Microclimate drives intraspecific thermal specialization: conservation perspectives in freshwater habitats

Downscaling the study of the endangered relict snail, Melanopsis etrusca, at the level of populations, inhabiting geographically separated thermal springs in central Italy, we identified intraspecific mechanisms of local specialization. We advocate that the fit of physiological and behavioural traits to different microclimatic niches should be included to design effective conservation actions.


Introduction
Temperature and its variability across space and time are leading physical factors shaping the complexity and distribution of life (Angilletta, 2009). Virtually, all species are characterized by a thermal niche (i.e. the temperature range within which fitness is maintained), the breadth of which is directly influenced by abrupt, daily and seasonal temperature fluctuations (Tracy and Christian, 1986;Huey, 1991). Ectothermic organisms, in particular, are strictly subjected to thermal regimes (i.e. the temporal and spatial variation of environmental temperatures), which influences, to a different extent, all biological processes, such as growth efficiency, reproductive output, cost of maintenance or mobility and metabolic performances (Angilletta et al., 2002;Angilletta, 2009;Huey, 1991;Pörtner, 2010). On an evolutionary time scale, natural selection shapes the breadth of thermal niches, promoting the differentiation of thermal specialist vs. generalist species as a function of the higher or lower variation of the thermal regime (Gilchrist, 1995;Verberk et al., 2016). In this regard, the understanding of the thermal requirements of different species becomes pivotal in order to accurately define their microclimatic niche and consequently to make reliable predictions of their vulnerability to environmental changes and to adopt precise conservation measures (Huey, 1991).
Recently, the importance of downscaling climatic data collection to match the environmental variation actually experienced by species has been proposed (Potter et al., 2013;Varner and Dearing, 2014;Pincebourde et al., 2016). We believe that this guidance should be paralleled by downscaling the characterization of physiological processes to the level of individual microclimatic niches. This approach often remains overlooked, despite its potential to both clarify the natural history of distinct populations and drive effective conservation strategies.
Spring habitats, both cold and hot, often represent ecological refuges from past environmental changes, such as those associated to the quaternary climate cycles. Thus, these relict or extreme ecosystems constitute optimal models to test adaptive responses of aquatic organisms to local or harsh environments (Specziár, 2004). At an evolutionary scale, the segregation of populations within spring habitats, imposed for example by the onset of adverse environmental conditions, may determine the diversification of phenotypes and life traits (Brown and Feldmeth, 1971;Mladenka and Minshall, 2001), ultimately causing speciation (Hershler et al., 2007). Examples include endemic fish species of the genus Eremycthis and Cyprinodon from north American deserts (Vinyard, 1996;Lema and Nevitt, 2006) and the springsnails of the genus Pyrgulopsis and Trochidrobia from desert springs in North America and Australia, respectively (Liu and Hershler, 2012;Murphy et al., 2012). The extremely localized distribution range of endemic species dwelling in spring habitats poses severe conservation challenges, both due to the intrinsic vulnerability of the populations to natural stochastic extinction events and to human impacts locally endangering spring ecosystems.
The poor conservation status of M. etrusca is the result of several anthropogenic pressures, which have historically determined its decline, locally causing population extinctions (Cianfanelli et al., 1991).
Populations of M. etrusca are currently known to occur in five thermal streams in southern Tuscany. For geological reasons, the springs that feed these streams differ from each other in terms of their physical features and chemical composition (Bartolini et al., 2017). In particular, water thermal regimes range from temperate, with seasonal fluctuations of water temperature (12-27 • C), to constantly hot (35-38 • C). The streams inhabited by M. etrusca are therefore, due to their sizable differences in thermal regimes, an ideal case study to apply an ecophysiological comparison downscaled to the microclimatic level. In parallel, the understanding of the degree of specialization attained by separate populations in response to local conditions is of extreme importance to develop effective measures for the conservation of this threatened species.
