Effects of Human Settlements on Abundance of Podocnemis unifilis and P. expansa Turtles in Northeastern Bolivia
ABSTRACT
Podocnemis unifilis and P. expansa abundance was compared at 3 sites in eastern lowland Bolivia in 2000. Two of the sites were adjacent to human communities while the third was not. The comparison was made to examine effects of hunting pressure on turtle abundance. Data suggest that hunting pressure has a negative effect on the abundance of Podocnemis turtles and it is more severe closer to human communities. The greater abundance at a less-hunted site suggests that turtle populations facing reduced hunting pressure may be recovering.
The forests and rivers of Amazonia contain tremendous faunal and floral diversity, which includes many species that are important resources for humans (Wilson 1988; Redford and Padoch 1992; Terborgh 1992; Smith 1999). River turtles, including turtles in the genus Podocnemis, are important among these resources used by humans (Aramayo Cuenca 1989; Moll and Moll 2000). Numerous surveys demonstrate the importance of wildlife as food resources utilized by humans (Santos Brito and Ferreira 1978; Robinson and Redford 1991; Fa et al. 1995; Edwards and Abivardi 1998; Funes and Novaro 1999; Bennett 2002). Specifically, Santos Brito and Ferreira (1978) quantified Amazonian wildlife consumption in Manaus in 1976 and 7 additional cities in 1977. Their studies identified species preferred in restaurants, with both P. unifilis and P. expansa mentioned by participants as desirable items. This information quantifies some of the pressure faced by animals in the wild.
In lowland eastern Bolivia, near Noel Kempff Mercado National Park (Park), the river turtles P. unifilis and P. expansa are among highly desirable wildlife species in local diets. Sale of bushmeat, which was outlawed in Brazil in 1967 (Law 5,197 3 January 1967) and is restricted within and near the Park, continues today (Johns 1987; K. Conway-Gómez, pers. obs., 1999–2001). This translates into greater than subsistence hunting pressures. Decades of turtle harvest in this area have led to reduced turtle populations, according to local residents, scientists, and Park managers. Based on these observations, protecting the turtle populations is a priority of the Park management (Ministerio de Desarollo Sostenible y Medio Ambiente 1996).
Assessment of turtle abundance has implications for the future of these wildlife resources and the human communities that depend on them. In areas where wildlife is harvested, monitoring harvest impacts is critical. In eastern lowland Bolivia, impact monitoring is necessary to determine the effects of harvest on abundance. This information can help determine whether hunting pressure poses a threat to turtle populations, and it is an important precursor to conservation and management programs. Abundance counts of P. unifilis and P. expansa were collected at 3 sites to compare hunted and less-hunted sites (Fig. 1). Two hypotheses were tested to examine potential relationships between humans and turtle abundance: turtles will be less abundant closer to human communities and there will be more turtles at Mangabalito (less hunting pressure) than at Remanso (more hunting pressure).
Citation: Chelonian Conservation and Biology 6, 2; 10.2744/1071-8443(2007)6[199:EOHSOA]2.0.CO;2
METHODS
Three 40-km transects were established along 2 rivers in lowland eastern Bolivia—2 on the Iténez River and 1 on the Paraguá River. Two of these transects (Piso Firme and Remanso) straddled human communities while the third transect (Mangabalito) was adjacent to an abandoned human community. Piso Firme and Remanso are located in the buffer zone of the Park. Piso Firme (−61.735, −13.62861) lies on the Paraguá River while Remanso (−61.87056, −13.52111) lies on the Itenéz/Guaporé River, ca. 30 km northwest of Piso Firme. Mangabalito (−60.5525, −13.77917), also on the Itenéz River, lies in the Park ca. 150 km upriver from Remanso. The central docks in Piso Firme and Remanso served as the middle points in the 40-km stretches of river used to collect data.
Piso Firme and Remanso were selected because 1) both are on rivers within the ranges of P. unifilis and P. expansa; 2) P. expansa is presently found in the Itenéz River and fishers report that the species was formerly abundant in the Paraguá River; 3) I determined that residents in these communities eat turtle meat; and 4) the leaders of each community were willing to allow me to conduct my research in their communities.
