Sexual Dimorphism in the Greek Tortoise: A Test of the Body Shape Hypothesis
ABSTRACT
In most animal species, it is expected that females should exhibit a greater abdominal volume than males to hold the progeny, when compared with females, males should exhibit more developed attributes that enhance mobility. We tested this hypothesis in the Greek tortoise. In chelonians, a reduction of the openings in the shell improves protection against predation but also constrains the abdominal volume and limits the space available to move the limbs. As expected, our data show that the shell provides a larger abdominal volume relative to tortoise size in females than in males. In males, deep notches in the shell and a reduction of several plastron plates offer more freedom to the limbs and to the tail; these characteristics presumably enhance mating success. Further studies are necessary to assess the applicability of these results in other chelonians, notably freshwater and marine turtles.
Recently, it has been suggested that sex divergence in body shape may be easier to interpret than sexual size dimorphism (SSD) at the species level, at least in the absence of parental care (Bonnet et al. 1998, 2001). Whatever the zoological group, and irrespective of body size, several morphologic traits should be consistently more developed in one sex relative to the other. For instance, females should exhibit a large intra-abdominal volume to enhance their capacity to hold a large clutch. However, the male's reproductive success increases with its mate-searching ability. In this sex, we may expect propulsive structures, such as locomotor musculature, or any traits that improve mobility, to be more developed (Bonnet et al. 1998).
Terrestrial tortoises offer an opportunity to address these simple and testable hypotheses. The shell constrains the intra-abdominal volume; whereas, the size of the openings in the shell limits the movements of the head, tail, and limbs. Theoretically, females should have a relatively greater shell volume relative to shell length; by contrast, males should have larger openings in the shell to enhance their mobility (see Table 1 in Bonnet et al. 2001). Studies on the steppe tortoise (Testudo horsfieldii) provided strong support to these simple hypotheses (Bonnet et al. 2001). An article on 3 closely related European tortoises, all from the Testudo genus, provided additional support; however, the data were not specifically designed to assess the sexual body shape dimorphism (Willemsen and Hailey 2003). For instance, the size of the limbs and head were not measured, and the shape of the plastron was not entirely characterized (only one plastron dimension was available). More information, nonetheless, is important to fully characterize the morphology (hence, the overall body shape) of the tortoises. For example, if males exhibit a smaller relative limb size compared with females, then the sexual dimorphism in the body-shape hypotheses would be seriously challenged. Similarly, the relative small size of the plastron in males may be a byproduct of its concavity solely and/or to the more flatter shell of the females; consequently, characterizing the notches of the plastron is essential to assess the mobility hypotheses.
The aim of this article is to provide such additional information on the body shape of the Greek tortoise (Testudo graeca) to perform straightforward comparisons with the steppe tortoise. Besides this test, the current data were gathered in an African population of Greek tortoises (Morocco), providing an opportunity for comparisons with the European populations studied by Willemsen and Hailey (2003) in Greece. The 2 sets of populations experience different environmental conditions (i.e., the very arid and Mediterranean climate in South Morocco vs. the mild climate on Greek islands). Growth rates and sexual maturity are strongly influenced by environmental conditions, notably food resources; such phenotypic plasticity determines the respective body size of each sex, and, consequently, the direction and degree of SSD (Shine 1991; Madsen and Shine. 1993; Pearson et al. 2002a, 2002b). By contrast, we expect that the direction of sexual dimorphism in body shape will remain relatively less affected, because selection for mobile and light males vs. heavy and highly fecund females (e.g., large abdominal volume) should apply over large ranges of body sizes.
METHODS
Animals and Study Area
The Greek tortoise (Testudo graeca graeca) is a medium-sized testudine (Fig. 1a), largely distributed in the Mediterranean region (Lambert 1983; Iverson 1992a; Ernst and Barbour 1989; Andreu et al. 2004). Most of the populations show a marked decline in the whole distribution range. The study site is a 32-ha area located in the central Jbilet mountains about 25 km north of Marrakech, in western Morocco (lat 31°37′N, long 8°02′W, altitude: 580 m above sea level). The study site is located in an arid region, with mean annual rainfall of about 240 mm and with most precipitation occurring between September and February (El Mouden et al. 1999; Znari et al. 2002; Ben Kaddour 2005). The average air temperature in the hottest month (July) can reach 39°C, and the minimal annual temperature is normally above 0°C (Emberger 1933; Le Houérou 1989). Vegetation mainly consists of scattered bushes of Jujube (Ziziphus lotus), with some widely separated Acacia (Acacia gummifera) (typical arid vegetation) and Retams (Retama monosperma). Most tortoise habitat is open, hard, bare ground, with stony soils on the flats and low hillsides that surround small sandy, pebbly, or stony wadis (riverbeds). Overgrazing by domestic livestock (sheep and goats) strongly affects the vegetation structure. The activity pattern of the tortoises is strongly shaped by climatic conditions. In the central Jbilets (T. Slimani and H. El Mouden pers. obs.) as in Spain (Diaz-Paniuagua et al. 1995; Braza et al. 1981; Andreu et al. 2000), tortoises are active from February to June and from September to November. Winter and summer activity remains highly sporadic under such a harsh climate (Lambert 1983; Bayley and Highfield 1996).
