The endangered yellow-spotted river turtle (Podocnemis unifilis), a member of the family Podocnemididae, is distributed in the Amazon and Orinoco river basins where it is disappearing because of human persecution and habitat destruction (Rueda-Almonacid et al. 2007; Turtle Taxonomy Working Group 2011; Convention on International Trade in Endangered Species of Wild Fauna and Flora [CITES] 2017). Beyond its ecological role in predation and scavenging, it has religious, cultural, and economic significance for local communities within its range. Although in Ecuador this species is culturally important to the Kichwa and Waorani indigenous groups, it is among the most threatened aquatic fauna because of water pollution caused by oil industry activities and human settlement expansion, increased boat traffic along major waterways, and wildlife trafficking.

Despite its conservation status and socioecological importance, current conservation efforts for the yellow-spotted river turtle are hindered by the lack of information on its autecology. The natural history of the yellow-spotted river turtle has been partially described, with most contributions focusing on reproductive biology, genetic structure, and egg harvesting (Vogt 2008; Escalona et al. 2012; Miorando et al. 2015; Trebbau and Pritchard 2016). Spatial ecology remains almost completely undocumented; home range estimates for species of the genus Podocnemis are not available, and before the present study, there were no robust home range estimates for the yellow-spotted river turtle. Available information on linear ranges of the species was obtained from low sample sizes (Bock et al. 1998; Padovani et al. 2015).

Spatial ecology of yellow-spotted river turtles is crucial when ensuring long-term conservation success (Sung et al. 2015; Walston et al. 2015). Understanding where or even whether yellow-spotted river turtles congregate, how far they travel, and how much space they use will enable conservationists to understand where to focus conservation efforts. We describe the space use of yellow-spotted river turtles in the Napo River, in the northern section of Yasuní National Park, Ecuadorian Amazon. We used VHF radiotelemetry to estimate 1) linear and home range sizes, 2) home range overlap, and 3) site fidelity.

METHODS

Study Area

The study area is located within the Orellana and Sucumbíos provinces in eastern Ecuador. We conducted fieldwork in a 60-km section of the Napo River, between Yasuní National Park and Limoncocha Biological Reserve (Fig. 1). Annual mean temperature and precipitation in the area are 25°C and 2,786 mm, respectively (Ministerio del Ambiente 2013). The Napo River flows from the eastern slopes of the Andes in Ecuador, descending eastward to the Peruvian border, a few kilometers downstream of the town of Nuevo Rocafuerte (0°56′N, 75°24′W). The Napo is a whitewater, meandering river with a large floodplain. The Napo has a mean width of 1000 m and is up to 7 m deep; the annual water level fluctuation is approximately 8 m (Puhakka et al. 1992). Its tributaries, mainly blackwater creeks, range from 20 to 30 m in width. Currently, the Napo is the main aquatic transportation artery in eastern Ecuador, providing access to the largest oil concessions in the Ecuadorian Amazon.

View Figure

• Download as Powerpoint
• Download Figure

Data Collection

We performed fieldwork 2 wks per month from August 2015 to February 2017. We captured 63 yellow-spotted river turtles (10 males, 38 females, and 15 of undetermined sex) by hand, either by snorkeling for them in the river or when females crawled up sandbanks during the nesting season. Male and female turtles with straight-line carapace lengths ≥ 18 cm and ≥ 27 cm, respectively, were considered sexually mature (Foote 1978; Escalona et al. 2012). Captured turtles were measured (straight-line carapace length and body mass) and fitted with a radio-tracking transmitter (TXF-317S, Telenax, Mexico) and released at their capture locations. Transmitters were affixed using Rally Epoxy (Disma C Ltda). Turtles were tracked daily by boat between 0700 and 1800 hrs using a handheld receiver (RX-RLNX, Telenax, Mexico) and a 3-element Yagi antenna. Turtles were located as often as possible, and each location was georeferenced with a handheld GPS unit (Garmin Oregon 600t).

Data Analyses

We could not determine the minimum number of fixes and survey effort to estimate home ranges based on the variograms of each individual; therefore, we included in the analyses all the turtles that had values greater than the median for tracking periods and number of fixes. For each turtle, we estimated linear range size, 95% and 50% autocorrelated kernel density estimate (AKDE) home range, and home range overlap (Fachín-Terán et al. 2006; Ghaffari et al. 2014). Turtles with ≥ 20 locations were used to estimate AKDE, and we measured home range overlap using Bhattacharyya's affinity index (Fieberg and Kochanny 2005). All the analyses were carried out using ArcGIS 10.1 (Esri 2011) and the R package ctmm (Calabrese et al. 2016). All measurements and estimations were done by measuring river and tributary courses and removing the land (islands) within them.

