Turtles are animals with moderate investment in reproduction and reproductive parameters, which are generally influenced by environmental factors such as climate (Souza and Abe 2001; Vitt and Caldwell 2014). Especially for species with wide distributions, characteristics such as size at maturity, nesting period, clutch size, and frequency of reproductive events tend to vary according to temperature and precipitation gradients, reflecting genetic differences and/or resource availability in the environment (Gibbons et al. 1981; Gibbons and Greene 1990). Comparative data from different populations are essential to understanding plasticity in species reproduction. Population studies in areas with different climatic regimes are especially valuable because they can provide conditions to evaluate geographical ecology patterns (MacArthur 1972; Aresco 2004). Information regarding the population structure of freshwater turtles is needed for management and conservation purposes (Gibbons et al. 2000; Primack 2012; Stanford et al. 2020). However, such data are still scarce for the Neotropical region (Marques et al. 2013). In addition, these studies can also provide a powerful means of refining and testing models related to the life history theory and the influence of environmental variation on species (Dunham and Overall 1994). Characterization of the age structure, sex ratio, and morphological variations between sexes are important tools in interpreting the ecological processes that act on natural populations (Gibbons et al. 2001; Aponte et al. 2003; Silveira et al. 2012).

Species that occupy large areas with significant environmental variation may vary significantly in their autoecology (Clavijo-Baquet et al. 2010; Litzgus and Smith 2010). Morphological variation is commonly observed in turtle populations (Lucas et al. 2020). However, it is difficult to quantify and identify the evolutionary processes that have shaped such variation due to the complex interactions between the phenotype and the behavior of these animals (Delmas et al. 2007; Litzgus and Smith 2010). This variation can be promoted by selective pressures or ecological factors, such as predation. In addition, the availability of food resources and reproductive behaviors may also influence the morphology of chelonians (Willemsen and Hailey 2003; Barros et al. 2012).

Phrynops geoffroanus has a wide geographic distribution in South America, from the Colombian Amazon to southern Brazil and northern Argentina (Rueda-Almonacid et al. 2007; Ferronato et al. 2009). It is found in the main hydrographic basins of the region and occurs in varied aquatic environments such as rivers, streams, and lakes (Costa et al. 2018; Pereira et al. 2018; Abrantes et al. 2021) and is even commonly found in urban environments (Martins et al. 2010; Andrade et al. 2020; Santana et al. 2020). Although it is a widely distributed species, there remain some questions about its taxonomy, regarding the occurrence of cryptic subspecies or species within the group, with some authors considering this freshwater turtle to be a “geoffroanus complex” (McCord et al. 2001; Vogt 2008). Recently, Carvalho et al. (2022) tested taxonomic hypotheses using genomic data and were able to provide support for four species/lineages within the P. geoffroanus species complex. According to the authors, these species/lineages are restricted to specific river basins and biomes but are morphologically cryptic, and specimens from the Northeast Brazil correspond to P. geoffroanus lineage 3. Phrynops geoffroanus is considered a medium-sized species (Rueda-Almonacid et al. 2007; Rodrigues et al. 2019), which nests in sandy or clay soil, usually covered by shrub vegetation, and preferably in places with good exposure to the sun (Medem 1960). In natural conditions (Souza and Abe 2001), clutches are laid in the driest period of the year while hatching occurs during the rainy season.

Mesoclemmys tuberculata is also a medium-sized species (Vanzolini et al. 1980; Ernst and Barbour 1989; Vetter 2005), and its neck is covered with conical tubercles, reflected in the name of this species (Ernst and Barbour 1989; Bonin et al. 2006). It is a freshwater turtle that is endemic to Brazil, occurring mainly in the semiarid region of the northeast (Vanzolini et al. 1980; Iverson 1992). It is generally associated with the Caatinga and Atlantic Forest, mainly in areas surrounding the São Francisco River basin (Batistella et al. 2008; Santos et al. 2008; Moura et al. 2014). However, it has also been recorded in the Caatinga-Cerrado ecotone of the São Francisco River basin, northwest Minas Gerais State (Silveira and Valinhas 2010). Most of the ecology of M. tuberculata remains unknown (Souza 2004; Santana et al. 2016, 2019). Reproductive parameters such as clutch size, egg parameters (size, volume, mass), average clutch mass, incubation time (days), reproductive season, and description of the nest location are unknown in natural environments (Santana et al. 2016).

