2009 NORTHEASTERN NATURALIST 16(3):321–338 Nest-site Selection by Wood Turtles (Glyptemys insculpta) in a Thermally Limited Environment Geoffrey N. Hughes1, William F. Greaves1, and Jacqueline D. Litzgus1,* Abstract _ In oviparous species that lack parental care, fitness of the mother depends on the selection of a high-quality nest site, as mothers do not compensate for poor incubation environment post-hatching. Near the northern range limit of Glyptemys insculpta (Wood Turtle), short summers and cool temperatures may be factors that limit population persistence because potential nest sites may not provide adequate conditions for successful egg incubation in some years. We quantified nest-site selection by examining soil temperature and substrate composition of real Wood Turtle nests (n = 5) and constructed false nests. False nests comprised two treatments: negative-test false nests (n = 5) constructed on beaches not used by females, and positive-test false nests (n = 5) constructed on beaches used by females but in microsites not chosen by females. Temperature was measured as total thermal units and mean temperature during the diel cycle. Soil composition was quantified using moisture content, organic content, and grain-size distribution. Soil temperature was the most important factor in nest-site selection. Temperatures and total thermal units were significantly higher and more variable in real nests than in false nests, except during the night. Soil composition was not significantly different among treatments. Grain sizes ranged from fine to gravel, and real nests contained mainly (58% to 96%) medium sand or larger grains. There was little variation in soil moisture among real nests, suggesting that females were choosing specific humidity conditions for nesting. Our findings can be directly applied to protecting nesting beaches for Wood Turtles, which are considered a species at risk. Introduction According to life-history theory, the amount of energy directed to reproduction should maximize parental fitness through offspring survival (Doughty and Shine 1997, Rollinson and Brooks 2007, Zug et al. 2001). In oviparous species that lack parental care, such strategies can include allocating more energy and lipid reserves to improve the quality of the eggs, or the selection of a high-quality nest site (Kamel and Mrosovsky 2005, Nagle et al. 2003). A high-quality nest environment is critical for species lacking parental care because the parent(s) do not compensate for a poor choice of nesting environment post-hatching (Kamel and Mrosovsky 2005, Kolbe and Janzen 2002, Shine and Harlow 1996). The incubation environment can alter neonate phenotypes and thus affect offspring and maternal fitness (Kolbe and Janzen 2002). For turtles, whose life histories are characterized by great longevity, delayed sexual maturity, iteroparity, and high egg and hatchling mortality, and no parental care, nest-site selection may be especially 1Department of Biology, Laurentian University, Sudbury, ON, P3E 2C6, Canada. *Corresponding author - jlitzgus@laurentian.ca. 322 Northeastern Naturalist Vol. 16, No. 3 important for population persistence because the nest site may directly influence next success and at leas some hatchlings must survive to offset adult mortality (Congdon et al. 1983, 1987; Horne et al. 2003). Nest-site selection in northern locales may be necessary to compensate for shorter incubation periods and cooler average temperatures relative to those in more southern regions (Litzgus and Mousseau 2006, Parker and Andrews 2007). These incubation temperature restrictions in northern regions are hypothesized to be a limiting factor for the distribution of turtle species into the north (Compton 1999, Lavigne et al. 1989). Despite the importance of nest-site selection in turtles, it is often overlooked in studies of maternal investment in favor of other life-history traits, such as egg size, egg mass, and clutch size (Kamel and Mrosovsky 2005). Rarely do studies on nesting involve examination of multiple physical parameters of the nesting environment (Kolbe and Janzen 2002). Previous research on nest-site selection includes studies done on Eretmochelys imbricata Linnaeus (Hawksbill Sea Turtle; Kamel and Mrosovsky 2005), Emydoidea blandingii Holbrook (Blanding's Turtle; Congdon et al. 1983, Gutzke and Packard 1987, Standing et al. 1999), Chelydra serpentina L. (Snapping Turtle; Congdon et al. 1987, Kolbe and Janzen 2002), Graptemys flavimaculata Cagle (Yellow-blotched Map Turtle; Horne et al. 2003), and Malaclemys terrapin Schoepf (Diamondback Terrapin; Burger and Montevecchi 1975). Glyptemys insculpta LeConte (Wood Turtle), Family Emydidae, are medium-sized freshwater turtles, with a mean carapace length of ≈200 mm for males and ≈182 mm for females (Harding and Bloomer 1979, Ernst and Lovich 2009, Walde et al. 2003). In Ontario, Wood Turtles are restricted to three regions that contain multiple populations: one in northern Ontario (which includes the current study population in the Sudbury District; Greaves and Litzgus 2007, 2008), one in southwestern Ontario (Foscarini 1994), and one in central Ontario (Brooks et al. 1992). In Canada, Wood Turtles also occur in Québec, New Brunswick, and Nova Scotia (COSEWIC 2007, Ernst and Lovich 2009, Ernst et al. 1994). The species is currently listed under Appendix II of CITES (Buhlmann 1993) and was recently uplisted from Special Concern to Threatened by the Committee on the Status of Endangered Wildlife in Canada (COSEWIC 2007). In Ontario, the Wood Turtle is considered Endangered under the recently revised provincial Endangered Species Act (Royal Ontario Museum August 2007). The nesting season for Wood Turtles occurs between May and July, depending on geographic location (Ernst et al. 1994, Harding and Bloomer 1979, Walde et al. 2007). Females have been observed constructing nests on sandy beaches, in railway embankments, in agricultural fields, and in gravel quarries (Foscarini 1994, Harding and Bloomer 1979, Walde et al. 2007). Nesting activities usually occur during the early morning and evening (Harding and Bloomer 1979, Walde et al. 2007). Nesting females will wander potential nesting areas, seemingly at random, occasionally throwing sand onto their carapace or stopping to smell the sand (Harding and Bloomer 1979). This behaviour can span several days, with each bout of activity lasting for a number of hours (Harding and Bloomer 1979, Walde et al. 2007). 