The precise location where a female turtle deposits her eggs has huge consequences for her fitness (e.g., Restrepo et al. 2006; Refsnider and Janzen 2010), especially in the majority of turtle species that exhibit temperature-dependent sex determination (TSD, where sex is determined by nest temperatures; Valenzuela and Janzen 2001). Not surprisingly, nest-site choice has been relatively well studied on regional and global scales (especially in marine turtles; Garmestani et al. 2000, and references therein). However, nest-site choice has been much less well studied on the microenvironmental scale, especially regarding the specific cues that a female turtle uses to choose a precise (rather than general) nest site.

Previous studies found support for the use of various environmental characteristics by female turtles in choosing nest sites, including the amount of solar insolation (Janzen 1994; Roosenburg 1996; Hughes and Brooks 2006; Micheli-Campbell et al. 2013), slope and aspect (Schwarzkopf and Brooks 1987; Garmestani et al. 2000; Wood and Bjorndal 2000; Hughes and Brooks 2006; Ficetola 2007), distance to water (Refsnider et al. 2014, but see Refsnider et al. 2015), ground temperatures (Stoneburner and Richardson 1981; Schwarzkopf and Brooks 1987; Roosenburg 1996, but see Morjan and Valenzuela 2001; Doody et al. 2003b), soil type (Stancyk and Ross 1978; Mortimer 1990; Doody et al. 2003b), soil moisture (Doody et al. 2003b), and vegetation cover (Hays and Speakman 1993; Wilson 1998; Janzen and Morjan 2001; Ficetola 2007). In addition, nest-site choice for predator avoidance (Spencer 2002; Spencer and Thompson 2003) and via social facilitation based on visual cues (Escalona et al. 2009, and references therein) have been supported. Observations of painted turtles (Chrysemys picta) on the Crescent Lake National Wildlife Refuge in the Nebraska Sandhills over the past 35 yrs (Iverson and Smith 1993) suggested that females often chose nest sites immediately adjacent to the earlier nest sites of other females and might be attracted to the odors left by those other nesting females. However, no previous study has attempted to evaluate the role of conspecific olfactory cues in nest-site choice in turtles.

This olfactory hypothesis is based on several lines of evidence. First, females on nesting forays frequently employed “ground-nuzzling” behavior prior to final nest-site selection (see review in Morjan and Valenzuela 2001), whereby females appeared to test the substrate with their snouts (https://www.youtube.com/watch?v=vxsRwoY3Sbc). However, the only published experimental study of nest-site choice (Morjan and Valenzuela 2001) did not test olfaction as a cue. Second, female painted turtles in our population often followed precisely the same path across known nesting areas on their way to an eventual nest site as a female from the previous day or earlier in that evening. Such journeys occasionally involved ground-nuzzling and soil-kicking at the same sites as the earlier female. Third, females often tried vigorously (and unsuccessfully) to excavate nests through wire predator screens placed over previous nests (for another possible example of ground-nuzzling of a previous female's nest site in Elusor macrurus, see also Micheli-Campbell et al. 2013, fig. 3c). Indeed, many females constructed nests immediately adjacent to previous covered nests (Fig. 1), even in the midst of large areas of apparently similar habitat (as evidenced by nest sites in previous years).

View Figure

• Download as Powerpoint
• Download Figure

For example, in a 287-m² area of level, seemingly uniform lawn on which females occasionally nested, only 2 females nested in 2013 (Mark's; see below). The first female nested 6 June in the middle of the area and the nest was covered immediately with hardware cloth. On 26 June, a different female repeatedly attempted to nest on the wire and ultimately nested at the very edge of the screen.

