Podocnemis erythrocephala (Spix 1824) is classified by the International Union for Conservation of Nature (IUCN) as vulnerable (Hilton 2000). This is because of the pressure put on this species by local inhabitants who collect eggs and adult turtles for local consumption (Vogt 2001). Knowledge of the reproductive biology of a population is essential for understanding species population dynamics (Ramo 1982). Some initial studies of the reproductive biology of P. erythrocephala from other areas have been published (Mittermeier and Wilson 1974; Vanzolini 1977; Castaño-Mora 1997; Vogt et al. 2001; Castanõ-Mora et al. 2003).

Turtle nests are subject to predation by humans and other animals, as well as adverse climatic conditions, e.g., flash floods, which occur after unseasonable heavy rainfall during the dry season (Alho 1982; Fachín-Terán 1982; Soini 1995; Pezzuti and Vogt 1999). Because of these variables, the selection of a nest site has a direct influence on the survivorship of the eggs (Soini 1995).

Little is known about the impact of natural phenomena and egg collection on the survivorship of nests of Amazonian turtles. Soini (1995) studied nesting of Podocnemis expansa, Podocnemis unifilis, and Podocnemis sextuberculata for 10 years in the Pacaya-Samíria National Reserve in Peru, noting predation by ants, lizards, birds, and mammals. He found a variation of 1%–100% nest loss from flooding. Vogt et al. (1994) documented 60.5% mortality of Peltocephalus dumerilianus hatchlings because of fly larvae attacking recently hatched turtles before they left the nest; whereas, Pezzuti and Vogt (1999) noted that only 2% of the nests of P. sextuberculata were attacked by fly larvae.

Studies of nest survivorship and identification of principal nest predators are important to establish successful management strategies. It is also important to study other aspects of reproductive ecology, including spatial and temporal distribution of nests, influence of vegetation cover on nest-site selection, hatchling success, morphological characteristics of nests, and primary and adult sex ratios (Vogt and Bull 1982; Alho et al. 1984; Vogt 1994).

The objective of this study was to investigate P. erythrocephala nesting ecology in campinas (savanna patches of white sands within the Amazonian forest) in the Ayuanã River basin near Santa Isabel do Rio Negro, Amazonas State, Brazil. We studied 1) spatial and temporal distribution of nests, 2) influence of vegetation cover on nest-site selection, 3) nest survivorship, and 4) morphological characteristics of nests, eggs, and hatchlings.

METHODS

The study area is located in the municipality of Santa Isabel do Rio Negro (lat 00°24′82.8″S; long 65°01′17.3″W), 781 km upriver from Manaus, Amazonas State, Brazil. The mean annual temperature is 27°C, with an annual precipitation of 2800 mm, distributed unevenly over the year, with the heaviest rains occurring between May and July. The climate is considered humid tropical rainforest (Walter 1986). There are 2 annual well-defined seasons in the region: the dry season, from August to February, and the flood season is from March to July. The intensity and duration of each season varies from year to year (Moran 1990; Oliveira et al. 2001). The nesting sites of P. erythrocephala are normally located in campinas (Mittermeier and Wilson 1974). The campina is a low forest found on white sand in the Rio Negro Basin, the trees are separated such that they do not make a closed canopy. Many of the trees that occur in this habitat are endemic (Oliveira et al. 2001). The physiographic and floral composition is determined by the soil drainage parameters (Daly and Silveira 2002).

The study areas were 4 campinas near the guard post of the Ayuanã River (lat 00°31′55.6"S; long 064°55′11.5"W), a tributary on the right margin of the Rio Negro, 17 km downstream from Santa Isabel do Rio Negro. There are 2 rural communities near the study areas (São João and Monte Alegre), which normally use this river for subsistence hunting and fishing (Fig. 1). The nesting areas were monitored from 28 September to 20 December 2002. Each morning, the nesting areas were searched for nests that were deposited the night before. Nests were located by following the tracks of females or the presence of a disturbed area in the soil. Rains hindered the success of this method by obscuring the tracks and the soil disturbances. Each nest was marked with a numbered stake located 1 m west of the nest and the location registered with 3-m precision global positioning system (GPS) (Garmin 12). These data were used to map the distribution of nests and calculated the area used for nesting in each campina by using the program GPS Track Maker v. 11.8 (Ferreira 2003). Each campina was divided into 25-m² plots, each plot was classified according to vegetation cover: 0%–25%, 25%–50%, 50%–75%, and 75%–100%. We noted the nesting date, depth to first egg (nest chamber top), depth to ultimate egg (nest chamber bottom), clutch size, distance to and height of the nearest plant, and distance to the margin of the campina.

