(a) Gene expression

Stressful environmental conditions such as heating induce the expression of several genes that control the activity of the heat-shock proteins (HSPs). The reviewed papers that reported genetic responses of macroinvertebrates to temperature changes all recorded upregulation or downregulation of different genes, including HSP genes (Karouna-Renier & Zehr, 1999; Lencioni et al., 2013). For example, when subjected to heat-induced stress, Chironomus riparius (Diptera) did not activate or repressed some HSP genes (e.g. HSP22) while others were activated (HSP23, HSP24, HSP34, HSP27 and HSP70) suggesting that the HSP subfamily possesses remarkable functional differentiation in response to stressful temperature conditions (Martín-Folgar et al., 2015). Similarly, Chou et al. (2018) observed that Neocloeon triangulifer (Ephemeroptera) larvae bred at a chronic threshold (30 °C) upregulated indicators of thermal stress (HSP90) but not genes sensitive to hypoxia [egg laying defective 9 (EGL-9) and lactate dehydrogenase (LDH)], indicating that the upper chronic thermal limit is not set by oxygen availability. Chronic thermal stress can lead to reductions in body size and fitness through reduced food intake, which results from the upregulation of genes producing histamine and dopamine (Chou et al., 2018). Upregulation of HSP70 has also been observed in stenotherm species [Lednia sp. (Plecoptera) and Crunoecia irrorata (Coleoptera)] in their natural temperature range, indicating that the thermal niche they occupy may not be optimal due to other limiting factors such as biotic interactions or resource availability (Hotaling et al., 2020; Ebner, Ritz & von Fumetti, 2019). This challenges the assumption that the distribution of insects in cold habitats reflects evolved preferences for those temperature conditions. Teets et al. (2013) reported upregulation of genes involved in both glycogenolysis and gluconeogenesis in Belgica antarctica midges in response to heat and cold stress, suggesting that insects exposed to extreme environmental conditions mobilize carbohydrate energy stocks to allow rapid shifts in metabolism. Hotaling et al. (2020), studying high-altitude stoneflies exposed to their CT_max, identified upregulation of genes involved in the developmental transition [ATP binding cassette subfamily A member 3 (ABCA3) and hexamerins (HEXA)]. Studies on gene expression allow us to understand the physiological mechanisms underlying organismal responses to temperature changes and are imperative for correct interpretation of the causes driving biological responses at different levels, for example, to disentangle behavioural and evolutionary responses (Hotaling et al., 2020; Schmeller et al., 2018). As stated by Clarke (2003) we can identify the relationships between cellular thermal physiology and organismal physiology as well as between some macroecological patterns and temperatures, however, we are still unable to relate thermal physiology to ecology at the community scale, despite this link likely being a strong determinant of life-history traits, food-web dynamics, and biological diversity.

(b) Osmoregulation ability

Temperature affects the regulation of haemolymph osmotic and ionic concentrations in invertebrates. In general, increasing temperatures increase ion transport rates (Orr & Buchwalter, 2020). We found only one study on macroinvertebrate osmoregulation in which Colburn (1983) observed that larvae of Limnephilus assimilis (Trichoptera) exposed to a wide salinity range (0–25%) could complete their development at low temperature because cellular Cl⁻ and Na⁺ were maintained at low concentrations. On the contrary, at high temperatures (for example in hydrothermal water) they were unable to control Cl⁻ intake, leading to lower survival and decreased ability to complete development.

(c) Respiration

As for all biological processes, respiration rate is positively correlated with temperature (Sinsabaugh, 1997), hence higher temperatures enhance the oxygen consumption of invertebrates, as shown by Bergström et al. (2010) for species in lake sediments. At higher temperatures, larger amounts of energy are required for metabolic maintenance, for both respiration and assimilation, compared to at the thermal optimum (Sweeney & Vannote, 1978; Vannote & Sweeney, 1980). However, due to the decrease in oxygen solubility with increasing temperature, oxygen availability is reduced simultaneously with this greater respiratory requirement (Forster, Hirst & Atkinson, 2012). The sensitivity of species to this decrease in oxygen availability varies depending on the taxon. Some species such as Leuctra hippopus (Plecoptera) and Asellus aquaticus (Isopoda) can maintain a constant respiration rate independent of ambient oxygen levels (below a critical limit) (Rotvit & Jacobsen, 2013; Kim et al., 2017), whereas others, such as Isoperla spp. (Plecoptera) show a higher oxygen consumption with increasing temperature and a respiratory rate that is proportionally greater in larger species (Modlin & Jayne, 1981). Some chironomid species [e.g. Chironomus anthracinus (Diptera)] can shift from aerobic metabolism to partially anaerobic (Hamburger, Dall & Lindegaard, 1994) as temperature increases (from 2 to 20 °C).

