Effects of the Hydrological Cycle and Human Settlements on the Population Status of Podocnemis unifilis (Testudines: Podocnemididae) in the Xingu River, Brazil
Abstract
Podocnemis unifilis is found throughout the Amazon basin, where it has been harvested as a highly valued source of animal protein since precolonial times. Harvesting rates are also lower during the flooding season due to the availability of habitat for turtle dispersal. This study investigated the effects of the hydrological cycle and human impacts on the abundance and population structure of the species on the middle Xingu River between September 2007 and March 2008. The highest density of basking turtles was recorded during the flooding period, while the number of specimens captured in hand nets was greater during low water. Density and capture rates varied positively with increasing distance from Altamira, the study area's main urban center. Consequently, the abundance of P. unifilis was affected negatively by proximity to urban centers. In the Great Bend area of the Xingu River, densities and abundance were lower, and the mean size of the animals was smaller. These results likely reflect the effects on local stocks of the presence of local gold-prospecting operations and, thus, more harvesting pressure on the P. unifilis population. The sex ratio was biased in favor of males (1.90∶1), possibly as a result of preferential harvesting of females, which are larger in size than males and are often laden with eggs. Further long-term studies are needed to better understand the impact of anthropogenic pressures on long-lived organisms such as turtles. In addition, such information would aid the ecological sustainability of turtles in the Amazon Basin that provide rural communities with subsistence resource.
The podocnemidids, in particular Podocnemis unifilis, are found throughout the Amazon basin (Pritchard and Trebbau 1984). These turtles were used by local indigenous populations as a source of protein long before the arrival of Europeans and were widely traded during the colonial period (Pezzuti et al. 2010), which subsequently led to the gradual exhaustion of populations, in particular those of Podocnemis expansa (Bates 1892; Smith 1979). This trade use eventually led to exploitation of smaller-bodied species, such as P. unifilis and Podocnemis sextuberculata, in particular adult females (Smith 1979; Rebêlo and Pezzuti 2000).
The differential harvesting of females can result in alterations of sex ratio and a decline in the abundance and size of adults in a population (Fachín-Terán et al. 2003). As a result of such exploitation, P. unifilis has been classified as vulnerable by the International Union for Conservation of Nature (IUCN 2010), although it has not been included in the official Brazilian listing of endangered animal species. Studies in the Amazon Basin (Conway-Gómez 2007; Félix-Silva 2009) have found a clear decline in the abundance of P. unifilis in areas adjacent to human settlements, and a similar pattern has been recorded in populations of Trachemys scripta, in the United States (Dreslik and Kuhns 2000).
Another factor that plays a fundamentally important role in the life cycle of the fauna of the Amazon Basin is the Flood/Ebb dynamic of the rivers in floodplain areas (Sioli 1991). The distribution, density, and abundance of P. unifilis are influenced by the regional hydrological cycle (Junk and Mello 1990) that modifies the availability of available habitat and the distribution of feeding resource and breeding sites (Bury 1979).
In this study, that focused on the middle Xingu River, in the Brazilian state of Pará, the following hypotheses were tested: 1) the density and abundance of P. unifilis vary significantly according to the different phases of the hydrological cycle, more specifically, in relation to river height levels; 2) anthropogenic pressures cause significant variations in the density and abundance of P. unifilis between the 2 sampling zones; 3) decreasing anthropogenic pressure is reflected in increasing density, abundance, and body size of P. unifilis with increasing distance from the town of Altamira (the main local market); and 4) the preferential harvesting of females has significantly altered the sex ratio of the P. unifilis populations in the study area.
METHODS
Study Area
Fieldwork was conducted in the municipality of Altamira, in the Brazilian state of Pará. The study focused on the middle Xingu River (Fig. 1) between its confluence with the Iriri River (lat 03°49′12.78″S, long 52°36′22.89″W) and the stretch of the Xingu known locally as the Great Bend (lat 03°22′00.7″S, long 51°44′18.1″W). The town of Altamira is located on the left bank of the Xingu central to this area (lat 03°12′41.13″S, long 52°12′43.75″W) and is the region's principal urban center, economic hub, and market.
