Aim The climate variability hypothesis (CVH) states that a positive relationship may exist between the breadth of thermal tolerance range and the level of climatic variability experienced by taxa with increasing latitude, especially in terrestrial ectotherms. Under CVH, we expected to find a correspondence between both thermal tolerance limits (CTmax and CTmin), ambient extreme temperature and the range sizes of species. We examined the validity of these predictions in a lowland tropical and a temperate tadpole assemblage.Location Lowland Neotropics (Bahia, Brazil) and Palaearctic (Iberian Peninsula and North Africa).Method We employed phylogenetic eigenvector regression (PVR) and Pagel's lambda to analyse phylogenetic signals in CTmax and CTmin. We used phylogenetic regression analyses (PGLS) to test the relationships between thermal limits, range size and temperature predictors (measured at the macroscale and microhabitat levels) and phy-ANOVA to compare both the physiological traits and thermal regimen in both tropical and temperate assemblages.Results We documented moderate-to-strong phylogenetic signal in both heat and cold tolerance. Temperate-zone tadpoles had broader thermal tolerances than tropical ones. Thermal tolerance range was correlated with range sizes and was explained by seasonal thermal range predictors at the global scale. Both macro-and microclimate temperature variables provided the best predictive multivariate models of thermal limits at the global scale. Microclimatic predictors, however, were the main determinants of CTmax and CTmin variation at the local level of tropical and temperate communities respectively.Main conclusions Thermal tolerance range increases with latitude in tadpoles due to the higher increase in cold tolerance in temperate tadpoles. At the global scale, both macro-and microenvironment thermal information were reliable predictors of critical thermal limits and thermal tolerance range, as CVH predicts. However, thermal limits were best predicted by temperatures of the micro-habitat at the regional level, thus suggesting that physiological thermal boundaries may be governed by thermal selection.

The climate variability hypothesis posits that increased environmental thermal variation should select for thermal generalists, while stable environments should favor thermal specialists. This hypothesis has been tested on large spatial scales, such as latitude and elevation, but less so on smaller scales reflective of the experienced microclimate. Here, we estimated thermal tolerance limits of 75 species of amphibian tadpoles from an aseasonal tropical mountain range of the Ecuadorian Andes, distributed along a 3500 m elevational range, to test the climatic variability hypothesis at a large (elevation) and a small (microhabitat) scale. We show how species from less variable thermal habitats, such as lowlands and those restricted to streams, exhibit narrower thermal tolerance breadths than highland and pond‐dwelling species respectively. Interestingly, while broader thermal tolerance breadths at large scales are driven by higher cold tolerance variation (heat‐invariant hypothesis), at local scales they are driven by higher heat tolerance variation. This contrasting pattern may result from divergent selection on both thermal limits to face environmental thermal extremes at different scales. Specifically, within the same elevational window, exposure to extreme maximum temperatures could be avoided through habitat shifts from temporary ponds to permanent ponds or streams, while minimum peak temperatures remained invariable between habitats but steadily decreased with elevation. Therefore an understanding of the effects of habitat conversion is crucial for future research on resilience to climate change.

Among vertebrates, amphibians have particular characteristics (e.g., permeable skin) that make them extremely dependent on abiotic factors such as temperature and humidity (Duellman & Trueb, 1994).Amphibians are the most endangered group of vertebrates in the world (Alroy, 2015;Ceballos et al., 2017;Stuart et al., 2004). Recent studies estimate that approximately 200 amphibian species (2.4% of all global diversity) are already extinct, and direct and indirect factors related to human activities threaten 41% (Alroy, 2015;Hoffmann et al., 2010; IUCN, 2020). Even those classified as Least Concern (LC) by the IUCN (International Union for Conservation of Nature-an

Critical thermal limits (CT max and CT min ) decrease with elevation, with greater change in CT min , and the risk to suffer heat and cold stress increasing at the gradient ends. A central prediction is that populations will adapt to the prevailing climatic conditions. Yet, reliable support for such expectation is scant because of the complexity of integrating phenotypic, molecular divergence and organism exposure. We examined intraspecific variation of CT max and CT min , neutral variation for 11 microsatellite loci, and micro‐ and macro‐temperatures in larvae from 11 populations of the Galician common frog ( Rana parvipalmata ) across an elevational gradient, to assess (1) the existence of local adaptation through a P ST ‐F ST comparison, (2) the acclimation scope in both thermal limits, and (3) the vulnerability to suffer acute heat and cold thermal stress, measured at both macro‐ and microclimatic scales. Our study revealed significant microgeographic variation in CT max and CT min , and unexpected elevation gradients in pond temperatures. However, variation in CT max and CT min could not be attributed to selection because critical thermal limits were not correlated to elevation or temperatures. Differences in breeding phenology among populations resulted in exposure to higher and more variable temperatures at mid and high elevations. Accordingly, mid‐ and high‐elevation populations had higher CT max and CT min plasticities than lowland populations, but not more extreme CT max and CT min . Thus, our results support the prediction that plasticity and phenological shifts may hinder local adaptation, promoting thermal niche conservatism. This may simply be a consequence of a coupled variation of reproductive timing with elevation (the “elevation‐time axis” for temperature variation). Mid and high mountain populations of R. parvipalmata are more vulnerable to heat and cool impacts than lowland populations during the aquatic phase. All of this contradicts some of the existing predictions on adaptive thermal clines and vulnerability to climate change in elevational gradients.