Combined threats of climate change and contaminant exposure through the lens of bioenergetics

Organisms face energetic challenges of climate change in combination with suites of natural and anthropogenic stressors. In particular, chemical contaminant exposure has neurotoxic, endocrine‐disrupting, and behavioral effects which may additively or interactively combine with challenges associated with climate change. We used a literature review across animal taxa and contaminant classes, but focused on Arctic endotherms and contaminants important in Arctic ecosystems, to demonstrate potential for interactive effects across five bioenergetic domains: (1) energy supply, (2) energy demand, (3) energy storage, (4) energy allocation tradeoffs, and (5) energy management strategies; and involving four climate change‐sensitive environmental stressors: changes in resource availability, temperature, predation risk, and parasitism. Identified examples included relatively equal numbers of synergistic and antagonistic interactions. Synergies are often suggested to be particularly problematic, since they magnify biological effects. However, we emphasize that antagonistic effects on bioenergetic traits can be equally problematic, since they can reflect dampening of beneficial responses and result in negative synergistic effects on fitness. Our review also highlights that empirical demonstrations remain limited, especially in endotherms. Elucidating the nature of climate change‐by‐contaminant interactive effects on bioenergetic traits will build toward determining overall outcomes for energy balance and fitness. Progressing to determine critical species, life stages, and target areas in which transformative effects arise will aid in forecasting broad‐scale bioenergetic outcomes under global change scenarios.

We reviewed the literature to assess evidence for interactive effects involving four climate change-sensitive environmental variables: resource availability, temperature, predation risk, and exposure and susceptibility to parasitism. Although other shifts in environmental conditions are relevant (e.g., ocean acidification, drought), the abiotic and biotic environmental factors that we focus on are likely to exert effects across ecosystem types. We reviewed evidence for interactive effects on multiple traits within domains 1-3 (Supplementary Material; Table S1 for details), and classified interactive effects via the framework explained in Figure 2 (Supplementary Material for details). For domains 4 and 5 our discussion remains more hypothetical, and focuses on generating testable hypotheses.
We were interested in understanding contaminant-byenvironment interactive effects on Arctic endotherms. Climate change is likely to disrupt bioenergetics in Arctic ecosystems, due to rapid warming (IPCC, 2021), extensive trophic reorganization (Frainer et al., 2017), organismal adaptations to cope with cold stress (Gilg et al., 2012;Oswald & Arnold, 2012), and alterations in biochemical cycling that modify contaminant concentrations in biota (Borgå et al., 2022;Macdonald et al., 2005). For instance, melting permafrost, and increased precipitation and terrestrial runoff may elevate exposure to POPs in aquatic ecosystems (Borgå et al., 2022).
Although isolated from contamination sources, the Arctic acts as a sink for volatile contaminants, which persist in cold environments and increase in concentration at higher trophic positions through biomagnification (AMAP, 2021a(AMAP, , 2021b. Although levels of legacy POPs (e.g., PCBs) show decreasing trends in the Arctic, these trends are slowing or leveling off, and other POPs (e.g., PBDEs) are often increasing, as are chemicals of emerging Arctic concern (CEAC; e.g., per-and polyfluoroalkyl substances [PFASs]; flame retardants).
Another important contaminant in the Arctic, mercury (Hg), displays decreasing atmospheric trends, but both increasing and decreasing trends have been observed in biota (AMAP, 2021a;Chastel et al., 2022). Thus, cumulative contaminant impacts are likely to increase in the Arctic in coming decades (AMAP, 2021b). We used the Arctic Monitoring and Assessment Programme report, listing contaminants of concern in the Arctic (AMAP, 2021a(AMAP, , 2021b, to guide our literature search (Supplementary Material for details). Given the few studies assessing interactive effects on Arctic organisms (Table 1), we expanded our review to encompass all animal taxa, while frequently incorporating discussions of how interactive effects might affect Arctic endotherms. Table 1 summarizes studies on contaminant-by-environment interactive effects across bioenergetic domains. This table highlights a deficit of research involving endotherms in general, and Arctic endotherms in particular, and a bias toward testing interactive effects with temperature and metals. Results suggest relative balance between synergistic and antagonistic effects and more unpredictable interactions than simple additive or nonsignificant effects. However, our emphasis was not on comparing the frequency of interactive effects, as we did not formally calculate effect sizes. Below, we discuss interactive effects within each bioenergetic domain and underlying mechanisms.

