Elemental stoichiometry and insect chill tolerance: evolved and plastic changes in organismal Na⁺ and K⁺ content in Drosophila

Understanding the pace, magnitude, and mechanisms of organismal responses to changing climates is a major goal in biology with profound implications for the conservation of biodiversity (Urban, 2015). Organismal traits and their underlying physiology are critical determinants of both the distribution of biodiversity and population growth rates. Traditionally, phenotypic responses to rapidly changing environments have focused on measuring phenotypic plasticity, as these plastic shifts occur within a generation (Pigliucci, 2001). Many organisms exhibit profound phenotypic plasticity, including numerous cases of acclimation in response to abiotic change (Ghalambor et al., 2007; Nylin and Gotthard, 1998). Similarly, there are many examples of spatial intraspecific genetic variation indicative of adaptation to abiotic factors that vary over space (Des Roches et al., 2020; Hoffmann et al., 2002; Rajpurohit et al., 2018). In the past few decades, there has been a growing recognition that populations can also exhibit evolutionary phenotypic change in response to environmental shifts over short timescales (Hairston et al., 2005; Hendry and Kinnison, 1999; Reznick et al., 1990). This includes evidence that selection can be sufficiently strong to drive adaptive phenotypic tracking over contemporary timescales (Barrett et al., 2008; Behrman et al., 2015; Rennison et al., 2019). Theory on the relationship between plastic and evolved responses to environmental stress creates several predictions (Chevin et al., 2010), including that trait changes should occur by similar underlying mechanisms (Pigliucci et al., 2006; Waddington, 1957). Yet, the relative importance and the similarity of the underlying physiological mechanisms of trait change stemming from acclimation and adaptation are still largely unknown.

Documenting adaptation in response to changes in climate requires demonstration that phenotypic shifts are genetically based (often determined using common garden experiments) and that the phenotypic change leads to an increase in fitness (Merilä and Hendry, 2014). Links between phenotypic change and fitness change are difficult to empirically measure, but cases of parallel phenotypic evolution across replicate populations or environmental gradients present strong evidence for adaptation (Losos, 1992; Schluter and Nagel, 1995). Most of what is known about adaptation comes from examining phenotypic differences among divergent populations or species using phylogenetic comparative methods or comparing spatially separated populations (Futuyma and Kirkpatrick, 2005; Kellermann et al., 2012). These approaches are powerful and essential when adaptation occurs too slowly to observe directly, but they don't allow for precise estimates of evolutionary rates that are needed to assess responses to rapidly changing environments (Gingerich, 1993). If adaptation occurs rapidly, it is possible to use longitudinal studies and manipulative experiments to directly measure the extent and pace of adaptation in response to environmental change, termed adaptive tracking (Botero et al., 2015; Grant and Grant, 1995; Rudman et al., 2022). Combined with assessments of adaptation arising across populations, which results from dozens to thousands of generations of evolutionary divergence, direct observation of adaptive tracking provides the temporal resolution needed to compare the impact of evolution and phenotypic plasticity on rapid trait change.

With climates changing rapidly due to anthropogenic disturbance, it is particularly important to understand the phenotypic responses, both plastic and evolved, to physiologically challenging temperatures. In insects, a primary physiological challenge linked to distribution is chill susceptibility (Andersen et al., 2015, 2018, 2023). Chill-susceptible insects are those that can withstand cool temperatures above their freezing point for an extended period but accumulate chilling injuries while doing so (Andersen et al., 2015; Bale, 1996; Overgaard and MacMillan, 2017). In a variety of distantly related insects (including flies, locusts, crickets, and firebugs), a primary mechanism of these injuries is the disruption of organismal ion homeostasis leading to cell death. Exposure to cold temperatures is associated with a loss of Na⁺ and water balance that contributes to an increased concentration of K⁺ in the hemolymph (hyperkalemia) (Bayley et al., 2018; MacMillan et al., 2015c; Olsson et al., 2016). The combination of cold and hyperkalemia leads to cell membrane depolarization, which triggers calcium influx into cells, activating signaling pathways that initiate cell death (Bayley et al., 2018; Carrington et al., 2020; MacMillan et al., 2015b). The increasingly well-resolved physiological basis of chilling injury (see, for example, Overgaard et al., 2021) makes it possible to test both the extent of plastic and evolved responses to cold temperatures and, further, to test specific hypotheses about whether trait changes have similar physiological underpinnings.

