Methods

This research represents a portion of a long-term study of painted turtle nesting ecology (Janzen 1994) at the Thomson Causeway Recreation Area (TCRA) near Thomson, Illinois. The Thomson Causeway is an ∼450- × 900-m island on the eastern bank of the Mississippi River, and it contains an ∼1.5-ha nesting area that is bordered on the east side by a 200-m-wide slough from which most turtles emerge to nest. The site is managed and maintained by the United States Army Corp of Engineers (USACE; see Kolbe and Janzen 2002a for a more complete site description). The TCRA is a popular destination for recreationists with motor homes (recreational vehicles [RV]) during the spring and summer months. Use of the area is variable, with most activity occurring on weekends and holidays (Bowen and Janzen, pers. obs.). The nesting area is interspersed with concrete pads for parking RVs and has a paved road running through it. This setting provides an opportunity to investigate the nesting responses of painted turtles to different levels of human recreation throughout the nesting season.

Chrysemys picta is a small- to medium-sized freshwater turtle that ranges from southern Canada to New Mexico and from the Atlantic to the Pacific Oceans (Ernst et al. 1994; Starkey et al. 2003). At TCRA, the mean clutch size is 10.5 ± 2.0 standard deviation (SD) eggs, and individual females may lay up to 3 clutches in a nesting season from late May to early July (Morjan 2003). Chrysemys picta has temperature-dependent sex determination, with cooler temperatures producing males (Ewert and Nelson 1991). Hatchlings remain in the nest during the winter and emerge from the nest and enter water in the spring (Ernst et al. 1994; Weisrock and Janzen 1999).

The nesting beach at TCRA was monitored for nesting turtles from early to late June in 2001 and from late May to early July in 2002 and 2003. The turtles were individually marked by using a series of notches in the marginal scutes of the carapace (Cagle 1939). Each year we observed females leave the water, construct nests, and lay eggs. Once a female had finished nesting, we temporarily mapped the location of the nest by using nearby landmarks (e.g., trees, posts, and RV sites), and determined the amount of overstory vegetation (% shaded) in all 4 cardinal directions over the nest by using a spherical densiometer (see Janzen 1994; Weisrock and Janzen 1999). Once the nesting season was complete, we returned to the location and established precise Cartesian coordinates for each nest by using the program INTERPNT (Boose et al. 1998). Geographic information system coverages were created by using these coordinates in ArcView® (ESRI Inc., Redlands, CA). We used these location data to determine the distance of each nest from the water. Error in this mapping and measuring process ranges from 0 to 15 cm (see Kolbe and Janzen 2002a for more detail). We also followed the fate of each nest through the third week of September (i.e., until all hatching was completed) by noting which nests had been depredated and which remained intact.

As an indicator of the level of human disturbance at TCRA during the nesting season, we used data on the number of RVs present daily at the nesting beach and nearby sites for the months of May through July in the years 2001–2003 obtained from the USACE. We used nested analysis of variance (ANOVA) to determine the effect that varying numbers of RVs had on the decision of turtles to nest. We used the number of nests constructed on a given day as an indicator of the decision of turtles to emerge from the water and nest. Effects in the nested ANOVA were the number of RVs, the Julian date, and the interaction between these two (all nested within year). The Julian date was included in an attempt to control for seasonal changes in nesting behavior that might be independent of human activity.

We hypothesized that the number of turtles emerging to nest would decrease as the number of RVs increased. This relationship was considered important because mass nesting events in response to decreased human presence might increase nest density on small scales and serve as cues for nest predators. Higher nest densities are known to result in higher probabilities of nest depredation at TCRA (Valenzuela and Janzen 2001).

We used nested ANOVAs to examine the effect of human recreation on habitat selection of nesting turtles by comparing the number of RVs present on a given day to the distance from water of nests laid on that day and by comparing the number of RVs present on a given day to the south and west overstory vegetation cover of nests laid on that day. Effects for each nested ANOVA were the number of RVs, Julian date, maximum air temperature on the day of nesting, water temperature on the day of nesting, and the interactions between these variables (all nested within year).

