In EffetsCumulatifsNavigation/ceanav: Research compendium for project on the assessment cumulative effects of marine vessel activities in the St. Lawrence
Plant and wildlife habitats
Natural environments
Several exercises to assess the vulnerability of different habitat types to various environmental stressors have been published in Canada [e.g. @ban2010; @clarkemurray2015; @clarkemurray2015a] and elsewhere in the world [@teck2010; @kappel2012]. A similar process led by Fisheries and Oceans Canada is currently underway to characterize the vulnerability of West Atlantic habitats to approximately 50 stressors. Originally developed by @halpern2007 to characterize habitat vulnerability on a global scale, this approach uses exposure and sensitivity metrics to assess five ecosystem vulnerability criteria (i.e. habitats in our study): spatial scale, frequency, trophic impact, resistance, and resilience (Table \@ref(tab:critereTeck)).
tab <- data.frame(critere = c("Spatial scale", "Frequency",
"Trophic impact","Resistance",
"Recovery time"),
descr = c(
"The spatial scale ($km^2$) at which a stressor impacts the habitat, both directly and indirectly",
"The mean annual frequency (number/year) of a stressor at a particular location within a given region",
"The degree to which marine life is affected by a stressor within a given ecosystem and region (i.e., one or more species, one or more trophic levels, or an entire ecosystem)",
"The degree (in %) to which species, trophic levels, or the natural state of an entire ecosystem is affected by the stressor",
"The mean time (in years) required for the affected species, trophic level(s), or entire community to return to its former natural state following disturbance caused by the stressor"))
tab %>%
knitr::kable(
col.names = c("Vulnerability criterion", "Description"),
row.names = FALSE,
caption = "Description of the five vulnerability criteria used to assess the relative vulnerability of habitats to each stressor [adapted from @teck2010]."
)
These allow the use of quantitative factors that provide greater transparency than the qualitative approach [@teck2010; @kappel2012]. The vulnerability scores $\mu$ are the weighted sum of the five vulnerability criteria:
$$mu_{D_i,E_j} = \sum_{k = 1,...,5} W_k * D^j_{i,k}$$
where $D^j_{i,k}$ is the value of stressor $i$ for criterion $k$ within ecosystem $j$ and $W_k$ is the weight associated with criterion $k$, with the sum of all weights equal to 1. The weight $W_k$ is thus used to weight the value attributed to individual criteria according to their individual importance. This additive and linear model assumes that vulnerability is monotonic for all vulnerability criteria, i.e. that vulnerability increases with an increase in the value of each criterion. A transformation of the $W_k$ values is required to ensure that the range of values is comparable for the different criteria. This equation ultimately generates a matrix of vulnerability scores for all stressor-ecosystem combinations.
@teck2010 and @kappel2012 used expert surveys to assess the values and weights associated with each vulnerability criterion for each stressor-habitat combination. A total of 107 experts assessed the relative vulnerability of 19 habitat types to 53 environmental stressors. The resulting vulnerability matrices have been used and adapted many times since [e.g. @ban2010; @clarkemurray2015]. These vulnerability matrices also provide a starting point for our assessment; indeed, all of the environmental stressors, and a significant portion of the habitats considered for our assessment, are found in the list presented by @teck2010, @kappel2012 and @clarkemurray2015. Specifically, we used the vulnerability matrix from @kappel2012 to inform the vulnerability assessment of habitats for which a clear match could be made, i.e., natural environments. Refer to Appendix 6 and Appendix 7 for the matches used between our study and @kappel2012 for stressors and valued components, respectively.
Some habitats considered for our assessment, however, did not have a clear match with the matrix of @kappel2012. In these cases, we adopted the approach used by @clarkemurray2015 and selected the habitat(s) presented by @kappel2012 that were most like those for which a clear match was unavailable. For example, the @kappel2012 shallow soft sediment and salt marsh habitat scores were used for mollusk beds and fluvial terraces, respectively. If a habitat under consideration could aggregate multiple @kappel2012 habitats, we instead used the average score of these habitats to obtain a vulnerability score. Vulnerability scores assigned to floodplains were thus obtained based on the average of the vulnerability scores assigned to rocky intertidal, intertidal, and salt marsh environments in the @kappel2012 matrix. Similarly, the government and research shipping category includes a significant diversity of vessel types such as military craft, patrol vessels, and scientific research vessels. The vulnerability of habitats to this type of shipping was thus assessed using the average of the scores assigned to shipping, military, research, and scientific sampling from @kappel2012. Finally, for the 2023 update, the vulnerability scores for eelgrass and algal zones were used for aquatic grass bed habitats.