2009 G.N. Hughes, W.F. Greaves, and J.D. Litzgus 323 Previous studies have noted that females appear to select specific nesting habitats: sandy, elevated, well-drained, and exposed areas (Buech et al. 1997, Farrell and Graham 1991, Harding and Bloomer 1979); however, nest-site selection has not been assessed quantitatively in Wood Turtles. The extensive period of activities related to nesting suggests that females are choosy about where they lay their eggs. Presumably females are choosing the best conditions available to maximize their fitness through offspring survival (Brown and Shine 2004, Kolbe and Janzen 2002, Wilson 1998). Wood Turtles are a rarity among turtles in that they utilize genetic sex determination rather than temperature-dependent sex determination (TSD; Ewert and Nelson 1991). This makes Wood Turtles a model species for studies of nest-site selection, as selective pressures on females should not be working to balance sex ratios in the clutch, but rather to maximize hatching success. The objective of our study was to examine nest-site selection at macro and micro scales in a population of Wood Turtles found near the northern extreme of the species’ range. We hypothesized that female turtles select specific environmental variables to promote successful incubation. We examined nest-site selection by measuring environmental variables including soil composition (moisture content, organic content, and grain-size distribution) and incubation temperature. We examined nest success by counting the number of eggs that hatched out of the total number of viable eggs. We predicted that females would select for stable, warm temperatures to achieve the required total thermal units to promote hatching success by choosing sites with a high degree of sun exposure. We also predicted that turtles would select soil with large grain sizes for better water drainage. Finally, we predicted that females would choose sites at moderate distances from the river, to protect against flooding while also allowing the hatchlings ready access to the safety of the water. Data collected during our study can be directly applied to identifying and protecting nesting beaches for a species at risk. Field-site Description The study site is located along a meandering river in Sudbury District, ON, Canada (N46, W81). For the protection of the Wood Turtle population, the name and exact location will be kept confidential. The river is clear and shallow, with a bottom composed largely of sand and fine gravel. The forest surrounding the river is typical of mixed Boreal and Great Lakes-St. Lawrence Forest, and includes large plantations of Pinus resinosa Solander ex. Aiton (Red Pine) and Pinus banksiana Lambert (Jack Pine). The river is characterized by its beach formations; as the river curves, it produces large, exposed beaches on the inside of the curve and high bluffs (some reaching 20 m in height) along the outside of the curve. The beaches can be elevated up to 3 m above the river. The insides of the curves are characterized by open, sandy beaches, which give way to riparian Salix spp. (willow) and/or Cornus spp. (dogwood) thickets backed by forest. The exposed beaches are the primary nesting habitats used by Wood Turtles. The tops of the sand cliffs on the outside of the curves are forested. 324 Northeastern Naturalist Vol. 16, No. 3 Methods Study techniques Field studies began on 2 May 2006. We outfitted female turtles (n = 8) with 20-g Lotek (Lotek Inc., Newmarket, ON) or 15-g Holohil (Holohil Systems Ltd., Carp, ON) radio-transmitters and tracked each individual 3–4 times within 10-day periods. Females were taken to a local veterinary clinic on 31 May (n = 8) and 5 June (n = 5) for X-radiographs, to determine reproductive status and clutch size (Gibbons and Greene 1979). On 1 June, when nesting behavior was first observed, we tracked females daily, during the early morning and late evening. When nesting females were located, we carefully watched them from concealed positions. Caution was taken to avoid disturbing females, as nesting Wood Turtles are known to be extremely sensitive to human presence, and often abandon the nest if disturbed (Walde et al. 2007). After a female had nested, we measured the nest’s distance from shore and distance from the nearest vegetation using a 50-m measuring tape (±0.1 cm, Lufkin Ind., Lufkin, TX), and recorded the cardinal direction the beach was facing. We excavated the nest and recorded clutch size and the length, width, and mass of each egg for another study. After the nest was processed, the eggs were reburied, and a chicken-wire cage was placed over the nest. The wire was cut into squares approximately ≈70 cm x 70 cm and shaped into domes, then the edges were buried around the nest chamber and the wire cage was secured with 50-cm long wooden stakes. The cages served to protect the nest from large predators, but the large gauge of the wire allowed the hatchlings to leave upon hatching. We obtained data from five real nests. The identity code of the female Wood Turtle was used to identify each nest; for example, the nest of Turtle #3 (T3) was called N3; this nomenclature is used throughout this paper. Ten false nests were constructed during the nesting season; they were placed haphazardly on non-randomly chosen beaches. The purpose of constructing these false nests was to examine macrosite (beach) and microsite (nest pit) selection. Half of the false nests (n = 5; F1–F5) served as negative tests, and were placed on beaches on which no nesting activities were observed (negative macrosite). The remaining false nests (n = 5; F6–F10) served as positive tests, and were placed on beaches where nesting activities were observed, but where the exact locations of nests were unknown (positive macrosite, negative microsite). False nests were not covered with a wire cage. On 13 June, when all observed nesting activity had ceased, waterproofed iButton temperature