Assessing the population status of an imperiled species carries vital importance for its conservation and management (International Union for Conservation of Nature [IUCN] 2016). In addition to land clearing, habitat degradation, wildfires, and other threats (Moreira and Russo 2007; Couturier et al. 2011; Maxwell et al. 2016), overexploitation of reptiles for the pet trade may cause serious population declines (Auliya et al. 2016). Within reptiles, turtles and tortoises are the most threatened group (Turtle Taxonomy Working Group 2014). About 2 million wild tortoises and turtles were traded over the past 20 yrs (Luiselli et al. 2016). One of the target trade species, Hermann's tortoise (Testudo hermanni Gmelin 1789), comprises 13% of the world Testudo trade (Türkozan et al. 2008). Testudo hermanni is listed as “near threatened” at the global scale (IUCN 2018). Furthermore, T. hermanni is listed by the Bern Convention and European Habitat Directive, and international trade of the species is regulated by Convention for International Trade of Wildlife Fauna and Flora.

Hermann's tortoises are limited to Europe, the Balkans, and Turkish Thrace (Bertolero et al. 2011). There are 2 recognized subspecies: T. h. hermanni and T. h. boettgeri (Fritz et al. 2006). Testudo h. hermanni inhabits the western part of the Po Valley, while T. h. boettgeri inhabits European Turkey in the easternmost distribution of the species. The biology of the species is poorly known from European Turkey compared with elsewhere (Hailey and Loumbourdis 1988; Filippi et al. 2010; Cutuli et al. 2013; Couturier et al. 2014; Berardo et al. 2015). One exception is where Lambert (1982) reported on the growth, structure, and abundance of Testudo graeca ibera in south Anatolia (Antalya-Kemer region) and Thrace (Tekirda and Çanakkale). Later, C¸evik (1982) studied the taxonomy and morphology of reptiles in European Turkey, reporting on the general distribution and morphology of T. h. boettgeri. A brief study (15 d; Türkozan et al. 2005) on T. h. boettgeri did not assess the population status of the species.

The Turkish population was overharvested for the pet trade between 1974 and 2005 (Türkozan and Kiremit 2007), and human activities have largely destroyed their habitat as well. To bolster information for conservation of the species, we investigated 1) the current distribution and population status of this species in European Turkey (Fig. 1) and 2) the current demographic structure of the species, and we tested 3) the hypothesis that overharvesting from the illegal pet trade caused a reduction in body size.

View Figure

• Download as Powerpoint
• Download Figure

METHODS

We conducted this study in European Turkey (Fig. 1) from April to October 2014–2016 with a team of 3 researchers, spending 100 d in the field. Surveys were conducted between 0800 and 2000 hrs. The monthly surveys were consistent for each year and carried out between the 14th and 24th of each month. We surveyed almost all possible habitats where T. h. boettgeri was expected. We recorded the tortoise capture localities to within 5-m resolution using a Garmin GPS (Monterra) and mapped localities using ArcGIS 10.4. For the calculation of species occupancy, we made multiple surveys (at least 6 per site) at the same localities (22 sites) and gave the same effort for each survey site. The survey sites we covered were almost the same size, and we coded “0″ for the absence of species and “1″ for the presence of the species. We used a survey specific model (based on higher ΔAIC [Akaike information criterion] and lower AICwgt value) of PRESENCE 7.5 software (MacKenzie et al. 2006), with a single season model for estimating site occupancy probability and probability of detection of T. hermanni. We also recorded size measurements (length, mass) and estimated age of each individual. Tortoises were released at their capture sites. We log-transformed mass and straight carapace length (SCL) and carried out analysis of covariance on whether male and female regressions have the same rate of change in mass per unit size. The juvenileto-female ratio was used as a stability index in the T. hermanni population (Hailey et al. 1988) and recorded as 0.1 at the most disturbed site and as 4 in a protected area.