Fourth, each year we placed a small box of mothballs (Napthalene) under our vehicle (to discourage rabbit damage to wiring). Painted turtles on nesting forays occasionally walked under our vehicle, in close proximity to the Napthalene, which seemed to disorient them, making some lethargic enough that we returned them to the wetland for “recovery” (although these females returned to nest normally on subsequent nights). Fifth, recent studies suggested that olfactory proficiency may be underestimated in turtles. Lohmann et al. (2013) provided evidence that odor is important for homing of marine turtles to natal beaches. Southwood et al. (2008) reviewed the importance of water-borne chemical cues in eliciting feeding behavior in marine turtles, and Endres et al. (2009) even showed that airborne food odors can be perceived by Loggerhead turtles (Caretta caretta). Furthermore, Ibáñez et al. (2012, 2013, 2014) demonstrated the importance of olfaction in mate choice in Spanish terrapins (Mauremys leprosa) (see also Auffenberg 1978; Augustine and Mendyk 2012).

When all of these observations are viewed in the context of recent whole genome studies that suggested that turtles have a suite of olfactory genes that rival those of mammals (Wang et al. 2013), the argument for the importance of olfaction in nest-site selection in turtles is reasonable, if not expected. Thus, we sought to test the hypothesis that odors of previously nesting females are an important cue for nest-site choice in C. picta. We predicted that the average distance between nests in primary nesting areas should be significantly shorter than the average distance between random sites in that same nesting area and that females on nesting forays should exhibit increased interest in sites treated with urine from other nesting females.

METHODS

Our painted turtle nesting study site was located immediately to the north of Gimlet Lake, a shallow, sandhill lake on the Crescent Lake National Wildlife Refuge in western Nebraska (see Iverson and Smith 1993, and references therein for additional details). We monitored a total nesting area of about 12,000 m², but this study focused on 5 restricted areas totaling approximately 862 m² (only approximately 7.2% of that area), where nesting was more concentrated (45% to 54% of all nests located therein each year) and surveillance was nearly continuous: Jeep (136 m²; south edge of area 49 m from wetland edge); Mark's (287 m²; 105 m from wetland); Marlin's (88 m²; 37 m from wetland); Parking (226 m²; 97 m from wetland); and Shop (125 m²; 68 m from wetland). These nesting areas lacked vegetation directly overhead, had mostly open soil (except grass covered at Mark's) and received direct sunlight for 50%–90% of the day. The Shop site differed because of the presence of a water faucet in the center of the area, where refuge personnel frequently washed residue (chemicals, mud, vegetation, etc.) from their equipment. This was the only site likely to have had its array of odors compromised frequently. Most of the other nests each year were deposited along 1-dimensional edges (e.g., road margins).

Nesting has never been observed in the morning at this site in 35 yrs of fieldwork. Therefore, we (2–5 persons) monitored the nesting areas from at least 1600 hrs to dark each day while at the study site (27 May to 26 June 2012; 16 May to 2 July 2013; 19 May to 7 July 2014), following females from a distance using binoculars to avoid disturbing them.

Females in our population typically nest at least twice each year, with most producing 3 clutches per year (mean per female 2.78; Iverson and Smith 1993). Females were generally allowed to nest undisturbed, although some (< 10%) that moved beyond our overall monitoring area (and likely to be lost from sight) were returned with minimal disturbance to the lakeside edge of the monitoring area and released to return to the wetland or continue their nesting foray; however, most of the latter nested outside our restricted sites and, hence, did not bias the results presented here. Upon nest completion, females were captured for identification and measurement, and nests were covered immediately (to reduce depredation) with 75 × 75-cm hardware cloth (12.5-mm mesh) secured to the ground with 20 20-cm nails. Because these screens precluded subsequent females from nesting in precise proximity (especially if more than one female has nested in close proximity), our measures of nearest nest distance were likely underestimates of natural nest proximity. Furthermore, we acknowledge that a few nests may have been deposited before our arrival in 2012, and some may have been laid after we left the site each year in late June (2012) or early July (2013, 2014); however, we have no reason to suspect that the locations of a few potential missed nests would have biased our results. Similarly, females sometimes abandoned partially constructed nests if disturbed by vehicles or if immoveable rocks were encountered (e.g., Fig. 1). Hence, although every first choice of a nesting site (i.e., if abandoned) was not mapped for every nesting female in a given year, our sample sizes are adequate for hypothesis testing and, if anything, our nearest neighbor data sets are likely to underestimate preferred nest proximity.

