Turkey harbors a substantial part of the genetic and taxonomic diversity of the widely distributed European pond turtle (Emys orbicularis; Fritz 2003; Fritz et al. 2009; Vamberger et al. 2015; Pöschel et al. 2018). The nominotypical subspecies Emys orbicularis orbicularis (Linnaeus 1758) is distributed across most of Turkey, while 2 endemic subspecies are restricted to southern and southeastern Turkey (Fritz et al. 2009). Along the southcentral coast of Turkey, an undescribed subspecies occurs. Further east, near the Syrian border, Eiselt's pond turtle (Emys orbicularis eiseltiFritz et al. 1998) is distributed, which is one of the least-known subspecies of the European pond turtle.

According to the original description (Fritz et al. 1998), Eiselt's pond turtle is a small-bodied and dark-colored subspecies endemic to the region east of the Amanos Mountains (Nur Dağları). This subspecies was described from Gaziantep Province, based on only 4 museum specimens collected at 2 different sites in 1966 and 1972 (Appendix 1). Despite some morphological differences (see Fritz et al. 1998), the only additional record that most likely also refers to Eiselt's pond turtle is another museum specimen collected in 1980 nearby in Syria (Kinzelbach 1988), about 40 km from the Turkish border. It represents the southernmost unambiguous record for European pond turtles in this region (Fritz et al. 1998). This juvenile Syrian specimen originates from the Asi River drainage. During the present study, we found in this drainage typical representatives of E. o. eiselti farther north, so it is likely that its somewhat divergent morphology represents simply ontogenetic variation.

Except for a genetic assessment based on mitochondrial DNA (mtDNA) sequences (Fritz et al. 2009), no additional information on E. o. eiselti has been published since its description. Based on genetic samples of the specimens reported herein, Fritz et al. (2009) discovered that a distinct mtDNA lineage corresponds to E. o. eiselti; yet, unpublished data using nuclear genomic microsatellite loci confirm its genetic distinctiveness (M. Vamberger et al., unpubl. data). In this study, we present our results of a 20-yr-long search for E. o. eiselti in southeastern Turkey and report for the first time additional morphological data.

METHODS

Study Area and Survey Approaches. — Fieldwork in the presumed distribution area of E. o. eiselti was conducted in the years 2001 (5–10 September), 2007 (18–23 June), 2008 (12–17 August), 2011 (1–5 September), 2012 (29 and 30 June, 1 and 2 August), 2013 (22 and 23 June, 7 and 8 September), 2015 (27 May–4 June, 23 June–3 July, 1–10 September), 2017 (4–16 July, 10–25 August, 9–12 October), and 2019 (10–12 July), including extensive explorations for suitable habitat (Appendix 1). In total, we spent 99 days in the field searching for Eiselt's pond turtle. In addition, we talked with locals to collect information on the possible occurrence of pond turtles.

Our fieldwork focused on the region east of the Amanos Mountains (i.e., the Amik Maraş Rift Valley, where the type series originated), and extended eastward to the Euphrates River (Fig. 1). The type locality (14 km NE of Fevzipaşa, Gaziantep Province), where the holotype and one paratype were collected in 1966 (Eiselt and Spitzenberger 1967), and the collection area of the two other paratypes from 1972 (between Kömürler and Sakçagözü villages, Gaziantep Province; Fritz et al. 1998) were visited in 4 yrs (2001, 2007, 2008, 2011). In addition, we searched for E. o. eiselti in many other sites in southeastern Turkey and visited several localities repeatedly (Appendix 1). Except for the collection sites of the type series, potentially suitable fieldwork locations were preselected (e.g., because of the presence of water bodies according to maps). In total, 109 different sites were examined. If freshwater turtles were recorded, they were captured either by hand or in fyke nets. Captured turtles (besides Eiselt's pond turtles, Mauremys caspica and M. rivulata) were released to their natural habitats after measurements and blood samples were taken. Genetic results for Mauremys have been published elsewhere (Fritz et al. 2008; Vamberger et al. 2013, 2014, 2017).

View Figure

• Download as Powerpoint
• Download Figure

Inference of Historical and Present Distribution Range. — The extent of the historical as well as the presently remaining distribution range of E. o. eiselti was estimated using a hydrologic unit compartment (HUC) approach. Hydrobasin layers with a resolution of 15 arc-seconds (approx. 500 m at the equator) were obtained from https://www.hydrosheds.org (Lehner and Grill 2013). Following Buhlmann et al. (2009), HUCs with present and historical occurrences of E. o. eiselti were selected and connected to HUCs within the same watersheds or to physiogeographic regions with similar habitats and elevations.