In the present study, we tested the hypothesis that intraspecific factors are more important than differences in environments in determining the ecological parameters of P. geoffroanus and M. tuberculata in areas of Caatinga and Atlantic Forest in northeast Brazil. More specifically, we evaluated the sex ratio and the presence of sexual dimorphism, and we described aspects of morphometry and reproduction of both species. Taking into account that most turtle species can be strongly influenced by climate, our prediction was that differences in the two environments are not sufficient to promote variations in ecological patterns.

METHODS

Study Sites. — Populations of P. geoffroanus and M. tuberculata (Fig. 1) were studied in six localities (three in the Caatinga and three in the Atlantic Forest) in the State of Sergipe, Brazil (Fig. 2). The locations selected in the Caatinga area were the Monumento Natural do Rio São Francisco (09°41.202′S, 037°42.963′W, 189 m elevation; municipality of Poço Redondo), Monumento Natural Grota do Angico (09°40.852′S, 037°41.408′W, 170 m elevation; municipality of Poço Redondo), and Bacia do Rio Real (11°10.730′S, 038°02.300′W, 158 m elevation; municipality of Tobias Barreto). The Atlantic Forest areas selected were the Parque Nacional Serra de Itabaiana (10°45.251′S, 037°20.511′W, 208 m elevation; municipality of Areia Branca), Reserva Biológica de Santa Isabel (10°35.766′S, 036°40.366′W, 13 m elevation; municipality of Pirambú), and Refúgio de Vida Silvestre Mata do Junco (10°31.895′S, 037°03.142′W, 118 m elevation; municipality of Capela).

View Figure

• Download as Powerpoint
• Download Figure

View Figure

• Download as Powerpoint
• Download Figure

Turtle Collection. — Samples were obtained in semiannual surveys (dry and rainy season) over a year, lasting 10 consecutive days at each of the six sampling sites, with three in the Caatinga and three in the Atlantic Forest, totaling 120 d of sampling. The study was conducted in 2014, and we defined the rainy and dry seasons based on historical averages of the distribution of precipitation in these areas (Hijmans et al. 2005). The months from April to August were considered the rainy season, and the remaining months were assigned to the dry season.

We used 20 hoop net traps placed in rivers, streams, and/or lagoons to capture individuals at each sampling point. In order to attract turtles, the traps were baited with beef placed in the water from 0600 to 1800 h and were checked every 3 h. There were occasional nocturnal collections. In environments with less-turbid waters, an active search (free diving with the aid of a mask, snorkel, and fins) was performed between 0700 and 1700 h.

Procedures. — We weighed the captured animals with a Pesola® scale (0.5-g precision) and performed 22 measurements with digital calipers (0.5 mm) and a tape measure (1 mm), the latter for curvilinear measurements: maximum straight-line carapace length (SCL), maximum straight-line carapace width (SCW), length of third central scute (LC3), width of third central scute (WC3), curvilinear carapace length (CCL), curvilinear carapace width (CCW) between the union of the second and third vertebral scute to the carapace edge, maximum plastron length (MPL), maximum plastron width (MPW), anterior lobe plastron width (LBA), posterior lobe plastron width (LBP), opening between the distal ends of the anal scutes plastral (ASP), carapace and plastron terminal distance (CPD), cephalic width (CW) measured over the tympanic membrane, cephalic length (CL), mouth width (MW), right and left barb length (RBL, LBL), right and left barb width (RBW, LBW), total tail length (TTL), distance between cloacal opening and tail end (COT), and maximum carapace height (MCH) (Fig. 3). We selected these morphometric measurements following the methodology described by Bager et al. (2010) and Fagundes (2010a, 2010b) and adapted them to the present study.

View Figure

• Download as Powerpoint
• Download Figure

We sexed each adult captured according to external morphological characteristics and secondary sexual characteristics, such as plastral concavity and tail length (Rueda-Almonacid et al. 2007; Brito et al. 2009; Molina et al. 2012).

We collected 5 individuals of each species, at each location, in each of the surveys, with the purpose of analyzing the reproductive cycle. Following measurements, the animals were partially anesthetized by being placed in a vessel containing water and ice at a temperature of about 5°C for approximately 20 min, following accepted animal care procedures (Instituto Chico Mendes de Conservação da Biodiversidade – ICMBio permission no. 38724-1/2013). Then, they were euthanized in the field by injection of 2% lidocaine, fixed in 10% formalin, and preserved in 70% alcohol (Costa et al. 2018). The reproductive condition was determined through direct observation of the gonads. Females were considered reproductive when they had vitellogenic follicles and calcified eggs, while males were considered reproductive when they had developed testicles and folded epididymides. When present, we measured egg length and width in the oviduct in females and testicle length and width in males. The reproductive period was associated with the season and the reproductive condition of adult individuals.