dataloggers (Dallas Semiconductor, Sunnyvale, CA) were placed in all real and false nests and simultaneously recorded temperatures at 1-h intervals during the incubation period until 6 September. The dataloggers were individually epoxied to a paint stir-stick and pushed through the soil to a depth of 15 cm, the approximate mean nest depth for Wood Turtles in northern locales (Compton 1999, Foscarini 1994, Walde 1998, this study). In the real nests, dataloggers were placed 10 cm away from the nest chamber to avoid damaging any eggs. Air-temperature data from another study (W.F. Greaves, unpubl. data) were acquired for use in our study. Air temperatures were recorded every 1.5 h from 1 July to 13 2009 G.N. Hughes, W.F. Greaves, and J.D. Litzgus 325 October using iButton dataloggers and were compared to the nest temperatures during the incubation period. Soil samples (≈1 kg) were collected ≈10 cm from all real and false nests at nest depth and used to quantify moisture content, organic content, and grainsize distribution. Moisture was determined by weighing a 50-g soil sample in a small tin, placing it in an Isotemp drying oven (Fisher Scientific, Ottawa, ON, Canada) for 6 days and then weighing it again to determine how much moisture was lost. Distribution of grain sizes was determined by sieve analysis; the sieve sizes used were #8 (Endecotts Test Sieves Ltd., London, UK), #16, #30, #50, #100, and #200 (Dual Manufacturing Co., Chicago, IL). The sieves were weighed individually using a Mettler PE22 digital scale (±0.05 g, Mettler- Toledo, Inc., Columbus, OH) and placed in a stack. A 50-g wet soil sample was placed in the stack and put on a sieve shaker for 10 minutes. The sieves were then individually weighed again to determine how much soil was retained in each. Using the USDA soil classification system (Soil Survey Division Staff 1993), the soil sample was divided into the following categories: gravel (≥2 mm grain diameter), very coarse sand (1–2 mm grain diameter), coarse sand (0.5–1 mm grain diameter), medium sand (0.25–0.5 mm grain diameter), fine sand (0.15–0.25 mm grain diameter), very fine sand (0.074–0.15 mm grain diameter), and fines (all silt, clay, and sand particles ≤ 0.074 mm grain diameter). Organic content of the soil was quantified by first placing empty porcelain crucibles in a muffle oven (Lindberg/Blue M, Waltham, MA) for 2 h followed by placement in a desiccator (Corning Inc., New York, NY) for 1 h to cool without absorbing moisture. The crucibles were weighed with a Mettler AE 200 scale (±0.0005 g) (Mettler-Toledo, Inc., Columbus, OH). Soil samples were sieved using a #25 sieve (W.S. Tyler Company of Canada, Ltd., St. Catharines, ON, Canada) to create a more homogenized sample, and ≈5 g of dried soil was placed in the crucible, after which the crucible was reweighed and replaced in the muffle oven for 3 h. The crucibles were removed and allowed to cool in the desiccator, and were then weighed a final time to determine the organic content. As hatching time approached, after ≈65 days of incubation (Tuttle and Carroll 2003), nests were patrolled daily when possible to determine the date of emergence so that incubation times could be determined. On 13 October 2006, after we felt that successful emergence was no longer possible due to snow cover and frequent frosts, all of the nests were excavated to determine the number of eggs that had successfully hatched (Walde et al. 2007). Eggshells present in the nest chamber were considered to indicate successful emergents, and unhatched eggs were considered unsuccessful. The unhatched eggs were brought back to the lab and dissected to determine if there was any embryonic development. Eggs with no development were considered unfertilized and were not included in the assessment of hatching success. We used hatching success as an indicator of nest success. Due to uncontrollable circumstances, N29 (real nest), and the false nests F3 (negative-test false nest) and F10 (positive-test false nest) were excluded from the temperature analyses. Thus, sample size for each treatment for the temperature analyses was n = 4. N29 and F3 were destroyed because of human interference, and the data sets from the data loggers were incomplete. 326 Northeastern Naturalist Vol. 16, No. 3 The data logger in F10 malfunctioned and all temperature data were lost. Sample sizes were n = 5 for each treatment for all other analyses. Statistical analyses Nests in this study belonged to one of three treatments: real nests (n = 5), positive-test false nests (n = 5), and negative-test false nests (n = 5). The variables compared among the three treatments were divided into three subcategories: (1) soil composition, including soil moisture, organic content, and grain-size distribution; (2) soil temperature, including total thermal units and mean temperature; and (3) external features of real nests, including distance from shore and distance from the nearest vegetation. We calculated total thermal units as the arithmetic sum of all of the temperature data during the incubation period. Because of small sample sizes, a repeated-measures ANOVA did not suit our data, therefore all statistical analyses were conducted using nonparametric Kruskal-Wallis (H) tests and multiple comparisons (MC) of probability values. A paired F-test and an ANOVA were used post-hoc to analyze differences among variances for soil moisture content and mean incubation temperature, respectively. We conducted all analyses using Statistica 6.1 (StatSoft Inc., Tulsa, OK). We examined grain-size distribution by testing for differences in the percent composition of each of the different grain-size classes among the treatments. Temperature was divided into 6-h intervals for comparison across the diel cycle; the intervals were labelled night (00:01h– 06:00h), morning (06:01h–12:00h), afternoon (12:01h–18:00h), and evening (18:01h–24:00h). Temperature data in each of these diel categories were then examined for differences among the three treatments. Results Nest-site descriptions Five real turtle nests were found during the nesting period. T3 nested on the evening of 5 June on a north-facing beach. The beach had an open canopy with small