(d) Body size, growth rates and size at emergence

Growth and adult body size depend on several processes regulated by temperature such as rates of ingestion, assimilation, metabolism, and excretion. Sweeney & Vannote (1978) conceptualized this temperature–growth–size relationship in their TEH according to which maximum adult size is achieved at a thermal optimum while outside this optimal range body size is reduced. Several studies have demonstrated that higher temperatures can cause acceleration of metabolism and consequently lower investment in growth, leading to premature adult development. On the contrary, at low temperatures, metabolic activity is slowed down, allowing a greater proportion of adult tissue maturation (Vannote & Sweeney, 1980; Sweeney & Vannote, 1978, Brittain, 1983; Rempel & Carter, 1987; Sweeney, Vannote & Dodds, 1986a). The size–temperature relationship is generally assessed in laboratory studies, in which organisms are bred at a constant temperature and measured and weighed at frequent intervals (typically 1–3 days). Morphological traits considered include total length, head capsule width, thorax length, pronotal length, wing length, leg length, antennal length, and body mass, depending on the taxon and developmental stage. The observed temperature–body size relationship often follows an exponential curve (Brittain, 1983; Giberson & Rosenberg, 1992a; Sweeney & Vannote, 1984; Reynolds & Benke, 2005; Rempel & Carter, 1987). Growth rates are calculated from the change in body size (both length and mass) for specific intervals of the developmental period. The temperature–body size relationship can also be studied in relation to sex and or life-cycle phase. Some experiments have shown that at high temperatures females reach smaller adult sizes than males suggesting that somatic growth is traded off against reproductive capacity (Rempel & Carter, 1987; McKie, Cranston & Pearson, 2004) while in other studies sexual dimorphism appears unaffected by temperature, with other factors such as sexual selection or fertility playing an important role (Lande, 1980; Encalada et al., 2019).

Although some evidence shows that higher temperatures lead to smaller adult body size (in agreement with the TEH), other studies on both terrestrial and aquatic ectotherms found a maximal adult size only at the coldest extreme of the species' thermal tolerance range and not at some intermediate temperature (conflicting with the TEH) (Atkinson, 1994, 1995; Sweeney et al., 2018). Such observations led to the development of the temperature size rule (TSR) (Atkinson, 1994), which was reformulated by Kingsolver & Huey (2008) as ‘hotter is smaller’. It seems to represent a special case of Bergmann's (1847) rule according to which populations/species of larger size are found in colder environments. Subsequent studies (Forster et al., 2012; Horne, Hirst & Atkinson, 2015), reviewing a large number of temperature–body size experiments involving freshwater, marine and terrestrial species have confirmed the TSR hypothesis and showed that warming-induced reductions in adult body size are larger for aquatic ectotherms than for terrestrial ones. Recently studies have begun to investigate the drivers that explain the TSR rule. Although temperature responses appear to be outcomes of phenotypic plasticity, latitudinal size gradients could depend also on genetic factors (Horne et al., 2015). Insect temperature–body size trends observed across latitudinal clines have not been replicated across altitudinal gradients (Horne, Hirst & Atkinson, 2018). TSR explanations have focused on physiological processes (such as oxygen limitation and resource availability) and responses (shorter developmental times due to higher mortality at higher temperatures), and on ecological and evolutionary mechanisms (adaptation to temperature to maximize fitness). Many of these studies support oxygen as a significant factor (Forster et al., 2012; Verberk et al., 2021). The higher cost of oxygen uptake in warmer water and the greater demands on large bodies to maintain aerobic scope in warmer environments both play important roles in determining adult size (Woods, 1999) and could explain the different temperature–size responses between aquatic and terrestrial organisms (Forster et al., 2012) and across latitude and altitude (Horne et al., 2018).

