Postfire Asymmetric Reptile and Amphibian Responses in a Mediterranean Forest Ecosystem

Abstract

In August 2023, a large forest fire burned more than 60% of the Dadia–Lefkimi–Soufli Forest National Park in northeastern Greece, following another large fire in 2022. To quantify the effects of these fires on local herpetofauna, we analyzed community composition, abundance, and diversity before and after the 2023 event. Standardized visual encounter surveys were conducted across 29 sites between 2015 and 2024, spanning burned and unburned areas. Species richness, abundance, and diversity metrics, together with Bray–Curtis community dissimilarities, were compared across sampling periods and fire-severity classes. Amphibian assemblages showed high postfire persistence, with 82% of regional species still detected and no significant changes in diversity indices, likely reflecting the buffering role of perennial streams and other hydrologically stable refugia. In contrast, reptile communities showed clear compositional shifts and experienced severe declines: overall reptile species richness decreased to 30% of prefire levels and diversity indices dropped significantly. Tortoises (i.e., Testudo graeca, T. hermanni) declined by nearly 90% relative to prefire estimates, indicating high vulnerability of low-mobility, long-lived species. Snakes were not detected in any burned sites, whereas only a few small-bodied lizards and the freshwater turtle Mauremys rivulata persisted locally. These findings demonstrate that extreme, landscape-scale fires can restructure reptile communities in Mediterranean forests, particularly where long-term habitat change and drought had already reduced population resilience. The study underscores the need for targeted postfire restoration, conservation planning for slow-dispersing taxa, and long-term biodiversity monitoring under increasingly frequent fire regimes.

1. Introduction

Fire is a major ecological process shaping the structure and functioning of ecosystems worldwide, influencing community composition, vegetation dynamics, nutrient cycling, and long-term landscape patterns [1,2,3]. In fire-prone regions such as the Mediterranean Basin, fire regimes, defined by their frequency, intensity, severity, seasonality, and spatial extent, create shifting habitat mosaics and landscape patches that may temporarily increase environmental heterogeneity [4], but can also fragment landscapes, simplify vegetation structure and reduce resource availability when fires are severe or recurrent [5,6]. Over recent decades, fire regimes in Mediterranean landscapes have changed markedly [6,7,8], driven by climate change, rural land abandonment, fuel accumulation, and fires suppression practices. These forces have contributed to larger, more frequent, and more intense fires, including megafires [7,9,10,11] now recognized as emerging threats to biodiversity worldwide. However, animal responses to these novel fire regimes remain highly variable and context dependent [12,13,14]. Understanding how wildlife communities reorganize after fire, and which ecological traits mediate persistence, is essential for predicting biodiversity trajectories and guiding effective conservation in fire-prone ecosystems [2].

Among vertebrates, reptiles and amphibians may be particularly sensitive to altered fire regimes due to their ectothermy, limited dispersal ability, and dependence on specific microhabitats for thermoregulation, shelter, and reproduction [13,15,16]. Fire can affect these taxa through multiple pathways: direct mortality during burns, loss of refuges such as leaf litter, coarse woody debris, and rock crevices, shifts in canopy structure and understory vegetation that modify thermal and hydric microclimates, and changes in prey availability or predation risk [17,18,19]. Yet empirical findings are mixed. Some reptiles, particularly small heliophilic lizards, may increase in abundance following low-severity burns that open habitat structure [16], whereas severe or recurrent fires often depress populations of habitat specialists, long-lived reptiles, and amphibians in landscapes with limited hydrological refugia [20,21]. These contrasting outcomes highlight strong trait- and context-dependent responses [13,22]. Despite growing interest in fire impacts on herpetofauna [23,24,25,26], community-level studies remain scarce in Mediterranean forest ecosystems, especially under extreme fire events [13]. This is especially true for amphibians, whose responses to fire remain comparatively poorly studied and often underrepresented in fire-ecology research [27,28,29]. In contrast, low-mobility reptiles such as tortoises have received more attention and are consistently identified as highly vulnerable to high-severity or frequent fires [30,31]. Such knowledge gaps underscore the need for integrated assessments of entire reptile and amphibian assemblages under contemporary fire regimes.