Contaminant-by-temperature interactive effects on foraging
Empirical demonstrations of contaminant-by-temperature interactive effects on foraging remain limited, with only six studies, and none on endotherms (Table 1; Table S2). However, many contaminants interfere with the thyroid hormone system, which is central to thermoregulation (Box 1), and high temperatures may increase chemical toxicity (Hooper et al., 2013). Thus, contaminant exposure could amplify the need to forage in reduced spatial ranges or temporal windows given thermal challenge (Rabaiotti & Woodroffe, 2019), resulting in negative synergy and magnifying effects on energy intake. Indeed, Hg exposure was associated with more time spent inactive in the shade in great egrets (Ardea alba) (Bouton et al., 1999).
Combined effects of thermal stress and contamination might be especially pronounced in species occupying niches close to the upper critical limit of thermal neutrality, in regions that are warming rapidly (e.g., the Arctic; AMAP, 2021a, 2021b), and in high latitude or elevation species poorly adapted to dissipate heat (Blix, 2016).
Contaminants affect hormonal axes via diverse mechanisms, including interference with receptors, carrier proteins, hormone release, or negative feedback. Metal and POP exposure can both elevate GCs and suppress HPA sensitivity (Monclús et al., 2018;Tartu et al., 2014). OC and PCB exposure may decrease THs via direct action on the thyroid, interfering with transport proteins, or disrupting deiodinases, such as those converting T 4 to T 3 (Leemans et al., 2019;Wada et al., 2009). Conversely, PFASs have been positively associated with THs (Ask et al., 2021;Melnes et al., 2017). In black-legged kittiwakes (Rissa tridactyla), PFASs positively correlated with T 4 in males and T 3 in females, with the mechanism involving competition with THs for binding proteins (e.g., albumin). Increased THs could elevate BMR, interfere with thermal acclimation, and reduce tolerance to cold or heat (Ask et al., 2021).

Contaminant-by-resource availability interactive effects on foraging
In endotherms, six studies tested for contaminant-by-resource availability interactive effects on foraging behavior (Table 1; Table S2).
This dearth of research is regrettable, given extensive potential for these effects. Climate change is affecting resource availability and distribution, causing shifts in trophic dynamics (Lurgi et al., 2012;Parmesan, 2006). For example, in the Arctic, climate change is reducing availability of sea ice-associated prey, such as F I G U R E 1 The effects of climate change-associated stressors and contaminants on bioenergetic traits may be additive, with individuals differing in contaminant load displaying parallel reaction norms across environmental gradients. Alternatively, effects may be interactive. First, contamination may modify the slope of reaction norms with respect to climate change-associated stressors, as proposed by toxicantinduced climate sensitivity (Hooper et al., 2013). For example, as temperature increases endotherms may decrease metabolic rate to conserve energy, but this response may be dampened if detoxification costs elevate metabolic rate (a). Here, since metabolic rate changes less steeply with environmental conditions within more contaminated individuals, antagonistic (<additive) effects result. Note that metabolic rate may be modified when temperature either increases or decreases away from an optimal temperature, or temperature range (thermal neutral zone), and contamination might interfere with such modulation. Additionally, climatic stress may alter effects of contamination on bioenergetic traits, as proposed by climate-induced toxicant sensitivity (Hooper et al., 2013) (b). Here, metabolic rate increases more steeply with contaminant exposure at higher temperatures, which could reflect heighten toxicity and detoxification costs, resulting in synergistic (>additive) effects. Note that climate-driven and contaminant-driven processes in reality operate simultaneously, as visualized by reaction norms in three-dimensional space (c). We consider the potential for such interactive effects within five bioenergetic domains. These processes may affect fitness outcomes, especially in challenging environments (bottom).
lipid-rich copepods and fish (e.g., Arctic cod; Boreogadus saida) utilized by seabirds (Amélineau et al., 2019;Divoky et al., 2015;Gaston et al., 2009), and seals preyed on by polar bears (Ursus maritimus) (Stirling & Derocher, 2012). Evolving precipitation and wind regimes also pose challenges. For example, as wind speed increases, flight costs increase and seabirds may deliver less energy to nestlings (Elliott et al., 2014), and freezing rain in the Arctic elevates energetic costs for terrestrial animals seeking plants or small rodents (Gilg et al., 2012). Given decreased resource availability or quality, organisms may increase foraging effort and DEE to stabilize energy intake (Welcker et al., 2009), or decrease activity and energy intake to conserve energy (Stirling & Derocher, 2012). Box 2 discusses mechanisms by which contaminant exposure might disrupt such responses, potentiating interactive effects.
Examples of contaminant-by-resource availability interactive effects in endotherms include a study on Arctic thick-billed murres (Uria lomvia) exposed to Hg, which displayed altered T 3 only during years with suboptimal conditions associated with advanced sea ice breakup. Elevated T 3 was associated with reduced diving, which could result in negative synergistic effects on energy intake . As T 3 is positively associated with MR in murres (Elliott et al., 2014), this effect could reflect a tradeoff between increased oxygen consumption and energy-intensive behavior.
Contamination may dampen increases in foraging effort, undermining ability to maintain energy intake given resource shortage, or increase intake to exploit abundant resources, resulting in positive antagony. For example, Arctic organisms contaminated by pollutants such as Hg, a contaminant of concern in the Arctic, may be unable to employ energetically intensive behaviors required to track spatial changes in resources associated with the cryosphere (Meredith, 2019;Box 3). Contamination could also dampen functional responses that enable organisms to capitalize on transient resource peaks, such as algae blooms triggered by melting sea ice (Eamer, 2013;Meredith, 2019). Although evidence from the Arctic remains lacking, Hg contamination dampened predator functional responses of great egret (Ardea albus) to increased prey detectability (Bouton et al., 1999).
Contamination could also affect animals' ability to phenologically track shifting resource peaks, for instance by deeming individuals unable to advance breeding under warmer conditions (Charmantier et al., 2008). Evidence exists of contaminated birds delaying egg laying date (Bustnes et al., 2007;Helberg et al., 2005), but interactive effects with environmental conditions remain largely unexplored.
One study reported a nonintuitive result. Common eiders (Somateria mollisma) laid later in warmer years, with this effect less pronounced in Hg-exposed females . The mechanism underlying this effect was postulated to involve disruption of the HPA axis by Hg .
Furthermore, contaminants could limit downregulation of foraging in response to resource shortage, which can conserve energy (Cuyler & Øritsland, 1993;Stirling & Derocher, 2012). In Arctic F I G U R E 2 Framework for classifying interactive effects between stressors when effects are in the (a) same or (b) opposite directions. Single stressor effects (E1 and E2) appear below the horizonal line, and cumulative effects classified as additive (Add), antagony (Ant), synergy (Syn), or dominant (Dom) above. Given effects in the same direction and an additive null model [vertical line in (a)], classifying effects as antagony or synergy (less or more in absolute value than the additive prediction) is straightforward. Given effects in opposing directions, classification becomes contentious. We classified interactive effects as antagony if less in absolute magnitude than the single stressor effect in the same direction [outer vertical lines in (b)]. The effect is called positive antagony if more negative than the additive prediction (positive effect antagonized), and negative antagony if more positive than the additive prediction (negative effect antagonized). Some have instead labeled effects greater in magnitude than the additive prediction and in the same direction synergy [negative antagony bar in (b)]. We classified effects as synergy only given an absolute magnitude greater than the single stressor effect in the same direction. The effect was called positive synergy if greater than a positive single stressor effect, and negative synergy if less than a negative single stressor effect. Dominant effects are of the same magnitude as one of the single stressor effects, and can also be viewed as a form of antagony (other stressor effect is antagonized). Framework derived from Piggott et al. (2015).  Note: Some studies involved multiple types of contaminants, environmental stressors and classes of interactive effects. Numbers for interactive effect classes refer to the number of studies reporting such an effect, but some studies may have reported the same type of effect for multiple variables. Some studies did not statistically test the interactive effect (e.g., because separate experiments were conducted under two conditions or only a one-way ANOVA was used). Of these studies, graphical depictions of results mostly allowed verification that the interactive effect was significantly different than the additive prediction. However, in a few cases this was difficult. We note these cases in parentheses when reporting classes of interactive effects. See Supplementary Material for the methodology of the literature review.
animals, such as the Arctic fox (Vulpes lagopus), decreased activity and MR serve as adaptations to winter food scarcity (Prestrud, 1991).
Sustained foraging effort despite resource shortage could result in a negative antagonistic effect on foraging activity, but a negative synergistic effect on energy balance. Such an effect remains undocumented, but could arise due to stimulatory effects of some contaminants. For instance, Hg-exposed golden shiners (Notemigonus crysoleucas) displayed hyperactivity, increasing energy expenditure (Webber & Haines, 2003).
Finally, climate change-associated declines in resource availability or quality could elevate contaminant exposure and/or toxicological effects, inducing interactive effects. Contaminant concentrations often rise with declines in body mass and adipose or muscle tissue (Bustnes et al., 2010;Peterson et al., 2018). In northern elephant seals (Mirounga angustirostris), blood Hg increased by 103% during the breeding season fast in association with increased blood volume and muscle tissue (Peterson et al., 2018), and lipophilic PCBs were mobilized from blubber to blood during the postweaning fast ( (Lovvorn & Gillingham, 1996).

