The study species belong to the Pyrenean clade of the Western Mediterranean lineage of the tribe Leptodirini (Coleoptera, Leiodidae, Cholevinae), distributed in different areas of the Pyrenees and Cantabrian Mountains, from Navarra to Catalonia (Fresneda et al. 2024). We selected four species of this clade (Table 1) in accordance with the following criteria: i) abundance in each of their localities, ii) species of which thermal tolerance and phylogenetic data are available from previous studies (Rizzo et al. 2015; Sánchez-Fernández et al. 2016, 2018; Pallarés et al. 2021; Colado et al. 2022a; Fresneda et al. 2024), and iii) ease of access to the caves.

Adult specimens were collected between March and July 2023. In each one of the caves, a Bluetooth equipped Data-Logger (HOBO MX2301, Onset Computer Corporation, Bourne, USA) was used to record temperature and relative humidity. These were located at the areas with the highest density of individuals, and kept recording for 2 to 3 hours, while the specimens were captured.

Specimens were captured using mouth aspirators and immediately transferred into recipients with moss to maintain a high humidity. The recipients were transported to the laboratory in an electric portable refrigerator at the same temperature of the caves, which was continuously monitored.

Upon arrival at the laboratory in the University of Murcia, specimens were kept under conditions like those of their localities (11ºC and 100% RH) for 72 hours, for initial acclimation. After that, 10 specimens per combined treatment of temperature and humidity (see below) were transferred to the experimental units consisted of plastic trays with the bottom covered in a 3 cm layer of plaster of Paris. These were placed in climatic chambers with temperature and humidity control.

Temperature treatments encompass a thermal range from current temperatures in the collection caves (11ºC), to temperatures that have been shown to be sublethal (20ºC) or lethal (23ºC, 25 °C) within a 7 days exposure period at 100% RH in beetle species of this lineage within the same study area, according to previous studies (Rizzo et al. 2015; Pallarés et al. 2020; Colado et al. 2022a). Such studies used the same methodology employed here to estimate upper thermal limits and all the species tested showed reduced or lack of mortality at 20ºC for 7 days and died within a few days at 23–25ºC. Given the phylogenetic relatedness with the species tested here (Fresneda et al. 2024), we assumed a similar thermal response, which is supported by our results (see Results section). Two humidity conditions (100% and 75% RH) were combined with the four temperatures, resulting in eight combined treatments of temperature and humidity. The 11ºC and 100% RH treatment was used as a standard control for all species (non-stressful conditions). Although the control temperature does not match exactly the habitat temperature of all the species (see Table 1), we made sure that such small temperature difference did not cause physiological stress to the species. We kept a minimum of 10 specimens of each species under control conditions after the experiment and they remained alive for months (up to 12 months in the case of T. impellitieri) showing normal activity and even mating behaviour.

We used programmable climatic chambers with temperature and humidity control (Memmert M360, Germany and Panasonic MLR-352H-PE, Japan). The 100% RH experimental units were covered with a plastic film to prevent desiccation and small holes to provide oxygenation. The 75% RH experimental units were covered with a plastic mosquito mesh, to prevent the beetles from escaping, while maintaining the RH at the 75% value set in the chambers. The humidity in all treatments was monitored in real time (5-minute intervals, for the entire duration of the experiment) with the same Bluetooth equipped Data-Loggers (HOBO MX2301), and the resulting data was exported into .xlsx sheets to check that the conditions did not vary significantly. In every unit, a cotton disc was placed in the centre, to provide a hiding place for the beetles, simulating small crevices or the humid areas beneath rocks that these species encounter in the caves.

The duration of the experiments was 7 days, as in previous thermal tolerance studies with subterranean species of the same lineage and same genera (e.g. Colado et al. 2022a, b). Beetles were fed daily throughout the experiment with humidified fish food pellets. Survival was checked every 24 hours.

