Measuring biological effectiveness across a very large, coherent network of coastal marine protected areas.

We evaluated the performance of California's coastal marine protected area (MPA) network, the largest scientifically designed network of its kind, using SCUBA survey data from a large-scale, long-term kelp forest monitoring program. To comprehensively evaluate MPA performance, we employed multiple metrics across various scales of taxonomic and spatial aggregation and four key results emerged. First, population responses to MPAs varied greatly across the large network, whether evaluated as combined taxa or individual focal species or at single MPA or regional scales. Second, while responses varied widely, effect size analysis revealed differences in magnitude of MPA performance over time, with targeted species increasing or declining less within MPAs compared to non-targeted species in all regions except the North Coast. Third, average response ratios further supported these findings by providing a cumulative perspective on MPA performance. Specifically, 80% of MPAs showed positive response ratios for targeted species, with 12 of these cases being significant, indicating generally higher biomass inside MPAs. Finally, using our most conservative performance framework of divergence between MPA and non-MPA (Ref) sites and between targeted and non-targeted species, only the South Coast and Northern Channel Islands regions exhibited a significantly positive trend in biomass ratios (MPA/Ref) over time for targeted species. While biomass ratios increased significantly over time within MPAs for targeted species, divergence between targeted and non-targeted species' trends was largely absent, except in the Northern Channel Islands where a marginally non-significant effect was noted. Taken together, these results demonstrate that MPA performance is variable across regions and taxa, and different aspects of performance were revealed depending on the metric used. This underscores the need for comprehensive evaluation approaches when assessing the effectiveness of large, networked MPAs.