Contaminant-by-predation risk interactive effects on foraging
Evidence also exists for contaminant-by-predation pressure interactive effects on foraging behavior, although only one study focused on endotherms (Table 1; Table S2). Heightened risk sensitivity in BOX 1 Endocrine pathways through which contaminants may disrupt metabolism.
Contaminants can interfere with physiological mechanisms, including the hypothalamus-pituitary-adrenal (HPA) and hypothalamuspituitary-thyroidal (HPT) axes, which mediate adjustment of metabolism to conditions. Steroid glucocorticoids (GCs) released via the HPA axis support glucose metabolism, mediate vertebrate stress responses (McEwen & Wingfield, 2003;Sapolsky et al., 2000), and aid in temperature-dependent regulation of metabolic rate (MR) (Jimeno et al., 2018). Corticotropin-releasing hormone (CRH) is involved in primary thermogenesis through its effect on thyroid-stimulating hormone (TSH) (De Groef et al., 2006). The HPT axis mediates oxygen consumption, ATP use, and thermogenesis through release of thyroid hormones (THs), thyroxine (T 4 ), and 3,5,3-triiodothyronine (T 3 ) (Leemans et al., 2019;Ruuskanen et al., 2021;Yen, 2001). Red arrows indicate possible targets of contaminant effects. Hg-exposed zebra finch induced larger reductions in foraging under predation risk, leading to a negative synergistic effect on energy intake . Alternatively, as suggested for zinc (Zn)-exposed damselfly larvae (Ischnura elegans), contaminated individuals with heightened energetic costs might be unable to tradeoff foraging against vigilance, canalizing reaction norms and generating negative antagony . A similar outcome is predicted if contaminants interfere with predator cue detection (Ehrsam et al., 2016). Conversely, positive antagony might arise if energetically stressed contaminated individuals increase foraging effort only in the absence of predation risk, as in Cd-exposed mayfly nymphs (Baetis tricaudatus) . Importantly, climate change can either intensify or ameliorate predation risk. For instance, sea ice loss reduces predation on ground-nesting seabirds by denying Arctic foxes access to island colonies (Chaulk et al., 2007;Eamer, 2013), but increases predation from polar bears, which shift from marine to terrestrial resources earlier in the season (Iverson et al., 2014). Actions of contaminants on risk sensitivity, or ability to balance tradeoffs, may mediate Arctic animals' responses to shifting predation pressures, but empirical examples remain lacking.

