Structure of a Population of Hydromedusa maximiliani (Testudines, Chelidae) from Parque Estadual da Serra do Mar, an Atlantic Rainforest Preserve in Southeastern Brazil
Abstract
We studied the structure of a population of Hydromedusa maximiliani associated with a stream in Parque Estadual da Serra do Mar, Núcleo Itutinga-Pilões, southeastern Brazil, between October 2004 and October 2005. Twenty-five individuals were captured, and a population size of 43.72 ± 23.7 individuals was estimated. This value is similar to that of the population of Parque Estadual Carlos Botelho, another Atlantic forest reserve of southeastern Brazil. Males were recaptured more frequently than females, suggesting higher activity and/or greater movement of males.
The freshwater chelid turtle Hydromedusa maximiliani (Mikan 1820) is endemic to the Brazilian Atlantic Rainforest (Yamashita 1990; Guix et al. 1992; Souza 1995a, 1995b; Souza and Abe 1995), one of the main South American biodiversity hotspots (Mittermeier et al. 1999). This biome is now highly fragmented because of forest destruction for cattle farming and sugarcane, coffee, and cocoa plantations, so that less than 9% of the original Atlantic coast forest remains (Myers 1986; Fearnside 1990, 1996; Dean 1996; Mittermeier et al. 1999). The best preserved remnants are in the coastal regions of south and southeastern Brazil (Mamede et al. 2004), where the largest Brazilian cities, São Paulo and Rio de Janeiro, continue to impose an enormous anthropogenic pressure on the forest (Por 1992; Dean 1996).
Despite the extreme importance of the Atlantic Rainforest as a refuge for different taxa, from plants to vertebrates, information on the biodiversity of many localities is still lacking. This is the case with the municipality of Cubatão, one of the most impacted areas of southeastern Brazil, because of the intense process of degradation that includes the construction of a complex of highways, the irregular use of the land, and the presence of a petrochemical refinery (Leitão-Filho 1993; Gutberlet 1996; Targa et al. 2001).
Chelonians are typically long-lived animals, and long-term studies provide the best data for conservation measures (Henry 2003; Litzgus 2006; Smith et al. 2006). Hydromedusa maximiliani was intensively studied by F.L. Souza and co-workers in Parque Estadual Carlos Botelho, another Atlantic Rainforest preserve in southeastern Brazil. They obtained data on natural history (Souza 1995a, 1995b; Souza and Abe 1995), ecology (Souza and Abe 1997a, 1997b, 1998; Souza and Martins 2006), genetics and molecular biology (Souza et al. 2002a, 2002b, 2003), and growth rates, which provided estimates of the age of the animals (Martins and Souza 2008). Other data on natural history are in Yamashita (1990) and Guix et al. (1992). Information obtained in different localities, however, is essential to verify if population characteristics represent a local pattern or if there is variation across the geographical distribution of the species.
Hydromedusa maximiliani is considered Vulnerable by IUCN (2011) and Critically Endangered in the state of Minas Gerais and Vulnerable in the states of São Paulo and Espírito Santo, southeastern Brazil (Martins and Molina 2008).
The aim of this study was to determine the population structure of H. maximiliani in an Atlantic rainforest fragment in Cubatão, southeastern Brazil, by means of mark–recapture techniques.
Methods
Field work was carried out in Parque Estadual da Serra do Mar (PESM), Núcleo Itutinga-Pilões, a 315,390-ha reserve of Atlantic Forest located in the municipality of Cubatão, São Paulo State, southeastern Brazil (lat 23°54′S, long 46°31′W) (Clauset 1999). A dense forest dominates the study site, with a canopy composed of 10- to 30-m–tall trees and a 2-m–tall understory. Lianas and epiphytes growing on tree trunks filter the sunlight reaching the forest floor (ECOVIAS 2002). In PESM, a segment of stream 1035 m long, 0.5–7 m wide, and 3–230 cm deep was selected to investigate the population structure of H. maximiliani. The stream was divided into 6 stretches separated by small waterfalls that form downstream plunge pools.
Fieldwork was conducted weekly between October 2004 and October 2005. Turtles were located visually during diurnal walks along the stream, twice a day (upstream to downstream and vice versa). Turtles were captured with a dip net and marked by filing notches in marginal scutes of the carapace or plastron (Cagle 1939).
