Stock identification provides a basis for understanding population dynamics and makes the stock assessment process more robust, thereby developing fisheries management strategies. Multiannual Management Plans under the Common Fishery Policy (EU Reg. 1380/2013) are tools for managing shared stocks in the long term and thus this requires improving our knowledge on biological stock units and fishery management units. Methods for delineating stocks advanced considerably in recent years and include genetic techniques, otolith shape and chemistry, acoustic telemetry, tagging, demographic analysis and meristic data. The integration of multiple techniques that operate over different temporal and spatial scales makes it possible to overcome many of the limitations of single technique approaches and strengthens the inference available from stock structure studies (Cadrin et al., 2013). The identification of fishing grounds is an essential information to delineate the fishing footprints on the fish and shellfish stocks. To identify fishing grounds different methods are available, based on Automatic Identification System (AIS) and Vessel Monitoring System (VMS). Linking information on stock boundaries with the one on the localization of the fishing grounds is a key step for the identification of spatial units for fishery management. The overall objective of the MED_UNITs project is to identify and match biological and management stock units of several important demersal species in the Mediterranean: European hake (Merluccius merluccius), red mullet (Mullus barbatus), deep water rose shrimp (Parapenaeus longirostris), giant red shrimp (Aristaeomorpha foliacea), blue and red shrimp (Aristeus antennatus) and Norway lobster (Nephrops norvegicus). The study covers the Geographical Sub-Areas (GSAs) 1-27 (Mediterranean Sea). The project structure consists of 5 Work-Packages (WPs) and 16 Tasks. WP0-Project management and coordination; WP1-Population genetics and phylogeographic studies for identification of biological units of priority species; WP-2 Otolith shape and microchemistry analyses; WP3-Delineate fishing grounds and stock assessment; WP4-Synthesis and proposals. The analyses undertaken in WP1 and WP2 delineate the population units from a biological perspective. WP3 defines the fisheries footprints not necessarily within the boundaries of the current GSAs. The integration of this information takes place in WP4, supported by the explanatory role of ecological/environmental profiles at spatial scale. Overall, this approach is expected to advance fisheries assessment and improve the management advice, reducing the bias associated with the assumption of a given stock unit, when instead multiple stocks are assessed as a single unit or only a portion of a stock is assessed as a closed unit. WP0 - Project management and coordination managed the reporting, institutional meetings, internal meetings and the coordination among the several project activities. Virtual tools were made available to the project for supporting the work organization and sharing documents and data. The coordination of the samplings for the activities of WP1 and WP2 was also carried out, establishing a Sampling Procedure with a Genetic Hub to manage the exchange of samples for the genetic analyses and an Otolith Hub for the exchange of otoliths. These two hubs managed the transfer among laboratories of approximately 10500 samples of tissues for genetics and 3700 otolith samples. Sampling protocols were distributed to the project partners and to the cooperating Institutes. The sampling strategy was designed to provide a wide coverage also balancing the geographical locations of samples. Most of the 27 Mediterranean GSAs were divided in subunits (overall 63). MEDITS survey was the main source of samples for the European Member States. A liaison between the Directorate-General for Maritime Affairs and Fisheries (DG MARE), the General Fisheries Commission for the Mediterranean (GFCM) and the Food and Agriculture Organization (FAO) Regional Projects was established to support the collection of samples also from south and east Mediterranean Countries through the Data Collection Regulation Framework (DCRF). Samples were collected from Algeria, Tunisia, Egypt, Palestine, Lebanon and Turkey. Data collected at individual level were stored in a database and complementary information on environmental variables from Copernicus products as well. Overall, 1984 individuals were sampled for M. merluccius, 2209 for M. barbatus, 1470 for A. antennatus, 1693 for A. foliacea, 1537 for N. norvegicus and 1751 for P. longirostris. WP1 - Population genetics and phylogeographic studies for the identification of biological units of priority species aimed to: review the genetic data available for the six target species; acquire information for the sampling design; deliver a comprehensive review of genetic methods and tools; perform experimental studies on the genetic distribution among and within Mediterranean sub-basins, and provide a detailed protocol for routine sampling and genetic monitoring. The review of population genetic studies identified preliminary considerations on the background related to