Turtles are oviparous and females construct nests in particular habitats that provide the biophysical conditions necessary for successful egg incubation. Because nest locations and microclimates affect offspring survival (Wilson 1998; Spencer 2002), sex (Janzen 1994; Roosenburg 1996), size (Brooks et al. 1991; Packard 1999), and performance (Finkler 1999; Kolbe and Janzen 2002), the choices that females make during nest site selection have profound effects on fitness. Successful nests are those that remain undetected and unmolested by predators; do not suffer flooding, erosion, or root encapsulation; and provide appropriate thermal regimes and hydration levels for developing embryos (Wilbur and Morin 1988; Congdon et al. 2000).

As a consequence of particular habitat requirements, female turtles are discerning in their search for nest sites. Habitat characteristics that females appear to evaluate when selecting nest sites include levels of canopy and ground cover (Jackson and Walker 1997; Janzen and Morjan 2001; Hughes et al. 2017), slope and aspect (Burger and Montevecchi 1975; Hughes and Brooks 2006; Micheli-Campbell et al. 2013), height above floodplains (Plummer 1976; Doody et al. 2004; Pignati et al. 2013), and substrate composition (Rasmussen and Litzgus 2010). Habitats that provide suitable nesting conditions are often limited across a variety of landscapes (Steen et al. 2012). For many adult female turtles, long-distance nesting migrations coupled with substantial risks along migratory pathways make accessibility to nesting habitats difficult, costly, and dangerous (Gibbs and Shriver 2002; Moll and Moll 2004Steen et al. 2006).

As long-lived, iteroparous vertebrates, turtles have reproductive lifetimes that span decades (Gibbons 1987; Congdon et al. 2001, 2003), providing opportunities to utilize the same nesting habitats repeatedly. Selection favors fidelity to nest sites that are predictable and benefit the female directly via her own survival (Spencer 2002) and through the survival and quality of her offspring (Lindeman 1992). Nest site fidelity is advantageous when nesting areas are spatially limited or difficult to find, or when substantial seasonal migrations are necessary to reach them (Obbard and Brooks 1980; Rowe et al. 2005).

Northern map turtles (Graptemys geographica) are a highly aquatic riverine species of the east-central United States and Southern Canada (Lindeman 2013). In central Pennsylvania, G. geographica inhabit the Susquehanna River and its major tributaries, including the Juniata River (Nagle 2010). This population east of the Allegheny Mountains is geographically isolated from the main region of the species distribution in the Midwestern United States (Lindeman 2013). Graptemys geographica is a species of conservation concern in Pennsylvania and an endangered species downstream in Maryland (Nagle 2010; Richards-Dimitrie et al. 2013).

We examined nest site fidelity of G. geographica at Mount Union, the largest reported riverine turtle nesting area in Pennsylvania (Nagle and Congdon 2016). During the 20th century, a large coal tailings pile along the Juniata River provided turtle nesting habitat at Mount Union. Following removal of some of the coal in 1998 to construct a new highway, subsequent mitigation activities helped protect adult female turtles from road mortality and replace lost nesting habitat. Because females of other Graptemys species reportedly show fidelity to general nesting areas (Vogt and Bull 1982), we examined the presence and magnitude of nest site fidelity in G. geographica at Mount Union and implications for species' risk assessment and habitat management. Ongoing conservation efforts in central Pennsylvania will benefit from a better understanding of nest site fidelity in northern map turtles.

METHODS

Study Site. — During 1998, the Pennsylvania Department of Transportation (Penndot) removed approximately 290,000 m³ of coal tailings from the eastern edge of Mount Union along the Juniata River to build a new highway. Following the opening of the highway in the spring of 1999 and substantial road mortality of turtles in June of that year, Penndot erected an 1150-m-long chain-link fence between the river and the new highway to prevent turtles from moving onto the road (Fig. 1). Penndot also deposited ∼ 1000 m³ of sand and shale in several long mounds along the river side of the turtle fence to mitigate the loss of nesting habitat.

