Sexual dimorphism and morphometric study of Caspian pond turtle, Mauremyscaspica, (Testudines: Geoemydidae) in Golestan Province, southeast of the Caspian Sea

Yazarloo M, KamiH.G*, Bagherian Yazdi A

Department of Biology, Faculty of Sciences, Golestan University, Gorgan, Iran.

* Corresponding author’s E-mail:hgkami2000@yahoo.com

(Received: Apr. 23. 2017 Accepted: Oct. 11. 2017)

The Caspian pond turtle, Mauremyscaspica, is a terrapin belonging to family Geoemydidae. Totally 130 specimens (67 males and 63 females) were collected manually and by net from different aquatic habitats in Golestan Province from 2016 through 2017. Morphometric characteristics such as length were measured using digital caliper and weight by electronic weighting scale. Skin of this turtle was dark olive green in color and had rows of longitudinal yellow stripes around the head and neck. Their carapace was relatively flat, the bridges between the dorsal and ventral shells were bony, and the ends of anal scutes were found to be pointed. Fingers and toes had swimming membranes. Sexes were identified using the location of vent on the tail. Sex ratios of males to females were nearly 1:1. The maximum straight carapace length (SCL₂) was 80.66- 230.16 (156.72 ± 42.93) and 56.96 - 236.84 (147.02 ± 50.76) in males and females respectively. Ratios of SCL₂ to maximum plastron length (PL₂) were found to be 1.013 - 1.32 (1.15 ± 0.04) in males and 1.01 - 1.15 (1.08 ± 0.03) in females; ratios of SCL₂ to straight carapace width (SCW) were 1.24 - 2.60 (1.40 ± 0.16) in males and 1.20 -1.47 (1.35 ± 0.07) in females; ratios of SCL₂ to carapace height (CH) were 2.28 - 3.55 (2.79 ± 0.22) in males and 2.05 - 8.78 (2.87 ± 1.11) in females; ratios of SCL₂ to tail length₂ (TL₂) were 2.33 - 7.59 (3.76 ± 0.91) in males and 2.34 - 4.78 (3.06 ± 0.62) in females. Ratios of straight width of femuro-anal suture (FASW) to anal seam length (AnSL) were 2.95 – 5.89 (3.78 ± 0.54) in males and 2.65 - 4.13 (3.23 ± 0.40) in females. The ratio of TL₂ to TL₁ and TL₁ to RBrL are found to be the fast and simple grouping index to determine sex of the specimens.