(e) Assimilation/excretion

High temperatures cause an increase in the fraction of energy needed for metabolism maintenance (Sweeney & Vannote, 1978), which requires greater food consumption and leads to faster gut clearance (Zimmerman & Wissing, 1978). At high temperatures, some organisms, such as Hydropsiche betteni (Trichoptera), seek better-quality food (animal material or algae instead of detritus) to cope with higher energy demands (Fuller & Fry, 1991), whereas Chironomus riparius (Diptera) and Mesogomphus lineatus (Odonata) do not show dietary changes depending on temperature (Péry & Garric, 2006; Pandian, Mathavan & Jeyagopl, 1979). Food uptake and assimilation rates increase with temperature up to the thermal optimum (Culler, McPeek & Ayres, 2014; Van Doorslaer & Stoks, 2005a; McCauley, Hammond & Mabry, 2018; Stoks, Swillen & De Block, 2012; Pandian et al., 1979; Péry & Garric, 2006).

(f) Thermal limits

Thermal limits are usually measured in laboratory studies (Fig. 6A) by exposing organisms to temperatures increasingly distant from their optimal temperature range (Sherberger et al., 1977). Organismal death occurs when the water temperature reaches the critical thermal limits (Sherberger et al., 1977; Rogowski & Stewart, 2016; Chou et al., 2018; Sweeney et al., 1986a; Rosillon, 1988). The upper thermal tolerance can be determined by the LT₅₀ test: this threshold represents the lethal upper temperature at which 50% of individuals die in a specified time. By contrast, the incipient lethal temperature (ILT) thermal limits are based on the most extreme temperatures at which 50% of the test organisms survive indefinitely after being transferred from an acclimation temperature directly into a constant-temperature tank where time to death is measured (Brett, 1952). A less time-consuming approach that requires smaller samples is the critical thermal method (CTM) which consists of assessing the behavioural stress response, defined as the ‘arithmetic mean of collected thermal points at which locomotor activity becomes disorganized to the point at which the organism loses its ability to escape conditions that will promptly lead to its death’ (Lowe & Vance, 1955, p. 2). For aquatic macroinvertebrates, the response includes the inability to remain attached to the substrate and hyposensitivity to stimuli. All these methods have been employed in studies of thermal biology and a review focused on terrestrial animals comparing these different approaches is available (Lutterschmidt & Hutchison, 1997). For aquatic insects, the upper thermal limit evaluated at 96 h (96-LT₅₀) and the CT_max are related by a significant positive linear relationship, establishing the CTM method for use (Dallas & Ketley, 2011). There have been various attempts to define the thermal threshold of different aquatic macroinvertebrate taxa based on laboratory experiments on individual species or using the relationship between the macroinvertebrate assemblage and the temperature regime of the water bodies where they are found. Stewart et al. (2013) defined the upper thermal tolerance of 13 taxonomic groups (mainly at order level) of southwestern Australian macroinvertebrates by reviewing the existing literature and measuring LT₅₀ for four key species. Dallas & Rivers-Moore (2012), using the CTM, determined the upper thermal limits for 27 families of South African macroinvertebrates. Polato et al. (2018) and Shah et al. (2017) quantified CT_max and CT_min of 62 EPT species from Colorado (USA) and the Andes, showing that the tropical (Andean) species had a narrower thermal tolerance than the temperate ones. Niedrist & Füreder (2020) redefined the temperature optima and thermal ranges for different species of EPT and chironomids (Diptera) using regression models for long series of water temperature data and showed that alpine benthic communities had moved to higher altitudes in the last decade due to glacial retreat.

(a) Total time of development

Several studies have shown that increasing temperature leads to shorter developmental time. Sweeney et al. (1986a) showed that the larval development of Leptophlebia intermedia (Ephemeroptera) is shorter at higher temperatures and Sweeney & Vannote (1984) reported the same for Cloeon triangulifer (Ephemeroptera). Other studies confirmed that developmental time, within the tolerance range, decreases with increasing temperature for eurythermal species (Sarvala, 1979; Frouz, Ali & Lobinske, 2002; Bayoh & Lindsay, 2003; Imholt et al., 2010; McCauley et al., 2015). By contrast, for stenothermal species like Soyedina carolinensis (Plecoptera), the shortest developmental time (~92 days) was observed at an optimal temperature (10 °C), increasing at both higher (15 °C) and lower temperatures (5 °C) (~109 and 141 days, respectively) (Sweeney, Vannote & Dodds, 1986b). The same pattern was observed for Eukiefferiella ikleyensis (Diptera), with the shortest larval stage at 14 °C compared to both higher (18 °C) and lower (9 °C) temperatures (~71, 74 and 110 days, respectively) (Storey, 1987). The relationship between temperature and developmental time for stenothermal species can be described by a parabolic curve (Sweeney et al., 1986b; Elliott, 1987) while for eurythermal species the trend typically follows a negative exponential model (Marten, 1990) or an inverse asymptotic correlation (McKie et al., 2004). Frouz et al. (2002) reported that under increasing temperatures chironomid males developed faster than females.

