Modifications to the text - 2020-03-13

[david-beauchesne]

Modifications to do and discuss:

• [x] Food web & interactions in introduction
• [x] Terminology in concept (impact vs effect)
• [x] Figure with unitary pathways
• [x] Add statement on the structure of the paper at the end of introduction
• [x] Modify figures with proper terminology
• [x] Try new amplification metric
• [x] Concept figure D part on cod
• [x] Methods for empirical illustration
• [x] Discussion within the text rather than separate section at the end (answers to questions)
• [x] Probability in last equation
• [x] Review list of comments from Kev's latest pull request
• [x] what it all means? -- suggestions:

  1. We shall tackle the challenge of complexity
  2. We can tackle this challenge
  3. From the cellular level to the entire ecosystem there are synergies and here we show how to use ecological networks to better understand how impacts of stressors at the species level may affect an entire community.
  4. Highlight punchy results.
  5. Needed for conservation, so we don't miss important synergies (would be cool to add an study case where focusing on 1 stressor and 1 species was damageable for the rest of ecosystem)

[david-beauchesne]

Comme fin d'introduction @KevCaz :

Here, our objective is to theoretically and holistically investigate the role of species and their interactions (i.e. food web topology) in driving species sensitivity to stressors and how they may buffer against or amplify the impacts of multiple stressors. In doing so, we seek to answer questions of particular significance to management and the application of holistic environmental approaches: 1) should species interactions be considered in impact assessments, 2) should the effects of stressors be evaluated separately or in combination, and 3) which species are most sensitive to stressors based on their trophic position. The paper is divided in two parts. In the first part, we conceptualize how multiple stressors permeate complex ecological communities. We then use numerical simulations to explore the many pathways through which species may be impacted in complex communities. In the second part, we illustrate our framework by using simulations as heuristics to infer the sensitivity of species in the St. Lawrence System in Eastern Canada.

[KevCaz]

The paper is divided in two parts. In the first part, we conceptualize how multiple stressors permeate complex ecological communities. We then use numerical simulations to explore the many pathways through which species may be impacted in complex communities.

"we conceptualize how multiple stressors permeate complex ecological communities." Je pense qu'on peut etre plus précis. En disant qu'on utilise les motifs à trois espèces et l'impact des stresseurs sur différentes caractéristiques dynamiques. Juste être un peu plus précis, c'est ça mon point.

In the second part, we illustrate our framework by using simulations as heuristics to infer the sensitivity of species in the St. Lawrence System in Eastern Canada.

j'enlèverais "by using simulations as heuristics" et je dirais un peu plus sur l'exemple: combien d'espèce, combien de stresseurs, données réelles?

[david-beauchesne]

Pour le paragraphe sur les interactions dans l'introduction @KevCaz :

Beyond their obvious direct impacts, stressors may also permeate entire ecological communities through the indirect pathways provided by the complex web of interactions in which species are embedded [@wootton2002; @bascompte2009a; @montoya2009; @ogorman2009; @ogorman2012]. Surprising observations may arise from indirect effects, such as a predator having a positive effect on its own prey [@abrams1992; @montoya2009]. Empirical evidence of trophically mediated indirect effects are abundant throughout all types of ecosystems globally [@paine1980; @estes2011]. Classic examples include the protection -- or lack thereof -- of kelp forests by the consumption of sea urchins (Strongylocentrotus sp.) by sea otters [Enhydra lutris; @estes1974], the indirect effect of largemouth bass (Micropterus salmoides) on phytoplankton in Long Lake [@carpenter2001], and the release of cottonwood (Populus spp.) and willows (Salix spp.) from elk (Cervus elaphus) browsing by the reintroduction of wolves (Canis lupus) in Yellowstone National Park [@ripple2001]. The number, the strength and the types of interactions structuring ecological communities govern its dynamic stability and how the impacts of stressors will propagate [@wootton2002; @montoya2009]. Modify any of these parameters, and the dynamics of the community -- and how stressors propagate -- will also change.

[KevCaz]

The number, the strength and the types of interactions structuring ecological communities govern its dynamic stability and how the impacts of stressors will propagate [@wootton2002; @montoya2009]. Modify any of these parameters, and the dynamics of the community -- and how stressors propagate -- will also change.