Drosophila melanogaster are amenable for both field and laboratory studies and they inhabit a broad range of thermal environments. In the laboratory, both plastic and evolved increases in basal chill tolerance have been linked to an improved ability to maintain ion homeostasis in the cold. For example, Drosophila species that are more chill tolerant and flies that are cold acclimated both avoid hyperkalemia, in part, through improved renal function in the cold (Andersen et al., 2017; MacMillan et al., 2015c). In both cases, these changes to thermal tolerance have been linked to decreased reliance on sodium as an extracellular osmolyte (MacMillan et al., 2015a; Olsson et al., 2016), suggesting that acclimation and evolved differences in chill tolerance are linked to both sodium and potassium regulation. Patterns of evolution in D. melanogaster thermal phenotypes have also been well studied; evolution of D. melanogaster populations from high- and low-latitude locations of the east coast of North America demonstrate that populations inhabiting different thermal environments exhibit genetic differences that are likely adaptive in relation to thermal tolerance traits (Mathur and Schmidt, 2017). Repeated sampling and common garden rearing of D. melanogaster from an orchard in Pennsylvania demonstrated putative adaptive tracking of chill coma recovery from spring to autumn, with spring flies showing faster rates of recovery (Behrman et al., 2015).

Our current understanding of chill coma recovery in D. melanogaster makes it possible to combine our laboratory, field experimental and field observational approaches to compare the relative phenotypic effects of acclimation and adaptation, and the underlying physiological mechanisms, across different timescales. To do so, we first assessed the role of acclimation on the phenotypic response to cold exposure using an isofemale line originally derived from wild-collected flies in London (Ontario, Canada) that is well studied in relation to the mechanisms of cold acclimation (Marshall and Sinclair, 2010). To compare the magnitude of this variation with that stemming from hundreds of generations of clinal adaptation (Fabian et al., 2012; Schmidt et al., 2005), we collected flies from locations on a well-studied latitudinal gradient in eastern North America (Florida, Pennsylvania, and Maine). Finally, we used a replicated field experiment in which an outbred founder population was introduced into five independent field enclosures to assess adaptive tracking from mid-summer to mid-autumn. For each contrast we assessed the fitness-associated phenotype of chill coma recovery time (CCRT) as well as the Na⁺ and K⁺ concentrations that can underlie loss of homeostasis in response to chilling in both males and females. We predicted that the whole-body Na and K content will relate to the cold tolerance (i.e. CCRT) of D. melanogaster in a consistent manner, and that both sexes would consistently differ in ion content because of sexual dimorphism. Moreover, we predicted that: (1) acclimation would produce a strong increase in cold tolerance; (2) evolved populations, both from different climates (‘spatial adaptation’) and different seasons (‘temporal adaptation’), would show an increase in tolerance; and (3) similar mechanisms of Na and K content association with improved chill coma recovery would be visible across all three contrasts, indicating that the mechanistic basis of shifts in chill coma recovery are constrained.

We quantified cold tolerance in all lines and both acclimation groups using CCRT. In the acclimated flies, there was a significant main effect of cold acclimation on CCRT (P<0.0001), as well as a main effect of sex (P=0.008; Table S1; Fig. 1A).

In the spatial adaptation flies, there was no significant main effect of origin location on CCRT (P=0.200). There was, however, a significant interaction in the effects of location and sex on CCRT (P=0.045; Table S1; Fig. 1B). Female flies from the more poleward climate (Maine; ME) recovered ∼5 min faster from chill coma than Florida (FL) females reared under the same conditions. By contrast, in male spatial adaptation fly lines, the Pennsylvania (PA) lines had the fastest CCRT while the FL males recovered slowest (a difference of ∼12 min).