The Dadia–Lefkimi–Soufli Forest National Park (hereafter Dadia Forest) in northeastern Greece (Figure 1) is one of Europe’s most important biodiversity refuges and a core component of the NATURA 2000 network. Its heterogeneous landscape, comprising oak (Quercus spp.), pine (Pinus nigra, P. brutia) and mixed forests, shrublands and riparian ecosystems, supports a rich herpetofauna, including 11 amphibian and more than 25 reptile species [32,33,34,35,36]. Many of these taxa are of high conservation concern, being listed under Annexes II and IV of the European Union (EU) Habitats Directive (92/43/European Economic Community), protected by the Bern Convention and national legislation, and included in international agreements such as Convention on International Trade in Endangered Species of Wild Fauna and Flora (CITES). In the summers of 2022 and 2023, Dadia Forest experienced two consecutive large fires, with the 2023 event burning more than 60% of the Park and along with the fire flame front in South Evros region represent the largest fire ever recorded in the European Union [37]. These large fires resulted in extensive habitat loss, structural simplification, and widespread faunal mortality, raising urgent conservation concerns for low-mobility and habitat-specialist species. Although vegetation showed signs of early natural regeneration, full ecosystem recovery will likely require decades.

The ecological significance of Dadia Forest and the magnitude of these fires provided a unique opportunity to evaluate the postfire responses of its herpetofaunal communities. In this study, we assessed the effects of 2023 fire on reptile and amphibian assemblages in Dadia Forest. We analyzed pre- and postfire field data collected across burned and unburned sites and compared species richness, abundance, and diversity indices across fire-severity classes. Surveys overlapped previous sampling locations to ensure temporal comparability, and unburned sites served as reference conditions. Special focus was given to taxa of high conservation concern, including tortoises and amphibians with limited dispersal ability. By quantifying short-term community responses to consecutive large fires in a Mediterranean protected area, this study provides critical baseline information for assessing herpetofaunal vulnerability to extreme fire events and informing postfire restoration and management strategies in fire-affected landscapes.

2. Materials and Methods

2.1. Sampling Area

Fieldwork was carried out across an extensive area of the Dadia–Lefkimi–Soufli Forest National Park (Figure 1) during 2022 and 2023, prior to the August 2023 fire. Sampling stations were selected based on habitat suitability for local reptile and amphibian communities, and in total 29 locations (Supplementary Material Table S1 and Figure 2) representing a variety of habitat types were surveyed. Particular emphasis was placed on streams, springs, ponds, and other water bodies, which serve as key refuges for many amphibians and reptile species associated with humid environments. The study area included coniferous, deciduous, and mixed forests, as well as forest openings, rocky substrates, and shrublands.

To collect postfire data, four seasonal field surveys were conducted from November 2023 to November 2024, each lasting four to five days. Survey timing was designed to account for seasonal variation in activity and detectability. In Mediterranean systems, amphibians are primarily active and detectable during spring breeding periods and around aquatic habitats, whereas reptiles show peak surface activity from spring to summer, with reduced detectability outside these periods [32]. Earlier access to the burned area was not permitted due to safety restrictions and persistent underground heat following the fire. Prefire datasets consisted of records collected at the same sites in April 2015 (provided by the Natural Environment and Climate Change Agency; NECCA), spring 2022, and July 2023; one month before the fire event (Supplementary Material Table S1). This sampling design enabled direct temporal comparisons of reptile and amphibian abundance, distribution, and community composition.

In addition, nine supplementary sampling locations situated outside or near the 2023 fire perimeter were surveyed to provide spatial reference conditions where fire impacts were absent or minor (Supplementary Materials Table S1). These unburned areas served as temporal and spatial controls, allowing us to distinguish fire-driven changes from natural interannual variability (Figure 3). After the fire, surveys were repeated both in previously sampled stations and in new locations within the burned area to document postfire assemblage responses. For evening- or nocturnally active species, such as Rana dalmatina, Salamandra salamandra, Bufotes viridis, Bufo bufo, and Vipera ammodytes, night surveys and targeted searches near water bodies were also conducted.