(b) Time and length of hatching

Temperature is a crucial determinant of invertebrate hatching time. In general, temperatures far from the optimal range induce diapause (Danks, 1987), an adaptation evolved by some organisms to extend the embryogenesis period until the environmental conditions are suitable (Pritchard, Harder & Mutch, 1996). The relationship between hatching time and temperature follows a decreasing trend best described by a power function (Brittain, 1977, 1982; Bohle, 1972; Brittain & Campbell, 1991; Elliott, 1986, 1972, Giberson & Rosenberg, 1992b; Humpesch, 1980a,b; Mendonça et al., 2018), or a hyperbolic power function (Elliott, 1978; Friesen, Flannagan & Lawrence, 1979), at least within the temperature tolerance range. Hatching time decreases at higher temperatures, more sharply in warm-adapted species such as Odonata than in cold-adapted species such as Plecoptera (Pritchard et al., 1996; Bouton, Iserbyt & Van Gossum, 2011). Diapause is longer at high temperatures for stenothermal species; eurytherms can survive low temperatures by remaining dormant (Pritchard et al., 1996). Embryonic period is positively correlated with egg size and the relationship between these does not seem to change with temperature (in the range 10–25 °C) in both univoltine and multivoltine species of mayflies, stoneflies, caddisflies, Coleoptera, Hemiptera and dragonflies (Gillooly & Dodson, 2000b).

(c) Time and length of emergence

Increasing temperatures typically lead to earlier emergence of insects (Nebeker, 1971; Rempel & Carter, 1987; Vannote & Sweeney, 1980; McCauley et al., 2018). In aquatic environments characterized by a variable temperature regime, the pivotal factor regulating emergence is temperature while in constant-temperature habitats photoperiod plays a major role (Ivković et al., 2013). Water temperature is the primary driver that determines the timing of emergence for holometabolous insects (where the pupae are submerged) while other variables (such as humidity and air temperature) are involved for hemimetabolous insects (Trottier, 1973b; Ivković et al., 2013). In recent decades the emergence of Odonata adults takes place earlier in the year due to increased temperatures. According to Hassall & Thompson (2008), British Odonata have advanced their emergence by about 1.15 days per decade and 3 days per degree between 1960 and 2004, showing a phenological response to climate change similar to those observed for terrestrial taxa (Lepidoptera, amphibians, birds and plants) (Sparks, Jeffree & Jeffree, 2000). A similar pattern was reported for Dutch Odonata (Dingemanse & Kalkman, 2008) and the German population of Gomphus vulgatissimus (Odonata) (Richter et al., 2008). Although Odonata is the best-studied group in terms of temperature-related emergence, there are similar findings for EPT and Diptera (Nebeker, 1971; Čmrlec et al., 2013; Dickson & Walker, 2015; Chacón, Segnini & Briceño, 2016; Cheney et al., 2019; Nyquist, Vondracek & Ferrington, 2020).

(d) Voltinism

In response to different temperature conditions, aquatic macroinvertebrates can show phenotypic plasticity that can speed up or slow down the development of adaptive strategies (Pritchard et al., 1996). For example, some stoneflies (e.g. Nemoura cinerea) are able to shift from a univoltine to a semivoltine life cycle when the eggs are exposed to a low temperature (10 °C) (Brittain, 1974). Under increasing temperature, some stoneflies (e.g. Leuctra nigra) and mayflies (e.g. Ephemerella danica) shift from semivoltine to univoltine, showing highly plastic phenology (Elliott, 1987; Everall et al., 2015). Many species have a synchronous life cycle coordinated by water temperature (Humpesch, 1980a; Sweeney & Vannote, 1984). For example, Beatis alpinus (Ephemeroptera) has a trivoltine/bivoltine or univoltine life cycle depending on altitude (Humpesh, 1979; Erba, Melissano & Buffagni, 2003) although Bottová, Derka & Svitok (2013b) found asynchronous life cycles in specimens maintained at constant temperature conditions. By contrast, a recent study carried out in a karstic spring of the Western Carpathians (Beracko & Revajová, 2019) investigating more than 40 benthic species did not support the proposal that constant water temperature promotes asynchronous life cycles and reported different phenological responses. Some Plecoptera species (e.g. Protonemura auberti and Leuctra albida) had an additional winter cohort instead of entering diapause, other species from various orders [e.g. Gammarus fossarum, (Amphipodae) Rhyacophila tristis (Trichoptera) and Protonemura austriaca (Plecoptera)] showed an unchanged or even a longer nymphal development while others maintained fixed voltinism [Ephemerella mucronate (Ephemeroptera), Isoperla sudetica (Plecoptera)]. Odonata species tend to show a clear voltinism gradient along latitude and temperature clines: voltinism decreases from Southern to Northern Europe ranging from one generation every 1–2 years in the south to one generation every 5 years in the north (Söndgerath, Rummland & Suhling, 2012), indicating that higher temperatures correlated with increasing voltinism (Braune et al., 2008). Univoltine species are likely to be negatively impacted by increases in temperature extremes whereas multivoltine species are likely to be advantaged (Rivers-Moore, Dallas & Ross-Gillespie, 2012).

