The first study site (Fig. 1a), “Le Porge”, is located on the Gironde coast (44°57'07", 1°12'23"). Characterised by an oceanic climate and sea spray, the area remains humid throughout the year, except during the summer months (Priol 2015). The study area covers 28 ha and is divided into four areas: i) the Beach; ii) the White dune, the upper grassy part of the artificial dune; iii) the Grey dune, a shrub-dominated zone located along both the crest and the lower slopes of the dune; iv) the Forest, which includes both an unmanaged old-growth forest characterized by standing pines and abundant coarse woody debris, and an actively managed plantation of maritime pine (Pinus pinaster) subjected to regular harvesting. The breeding pond located on the study site is permanent, appears relatively natural rather than artificial, and covers an area of approximately 15 m².

The second study site (Fig. 1b), “Le Plan”, is located in the municipality of Oppède (43°51'27", 5°09'09") in southeastern France, in an agricultural landscape. The study area covers 14.5 ha and consists of a former gravel pit on the banks of the river Calavon surrounded by farmland. The site includes a vineyard and a lavender field, both subject to mechanical soil tilling several times a year, and a bushy fallow area. The site hosts a single temporary breeding pond, originating from former gravel extraction activities, which is relatively deep and covers an area of approximately 1.3 ha. It tends to flood more frequently in spring (approximately one drought year per decade) than in autumn (around four drought years per decade). Furthermore, the study site is crossed by a secondary road with heavy traffic, averaging 2,028 vehicles per day across seasons and 25 vehicles per hour during the night (Travers 2023), a feature that distinguishes it from the two other study sites, which lack road infrastructure.

The third study site (Fig. 1c), called “Les Tours”, is located in the municipality of Goult (43°51'33", 5°16'46") in south-eastern France, within an agricultural landscape. The study area covers 21 ha and is located along the Calavon river. It consists of riparian forest, open clearings with low vegetation, dry grassland and agricultural plots primarily used for market gardening. The area includes six permanent and temporary ponds with an average area of 1,800 m², which are more often filled with water in spring than in autumn.

Populations of Western spadefoot toad at all three study sites were monitored using the same Capture-Mark-Recapture (CMR) design.

The population at Le Porge was monitored in 2010 and 2011, with a total of 22 field sessions, whereas at Le Plan and Les tours, monitoring occurred between 2019 and 2021, with 30 and 26 sessions respectively, conducted between March and November (Table 1). Sessions conducted between March and May were classified as spring sessions, whereas those conducted between September and November were classified as autumn sessions. The summer period (June to August) and the winter period (December to February) were excluded, as the dry summer conditions are unsuitable for amphibian activity in Mediterranean regions (Segura and Palomar 2023) and winter temperatures are similarly limiting. Intervals between spring and autumn sessions were consistently based on a three-month gap (June, July, and August), as were intervals between autumn and spring sessions (December, January, and February).

Each capture session was conducted at night exclusively within terrestrial habitats, under light rainfall or relatively high humidity conditions (> 90%), using headlamps. This choice was motivated primarily by our focus on the species’ annual activity patterns rather than its breeding activity. In addition, the terrestrial activity period extends over a much longer timespan than the breeding period, which is relatively short and “explosive” in aquatic environments for this species, making it more suitable for a CMR study aimed at maximising recapture opportunities. Furthermore, we deliberately avoided working in aquatic habitats in order to minimise disturbance and reduce potential impacts on the egg clutches and larval stages of this rare species.