View Figure

• Download as Powerpoint
• Download Figure

All eggs were weighed with an electronic digital balance accurate to 0.1 g. The length and width were noted with digital calipers accurate to 0.1 mm. The volume of each egg was determined by the equation: V = πxy²/6 where x is the egg width, and y is the length (Vanzolini 1977).

When flood waters reached nests, the nests were transplanted to the beach at Itaubal, one of the highest points in the region (lat 00°32′06.4″ S, long 064°54′23.0″ W). For analysis, these nests were considered a total loss to flood waters. The hatchlings from the transplanted nests were measured: straight-line carapace length, width, and height, maximum plastron length, femoral scute length, head width, and mass to 0.1 g after they absorbed the yolk sac. After the termination of nesting, the sites were visited only for 1 hour every 3 days to minimize anthropomorphic effects on nest predation. Predations of nests were identified, and the predator was determined by tracks or tooth marks on the egg shells.

We compared the mean number of nests in the grid of vegetation cover with a χ² test. The relationship between nesting, water level, and the phases of the moon was calculated by using the Pearson correlation coefficient. The mean number of eggs per nest was correlated with the nesting date by a simple regression. We tested the relationship between mean egg mass and mean hatchling mass with a simple regression. We tested the relationship between the mean egg volume and mean nest volume with a simple regression. The mean egg volumes per nest were compared among the campinas and between the nesting periods with analysis of variance (ANOVA). The depth to the first egg, distance from the nest to vegetation, height of the nearest plant, distance to the margin of the campina, and number of days after nesting that predation occurred were correlated with a logistic regression, with the 2 principal nest predators. The dependent variable of predator was categorized as 0 (no predations) or 1 (nest predation). All analyses were done by using Systat 9.0 (Wilkinson 1990).

RESULTS

The first nest of P. erythrocephala was found on 2 October in the campina Itaubal, which was the first nesting area to appear as the water level receded. The last site to receive nesting was Armando II, on 9 October. There were 2 nesting periods separated by flash floods (Fig. 2). The first was from 2 to 13 October when 67 nests were located. The second was from 26 November to 7 December when 50 nests were recorded. Most of the nesting occurred during the dark phases of the moon, (46% and 48% of all of the nests). Only 5% occurred in the crescent moon and just 0.9% during the full moon. There was no statistical relationship between nesting and moon phases (r = 0.2571, p = 0.0144), but there was a positive correlation between nesting and water level (r = 0.8347, p = 0.0059).

View Figure

• Download as Powerpoint
• Download Figure

During the monitoring period, we found 117 P. erythrocephala nests distributed in the 4 campinas (5.6 ha), with a mean density of 19.9 nests/ha. The lowest nest density was found in Armando II, 11.1 nests/ha, and the highest in Itaubal, 29.2 nests/ha (Table 1). In addition to the 117 nests found in the campinas, we found 2 nests on the beach near Itaubal; they are included in the analysis. There were small differences among the 4 campinas in relationship to the percentage of vegetation cover and distance to the river. The campina Itaubal is the largest in area with only a few trees, with a maximum height of 3 m, it has the lowest percentage of vegetation cover and is situated about 150 m from the Rio Negro. The campina Macaco Apu has a higher percentage of vegetation, has trees reaching up to 5 m, and is about 50 m from the Ayuanã River. The other 2 campinas have denser vegetation: campina Armando I is the smallest, with most of the area covered with grasses and trees up to 5-m tall, located about 10 m from Ayuanã River; campina Armando II has less grass density and was the only site located within the flooded forest, with a small sandy beach at its confluence with the Ayuanã River.