Citation: Chelonian Conservation and Biology 12, 1; 10.2744/CCB-0954.1
For this study, 2 zones were defined based on their different patterns of human occupation and density and degree of anthropogenic impact. Zone 1 is located between Altamira and the mouth of the Iriri River, a distance of approximately 100 km, and represents a relatively well-preserved stretch of the river, with sparsely populated margins and reduced pressure from local fishermen. Zone 2 is located between Altamira and the Great Bend, close to the mouth of the Bacajá River, at a distance of approximately 100 km. This area is subject to much greater anthropogenic pressure, with 2 riverside communities and extensive gold prospecting operations. Both zones present similar conditions, with abundant rock formations and a series of rapids that are exposed during the driest part of the year.
Procedures
The study was conducted during 3 distinct phases of the hydrological cycle: low water (September 2007), flooding period (December 2007), and high water (March 2008). The collection of data thus encompassed 2 extremes of river level and discharge and the availability of aquatic habitats on its floodplain. Two sampling methods were used: visual counts of basking turtles and trapping.
Visual Counts
Podocnemis unifilis requires an external source of warmth for thermoregulation, and animals can often be observed basking along the margins of rivers (Lacher et al. 1986; Moll and Moll 2004). Counting the number of animals engaged in this activity during the hottest hours of the day, between 1000 and 1500 hrs, is a reliable form of active sampling that has been used to estimate the density and abundance of populations of P. unifilis at a number of sites in the Amazon Basin (Conway-Gómez 2007; Félix-Silva et al. 2008) as well as for surveys of other turtles species in Europe (Segurado and Figueiredo 2007) and North America (Dreslik and Kuhns 2000; Peterman and Ryan 2009).
Twenty-four independent 5-km transects were selected randomly along the river in both zones during each phase of the hydrological cycle with a total of 72 transects separated by more than 10 km (Fig. 1). Different stretches of the river were surveyed during the different phases in order to avoid pseudoreplication in sampling. The transects were surveyed in an aluminum boat with a 15-horsepower outboard motor, cruising at an average speed of 7–15 km/hr perpendicular to the riverbank at a distance of 5–10 m from the margin. Basking animals were observed and identified using a pair of 8 × 40 Nikon binoculars. The exact location of each individual or group of turtles was recorded using a handheld GPS receptor (Garmin model 60c).
The density of P. unifilis was defined as the number of animals observed per kilometer surveyed. Possible differences in the density of the species among the different phases of the hydrological cycle were tested using 1-way analysis of variance (ANOVA) followed by the Tukey post hoc test, when significant differences were found. Student t-test was used to test possible differences in the density of animals between zones 1 and 2. A simple linear regression was used to test the hypothesis that the density of P. unifilis increased with increasing distance from Altamira, that is, as a function of decreasing anthropogenic pressure. The density values were log10 transformed prior to all analyses.
Trapping and Morphometry
Data on the structure of the P. unifilis population within the study area was obtained by capturing the animals (Rebêlo et al. 2005) using a hand net that consisted of a wooden shaft attached to a metal ring supporting a bag-shaped net with a small mesh. This trapping method has been developed and perfected by the local river dwellers of the middle Xingu River (Pezzuti et al. 2008). In comparison with other trapping techniques (Fachín-Téran et al. 2003; Fachín-Téran and Vogt 2004; Rebêlo et al. 2005), this procedure is not selective, and turtles of all sizes and both sexes are captured. The same experimental design used for the boat surveys was adopted, with the trapping being conducted at sites known locally as “boiadouros” (literally, “floating areas”), which are areas of relatively deep water where turtles tend to congregate during low water (Fig. 1).
Each adult animal captured was weighed (± 1.0 g) and measured (± 1 mm): straight carapace length (SCL) and curved carapace length, head width, and carapace height. The specimens were marked with a triangular incision in the center of a marginal scute, following a preestablished coding system. The sex was identified through the analysis of secondary sexual characteristics, such as the size and color of the head, width of carapace, and the length and thickness of the tail (Pritchard and Trebbau 1984). Abundance was estimated based on capture per unit effort, defined as the number of individuals captured per hour of netting (Rebêlo et al. 2005).