INTRODUCTION

Testudines are reptiles that typically exhibit a long lifespan, delayed sexual maturity, and slow growth (Rueda-Almonacidet al. 2007; Moura et al. 2015). These characteristics have been associated with a low rate of individual substitution among populations, making this group more susceptible to anthropogenic threats and environmental weathering (Rodrigues 2005; Moura et al. 2012). There is strong relation between population dynamics of testudines and climatic variation, resulting in a seasonal reproductive cycle (Souza 2004).The duration and level of biological activities such as foraging, thermal regulation and migration are determined by genetic interactions and environmental conditions (Molina 1992). These aspects directly impact the population density and sex ratio; they are responsible for variations in the population dynamics of testudines (Gibbons 1990; Brito et al. 2009).Nine species and six subspecies of turtles, terrapins and tortoises belonging to nine genera and six families (Rastegar-Pouyani, et al. 2008) and 12 species of turtles and tortoises have been reported in Iran, two of them are exotic (Safaei-Mahroo et al. 2015).The Caspian pond turtle, Mauremys caspica (Gmelin 1774), is a medium-sized freshwater Geoemydid turtle that is  widespread throughout the Middle East, where it is distributed from Inner Anatolia and Syria over the Caucasus region to Iraq and Iran; isolated relict populations are known from Bahrain and adjacent Saudi Arabia (Yadollahvand & Kami,  2014). Caspian pond turtle has three subspecies in Iran. M. c. caspica (Gmelin 1774) is widely distributed in Golestan, Mazandaran, Guilan, Ardabil, East and West Azarbaijan provinces; M. c. siebenrocki Wischuf & Fritz 1997 in Bushehr, Kordestan, Kermanshah, Lorestan, Ilam, Khuzestan, Fars and Chahar Mahal Va Bakhtiari provinces and M. c. ventrimaculata in Fars and Esfahan provinces (Fritz & Wischuf 1997; Kami et al. 2006; Fritz et al. 2007; Rastegar-Pouyani, et al. 2008; Safaei-Mahroo et al. 2015). Researchers often understand body size as a linear measurement based on anatomical landmarks (Gosler et al. 1998; Regis & Meik 2017), for example snout-vent length in most amphibians and reptiles, total length in others (Olalla-Tárraga & Rodríguez 2007; Meiri 2010), and different other linear measurements are often used among invertebrates. To other researchers, body size is compared with some other physical characters such as mass or volume. Recently, progresses in geometric morphometrics have allowed three-dimensional evaluations of size and shape of biological structures (Chiari et al. 2008).Carapace length often is used as a stable measurement of size across turtles, with little or no apparent daily or seasonal variation. However species vary, in the relative size and shape of the shell when compared with other aspects of body size (Pritchard et al. 1983; Bolten 1999).There is a difference in mean body size between males and females in many species of turtles. The development of external (e.g., elongate tail, bright colors, elongate claws, etc.), or internal (e.g., biochemical and brain structure) characters between the sexes is common (Bonnet et al. 2001; Harvey & Bradbury 1991; Leuteritz& Ganz 2013). These differences in size and special characters are referred to sexual size dimorphisms (SSD) (Harvey & Bradbury 1991). There are two major theories to explain SSD. One theory is based on sexual selection, which was first proposed by Darwin (1871). This states that sexual selection can result from either intra- or intersexual selection. In intrasexual selection, selection acts on traits that give an advantage to the possessor in competition among members of the same sex. In intersexual selection, selection acts on traits that make the possessor more likely to be chosen by the opposite sex (Berry & Shine 1980; Gibbons &Lovich 1990; Harvey & Bradbury 1991; Lovich& Gibbons 1992). Another theory to explain SSD as a result of natural selection (Darwin 1859). Differences in size can be related to differential interactions of each sex with its environment as the result of natural selection, or ecological forces (Slatkin 1984; Lovich & Gibbons 1992).Turtles exhibit a wide variety of size differences between the sexes (Berry & Shine 1980). It was suggested that female optimal body size may depend primarily on the number of eggs she lays, and secondarily on the degree of nest predation occurring. These factors may be similar between females of closely - related species, but are of course different than the factors affecting male body size. This suggests that the mating system of males is responsible for the variability in direction and degree of interspecific SSD (Leuteritz& Ganz 2013).In addition to body size, other different characters exist that have often been used to determine the sex of turtles, including shell morphology, claw length, eye color, tail size, and distance of cloacal vent to posterior body margin (Gibbons & Lovich 1990; Bonnet et al. 2010).There is a concavity in the abdominal and femoral scutes of plastral shell that is indicative of males in many species (Carr 1952; Ernst & Barbour 1989; Ernst &Lovich 2009).Results of some researches  were reported on Caspian pond turtle in Iran, including morphologic and morphometric, biological study (`Hojjatiet al. 2005;Kami et al. 2006; Karimpouret al. 2011; Kami et al. 2012; Yazarlooet al. 2016), cytogenetic characterization (Yadollahvand et al. 2013), and assessment trace element concentrations in its tissues (Yadollahvand et al. 2014),  and  impacts of habitat change on the  turtle (Yadollahvand & Kami  2014)  as well as phylogeographies (Fritz et al. 2007), conservation genetics and phylogeography (Vamberger et al. 2012).The purpose of this study was to investigate sexual dimorphism in terms of much more specimens and characters that have not been carried out yet and set a simple index to identify sexes for all specimens. Previous investigations have surveyed maximum roughly one - third of these characters and no reported index.

MATERIALS AND METHODS

According to distribution of this species in the north of Iran, 130  specimens were collected from 14 stations (Fig. 1) such as Sijoval, Mohammad-Abad, Imer, AqQala, Voshmgir dam, Anbar-Olum, Gharasu, Niazabad, Kordkuy, Khanbebin, Kalaleh, Galikesh, Maraveh Tappeh and Gonbad Kavus in aquatic habitats including rivers, ponds, pools and fish farms in Golestan Province. As the species have hibernation in winter, specimens were collected in spring, summer and fall during 2016-2017 by hand and net. The turtles were transported alive to the laboratory in the Golestan University and then morphometric characters were measured using digital caliper to the nearest 0.1 mm (Figs. 2, 3, 4 and 5) based on some references (Terentev & Chernov 1949; Perälä, 2001) Quantitative data was recorded  including  25 characters (Table 1) and their weights were measured using electronic weighing scale. Sexes were determined using visual observation of morphological characters. All specimens were finally returned to their habitats. Statistical analyses were carried out using Excel and SPSS (version 22) and also paleontological statistics (PAST) (version 3.14) software to study sexual dimorphism in males and females. Results were considered significant at P< 0.001.

Fig. 1. 14 collecting stations of specimens in Golestan Province.

Normal distributions of the characters were tested by Shapiro-Wilk separately for males and females. Significance of differences between sexes in mean value of the parametric characters was tested by t-student test. Sexual

differences in nonparametric characters were tested by Mann-Whitney U test. Multivariate analysis of variances (MANOVA) were performed for comparing mean vectors of sexes.