Citation: Chelonian Conservation and Biology 7, 1; 10.2744/CCB-0649.1
Captures and Measurements
Tortoises were searched for during the day and captured by hand. Immediately after capture, sex and morphologic measurements were recorded. Sex determination was possible in adults only. In tortoises, maturity is typically associated with a reduction of the width of the successive rings deposited each year on each scute (Castanet and Cheylan 1979; Stubbs and Swingland 1985; Germano 1994; Germano and Bury 1998; Willemsen and Hailey, 1999; Lagarde et al. 2001; Ben Kaddour et al. 2005). At maturity, males exhibit a larger tail, a typical concave plastron, and a convex supracaudal scute. Females have a short tail, a flat plastron, and a relatively flat supracaudal scute (Fig. 1b). By using scute annuli width as an indicator for maturity, we observed that, in the central Jbilets mountains, male T. graeca reach maturity at a smaller body size (mean, standard deviation, and range estimated carapace length: 109.7 ± 10.0 mm, 91.4–131.4 mm) compared with females (146.2 ± 16.5 mm, 114.6–171.8 mm) (Ben Kaddour et al. 2005). Based on sexual behavior, the minimum size at maturity was 104.8 mm (N = 9) for males and 152.0 mm (N = 8) for females (Ben Kaddour et al. 2005); although, this latter method likely missed most of the sexual behaviors because of small sample sizes and therefore, underestimated the size for maturity, the values remained comparable with those obtained with scute annuli width. The only method to reliably measure body size at maturity is to regularly monitor each individual in the course of the acquisition of sexual maturity (i.e., assessing sexual behaviors, gonad development, morphology) during a period that extends over months or years. This is an insurmountable task. Therefore, we rely on untested estimates and classify individuals as mature only when morphologic criteria (e.g., reduced annuli as a consequence of reduced growth over more than a year) become obvious and not necessarily when the individuals are actually ready for reproduction. Therefore, for simplicity, we included in the analyses only individuals larger than 100 mm in carapace length. By adopting such simplification, the only risk was an artificial attenuation of the observed SSD: several small mature males (<100 mm) were ignored and several immature small females (<152 mm) were incorporated in the analyses. However, our results suggest that the studied population exhibits a pronounced SSD compared with other populations. Consequently, our method was conservative with respect to the conclusions presented.
Each individual was permanently marked by using a code of notches in the external parts of the marginal scutes (Cagle 1939; Stubbs et al. 1984). Body mass was recorded with an electronic scale (1g accuracy). The size and the shape of the main morphologic characteristics of each individual were recorded as follows (Fig. 2):
Citation: Chelonian Conservation and Biology 7, 1; 10.2744/CCB-0649.1
-
Carapace or shell length (CL): maximal anteroposterior shell length.
-
Carapace or shell width (CW): maximal width at the level of the sixth (CW6) and eighth (CW8) marginal scute.
-
Carapace, or shell height: maximal height of the shell (H).
-
The curvilinear dorsal length of the shell carapace length (CCL): from the anterior tip of the shell to the end of the anal scute.
-
Plastron size: we recorded the maximal anteroposterior length of the plastron (Plmax), the minimal midline length of the plastron recorded in the notches (Plmid).
-
An estimate of the space available to move the tail laterally, and backward and forward between the posterior parts of the anal scutes (W), and between the rear parts of the plastron and the supracaudal scute (E). This later measurement also provided the most restrictive dimension for passing eggs (E).
-
Limb length was estimated by using the left antebrachium length (fore leg) from the proximal end of the ulna to the flat, ventral surface of the manus, and the left crus (hind leg) from the proximal end of the tibia to the flat, ventral surface of the pes.
-
Tail length was measured along its ventral edge, from the base of the external portion of the tail to the tip.
-
Head length, width, and depth correspond to the maximal external cranial length (from the base of the skull to the tip of the snout), to the widest part of the head, and to the maximal height of the head, respectively.