Weather variables were included in an attempt to account for the effects of weather on habitat selection. Maximum air temperature and water temperature were chosen because these variables appear to affect the decision of turtles to emerge from the water and nest at TCRA (Bowen et al. 2005). We obtained weather data for the nesting seasons from the USACE Lock and Dam 13, ca. 12 km south of the study site.

We hypothesized that as the level of human activity increased females would perceive a greater risk to themselves, and the distance of nests from the water would decrease (Spencer 2002). This relationship was considered important because rates of nest depredation are higher along the water edge in most years at TCRA (Kolbe and Janzen 2002a). Sites near the water are less shaded at TCRA; therefore, we hypothesized that nest overstory vegetation cover would decrease with increasing levels of human recreation (Kolbe and Janzen 2002a). This relationship was considered important because south and west overstory vegetation cover is a predictor of nest temperature and offspring sex ratio at this site in most years (Janzen 1994; Morjan and Janzen 2003). Nesting turtles are known to choose nesting sites nonrandomly with respect to overstory vegetation at TCRA (Janzen and Morjan 2001).

Although some females at TCRA lay multiple clutches within a nesting season (Morjan 2003), we used only the first nest for each female within years in our analyses. We did so to ensure that each nesting event was independent within years (i.e., females may exhibit different nesting behavior during their second attempt, based on what they experience during their first attempt). Adding data from second and third nests did not change our findings (results not shown). All statistical analyses were performed by using JMP (SAS Institute Inc).

Results

Nesting began by 7 June in each year and the nesting beach was monitored until at least 1 July. Either 23 (2001 and 2003) or 35 (2002) nesting days for each year were included in the analyses. The number of nesting days varied, in part, as a result of weather conditions that inhibited nesting on some days. Data were analyzed from 147 nests in 2001, 158 nests in 2002, and 218 nests in 2003. Nest parameters were highly variable during the nesting season in all years studied, as was the number of RVs present at TCRA (ranging from 0 to 64 RVs per day; Table 1). Overall, substantial within- and among-year variation existed in the data set for the response and predictor variables of interest in this study.

Table 1. Descriptive statistics (mean ± 1 SD) and sample sizes for human recreational activity and nest variables of the painted turtle (Chrysemys picta) at a nesting beach in northwestern Illinois.

View Fullscreen

The number of nests laid on a given day was not affected by any of the predictor variables (overall p = 0.7993, r² = 0.09; Table 2). A similar pattern was observed for overstory vegetation (overall p = 0.4360, r² = 0.06; Table 3). The nested ANOVA for the distance of the nests from water was statistically significant (overall p = 0.0009), but none of the individual effects approached statistical significance and the r² value was small (r² = 0.12; Table 4). The biological significance, therefore, is likely to be negligible.

Table 2. Results for individual effects from a nested analysis of variance to determine the effect of the number of recreational vehicles (RV) on the number of painted turtle (Chrysemys picta) nests constructed on a nesting beach in northwestern Illinois during the years 2001–2003.^a An asterisk (*) signifies an interaction between terms.

View Fullscreen

Table 3. Results for individual effects from a nested analysis of variance to determine the effect of the number of recreational vehicles (RV) on the south and west overstory vegetation of painted turtle (Chrysemys picta) nests constructed on a nesting beach in northwestern Illinois during the years 2001–2003.^a

View Fullscreen

Table 4. Results for individual effects from a nested analysis of variance to determine the effect of the number of recreational vehicles (RV) on the distance from water of painted turtle (Chrysemys picta) nests constructed on a nesting beach in northwestern Illinois during the years 2001–2003.^a

View Fullscreen

Discussion

Human recreational activity, as measured by the number of RVs on and near a major nesting beach, did not appear to affect large-scale patterns in the nesting ecology of the population of painted turtles studied here. Variables that represented both the decision to emerge from the water to nest and habitat selection had no biologically significant relationship with the number of RVs present.