Table 1. Nest density of Podocnemis erythrocephala in 4 campinas in the Ayuanã River, Santa Isabel do Rio Negro, Amazonas State, Brazil in the 2002 reproductive season.

View Fullscreen

The nest distribution maps show the greatest density in the open areas of the campinas (Figs. 3 and 4) (χ² = 76.35915; p < 0.001; Table 2), which showed a significant preference for nesting sites with less than 25% vegetation cover.

View Figure

• Download as Powerpoint
• Download Figure

View Figure

• Download as Powerpoint
• Download Figure

Table 2. Number of nests of Podocnemis erythrocephala observed and expected in each category of vegetation cover in the 4 campinas monitored.^a

View Fullscreen

The nest of P. erythrocephala consists of an elliptical flask-shaped cavity with one of the walls more concave than the others. The opening is circular and shorter than the interior pocket, with a diameter of 7 ± 1.2 cm (range 4.5–10 cm; n = 119). The depth to the first egg is 5.1 ± 2 cm (range 0.3–10 cm), and the nest chamber depth is 10.8 ± 2 cm (range 7–22.5 cm). The clutch size was 8.7 ± 2.1 eggs (range 2–16). The mean distance from nests to plants was 74.3 ± 86.2 cm (range 5– 400 cm).

The values obtained for the egg and hatchling morphmetrics can be found in Table 3. There was no variation in the clutch size during the 23 days nesting period (R² = 0.863, p < 0.001, F = 24.23, n = 88; Fig. 5A). Heavier eggs produced heavier hatchlings (R² = 0.641, p < 0.001, F = 20.26, n = 55; Fig. 5B). Eggs with the largest volume were found in the clutches with the largest volume (R² = 0.402, p < 0.001, F = 16.621, n = 88; Fig. 5C).

Table 3. Morphometrics of eggs and hatchlings of Podocnemis erythrocephala in the 4 campinas monitored.

View Fullscreen

View Figure

• Download as Powerpoint
• Download Figure

The mean volume of each egg per nest did not vary significantly among the 4 sites (1-way ANOVA: F_0,886 = 0.546), the highest mean volume was in campina Armando II (15.3 cm³) the lowest in campina do Armando I (14.1 cm³). There was no significant difference in the mean volume of the eggs per nest between the nesting periods, 15 cm³ (1-way ANOVA: F_0,126 = 0.724).

Nest loss in 2002 at these 4 sites on the Ayuanã River was 100%. The principal cause of mortality was flooding (Fig. 6). Two individual predators, tayra weasels, Eira barbara (Mustelidae), were the major nest predators that consumed the eggs in 14 nests: 10 in campina Macaco Apu and 4 in campina Armando I. Nests were taken by E. barbara only within the first 10 days after laying ( p = 0.002; Table 4). Six nests were collected by local fishermen, 3 on the first day of the study in campina Itaubal by people from Santa Isabel do Rio Negro and another 3 in campina Armando II. Two nests were taken by the lizard Crocodilurus lacertinus (Teiidae) in campina Itaubal. There was predation by the falcon Daptrius ater (Falconidae) of 1 nest in campina Armando II. There was predation by the lizard Ameiva ameiva (Teiidae) of nests that were within 10 m of the campina margin and less than 15 days old. Logistic regression indicated that 2 variables were significantly correlated with nest predation: distance to the campina margin and number of days after nesting p = 0.04 and p = 0.003, respectively, Table 5).

View Figure

• Download as Powerpoint
• Download Figure

Table 4. Logistic regression of the independent variables.^a

View Fullscreen

Table 5. Logistic regression of the independent variables.^a

View Fullscreen

DISCUSSION

The nesting period of P. erythrocephala at the mouth of the Ayuanã River began in October and terminated in December (Fig. 2). Mittermeier and Wilson (1974) mentioned that the nesting period in the Rio Negro extended from August to the end of November. Rebêlo (1991) reported that nesting extended to December in the region of Barcelos (Amazonas State, Brazil). In Colombia, nesting occurs from November and January (Castaño-Mora 1997; Castaño-Mora et al. 2003). Vanzolini (1977) commented that P. erythrocephala nested until the first flooding occurred, normally until November. There is obviously local and annual variation in nesting in different areas along the Rio Negro, depending on water levels; thus, the variance in the reporting of the nesting period for this species (Vogt 2001).