Mangabalito, currently with no village within 40 km, was selected as the less-hunted site in the Park. When the Park was expanded to its present extent in 1996, the 12 families living at Mangabalito abandoned the site (Juan de Dios Peña, pers. comm., 2000). While the community of Mangabalito was smaller than that of either Piso Firme (70 families) or Remanso (97 families), the fact that people lived there 4 years prior to data collection posed the potential for observing the response of turtles to reduced hunting pressure.
Survey Procedure
Turtles were counted using a basking-observation technique. This technique is noninvasive, can yield large samples, and produces less variability in data compared to hand and hoop net capture techniques (Koper and Brooks 1998). Among these 3 techniques, results from basking surveys provide more accurate population estimates although with more variability (from underestimation by 32% to overestimation by 13%) in relation to true population size (Koper and Brooks 1998). Basking surveys also result in lower heterogeneity of capture for the sexes than hand capture, but more accurate estimates of female populations than male, suggesting that basking surveys may offer a relatively unbiased form of estimating population size and obtaining large samples over a short time (Koper and Brooks 1998). The basking technique is limited by curves in the river and angles of basking logs that prevent all turtles from being spotted before they disperse as the boat approaches. I conducted all surveys and alternated the start side of the river on each run, so any bias should be consistent and the abundance counts for the 3 areas comparable to each other (as suggested in Caughley 1977).
Transects were set up along 40 km of river length traveling down the middle of the river. The 40-km length was used because fishermen indicated they would travel more than 10 km for fishing/hunting trips of more than 1 day and 40 km was the maximum physical distance that could be covered in the 4.5-hour period during which surveys were conducted. This distance permits a within-site comparison of areas that have different hunting pressure. The window of survey time (1100–1530 hours) was delimited by the peak hours of sunshine, a significant factor influencing turtle basking (Koper and Brooks 1998).
Basking surveys were conducted during the dry season, between August and October 2000. During the dry season there are more basking spots available along river edges because the river is low and turtles are seen along the main river channel. During the wet season turtles more commonly frequent bays and flooded forests adjacent to rivers (Soini and Cóppula 1995) and are rarely seen along the main river channels.
The 2-way surveys were conducted from an aluminum johnboat with a 25-horsepower outboard motor, traveling at an average speed of 18–24 km/h. An assistant drove the boat ca. 5 m from the riverbank. Waiting periods of at least 30 minutes were observed a few kilometers beyond the endpoint of each survey to minimize disturbance in the count on the opposite side of the river. Turtles were observed basking in areas we had passed on the first run 30 minutes earlier, so this was presumed a sufficient waiting period. The side of the river and direction of travel were alternated between surveys. All basking turtles and those seen in the water were counted. For each turtle observed the 1-km segment in which it was seen was recorded.
I divided transects into 2 distance categories in relation to the communities—far and close. “Far” segments were those more than 10 km upstream or downstream from the center point while “close” segments were those within 10 km of the transect center (Fig. 2). This distance represents a cut-off point for 1-day fishing trips according to local fishers and was the main criterion in making this delimitation. In an Amazon Basin-wide study of hunting pressure, Peres and Lake (2003) found a 9-km threshold for hunting activity around villages, which further suggests this approximate distance as a critical division point.
Citation: Chelonian Conservation and Biology 6, 2; 10.2744/1071-8443(2007)6[199:EOHSOA]2.0.CO;2
I conducted 30 surveys in the period from 14 August to 3 October 2000. The sum of these surveys provided 8 complete counts of turtles in 40-km transects at each site. It was not always possible to survey the full 40-km transect on each river within each assigned survey day due to weather, logistic constraints, equipment failure, or assistant availability. Final survey coverage provided effectively equal effort in all sections over the course of the study. Counts of turtles observed in each kilometer section were summed over 10-km intervals, and the resultant turtle counts were compared among the different rivers and between the close and far distance categories.
RESULTS
Survey results indicated a disparity in numbers of turtles observed along the 40-km transects, with fewer turtles seen at the 2 sites where there are human communities, Piso Firme and Remanso (Table 1, Fig. 3). There was a clear decline in the number of turtles around the community-based center points of the 2 human-inhabited transects.
