Environmental conditions and body temperature of anurans in mountain regions of the Atlantic Forest, Brazil

1 Introduction

Anuran behavior and physiological functions, such as growth, development, and reproduction, are strongly influenced by thermal conditions, as their body temperature (T_b) is largely determined by environmental temperatures (e.g (Huey, 1982; Abram et al., 2017; Bovo et al., 2018)). Maintaining T_b within a physiologically suitable range is essential for performance and fitness, as even small deviations beyond thermal limits can be detrimental or lethal (Navas, 2002; Algiriyage et al., 2020).

Mountainous regions in tropical areas experience greater daily thermal fluctuations than lowland environments, resulting in pronounced microclimatic heterogeneity (Navas, 1996, 1997; Siqueira and Rocha, 2013). This complexity creates a mosaic of thermal patches that allow ectotherms to select specific microhabitats to maintain their T_b within a range that maximizes thermal fitness, defined as the capacity to optimize physiological performance, distribution, and long-term persistence under varying environmental conditions (Navas, 1996; Abram et al., 2017). However, these regions are currently undergoing significant shifts due to elevation-dependent warming (EDW), a phenomenon characterized by accelerated temperature increases at higher altitudes compared to lower regions (Mountain Research Initiative EDW Working Group, 2015). Such warming trends pose a severe threat to tropical ectotherms, which often exhibit narrow thermal tolerance breadths and are physiologically specialized to stable thermal regimes (Tewksbury et al., 2008; Laurance et al., 2011; Nowakowski et al., 2017). Consequently, understanding the physiological responses of these species to climate warming in montane ecosystems is critical for predicting species’ vulnerability and potential for distributional shifts.

Because most anurans are thermoconformers, their body temperature closely tracks environmental conditions, limiting their capacity for active thermoregulation (Brattstrom, 1963). As a result, thermal variation along elevational gradients may be associated with selective pressures that drive evolutionary responses such as increased tolerance to temperature fluctuations or restriction of activity to specific microhabitats or time periods (Navas, 1996; Lambrinos and Kleier, 2003). Elevational gradients, therefore, provide a valuable spatial framework for understanding how species may respond to ongoing climate change.

In addition to environmental conditions, body size and mass are critical traits influencing thermal dynamics in ectotherms by dictating thermal inertia, i.e., the capacity to resist changes in T_b through heat storage, and influencing the rate of heat exchange with the surrounding environment (Carey, 1978; Tracy et al., 2010). Because mass-specific metabolic rates and surface area-to-volume ratios are non-linearly related to total body mass, larger individuals are expected to exhibit more stable T_b under fluctuating conditions due to higher thermal inertia (Carey, 1978; Tracy et al., 2010). Conversely, smaller individuals, which possess higher mass-specific surface areas, may facilitate more rapid thermal equilibration, allowing them to track environmental temperature shifts more closely (Olalla‐Tárraga and Rodríguez, 2007). Furthermore, body mass is often correlated with physiological capacity and energy expenditure, which can constrain thermoregulatory behavior in complex montane landscapes (Pounds et al., 1999; Olalla‐Tárraga and Rodríguez, 2007; Mokhatla et al., 2019). Despite these physiological implications, the extent to which body mass modulates T_b and its variability in natural populations remains unclear, particularly in tropical montane environments where high environmental heterogeneity places distinct constraints on the thermal budgets of ectotherms.

