Evaluating Indicators for Assessing Marine Protected Area Effects on Conservation and Catches

Evaluating common indicators for assessing marine 1 protected area effects on conservation and catches 2 Daniel Ovando 3 Cori Lopazanski 4 1 Title Page5 Manuscript title: Evaluating common indicators for assessing marine protected area effects on 6 conservation and catches7 Short Title: Evaluating common indicators of MPA effects8 Authors: Daniel Ovando1* & Cori Lopazanski2,3 9 1 Inter-American Tropical Tuna Commission, Ecosystem & Bycatch Group, 8901 La Jolla Shores 10 Drive, La Jolla, CA, 92102, USA11 2 Duke University Marine Laboratory,Nicholas School of the Environment Beaufort, NC, 28516, 12 USA13 3 University of California Santa Barbara, Bren School of Environmental Science & Management, 14 Santa Barbara, CA, 93106, USA15 *Corresponding author: [email protected]16 Open Research statement: No data were collected for this study. This submission uses novel code, 17 which is provided, per our requirements, in an external repository to be evaluated during the peer 18 review process. The link to this external repository is https://github.com/DanOvando/mpa-effects-19 and-indicators. A pre-print archive of the code along with all results needed to reproduce this paper 20 is also available at Zenodo https://doi.org/10.5281/zenodo.17308920 . This Zenodo repository will 21 be upgraded to a permanent archive of the final results and paper upon publication.22 Key Words: Marine Protected Areas, Conservation effectiveness, Bio-economic modeling, Pro23 tected area evaluation, Fisheries management24 2 Abstract 824 Marine protected areas (MPAs) are spatial management tools designed to ideally benefit diverse 825 conservation and social objectives, including biodiversity, food security, and climate resilience. 826 Because these objectives are often difficult to measure directly, evaluations of MPAs typically 827 rely on simplified empirical indicators—such as biomass inside versus outside MPAs, or gradients 828 in biomass outside boundaries—as proxies for broader MPA effects. Yet in complex social– 829 ecological systems, even well-measured indicators may fail to capture the true causal effects of 830 MPAs. 831 Using a large-scale simulation framework, we explored variability in MPA effects and assessed 832 whether common indicators reliably tracked intended objectives of protected areas. We show that 833 similarly sized MPAs can produce vastly different effects for conservation (generally positive) and 834 fishery catches (more frequently negative) depending on factors such as fishing fleet dynamics 835 and habitat heterogeneity within and across species. The degree of correlation between common 836 empirical indicators of MPA performance and the true effects of an MPA also varied widely. 837 Indicators such as Biomass Density Response Ratios were positively correlated with the effect of 838 MPAs on biomass inside their borders (Spearman’s 𝜌 = 0.56), but slightly negatively correlated 839 with the effect of MPAs on total fishery catches ( 𝜌 = -0.07). Common indicators of spillover, such 840 as gradients in biomass density near MPAs relative to far from MPAs, had little correlation with 841 any simulated conservation or catch outcome (absolute value of Spearman’s 𝜌all ≤0.2). 842 The high variability in simulated MPA effects highlights the importance of effective ecological 843 and economic monitoring programs. We found that many indicators currently measured in and 844 around MPAs may contain little information on macro-level effects such as total changes in fish 845 populations or fishery catches. The next generation of MPA research needs interdisciplinary 846 research to develop practical and reliable methods for tracking the effects of MPAs. Doing so will 847 help ensure that the planned rapid and global expansion of MPAs has the best chance of delivering 848 positive and equitable outcomes for nature and people. 849 45 Introduction 25 Various forms of spatial management have been used to conserve and manage marine ecosystems 26 across cultures throughout human history (e.g. Johannes 2002). The 1990s marked a period of 27 expanded interest in both the science and use of spatial management tools, specifically the general 28 concept of “marine protected areas” (MPAs) (Carr et al. 2019; Humphreys and Clark 2020). MPAs 29 are areas of the ocean where human activities are restricted or prohibited to promote biodiversity 30 conservation and potentially achieve other complementary objectives (Grorud-Colvert et al. 2021). 31 These objectives can include preserving cultural heritage, rebuilding populations inside protected 32 borders, providing conservation benefits outside in fished waters, and supporting surrounding 33 fisheries (Gaines et al. 2010). International movements such as the Montreal–Kunming Global 34 Biodiversity Framework’s call for protection of 30% of the oceans, and the Agreement under 35 the United Nations Convention on the Law of the Sea on the Conservation and Sustainable Use 36 of Marine Biological Diversity of Areas Beyond National Jurisdiction (the High Seas Treaty or 37 BBNJ) have provided substantial momentum for the expanded use of MPAs in the world’s oceans 38 in pursuit of these goals. 39 Over 9% of the ocean is now covered by MPAs (as of October 2025; World Database of Protected 40 Areas), and the recent global expansion of MPAs has been matched by growing efforts to empir41 ically evaluate their performance (e.g. Osenberg et al. 2011; Edgar et al. 2014; Di Lorenzo et 42 al. 2020; Lynham and Villaseñor-Derbez 2024; Hopf et al. 2024) . Since many of the benefits 43 that MPAs are expected to deliver are difficult to measure directly, evaluations of MPA effects 44 often rely on tracking simpler indicators as proxies for broader MPA effects (Pelletier 2011; Hopf 45 et al. 2024). These indicators enable assessments to be carried out across diverse contexts, and 46 have become a foundational tool for attempting to understand MPA effects in both scientific and 47 policy settings. Many indicators compare attributes within MPAs to nearby unprotected areas 48 [i.e., response ratios; Smith et al. (2024); Lester et al. (2009)], or from unprotected areas that are 49 near to MPA borders compared to those farther away (i.e., gradients, Halpern, Lester, and Kellner 50 3 2009). More causally-robust approaches incorporate measurements both before and after MPA 51 implementation [i.e., before-after-control-impact or BACI; Medoff, Lynham, and Raynor (2022); 52 Lynham and Villaseñor-Derbez (2024); Kerr, Kritzer, and Cadrin (2019); Ovando et al. (2021)].53 While these empirical indicators differ in their complexity, each is presumably intended to deter54 mine the effect of the MPA on a particular observed metric: higher biomass in MPAs compared 55 to unprotected areas is interpreted as conservation gains