Integrating water balance mechanisms into predictions of insect responses to climate change Free

Insects are successful in terrestrial environments partly because they have overcome the challenges of a small surface area to volume ratio to maintain water balance, even in some of the driest places on Earth (Edney, 1977; Sømme, 1995; Chown and Nicolson, 2004; Harrison et al., 2012; O'Donnell, 2022; Benoit et al., 2023). Nevertheless, water availability and humidity can limit fecundity (e.g. Dahl and Renault, 2022), modify development (e.g. Tamiru et al., 2012), shift dietary targets (e.g. Becker and McCluney, 2021), alter activity patterns (Johnson et al., 2023), determine responses to heatwaves (Maurya et al., 2021) and change the top-down influence of predators in food webs (McCluney, 2017). Thus, water balance is critical to the survival and success of insects in a changing climate (Benoit et al., 2023; Brown et al., 2023). Although behaviour can mitigate exposure to desiccation (e.g. Klok and Chown, 1999; Pincebourde and Woods, 2012; Woods et al., 2022; Johnson et al., 2023), and some insects can manipulate the humidity of their environs (e.g. Lüscher, 1961; Nicolson, 2009), water homeostasis remains essential to surviving and thriving in desiccating environments (Benoit et al., 2023; Halberg and Denholm, 2024). The physiology, behaviour and importance of insect water balance have been recently reviewed (O'Donnell, 2022; Benoit et al., 2023; Brown et al., 2023; Halberg and Denholm, 2024), and several classic books summarise the general mechanisms and relationships of organismal-level water balance (Edney, 1977; Hadley, 1994; Sømme, 1995; Chown and Nicolson, 2004; Harrison et al., 2012), and its evolution and role in macrophysiology have been reviewed (Addo-Bediako et al., 2001; Chown, 2002; Chown et al., 2011). In this Commentary, we explore how our knowledge of the mechanisms underlying insect water balance can be incorporated into projections of insect responses to climate change and used to develop nuanced predictions at diverse spatial and temporal scales.

Insect water loss rate (WLR) is a function of the physical properties of water, air and solutes, their interactions, and the ways these interactions change with temperature. Relative humidity (RH; see Glossary) expresses the proportion of the holding capacity of the air that is occupied by water vapour (Fig. 1). The absolute difference between holding capacity and current water vapour pressure (WVP; see Glossary) of the air is the vapour pressure deficit [VPD, also saturation deficit (SD); see Glossary]. Temperature, atmospheric pressure and dissolved solutes all affect water activity (w_a; effectively, the average energy state of the water molecules; see Glossary) in an aqueous solution, and net flow of water molecules is down w_a gradients, which can be osmotic gradients or gradients between a liquid and air at different w_a (Bradley, 2009). The w_a of an insect's haemolymph (driven by haemolymph osmolality) can therefore be recalculated as the vapour pressure (VP) of the haemolymph (Wharton, 1985). The VPD of the air surrounding the insect usually exceeds the haemolymph VP, driving water out of the insect (Fig. 2). If the surrounding air is very humid, or the haemolymph concentration is very high, then haemolymph VP can equal (or even exceed) air VPD, and there will be no net outflow of water. Thus, because haemolymph concentration increases during dehydration, water loss rate should decrease toward an asymptote. The cuticle impedes flow down these gradients, reducing water loss from the insect, but also prevents atmospheric water vapour from entering (Hadley et al., 1986; Machin and Lampert, 1989), and the VPD of the air and cuticular permeability are more important drivers of water loss rate than haemolymph concentration in most circumstances.