The data distribution could not be normalized; thus, the differences in abundance among the phases of the hydrological cycle were examined using a Kruskal-Wallis test (H). We employed the a posteriori Dunn test for statistically significant results. Mann-Whitney tests (U) were used to evaluate possible differences in abundance between the 2 zones. The possible relationship between abundance and the distance from Altamira was examined using Spearman's rank correlation coefficient (rs).
A simple linear regression model was used to determine whether the distribution of body size (SCL) varied as a function of the distance from Altamira in both zones. Between-sex differences in total length of the carapace were evaluated using Mann-Whitney U-test, which was also used to evaluate possible differences between males and females in the 2 zones. The Spearman rank coefficient (rs) was used to compare the relationship of the sex ratio with the distance from Altamira in the 2 zones. For the variables that presented a normal distribution, the variation in the sex ratio between the 2 zones was tested using a the Student t-test. Similarly, a 1-way ANOVA with the Tukey post hoc test was used to test whether the sex ratio varied according to the phase of the hydrological cycle. All analyses were run in BioEstat 5.0 (Ayres et al. 2007).
RESULTS
Visual Counts
A total of 689 P. unifilis were observed basking during the visual surveys, of which 241 (35.0%) were female and 81 (11.8%) male, while the sex of the remaining 366 (53.2%) was undetermined during the visual counts of basking animals. Given that the sex was unknown for such a large proportion of the individuals, the sex ratio of this sample was not statistically examined. The density of P. unifilis did not vary significantly between the 2 zones (t = 0.275, n = 109, df = 107.000, p = 0.784), while the mean density of P. unifilis was 0.32 ± 0.23 individuals per kilometer (range: 0.00–0.79 individuals/km) in zone 1 and 0.34 ± 0.25 individuals/km (range: 0.00–0.86 individuals/km) for zone 2.
By contrast, the hydrological cycle had a significant effect on the density of P. unifilis based on these counts. The highest densities were recorded during the flooding period, with 0.40 ± 0.25 individuals/km (range: 0.00–0.83 individuals/km), whereas the lowest densities were recorded during low water at 0.21 ± 0.27 individuals/km (range: 0.00–0.86 individuals/km; F2,106 = 5.633, n = 109, p = 0.005; Fig. 2). The Tukey post hoc test indicated that abundance was significantly higher during the flooding period in comparison with the other phases (p < 0.01). Densities also increased significantly with increasing distance from Altamira in zone 1 (r2 = 0.24, F1,70 = 23.4, p < 0.0001; Fig. 3); to show this fluctuation, we used a polynomial trend line with the adjustment being indicated by the regression coefficient r2. A polynomial trend line could be useful when the relationship between variables is not linear and when data fluctuate. No significant pattern was found in zone 2 (r2 = 0.0068, F1,36 = 1.25, p = 0.2693).
Citation: Chelonian Conservation and Biology 12, 1; 10.2744/CCB-0954.1
Citation: Chelonian Conservation and Biology 12, 1; 10.2744/CCB-0954.1
Trapping and Morphometry
Of the total of 600 P. unifilis captured using hand nets, 194 (32.3%) were female and 368 (61.4%) male, while the sex of 38 (6.3%) specimens was not determined due to the lack of well-defined secondary sexual characteristics because they were juveniles. The capture rates were significantly different among the phases of the hydrological cycle (H = 82.446, n = 110, df = 2, p = 0.000), and the post hoc test (Dunn) indicated that it was significantly higher during low water in comparison with the other phases (p < 0.05; Table 1; Fig. 4).
Citation: Chelonian Conservation and Biology 12, 1; 10.2744/CCB-0954.1
The analysis of the effect of the distance from Altamira on abundance was conducted based the data only for the flooding period in order to standardize the sample between and within zones and because this period was the most adequate for the trapping technique employed. Abundance was significantly higher in zone 1 (25.2 ± 36.4 individuals captured per hour) compared with zone 2 (16.7 ± 29.4 individuals/hr; U = 178, p = 0.0109; Fig. 5). We also found a significant correlation between the abundance of P. unifilis and the distance from Altamira in zone 1 (rs = 0.5894, n = 31, p = 0.0005; Fig. 6), but no clear pattern was observed in zone 2 (rs = 0.0722, n = 21, p = 0.7558).