In the temporal adaptation flies (summer vs autumn), there was no significant effect of adaptive tracking across season (P=0.739), but there was a significant effect of sex on CCRT (P=0.001; Table S1; Fig. 1C). Across all three groups (acclimation, spatial, and temporal adaptation), we noted a clear pattern of females recovering from chill coma 5-15 min faster than males within the same line/acclimation group (Fig. 1). Cold acclimation reduced CCRT by approximately 19 min in males and 16 min in females, climate adaptation reduced CCRT by approximately 9.5 min in males and 5 min in females (FL vs ME), and seasonal adaptation altered CCRT by less than 1 min in both males and females.

We quantified Na and K content in D. melanogaster (expressed per mg of dry mass) using flame photometry on individual flies that were weighed before and after drying to determine water content. Neither thermal acclimation (P=0.922) nor sex (P=0.517) influenced Na content in acclimated flies (Table S2; Fig. 2A). By contrast, K content was significantly higher in cold-acclimated flies (P<0.0001). There was no main effect of sex (P=0.482), nor was there an interactive effect of acclimation temperature and sex on K content (P=0.468; Table S2; Fig. 2D). K content increased by ∼0.1 µmol mg⁻¹ (∼19% increase) in both the male and female flies that acclimated at 15°C compared to the male and female flies that acclimated at 25°C. There were no significant effects of acclimation temperature (P=0.148) or sex (P=0.122) on water content in the acclimated flies (Table S2; Fig. 2G).

In the spatially adapted flies, there was a significant interactive effect of sex and origin on Na content (P<0.0001; Table S2; Fig. 2B). In contrast to the limited effect of acclimation on Na content, flies from PA had a higher sodium content than flies from ME or FL. There was also a significant main effect of sex (P=0.011); however, unlike in the cold acclimated flies, spatially adapted females tended to have a lower Na content across all climates. In these flies, K content displayed a similar trend to Na, where flies from PA contained more K and those from FL contained less (by ∼0.05 µmol mg⁻¹). There was a significant interactive effect of origin location and sex on K content (P<0.0001; Table S2; Fig. 2E), but neither location (P=0.348) nor sex (P=0.156) had a main effect on K content. The water content of flies collected from different locations was significantly affected by an interaction between location and sex (P<0.0001; Table S2; Fig. 2H); males from PA and FL carried more water than females, but males from ME carried less water than females from the same location. Lines collected in FL tended to carry less water than those from the more poleward climates, and both male and female flies from PA carried the most water.

In temporal adaptation fly lines, we found a significant effect of sex (P<0.0001), but not collection season (P=0.980) on Na content (Table S2; Fig. 2C). Female flies had higher Na content across both seasons when compared to the male flies. Season and sex interacted to influence K content (P=0.010); females had greater K content in the summer, but this difference was absent in flies collected in the autumn (Fig. 2F). There were no main effects of season (P=0.780) or sex (P=0.123) on K content (Table S2; Fig. 2F). The water content of temporal adaptation flies differed only according to sex (P<0.0001; Table S2; Fig. 2I), with males holding more water than females regardless of the season during which the lines were sampled.

Changes in ion and water content can both influence physiologically relevant ion concentrations, and Na and K are almost exclusively retained in animals in their free (unbound) ionic forms (Na⁺, K⁺). In Fig. 3, we examined the relationships between cold tolerance (CCRT) and whole-body average [K⁺] ([Na⁺] in Fig. S2). Improvements in cold tolerance with cold acclimation were associated with an increase in [K⁺]. However, the opposite was true with spatial adaptation (Fig. 3) – where higher whole-body average [K⁺] was associated with a longer CCRT.

The relationship between [Na⁺] and cold tolerance displayed a qualitatively similar pattern to [K⁺] (Fig. S2). The acclimated flies overall tended to have a negative relationship between the [Na⁺] and cold tolerance, but there was only a ∼5 mM difference in concentration between the warm- and cold-acclimated flies (Fig. S2). As with the potassium/cold tolerance relationship, the spatial adaptation flies displayed a positive trend between the sodium concentration and CCRT (Fig. S2).

The temporal adaptation flies had no distinct relationship between the potassium or sodium concentration and CCRT (but also very little difference in CCRT).

