Chelonians (order Testudines) are long-lived vertebrates (Gibbons 1987; Wang et al. 2025). Among terrestrial chelonians, the genus Testudo includes tortoises inhabiting Europe, North Africa, and Western and Central Asia (Turkozan et al. 2008; Speybroeck et al. 2016). Testudo hermanni Gmelin, 1789, or Hermann’s tortoise, is a species found in southern Europe (Speybroeck et al. 2016; Luiselli 2024), ranging from Mediterranean and sub-Mediterranean regions to more inland areas (see Koller Šarić et al. 2023). The lifespan of this species can reach up to 79 years (Reinke et al. 2022). The population age structure could be dominated by older age classes if nest predation is high and recruitment is low (Stubbs and Swingland 1985). Two subspecies are currently recognized: the western Hermann’s tortoise (Testudo hermanni hermanni Gmelin, 1789) is distributed in northeastern Spain, the Balearic Islands, southeastern France, Corsica, the coastal plains and hilly areas of the Italian Peninsula, and Sardinia; the eastern Hermann’s tortoise (Testudo hermanni boettgeri Mojsisovics, 1889) is distributed in northeastern Italy, northwest of the Po River, and throughout the Balkan Peninsula (Speybroeck et al. 2016; Luiselli 2024). The western subspecies is strictly confined to typical Mediterranean ecosystems, heavily utilizing dense coastal maquis, garrigue scrublands, and cork oak woodlands, and has a relatively small body size and low fecundity (Stubbs and Swingland 1985; Livoreil 2009). The eastern subspecies inhabits complex habitat systems, has a larger average body size than the western subspecies, and produces larger clutches (Koller Šarić et al. 2023 and references therein).

Although well protected by their armor, chelonians are not safe from injuries. Shell injuries may be frequent and may be caused by various factors, both natural (e.g., depredation and accidental falls) and anthropogenic (vehicles, agricultural machinery, intentional fires for maintaining and creating local environments, e.g., pastures, falls from human-made structures, and attacks by allochthonous species), as reviewed in Biaggini and Corti (2018). Specific shell injuries, such as partially or completely peeled dermal plates, result from wildfires, which tortoises may or may not survive (Couturier et al. 2014; Vujović et al. 2015; Badiane et al. 2017). The detrimental effects of wildfires on local populations of eastern Hermann’s tortoises have been reported mainly from the Mediterranean and sub-Mediterranean parts of the subspecies’ distribution range in Greece (Hailey 2000) and Montenegro (Vujović et al. 2015), as well as from Bulgaria (Popgeorgiev 2008 and references therein; Slavchev et al. 2014) and Romania (Rozylowicz and Dobre 2010).

A rapid increase in non-lethal shell injuries was observed over 15 years (2010–2024) in a continental population of eastern Hermann’s tortoises situated in a partially human-altered habitat system within the “Sićevo Gorge” Nature Park in southeastern Serbia. Therefore, changes in the frequency of non-lethal shell injuries were quantified over the monitoring period in an open population exposed to anthropogenic impacts, and shell condition was evaluated to determine whether it had deteriorated. Moreover, some shell injuries were undoubtedly caused by wildfires, which have increased in frequency and intensity within the study area: five wildfire events were recorded within the Nature Park between 2016 and 2025, four of which affected areas greater than 0.10 ha, while one was smaller (Biological Society “Dr. Sava Petrović,” unpublished). Therefore, whether there was a significant change in the frequency of non-lethal shell injuries specifically resulting from wildfires was assessed.

Monitoring of the open population of the eastern Hermann’s tortoise was conducted between 2010 and 2024 in southeastern Serbia, in the surroundings of Kunovica village (43.301°N, 22.078°E), located 17 km east of the city of Niš. The study area partially falls within the “Sićevo Gorge” Nature Park (https://zzps.rs/парк-природе-сићевачка-клисура/), with most of the land privately owned. Tortoises were surveyed within an area of approximately 23 ha (see the map and further details in Stojadinovic et al. 2017a, 2017b). Tortoises were recorded on the southeastern, southern, and southwestern slopes of the hilly terrain (324–462 m above sea level), which is primarily overgrown with deciduous forest of Quercetum farnetto-cerris and interspersed with grassland patches, active and abandoned vineyards, orchards, and gardens (for details, see Stojadinovic et al. 2017a, 2017b). Monitoring was conducted twice per year—six consecutive days in the fourth week of May and seven consecutive days in the third week of July (for details, see Stojadinović et al. 2013)—from 8:00 to 17:00 h, depending on weather conditions. The number of researchers in the field varied from two to eight per day, with four on average. At least one experienced researcher was present during each monitoring session.