Tolerance of high temperatures and dehydration may be confounded for insects under some circumstances. Both the w_a and the water holding capacity (see Glossary) of air increase with increasing temperature (Fig. 1B), which means that an insect's SD relative to air (SD_insect) will change during the day, even under constant RH. Thus, the actual desiccation stress experienced by insects is dynamic on multiple temporal and spatial scales (Brown et al., 2023), and RH is not a useful metric of VPD except under constant temperatures. High temperatures will be accompanied by increased metabolic rate (increasing respiratory water loss, RWL; see Glossary; Chown, 2002) as well as faster cuticular water loss (CWL; Gibbs et al., 1998; see Glossary). Many insects can behaviourally thermoregulate, and the effect of temperature on VP can potentially magnify water loss in these cases (Fig. 2E). Prolonged exposure to high temperatures can cause death from dehydration (Santos et al., 2011), and we explore this temperature–water loss interaction below.

Insect water balance is plastic, varies among and within species, and is often tied to the hygric environment (see Glossary). For example, monsoons in seasonally wet tropical and sub-tropical environments can cue diapause entry or exit (Denlinger, 2022) and lead to changes in water balance (Parkash and Ranga, 2014); brief exposure to desiccation can lead to rapid desiccation hardening (e.g. Hoffmann, 1990; Bazinet et al., 2010; Bosua et al., 2023); and spatial gradients of desiccation pressure are often accompanied by intraspecific variation in dehydration tolerance (Gibbs et al., 1991; Hoffmann, 1991). Indeed, unpredictable water availability has driven the evolution of anhydrobiosis in Polypedilum midge larvae (Gusev et al., 2014; Shaikhutdinov et al., 2023). By contrast, in temperate and polar environments during winter, water locked in ice has very low w_a and is biologically unavailable, generating a strongly desiccating environment for overwintering insects (Danks, 2000; Sinclair et al., 2013). Some insects co-opt this desiccation to enhance their cold tolerance by reducing the amount of freezable water in their bodies (Sformo et al., 2010; Duell et al., 2022). Thus, the hygric environment clearly exerts selective pressure on insect water balance traits.

Fluid-feeding insects face a different set of water balance challenges. The nutrients in fluids – xylem, phloem and nectar from plants, or blood from animals – are often dilute or in ratios that do not match insect needs (e.g. phloem sap is sugar-rich, but depauperate in protein). Furthermore, drinking blood from endotherms is thermally challenging (Benoit et al., 2011; Lahondère et al., 2017), and insects must rapidly reduce the volume (i.e. mass) of their ingested meal to allow them to escape retribution from their prey (Roitberg et al., 2003). Thus, although fluid feeders do not suffer desiccation stress while they are feeding, the water balance machinery must process the huge quantities of fluid needed to satisfy their nutritional requirements. The resultant water excretion is sufficient to evaporatively cool mosquitoes and large cicadas (Sanborn et al., 1992; Lahondere and Lazzari, 2012). Aquatic life stages of freshwater insects have access to unlimited water, but must counteract a net influx of water; by contrast, larvae in seawater or other high osmotic environments experience a net efflux of water similar to that with which terrestrial insects contend (Bradley, 2009). Exploring these topics is beyond the scope of this Commentary, but we note that water balance in fluid-feeding and aquatic species is similar to that of other insects when they not feeding, or when in a terrestrial adult stage.

Here, we summarise how insects maintain water balance in relation to their environment, and how water relations might mediate the impacts of climate change on insects. We then focus on the inextricable relationship between water relations and heat tolerance, and propose one way to integrate water balance into the current temperature-focused predictions of the risks that insects face in a changing world.

Water balance, as the phrase implies, is the maintenance of adequate body water by modulating water intake and water loss (Fig. 3; Wharton, 1985). Acquiring water during dehydration prevents or delays desiccation stress. Drinking liquid water or eating moist food are the most obvious water sources, and some Namib tenebrionid beetles can even harvest drinkable water from fog (Hamilton and Seely, 1976). There are also several metabolic mechanisms that insects can use to liberate osmotically inaccessible water already within the body. Metabolising stored glycogen liberates water molecules that are hydrogen-bonded to the glycogen molecule (Marron et al., 2003; Kalra and Gefen, 2012). Oxidising any substrate produces metabolic water. Metabolic water may supply all necessary water for insects with very efficient water retention, such as desert tenebrionid beetles (e.g. Nicolson, 1980). A few insects can absorb water vapour directly via concentrated solutes that create a w_a gradient between humid air and some part of the animal that is sufficiently permeable. This absorption occurs at the mouthparts in desert cockroaches (O'Donnell, 1977), the rectum in the larvae of tenebrionid beetles and firebrats (Noble-Nesbitt, 1970; Hansen et al., 2006), and across the highly permeable cuticle of some Collembola (Holmstrup et al., 2001).