2.2. Sampling Methodology

Amphibian and reptile populations were surveyed using visual encounter surveys along random line transects [38,39]. This method is effective across a variety of habitat types, has minimal impact on site conditions, and poses no risk to the welfare of individual animals. Each transect was 200 m in length, allowing coverage of multiple microhabitats within each sampling site. Observers recorded all individuals detected within 2.5 m on either side of the transect line, resulting in a total surveyed area of 1000 m² per transect. During surveys, good- and high-quality microhabitats up to 5 m from either side of the line were also searched. The number of transects per site varied according to habitat suitability, microhabitat complexity, and geomorphology, resulting in an average survey effort of 2.8 ± 0.8 transects per location. In addition, random quadrat sampling points were established along transects to increase detection probability. Time-constrained searches (~30 min per observer) were conducted at each site to detect species likely to be overlooked during transect walks, particularly cryptic reptiles. For amphibians, targeted searches were conducted at sampling stations near aquatic habitats. Population density was estimated using the formula:

Density per hectare = (N × 10,000)/A,

where N is the total number of observed individuals across all transects and A is the total surveyed area (in m²).

Visual encounter surveys along line transects were used as a standardized and widely applied method for surveying reptile and amphibian assemblages across heterogeneous habitats. Although this approach does not explicitly account for imperfect detection neither allow estimation of absolute population size, it is well suited for comparative assessments of community-level patterns [38,39]. In this study, we focused on evaluating relative changes in species richness, diversity, and community composition across pre- and postfire periods and between burned and unburned sites. More robust occupancy- or detection-based models [40,41] were not applied because the historical prefire datasets were not structured to provide the repeated, temporally independent surveys required for reliable estimation of detection probability.

2.3. Diversity Indices

For each sampling station, we calculated two widely used ecological diversity metrics: the Shannon-Wiener index (H′) [42,43] and the Simpson dominance index (Simpson’s 1–D) [44]. These metrics were chosen because they capture different ecological aspects of the communities. The Simpson index incorporates species richness and dominance and was calculated as:

$$D=\sum p_i^2$$

where $p_i$ represents the relative abundance of species i. Values close to 0 indicate high diversity, whereas values approaching 1 indicate dominance by few taxa. Shannon diversity, more sensitive to rare species and richness patterns, was calculated as follows. Higher H′ values represent greater community diversity [45]:

$$H=-\sum p_i\mathrm{l}\mathrm{n}{p}_i$$

2.4. Data Analyses

To evaluate changes in herpetofaunal abundance, diversity, and community structure before and after the 2023 fire, we analyzed species-level abundance data from all sampling sites and years. Shannon (H′) and Simpson (1–D) indices were calculated for each site by year combination. Because preliminary normality assessments indicated non-normal distributions, differences between prefire (2015, 2022, July 2023) and postfire (late 2023–2024) conditions were tested using non-parametric Kruskal–Wallis tests, followed by pairwise Wilcoxon rank-sum comparisons with Benjamini–Hochberg correction when appropriate. To examine how fire affected community composition, we applied Analysis of Similarities (ANOSIM) using the vegan v.2.8-0 package in R 4.3 [46]. Species abundance data were square-root transformed, and Bray–Curtis dissimilarities were calculated among all samples. Treatment (prefire vs. postfire) served as the grouping variable, and significance was assessed using 999 permutations. ANOSIM was conducted for the entire herpetofaunal assemblage and separately for reptiles and amphibians. To account for potential temporal variation unrelated to fire, we additionally fitted generalized linear mixed models (GLMMs) with Treatment (prefire vs. postfire) as a fixed effect and Year as a random effect. Sites with zero detections were retained and assigned diversity values of zero to avoid inflating postfire diversity estimates. All statistical analyses were conducted in R 4.3 [47].