Whereas several aspects of anuran thermal ecology have been extensively investigated, including the effects of temperature on disease susceptibility, larval development, and reproductive cycles (e.g (Berger et al., 2004; Galindo et al., 2018; Sauer et al., 2018), the role of morphological traits in shaping T_b dynamics remains less understood. Here, we evaluate how environmental conditions (air and substrate temperatures, T_a and T_s), elevation, body size, body mass, sex, and microhabitat influence T_b and its variability in anurans inhabiting two mountainous regions of the Brazilian Atlantic Forest. We hypothesize that: (1) anuran T_b is primarily determined by environmental temperatures (T_a and T_s) along the elevational gradient, reflecting a thermoconforming strategy; (2) body size and mass influences mean T_b, reflecting morphological constraints on heat exchange rates and thermal capacity; (3) activity period (diurnal vs. nocturnal) acts as a primary mediator of thermal exposure, with diurnal species exhibiting greater variation in T_b due to higher daily thermal heterogeneity; and (4) microhabitat type acts as a key mediator of thermal status, with distinct environmental properties at the micro-scale modulating anuran T_b. Our focal species assemblage includes several taxa of conservation concern, including range-restricted endemics such as Ischnochnema parnaso, which is classified as Vulnerable by the IUCN Red List (IUCN, 2021). Understanding the thermal constraints of these species is thus critical for predicting their persistence in the face of ongoing climate change.

2 Material and methods

Thermal data were collected across two major mountain chains in the Brazilian Atlantic Forest: the Serra da Mantiqueira (Itatiaia National Park; PARNA Itatiaia) and the Serra do Mar (Serra dos Órgãos National Park; PARNASO). At PARNA Itatiaia, sampling was conducted in November and December 2016 at elevations ranging from 2,100 to 2,450 m a.s.l. (coordinates: 22°21’22.15”S, 44°44’07.58”W to 22°23’07.41”S, 44°40’44.94”W). At PARNASO, sampling took place across multiple years (October–December 2015; January, February, and October 2016; and October 2017) at elevations ranging from 980 to 2,200 m a.s.l (coordinates: 22°27’26.09”S, 42°59’54.15”W). This sampling design intentionally integrated multi-year observations at PARNASO with a focused snapshot at PARNA Itatiaia, providing a robust assessment of thermal biology that captures both temporal breadth and distinct spatial thermal regimes across these two representative montane sites. The elevation gradient served as a proxy for the broad thermal heterogeneity characterizing montane Atlantic Forest ecosystems, allowing us to capture significant variation in T_a and T_s. All procedures were conducted under the federal SISBIO license number 51410-2. The protocol was submitted to the institutional ethics department of the State University of Rio de Janeiro (CEUA) and was considered exempted because this study involved strictly field-based sampling.

Two to four researchers carefully searched the animals under and on leaf litter, trunks, branches, roots, and stones, and on shrub and herbaceous vegetation, on the ground and around temporary and permanent ponds and streams. Although sampling was primarily conducted between 30 min and 3 h after dusk to standardize nocturnal encounters (Rocha et al., 2015), Melanophryniscus moreirae was sampled during its peak diurnal activity period in high-altitude rocky outcrops. All individuals captured were active (e.g., calling, moving, or foraging) at the time of measurement. Inactive or sheltered individuals were excluded to ensure that T_b reflected realized body temperatures during activity (Hertz et al., 1993; Navas et al., 2008). Species were classified according to their primary activity period (nocturnal vs. diurnal) based on Haddad et al. (2013).

We captured the individuals by hand and measured frog T_b from the dorsal skin surface and T_s using a portable infrared digital thermometer (scale: -50 to 550 °C, to the nearest 0.1 °C). While infrared thermometry is subject to potential bias from environmental reflectance and emissivity in small-bodied ectotherms, it was chosen to allow for rapid, non-invasive measurements, minimizing handling time and the resulting physiological stress that can inflate T_b readings (Rowley and Alford, 2007; Taylor et al., 2021). To mitigate potential errors related to environmental reflectance, skin emissivity, and animal posture, we maintained a fixed distance of 30 cm and verified that the thermometer’s field of view focused exclusively on the animal’s dorsal surface, thereby avoiding interference from the background substrate (Rowley and Alford, 2007; Taylor et al., 2021). Air temperature was measured 1 cm above the substrate using a Schultheis^® quick-reading thermometer (0.2 °C accuracy). Substrate and air temperatures were measured immediately after capture at the same location. We measured frog body size (snout-vent length) with a digital caliper, body mass with a digital scale (0.01 g accuracy), and elevation using a Garmin GPSmap 60CSX immediately after thermal trait measurements. Additionally, we recorded individual sex and the specific microhabitat type at the site of capture (i.e., arboreal, bromeliad-dwelling, riparian, or terrestrial).