Citation: Chelonian Conservation and Biology 12, 1; 10.2744/CCB-0954.1
Citation: Chelonian Conservation and Biology 12, 1; 10.2744/CCB-0954.1
In zone 1, the mean carapace length (SCL) of P. unifilis varied positively with the distance from Altamira (r2 = 0.3781, F1,9 = 7.07, p = 0.025; Fig. 7). However, only 37.8% of the variation in carapace length is explained by the distance from the urban center. In zone 2, however, there was no clear pattern (r2 = −0.1088, F1,9 = 0.019, p = 0.88). Mean body length (SCL) in the P. unifilis females was significantly greater than in males (U = 22,669, p < 0.0001; Fig. 8). However, neither females (U = 3704, p = 0.5093) nor males (U = 11,548, p = 0.0853) were significantly different in size between zones.
Citation: Chelonian Conservation and Biology 12, 1; 10.2744/CCB-0954.1
Citation: Chelonian Conservation and Biology 12, 1; 10.2744/CCB-0954.1
The overall sex ratio, based on the captures was 1.90∶1 and therefore strongly biased toward males (Table 2). The sex ratio also varied significantly among the different phases of the hydrological cycle (F2,63 = 3.179, n = 66, p = 0.048; Table 3). The Tukey test indicated that males were significantly more abundant (p = 0.05) during the low-water period. However, there was no significant difference in the sex ratio between zones (t = −1.327, df = 61, p = 0.189), nor did the ratio varied according to the distance from Altamira in either zone 1 (rs = 0.1218, n = 31, p = 0.5139) or zone 2 (rs = 0.0359, n = 20, p = 0.8807).
DISCUSSION
Seasonal Variation in Density and Abundance
Like many aquatic animals, turtles are affected by hydrological cycles in river systems that typically involve fluctuations in the water level height that result in major variation in the availability of environments for feeding and reproduction (Bury 1979; Junk et al. 1989; Junk and Mello 1990). Given this, many turtles typically undertake substantial migrations in order to reproduce (Gibbons 1986; Fachín-Terán et al. 2006). When the waters ebb, the P. unifilis migrate from their feeding areas on the floodplain to the main river channel, where they build their nests in different seasonal habitats associated with the margins. In the middle Xingu Basin, the species is distributed throughout the whole length of the river, although its density and abundance fluctuate considerably according to the variation in the region's hydrological cycle.
The visual counts recorded the highest densities during the flooding period, whereas the abundance recorded during trapping was greater during the low-water period. This apparent contradiction is easily accounted for by the fact that the concentration of individuals in the boiadouros during low water facilitates capture. At high water, the turtles disperse over a wider area and are thus less easily captured.
The results of the present study were similar to those of podocnemidids at some other sites in the Amazon Basin. In the Mamirauá Sustainable Development Reserve, Fachín-Terán et al. (2003) collected larger quantities of Podocnemis spp. during the low-water period, and the same pattern was recorded on the Guaporé River by Fachín-Terán and Vogt (2004) and on the Iténez and Paraguá rivers in Bolivia by Conway-Gómez (2007). On the Tocantins and Tapajós rivers in the Brazilian state of Pará, by contrast, Félix-Silva et al. (2008) and Ullmann-Leite (2010) recorded higher densities of podocnemidids during the flooding period.
Ullmann-Leite (2010) found that the higher densities of Podocnemis spp. recorded at Lago Verde during the flooding period were related to increased foraging behavior given that the availability of fruit and seeds, the principal food of these turtles, is greatly reduced during the dry season (low water). The abundance of food at high water leads to an increase in both foraging behavior and the numbers of turtles basking for thermoregulation (Fachín-Terán et al. 2006). This is also a plausible explanation for the results obtained in the present study.