An explorative principle component analysis (PCA) based on correlation matrix extraction was performed to explore patterns of co-variation in the data matrix and its relationships with sexual dimorphism. The stepwise discriminant analysis with cross-validation was carried out to examine validation of grouping the specimens by their sexes and extract the most discriminative character(s) between sexes.

Table 1. Measurements and conventional abbreviations used in this study.

Fig. 2. Measured parameters of carapace.

Fig. 3. CH: Maximum carapace height

Fig. 4. Measured parameters of plastron.

Note: If the left and right scute are different in length, as is the case here, then take the mid for placing the caliper.

Fig. 5. Measured parameters of tail length.

RESULTS

The carapace of Caspian pond turtles, Mauremys caspica, was relatively flat, having two supra-caudal scutes; Their fingers and toes had swimming membranes, and also five and four claws respectively. Plastron lacks mobility, with axillary and inguinal scutes. The bridge between the dorsal and ventral shells was bony. The skin of this turtle was dark olive green in color and had rows of longitudinal yellow stripes around of head and neck. The ends of anal scutes were to be pointed. The presence of a plastral concavity located at the abdominal and femoral scutes was indicative of males in this species; this is due to the ease of mating. As an adaptation for copulation, male turtles also had longer, thicker tails with the cloacal vent located more distally as compared to females. In this study, sex ratio of males to females was nearly 1: 1.

Results of mean, minimum, maximum and standard deviation (SD) of characters in males and females are presented in Table 2, and also ratios of some parameters in Table 3.

All morphometric characters showed normal distribution (P> 0.001) except for weight in females (p = 0.001).  Four characters (W, PL₁, PFLW and PHLW) did not show normal distribution in males (P<0.001).

In the parametric compare mean analysis, mean values of characters TL₁ (p< 0.000), AnSL (p< 0.005) and TL₂ (p<0.005) were significantly different between the sexes. None of the four nonparametric characters showed

significant differences between males and females (p> 0.66). Characters with higher mean values in male than in female included: Tl₁ (56%), SCL₁ and SCL₂ (6.5%), FASW (4.8%), CH (4.5%), GU-w (7%) and lower mean values included: AnSL (12.7%), W (7.8%), PSL (8.5%) and RBrL (7.6%).

Mean vectors of characters differed significantly between sexes (F= 16.7; p<0.000).The first four component of principle analysis represented over 95% of variances. The first component  exhibiting  a high positive correlation with all the characters, showed variation  in overall sizes of the specimens, while the second one showed sexual dimorphism in  size (Fig. 6).

Characters TL₁, ASO-h exhibited a higher positive correlation with second component (Correlations 8.26 and 0.106 respectively) whereas AnSL, PSL and RBrL showed negative correlation (Correlations 0.36, 0.18 and 0.14 respectively). Linear discriminant analysis (LDA) with stepwise data reduction extracted one significant function (Eigen value = 4.38; Wilks' Lambda = 0.186; p<0.000) that classified 98.4% of cases to the correct memberships with cross-validation (Table 4), whereas the discriminant function grouped all the membership correctly but based on posterior probabilities of membership two specimens showed a doubtful grouping with p = 0.47 and 0.58 (Fig. 7). Included characters were PL₂, RBrL and TL₁. The most discriminative characters were TL₁, AnSL, ASO-h, PSL, RBrL, SCL₁ and SCL₂.  The ratio TL₂  to TL₁ is simple grouping index to determine sex of the specimens with mean value nearly 1.2 for males and 3 for females  in 63% of specimens (n = 81) which have complete tail tips. The index score for these specimens is: males < 2.13 = TL₂/TL₁ = 2.21 2), thus we found also an alternative index_, the ratio of TL₁ to RBrL. The index determined sex of the specimens with 100% confident on all specimens (n=127) with following sores: Females < 0.46 = TL₁/RBrL = 0.50 < males.

Table 2.Values are Min-Max (mean ± SD) of characteristics in males and females.

MIN= Minimum

MAX= Maximum

SD= Standard Deviation

Table 3.Min-Max (mean ± SD) of seven ratios calculated in males and females.

Min= Minimum

Max= Maximum

SD= Standard Deviation

Fig. 6. Scatter plot of Caspian pond turtles based on first two components.

a. 99.2% of original grouped cases correctly classified.

b. Cross validation is done only for those cases in the analysis. In cross validation, each case is classified by the functions derived from all cases other than that case.

c. 98.4% of cross-validated grouped cases correctly classified.

Fig. 7. Scatter plot of specimens based on LDA score and posterior probability of memberships.