations found them to bask for shorter periods (i.e. they could declining sub‐populations it is prudent to minimize disturbance by achieve Topt more quickly). However, in addition to the low efficiency managing watercraft more effectively. Because boat wakes are a of basking as a warming strategy early in the season, various lines of major source of basking disruption for the ringed sawback, for exam- evidence suggest that ectotherms perceive greater predation risk at ple, the entire PRWMA set aside to protect this species should be low Tb that might affect their willingness to bask aerially. designated for ‘no wake’ boating. Doing so would not only minimize Except where habituated to frequent human activity, adult, wake‐induced termination of basking but would also reduce the aerially‐basking turtles appear to be quite fearful. They tend to avoid destruction of nests caused by the erosion of sandbars (Selman basking perches that connect to the shoreline whereby terrestrial et al., 2013). After site‐specific further study, restrictions on boaters predators may reach them (Waters, 1974), they may refuse to bask could be limited to the times of year when basking maximizes if unable to view their surroundings for approaching threats (Boyer, metabolism directly affecting nutrient assimilation, egg formation, 1965), they are more likely to join basking groups than use unoccu- and somatic growth. Also, basking refugia could be established by pied structures nearby (Lindeman, 2013), and if one turtle suddenly restricting boat access from inlets of main river channels, perhaps with abandons basking, a chain reaction of fleeing by other group mem- floating, basking structures, anchored so that they remain above water bers sometimes occurs (Selman & Qualls, 2011), even in the absence during flooding. Closing road access to river sections with critical of a potential predator. As ectotherms, turtles should hypothetically basking and nesting areas might reduce disturbance by making them prefer to bask under environmental conditions thermally favourable less convenient places for human recreation. to motor function so that they can escape predators quickly (Huey, Worldwide, most threatened turtle species have declined because 1982). Indeed, cooled ectotherms may refuse to leave shelters in of habitat destruction, and unsustainable harvest for human consump- order to bask if predation risk is perceived to be high (Martín & tion and for the pet trade (Turtle Conservation Coalition, 2018). Even López, 1999). The ringed sawbacks in the present study were more when actively protected from these factors, however, populations may likely to be observed basking as Ta increased, suggesting that they suffer disturbance effects from other human uses of protected are unwilling to leave the relative safety of the water to bask at areas. As both ecotourism (Hogenson, 2017) and recreational boat cooler Ta when they might be less efficient at movement (Hertz, use (e.g. Bulté et al., 2010; Gonson et al., 2016; Steckenreuter, 10 HEPPARD AND BUCHHOLZ

Harcourt, & Möller, 2012) continue to grow internationally, research Bulté, G., Carriere, M.‐A., & Blouin‐Demers, G. (2010). Impact of recrea- efforts to understand their impact, and management to minimize their tional power boating on two populations of northern map turtles (Graptemys geographica). Aquatic Conservation: Marine and Freshwater adverse direct and indirect effects on threatened riverine species, Ecosystems, 20,31–38. https://doi.org/10.1002/aqc.1063 must keep pace for susceptible aquatic organisms to persist in parks Burger, J. (2001). The behavioral response of basking Northern water and reserves. (Nerodia sipedon) and Eastern garter (Thamnophis sirtalis) snakes to pedestrians in a New Jersey park. Urban Ecosystems, 5, 119–129. ACKNOWLEDGEMENTS https://doi.org/10.1023/A:1022339704784