Local riverside residents also affirm that, during the dry season, the water is warmer due to the low level of the river. This may suggest that the turtles would not have to sun themselves for thermoregulation, for which they could prefer less exposed aquatic microhabitats. During other parts of the year, the water is cooler, forcing the turtles to leave the river more frequently. A similar behavior pattern has been recorded by Lacher et al. (1986) in captive P. expansa and Kinosternon scorpioides. Preferring aquatic environments for thermoregulation has obvious benefits in terms of avoiding potential predators, but it also has practical implications for population surveys, as found in this study.
Impact of Harvesting
The results of the present study support the hypotheses raised with regard to the effects of harvesting, that is, a decrease in the density and abundance of P. unifilis with increasing proximity to the study area's main urban center. This pattern was especially clear in zone 1, where a significant correlation was found, although in zone 2, the results suggest that the species could be being exploited intensively throughout the area as a source of food and illegal trade. These activities are widespread in the region (Pezzuti et al. 2008).
The lack of a clear spatial pattern in zone 2 (and the reduced densities of P. unifilis in comparison with zone 1) may also be related to the presence of riverside communities and intense illegal gold prospecting operations that impact the aquatic ecosystem through the disturbance of the riverbed and pollution from discarded mercury (Costa 2005; Switkes and Sevá 2005). The harvesting of turtles is not only a subsistence activity in this area but also a source of income. In particular, gold prospectors pay local fishermen the equivalent of US$18.00 per d to catch turtles (Pezzuti et al. 2008).
Studies in Brazil (Félix-Silva 2009), Bolivia (Conway-Gómez 2007), and the United States (Dreslik and Kuhns, 2000) found that hunting pressure has a negative effect on the abundance of turtles, an effect that was more intense with proximity to human settlements. In Africa, Luiselli (2003) also found that the relative abundance of Kinixys homeana and Kinixys erosa was greater in areas with no hunting pressure. The harvesting of native fauna is especially intense in the proximity of human settlements and can often lead to local extinctions (e.g., Peres 2000; Sirén et al. 2003; Mendonça 2009). In the present study, however, P. unifilis was clearly less abundant close to the urban center of Altamira, although there is no conclusive evidence that the species is declining significantly throughout the region. This is possibly because the effects of exploitation are diluted with increasing distance from the town. The results of the present study confirm a decline in P. unifilis numbers in the area of the Great Bend, and this is agreed on by residents of local communities, including members of the Juruna indigenous people (Vieira et al. 2009).
Population Structure
Anthropogenic impacts on the study population of P. unifilis are reflected in a greater proportion of smaller animals and fewer adults in the population, which may have serious implications for demographic processes such as the recruitment and survival of juveniles. This was also observed by Félix-Silva (2009) on the Tocantins River. Female-biased sexual dimorphism has also been recorded in P. sextuberculata (Fachín-Terán et al. 2003). Fachín-Terán and Vogt (2004) captured female and male P. unifilis with mean carapace lengths of 35.0 ± 7.7 and 26.4 ± 3.3 cm, respectively, while Félix-Silva (2009) recorded mean lengths of 23.7 ± 8.7 cm (females) and 17.7 ± 2.7 cm (males). Félix-Silva (2009) concluded that the smaller size of individuals in this population may be a consequence of the selective harvesting of larger turtles, a pattern also observed in this study.
In a population of P. sextuberculata, Fachín-Terán et al. (2000) found that all age classes had been affected by harvesting, whereas in P. unifilis, the larger adults were harvested preferentially. Studies of other vertebrates have shown a similar pattern of a reduction in the mean size of individuals resulting from intense hunting pressure (Peres 1990; Garcia 2006). On the Purus River, Mendonça (2009) observed that the reduced abundance of adults in the local population of Melanosuchus niger reflected intense local hunting pressure.
Sex Ratio
The male-biased sex ratio appears to be a consequence of the historical preferential harvesting of females, which are larger in size and thus more valuable commercially because they provide greater amount of meat and eggs. Similar results have been obtained in studies of podocnemidids in Venezuela (Ramo 1982) and Brazil, including the Crixás-Açú River (Bataus 1998), Mamirauá (Fachín-Terán et al. 2003), the Guaporé River (Fachín-Terán and Vogt 2004), and the Tocantins (Félix-Silva 2009). The overexploitation of females is especially intense during the nesting period. Large females, often laden with eggs, are preferred by rural populations throughout the Amazon Basin (Rebêlo and Pezzuti 2000; Vieira et al. 2009; Pezzuti et al. 2010).