within MPA borders (Smith et al. 2024; 56 Lester et al. 2009), while gradients in fisheries catch or effort are thought to signal evidence of 57 spillover of adult or larval fish from the MPA to surrounding waters (Halpern, Lester, and Kellner 58 2009; Medoff, Lynham, and Raynor 2022; Lynham and Villaseñor-Derbez 2024; Roberts et al. 59 2001). However, these indicator results are sometimes then assumed to also imply evidence for an 60 unobserved effect, such as changes in total population size or total fisheries catch.61 The assumption that specific indicators can stand in for broader and harder to observe ecological 62 or social effects has not been to our knowledge comprehensively tested, despite their widespread 63 use (though works such as Hopf et al. (2024), Kerr, Kritzer, and Cadrin (2019), and Hilborn et al. 64 (2024) evaluated specific combinations of indicators and effects). Understanding the reliability of 65 indicators, and the variability of the underlying MPA effects they aim to measure, is critical for 66 interpreting existing evaluations, designing effective monitoring strategies, and making evidence67 informed policy decisions.68 As MPA usage expands around the world, establishing expectations around potential MPA effects 69 and our ability to accurately measure them is essential to tracking progress towards outcomes 70 achieved, not just area protected. To that end, this paper addresses two questions. We first anal71 ysed whether MPA effects on conservation and catches are likely to be variable enough to justify 72 the need for empirical monitoring. We then tested the ability of several commonly used empirical 73 indicators to reflect the actual effects of MPAs. We found that MPAs can have highly variable 74 effects on conservation and catches depending on the specific dynamics of the system in question, 75 and that while some indicators reliably track this variability in some MPA effects, many do not.76 4 Methods 77 MPA Experiment Overview 78 We used a simulation framework to model both the variability of MPA effects and the performance 79 of empirical MPA indicators. The general structure of this simulation experiment is as follows. 80 We created a fishery system with a set of bio-economic traits. We then ran an MPA experiment 81 on that system by generating two paired simulations identical in every way, except that one 82 system contained a no-take MPA and the other did not. We then calculated the simulated effects 83 of the MPA on a range of objectives based on the differences between the simulation with an 84 MPA relative to the paired simulation without an MPA. Lastly, we calculated common empirical 85 indicators of MPA performance from the simulation with the MPA, and compared these indicators 86 to the true simulated effects. We then repeated this process multiple times with different randomly 87 generated fishery states. The details of this process are explained below and illustrated in Figure 1. 88 Simulating Fisheries 89 Fisheries systems were simulated using the marlin model presented in Ovando et al. (2023) and 90 Ovando (2025). marlin simulates a user-specified number of age-structured fish populations fished 91 by a user-specified number of fishing fleets, all operating in a two-dimensional space. This spatial 92 domain represents the surface area of the seascape, broken up into a series of “patches”, each 93 with a defined spatial area. For this paper, we modeled a 21 x 21 grid of patches (for a total of 94 441 patches), where each patch had a surface area of 25KM 2 (5 km x 5km), for a total simulated 95 seascape area of 11,025 KM2.96 To explore MPA effects and indicators across a range of species, we modeled four different species 97 archetypes made from a set of fixed and variable parameters. These simulated species were broadly 98 based on a tuna (Thunnus albacares), shark (Carcharhinus falciformis), grouper (Epinephelus 99 fuscoguttatus), and reef fish (Seriola quinqueradiata).Fixed parameters included those relating to 100 5 Figure 1: Conceptual illustration of the process of simulating effects of MPAs, calculating empirical indicators, and comparing indicators to effects. Colored polygons in Step 2 indicate habitat distributions for each simulated species. See Methods section for detailed explanations of these steps. 6 growth, mortality, fecundity, recruitment, and movement, with values for each species based on 101 available literature and our own judgement (Table 1). We fixed these parameters in order to not 102 create unrealistic combinations of for example growth and mortality (Prince et al. 2015). 103 One of the key features of this model is the ability to simulate fish movement across the two104 dimensional seascape through both passive diffusion and active taxis in response to habitat gra105 dients, in the manner of J. T. Thorson et al. (2021) (see Ovando (2025)). Empirical estimates of 106 movement rates in the same units as those used in our simulation model were not readily available. 107 As such, we fixed movement rates of recruits and post-recruit fish such that 95% of animals were 108 within a given linear distance of their point of origin after one year of diffusion and taxis. We 109 grouped these distances into low (2.5km), medium (25km), or high (250km) dispersal distances, 110 and assigned these dispersal distances to the recruit and post-recruit life stages of each of our simu111 lated species based on our best judgement and a desire to reflect a range of movement dynamics 112 (e.g. sedentary adults and highly dispersed larvae, and vice versa) (Table 1). 113 For each simulation we augmented these fixed parameters with random draws of variable parame114 ters that describe other less-certain aspects of the simulation. These include the ecological traits 115 such as the uniformity of species-level habitat across the simulated seascape, the correlation in 116 habitat across species, seasonal movement shifts, spawning aggregations, the timing of density 117 dependence, the degree of recruitment deviation, and the temporal and cross-species correlations 118 in recruitment deviates (Table 2). Variable traits also included fishery characteristics such as the 119 magnitude of fishing pressure on each species, the economic value of each species for each fleet, 120 the presence of fishing ports affecting spatial fishing choices, and the selectivity of each fishing 121 fleet for each fished species (Table 2). 