Monitoring was structured as three consecutive 90-minute surveys at “Le Porge” and “Le Plan” (survey time divided among observers, typically one at “Le Porge” and four at “Le Plan”) and as two consecutive 90-minute surveys at “Les Tours” (conducted by two observers), with a 30-minute interval between surveys. The time spent handling individuals was systematically subtracted from the total survey duration. During each survey, individuals were located by random active searching within a 300 m radius around the breeding ponds. Each captured individual was sexed based on the presence of a glandular pad on the dorsal surface of the forearm in males (Recuero 2014) and assigned to one of two age classes following criteria established by skeletochronological analysis by Leclair et al. (2005), which considers mature adults as individuals with a snout-vent length (SVL) ≥ 40 mm and juveniles as those with an SVL of 34–39 mm. SVL was measured using a caliper (Wiha dialMax, accuracy 0.1 mm) and body mass was recorded with a spring scale (100 g, accuracy 0.3 g; Pesola). Individuals were marked with sterile PIT-tags (1.4 × 9 mm; 0.03 g; 134.2 kHz; Biolog-ID) implanted subcutaneously along the internal face of the thigh beside the femur and photographed before being released at the capture site. The mass of PIT-tags (0.03 g) falls well below both the widely accepted 5% body mass threshold for terrestrial vertebrate-attached devices (Wilson and McMahon 2006) and the more conservative limits of 1.6–3% proposed in the literature (Wilson et al. 2021). The transponders were monitored using a handheld RFID reader (HHR-3000 and RS100 V8 Biolog-ID), allowing individual identification without physically handling the animal during recapture. This marking technique has proven highly reliable for studying the Western spadefoot toad, with a 100% retention rate and no detectable effect on body mass following marking (Renet et al. 2021).

Ministerial authorisations were granted to the field personnel by the Direction régionale de l’environnement, de l’aménagement et du logement (DREAL) of Nouvelle-Aquitaine and the DREAL de Provence-Alpes-Côte d’Azur (PACA).

The CMR data were analysed using the Pradel Robust Design Survival and Seniority model (Pradel 1996) which allows estimation of capture probability (p), apparent survival (φ), recruitment (b, inferred from the seniority as b = 1-seniority (γ)) and population sizes (N), for each site separately. All intra-night sessions were merged to create a single night dataset (secondary session) and capture sessions were grouped into two seasonal periods (primary session) for analysis, spring (from March to May) and autumn (from September to November).

The goodness-of-fit (GOF) was evaluated using the Cormack–Jolly–Seber (CJS) as no specific GOF is available for the Robust Design model (Kendall et al. 1995). We performed exact tests, as commonly recommended for capture–recapture data, using U-CARE (Choquet et al. 2009), and overdispersion was evaluated using the variance inflation factor ĉ, estimated as the ratio of the global χ² statistic to its degrees of freedom (Burnham and Anderson 2002). Because only a few juvenile individuals were tagged at Le Porge (n=6) and Les Tours (n=4), and most juveniles at Le Plan (n=110) were captured in only one season, we excluded juveniles from the analyses to avoid transience effects. We only considered adults with a determined sex for analyses, excluding a small number of individuals at Le Plan (n=36).

For the models, capture and recapture probabilities were considered equivalent because there was no trap-dependence in the data (see Results).

Capture probabilities are expected to vary over time, as they generally depend on sampling effort, survey method, sampling period, and observer in amphibians (Mazerolle et al. 2007). We thus fitted models in which capture probability varied with season (season), additive effects of season and sex (season+sex), additive effects of season and year (season+yr) or additive effects of season, sex, and year (season+sex+yr).

Survival and recruitment were considered either constant (.), variable according to year (yr), season (season), spring or autumn period (S/A), variable according to sex (sex), or variable according to various combinations of these effects (season + sex; season + yr; season * sex; season * sex + yr; S/A + sex; S/A + yr; S/A * sex; S/A * sex + yr).

The number of models using all combinations of the effects we tested was too high to be fitted exhaustively. Accordingly, we proceeded step by step as follows:

First, we tested all the combinations on capture probability, while the other parameters depended on season*sex+year. Second, we kept the best model on capture probability, considered recruitment as season*sex+year dependent, and tested the different effect combinations on the survival probability. Finally, we kept the best models on capture and survival probabilities, then tested the different hypotheses on recruitment. We thus obtained a set of 28 models for the three sites.

All models were fitted using the software MARK 6.2 (White and Burnham 1999). Models were compared using AICc adjusted for small sample sizes (Burnham and Anderson 2002). Because several models received similar support, we used model averaging to obtain the average estimates (Burnham and Anderson 2002). All estimates were provided with their 95% confidence intervals (95%CI). Non-overlapping 95% confidence intervals were interpreted as indicative of strong differences among estimates (White and Burnham 1999).