Individual females only occasionally placed a subsequent nest in the same restricted nesting area as their previous nest, and nest-site philopatry was not evident (contrary to Lindeman 1992). For example, mean distance between pairs of nests from individual females in 2012 averaged 71.1 ± 51.1 m SD (range, 1–193; n = 21); in 2013, 85.2 ± 82.9 m SD (1–387; n = 43); and in 2014, 49.4 ± 27.9 m SD (1–120; n = 26) and did not differ significantly by year (1-way ANOVA F_2,87 = 2.51, p = 0.09; 1-way ANOVA by rank, Kruskal-Wallis H = 3.81, p = 0.15; STATVIEW^TM software; see also Janzen and Morjan 2001). Hence, our data are not biased by precise nest-site philopatry of individual females. Furthermore, these large internest distances complicated our ability to detect individual female preferences for nesting in proximity (especially when they nested outside of our five restricted nest areas). However, as a preliminary test of individual preferences, we scored females that nested within 1 m of a previous nest (“close”) and those that nested at least 2 m (“far”) from a previous nest (in the same year) and compared their tendencies to nest repeatedly in close proximity to other nests. We also compared carapace lengths of close versus far nesters to test the hypothesis that smaller (typically younger) females might tend to place nests near other nests. Each carapace length sample except for far nesters in 2012 was normally distributed based on a Shapiro-Wilk normality test at p < 0.01. Samples with normal distributions were compared with 1-way ANOVA, whereas the 2012 far sample was compared with a Mann-Whitney U-test (STATVIEW).

During each nesting season (2012–2014), each restricted nesting area and each nest placed therein was mapped to the nearest 10 cm. These maps were scanned into ArcMap 10.1, georeferenced, and the nearest neighbor function in ArcTools was used to test for clumping of nests. This tool compares the distribution of observed nearest neighbor distances in an area of defined size and shape with that of a random distribution for that same area (see Ebdon 1985).

During the 2015 nesting season, we initiated a preliminary field test of our olfactory hypothesis. On 3 June, we established 3 experimental sites in our 3 main nesting areas (Jeep, Parking, and Shop), near the center of each area, where physical conditions (slope, soil, insolation, etc.) appeared to be uniform. Two microsites (3 m apart) were chosen in each of the 3 areas and covered with hardware cloth as described above. The microsites within the nesting sites were equidistant from the wetland. One of the two microsites was randomly chosen to receive treatments of urine from female turtles moving to nest off-site. The second microsite was treated with an equal volume of water on the same schedule. Each site was re-treated twice at intervals of 3–5 d. Our intent was to score the behavior of each female that crossed a microsite screen as 1) no interest (crossed the screen without stopping); 2) some interest (stopped briefly on the screen); or 3) high interest (ground-nuzzled the screen and/or moved back and forth across the screen while scratching with its claws).

RESULTS

Evidence for clustering of nests was mixed across years and restricted sites (Table 1; Fig. 2). We found no evidence of clustering in any year at either the Shop (chemically challenged) or the Parking sites. However, at 2 sites that were not used annually for nesting (Mark's and Marlin's), clustering was strongly evident. Finally, at our most-used nesting area (Jeep), nests were significantly clustered in 1 of 3 yrs, as well as during a second year if one extreme outlying nest (by a 170-mm carapace length female) was not included in the calculations (Table 1).

Table 1. Nearest neighbor distance analysis of painted turtle nests over 3 yrs in 5 nesting areas north of Gimlet Lake, Crescent Lake National Wildlife Refuge, Nebraska. Obs. = observed; Exp. = expected. Significant negative Z-scores indicate clustering of nests; positive scores indicate overdispersion of nests. No turtles nested at Marlin's site in 2014 or at Mark's in 2013 or 2014. Because of one extreme outlier nest in 2013 in the Jeep area, the data were also analyzed without that nest. Significant p-values are in bold.