We compared the thermal variance between M. moreirae and the pooled nocturnal species using Levene’s test to assess differences in intraspecific thermal variation (Levene, 1960), implemented in the car package (Fox and Weisberg, 2018) in R v3.5.1 (R Core Team, 2026). Then, to evaluate the influence of environmental predictors on T_b, we employed a Linear Mixed-Effects Model (LMM) implemented in the lme4 package (Bates et al., 2014) in the R statistical environment v3.5.1 (R Core Team, 2026). To ensure the ecological consistency of our thermal analysis, the model was restricted to nocturnal species, which represent the core of our dataset and share similar physiological constraints. We excluded the diurnal species M. moreirae from this LMM to prevent statistical bias arising from the combination of distinct diel thermal strategies and the significant imbalance in sample sizes (n=1 diurnal vs. n=15 nocturnal species).

We included species identity as a random intercept to account for the non-independence of data points from the same taxon. Body mass, SVL, T_a,T_s, sex, and microhabitat type were included as fixed effects to account for potential biological and niche-based variation in thermal requirements. The significance of fixed effects was assessed using Satterthwaite’s degrees of freedom approximation via lmerTest (Kuznetsova et al., 2017). We assessed the normality of residuals, homogeneity of variance, and multicollinearity using the performance package (Lüdecke et al., 2021). The model demonstrated appropriate fit, with no significant deviations from normality or heteroscedasticity. To test for potential residual effects of elevation independent of microclimate, we conducted a supplementary LMM considering T_b as the response variable, with T_a, T_s, and altitude as fixed effects and species identity as a random effect. Finally, we tested whether morphological traits influence thermal variability by regressing species-specific T_b variability (quantified as the standard deviation of T_b per species) against mean body mass using a linear regression model.

3 Results

We measured T_b of 146 individuals from 16 anuran species, of which only Melanophryniscus moreirae was diurnal (Table 1). Mean body temperature ranged from 12.2 °C in Ischnocnema melanopygia to 22.1 °C in Cycloramphus brasiliensis. The diurnal species Melanophryniscus moreirae exhibited significantly higher intraspecific variance in T_b compared to the nocturnal assemblage (Levene’s test: F_1,142 = 6.98, p = 0.009). Among the nocturnal species, greatest intraspecific variation in T_b was observed in Aplastodiscus arildae (19.8 ± 5.4 °C; Table 1).

Table 1. Body (T_b), air (T_a) and substrate (T_s) temperatures for 16 anuran species, along with demographic (sex; M= Male, F=Female, NA = Not Available) and ecological (microhabitat) information at Parque Nacional da Serra dos Órgãos and Parque Nacional de Itatiaia, Rio de Janeiro, Brazil.

Data are presented as mean ± SD (range) for species with n > 2, and as individual values for species with n ≤ 2.

Correlation analyses revealed a significant positive relationship between body mass and environmental temperatures (r = 0.60, p < 0.001 for T_s; r = 0.59, p < 0.001 for T_a). Given this strong collinearity, we assessed the effect of morphological and environmental variables in the full LMM. The model revealed that T_s was the primary significant predictor of T_b in nocturnal anurans (t = 2.80, p = 0.0067; Figure 1, Table 2). In contrast, T_a did not significantly influence T_b when additional covariates were included in the model (t = 1.52, p = 0.1327; Table 2). We also found that microhabitat specialization significantly influenced T_b, with individuals captured in riparian habitats exhibiting higher body temperatures compared to those in other microhabitats (t = 2.48, p = 0.0158; Table 2).

Table 2. Summary of the Linear Mixed-Effects Model (LMM) assessing the predictors of anuran body temperature (T_b).

Species identity was included as a random effect. Significant predictors (p < 0.05) are indicated in bold.