Both Na⁺ and K⁺ are critically important to water balance as well as cold tolerance. We noted that relationships between these ions and water content differed among experimental groups, and this was most striking in regard to acclimation and in female flies (Fig. 4; Table S3). Warm-acclimated females preferentially retained K⁺ along with water; there was a positive relationship between water content and K⁺ content but not Na⁺ content (Fig. 4A,C). However, in the cold-acclimated flies, the opposite relationship is clear; Na⁺ content and water content had no distinct relationship, whereas K⁺ content increased along with water content (Fig. 4A,C). This discrepancy is clear from significant interactions in the effects of acclimation temperature and water content on ion content (P<0.006) in all cases except for Na in males where it was near-significant (P=0.062) and following the same trend. To remove any possible influence of body size variation, we calculated the ratio of Na:K and found that specifically in females, cold acclimated flies retain more Na relative to K as whole-body water content increases, while warm acclimated flies do the opposite (P<0.001; Fig. 4E). While the trend was similar, there was no significant effect of water content on the Na:K ratio in male acclimated flies (P=0.107).

Notably, in both the temporal and spatial adaptation fly lines, we found no such tendency for one ion to be retained over the other in flies that had higher water content; both ions increased in abundance in flies that carried more water and there was no significant effect of origin location or season in either sex (Figs S3 and S4, Table S4). In both temporal and spatial adaptation lines, K content was more tightly associated with water content than Na (Table S4).

Here, we compared the magnitude of phenotypic change and underlying physiological differences associated with long-term thermal acclimation, spatial adaptation, and temporal adaptation using an acclimation experiment, a field study, and a replicated field experiment. We used a common chill tolerance assay (CCRT; Fig. 1) to determine the overall phenotypic effects and measured whole-body Na⁺ and K⁺, two ions that play key roles in the cold tolerance of Drosophila (Fig. 2) (MacMillan et al., 2015a; Overgaard and MacMillan, 2017). Long-term acclimation at 15°C decreased CCRT, and this effect was greater than that observed from spatial adaptation (Fig. 1). We saw little evidence for variation in CCRT arising from temporal adaptation (Fig. 1). Variation in chill tolerance is well studied in Drosophila in relation to both acclimation (Findsen et al., 2013; Koštál et al., 2011; Overgaard et al., 2008; Sinclair and Roberts, 2005) and climatic and seasonal variation (Gibert et al., 2001; Hoffmann et al., 2002; Kellermann et al., 2012; MacLean et al., 2019; MacMillan et al., 2016; Overgaard et al., 2011), and the variation observed is a prerequisite for asking mechanistic questions about how ion balance and chill tolerance co-vary.

We sought to determine whether physiological mechanisms of variation in cold tolerance in response to acclimation and owing to adaptation were similar. In cases of adaptation through genetic assimilation environmentally induced trait change (e.g. acclimation) becomes constitutively expressed (Pigliucci et al., 2006; Waddington, 1957). If adaptation occurs through genetic assimilation, we would predict similar underlying physiological mechanisms of acclimation and adaptation. Given the demonstrated importance of ion balance in Drosophila chill tolerance, this similarity in mechanisms is likely to manifest as differences in Na and K stoichiometry. As is the case with CCRT, our warm- and cold-acclimated d flies also displayed the largest differences in total ion content when compared to populations originating from different latitudes or seasons. This was most evident from whole-body potassium content, which increased in cold-acclimated flies (Fig. 2D), while they maintained the same water content as the warm-acclimated flies (Fig. 2G). This (more K in the same amount of water) drove an overall increase in whole-body average [K⁺] in both male and female cold-acclimated flies. In contrast to predictions from adaptation by genetic assimilation, this pattern is wholly different from how changes in ion balance matched with the chill tolerance of flies adapted to different climates; in this group, whole-body K content was largely stable regardless of origin location (Fig. 2E), while water content varied (Fig. 2H). This pattern resulted in flies from cooler climates having overall lower, rather than higher, whole-body [K⁺] (the same amount of ions in a larger water volume; Fig. 3B). Taken together, acclimation and latitudinal adaptation of basal tolerance appear to be related to whole-animal average [K⁺] in opposing directions: while improved cold tolerance is associated with higher [K⁺] in acclimated flies, it is instead associated with lower [K⁺] in cold climate-adapted flies.

