Small-scale fisheries catch more threatened elasmobranchs inside partially protected areas than in unprotected areas

The study investigates how small-scale fisheries (SSF) interact with and impact coastal elasmobranchs (sharks, skates, rays) in the Mediterranean Sea, particularly within Marine Protected Areas (MPAs) that are largely composed of partially protected areas (PPAs). Elasmobranchs have undergone global declines driven by overfishing, exacerbated by life-history traits that increase vulnerability. SSF, which employ most of the world’s fishers and contribute substantially to global catch, are understudied relative to industrial fleets, and their impacts on elasmobranchs are poorly quantified. MPAs, including PPAs, are touted as conservation tools with potential benefits for mobile predators, yet their effectiveness for elasmobranchs in the Mediterranean remains uncertain. The authors hypothesize that, due to generally lower fishing pressure and restrictions in PPAs relative to surrounding unprotected areas (UPAs), elasmobranch catch per unit effort (CPUE) may be higher in PPAs, reflecting refuge effects, and aim to disentangle protection effects from environmental and fishing covariates. The research assesses species composition, abundance (NCPUE), biomass (BCPUE), and maturity structure of elasmobranchs caught by SSF inside PPAs versus UPAs across 11 locations in six Mediterranean countries.

Prior work documents substantial global and regional declines in elasmobranch abundance, with underreporting in official statistics due to IUU catches and lack of species-specific data. MPAs can confer biodiversity benefits and may aid conservation of mobile predators, though effectiveness varies with size, enforcement, and species’ movement ecology. Large fully protected areas have shown potential benefits for pelagic elasmobranchs, yet most MPAs are small and multi-use PPAs, and evidence of their effectiveness for elasmobranchs is limited. The Mediterranean is a biodiversity hotspot with a long history of exploitation, where elasmobranch status has worsened faster than global trends, and SSF comprise the majority of vessels and can significantly impact vulnerable megafauna. Previous studies have highlighted declines in both large predators and mesopredators in the region, and the need for improved data collection and management. There is a gap regarding the role of PPAs in protecting Mediterranean elasmobranchs, motivating this study.

Study area: Eleven MPAs in six Mediterranean EU countries (France: Bonifacio, Cap Roux, Côte Bleue; Italy: Portofino, Egadi, Torre Guaceto; Spain: Es Freus, Cabo de Palos; Croatia: Telascica; Slovenia: Strunjan; Greece: Zakynthos Island) were sampled between June 2017 and October 2018. Comparisons of PPAs vs UPAs were conducted for nine locations (Cap Roux and Côte Bleue excluded due to lack of PPA with restrictions). Data collection and sampling: A total of 1,256 SSF fixed-net operations (primarily trammel nets ~95%, gillnets 4%, mixed 1%) were monitored at depths up to 150 m, totaling 737.71 km of nets deployed. Fishing operations spanned 37 to 162 per location. Fishers (<12 m LOA vessels, no towed gears) voluntarily participated; catches were photo-sampled at landing to minimize handling and expedite data capture. Operators photographed entire catches with a ruler for scale and unique IDs, ensuring all specimens visible. Net type and length, deployment position/depth (mapped via fishers), and gear parameters (mesh size, soak time) were recorded. Species identification and measurements: Trained operators analyzed images using ImageJ. Individuals were identified to species, measured for total length to 1 mm precision, and biomass estimated via length–weight relationships (FishBase). Each species was assigned its Mediterranean IUCN Red List category. Standardization and indices: CPUE was standardized per 1000 m net length as NCPUE (number per 1000 m) and BCPUE (grams/kg per 1000 m). A maturity index was computed as Size_ind/L50 (using the lower of sex-specific L50 from literature), classifying individuals as immature (<1) or mature (≥1). Species were grouped into Threatened (THR: CR/EN/VU), Non-threatened (NTH: NT/LC), and Data Deficient (DD) for maturity analyses. Covariates: Environmental (chlorophyll a, SST, SSS, dissolved oxygen, phosphate, nitrate, habitat), geographic (location, latitude, longitude), temporal (season), bathymetric (depth), anthropogenic (cumulative human impacts), and fishing operation features (depth, net length, mesh size, soak time) were compiled from mapped coordinates and datasets. DO and nitrate were excluded from models due to multicollinearity. Statistical analyses: Zero-inflation was assessed; generalized additive models for location, scale, and shape (GAMLSS) with zero-adjusted Gamma (ZAGA) distributions were selected for BCPUE and NCPUE, modeling two components: (1) logit (presence/absence of any elasmobranch in catch) and (2) log (positive part excluding zeros). Random intercepts for location (9 levels) and habitat nested within location (6 levels) accounted for the nested design. Model selection used StepGAIC with Schwarz Bayesian Criterion, diagnostic residual tools (worm plots), and partial residual plots; pseudo R^2 via Cragg-Uhler. Separate GAMLSS were run for counts of immature and mature individuals per IUCN group (comparing ZIP vs ZAGA; ZAGA selected by AIC). Multivariate analysis: Partial distance-based redundancy analyses (pRDA) on Hellinger-transformed BCPUE and NCPUE across the six IUCN categories (CR, EN, VU, NT, LC, DD) evaluated protection effects and environmental predictors, with location and habitat as conditional variables. Significance used Monte Carlo permutation (999–1000 iterations). Software and data/code availability: Analyses performed in R (gamlss, ggplot2, vegan). Data and R code deposited in Figshare repositories cited in the article.