(e) Colour

McCafferty & Pereira (1984) noted that in larvae of Hexagenia limbata and Stenacron interpunctatum (Ephemeroptera) the colour of the body and wings, as well as the spotting pattern, depended on the temperature regime of the water in which larvae developed. The colour of the compound eyes and legs was independent of temperature. Abbott (2013) conducted experiments on female Ischnura elegans (Odonata), a three-colour polymorphic species, to investigate whether colour polymorphism was correlated with thermal performance. He found that life-history traits varied across colour morphs, suggesting that thermal performance was more associated with morphospecies rather than local thermal adaptation.

(a) Fecundity and hatching success

In most invertebrates, fecundity is directly proportional to female body size (Rempel & Carter, 1987). High temperatures reduce the capacity of organisms to exploit resources from the ecosystem (Marten, 1990), leading to a decrease in the energy available for egg production, and thus to lower fecundity (Sweeney & Vannote, 1978; Rempel & Carter, 1987; Rosillon, 1988; Pritchard et al., 1996; Dallas & Ross-Gillespie, 2015). Increasing temperature also leads to faster hatching and lower egg survival (Bouton et al., 2011), partly due to a greater risk of infection by fungi and bacteria (Harvell et al., 2002; Marcogliese, 2016). In response to stressful temperature conditions, aquatic insects face a trade-off between growth and reproduction. According to the TEH, fecundity varies with altitude and latitude, declining as temperatures move away from the optimum. For example, Van Doorslaer & Stoks (2005b), studying two congeneric damselflies Coenagrion hastulatum and C. puella (Odonata) widespread in northern and central Europe respectively, identified the evolution of latitudinal compensation to low temperature in line with predictions of the TEH, but only at the embryogenic stage and not at the larval stage. This observation stresses the importance of assessing thermal responses at different life-history stages. Each species has a specific thermal threshold for egg hatching and development, which will affect both population size and species distribution (Elliott, 1988; Lambret, Hilaire & Stoks, 2017). Optimal temperatures promote the largest broods and eggs, higher hatching success and greater reproductive success (Bovill, Downes & Lancaster, 2015), while higher temperatures have significantly negative effects on egg survival and overall fitness (Starr & McIntyre, 2020). Low temperatures prolong dormancy and delay hatching (Danks, 2002; Lencioni, 2004).

(b) Larval recruitment

With increasing temperature the nymph recruitment increases while growth rates increase exponentially (Wright, Mattice & Beauchamp, 1982; Humpesch, 1980a,b; Strange, 1985; Marten, 1990; Corkum & Hanes, 1992; McCauley et al., 2018; Chavez et al., 2015; Lencioni et al., 2013; Ingram, 1976; Van Doorslaer & Stoks, 2005b). Some studies show that survival rates differ between the sexes, suggesting an interaction between sex and temperature. Other factors may also play important roles in larval recruitment, such as the ability to reproduce parthenogenetically (Wright et al., 1982).