We are grateful to Robert Jones, Mississippi Department of Wildlife, do Amaral, J. P. S., Marvin, G. A., & Hutchison, V. H. (2002). The influence of bacterial lipopolysaccharide on the thermoregulation of the box Fisheries, and Parks, whose expertise in Graptemys and field turtle Terrapene carolina. Physiological and Biochemical Zoology, 75, ‐ techniques were invaluable. Sofia Jain Schlaepfer kindly provided the 273–282. https://doi.org/10.1086/341816 simulation R coding. Other assistance was provided by Barrett Dubois, Y., Blouin‐Demers, G., Shipley, B., & Thomas, D. (2009). Thermo- Aldridge, Scotlynn Farmer, Darian Raucher, Priya Sanipara, Chaz regulation and habitat selection in wood turtles Glyptemys insculpta: Hyseni and Stephanie Burgess. Funding and equipment were provided Chasing the sun slowly. Journal of Animal Ecology, 78, 1023–1032. ‐ by the University of Mississippi Graduate School, the Department of https://doi.org/10.1111/j.1365 2656.2009.01555.x Biology, and the Animal Behavior Research Fund of the University of Dzialowski, E. M. (2005). Use of operative temperature and standard oper- Mississippi Foundation. Brice Noonan, Carol Britson, Will Selman, ative temperature models in thermal biology. Journal of Thermal Biology, 30, 317–334. https://doi.org/10.1016/j.jtherbio.2005.01.005 and Grégory Bulté helped improve the manuscript. This is publication Dzialowski, E. M., & O'Connor, M. P. (2001). Thermal time constant estima- number 4 of The Center for Biodiversity and Conservation Research at tion in warming and cooling ectotherms. Journal of Thermal Biology, 26, the University of Mississippi. 231–245. https://doi.org/10.1016/S0306‐4565(00)00050‐4 Edwards, A. L., & Blouin‐Demers, G. (2007). Thermoregulation as a func- ORCID tion of thermal quality in a northern population of painted turtles, – Richard Buchholz https://orcid.org/0000-0003-3336-2676 Chrysemys picta. Canadian Journal of Zoology, 85, 526 535. https:// doi.org/10.1139/Z07‐037 ESRI. (2014). ArcGIS. Redlands, California, United States: ESRI. REFERENCES Frid, A., & Dill, L. (2002). Human‐caused disturbance stimuli as a form of Arlettaz, R., Patthey, P., Baltic, M., Leu, T., Schaub, M., Palme, R., & predation risk. Conservation Ecology, 6, 11. https://doi.org/10.5751/ Jenni‐Eiermann, S. (2007). Spreading free‐riding snow sports represent es-00404-060111 a novel serious threat for wildlife. Proceedings of the Royal Society of London B: Biological Sciences, 274, 1219–1224. https://doi.org/ Galois, P., & Ouellet, M. (2007). Traumatic injuries in eastern spiny softshell 10.1098/rspb.2006.0434 turtles (Apalone spinifera) due to recreational activities in the northern Lake Champlain basin. Chelonian Conservation and Biology, 6, 288–293. Ben‐Ezra, E., Bulté, G., & Blouin‐Demers, G. (2008). Are locomotor perfor- https://doi.org/10.2744/1071‐8443(2007)6[288:TIIESS]2.0.CO;2 mances coadapted to preferred basking temperature in the Northern Map Turtle (Graptemys geographica)? Journal of Herpetology, 42, Gatten, R. E. Jr. (1974). Effect of nutritional status on the preferred body 322–331. https://doi.org/10.1670/07‐1881.1 temperature of the turtles Pseudemys scripta and Terrapene ornata. Copeia, 4, 912–917. https://doi.org/10.2307/1442590 Blumstein, D. T. (2006). Developing an evolutionary ecology of fear: How life history and natural history traits affect disturbance tolerance in Gonson, C., Pelletier, D., Gamp, E., Preuss, B., Jollit, I., & Ferraris, J. (2016). birds. Animal