Conversely, biological traits, including body mass, body size, and sex, did not show significant effects on T_b (p > 0.05 for all). The random effect variance attributed to species identity was negligible (variance = 0.000), indicating that the thermoconforming strategy is a generalized physiological pattern across the studied nocturnal assemblages, rather than a trait specific to individual species. The supplementary analysis indicated that elevation did not significantly predict T_b (t = −1.605, p = 0.111), highlighting no additional signal from elevation at the studied scale. Lastly, we found no significant relationship between species-specific T_b variability and mean body mass (F_1,9 = 0.2079, p = 0.6592), indicating that larger individuals did not exhibit greater T_b stability in our dataset.

4 Discussion

Our results indicate that anuran T_b is primarily driven by local environmental conditions, confirming a predominantly thermoconforming strategy in the studied assemblages (Brattstrom, 1963; Navas, 1996). Although some behavioral thermoregulation has been documented in amphibians, physiological constraints associated with cutaneous respiration and water balance likely limit their capacity to buffer T_b against environmental variation (Tracy, 1976; Tracy et al., 2010; Tattersall et al., 2012). However, behavioral choices remain ecologically significant. For example, maintaining even slightly elevated T_bs can influence fitness outcomes, such as providing protection against infections like the chytrid fungus (Rowley and Alford, 2013; Sauer et al., 2018). Therefore, while our study reinforces a thermoconforming trend, behavior remains a critical component of the thermal niche.

In the relatively preserved montane forests of our study sites, dense canopy cover may act as a thermal buffer, dampening the amplitude of microclimatic fluctuations (Pounds et al., 1999; Sheldon et al., 2011). Our results provide partial empirical support for this mechanism. Microhabitat type significantly influenced anuran T_b, with individuals in riparian habitats exhibiting higher body temperatures compared to those in other microhabitats, suggesting that riparian environments may offer distinct thermal opportunities or constraints that modulate T_b. However, as we did not directly quantify canopy cover, the broader canopy-induced buffering mechanism remains a hypothesized contributor to the thermal stability observed. This high sensitivity to the thermal environment implies a precarious vulnerability in anthropized or modified landscapes (Deutsch et al., 2008; Laurance et al., 2011). In areas where forest cover is reduced or fragmented, the loss of canopy-induced buffering would likely lead to significantly higher microclimatic thermal extremes, threatening amphibian persistence (Nowakowski et al., 2017). Given that these species lack the physiological capacity to buffer their internal temperatures through active thermoregulation, they may become susceptible to thermal stress and dehydration in open, modified environments (Navas, 1996; Almeida-Gomes et al., 2015).

The highest intraspecific range in T_b was observed in the nocturnal species Aplastodiscus arildae, occurring at intermediate elevations. This pattern contrasts with expectations that high-elevation species experience greater thermal variability (Siqueira and Rocha, 2013), although it may be influenced by a limited sample size. Although our sampling design covered a broad elevational gradient, our results highlight that anuran thermal variance is directly linked to local microclimatic conditions (T_a and T_s). This elevational pattern is consistent with findings from other mountain systems, such as the eastern Nepal Himalaya, where thermal variation along gradients also underscores the tight coupling between environment and physiology (Khatiwada et al., 2020). While elevation is a key distal factor that modulates temperature (Navas et al., 2013), our findings suggest that T_s serves as the proximal driver of anuran T_b. This implies that the physiological constraints of these species are coupled to the immediate thermal environment, which varies along the elevational gradient but is experienced by the organisms at the scale of substrate contact.

Our results for the diurnal species M. moreirae suggest that diurnal activity is associated with higher T_b variability, consistent with exposure to daily thermal heterogeneity (Putnam and Bennett, 1981), whereas the nocturnal species experience more constrained and stable thermal regimes. However, this observation is based on a single diurnal species and requires testing in a broader range of diurnal taxa to assess the generality of this pattern across montane anuran assemblages. While greater thermal flexibility in diurnal species, such as M. moreirae, may potentially confer resilience to short-term fluctuations (Canale and Henry, 2010), the lower intraspecific variation observed in nocturnal species suggests more constrained thermal regimes and potentially higher sensitivity to nocturnal warming trends (Tewksbury et al., 2008).