At Le Porge, the dataset comprised 726 captures of 328 individually marked toads (214 males, 104 females, 6 juveniles, and 4 unsexed adults). At Le Plan, it consisted of 1,496 captures of 741 individually marked toads (279 males, 316 females, 110 juveniles and 36 unsexed adults). At Les Tours, the dataset included 233 captures of 111 individually marked toads (43 males, 64 females and 4 juveniles). The Pradel Robust-Design dataset included 713 capture-recapture events involving 318 adult individuals at Le Porge, 1320 events involving 595 adults at Le Plan, and 229 events involving 107 adults at Les Tours.

The observed sex ratio (male number/female number) was male-biased at Le Porge (2.06) and female-biased at both Les Tours (0.67) and Le Plan (0.88). Estimated sex ratios did not vary across seasons at Le Porge, as indicated by the overlap of confidence intervals, whereas seasonal variation was observed at the two other sites (Table 2).

There was no trap-dependence in the data (p=0.56 for Le Porge; p=0.056 for Le Plan and p=0.93 for Les Tours).

After accounting for transience, goodness-of-fit tests indicated an excellent fit of the CJS model (χ² = 3.35, df = 12, P = 0.99) with no evidence of residual overdispersion (ĉ set to 1) for Les Tours.

The GOF tests were however, highly significant for the two other sites (χ² = 15.22, df = 4, P = 0.004 for Le Porge and χ² = 48.185, df = 14, P < 0.001 for Le Plan). To account for overdispersion, a variance inflation factor (ĉ = 3.80 for Le Porge and ĉ = 3.44 for Le Plan) derived from U-CARE goodness-of-fit tests was applied in Program MARK, and model selection was based on the Quasi-likelihood Akaike Information Criterion corrected for small sample size (QAICc).

For Le Porge, three models were considered equivalent (ΔAICc<2) (Suppl. material 1: table SS1). These models accounted for (i) a seasonal effect on capture probability; (ii) an interaction effect of year or period (spring or autumn) on survival probability; and (iii) an interaction effect of sex and season with an additive effect of year on seniority probability.

For Le Plan, four models were considered equivalent (ΔAICc<2) (Suppl. material 1: table SS2). These models accounted for (i) an effect of session, sex, and year on probability of capture; (ii) an effect of period (S/A) with sex or year on survival probability; and (iii) an interaction effect of sex with season on seniority probability.

For Les Tours, two models were considered equivalent (ΔAICc<2) (Suppl. material 1: table S3). These models included (i) a seasonal effect on capture probability; (ii) an effect of sex and season on survival probability; and (iii) an effect of sex or a combined effect of sex and period (S/A) on seniority probability.

Capture probabilities depended on capture season across all three sites, as well as on sex and year at Le Plan. The capture probabilities ranged from 0.07 (95%CI 0.05–0.10) to 0.22 (95%CI 0.13–0.34) (Fig. 2), and were highest at Le Plan [mean 0.17 (95% CI 0.11–0.24)], followed by Le Porge [mean 0.12 (95%CI 0.07–0.22)], and Les Tours [mean 0.11 (95% CI 0.06–0.19)].

Estimated seasonal population sizes were relatively similar between Le Porge and Le Plan, ranging from 200 to 350 individuals per season, whereas Les Tours supported only 50 to 100 individuals, i.e. up to six times fewer than the other sites. Across the three sites, estimated population size was generally higher in spring than in autumn, except in 2019 at Les Tours (Fig. 2). At Le Porge, males were also more numerous than females.

Survival probabilities depended on the year or period at Le Porge, on the period and sex at Le Plan, and on the season and sex of individuals at Les Tours. They ranged from 0.33 (95%CI 0.11–0.65) to 0.99 (95%CI 0.98–1) (Fig. 3). For descriptive purposes, we also report site-specific mean survival estimates obtained from constant models to facilitate broad comparisons among populations. These estimates suggest lower mean survival at Le Plan [0.63 (95%CI 0.55–0.67)] than at Le Porge [0.76 (95%CI 0.68–0.86)] and Les Tours [0.83 (95%CI 0.70–0.91)], with non-overlapping confidence intervals.

Survival probabilities were marginally higher for males than for females at Le Porge and Le Plan, whereas the opposite pattern was observed at Les Tours. Survival probabilities were also slightly lower between spring and autumn than between autumn and the following spring at the Mediterranean sites, and especially at Le Plan (Fig. 3), despite overlapping confidence intervals.