Morphology and Descriptive Statistics. — The 14 recorded morphometric traits for E. o. eiselti were those of Fritz (1994, 1995). Measurements were taken to the nearest 0.1 mm using a caliper: SCL (maximum straight-line carapace length), CW (maximum carapace width), CH (maximum carapace height), PL (midline plastron length), GuL (intergular suture length), HumL (interhumeral suture length), PecL (interpectoral suture length), AbdL (inter-abdominal suture length), FemL (interfemoral suture length), AnL (interanal suture length), HL (head length), HW (head width), NuL (nuchal length), and NuW (nuchal width). All measurements were straight-line. PL/ GuL × 10 ratios were calculated for comparative purposes. Coloration was classified according to the categories of Fritz (1994, 1995). Sexual dimorphism of Eiselt's pond turtles was tested for normality using a Kolmogorov-Smirnov Z-test. If the data were normally distributed, an independent t-test was performed; otherwise, a Mann-Whitney U-test was used for comparing sexes. The significance level was set at p = 0.05. All statistical analyses were run in SPSS 18.0 software.

RESULTS

Survey Results. — We recorded E. o. eiselti only at 4 sites in the region of the Asi (Orontes) River mouth. In the same habitats there were stripe-necked terrapins (M. rivulata), which outnumbered by far the pond turtles. Although we found only 18 E. o. eiselti across 5 fieldwork years (2001, 2007, 2008, 2011, 2012), we captured no less than 390 M. rivulata in the same water bodies. We marked all pond turtles with marginal notches and did not recapture any individual. In contrast, we cannot exclude recaptures for the stripe-necked terrapins. The remaining 105 study sites harbored either no turtles (because no suitable habitat remained) or only M. caspica or M. rivulata were present (Fig. 1; Appendix 1). All E. o. eiselti were fully grown adults, 10 males and 8 females, and no additional adults or juveniles were sighted.

We did not encounter any pond turtle at the two sites where the type material was collected in 1966 and 1972. These localities are completely devoid of surface water today (Fig. 2). Also, our efforts to find E. o. eiselti at other potential occurrence sites failed, except for the vicinity of the Asi River mouth, where we captured the 18 adults at 4 distinct sites. Our conversations with locals suggest that E. o. eiselti presently does not occur elsewhere, even though people were generally familiar with turtles (but only with Mauremys). Therefore, we believe that E. o. eiselti currently is restricted to the Asi River mouth and that the number of mature individuals does not exceed 100. This estimate is based on the obviously very low local population density and our extensive expertise with mark–recapture studies for E. orbicularis elsewhere in Turkey.

View Figure

• Download as Powerpoint
• Download Figure

Inference of Historical and Present Distribution Range. — In addition to sites in the presumed former range of E. o. eiselti in the Amik Maraş Rift Valley, we also searched farther east, up to the Euphrates River (Fig. 1; Appendix 1), but without success. The whole region suffers from large-scale alterations of the landscape, especially from the construction of dams and channels, which are densely inhabited by M. caspica. In any case, these human disturbances and possibly the direct competition with M. caspica may have contributed to the disappearance of E. o. eiselti, assuming it had once been present in these more eastern regions. However, it remains questionable whether E. o. eiselti ever occurred east of the Amik Maraş Rift Valley. For stripe-necked terrapins (M. caspica, M. rivulata), living in Turkey widely syntopic with E. orbicularis, the watershed between the Asi River and other drainage systems in the Amik Maraş Rift Valley and the Euphrates constitutes the divide between M. rivulata and M. caspica, respectively (Fritz and Wischuf 1997; Vamberger et al. 2017). This separation suggests that the original distribution of E. o. eiselti in the region paralleled that of M. rivulata and that E. o. eiselti did not occur in the Euphrates drainage. Therefore, we based our estimate of the former distribution range only on the drainage systems of the rift valley (i.e., west of the watershed; Fig. 1). This area corresponds to 11,612 km², whereas the presently occupied area covers only 206 km².

Morphology and Descriptive Statistics. — Our new data confirm that E. o. eiselti is small-sized, even though Pritchard (1966) mentions a larger female (15 cm shell length) from the vicinity of Reyhanlı (Hatay Province; see Appendix 1). However, it is unclear whether Pritchard (1966) reported a straight-line measurement or whether he measured the 15 cm over the curve of the carapace. In our males the mean SCL was 109.3 ± 2.7 mm (range, 95–127 mm; SD 8.7). Females were slightly larger, with a mean SCL of 118 ± 4.3 mm (range, 90–130 mm; SD 12.2). These data are similar to those of the type specimens (males: SCL 115 and 129 mm; females: SCL 122 and 131 mm; Fritz et al. 1998). As in the type material, the gular scutes were small in all 18 new specimens, with a midline seam length of 14.5 ± 0.4 mm (range, 13–16.3 mm; SD 1.2) in males and 16.8 ± 0.6 mm (range, 13.1–18.4 mm; SD 1.7) in females. The mean ratio PL/GuL × 10 for all measured specimens was calculated as 64.7 ± 1.3 (range, 56–72; SD 5.34), matching the value of the type series (compare fig. 6 in Fritz et al. 1998: approx. 64). Descriptive statistics for morphometric characters are summarized in Table 1. Some characters (NuL, NuW, GuL, FemL, HW, HL) were normally distributed and others were not (SCL, CW, CH, PL, HumL, PecL, AbdL, AnL). Sexual size dimorphism was identified for 8 of 14 measurements, with males being significantly smaller than females (SCL: Mann-Whitney U-test, U = 26, p = 0.025; CH: U = 23, p = 0.014; PL: U = 18.5, p = 0.004; GuL: ttest, t = 2.98, df = 20, p = 0.007; HumL: U = 23, p = 0.014; PecL: U = 26, p = 0.025; AbdL: U = 11, p < 0.001; AnL: U = 10, p < 0.001).

