The drastic concentration process of human populations in large cities started with the beginnings of industrialization, and it accelerated during the second half of the previous century. At different times, this process has caused drastic changes in natural landscapes across the world. Urbanization and establishment of industrial enclaves involve partial or total habitat destruction, introduction of exotic species, pollution, and direct persecution of certain species. In recent years, important works have shown the effects of urbanization on herpetological communities (Andrews and Gibbons 2008; Ner and Burke 2008; Windmiller et al. 2008; Hamer et al. 2018). The undesirable effects of urbanization have led to a decrease of populations and local extinction of some turtle species, depending on their respective requirements and adaptability (Eskew et al. 2010a; Stokeld et al. 2014).

Variables such as abundance, sex ratio, size class distribution, and reproductive success are key factors to evaluate how human settlements and industrial enclaves affect freshwater turtle populations (Bujes et al. 2011; Bernardes et al. 2014). These parameters can be negatively affected due to reduction of nesting sites, higher mortality rates due to collisions on roads, and increased predator abundance, among other factors, that particularly affect immature individuals and adult females (Marchand and Litvaitis 2004; Gibbs and Steen 2005; Riedle et al. 2017; Bowne et al. 2018).

The species studied in the present work is the South American snake-necked turtle, Hydromedusa tectifera. It is widely distributed through Uruguay, eastern Paraguay, southeastern Brazil, and northeastern and central Argentina (Cabrera 1998; Prado et al. 2012). Although it is commonly found across its distribution, so far only 1 study has been published with population parameters of this species (Lescano et al. 2008). In that study, the population of H. tectifera inhabited unpolluted sections of the Toro Muerto stream (Lescano et al. 2008). Here, we studied a population of H. tectifera in a highly anthropogenic environment in Argentina.

Our main objective was to study how the abundance, sex ratio, body condition, and size class distribution of H. tectifera vary across 3 sections of an anthropogenic stream, Rodriguez stream, with different degrees of urbanization along its margins. Additionally, we compared our population parameters from an anthropogenic stream to a population inhabiting an unpolluted stream reported in Lescano et al. (2008).

METHODS

Study Area. — Fieldwork was conducted in a stream located in the suburbs of La Plata city (Rodriguez stream, Buenos Aires province, Argentina). Hydromedusa tectifera is the most common freshwater turtle in the headwaters and middle course of this stream (Fig. 1). The other turtle species present in the stream is Phrynops hilarii, but it occurs mainly near the mouth of the Rio de la Plata River and rarely within the headwaters. Indeed, in the entire time that we worked in Rodriguez stream, we detected only 1 specimen of the latter species, in the midsection.

View Figure

• Download as Powerpoint
• Download Figure

We chose 3 stream sections with different degrees of urbanization along the margins (considering a 10-ha rectangular area): site 1 (S1), upstream, rural zone without housing on the margins; site 2 (S2), midsection, sparsely populated (urbanization density intermediate between S1 and S3: 3 houses/ha); and site 3 (S3), midsection downstream from S2, higher urbanization density with 5 houses/ha (Fig. 2). S1 and S2 are separated by 4.5 km of stream course (not in a straight line) and S2 and S3 by 4.2 km. These sections are also characterized by different numbers of streets and intensity of traffic in the neighboring 10 ha surrounding each site: S1 is traversed by a small foot path (not enabled for vehicles); S2 is traversed by a grid of streets comprising 320 m of well-transited, paved street and 450 m of low-transited, unpaved street; and S3 is crossed by 700 m of paved street. Motor vehicle traffic was measured during 5 min between 800 and 900 hrs (moment of intense traffic) at the most transited streets within S2 and S3 (S1 lacks streets): 13 vehicles/min were recorded for S2 and 22 vehicles/min for S3. Turtles run into by cars (we verified 5 cases in a 3-yr period, 1 at S2 and 4 at S3), predation by dogs, and barriers such as rhomboid wire-mesh perimeter fences and walls are indisputably undesirable effects of urbanization that hinder the movements of females during the search for appropriate nesting sites.

View Figure

• Download as Powerpoint
• Download Figure

The 3 stream sections studied present vegetated and nonurbanized margins that constitute available nesting areas for H. tectifera. Reports indicate that in Brazilian populations of this species, eggs are laid in low slope areas and very close to water bodies (Fagundes and Bager 2007; Bager and Rosado 2010); however, the population studied here does not necessarily share these nesting requirements. In any case, there is a 45-ha military area slightly downstream from S3 in which the environment remains almost pristine. This area provides a larger zone for the terrestrial reproductive incursions made by females.