3. Results

Across the prefire surveys conducted in April 2015, spring 2022, and July 2023, 10 of the 11 amphibian species (91%) known from Dadia Forest and 19 of the 27 reptiles (70%) were recorded (

Table 1. Summary table of amphibian and reptile species and the number of individuals recorded at the sampling sites during the pre-fire surveys (2015 and 2022), prior to the 2023 fire. H′ refers to the Shannon diversity index and 1–D to the Simpson diversity index. Gray-shaded cells indicate sites located within the burned area.

Taxa	S1	S2	S3	S4	S5	S6	S7	S8	S9	S10	S11	S12	S13	S14	S15	S16	S17	S18	S19	S20	S21	S22	S23	S24	S25	S26	S27	S28	S29
Amphibians
Bombina variegata				10	12							6								2		4	2			100
Bufo bufo				2	8							4								1	7		1
Bufotes virids				1																1						100
Hyla orientalis				12	3							2								3
Lissotriton schmidtleri				50	100				1			1					2	4		60		5	40			10			1
Pelophylax sp.				100	150							15				2		10		130	50		70		50	10			15
Rana graeca				1
Rana dalmatina					3								1					2		1			3
Salamandra salamandra				1	32	1						1	1				19						4
Triturus ivanbureschi					5				21			1						1		14
Reptiles
Emys orbicularis																				2
Mauremys rivulata					1							4								3
Testudo graeca		1		7	9	1	2					1								1			3			6	1	2
Testudo hermanni	2			1	1	2						2				3				6			4			3
Ablepharus kitaibelii						1	1						1										2
Lacerta trilineata		15		2	10	4		8	3			2	6							5				18		13	5	1	9
Lacerta viridis	5			20	22				4			6								15	21		8		10	7		2
Mediodactylus kotschyi				2		1
Ophisops elegans	6	4	2	1		2			2											1
Podarcis sp.			1						4														1	8
Podarcis muralis	9	8		4		9	15	19	21				4							6				9		4			2
Pseudopus apodus	1																						1			1	2
Dolichophis caspius																					1						1
Elaphe quatuorlineata												1
Malpolon insignitus																				1
Natrix natrix				2	5				1				1							1	2
Natrix tessellata												1								1
Xerotyphlops vermicularis																										1			2
Zamenis longissima																													1
Species number	5	4	2	16	14	8	3	2	8	-	-	14	6	-	-	2	2	4	-	19	5	2	12	3	2	11	4	3	6
1–D	0.72	0.61	0.44	0.72	0.74	0.75	0.29	0.42	0.71	-	-	0.84	0.71	-	-	0.48	0.17	0.58	-	0.67	0.54	0.49	0.66	0.62	0.28	0.69	0.62	0.64	0.65
H′	1.40	1.09	0.64	1.70	1.73	1.71	0.56	0.61	1.52	-	-	2.20	1.48	-	-	0.67	0.31	1.07	-	1.62	1.01	0.69	1.47	1.03	0.45	1.49	1.15	1.06	1.30

). Specifically, prefire surveys detected 8 amphibian and 15 reptile species in 2015, 8 amphibian and 17 reptile species in 2022, and 10 amphibian and 9 reptile species in July 2023. On the other hand, the postfire sampling conducted between late 2023 and 2024 revealed a significant decline in reptile richness, with only 8 reptiles (30% of the regional fauna) and 9 amphibian (82%) species detected. This shift in species detections was statistically significant (chi-square test; χ² = 19.31, p < 0.001). Despite the more intensive postfire sampling effort, comprising four field surveys between September 2023 and November 2024, both reptile abundance and species richness were substantially lower than in any prefire year. These patterns were consistent even when pre- and post- fired datasets pooled together.