In addition to the above measurements required for comparison with previous results (Bonnet et al. 2001; Willemsen and Hailey 2003), we also measured the length of each of the 6 scutes along the ventral line of the plastron. We expected that the first (1, 2) and the last (5, 6) plastral scutes would be relatively shorter in males to provide space for the legs and the tail. These scutes cover the front part of the plastron (gular + humeral scutes cover the epiplastron and the front part of hypoplastron, based on examination of museum specimens) and the distal parts of the plastron (femoral and anal scutes cover the xiphiplastron and caudal part of the hypoplastron). They are relatively free because they are not attached laterally to the dorsal part of the shell. The intermediate plastral scutes (3, 4 = abdominal and pectoral scutes) are less directly involved in the determination of the openings, they are attached laterally to the dorsal part of the shell and contribute to the rigidity of the whole shell. Morphologic changes of the plastron should be less constrained on relatively free part of the plastron (1, 2, 5, and 6 plastral scutes) compared with the median part of the plastron (scutes 3 and 4). Linear measurements (e.g., plastron length or shell height) were measured with a calliper (precision 0.1 mm), the nonlinear measurements (CL) were gathered with a flexible rule placed along the midline of the shell (precision 1 mm).
Analyses
Each individual was represented only once in the data set. Comparisons of the body shape between the sexes where performed by using Analyses of Covariance (ANCOVAs), with sex as the factor (Garcia-Berthou 2001). However, we did not systematically use the linear shell length (CL) as a covariate. For example, in the presence of a significant size dimorphism (which was the case in the present study [see Results section]), the relative width of the head is better characterized relative to head length; similarly, the relative depth of the notches in the plastron would be better appreciated relative to plastron length. Although the total number of adult tortoises sampled was relatively large (117 females and 131 males), comprehensive measurements were not recorded on all individuals. For instance, the measurement of the length of the limbs was not easy, especially without hurting the animal (i.e., by pulling the limbs too firmly). Consequently, some of our results may have been affected by the small sample size. Therefore, we performed 2-tailed power analyses to gauge the β error of the tests and to calculate the required sample size to reach conventional statistical significance (α <0.05). We are aware that power analyses are primarily designed to set up designs and cannot compensate for small sample sizes. They, however, are useful to appreciate our capacity to reject the null hypothesis that there is no sex difference for a given trait.
RESULTS
Sexual Size Dimorphism
On average, females attained larger maximal and mean sizes than males (Table 1). This translated into greater absolute values for all the traits we measured, except for the tail (no difference) and the space available to move the tail (W + E were larger in males) (Table 1).
Sexual Dimorphism in Body Shape
The comparisons between the sexes of the size-corrected morphologic traits (= body proportions or body shape) are summarized in Table 2. Females displayed a more voluminous shell relative to the males. Notably, they exhibited a wider (CW6), and higher (H) shell relative to the length of the shell (CL), thereby providing a larger internal volume. Females were also heavier relative to the overall body size, and they exhibited a relatively wider head relative to head length.
In males, the plastron was smaller (Plmax relative to CL) and more indented (Plmid relative to Plmax) than in females; the openings between the plastron and the shell (W and E) were also more developed. The reduction of the plastron in males was attributable to a reduction of various plates relative to shell length (S2, S3, S4, and S6). The marked indentation of the plastron itself was mostly attributable to the reduction of the rear plates of the plastron (S6) providing additional space to move the tail in an already small plastron (Fig. 1b). Overall, the spaces available to move the legs, the head, and the tail (Plmid relative to Plmax) were larger in males.
We did not find any sex difference in the relative size of the legs. The large sample sizes required to reach statistical significance (Table 2) suggest that males and females actually exhibit similar relative limb sizes with a very weak trend for greater values in males. Females tended to have (not significantly, however) larger hind limbs relative to the fore limbs; the sample size for this analysis was limited, but the required sample size to obtain a significant effect was small.
Comparisons Between Populations and Species
At the population level, the degree of divergence in SSD was more variable than the degree of divergence in sexual body shape: female T. graeca were relatively larger than males in the Moroccan population compared with the Greek population (on average, the difference in sex divergence in size [shell length and mass] between the 2 populations was of 29%; Table 3). By contrast, all sex divergences in body proportion were more stable (3% on average for the 7 measurements of body proportions reported in Table 3).
