In recent decades, a decline in reptile populations has been observed on a global scale, with nearly 1 in 5 reptilian species threatened with extinction (Böhm et al. 2013). Several sometimes concomitant factors are believed to be the main drivers of the regression of reptile species, such as habitat destruction and fragmentation, environmental pollution, disease, climate change, and the introduction of invasive alien species (Gibbons et al. 2000). Despite protection strategies, this pattern of decline is detected even in some protected areas (Gibbons et al. 2000; Reading et al. 2010). Like many reptile species, greater than 50% of turtle species are threatened (Böhm et al. 2013).

Relocations, repatriations, and reintroductions are becoming increasingly popular tools to reverse the negative trends observed for many species worldwide (Griffith et al. 1989; Dodd and Seigel 1991). Although long-term monitoring is necessary to assess the outcomes of such programs, in practice it is rarely conducted (Bertolero and Oro 2009). Despite the lack of empirical data to assess the effectiveness of translocation and reintroduction programs, they are still carried out for many taxa, including reptiles (Seddon et al. 2005). In terms of turtles, there have been several reintroduction programs of, for instance, the Seychelles giant tortoise, Geochelone gigantea (Hambler 1994), the gopher tortoise, Gopherus polyphemus (Turberville et al. 2005), the Hermann's tortoise, Testudo hermanni (Bertolero et al. 2007), and the Spanish pond turtle, Mauremys leprosa (Bertolero and Oro 2009). Yet one-third of all animal reintroduction programs fail to create a viable population or successfully augment an existing one (Tavecchia et al. 2009). Moreover, some studies have reported lower success rates for reptiles and amphibians (19%) compared with mammals and birds (44%; Dodd and Seigel 1991). These success rates may even be an overestimate given that not all failures may be reported.

The success of a reintroduction program depends on several ecological factors, such as the quality of the habitat chosen for the reintroduction and the genetic pool of the reintroduced individuals (see references in Fischer and Lindenmayer 2000). However, the low number of reptile reintroduction programs (Seddon et al. 2005) and the rarity of published results reporting the causes of their failure or success make it difficult to highlight the general principles for effectively reintroducing reptiles.

One crucial element in permitting the viability of the reintroduced species is to provide favorable habitat conditions (good intrinsic environmental quality in terms of vital resources, limited predation, moderate competition, and so on). In France, in order to obtain state authorization for reintroduction, whatever the species, the quality of the chosen site must be assessed before release. This evaluation of site quality is carried out following International Union for Conservation of Nature reintroduction guidelines (International Union for Conservation of Nature 1998; Bertolero and Oro 2009). However, to be relevant, this process requires good knowledge of the chosen site, as well as of the species' biology and ecology, which is often lacking for reptiles (Burke 1991).

Another factor to consider is that some reptiles, especially chelonians, are long-lived species with a slow population growth rate. Thus a true assessment of the success of a reintroduction program requires population monitoring over decades (Congdon et al. 1993). However, short-term evaluation is necessary to determine if it is worth pursuing the program or if it will eventually fail. With that aim, several postrelease studies have simultaneously integrated demographic parameter estimates and an evaluation of the use of space by the released individuals, which can quickly indicate potential acclimation problems and whether or not the site covers all the requirements of the species (e.g., Bennett et al. 2012; Bodinof et al. 2012; Nussear et al. 2012). Such studies can also signal aberrant behavior such as dispersal outside the release site, large movements, or in contrast, underutilization of the available space, behaviors that can be an early warning of potential reintroduction failure (Germano and Bishop 2009). Short-term evaluations are particularly relevant if it is possible to compare reintroduced individuals to wild populations (e.g., Rittenhouse et al. 2007). They also allow rapid reaction to problems by means of habitat management actions or even stopping the program if the basic species' requirements do not seem to be fulfilled.

The European pond turtle, Emys orbicularis, is the only Old World member of the diverse family Emydidae. It is widely distributed across Europe and western Asia (Spinks and Shaffer 2009). Present in Europe since the late Pliocene, today this turtle is in a critical demographic situation in different parts of its European distribution (Fritz et al. 2009). In central and southern European countries, the destruction or progressive alteration of wetlands due to the intensification of agricultural practices, drainage, urbanization, and damming of rivers is the main cause of this species' decline (Roques et al. 2006). Matson et al (2005) also suggest that pollution may reinforce its decline, and Cheylan and Poitevin (1998) point out that Mediterranean populations are additionally threatened by wildfires. Furthermore, the recent introduction of the red-eared slider, Trachemys scripta elegans, appears to be an aggravating factor in many countries (Cadi and Joly 2004). The red-eared slider occupies the same ecological niche as the European pond turtle and directly competes with it for food, nesting sites, and basking places (Cadi and Joly 2004).

