Methods

Study Communities

Figure 1. Map of study communities.

Results

Saint Paul

Digraph

Figure 2. Digraph showing relationships (edges) between variables (nodes) for the St. Paul community model.

Community matrix

Figure 3. Representation of the Saint Paul community matrix in which blue cells indicate a negative effect of column node on row node, and orange indicate a positive effect. Diagonals in blue indicate that each node in the system is assumed to be self-limiting.

Press perturbations

Figure 4. Impact barpot of simulated responses to an increase in halibut abundance for St. Paul community model. The proportion of simulated stable (retained) matrices that had a negative response at the node in question is shown in blue, and the proportion of stable matrices in which that node had a positive response is shown in orange.

Figure 5. Impact barpot of simulated responses to a decrease in halibut bycatch limits for St. Paul community model. The proportion of simulated stable (retained) matrices that had a negative response at the node in question is shown in blue, and the proportion of stable matrices in which that node had a positive response is shown in orange.

Impact table

Figure 6. Mean simulated effect on row variable of positive perturbation of column variable.

Node similarity across perturbations

Figure 7. Dendrogram showing node similarity across perturbations (Euclidean distance). The more similarly a pair of nodes responds to a press perturbation of any kind, the closer they will be grouped in the dendrogram.

Perturbation similarity

Figure 8. Dendrogram showing perturbation similarity (Euclidean distance). The more similarly a pair of nodes affects all other nodes when perturbed (positively?), the closer they will be grouped in the dendrogram.

Savoonga

Digraph

Figure 9. Digraph showing relationships (edges) between variables (nodes) for the Savoonga community model.

Community matrix

Figure 10. Representation of the Savoonga community matrix in which blue cells indicate a negative effect of column node on row node, and orange indicate a positive effect. Diagonals in blue indicate that each node in the system is assumed to be self-limiting.

Press perturbations

Figure 11. Impact barpot of simulated responses to an increase in halibut abundance for Savoonga community model. The proportion of simulated stable (retained) matrices that had a negative response at the node in question is shown in blue, and the proportion of stable matrices in which that node had a positive response is shown in orange.

Figure 12. Impact barpot of simulated responses to a decrease in halibut bycatch limits for Savoonga community model. The proportion of simulated stable (retained) matrices that had a negative response at the node in question is shown in blue, and the proportion of stable matrices in which that node had a positive response is shown in orange.

Impact table

Figure 13. Mean simulated effect on row variable of positive perturbation of column variable.

Node similarity across perturbations

Figure 14. Dendrogram showing node similarity across perturbations (Euclidean distance). The more similarly a pair of nodes responds to a press perturbation of any kind, the closer they will be grouped in the dendrogram.

Perturbation similarity

Figure 15. Dendrogram showing perturbation similarity (Euclidean distance). The more similarly a pair of nodes affects all other nodes when perturbed (positively?), the closer they will be grouped in the dendrogram.

Unalaska/Dutch Harbor

Digraph

Figure 16. Digraph showing relationships (edges) between variables (nodes) for the Unalaska/Dutch Harbor community model.

Community matrix

Figure 17. Representation of the Unalaska/Dutch Harbor community matrix in which blue cells indicate a negative effect of column node on row node, and orange indicate a positive effect. Diagonals in blue indicate that each node in the system is assumed to be self-limiting.

Press perturbations

Figure 18. Impact barpot of simulated responses to an increase in halibut abundance for Unalaska/Dutch Harbor community model. The proportion of simulated stable (retained) matrices that had a negative response at the node in question is shown in blue, and the proportion of stable matrices in which that node had a positive response is shown in orange.

Figure 19. Impact barpot of simulated responses to a decrease in halibut bycatch limits for Unalaska/Dutch Harbor community model. The proportion of simulated stable (retained) matrices that had a negative response at the node in question is shown in blue, and the proportion of stable matrices in which that node had a positive response is shown in orange.

Impact table

Figure 20. Mean simulated effect on row variable of positive perturbation of column variable.

Node similarity across perturbations

Figure 21. Dendrogram showing node similarity across perturbations (Euclidean distance). The more similarly a pair of nodes responds to a press perturbation of any kind, the closer they will be grouped in the dendrogram.

Perturbation similarity

Figure 22. Dendrogram showing perturbation similarity (Euclidean distance). The more similarly a pair of nodes affects all other nodes when perturbed (positively?), the closer they will be grouped in the dendrogram.

Amendment 80

Digraph

Figure 23. Digraph showing relationships (edges) between variables (nodes) for the Amendment 80 community model.

Community matrix

Figure 24. Representation of the Amendment 80 community matrix in which blue cells indicate a negative effect of column node on row node, and orange indicate a positive effect. Diagonals in blue indicate that each node in the system is assumed to be self-limiting.

Press perturbations

Figure 25. Impact barpot of simulated responses to an increase in halibut abundance for Amendment 80 community model. The proportion of simulated stable (retained) matrices that had a negative response at the node in question is shown in blue, and the proportion of stable matrices in which that node had a positive response is shown in orange.

Figure 26. Impact barpot of simulated responses to a decrease in halibut bycatch limits for Amendment 80 community model. The proportion of simulated stable (retained) matrices that had a negative response at the node in question is shown in blue, and the proportion of stable matrices in which that node had a positive response is shown in orange.

Impact table

Figure 27. Mean simulated effect on row variable of positive perturbation of column variable.

Node similarity across perturbations

Figure 28. Dendrogram showing node similarity across perturbations (Euclidean distance). The more similarly a pair of nodes responds to a press perturbation of any kind, the closer they will be grouped in the dendrogram.

Perturbation similarity

Figure 29. Dendrogram showing perturbation similarity (Euclidean distance). The more similarly a pair of nodes affects all other nodes when perturbed (positively?), the closer they will be grouped in the dendrogram.