Regulating movement of water between the gut and haemocoel is key to insect water balance, allowing insects to modulate excretory water loss (O'Donnell, 2022; see Glossary). Water is generally absorbed at the midgut. Although this absorption receives little attention in insect water balance papers, the midgut's importance in nutrient absorption, disease vectoring and RNAi-mediated biocontrol mean its structure and function are well known (Caccia et al., 2019; Terra et al., 2019; Kunte et al., 2020; Terra and Ferreira, 2020; Capriotti et al., 2021; Hajkazemian et al., 2021). The mechanisms and regulation of urine secretion by the Malpighian tubules (first described by Wigglesworth, 1931a,b,c) continue to be well studied (Dow et al., 2022; Farina et al., 2022; O'Donnell, 2022; Halberg and Denholm, 2024). Secretion by the Malpighian tubules is easily measured via the Ramsay assay (Ramsay, 1953; see Glossary), facilitating comparative studies (e.g. Halberg et al., 2015) and fine-scale studies of regulation and detoxification (Chahine and O'Donnell, 2011). Water reabsorption across the hindgut is as important as secretion by the Malpighian tubules; however – perhaps because measuring it is technically challenging (Andersen and Overgaard, 2020) – hindgut re-absorption receives less attention. Nevertheless, the mechanistic underpinnings and regulation of absorption by the hindgut and rectum have been well explored (Phillips et al., 1987; Coast et al., 2002), and the interactions between the Malpighian tubules and rectum in the cryptonephridial complex remain the subject of active research (e.g. Kapoor et al., 2021). Recent work on the low temperature properties of the hindgut reflects the emerging role of water balance in cold tolerance (e.g. Andersen and Overgaard, 2020; Brzezinski and MacMillan, 2020; El-Saadi et al., 2023).

Insects lose the most water across the relatively large surface area of their cuticle (Chown and Nicolson, 2004). However, cuticular waterproofing from both structural components (Ramniwas et al., 2013; King and Sinclair, 2015) and secreted hydrocarbons (Gibbs, 1998) means that most insects lose relatively little water even at high VPD. Physical abrasion and lipid melting at high temperatures can dramatically increase CWL (Gibbs, 2002a; Johnson et al., 2011). The synthesis and regulation of cuticular hydrocarbons (CHCs; see Glossary) which waterproof the cuticle (and also have communication and defence roles) is reasonably well understood (Blomquist and Ginzel, 2021), and the underlying genes and enzymes are being identified and linked specifically to CWL (e.g. Moriconi et al., 2019; Holze et al., 2021; Xin et al., 2022). The structure of the cuticle, through melanisation and sclerotisation, is also thought to modulate CWL (Hepburn, 1985); in particular, melanism is associated with low CWL in many insects (Rajpurohit et al., 2008; Ramniwas et al., 2013; King and Sinclair, 2015; Farnesi et al., 2017). The pathways regulating cuticle tanning are also reasonably well known (Anderson, 1985), and manipulating these genes directly affects CWL (Noh et al., 2015; Bai et al., 2022).

RWL typically accounts for <15% of total water loss (Chown, 2002), but has been well explored because of its possible association with discontinuous gas exchange (DGE; see Glossary), wherein restricted spiracular opening reduces the opportunities to lose water via the tracheal system. The actual importance of DGE for water balance is unclear: arguments both for and against the role of water balance in the evolution of gas-exchange cycles have been well articulated (Chown et al., 2006; White et al., 2007; Terblanche et al., 2010; Contreras et al., 2014; Matthews, 2018; Terblanche and Woods, 2018; Oladipupo et al., 2022). Insects can also reduce RWL by creating humid microenvironments – for example, under the elytra of desert tenebrionid beetles (Byrne and Duncan, 2003; Chown and Nicolson, 2004) – or possibly by simply reducing their metabolic rate (Zachariassen, 1996).