To limit the large-scale and rapid decline of the European pond turtle, many protection programs have been established, especially in protected areas that host large numbers of the species (Ficetola et al. 2004). In addition, many reintroduction programs have been implemented across Europe, notably in Hesse, Germany (Schweitzer et al. 2008), Switzerland (Monney and Meyer 2008), Italy (Ficetola et al. 2013), Slovakia (Bona et al. 2013), and the Savoy region of France (see below). In Germany, the reintroduction program has been evaluated using radio-tracking and capture–recapture methods. Several parameters were monitored, such as body mass development and daily and seasonal activity and dispersal. In the French program conducted in Savoy, radio-telemetry monitoring allowed survival rate, area fidelity, and distance covered to be evaluated. In both Germany and France, the results are encouraging because the reintroduced individuals permanently settled in the chosen sites.

In France, the decrease in the number of European pond turtle populations and their high fragmentation has led to the establishment of a national conservation plan for the species (Thienpont 2011). One of the actions promoted by this plan is the augmentation of existing populations and the reintroduction of the species. In Savoy, 37 turtles were reintroduced between 2000 and 2002 south of Lake Bourget and a second group of 28 individuals was introduced in the spring of 2009 north of the lake (Thienpont 2011). In 2006, a population reinforcement program was launched in the Mazière Nature Reserve (in the region of Aquitaine) with about 100 individuals released (Thienpont 2011). Another reintroduction program is currently underway in Alsace, where the first nucleus of a population was released in 2011 at the site of Woerr in Lauterbourg (Department of Bas-Rhin, France).

Our study focused on a recent reintroduction program in Languedoc-Roussillon (southern France) at the Estagnol Nature Reserve, where 30 individuals were released in 2008 and 2009. Our aim was to provide an initial short-term evaluation of the reintroduction program. To this end, we studied the fixation rate, spatial use of the site, and movement patterns of the released individuals over 2 yrs of monitoring at the site. We then compared our results with those obtained in previous studies of wild populations to determine whether or not the reintroduced individuals had adapted to the release habitat.

METHODS

RESULTS

The mean of number of relocations per turtle was 45.5 (SD  =  24.07, range  =  13–86). For lot 1, the mean number was 60.6 (SD  =  25.03, range  =  13–86), while the mean for lot 2 was 31.07 (SD  =  8.53, range  =  15–51). The mean duration of individual monitoring was 13.93 mo (SD  =  7.04, range  =  2–29). For lot 1, the mean was 16.53 mo (SD  =  8.11, range  =  2–29). For lot 2, the mean was 11.93 mo (SD  =  4.30, range  =  2–16).

Fixation Rate

The fixation rate is the percentage of released individuals that are still present on the site. For the first group of 15 individuals released in April 2008, 11 individuals were seen in the nature reserve at least once between November 2008 and January 2009 (i.e., a fixation rate of 0.73). Among this first group, 12 were also seen between April and June 2009 (i.e., a fixation rate of 0.80 1 yr after release). Among these 15 individuals, only 5 were seen in March and April 2010. This low number was expected since the radio-transmitters' maximum longevity was estimated at 18 mo. Among this first group, 1 individual was recovered dead, having been run over by a car (July 2008). The discrepancy observed between 2008 and 2009 regarding the number of turtles (11 in 2008 and 12 in 2009) is due to the fact that 1 turtle temporarily left the study site in 2008 and came back in 2009.

For the second group of 14 individuals released in April 2009, 10 were observed in the nature reserve between November 2009 and January 2010 (i.e., a fixation rate of 0.71). In March and April 2010, 8 of these individuals were detected (i.e., a fixation rate of 0.57 1 yr after release). Note that in the second year, the field session stopped in April, so some individuals might have been detected later in the season. Due to transmitter failures, all these values are the minimal possible fixation rates.

Traveled Distances

All turtles were included in the distances analysis, namely 30 turtles. The model that best fit the whole data set included an interaction between month and year; the the second-best model added the effect of sex in interaction with the best model (3-way interaction with month and year; Table 1). The inclusion of sex as an effect decreased by more than 2 points the AIC and could thus be considered as significant (Burnham and Anderson 2002). This suggests that the traveled distances varied highly over the months, that males and females exhibited a slightly different pattern of monthly displacement variation, and that these relationships differed between the 2 studied years. The analyses conducted separately for the 2 studied years confirmed that in the first year, the 2 sexes exhibited different monthly traveled distances (best models included the interaction between sex and month, Table 1), while in the second year, the 2 sexes behaved similarly (no sex effect in the best model, Table 1). The comparison of the traveled distance patterns exhibited by the 2 released groups in the second year, i.e., when they were simultaneoulsy present in the reserve, shows no difference between the 2 groups (Table 1).