Wide-ranging literature documenting rapid adaptation has shaped thinking about the importance and pace of adaptation in response to environmental change (Sanderson et al., 2022). This includes a seasonal adaptation experiment from which the temporal adaptation flies studied here were taken, which found rapid phenotypic adaptation in response to both seasonal change and the presence of a competitive species (Grainger et al., 2021). A growing number of experiments tracking Drosophila populations through time have found evidence of rapid adaptation (Rajpurohit et al., 2018; Rudman et al., 2019, 2022). Moreover, patterns of spatial intraspecific genetic variation across Drosophila populations are well documented, and the resulting phenotypic variation between populations can be considerable (Fabian et al., 2012; Hoffmann et al., 2002; Schmidt et al., 2005). Here, we find modest evidence for spatial adaptation in cold tolerance phenotypes and no evidence for rapid temporal adaptation in response to an autumn acclimatization. There are several putative explanations for the lack of cold adaptation in seasonally cooling conditions – chief among them that the temperature drop prior to sampling had been modest relative to what is observed later in the autumn (Fig. S2). Indeed, a recent paper in which rapid evolution of chill coma recovery was observed across seasonal time did not find significant evolution over the window sampled for this study (Rudman et al., 2022). Perhaps cooling temperatures was not a strong enough agent of selection over a long enough duration to drive adaptation during the period studied here. More broadly, it is also possible that high-profile cases of rapid adaptation are creating an expectation of ubiquitous adaptation: A recent meta-analysis of experiments found no evidence for consistent life history trait adaptation in response to manipulative climatic warming (Grainger and Levine, 2022). Our data support a view in which acclimation is a singular mode by which phenotypes shift in response to rapid environmental change. Future synthetic work comparing the predictability and magnitude of plastic and evolved changes will be key in identifying the mechanisms by which organisms are most likely to respond to abiotic challenges.

The observation that thermal plasticity and adaptation of basal thermal tolerance are associated with changes in ion stoichiometry may point us toward important and undescribed mechanisms of thermal performance in insects. Ultimately, however, connecting the observed changes in K and water content to current knowledge on the physiological mechanisms of cold tolerance is challenging when the changes are observed at the whole organism level. One of the most pressing questions that emerges from this work, therefore, is what organs in the body are driving the observed variation in stoichiometry. One factor that may be highly relevant here, and explain differences in acclimation effects we observed between sexes, is the possibility or females entering a state of reproductive dormancy or egg development being slowed at 15°C relative to 25°C. This change could drive substantial changes in body composition as mature eggs make up a substantial portion of female body mass. Notably, however, the induction of reproductive diapause typically requires changes in both temperature (∼12°C) and light cycle (<13 h of light) indicative of seasonal change (Saunders and Gibert, 1990), and in this case only temperature was changed to induce acclimation to 15°C.

In Drosophila, Na⁺ is the primary extracellular ion and an important osmolyte, and intracellular [Na⁺] levels are lower in the cell cytoplasm (∼18 mM) than in the hemolymph (∼70 mM; MacMillan et al., 2015a). By contrast, [K⁺] is higher inside cells (∼120 mM) and maintained at low levels (∼10-15 mM) in the hemolymph. When chilled, chill-susceptible insects like D. melanogaster lose the ability to maintain these gradients, and ion balance is progressively lost as Na⁺ leaks across epithelia (osmotically taking water with it) and [K⁺] concurrently rises in the hemolymph (Andersen and Overgaard, 2019; Koštál et al., 2006; MacMillan et al., 2014). If severe enough, high [K⁺] in the hemolymph depolarizes cells, causing catastrophic Ca²⁺ influx and cell death (ultimately causing organismal death; Bayley et al., 2020; Carrington et al., 2020; MacMillan et al., 2015b). Similarly, a rapid loss of ion balance occurs within the central nervous system when insects like D. melanogaster are sufficiently cooled, albeit it via an unknown, different physiological mechanism with unknown consequences for survival and fitness (Robertson et al., 2017, 2020).