122 While all traits can have an impact on simulated effects, spatial variables are particularly important 123 for MPA effects, so we explain them in more detail here. The simplest assumption one can make in 124 an MPA model is that all else being equal fish biomass and fishing pressure are constant in space 125 and time, baring a policy intervention. Under these circumstances, in the absence of an MPA, fish 126 7 biomass and fishing effort are always the same in each patch. Following MPA implementation 127 in such a system, any differences in biomass or fish catches in any patch must be due to the MPA 128 itself.129 In reality, fish populations can be heterogeneously distributed in space at any given time step, 130 and their species distributions can exhibit positive, negative, or no correlations with other fished 131 species in the region. These spatial dynamics of species distributions are so pronounced in marine 132 systems that an extensive literature of Species Distribution Models (SDMs) has developed to 133 describe and model them (J. T. Thorson and Barnett 2017; Karp et al. 2025; Brodie et al. 2021) 134 (Lopez et al. 2024). This means that an MPA placed on core habitat can have very different effects 135 from a equally sized MPA placed on marginal habitat. Similarly, given that not all species have 136 the same habitat distributions, an MPA placed on the core habitat of one species may be protecting 137 marginal habitat of another.138 Following methods described in Ovando et al. (2023) and Ovando (2025), for each species within 139 each simulation we randomly set a parameter 𝜅 that describes the intrinsic heterogeneity of their 140 habitat across the simulated seascape; lower values of 𝜅 result in a smoother habitat distribution, 141 higher values of 𝜅 result in a more patchy habitat distribution. At the same time, we randomly 142 generated a habitat correlation matrix between each of the species in that seascape in a given 143 simulation. Using the methods described in Ovando (2025) based on J. T. Thorson and Barnett 144 (2017), these 𝜅 values for each species, along with the correlation matrix among all species, 145 are used to construct a covariance matrix. which in turn is used to simulate habitat in space as 146 a function of a multivariate normal distribution. We used this multivariate normal distribution 147 to generate random draws of species distributions that can vary across simulations both in their 148 heterogeneity and in their correlation across species, with for example some simulations resulting 149 in reef fish and groupers having pathcy and highly positively correlated habitats, and in others 150 smooth and highly negatively correlated habitats.151 Along with habitat heterogeneity, our model accounts for some of the complex behaviors of fishing 152 8 extreme levels of depletion are rare in marine fish populations and produce extreme results when 245 considering MPA effects on a percentage based scale. Similarly, we removed any simulations 246 in which any species had total biomass values less than one, or any fishing fleet had catches less 247 than one. These simulations were removed since they resulted in improbably high percentage 248 effect sizes (e.g. increases in catches of of 900% when catches increased from 0.1 MT to 1 MT). 249 Post-filtering, the combination of each simulated fishery with each MPA design resulted in a total 250 of 10,896 unique MPA experiments. 251 Calculating Empirical Indicators 252 The prior steps of simulating a fishery system under a range of MPA experiments create a distri253 bution of causal effects of MPAs on the selected metrics. We compared these true effects with 254 estimates from common empirical indicators of MPA performance. To isolate the fundamental 255 performance of the indicator from questions around data quality and bias, all indicators were 256 calculated using the simulated data without any observation error. 257 We evaluated six empirical indicators commonly used to assess MPA performance; Biomass 258 Density Before-After-Control-Impact ( BACI), Biomass Density Response Ratio,Mean Length 259 Response Ratio,Effort Gradient,Biomass Density Gradient,Biomass Density Before-After Gradi260 ent (Table 3). These indicators vary in their underlying response variable (biomass density, mean 261 length, or fishing effort), their spatial design (inside-outside or nearfar), and whether they include 262 temporal comparisons (before-after vs. after-only). These indicators were selected to represent a 263 range of approaches described in the empirical MPA literature to infer conservation and fisheries 264 effects. A full description of each indicator and its stylized equation is provided in Table 3.265 All models were fit using a log-normal distribution, with the dependent variable measured at the 266 resolution of species, patch, and time step (and by fleet where applicable). Coefficients from these 267 models can be interpreted as multiplicative effects; effect sizes are reported as percentage changes 268 using the transformation 100 × (𝑒𝛽𝑀𝑃𝐴 − 1) . All indicators were calculated at the species level, 269 15 as results were robust to aggregating biomass or catch across species (Figure S30).270 Indicators based on inside-outside comparisons (e.g., Biomass Response Ratio, Biomass BACI, 271 Mean Length Response Ratio) require the selection of a comparable reference site outside the 272 MPA. In empirical studies, reference sites are often selected to match the habitat and environ273 mental characteristics within the MPA (Chapman and Kramer 1999; Claudet and Guidetti 2010; 274 Osenberg et al. 2011; Smith et al. 2024). We approximated this reference site selection approach 275 by controlling for unfished biomass per patch and weighting regressions by distance from the MPA 276 border. Specifically, we calculated the linear distance of each patch from the nearest MPA border 277 and weighted models such that the the model pays the most attention to patches deepest inside the278 MPA and those farthest outside the MPA.279 Gradient (near-far) indicators (e.g., Effort Gradient, Biomass Gradient, Biomass Before-After 280 Gradient) compare metrics outside of MPAs based on distance from MPA borders. Following the 281 general approach of Medoff, Lynham, and Raynor (2022) and Lynham and Villaseñor-Derbez 282 (2024), we calculated the linear distance from each patch from the MPA border, and assigned 283 patches in the lowest 20th percentile of distance as Near and those in the top 80th percentile as 284 Far.Aswithinside-outside style analyses, we also controlled for unfished biomass per species per 285 patch.286 All models were fit using linear models in R (R Core Team 2024). For each indicator, variables 287 were measured without error at the resolution of spatial patch pand time step t. The spatial reso288 lution of the system is 21 by 21 patches, meaning that any given time step has 441 patches. For 289 models fit to a single time step then (e.g. Biomass Density Response Ratio), the number of data 290 points used to fit the model were N = 441. For models fit to multiple time steps of before and after, 291 (e.g. Biomass Density BACI), the number of data points used to fit the model were N = 882.292 16 Table 3: General structure of tested indicators of MPA effects. Bolded 𝛽parameter