We marked individual specimens with perforations in the marginal shield, following the methodology described by Cagle (1939). This marking of the animals was used to prevent data from the same specimen being recorded more than once in the same survey and to allow registration of individuals in different periods of the study. The collected chelonians were processed in the Herpetology Laboratory and housed in the Herpetological Collection of the Universidade Federal da Paraíba (CHUFPB).

Statistical Analysis. — Initially, all data were assessed for normality using the Shapiro-Wilk test to determine the suitability of parametric or nonparametric analyses. Whenever necessary we log₁₀-transformed morphometric measures and mass to meet the premises of normality and reduce scale effects (Zar 1999). We created a variable called “body size,” defined as a variable resulting from the product of an isometric vector (Rohlf and Bookstein 1987), with values of p^–0.5 obtained by n × p matrix of log₁₀-transformed morphometric data, where n is the number of observations and p is the number of variables (Jolicoeur 1963; Somers 1986). To remove the effect of size on the transformed variables in log₁₀, we used the Burnaby method (Burnaby 1966), multiplying the n × p matrix of log-transformed data by a symmetric matrix (L) according to the following equation:

where I_p is the identity matrix p × p, V is the isometric vector defined above, and V^T is the transposed matrix of V (Rohlf and Bookstein 1987). The Wilcoxon test was performed to verify differences in body size (isometric vector) for SCL (mm) and mass (g), between sexes (adult individuals) of the same species between biomes.

Adult individuals were defined based on dissected individuals and specialized literature (Souza and Abe 2001; Moura et al. 2015; Santana et al. 2015). These adult individuals from the two populations were used to investigate possible variations in sex ratios during the study period. Population structure was analyzed through the distribution of SCL size classes and the interval of size class (C) of adults was calculated according to the empirical formula described by Sturges (1926), defined by:

where R is the range of values and n represents the sample size. Then, the proportions of each size class were adopted for this study.

Due to the rarity of these animals, reproductive data were analyzed independently of biome. We used an ellipsoid formula to estimate the volume (V) of eggs and testicles.

where w is egg/testicle width and l is its length.

We also performed regressions between the testicular volume and the SCL of males and the volume of eggs and clutch size (number of eggs) in relation to the SCL of females. We established the degree of sphericity of the eggs through the ratio of the smallest diameter divided by the largest. We used a chi-square test (χ²) to compare the sexual proportions (sex ratio) in the sampled environments, as well as a Fisher's Exact test, the exact version of the chi-square test employed for small samples.

All statistical analyses were performed using R (R Development Core Team 2011) and the significance level for hypothesis tests was 5%.

RESULTS

In total, we captured 154 freshwater turtles comprising 99 P. geoffroanus (11 males, 24 females, and 64 juveniles) and 55 M. tuberculata (24 males, 26 females, and 5 juveniles). The captures were more abundant in the Caatinga, with 133 individuals being 93 P. geoffroanus (8 males, 22 females, and 63 juveniles) and 40 M. tuberculata (18 males, 21 females, and 1 juvenile). In the Atlantic Forest, only 21 individuals were captured, with 6 P. geoffroanus (3 males, 2 females, and 1 juvenile) and 15 M. tuberculata (6 males, 5 females, and 4 juveniles). We considered as juveniles P. geoffroanus and M. tuberculata with an SCL less than 12 cm, the size of the smallest reproductive individual. Due to the low number of recaptures (only one recapture of a juvenile of P. geoffroanus and none of M. tuberculata), it was not possible to estimate population size, growth rate, survival, or migration rate.

Males and females of P. geoffroanus did not differ in size (Wilcoxon W = 151; p = 0.517), in SCL (Wilcoxon W = 158; p = 0.365), or mass (Wilcoxon W = 167.5; p = 0.214). Mesoclemmys tuberculata males and females presented significant differences in size (Wilcoxon W = 515; p < 0.001), in SCL (Wilcoxon W = 539; p < 0.0001), and mass (Wilcoxon W = 545.5; p < 0.0001), where females were larger and heavier than males. We did not find significant differences in the size, SCL, or mass of either species (P. geoffroanus and M. tuberculata), irrespective of sex and environments (Table 1; Fig. 4). Detailed information regarding the morphometry and mass of adults of P. geoffroanus and M. tuberculata, by correlated environment, is described in Tables 2 and 3, while juveniles were analyzed independently of the environment (Table 4). A juvenile of M. tuberculata showed the characteristics of a hatchling, such as its morphology (body proportions) and presenting a different color pattern than adults (Santana et al. 2015). The sex ratio was not significantly different from 1:1 in either species in the environments although, apparently, the females are more numerous in the Caatinga (Table 5).