Differences between the sexes in linear range size and home range size were examined using Wilcoxon rank-sum tests (McDonald 2014). We explored the relationship of body mass, linear range size, and home ranges using Spearman rank correlation (McDonald 2014). We evaluated turtles' site fidelity by comparing observed home ranges (OHR) and simulated home ranges (SHR; Fujisaki et al. 2014; Ghaffari et al. 2014). We used correlated random walk models to create SHR. Data were generated using the function movement.simplecrw in the Geospatial Modelling Environment (Beyer 2012). Movement distances and turning angles were based on normal and uniform distributions, respectively. We compared OHR and SHR using Wilcoxon signed-rank test and Bhattacharyya's affinity index. Turtles were deemed to exhibit site fidelity if OHRs were significantly smaller than SHRs, along with a high overlap (Spencer et al. 1990). Statistical significance was set at α = 0.05 level for all tests.

RESULTS

During 16 mo of continuous tracking, we recorded 1350 locations from 63 yellow-spotted river turtles. Tracking period ranged from 1 to 16 mo (x̄ = 8, SD = 5, median = 9), with 1−70 fixes per turtle (x̄ = 21, SD = 18, median = 17). Thirty-one turtles (49%; 6 adult males, 23 adult females, and 2 immatures females) were located more than 17 times (x̄ = 37 times) and were included in the analysis.

Mean linear range size was 16.2 km (n = 31, SD = 11.1, range = 2−50.1). Home range sizes averaged 5.2 km² (n = 27, 95% CI = 3.8−7.0) and 1.0 km² (95% CI = 0.5−1.5) for 95% and 50% AKDEs, respectively (Table 1). Sixteen home ranges had one activity center, 8 had two, and 3 had three. Home range overlap was low; the mean value of Bhattacharyya's affinity index for 351 pairs, of 27 individuals, was 0.32 (95% CI = 0.29−0.36).

Table 1. Radio-tracked individuals of yellow-spotted river turtle (Podocnemis unifilis), from August 2015 to February 2017, in Amazonian Ecuador. AKDE = autocorrelated kernel density estimate for home range (km²). Numbers in parentheses are 95% confidence intervals.

View Fullscreen

There were no differences in linear range sizes (n = 6, U = 15, p = 0.70; Fig. 2) or home ranges between the sexes (95% AKDE: n = 6, U = 17, p = 1.00; 50% AKDE: n = 6, U = 21, p = 0.70; Fig. 2). We did not find correlations of linear range size with body mass (n = 31, r_s = 0.170, p = 0.40) or of home range sizes with body mass (95% AKDE: n = 27, r_s = 0.16, p = 0.34; 50% AKDE: n = 27, r_s = 0.113, p = 0.57).

View Figure

• Download as Powerpoint
• Download Figure

Yellow-spotted river turtles exhibited site fidelity in our study area (Fig. 3), as evidenced by the fact that OHRs were significantly smaller than SHRs (95% AKDE: medians = 2.3 and 5.1, respectively, W = 94, p = 0.021; 50% AKDE: medians = 0.4 and 1.2, respectively, W = 35, p < 0.0001). Bhattacharyya's affinity index for OHRs and SHRs averaged 0.73 (95% CI = 0.68−0.78), describing a high overlap.

View Figure

• Download as Powerpoint
• Download Figure

DISCUSSION

Home range estimates for species of the genus Podocnemis are not available, and before this study, there were no robust home range estimates for the yellow-spotted river turtle. Available information on linear ranges of this species was obtained from low sample sizes (Bock et al. 1998; Padovani et al. 2015). The species is considered sedentary because of its short linear range of up to 10 km (Bock et al. 1998; Gallego 2012; Padovani et al. 2015). Although some of our observations (35%) fall within this range, our data suggest that P. unifilis could be treated as a short-distance facultative migrant with local seasonal movements (Dingle and Drake 2007). This characterization is based on the evidence that 18% of the turtles in our analyses lived along tributaries and only moved to the Napo River during the nesting season and that 41% of home ranges had more than one center of activity.

Mean home range size of the yellow-spotted river turtle was greater than the mean home ranges of 26 species of freshwater turtles (0.32 km²; Slavenko et al. 2016). According to the authors, there is much intraspecific and intrapopulation variation in home range sizes in freshwater turtles and our home range estimates were consistent with these findings. Differences in home range size within individuals of the same population are associated with habitat quality (larger home ranges being an indicator of poor habitat quality), habitat availability, and environmental productivity (Plummer et al. 1997; Galois et al. 2002; Slavenko et al. 2016). We did not measure any of these variables, but we hypothesize that home range sizes of the yellow-spotted river turtle in this study might have been correlated with habitat availability and quality, or an interaction between them.