Shannon diversity index (H′; F = 21.163, p < 0.001) and the Simpson index (F = 24.598, p < 0.001) showed statistically significant declines following the 2023 fire (

Table 2. Summary table of amphibian and reptile species and the number of individuals recorded at the sampling sites of the study area during the post-fire surveys (2023–2024). H′ refers to the Shannon diversity index and 1–D to the Simpson diversity index. Gray-shaded cells indicate sites located within the burned area.

Taxa	S1	S2	S3	S4	S5	S6	S7	S8	S9	S10	S11	S12	S13	S14	S15	S16	S17	S18	S19	S20	S21	S22	S23	S24	S25	S26	S27	S28	S29
Amphibians
Bombina variegata					2									7								4
Bufo bufo					1										1						7
Bufotes virids																				2
Hyla orientalis					3															100	1
Lissotriton schmidtleri					100						4		7	9			2	4		150	100	27	100
Pelophylax sp.					100					10	18			25	50	2		10	14	100	100	9	100	15	50				50
Rana graeca
Rana dalmatina					2													2					3
Salamandra salamandra					7					63	1		3	2			19						4
Triturus ivanbureschi																		1		14
Reptiles
Emys orbicularis
Mauremys rivulata	1				23									15							21				19
Testudo graeca									1																	1
Testudo hermanni					1
Ablepharus kitaibelii
Lacerta trilineata	2																			3						1	3		14
Lacerta viridis		3			11
Mediodactylus kotschyi					1
Ophisops elegans					5																2
Podarcis sp.
Podarcis muralis			3											6															1
Pseudopus apodus
Dolichophis caspius
Elaphe quatuorlineata
Malpolon insignitus
Natrix natrix
Natrix tessellata
Xerotyphlops vermicularis
Zamenis longissima
Species number	2	1	1	-	12	-	-	-	1	2	3	-	2	5	3	1	2	4	1	6	6	3	4	1	2	2	1	-	3
1–D	0.44	0.00	0.00	-	0.68	-	-	-	0.00	0.24	0.36	-	0.42	0.75	0.04	0.00	0.17	0.58	0.00	0.69	0.62	0.48	0.53	0.00	0.40	0.50	0.00	-	0.36
H′	0.64	0.00	0.00	-	1.45	-	-	-	0.00	0.40	0.63	-	0.61	1.56	0.10	0.00	0.31	1.07	0.00	1.27	1.11	0.83	0.84	0.00	0.59	0.69	0.00	-	0.60

). Shannon values decreased from 1.18 ± 0.51 before the fire to 0.55 ± 0.49 afterwards, while Simpson values declined from 0.59 ± 0.17 to 0.32 ± 0.26 (see Table 1 and Table 2). We found no significant differences between prefire years (i.e., 2015 and 2022) for either Shannon (t = 1.28, df = 11.14, p = 0.228) or Simpson (t = 1.10, df = 10.54, p = 0.297) indicating that overall herpetofauna diversity remained relatively stable prior to the fire. However, the lack of change in composite diversity indices during 2015–2022 masks substantial population declines in certain species, particularly tortoises, whose populations had already decreased markedly prior to the fire.

Multivariate analyses further confirmed a shift in community structure associated with the fire. ANOSIM based on Bray–Curtis dissimilarities revealed a statistically significant difference in herpetofaunal community composition between prefire and postfire samples (R = 0.132, p = 0.001). Although the effect size was modest, suggesting partial overlaps between pre- and postfire assemblages, between-group dissimilarities (median rank; 1443.5) were consistently higher than within-group dissimilarities. In addition, postfire communities (median rank; 1391.0) exhibited greater internal variability than prefire communities (median rank; 1128.3), suggesting increased spatial heterogeneity in species composition across burned areas. Together, these findings indicate that while overall postfire diversity was sharply declined, the fire also generated shifts in community structure, with different taxonomic groups responding in distinct ways.