Sampling Methods. — Between March 2017 and January 2018, we visited the stream monthly or bimonthly, according to the temperature: 3 times in autumn, 1 time in winter, 2 times in spring, and 1 time in summer, totaling 7 workdays. Turtles were caught by hand between 1000 and 1600 hrs while they were resting on the bottom of the stream. We also checked for turtle presence in diverse possible shelters, such as under rocks and in marginal aquatic vegetation, garbage accumulations, trunks, and hollows on the streambanks. We collected turtles using the following procedure: both stream margins were divided into 5 transects by site (each transect was 30 m long with a 15-m separation between consecutive transects), and each transect was inspected simultaneously by 2 people (1 at each stream margin). For each turtle caught, the following variables were recorded: 1) site, transect, and margin; 2) sex; and 3) weight and straight carapace length (SCL: straight distance between anterior margin of nuchal scute and posterior margin of supracaudals). Turtles were sexed according to sexually dimorphic characteristics (Cabrera 1998) and, in the case of smaller specimens with no clear plastron concavity, checking for presence of a penis within the cloaca following Rodrigues et al. (2014). Finally, turtles were marked individually according to Cagle (1939) before being released at the point of capture.

Air and water temperatures were measured using a digital thermometer (Silcook, accuracy ± 1°C). We recorded water temperature at 10-cm depth in the middle of the stream course and air temperature at the same point but 1 m above the water surface.

Data Analysis. — Differences in turtle abundance among the 3 sampling sites were evaluated by conducting 1-way analysis of variance (ANOVA) with the number of turtles as the dependent variable and stream site (S1, S2, and S3) as the factor. The transects were the sampling units. Data were fourth-root transformed to approach the assumptions of the test. Recaptured turtles were not included in the analysis to avoid data pseudoreplication.

A chi-square (χ²) test (Zar 1999) was used to analyze contingency tables of association of sampling site with size class and season. In order to facilitate comparisons with the previous work by Lescano et al. (2008), we established 5 size classes (SC) based on SCL: SC I = 30–100 mm; SC II = 100.1–150 mm; SC III = 150.1–200 mm; SC IV = 200.1–250 mm; and SC V = > 250 mm.

We also used a χ² test to evaluate whether the male: female sex ratio at each sampling site diverged from the 1: 1 expected value. We calculated the body condition index using the formula proposed by Bjorndal et al. (2000) to allow statistical comparison of this index between sexes and among sites using 2 1-way ANOVAs.

All analyses were performed using the free trial of software Xlstat Version 2018.1 (Addinsoft SARL, Paris, France); a p-value was considered significant when < 0.05.

RESULTS

We found 109 individuals of H. tectifera (mean ± SD = 15.71 ± 3.68 individuals/survey; n = 7): 56 males, 46 females, and 7 hatchling turtles for which sex could not be determined. A total of 14 turtles were recaptured during fieldwork, all of which were caught at the same stream section where they had been first captured. Females (9) were recaptured more frequently than males (5). The mean male: female sex ratio was 1: 22, with the ratio being slightly male biased at S3 and S2 (1: 56 and 1: 18, respectively), while it was female biased at S1 (0: 86), but the differences were not statistically significant among sites (χ² = 0.096, df = 2, p = 0.953).

The number of males and females captured did not show significant differences among seasons (χ² = 0.880, df = 3, p = 0.83). Table 1 details the number of males and females caught at each site for each season.

Table 1. Number of males and females caught at each sampling site by season.

View Fullscreen

Most turtles were captured at S3, the most urbanized site we studied. The relative abundance at that sector was 0.42, while at the other sites it was lower: 0.26 for the nonurbanized S1 section (rural zone) and 0.32 for the intermediate urbanized S2 section. However, the ANOVA did not find significant differences in the number of turtles among sites (F_2,12 = 0.861, p = 0.447) nor any significant variation in the number of turtles caught at the different sampling sites between seasons (χ² = 12.074, df = 6, p = 0.06). Turtle size ranged from 53 to 258.5 mm (mean ± SD = 180 ± 44.37 mm), and weight ranged from 21 to 1956 g (mean ± SD = 739.77 ± 417.32 g). At S2, there was predominance of 150–200 mm (SC III) individuals, whereas at sites S1 and S3, 150–250 mm (SC III–IV) was the most frequent size class; turtles of SC V were found only at S3 (Fig. 3). Both the smallest and the largest sizes were recorded at S3. However, comparisons among the 5 size classes showed no significant variation across sampling sites (χ² = 11.021, df = 8, p = 0.2).