The beginning and termination of nesting by P. erythrocephala in the 4 campinas studied was influenced by the variation in water level. Podocnemis unifilis and P. expansa differ by nesting only when the water level has reached its lowest level. Vogt (2001) found P. erythrocephala nesting in campinas up to 200 m from the Itu River. The local people believe that nesting coincides with the new and last quarter moons to avoid predators. The data collected do not confirm this popular belief (r = 0.2571, p = 0.0144). However, the results of this work indicate that the water level influences the nesting (r = 0.8347, p = 0.0059) (Fig. 2). Soini (1995) noted that weather characteristics, not the moon's phases, affected P. expansa nesting.

We observed small 1-day peaks in nesting in 2 of the 4 campinas monitored: in Itaubal, we found 9 of the 29 nests on 11 October, and, in Macaco Apu, we found 9 of the 19 nests on 8 October. The highest densities of nests were found in these 2 areas. These occurred during the first nesting period and were not repeated in the second period. Vanzolini (1977) commented that P. erythrocephala nests solitarily during the entire nesting season. It is necessary to repeat this study in other nesting areas to confirm whether the peaks in nesting activity represent aggregations of turtles synchronized to nest at the same time by water levels and moon intensity.

A density of 19.9 nests/ha (Table 1) of P. erythrocephala could be considered to be at more natural levels compared with 3.47 nests/ha found for P. sextuberculata in a nesting beach in the Reserva de Desenvolvimento Sustentável Mamirauá where human predation pressure has been much higher (Pezzuti and Vogt 1999).

In the 4 nesting areas P. erythrocephala nested in areas with up to 50% vegetation cover, and the greatest densities of nests were found in areas with less than 25% cover (Figs. 3 and 4; Table 2). Schwarzkopf and Brooks (1987) observed Chrysemys picta selecting warmer areas for nesting to lower egg mortality by decreasing incubation time. Janzen (1994) found that females of this same species selected nest sites based on vegetation cover; these results coincide with the tendency we found for P. erythrocephala. The nest measurements and characteristics are similar to those described for Podocnemis vogli (Ramo 1982), perhaps because the body sizes of these 2 species are similar.

Vogt (2001) found a variation of 4 to 18 eggs per nest in P. erythrocephala, in the region of Barcelos, Amazonas State, Brazil. Mittermeier and Wilson (1974) reported a range of 5 to 14 eggs from data collected in the Rio Negro. Castaño-Mora (1997) found 5–12 eggs per nest in Colombia. Castaño-Mora et al. (2003) found 5–10 eggs. Our data (2–16 eggs) are within the range of variation reported for this species. Positive relationships between clutch size, egg volume, and carapace size have been reported for many turtles species (Wilbur and Morin 1988; Vanzolini 1977; Ramo 1982; Bernhard 2001). We did not find any difference in the egg volume among the 4 sites studied nor during different portions of the nesting period (Fig. 5A). These results indicate that females of all sizes are nesting in the 4 campinas during the entire nesting season. Eggs of greater mass produced larger hatchlings (Fig. 5B), and eggs with greater volume were found in clutches with the largest volume (Fig. 5C). The mean size of P. erythrocephala hatchlings is the lowest reported for the genus (Pritchard and Trebbau 1984). Most studies of natural turtle populations find high nest mortality from both weather and predation (Wilbur and Morin 1988). Flooding was the principal cause of mortality in our 4 sites. High nest mortality because of flooding has been reported for many other areas and other species as well: P. expansa (Roze 1964; Hildebrand et al. 1988; Soini 1995), P. unifilis (Mitchell and Quiñones 1994), P. sextuberculata (Pezzuti and Vogt 1999), and Trionyx muticus (Plummer 1976). Soini (1995) found variation in nest loss from flooding to vary extensively year to year from 1% to 100% in a long-term study in Reserva Pacaya-Samiria, Peru, and in Colombia. Hildebrand et al. (1988) also found high annual variation in nest loss from flooding. Long-term studies are needed to establish the level of variation that occurs in nest loss from flooding in the Ayuanã River.