At the first capture, every tortoise was individually marked by notching the marginal plates of the carapace with a hacksaw, following the scheme described in Stojadinović (2017). Age (adult, subadult, or juvenile), sex (male, female, or undetermined), and shell condition were recorded upon the first capture and at subsequent recaptures. Tortoises were immediately released afterward at the locations where they were found.

Shell condition was assessed visually, recorded in the protocol, and assigned to one of four categories: 0—undamaged shell (Fig. 1); 1—shell with minor damage (less than 30% of the shell surface area damaged), without partially or completely broken plates and with no more than five bite marks on the shell (Fig. 2); 2—shell with major damage (30% or more of the shell surface scratched or at least one shell plate with a broken peripheral part; Fig. 3); and 3—shell with severe damage, including stubs and completely broken plates, mostly marginal plates (Fig. 4). Signs of charring and peeling of the dermal plates indicated that a tortoise had experienced a wildfire (Hailey 2000). The extent of wildfire damage contributed to the evaluation of general shell condition (see the legends of Figs 2, 3, 4); however, the impact on shell condition (affected or unaffected by wildfire) was recorded as a separate variable.

Only the first annual record of each tortoise was analyzed. Five variables were selected for statistical analyses: year of monitoring, sample size per year, status of the recorded tortoise in a particular year (N—“new,” that is, previously unmarked, or R—“marked”), shell condition category (categories 0–3; see the previous paragraph), and shell damage caused by wildfire (0—shell not affected by wildfire; 1—shell affected by wildfire).

All variables were tested for normality using the Kolmogorov–Smirnov test. As the variables were not normally distributed, the nonparametric Kruskal–Wallis test and Spearman’s rank correlation were applied, with the results considered significant at P < 0.05. The analyses were conducted using STATISTICA 8.0 (StatSoft Inc., Tulsa, OK, USA).

From 2010 to 2024, 746 tortoises were recorded and marked within the study area. Annually, the proportion of recaptured tortoises ranged from 43% to 84% (Fig. 5). On average, 74% ± 9.88% (range: 43.17–83.84%) of the tortoises recorded annually were individuals that had already been marked in previous years. The variation in the number of captured tortoises per year was statistically significant (Kruskal–Wallis test: H [14, N = 2218] = 479.48, p < 0.01; Spearman’s R = –0.71, p < 0.01; Fig. 5), but the annual proportion of recaptured tortoises showed no significant variation over the years (Kruskal–Wallis test: H [14, N = 1473] = 0.00, p > 0.05). Only seven tortoises (<1% of all tortoises) were found dead after being marked (one each in 2013, 2016, and 2018, and two each in 2023 and 2024). In all cases, the cause of death was unknown.

The highest (91.4%) and lowest (1.0%) proportions of tortoises with intact shells were recorded in the fourth (2013) and twelfth (2021) years of monitoring, respectively (Fig. 6). During the first seven years of monitoring, at least 80% of the tortoises recorded per year had no scratches or injuries on their shells. This percentage decreased to less than 20% of the annual sample in the eighth year, continued to decline in subsequent years, and fell to less than 10% of the annual sample in the last six years of monitoring (Fig. 6). Over the years, the proportion of tortoises with intact shells decreased (shell condition 0; Spearman’s R = –0.90, p < 0.01), while the proportion of those with injured shells increased. During the first seven years of monitoring, tortoises with shell condition 1 and shell condition 2 each made up 10% or less of the annual sample (Fig. 6); however, their proportions increased over the last eight years of monitoring (Spearman’s R = 0.73, p < 0.01, and R = 0.88, p < 0.01, for shell condition 1 and shell condition 2, respectively; Fig. 6). When these two shell categories were pooled, they reached or exceeded 60% during the last eight years of monitoring. The proportion of tortoises suffering severe mechanical shell damage (shell condition 3; Fig. 6) fluctuated between 2% and 4% during the first seven years of monitoring but increased to over 10% in the ninth year and fluctuated between 10% and 22% during the last seven years of monitoring. This proportion also increased significantly over the years (Spearman’s R = 0.80, p < 0.01; Fig. 6).