View Fullscreen

View Figure

• Download as Powerpoint
• Download Figure

Females that were scored as close nesters (< 1 m) tended to have a smaller carapace than females scored as far nesters (> 2 m) (Table 2). However, these data are potentially biased because some females produced a close and a far nest in a year and were included in both categories.

Table 2. Comparison of carapace length (CL in mm ± 1 SD) for female painted turtles considered “close” nesters (< 1 m) versus those considered “far” nesters (< 2 m; by ANOVA or Mann-Whitney U-test*) at Gimlet Lake, Nebraska.

View Fullscreen

Furthermore, of 11 individual females that were considered close nesters and placed 2 nests in one or more of our restricted nesting areas within a year, only 3 placed their second nest < 1 m from another nest. The remaining 8 placed 1 nest close (< 1 m) and 1 nest far (> 2 m) from another nest.

Despite the location of our experimental sites in the center of our 3 main nesting areas, only 3 of 81 females observed on nesting forays crossed the microsites during the nesting season. One crossed the water-moistened site without stopping, and two others crossed urine-treated sites, one showing medium interest and one showing high interest. Although the results match our predictions, the sample size is too small to permit evaluation with statistical confidence.

DISCUSSION

Placement of nests in close proximity (i.e., ≪ 2 m) has only rarely been reported for freshwater turtles (> 1 m, Burger 1977; > 1 m, Robinson and Bider 1988; > 1.3 m, Lindeman 1992), and reported internest distances for painted turtles are typically much greater (e.g., Schwartzkopf and Brooks 1987; Rowe et al. 2005). On the contrary, painted turtles at our study site often nested in precise nonrandom proximity (Table 1; Fig. 2), despite the availability of over 12,000 m² of previously used nesting habitat (as determined by the combined nest locations of over 30 yrs of observation). Furthermore, despite the very small sample size, our experimental study provided preliminary support for odor as a cue in nest-site choice. However, although our data provide some support for very precise, proximate placement of nests in our population (supporting the use of olfactory cues), they are insufficient to reject the possibility of the simultaneous use of other subtle very fine-scale environmental cues (temperature, soil moisture, etc.) by females as the basis for proximate nest placement. Nevertheless, given the clustering of nests that we observed within small, seemingly uniform nesting areas, it seems unlikely that variation in those other factors on a scale of less than a meter could explain the pattern we observed (e.g., see Fig. 1). Hence, the detection of odors from previous nests by subsequent nesting females is a reasonable hypothesis explaining our observations, but it is certainly not the only cue.

It is clear that female painted turtles use multiple cues for nest placement, but we contend that odors from earlier nests are more important than previously appreciated. However, the question remains as to why a female would choose to nest adjacent to another female's nest? Obviously, selection should favor a female's ability to detect cues for nest sites that optimize her and her offspring's fitness (Resetarits 1996), but the advantages of proximate nesting are not yet clear. It is tempting to invoke the advantages of “group” behavior (such as predator confusion or satiation, or safety in numbers; Doody et al. 2009). For example, as noted in the introduction, in a 287-m² area of level, seemingly uniform lawn on which females occasionally nested (Mark's), only 2 females nested in 2013. The first female nested 6 June in the middle of the area, and the nest was covered immediately with hardware cloth. On 26 June, a different female repeatedly attempted to nest on the wire and ultimately nested at the very edge of the screen (Fig. 2, panel K). The second nest was depredated from under its protective screen by a badger the same night it was constructed, but the first nest was missed and later hatched successfully. However, most of the published data for turtles argue that clumping of nests results in increased risk of depredation (Burger 1977; Robinson and Bider 1998; Valenzuela and Janzen 2001; Marchand et al. 2002; Spencer 2002; Doody et al. 2003a; Marchand and Litvaitis 2004; but see Fowler 1979; Eckrich and Owens 1995; Burke et al. 1998).