Citation: Chelonian Conservation and Biology 6, 2; 10.2744/1071-8443(2007)6[199:EOHSOA]2.0.CO;2
Numbers of Turtles
I compared the mean number of turtles observed in 10-km segments between and within sites. Turtle counts in the 10-km segments used for analyses are shown in Figs. 4 and 5. Descriptive statistics and results of the analyses are found in Tables 2–4 below. The results in Table 2 indicate that the larger median number of turtles seen in far distance categories was not due to chance.
Citation: Chelonian Conservation and Biology 6, 2; 10.2744/1071-8443(2007)6[199:EOHSOA]2.0.CO;2
Citation: Chelonian Conservation and Biology 6, 2; 10.2744/1071-8443(2007)6[199:EOHSOA]2.0.CO;2
Comparison of medians within sites indicated that the differences seen in close vs. far segments were significant at Remanso only (Table 3). The median values indicated that more turtles were observed at distances beyond 10 km from the center. On a site-by-site basis, the number of turtles seen close to human settlement was significantly smaller at Remanso only, suggesting that hunting pressure may be stronger at Remanso than Piso Firme. Mangabalito, which has no constant human presence, had more turtles overall and did not exhibit a pattern like Remanso. This may be indicative of a more robust turtle population.
The results of the median frequency test showed patterns at the 3 sites. At Piso Firme only 1 median count was above the median value for far segments while 7 were below it and 3 of 8 sample counts were greater than the median calculated value for close segments (Table 4). These values indicated that in 12 of 16 cases the number of turtles observed was less than expected, suggesting a small population. From the graph of turtles counted at Piso Firme it is evident that there were more turtles downstream than upstream (Fig. 3). This is possibly a combined result of hunting pressure and ecological factors since the Paraguá River is considerably smaller than the Itenéz River.
The observations were similar at Remanso, where the number of turtles counted in segments far from the community was above the median value in 3 of the 5 samples but never in the close segments. Again, this relationship is illustrated in the graph of turtles counted, where there is a clear dip in the numbers around the center of the transect (Fig. 3).
In stark contrast, the mean number of turtles observed at Mangabalito was above the median at all distances for all samples. The graphs of numbers of turtles observed (Fig. 4) show that more turtles were observed downstream at this site. Clearly, turtle abundance at Mangabalito was greater than at Piso Firme or Remanso.
Finally, to observe differences between hunted and less-hunted sites, Mann-Whitney tests were used (Tables 5 and 6). It is worth noting again that Mangabalito and Remanso are on the same river, the Itenéz, which makes a stronger comparison between these 2 sites and the basis for testing the second research hypothesis.
The results indicate that the difference in mean numbers of turtles counted at Piso Firme was less than that at Mangabalito and the difference was not due to chance alone. The difference between Remanso and Mangabalito also indicates that the difference in mean numbers of turtles counted at Remanso was less than at Mangabalito. Hypothesis 2 states there will be more turtles at Mangabalito (suggesting less hunting pressure) than at Remanso (suggesting more hunting pressure), and based on the results of the Mann-Whitney test of difference, this was the case.
River Edge Habitat Characteristics
Similarities and differences among the 3 sites based on a habitat type classification provide additional comparative information between sites that may be an important additional consideration in factors leading to differences seen among the sites (Fig. 5). There were more steep barren cliffs, which ranged from 1 to 5 m high, at Remanso and Mangabalito than Piso Firme. This is likely due to the higher order of the Iténez River as well as geological differences. These physical characteristics may cause a concentration of turtles in the main river channel, which could contribute to the larger number of turtles seen in the Itenéz vs. the Paraguá River. A related characteristic is river depth. In some parts of the Paraguá River transect, the bottom at center was less than 1 m, while on the Itenéz River transects, the bottom was rarely to never less than 1 m deep. This characteristic may also contribute to the Itenéz River being more suitable habitat for the turtles and could partially explain the greater number of turtles seen on this river.