willow saplings located ≈1 m from the nest. There were also scattered clumps of grasses around the beach, and some pieces of driftwood. T20 nested on the evening of 5 June on a south-facing beach. The beach was highly elevated above the river, and had large amounts of Equisetum arvense L. (Field Horsetail), with one horsetail plant less than 10 cm from the nest. T52 nested during the morning of 8 June on a southwest-facing sandy bluff. The bluff was free of vegetation and very steep, but had Comptonia peregrine L. (Coulter) (Sweet Fern) and Alnus incana L. (Moench) (Speckled Alder) along the edges, with roots running in the sand. T31 nested during the evening of 8 June on the same bluff as T52. This nest (N31) was located at the base of the bluff in the vicinity of willow saplings, Solidago canadensis L. (Goldenrod), and Pteridium aquilinum L. (Kuhn) (Bracken Fern). T29 nested during the evening of 8 June on the same bluff as T52 and T31. This nest (N29) was located near the base of the bluff, but higher up than N31, with similar floral assemblages. 2009 G.N. Hughes, W.F. Greaves, and J.D. Litzgus 327 Soil temperature Incubation temperature was the factor that differed most among nest treatments. Mean real nest temperatures were significantly warmer than negative and positive-test false nests (MC: H = 10828.8, df = 11, 24480, P < 0.001), and negative-test false nests were significantly warmer than positive-test false nests (MC: P < 0.001, Fig. 1). Variance among treatments was significantly different (F = 5.97, df = 2, 9, P = 0.022); however, we lacked the statistical power to conduct a post-hoc test. Examination of the data by eye indicates that real nests had more variable soil temperatures than both false-nest treatments, but that there was little or no difference between the false-nest treatments. Total thermal units also differed among treatments (Fig. 2). Total thermal units were significantly higher in real nests than in negative-test false nests (MC: H = 6.5, df = 2, 12, P = 0.04), but not higher than in positive-test false nests (MC: P > 0.05). The two false-nest treatments were not significantly different from one another (MC: P > 0.05). Temperature differences were found among the three treatments when divided into 6-h diel intervals (MC: H =1620.3, df = 2, 2040, P < 0.001). During the night interval, temperatures in the real Figure 1. Mean incubation temperatures (central line; in °C), standard error (box), and 95% confidence interval (whiskers) for four Wood Turtle (Glyptemys insculpta) nests (range = 8.5–41.0 °C), four positive-test false nests (haphazard sites on beaches used by turtles; range = 10.5–32.0 °C), and four negative-test false nests (haphazard sites on beaches not used by turtles; range = 11.5–35.0 °C), from a site in Sudbury District, ON, Canada during the period 13 June to 6 September 2006. n = 8160 temperature data points in each nest treatment. 328 Northeastern Naturalist Vol. 16, No. 3 nests were not significantly different from the negative-test false nests (MC: P > 0.05), and the positive-test false nests were significantly warmer than real and negative-test false nests (MC: P < 0.001; Fig. 3). During the morning interval, real nests were significantly warmer than negative- and positive-test false nests (MC: P < 0.001); the false-nest treatments were not significantly different from each other (MC: P > 0.05; Fig. 3). During the afternoon interval, real nests were significantly warmer than negative- and positive-test false nests, and negative-test false nests were significantly warmer than positive-test false nests (MC: P < 0.001 in all cases; Fig. 3). During the evening interval, real nests were significantly warmer than both false-nest treatments (MC: P < 0.001 in both cases); the false-nest treatments were not significantly different from each other (MC: P > 0.05; Fig. 3). Nest temperatures were warmer than air temperatures. Mean air temperature during the nest incubation period was 13.2 °C (± 7.0 °C SD; range = -5.5 to 31.0 °C). Soil composition Soil moisture was low in all treatments (Fig. 4). Mean soil moisture did not differ among the treatments (MC: H = 2.6, df = 2, 15, P = 0.28), although variation among real nests was less than that in the two false-nest treatments Figure 2. Total thermal units (central line; in °C), standard error (box), and 95% confidence interval (whiskers) for the incubation period of four Wood Turtle (Glyptemys insculpta) nests, four positive-test false nests, and four negative-test false nests, from a site in Sudbury District, ON, Canada during the period 13 June to 6 September 2006. Total thermal units were calculated as the arithmetic sum of all temperature data points from the nest. 2009 G.N. Hughes, W.F. Greaves, and J.D. Litzgus 329 (Fig. 4). This difference in variation was significant between real nests and negative-test false nests (F = 22.00, df = 2, 4, P < 0.01), but not between real nests and positive-test false nests (F = 5.75, df = 2, 4, P > 0.05). Organic content was low in all samples and did not differ among treatments (MC: H = 0.2, df = 2, 15, P = 0.89). Grain size did not differ among treatments with respect to proportion of gravel (MC: H = 5.3, df = 2, 15, P = 0.07), very coarse sand (MC: H = 2.5, df = 2, 15, P = 0.28), coarse sand (MC: H = 3.1, df = 2, 15, P = 0.21), medium sand (MC: H = 2.0, df = 2, 15, P = 0.38), fine sand (MC: H = 2.6, df = 2, 15, P = 0.28), very fine sand MC: H = 1.2, df = 2, 15, P = 0.56) and fines (MC: H = 0.4, df = 2, 15, P = 0.82). However, a visual comparison of the grain size distribution revealed a potential pattern of selection for grain size among real nests, when compared to the seemingly random grain size composition in both false-nest groups (Fig.5). External features Distances from the river and from the nearest vegetation were variable within and among treatments. For real nests, mean distance from the river was 10.4 m ± 1.7 m (SE), and mean distance from the nearest vegetation was 1.4 m ± 0.2 m (SE). For positive-test false nests, mean distance from the river was 7.9 m ± 0.9 m (SE), and mean distance from the nearest vegetation was 1.3 m ± 0.7 m (SE). For negative-test false nests, mean distance from the Figure 3. Changes in mean temperature over diel time intervals (night [00:01h– 06:00h], morning [06:01h–12:00h], afternoon [12:01h–18:00h], and evening [18:01h–24:00h]) in four Wood Turtle (Glyptemys insculpta) nests (diamonds and solid line), four positive-test false nests (triangles and dotted line), and four negativetest false nests (squares and dashed line), from a riverine site in the Sudbury District, ON, Canada in the summer of 2006. 