An insect ultimately dies after losing some unsustainable amount of water (Hadley, 1994), so beginning dehydration with more water delays reaching that lethal point (Fig. 3B). During dehydration, both Drosophila melanogaster and arid-adapted African tenebrionid beetles appear to preferentially lose water from the haemolymph to preserve cell volume (Zachariassen and Einarson, 1993; Folk et al., 2001; Albers and Bradley, 2004). Storing water in the haemolymph creates osmotic challenges; Drosophila increase their initial water content without these osmotic impacts by accumulating glycogen, to which large amounts of water is hydrogen bonded (Marron et al., 2003). Although laboratory experiments usually commence with fully hydrated animals, the hydration state of insects varies considerably in nature (Riddell et al., 2023), and affects water balance. This variation in hydration state could be confounded with other ecologically relevant parameters, such as body condition and feeding history. For example, both energy stores (glycogen) and water will determine the initial water content of Drosophila (Marron et al., 2003), and the opportunity to replenish carbohydrate stores between desiccation bouts will presumably determine their response to repeated dehydration events.

The amount of water an insect can survive losing varies among species. For example, water content at death varies threefold among similar-sized Ceratitis flies, depending on sex, temperature and ambient humidity (Weldon et al., 2016), whereas water content at death barely changes across multiple Glossina spp. (Bursell, 1959) and adult Drosophila spp. (Gibbs, 2002b). Nevertheless, some insects can tolerate considerable dehydration. For example, the semiaquatic beetle Peltodytes muticus remains active after losing 89% of its total water content (Arlian and Staiger, 1979). At the extreme, some Polypedilum midges are anhydrobiotic, withstanding the loss of >99% of their body water (Hinton, 1960; Watanabe et al., 2002). Such extreme tolerance requires unusual cellular adaptations including the accumulation of cytoprotectants and dehydration-related proteins (Sogame and Kikawada, 2017). Freeze-tolerant insects appear to survive extensive cellular dehydration, so the hypothesised causes of mortality in frozen insects [e.g. molecular crowding, protein misfolding, and high [Ca²⁺] initiating apoptosis (Toxopeus and Sinclair, 2018)] could well overlap with the causes of dehydration mortality (see also Sinclair et al., 2013).

Thus, we have a broad, and often deep, understanding of the mechanisms associated with water balance at both the organismal and sub-organismal levels. However, we specifically lack a cellular level explanation for why insects die of dehydration, and although the traits associated with water balance are well understood, the sources of variation in those traits are less clear. As is the case with other stressors (Hoffmann et al., 2021), we are some way from identifying genomic markers that predict an insect's water balance characteristics and responses to dehydration. Furthermore, although we know that behaviour modulates exposure to desiccation, we lack sufficient data to incorporate field behaviour into a water budget (but see Woods and Bernays, 2000; Becker et al., 2021; Johnson et al., 2023; Riddell et al., 2023).