Mon sentiment c'est qu'il faut appuyer un peu plus fort et pour souligner que sans ça on peut manquer des choses. Un truc un peu fort en disant qu'evaluer les impacts des stresseurs, meme si on les considère tous et comme il faut par especes, si on ignore le reseau, on pourrait faire de lourde erreur quand à l'effet sur l'ensemble de l'ecosysteme.

[david-beauchesne]

Fin intro @KevCaz :

In the first part, we conceptualize how multiple stressors permeate complex ecological communities. We then simulate the impacts of stressors on the equilibrium dynamics of the most prevalant three-species motifs in food webs (i.e. food chain, omnivory, exploitative competition, and apparent competition) to explore the many pathways through which species may be impacted in complex communities. In the second part, we illustrate our framework by inferring the sensitivity of species in the St. Lawrence System in Eastern Canada using data from three empirical food webs describing different regions of the St. Lawrence and exposed to up to eight different sources of stress.

[KevCaz]

Ca marche bien!

most prevalant three-species motifs

tu veux pas mettre "all" parcequ'on fait pas le bidirectional?

[david-beauchesne]

Il y a aussi un unidirectionel qu'on n'analyse pas. Mais le but c'est de dire qu'on explore les plus fréquents. Je vais remplacer the most prevalent par the most frequent

[david-beauchesne]

The number, the strength and the types of interactions structuring ecological communities govern its dynamic stability and how the impacts of stressors will propagate [@wootton2002; @montoya2009]. Modify any of these parameters, and the dynamics of the community -- and how stressors propagate -- will also change.

Mon sentiment c'est qu'il faut appuyer un peu plus fort et pour souligner que sans ça on peut manquer des choses. Un truc un peu fort en disant qu'evaluer les impacts des stresseurs, meme si on les considère tous et comme il faut par especes, si on ignore le reseau, on pourrait faire de lourde erreur quand à l'effet sur l'ensemble de l'ecosysteme.

Je réfléchis à ce commentaire depuis hier et je suis debout depuis 4h30 à y penser. Je n'arrive pas à trouver quelque chose de satisfaisant. L'argument faible est de simplement dire que différentes configurations de réseaux vont avoir différentes dynamiques. Un argument plus fort se situerait dans le contexte des impacts de stresseurs. Ma réflexion revient toujours vers la perte d'espèces, les extinctions secondaires et la robustesse des réseaux. Mon problème avec ça c'est qu'on ne parle pas de perte d'espèce dans le reste de l'article. Bref, je ne suis vraiment pas certain de comment bien présente cet aspect..

[KevCaz]

En fait, je ne pense pas qu'il faille que ce soit plus un casse-tete.

L'argument faible est de simplement dire que différentes configurations de réseaux vont avoir différentes dynamiques.

Ça suffit. Mais le changement de dynamique, c'est ce qui peut mettre des espèces en danger, c'est ce qui peut empêcher que la pêche soit durable, que les baleines reviennent ou non chaque année.... On ne répond pas à ces questions dans le papier, de toute manière on ne bouscule pas trop les paramètres, il me semble. Mais si on ajoutait de la stochasticité à ce genre d'équation, on pourrait evaluer des risques. C'est pas le but ici, donc on le fera pas, mais ça ne nous empêche pas de dire que si on ne prend pas en compte les interactions, on loupe qqchose. Pour moi c'est important sinon y'a pas tant d'intérêt à faire autant d'effort pour aller chercher le réseau, sinon on peut faire du muti-especes/ multi-stresseur sans réseau.

[david-beauchesne]

Fair enough! J'aurais dû dormir ;p

[david-beauchesne]

Avant d'aller plus loin:

Beyond their obvious direct impacts, stressors ripple through ecological communities by way of the interactions structuring the complex network in which species are embedded [@wootton2002; @bascompte2009a; @montoya2009; @ogorman2009; @ogorman2012]. Surprising observations arise from complex networks, such as a predator positively affecting its own prey [@abrams1992]. Ample empirical evidence exist of such trophically-mediated effects across ecosystems globally [@paine1980; @estes2011]. Classic examples include sea otters (Enhydra lutris) indirectly shielding kelp forests from browsing by sea urchins [Strongylocentrotus sp.; @estes1974] and the release of cottonwood (Populus spp.) and willows (Salix spp.) from elk (Cervus elaphus) browsing following the reintroduction of wolves (Canis lupus) in Yellowstone National Park [@ripple2001].