stock boundaries, connectivity patterns and population genetic structure of the target species. This information constituted the basis for the sampling design and the new genetic analyses. The review of genetic methods and tools examined each method/tool for its peculiarities, appropriateness, robustness and accuracy in stock identification. Following the aforementioned evaluation, a highly reproducible pipeline for the analysis of the six species of MED_UNITs was proposed and used within the project. A first batch of analyses of tissues was carried out in the Pilot Genetic studies’ part for the optimization of protocols. Based on the knowledge from the Pilot Genetic studies, the totality of the specimens sampled for each species was analysed in the Comprehensive genetic studies’ part, allowing an unprecedented biogeographic analysis. The genomic method applied was a Genotyping-by-Sequencing (GBS) methodology, constructing reduced-representation libraries in each species. Single Nucleotide Polymorphism markers (SNPs) newly isolated following the double-digest Random Amplified DNA (ddRAD) sequencing were used. The main results by species can be summarized as follows. For the giant red shrimp (Aristaeomorpha foliacea), the final dataset was composed of 771 samples (for 30 localities) and 443 high quality SNPs. The results point out an evident lack of genetic differentiation and are generally in agreement with previous studies conducted at smaller geographical scales and less extended sampling points in the Mediterranean Sea. However, using a statistical tool as the hierarchical Analysis of Molecular Variance (AMOVA), we detected the existence of very weak differentiation between Western, Eastern and Central Mediterranean samples. For the blue and red shrimp (Aristeus antennatus), a total of 1253 SNPs were retained for 886 samples. As for A. foliacea, much of the genetic variation was distributed among individuals in the populations, with a slight support for three groups corresponding to Western Mediterranean, Central Mediterranean, and Eastern Mediterranean. For European hake (Merluccius merluccius) a total of 665 high quality SNPs was retained (for 1,667 samples) and used for all downstream differentiation analyses. The strongest differentiation was between Atlantic and Mediterranean, but also within the Mediterranean populations following a West to East pattern. For red mullet (Mullus barbatus), the final dataset was composed of 771 samples (for 30 localities) and 853 high quality SNPs. Very low genetic differentiation values were measured, indicating the absence of clear population structure in the Mediterranean. In Norway lobster (Nephrops norvegicus), the final dataset was composed of 890 samples (for 27 localities) and 730 high quality SNPs. The results showed a significant differentiation of the samples eastern of the Adriatic Sea (GSA17) against the others from the central and Western Mediterranean. Additionally, relatively high and significant values were also encountered for the separation of the Adriatic Sea (GSA17 to 19) from the neighbouring basins to the west (GSA1 to 11) and the east (GSA22). For deep-water rose shrimp (Parapenaeus longirostris), a total of 1,225 SNPs was retained for 782 individuals. Genetic clustering methods and AMOVA analyses indicated the existence of differentiation, and support for a three-groups scenario: a “western-central” group including samples from Western and Central Mediterranean up to the Strait of Sicily, a “central” group including the remaining samples from the Central Mediterranean except the easternmost Ionian sample, and an “eastern” group including the samples from Eastern Mediterranean and the remaining Ionian. Overall, the MED_UNITs sampling was satisfactory for the spatial coverage and number of specimens collected with a total of 10,670 specimens for all six species. In general, DNA of good quality was easily obtained from fresh tissues, while frozen tissues generated larger fraction of medium quality or even unusable DNAs for the methodology applied. The genotype success of the ‘good DNA quality’ tissues was high (82%), while only 61% of the ‘medium quality’ DNAs included in the libraries ended up producing genotype data. We concluded that the sampling should be designed in order to fulfil the needs of ‘genomics’ requirements (in terms of quality of samples, procedures, storage, timing). Under task 1.6 it was highlighted that the scientific surveys at sea (as MEDITS or other similar surveys) are a good opportunity to implement proper sampling design for genomic analyses. However, the sampling for genomic analyses cannot be a collateral activity, and adequate resources should be allocated for this activity. The timeframe for collecting tissues should be carefully defined, and the timing for the experimental phases cannot be too strict. The collection of tissues should preferentially be realized from alive/freshly caught individuals and the sampling realized as soon as possible. A specific protocol addressing these points was delivered in this project. WP2 - Otolith shape and microchemistry analyses conducted a literature review scrutinising about 600 papers to obtain a