| Assimilation and metabolic efficiency
Studies documenting contaminant-by-environment interactive effects on assimilation (AE) and metabolic (ME) efficiency were exclusively on ectotherms, and primarily involved interactions with temperature (

BOX 2 Mechanisms of contaminant-linked effects on upregulation of foraging effort.
Neurotoxic effects can compromise locomotion, reducing foraging efficiency (Ecke et al., 2017;. Contaminant exposure can also affect foraging motivation through actions on appetite-inducing or suppressing neuropeptides, and interfere with olfactory detection of prey (Groh et al., 2015). In endotherms, reduced appetite and voluntary food intake at high temperatures, which adaptively reduces digestion-related heat production, could magnify these effects (Youngentob et al., 2021). Contaminants may also influence foraging effort by affecting metabolic parameters. For example, Hg can reduce metabolic scope (Clarke & Pörtner, 2010) and impair performance by suppressing hemoglobin production and oxygen carrying capacity, and binding with thiol-containing enzymes and carrier proteins (Gerson, Cristol, et al., 2019;Olsen et al., 2000;Seewagen, 2020). Contaminants' effects on metabolism and foraging could additionally be mediated through body condition. Arctic belugas (Delphinapterus leucas) in poor condition had lower oxygen stores, which may affect diving depth and aerobic dive limits (Choy et al., 2019). Finally, effects on cognition such as reduced spatial memory, inability to assimilate information, or compromised neurological processes could increase search costs Swaddle et al., 2017).
A few studies explored interactive effects of contaminants with predation risk and parasitism. One reported a significant interaction (Table 1; Table S3), wherein two POP pesticides, atrazine and endosulfan, combined with predation risk to have a negative synergistic and negative antagonistic effect, respectively, on the AE of damselfly larvae (Coenagrion puella) . Resource availability could also interact with contamination to affect AE and ME, although no empirical demonstrations exist. Climate changerelated shifts in prey use or availability may reduce AE, exacerbating contaminant effects and reduce trophic energy transfer. Thick-billed BOX 3 Hypothetical inhibition of foraging plasticity by Hg exposure.
Little auks (Alle alle) depend on sea ice-associated copepods, which are being replaced by less lipid-rich species due to climate change-linked sea ice declines. Birds can travel farther and use novel foraging areas along melting glaciers to stabilize energy intake (Grémillet et al., 2015;Jakubas et al., 2013). However, Hg in copepods in Greenland have been increasing with climate change, driving increases in Hg in little auks (Fort et al., 2016). Neurotoxic effects of Hg may dampen reaction norms, resulting in a positive antagonistic interaction between warming and Hg contamination on foraging behavior. In particular, increasing flying and diving, which have terfere with the ability to adjust MR to temperature, generating interactive effects. For example, mice exposed to arsenic (As) dis- Furthermore, contaminants, such as methylmercury, can limit mitochondrial oxidation and ATP production, which could suppress MR, limiting capacity for upregulation (Gonzalez et al., 2005).

Other contaminant-by-environment interactive effects on metabolism
Animals often adjust MR in response to resource levels, and contaminant exposure may interfere with these responses, as indicated by some research, limited to ectotherms (Table S4). For example, lake chubsuckers (Erimyzon sucetta) exposed to metals reduced standard metabolic rate (SMR) less when subjected to food limitation than unexposed fish, resulting in negative antagony . SMR could be elevated by detoxification costs (Verreault et al., 2007), limiting downregulation and interfering with an important energy conservation mechanism. In contrast, food limitation magnified suppressive effects of PCBs on lipid metabolism in Arctic char (Salvelinus alpinus), resulting in a negative synergistic effect that could compromise energy balance .
Changes in predation risk linked to climate change could also induce changes in metabolism that are sensitive to contaminant exposure, with possible responses including acute increases to support fight-flight responses, or decreases reflecting suppressed activity and increased refuge use . However, neither of two studies examining contaminant-by-predation risk effects on metabolism reported significant interactions (Burraco et al., 2013;.
Finally, in the only study testing contaminant-by-parasitism interactive effects on metabolism, parasitism elevated activity of the electron transport system in Cockle (Cerastoderma edule), but arsenic (As) exposure reduced this effect . This positive antagonistic effect could limit the ability to elevate energy expenditure to combat infection.  Table S4). In ectotherms, considerably more work has been done (Table 1; Table S5). For instance, exposure to copper (Cu) and the legacy POP dichlorodiphenyltrichloroethane (DDT) narrowed the thermal tolerance window in medaka fish (Oryzias melastigma), and exposure to high temperatures and the contaminants synergistically decreased heat shock defense mechanisms . In addition, carabid beetles (Platynus assimilis) exposed to an insecticide were unable to maintain body temperature and survive as temperature increased .  (Speakman, 2018). In lieu of effective heat dissipation, hyperthermia results, which can damage proteins, induce oxidative stress or be fatal (Rogers et al., 2021;Speakman & Król, 2010;Tapper et al., 2020). Heat dissipation capacity has strong effects during energy-intensive life-history stages, such as lactation (Speakman & Król, 2010;Zhao et al., 2020), or nestling provisioning (Tapper et al., 2020), and may limit DEE in Arctic animals adapted to retain heat (Blix, 2016), and large animals with low surface area to volume ratios (Rogers et al., 2021). Box 4 discusses physiological mechanisms that could underlie disruption of thermoregulatory responses by contaminants, generating interactive effects. Contaminant exposure may interfere with thermoregulation via endocrine disruption, with effects on the HPT axis particularly relevant, due to its role in mediating thermoregulation in vertebrates (Verreault et al., 2004(Verreault et al., , 2007Wada et al., 2009). Contaminants could also impact thermoregulation via effects on body composition (more lean vs. adipose tissue) and the integument (feather, fur quality Poor-quality feathers or fur could also increase thermal conductance, increasing need to elevate MR to maintain body temperature.