3.1. Effects of Fire on Reptilian Populations

We found substantial declines in reptiles’ abundance and diversity following the 2023 fire. Shannon diversity decreased from 1.08 ± 0.54 before the fire to 0.30 ± 0.38 afterwards (Figure 4A), while Simpson diversity declined from 0.55 ± 0.24 to 0.19 ± 0.23 (Table 1 and Table 2). Both indices showed statistically significant reductions (Kruskal–Wallis test; Shannon: U = 11.64, p < 0.001; Simpson: U = 12.16, p < 0.001). These differences were driven exclusively by sites located within burned areas, where diversity decreased sharply between prefire and postfire surveys (Shannon: U = 10.07, p = 0.001; Simpson: U = 10.07, p = 0.001). In contrast, unburned reference sites showed no significant changes in either index (both p > 0.10). Reptile species richness also declined markedly after the fire, with only eight species detected in the postfire period compared to nineteen species documented across prefire surveys. This reduction was reflected in ANOSIM where reptile community composition differed significantly between pre- and postfire assemblages (ANOSIM; R = 0.294, p < 0.001), indicating a clear shift in reptile assemblages associated with the 2023 fire. Postfire reptile communities displayed increased internal dissimilarity compared to prefire communities, suggesting greater spatial variability in species occurrence across burned sites. The magnitude of the observed decline is consistent with the seasonal life-history and activity patterns of reptiles, which are most surface-active during spring and summer for mating and reproduction and retreat to underground refuges later in the year [32]; notably, the August 2023 fire coincided with this period of peak activity.

Declines in population abundance were most pronounced for low-mobility, long-lived reptiles. Tortoise populations experienced severe reductions: observations of Testudo graeca declined from 23 individuals in spring 2022 to 11 individuals in July 2023 (prefire), and to only 2 individuals during postfire surveys. Testudo hermanni showed a similar pattern, with counts decreasing from 17 (2022) and 7 (July 2023) to only 1 individual after the fire. Generalized linear models confirmed substantial declines in tortoise counts (GLM: χ² = 9.14, p = 0.002). Estimated population densities fell from 13.8 individuals/ha in 2015 and 6.9 individuals/ha in 2023 to only 1.03 individuals/ha after the fire, representing an overall reduction of 92.5% since 2015 and 85.1% since 2022. Notably, an approximately 50% decline had already occurred between 2015 and 2023, prior to the fire, indicating ongoing population deterioration. Finally, no snakes were recorded inside burned areas during any of the four postfire surveys (from November 2023 to November 2024), despite extensive sampling effort.

3.2. Effects of Fire on Amphibian Populations

Amphibians showed comparatively limited responses to fire. Diversity indices did not differ significantly between prefire and postfire periods. Shannon diversity remained largely unchanged (prefire: 0.72 ± 0.53; postfire: 0.52 ± 0.45; U = 1.07, p = 0.301), as did Simpson diversity (prefire: 0.39 ± 0.27; postfire: 0.31 ± 0.26; U = 0.99, p = 0.318) (Figure 4B). All amphibian species detected before the fire were also detected afterward, and species richness did not differ significantly between years (χ², p = 0.37). Further analyses evaluating burned versus unburned sites showed no significant changes in amphibian diversity within either category (all four ps > 0.05). ANOSIM results for amphibians likewise indicated no significant difference in community composition between prefire and postfire samples (ANOSIM; R = 0.067, p = 0.932). Amphibian assemblages thus remained compositionally stable across sampling periods.

This stability is consistent with the broadly shared life-history and activity patterns of amphibians in the study area, which reproduce in aquatic habitats during spring, with eggs and larvae developing in water, while adults remain closely associated with perennial or semi-permanent water bodies or reduce activity through aestivation during hot summer periods [32]; consequently, the August 2023 fire coincided with a phase when many individuals were either aquatic or sheltered in moist refugia.

4. Discussion

Our study examined how the fire in Dadia-Lefkimi-Soufli Forest National Park—one of the most extensive recent fires in southeastern Europe—affected amphibian and reptile assemblages. We hypothesized that a fire of such magnitude would impose pronounced negative effects across herpetofaunal diversity. Instead, our findings revealed a highly asymmetric response: reptiles underwent substantial declines in abundance, richness, and diversity, whereas amphibian assemblages remained largely unaffected and stable across the landscape. These divergent outcomes emphasize that fire magnitude and extent do not uniformly predict biological impacts ([48], but see [49]); rather, the ecological traits, behavioral patterns, and microhabitat associations of different taxa fundamentally shape their vulnerability or resilience [13,16,28,50].