Melanopsis etrsuca belongs to the Caenogatsropoda (Colgan et al., 2007), a group composed by operculate and gillbreathing snails. In general, a more limited tolerance to desiccation is observed in gill breather compared to pulmonate snails, which are able to use atmospheric oxygen for respiration (Poznańska et al., 2015;Koopman et al., 2016). However, caenogastropods are usually have thick shell and an operculum, allowing species dwelling near-shore habitat subjected to water level fluctuation to avoid desiccation through aestivation (Poznańska et al., 2015). Aestivation is the behaviour of withdrawal into the shell, which associates with aerobic hypometabolism (adaptive, as compared to metabolic depression, which implies either an adaptive or a capacity limited response; e.g. Marshall and McQuaid, 2020). Under severe environmental conditions, snails can even switch to anaerobic facultative metabolism while aestivating (e.g. Storey and Storey, 2012;Ferreira et al., 2003 The aim of our study was to downscale the characterization of the thermal physiology of M. etrusca to population level, focusing on two streams with different thermal conditions. Specifically, we compared the physiological thermal sensitivity and the behavioural response to heat between a population living in a temperate stream, characterized by a thermal regime that fluctuates on a seasonal basis, and a population confined to a constantly hot stream. Implications for the conservation of the species are discussed in light of the experimental evidence.

Melanopsis etrusca distribution and habitat
The freshwater gastropod M. etrusca (Brot, 1862) (Gastropoda; Mollusca) is found in five locations in southern Tuscany (Venturina, Aronna, Bruna, Poggetti Vecchi and Roselle; Fig. 1), inhabiting the upper course of streams characterized by thermo-mineral waters (Bencini et al., 1977;Celati et al., 1991). One of these populations, Poggetti Vecchi, is severely threatened by anthropogenic stream desiccation, running the risk of local extinction, as has occurred to six other populations in recent years (Neiber et al., 2020). The thermal regimes of the streams inhabited by M. etrusca are highly heterogeneous (Bartolini et al., 2017). Specifically, streams are characterized by water temperatures ranging between 30 and 35 • C in Venturina, 19-23 • C in Aronna, and a higher water temperature (33 • C, Bartolini pers. observ.) in Poggetti Vecchi, but the annual temperature variation in these locations has not yet been recorded. The stream in Roselle has the highest water temperature among all locations and a constant thermal regime throughout the year (35-38 • C) (Bartolini and Giomi, in prep). The stream in Bruna is an effluent of Accesa Lake, which is fed by a system of cooler springs from the lake bed. This stream is characterized by broad seasonal temperature variation that spans from 12 • C in winter (February) to 27 • C in summer (August; Bartolini and Giomi in prep).
To investigate the degree of specialization of M. etrusca in these heterogeneous freshwater habitats, we measured the physiological (oxygen consumption rate and mobility performance) and the behavioural responses to a range of experimental thermal conditions, focusing on snails from the two most diverse streams: the constantly hot stream (Roselle) and the seasonally variable stream (Bruna).

Experimental protocol
Animal sampling and the subsequent experiments on the Bruna population were carried out on two cohorts of snails (shell length, 2.5-16.6 mm) collected twice during the year, to encompass the seasonal variation of the thermal niche from 20 • C in spring (2010) to 27 • C in summer (2011). Due to the steady nature and narrow breadth of the thermal niche experienced by the population in Roselle, sampling was only conducted once at this site in spring when water temperatures of 36-38 • C were recorded (2010).
We transported live snails from the stream to the laboratory in polystyrene boxes, in order to limit thermal stress; the maximum thermal excursion during transportation was less than 2 • C over 1 hour. Subsequently, snails were housed at the same temperature measured in the stream at the moment of their collection. Water temperature inside 15 l aquaria was adjusted by means of heating cords connected to a precision thermostat and the systems were thermally isolated by polystyrene panels mounted on the lateral and bottom walls of the aquaria. This setup allowed fine-scale water temperature regulation and compensation with a precision of ±0.2 • C. Snails were provided with continuously aerated water and natural stones, collected in their respective streams to reflect the chemical features of the individual sites and minimize the effect of sampling. All animals were acclimated for 2 days to laboratory conditions prior to experimentation, which also allowed for their recovery from translocation and handling stress.