How ecological networks are structured, i.e. the number, configuration and strength of interactions between species, influences the propagation of stressors and the stability and resilience of whole systems [@refs]. Stressors can rewire entire food webs by modifying its structure and altering the flow of energy in the system [@blanchard2015; @kortsch2015; @bartley2019]. Links can be added or removed from food webs [i.e. topological rewiring; @bartley2019] through primary and secondary species extinctions [e.g. @ref], climate-related distributional shifts [e.g. @kortsch2015; @bartley2019] or invasive species introductions [e.g. @ref]. Consumer foraging can also be altered through local shifts in space and resource use [i.e. interation strength rewiring; @bartley2019; @refs]. Rewiring network structure can lead to dramatic in whole community dynamics, such as [...].

[KevCaz]

So far so good!

[david-beauchesne]

We define a variation in species abundance resulting from the impact of stressors on a food web as a species trophic sensitivity ($S$; Figure \ref{concept}B):

$$S_{m,Kj} = \frac{a{m,K_j} - a_m}{a_m}\text{,}$$

Suggestion:

Disturbances to individual biological processes arise though unitary pathways of effect ($j$; Figure \ref{concept}B1-3, \ref{pathwaysmethod}). In reality, those unitary pathways of effect combine to impact our system simultaneously to the benefit of cod and the detriment of capelin and beluga (Figure \ref{concept}B4). We define the ensemble of unitary pathways of effect ($j$) impacting a system simultaneously as an integrative pathway of effect ($K_j$). In the remainder of the text, if we use the term pathway of effect without a qualifier (i.e. integrative or unitary), we mean an integrative pathway of effect. We further define the net impact of an integrative pathway of effect ($Kj$) on a focal species $m$ as its trophic sensitivity ($S{m,K_j}$; Figure \ref{concept}B):

$$S_{m,Kj} = \frac{a{m,K_j} - a_m}{a_m}\text{,}$$

where $am$ and $a{m,Kj}$ are the pre- and post-stressor abundances of species $m$, respectively. Note that by definition $S{m,K_j}$ is bounded negatively to -1, as species abundances cannot be negative.

[KevCaz]

Disturbances to individual biological processes arise through unitary pathways of effect ($j$; Figure \ref{concept}B1-3, \ref{pathwaysmethod}). In reality, those unitary pathways of effect combine to impact our system simultaneously to the benefit of cod and the detriment of capelin and beluga (Figure \ref{concept}B4). We define the ensemble of unitary pathways of effect ($j$) impacting a system simultaneously as an integrative pathway of effect ($K_j$). In the remainder of the text, if we use the term pathway of effect without a qualifier (i.e. integrative, additive or unitary), we mean an integrative pathway of effect. We further define the net impact of an integrative pathway of effect ($Kj$) on a focal species $m$ as its trophic sensitivity ($S{m,K_j}$; Figure \ref{concept}B):

[KevCaz]

Je suis toujours pas bien sur de comprendre le $j$, il faudrait qu'on s'en parle, c'est mineur mais il est temps d'etre ok sur la notation!

[david-beauchesne]

C'est bon! Je me demande si je ne devrais pas inclure la figure d'omnivorie avec les paramètres avec la figure concept sous le réseau, soit directement dans la partie A, ou dans une partie B.. Dans tous les cas je sis d'accord il faut être ok sur la notation!

En attendant (Liam se réveille), je trouvais que tout ça était trop lourd alors j'ai essayé de modifié le début de la section pour intégrer les notations à la narration du texte directement:

We focus on the omnivory interaction connecting cod, beluga and capelin in our system to explore the dynamic impact of multiple stressors (Figure \ref{concept}B). Net effects are typically measured as variations in species abundances or densities in food webs, which integrate all trophically-mediated effects operating on the system collectively [@wootton2002; montoya2009]. We measure the net impact of stressors in the same way, i.e. by evaluating how pre-stressor abundances at equilibrium shift after the permanent appearance of stressors in a system. Each biological process disrupted by stressors constitute a unitary pathway of effect ($j$) that can induce contrasting population trajectories. The impact of fishing on capelin mortality ($j = r_x$; Figure \ref{pathwaysmethod}) favours cod and results in reduced abundances for capelin and beluga (Figure \ref{concept}B1). In this scenario, cod are likely released from beluga predation due to their drop in numbers [i.e. mesopredator release; ritchie2009]. This trophically-mediated effect could ultimately exacerbate the impact of fishing on capelin by favouring one of its predators. Meanwhile, impacting cod mortality ($j = my$; Figure \ref{pathwaysmethod}) results in the growth of the capelin and beluga populations (Figure \ref{concept}B2). Surprisingly, the cod population remains relatively unchanged (Figure \ref{concept}B2), likely bacause the increase in prey availability offsets the impact of fishing [i.e. compensatory dynamics; @gonzalez2009]. Finally, the beluga population appears insensitive to the impact of shipping ($j = \alpha{xz}$ and $j = \alpha_{yz}$; Figure \ref{pathwaysmethod}); yet shipping likely disrupts the top-down control of beluga on cod to the benefit of cod and to the detriment of capelin (Figure \ref{concept}B3).

Unless a single biological process is impacted, unitary pathways of effect combine to form an integrative pathway of effect ($j \in K_j$). In our system, the impacts of shipping and fishing combine to form an integrative pathway of effect ($K_j = r_x, my, \alpha{xz}, \alpha{yz}$) that benefits cod and reduces capelin and beluga (Figure \ref{concept}B4). We define as a species $m$ trophic sensitivity $S{m,K_j}$ the net impact -- i.e. the pre- and post-stressors variation in abundance -- resulting from an integrative pathway of effect $K_j$ (Figure \ref{concept}B):

$$S_{m,Kj} = \frac{a{m,K_j} - a_m}{a_m}\text{,} \label{eq1} \tag{1}$$

where $am$ and $a{m,Kj}$ are the pre- and post-stressors abundances of species $m$, respectively. Note that by definition $S{m,K_j}$ is bounded negatively to -1, as species abundances cannot be negative. In the remainder of the text, if we use the term pathway of effect without a qualifier (i.e. integrative or unitary), we mean an integrative pathway of effect.

[david-beauchesne]

Bon voici ce que ça devient:

We explore the dynamic impacts of multiple stressors with the omnivory interaction connecting cod, beluga and capelin in our system (Figure \ref{concept}B). Net effects are typically measured as variations in species abundances or densities in food webs, which integrate all trophically-mediated effects operating on the system collectively [@wootton2002; @montoya2009]. Likewise, we evaluate how pre-stressor abundances at equilibrium shift after the permanent appearance of stressors in a system as a measure of net impact.

Impacts to single biological processes travel through unitary pathways of effect ($k$), such as an increase in cod mortality ($k = {m_y}$; Figure \ref{concept}B). Unitary pathways of effect that can induce contrasting population trajectories. The impact of fishing on capelin mortality ($k = {r_x}$; Figure \ref{pathwaysmethod}) favours cod and results in reduced abundances for capelin and beluga (Figure \ref{concept}B1). In this scenario, cod are likely released from beluga predation due to their drop in numbers [i.e. mesopredator release; @ritchie2009]. This trophically-mediated effect could ultimately exacerbate the impact of fishing on capelin by favouring one of its predators. Meanwhile, impacting cod mortality ($k = {my}$; Figure \ref{pathwaysmethod}) results in the growth of the capelin and beluga populations (Figure \ref{concept}B2). Surprisingly, the cod population remains relatively unchanged (Figure \ref{concept}B2), likely because the increase in prey availability offsets the impact of fishing [i.e. compensatory dynamics; @gonzalez2009]. Finally, the beluga population appears insensitive to the impact of shipping ($k = {\alpha{xz}} and k = {\alpha_{yz}}$; Figure \ref{pathwaysmethod}); yet shipping likely disrupts the top-down control of beluga on cod to the benefit of cod and to the detriment of capelin (Figure \ref{concept}B3).