Our analyses demonstrated that T_s were the main drivers of T_b variation, while SVL, body mass, and sex were not statistically significant. This aligns with observations in other systems, such as in southern African anurans, where ambient temperature similarly overrides body mass in determining thermal status (Mokhatla et al., 2019). Additionally, the lack of a significant effect of sex suggests that thermal constraints are experienced similarly across demographics in these assemblages. The strong collinearity between body mass and environmental temperatures found here indicates that the environmental variables effectively capture the thermal variance that might otherwise be attributed to morphological traits, thus explaining why body mass lost significance in the full model. These results suggest that environmental forcing exerts stronger influence on T_b than subtle morphological buffering capacity at the spatial scale considered here (Woods et al., 2015), confirming that these anurans function predominantly as thermoconformers within these montane assemblages (Ohlberger, 2013). It is important to clarify, however, that our LMM specifically assessed the influence of body size and mass on the mean T_b. Therefore, to explicitly test the thermal inertia hypothesis, we assessed whether T_b variability (quantified as the standard deviation of T_b per species) varied as a function of body mass. Our analysis revealed no significant relationship (p = 0.6592), indicating that larger individuals did not exhibit significantly more stable T_b in our dataset. While thermal inertia suggests that larger individuals might theoretically maintain more stable T_b (i.e., lower variance), our empirical results indicate that these morphological traits do not significantly predict mean T_b within these assemblages. Future studies targeting T_b variance over time would be required to explicitly test the hypothesis of thermal stability mediated by body size. In this sense, under colder conditions, smaller body sizes may be favored due to faster heat exchange rates, whereas larger individuals may face constraints on heat acquisition and maintenance, particularly in thermoconforming organisms (Tracy et al., 2010).

While our results strongly support a thermoconforming strategy, several ecological and physiological factors may influence these patterns. First, phylogenetic constraints may underlie thermal requirements (Von May et al., 2017). Although we included species identity as a random intercept in our LMM, we did not explicitly quantify phylogenetic signal. Future comparative studies using phylogenetically informed models are necessary to disentangle evolutionary history from current ecological drivers. Second, our snapshots of T_b likely capture a composite of immediate thermoconforming responses and potential long-term physiological acclimation to local conditions (Hertz et al., 1993). Finally, although we observed a high degree of thermoconformity, we cannot exclude the influence of behavioral strategies, such as postural adjustments or fine-scale microhabitat selection, that may partially decouple T_b from ambient conditions (Navas, 1996; Navas et al., 2013). These behaviors may introduce additional variation into the relationship between environmental and body temperatures, suggesting that our findings represent a conservative estimate of the degree of thermoconformity in these assemblages.

We acknowledge that uneven sampling effort across species limits both mean niche estimation and intraspecific variance analysis. Species with n ≤ 3 provide limited species-level inference, resulting in a model primarily driven by more abundant taxa such as Bokermannohyla circumdata (n = 22), Scinax duartei (n = 14), and Cycloramphus brasiliensis (n = 13). It is worth noting that our most abundant species, Melanophryniscus moreirae (n = 50), was excluded from the main LMM analysis to avoid conflating distinct thermal strategies. In this sense, although our use of Linear Mixed-Effects Models (LMMs) with species as a random effect helps mitigate the impact of unequal sample size (Bolker et al., 2009), specific thermal niche inferences for sparsely sampled species remain constrained. Similarly, our estimates of intraspecific thermal variance are sensitive to sample size, so the high T_b range observed in Aplastodiscus arildae (n = 3) must be interpreted as a preliminary observation rather than a definitive characterization of its thermal niche.

We also acknowledge the inherent limitations of infrared thermometry for estimating T_b in small ectotherms. Our approach prioritized minimizing capture-induced thermal bias, which is crucial for field-based ecological studies, and while we kept a fixed distance and focused on the dorsal surface to mitigate biases related to environmental reflectance, emissivity, and animal posture, we recognize that potential measurement noise due to the physiological and morphological characteristics of the skin may occur (Taylor et al., 2021). Therefore, this trade-off should be considered when interpreting precise thermal values at fine scales.