Habitat availability is greater in the main (Napo) river than in tributaries (Añangu and Indillama rivers), which may be the reason that smaller home ranges were observed in tributaries. Regarding habitat quality, 77% of turtles with large home ranges (> 1 km²) were observed within the Napo River, which is highly affected by outboard motorboat traffic, whereas smaller home ranges were found in tributaries, where outboard motor boat traffic does not occur. The negative impact of motorboat activities and transit on freshwater organisms can be summarized as pollution and impoverishment of habitat quality (e.g., washing, erosion, turbulence, turbidity) and direct disturbance and traumatic injuries (Liddle and Scorgie 1980). In the case of freshwater turtles, there is evidence that boating increases physiological stress and decreases basking behavior with negative consequences on demography (Bulté et al. 2010; Selman et al. 2013).

Extrinsic factors affecting home range size, such as those mentioned above, also have effects on home range overlap. Overlap is an indicator of species territoriality (Powell 2000) or habitats offering sufficient resources (Makowski et al. 2006). The overall home range overlap was low for our study species; because of the social behavior of Podocnemis turtles, we discard the hypothesis of high territoriality when a low degree of overlap is observed (Powell 2000). Yellow-spotted river turtles' home range overlap was low in the Napo River (x̄ = 0.32) and high in tributaries (x̄ = 0.61), a finding consistent with the hypothesis that habitat quality influences home range size.

Home range in reptiles can differ between the sexes (Vitt and Caldwell 1999). For instance, Fachín-Terán et al. (2006) and Doody et al. (2002) found that linear range size was larger in females than in males of six-tubercled Amazon River turtles (Podocnemis sextuberculata) and pig-nosed turtles (Carettochelys insculpta), which was explained in both cases as a reproductive strategy (nesting excursions; Morreale et al. 1984; Gibbons et al. 1990). We did not find differences in linear and home range sizes between the sexes. These results could be a consequence of a low sample size, a species-specific trait (Gibbons et al. 1990), or sex ratio; however, we do not have quantitative data to support the last of these hypotheses. We attribute the differences between the sexes we observed to intrapopulation variability in turtles' home range sizes, as suggested by Slavenko et al. (2016).

We found no significant correlations of either linear range length or home range size with body mass in the yellow-spotted river turtle. Significant positive correlation of home range size with body size has been found to be a pattern in aquatic, semiaquatic, and terrestrial turtles (Slavenko et al. 2016) and has been related to energetic requirements (Reiss 1988). Nonetheless, this hypothesis was postulated for terrestrial organisms living in a 2-dimensional space, whereas aquatic organisms live in a 3-dimensional space. Freshwater turtles would benefit from water flow to manage energetic expenditure (Sung et al. 2015). Recently, Slavenko et al. (2016) suggested that home range size in turtles is driven by habitat characteristics rather than animal traits.

The migration distances of reptiles range from a few hundred meters to thousands of kilometers, and in the case of freshwater turtles, migration is related to environmental conditions and the availability of suitable nesting sites (Russell et al. 2005). Migration patterns still remain unknown for most Neotropical freshwater turtle species. The members of Podocnemididae are considered to have large dispersal capacity based on the available information from the giant South American river turtle (Podocnemis expansa) and the six-tubercled Amazon River turtle (Fachín-Terán et al. 2006; Gallego 2012). In addition, migratory behavior has been reported for these species, as well as for the yellow-spotted river turtle (Fachín-Terán et al. 2006; Vogt 2008; Escalona et al. 2009; Ferrara et al. 2014).

Our results suggest nest-site fidelity because 50% of females we radio tracked returned periodically to the nesting beaches. This result was expected because nesting site fidelity and genetic structure within populations have been documented for P. unifilis in other countries (Vogt 2008; Escalona et al. 2009). The existence of site fidelity does not exclude the potential migration of this species; many migratory species have site fidelity at different moments within their annual cycles (Iverson and Esler 2006). Site fidelity is an indicator of local knowledge of resource distribution and stability (Iverson and Esler 2006; Marmet et al. 2009) and should be taken into account for population management and conservation.

Moving forward, the conservation of the yellow-spotted river turtle needs to take into account site fidelity and migratory behavior of the species. Escalona et al. (2009) emphasized that in the presence of site fidelity, populations of the species need to be treated as demographically independent units, and egg translocation among rivers should be avoided because of potential genetic contamination. Although our study area is located between two natural protected areas, most home ranges and nesting beaches of the yellow-spotted river turtle are located along the Napo River and are not protected. A series of conservation strategies are needed for yellow-spotted river turtle protection. Habitat conservation should be prioritized to avoid displacing turtles, regulations for navigation in the Napo River need to be improved to control high-speed outboard motor boats that cause shore and nesting beach erosion, and dredging to facilitate navigation must be controlled to protect nesting beaches. Finally, we propose the inclusion of the yellow-spotted river turtle in the Convention on the Conservation of Migratory Species of Wild Animals to improve and strengthen conservation efforts for the species.