Despite the extreme, landscape-scale extent of the 2023 fire [37], 82% of the regional amphibian fauna were still detected in the postfire surveys. As such, diversity metrics (Shannon, Simpson) and community-level analyses (ANOSIM) showed no significant departures from prefire assemblages. This relative stability aligns with recent global syntheses indicating that amphibian responses to fire are highly variable and strongly dependent on local environmental conditions [13,16,28,51]. In many systems, the postfire persistence of amphibian populations is facilitated by hydrologically buffered microrefugia, such as streams, ponds, and/or shaded riparian corridors, which can drastically reduce direct mortality, providing shelter during and after fire events. Renken [16] for instance, reported that moist microhabitats and the burrowing behavior of many amphibians can limit fire-related mortality in temperate forest ecosystems. A similar mechanism likely occurred in Dadia Forest. The combination of estivating underground during the hottest summer months and the continued flow of water in many stream sections within the burning sites, likely provided amphibians with cool, humid refuges during and after the fire, facilitating their high postfire detectability and contributed to the apparent resilience of the amphibian assemblage [52]. However, it is important to note that our assessment represents only the first year after the fire, capturing the initial, short-term phase of ecological response [28]. A growing body of literature cautions that early postfire surveys may not reflect longer-term outcomes, as delayed population declines, gradual shifts in community composition, and habitat-driven changes often emerge only after several years [28,53,54]. Studies reporting little or no immediate fire effects—particularly for amphibians—frequently attribute such patterns to the short temporal window of sampling, which can underestimate or entirely miss longer-term ecological trajectories [53,54,55]. Thus, while our findings describe the immediate postfire stage in Dadia Forest, continued monitoring will be critical to determine whether these patterns persist, intensify, or reverse over longer time scales.

Contrary to amphibians, reptile communities exhibited pronounced postfire declines across the burned sites of Dadia Forest. Only 30% of the region’s reptile fauna was detected after the fire, and both abundance and diversity indices decreased sharply relative to prefire years. These findings come in contrast to numerous studies reporting neutral or even positive reptile responses to fire [16,25,51,56,57], particularly in small, heliophilic lizard species that often increase in abundance following low-severity prescribed burns. Such burns typically create fine-scale habitat mosaics, preserve pockets of woody debris, maintain refugia, and increase ground-level temperatures, thereby enhancing basking and foraging opportunities for disturbance-tolerant reptiles [58]. The discrepancy between these outcomes and the severe declines observed in Dadia likely reflects fundamental differences in fire severity, spatial extent, and landscape context. The 2023 Dadia fire burned for nearly fourteen consecutive days, producing a landscape-scale event that consumed large tracts of forest with minimal unburned refuges [59]. Dadia fire was characterized by high-severity crowning which removed canopy cover, eliminated shrub and herb layers, and fully consumed the litter and coarse woody debris that many reptiles rely on for thermoregulation, predator avoidance, and shelter [17]. Under such conditions, even relatively mobile species face limited opportunities to escape advancing flames or recolonize burned areas.

But even within reptiles, these extreme fire characteristics affected different taxa disproportionally. Tortoises, i.e., Testudo graeca and T. hermanni, slow-moving, long-lived species with restricted dispersal abilities, experienced the most severe declines [25], with densities reduced to roughly one-tenth of prefire levels (Table 1 and Table 2). Although high fire-induced mortality in Mediterranean chelonians is well documented [30,31,60,61] our results show that the 2023 fire acted on populations that were already declining. A substantial decline of approximately 50% had already occurred between 2015 and July 2023, suggesting substantial demographic erosion predating the fire. We assume that long-term processes such as land abandonment, reduced grazing pressure, and resulting vegetation encroachment have progressively reduced habitat openness, creating cooler and more shaded microclimates that hinder thermoregulation, movement, and nesting [62]. These habitat changes, combined with multi-year drought and reduced precipitation between 2016 and 2022 (Standardized Precipitation Index; Hellenic National Meteorological Service), likely constrained foraging and reproductive output [62]. Τhese findings suggest that the 2023 fire did not initiate the tortoise collapse but accelerated and magnified ongoing negative demographic trends.