Behaviour, 71, 389–399. https://doi.org/10.1016/j. Decadal increase in the number of recreational users is concentrated in anbehav.2005.05.010 no‐take reserves. Marine Pollution Bulletin, 107, 144–154. https://doi. org/10.1016/j.marpolbul.2016.04.007 Blumstein, D. T. (2016). Habituation and sensitization: New thoughts about old ideas. Animal Behaviour, 120, 255–262. https://doi.org/ Hammond, K. A., Spotila, J. R., & Standora, E. A. (1988). Basking behavior 10.1016/j.anbehav.2016.05.012 of the turtle Pseudemys scripta: effects of digestive state, acclimation temperature, sex, and season. Physiological Zoology, 61,69–77. Bodie, J. R., & Semlitsch, R. D. (2000). Spatial and temporal use of flood- https://doi.org/10.1086/physzool.61.1.30163738 plain habitats by lentic and lotic species of aquatic turtles. Oecologia, 122, 138–146. https://doi.org/10.1007/PL00008830 Heppard, J. M. (2018). Impact of human disturbance on the behavior and physiology of the endangered ringed sawback turtle (Graptemys Boyer, D. R. (1965). Ecology of the basking habit in turtles. Ecology, 46, oculifera). (MS thesis), University of Mississippi, Mississippi. 99–118. https://doi.org/10.2307/1935262 Hertz, P. E., Huey, R. B., & Nevo, E. (1982). Fight versus flight: Body tem- Brattstrom, B. H. (1965). Body temperatures of reptiles. American Midland perature influences defensive responses of lizards. Animal Behaviour, Naturalist, 73, 376–422. https://doi.org/10.2307/2423461 30, 676–679. https://doi.org/10.1016/S0003‐3472(82)80137‐1 Buchholz, R., & Hanlon, E. (2012). Ecotourism, wildlife management, and Hogenson, S. (2017). The case for responsible travel: Trends & Statistics 2017 behavioral biologists: Changing minds for conservation. In U. Candolin, (5th ed.). Center for Responsible Travel, Washington, D. C. www. & B. B. M. Wong (Eds.), Behavioural responses to a changing world: responsibletravel.org Mechanisms and consequences (pp. 234–249). Oxford, UK: Oxford University Press. https://doi.org/10.1093/acprof:osobl/978019960 Huey, R. B. (1982). Temperature, physiology, and the ecology of reptiles. 2568.003.0017 Biology of the Reptilia, 12,25–92. Bulté, G., & Blouin‐Demers, G. (2010). Estimating the energetic signifi- Jain‐Schlaepfer, S. M., Blouin‐Demers, G., Cooke, S. J., & Bulté, G. (2017). cance of basking behaviour in a temperate‐zone turtle. Ecoscience, 17, Do boating and basking mix? The effect of basking disturbances 387–393. https://doi.org/10.2980/17‐4‐3377 by motorboats on the body temperature and energy budget of the HEPPARD AND BUCHHOLZ 11

northern map turtle. Aquatic Conservation: Marine and Freshwater R CoreTeam (2017). R: A language and environment for statistical computing. Ecosystems, 27, 547–558. https://doi.org/10.1002/aqc.2693 Vienna, Austria: R Foundation for statistical computing. URL https:// ‐ Jones, R. L. (1996). Home range and seasonal movements of the turtle www.R project.org/ Graptemys flavimaculata. Journal of Herpetology, 30, 376–385. https:// Rosser, A. M., & Mainka, S. A. (2002). Overexploitation and species extinc- doi.org/10.2307/1565175 tions. Conservation Biology, 16, 584–586. https://doi.org/10.1046/ ‐ Jones, R. L. (2006). Reproduction and nesting of the endangered ringed j.1523 1739.2002.01635.x map turtle, Graptemys oculifera, in Mississippi. Chelonian Conservation Rowe, C. L., Woodland, R. J., & Funck, S. A. (2017). Metabolic rates are and Biology, 5, 195–209. https://doi.org/10.2744/1071‐8443(2006)5 elevated and influenced by maternal identity during the early, [195:RANOTE]2.0.CO;2 yolk‐dependent, post‐hatching period in an estuarine turtle, the