in Equation column indicates parameter interpreted as the effect of the MPA. Brefers to biomass, Erefers to effort, ML refers to mean length. INSIDE refers to patches plocated inside MPA borders (whether before or after MPA implementation), OUTSIDE to patches outside. AFTER refers to time periods tafter MPA implementation, BEFORE is before MPA implementation. NEAR are patches p outside but near the border of an MPA (whether before or after MPA implementation), FAR is outside and farther from the border of an MPA. Note that all regressions also include covariates for unfished total biomass across all species in each patch p, and weighting by absolute distance from MPA border. However we omit these ancillary terms from the table for clarity. Any indicator lacking a time term tis measured in the final time-step of the simulation, 20 years following MPA implementation. Indicator Estimated Effect and Key Assumptions Example Use Equation Biomass Density Before-After -Control-Impact (inside-outside) Ratio of change in Bafter MPA implementation inside MPA relative to outside MPA. Assumes Parallel trends in Binside and outside MPA preand post-implementation. Hopf et al. (2024) 𝑙𝑜𝑔(𝐵𝑝,𝑡)∼𝛽 0+ 𝛽1𝐼𝑁𝑆𝐼𝐷𝐸𝑝+𝛽 2𝐴𝐹𝑇𝐸𝑅𝑡 +𝛽3 𝛽3 𝛽3𝐼𝑁𝑆𝐼𝐷𝐸𝑝× 𝐴𝐹𝑇𝐸𝑅𝑡 Biomass Density Response Ratio (inside-outside) Post-MPA Ratio of B inside MPA relative to outside MPA. Assumes no pre-existing differences in Bbetween inside and outside sites Smith et al. (2024) 𝑙𝑜𝑔(𝐵𝑝)∼ 𝛽0+𝛽 1 𝛽1 𝛽1𝐼𝑁𝑆𝐼𝐷𝐸𝑝 Mean Length Response Ratio (inside-outside) Post-MPA ratio of ML inside MPA relative to outside MPA. Assumes no pre-existing differences in ML between inside and outside patches. Halpern (2003) 𝑙𝑜𝑔(𝑀𝐿𝑝)∼ 𝛽0+𝛽 1 𝛽1 𝛽1𝐼𝑁𝑆𝐼𝐷𝐸𝑝 Effort Gradient (near-far) Post-MPA ratio of Enear MPA relative to far. Assumes no pre-existing differences in Ebetween near and far patches. Goni et al. (2008) 𝑙𝑜𝑔(𝐸𝑝)∼ 𝛽0+𝛽 1 𝛽1 𝛽1𝑁𝐸𝐴𝑅𝑝 17 Indicator Estimated Effect and Key Assumptions Example Use Equation Biomass Density Gradient (near-far) Post-MPA ratio of Bnear MPA relative to far from MPA. Assumes no pre-existing differences in Bbetween near and far patches. Halpern, Lester, and Kellner (2009) 𝑙𝑜𝑔(𝐵𝑝)∼ 𝛽0+𝛽 1 𝛽1 𝛽1𝑁𝐸𝐴𝑅𝑝 Biomass Density Before-After Gradient (near-far) Ratio of change in Bafter MPA implementation near to MPA relative to far from MPA. Assumes Parallel trends in Bdensity near and far from MPA preand post-implementation. Lynham and Villaseñor-Derbez (2024) 𝑙𝑜𝑔(𝐵𝑝,𝑡)∼𝛽 0+ 𝛽1𝑁𝐸𝐴𝑅𝑝+𝛽 2𝐴𝐹𝑇𝐸𝑅𝑡+ 𝛽3 𝛽3 𝛽3𝑁𝐸𝐴𝑅𝑝× 𝐴𝐹𝑇 𝐸𝑅𝑡 We do not report uncertainty of estimates, as our primary objective is to evaluate relative perfor293 mance across indicators rather than to make statistical inference. Given that this is a simulation294 based analysis with arbitrarily defined sample sizes, conventional statistical significance testing is295 not meaningful (White et al. 2014), so we do not evaluate indicators against statistical significance 296 thresholds (e.g., p < 0.05). Similarly, we do not account for statistical complexities such as spatio297 temporal clustering (J. T. Thorson and Kristensen 2024) or zero-inflation (Zuur 2009). Applied use 298 of empirical indicators would need to be tailored to the local context, following best practices in 299 modeling and causal inference based on observational data from social-ecological systems.300 California Response Ratio Case Study301 The 10896 MPA experiments simulated in this study were generated to reflect a range of possible 302 bio-economic states and MPA design strategies. However, as a result there is no guarantee that 303 the distribution of simulated MPA effects within this range is reflective of the values seen in the 304 real world. To partly address this challenge, we created a subset of our simulations selected to 305 match the estimated response ratios from a network of MPAs along the coast of California, USA, 306 reported in Smith et al. (2024). These response ratios are reported at the level of individual MPAs, 307 with separate ratios for the total biomass density of targeted and non-targeted fishes. In order to 308 18 match these empirical response ratios, we sampled 2500 of our MPA experiments with replacement 309 using a weighting scheme designed to produce a distribution of Biomass Density Response Ratios 310 that matched those reported in Smith et al. (2024) (see Figure S27 ). We restricted the candidate 311 simulations for this process to MPA sizes between 10% and 40%, and to multi-species levels of 312 complexity, to more closely reflect plausible scenarios for the California system. This allowed us 313 to examine what range simulated MPA effects could be associated with the distribution of Biomass 314 Density Response Ratios measured in the California MPAs studied in Smith et al. (2024). Note that 315 we do not claim these as being simulations of the California MPA system itself, but rather a set 316 of simulations constrained to match a distribution of Biomass Density Response Ratios seen in a 317 real-world system. 318 Comparing MPA Effects and Indicators 319 The prior steps provide simulated MPA effects and indicator values for each MPA experiment. 320 We then assessed the ability of each indicator to track each outcome using a range of metrics. 321 We measured the correlation between each indicator and each outcome using Spearman’s rank 322 correlations ( 𝜌 ), given the potential for highly different scales between the indicator and the effects 323 and the potential for non-linear relationships between the indicator and the effects. We converted 324 these ρ values into ρ 2 values that reflect the proportion of the rank-order variance in the outcome 325 that is explained by the indicator. These Spearman’s rank correlation values provide a general 326 sense of the extent to which a higher indicator value in one MPA experiment and a lower indicator 327 value in another MPA experiment corresponds to a relatively higher outcome value at the first 328 MPA experiment and a relatively lower outcome value at the second MPA experiment. 