(a) Migration and drift

Temperature varies seasonally and within a waterbody, especially in rivers and deep lakes. As ectotherms, aquatic macroinvertebrates must maintain their metabolic and physiological processes at levels high enough to survive and reproduce (Vannote & Sweeney, 1980). Aquatic species can track suitable thermal niches by dispersal through drift or active swimming, with drift being the most common dispersal type in rivers (Waters, 1965). There are two types of drift: catastrophic (mainly due to disruptive floods or hydropeaking as well as drought, high temperature and pollution) and behavioural, occurring when macroinvertebrates voluntarily leave their substrate in response to stress conditions that include temperature, predation or resource scarcity (Muller, 1954; Waters, 1965; Wiley & Kohler, 1980). A variety of studies have recorded distinct drift in benthic invertebrates exposed to thermal and discharge waves caused by sudden water release from hydropower plants, with catastrophic drift due to hydropeaking and behavioural drift caused by thermopeaking. Chironomidae, Simuliidae (Diptera) and Baetidae (Ephemeroptera) resulted the most abundant drifting taxa (Bruno et al., 2012; Carolli et al., 2012; Schülting, Feld & Graf, 2016). Temperature can influence drift, for example, Wojtalik & Waters (1970) observed that increased temperature resulted in nocturnal drift in Baetis vagans (Ephemeroptera) but not Gammarus pseudolimnaeus (Amphipoda), while at constant temperature conditions neither species drifted. Scherr, Wooster & Rao (2010) reported greater drift in the mayfly E. alberta at a high temperature (28 °C). High water temperatures can also promote emergence events as shown by Trottier (1973b) for the climbing speed of Anax junius (Odonata).

(b) Predation

High temperatures may disproportionately influence organisms at higher trophic levels which are more strongly affected by alterations of energy fluxes across the food web (Vasseur & McCann, 2005; Gilman et al., 2010). Thus, theoretically predators may be more vulnerable to increasing temperatures than their prey. However, McKie & Pearson (2006) showed that predation of Australian chironomids by Australopelopia prionoptera (Diptera) was not influenced by different temperatures (12, 18 and 26 °C). This suggests that in macroinvertebrates characterized by broad physiological tolerances the predator–prey relationship may be unaffected by temperature. Thermal shocks did not alter predation of Ephemeroptera (Sherberger et al., 1977), with mortality of individuals of Isonychia sp. at 33 °C for 30 min due to the presence of a predatory fish (Cottus carolinae) similar to that for the control group (14 °C). By contrast, Smolinský & Gvoždík (2014) found that during daily temperature extremes predation rates on newt larvae diminished, despite increased predator (dragonfly larvae) movement. Predation pressure may be lower at high seasonal temperatures or where fauna have a broad thermal tolerance range (Hildrew & Giller, 1994; Reice, 1994; McKie & Pearson, 2006).

In boreal freshwater systems, predator–prey interactions are particularly sensitive to thermal changes due to the simpler trophic web and to the presence of stenothermal species (Thompson, 1978; Moore & Townsend, 1998). For example, Kishi et al. (2005), studying the trophic chain of a boreal stream composed by a predatory fish (Salvelinus malma), an herbivorous caddisfly (Glossosoma) and periphyton, observed that thermal habitat alteration can change food-web structure via combinations of direct and indirect trophic interactions. Indeed, at high temperature (21 °C) Glossosoma larvae were promoted due to both the lower salmonid predation and the greater availability of periphyton. On the other hand, high temperature can reduce the ability to build cases in Trichoptera larvae due to the high energetic cost (Mondy et al., 2011) leaving them more exposed to predators. For example, Rogowski & Stewart (2016) observed decreased retreat building and higher mortality in Leptonema sp. (Hydropsichae, Trichoptera) at 22 °C.

(c) Feeding

A variety of studies have shown that key consumers in freshwater ecosystems change their feeding behaviour depending on temperature conditions. Metabolism is enhanced by increased temperatures, and this generates the requirement for a greater energy intake (Vannote & Sweeney, 1980). Greater feeding efficiency can be achieved either by targeting resources that are more easily assimilated or by seeking higher quality food. For example, in a geothermal stream network characterized by a large temperature range (5–23 °C), at warmer temperatures (20 °C) the snail Radix balthica (Gastropoda) shifts to a more specialized diet while the black fly Simulium aureoum (Diptera) switches from active collection of sessile diatoms to passive filter-feeding on motile diatoms. On the contrary, the chironomid Eukiefferiella minor (Diptera) becomes more generalist at warmer temperatures (Gordon et al., 2018). Diet and temperature may interact: food quality influences both growth rates and body size in shredders, scrapers and grazers of EPT and Diptera (Fuller & Fry, 1991; Giberson & Rosenberg, 1992a; Rosillon, 1988; Storey, 1987; Sweeney et al., 1986a,b). The interactions among food, temperature, developmental time and fecundity suggest that the TEH should be adapted to include both food quality (Rosillon, 1988) and availability.

