The habitat of the giant Amazon turtle (Podocnemis expansa) includes large rivers as well as meanders formed by clay–sand unconsolidated alluvium (Cantarelli and Alves 1999). In these environments sand beaches are normally formed by the exposition of sand banks (Vanzolini 1967; Von Hildebrand et al. 1988; Soini 1997).

Chelonians may use a combination of multiple factors for nest site selection, such as beach height in relation to the water surface and sand depth, consistency, and humidity (Pritchard and Trebbau 1984; Ratterman and Ackerman 1989; Mortimer 1990; Speakman et al. 1998; Seigel and Dodd 2000; Wood and Bjorndal 2000). Embryogenesis is significantly affected by those factors (Tucker and Warner 1999). Gaseous exchange is also affected by the substrate physical structure (Prange and Ackerman 1974; Milton et al. 1997) and can determine embryo survivorship (Bobyn and Brooks 1994; Milton et al. 1997; Packard et al. 1999).

Studies concerning the physical substrate structure at freshwater turtle nest sites should guide chelonian management programs based on egg collection (Ojasti 1967; IBAMA 1989; Martinez and Rodriguez 1997) either for artificial incubation (Plummer et al. 1994; Rostal et al. 1994; Souza and Vogt 1994; Molina and Gomes 1998) or egg translocation (Alho et al. 1979; Von Hildebrand et al. 1988; Stancyk 1995). In the present study we determined the relationship between incubation success and some physical characteristics of Podocnemis expansa nests in the southern Amazon basin, Brazil.

Materials and Methods

This study was carried out on 4 sand beaches of the Araguaia and Crixás-Açu rivers in the southern Amazon basin on the border between the states of Goiás (GO) and Mato Grosso (MT) in Brazil. We sampled 5 nests per beach for a total of 20 nests in both rivers.

Sand samples were collected 0700 to 0930 hours or 1600 to 1830 hours during the first 7 days after egg-laying. The nests were carefully excavated until the egg chamber appeared. The first egg layer was then taken off the nest and a cylindrical metal ring (volume 18.64 cm³) was introduced on the border of the egg chamber in order to collect sand as close as possible to the eggs. Two samples were taken from each nest, weighed on a digital balance (0.1 g/400 g), and stored in zip-lock plastic bags.

The remaining eggs were then removed from the egg chamber for counting, and then reintroduced randomly. After this procedure, the eggs were manually recovered with sand and the nests were protected by a surrounding plastic net and tapped by a metallic fence. Two control samples were taken by the same metallic cylinder by the side of each nest at the same depth, weighed, and returned to the nest. Nests were reopened at day 45 after egg laying for hatching monitoring.

The following physical characteristics of the sand were analyzed at the Laboratório de Física do Solo from Escola Superior de Agricultura “Luiz de Queiroz”, University of São Paulo, according to Camargo et al. (1986) and Gee and Bauder (1986): 1) sand grain distribution (in percent): very fine (<0.106 mm), fine (0.107–0.25 mm), medium (0.25–0.4 mm), gross (0.5–1.17 mm), very gross (>1.18 mm), and total sand; 2) moisture in the sand (%) = (humid sand weight − dry sand weight) × 100/(dry sand weight); 3) apparent density (D) (g/cm³) = dry sand weight/18.64.

For data that did not present normal distribution we used nonparametric statistics (Kruskal-Wallis and Mann-Whitmann) in order to test the relationships between the variables studied (Sokal and Rholf 1995). Otherwise, we used parametric tests.

Results

Results are presented in Table 1. Sand varied among beaches in terms of all kinds of grain size (Kruskal-Wallis: p < 0.05, df = 5), but not for fine-grained and total sand percentages (Kruskal-Wallis: p > 0.05, df = 5). There was no consistent relationship between hatching success and the physical characteristics of the nests, the sand beach where the female laid eggs, and the sand apparent density, relative humidity, and grain size distribution (Kruskal-Wallis: p > 0.100, n = 20). There was no consistent relationship between hatching success and incubation period (ANOVA: P = 0.196, n = 20) either. However, there was a consistent relationship between the sand beach where the females nested and clutch size (ANOVA: p = 0.004; 18GO: 100.15 ± 14.09; 6MT: 76 ± 17.84 eggs, n = 20), and incubation period (ANOVA: p < 0.001; 18GO: 54.769 ± 0.725; 6MT: 57 ± 0.756 days, n = 20).