Water balance can determine insect population dynamics, geographic distribution and seasonality. For example, Antarctic Collembola appear to partition habitat based on water availability (Hayward et al., 2004; Sinclair et al., 2006b,a); monsoons determine the seasonality of many tropical insects (Denlinger, 1980), including both adult mosquitoes and the diseases they vector (Brown et al., 2023). Furthermore, anthropogenic changes in water availability – such as irrigation – deeply alter the outcomes of modelled species ranges or population dynamics (Macfadyen and Kriticos, 2012; de Villiers et al., 2017; Barton et al., 2019). Finally, water availability and rainfall seasonality can indirectly affect insects via host plant quality (e.g. Liang et al., 2022). Thus, insect water balance is likely key to predicting insect responses to their environment. The hygric environment is implicitly incorporated (typically as precipitation) into correlative bioclimatic envelope models (Phillips et al., 2006). Although water is not explicitly included in temperature-focused, trait-based models of insect responses and distributions (e.g. Deutsch et al., 2008; Sunday et al., 2012; Rezende et al., 2014, 2020; Jørgensen et al., 2022), one might argue that the empirical input data already implicitly include the effects of temperature on desiccation stress. More pointedly, the sheer complexity of water balance–temperature–microenvironment interactions makes reducing water balance traits to a simple parameter or index difficult compared with convenient endpoint measures such as the critical thermal maximum (CT_max; see Glossary).

How important is it to use water balance when models linking temperature and exposure time to mortality are already strikingly successful? We know that climate change will affect the hygric environment (Konapala et al., 2020), and that interactions between thermal tolerance and water balance could substantially modify the impacts of a warming world on terrestrial ectotherms (e.g. Riddell et al., 2017; Rozen-Rechels et al., 2019; Henry et al., 2022; Brown et al., 2023). Thermal death time (TDT; see Glossary) models incorporate the effect of exposure time on mortality at a given temperature (Rezende et al., 2014; Jørgensen et al., 2021), but we contend that this does not equate to implicit inclusion of temperature–humidity interactions, even though humidity always changes with temperature (Fig. 1), because insects can actively change their water balance in relation to the conditions, and WLR itself is non-linear over time. Thus, TDT input data collected under constant conditions may yield a model that does not apply to an insect exposed to different conditions, and we are not aware of an exploration of whether TDT models are robust to variation in environmental factors other than temperature.

We believe that considering a priori the mechanisms underlying mortality during a high temperature exposure demonstrates a role for water balance. In dry environments, dehydration (not heat) could be the cause of death if water loss is very high. By contrast, water loss will not substantially contribute to mortality in humid conditions, so cellular and organismal susceptibility to high temperatures will take primacy. Furthermore, because water loss is usually relatively slow (even in animals with large surface areas), desiccation is more likely to affect thermal limits at very slow heating rates or during longer exposures (Santos et al., 2011), or when strenuous activity (e.g. flight) increases water loss rate (Johnson et al., 2023). However, data on how humidity modifies thermal tolerance are surprisingly equivocal. Some studies show that thermal limits (e.g. CT_max, knockdown time or a sublethal measure) can be substantially affected by hydration state (e.g. Riddell et al., 2023) or humidity of the environment during the assay (e.g. Buxton, 1931; Mellanby, 1932), while others detect no impact of water balance on thermal limits (e.g. Terblanche et al., 2011; Overgaard et al., 2012).

Subtle methodological differences probably account for many of the discrepancies among experiments evaluating the effect of water balance on thermal tolerance. For example, Terblanche et al. (2011) found that humid conditions increased high temperature knockdown time when D. melanogaster adults were held at a static temperature, but not when the temperature was increased over a slow ramp. Indeed, even the size of the container in which the assay is conducted could modulate the influence of the hygric environment – as insects lose water, the WVP of the air in their container increases – in small containers, this can be sufficient to meaningfully decrease the SD (X. Xu and B.J.S., unpublished observations). Measuring and reporting VPD (rather than RH) will bring these effects into sharper focus. Because the water content of air can be tightly controlled in flow-through systems (Hunt, 2003), it should be possible to separate the effects of temperature and water loss experimentally. This could yield information about the circumstances when temperature and humidity interact to affect survival and help identify whether insects experience these conditions in nature. We can identify key traits that might determine this relationship. For example, body size affects CWL (larger insects have a smaller surface area to volume ratio; e.g. Terblanche et al., 2011); hindgut structure determines the capacity to reduce excretory water loss (Hadley, 1994); and the capacity and propensity to store glycogen modify the amount of stored water (Marron et al., 2003).