329 To assess how accurately each indicator reflected the true outcome, we calculated the root mean 330 squared error (RMSE) and the mean error (Bias) for each indicator-outcome pair. These metrics 331 are expressed in percentage points, not percent differences. For example, a Bias of -20 means 332 that the indicator underestimated the true outcome by an average of 20 percentage points (e.g., 333 19 estimating 30% when the true value is 50%), not that it was 20% lower. Similarly an RMSE of 20 334 percentage points indicates typical deviations of about +/- 20 percentage points, not that the error 335 was within 20% of the true outcome.336 Reproducing Results337 All code needed to fully reproduce the results and manuscript are publicly available at https: 338 //github.com/DanOvando/mpa-effects-and-indicators https://doi.org/10.5281/zenodo.17308920.339 Results340 Variability of MPA Effects341 Our simulated fisheries varied in their behavior as a function of the fixed (Table 1) and variable 342 (Table 2) simulation parameters. The degree of resulting fishing pressure, and subsequent deple343 tion, is one of the most important determinants of MPA effects. The median depletion ( 𝐵/𝐵0 ) 344 value across all our included simulations was 0.38, mean value 0.41, ranging from a minimum of 345 0.01and a maximum of 1.29 (noting that values above 1 are possible given recruitment variation).346 Our simulated MPAs had a wide range of effects on species-level biomass and catch, with both 347 positive and negative causal effects possible for all measured effects (Figure 2). Increasing MPA 348 size generally expanded the range of possible effects. MPAs caused an increase in Biomass 349 Inside their borders in nearly all simulations (96%), though negative effects were also observed 350 in 4%. In contrast, Biomass Outside MPA borders increased in 48% of simulations, indicating a 351 roughly even likelihood of positive or negative effects on biomass beyond MPA borders across 352 our simulations. This asymmetry (consistent biomass gains inside and variable effects outside) 353 resulted in a net increase in Total Biomass in 88% of simulations and a net decrease in 12% of 354 simulations. Negative MPA effects for Biomass Inside and Total Biomass were more common 355 when the MPA placement strategy was to avoid fishing and when fishing effort was displaced 356 20 by the MPA, as these scenarios generally resulted in the concentration of fishing effort in prime 357 habitat. The possibility of the MPA causing net losses in biomass increased with the level of 358 fishing pressure (Figure 2). 359 MPAs resulted in a net increase in speciesand fleet-level Catch in 27% of simulations, and a 360 net decrease in Catch in 73% of simulations. Smaller effect sizes on Catch (absolute effect sizes 361 less than 25%) were much more common than larger effect sizes. Using the Avoid Fishing MPA 362 strategy resulted in slightly lower frequency of net catch losses than the Target Fishing strategy, 363 but only slightly less so. While negative catch effects still occurred across all simulated levels 364 of fishing pressure, the number of simulations with positive catch effects increased with the 365 counterfactual degree of fishing pressure (Figure 2). 366 21 Total Biomass Catch Biomass Inside Biomass Outside 0-15% 15-30% 30-60% 0-15% 15-30% 30-60% -100% 0% 100% ≥200% -100% 0% 100% ≥200% -100% 0% 100% ≥200% -100% 0% 100% ≥200% Seascape in MPA MPA Effect B/B0 >50% 25-50% 0-25% Quantile 100% 80% 50% Figure 2: Distribution of simulation results across MPA effects (panels), proportion of seascape in MPA (columns), and level of baseline depletion in the absence of MPAs (colors, biomass B divided by unfished biomasss B0). Y-axis shows the “MPA effect”, the percent change in the outcome in question caused by the MPA. Biomass Inside refers to total individual species biomass inside MPA borders. Biomass Outside refers to total individual species biomass outside MPA borders. Total Biomass refers to total individual species biomass both inside and outside MPA borders. Catch refers to total catch per species and fleet outside the MPA. 22 Performance of Empirical Indicators of MPA Effects 367 The ability of empirical indicators to track MPA effects varied widely (Figure 3A). All three 368 inside-outside indicators had positive Spearman’s 𝜌 with Biomass Inside and Total Biomass,with 369 Biomass Density BACI having the highest 𝜌 values for both ( 𝜌 = 0.64 and 𝜌 = 0.45 respectively), 370 followed by Biomass Density Response Ratio ( 𝜌 = 0.56and 𝜌 = 0.38 respectively). All three 371 inside-outside indicators had negative Spearman’s 𝜌 values with Biomass Outside, meaning that 372 simulations with relatively higher biomass density response ratios were associated with simulations 373 with relatively lower Biomass Outside effects. 374 Gradient indicators exploiting near-far patterns around MPAs had much lower Spearman’s 𝜌375 values across all effects (all |𝜌| ≤ 0.2 ), explaining almost none of the rank-level variation in any 376 of the evaluated MPA effects. Among the near-far indicators, Effort Gradients had the highest 377 correlation values, (𝜌= 0.2 for Biomass Inside and 𝜌= 0.16 for Total Biomass) (Figure 3). 378 None of the evaluated indicators were meaningfully correlated with the effects of MPAs on total 379 catches, though all estimated correlations were negative. Biomass Density BACI had the clearest 380 negative correlation with MPA effects on catch, with 𝜌 = −0.14 (Figure 3). 381 The 𝜌 values shown in Figure 3indicate correlation between the indicator value and the outcome 382 value. These correlations tell us how reliably an indicator can be used to rank different MPAs in 383 terms of performance related to a specific MPA effect. We also examined how well raw indicator 384 values represented raw MPA effects by calculating measures of error (root mean squared error) and 385 bias (mean absolute error), where positive bias values indicate that, on average, indicator values 386 were higher than true values, and vice versa. These metrics are more useful for determining how 387 well the value of a given indicator at one MPA translates into the value of a given effects at the 388 same MPA. The average RMSE across all indicators was 93 percentage points, with an average 389 bias across all indicators of 13 percentage points. Most indicators were positively biased relative to 390 the true effect size, though some indicators were negatively biased for Biomass Inside and Total 391 Biomass.392 23 All of the results shown in Figure 3are based comparisons at the level of individual species and 393 where applicable fleets, for example comparing the Biomass Density BACI value for reef fish 394 to the true effect of the MPA on reef fish catches by an individual fleet. We also ran our results 395 on total values, comparing for example the Biomass Density BACI value aggregated across all 396 simulated species to the true effect of the MPA on all catches across all fleets (Figure S31). Doing 397 so had no substantial impact on the core results presented in the body of the paper.398 24 MPAs given uncontrolled for violations to the core assumptions of these indicators such as differ488 ences in baseline habitat inside versus outside, recruitment shocks, and fishing fleet responses, are 489 common in marine social-ecological systems (Ferraro, Sanchirico, and Smith 2018; Larsen, Meng, 490 and Kendall 2019; Hilborn et al. 2022; Ovando et al. 2021; Hopf et al. 2024). Our results show 491 that even when faced with these biasing factors, inside-outside indicators still could be reasonably 492 reliable proxies for the relative conservation performance of different MPAs, as evidenced by the 493 relatively high Spearman’s 𝜌 values (Figure 3). However, these inside-outside indicators were 494 still generally imprecise (high RMSE) and biased, indicating that one cannot reliably interpret the 495 specific values of an inside-outside indicator as the numerical effect on an MPA on conservation. 