Some modifications were also made to the vulnerability scores reported by @kappel2012. First, the vulnerability values assigned to wetlands were modified to appropriately capture the fineness of the dataset available for our study. The characterized wetlands reflect the tiered nature of wetlands, with shallow water wetlands, followed by marshes, then swamps, with the potential exposure of these environments to marine-based stressors gradually decreasing. We thus decreased the vulnerability of wetlands to all stressors proportionally by a factor of 5% for marshes and 10% for swamps. We retained the @kappel2012 assessment for the wetland habitat category containing wetlands of unknown type as a precaution. Second, accidental spills were categorized to differentiate hydrocarbon spills from other types of spills, meaning that we decrease the spill vulnerability values by a factor of 10% for other spills. Third, the vulnerability scores for aquatic grass beds was decreased by 10% per biovolume category under the assumption that higher biovolumes are more important, and thus have a higher vulnerability: high biovolume (100%), moderate biovolume (90%), and low biovolume (80%). As with the wetlands, we kept the vulnerability value intact for spills of unknown content as a precaution. Lastly, a relative score was estimated by normalizing the total scores by the maximum observed vulnerability value among the habitats considered (i.e. 3.6) to obtain a score ranging from 0 to 1. The resulting vulnerability matrix can be consulted in Table \@ref(fig:habitatvuln).
knitr::include_graphics("./figures/figures-vulnerability/habitat.png")
Habitats of importance to wildlife and plant species
The vulnerability of habitats identified for their importance to life cycles and special status species cannot be assessed using the same approach used for natural environments. These sites differ from the habitats presented above in that they are environments whose characteristics may vary, but which perform essential functions for the maintenance of certain species, or are deemed important for the conservation of certain special status species. As such, the vulnerability of these sites cannot be assessed in the same way as the assessment presented in the previous section. An alternative approach using species-based criteria rather than habitat-based criteria was used to assess their relative vulnerability to the effects of the stressors under consideration.
Exercises to assess species vulnerability to the effects of stressors are less common in cumulative effects assessments, although some assessment processes do exist [e.g. @maxwell2013; @trew2019; @ohara2021]. These assessments primarily involved marine megafauna species (e.g. whales and sharks) and species at risk. However, these publications do not cover all species and stressors included in this assessment. They do provide a set of criteria for assessing species vulnerability to stressors. We used a similar process to determine the vulnerability to marine vessel activities of environments of importance to species in the St. Lawrence and Saguenay rivers.
The criteria used to assess species vulnerability to the effects of marine vessel activities are taken from @maxwell2013: 1) the frequency of stressors, 2) whether the effect of the stressor is direct or indirect, 3) the resistance of the species to the effects of the stressor, 4) the recovery time of the species, 5) the relative effect on reproduction, and 6) the relative effect on the population (Table \@ref(tab:vulnerabilitySpecies)). We added two criteria to this list in order to consider regional particularities specific to the St. Lawrence and Saguenay rivers, namely 7) the status of the species and 8) whether the species is resident in the study area (Table \@ref(tab:vulnerabilitySpecies)).
meta <- read_yaml("../data/data-metadata/vulnerability_mammiferes_marins.yml")
criteres <- meta$criteres
tab <- list()
lin <- numeric()
for(i in 1:length(criteres)) {
dat <- meta$details_en[[criteres[i]]]
nm <- character(length(dat$description$rank))
nm[1] <- glue("***{dat$name}***")
nm[2] <- glue("*{dat$question}*")
nm[3] <- ifelse(!is.null(dat$specification), dat$specification, "")
crit <- as.data.frame(dat$description)
tab[[i]] <- cbind(nm,crit)
lin[i] <- length(nm)
}
lin <- cumsum(lin)
tab <- dplyr::bind_rows(tab)
# -----
tab %>%
knitr::kable(
col.names = c("Vulnerability criteria", "Rank", "Categories", "Description", "Examples"),
row.names = FALSE,
caption = "Criteria considered for assessing the vulnerability of valued components with regard to species – i.e. - important environmental conditions for life cycles, special status species and marine mammals – to the stressors considered for the CEA of marine vessel activities in the St. Lawrence and the Saguenay river. The criteria were adapted from @maxwell2013, to which species special status and residence criteria were added in order to account for the regional particularities of the St. Lawrence and the Saguenay river."