Jones, R. L. (2017). Long‐term trends in ringed sawback (Graptemys diamondback terrapin (Malaclemys terrapin). Comparative Biochemistry oculifera) growth, survivorship, sex ratios, and population sizes in the and Physiology Part A: Molecular & Integrative Physiology, 204, – Pearl River, Mississippi. Chelonian Conservation and Biology, 16, 137 145. https://doi.org/10.1016/j.cbpa.2016.11.015 215–228. https://doi.org/10.2744/CCB‐1268.1 Ryan, T. J., & Lambert, A. (2005). Prevalence and colonization of Jones, R. L., & Hartfield, P. D. (1995). Population size and growth in the Placobdella on two species of freshwater turtles (Graptemys turtle Graptemys oculifera. Journal of Herpetology, 29, 426–436. geographica and Sternotherus odoratus). Journal of Herpetology, 39, https://doi.org/10.2307/1564994 284–287. https://doi.org/10.1670/180‐04N Jones, R. L. & Selman, W. (2009). Graptemys oculifera (Baur 1890)—Ringed Şekercioğlu, Ç. H., Anderson, S., Akçay, E., Bilgin, R., Can, Ö. E., Semiz, G., … map turtle, ringed sawback. Conservation Biology of Freshwater Sağlam, İ. K. (2011). Turkey's globally important biodiversity in crisis. Turtles and Tortoises: A Compilation Project of the IUCN/SSC Tortoise Biological Conservation, 144, 2752–2769. https://doi.org/10.1016/j. and Freshwater Turtle Specialist Group. Chelonian Research Mono- biocon.2011.06.025 graphs, DOI: https://doi.org/10.3854/crm.5.033.oculifera.v1.2009 Selman, W., & Qualls, C. (2009). Graptemys flavimaculata (yellow‐blotched Kramer, D. L. (1988). The behavioral ecology of air breathing by aquatic map turtle). Basking and parasite removal. Herpetological Review, 40, animals. Canadian Journal of Zoology, 66,89–94. https://doi.org/ 78–79. ‐ 10.1139/z88 012 Selman, W., & Qualls, C. (2011). Basking ecology of the yellow‐blotched Lindeman, P. V. (1999). Surveys of basking map turtles Graptemys spp. in sawback (Graptemys flavimaculata), an imperiled turtle species of the three river drainages and the importance of deadwood abundance. Bio- Pascagoula River system, Mississippi, USA. Chelonian Conservation and logical Conservation, 88,33–42. https://doi.org/10.1016/S0006‐ Biology, 10, 188–197. https://doi.org/10.2744/CCB‐0886.1 ‐ 3207(98)00093 7 Selman, W., Qualls, C., & Owen, J. C. (2013). Effects of human disturbance Lindeman, P. V. (2013). The map turtle and sawback atlas: Ecology, evolution, on the behavior and physiology of an imperiled freshwater turtle. The distribution, and conservation. Norman, Oklahoma: University of Okla- Journal of Wildlife Management, 77, 877–885. https://doi.org/ homa Press. 10.1002/jwmg.538 Litzgus, J. D., & Hopkins, W. A. (2003). Effect of temperature on metabolic Steckenreuter, A., Harcourt, R., & Möller, L. (2012). Are speed restriction rate of the mud turtle (Kinosternon subrubrum). Journal of Thermal zones an effective management tool for minimising impacts of boats Biology, 28, 595–600. https://doi.org/10.1016/j.jtherbio.2003.08.005 on dolphins in an Australian marine park? Marine Policy, 36, 258–264. Lynch, A. (2008). Lakes plan still alive, despite vote? Jackson, Mississippi: https://doi.org/10.1016/j.marpol.2011.05.013 Jackson Free Press. http://www.jacksonfreepress.com/news/2008/ Steyermark, A. C., & Spotila, J. R. (2000). Effects of maternal identity and mar/26/lakes‐plan‐still‐alive‐despite‐vote/ incubation temperature on snapping turtle (Chelydra serpentina) metab- – Lynch, A. (2009). Pearl wetlands worth saving? Jackson, Mississippi: Jackson olism. Physiological and Biochemical Zoology, 73, 298 306. https://doi. Free Press. http://www.jacksonfreepress.com/news/2009/jun/17/ org/10.1086/316743 pearl‐wetlands‐worth‐saving/ Stolpe, B., Guo, L., Shiller, A. M., & Hassellöv, M. (2010). Size and compo- Martín, J., & López, P. (1999). When to come out from a refuge: Risk‐ sition of colloidal organic matter and trace elements in the Mississippi sensitive and state‐dependent decisions in an alpine lizard. Behavioral River, Pearl River and the northern Gulf of Mexico, as characterized Ecology, 10, 487–492. https://doi.org/10.1093/beheco/10.5.487 by flow field‐flow fractionation. Marine Chemistry, 118, 119–128. https://doi.org/10.1016/j.marchem.2009.11.007 Moore, M. J., & Seigel, R. A. (2006). No place to nest or bask: Effects of human disturbance on the nesting and basking habits of yellow‐ Tablado, Z., & Jenni, L. (2017). Determinants of uncertainty in wildlife blotched map turtles (Graptemys flavimaculata). Biological Conservation, responses to human disturbance. Biological Reviews, 92, 216–233. 130, 386–393. https://doi.org/10.1016/j.biocon.2006.01.001 https://doi.org/10.1111/brv.12224 Müllner, A., Linsenmair, K. E., & Wikelski, M. (2004). Exposure to ecotour- Taylor, A. R., & Knight, R. L. (2003). Wildlife responses to recreation and ism reduces survival and affects stress response in hoatzin chicks associated visitor perceptions. Ecological Applications, 13, 951–963. (Opisthocomus hoazin). Biological Conservation, 118, 549–558. https:// https://doi.org/10.1890/1051‐0761(2003)13[951:WRTRAA]2.0.CO;2 doi.org/10.1016/j.biocon.2003.10.003 The Associated Press (2015). Conservationists: 'One Lake' endangers Pearl National Weather Service (2018). Jackson, Mississippi Climate Records. River. Jackson, Mississippi: Jackson Free Press. http://www. www.weather.gov/jan/climatejan jacksonfreepress.com/news/2015/apr/07/conservationists‐one‐lake‐ endangers‐pearl‐river/ Polich, R. L., & Barazowski, M. (2016). Flight initiation distance in a freshwater turtle, Chrysemys picta. Chelonian Conservation & Biology, 15, Trevors, J. T., & Saier, M. H. (2009). Testosterone: The cause of our world's 214–218. https://doi.org/10.2744/CCB‐1164.1 problems? Water, Air, & Soil Pollution, 200,1–2. https://doi.org/ 10.1007/s11270‐006‐9278‐8 Polo‐Cavia, N., López, P., & Martín, J. (2010). Competitive interactions during basking between native and invasive freshwater turtle species. Turtle Conservation Coalition. (2018). Turtles in trouble: The world's 25+ Biological Invasions, 12, 2141–2152. https://doi.org/10.1007/s10530‐ most endangered tortoises and freshwater turtles. Report, 2. http:// 009‐9615‐0 www.iucn‐tftsg.org/turtles‐in‐trouble‐2018/ 12 HEPPARD AND BUCHHOLZ

Waters, J. C. (1974). The biological significance of the basking habit in the Yagi, K. T., & Litzgus, J. D. (2013). Thermoregulation of spotted turtles black‐knobbed sawback, Graptemys nigrinoda Cagle. (MS thesis), (Clemmys guttata) in a beaver‐flooded bog in southern Ontario, Canada. Auburn University, Alabama. Journal of Thermal Biology, 38, 205–213. https://doi.org/10.1016/j. jtherbio.2013.02.010 Weatherhead, P. J., & Robertson, I. C. (1992). Thermal constraints on swimming performance and escape response of northern water snakes (Nerodia sipedon). Canadian Journal of Zoology, 70,94–98. https://doi. How to cite this article: Heppard JM, Buchholz R. Impact of org/10.1139/z92‐014 human disturbance on the thermoregulatory behaviour of the Williamson, L. U., Spotila, J. R., & Standora, E. A. (1989). Growth selected endangered ringed sawback turtle (Graptemys oculifera). temperature and CTM of young snapping turtles, Chelydra serpentina. Aquatic Conserv: Mar Freshw Ecosyst. 2019;1–12. https://doi. – ‐ Journal of Thermal Biology, 14,33 39. https://doi.org/10.1016/0306 org/10.1002/aqc.3106 4565(89)90027‐2