The experimental protocol for all analytical set-ups involved stepwise procedures, either decreasing or increasing temperature, at an average rate of 2 • C * h −1 . Specifically, starting from each particular control temperature, snails were cooled or heated to the lowest or highest value setting the assay temperatures along the temperature ramp every 3-4 • C for the physiological trials and every 1 • C for the behavioural observations. A similar rate of temperature variation for the two experiments has low ecological relevance for the species, since it inhabits environments particularly stable on a short time scale. M. etrusca experiences sporadic variation in its thermal environment only when individuals emerge on the stream banks or crawl in very shallow waters. However, we adopted this stepwise rate of temperature variation in order to meet the criteria for reliable comparison with similar case studies Leung et al., 2019). We also chose to test the spring cohort of Bruna between 8 and 32 • C, covering a very large thermal window, in order to explore the physiological response at low temperatures. The summer cohort of Bruna was tested considering the upper part of the thermal tolerance window, between 28 and 36 • C. Finally, the population from Roselle was tested between 25 and 40 • C.
Due to the endangered conservation status of this species, shell length was assumed as a proxy of body mass (e.g. Chaparro et al., 2019;McClain et al., 2006)

Oxygen consumption rate and mobility performance
Respirometric measurements were carried out using the electrode micro-respiration system MRS-8 (Unisense, Aarhus, Denmark), provided with eight, individually-stirred, microrespiration glass chambers (approximate volume 4 ml). Respirometric chambers were filled with filtered (0.2 μm) and sterilized stream water to reduce the confounding effect of respiration by micro-organisms.
When the assay temperature was reached, the glass chambers, each housing an individual snail (shell length, 3.5-14.1 mm), were sealed and oxygen-concentration was measured. A second measurement was performed after 15 min and the oxygen-consumption rate was extrapolated from the difference between the two measurements. Values of oxygen consumption rate were standardized for snail shell length. For each assay temperature, different groups of 24 specimens were tested.
Mobility performance was tested by snail righting trials performed at each assay temperature. Snails (shell length, 2.6-15.8 mm) were placed individually in small plastic containers with the shell opening facing upwards. After a short period of immobility (less than 30 seconds), probably due to the disturbance caused by handling, this position stimulated an instinctive reaction of the snails to return to the natural crawling position. The time spent for snails to right themselves was recorded and considered as a proxy of snail mobility performance. For each experimental temperature, we tested separate groups of 30 specimens.

Behavioural response
Since many individuals of the Roselle population were observed aestivating a few centimetres above the water level on the stream banks, a behavioural experiment was designed to test if water temperature is the driving factor for this migration. Snails from Roselle, and snails from Bruna for comparison, were acclimated at 20 • C for 5 days before the start of the experiment. This unnatural acclimation protocol, which is de facto a 'common garden' approach, allowed us to compare the response of the two groups across the same range of temperatures. Thirty specimens (shell length, 2.8-16.6 mm) of each population were placed in a circular plastic container (diameter, 15 cm; depth, 10 cm) and the water temperature was increased at a rate of 2 • C * h −1 until the maximum temperature was reached. The snails were continuously monitored and at each temperature the number and the size of specimens that had migrated outside the water up the wall of the container and those that were inactive in water were recorded. Once a snail emerged from the water or withdrew inside its shell, it was removed from the experimental container. Each trial was replicated three times and a total of 224 snails were tested.

Statistical analysis
The null hypothesis of no differences among respiration rates and righting times for different populations and temperatures was tested using a univariate permutational analysis of variance (PERMANOVA; Anderson, 2001).
To test the null hypothesis that the frequency of inactive (Bruna) or migrating (Roselle) specimens were not different across experimental temperatures, we performed PERMANOVA, taking into account that the measures were repeated on the same experimental specimens. To do this, the experimental design included the following factors: 'temperature' (number of levels were equal to the number of experimental temperatures in each trial) and 'replicate' (three levels; random and nested in 'temperature').
Similarity matrices were computed by using Euclidean distance on square-root-transformed data or arcsin of square root in cases of data expressed as percentages. All analyses were based on 999 restricted permutations under a reduced model of raw data and Type III sums of squares (Anderson et al., 2008). Post hoc pair-wise t-tests were used for multiple comparisons of means when PERMANOVA showed significant differences.
Linear regression was used to relate average shell length (mm) of snails becoming inactive in water (Bruna) or migrating outside the water (Roselle) with increasing experimental temperature. Statistical analyses were performed using PAST (Hammer et al., 2001) procedures and fitting of linear regressions were implemented with SigmaPlot v.11.