Kaplan-Meier survival curves were fitted to visualise and estimate survival probabilities of each species in the different temperature and relative humidity treatments along the duration of the experiment. The average survival time was calculated for each species and treatment, using the restricted mean survival time (RMST) up to a specified time (tau = 7). This RMST represents the area under the Kaplan-Meier curve within that time window. This calculation accounts for the presence of right-censored data (i.e. individuals that survived on day 7). The survival rate of each species in each treatment was also obtained, expressed as a proportion of survival at the end of the experiment. We fitted a Cox proportional - hazards model (Therneau and Grambsch 2000) to assess the effects of temperature, relative humidity and their interaction on the probability of mortality, using the R packages “survival” version 3.8.3 (Therneau et al. 2024) and “survminer” version 0.5.0 (Kassambara et al. 2024). The Cox proportional - hazards model is a semiparametric model used in survival analysis to examine the effect of covariates (predictor variables) on the hazard rate: the rate at which events happen over time (in this case, death). The hazard function h(t) is the instantaneous rate of the event occurring at time t, given survival up to time t. The higher the hazard rate, the greater the impact of the covariates involved in the events. For each predictor, the model returns a regression coefficient (coef), its standard error (se), a Wald test statistic (z), and an associated p-value. The exponentiated coefficient (exp(coef)) represents the hazard ratio: values greater than 1 indicate increased risk of mortality relative to the reference level, while values below 1 indicate reduced risk (Therneau and Grambsch 2000). For example, a positive coefficient for temperature means that higher temperatures increase the hazard of death, whereas a negative coefficient for the temperature × humidity interaction indicates that the effect of temperature differs depending on humidity level. The humidity coefficient compares survival at 75% relative humidity against 100%, with large positive values reflecting substantially higher mortality risk at 75%. Significance was assessed using Wald tests for individual coefficients, and global model fit was evaluated with likelihood ratio, Wald, and log-rank score tests.

To test whether a reduction in humidity alone influenced survival at 11 °C, we compared survival curves between 100% and 75% RH within each species. Survival data were analysed using Kaplan–Meier estimates and compared with the log-rank test, which evaluates whether the overall survival distributions differ between groups. In addition, Cox proportional hazards models with humidity as the sole predictor were fitted to estimate hazard ratios, although in cases of complete or near-complete separation (i.e., when survival in one humidity level was uniformly higher or lower), these models produced unstable estimates. Therefore, interpretation was based primarily on the log-rank tests, which are robust to these situations and directly test the null hypothesis of equal survival across treatments.

Finally, we estimated upper thermal tolerance as the mean lethal temperature (LT₅₀), i.e., the temperature at which 50% of the individuals died after the 7 days of experiment, allowing comparison with values reported for other Leptodirini species in previous studies (Colado et al. 2022a). To obtain the LT₅₀, survival data were fitted to both binomial and quasi-binomial generalized linear models, using the package “brglm” version 0.7.2 (Kosmidis 2021), with the same methodology as in Colado et al. (2022a). The final selected model to obtain the LT₅₀ value was a binomial model with bias reduction. Due to the lack of survivors in the 75% RH treatments at the end of the experiments (see Results section), the LT₅₀ could only be obtained for the 100% RH treatment. All statistical analyses were conducted using R software, v. 4.4.2 (R Core Team 2024).

All the datasets and Rscripts used in the analyses are available at Figshare (DOI: 10.6084/m9.figshare.28902560, link: https://figshare.com/s/48a6b6a148a78c444a0c).

Our findings show that subterranean species are also highly vulnerable to desiccation stress, an overlooked factor in physiological studies of subterranean species. In the absence of heat stress, all the studied species showed a significant and rapid survival decrease under reduced humidity (75% RH) compared to 100% RH. Insects exhibit a wide range of desiccation resistance mechanisms and strategies (Chown et al. 2011), which are shaped by the environmental conditions of their habitats. Species living in arid environments typically evolve features such as thickened cuticles, increased production of waterproofing cuticular hydrocarbons, reduced respiratory water loss, or behavioural adaptations like burrowing or nocturnal activity to minimize exposure to dry conditions (e.g., Hadley 1978; Gefen et al. 2015). In contrast, insects adapted to wetter habitats often show greatly reduced investment in these protective traits (Menzel et al. 2017). Our results suggest that this is the case for subterranean arthropods. This hypothesis is supported by the few available cuticular studies on subterranean taxa. For instance, the cave spider Lycosa howarthi exhibited a lower density of cuticular lipids and a reduced abundance of hydrocarbon classes typically associated with waterproofing compared to its epigean relatives (Hadley et al. 1981).