general overview on applications and methodologies and their potential use. Ten studies were retained that distinguish Merluccius spp. stocks in different areas. Mediterranean population of M. merluccius was evaluated in one study combining a multiple approach (chemical and shape). With reference to M. barbatus, five studies were performed and only one was carried out in the Mediterranean Sea. For both species, the protocols are common for image acquisition process, extraction of the external outline information and otolith shape analysis as well as for otolith preparation, processing and analysis for microchemistry trace element detection. For each species, the first sexual maturity and the sex could be potential sources of variation. Consequently, only 1 life stage and 1 sex were used by each species. For otolith shape analysis, the used data sets (left and right otoliths) were composed by 1845 otoliths of red mullet from 37 geographical subunits and by 1868 of European hake otoliths from 39 geographical subunits. In addition to the geographical subunits in the Mediterranean, samples were also gathered from 4 ICES areas for comparison purposes. For both species, two complementary analyses were performed using the Linear Discriminant Analysis (LDA) with Jacknifed prediction (Supervised Machine Learning) and the hierarchical clustering (Unsupervised Machine Learning). For European hake, the optimised classification (correct classification rate = 39.61%) presented 4 groups, distributed as follows: Atlantic Ocean (from ICES IV to ICES VIII), Western Mediterranean Sea (from GSA1 to GSA13), Adriatic Sea with Central Mediterranean Sea (from GSA16 to GSA20) and Eastern Mediterranean Sea (from GSA22 to GSA27). For red mullet, the optimised classification (correct classification rate = 37.56%) presented three groups, distributed as follows: Western Mediterranean Sea (from GSA1 to GSA16), Adriatic Sea with Central Mediterranean Sea (from GSA17 to GSA20) and Eastern Mediterranean Sea (from GSA22 to GSA27). The analysis of otolith microchemistry was based on a subset of the otolith samples used for the otolith shape analysis: 279 otoliths for European hake from 10 Mediterranean subunits, plus two additional areas in the NE Atlantic added for comparison, and 250 otoliths from 10 different Mediterranean GSAs for red mullet. For European hake, 25 isotopes corresponding to the otolith core and otolith edge for each individual were obtained and 19 isotopes for red mullet. In spite that GSAs differed in otolith microchemistry, the attribution of individuals to the GSA of origin has been correctly predicted from only 30% of the European hake individuals when using otolith edge data. The percentage of correct classification increased to 63% when using only Western, Central and Eastern pooled areas but this increase should be interpreted with caution, given the small number of sampled areas for microchemistry analyses. For red mullet, correct allocation was 63% for edges and 66% for cores, albeit a high individual variability that decreased the classification power. Overall, these results can be explained by at least three compatible hypotheses: (1) otolith microchemistry may only reflect water mass features at another spatial scale; (2) the limits of biological populations may include several management units; and (3) alternative processes related with growth rate may also be affecting the microchemical composition and mask the link between water mass features and otolith composition. The outcomes of otolith shape and chemical composition analyses were combined under the hypothesis that this improves the capability of predicting GSA membership of a given fish from its otolith features. As regards European hake, data for both otolith shape and chemistry of 159 fish from 10 GSA subunits were used, while for red mullet 237 fish from 10 GSA subunits were used. Cross-validated correct predictions of population membership inferred from Linear Discriminant Analysis (LDA) increased to 42.2% after merging shape and chemistry data for European hake (with only shape data it reached 34.6%), while for red mullet it was 44.7% after merging shape and chemistry data, a bit less than using only the chemical composition of the otolith edge (47.2%). Therefore, combining the two sources of data implied a slight improvement of accuracy for hake but a slight decline for red mullet. All the trials showed moderate success and it is necessary to increase the spatial coverage and the total number of individuals to improve the stock identification using otolith shape and chemical composition. Analysis of otolith shape alone was performed on much larger samples for both species. The time and cost of otolith microchemistry analysis precluded the same sampling coverage. WP3 - Delineate fishing grounds and stock assessment aimed at identifying and characterizing fishing grounds over the Mediterranean waters, including where possible non-EU fleets (which is an aspect of great importance), even combining different data sources (e.g. Automatic Identification System - AIS, Vessel Monitoring System - VMS, experimental data from surveys, fisheries statistics data), processing and