DISCUSSION
As predicted, the direction of SSD varied more than sexual body shape dimorphism across species. In females, selection for fecundity favors characteristics that enhance the production of large clutches and/or large eggs (Wilbur and Morin 1988; Gibbons and Greene 1990; Iverson 1992b). In chelonians, a reduction of egg size, an increase of maternal body size, or a rounding of the maternal abdomen can convey such an increase in fecundity. In males, body size is influenced by sexual selection. Because large males are superior during male-to-male combats, a larger average body size in males relative to females is favored; such a trend has been documented in tortoises (Berry and Shine 1980) and in other taxa (Anderson 1994; Shine 1978, 1994). Despite the fact that male-to-male combats occur in the 4 species of Testudo studied (Lagarde et al. 2001, 2002; Willemsen and Hailey 2003; T. Slimani and H. El Mouden, pers. obs.), the direction of SSD was inconsistent across the species: females were larger in 3 species (T. horsfieldi, Testudo hermanni, and T. graeca) but a reverse trend was observed in the third species (Testudo marginata), with males attaining a greater maximal size than females (Willemsen and Hailey 2003). The degree of divergence in SSD was even more pronounced between the 2 populations of T. graeca (Morocco vs. Greece) than between species (e.g., T. graeca vs. T. hermanni in Greece; Table 3). Such instability in the degree of SSD highlights the difficulties to predict the direction, and degree of SSD at a small taxonomic scale (i.e., genus, populations). In contrast, the 3 independent studies constantly reported that males exhibit larger openings in the shell (notably a short plastron), which provides more space to move the limbs and the tail, and that they are also relatively lighter as expected from the male's mobility hypothesis (Bonnet et al. 2001; Willemsen and Hailey 2003; this study). Similarly, the 3 studies reported that females exhibit a greater relative body mass and a more rounded body shape associated with a larger relative volume of the shell compared with the males, especially in the rear part of the shell that holds the clutch (Bonnet et al. 2001; Willemsen and Hailey 2003; this study). Therefore, comparisons of body shape between the sexes provide a straightforward way to test the hypotheses related to the influence of sexual or fecundity selection on body shape at a taxonomic scale. There is no obvious advantage for males to develop a round body shape (e.g., wide shell + flat plastron) and for females to increase the size of the openings in the shell, thereby reducing their main antipredator protection. The sexual body shape dimorphism hypothesis can be confidently considered as the more parsimonious explanation for the patterns we observed. We further predict that reverse sexual dimorphism in body shape is likely to be far less common than reverse SSD.
No conflict arose from the comparison between the current analyses and the data gathered on the steppe tortoise (Table 2). However, the results were not fully identical: several sex differences observed in the steppe tortoise were not found in the Greek tortoise. The larger relative leg size, the larger relative head size, and the more domed shell of the males were solely observed in the steppe tortoise. Power analyses suggest that this was partly because of the small size of several samples but also suggest actual differences between the species (Table 2). Male steppe tortoises travel greater distances than females during the short period of activity (Lagarde et al. 2003b); by contrast, no such sex difference was detected in the Greek tortoise in Morocco (unpublished radio-tracking data). Consequently, although mobile and agile males should always have an advantage during courtship and copulation in chelonians, the potential advantage linked to the ability to travel great distances to find mates might not be a consistent rule. Male steppe tortoises display extremely vigorous sexual activity during the mating season, including violent combats with many males being flipped onto their backs, intensive head bobbing, strong bites, prolonged pursuits, and assiduous vocalizations. All these traits potentially favor the development of longer legs and of a larger head (Bonnet et al. 2001; Galeotti et al. 2005). On average, male steppe tortoises devote 30% of their activity to sexual behaviors; females display sexual activity during 3% of their activity time (Lagarde et al. 2003a). By comparison, male Greek tortoises are far less exuberant and spent less than 0.1% of their activity for sexual behaviors and do not differ in this respect to conspecific females (Lagarde et al. 2007). Such behavioral differences may explain the morphologic differences in relative leg and head size observed between the 2 species.
Conclusions
Although almost all studies on sexual dimorphism concentrated on body size, few focused on body shape. Body size (or body mass) is a very crude descriptor of the morphology. Greater attention to sexual dimorphism in body shape should be paid to better understand how selection shapes the respective morphology of the sexes.
(a) Adult female Testudo graeca graeca pictured in the central Jbilets mountains. (b) Plastral view of an adult male (right) and female (left). (Photos by Ben Kaddour).
Ventral view (a) and lateral view (b) of the shell of an adult male of Testudo graeca graeca in the central Jbilets. The main measurements recorded on the shell (see text for abbreviations) are indicated (drawings based on a corpse found in the study site). CL = Carapace length; CW6 = Carapace width at the level of the sixth marginal scute; CW8 = Carapace width at level of the eighth marginal scute; H = Carapace height; P/max = maximal antero-posterior length of the plastron; P/mid = minimal midline length of the plastron recorded in the notches; W = space available between the posterior parts of the anal scutes; E = space available between the rear parts of the plastron and the supracaudal scute.