We measured SCL as the length between the most distal projection of the cervical scute and the most distal projection of the posterior supracaudal or marginal scutes. Measurements of SCL were taken with either wooden (± 1 mm) or dial (± 0.2 mm) calipers. We weighed tortoises using a digital scale (± 0.2 g) and used the massto-SCL relationship to calculate a condition index (CI) of each tortoise as described in Willemsen and Hailey (2002). The CI was calculated from body mass (M) of a tortoise compared with that predicted (M′) from the relationship between mass and length (L). Log(M/M′) is the best CI based on body mass that is equal to residuals from the regression of log M on log L (Willemsen and Hailey 2002). CI = 0 indicates that the observed mass is equal to the predicted mass and relative mass (i.e., 100 × observed mass/predicted mass). Less than about 80% is an indication of poor condition (Hailey 2000). Because distribution of residuals did not deviate from normality, we used parametric tests (general linear models analysis of variance) to assess CI differences between sexes and among months.

We used secondary sexual characteristics (concavity of plastron, tail length, and curve of supracaudal scute) to identify each animal's sex except individuals that we could not confidently differentiate. Shell damage and injuries of tortoises were also noted. We did not perform any additional health assessments on the tortoises. Furthermore, we counted the growth annuli on the carapace assuming that 1 growth annulus was added each year (Germano and Bury 1998). Growth annuli counts are reliable methods for assessing the tortoise ages in T. hermanni (Bertolero et al. 2011) up to 20 yrs. The first ring on the scute was considered age 0, and then each scute was counted and recorded as the age of tortoise. After that, age was categorized as 0–5, 6–10, 11–15, 16–20, and > 20 yrs.

We used SPSS 16.0 and Statistica 8.0 for all statistical analysis and graphs were produced using Prism 6. All data were checked for normality before the application of parametric tests. Means and standard deviation are reported (mean ± SD).

RESULTS

Distribution and Habitat. — We collected data from 283 T. hermanni (135 males, 102 females, 45 juveniles, and 1 unknown) from 34 localities (Fig. 1). Testudo hermanni was syntopic with T. graeca at 23 sites. The general distribution included areas around human settlements, industrial zones, and fields but was limited to habitat with shrubs and within forest zones, including forest edges. The most common bush observed in the distribution area was crown of thorns, Palluris spinacristii. Elevation of capture sites ranged from 22 to 380 m above sea level (ASL). The southernmost distribution of T. hermanni was Terziköy, and no individuals were observed on the Gallipoli Peninsula. Tortoises were not found in the central part of European Thrace owing to extensive development of industrial zones. The probability that the species was detected at least once at survey sites (22 sites) was 94%, and 29% of the sites had at least 1 T. hermanni. We recorded 2.8 ± 4.9 tortoises/d, ranging from 0 to 25 tortoises.

Most tortoises (n = 92) were captured in May (35%) and July (64 tortoises; 24%). Males were active in almost all months before midday, and some were observed in the afternoon. However, females were most active in May before midday and had decreased activity observed in the afternoon.

Population Structure and Morphometry. — The sex ratio of T. hermanni was male biased with 1.3 males per female (p < 0.05). The mean SCL was 153 ± 16 mm for males and 175 ± 22 mm for females (Table 1). The peak frequency distribution of SCL ranged between 141 and 160 mm (43%) for males and 161 and 180 mm (41.2%) for females (Fig. 2). The mean mass was 851 ± 226 g for males and 1222 ± 343 g for females, with the peak frequency distributions ranging between 751 and 1000 g (43%) for males and 1001 and 1250 g (28.4%) for females (Fig. 3). Females were larger (t = 8.34, p < 0.0001, n = 227) and heavier (t = 9.64, p < 0.0001, n = 218) than males. The juvenile–to–adult female ratio (stability index) in the present study was 0.44.

Table 1. Mean, range, standard deviation (SD), and n for male, female, and juvenile Testudo hermanni boettgeri shell size, body mass, environmental temperatures, and relative humidity when detected in the field during 2014, 2015, and 2016 surveys of European Turkey.