There were more steep cliffs with aquatic vegetation (that constitutes some of the turtles' diets [Bruno Coca 1999]) and branches (important basking platforms) at Mangabalito than either Remanso or Piso Firme. The presence of more foraging opportunities and basking platforms may partially explain the greater abundance at Mangabalito.
Additionally, there were more beaches at Remanso and Mangabalito than at Piso Firme. This may reflect more nesting habitat, particularly for P. expansa, which nests exclusively on sandy beaches (Vanzolini 1967), making it more suitable for the larger species.
In summary, there were subtle differences in river edge habitat between the sites. However, the habitat differences observed were not interpreted as providing major differences between sites in terms of adequate turtle habitat. The proportions of each vegetation type were relatively similar across transects. This lends further strength to the suggestion that the main factor leading to difference in abundance between the sites was human presence and hunting pressure.
DISCUSSION
The abundance data presented here suggest that human activities, including predation, may play roles in reduction of turtle numbers. For the first hypothesis (turtles will be less abundant closer to human communities) there was statistically significant evidence to indicate differences in turtle abundance closer to humans vs. farther away from humans, with fewer turtles counted closer to humans at the hunted sites. For the second hypothesis, there will be more turtles at Mangabalito (less hunting pressure) than at Remanso (more hunting pressure), there was also evidence for differences being due to proximity to humans at Remanso.
If hunting is the main reason for the distance-abundance relationships, my data suggest that current practices of turtle hunting are working to contribute to the depression of turtle abundance. The fewer turtles seen close to the communities at Piso Firme and Remanso may be a result of hunting pressure. The lower turtle counts on the Paraguá River, around Piso Firme, suggest that the effect of humans on turtle abundance may be greater at this site. This may be a result of the synergy between a smaller river and hunting pressure (the greater hunting pressure leading to a greater decimation of a smaller population). The contrast of the 2 human-inhabited sites with uninhabited Mangabalito, which has greater numbers of turtles at all distances in 100% of the samples, suggests even more clearly that the cause of the difference is human activity. A comparison of the 2 sites on the Itenéz River, Remanso and Mangabalito, provides further support for the hypothesis that hunting pressure is the main difference between these sites that leads to fewer turtles close to humans because the confounding factor of different rivers is not present.
Redford and Robinson (1985) showed that depletion of game species within the catchment area of villages led to decline of game. Peres and Lake (2003) quantified a correlation between proximity to humans and reduced prey abundance in the Amazon Basin, with 9 km being the threshold beyond which hunting did not negatively affect prey abundance. The 10-km threshold from humans used here was based on information from fishers that they do not typically travel more than that distance on 1-day fishing trips. Thus, my results concur with Peres and Lake's (2003) large-scale work and the conclusion of hunting pressure within a catchment area leading to degradation of natural resources closer to humans, as suggested by Redford and Robinson (1985). Like Bodmer et al. (1997) my results show that human activities, which may include hunting pressure, can negatively influence game species—in this case river turtles of the genus Podocnemis.
My results show that there are fewer turtles close to human communities. The lesser abundance of turtles may be the result of hunting pressure. The greater turtle abundance near recently abandoned Mangabalito compared to 2 currently hunted sites may demonstrate turtle population recovery from hunting pressure. If there has been a recovery of the population at Mangabalito in the 4-year absence of humans, the reduced hunting pressure at this site may be sufficient for turtle recovery. The implication that recovery is possible is exciting for conservation efforts.
Map of study site within Bolivia.
Schematic diagram of the 40-km transects illustrating location of Piso Firme and Remanso communities at the center and categorization of distance categories.
Number of turtles counted along the Paraguá River (Piso Firme) and the Itenéz River (Remanso and Mangabalito) shown in 10-km segments. Each line represents a complete survey along the 40-km transect.
Average number of turtles seen in each kilometer over 8 samples. Top: at populated sites (Piso Firme and Remanso). Bottom: at the unpopulated site (Mangabalito). One standard deviation is indicated by bars on each kilometer. Note the different y-axis scales used for the populated vs. unpopulated sites.
Quantity (in m) of each habitat type at the 3 sites. A. Piso Firme, B. Remanso, C. Mangabalito.