The responses of the study species to temperature increase under non-stressful humidity conditions (100% RH) were consistent with previous observed thermal tolerance patterns in other species of the Pyrenean Leptodirini clade (Rizzo et al. 2015; Sánchez-Fernández et al. 2016, 2018; Pallarés et al. 2020; Colado et al. 2022a), which are considered among the most heat-sensitive terrestrial arthropods known. Indeed, LT₅₀ values for all species were below 20ºC, and none of the species survived at temperatures > 20ºC (23 or 25ºC) for the 7 days of exposure, regardless of humidity levels. Notably, individuals exposed to reduced humidity died in a timespan ranging from 24 hours to 6 days, even at the control (cave) temperature (see Tables 3–5). The deaths in 11ºC and 75% RH can only be attributed to the decrease in relative humidity, as the temperature remains at a non-stressful level. This means that their physiological limits are greatly determined by the relative humidity. These results are consistent with a previous study that showed that the upper thermal limits of click beetles reduce with decreasing humidity (Riddell et al. 2023). The authors suggested this is likely due to a physiological trade-off between limiting spiracular water loss in dry environments and maintaining adequate oxygen intake, ultimately reducing thermal tolerance. Further research measuring cuticular and respiratory water loss in subterranean species would be valuable to elucidate the physiological mechanisms behind the humidity-dependent reduction in thermal tolerance observed here.

Regardless of the underlying mechanisms, our findings have important conservation implications. Reduced heat tolerance, coupled with a limited thermal plasticity (Pallarés et al. 2021), has raised concerns about the high vulnerability of subterranean invertebrates to climate change. Our results suggest that such vulnerability may, in fact, be underestimated, as increasing aridification could further compromise thermal limits and threaten the persistence of these highly specialized species.

The different sensitivities to environmental changes observed in our experiment among the study species could be related to differences in their evolutionary history and subtle differences in their degree of subterranean specialization, reflected in their morphology. All these four species are specialists of the deep subterranean habitat, and they share the typical troglomorphic traits of the cave-dwelling Leptodirini species (Fresneda et al. 2024), i.e. anophthalmia, absence of wings, reduced pigmentation, elongation of appendages and a reduction of the larval stages (contract or semi-contract life cycle – Cieslak et al. 2014).

Euryspeonomus eloseguii was the most sensitive species, exhibiting the shortest survival times under both humidity treatments and the lowest LT₅₀ value, indicating low tolerance to both desiccation and heat. These results align with those previously obtained for its sister species, E. breuili (Colado et al. 2022), which shares a similar highly specialized habitat and distribution. T. hustachei and T. impellitieri showed intermediate sensitivity, with nearly identical LT₅₀ values, likely due to their closer phylogenetic relatedness. The least sensitive species to reduced humidity was S. ribagorzanus, which had the longest average survival time, although its LT₅₀ value did not very similar to those of the Troglocharinus species. While S. ribagorzanus appears to be the most resistant to both increased temperature and decreased humidity, it is important to highlight that the observed differences are subtle, and this species is still highly sensitive to both temperature increase and reduced humidity.

Our results suggest that even among deep subterranean specialists, small differences in subterranean specialization could be linked to differing physiological tolerances. Future research comparing physiological and morphological data — particularly including species with lower degrees of subterranean specialization (such as those from shallow subterranean habitats, which typically exhibit reduced body size and shorter appendages) — would shed light into the evolutionary relationship between troglomorphy and environmental stress tolerance.