| Thermoregulation
In addition, contamination could interfere with microvasodilation or constriction responses that affect cutaneous evaporation, impair temperature regulation, which could interfere with responses to environmental conditions. For instance, hypothermia often occurs in rodents exposed to contaminants, and heat stress may exacerbate toxicity by overriding this response (Gordon et al., 1990;Holmstrup et al., 2010;Leon, 2008).

Other types of contaminant-by-environment interactive effects on thermoregulation
Contaminant exposure might also interact with environmental variables besides temperature to affect thermoregulation (Table 1;   Table S5). For example, a positive synergistic interaction occurred between radioactivity exposure and nutritional stress in barn swallows (Hirundo rustica) (Boratyński et al., 2021). Birds in poor condition upregulated body temperature more with radioactivity exposure, compromising energy balance and increasing risk of overheating.
Similarly, in damselfly larvae (Ischnura elegans), exposure to an OP (chlorpyrifos), a CEAC, reduced heat tolerance, but only when competition for resources was high (Op de Beeck et al., 2018). Predation risk and parasitism could also combine with contamination to affect thermoregulatory ability via effects on behavior and energy balance.
However, the two studies testing for such effects reported additive results (Table 1; Table S5).

Contaminant-by-environment interactive effects on behavioral thermoregulation
A limited number of studies on ectotherms have reported contaminant-by-environment interactive effects on behavioral thermoregulation (Table S5). Neurotoxic or endocrine-disrupting effects of contaminants can interfere with behavioral plasticity that facilitates thermoregulation, such as foraging during crepuscular periods to avoid thermal stress (Hetem et al., 2014). In addition, behavioral thermoregulation strategies often entail tradeoffs, with which contaminated individuals might be unable to cope. For instance, microhabitats selected for thermoregulatory advantages may have lower food availability or increased predation risk (Hetem et al., 2014).
Furthermore, the behavioral thermoregulatory strategy that proves adaptive may be altered when temperature and contaminant stress are experienced in combination, relative to in single stress scenarios.
For example, Mongolia racerunners (Eremias argus) exposed to the insecticide abamectin displayed higher body temperature preference as a detoxification mechanism. However, lizards jointly exposed to abamectin and heat stress simultaneously face elevated MR and repair costs linked to contaminant and heat exposure, and decreased locomotion and predatory performance caused by contamination.
Under this extreme stress, lizards shift to display a heat-avoidance reaction, resulting in a positive synergistic effect on thermal shelter use, which is a short-term means of conserving energy, but might ultimately compromise energy balance .

| Physical activity
Daily energy expenditure correlates with physical activity in freeliving animals and can predict fitness (Grémillet et al., 2018;Wilson et al., 2020), suggesting importance of contaminant-by-environment interactive effects on activity. Table S1 gives relevant examples for foraging activity. Examples pertaining to activity in other contexts mostly involved interactive effects of contaminant exposure with predation risk or temperature, and all but two were on ectotherms (Table 1; Table S6).

Contaminant-by-temperature interactive effects on activity
Eleven studies (one on endotherms) reported significant contaminant-by-temperature interactive effects on activity, with no one interaction type clearly dominating. For instance, female mice (Mus musculus) prenatally exposed to methylmercury showed inactivity during an open field test, but this effect was dampened in combination with thermal stress, resulting in negative antagony .