The proportion of annually recorded tortoises with shell injuries due to exposure to fires was further estimated. This proportion was less than or equal to 1% during the first eight years of monitoring. It was 1% in 2016 and continued to rise in the following years, albeit with some fluctuation, exceeding 50% of the annual sample during the last 3 years of monitoring (Fig. 7). The analysis revealed a significant change over the years in the annual proportion of tortoises affected by fire (Kruskal–Wallis test: H [14, N = 2218] = 1262.67, p < 0.01; Spearman’s R = 0.95, p < 0.01).

During 15 consecutive years of monitoring Hermann’s tortoises within the study area, the proportion of those with heavily damaged shells (shell condition 3) ranged from 2.1% to 26.5%. For comparison, Meek and Inskeep (1980) reported that only 4% of examined Hermann’s tortoises in one population from southern Montenegro had serious injuries to the carapace, plastron, and limbs and that “such serious injuries to the shells or limbs…were not particularly common.” A similar conclusion could be drawn for tortoises in the study area if relying only on the records collected during the first seven years of monitoring, supporting the need for long-term studies that can identify such changes. However, analysis of records collected from 2017 to 2024 indicates that the annual proportion of tortoises with major shell injuries increased approximately five times during the last eight years of monitoring.

The causes of this rapid increase in shell injuries within the population may be natural, anthropogenic, or both (Meek 2007; Biaggini and Corti 2018). Natural causes may include an increase in the population density of autochthonous predators, which are a common threat to T. hermanni throughout its distribution range (Luiselli 2024). A wide range of potential tortoise predators has been mentioned in the literature: foxes (Vulpes vulpes), rodents, and birds of prey in southern Montenegro (Meek and Inskeep 1980); wild boars (Sus scrofa), foxes, dogs—particularly feral dogs (Canis familiaris)—martens (Martes spp.), badgers (Meles meles), and polecats (Mustela putorius) in Romania (Rozylowicz and Dobre 2010; Moraru et al. 2018); wild boars and some Mediterranean carnivorous mammals in Croatia (Koller Šarić et al. 2023); golden eagles (Aquila chrysaetos) in Bulgaria (Milchev 2022); and beech martens (Martes foina) and badgers in Greece (Hailey et al. 1988). The golden jackal (Canis aureus) should be added to the list as an allochthonous/invasive carnivore species with an increasing presence in this part of Europe (Cunze and Klimpel 2022). Although tortoises have not been specifically confirmed as prey of jackals, reptiles in general have been (Penezić and Ćirović 2015). Birds of prey—predominantly buzzards (mainly Buteo buteo)—as well as corvids—mostly ravens (Corvus corax)—were recorded in the habitat of this study population of the eastern Hermann’s tortoise; moreover, the nearest nest of the golden eagle (Aquila chrysaetos) was recorded at a distance of 3 km in a straight line from the study site, and the adult birds are regularly seen in this area searching for prey (M. Nikolić, unpublished). Traces of wild boar, badger, and possibly wolf (Canis lupus) were noted during the monitoring sessions (J. Crnobrnja-Isailović, unpublished). Moreover, certain carapace injuries could represent traces of predator bites because they were numerous, round, small, shallow, and uniform in size (approximately 4 mm in radius) (Crnobrnja-Isailović, unpublished field observations).