Based on model selection, adult recruitment probabilities depended on both season and sex at Le Porge and Le Plan, and on sex only at Les Tours (Suppl. material 1: table SS1–S3). Estimates ranged from 0.01 (95%CI 0.00–0.03) to 0.81 (95%CI 0.67–0.90) (Fig. 3). To summarise overall differences among populations, we additionally derived site-specific mean recruitment estimates from constant models. These estimates were higher at Le Plan [0.40 (95% CI: 0.32–0.47)] than at the other sites [Le Porge: 0.11 (95% CI: 0.03–0.42); Les Tours: 0.12 (95% CI: 0.05–0.27)], with confidence intervals showing, respectively, little to no overlap between Le Plan and the other sites. Recruitment probabilities were also slightly higher for males than for females at Les Tours, without a clear pattern emerging at the other sites (Fig. 3).

Capture probabilities did not differ between seasons (spring and autumn) but varied among sites. Probabilities were highest at Le Plan, followed by Le Porge and Les Tours. This variation can be explained by differences in sampling effort, with shorter sessions at Les Tours (two intra-night sessions versus three at the other sites) and fewer observers at Le Porge (one versus four at Le Plan), even though total time spent was similar. At Le Porge, searches were spread throughout the night, sometimes extending until dawn, a period less favourable for amphibian activity (Priol 2015).

These capture probabilities obtained in the present study are comparable to, and slightly higher than, those reported by Lorrain-Soligon et al. (2022) for the same species using similar methods but with more observers. Nevertheless, the relatively low capture probabilities highlight the importance of multiple-night monitoring for spadefoot toads, whose individuals are not active every night (Ryan et al. 2015), as also reported in radiotracking studies (Eggert 2000; Timm et al. 2014; Priol 2015). Indeed, seasonal estimates of active adults represented approximately 70–100% of marked individuals at Le Porge and Les Tours, but less than 50% at Le Plan, reflecting differences in surface activity and detection probability among sites. Activity patterns within the genus Pelobates are strongly shaped by thermohygrometric constraints, as individuals exhibit sensitivity to both air and soil temperature and humidity (Priol 2015). In particular, warmer and drier nocturnal conditions have been shown to reduce surface activity (Székely et al. 2017; Natchev et al. 2022). These findings underscore the necessity of sustained monitoring to study the population dynamics of elusive burrowing species. For CMR studies on P. cultripes, we recommend conducting at least five capture sessions per season, each lasting approximately three hours, ideally with several observers present (sharing the survey effort) during the species’ two peak activity periods in spring and autumn. This effort should, of course, be adjusted to each site and population density, depending on the estimated capture rates.

Across the three sites, especially at the two largest populations (Le Porge and Le Plan), individual abundance was higher in spring than in autumn (by several dozen individuals), indicating greater surface activity in spring, regardless of short-term meteorological variation. This pattern suggests that, at the studied latitudes, the main activity period for this species occurs in spring. Although our data concern terrestrial activity and not reproductive activity per se, this seasonal pattern is consistent with the biannual breeding phenology described for P. cultripes in Mediterranean intermittent streams, where spring breeding events are more regular and successful than autumn ones (Duguet 2017). Estimated seasonal population sizes at Le Porge and Le Plan (200–350 active adults per season) were considerably larger than at Les Tours (50–100 adults), and are broadly comparable to annual adult population size estimates (304–352 adults) reported for a Mediterranean population in central Spain (Valdemanco) using the same CMR approach (Gutiérrez-Rodríguez et al. 2017).

Sex ratios fluctuated seasonally and differed among sites, being male-biased at Le Porge, variable at Le Plan, and female-biased at Les Tours. These differences likely reflect a combination of phenological variation, sex-specific site attendance, and skipped breeding events among females (Frétey et al. 2004; Muths et al. 2010). The male-biased sex ratio at Le Porge is consistent with Lorrain-Soligon et al. (2022), who showed that distance to the coastline is associated with variation in sex ratio, age ratio, and morphology in P. cultripes, suggesting that salinity and sea-spray deposition may influence population structure and potentially carry demographic consequences for coastal populations. Nevertheless, a comparable male bias was also systematically reported in the Mediterranean population of Valdemanco (166–186 males vs. 138–166 females; Gutiérrez-Rodríguez et al. 2017), suggesting that this skew may not be exclusive to oceanic environments. Taken together, these comparisons suggest that sex ratios in P. cultripes are strongly context-dependent, varying not only across seasons but also among sites differing in landscape context and local environmental conditions.