We speculate that water balance will modify both the elevation and slope of insects' TDT surfaces in nature in a humidity-specific manner. One way to account for water balance would be to parameterise the TDT surface for a range of hygric conditions, adding a humidity parameter to the (now multi-dimensional) survival surface. Like other multiple stressor problems (for discussion, see Todgham and Stillman, 2013; Kaunisto et al., 2016), this approach instantly multiplies the number of experiments required to parameterise the model, highlighting a central challenge with models that require phenotypic input. Other modelling approaches can be used to predict thermal tolerance whilst incorporating water balance. Riddell et al. (2023) showed that plasticity in water loss rate is associated with thermal limits in a click beetle, such that water-replete individuals have higher water loss rates and a lower CT_max. From these parameters, they predicted that climate change will reduce the ability of the beetles to remain active in full sun during the summer. Riddell et al.'s (2023) model uses a known (and directional) connection between water loss rate and thermal tolerance to include water balance; but because the mechanisms underlying the connection are unknown, the approach still requires a large phenomenological dataset. Of course, the exact relationships between water balance and thermal tolerance will vary by species and population, and will also vary as a result of plasticity (Hoffmann, 1990, 1991; Hoffmann and Watson, 1993; Weldon et al., 2018; Bosua et al., 2023; but see Hoffmann et al., 2003). Therefore, an elegant empirical model for one species will probably not apply even to its close relatives. Thus, an ultimate goal is to derive models from underlying mechanisms, rather than parameterising them from exhaustive phenomenological measurements.

Temperature clearly affects an insect's water loss rate regardless of whether water balance affects thermal tolerance. Thus, a minimal mechanistic model of thermal tolerance must account for a species' univariate responses to both temperature and water loss (and recognise that there are some circumstances where one or other has a negligible effect on survival). The TDT framework generates standardised parameters for thermal tolerance from laboratory data (Rezende et al., 2014; Jørgensen et al., 2021), but we have not yet converged on standardised metrics of water balance for insects (the plant science community has begun to tackle this problem; e.g. Fatichi et al., 2016; Duursma et al., 2019; Kannenberg et al., 2022; De Cáceres et al., 2023). We know that water loss rate is temperature and SD dependent (Fig. 2), that water loss rate increases at the temperature where cuticular lipids melt (Gibbs, 1998), and that insects vary in both their initial water content and water content at death (Hadley, 1994). All of these parameters can be routinely measured (Moretti et al., 2016). Furthermore, these parameters should (in principle) be sufficient to identify the threshold where a TDT model switches from a temperature-focused survival surface to a desiccation-focused one.

However, adding additional traits to an empirical parameterisation does not make a mechanistic model. Extrapolating beyond model species is an overarching challenge when attempting to unravel biological responses to multiple interacting stressors (Todgham and Stillman, 2013). If the mechanisms are well understood, then comparing, for example, Malpighian tubule function or CWL among species, life stages or acclimation states allows some understanding of the role of that mechanism (water loss regulation) in determining the trait (water loss rate) and then the phenotype (desiccation survival). Of course, we cannot yet predict phenotype a priori for either thermal tolerance or water balance; thus, identifying markers that allow for the prediction of a species' physiology (and thence climate change responses) remains a Grand Challenge (see Mykles et al., 2010; Somero, 2010). Nevertheless, detailed biophysical models have been made for water relations in plants (Fatichi et al., 2016) and insects (e.g. Woods et al., 2005), so routes exist to connect mechanism to phenotype, and thence to move beyond simply parameterising descriptive models.