View Fullscreen

View Figure

• Download as Powerpoint
• Download Figure

View Figure

• Download as Powerpoint
• Download Figure

There was a relationship between growth annuli and SCL in the form of exponential growth for males (F_1,64 = 9.82, R² = 0.13, p < 0.01) and females (F_1,38 = 11.59, R² = 0.23, p < 0.01); juveniles showed a stronger relationship (F_1,31 = 88.99, R² = 0.74, p < 0.001) (Fig. 4). There was a positive correlation between log SCL and log mass (F_1,213 = 4147, R² = 0.95, p < 0.001), with no effect of sex and SCL interaction on mass (F_1,212 = 2.77, R² = 0.01, p > 0.05). Furthermore, the regressions for males and females have a significantly different rate of change in animal mass per unit size (F_1,213 = 14.15, p < 0.001). meaning that males and females gain mass at different intervals in their life history (Fig. 5).

View Figure

• Download as Powerpoint
• Download Figure

View Figure

• Download as Powerpoint
• Download Figure

The CI of tortoises ranged from -0.18 to 0.14 with a mean CI of 0.48 ± 0.41 in females and from -0.09 to 0.15 with a mean CI of 0.69 ± 0.36 in males (Fig. 6). The mean CI varied between females and males (F_1,214 = 19.75, p < 0.001).

View Figure

• Download as Powerpoint
• Download Figure

Based on annuli, 108 tortoises (38.3%) were within the age group of 11–15 yrs, while 16 tortoises (5.7%) were older than age 20. Only 25 (8.9%) tortoises were within the 0–5 age group. Age was normally distributed in the population and did not differ between sexes (x²₄ = 3.043, p < 0.05). The interaction between sex and age group was important (F_4,201 = 1.384; p < 0.05).

Of the 283 tortoises encountered, 254 (89.8%) appeared to be physically healthy, 12 (4.2%) were injured, and 17 (6%) were dead. The mean SCL of dead tortoises was 162.8 ± 27.8 mm (range = 105–205), meaning that they were mostly adults. Of the dead tortoises, 7 were male, 7 were female, 2 were juvenile, and 1 was of unidentified sex.

DISCUSSION

Distribution and Habitat. — The habitat types of T. hermanni were similar to those recorded in previous studies (Rozylowicz and Popescu 2013; Couturier et al. 2014; Berardo et al. 2015). The highest occurrence of T. hermanni was in patchy landscapes and shrublands, while the lowest occurrence was in artificial areas, such as vineyards and arable lands (Couturier et al. 2014). The vertical distribution of some eastern European populations occurs up to 1300 m but occurs mostly under 500 m (Bertolero et al. 2011), which is similar to findings from Willemsen and Hailey (2001). Within our study sites, the distribution range of T. hermanni was under 500 m ASL. The highest elevation on the European side of Turkey does not exceed 1000 m. The vertical distribution range of mostly under 500 m in Europe may be related to more forage provided by field crops on the plains.

As reported in a previous work (Türkozan et al. 2005), we did not observe T. hermanni on the Gallipoli Peninsula, while T. graeca occurred in the area. Otherwise, both species show syntopic occurrence in many parts of European Turkey. The southernmost range was Terziköy. The absence of T. h. boettgeri from the Gallipoli Peninsula could be attributable to different climatic characteristics. Based on ecological zones (Atalay 2002) corresponding to main climatic types, the northern border remains in the Marmara transitional zone, while the southern border, including the Gallipoli Peninsula, remains in the Aegean ecological zone.