Measurements were taken of curvilinear carapace length, curvilinear plastron length, carapace width, plastron width, and tail length, using a tape to the nearest 1 mm (Bandas and Higgins 2006). A sexual dimorphism index was calculated by dividing mean male carapace and plastron length by mean female carapace and plastron length (Lovich and Gibbons 1992). Ratios were based on the sample median, mode, and maximum value of carapace and plastron length, and also on the mean of 3 and 5 of the largest adult individuals of each sex. External morphological characters distinguished males, females, and juveniles. Males are larger than females (Guix et al. 1992) and have a concavity in the plastron and a tail that is significantly longer than wide (Ernst and Barbour 1989). Juveniles were identified by carapace shape and plastron coloration (Souza 1995b) and by the serrated shape of the posterior carapace margin (Yamashita 1990; Souza 1995b). At each capture, animals were screened for external parasites and injuries and defensive behaviors were recorded.
Population size was estimated by the mark–recapture method of Jolly-Seber (Krebs 1998) using the software Krebs/Win version 0.94, available at http://www.biology.ualberta.ca/jbrzusto/krebswin.html. Operational sex ratio was calculated as the number of adult females to the number of adult males. Biometrical measurements were compared by ANOVA.
Results
During the study period, 25 individual turtles were marked, comprising 10 males (40%), 8 females (32%), and 7 juveniles (28%); the total number of captures was 63 (Fig. 1). Most captures (41) occurred in the rainy season (October 2004 to January 2005); females were only captured in this period. One male was recaptured 9 times, and 2 other males were recaptured 7 times each. The minimum period between captures was 1 day (for a male) and the maximum was 314 days (for a juvenile). Animals recaptured after long periods (4–5 months) were muddy, and a muddy juvenile smelled, suggesting an estivation period. Population size was estimated as 43.72 ± 23.7 individuals. Operational sex ratio for the period of study was 0.8∶1.



Citation: Chelonian Conservation and Biology 10, 1; 10.2744/CCB-0841.1
The size of the turtles (curvilinear carapace length) varied from 9.5 to 23 cm (16.87 ± 4.06; n = 25), with a bimodal distribution (Fig. 1). Juveniles ranged from 9.5 to 14 cm (11.64 ± 1.82; n = 7); whereas, adults ranged from 14 to 23 cm (females: 16.25 ± 1.44, n = 8; males: 20.13 ± 1.88, n = 10) (Fig. 2). Table 1 shows that the measurements differed significantly among adults and juveniles in all biometric parameters (p < 0.005), except for carapace width when comparing females and juveniles. Males and females were different only in tail length (p < 0.005). The sexual dimorphism index calculated with mean values was 2.24 to carapace length and 2.13 to plastron length (Table 2).



Citation: Chelonian Conservation and Biology 10, 1; 10.2744/CCB-0841.1
Two defensive behaviors were observed during handling of animals. Cloacal discharge (contents of the cloaca were expelled, including feces and urine) occurred in 10% of the captures of males and females; a single juvenile displayed this behavior. Males everted the penis totally or partially in 28% of the captures.
Some individuals had shell or limb injuries. The right forefoot of a male was missing, and it had injuries in the posterior left marginal scutes. A female had small holes in the anterior marginal scutes. Another male was recaptured several times in a period of 2 months, showing no injuries, but after 6 months it had lost the digits of the right forelimb and 2 claws of the left forelimb, and it was bleeding. This male was recaptured several times in the following months. No dead turtles were found in the area during the study period. The only evidence of the presence of a turtle predator in the study site were tracks of a cat (probably the ocelot, Leopardus pardalis). The only ectoparasite detected was a leech, which was found attached to the plastron of a male.
Discussion
The estimated population size of 43.72 individuals in PESM is similar, in terms of number of individuals per kilometer, to that of the Parque Estadual Carlos Botelho (PECB), another Atlantic forest reserve of southeastern Brazil, where Souza (1995a) and Souza and Abe (1997a) calculated a population size of 103 individuals in a stream 5000 m long; that is, roughly 21 individuals/km (25 in the PESM stream). Males were recaptured more frequently than females, suggesting higher activity, longer distances travelled, or larger home ranges of males. In fact, males of this population are more active and travel longer distances than females (Famelli et al., unpubl. data, 2010). This pattern may represent a male tactic to find females for mating, a common strategy adopted by chelonians (Berry and Shine 1980). These data do not corroborate data obtained by Souza (1995a) in PECB, where females seem to move more than males. However more recent data (2009–2010) obtained in PECB using radiotelemetry and thread bobbins show that males have larger home ranges (Famelli et al., unpubl. data) and exhibit longer movements than females (Adriano et al. 2009), corroborating the present data.