330 Northeastern Naturalist Vol. 16, No. 3 river was 6.1 m ± 1.2 m (SE), and the mean distance from the nearest vegetation was 1.3 m ± 0.8 m (SE). Real nests were significantly further from shore than negative-test false nests (MC: H = 4.0, df = 2, 15, P = 0.047) but not positive-test false nests (MC: H = 4.0, df = 2, 15, P = 0.25), and the false-nest treatments were not significantly different from each other (MC: H = 4.0, df = 2, 15, P = 0.46). There were no significant differences among treatments in regards to distance from the nearest vegetation (MC: H = 0.59, df = 2, 14, P = 0.74). Nest success The hatching success of each real nest was expressed using the ratio of the number of eggs that hatched out of the total number of viable eggs, and was used as an indicator of nest success. Mean nest success for the 5 nests Figure 4. Moisture content (central line; in %), standard error (box), and 95% confidence interval (whiskers) of soil samples taken from five Wood Turtle (Glyptemys insculpta) nests, five positive-test false nests, and five negative-test false nests, from a riverine site in the Sudbury District, ON, Canada in the summer of 2006. Figure 5 (opposite page). Percentage grain-size distribution of five real nests (A), five negative-test false nests (B), and five positive-test false nests (C) of soil samples taken from Wood Turtle (Glyptemys insculpta) nesting beaches along a river in the Sudbury District, ON, Canada in the summer of 2006. From bottom to top of each bar, gravel (≥2 mm grain diameter), very coarse sand (1–2 mm grain diameter), coarse sand (0.5–1 mm grain diameter), medium sand (0.25–0.5 mm grain diameter), fine sand (0.15–0.25 mm grain diameter), very fine sand (0.074–0.15 mm grain diameter) , and fines (≤0.074 mm grain diameter). 2009 G.N. Hughes, W.F. Greaves, and J.D. Litzgus 331 332 Northeastern Naturalist Vol. 16, No. 3 was 92.8 % ± 4.6% (SE). N52 had 100% nest success (7 hatched eggs out of 7 viable eggs), and first emergence of hatchlings was 82 days after oviposition (Table 1). Three eggs were depredated after first emergence, and these were not counted in the success value, as it is impossible to determine if the eggs were viable or not. Campers tampered with N52 before first emergence, which may have allowed the predator (suspected to be Vulpes vulpes L. [Red Fox]) to detect the nest. N29 had 100% hatching success (8 hatched eggs out of 8 viable eggs), and first emergence occurred 82 days after oviposition (Table 1). N3 had 100% nest success (5 hatched eggs out of 5 viable eggs), and first emergence was 79 days after oviposition, which was the shortest incubation time in the study (Table 1). N31 had 86% hatching success (6 hatched eggs out of 7 viable eggs), and emergence was between 84 and 89 days after oviposition (Table 1). N20 had 78% nest success (8 hatched eggs out of 9 viable eggs), and emergence was 87 to 92 days after oviposition; this nest had the longest incubation time (Table 1). Although one of the “dead” eggs had hatched, and the hatchling was found alive in the nest at the final excavation on 13 October, it would likely not have survived the winter (Parren and Rice 2004), and was thus not counted as a successful hatch. Discussion Wood Turtle females in the Sudbury District selected for high, variable temperatures for their nest sites. On a macrosite scale, turtles chose elevated sandy beaches with little or no vegetation cover, whereas on a microsite scale, they chose sites with large sand grain sizes, high incubation temperatures, low moisture content, and low organic content. Soil temperature Temperature appears to be the most important factor in the selection of nest sites by female Wood Turtles in the Sudbury District. The mean temperatures in real nests were significantly warmer than those in the haphazardly placed false nests, and the maximum temperature experienced among the real nests was 41.0 °C (Fig. 1). During the morning, afternoon, and evening, real turtle nests were warmer than the false-nest groups (Fig. 3). Bodie et al. (1996) found a similar pattern of diel nest temperatures in both Pseudemys floridana Table 1. Summary of nest-site success with respect to total number of eggs, number of hatched eggs, number of viable eggs, and incubation length of nests of the Wood Turtle (Glyptemys insculpta), from a river in the Sudbury District, ON, Canada in the summer of 2006. Nest success was calculated as the proportion of hatched eggs out of the number of viable eggs. Incubation Nest Total eggs Hatched eggs Viable eggs Nest success length (days) N3 8 5 5 1.00 79 N20 10 8 9 0.89 87–92 N29 8 8 8 1.00 82 N31 14 6 7 0.86 84–89 N52 10 7 7 1.00 82 2009 G.N. Hughes, W.F. Greaves, and J.D. Litzgus 333 Stejneger (Florida River Cooter) and Kinosternon subrubrum Lacépéde (Eastern Mud Turtle) nests. Also of interest is the higher standard errors and ranges in temperature in the real nests compared to both false-nest treatments (Fig. 1); thermal stability seems to be less important than high temperatures in nest-site selection, which does not support our original hypothesis of selection for stable temperatures. Northern female Wood Turtles in our study appear to be selecting for warm and variable nest temperatures. The highest recorded incubation temperature for Natator depressus Garman (Flatback Turtle) during a study conducted in Australia was 36.4 °C, despite their nesting in a tropical environment (Hewavisenthi and Parmenter 2002). The Flatback Turtle has TSD (Hewavisenthi and Parmeter 2002) while the Wood Turtle does not (Ewert and Nelson 1991), allowing the Wood Turtle to select for high incubation temperatures to promote embryonic development without skewing sex ratios. Compton (1999) found that more variable incubation temperatures