View Figure

• Download as Powerpoint
• Download Figure

Chi-square showed that class sizes varied significantly with sex (χ² = 32.356, df = 3, p < 0.0001). For females, SCL of 52.17% of individuals ranged between 150 and 200 mm, whereas for males, SCL of 57.14% of individuals ranged between 200 and 250 mm (Fig. 4). The body condition index (see mean values in Table 2) varied significantly between sexes (F_1,103 = 50.434, p < 0.0001) but not among stream sites (F_2,102 = 0.708, p = 0.495). The body condition index of females was higher than that of males (Table 2).

View Figure

• Download as Powerpoint
• Download Figure

Table 2. Body condition index (BCI) calculated for males and females at each sampling site and mean values from overall stream.

View Fullscreen

DISCUSSION

The present study is the first to characterize a population of H. tectifera in an anthropogenic stream. There is evidence about how different degrees of urbanization impact freshwater turtle populations, with most works dealing with North American turtles (Aresco and Gunzburger 2004; Bury 2008; Mitchell and Brown 2008; Eskew et al. 2010a, 2010b; Patrick and Gibbs 2010). Considering the diversity of Neotropical turtles, few works have focused on the population ecology of South American freshwater species (Souza and Abe 1997, 2001; Fachín-Terán et al. 2003; Fachín-Terán and Vogt 2004; Brito et al. 2009; Martins and Souza 2009; Fagundes et al. 2010; Bujes et al. 2011; Famelli et al. 2011; Bernhard and Vogt 2012; Marques et al. 2013; Bernardes et al. 2014; Moura et al. 2015; Rodrigues and Silva 2015), and most works have investigated turtle populations in rural and conserved areas.

McKinney (2002, 2008) reviewed the response patterns of numerous plant and animal species (including reptiles) in an urban-to-rural gradient and concluded that species present different reactions to urbanization. Blair (2001) proposed 3 categories according to the species' response to urban modified environments: urban avoiders, adapters, and exploiters. Our study suggests that H. tectifera is not an urban avoider, and this species could be characterized as an urban adapter species or even as an urban exploiter in future studies. Recent publications have demonstrated that H. tectifera is capable of surviving in polluted waters. For example, this species is tolerant of altered landscapes, including agricultural and silvicultural areas, pasture mosaics, and urban- and industrial-impacted streams (Molina et al. 2016; Semeñiuk et al. 2017). Our study demonstrates that moderate urbanization occurring along the Rodriguez stream appears not to have negatively impacted the population of H. tectifera.

Published evidence from several sources indicates that population parameters (e.g., average population body size and sex ratio) may change in response to particular features and environmental pressures (Ashton 2001; Reznick and Ghalambor 2001; Caruso et al. 2015). Thus, the search for “normal” population parameters is a fruitless effort. Instead, a productive approach is the search for environmental contexts that better explain the parameters we observed under different scenarios. In this context, several studies indicate that turtle populations respond neutrally or somewhat positively to suburban or urban development (Eskew et al. 2010a; Roe et al. 2011). According to Stokeld et al. (2014), certain species are resilient to habitat degradation and fragmentation resulting from urbanization. In this sense, these authors argue that particular urban factors may subsidize food and water requirements, increase availability of nesting sites, and decrease presence of predators, among others features that favor occurrence of turtles (Souza and Abe 2000; McKinney 2002; Stokeld et al. 2014; Riedle et al. 2017). In addition, Rees et al. (2009) and Roe et al. (2011) have suggested that suburban water bodies represent higher-quality habitats than those within nature reserves. This difference in habitat quality is due to the fact that during droughts, the water levels in natural areas drop substantially, whereas suburban waters fluctuate relatively little and remain mostly flooded, providing refuge to the turtles during these environmental fluctuations. Similarly, the water level of our study area never dropped in summer, while in other unpolluted streams in the region (outside the city, e.g., headwaters of Tubichamini stream), the water level drops in summer. This point was highlighted by Ferronato et al. (2017), who showed the strong influence of climatic conditions on turtle population dynamics and resilience and how these effects are moderated by urbanization factors. We suggest that urbanization, which causes water permanence in the midsection of Rodriguez stream, mitigates strong climatic conditions and influences population structure and dynamics at our study sites.

