Future research directions
Here, we have focused primarily on ecological scenarios where organisms are exposed to two primary stressors, with a third stressor (predation risk) possibly arising as a consequence of increased energy acquisition behaviour (trade-off 2). However, to develop a deeper understanding, including means of accurately predicting the nature and strength of interactive effects of multiple stressors, we must extend our focus to scenarios involving more than two primary stressors, where higher-order interactions can occur. While empiricists are incorporating increasingly larger numbers of stressors into experiments (e.g. Beermann et al. (2018), and fields outside of ecology have developed conceptual models to define higher-order interactions among three or more stressors (e.g. interactions among pharmaceutical drugs, Beppler et al.(2016)), our current framework suggests that explicitly considering animal behavior in these efforts will generate novel insights relevant to ecology and conservation. Furthermore, in light of recent advances in simultaneously collecting large volumes of data on animal behavior and multiple environmental variables in situ , such an integration is both timely and likely to illuminate additional stressor-related trade-offs and alternative behavioral responses used by organisms to navigate such tradeoffs.
With regard to understanding variation in how organisms respond to multiple environmental factors, for over a decade, there has been mounting interest in the importance of consistent individual differences in animal personalities or behavioural syndromes (e.g., in aggressiveness, boldness or exploratory tendency; (Sih et al.2004; Reale et al. 2007; Sih et al. 2012) including dispersal tendency (Cote et al. 2010), physiology (Biro & Stamps 2008) and life histories (Reale et al. 2010). Promising topics that remain understudied include how individual differences in suites of phenotypic traits relate to variation in how organisms balance the four trade-offs discussed here.
Another fruitful direction for future studies would be to examine the influence of a mix of genetic adaptation and transgenerational and within-generation developmental plasticity (including learning) in shaping an integrated response to multiple stressors. Examining ecological and social factors that, in the past or present, shaped the overall integrated response to multiple stressors could help identify genetic or developmental constraints that affect the speed or trajectory of adaption to multiple stressors (De Coninck et al. 2013; Cambronero et al. 2018). In particular, understanding epigenetic or developmental effects can reveal otherwise hidden mechanisms of multiple stressor effects discussed in Box 4. With multi-generational transgenerational plasticity, behavioural responses to stressors in one generation can influence impacts of those stressors on others generations into the future (Bell & Hellmann 2019).
Finally, while we have focused primarily on individual responses to multiple stressors, in Box 5 and Figure 5 we expand our scope to consider stressors, and their physiological and behavioral effects, in the context of natural communities. Specifically, we outline how stressors can affect species interactions and how the nature, strength, and trophic position of these affected interactions can determine qualitatively distinct outcomes for communities and ecosystems. Further theoretical and empirical investigations of how our comprehensive framework on stressor effects (Figure 1) could help explain the structure and dynamics of natural communities offer a timely and promising avenue for future study.
Acknowledgements
We thank Anne E. Todgham for her feedback on the initial conceptual framework. This overall endeavor was supported by National Science Foundation grant IOS 1456724 to AS. In addition, LKL was supported by an Australia Awards Endeavour Fellowship, MAG was supported by a NSF Postdoctoral Fellowship and University of Colorado Chancellor’s Postdoctoral Fellowship. PCT was partly funded by the German Research Foundation (DFG) as part of the SFB TRR 212 (NC³).  IYL was supported by National Science Foundation grant 1557836.