Table 1. Descriptive statistics for morphometric measurements (in mm) of new material of Emys orbicularis eiselti (see “Methods” for definition of abbreviations).

View Fullscreen

As outlined in the original description of E. o. eiselti for the type series (Fritz et al. 1998), all of our 18 new turtles were also very darkly colored, with a predominantly black carapace and plastron (‘orbicularis coloration morph’), with the exception of one male that had a predominantly light-colored plastron (Fig. 3). Otherwise, females had larger yellow portions on the plastron than did males, and the light shell pattern in males, if present, consisted of spots, whereas that of the females was rather streaked. Older individuals were more darkly colored than were younger adults. The irises in males were intensely reddish, while they were yellowish to whitish in females.

View Figure

• Download as Powerpoint
• Download Figure

DISCUSSION

It seems likely that Eiselt's pond turtle was originally widely distributed across the Amik Maraş Rift Valley east of the Amanos Mountains (i.e., in the northernmost part of the Dead Sea Rift). The Amik Maraş Rift Valley once harbored large marshlands. Lake Amik (Lake Antioch) was located in the central rift valley and was originally fed by the Afrin and Karasu rivers (Fig. 1). The type series of E. o. eiselti was collected in the catchment basin of the latter river and 2 records from 1963 (Pritchard 1966; 10 miles N Reyhanlı, Hatay Province) and 1966 (Eiselt and Spitzenberger 1967; 14 km E Kırıkhan, Hatay Province) originated from the vicinity of Lake Amik. Originally, Lake Amik, its tributaries, and the surrounding wetlands formed an important marshland system that was connected to the Asi River. Today, this marshland system is largely destroyed.

For nearly 80 yrs, wetlands in Turkey have been systematically drained for various reasons, such as agriculture, urbanization, road construction, and prevention of malaria. Lake Amik exemplifies this situation. According to Kumerloeve (1988), the lake covered an area of approximately 35,000 ha in the 19th century and receded to 30,000 ha at the beginning of the 20th century. The draining of Lake Amik intensified in the 1960s. In 1966, the surface area of the lake was reported as 7000 ha by the State Hydraulic Works (Çalışkan 2008; Özelkan et al. 2011). Water from the lake and its tributaries was discharged to the Asi River by creating new channels toward the river, and the habitats of E. o. eiselti and other freshwater species dwindled away. In 2007, Hatay Airport was constructed in the center of the former lakebed. Today, only the Asi River and its tributaries serve as a refuge for many freshwater biota. Our new records for E. o. eiselti are all restricted to the zone of the river mouth, where the Asi River discharges into the Mediterranean Sea. In Syria, large parts of the Asi River are channelized and the neighboring wetlands have been transformed to intensely used farmland and settlements, including the site where a pond turtle was collected in 1980 (Kinzelbach 1988). As satellite photos on Google Maps show, only the immediate border region to Turkey could offer suitable habitat for E. o. eiselti, suggesting that an isolated population could still exist there. Outside the Asi River drainage, we could not confirm the presence of E. o. eiselti in its former northern distribution range; thus, the subspecies seems to be extinct there.

Based on our field data and the inferred historical and present area of occupancy, E. o. eiselti is now restricted to a region of only 206 km² compared with 11,612 km² prior to the draining of the wetlands in the Amik Maraş Rift Valley (Fig. 1; i.e., only approximately 2% of the historical range still harbors Eiselt's pond turtles). Even if the remaining range were to double, if pond turtles should still live in Syria near the Turkish border, it would not change the general situation. With an estimated number of less than 100 mature individuals left in the wild in Turkey, the subspecies fulfills the International Union for Conservation of Nature (IUCN) criteria for the threat category Critically Endangered (IUCN Species Survival Commission 2012), to which it should be assigned.

Our data indicate that E. o. eiselti is on the brink of extinction. The Asi River mouth is its last known bastion and we suggest that conservation and management measures be developed by the respective authorities and stakeholders with utmost priority to protect these habitats before it is too late. Additional surveys for suitable habitat should be undertaken, especially along the Asi River in the Turkish–Syrian border region and in adjacent Syria, as soon as the political situation allows. In addition to these in situ measures, establishing a captive breeding program could be a promising security strategy for rescuing Eiselt's pond turtle from extinction.