The proportion of juveniles in our population was close to values presented by Bury (1979), who found an average of 31% juveniles in a composite of samples, including mainly Trachemys scripta and Chrysemys picta (Emydidae). This suggests that the PESM population is healthy, or at least shows no evidence of decreased recruitment, despite the habitat degradation that continues in the area. However, the present study focused only on a segment of a single stream, which makes generalizations to other streams difficult, including other stretches of our study stream. More information on population structure and dynamics obtained by long-term studies developed along the entire range of the species are needed to more fully evaluate its conservation status.
Operational sex ratio in PESM (0.8∶1) was not similar to that of PECB (2∶1), but this difference may be due to our small sample size (18 adults) compared with that of PECB (n = 98) (Martins and Souza 2009).
The reproductive biology of H. maximiliani remains poorly known (Guix et al. 1992; Souza 2004; Souza et al. 2006). According to Souza (2004), the seasonal activity cycle of chelid turtles is strongly associated with their reproductive periods, when males actively search for females and females search for nesting sites; as a consequence males and females have complementary activity periods. In PESM, H. maximiliani was more active during the rainy season (October and January) and female activity was restricted to this period, which corroborates the pattern described for PECB animals (Yamashita 1990; Souza 1995a; Souza and Abe 1997a) and suggests this may be the mating season of the species (Yamashita 1990; Guix et al. 1992; Souza 1995a, 2004; Souza and Abe 1997a).
According to Guix et al. (1992), females of H. maximiliani may build nests away from stream margins to avoid flooding. Nesting seems to occur in January–February (Souza 1995a, 2004). In the present work, females were captured only between October and January, which suggests they may estivate after that. Incubation of eggs lasts 250–330 days (Souza 1995a, 2004), so hatchlings would be expected to appear in September–October. Information on hatchlings is very useful to conservation practices, since it permits estimates of population size and growth and insights on the potential for recovering a population (Pike 2006).
Estivation is a critical stage of turtle life, and knowledge of habitat requirements in this stage is fundamental to turtle conservation (Burke and Gibbons 1995). The presence of mud on the shell of some individuals after long periods of absence suggests H. maximiliani buries itself in the mud, as described for other side-necked turtles, like H. tectifera (Lema and Ferreira 1990), Acanthochelys macrocephala (Artner 2007), A. pallidipectoris (Cabrera 1998; Richard 1999), Platemys platycephala (Métrailler and Le Gratiet 1996), Mesoclemmys gibba (Mittermeier et al. 1978), and M. tuberculata (Bour and Zaher 2005). This behavior was also recorded in Trachemys adiutrix, which remains under the sand during the dry season and emerges after the first rains (Vanzolini 1995).
The shell injuries and the absence of a forefoot in some individuals were also reported for the turtles of PECB by Souza (1995a) and Souza and Abe (1997a), who suggested these anomalies may have resulted from the low temperatures experienced by embryos during incubation. The loss of digits and claws observed in a male, however, may have resulted from a predation attempt. A small cat, probably ocelot, was present in the study area and may prey upon H. maximiliani. This felid is known to prey on side-necked turtles (Lamar and Medem 1982) and other reptiles (Emmons 1987; Chinchilla 1997; Wang 2002).
Cloacal discharge and penis eversion were exhibited by some individuals when handled. These behaviors are commonly observed in captive side-necked turtles (F.B. Molina, pers. obs.) and may be related to defense against predators. Cloacal discharge and hemipenis eversion are common primitive defensive behaviors of many snake and lizard species (Greene 1988, 1997; Martins 1996).
Cryptic coloration of H. maximiliani makes it difficult to detect the animals in the field (Guix et al. 1992; Souza 1995a, 2005; Souza and Abe 1997a). The turtles resemble dead leaves, which are very similar in shape and color to the turtle shell, on the stream bottom, especially those of Alchornea triplinervia (Euphorbiaceae). Similar cryptic coloration has been described for Chelus fimbriatus, which resembles the leaves of Montrichardia arborescens, an Araceae (Métrailler and Le Gratiet 1996). The only ectoparasite found attached to the turtles was an unidentified leech, a group containing many species commonly associated with turtles (Siddall and Gaffney 2004; Tucker et al. 2005; Readel et al. 2008).
The ecological requirements of H. maximiliani, along with its restricted distribution in the Brazilian Atlantic Rainforest (Ernst and Barbour 1989), one of the most threatened biomes in the world (Mittermeier et al. 1999), justify more detailed studies of this species.

Number of turtles captured in Parque Estadual da Serra do Mar between October 2004 and October 2005. Males: white bars; females: black bars; juveniles: shaded bars.

Population structure of Hydromedusa maximiliani at Parque Estadual da Serra do Mar, southeastern Brazil, between October 2004 and October 2005. Males: white bars; females: black bars; juveniles: shaded bars.