Fig 1.
Schematic overview of the research approach utilized in this study.
Fig 2.
Location of household surveys in Kenya and Tanzania.
Yellow circles indicate locations where household survey data was collected.
Table 1.
Parameter values used in iterative optimization of classification models and final values used for each parameter for each country-specific model.
Table 2.
Gender-specific number of surveys (N) from each site in Kenya and Tanzania that were included in analyses.
The “% Effective” column represents the number of female or male respondents from each site that indicated that they believed management was effective.
Fig 3.
Relative importance and distribution of responses for indicators most strongly associated with differences in perceived management effectiveness in Kenya.
(A) Relative importance of features to differences in perceived effectiveness. The length of solid bars along the x-axis represents the average feature importance for each variable across 100 model iterations, and the color of bars indicates relationships between increases in each metric and more positive perceptions of management effectiveness (blue: positive; red: negative; gray: variable). Error bars represent the standard deviation of importance across model runs. (B) Density plots of individual responses for the top 10 most important metrics, separated by individuals that believed management was effective (yellow) and not effective (purple). Brown areas represent overlapping values between individuals that believed management was effective and not effective.
Fig 4.
Distributions and interactions between indicators with the strongest relationship to differences in perceived management effectiveness in Kenya.
Density plots demonstrating the relative proportion of individuals responding that they believed management was either not effective (left) or effective (right) across different values of: (A) the top three most important indicators (WAMR: Women’s Access to Marine Resources; TM: Trained in Management; WNG: Women’s Nature Groups); (B-D) two-way interactions between the top three indicators; and (E) three-way interactions between all indicators (x-axis: (NE) not effective; (E) effective). Numbers on either side of density plots represent the total number of individuals for each feature or combination of features that believed management was effective (right side) or not effective (left side).
Fig 5.
Relative importance and distribution of responses for indicators most strongly associated with differences in perceived management effectiveness in Tanzania.
(A) Relative importance of features to differences in perceived effectiveness. The length of solid bars along the x-axis represents the average feature importance for each variable across 100 model iterations, and the color of bars indicates relationships between increases in each metric and more positive perceptions of management effectiveness (blue: positive; red: negative; gray: variable). Error bars represent the standard deviation of importance across model runs. (B) Density plots of individual responses for the top 10 most important metrics, separated by individuals that believed management was effective (yellow) and not effective (purple). Brown areas represent overlapping values between individuals that believed management was effective and not effective.
Fig 6.
Distributions and interactions between indicators with the strongest relationship to differences in perceived management effectiveness in Tanzania.
Density plots demonstrating the relative proportion of individuals responding that they believed management was either not effective (left) or effective (right) across different values of: (A) the top three most important indicators (IFL: Influence on Governance; IFO: Frequency of Information from Governing Bodies; EDU: Education Level); (B-D) two-way interactions between the top three indicators; and (E) three-way interactions between all indicators (x-axis: (NE) not effective; (E) effective). Numbers on either side of density plots represent the total number of individuals for each feature or combination of features that believed management was effective (right side) or not effective (left side).