Contaminant-by-predation risk interactive effects on activity
Seven studies on ectotherms (no studies exist on endotherms) reported negative antagonistic interactions between contaminant exposure and predation risk (Table 1; Table S6). For example, metalexposed round goby (Neogobius melanostomus) downregulated activity less than unexposed individuals under predation risk

| Immune defense
The immune system is a complex, costly component of energy demand (Sheldon & Verhulst, 1996) which is vulnerable to contaminantby-environment interactive effects. Of four studies testing for such effects in endotherms (Table 1; Table S7) In ectotherms, studies have reported both synergistic and antagonistic interactions between contaminant exposure and other environmental stressors (Table 1; Table S7). For example, in black sea bream (Acanthopagrus schlegelii) positive effects of Zn on phagocytosis and lysosome activity were increased at higher temperature, which could prime organisms to cope with pathogen challenge, but magnify effects on energy demand . On the other hand, in Pond snails (Physa gyrina), a stimulatory effect of the herbicide atrazine on hemocyte numbers disappeared under predation pressure . This positive antagonistic interaction could reflect reduced energy intake under predation risk, deeming increased energy investment in response to contaminant exposure impossible or unbeneficial.

| Energy storage
The balance between energy supply and demand affects energy storage capacity, which is critical during particular life-history stages, such as prior to migration or hibernation (Guglielmo, 2018;Weitten et al., 2018), before breeding in capital breeders (Gauthier et al., 2003), and in animals preparing for winter resource scarcity Two studies tested for contaminant-by-environment interactive effects on variables related to energy storage in endotherms, both involving metal contamination in birds (Table 1; Table S8). First, there was no interactive effect between metal contamination and weather conditions, including temperature, on body condition of tree swallows (Beck et al., 2015). Second, Hg-exposed zebra finches lost more mass than controls only under predation risk, resulting in a negative synergistic interaction that was mediated by increased risk sensitivity . Given climate change-linked elevation in predation risk, such an effect could further undermine contaminated organisms' ability to store energy.
In ectotherms, studies reporting contaminant-by-environment interactive effects on energy storage all involved temperature (Table 1; Table S8). In Cope's gray tree frog (Hyla chrysoscelis), metal exposure and elevated temperatures synergistically reduced body condition, reflecting combined effects of elevated MR at high temperatures and detoxification costs ). In contrast, in goldfish (Carassius auratus), a negative antagonistic effect occurred in which a POP-containing pesticide mixture decreased the hepatosomatic index (reflecting energy reserves) and muscle protein content at control, but not elevated temperatures . Studies in ectotherms also assessed combined effects of contamination and parasitism, predation risk and food shortage, but none reported significant interactions (Table 1; Table S8).

| Energy allocation
Assuming limited energy budgets, organisms must tradeoff allocating energy between competing functions, such as survival and reproduction (Van Noordwijk & de Jong, 1986). Climate change and contaminant exposure may combine in diverse fashions to interactively affect energy allocation.

| Shifting investment toward self-maintenance
First, climate change might intensify energetic tradeoffs between survival and reproduction by shifting environmental conditions away from the optimum, reducing resource availability or quality, leading to mismatches between breeding phenology and resource peaks (Thomas et al., 2001), and increasing energy demands, such as immune costs. In this scenario, organisms may enter an emergency life-history stage, and reduce energy investment into reproduction (McEwen & Wingfield, 2003;Wingfield et al., 1998). Contaminant exposure could lower the threshold in environmental challenge (e.g., resource shortage) at which individuals terminate reproduction, by impairing energy reserves and resource garnering capacity, leading to a negative synergistic effect. Contaminant exposure has been associated with reduced incubation attentiveness (Tartu et al., 2015), offspring provisioning (Evers et al., 2008), and reproductive success (Tartu et al., 2016), although interactive effects with climate changelinked environmental variables remain undocumented.
Contaminant effects on reproductive effort may be linked to effects on hormones that mediate reproductive effort, including CORT, prolactin, and luteinizing hormone. For instance, prolactin and Hg levels negatively correlated in two polar seabirds, the black-legged kittiwake and snow petrel (Pagodroma nivea), with breeding success in kittiwakes also suppressed (Tartu et al., 2015(Tartu et al., , 2016. Contaminant-linked disruption of hormones could undermine adjustments in reproductive effort with environmental change. Research suggests synergistic effects of environmental conditions and contamination on reproductive investment and success. For example, south polar skuas (Catharacta maccormicki) exposed to mirex (an organochlorine insecticide) and DDT reduced nest defense and displayed increased reproductive failure only in a year of high environmental stress, characterized by high baseline CORT (Goutte et al., 2018). Baseline CORT is a proxy of food availability and energetic challenge in seabirds (Kitaysky et al., 2007). Furthermore, reproductive success of Hg-exposed tree swallows declined only under warm conditions, which could be mediated by disruption of THs and thermoregulatory capacity . Contaminant exposure may disrupt metabolic processes and behavioral plasticity facilitating energy storage, and potentially exacerbating climate change-linked stress through cumulative effects that may be additive or interactive. For instance, some flame retardants and PCBs promote lipid accumulation, disrupting body mass regulation (Grün & Blumberg, 2009). These obesogens interfere with endocrine regulation of metabolism and neuroendocrine effects on appetite and feeding, and are implicated in metabolic syndromes and obesity in humans (Grün & Blumberg, 2009), and inappropriate weight gain in animals (Marteinson & Fernie, 2019).
Obesogens have potent effects early in life, predisposing individuals to altered metabolic profiles (Muscogiuri et al., 2017), and in the wild, where decreased maneuverability jeopardizes fitness (Marteinson & Fernie, 2019). Altered predation pressure under climate change, such as that of glaucous gulls (Larus hyperboreus) on little auks (Wojczulanis et al., 2005), could favor decreases in lipid reserves to increase maneuverability. Obesogens could block downregulation of lipid storage under elevated predation risk, resulting in an antagonistic effect on lipid storage (left), but synergistic effect on survivorship (right) (C = contaminated, UC = uncontaminated). Conversely, Cd has disruptive effects on lipid metabolism and storage in fish, mammals and birds (Larregle et al., 2008;Lucia et al., 2010;Pierron et al., 2007), and methylmercury also interferes with lipid metabolism (Moreira et al., 2012). These effects may be magnified by climate change if warmer temperatures increase chemical toxicity, resulting in a synergistic interaction. In addition, contaminant-linked appetite suppression may block increased feeding and accumulation of energy reserves, with this effect magnifying negative effects on energy storage under low resource availability (Groh et al., 2015).
as Hg-exposed black-legged kittiwakes, also skip breeding more often (Tartu et al., 2013). Deteriorating conditions for breeding and concomitant increases in contaminant exposure or toxicity might additively or synergistically reduce breeding incidence and population stability, with synergistic effects potentially induced if the combination of the two stressors caused organisms to surpass a threshold past which breeding activity could not be sustained.  Abundant resources may enable an independence or performance EMS. In independent energy management, no restriction on energy expenditure exists, and BMR and activity additively affect DEE.