) #%>%
#kableExtra::row_spec(lin, extra_css = "border-bottom: 2px solid")
The criteria used varied depending on the habitat types assessed. The frequency of environmental stressors was used for all habitats of importance to the species. The vulnerability of spawning and nursery sites considered the frequency of environmental stressors and the effects on reproduction and population. The vulnerability of habitats of importance to seabirds was assessed by considering the frequency of stressors and the effects on populations. Lastly, the vulnerability of habitats of importance to special status species was assessed using the frequency of environmental stressors, effects on populations, and species status. It should be noted that the vulnerability of sites of importance to special status wildlife and plant species was considered in the same way. Although wildlife and plant species respond differently to environmental stressors, our assessment uses occurrences of species at risk as a proxy for vulnerable environments rather than a species-specific assessment. The general characterization used did not allow us to differentiate between species, so we focused on criteria specific to species status.
For each combination of valued component $CV_i$ and stressor $S_j$, the final vulnerability score $\mu_{CV_i, S_j}$ is obtained from the sum of the scores for each criterion:
$$mu_{CV_i, S_j} = \sum_{k \in K} C^i_{j,k}$$
where $C^i_{j,k}$ is the value of the valued component $i$ for the stressor $j$ and the criterion $k$, and $K$ are the set of criteria considered to assess the vulnerability of the valued component $i$.
Lastly, a relative score was assessed by normalizing the total scores by the maximum observed vulnerability value per habitat group assessed using the same criteria. Spawning and nursery sites were the first group, seabirds the second, and special status species the third. The final vulnerability scores are available in Table \@ref(fig:speciesvuln) and the individual criteria assessment tables per habitat category are available in Appendix 8.
knitr::include_graphics("./figures/figures-vulnerability/faune_flore.png")
EffetsCumulatifsNavigation/ceanav documentation built on April 17, 2023, 1:02 p.m.
R Package Documentation
Browse R Packages
We want your feedback!
Note that we can't provide technical support on individual packages. You should contact the package authors for that.
Plant and wildlife habitats
Natural environments
Several exercises to assess the vulnerability of different habitat types to various environmental stressors have been published in Canada [e.g. @ban2010; @clarkemurray2015; @clarkemurray2015a] and elsewhere in the world [@teck2010; @kappel2012]. A similar process led by Fisheries and Oceans Canada is currently underway to characterize the vulnerability of West Atlantic habitats to approximately 50 stressors. Originally developed by @halpern2007 to characterize habitat vulnerability on a global scale, this approach uses exposure and sensitivity metrics to assess five ecosystem vulnerability criteria (i.e. habitats in our study): spatial scale, frequency, trophic impact, resistance, and resilience (Table \@ref(tab:critereTeck)).
tab <- data.frame(critere = c("Spatial scale", "Frequency", "Trophic impact","Resistance", "Recovery time"), descr = c( "The spatial scale ($km^2$) at which a stressor impacts the habitat, both directly and indirectly", "The mean annual frequency (number/year) of a stressor at a particular location within a given region", "The degree to which marine life is affected by a stressor within a given ecosystem and region (i.e., one or more species, one or more trophic levels, or an entire ecosystem)", "The degree (in %) to which species, trophic levels, or the natural state of an entire ecosystem is affected by the stressor", "The mean time (in years) required for the affected species, trophic level(s), or entire community to return to its former natural state following disturbance caused by the stressor")) tab %>% knitr::kable( col.names = c("Vulnerability criterion", "Description"), row.names = FALSE, caption = "Description of the five vulnerability criteria used to assess the relative vulnerability of habitats to each stressor [adapted from @teck2010]." )
These allow the use of quantitative factors that provide greater transparency than the qualitative approach [@teck2010; @kappel2012]. The vulnerability scores $\mu$ are the weighted sum of the five vulnerability criteria:
$$mu_{D_i,E_j} = \sum_{k = 1,...,5} W_k * D^j_{i,k}$$
where $D^j_{i,k}$ is the value of stressor $i$ for criterion $k$ within ecosystem $j$ and $W_k$ is the weight associated with criterion $k$, with the sum of all weights equal to 1. The weight $W_k$ is thus used to weight the value attributed to individual criteria according to their individual importance. This additive and linear model assumes that vulnerability is monotonic for all vulnerability criteria, i.e. that vulnerability increases with an increase in the value of each criterion. A transformation of the $W_k$ values is required to ensure that the range of values is comparable for the different criteria. This equation ultimately generates a matrix of vulnerability scores for all stressor-ecosystem combinations.