Anthropogenic impacts on the study area have likely occurred for a long time, as the south–southwest-oriented slopes have been attractive for growing grapes, other fruits, and vegetables (Turnšek 2006). Following the Second World War, industrialization in the region prompted significant migration from agricultural areas to urban centers, as people sought better-paid employment and improved living conditions (Radivojević et al. 2014; Milovanović 2016); human-altered habitat patches within the study area became overgrown by primary vegetation (Turnšek 2006). However, during the study, an increase in anthropogenic pressure in the habitat was observed in the form of the gradual renovation of some abandoned cottages, mostly located along the local paved road, as well as the renewal of several gardens and vineyards. An increase in agricultural activities in an area could intensify the negative impact of vehicles on tortoises (Saumure and Bider 1998, Saumure et al. 2007). Additionally, intensive construction work on a section of the international road to Bulgaria (the construction of the A4 highway from the city of Niš to the Gradina border crossing) started in 2012 and was completed in 2018. One section of this highway is less than 1 km in a straight line from the study area. Although the average home range of Hermann’s tortoises was calculated as 3.2 ha in Turkey (Turkozan et al. 2019) and 3.79 ha in Romania (Rozylowicz and Popescu 2013), an earlier study of the population studied here revealed that females could move within an area as large as 90 ha (Stojadinovic et al. 2017a). Transportation intensity clearly increased during the A4 highway construction work (Glavic 2025), which could explain the rise in the number of tortoise–human encounters and, consequently, more frequent, although not necessarily lethal, tortoise casualties. Indeed, the number of tortoises with severe shell injuries has increased in the study area since 2017–2018. Another anthropogenic impact that should not be neglected is sport hunting. In addition to being part of the Nature Park, the study area has also been a regular hunting ground managed by the local hunting society since the end of the 19^th century (https://www.lunis.rs/). Interactions with hunting dogs could result in different kinds of injuries to tortoises’ shells and limbs (see Carlson et al. 2026), but they could not be the main factor because such interactions also occurred before the shell condition of this particular tortoise population deteriorated.

In the study area, the proportion of eastern Hermann’s tortoises affected by wildfires from 2010 to 2018 was negligible compared with cases reported for the western subspecies (Couturier et al. 2014; Badiane et al. 2017) or those known from other parts of the Balkan Peninsula (Hailey 2000; Popgeorgiev 2008; Slavchev et al. 2014; Vujović et al. 2015). This analysis revealed that the proportion of tortoises affected by fire in the study area apparently increased in 2018 and continued to rise, reaching more than 80% of all recorded individuals in 2024. However, as the overall number of tortoise carcasses found in the studied part of the Nature Park from 2018 to 2024 was negligible and the causes of their deaths were indeterminate, the evidence of the severity of the impact of these wildfires on tortoise population size is unclear. Couturier et al. (2014) observed that the numbers of Hermann’s tortoises recorded at sites that had experienced wildfires over the previous 50 years were approximately 31% lower than those at unburned sites.

Non-lethal injuries to tortoise shells could accumulate with age. Therefore, an increase in the proportion of individuals with damaged shells in the population over the years could simply be a consequence of predominantly observing subadults and adults, as juveniles are much more difficult to spot in dense vegetation because of their small size. Carlson et al. (2026) warned of another threat related to tortoise longevity: a distinct decline in female shell quality occurs at a very old age because of osteoporosis, whereby the scutes on the carapace, as well as the underlying bone, become thin, porous, and more susceptible to breakage. However, the shell condition of an old tortoise does not necessarily deteriorate over time (Crnobrnja-Isailović, unpublished data). That is, in an “open” (not isolated) tortoise population, even if the age distribution is “stable” (the youngest age class predominates), an increased proportion of tortoises with non-lethal shell injuries, coupled with a decrease in the annual population sample size, should be a sign of an existing threat. Shell injuries make tortoises more vulnerable to further injuries and depredation (Berry 1974; Harding 1985). Moreover, lesions resulting from trauma appear to make tortoises more susceptible to anemia (Christopher et al. 2003). Despite the scarcity of records of dead tortoises in the studied population, this study revealed that the annual sample size of recorded tortoises has decreased over the years. This strongly suggests that conservation measures must be strengthened. In this particular case, such measures would include the prevention of wildfires through improved monitoring of the protected area and increased awareness among inhabitants and visitors. In addition, increased communication and cooperation among the local hunting society, the forest department, and conservation biologists would improve knowledge of the impact of natural predator species on Hermann’s tortoises and help design sustainable actions.

Fifteen years of monitoring an eastern Hermann’s tortoise (Testudo hermanni boettgeri) population clearly indicated a progressive decline in shell condition, which can be interpreted as a direct response to the synergistic effects of natural and anthropogenic factors. Intensification of anthropogenic pressure appears to have coincided with an increased frequency of fires and possibly heightened predator activity. The results suggest that current habitat protection measures may be insufficient to offset the cumulative damage, highlighting the urgent need for active population management to prevent the population’s further decline.