Table 1. Values (± standard deviation [SD]) of body size (isometric vector), SCL (mm), and mass (g) of females (♀) and males (♂) of Phrynops geoffroanus and Mesoclemmys tuberculata in areas of Atlantic Forest and Caatinga in northeastern Brazil.

View Fullscreen

Table 2. Descriptive statistics of the morphometric measurements of adult individuals of Phrynops geoffroanus from Atlantic Forest and Caatinga in northeastern Brazil. Linear measurements are in millimeters and mass in grams. Variables: Maximum straight-line carapace length (SCL), maximum straight-line carapace width (SCW), length of third central scute (LC3), width of third central scute (WC3), circular carapace length (CCL), circular carapace width (CCW), maximum plastron length (MPL), maximum plastron width (MPW), anterior lobe plastron width (LBA), posterior lobe plastron width (LBP), anal scutes plastral (ASP), carapace and plastron terminal distance (CPD), cephalic width (CW), cephalic length (CL), mouth width (MW), right barb length (RBL), left barb length (LBL), right barb width (RBW), left barb width (LBW), total tail length (TTL), distance between cloacal opening and tail end (COT), and maximum carapace height (MCH).

View Fullscreen

Table 3. Descriptive statistics of the morphometric measurements of adult individuals of Mesoclemmys tuberculata from Atlantic Forest and Caatinga in northeastern Brazil. Linear measurements are in millimeters and mass in grams (see Table 2 for definition of abbreviations).

View Fullscreen

Table 4. Descriptive statistics of the morphometric measurements of juvenile individuals of Phrynops geoffroanus and Mesoclemmys tuberculata. Linear measurements are in millimeters and mass in grams (see Table 2 for definition of abbreviations).

View Fullscreen

Table 5. Sex ratio verified for Phrynops geoffroanus and Mesoclemmys tuberculata in Atlantic Forest and Caatinga, in northeastern Brazil.

View Fullscreen

View Figure

• Download as Powerpoint
• Download Figure

For the analysis of reproductive condition, we collected 61 turtles, 26 of which were P. geoffroanus (6 males, 12 females, and 8 juveniles) and 35 M. tuberculata (14 males, 17 females, and 4 juveniles). In P. geoffroanus, we found eggs in 41.7% (n = 5) of females from 224 to 285 mm in carapace length (mean = 254.8 ± 22.1 mm). (All values are presented as mean ± SD.) In total, we collected 42 eggs, ranging from 7 to 9 per female (8.4 ± 0.9 eggs), with an average length of 27.8 ± 1.4 mm and a width of 26.0 ± 1.2 mm. We registered the presence of eggs and oocytes exclusively in females collected during the rainy season (June to August). All reproductive females simultaneously presented eggs and developed vitellogenic follicles, indicating the production of sequential clutches during the reproductive season. The females collected in the dry season were not reproductive.

In M. tuberculata, we found eggs in 41.2% (n = 7) of females of between 132.6 and 220.8 mm in carapace length (mean = 221.2 ± 25.8 mm). In total, we collected 46 eggs, ranging from 3 to 9 per female (6.6 ± 1.99 eggs), with an average length of 30.6 ± 2.4 mm and width of 24.9 ± 1.9 mm. As in P. geoffroanus, females of M. tuberculata had eggs exclusively during the rainy season, where all females with eggs also had vitellogenic follicles. Three females collected in the dry season showed only vitellogenic follicles. Of the 9 females collected during the rainy season, only 2 had neither eggs nor vitellogenic follicles.

In P. geoffroanus, we did not find a significant relationship between clutch size (number of eggs) and SCL of females (R = 0.368, F_1,3 = 3.33, p = 0.165), although egg volume was positively related to SCL (R = 0.9669, F_1,3 = 117.8, p < 0.01). We also found no significant relationship in M. tuberculata between clutch size and female carapace length (R = 0.179, F_1,5 = 2.31, p = 0.189) or egg volume (R = 0.078, F_1,5 = 1,508, p = 0.274). With regard to the shape of the eggs, the degree of sphericity of P. geoffroanus eggs was 0.936 (n = 35) while that of M. tuberculata was 0.813 (n = 42).