Le Plan showed the lowest survival (mean 0.63) compared to Le Porge (0.76) and Les Tours (0.83). These differences may reflect interannual environmental variation, as the study sites were not monitored during the same periods, particularly Le Porge, which was surveyed several years earlier than the other sites. They may also be associated with contrasting environmental contexts among sites. In this second case, the lower apparent survival observed at Le Plan could be consistent with the stronger anthropogenic pressures affecting this site. Indeed, repeated soil ploughing, agrochemical exposure, and the immediate proximity of a busy road (average 2,028 vehicles/day and average 25 vehicles/hour during the night; Travers 2023) may contribute to increased adult mortality, in line with patterns documented in other amphibian populations inhabiting agricultural landscapes (Guillot et al. 2016; Berger et al. 2018; Boissinot et al. 2019; Renoirt et al. 2024) and in close proximity to a road network (Hels and Buchwald 2001; Gibbs and Shriver 2005; Glista et al. 2008). At Le Porge, apparent survival estimates were slightly lower than at Les Tours, although confidence intervals largely overlap, indicating considerable uncertainty in this difference. If such a difference exists, it may reflect processes such as permanent emigration within a connected pond network or local environmental constraints (e.g. coastal influences including increasing salinity), which can affect amphibian reproduction and survival (Thirion 2014; Albecker and McCoy 2017; Lorrain-Soligon et al. 2024, 2025). Conversely, at Les Tours, higher survival estimates may be consistent with the presence of wide buffer zones between breeding habitats and roads and the lack of use of moderately disturbed zones (moderate tillage) by P. cultripes. However, the small population size and low connectivity still represent substantial vulnerabilities and increase extinction risk for this population (Caughley 1994; Gardner 2001). Populations maintained at low numbers are precisely those most vulnerable to demographic stochasticity and loss of genetic diversity (Caughley 1994), processes that can lead to an extinction vortex, whereby demographic, genetic, and environmental factors interact synergistically and reinforce each other, increasing the likelihood of population extinction (Gilpin and Soulé 1986). Although these contrasts strongly suggest that anthropogenic disturbance is a major driver of survival differences among sites, differences in monitoring periods, potentially associated with interannual weather variation, may also have contributed to the observed patterns.

Across all sites, survival between spring and autumn was slightly lower than between autumn and spring. Although this pattern should be interpreted with caution given the overlap in confidence intervals, it may be consistent with the energetic and physiological costs associated with reproduction. This difference appeared more pronounced at Le Plan, although seasonal differences were not robustly supported at this site. If such a seasonal pattern exists, it could be related to local environmental pressures, particularly at Le Plan, where intensively managed agricultural practices may exacerbate post-breeding mortality. Indeed, this fossorial species occurs at high density at Le Plan, where it inhabits intensively managed agricultural areas that are regularly ploughed and treated with pesticides. While evidence for sex-related variation in survival remained weak, the observed tendency towards lower female survival at Le Porge and Le Plan is consistent with mechanisms previously described in amphibians. Female-specific energetic and mechanical costs, including ovum production, migration to breeding ponds, and male harassment, have been associated with reduced female survival (Loman and Madsen 2010; Dittrich and Rödel 2023). Alternative explanations, such as sex-specific site-use strategies, may also be involved (Sinsch 1992; Staufer et al. 2023). In contrast to survival, adult recruitment probabilities were substantially higher at Le Plan (mean = 0.40) than at Le Porge and Les Tours (0.11–0.12), which may be indicative of faster adult turnover at the site exposed to the highest anthropogenic pressure. Although direct causality cannot be established from our observational data, this pattern is compatible with compensatory recruitment mechanisms described in other amphibian species exposed to anthropogenic pressures (Muths et al. 2011; Cayuela et al. 2022). For instance, in Bombina variegata, reduced adult survival has been associated with increased recruitment in human-altered environments (Cayuela et al. 2022). However, given the marked differences in life-history strategies between B. variegata and P. cultripes, the relevance of such a mechanism to our study system remains uncertain. If demographic compensation occurs in these populations, it could arise from density-dependent processes operating during larval or juvenile stages, which may buffer variation in early-life survival (Berven 1990; Vonesh and De la Cruz 2002). Under this hypothesis, the higher recruitment rate observed at Le Plan could partly offset lower survival, whereas the combination of lower recruitment and higher survival at Le Porge and Les Tours may reflect alternative, slower demographic strategies.