However, Félix-Silva (2009) has stated that deviations in the sex ratio may also be a consequence of modifications in the characteristics of nesting sites or of recent climatic change. Pezzuti and Vogt (1999) report that 78% of the natural nests of P. sextuberculata monitored on the Japurá River in 1996 produced only male neonates due to the low incubation temperatures recorded during the annual nesting period. Two years later, 5 nests monitored at the same nesting site presented a sex ratio of 67% males (E.M. von Mulhen, unpubl. data, 1998). This indicates that the consequences on annual climatic variations and their influence in hatchling sex ratios should be continuously monitored, and care must be taken to interpret results from a single year. There are no studies regarding hatchling sex ratios in the Xingu River and also no evidence that our results could be a consequence of a bias in hatchling sex ratio production.
While the present study did not record a major overall decline in the region's P. unifilis population, the decreasing abundance in the proximity of settlements and the male-biased sex ratio may have implications in the future, considering that the recruitment of stocks is determined by the number of adult females present in the population. If female numbers continue to decline, recruitment may also decrease, and stocks may eventually be compromised. Moll and Moll (2004) considered the preferential harvesting of females to be the worst possible impact for turtle populations.
The threats to P. unifilis in the Xingu River may be intensified in the near future, especially considering the proposed construction of the Belo Monte hydroelectric power station on the lower Xingu River, which may have a major impact on the hydrological regime of the Great Bend. Without adequate monitoring and management policies to compensate for these impacts, local P. unifilis populations risk declining further in the future.
The study area on the Xingu River in the Brazilian state of Pará. Zone 1 is upriver from Altamira, as far as the mouth of the Iriri River; zone 2 is downriver, as far as the Great Bend. Highways are marked BR (federal) or PA (state). The black line and the circles are the schemes of transects and “boiadouros” that were selected randomly along the river. Location of study area within Brazil shown in inset map.
Density of sightings of basking Podocnemis unifilis along the middle Xingu River in Pará (individuals per kilometer surveyed) during different phases of the hydrological cycle between September 2007 and March 2008 on the middle Xingu River, Pará (F2,106 = 5.633, n = 109, p = 0.005).
Relationship between the density of Podocnemis unifilis (animals basking per kilometer surveyed) in zone 1 and the distance from Altamira in transects of 5 km during 3 phases of the hydrological cycle on the middle Xingu River in the state of Pará between September 2007 and March 2008 (r2 = 0.24, F1,70 = 23.4, p < 0.0001).
Number of Podocnemis unifilis specimens captured per hour during trapping in the 3 phases of the hydrological cycle between September 2007 and March 2008 on the middle Xingu River, Pará (H = 82.446, n = 110, df = 2, p = 0.0001).
Number of Podocnemis unifilis specimens captured per hour during trapping in zones 1 (Altamira-Iriri) and 2 (Altamira-Great Bend) between September 2007 and March 2008 on the middle Xingu River, Pará (U = 178, p = 0.0109).
Relationship between the number of Podocnemis unifilis specimens captured per hour in zone and the distance from Altamira during the flooding period between November 2007 and December 2007 on the middle Xingu River, Pará (rs = 0.5894, n = 31, p = 0.0005).
Relationship between body size (straight carapace length) of the Podocnemis unifilis specimens captured during trapping in zone 1 and the distance from the capture site to the town of Altamira during the flooding period between November 2007 and December 2007 on the middle Xingu River, Pará (r2 = 0.3781, F1,9 = 7.07, p = 0.025).
Distribution of body sizes (straight carapace length, SCL) by sex of 562 Podocnemis unifilis specimens (194 female, 368 male) captured on the middle Xingu River in the Brazilian state of Pará between September 2007 and November 2007 (U = 22,669, p < 0.0001).