View Figure

• Download as Powerpoint
• Download Figure

Forty meters south of the mitigation area, a remnant coal tailings pile ∼ 500 m long by 40–70 m wide provides additional turtle nesting habitat (Fig. 1). The coal had been mined 30 km south on Broad Top Mountain and the tailings were refuse, discarded in the early 20th century by a coal-cleaning plant at the junction of the East Broad Top Railroad and mainline of the Pennsylvania Railroad. The tailings are attractive to turtles as a nesting substrate because they are uniformly black, sparsely vegetated, and easy to excavate (Nagle and Congdon 2016). Turtle nests placed in coal tailings can experience extremely high incubation temperatures and potential chemical insults from exposure to low pH and elevated levels of heavy metals (Nagle et al. 2018).

In the mitigation area, a unique aspect of our study was the very linear nesting area within the turtle exclusion fence. Although the length of this mitigation area spanned 1150 m along the turtle fence, the width of available nesting habitat was only about 3 m, and most turtles nested < 1 m from the fence (Nagle and Congdon 2016). The support posts of the chain-link turtle fence were uniformly spaced (3.05 m apart), and we created a grid system by consecutively numbering the fence posts with waterproof decals. The linear nesting area along the turtle fence and location posts allowed us to accurately calculate distances between nests of individual females within the mitigation habitat.

We analyzed data only for completed nests that were verified to contain eggs or hatchlings, not capture locations or partially excavated nest cavities. Thus, unlike many studies that defined nest site fidelity as the tendency of females to return to general nesting areas, we define nest site fidelity as the tendency of females to return to the same specific location and construct nests.

Field Methods. — We monitored the mitigation area and adjacent pile of coal tailings each year from 2000 to 2003, 2006 to 2007, and again from 2017 to 2019. We also monitored the mitigation area during 2005 and conducted limited sampling of both areas in 2008, 2010, and 2016. Because most northern map turtles nest during the morning hours (Nagle and Congdon 2016), researchers typically patrolled the site from 0800 to 1200 hrs. Turtles were captured during terrestrial movements to or within nesting areas, or immediately after they had completed nesting.

We measured straight-line carapace lengths using modified tree calipers and gave individual turtles unique identification codes by notching or drilling marginal scutes (Nagle et al. 2017). Gravid females were additionally marked with small paint numbers on the second vertebral scute, so that during subsequent observations, they could be identified from a distance without disturbance. Observed nests were protected from predators using cylindrical 2.5 × 5-cm wire-mesh cages (Nagle et al. 2004). Nest locations along the turtle fence were recorded to the nearest 0.5 fence posts (an accuracy of approximately ± 1.5 m). From 2016 to 2019, locations of turtles were recorded using a Garmin GPSmap 76CSx handheld meter (Garmin Ltd., Schaffhausen, Switzerland; also an accuracy of approximately ± 1.5 m or less).

We also used radiotelemetry to determine the distances that turtles moved in the river following nesting and prior to subsequent nesting events. We attached radio transmitters (AI-2, Holohil Ltd., Carp, Ontario, Canada) to 5 female G. geographica after they had nested: 3 in late June 2000 and 2 in late June 2001. Transmitters weighed 20 g (<2% of female body mass) and were attached with copper wire to 2 small holes drilled in posterior marginal scutes. Turtles were tracked approximately weekly for 1 mo following nesting, approximately bimonthly through winter and early spring, and then at least weekly during May and June. Locations in the river were determined using triangulation and recorded in detailed notes based on mile markers along roads, major landmarks, or unique topographic features such as islands in the river or stream inlets. We used the measuring tool in Google Earth^™ to calculate the distances moved by turtles.

Data Analysis. — Internest distances for individual females were calculated within season (first and second clutches) and among years. One such distance between a pair of within-season nests was measured directly, as the nests were adjacent to each other. We also used the measuring function in Google Maps^™ to calculate the distance between 2 recent nests for which we obtained global positioning system coordinates.