@teck2010 and @kappel2012 used expert surveys to assess the values and weights associated with each vulnerability criterion for each stressor-habitat combination. A total of 107 experts assessed the relative vulnerability of 19 habitat types to 53 environmental stressors. The resulting vulnerability matrices have been used and adapted many times since [e.g. @ban2010; @clarkemurray2015]. These vulnerability matrices also provide a starting point for our assessment; indeed, all of the environmental stressors, and a significant portion of the habitats considered for our assessment, are found in the list presented by @teck2010, @kappel2012 and @clarkemurray2015. Specifically, we used the vulnerability matrix from @kappel2012 to inform the vulnerability assessment of habitats for which a clear match could be made, i.e., natural environments. Refer to Appendix 6 and Appendix 7 for the matches used between our study and @kappel2012 for stressors and valued components, respectively.
Some habitats considered for our assessment, however, did not have a clear match with the matrix of @kappel2012. In these cases, we adopted the approach used by @clarkemurray2015 and selected the habitat(s) presented by @kappel2012 that were most like those for which a clear match was unavailable. For example, the @kappel2012 shallow soft sediment and salt marsh habitat scores were used for mollusk beds and fluvial terraces, respectively. If a habitat under consideration could aggregate multiple @kappel2012 habitats, we instead used the average score of these habitats to obtain a vulnerability score. Vulnerability scores assigned to floodplains were thus obtained based on the average of the vulnerability scores assigned to rocky intertidal, intertidal, and salt marsh environments in the @kappel2012 matrix. Similarly, the government and research shipping category includes a significant diversity of vessel types such as military craft, patrol vessels, and scientific research vessels. The vulnerability of habitats to this type of shipping was thus assessed using the average of the scores assigned to shipping, military, research, and scientific sampling from @kappel2012. Finally, for the 2023 update, the vulnerability scores for eelgrass and algal zones were used for aquatic grass bed habitats.
Some modifications were also made to the vulnerability scores reported by @kappel2012. First, the vulnerability values assigned to wetlands were modified to appropriately capture the fineness of the dataset available for our study. The characterized wetlands reflect the tiered nature of wetlands, with shallow water wetlands, followed by marshes, then swamps, with the potential exposure of these environments to marine-based stressors gradually decreasing. We thus decreased the vulnerability of wetlands to all stressors proportionally by a factor of 5% for marshes and 10% for swamps. We retained the @kappel2012 assessment for the wetland habitat category containing wetlands of unknown type as a precaution. Second, accidental spills were categorized to differentiate hydrocarbon spills from other types of spills, meaning that we decrease the spill vulnerability values by a factor of 10% for other spills. Third, the vulnerability scores for aquatic grass beds was decreased by 10% per biovolume category under the assumption that higher biovolumes are more important, and thus have a higher vulnerability: high biovolume (100%), moderate biovolume (90%), and low biovolume (80%). As with the wetlands, we kept the vulnerability value intact for spills of unknown content as a precaution. Lastly, a relative score was estimated by normalizing the total scores by the maximum observed vulnerability value among the habitats considered (i.e. 3.6) to obtain a score ranging from 0 to 1. The resulting vulnerability matrix can be consulted in Table \@ref(fig:habitatvuln).
knitr::include_graphics("./figures/figures-vulnerability/habitat.png")
Habitats of importance to wildlife and plant species
The vulnerability of habitats identified for their importance to life cycles and special status species cannot be assessed using the same approach used for natural environments. These sites differ from the habitats presented above in that they are environments whose characteristics may vary, but which perform essential functions for the maintenance of certain species, or are deemed important for the conservation of certain special status species. As such, the vulnerability of these sites cannot be assessed in the same way as the assessment presented in the previous section. An alternative approach using species-based criteria rather than habitat-based criteria was used to assess their relative vulnerability to the effects of the stressors under consideration.