Understanding the sub-organismal mechanisms determining dehydration tolerance and thermal limits will shed light on the causes of additive, synergistic and antagonistic interactions such that they can be predicted a priori (Sinclair et al., 2013; Todgham and Stillman, 2013; Kaunisto et al., 2016). In the absence of clearly tested (or perhaps testable) hypotheses, we suggest that modern comparative -omics-based tools could help identify commonalities in the physiological mechanisms underlying the responses to dehydrating conditions (Torson et al., 2020; Hoffmann et al., 2021). For example, if transcriptomic responses to the water×temperature interaction are consistent among species, then the underlying mechanisms are likely shared. If those shared responses yield similar outcomes (i.e. desiccation effects on the TDT surface are consistent among species), then the interaction may be generalisable among species (Kaunisto et al., 2016). In turn, generalisable predictions will allow expansion from temperature-only (or water balance-only) models (e.g. Rezende et al., 2020; Jørgensen et al., 2022) to incorporate the temperature–dehydration landscape (see also Brown et al., 2023; Riddell et al., 2023).

The predictive value of any model depends on the quality of input environmental data. This is glossed over for temperature by relying on meteorological data (Duffy et al., 2015), but tools to extrapolate from macroclimate to microclimate are available (e.g. Kearney and Porter, 2020), and increasingly effective mechanistic models of microclimate are being developed (e.g. Pincebourde and Woods, 2012; Koussoroplis et al., 2017; Pincebourde and Woods, 2020; Maclean and Klinges, 2021; Pincebourde et al., 2021; Srygley et al., 2023). Operative temperature models mimic the morphology and absorptivity of an animal for steady-state heat transfer (Dzialowski, 2005), and can be used to derive null models, evaluate biophysical and behavioural drivers of thermal biology, and forecast potential risks of climate change (e.g. O'Neill and Rolston, 2007; Harris et al., 2015). Operative models of evaporation evaporative water loss (EWL) have been widely used for amphibians (e.g. Navas and Araujo, 2000; Tracy et al., 2007; Riddell et al., 2017) but not, to our knowledge, for insects. While operative models of water loss may be unnecessary if we can use an insect's water balance traits in concert with temperature and vapour pressure to predict dehydration stress, they can also reveal fine-scale information about the habitat's hygric heterogeneity. Hygric heterogeneity was being explored in Journal of Experimental Biology as far back as the 1930s (Ramsay et al., 1938), and some data are already available (e.g. in relation to leaf surfaces; Ferro and Southwick, 1984; Pincebourde and Woods, 2012).

The final piece in the puzzle of how water balance and temperature will interact to determine insects' responses to climate change is beyond the scope of comparative physiology. The available predictions of how climate change will alter the hygric environment are weak at the level of monthly precipitation, and non-existent for the microclimate-scale vapour pressure information necessary to match the scale and precision of our understanding of insect responses. For temperature data, authors often eschew the complexity of changing extremes, night-time temperatures and seasonality (IPCC, 2022) to simply increase currently observed temperatures by a global average number predicted from IPCC models. We expect that the most expedient approach for including water in projections of insect responses to climate change is to model responses to plausible high and low humidities under the predicted future regime to circumscribe the range of potential responses. This will constitute something of a sensitivity analysis that reveals the importance of water balance in the model. More complicated models might derive hygric conditions from projected precipitation patterns (see Ferro and Southwick, 1984; Barton et al., 2019; Riddell et al., 2023), and perhaps leverage existing work on microclimate humidity heterogeneity (e.g. Ramsay et al., 1938; Ferro and Southwick, 1984; Pincebourde and Woods, 2012; Pincebourde and Casas, 2019).

Water balance is key to insect success, and likely their responses to climate change, but is seldom incorporated into models of insect responses to climate change. We suggest that deriving mechanistic models that predict the circumstances under which humidity changes insects' responses to the thermal environment is a good first step to using water balance in climate change predictions. A longer-term goal is to predict water balance (and thermal biology) traits a priori to allow broader predictions of water–temperature interactions across a range of species. However, using this knowledge in predictions remains hampered by a dearth of quality microenvironmental data and predictions of the effect of climate change on the hygric environment at a scale relevant to insects.

We are grateful to Erika Huisamen for bibliometric summaries that helped refine the scope of this Review and our thinking on the topic, and to Sylvain Pincebourde, JEB editorial staff and two anonymous reviewers whose constructive suggestions and critical evaluation improved all aspects of this manuscript.