Additionally, while we incorporated categorical microhabitat types into our analysis, this study did not include continuous fine-scale microhabitat covariates such as relative humidity, canopy cover density, or leaf litter depth. While these variables are known to mediate thermal exposure (Huey et al., 2012; Scheffers et al., 2014), our sampling design focused on the relationship between environmental temperatures and anuran T_b across a broad elevational gradient. Consequently, our findings should be viewed as a foundational assessment of thermoconforming strategies in this system. We, therefore, recommend that future studies explicitly incorporate these microhabitat factors to disentangle the complex interplay between thermoconformity and active microhabitat selection in tropical montane anurans.

Similarly, our study was conducted within two national parks in the Brazilian Atlantic Forest, an ecoregion characterized by specific climate regimes and complex topography (Colombo and Joly, 2010; Mendonca Dantas et al., 2011). Consequently, these findings may not generalize to other montane systems or lowland environments with distinct thermal or moisture profiles. We encourage future research to replicate this framework in broader contexts to assess whether the thermoconforming patterns observed here are universal or context-dependent, particularly across varying biogeographic zones.

Taken together, our results indicate that amphibian T_b is primarily associated with local environmental conditions, supporting a predominantly thermoconforming strategy in the studied assemblages. From a climate change perspective, species with narrower thermal variation and stronger reliance on environmental temperatures may be more susceptible to increases in thermal extremes (Pottier et al., 2025), particularly in montane systems where upward shifts in temperature regimes may rapidly reduce suitable thermal habitat (Navas et al., 2013; Mountain Research Initiative EDW Working Group, 2015). However, we emphasize that our study provides baseline thermal data and does not directly assess vulnerability metrics such as critical thermal limits, thermal safety margins, or individual acclimation capacity, which should be the focus of future assessments.

The high thermal sensitivity observed in our study is particularly concerning for species already at risk of extinction. Given that our focal assemblage includes threatened endemics like M. moreirae, the reliance on local microclimatic stability and the subsequent risk posed by forest degradation highlights an urgent need for protective measures that safeguard both thermal refugia and habitat connectivity (Scheffers et al., 2014; Morelli et al., 2016). To translate these findings into actionable conservation strategies, we propose a tiered approach that prioritizes maintaining thermal heterogeneity. First, based on our finding that microhabitat specialization influences T_b, we recommend that land management prioritize the protection of riparian buffers and areas with dense, multi-strata canopy cover, which serve as essential thermal refugia by dampening microclimatic extremes (De Frenne et al., 2019; Zellweger et al., 2020). Second, where topographic conditions allow, maintaining elevational connectivity may facilitate local movements among microclimatically suitable habitats under future warming (Heller and Zavaleta, 2009; Brasil-Godinho et al., 2024). While these recommendations are informed by our data on microclimate-driven thermal sensitivity, they should be integrated into broader systematic conservation planning, alongside long-term monitoring of species-specific physiological tolerances to ensure that protected area boundaries remain climatically relevant under future warming scenarios.

Overall, our findings highlight that amphibian thermal biology in these montane environments is strongly governed by external thermal conditions, with limited morphological buffering capacity at the spatial scale examined. Given the rapid warming projected for tropical montane regions, shifts in thermal landscapes may disproportionately affect thermally constrained nocturnal species, potentially altering community structure and elevational distributions. To mitigate these risks, we suggest that conservation strategies focus on identifying and protecting high-altitude microclimatic refugia where canopy structure provides thermal buffering and maintaining connectivity in montane corridors to facilitate potential elevational range shifts. Integrating these landscape-level management actions with physiological data is essential for improving predictions of species vulnerability and guiding conservation strategies under ongoing climate change.

Generative AI statement

The author(s) declared that generative AI was not used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Footnotes

Edited by: Susan C. Walls, United States Department of the Interior, United States

Reviewed by: Rodrigo Barbosa Fontana, Federal University of Santa Maria, Brazil

Santosh Mogali Mogali, Karnatak University Dharwad, India