Because a loss of systemic ion balance is an important driver of low temperature injury and death (Overgaard and MacMillan, 2017), variation in chilling tolerance within and among species is thought to be closely tied to the magnitude of ion (most notably Na⁺ and K⁺) gradients across membranes and epithelia before a stress is applied, as well as how cells, tissues, and organs responsible for maintaining ion gradients at the local and organismal level respond during chilling (Andersen and Overgaard, 2020; MacMillan et al., 2015c, 2017). Specifically, the status of [Na⁺] gradients before chilling occurs appears relevant to both cold acclimation and cold adaptation among species; cold-acclimated D. melanogaster and cold-adapted Drosophila species maintain lower [Na⁺] gradients without altering hemolymph osmolarity (MacMillan et al., 2015a; Olsson et al., 2016), such that water balance may be better maintained in the cold. For this reason, we expected to see adaptive changes in whole-body Na content in the present study, but our results instead suggest that changes in these gradients may occur without any gain or loss of organismal ion content. Total body Na content may nonetheless intersect in important ways in insects to influence thermal performance and other fitness-related traits in insects; dietary Na⁺ can alter insect growth and may be limiting in some food webs (Peterson et al., 2021; Welti et al., 2019).

Given that a loss of K⁺ balance has been identified as a direct cause of chilling injury in a variety of insects, we did not expect cold-acclimated flies to retain a greater amount of K than their warm-acclimated counterparts (Fig. 3). Whether or how this additional K is contributing to improved chill tolerance by altering K⁺ gradients entirely depends on the location in which it is being accumulated. Increased intracellular [K⁺] without any change in membrane ion permeability or extracellular [K⁺] would cause hyperpolarization of nerve and muscle cells, which may protect against the cold-induced depolarization responsible for chilling injury. Establishing and maintaining this state would likely come at an energetic cost (and many secondary consequences) that might not be favorable to maintain constitutively when conditions are warmer (as in the case of cold-adapted fly lines maintained at constant 25°C). By contrast, additional K⁺ in the lumen of the digestive tract could establish gradients that make K⁺ clearance from the hemolymph via the gut epithelia and renal system less energetically favorable, which would be detrimental to chill tolerance. Similarly, storing additional K⁺ in the hemolymph seems maladaptive (as survival depends on preventing hemolymph hyperkalemia), but recent findings suggest that the depolarizing effects of a putative cold-acclimation-induced increase in hemolymph K⁺ concentration could be mitigated by an increased expression of K⁺ channels in the cell membranes (Bayley et al., 2020). Lastly, there is the possibility that the increase in whole-fly K⁺ concentration after cold acclimation is the result of a wholesale increase in K⁺ content in all bodily compartments, which, in the light of the current conceptual model for insect chill susceptibility (Overgaard and MacMillan, 2017), would be detrimental for the maintenance of membrane potential, and by extension survival, unless counterbalanced by a suite of other (potentially costly) adaptive responses (e.g. differential expression of channels and/or ion transporters). Changes in ion content at the organismal level may arise not just from changes in ion concentrations, but also from changes in the relative size of organs and extracellular fluids within the body. In this scenario, cold-acclimated or cold-adapted flies may retain more or less total K (respectively) because of underlying variation in the relative investment in organs that contain more or less K per unit mass.

Overall, the findings presented here arise from complimentary field and laboratory approaches and past perspectives from both comparative evolutionary genomics and integrative and comparative physiology. We found that phenotypic plasticity and spatial adaptation can both shape cold tolerance, but that acclimation produced the greatest magnitude change in this trait in D. melanogaster. Given that ion balance is critical to chill tolerance, changes to ion uptake and elimination processes may drive the differences in whole-organism ion stoichiometry that we observed here, and thereby alter thermal performance. While we have described these changes and how they can differ at the whole-animal level, there remains a significant gap in our understanding of how these changes relate to ion gradients across cell membranes and epithelia, and how, precisely, such changes modulate low temperature performance traits. Importantly, the patterns we observed here in D. melanogaster may not reflect the ways Na and K stoichiometry changes in response to acclimation or thermal adaptation in other insect species with different evolutionary history, diet, or phenology. Differences in ionic relationships associated with cold acclimation and adaptation present a useful model to ask broader questions about whether plastic changes in gene expression involve different genes and gene networks than those under selection during adaptation to cold temperatures. In the present study, we treated acclimation and special and temporal variation in basal tolerance as separate, but it remains unclear how the capacity for plasticity can itself evolve in response to change, and to what degree the same physiological mechanisms underlie differences in acclimation capacity and basal tolerance.