496 None of indicators evaluated in this study reliably tracked the effects of MPAs on fishery catches, 497 demonstrated through low and negative Spearman’s 𝜌 values (all |𝜌| values < 0.15), high RMSE, 498 and generally positive bias (Figure 3). This highlights an urgent need for research on effective 499 indicators of the effects of MPAs on fishery catches, as our simulation results indicate that negative 500 effects can be common. Collection of actual data on the economic performance of fisheries (rather 501 than indirect indicators such as spillover gradients) could help provide better estimates of catch 502 effects, though finding suitable “control” fisheries to isolate the causal effect of MPAs on total 503 catch from other exogenous shocks (e.g. changes in fuel prices or market demand) will be difficult. 504 Near–far indicators, used as a measure of spillover effects, failed to track any MPA effects, despite 505 strong gradients in simulated biomass and effort being produced by our model (Figure S28). Both 506 simulation modeling and empirical evidence have confirmed that some level of spillover will 507 almost always occur at one or more life stages given the movement dynamics of marine organisms 508 (Cudney-Bueno et al. 2009; Di Lorenzo et al. 2020; Franceschini, Lynham, and Madin 2024; 509 Gaines et al. 2010; Hilborn et al. 2004; Hilborn et al. 2024). The more challenging question 510 is what does observing a spillover gradient tell us? Our results show that under our simulated 511 conditions, the presence of true spillover gradients in for example biomass density or fishing effort 512 caused by an MPA, is not generally necessary nor sufficient evidence for the effects of MPAs on 513 31 Biomass Inside,Biomass Outside,Total Biomass, or Catch.514 Near-far indicators exploiting gradients outside of MPAs methods have a long history in MPA 515 science in part because they can be calculated using only post-MPA fishery-dependent data, and do 516 not necessarily require establishing monitoring programs to track conditions pre-implementation or 517 survey within MPAs (Halpern, Lester, and Kellner 2009; Lynham and Villaseñor-Derbez 2024; 518 Medoff, Lynham, and Raynor 2022; Roberts et al. 2001; Di Lorenzo, Claudet, and Guidetti 2016; 519 Methratta 2020). While gradient methods were not effective indicators of any of the MPA effects 520 evaluated in this study, research is needed on whether there are modifications to these designs that 521 are effective for some use cases, given that fishery-dependent near-far data are often all that are 522 available and as such are likely to be used despite any associated warnings from works such as this. 523 For example, perhaps conditional on the life history of the species gradients at specific distances 524 are more meaningful than others. Ensembles of near-far indicators might perform better than any 525 individual one (Anderson et al. 2017).526 There are also many other potential objectives of MPAs besides those evaluated here, which may 527 perhaps be reliably tracked by gradient-based methods. For example, increases in the size and 528 abundance of fish near MPA borders may be of high value to recreational fishers (Franceschini, 529 Lynham, and Madin 2024), even if those increases on the border do not necessarily equate to 530 commensurate benefits at the scale of the total population or fishery.531 The poor performance of near-far indicators as proxies for broader MPA effects in our model 532 presents a major challenge for evaluating MPAs where only fishery-dependent data from outside 533 MPAs are available. Evaluations of large high-seas MPAs created through efforts such as BBNJ 534 are likely to depend solely on fisheries data from outside the MPA (Medoff, Lynham, and Raynor 535 2022; Lynham and Villaseñor-Derbez 2024), as the costs to conduct fishery-independent surveys in 536 large and remote pelagic MPAs are likely to be prohibitive. Hampton et al. (2023) provides a good 537 example of integration of fishery-dependent data with process-based models to estimate the effects 538 of high-seas MPAs on tuna fisheries. Novel sources of data such as the acoustic signals collected 539 32 by Fish Aggregating Devices used in tuna fisheries could also be explored (Moreno et al. 2019). 540 It is important to note that even among the indicators with relatively higher correlations (inside541 outside), the sign of the correlation coefficient was not consistent. This means that for example 542 when comparing two MPAs, the MPA with the higher Biomass Density Response Ratio may 543 have the relatively better outcome for Biomass Inside and Total Biomass, but the relatively worse 544 outcome for Biomass Outside and Catch. To illustrate, Biomass Density Response Ratios observed 545 in targeted fish species around California MPAs were generally positive, often substantially so 546 (Smith et al. 2024). Our simulation results show that while higher response ratios were generally 547 produced by better Biomass Inside outcomes, higher Biomass Density Response Ratios were 548 associated with relatively worse Biomass Outside and Catch effects (Figure 4). 549 We concur with Hopf et al. (2024) that while purely empirical methods have value, efforts to esti550 mate the effects of MPAs should ideally be placed in the context of the complexities and dynamics 551 of marine social-ecological systems. Bayesian estimation techniques provide a natural means for 552 encoding prior information based on other studies or bio-economic theory. Use of spatio-temporal 553 modeling to appropriately standardize spatial observations (J. T. Thorson 2019) so that they can 554 be passed to or integrated into spatial stock assessments (Punt 2019) could provide statistically 555 rigorous estimates of changes in fishing mortality rates and biomass relative to reference points as 556 a function fo MPA implementation, though these methods are highly dataand expertise-intensive. 557 Caveats 558 Our results depend on the assumptions embedded in the marlin operating model (Ovando et 559 al. 2023). While alternate assumptions could be used, the core of our results are a function of 560 assumptions that are broadly representative of real fisheries: both fish and fishers move, and 561 there is often spatial and temporal heterogeneity in life history, movement, habitat use, and fleet 562 dynamics. While we have aimed to produce plausible simulated states (e.g. by constraining the life 563 history of the species to common archetypes), we cannot assign real-world likelihood to any of our 564 33 simulated states, and some states and their associated MPA effects and indicator performances may 565 be more typical in reality than others.566 The model also omits certain ecological processes, such as trophic interactions or temporal trends 567 in habitat or life history. These temporal trends are likely to become increasingly important to 568 account for given the effects of climate change on global oceans (Fredston-Hermann et al. 2020). 