Our latest work¹ led by David Beauchesne is out since yesterday in Ecology Letters 🥳! This paper introduces a framework that allows the exploration of the impact of multiple stressors on food webs.

Proud to have published part of my PhD on the sensitivity of food webs to multiple stressors in @Ecology_Letters: https://t.co/jjbQjwi6u2. Thank you @LauraEllenDee , @KCazelles, @philbenthos & @GravelDom for your collaboration!

— David Beauchesne (@d_beauchesne) July 20, 2021

In the Anthropocene, human activities impact species is various ways. Multiple stressors frameworks attempt to account for their diversity and their joint effects (stressors may have antagonistic or synergistic effects). Most of such frameworks (if not all) are species-based, meaning that they account for the sensitivity of species to the stressors and the intensity of the stressor (e.g., Halpern 2019²) but not for the ecological relationships among species. However, it is a truism to say that species are not independent and therefore approaches that neglect ecological interactions may lead to inaccurate predictions. That is why we’ve put a lot of efforts in combining food web concepts in a multiple stressors approach.

Adding another layer of complexity in an already complex framework proved challenging: it required creativity and work! IMHO, David used plenty of both to develop the framework and convincingly showed how fruitful it could be. While working on this, we realized that some new concepts were needed, and several terms were thus coined. A major concept is what David called the pathway of effect, which is “the collection of ecological processes through which stressors directly and indirectly affect ecological communities”¹. Using generalized Lotka-Volterra models (hereafter LV models) for all 3-species motifs, we systematically evaluated the impact of all pathways of effects on species abundances. We did so by slightly modifying the parameters of the LV models³, and comparing the abundances at equilibrium before and after the perturbation. For a given pathway of effet, this difference was called the trophic sensitivity, and we further defined the trophic amplification, which basically is the difference between the observed trophic sensitivity and the sum of trophic sensitivities for the all unitary pathways of effects ( pathways of effect for which only one parameter of the LV model is perturbed). Note that the latter quantifies the non-additivity of the pathway of effect.

The interpretation of these two quantities is rather straightforward. For a given pathway, a species with a high trophic sensitivity is strongly affected by stressors (if the value is negative, then abundances are negatively impacted and if the value is positive then species are positively impacted). Also, high trophic amplification values mean that there are strong non-additive effects (though the exact interpretation of the signed values depends on the sign of the trophic sensitivity). So we had these two quantities for the three species in every motif. Once those metrics defined, we show the relationship between the two for all pathway of effects and there is actually a lot of variation among motifs ⬇️.

In order to obtain the trophic sensitivity and the trophic amplification for species in a complex network, for each species, we performed a motif census (see motifcensus for an implementation) and then sum all the corresponding values of trophic sensitivity and trophic amplification ⬇️.

We then apply this framework on real marine ecological networks that represent the St-Lawrence system in the mid-1980s. Notably, we show that applying our framework can reveal species (or group of species) particularly vulnerable to multiple stressors ⬇️.

We actually say much more in the paper, so if you want to learn I suggest you read it through! This was a lot of work, we had a lot of feedback from the reviewers to make the paper as clear as possible. I hope we’ve succeeded in doing that, and I hope you’ll be as excited as we are about the research avenues this study unveils! By the way, David did his best to make this study reproducible, see ➡️ https://zenodo.org/record/5014237#.YPhE8iUpD0o so you know where to start if you want to work on this 😉.