We found reproductive males of both species throughout the study (dry and rainy seasons), comprising 50% of the P. geoffroanus collected, between 199.0 and 270.0 mm (mean = 244.7 ± 40.2 mm) and 86.7% of M. tuberculata, with carapace lengths between 139.5 and 187.0 mm (mean = 163.9 ± 14.4 mm). Testicular volume was positively correlated with SCL in males of both species: P. geoffroanus (R = 0.735, F_1,5 = 14.88, p < 0.001) and M. tuberculata (R = 0.339, F_1,13 = 7.69, p = 0.017).

DISCUSSION

Variations in ecological aspects, such as morphological characteristics, often result from different seasonal regimes and have been described for several vertebrates (Leclair and Laurin 1996; Bitto and Egbunike 2006), including reptiles (Clavijo-Baquet et al. 2010; Lucas et al. 2020). With regard to turtles, seasonal differences can drive different ecological pressures that affect their size (Clavijo-Baquet et al. 2010; Litzgus and Smith 2010; Lucas and Bager 2017). However, these differences were not significant in P. geoffroanus and M. tuberculata; their size in the Caatinga, an environment with a more stochastic rain regime, was similar to that in the Atlantic Forest, an environment with a more predictable and seasonal pattern (Nimer 1989). This absence of morphological variation may be related to relatively short geographic distances between the studied populations (∼ 150 km), despite drastically different environmental conditions, which may not restrict gene flow between them (Clavijo-Baquet et al. 2010), thus influencing the lack of morphological variation. These animals, which exhibit migratory behavior, can even disperse between different watersheds (Milam and Melvin 2001; Clavijo-Baquet et al. 2010).

Female Phrynops and Mesoclemmys are, in general, larger than males (Cunha et al. 2019; Müller et al. 2019; Santana et al. 2019; Abrantes et al. 2021; Carvalho et al. 2022), so the lack of sex differences found in P. geoffroanus in this study may be related to the small sample size of adults. The sex ratio found, of 1:1, and the pattern of females being larger than males in M. tuberculata, agrees with Berry and Shine (1980), who suggested that species with smaller males are associated with the absence of disputed territories for female's access. In addition, larger females may increase reproductive potential in terms of the number and size of eggs produced and storage capacity (Lovich and Gibbons 1992; Bager et al. 2007; Kaddour et al. 2008). This pattern may be even more pronounced in species with multiple clutches per reproductive season, such as those recorded here (Bager et al. 2007, 2010). Similar patterns have been verified for several Chelidae species (Cann 1998; Garbin et al. 2016), including P. geoffroanus (Molina 1998). The length of the smallest adult P. geoffroanus and M. tuberculata individuals were similar to that described by Rodrigues and Silva (2016) in Phrynops tuberosus from Caatinga in Ceará State, northeast Brazil.

The sex ratio represents a fundamental demographic parameter because it can affect population dynamics (Gibbons 1990; Lovich 1996). The sex ratio recorded for P. geoffroanus and M. tuberculata did not significantly differ from 1:1, which is similar to that found in many other studies, such as in P. geoffroanus in the Atlantic Forest (Souza and Abe 2001) and Caatinga (Moura et al. 2015; Abrantes et al. 2021). The occurrence of a higher proportion of males is rare in other Chelidae, but recorded to occur in P. tuberosus in Caatinga (Rodrigues and Silva 2016). There are also studies reporting a skewed sex ratio for females, such as in Mesoclemmys vanderhaegei in the Cerrado (Brito et al. 2009), and the only recorded sex ratio for M. tuberculata (Moura et al. 2015) was registered only for females. Bury (1979), after analyzing diverse turtle populations, observed that sex ratios varied within and between species, but in most cases tended to have a 1:1 ratio. Usually, the sex ratio found can be attributed to several factors, such as the influence of different sampling techniques, the result of habitat selection, differential mortality between sexes, incubation temperature, or a combination of these factors (Gibbons 1990; Fachín-Terán et al. 2003). We collected in all available habitats, at different times of the year, using two different methods, hoop net traps and active search and, in general, these methods did not show a trend of deviations in sex ratio (Sterrett et al. 2010; Rodrigues and Silva 2016). In this context, we consider that the sex ratios found could simply reflect the real proportion of males and females in the studied populations.