569 We also assessed a subset of possible MPA effects and their interactions, omitting potentially 570 important but more challenging to model objectives such as fishery profitability. This could have 571 important implications; for example in scenarios where MPAs decrease catch, the economic 572 condition and food security of a community could still go up in total if those decreases in catch 573 were offset by the economic benefits of tourism opportunities caused by an MPA (Ban et al. 2019). 574 In order to calculate our empirical indicators, we assumed that all data are observed without error. 575 Observation error and sample sizes will further affect how well indicators track MPA effects. All 576 simulated data were collected after 20 years of MPA protection. MPA indicators and effects both 577 evolve over time (Hopf et al. 2024; Ovando, Dougherty, and Wilson 2016) and so future studies 578 could further examine the ways in which different time periods of sampling affect results. We 579 assumed that MPAs are fully no-take and were perfectly enforced; violations of either of these 580 assumptions would further complicate studies of MPA effects (Claudet and Guidetti 2010).581 In general, real empirical evaluations of MPA effects should consider good practices in causal 582 inference in spatial social-ecological systems, see Ferraro, Sanchirico, and Smith (2018), McEl583 reath (2020), Zuur (2009), Byrnes and Dee (2025), Punt (2019), J. Thorson and Kristensen (2024), 584 Tredennick et al. (2021), Grace (2024), White et al. (2011), Hopf et al. (2024), Hilborn et al. 585 (2022), and Ovando et al. (2021) for useful insights in this space.586 Conclusions587 The coming decades are likely to see a rapid expansion in the use of MPAs of various forms, 588 both in coastal seas and in dynamic open-ocean systems. The simulation modeling informed by 589 34 bio-economic theory implemented here supports the basic conclusion that this expansion is likely 590 to produce conservation gains inside borders and at the total population level, though the exact 591 magnitude of these benefits is highly uncertain. However, the effects of these MPAs outside their 592 borders, on both biomass and catch, is far less certain and may be highly positive, highly negative, 593 or anywhere in between, with negative catch effects being more common in our simulations. 594 As such, it is important that we be able to measure the effects of MPAs on a range of objectives, 595 so that we can quantify the costs and benefits produced by MPAs and subsequently adapt our 596 understanding of effective design and use of MPAs as needed. Our results show that while imper597 fect, some empirical indicators, particularly inside-outside indicators such as response ratios, can 598 be reliable indicators of conservation outcomes of MPAs, particularly inside their borders. But 599 that few empirical indicators commonly in use reliably track the effects of MPAs outside their 600 borders, particularly on fishery catches. Gradient based near-far methods (e.g. those that assess 601 spillover based on biomass densities near relative to far from MPA borders), while commonly used, 602 performed particularly poorly as an indicator of any MPA effect evaluated in this paper. 603 The next generation of MPA research urgently needs collaboration across disciplines such as 604 conservation biology, ecology, fisheries, and social sciences to develop sufficiently reliable and 605 practical methods for tracking the effects of MPAs. Doing so will help ensure that future expansion 606 of protected areas in the world’s oceans stand the best chance of achieving positive and equitable 607 outcomes for nature and people. 608 Acknowledgments 609 This paper was greatly aided by discussions with participants of the Area-Based Management 610 working group of the “Helping science advance policy in ocean conservation” workshop convened 611 by the The Center for Sustaining Seafood at the University of Washington, March 2-3 2024. 612 35 Author Contributions613 DO and CL conceived of the study and wrote the manuscript. All analyses performed by DO.614 Conflict of Interest Statement615 The authors declare no conflicts of interest.616 Data Availability Statement617 Open Research statement: No data were collected for this study. This submission uses novel 618 code, which is provided, per our requirements, in an external repository to be evaluated during the 619 peer review process. The link to this external repository is https://github.com/DanOvando/mpa620 effects-and-indicators A pre-print archive of the code along with all results needed to reproduce 621 this paper is also available at https://doi.org/10.5281/zenodo.17308920. This zenodo repository 622 will be upgraded to a permanent archive of the final results and paper upon publication.623 36 References 624 Abbott, Joshua K., and Alan C. Haynie. 2012. “What Are We Protecting? Fisher Behavior and the 625 Unintended Consequences of Spatial Closures as a Fishery Management Tool.” Ecological 626 Applications 22 (3): 762–77. doi:10.1890/11-1319.1.627 Anderson, Sean C., Andrew B. Cooper, Olaf P. Jensen, Cóilín Minto, James T. Thorson, Jessica C. 628 Walsh, Jamie Afflerbach, et al. 2017. “Improving Estimates of Population Status and Trend 629 with Superensemble Models.” Fish and Fisheries 18 (4): 732–41. doi:10.1111/faf.12200.630 Ban, Natalie C., Georgina Grace Gurney, Nadine A. Marshall, Charlotte K. Whitney, Morena 631 Mills, Stefan Gelcich, Nathan J. Bennett, et al. 2019. “Well-Being Outcomes of Marine 632 Protected Areas.” Nature Sustainability 2 (6): 524. doi:10.1038/s41893-019-0306-2.633 Brodie, Stephanie, Briana Abrahms, Steven J. Bograd, Gemma Carroll, Elliott L. Hazen, Barbara 634 A. Muhling, Mercedes Pozo Buil, James A. Smith, Heather Welch, and Michael G. Jacox. 635 2021. “Exploring Timescales of Predictability in Species Distributions.” Ecography 44 (6): 636 832–44. doi:10.1111/ecog.05504.637 Byrnes, Jarrett E. K., and Laura E. Dee. 2025. “Causal Inference With Observational Data and 638 Unobserved Confounding Variables.” Ecology Letters 28 (1): e70023. doi:10.1111/ele.70023.639 Cabral, Reniel B., Benjamin S. Halpern, Sarah E. Lester, Crow White, Steven D. Gaines, and 640 Christopher Costello. 2019. “Designing MPAs for Food Security in Open-Access Fisheries.” 641 Scientific Reports 9 (1): 8033. doi:10.1038/s41598-019-44406-w.642 Carr, Mark H., J. Wilson White, Emily Saarman, Jane Lubchenco, Kristen Milligan, and Jennifer 643 E. Caselle. 