- Species and catches: From 1,256 fixed-net operations (737.71 km nets), 892 elasmobranch individuals of 24 species (4 demersal sharks, 20 batoids) were recorded, representing nearly one-third of Mediterranean elasmobranch species to 800 m depth. Elasmobranchs comprised 2.4% of overall NCPUE and 6.4% of overall BCPUE. More than one-third of species caught were IUCN Threatened (CR/EN/VU). CR species recorded: Dipturus batis, Leucoraja fullonica. EN: Raja radula, Rhinobatos rhinobatos, Rostroraja alba. VU: Dasyatis pastinaca, Myliobatis aquila, Mustelus mustelus, Mustelus punctulatus. Seven NT species were listed. DD and LC species formed the remainder. Threatened species had the highest mean BCPUE (507.73 ± 74.3 g/1000 m), DD the lowest (2.5 ± 1.0 g/1000 m). - PPAs vs UPAs sampling effort and totals (9 comparable locations): PPAs: 573 operations, 385.51 km nets, 517 individuals, 487.53 kg total biomass. UPAs: 511 operations, 352.20 km nets, 358 individuals, 223.07 kg biomass. - CPUE differences: Mean BCPUE and NCPUE were higher inside PPAs than UPAs. PPA: 1299.8 ± 177.6 g/1000 m and 0.63 ± 0.08 n/1000 m; UPA: 773.6 ± 197.3 g/1000 m and 0.57 ± 0.10 n/1000 m. - GAMLSS results: Protection significantly affected both BCPUE and NCPUE. Presence probability (logit) of any elasmobranch in a catch was higher in UPAs than PPAs (BCPUE logit: t = -2.95, p = 0.003; NCPUE logit: t = -2.82, p = 0.004). However, conditional on presence, the positive component (log) showed higher BCPUE and NCPUE inside PPAs (BCPUE log: t = -2.81, p = 0.001; NCPUE log: t = -2.93, p = 0.003). Chlorophyll a was positively related to both BCPUE (t = 6.96, p = 0.001) and NCPUE (t = 3.70, p = 0.001). SST was inversely related to BCPUE (t = 3.69, p = 0.001). Human impacts index was positively related to NCPUE (t = 2.62, p = 0.008) but not BCPUE. Deviance explained: 37.0% (BCPUE) and 41.2% (NCPUE). - Maturity structure: Kernel density analyses indicated immature individuals were more frequently caught than mature ones for THR and DD groups, while NTH catches had similar proportions of mature and immature. Across both PPAs and UPAs, higher numbers of immature individuals were caught overall. The proportion of mature individuals did not differ between PPAs and UPAs within THR, NTH, DD groups. However, immature individuals of THR and DD species were more frequently captured inside PPAs (THR: Z = -3.268, p = 0.0015; DD: Z = -2.006, p = 0.006). - Multivariate (pRDA): Endangered and Vulnerable categories were highly correlated with PPAs; protection emerged as a significant predictor (p < 0.001), supporting higher BCPUE/NCPUE of Threatened categories inside PPAs. - Management-relevant observations: Three species fully protected under regional agreements (D. batis, R. rhinobatos, R. alba) were nonetheless caught and landed by SSF, indicating awareness and enforcement gaps.

The findings confirm the hypothesis that, while elasmobranch presence in catches may be more likely in UPAs, the abundance and biomass of elasmobranchs per unit effort are higher within PPAs, consistent with PPAs acting as refuges that harbor greater densities and biomasses of threatened elasmobranchs. This suggests that restrictions within PPAs (reduced effort, gear regulations, reduced disturbance) can indirectly or directly benefit elasmobranch populations, potentially via improved habitat condition and increased prey availability. The higher frequency of immature THR and DD individuals in PPAs implies that PPAs may include nursery or critical juvenile habitats; thus, these areas can be important for recruitment. However, the removal of both immature and mature threatened individuals by SSF inside PPAs reveals a conservation paradox: PPAs may accumulate vulnerable life stages yet allow fishing that undermines conservation outcomes. Environmental covariates (chlorophyll a positive; SST negative for BCPUE) indicate productivity and temperature influence catch rates, while a positive association of human impact with NCPUE may reflect altered community dynamics (e.g., mesopredator or juvenile release) in heavily impacted areas. Overall, the work highlights that PPAs can contribute to elasmobranch conservation but require effective SSF management to translate ecological potential into realized protection.

This study provides region-wide, fishery-dependent evidence that small-scale fisheries capture significant numbers of threatened elasmobranchs and that CPUE (abundance and biomass) of these species is higher inside partially protected areas than in adjacent unprotected areas. PPAs likely encompass critical habitats and may function as refuges for elasmobranchs, including juveniles, but current SSF practices inside PPAs can compromise conservation objectives. The authors recommend: (1) closing data gaps through coordinated fishery-dependent and independent monitoring (including species-level, size, spatial and life-history data); (2) implementing fishing-effort controls, temporal/gear restrictions, and shark deterrent devices; (3) disseminating best-practice handling, release, and identification training for fishers and enforcement personnel; (4) identifying critical habitats to enable seasonal closures and systematically collecting SSF effort data. Strengthening PPA design, compliance, and governance is essential to leverage their potential within broader goals (e.g., 30×30) and to reduce bycatch mortality while allowing sustainable uses where appropriate.

Results are based on fishery-dependent data from fixed nets and assume CPUE is a reliable proxy for abundance and biomass in the absence of hyperstability or hyperdepletion. Species sex was generally indeterminable from photos; the maturity index used the lower sex-specific L50, potentially biasing maturity classification conservatively. Data on prey abundance and community structure were lacking, limiting inference on trophic mechanisms. SSF fishing-effort data (e.g., spatial and temporal effort, compliance) were generally unavailable. Two locations were excluded from PPA–UPA comparisons due to missing PPA restrictions. Environmental and anthropogenic covariates, while extensive, cannot capture all sources of variability, and multicollinearity required excluding some predictors. Findings pertain to Mediterranean coastal SSF and may not directly generalize elsewhere.