Results
Oxygen consumption rates were consistently positively correlated with temperature ( Fig. 2, Table 1). The rates of spring and summer cohorts of Bruna were recorded up to 32 and 36 • C, respectively. These temperatures represent the thermal tolerance limits for activity of the populations naturally acclimated at 20 (spring) and 27 • C (summer); beyond those limits, snails became inactive and withdrew inside their shells. Interestingly, no differences in oxygen consumption rates existed at 28 and 32 • C between different seasons (Table 1). The animals from Roselle showed an increasing oxygen consumption rate until 37 • C. At 40 • C, respiration rates did not increase further ( Table 1).
The snails from Bruna showed an optimal mobility performance at 20 • C in spring, coincident with the acclimatization temperature, while performance tended to decline at both higher and lower temperatures (Fig. 2). The slowest righting times recorded were at 8 and 32 • C, not differing from each other (Table 1). Mobility performance was enhanced in summer (acclimatization 27 • C), particularly at high temperatures, with snails showing the fastest righting times between 28 and 32 • C, although differences among experimental temperatures were not significant (Fig. 2, Table 1). At 36 • C, the average righting time tended to increase (Fig. 2, Table 1). The comparison among righting times of two cohorts from Bruna at 16, 20, 24, 28 and 32 • C revealed that the snails acclimatized to the summer temperature had a consistently faster mobility performance than those in spring (PERMANOVA, df = 4, F = 7.43, P < 0.01). The snails from Roselle performed optimally between 31 and 37 • C, with a significant increase  in righting time at 20 and 40 • C (Fig. 2, Table 1). Between the two populations, the comparison of performances at 20, 24-25 and 28 • C showed that animals from Roselle and from the summer cohort of Bruna performed better than those acclimatized to spring temperatures (Table 1).
The heat-induced migration outside the water was virtually absent in both cohorts from Bruna (Fig. 3, Table 1). In fact, snails tended to withdraw inside their shells remaining inactive in the water when the heat limit was reached. The onset of this behavioural response occurred at 36 • C for the spring cohort, when 61.15 ± 8.57% of individuals were induced to aestivate in the water. This heat limit appeared to shift at 38 • C for the summer cohort, with 58.70 ± 11.92% of individuals recorded as inactive in water. At 39 • C 2.38 ± 2.38% of snails migrated out of the water whereas 4.55 ± 3.20% were still active in the water. At 40 • C the percentage of active snails further decreased to 1.19 ± 1.19%. At 41 • C no active snails were observed in the water. The snails from Roselle showed a markedly different pattern of thermally induced migration outside of the water and the onset of this behaviour occurred at 38 • C (Fig. 3, Table 1). At 40 • C, 70.98 ± 7.93% of snails migrated outside the water. At 41 • C, the last snails remaining in the water (26.67 ± 9.09%) did not migrate but instead withdrew inside their shells (Fig. 3, Table 1).
The susceptibility to increasing temperature was demonstrated to be size dependent in both populations, as shown by the negative relationship between shell length and temperature ( Fig. 4; Bruna spring: F = 38.89, n = 64, P < 0.001; Bruna summer: F = 23.30, n = 97, P < 0.001; Roselle: F = 31.97, n = 63, P < 0.001). Specifically, large animals (shell length > 10 mm) from Bruna became inactive and those from Roselle emerged from the water at lower temperatures compared to snails of medium/small size (shell length < 10 mm; Fig. 4).

Thermal sensitivity of performance
The capacity to sustain aerobic metabolism during acute warming represents a pivotal mechanism regulating the breadth of their thermal niche and the limits of heatdependent performances in many aquatic ectotherms (Pörtner, 2010; Clark et al., 2013;MacMillan, 2019). The exponential rise of oxygen consumption ideally describes the thermal response to abrupt variations in the vast majority of water breathing organisms (Gillooly et al., 2001;Brown et al., 2004;Giomi and Pörtner, 2013). Several aquatic ectotherms shift their aerobic performance to match their habitat temperature range (e.g. Sokolova and Pörtner, 2003;Wittmann et al., 2008; and recently conceptualized in Sinclair et al., 2016). These shifts occur over short time scales when ectotherms acclimatize to seasonal temperature fluctuations and across multiple generations when species evolve into new thermal niches.