Exercises to assess species vulnerability to the effects of stressors are less common in cumulative effects assessments, although some assessment processes do exist [e.g. @maxwell2013; @trew2019; @ohara2021]. These assessments primarily involved marine megafauna species (e.g. whales and sharks) and species at risk. However, these publications do not cover all species and stressors included in this assessment. They do provide a set of criteria for assessing species vulnerability to stressors. We used a similar process to determine the vulnerability to marine vessel activities of environments of importance to species in the St. Lawrence and Saguenay rivers.
The criteria used to assess species vulnerability to the effects of marine vessel activities are taken from @maxwell2013: 1) the frequency of stressors, 2) whether the effect of the stressor is direct or indirect, 3) the resistance of the species to the effects of the stressor, 4) the recovery time of the species, 5) the relative effect on reproduction, and 6) the relative effect on the population (Table \@ref(tab:vulnerabilitySpecies)). We added two criteria to this list in order to consider regional particularities specific to the St. Lawrence and Saguenay rivers, namely 7) the status of the species and 8) whether the species is resident in the study area (Table \@ref(tab:vulnerabilitySpecies)).
meta <- read_yaml("../data/data-metadata/vulnerability_mammiferes_marins.yml") criteres <- meta$criteres tab <- list() lin <- numeric() for(i in 1:length(criteres)) { dat <- meta$details_en[[criteres[i]]] nm <- character(length(dat$description$rank)) nm[1] <- glue("***{dat$name}***") nm[2] <- glue("*{dat$question}*") nm[3] <- ifelse(!is.null(dat$specification), dat$specification, "") crit <- as.data.frame(dat$description) tab[[i]] <- cbind(nm,crit) lin[i] <- length(nm) } lin <- cumsum(lin) tab <- dplyr::bind_rows(tab) # ----- tab %>% knitr::kable( col.names = c("Vulnerability criteria", "Rank", "Categories", "Description", "Examples"), row.names = FALSE, caption = "Criteria considered for assessing the vulnerability of valued components with regard to species – i.e. - important environmental conditions for life cycles, special status species and marine mammals – to the stressors considered for the CEA of marine vessel activities in the St. Lawrence and the Saguenay river. The criteria were adapted from @maxwell2013, to which species special status and residence criteria were added in order to account for the regional particularities of the St. Lawrence and the Saguenay river." ) #%>% #kableExtra::row_spec(lin, extra_css = "border-bottom: 2px solid")
The criteria used varied depending on the habitat types assessed. The frequency of environmental stressors was used for all habitats of importance to the species. The vulnerability of spawning and nursery sites considered the frequency of environmental stressors and the effects on reproduction and population. The vulnerability of habitats of importance to seabirds was assessed by considering the frequency of stressors and the effects on populations. Lastly, the vulnerability of habitats of importance to special status species was assessed using the frequency of environmental stressors, effects on populations, and species status. It should be noted that the vulnerability of sites of importance to special status wildlife and plant species was considered in the same way. Although wildlife and plant species respond differently to environmental stressors, our assessment uses occurrences of species at risk as a proxy for vulnerable environments rather than a species-specific assessment. The general characterization used did not allow us to differentiate between species, so we focused on criteria specific to species status.
For each combination of valued component $CV_i$ and stressor $S_j$, the final vulnerability score $\mu_{CV_i, S_j}$ is obtained from the sum of the scores for each criterion:
$$mu_{CV_i, S_j} = \sum_{k \in K} C^i_{j,k}$$
where $C^i_{j,k}$ is the value of the valued component $i$ for the stressor $j$ and the criterion $k$, and $K$ are the set of criteria considered to assess the vulnerability of the valued component $i$.
Lastly, a relative score was assessed by normalizing the total scores by the maximum observed vulnerability value per habitat group assessed using the same criteria. Spawning and nursery sites were the first group, seabirds the second, and special status species the third. The final vulnerability scores are available in Table \@ref(fig:speciesvuln) and the individual criteria assessment tables per habitat category are available in Appendix 8.
knitr::include_graphics("./figures/figures-vulnerability/faune_flore.png")
R Package Documentation
Browse R Packages
We want your feedback!
Note that we can't provide technical support on individual packages. You should contact the package authors for that.
Embedding an R snippet on your website
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.