The present work revealed a balanced sex ratio, as was also reported for H. tectifera inhabiting unpolluted streams: Toro Muerto (Córdoba province; Lescano et al. 2008), Zapata, and Tubichamini (Buenos Aires province; Semeñiuk et al. 2017). The sex ratio of H. tectifera was highly male biased (7: 1) in the polluted Buñirigo stream (Buenos Aires province, Argentina). Semeñiuk et al. (2017) suggested that pollution in Buñirigo stream presumably affects females more than males. Sex ratio can be influenced by a variety factors, including trapping method (Ream and Ream 1966; Browne and Hecnar 2005; Bury et al. 2012), environmental causes (Smith and Iverson 2002; Gibbs and Steen 2005), or incubation temperatures (Ewert and Nelson 1991), among other causes. However, factors derived from urbanization (e.g., water pollution, predation by dogs, and turtles hit by cars) do not appear to influence the sex ratio of the H. tectifera population from Rodriguez stream.

Regarding the distribution of size classes, the population of H. tectifera in Rodriguez stream is composed mainly of medium-sized to large adult turtles (SC III–IV), as previously reported for other populations of H. tectifera and turtle species (Souza and Abe 2001; Lescano et al. 2008; Bujes et al. 2011; Famelli et al. 2011; Rodrigues and Silva 2015). The poorly represented SC I (mostly hatchlings) at our sites is more likely related to trapping bias than high mortality. Although the egg and hatchling age classes are the most vulnerable to predation and experience high mortality in turtles (Congdon et al. 1993, 1994), recruitment appears to be good given that the SC II turtles are well represented at all of our sites. Therefore, the best explanation for the low number of hatchlings at our sites is sampling bias. Our hand-capture method, like many other turtle trapping methods, was not ideal for capturing hatchlings likely due to their small size and camouflage on the stream bottom. On the other hand, as stated in the “Methods” section, suitable nesting areas are available at the 3 stream sites. This fact, together with occurrence of the smallest turtles (SC I) at S1 and S2 and of small turtles of the next size class (SC II) at all stream sections, suggests that reproduction occurs in all studied sections of the stream (see Fig. 3).

Our data also show that males are larger than females, with predominance in SC IV and SC III, respectively. These results differ from those reported for H. tectifera by Lescano et al. (2008), who observed that females and males belonged mostly to SC IV and SC III, respectively. Although it was not observed in our population in Rodriguez stream, other studies reported sexual size dimorphism with females being larger than males in other populations of H. tectifera (Cabrera 1998; Lescano et al. 2008; Semeñiuk et al. 2017; M.B.S., R.M.S., M.J.C., E.P., and L.A., unpubl. data, 2018). The predominance of larger males may be explained by factors that remove larger females from the population. For example, terrestrial nesting excursions by females expose them to predators (e.g., dogs and people in the cities) and cars, increasing mortality (Gibbs and Steen 2005). This may be a plausible explanation for our results that show males on average larger than females in the studied population. However, we recognize that many factors (e.g., environmental and spatial), other than the one suggested here, influence sexual size dimorphism (Agha et al. 2018).

We wish to highlight a concern about the potential lack of independence among the 3 sections of our study stream. Particularly, it is important to clarify the impact of turtle migration along stream sections and its possible impact on the present results. One approach to assess this issue was the recapture of marked turtles. Each of the 14 recaptured turtles was collected at the same stream section as their first capture. This finding suggests that minimum migration of turtles took place during our study (1 yr). Additional information supporting this lack of migration between sites comes from currently under way research projects, for which we have continued marking individual turtles at S3. Of a total of 221 turtles, 39 individuals (25 males and 14 females, most recaptured once but some 2 or 3 times) were recaptured at the same place they were first marked, suggesting that migration across stream sites is not common in brief time-scale periods (1–2 yrs).

In summary, the current degree of urbanization of Rodriguez stream seems not to negatively impact the population of H. tectifera. Some of the topics highlighted here represent the first contribution to understand how urbanization impacts this species. We recommend further research into their habitat requirements and the availability of microhabitats in this anthropogenic stream by taking into account factors such as substrate types, nesting sites, and aquatic plant cover, among others. In any case, a periodic monitoring program is clearly necessary, as there may be a lag period between the beginning of anthropogenic disturbance and its effect on turtle populations (Eskew et al. 2010b).