(a) Community richness, taxonomic composition, and density

Macroinvertebrate community composition varies with temperature at both micro- and macro-geographic spatial scales; temperature affects the selection and maintenance of different species in water bodies (Vannote & Sweeney, 1980). A clear trend of increasing richness occurs with increasing temperature (along both altitudinal and latitudinal gradients); Castella et al. (2001) showed this pattern for glacier-fed streams across Europe. In the Po catchment (Italy), a clear altitudinal pattern in macroinvertebrate community composition was identified. Assemblages inhabiting high-altitude sites were characterized mostly by Plecoptera, Trichoptera, Coleoptera, and Diptera, whereas macroinvertebrate communities inhabiting lowland sites included mostly non-insect orders such as Clitellata, Gastropoda and Bivalvia. At the temporal scale, annual thermal variability promotes seasonal dissimilarity in macroinvertebrate assemblages (Vannote & Sweeney, 1980; Ward & Stanford, 1982; Arai et al., 2015) while inter-annual water temperature variations affect community composition. For example, Fornaroli et al. (2020) found that in Northern Italy inter-annual flow and temperature regime variations affected community richness with higher alpha diversity in warmer years but lower EPT taxa abundance. Extreme temperatures can cause decreases in both species numbers and density (Vannote & Sweeney, 1980; Voelz, Poff & Ward, 1994; Glazier, 2012; Arai et al., 2015) inducing shifts in community composition (Arim, Bozinovic & Marquet, 2007a).

Climate warming may promote eurythermal and generalist species with a consequent expansion of these less-specialized macroinvertebrate communities (Domish, Jahnig & Haase, 2011). This is likely to affect springs and small streams more than large rivers (Haidekker & Hering, 2008). Increasing temperatures led to the upstream spread of eurythermal and rheophilic species from subalpine levels, causing homogenization of macroinvertebrate communities, especially for EPT (Timoner et al., 2020). Increasing temperatures thus result in movement of species upstream and an increase in invasive species (Jourdan et al., 2018).

(b) Distribution

Temperature affects species distributions both positively and negatively: some can be expected to increase their distribution in response to climate warming (‘winning’ species as defined by Domish et al., 2011) while others become more restricted (‘loser’ species) (Sweeney & Vannote, 1978; Arai et al., 2015; Jacobsen et al., 2014; Besacier et al., 2019). Adaptation to a specific temperature range restricts species zonation to particular ecological niches along the temperature gradient (Sweeney & Vannote, 1978; Arai et al., 2015).

In recent years, attempts have been made to predict the distribution of freshwater communities at national, continental, and global spatial scales by applying predicted air temperatures (sometimes together with precipitation predictions) to models using long-term aquatic monitoring data sets. Several indicators have been developed to assess the sensitivity and vulnerability of aquatic communities to climate change (Hering et al., 2009; Conti et al., 2014; Sandin et al., 2014; Mustonen et al., 2018). Among the environmental drivers that regulate taxa distribution, climatic drivers contribute substantially at a broad geographic scale but are insufficient to explain local community dynamics at catchment scale, where other variables such as habitat, geomorphology and land-use features play an important filtering role (Poff et al., 2010; Feld & Hering, 2007). Other factors influencing the macroinvertebrate community at the local scale include species thermal limits, adaptation capacity, drift propensity, resource use and interspecific interactions (Glazier, 2012).

(c) Food-chain length

The amount of energy available in an ecosystem influences its food-web structure (Odum, 1968) and sets an upper limit to the length of the food chain (Lindeman, 1942; Hutchinson & MacArthur, 1959). Two theories regarding the relationship between temperature and food-chain length have been proposed. According to the metabolic theory (Arim, Marquet & Jaksic, 2007b), body size and food-chain length are inversely correlated with environmental temperature: at high temperatures, metabolic demand increases so that lower levels of the trophic web consume more energy and energy flow to upper trophic levels is reduced. According to the thermal tolerance hypothesis (Brock, 1985), biochemical similarity among the organisms that constitute a specific ecological community will lead to similarity in their thermal tolerance. Therefore, below the upper thermal limit, the food-chain length is independent of temperature while close to the limit it is considerably reduced (Brock, 1985). Glazier (2012) showed that in springs (characterized by constant temperature and flow regimes), food-chain length decreases with increasing temperature, but the decline is not linear, broadly in support of the thermal tolerance hypothesis.