led to a measurably faster development rate in Wood Turtles. Compton’s (1999) findings were from laboratory studies, and our study shows that the same conditions may apply in the wild. The significantly lower nest temperatures during the early morning hours appear to be what causes the high temperature variation in the real nests, and may be important in promoting shorter incubation periods in Wood Turtles. Finding nesting sites with incubation temperatures that promote successful hatching is critical in the thermally limited, northern portion of the Wood Turtle’s range (Compton 1999, Shine 1999); females need to choose the best nest sites to maximize the survival of their offspring, and by extension, their own fitness (Doughty and Shine 1997, Rollinson and Brooks 2007, Zug et al. 2001). Adequate incubation temperatures may limit the northern range boundary of turtles (Compton 1999, Lavigne et al. 1989). Because Wood Turtles have genetic sex determination (Ewert and Nelson 1991), natural selection will not necessarily favor females that choose a narrow temperature range around a pivotal temperature to ensure a balanced sex ratio of neonates, as would be the case for a female from a species with TSD (Miller et al. 2004). A female Wood Turtle need only find a site with temperatures that would promote incubation. Soil composition The soil composition studies yielded no significant differences among treatments. Although not statistically significant (P > 0.07 in all cases), the grain-size distribution analysis showed a possible trend that may indicate selection for larger grain sizes in nest sites. The mean distributions showed that 86% of the substrate composition in real nests was medium sand or larger (>0.25 mm), while it was 47% in the positive-test false nests, and 63% in negative-test false nests (Fig. 5). We suspect that this may be a real trend, and that differences would be significant with a larger sample size. The soil-moisture analyses also revealed an intriguing pattern. Soilmoisture variation was low in real nests, higher in positive-test false nests, and higher still in negative-test false nests (Fig. 4). This implies that females are actively choosing macrosites and microsites with a specific, low moisture 334 Northeastern Naturalist Vol. 16, No. 3 content level. With a larger sample size, we suspect that differences in soil moisture between real and false nests would be significant. Organic content did not show any patterns among treatments, and Wood Turtles may be ignoring or actively selecting against organic content. The sandy beaches where the Sudbury District population of Wood Turtles nested are naturally low in organic content; however, Wood Turtles in other regions are known to nest in agricultural fields (Buech et al. 1997, Tuttle and Carroll 2003), and captive Wood Turtles have been observed nesting in peat (Farrell and Graham 1991). Organic content reduces nest success in Chrysemys picta Schneider (Painted Turtle; Hughes and Brooks 2006) and may reduce success in Lepidochelys olivacea Eschscholtz (Olive Ridley Turtle; Clusella Trallas and Paladino 2007). Our findings suggest that female Wood Turtles do not select specific organic content when selecting a nest site, although soil organic content affects nest-site selection by other turtle species (Clusella Trallas and Paladino 2007, Hughes and Brooks 2006). Dry, sandy soils warm up more quickly in the sun than wetter soils, and sandy soils do not hold water as well as organic-rich soils (Brady and Weil 2002). In the Sudbury District, females appear to select dry, low-humic soil conditions, to promote heating (Brady and Weil 2002), which in turn will promote embryo incubation. External features Wood Turtles appeared to select for nest sites relatively far from the river, likely to protect the eggs from flooding (Standing et al. 1999). Wood Turtles are known to prefer elevated, open, well-drained sites for nesting (Buech et al. 1997, Farrell and Graham 1991), which was supported by our findings. More than 25% of nests at a study site in Maine were flooded due to rainfall and water-release from dams (Compton 1999). Soil temperature on a maritime beach was significantly lower closer to the water (Wood and Bjorndal 2000). These observations imply a trade-off between choosing sites away from the river, which protects the eggs from flooding and offers higher soil temperatures, and sites close to the water, which allows emergent hatchlings to reach the relative safety of the water (Wood and Bjorndal 2000). Female Wood Turtles appeared to select sites away from vegetation for nesting. The exception was N20, which was located 9 cm from a stand of Field Horsetail. Vegetation near turtle nests can cause two problems: shade and invasion of the nest by roots. Shade reduces sunlight exposure and thus the heat units received by a nest (Hughes and Brooks 2006). Root invasions can lead to egg mortality (Behler and Castellano 2005, Congdon et al. 2000, Harding and Bloomer 1979). The beaches at our study site are at high elevations, and as such have some protection from flooding; thus, nests can be placed closer to the river, and have little vegetation, which reduces the risk of root invasion and shading of nests. Nest success Mean hatching success of nests was 92.8% for our study. The nests with the highest hatching success were N3, N52, and N29, each having 100% hatching of viable eggs. N31 and N20 had lower, but still high, hatching 2009 G.N. Hughes, W.F. Greaves, and J.D. Litzgus 335 success, with 86% and 78% hatching success respectively. These values of hatching success are relatively high for turtles. Standing et al. (1999) found lower rates of hatching success for Blanding's Turtles over their 3-year study; they did not count hatchlings that had not left the nest upon excavation as successful hatches, but they found many more hatchlings that had not left the nest than we did. The range of incubation times, defined as the period between oviposition and first emergence, was 79–92 days in our study. Wood Turtles are noted for having relatively short incubation periods compared to other freshwater turtles (Harding and Bloomer 1979). For example, Standing et