2019. “Marine Protected Areas Exemplify the Evolution of Science and Policy.” 644 Oceanography 32 (3): 94–103. https://www.jstor.org/stable/26760087.645 Chapman, Matthew R., and Donald L. Kramer. 1999. “Gradients in Coral Reef Fish Density and 646 Size Across the Barbados Marine Reserve Boundary: Effects of Reserve Protection and Habitat 647 Characteristics.” Marine Ecology Progress Series 181: 81–96. https://www.jstor.org/stable/ 648 24849598.649 37 Claudet, Joachim, and Paolo Guidetti. 2010. “Improving Assessments of Marine Protected 650 Areas.” Aquatic Conservation: Marine and Freshwater Ecosystems 20 (2): 239–42. 651 doi:10.1002/aqc.1087.652 Costello, Christopher, Daniel Ovando, Tyler Clavelle, C. Kent Strauss, Ray Hilborn, Michael 653 C. Melnychuk, Trevor A. Branch, et al. 2016. “Global Fishery Prospects Under Contrasting 654 Management Regimes.” Proceedings of the National Academy of Sciences 113 (18): 5125–29. 655 doi:10.1073/pnas.1520420113.656 Cudney-Bueno, Richard, Miguel F. Lavín, Silvio G. Marinone, Peter T. Raimondi, and William 657 W. Shaw. 2009. “Rapid Effects of Marine Reserves via Larval Dispersal.” PLoS ONE 4 (1): 658 e4140. doi:10.1371/journal.pone.0004140.659 Di Lorenzo, Manfredi, Joachim Claudet, and Paolo Guidetti. 2016. “Spillover from Marine 660 Protected Areas to Adjacent Fisheries Has an Ecological and a Fishery Component.” Journal 661 for Nature Conservation 32 (July): 62–66. doi:10.1016/j.jnc.2016.04.004.662 Di Lorenzo, Manfredi, Paolo Guidetti, Antonio Di Franco, Antonio Calò, and Joachim Claudet. 663 2020. “Assessing Spillover from Marine Protected Areas and Its Drivers: A Meta-Analytical 664 Approach.” Fish and Fisheries 21 (5): 906–15. doi:10.1111/faf.12469.665 Edgar, Graham J., Rick D. Stuart-Smith, Trevor J. Willis, Stuart Kininmonth, Susan C. Baker, 666 Stuart Banks, Neville S. Barrett, et al. 2014. “Global Conservation Outcomes Depend on 667 Marine Protected Areas with Five Key Features.” Nature 506 (7487): 216–20. doi:10.1038/na668 ture13022.669 Ferraro, Paul J., James N. Sanchirico, and Martin D. Smith. 2018. “Causal Inference in Coupled 670 Human and Natural Systems.” Proceedings of the National Academy of Sciences, August, 671 201805563. doi:10.1073/pnas.1805563115.672 Franceschini, Simone, John Lynham, and Elizabeth M. P. Madin. 2024. “A Global Test of 673 MPA Spillover Benefits to Recreational Fisheries.” Science Advances 10 (29): eado9783. 674 doi:10.1126/sciadv.ado9783.675 Fredston-Hermann, Alexa, Rebecca Selden, Malin Pinsky, Steven D. Gaines, and Benjamin S. 676 38 Halpern. 2020. “Cold Range Edges of Marine Fishes Track Climate Change Better Than 677 Warm Edges.” Global Change Biology 26 (5): 2908–22. doi:10.1111/gcb.15035.678 Gaines, Steven D., Crow White, Mark H. Carr, and Stephen R. Palumbi. 2010. “Designing Marine 679 Reserve Networks for Both Conservation and Fisheries Management.” Proceedings of the 680 National Academy of Sciences 107 (43): 18286–93. doi:10.1073/pnas.0906473107.681 Goni, R., S. Adlerstein, D. Alvarez-Berastegui, A. Forcada, O. Renones, G. Criquet, S. Polti, et al. 682 2008. “Spillover from Six Western Mediterranean Marine Protected Areas:: Evidence from 683 Artisanal Fisheries.” Marine Ecology Progress Series 366: 159–74. doi:10.3354/meps07532.684 Grace, James B. 2024. “An Integrative Paradigm for Building Causal Knowledge.” Ecological 685 Monographs 94 (4): e1628. doi:10.1002/ecm.1628.686 Grorud-Colvert, Kirsten, Jenna Sullivan-Stack, Callum Roberts, Vanessa Constant, Barbara Horta 687 e Costa, Elizabeth P. Pike, Naomi Kingston, et al. 2021. “The MPA Guide: A Framework 688 to Achieve Global Goals for the Ocean.” Science 373 (6560): eabf0861. doi:10.1126/sci689 ence.abf0861.690 Halpern, Benjamin S. 2003. “The Impact of Marine Reserves: Do Reserves Work and Does 691 Reserve Size Matter?” Ecological Applications 13 (sp1): 117–37. doi:10.1890/1051692 0761(2003)013[0117:TIOMRD]2.0.CO;2.693 Halpern, Benjamin S., Sarah E. Lester, and Julie B. Kellner. 2009. “Spillover from Marine Re694 serves and the Replenishment of Fished Stocks.” Environmental Conservation 36 (04): 268276. 695 doi:10.1017/S0376892910000032.696 Hampton, John, Patrick Lehodey, Inna Senina, Simon Nicol, Joe Scutt Phillips, and Kaon Tiamere. 697 2023. “Limited Conservation Efficacy of Large-Scale Marine Protected Areas for Pacific 698 Skipjack and Bigeye Tunas.” Frontiers in Marine Science 9. https://www.frontiersin.org/ 699 articles/10.3389/fmars.2022.1060943.700 Hilborn, Ray, Vera N. Agostini, Milani Chaloupka, Serge M. Garcia, Leah R. Gerber, Eric Gilman, 701 Quentin Hanich, et al. 2022. “Area-Based Management of Blue Water Fisheries: Current 702 Knowledge and Research Needs.” Fish and Fisheries 23 (2): 492–518. doi:10.1111/faf.12629. 703 39 Hilborn, Ray, Mark Fitchett, John Hampton, and Daniel Ovando. 2024. “When Does Spillover 704 from Marine Protected Areas Indicate Benefits to Fish Abundance and Catch?” Theoretical 705 Ecology 18 (1): 1. doi:10.1007/s12080-024-00596-2.706 Hilborn, Ray, Kevin Stokes, Jean-Jacques Maguire, Tony Smith, Louis W Botsford, Marc Mangel, 707 José Orensanz, et al. 2004. “When Can Marine Reserves Improve Fisheries Management?” 708 Ocean & Coastal Management 47 (3–4): 197–205. doi:10.1016/j.ocecoaman.2004.04.001.709 Hopf, Jess K., Victoria Quennessen, Jacob Ridgway, Caren Barceló, Fabio Prior Caltabellotta, 710 Sarah Farnsworth Hayroyan, Derek Garcia, et al. 2024. “Ecological Success of No-Take Ma711 rine Protected Areas: Using Population Dynamics Theory to Inform a Global Meta-Analysis.” 712 Ecological Applications 34 (7): e3027. doi:10.1002/eap.3027.713 Humphreys, John, and Robert W. E. Clark. 2020. “Chapter 1 - a Critical History of Marine 714 Protected Areas.” In Marine Protected Areas, edited by John Humphreys and Robert W. E. 715 Clark, 1–12. Elsevier. doi:10.1016/B978-0-08-102698-4.00001-0.716 Johannes, R. E. 2002. “The Renaissance of Community-Based Marine Resource Management 717 in Oceania.” Annual Review of Ecology and Systematics 33 (1): 317–40. doi:10.1146/an718 nurev.ecolsys.33.010802.150524.719 Karp, Melissa A, Megan Cimino, J Kevin Craig, Daniel P Crear, Christopher Haak, Elliott L 720 Hazen, Isaac Kaplan, et al. 2025. “Applications of Species Distribution Modeling and Future 721 Needs to Support Marine Resource Management.” ICES Journal of Marine Science 82 (3): 722 fsaf024. doi:10.1093/icesjms/fsaf024.723 Kerr, Lisa A, Jacob P Kritzer, and Steven X Cadrin. 2019. “Strengths and Limitations 724 of Before–after–control–impact Analysis for Testing the Effects of Marine Protected 725 Areas on Managed Populations.” ICES Journal of Marine Science 76 (4): 1039–51. 726 doi:10.1093/icesjms/fsz014.727 Larsen, Ashley E., Kyle Meng, and Bruce E. Kendall. 2019. “Causal Analysis in Control–impact 728 Ecological Studies with Observational Data.” Methods in Ecology and Evolution 10 (7): 729 924–34. doi:https://doi.org/10.1111/2041-210X.13190.730 40