In general, reproduction of freshwater turtles is related to precipitation, particularly in semiarid regions (Souza 2004; Rodrigues and Silva 2014, 2015). Usually, turtles are observed to nest during the driest periods of the year, with hatchlings emerging in the rainy season (Souza and Abe 2001; Souza 2004). The records of P. geoffroanus nesting in the driest time of the year were in natural conditions (Souza and Abe 2001), but this study was carried out in southeastern Brazil, a region that has different climatic conditions, which could considerably influence reproductive biology (Iverson et al. 1997; Souza et al. 2006). However, in another study in southeast Brazil, a greater number of P. geoffroanus nests were observed at the end of winter (Silva and Vilela 2008). The records of pregnant females found exclusively in the rainy season characterized a well-marked seasonal reproduction, although the occurrence of the reproductive period in the dry season is usually reported. Thus, our results reinforce the influence of environmental conditions on the reproduction of freshwater turtles. Comparing studies in different regions, we believe that the reproductive period of P. geoffroanus can vary depending on habitat characteristics and local climatic conditions.

The slightly ellipsoid eggs found in M. tubeculata correspond to the shape commonly recorded for several turtles (Rhodin and Mittermeier 1983; Cabrera 1998; Mocelin et al. 2008), including all Australian Chelidae (Harless and Morlock 1989). However, P. geoffroanus presented eggs that were more spherical in shape (Medem 1960; Molina 1998; Silva and Vilela 2008), and the shape found by Molina (1998) of a 0.947 sphericity is very similar to the results obtained in our results.

The simultaneous presence of eggs and vitellogenic follicles indicates more than one clutch per reproductive season (Bager et al. 2007; Bujes and Verrastro 2009). Records of multiple clutches in the same season frequently occur in different chelonians, whether freshwater turtles, sea turtles, or tortoises (Gibbons and Tinkle 1969; Georges 1983). It is difficult to compare this information with other populations because information regarding the reproduction of turtles is usually obtained through the analysis of nests. Thus, we believe that our data contribute considerably to elucidating this aspect of the reproduction of free-living P. geoffroanus and M. tuberculata.

The average clutch size of 8.4 eggs found for P. geoffroanus was larger than that of 6 eggs recorded in Mato Grosso, central-west Brazil (Silva and Vilela 2008), and smaller than the average clutch of 15 eggs (Souza and Abe 2001) in São Paulo, southeast Brazil. However, Souza et al. (2006) reported that P. geoffroanus presents the largest clutch size of the Chelidae family, with a record of 30 eggs, although details of precedence of the sample were lacking. Intraspecific variation related to clutch size is generally seen in species with greater geographic distributions (Iverson et al. 1997; Souza et al. 2006), such as P. geoffroanus, which covers areas influenced by different climates and/or habitats (Souza et al. 2006). Thus, differences in clutch and egg size between populations are probably the result of differences in the quality and availability of food, which also influences the size of specimens, growth rates, and age at maturity (Gibbons and Greene 1990; Iverson et al. 1993).

To date, reproductive parameters such as average size at sexual maturity, clutch size, egg parameters (size, volume), average clutch mass, and reproductive season remained unknown for M. tuberculata (Santana et al. 2016). We did not find a significant relationship between clutch size and volume with carapace length in M. tuberculata or clutch size with carapace length in P. geoffroanus, but it is possible that a larger sample of gravid females could confirm such predictions because these relationships are generally found in other turtles (Bager et al. 2007; Fagundes et al. 2010a), including P. geoffroanus (Souza 2004; Souza et al. 2006).

In conclusion, P. geoffroanus and M. tuberculata from Caatinga and the Atlantic Forest did not vary significantly between different environments in terms of morphology, dimorphism, or sex ratio. It is possible that the different seasonal regimes did not promote such variation due to the relatively short geographic distance between these environments, or due to the occurrence of gene flow between populations of different basins. In this context, morphometric analysis is essential, as it can provide information on the degree of divergence between species and populations (Fernandez and Rivera 2001; Leary et al. 2003; Lindeman 2003; Yadollahvandmiandoab et al. 2018).

The main objective of turtle conservation is to ensure that viable populations exist in the long term; so to be successful, data on ecology, demography, habitat management, genetics, and husbandry need to be incorporated into management plans for each species (Stanford et al. 2020). Thus, other studies involving genetic analysis, habitat management, and husbandry could broaden the perspective of this study, leading to greater clarification of the patterns found and helping the conservation of the studied species.