Based on the results of previous studies, the lack of effect found here was unexpected. For example, breeding and nest survival of colonial waterbirds may be negatively affected by human recreation (Yorio et al. 2001; reviewed in Carney and Sydeman 1999). Garber and Burger (1995) documented that 2 populations of the wood turtle (Glyptemys insculpta) declined 100% within 10 years after the opening of habitat to human recreation (foot traffic leading to opportunistic removal of turtles). Bury and Luckenbach (2002) found that a population of desert tortoises (Gopherus agassizii) that was subjected to human recreation in the form of off-road vehicles appeared to be less dense and less healthy than a population that was protected. Given that human recreation and disturbance can have adverse effects on breeding organisms in general and on turtle populations in particular, combined with the likelihood for some turtles to alter nesting behavior when disturbed or when they perceive danger (Iverson and Smith 1993; Spencer 2002; Spencer and Thompson 2003), one might assume that nesting turtles would suffer from human activity. It is important to note, however, that we studied behavioral responses and not population dynamics.

Our results give tentative support to the assertion of a number of investigators (Whittaker and Knight 1998; Miller and Hobbs 2000; Nisbet 2000) that wildlife responses to human recreation are difficult to generalize. Wildlife may respond to human recreation in many ways, thus studies of these effects should be done on a species-specific level (Miller and Hobbs 2000). A population-specific level may be necessary if some species are capable of habituating to disturbance by humans (i.e., they no longer respond to human disturbance; Whittaker and Knight 1998). In the future, we plan to test the hypothesis that our turtles have habituated to human presence by using comparative experimental studies with nearby populations.

There is an important caveat to consider in interpreting our results: the analytical methods used are capable of detecting only large-scale changes in turtle nesting behavior as a result of human recreation. We did not evaluate the responses of individual turtles per se, nor did we directly track population dynamics. We have observed turtles abandon nesting attempts as the result of direct human intrusion, adults and hatchlings killed by automobiles, and removal of turtles from the study area by recreationists. These types of situations are not accounted for in our analysis. Examining the response of an individual turtle under different conditions might be more instructive in determining the effects of human recreation. However, this experimental approach would be difficult to implement given that we cannot control when a turtle emerges to nest nor can we directly manipulate the intensity of human recreation.

What do our results mean for managers? Human recreation at TCRA does not appear to have effects on large-scale patterns of painted turtle nesting behavior, and the needs of these turtles and human recreationists may be reconcilable. However, we emphasize that these results should not be taken to suggest that human recreation does not affect freshwater turtles. Generalizations to other species and other forms of recreation should be avoided. Furthermore, even if painted turtles are unaffected by large-scale human activity, the actions of individual humans (removing or disturbing nesting turtles, road kills) should still be taken into account. Education of the public (Klein 1993; Taylor and Knight 2003) concerning the plight and sensitivity of turtles is a good first step. At TCRA, education over the past 15 years by both our research team and USACE park rangers has minimized, but not eliminated, individual human disturbance of painted turtles. Monitoring and enforcement of applicable laws will still be necessary in most cases.

We generally agree with other investigators (Boyle and Samson 1985; Nisbet 2000) that the large number of studies that suggest a negative relationship between human recreation and wildlife should not be applied to all species and all situations. Furthermore, studies that test explicit hypotheses and attempt to determine the fitness effects of recreation on wildlife (Boyle and Samson 1985) should be designed where feasible. Finally, although research on “secure” populations of freshwater turtles is important (Congdon et al. 2003), ecological and evolutionary studies on human-influenced populations (Kolbe and Janzen 2002b; Feinberg and Burke 2003) are equally crucial given the current rate of habitat alteration and the conservation status of most turtle species. Future studies should focus on the area around individual nesting turtles that must be kept inviolate, if they are to remain undisturbed (area of influence; Miller et al. 2001), and on the effects of recreation-related deaths and adult removals on population dynamics (Garber and Burger 1995).