Fisheries management impacts on target species status

Abstract

Fisheries management systems around the world are highly diverse in their design, operation, and effectiveness at meeting objectives. A variety of management institutions, strategies, and tactics are used across disparate regions, fishing fleets, and taxonomic groups. At a global level, it is unclear which particular management attributes have greatest influence on the status of fished populations, and also unclear which external factors affect the overall success of fisheries management systems. We used expert surveys to characterize the management systems by species of 28 major fishing nations and examined influences of economic, geographic, and fishery-related factors. A Fisheries Management Index, which integrated research, management, enforcement, and socioeconomic attributes, showed wide variation among countries and was strongly affected by per capita gross domestic product (positively) and capacity-enhancing subsidies (negatively). Among 13 management attributes considered, three were particularly influential in whether stock size and fishing mortality are currently in or trending toward desirable states: extensiveness of stock assessments, strength of fishing pressure limits, and comprehensiveness of enforcement programs. These results support arguments that the key to successful fisheries management is the implementation and enforcement of science-based catch or effort limits, and that monetary investment into fisheries can help achieve management objectives if used to limit fishing pressure rather than enhance fishing capacity. Countries with currently less-effective management systems have the greatest potential for improving long-term stock status outcomes and should be the focus of efforts to improve fisheries management globally.

Keywords: fisheries enforcement; fishery subsidies; marine conservation; resource management; stock assessment.

Fig. 1.

Summarized survey answers by dimension and country. Responses are weighted by both respondent expertise and confidence in individual answers provided, and are adjusted for observed differences among respondent background categories. Countries (n = 28) are sorted by FMI values, a composite of research, management, enforcement, and socioeconomics dimensions.

Fig. S1.

Summary statistics of survey answers by dimension. Responses for individual questions within dimensions have: (A) weighting by confidence score in the answers provided for individual questions; or (B) equal weighting. The FMI is a composite of research, management, enforcement, and socioeconomics dimensions with equal weighting. Diagonals show histograms of survey responses (n = 191) across all countries and respondents. Lower panels show scatterplots between pairs of dimensions, with correlation ellipses (black lines), loess smoothers (red lines), and bivariate medians (red circles) overlaid. Upper panels show Pearson correlation coefficients between pairs of dimensions with font size scaled to r.

Fig. S2.

Summarized survey answers and FMI values by country. FMI is a composite of research, management, enforcement, and socioeconomics dimensions with equal weighting. (A) Circles show mean FMI values of individual respondents’ answers, weighted by the confidence score provided for individual questions, and adjusted for respondent background category using the random effect conditional mode estimates shown in Fig. S5B. The grand mean across respondents for each country is shown by “x” symbols, weighed by the level of expertise of respondents. (B) Unweighted and unadjusted summarized survey answers by dimension and country. Similar summaries that were weighted by both level of expertise of respondents and confidence in the answers provided for individual questions and that were adjusted for the background category of respondents were shown in Fig. 1.

Fig. 2.

Influences of country-level factors on FMI values. Data points show FMI values for each respondent and country (n = 191 surveys), weighted by confidence in individual answers provided. Black lines show overall best fits with 95% confidence bands in gray, weighted by respondent expertise. Best-fit lines for respondent background categories are overlaid. Panels are sorted left to right by absolute values of t-statistics for predictor variables.

Fig. S3.

Mosaic plot of respondent background category and self-identified level of familiarity with the country’s fisheries. Bar widths are proportional to the number of respondents in each background category, ranging from 16 to 60 surveys (total n = 191 surveys completed by 182 individuals). Shading denotes levels of familiarity (A = expert, B = strong familiarity, C = moderate familiarity, D = limited familiarity).

Fig. S4.

Values by country of 12 numerical covariates considered in analysis. See Table S1 for a description of each variable including units. nei, not elsewhere included.

Fig. S5.

Effect sizes of predictor variables on logit-transformed FMI values. Responses are weighted by both level of expertise of respondents and confidence in the answers provided for individual questions. (A) Estimated coefficients of standardized numerical fixed-effect covariates are shown with SDs (thicker bars) and 95% confidence limits (thinner bars). (B) Conditional modes of respondent background categories, treated as random intercepts, are shown with estimated SEs. Values above data points in B show back-transformed FMI values on the linear scale calculated at mean values of fixed-effect covariates. Descriptions of covariates in A are given in Table S1. nei, not elsewhere included.

Fig. S6.

Standardized effects of country-level numerical covariates on logit-transformed FMI values. FMI responses are weighted by both level of expertise of respondents and confidence in the answers provided for individual questions. All predictions (red lines) intersect [0,0] and show the change in logit(FMI) associated with a change in a standardized numerical covariate after accounting for effects of other predictors including random intercepts for respondent background category. Partial residuals and 95% confidence bands around predictions are shown. Overlap of confidence bands with dashed lines at 0 suggests no significant effect (α = 0.05) of the covariate on the FMI. Panels are sorted left to right by absolute values of t-statistics for predictor variables. nei, not elsewhere included.

Fig. S7.

FMI values calculated under different weighting schemes and adjustments. FMI is a composite of research, management, enforcement, and socioeconomics dimensions with equal weighting. In all panels, values along horizontal axis are unweighted and not adjusted for respondent background category. (A) Values along vertical axis are weighted means of answers to individual questions within each dimension, weighted by the confidence scores provided for each answer (n = 191 surveys, r = 0.999). (B) Values along vertical axis are weighted by confidence scores in the answers provided for individual questions, by the self-identified level of familiarity of respondents with the country’s fisheries (“expertise”), or by both (n = 28 countries, all r > 0.998). Confidence scores and respondent expertise were collected as qualitative categories (A, B, C, D) and assigned values of: A = 1.0, B = 0.8, C = 0.6, D = 0.4. (C) Values along vertical axis are adjusted for respondent background category using the random effect conditional mode estimates shown in Fig. S5B.

Fig. 3.

Effects of fisheries management attributes in research (R), management (M), enforcement (E), and socioeconomics (S) dimensions on the current status and trends of biomass (B) and fishing mortality (F). Higher values of response variables indicate increasingly desirable states or trends toward desirable states with respect to management targets (i.e., high values of F do not indicate F > F_MSY, but rather F ≤ F_MSY). Line thicknesses are proportional to predictor variable importance scores from random forest analyses for each response variable. Panels are sorted left to right by the sum of standardized variable importance scores across all four response variables. Response and predictor variables are weighted by confidence in individual answers. All variables range from 0 to 1, but vertical axes are truncated. Rug marks at bottom show deciles of predictor variable values.

Fig. 4.

Comparison of country FMI or stock status values with published indices of fisheries management or status. (A) EBFM overall performance [Pitcher et al. (19)] (r = 0.68), overall management effectiveness in EEZs [Mora et al. (20)] (r = 0.44), and OHI–Food Provision from Wild Capture Fisheries [Halpern et al. (21)] (r = 0.15) compared with FMI, by country. (B) OHI–Fishery Status (21) (r = 0.07), and EPI–Fish Stocks, using Sea Around Us Project estimates (22) (r = 0.21, inverted such that greater values suggest lower percent overfished stocks) compared with stock status values (n = 28). Best-fit lines are shown for each comparison.

Fig. S8.

Detailed comparison and overlap of individual criteria from the present study with criteria or survey questions from previous studies (19, 20). Lines join related criteria/questions between the present study (in the center) and previous studies (on both sides); thick black lines show strong overlap and thin gray lines show weaker overlap between paired criteria. Phrasing of criteria is abbreviated in all studies. EBM, ecosystem-based management.