al. (1999) found a range of 80–128 days for Blanding's Turtle nests over the three years of their study. Congdon et al. (1983) found a slightly smaller range of 73–104 days for Blanding's Turtles over their 6-year study. Walde et al. (2007) found a range of 65–116 days in 1996 and 60–99 days in 1997 for Wood Turtles in Québec. The incubation time found in our study was not appreciably shorter than the above-mentioned studies; however, our small sample size may not have given us a representative sample. With respect to incubation time, the fastest developing nest was N3, from which the first emergence took place 79 days after oviposition, and 8 days before the first frost (31 August; Table 1). N3 had the highest total thermal units (45,884.5 °C), and the highest percentage (96%) of substrate composed of medium or larger sand grains among the real nests (Fig. 5). The slowest developing nest was N20, which had the longest incubation period (between 87 and 92 days; Table 1). N20 was characterized by having the lowest total thermal units (43,649.5 °C), and the lowest percentage (58%) of medium sand grains or larger (Fig. 5). N20 was also the closest to vegetation of all of the real nests (0.09 m), which suggests the possibility of shade interfering with the amount of sunlight received by the nest (Hughes and Brooks 2006). When the nest was excavated, no root damage to eggs was discovered. With respect to hatching success, the most successful nests were N29, N52, and N3, with 100% hatching success. The least successful was N20, with 78% hatching success. Conclusions and future directions Our hypothesis and predictions were partially supported. Soil temperature of the real nests was significantly higher than in the false nests. Therefore Wood Turtles appear to select for higher temperatures when constructing a nest; this selection for higher temperatures supports our prediction. However, in contrast to our prediction, females did not select for stable temperatures, instead preferring variable nest temperatures which may shorten incubation times (Compton, 1999) to ensure hatchling emergence before the first frost. Although soil data analyses were inconclusive, we suspect that with a larger sample size, a significant pattern of selection for soil composed primarily of medium and coarse sand would be found. Future studies should seek to increase sample sizes of all treatments, as small sample size was a limitation in our study. Future studies should also include a third false-nest treatment: nests constructed in the test digs and tracks made by females, which might represent sites actively rejected by the females (Wood and Bjorndal 2000). 336 Northeastern Naturalist Vol. 16, No. 3 Suitable nesting beaches may be critical for the presence of a Wood Turtle population, and understanding what constitutes suitable nesting habitat is important for conservation (Buech et al. 1997, Kolbe and Janzen 2002). Our research provides new data that identify environmental variables in nesting habitats that are important for the conservation of this at-risk reptile. Acknowledgments Financial support for this research came from the Endangered Species Recovery Fund of the World Wildlife Fund Canada and Environment Canada, the Ontario Ministry of Natural Resources, NSERC, and Laurentian University. We would like to thank A. Gallie, Y. Chen, and H. Ylitalo for the use of their labs and expertise while doing the soil analyses. This paper has benefited from comments by A. Breisch and two anonymous reviewers. This research was conducted by G.N. Hughes for his Honours Thesis at Laurentian University. The study was carried out under the guidelines of the Canadian Council on Animal Care and the Laurentian University Animal Care Committee (protocol numbers 2004-09-01 and 2004-11-01). Literature Cited Behler, J.L., and Castellano. 2005. Glyptemys insculpta (North American Wood Turtle). Nest-invading plants. Herpetological Review 36:311. Bodie, J.R., K.R. Smith, and V.J. Burke. 1996. A comparison of diel nest temperature and nest-site selection for two sympatric species of freshwater turtles. American Midland Naturalist 136:181–186. Brady, N.C., and R.R. Weil. 2002. The Nature and Properties of Soils, 13th Edition. Pearson Education Canada, Ltd. Prentice Hall, Upper Saddle River, NJ. Brooks, R.J., C.M. Shilton, G.P. Brown, and N.W.S. Quinn. 1992. Body size, age distribution, and reproduction in a northern population of Wood Turtles (Clemmys insculpta). Canadian Journal of Zoology 70:462–469. Brown, G.P., and R. Shine. 2004. Maternal nest-site choice and offspring fitness in a tropical snake (Trodipinophis mairii, Colubridae). Ecology 85:1627–1634. Buech, R.R., L.G. Hanson, and M.D. Nelson. 1997. Identification of Wood Turtle nesting areas for protection and management. Pp. 383–391, In J. Van Abbema (Ed.). Proceedings: Conservation, Restoration, and Management of Tortoises and Turtles - An International Conference. New York Turtle and Tortoise Society, New York, NY. Buhlmann, K.A. 1993. Legislation and conservation alert. Herpetological Review 24:125. Burger, J., and W.A. Montevecchi. 1975. Nest-site selection in the terrapin Malaclemys terrapin. Copeia 1975:113–119. Clusella Trallas, S., and F.V. Paladino. 2007. Micro-environment of Olive Ridley Turtle nests deposited during an aggregated nesting event. Journal of Zoology 272:367–376. Committee on the Status of Endangered Wildlife in Canada (COSEWIC) 2007. COSEWIC assessment and update status report on the Wood Turtle, Glyptemys insculpta, in Canada. Ottawa, ON, Canada. vii + 42 pp. Available online at www. sararegistry.gc.ca/status/status_e.cfm. Compton, B.W. 1999. Ecology and conservation of the Wood Turtle (Clemmys insculpta) in Maine. M.Sc. Thesis. University of Maine, Orono, ME. 91 pp. Congdon, J.D., D.W. Tinkle, G.L. Breitenbach, and R.C. van Loben Sels. 1983. Nesting ecology and hatching success in the turtle Emydoidea blandingii. Herpetologica 39:417–429. 2009 G.N. Hughes, W.F. Greaves, and J.D. Litzgus 337 Congdon, J.D., G.L. Breitenbach, R.C. van Loben Sels, and D.W. Tinkle. 1987. Reproduction and nesting ecology of Snapping Turtles (Chelydra serpentina) in southeastern Michigan. Herpetologica 43:39–54. Congdon, J.D., R.D. Nagle, O.M. Kinney, M. Osentoski, H.W. Avery, R.C. van Loben Sels, and D.W. Tinkle. 