Surprisingly, M. etrusca showed non-acclimation of temperature-related metabolism, since the temperaturedependent oxygen consumption rate was consistent, regardless of seasonality and habitat features. Specifically, the oxygen consumption rate of active individuals is not fitted to the thermal regimes of the specific locations nor to compensate for the seasonal shift of the thermal niche. The metabolic rates of the three cohorts differed exclusively for the maximum temperature at which they were sustained.  The spring and the summer cohorts of the Bruna population showed a sizable increment of the thermal limit for oxygen consumption from 32 to 36 • C, indicating that certain acclimatization may partially compensate for the seasonal expansion of the thermal niche at higher temperatures. Beyond these thermal limits, it was not possible to take additional measurements since the animals withdrew inside their shells. This response to thermal extremes is similar to the behaviour that the Bruna population adopts in winter, when the thermal regime shrinks to low temperatures and M. etrusca become inactive on the stream bed (Bartolini and Giomi, pers. obs.). On the contrary, the snails from Roselle did not withdraw inside their shells but they were able to sustain the increase of metabolic rate up to 38 • C, while it progressively declined when animals were subjected to further warming. This physiological response to warming, distinctive of the population in Roselle, is similar to the trends shown by the majority of aquatic ectotherms and suggests a progressive switch to anaerobic metabolism when the capacity to extract oxygen from the water becomes compromised (Pörtner, 2010;Giomi and Pörtner, 2013). The shift to anaerobic metabolism has been demonstrated in intertidal gastropods, which showed to be able to survive at temperature >4 • C beyond the limit of Arrhenius break temperature (Han et al., 2017). Interestingly, in contrast to the population of Bruna, the Roselle population rarely becomes inactive in the stream but instead regularly crawls outside of the hot water onto the banks (Bartolini and Giomi, pers. obs.).
The comparison of the righting time, as a proxy for mobility, between the spring and summer cohort of the Bruna population reveals a seasonal shift and widening of the thermal performance breadth. In fact, the summer cohort showed greatest activity at higher temperatures, maximizing their mobility performance more efficiently than the spring cohort. Interestingly, this trend also persists at low temperatures, representing the acclimatization condition of the spring cohort (around 20 • C), which showed consistently slow righting times. This observation is in agreement with the seasonal pattern of activity exhibited by the Bruna M. etrusca population. During winter months, snails undergo a suspension of somatic growth that resumes in spring (Bartolini et al., 2017).
Even though the Roselle population showed a thermal window of activity larger than the summer cohort of Bruna, with an upper limit over 37 • C, the righting time of the two groups overlapped for most of the temperatures. This result reveals the capability of the population that inhabits the seasonally variable environment to match the performance attained by the thermal specialist population. In this respect, the peak of the Roselle snails' mobility performance matches the constant high temperatures of this stream, which is fed by a hot thermal spring (annual range, 35-38 • C). Interestingly, the thermal optimum for mobility, represented by the fastest righting time, differs slightly from the critical temperature, suggesting that this population optimized its performance in the narrowest range close to the upper tolerance limit.
These findings provide an interesting example of the phenotypic plasticity of gastropods to heat tolerance and underline the importance to downscale physiological investigations to the most relevant local level. In fact, although metabolic rates of typical ectotherms increase with temperature, some species are able to restrain heat sensitivity through metabolic depression (Houlihan and Innes, 1982;Sokolova and Pörtner, 2001;Verberk et al., 2016). For example, intertidal gastropods have the capacity to adopt a temperature-insensitive metabolism, even below their limits of heat tolerance (Garrity, 1984;. Snails can rest inside their shell, lowering their metabolic rate and minimizing the energy loss that otherwise would be imposed by cyclical stressful conditions, such as daily peak temperatures. In the present study, we found evidence of both physiological mechanisms in geographically separated populations of the same species. Specifically, the snails in Roselle maximize their metabolism and performances close to the upper end of thermal tolerance, displaying a progressive decline when the heat limit is exceeded. Conversely, the snails in Bruna, active over a wider thermal window, respond to adverse temperature extremes, both at the low and high end of the niche, by retreating within their shells showing a good capacity to down-regulate cellular metabolic demand.