Several other possible advantages remain to be tested. First, because turtle nest depredation is often most intense during the first few days after nest construction (Christens and Bider 1987; Congdon et al. 1987; Spencer 2002; but see Snow 1982), females may gain an advantage by associating their nest with a previous nest that has escaped detection by predators. However, we often observed early nests that had escaped depredation for weeks that were destroyed soon after a second turtle nested in close proximity. Unfortunately, because we disturbed the nest sites (by covering them with wire and partially excavating them to place data loggers), we could not determine natural nest success in clustered versus dispersed nests.

Second, some evidence suggests that larger/older females may tend to deposit their first nest earlier in the season than smaller/younger females (Bowden et al. 2004; Rollinson and Brooks 2008). Our data for first clutches in 2013 clearly showed this pattern (n = 62; r = −0.46; p = 0.0002), although it was not evident in 2014 (n = 41; r = 0.02; p = 0.90 [the relationship could not be calculated for 2012 because we missed the first few days of nesting]). Hence, smaller females, presumably younger and less experienced nesters, may gain an advantage by detecting and learning where larger, presumably older, females nest. Again, because we disturbed nests during our work, our data would not permit us to study natural nest success as a function of female body size or age.

Third, selecting a previously used nest site might reduce the time spent searching for and selecting a nest site, thus reducing exposure times to terrestrial predators (e.g., Spencer 2002). However, our unpublished data on nest-site search times actually showed that larger females had shorter search times in 1 yr and no difference in times relative to smaller females in a second year (A. Brouillette, K. Hardy, A.R. Hedrick, and J.B. Iverson, unpubl. data, 2013).

Fourth, the use of olfactory cues could help explain how at least painted turtles are able to exploit new nesting areas or even new wetland habitats (i.e., by following other females). We noted that nest clustering in our study seemed to be most pronounced in nesting areas that were not used annually, but when used, they were repeatedly used. This pattern suggests that once a “new” or sporadically used area is chosen by a female, the area may become more attractive to subsequent females, and the latter females may then be more likely to nest in close proximity.

Finally, we have not studied relatedness among our females and therefore cannot address the possibility that females that nest in proximity might be related and, hence, gain an advantage by that behavior. However, our data also suggest that, although females tend to nest in the same 12,000-m² area year after year, they typically do not use the same local nesting area more than once in a nesting season, and placement of subsequent intraseasonal nests in precise proximity is very rare (see above, and Janzen and Morjan 2001).

Our observations suggest that odors from previous nests might remain detectable in the environment by other females for at least 3 wks (see Mark's yard, above). Whether odors might persist from year to year could not be directly addressed by our study. However, the occasional abandonment of well-used nesting areas in a given year argues against that persistence (e.g., Marlin's in 2014). Indeed, even the concentration of nests within our focal areas often changed from one year to the next (e.g., Fig. 2).

Our study also suggests that future work needs to examine precise nest-site placement relative to other nests in other populations and species, particularly at sites offering uniform environmental conditions (e.g., in slope, substrate, shade), to reduce the multitude of cues that females might use and further test the importance of olfactory cues in nest-site choice. In addition, expanded experimental studies of the effects of the application of bladder water from gravid female painted turtles on nesting patterns over uniform nesting areas could be another useful test of the importance of olfaction in nest-site choice, especially if the liquid is applied in the form of a trail to the more heavily scented microsites.

In summary, the range of sensory modalities among turtles is only now beginning to be appreciated (e.g., Ferrara et al. 2014), and the evidence for the proficiency of olfaction in turtles is mounting. In addition to its importance in feeding (Southwood et al. 2008; Endres et al. 2009), mate choice (Ibañez et al. 2012, 2013, 2014), and possibly homing to natal beaches (Lohmann et al. 2013), our nesting observations and study of nest-site clustering provide preliminary evidence that females may also use odors associated with previously nesting females to select their own nest sites in at least C. picta.