It could be argued that the humidity conditions tested in our experiments are not realistic in subterranean environments. However, low relative humidity values have been already recorded in some caves in Spain, particularly in warmer and dryer regions, with clear (and already observed) negative impacts on subterranean communities. For instance, in our field observations in the cave of the Sant Gervasi castle, located in the Montsec mountain range (Lleida, pre-Pyrenees region), in July 2023, we recorded 14ºC and a relative humidity of 82%, along with dry, dusty soil. We also observed a very low abundance of Speonomites aurouxi (Español, 1965), an endemic Leptodirini species which is often abundant in this small cavity (Fresneda and Salgado 2016). Another recent example of droughts affecting the ecological community of caves was documented by Sendra-Mocholí et al. (2025) in southeastern Iberian caves, some of them with historically rich subterranean faunas. In recent visits to these caves, a sharp decline of invertebrate species richness and abundance was observed, with only a few individuals found restricted to the scarce more humid microhabitats left in the otherwise dry (82–86% RH) and dusty caves. Notably, the historically abundant endemic leiodid beetle Anillochlamys tropica Jeannel, 1909 had almost disappeared in one of the caves. This situation is most certainly a consequence of the severe drought affecting the southeastern Spain, one of the longest and most intense ever recorded in the region. Similar patterns have also been described recently in other Mediterranean regions. A recent study by Galbiati et al. (2025) reported a drastic reduction in abundance and a complete absence of reproduction in populations of the cave spider Meta bourneti in north-western Italy following the severe 2021–2022 drought. This further reinforces the view that prolonged droughts, through their effects on subterranean humidity, can severely compromise both population persistence and ecological function in cave systems.

These field observations and studies support the findings of our study, where moderate reductions in relative humidity caused extreme mortality in cave beetles. Notably, abiotic stressors such as elevated temperatures and reduced humidity not only cause immediate mortality but also compromise the ecological fitness of organisms. Longer-term effects could manifest in altered behavioural patterns, suppressed reproductive and foraging activities or an increased susceptibility to pathogenic infections, and could occur at humidity values higher than those tested here. Therefore, further research, focused on the responses at sublethal values of RH (i.e., higher than 75%), would be extremely valuable to better define tolerance thresholds and understand the broader ecological consequences of humidity decline in subterranean species.

This study was intended to be a first exploration of the previously unknown topic of the impact of a reduction of humidity in the survival of the cave beetles of the Leptodirini tribe, which has a major relevance in the current context of global climate change and increasing droughts. As such, several limitations should be acknowledged, along with possible directions for future research. First, to ensure comparability with previous studies on heat tolerance in Leptodirini species (Colado et al. 2022a, 2022b; Rizzo et al. 2015; Sánchez-Fernández et al. 2016; Pallarés et al. 2019, 2021), we used the same temperature treatments, combined with two humidity levels. Due to other logistical reasons (limited number of specimens and experimental climatic chambers – see below) we were unable to test additional temperature or humidity conditions. Related to this, the selection of the same control temperature (11ºC) for all species could be questioned, as they were collected from caves with temperatures ranging from 9 to 13 ºC (see Table 1). Theoretically, this could induce physiological stress in the species living at higher or lower temperatures than those of the control treatment. Nonetheless, we kept a stock of 10 to 20 individuals of all the four species at 11ºC and 100% RH after the experiments, and almost all the specimens remained alive for 6 months in the case of E. breuili and T. hustachei, and 12 months in T. impellitieri. During this time, we even observed mating behaviour in all species, and they fed normally. Therefore, these conditions were apparently not stressful for the selected species, even in the long term. Future studies, however, should explore potential sublethal stress responses (e.g., oxidative stress; see Pallarés et al. 2020) that may arise from relatively small thermal changes. Similarly, testing responses at humidities between 75% and 100% RH would allow to define more precise humidity physiological thresholds for subterranean beetles. While 75% RH was a highly stressful humidity for the tested species in this study, we have also performed additional experiments with cave species at 80% RH (unpublished data), and the specimens show relatively high survival rates under these conditions. This indicates the existence of a narrow critical humidity threshold below which these species cannot persist. Finally, sample size is another important limitation, due to the difficulty to collect large number of specimens of these species from cave populations. However, the sample size in our experiments (N = 10–12 specimens per treatment) was sufficient for performing robust statistical analyses.