Unless a single biological process is impacted, unitary pathways of effect combine to form an integrative pathway of effect ($k \in K$). Shipping and fishing impact collectively form an integrative pathway of effect ($K = {r_x, my, \alpha{xz}, \alpha{yz}}$) that benefits cod and reduces capelin and beluga (Figure \ref{concept}B4). We define a species ($m$) trophic sensitivity ($S{m,K_j}$) as the net impact -- i.e. the pre- and post-stressors variation in abundance -- resulting from an integrative pathway of effect $K$ (Figure \ref{concept}B):

$$S{m,K} = \frac{a{m,K} - a_m}{a_m}\text{,} \label{eq1} \tag{1}$$

where $am$ and $a{m,K}$ are the pre- and post-stressors abundances of species $m$, respectively. Note that by definition $S_{m,K}$ is bounded negatively to -1, as species abundances cannot be negative. In the remainder of the text, if we use the term pathway of effect without a qualifier (i.e. integrative or unitary), we mean an integrative pathway of effect.

[david-beauchesne]

Pour la variance trophique:

A species trophic sensitivity -- or lack thereof -- can also arise from different mechanisms. Unitary pathways of effect may reinforce each other, whereas others may cancel each other out [@wootton2002; @montoya2009]. For example, the positive impact of cod mortality on capelin (Figure \ref{concept}B2) is offset by the negative impacts of capelin mortality and altered beluga behaviour (Figure \ref{concept}B1, B3, B4). Comparing the effective and expected impacts of a unitary pathway of effect -- i.e. the average impact of an integerative pathways of effect -- provides a measure of variance on trophic sensitivity to an integrative pathway of effect ($K$) that we define as a species ($m$) trophic variance ($V_{m,K}$):

$$V{m, K} = \sum{k \in K} \left(S{m, k} - \frac{1}{|K|} S{m, K} \right)^2\text{,} \label{eq3} \tag{3}$$

Low trophic variance arise from sets of unitary pathways of effect whose individual impacts are relatively similar, whereas high trophic variance identify sets of contrasting unitary pathways of effect. In our system, beluga ($V{beluga,K) = 0.22$) and capelin ($V{m,K} = 0.18$) are exposed to unitary pathways of effect that tend cancel each other out, whereas cod ($V_{m,K} = 0.09$) is exposed to unitary pathways of effect that reinforce each other.

[KevCaz]

Unless a single biological process is impacted, unitary pathways of effect combine to form an integrative pathway of effect ($k \in K$)

($k \in K$) => ($K$)

[KevCaz]

$S_{m,Kj}$ => $S{m,K}$

[KevCaz]

post-stressors => post perturbation?

[david-beauchesne]

Un des commentaires à Dom c'est qu'il n'aime pas l'utilisation des termes disturbance et perturbation, alors je les évite le plus possible. Le terme est consacré à autre chose dans la littérature

[david-beauchesne]

Intro:

Beyond their obvious direct impacts, stressors ripple through ecological communities by way of the interactions structuring the complex network in which species are embedded [@wootton2002; @bascompte2009a; @montoya2009; @ogorman2009; @ogorman2012]. Surprising observations arise from complex networks, such as a predator positively affecting its own prey [@abrams1992]. Ample empirical evidence exist of such trophically-mediated effects across ecosystems globally [@paine1980; @estes2011]. Classic examples include sea otters (Enhydra lutris) indirectly shielding kelp forests from browsing by sea urchins [Strongylocentrotus sp.; @estes1974] and the release of cottonwood (Populus spp.) and willows (Salix spp.) from elk (Cervus elaphus) browsing following the reintroduction of wolves (Canis lupus) in Yellowstone National Park [@ripple2001]. A species's susceptibility to trophically-mediated effects is influenced by its trophic role and position. For example, species with diversied diets (i.e. generalists) are more resilient than species with specialized diets [i.e. specialists; @montoya2009; @clavel2011], and top predators are generally more vulnerable to trophically-mediated effects [@ripple2015; @stier2016].

How ecological networks are structured, i.e. the number, configuration and strength of interactions between species, also influences the propagation of stressors and the stability and resilience of whole systems [@wootton2002; @montoya2009; @bartley2019]. Stressors can modify these structural properties and rewire entire food webs [@blanchard2015; @kortsch2015; @bartley2019]. Links can be added or removed [i.e. topological rewiring; @bartley2019] through primary and secondary species extinctions [e.g. @allesina2006; @eklof2006], climate-related distributional shifts [e.g. @kortsch2015; @bartley2019] or invasive species introductions [e.g. @vanderzanden1999; @david2017]. Alteration to the flow of energy arise when consumer modify their space and resource use [i.e. interaction strength rewiring; @bartley2019].