| Reduced energetic costs
BMR does not covary with activity and positively covaries with DEE.
In performance energy management, an intrinsic positive association between BMR, activity and DEE exists, and upregulating BMR mobilizes energy for activity. Conversely, given resource shortage or intense energy demand, a limit to DEE is approached, inducing a BMR-activity tradeoff, and constrained energy management, in which BMR and DEE do not covary (Careau & Garland, 2012;Halsey et al., 2019).
Contaminant exposure may interfere with animals' capacity to manage energy use given climate change-associated challenges. Conversely, contaminant-linked suppression of BMR might interfere with upregulation of BMR and performance energy management ( Figure 3b). If birds must feed chicks at higher rates to compensate for reduced prey quality (Senécal et al., 2021), or increase foraging time to maintain provision rates at low prey density (Harding et al., 2007), importance of performance energy management could increase.
Furthermore, contaminant exposure might affect when individuals adopt different EMSs. For example, contaminant exposure might magnify effects of climate change-linked energy shortages, such that contaminated individuals with elevated BMRs reach a threshold in DEE and switch between independent/performance and constrained EMSs at lower AEE, and spend more time employing constrained energy management (Figure 4), potentially reducing fitness.

| Future directions
We currently lack clear understanding regarding which contaminantby-environment interactive effects result in broad-scale changes in population and ecosystem dynamics. To build a comprehensive understanding of bioenergetic outcomes and predict large scale responses, we particularly encourage research in the following areas.
2.6.1 | How do lower level effects (e.g., on MR) combine to affect higher bioenergetic (e.g., DEE) and fitness traits? Effects exerted across multiple lower level bioenergetic processes, such as BMR and activity, might combine to determine higher level effects on energy balance, DEE, and fitness. First, effects of contamination on lower level traits could combine to amplify higher level effects ( Figure 5a). Conversely, effects on lower level traits could counteract one another, dampening higher level effects (Figure 5b).
In either case, considering effects on multiple traits will aid in predicting energetic outcomes.

| Which species, populations, and individuals?
Vulnerable species' responses will serve as fulcrums around which broader-scale contaminant-by-environment interactive effects become manifest. Thus, identifying traits that determine vulnerability is essential. High trophic level species experience biomagnification (Burger & Gochfeld, 2004;Michelutti et al., 2010), and may be more sensitive to contaminant-induced effects on bioenergetics. Species also differ in evolved detoxification mechanisms, affecting sensitivity to contamination (Ikemoto et al., 2004). Furthermore, vulnerability may be higher in long-lived species due to bioaccumulation (Burger & Gochfeld, 2004), and in species and populations occupying habitats or areas that serve as contaminant sinks (e.g., the Arctic; agricultural fields). Individuals may also differ in contaminant exposure due to feeding specialization and differences in spatial move- With respect to sensitivity to climate change, vulnerability may be elevated in species inhabiting regions experiencing rapid change (e.g., polar regions), environments close to thermal tolerance limits, or areas lacking options for spatiotemporal shifts in range or phenology (Parmesan, 2006). Range-edge populations may be preadapted to cope with climatic extremes (Rehm et al., 2015), but may also suffer from reduced genetic diversity, potentially limiting scope for genetic response (Eckert et al., 2008). Overlap between high vulnerability to contaminant-and climate-related stress will determine species and populations of concern. Finally, determining physiological, behavioral, and energetic characteristics that heighten sensitivity once exposure occurs will be important.