2000. Nesting ecology and embryo mortality: Implications for hatchling success and demography of Blanding’s Turtles (Emydoidea blandingii). Chelonian Conservation and Biology 3:569–579. Doughty, P., and R. Shine. 1997. Detecting life-history trade-offs: Measuring energy stores in “capital” breeders reveals costs of reproduction. Oecologica 110:508–513. Ernst, C.H., and J.E. Lovich. 2009. Turtles of the United States and Canada (2nd Edition). The John Hopkins University Press, Baltimore, MD. Ernst, C.H., J.E. Lovich, and R.W. Barbour. 1994. Turtles of the United States and Canada. Smithsonian Institution Press, Washington, DC. Ewert, M.A., and C.E. Nelson. 1991. Sex determination in turtles: Diverse patterns and some possible adaptive values. Copeia 1991:50–69. Farrell, R.F., and T.E. Graham. 1991. Ecological notes on the turtle Clemmys insculpta in northwestern New Jersey. Journal of Herpetology 25:1–9. Foscarini, D.A. 1994. Demography of the Wood Turtle (Clemmys insculpta) and habitat selection in the Maitland River valley. M.Sc. Thesis. University of Guelph, Guelph, ON, Canada. 139 pp. Gibbons, J.W., and J.L. Greene. 1979. X-ray photography: A technique to determine reproductive patterns of freshwater turtles. Herpetologica 35:86–89. Greaves, W.F., and J.D. Litzgus. 2007. Overwintering ecology of Wood Turtles (Glyptemys insculpta) at the species' northern range limit. Journal of Herpetology 41:32–40. Greaves, W.F., and J.D. Litzgus. 2008. Chemical, thermal, and physical properties of sites selected for overwintering by northern Wood Turtles (Glyptemys insculpta). Canadian Journal of Zoology 86:659–667. Gutzke, W.H.N., and G.C. Packard. 1987. The influence of temperature on eggs and hatchlings of Blanding's Turtles, Emydoidea blandingii. Journal of Herpetology 21:161–163. Harding, J.H., and T.J. Bloomer. 1979. The Wood Turtle, Clemmys insculpta. A natural history. Bulletin of the New York Herpetological Society 15:9–26. Hewavisenthi, S., and C.J. Parmenter. 2002. Incubation environment and nest success of the Flatback Turtle (Natator depressus) from a natural nesting beach. Copeia 2002:302–312. Horne, B.D., R.J. Brauman, M.J.C. Moore, and R.A. Siegel. 2003. Reproductive and nesting ecology of the Yellow-blotched Map Turtle Graptemys flavimaculata for conservation and management. Copeia 2003:729–738. Hughes, E.J., and R.J. Brooks. 2006. The good mother: Does nest-site selection constitute parental investment in turtles? Canadian Journal of Zoology 84:1545–1554. Kamel, S.J., and N. Mrosovsky. 2005. Repeatability of nesting preferences in the Hawksbill Sea Turtle, Erotmochelys imbricata, and their fitness consequences. Animal Behaviour 70:819–828. Kolbe, J.J., and F.J. Janzen. 2002. Impact of nest-site selection on nest success and nest temperature in natural and disturbed habitats. Ecology 83:269–281. Lavigne, D.M., R.J. Brooks, D.A. Rosen, and D.A. Galbraith. 1989. Cold, energetics, and populations. Pp. 403–432, In L.C. Wang (Ed.). Advances in Comparative and Environmental Physiology (Volume 4). Springer-Verlag, Berlin, Germany. Litzgus, J.D., and T.A. Mousseau. 2006. Geographic variation in reproduction in a freshwater turtle (Clemmys guttata). Herpetologica 62:132–140. 338 Northeastern Naturalist Vol. 16, No. 3 Miller, D., J. Summers, and S. Silber. 2004. Environmental versus genetic sex determination: A possible factor in dinosaur extinction? Fertility and Sterility 81:954–964. Nagle, R.D., M.V. Plummer, J.D. Congdon, and R.U. Fischer. 2003. Parental investment, embryo growth, and hatchling lipid reserves in Softshell Turtles (Apalone mutica) from Arkansas. Herpetologica 59:145–154. Parker, S.L., and R.M. Andrews. 2007. Incubation temperature and phenotypic traits of Sceloporus undulatus: Implications for the northern limits of distribution. Oecologica 151:218–231. Rollinson, N., and R.J. Brooks. 2007. Proximate constraints on reproductive output in a northern population of Painted Turtles: An empirical test of the bet-hedging paradigm. Canadian Journal of Zoology 85:177–184. Royal Ontario Museum. 2007. Ontario's Biodiversity: Species At Risk. Available online at http://www.rom.on.ca/ontario/risk.php?doc_type=fact&lang=&id=95. Accessed September 2007. Shine, R. 1999. Egg-laying reptiles in cold climates: Determinants and consequences of nest temperatures in montane lizards. Journal of Evolutionary Biology 12:918–926. Shine, R., and P.S. Harlow. 1996. Maternal manipulation of offspring phenotypes via nest-site selection in an oviparous lizard. Ecology 77:1808–1817. Soil Survey Division Staff. 1993. Soil Survey Manual. Soil Conservation Service, US Department of Agriculture Handbook 18. Washington, DC. Standing, K.L., T.B. Herman, and I.P. Morrison. 1999. Nesting ecology of Blanding’s Turtles (Emydoidea blandingii) in Nova Scotia, the northeastern limit of the species’ range. Canadian Journal of Zoology 77:1609–1614. Tuttle, S.E., and D.M. Carroll. 2003. Ecology and natural history of the Wood Turtle (Clemmys insculpta) in southern New Hampshire. Chelonian Conservation and Biology 2:447–449. Walde, A.D. 1998. Ecology of the Wood Turtle, Clemmys insculpta, Quebec, Canada. M.Sc. Thesis. McGill University, Montreal, PQ, Canada. 95 pp. Walde, A.D., J.R. Bider, C. Daigle, D. Masse, J.-C. Bourgeois, J. Jutras, R.D. Titman. 2003. Ecological aspects of a Wood Turtle, Glyptemys insculpta, population at the northern limit of its range in Québec. Canadian Field-Naturalist 117:377–388. Walde, J.D., J.R. Bider, D. Masse, R.A. Saumure, R.D. Titman. 2007. Nesting ecology and hatching success of the Wood Turtle, Glyptemys insculpta, in Québec. Herpetological Conservation and Biology 2:49–60. Wilson, D.S. 1998. Nest-site selection: Microhabitat variation and its effects on the survival of turtle embryos. Ecology 79:1884–1892. Wood, D.W., and K.A. Bjorndal. 2000. Relation of temperature, moisture, salinity, and slope to nest-site selection in Loggerhead Sea Turtles. Copeia 2000:119–128. Zug, G.R., L.J. Vitt, and J.P. Caldwell. 2001. Herpetology: An Introductory Biology of Amphibians and Reptiles, Second Edition. Academic Press, New York, NY.