(d) Community structure and trophic role

Temperature changes in freshwater ecosystems potentially alter macroinvertebrate community structure, modifying trophic interactions within the aquatic food web. Taxon-specific trait information (www.freshwaterecology.info database) can be used to investigate the mechanisms through which temperature affects community structure. For example, Jourdan et al. (2018) used long-term data (10–30 years) on macroinvertebrates from several streams in the UK, Germany, Finland, and Latvia to show that the composition of functional feeding groups was strongly impacted by warming temperatures and more intense precipitation events. In particular, grazers and scrapers appeared especially vulnerable at higher temperatures, as predicted by Pyne & Poff (2017) for the macroinvertebrate communities of the western USA. Trait information related to feeding, substrate and habitat specializations proved critical to understanding the responses of macroinvertebrates to temperature changes in Sweden (Sandin et al., 2014).

(e) Secondary production

The metabolic theory asserts that secondary production will be relatively temperature invariant, recognizing resource supply as the sole driver, and this has been validated by studies carried out in Iceland's geothermal streams (Nelson et al., 2017; Junker et al., 2020). However, basal resource dynamics depend on many variables including light, nutrient availability and temperature so apparent relationships between temperature and production can be explained by the positive effects of temperature on resource supply (Junker et al., 2020). Inland waters are heterotrophic ecosystems in which secondary production is strongly supported by allochthonous organic matter rather than by internal primary production. Climate change-related mechanisms may increase the inputs of allochthonous dissolved organic carbon (Pagano, Bida & Kenny, 2014; Porcal et al., 2009). In addition, primary production is strongly dependent on water temperature regime (Demars et al., 2011; Padfield et al., 2017) making a general trend of increased resource availability likely with global warming. However, in addition to an increased supply of both autotrophic and heterotrophic food resources at higher temperatures, decay rates of organic matter will also be accelerated, thus decreasing its availability to consumers (Rempel & Carter, 1986). In aquatic environments characterized by constant temperatures (springs), stonefly secondary production was found to be very high, possibly due to reliable resource supply due to stable temperature (Bottová et al., 2013a) and flow regimes (Zimmerman & Wissing, 1978).

(a) Genetic diversity

Temperature can also have effects at an evolutionary scale, causing thermal divergence in populations of the same species, promoting genetic diversity, and leading to temporal segregation. For example, temperate and tropical populations of the chironomids Echinoclaudius martini and Australopelopia pronioptera (Diptera) have diverged in developmental time; moreover, different populations of E. martini have diverged in oocyte production (greater in the temperate population) as well as body size, suggesting that temperature could facilitate population differentiation (McKie et al., 2004). Indeed, tropical species often have higher thermal limits than congeneric temperate species (Chapman, 2013). For EPT species of Ecuadorian Andean and USA Rocky Mountains, Polato et al. (2018) showed that tropical and temperate mountain stream insects have diverged in thermal tolerance and dispersal capacity due to different seasonal temperature variations. Tropical species had narrower thermal breadths, less gene flow, higher population divergence, higher cryptic diversity, and higher speciation rates, rendering them especially vulnerable to rapid changes in thermal environments. Johansson, Quintela & Laurila (2016) showed that the genetic population structure of the Icelandic freshwater gastropod Radix balthica living in contrasting geothermal habitats was influenced by both geographic distance and water temperature. Genetic variation decreased with increasing temperature, suggesting that natural selection had led to reduced genetic diversity in warm geothermal springs due to higher thermal specialization. Herzog & Hadrys (2017) used a 20-year data set on the genetic diversity of a population of Orthetrum coerulescens (Odonata) located in Crau (France) to identify a dramatic decline in genetic diversity caused by increased water temperatures mediated by the destruction of bank vegetation. At high altitudes, global warming is leading to the loss of glaciers, which promotes fragmentation, limits gene flow and leads to loss of genetic variation among populations of high-altitude freshwater invertebrates. For example, populations of Lednia tumana (Plecoptera) in the Glacier National Park (Montana, USA) showed reduced gene and nucleotide diversity and increased genetic isolation in response to glacier retreat during 1997–2010 (Jordan et al., 2016). Genetic loss has been predicted in some European meltwater species of stoneflies, caddisflies and mayflies from predictions of mitochondrial DNA variability under different climate change scenarios. Intraspecific (cryptic) genetic loss is a significant concern and should be included when estimating the effects of global warming on biodiversity loss (Bálint et al., 2011). Temperature and genetic variability may interact affecting mortality: dramatic effects of high temperature (20 °C) were observed in a population of Chironomus riparius (Diptera) with poor genetic variability (Vogt et al., 2007).