A similar pattern was observed for snakes, which were not detected at any burned site during the postfire surveys. Although this absence might suggest severe impacts, it should be interpreted with caution: snakes in Mediterranean forests typically occur at low densities, are highly cryptic, and rely heavily on shelter availability; characteristics that reduce detectability even when populations persist [32,63]. Thus, failure to record snakes does not necessarily indicate complete local extirpation. Nonetheless, their absence in postfire surveys aligns with the expectation that species with low mobility and strong dependence on structurally complex microhabitats are particularly vulnerable to high-severity, landscape-scale fires. Despite the latter, a few small-bodied lizard species (e.g., Lacerta trilineata, L. viridis, Podarcis muralis; Table 2) persisted in parts of the burned landscape. Their agility, small body size, and frequent use of microhabitats such as root systems, rock crevices, holes in the soil, and shrub bases may have offered partial protection during the fire, allowing some individuals to survive [17,64]. However, their persistence was limited in scope and did not offset the broader collapse observed across the reptile community. Another taxon that survived the fire was the freshwater turtle Mauremys rivulata, which remained present in several postfire sites but exclusively in close association with perennial streams. This pattern further supports our inference on amphibian results, where running water and riparian microhabitats functioned as critical fire refugia. The species’ strong aquatic dependence likely buffered it from direct fire mortality, while the thermal and hydric stability of streams provided shelter during the most intense phases of the burn. The survival of M. rivulata therefore reinforces the broader conclusion that hydrologically buffered habitats played a central role in structuring postfire vertebrate persistence across Dadia.

Overall, our findings highlight that the effects of large fires on herpetofauna are strongly taxa-specific and shaped by a combination of life-history traits, habitat associations, and the availability of refugia [22,65,66,67]. Reptiles, particularly low-mobility, habitat-specialist species, proved far more vulnerable to the extreme conditions of the 2023 Dadia fire than amphibians, whose persistence was supported by hydrologically buffered microhabitats. These patterns mirror emerging evidence from across the Mediterranean Basin showing that increasingly large, severe, and recurrent fires disproportionately impact ecological specialists and species reliant on structurally complex forest habitats, while favoring only a subset of disturbance-tolerant taxa [1,25]. Moreover, the Dadia Forest hosts a high diversity and density of raptor species [68,69], and the extensive postfire canopy openings likely increased prey detectability, potentially intensifying avian predation on reptiles. Such fire-induced structural simplification may therefore have indirectly contributed to reptile declines by facilitating more efficient foraging by avian predators. Although our conclusions are based on standardized visual encounter surveys that do not explicitly account for imperfect detection, our analyses focus on relative comparisons of species richness, diversity, and community composition across pre- and postfire periods and burned versus unburned sites. Occupancy-based approaches that incorporate detection probability [40,41] offer a powerful framework for inferring species occurrence and habitat use but require repeated, temporally independent surveys, which were not consistently available across the historical prefire datasets used here.

As climate warming, prolonged drought, and vegetation encroachment continue to intensify regional fire regimes, such large fire events are expected to become more frequent [70,71,72], further constraining recovery opportunities for slow-dispersing species such as Testudo spp. and many snakes. Future research and management in Dadia Forest and similar Mediterranean systems should integrate long-term, detection-corrected monitoring with assessments of fire severity, habitat structure, and climate trends, alongside management strategies that maintain habitat heterogeneity, protect riparian refugia, and prevent further loss of open habitats through targeted grazing or controlled vegetation management.