Accordingly, we hypothesize that population-specific behaviours have promoted the divergence of the thermal physiology of M. etrusca at an extremely local scale.

Thermoregulatory behaviour
For the population of Roselle, we demonstrated that 38 • C represents the temperature threshold inducing the onset of migration outside the water and aestivation phase, when snails are observed withdrawn inside the shell. This heat limit corresponds to the maximum temperature recorded in this stream (Bartolini et al., 2017). During several periods of field observation at Roselle, we documented solely the occurrence of animals active in hot water or laying inactive on the banks a few centimetres above the water level. To explain these observations, we hypothesized that periodical emersion behaviour would be the strategy adopted to minimize the costs associated with settlement in this extremely hot site (Fig. 5). The microhabitat occupied by the snails along the stream banks is indeed characterized by air temperatures that are generally lower than aquatic temperatures (Bartolini and Giomi, in prep). Conversely, within the stream, the marked temperature dependence of metabolic rate implies a high cost of maintenance diverting energy expenditure away somatic growth and reproductive investment (Brown et al., 2004;Nisbet et al., 2012). Thus, M. etrusca in Roselle commutes between two dissociated thermal niches to recurrently aestivate in the cooler environment. While resting on the river banks, the snails of Roselle are able to optimize their energy budget by means of a good capacity to down-regulate cellular metabolic demand. This behaviour likely maximizes  In the Bruna stream, temperature ranges between 12 and 27 • C according with a seasonal fluctuation: in winter, when the water temperature (Tenv) decreases below the optimal temperature for activity (Topt), M. etrusca suspends somatic growth and reduces the activity in water, wintering on the stream bed. With the seasonal rise of water temperature snails resume both activity and somatic growth. In the Roselle stream, snails experience constant temperature conditions of 35-38 • C coincident with the optimal temperature for metabolic and activity performances. However, to avoid continuous exposure to such a hot microclimate, snails emerge from the water on the stream banks, aestivating for periods in a milder thermal niche (Tpref ).
the snail individual fitness. Similarly, the active selection between habitats characterized by disjointed thermal regimes has been documented in several ectotherms. A separation between thermal niches, where the optimal temperature for performance differs from the preferred temperature that minimizes energy expenditure, has been observed in situations of limited resources, strenuous habitat conditions or under growth and reproductive constraints (Huey, 1991;Angilletta 2009). The sockeye salmon (Oncorhynchus nerka) performs daily migrations to deep cold water under conditions of food limitation to reduce metabolic costs, even though the optimal temperature for foraging is in the shallow warmer habitat (Bevelhimer and Adams, 1993). The behavioural strategy of the population in Roselle would allow for highly efficient foraging in hot water where the mobility performance is optimal and the maximization of growth when aestivating in a low energy demanding environment (Fig. 5).
The shell of gastropods allows several species to colonize unexploited habitats, constituting a physical barrier to harsh temperatures and adverse conditions and providing a retreat to endure physical stress and avoid predation. Aestivation is a well-known strategy showed by freshwater and terrestrial snails, which are able to avoid adverse environmental conditions, such as periodic aridity or deterioration of physicochemical water conditions. Aestivation is often accompanied by some degree of 'escape' from unsuitable environment, including horizontal or vertical migration. Horizontal migrations allow snails to actively disperse in search for more favorable areas (Poznańska et al., 2015;Jokinen, 1978). Vertical migrations include burial into wet sediment (Ferreira et al., 2003) or crawling outside water (Proum et al., 2017).
Intertidal marine snails are able to thermoregulate or to suspend their activity during low tide and become inactive during periods of emersion (Garrity, 1984;Munoz et al., 2005;Verberk et al., 2016). Several freshwater snails adopt aestivation as a strategy to survive desiccation caused by periodic drought, showing astonishingly capacities to survive in the absence of water. Pulmonate snails are able to withstand several months of environmental drought (Jokinen, 1978), with extreme examples in tropical aestivating species (e.g. Richards, 1967). Gill-breathers can tolerate shorter periods of desiccation ranging from few days to several weeks (e.g. Poznańska et al., 2015). Prolonged aestivation frequently involves transition into a hypometabolic state and anaerobic metabolism (Storey and Storey, 2012;Ferreira et al., 2003). On a short time scale, M. etrusca appears to periodically aestivate, purportedly to avoid the prolonged exposure to high water temperatures. We suggest that during this phase, M. etrusca undergoes to short-term metabolic adjustment (likely hypometabolism as described by Marshall and McQuaid, 2020). As a matter of fact, individuals are observed active in the water across every season (Bartolini et al., 2017). In other words, Roselle population shows an adaptive behavioural emergence from water, as opposed to emergence from the water in response to respiratory stress, as is observed in many aquatic animals when water PO 2 declines.