We used nonparametric Mann-Whitney U-tests to compare mean within-season and among-year internest distances to each other and to distances between randomly selected pairs of G. geographica nests from different individuals. The Anderson-Darling normality test was used to assess whether among-year internest distances were positively skewed. For females with at least 3 internest distances among years, we compared mean values among individuals using 1-way analysis of variance. Spearman's rank correlations were used to examine the relationships between internest distances and 1) the number of years between nesting events, as well as 2) female carapace lengths. All statistical analyses were completed using Minitab 17.0 (Minitab, Inc., State College, PA).

RESULTS

Over 20 yrs from 2000 to 2019, we captured 612 individual reproductive female G. geographica in the nesting habitats at Mount Union and made more than 1800 recaptures of those females. A total of 1553 captures of turtles were made along the turtle exclusion fence (Fig. 2), and 875 captures were made on the coal tailings pile. We recorded grid locations for 128 nests along the turtle fence and calculated internest distances for 52 pairs of nests.

View Figure

• Download as Powerpoint
• Download Figure

Internest distances of individual female G. geographica along the turtle fence varied substantially within and among years (Table 1). Within-season (i.e., first and second clutch) internest distances were significantly smaller than those between random pairs (U = 33.5; p = 0.003; z-score = 2.98). Among the 6 pairs of within-season first and second clutches, 5 were deposited ≤ 31 m from each other, but the sixth pair was spaced much farther apart (468 m). One female deposited her first and second clutches adjacent to one another (0.30 m apart); this distance was measured directly rather than calculated, as the female nested with her second clutch against the wire cage that protected her first clutch.

Table 1. Distances between paired nests of individual female northern map turtles (Graptemys geographica) within season (first and second clutches) and among years, and distances between randomly selected pairs of nests from different individuals (m).

View Fullscreen

We calculated internest distances among years for 46 pairs of nests from 24 individual females over deposition intervals ranging from 1 to 7 years. The frequency distribution of internest distances among years was positively, highly skewed, with most internest differences being <100 m (skewness = 1.68; A² = 3.61; p < 0.005; Fig. 3a). Thus, the majority of among-year internest distances were smaller than the mean value.

View Figure

• Download as Powerpoint
• Download Figure

Internest distances of among-year pairs were positively correlated with the number of years between nests (i.e., longer time periods were associated with greater distances, Spearman's r_s = 0.499; p < 0.001; n = 46; Fig. 3b). Internest distances among years were significantly smaller than those of randomly paired nests (U = 475; p < 0.00001; z-score = –4.55) and significantly larger than within-season (first and second clutch) paired nests (U = 50; p = 0.01; z-score = 2.01). For 7 females with at least 3 nests located along the grid in different years, mean internest distance (± SD) ranged from 42.7 ± 27.1 m SD to 358.1 ± 249.3 m SD, and varied significantly among females (F_6,20 = 2.81; p = 0.037). Mean internest distance among years for each female was not correlated with carapace length of females (Spearman's r_s = 0.128; p = 0.55; n = 24).

Of the 5 female turtles outfitted with radio transmitters, 2 remained in the river near their nesting areas during winter and the following spring (one near the mitigation area and the other near the coal tailings pile). The 3 remaining females moved downstream 4.3, 5.0, and 5.6 km within 1 mo following nesting, and remained near those respective locations throughout winter and early spring. By the following June, all 5 turtles returned to the same nesting areas at Mount Union where they had previously nested (all within < 150 m).

Females that nested on the coal tailings pile appeared to exhibit high fidelity to that area (Table 2). Fifty-seven females were first captured in the mitigation area and last captured on the coal tailings pile. Only 5 females exhibited the reverse trend (coal pile to mitigation area). Four females that nested in the mitigation area subsequently nested on the coal tailings pile in later years, yet no females were observed to nest on the coal pile and then nest later in the mitigation area.

Table 2. Numbers of reproductive female northern map turtles (Graptemys geographica) at Mount Union exhibiting different patterns of nesting habitat use.