Patterns of clinal variation in genotype and phenotype along the east coast of North America are amongst the best studied cases for spatial adaptation in D. melanogaster (Bergland et al., 2016; Schmidt et al., 2008; Telonis-Scott et al., 2011). We created five independent outbred populations from each of three locations (Florida, Pennsylvania and Maine) to assess the extent of genetic differentiation in thermal tolerance across latitude. To create each outbred population, we randomly selected ten isofemale lines collected from each latitude and combined ten males and ten females from each line into a small cage to allow for recombination. The F1 populations were allowed to expand to >5000 flies and we then collected 100 eggs from each of these populations for the F2 generation. Each population was then held at a minimum population of 300 flies until experimentation. We conducted this protocol five times for each latitudinal locality to generate independent outbred populations from each location.

Adaptive tracking (e.g. continuous adaptation in thermal performance in response to seasonal changes) within a given locality can also produce pronounced genotypic and phenotypic evolution (Behrman et al., 2015; Bergland et al., 2014; Bitter et al., 2024 preprint; Rudman et al., 2022; Schmidt et al., 2005). We conducted an experiment in an outdoor orchard facility located at the University of Pennsylvania aimed at assessing the extent of temporal adaptation in response to seasonal change (for more information on the experimental facility see Rudman et al., 2019). We released 1000 individuals from an outbred population constructed from 150 isofemale lines originally collected in Pennsylvania into five independent 2 m×2 m×2 m outdoor mesh enclosures on 9 July 2019 (Grainger et al., 2021). We fed each population 400 ml of modified Bloomington recipe three times weekly. To represent summer adapted flies, we collected ∼5000 eggs from each outdoor cage toward the end of summer (12 September) and kept them reproducing at density-controlled and constant abiotic laboratory conditions until the autumn sampling. We collected eggs from each cage again on 23 October to represent autumn-adapted fly populations. Prior to any measurements conducted in this study, Summer and Fall populations were held in common garden conditions for at least five generations to remove any environmental effects and isolate the impacts of rapid evolution. Temperatures recorded in the field during this time are shown in Fig. S1.

All fly lines were reared in the laboratory in 30 ml glass vials with 7 ml of a banana, corn syrup, agar, and yeast-based medium at 25°C and on a 12 h:12 h light:dark (L:D) cycle. Parental flies from each line were transferred into fresh vials and allowed to lay eggs at 25°C for 24 h, after which the adult flies were removed and the eggs were left under the same rearing conditions until adult emergence (at ∼10 days of age), at which point they were moved to vials containing fresh medium. Three days post-emergence, adult flies were sexed under light CO₂ anesthesia (less than 5 min to avoid effects of CO₂ on thermal performance; Nilson et al., 2006), and ∼50 flies of each sex were separated and placed in fresh vials.

A separate D. melanogaster strain, which was originally collected from London and Niagara on the Lake Ontario, Canada (Marshall and Sinclair, 2010) was also used to investigate long-term acclimation. Flies were maintained in 250 ml plastic bottles containing the same diet and under the same conditions (25°C, 12 h:12 h L:D) as described above. Egg laying and development of experimental flies occurred as described for the other lines. Upon emergence, however, adult flies for the acclimation experiment were split into two groups: half the flies remained at the same conditions (25°C), while the other half were transferred to a 15°C incubator with the same light cycle. Flies matured at their designated acclimation temperature for 3 days, were sexed (as above), placed in separate vials of ∼50 male and female flies and acclimated for an additional 3 days. Thus, all flies used for experiments were approximately 6-8 days post adult emergence, and acclimated flies spent 6 days at their respective acclimation temperature.