Table 1. Model comparison fitted on the distances traveled by European pond turtles reintroduced in the Estagnol Nature Reserve in 2008 and 2009. s(Month) is the smoothing covariate of the Generalized Additive Model.

View Fullscreen

The examination of the GAM estimates showed that the mean traveled distances were of the same order for the 2 sexes. It also showed that the pattern of monthly displacement variation was sharply different between the 2 yrs and rather similar among sexes. The first year, displacements reached a maximum in April and then rapidly decreased and stayed low until hibernation (Fig. 1a, c). The second year, travel distances were high in April, only slightly decreased thereafter, and reached a second peak in September (Fig. 1b, d). The main difference between sexes was detected the first year only. The females in the first year showed a peak of displacements in April, gradually decreasing until September–October (Fig. 1a). In contrast, the displacements of males over the first year rapidly decreased from a maximum in April to reach very low values in July and August, followed by a very small increase in October and a rapid decrease in November (Fig. 1b).

View Figure

• Download as Powerpoint
• Download Figure

Home Ranges

For the home range analysis, we used 20 turtles that were detected at least 25 times: 11 from the first group and 9 from the second group. The model that best fit the home ranges of the individuals the first year after their release included only a group effect (Table 2; note that this group effect was identical to a year effect). The mean home range of the first released group was 3.88 ha (95% confidence interval [CI] 1.86–8.13), while that of the second group was much larger at 14.23 ha (95% CI 7.15–28.22). In the second year of the study, when both groups were simultaneously in the study site, the best model included only a released group effect (Table 2). The group effect is significant because the addition of this effect in the model improved the AIC by more than 2 points (Burnham and Anderson 2002). The 2 groups had a different mean home range in this second year. The mean home range of the individuals released the year before was lower than that of the newly released individuals (5.90 ha, 95% CI 2.53–13.54, and 14.23 ha, 95% CI 7.15–28.22, respectively).

Table 2. Model comparison for the home range of European pond turtles reintroduced in the Estagnol Nature Reserve in 2008 and 2009.

View Fullscreen

Spatial Spread

Generally, the turtles had spread over a large proportion of the site within a few weeks after their release (Fig. 2). However, the spatial organization of individuals within the reserve evolved over the months. Turtles used the western, northern, and southern sides of the reserve, but avoided the eastern part of the site. This pattern of site use was already clearly observable 3 mo after release for each released group.

View Figure

• Download as Powerpoint
• Download Figure

DISCUSSION

Animal reintroduction programs have the long-term goal of establishing viable, self-sustaining populations (Griffith et al. 1989). However, the evaluation of their success in long-lived species may require decades of monitoring to be conclusive (Sarrazin and Barbault 1996), particularly since it requires assessing the viability of the population through demographic approaches (Seddon 1999). However, the analysis of space-use (movements and home ranges) could be used to assess some reintroduction programs in the short term (Bodinof et al. 2012). The study we conducted on 30 European pond turtles reintroduced to the Estagnol Nature Reserve in southern France showed that the animals rapidly settled in the site, using a large part of the site, and that they exhibited a typical pattern of home range size and seasonal movements. These results suggest that the first phase of the reintroduction (i.e., acclimation to the release site) has been successful.

The fixation rates are very encouraging. Six to eight months after release, at least 70% of the released individuals were still alive in the reserve. Moreover, this high rate is likely to be an underestimation of local survival due to the failure of some radio-transmitters. Only 1 individual was found dead. To obtain unbiased local survival rates, an intensive capture–recapture program could be used to recapture individuals after radio-transmitter failure, although such captures were unfortunately not made in our study. Since reintroduced European pond turtle survival is very high, we recommend such a capture–recapture program in the near future as it would also be useful for estimating population size and confirming recruitment.

The activity period for European pond turtles is usually characterized by a short period of little movement within the hibernation environment, followed by a period of rapid spread over the whole available habitat (Thienpont et al. 2004). The traveled distances then decline during the warmer months, followed by an increase in activity characterized by larger displacements in order to get back to hibernation sites. Here the 2 successive years were highly different in terms of the yearly displacement pattern for both sexes.