View Fullscreen

DISCUSSION

Our results suggest that female G. geographica at Mount Union exhibit nest site fidelity, as the frequency distribution of distances between nests of individual females was positively skewed, and distances between nests (both within and among seasons) were smaller than distances between randomly selected pairs of nests among different females (Schwarzkopf and Brooks 1987; Valenzuela and Janzen 2001; Rowe et al. 2005). Distances between nests of individual females also increased over time, likely due in part to vegetation succession and shading, which rendered portions of the mitigation habitat at Mount Union unsuitable for turtle nesting.

Historically, the type of nesting habitat required for map turtles—open canopy, well-drained substrates with sparse vegetation near major rivers—was probably rare in central Pennsylvania and in other heavily forested regions. Thus, in such areas, selection has likely favored female map turtles that tend to return to suitable, spatially limited, difficult to find nesting habitats. In addition, our results using radiotelemetry show that some riverine G. geographica make extensive seasonal migrations for nesting, another factor that indicates fidelity to particular nesting sites.

Studies of other Graptemys species have shown fidelity of females to general nesting areas. Among several insular nesting areas in the Mississippi River in Wisconsin, female Graptemys ouachitensis and Graptemys pseudo-geographica showed a tendency to return to the same beach area to nest (Vogt and Bull 1982). Graptemys kohnii from Reelfoot Lake in Tennessee exhibited strong fidelity to nesting areas, despite an experimental design in which some females were transplanted several kilometers away from their nesting locations (Freedberg et al. 2005).

Although G. geographica rarely nest far from water (mean distance from water in the mitigation area = 29.5 m; Nagle and Congdon 2016), the largest part of nesting migrations for some turtles occurs within the river course itself. Compared with turtles in lentic wetlands, major variation in water levels and flow rates of lotic habitats might influence nesting migrations and nest site fidelity of river turtles (Tucker 2001). Yet internest distances of G. geographica in our study and of riverine Trachemys scripta in Illinois (Tucker 2001) both fell within the range of values reported for freshwater turtle species that inhabit lentic wetlands.

Several studies have documented nest site fidelity in female Chrysemys picta. Female C. picta in northern Idaho placed within-season clutches in very close proximity to one another (1–10 m), and some females exhibited nest site fidelity over multiple years (Lindeman 1992). Among C. picta on Beaver Island, Michigan, some females showed strong fidelity to particular nesting sites among years, constructing nests within a few meters of previous locations, yet other females nested in very different areas among years (Rowe et al. 2005). Female C. picta in Illinois exhibited fidelity to both microgeographic sites and specific types of vegetative cover (Valenzuela and Janzen 2001).

Some individual female G. geographica at Mount Union exhibited strong nest site fidelity, placing their first and second clutches within a single nesting season or nests over multiple years within close proximity to one another. Other individuals regularly placed their nests farther apart, with some nests separated by hundreds of meters. During a 16-yr nesting study of Emys orbicularis in Poland, Mitrus (2006) reported that a few individual females exhibited strong nest site fidelity over a decade, other individuals for a few years, and yet other turtles did not exhibit the behavior at all. Congdon et al. (1983, 2011) reported that the majority of female Emydoidea blandingii on the E.S. George Reserve in southeastern Michigan showed fidelity to a general nesting area, yet other females regularly placed their nests at sites separated by great distances (> 1000 m).

What factors might drive females to place different nests at sites separated by considerable distances? First, distributing nests broadly provides increased offspring dispersal. For females that make extensive movements among different habitats, the benefits of offspring dispersal must outweigh the risks of increased exposure to predators as well as the increased risks to hatchlings of extensive overland movements from nest to water (Starrfelt and Kokko 2010; Delaney and Janzen 2018). Second, in some aquatic turtles that occur at low densities and have long lifespans and long generation times, substantial nesting migrations might also reduce the likelihood that offspring will share the same habitat as their parents, and ultimately prevent inbreeding (Congdon et al. 2011).