The results of this study highlight site-specific threats that call for targeted and timely conservation actions to ensure the persistence of P. cultripes, a species in severe decline across both Mediterranean regions (Durand 2015; Renet et al. 2023) and the Atlantic coast (Thirion 2002), and currently listed as “Endangered” (EN) or “Critically Endangered” (CR) in all French regions where it still occurs (Le Moigne and Jailloux 2013; LPO Rhône-Alpes 2015; Boissinot and Roques 2016; Marchand et al. 2017; Catil and Pottier 2026).

At Le Plan, where descriptive survival estimates were lowest, reducing road mortality represents a key conservation priority. Toad tunnel installation along major migration routes (Rytwinski et al. 2016; Helldin and Petrovan 2019; Marcelino et al. 2025), combined with the creation of new breeding ponds encompassing diverse hydroperiods (Fortuna et al. 2006; Moor et al. 2022), could contribute to enhancing the demographic resilience of this population under increasing climatic variability. Before-After Control-Impact (BACI) designed evaluations should accompany any mitigation structure (e.g., tunnels, drift fences) to ensure evidence-based refinement of management actions. It would also be interesting to study the habitat selection of P. cultripes in agricultural contexts before promoting fallows, as, similar to what Churko et al. (2024) observed for Epidalea calamita, this species may be particularly attracted to highly open areas intensively used by humans.

At Le Porge, regular monitoring of pond salinity and hydrological conditions is necessary to anticipate the future effects of increasing salinity and changes in hydrological regimes (Mazhar et al. 2022) on reproduction and juvenile survival. The creation of ponds further inland within the dune system could help anticipate these changes.

At Les Tours, the higher survival rates highlight the value of existing buffer zones and low-productivity agricultural land. However, the small population size and low connectivity with other populations warrant the enhancement of functional connectivity through the creation or restoration of stepping-stone ponds within dispersal distance. A study of 12 amphibian species indeed indicates that the probability of long-term pond occupancy more than triples when new ponds are created within 500 m of two to four occupied ponds (Moor et al. 2024). Movement data collected at Le Plan and Les Tours through CMR monitoring, and at Le Porge through radiotracking, revealed substantial individual variability, with maximum distances ranging from a few metres to 700 m. Mean maximum distances were approximately 44 m for males and 74 m for females at Le Plan (Renet et al. 2023), and 52 m (sexes combined; range: 15–253 m) at Le Porge (Priol 2015). These values are consistent with the low but non-negligible dispersal capacity reported for P. cultripes across its range, where movements exceeding 700 m have occasionally been recorded (Gutiérrez-Rodríguez et al. 2017; Sánchez-Montes and Martínez-Solano 2023). Importantly, in the Calavon watershed, newly created or restored ponds were successfully colonised within two years of their establishment (Renet et al. 2023), demonstrating that connectivity-enhancing measures may be rapidly effective when implemented within the species’ functional dispersal range. These ponds were located 300 m and 700 m, respectively, from a pond where reproduction of the species had been confirmed.

Across all sites, assessing soil management practices and water quality is crucial to understand pressures from pesticides, ploughing, and land-use changes.

Effective conservation of P. cultripes in France ultimately requires coordinated, long-term strategies underpinned by standardised monitoring protocols to detect demographic trends, prioritise actions, and evaluate their effectiveness (Borzée et al. 2025). In this context, the development of a dedicated National Action Plan (Plan National d’Actions, PNA) would provide a coherent operational framework to align stakeholder efforts and ensure species-level conservation at the national scale.