The local inhabitants along the Rio Negro consume great numbers of P. erytrocephala eggs (Vogt 2001). In the region of Santa Isabel do Rio Negro, it is common for entire families to visit the campinas during the nesting period to collect turtle eggs, they are efficient in locating nearly 100% of fresh nests. The same situation has been reported for other species in the Amazon Basin as well: Caqueta River, Colombia (Hildebrand et al. 1988), Manu Biosphere Reserve, Peru (Mitchell and Quiñones 1994), Pacaya-Samíria Reserve, Peru (Soini 1995), and Reserva de Desenvolvimento Sustentável Mamirauá, Brazil (Pezzuti and Vogt 1999). There is an illegal trade of eggs between people who live near the nesting beaches and those who live farther away and in the city of Santa Isabel. In the city, the eggs are sold for about US$ 0.70 per liter (about 30 eggs).

Besides humans, E. barbara was the principal nest predator documented during the study, responsible for destroying 12% of the total number of nests. The signs left by E. barbara at the nest site were very characteristic; in addition to their tracks, they removed all of the eggs from the nest to the nearest tree, ± 3.5 m (Table 4), where the contents were eaten and the shells discarded.

Predation occurred by the diurnal lizard Ameiva ameiva of 9% of the nests (Table 5). Moll and Legler (1971) reported that this lizard was one of the most important predators of turtle nests in the neotropics. The lizards consumed eggs of nests that were up to 10 days old and preferred nests that were near the border of the campina. Escalona and Fa (1998) found higher levels of predation of P. unifilis nests within 10 m of the vegetation around the border of the beaches studied in Venezuela. Fachín-Terán (1982) and Soini (1994) commented that nests near the vegetation had high incidences of predation by animals that foraged in the vegetation. Nests in open areas were shown to be much more vulnerable to human predation (Escalona and Fa 1998).

RESUMO

Foram estudados alguns parâmetros da ecologia reprodutiva e fatores que influenciam a sobrevivência de ovos de Podocnemis erythrocephala (1824 Spix) em quatro campinas da foz do rio Ayuanã, Município de Santa Isabel do Rio Negro, Estado do Amazonas, Brasil (S 00° 31′55.6"/ W 064° 55′11.5"). Entre outubro e dezembro de 2002 as quatro campinas foram visitadas diariamente durante o período de desova, cada ninho encontrado foi marcado com uma estaca numerada e georreferenciado com o auxílio de um GPS. Foram anotadas as seguintes informações: data da desova, largura e profundidade da câmara até o ovo da base, número de ovos, comprimento, largura e peso de cada ovo, a distância e altura da vegetação mais próxima e a distância do ninho até a margem da campina. Ocorreram dois períodos de desova, um no início de outubro e outro no início de dezembro, separados pelo alagamento repentino do rio. Ao todo foram registrados 117 ninhos, com densidade de 19,9 ninhos/ha, com uma média de 9 ovos por ninho. O volume médio por ovo foi de 14 cm³. As fêmeas de P. erythrocephala desovaram preferencialmente nas áreas mais abertas das campinas com até 50% de cobertura vegetal. Após o período de postura as campinas foram visitadas a cada três dias para detecção de possíveis ninhos predados. A perda de ninhos foi de 100%, sendo 70% por repiquete, 12% por predação por Eira barbara (Mustelidae), que atacou preferencialmente os ninhos mais novos, com até 10 dias de postura, 9% por predação por Ameiva ameiva (Teiidae), que normalmente destruiu os ninhos mais novos, com até 10 dias, e situados a até 10 m da margem das campinas. Humanos destruíram 5% dos ninhos, Crocodilurus lacertinus (Teiidae), predaram 2%, Daptrius ater (Falconidae), e 1% não tiveram o predador identificado.