The typical displacement pattern in E. orbicularis is that males cover long distances in March–April to find females for mating, while females make larger movements in May–July at the time of egg-laying (Duguy and Baron 1998; Schneeweiss et al. 1998). In the second year of our study, the pattern observed followed this typical pattern for both released groups (i.e., even for the newly released individuals). Cadi and Miquet (2004) also observed a peak of daily movement for females during the egg-laying period (June). The pattern observed in our study exactly followed this pattern. During the foraging period (May–June) in the second year, the mean traveled distance was 133 ± 178 m (min–max: 0–1269 m; n  =  8 males and 18 females). These values are extremely close to those observed in a wild population in Isère, France (143 ± 217 m, range  =  5–520 m; n  =  3 males and 5 females; Cadi et al. 2008).

The first year of the study, in contrast, differed from this typical pattern, with a very rapid decline in activity right after the spring. This year was extremely dry. The dryness induced an especially long estivation period for males, whose displacements sharply decreased to almost zero in July and August. For these males, the peak of displacements was detected later, in the autumn, which may correspond to displacement to reach hibernation sites. In 2009, the males exhibited movements during a much longer period than in 2008 since they were active until October. In 2008, the pattern observed in females differed significantly from that of males. Their movements decreased just after April, but they did not exhibit aestivation behavior as observed in males. This difference is probably due to laying behavior, which required females to prospect in the site in order to find nesting areas. However, the end of egg-laying is usually observed in mid-July for this species in Europe (Rössler 2000 in Novotný et al. 2004). Favorable nesting sites are sometimes far from aquatic environments, thus female turtles need to travel long distances to find these sites (Zuffi and Odetti 1998; Ficetola et al. 2004). Females always choose sunny places for egg-laying (Mitrus and Zemanek 2000), whereas estivation habitats are sometimes composed of the leaf litter provided by woodland (Ficetola et al. 2004). The European pond turtle can make exceptional terrestrial movements, migrating to upland environments for activities other than nesting (Ficetola and De Bernardi 2006), for example, for estivation (Utzeri and Serra 2001). It is thus likely that females were seeking estivation sites after egg-laying, which could explain their activity observed in August.

Regarding home range size, our results show that there was no difference between males and females in either year. Such homology between the sexes has already been observed in wild turtles by Cadi et al. (2004). Yet the comparison between the home range of individuals released in 2008 and 2009 showed large differences; this may be due to the estivation observed in 2008. In 2009, the 2 released groups behaved differently: the individuals released 1 yr before had a smaller home range than the individuals just released. This may suggest that newly released individuals explore more of the site than individuals already present on the site for a year. The home range for newly released individuals was 14.23 ha (95% CI 7.15–28.22), while it was only 5.90 ha (95% CI 2.53–13.54) for the individuals released 1 yr earlier. The home range of the released individuals was, however, of the same order of magnitude as those obtained by Cadi et al. (2004) (i.e., 7.74 ± 3.63 ha for males and 12.51 ± 8.38 ha for females.

After their release, the turtles rapidly spread over the whole site. However, after only a few months they appeared to avoid the eastern part of the reserve. This may be explained by this part of the site being covered by dense vegetation composed of trees and bushes, which limits the amount of solar radiation. Studies conducted on other chelonians have regularly demonstrated a strong dependence of habitat use on thermoregulatory opportunities (Compton et al. 2002; Dubois et al. 2009). Turtles usually select open habitats that allow access to solar radiation (Dubois et al. 2009). Management action to cut the vegetation in this part of the reserve unfavorable for turtles could be considered; however, it is a nesting site for birds, particularly the gray heron (Ardea cinerea).

In terms of the use of space by the reintroduced animals, our results were highly comparable to what is known in wild populations of the species. Females and males have a relatively similar home range size; however, they exhibit different yearly displacement patterns, which are explained by differences in mating and laying behavior. The newly released individuals seem to have explored the site more extensively, resulting in a larger home range, but this was only detected in the release year and may vary depending on environmental conditions. Indeed, such exploration was not perceptible the first year of the study when the weather was dry and warm.

While the general pattern is not sufficient to demonstrate that the reintroduction program will be a success in the long term, it does demonstrate that individuals have not been highly impacted by reintroduction and have developed the typical behavior of the species from the first year. This is a strong sign that at least the acclimation phase of the reintroduction program has been successful and thus that the program and its evaluation should be pursued.

Demonstrating the longer-term success of such a reintroduction program would require information on population dynamics. In long-lived species, this information is generally acquired over a minimum of 10 yrs and sometimes more than 20 yrs of intensive monitoring, using, for instance, capture–recapture methods, a much longer time scale than the study presented here (Dodd and Seigel 1991; Tuberville et al. 2008). Yet for reintroduction programs of long-lived species, in addition to a long-term demographic study, we recommend a fine-scale study of space-use modalities such as this one to evaluate the acclimation phase of the individuals in their new site in the short term. This can allow a rapid reevaluation of the project if failure at this stage is observed or can help in formulating appropriate management actions at the site.