| When and over which timescales?
Vulnerability to contaminant-by-environment interactive effects may increase during energy-intensive life-history stages, or periods of environmental challenge, including winter at high latitudes (Fort et al., 2009. Research demonstrates shifts in spatiotemporal variation in energy balance with climate change-associated environmental variation (Amélineau et al., 2018;Clairbaux et al., 2021).
Contaminant exposure now needs to be integrated into models.
Contaminant exposure may be concentrated during particular times of the year or life-history stages (Fort et al., 2014). Exposure may be higher on breeding or wintering grounds (Albert et al., 2021), and effects may increase with age due to bioaccumulation (Burger & Gochfeld, 2004). Studies are needed that span annual cycles and lifespans, and explore timeframes over which effects operate. There could also be lag-times between exposure and when effects become F I G U R E 3 Hypothetical relationships in individuals with high, versus low, contamination loads between (a) auxiliary energy demand and BMR (top) and activity (bottom), given constrained energy management and elevation of BMR by detoxification costs, and (b) auxiliary energy demand and BMR (top) and activity or FMR (bottom), given performance energy management and suppression of BMR by contamination. BMR, basal metabolic rate; FMR, field metabolic rate.
manifest, meaning that stressors need not occur simultaneously to produce interactive effects (Jackson et al., 2021). The timeframe of stress exposure (e.g., pulse vs. press disturbance; Bender et al., 1984) could also affect the nature of effects, and scope for evolutionary  (Kortsch et al., 2015).

| Thresholds and other nonlinearities?
Reaction norms describing effects of contaminant exposure and environmental stress on energetic traits may be nonlinear. Negligible effects may occur below thresholds of contamination that challenge detoxification mechanisms, or within the range of environmental conditions encompassed by acclimatization responses (Fisk et al., 2005). Beyond these thresholds, energetic responses may range in form from linear to exponential, and plasticity's ability to F I G U R E 4 Hypothetical relationship between contaminant exposure and time of switching between independent/performance and constrained energy management. When DEE is below a threshold (dashed line, (a), AEE is unconstrained by maintenance energy expenditure (BMR), or vice versa. As AEE increases, a threshold is reached, and reductions in AEE constrain DEE. Here, highly contaminated individuals (HC) have higher maintenance costs (BMR), surpass the threshold at lower AEE than individuals with low contamination loads (LC), and switch strategies sooner. The slope of the DEE-BMR relationship for the entire period (b) lies between 1 and 0 for both types of individuals, indicating partial energy constraint. However, the slope for HC individuals is closer to 0, indicating greater constraint. Contaminated individuals might also switch between an independent/performance EMS and constrained EMS at a lower threshold of DEE (imagine shifting the dashed line in a) down, but only for contaminated individuals), which could accentuate the difference in time between when contaminated and uncontaminated individuals switch. Extended from figure 2 of Halsey et al. (2019). AEE, auxiliary energy expenditure; BMR, basal metabolic rate; DEE, daily energy expenditure; EMS, energy management strategy.
buffer fitness may be surpassed. Research is needed to establish whether contaminant exposure lowers thresholds at which organisms become vulnerable to environmental change, and vice versa. In addition to thresholds, other nonlinearities in the impact of contaminant exposure or environmental stress deserve further attention, as they will complicate the nature of interactive effects. Examples include inverse dose-response curves (Vandenberg et al., 2012), and hormetic effects (Sebastiano et al., 2022).

| CON CLUS IONS
Climate change and contaminant exposure can have wide-ranging effects on bioenergetic traits. These effects may be additive in nature, but substantial evidence exists for both synergistic and antagonistic interactive effects. Our review revealed relative balance between antagonistic and synergistic interactions, and like other recent syntheses, suggested that unpredictable, nonadditive interactions are more common than additive effects (e.g., Crain Darling & Côté, 2008). Prevalence of synergistic interactions was once emphasized, but it has now been suggested that synergies do not prevail among interaction types (Côté et al., 2016;Darling & Côté, 2008 Predicting future ramifications of climate change and contaminant exposure on Arctic ecosystems, and globally, will depend on understanding mechanistic underpinnings of interactive effects, and how different effects will combine to affect fitness, species interactions, and community dynamics.

F I G U R E 5
Hypothetical example of reaction norms in multiple bioenergetic traits combining to determine reaction norms in higher level traits. Polar bears exhibit a seasonal fast during the summer and fall when sea ice disappears, whose duration is increasing with climate change. Simultaneous reductions in BMR and activity allow bears to reduce DEE (a; reaction norms in solid black) and survive the fast (Stirling & Derocher, 2012). However, contaminant exposure might interfere with both responses (a; reaction norms in dashed red) by elevating basal detoxification costs and causing hyperactivity through endocrine-disrupting effects (e.g., Polybrominated diphenyl ethers (PDBEs) cause hyperactivity in rats; Suvorov et al., 2009), especially as contaminants are released from lipid reserves in fasting bears (Polischuk et al., 2002), resulting in sustained elevated DEE, and deleterious effects on survival. Alternatively, reaction norms in one lower level trait may buffer against lack of response in another, maintaining the response in the higher level trait (b). For instance, if contaminated polar bears with heightened detoxification costs are unable to suppress BMR, activity levels may be more dramatically reduced than in uncontaminated individuals, such that a similar reaction norm in DEE is observed in contaminated and uncontaminated individuals, potentially serving as an adaptive buffering response to prevent starvation. BMR, basal metabolic rate; DEE, daily energy expenditure.

CO N FLI C T O F I NTE R E S T S TATE M E NT
The authors declare no conflicts of interest.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available from