Table 1. Physical characteristics of Podocnemis expansa nests.

View Fullscreen

There was no significant difference in apparent density between egg chamber and the sand around the nest at the same depth (Mann-Whitney: p = 0.094, n = 30). However, moisture at egg chamber was significantly higher than at the sand around the nest at the same depth (Mann-Whitney: p = 0.044, n = 30) during the first week after egg-laying. The apparent density of sand at the egg chambers consistently varied among sand beaches (Kruskal-Wallis: p = 0.006, df = 5); however, sand moisture levels did not (Kruskal-Wallis: p < 0.1).

Discussion

The Araguaia river basin has innumerous sand-belts, sandbanks, and sand beaches formed along its course by deposition of suspended materials carried by the water (Cantarelli and Alves 1999). Topography of the river borders and water speed determine the deposition pattern of large-grained sand particles along the beaches, whereas fine-grained particles are more easily carried away even by slow currents (Stancyk and Ross 1978). This is possibly the reason we did not find significant differences among beaches in terms of the fine-grained sand, but only in terms of the larger sand particles.

We found a relatively high rate of fine-medium sand grains, which is similar to the pattern described by Malvásio (2001) at the southernmost region of Araguaia River at Araguaia National Park. However, it differs from the pattern found in nesting beaches of the species at Trombetas, state of Pará, by Alho et al. (1979), where there is a higher rate of coarse grains, similarly to what has been described for marine turtles (Chelonia mydas) by Stancyk and Ross (1978). The percentage of very fine sand grains was the smallest component in this study as well as the other studies above.

Even though there was a significant variation among sand beaches in terms of apparent sand density of egg chambers, sand moisture levels did not significantly vary. There may be a significant variation among populations of the same species in terms of how temperature and humidity affect egg incubation (Bobyn and Brooks 1994; Packard et al. 1999).

The lack of significant variation in terms of sand apparent density between the egg chamber and the sand around for both rivers suggests that the egg chamber is not a less compacted “air-chamber.”

In laboratory experiments under controlled conditions, moisture affects chelonian hatching success (Packard et al. 1991, 1999; Bobyn and Brooks 1994). However, in this study this relationship was not significant. A possible explanation for this is that we only measured humidity at the initial period of incubation, not recording possible later variation. On the other hand, laboratory experiments normally do not consider physical variation suffered by nests under natural conditions related for instance to rainfall or sudden temperature drops (Wood and Bjorndal 2000). In the present study the egg chambers were significantly more humid than the substrate around the nest at the same depth. This is possibly due to the presence of mucus (or water) expelled by the female during egg-laying (Iverson 1990). The ecological implications of this are still unclear, but it is possibly related to egg protection against desiccation.

The site wherein the artificial nest is built can affect the incubation period of translocated clutches (Bonach et al. 2003). The present results suggest that the sand beach—and consequently the female capacity to choose it—can be more relevant to her reproductive success than her capacity to build an elaborate nest. Contrary to seashore sand beaches used by marine turtles to nest, sand beaches used by freshwater turtles are geologically unstable, possibly lasting shorter than one single turtle generation. This possibly prevents the occurrence of philopatry in this group as it more likely occurs in marine turtles (FitzSimmons et al. 1997).

The giant Amazon River turtle occurs over a vast geographic area. People have exploited it for its meat and eggs for centuries. Its conservation depends on the protection of its nesting sites. Future studies about the relationship between physical characteristics of the nest (i.e., sand texture, temperature, and soil moisture), and the reproductive success of the species can help accomplish this mission. Temperature sex-determination (TSD) patterns can also be influenced by the physical characteristics of the nest substrate and—besides reproductive success—should also be considered for future studies.