The existence of some itinerant reproductive females likely contributes to genetic diversity within and among many turtle populations. In the extreme example of individual sea turtles observed to nest thousands of kilometers from their previous nesting sites, Bowen and Karl (2007) dubbed them “gravid waifs,” and suggested that, compared with the strategy of absolute natal homing, which could result in species’ extinctions if employed over geologic time periods, such individuals have helped sea turtles flourish during their > 100-million-yr history.

Internest distances among females within a population can vary according to the amount and accessibility of suitable habitat. Najbar and Szuszkiewicz (2007) observed very small internest distances for Emys orbicularis in western Poland among years (mean = 8.8 m), and attributed the pattern to the presence of only a few very small canopy openings along the river at their study site. In our study, several large, linear areas of nesting habitat along the turtle fence at Mount Union were relatively continuous, although some areas near the center of the fence were shaded by vegetation or inaccessible due to steep river banks and large bridge abutments. Gaps in available nesting habitat contributed to some of the largest internest distances in our study.

Lindeman (1992) proposed a model of nest site fixity in which female turtles initially select a nest site based on a variety of physical characteristics, and then return to that site as long as suitable conditions remain. When conditions change such that it becomes unsuitable for nesting, females seek out and fixate on a new site. The initial nesting site of a female may be her natal nest site—if she has homing ability to return to that location (Lindeman 1992). During our study, female G. geographica appeared to fixate on the coal tailings pile, as many females were observed on the coal pile after being captured in the mitigation area, yet few females exhibited the reverse trend.

Black coal tailings, a primary nesting substrate at Mount Union present for more than a century, produce very high incubation temperatures, which likely result in female-biased hatchling sex ratios (Nagle et al. 2018). Although the coal tailings pile at Mount Union contains some sparse vegetation, the amount of open-canopy habitat remained stable during the course of our study compared with the mitigation area. If, from a turtle's perspective, coal tailings represent high-quality, persistently stable nesting habitat, then hatchling survivorship may be correlated with female-biased sex ratios (Reinhold 1998), and adult females may have high fidelity to nest sites on coal tailings compared to sites elsewhere.

Reinhold (1998) suggested that in species with environmental sex determination (such as G. geographica; Ewert and Nelson 1991), selection should favor females produced in rare, high-quality nest sites that return to the same nesting site. Strong natal homing is expected to result in female-biased sex ratios (Bull 1980) and perhaps enduring female-biased populations (Freedberg and Wade 2001).

Our finding that female map turtles display nest site fidelity indicates that maintaining long-term nesting habitat at Mount Union may be essential for conserving G. geographica in central Pennsylvania. More than 200 G. geographica migrate to Mount Union annually to lay eggs, and these turtles likely serve as a source population for the Juniata River (Nagle and Congdon 2016). Recent mitigation work includes control of invasive plants and other vegetation, and construction of large mounds of sand partially covered with dark shale—habitat that may compensate in part for vegetation succession and shading in the mitigation area, or perhaps for loss of habitat in the event the coal tailings are removed for use as fuel at a cogeneration power plant.

A potential problem with fidelity of turtles to the Mount Union nesting area is exposure to environmental contamination. Almost one-third of the adult female G. geographica at Mount Union have abnormally shaped shells, which may result from incubation of eggs in coal tailings or exposure of subadults to contaminants in the Juniata River (Nagle et al. 2018). During the late 19th and early 20th centuries, the area in and around the turtle nesting habitat at Mount Union served as an industrial site along the Juniata River, and the coal tailings and legacy chemicals that remain present risks for wildlife (Pennsylvania Bulletin 2002). If environmental contaminants are responsible for the morphological abnormalities in turtles, then the tendency of gravid females to return to Mount Union may exacerbate the exposure problem.

Nest site fidelity can benefit turtles if the habitat remains stable, results in high nest survivorship, and produces high-quality hatchlings, yet the behavior may be detrimental if it exposes adult females or hatchlings to significant risks. Pennsylvania's long-term conservation strategy for northern map turtles of the Juniata River must ensure suitable nesting habitat at Mount Union and mitigate risks associated with road mortality and contaminant exposure.