The male-biased population structure reported from Greece (Hailey et al. 1988), former Yugoslavia-Montegenero (Meek 1985), and Bulgaria (Ivanchev 2007) was attributed to the preference of males to inhabit open areas, making them easier to find. This was observed during the mating period (April and May) as males were looking for females. Furthermore, the higher density of males might have limited the occurrence of females (Hailey and Willemsen 2000). Because females are larger than males, there could be a female observation bias that is countered by males being more active in open areas. Based on the temperature-dependent sex determination mechanism in many turtles (Pieau 2002), higher temperatures during egg development produce more female hatchlings, while cooler temperatures produce more male hatchlings (Pieau 2002). It is possible that temperature may be producing a male bias. Furthermore, it could be the result of other factors, including differences in the timing of maturity between the sexes where the earlier maturing sex dominates numerically as adults (Lovich and Gibbons 1990). Another possibility is that the nail on the tip of the tail of males causes serious injuries in females during mating and increases the mortality rates of females (Hailey and Willemsen 2000).

The stability index in the T. hermanni population (Hailey et al. 1988) was 0.1 at the most disturbed site and as 4 in a protected area. Our index (0.44) is quite low in comparison to other studies (Hailey et al. 1988; Filippi et al. 2010), indicating a risk of population decline probably due to the suppression of recruitment within the region. Likewise, Ivanchev (2007) reported a 0.19 index ratio in Bulgaria and took this as a serious signal of population decline.

Population Structure and Morphometry. — Eastern T. hermanni populations (T. h. boettgeri) were reported to be larger than western (T. h. hermanni) populations (Bertolero et al. 2011). Sacchi et al. (2007) and Filippi et al. (2010) recorded that the size of T. h. hermanni follows Bergmann's rule in Italy, meaning that the size and shape of both sexes changed along a north-to-south cline. We found that females were significantly larger and heavier than males, and this is consistent with all previous studies (Meek and Inskeep 1980; Corti and Zuffi 2003; Filippi et al. 2010). The mean SCL of males was significantly smaller (t = –4.58, p < 0.001) than that of Yugoslavia-Montenegro T. h. hermanni (Meek 1985; Meek and Inskeep 1980), while no difference was observed in females (t = 0.307, p > 0.05). Conversely, both males (t = 5.93, p < 0.001) and females (t = 4.78, p < 0.001) were significantly larger than those from Croatia (Meek 1989). Willemsen and Hailey (1999) stated that the SCL of the T. h. boettgeri might be smaller in areas where they are syntopic with other tortoises because of character displacement. Close and Seigel (1997) found that different levels of harvesting in red sliders, Trachemys scripta elegans, created a significant difference in body size indicating smaller body size in harvested populations.

Our measurements are within the range of previous studies, and there is no clear negative impact of illegal trade on the size of T. h. boettgeri in European Turkey. However, this does not preclude the potential impact of illegal trade on the numbers and stability of populations in this region. Additional research on population size estimates and long-term population assessments are needed to fully assess the stability of European Turkey populations of T. h. hermanni.

The correlation of growth annuli and carapace length was previously reported in the form of allometric growth for different populations of T. hermanni (Meek 1985; Živkov et al. 2007). The CI of the present study is within the limit of CI index previously defined from different T. hermanni populations. Sexual and seasonal variation of CI was also reported from France (Sibeaux et al. 2016). Habitat conditions, food availability, activity, and dehydration during summer drives CI (Willemsen and Hailey 2002; Lecq et al. 2014; Sibeaux et al. 2016). Geographic variation in mass-to-length relationship did not affect the CI of T. hermanni in Greece (Willemsen and Hailey 2002).

Both Meek (1985) in Montenegro (then Yugoslavia) and Ivanchev (2007) in Bulgaria recorded mainly older individuals (> 20 yrs) in the population. In contrast, our population is composed primarily of the 11–15 yrs age group. The age at sexual maturity was found to be related more to size than to age (Hailey 1990). Adult males 130 mm or larger (8–11 yrs) are considered to be sexually mature (Hailey 1990); therefore, our population consists mainly of sexually mature individuals.

In conclusion, recent findings on the population ecology are not enough to prepare an effective conservation and management plan for the species. The genetic diversity of the population along with a test of inbreeding level should be investigated in order to see if there is a need for introduction of new individuals from neighboring populations that have similar genotypes.