We observed that most snails from Bruna became inactive in water and withdrew inside their shells when exposed to extreme temperatures; however, a small number of individuals migrated outside the water (overall 6 specimens out of 164 tested). This suggests that this behavioural trait remains inherent of the species and has promoted the intra-specific specialization to extreme thermal conditions by means of highly selective exaptation (Gould and Vrba, 1982). The fact that our experimental evidences come from a common-garden approach (snails acclimated at 20 • C for 5 days), corroborates the hypothesis that an adaptive behaviour to specific microclimate conditions has been selected in the Roselle population. Through downscaling our study to the population level, we recognized that the divergence of behavioural features clearly distinguishes the animals of Bruna and Roselle providing M. etrusca with distinct mechanisms to colonize an array of sites characterized by highly divergent thermal regimes. It is worth mentioning that body size is revealed to trigger these behavioural traits, being inversely related to the thermal tolerance. Indeed, both the emersion and the withdrawn behaviours were elicited at lower temperature in larger snails. Interestingly, another population of M. etrusca that lives in a constantly warm stream (30-35 • C; Venturina) does not exhibit clear emergence behaviour. In this site, snails attain high growth rates but have a short life span resulting in a population dominated by small specimens where a clear structure in size classes is absent (Bartolini et al., 2017). Conversely, populations from Bruna and Roselle appear well structured due to different strategies (quiescence and migratory behaviour, respectively), with a high frequency of large specimens (Bartolini et al., 2017). In other words, if a successful behavioural specialization is not fixed in a certain population when restrained to a hot spring, this results in limited demographic structure.

Conservation perspectives
In recent years, the knowledge of species physiology (Wikelski and Cooke, 2006;Cooke et al., 2013, Cooke et al., 2014 and behaviour (Anthony and Blumstein, 2000;Berger-Tal et al., 2011) has been integrated into conservation science to help discover practical solutions to counteract the loss of biodiversity. Within this framework, we downscaled the characterization of physiological and behavioural traits to populations specialized to distinct microclimatic niches. The study of M. etrusca thermal physiology highlights the need to focus on local microclimatic niches, especially when intraspecific specialization occur at a fine geographic scale. Similarly, conservation actions should take into account the occurrence of specific traits that enable populations to cope with local variability, particularly in extremely selective environments. We believe that our findings have specific practical implications for the conservation of M. etrusca and generally support the need to downscale conservation efforts to the most relevant evolutionary and ecological level.
In the future, introduction/reintroduction programmes could represent a strategy for restocking M. etrusca both in streams where local populations become extinct, once the causes of extinction have been mitigated or removed, and in new streams (IUCN, 2013). However, populations from temperate streams with seasonal variation of the thermal regime may lack of physiological and behavioural traits enabling the colonization of hot streams. Therefore, beside genetic consideration concerning the most appropriate population suited for a potential introduction (Neiber et al., 2020), it would be necessary to select the most appropriate stock according to rigorous physiological and behavioural criteria.
Melanopsis etrusca populations appear to be currently isolated by geographical barriers. Recent genetic evidence is consistent with a pattern of isolation by distance and shows a differentiation into a western, a central (including Bruna) and an eastern (including Roselle) group of populations (Neiber et al., 2020). In this scenario, an assessment of possible detrimental genetic processes (e.g. genetic loads) affecting isolated populations or a group of populations would be required (Martin et al., 2012). This information, combined with the evidence of adaptive innovations, would pinpoint the necessity of specifically designed conservation actions. In fact, our study suggests that the Roselle and Bruna populations are not ecologically exchangeable (Crandall et al., 2000) and thus could be considered partial Evolutionary Significant Units (de Guia and Saitoh, 2007).