Our CCRT assay followed previously described methods (MacMillan et al., 2015c). Briefly, flies from each experimental condition (n=15 flies) were collected and separated individually into 5 ml glass screw-top vials. The vials were sealed, put inside a plastic bag, and then submerged in a mixture of ice and water (0°C) for 6 h. The CCRT was recorded as the time it takes for individual flies to recover a standing position once removed from the ice-water bath and placed at room temperature (22°C) without being disturbed.

Individual flies of each experimental population (n=15-20 flies) were collected into pre-weighed individual 0.2 ml heat-resistant PCR tubes. Each fly was weighed (inside the tube) using a Sartorius ME5 microbalance (Sartorius Lab Instruments, Goettingen, Germany), and this mass was subtracted from the empty tube mass to determine a wet mass (WM) of the fly. The flies were then dried for ∼24 h in a drying oven at 60°C then reweighed (inside the tube) using the microscale to obtain a dry mass (DM). WM and DM were then used to determine total water content of individual flies.

Whole-body ion contents were measured using flame photometry. First, a stock solution of 100 ppm Li⁺ (Cl was always the counter ion) was prepared from a 3 M Li by dilution with Milli-Q water. Five standard solutions of K⁺ and Na⁺ between 0.1 mM and 5 mM were prepared by dilution with 100 ppm Li, after which 750 µl of each standard was transferred to a conical tube and diluted up to a final volume of 10 ml with 100 ppm Li solution. Standards were prepared in triplicates for measurement in the Sherwood Model 420 Flame Photometer (Sherwood Scientific, Cambridge, UK). After calibration with a blank solution of 100 ppm Li, K⁺ and Na⁺ concentrations were recorded from five prepared standards to create K⁺ and Na⁺ standard curves. Fly samples were prepared by transferring the dried flies to 1.7 ml microcentrifuge tubes and grinding each fly using a mortar and pestle before adding 200 µl of 100 ppm Li to the tube. The samples were homogenized using a sonicator (Qsonica, Newton, CT, USA), until the solution was visibly homogenous. The tubes were then centrifuged at 10,000 g for 5 min at 4°C. After centrifugation, 150 µl of the supernatant was transferred to a 2 ml microcentrifuge tube and diluted up to a final volume of 2 ml with 100 ppm Li solution. The pellet was discarded. Na⁺ and K⁺ concentrations in the diluted supernatants were measured using the flame photometer and converted to [Na⁺] and [K⁺] in each fly (interpreted as an averaging of all concentrations in all parts of the fly) by reference to the standard curves and accounting for initial water content. Na and K content were also expressed relative to dry mass (rather than water content) to elucidate whether ion content or water content was driving differences in mean concentration. This was done by converting concentrations to molar quantities of each element and accounting for the individual fly's dry mass.

All data analysis was done in the R language for statistical computing (version 3.0.1) (http://www.r-project.org). We tested for differences in CCRT among acclimated, climate-adapted or seasonally adapted D. melanogaster using mixed-effects models using the lme() function (https://cran.r-project.org/web/packages/nlme/index.html) with treatment/climate/season, and sex treated as fixed effects and with line and sampling date treated as random effects. Similarly, we tested for differences in ion concentration (in mM) and ion content (in mmol/mg) of sodium and potassium obtained by flame photometry using mixed-effects models with the same model form. We also tested for differences in water content (as a proportion of body mass) using measurements of wet and dry mass and analyzed these differences using the same mixed-effects models. In all cases, sampling date had no effect on our results, so we simplified our analysis by removing collection date as a random effect. We found intriguing patterns in the relationship between ion and water content in acclimated flies and tested whether ion content was tied to variation in water content (as a proportion of body mass) separately in male and female acclimated files using generalized linear models with water content and acclimation temperature treated as factors. To test for the same patterns in climate and seasonally adapted flies we used mixed effects models with treatment/climate/season, and sex treated as factors and with line treated as a random effect. All data from this study are available as supplementary material (Dataset 1).