modelling approaches and devoting part of the work to the quantitative comparison of results. The main steps were: (i) estimation of the potential fishing grounds by processing AIS data and related hot spot analysis; (ii) Multi-Criteria Decision Analysis (MCDA) applied on environmental and fleet data for both large and small-scale components of the fleets and (iii) Cascaded Multilayer Perceptron Network (CMPN) applied on environmental and fleet data for the large-scale component of the fleets. The produced outcomes emphasize the importance of applying and comparing different methods. The effort dataset was reconstructed by processing AIS data and complemented using estimated (modelled) effort data in the southern/eastern basin, where the AIS is still poorly adopted and received. Outputs were validated against those freely provided by other projects (EMODnet, Global Fishing Watch) and, in some areas, by comparing the pattern obtained using only AIS data with the one obtained integrating AIS and VMS data (H2020 Project MINOUW). It must be acknowledged that this “new” fishing dataset covers for the first time the whole highly productive Mediterranean basin and may inform the wider scientific community, as well as those involved with policy and management, on fishing footprints in Mediterranean ecosystem. WP3 also delineated fishing grounds by species or group of species. A methodological framework, based on spatial analytical techniques, for combining the fishing effort (for bottom trawl, longlines, gillnets and trammel nets) and the spatial distribution of biomass for the investigated species, was applied. The main steps followed were: (i) estimation of the potential fishing grounds by species; (ii) hot spot analysis and (iii) the production of aggregated hot and cold spots. The outcomes demonstrate the possibility for fishers to harvest a certain species in a specific area, while providing spatial information about the number of exploited species in several fishing grounds. The main results revealed important fishing grounds over the entire Mediterranean Sea, namely in certain areas of the Adriatic, Tyrrhenian, Strait of Sicily, Aegean, eastern Ionian, Balearic, Alboran, Libyan and Levantine Seas. Within WP3 thirteen novel stock assessments were carried out and compared, when possible, with the assessments routinely carried out at the level of single GSA or combination of GSAs by GFCM or STECF. The new stock configurations for the six target species were based on WP1 and WP2 outputs. The results of the analyses show that, in most cases, the new stock assessments do not present particular improvements in term of diagnostics and model accuracy. The lack of improvement can be due to several reasons, apart from the new stock configurations (increased data heterogeneity when the number of aggregated GSAs is increased, model settings, etc.). However, it must be acknowledged that these trials represent a first and promising approach to the assessment of the new stock configurations and further investigation shall be implemented before scientific advice can be provided in a reliable and robust way. WP4 - Synthesis and proposals aimed at synthetizing and combining the results obtained in the WPs 1-3. A SWOT analysis provided an evaluation of the internal (Strengths/Weaknesses) and external (Opportunities/Threats) factors related to the methods applied in the WPs 1-3. The results stressed that the genome-wide methods can provide an accurate picture of the genetic differentiation, but if sample processing is not properly followed the quality of results can be greatly reduced. Thus, dedicated samplings are needed. The time to achieve all the process is rather long, about 1 year. The transfer of genetic samples can pose some difficulties, owing to the differences in the legal/bureaucratic aspects among countries. Otolith shape is not much demanding in terms of equipment and costs, while microchemistry is expensive. Samples of otoliths can be obtained from existing data collection campaigns, but otolith shape and microchemistry analyses are applicable only to fish. The identification of fishing grounds has fast progressed and the methodology is consolidated. Access to AIS time series needs a dedicated budget and the access to VMS data is not easy. The SWOT analysis highlighted the presence of advantages in each method, while at the same time limitations emerged for each approach, pointing out to the need for a continuous integration process with an exchange of knowledge and achievements among the research groups. A data collection for stock identification needs the strengthening of the cooperation and the necessity of investments in capacity building. It would be useful to develop a Regional Sampling Plan in cooperation between the Regional Coordination Group of the DCF and the GFCM. The explanatory role of environmental variables at spatial scale in delineating the population structures that emerged from the genetic and otolith analyses was explored using environmental data from Copernicus Marine Environment Monitoring Service (CMEMS, http://marine.copernicus.eu). The spatial origin of each individual was considered to reconstruct the spatial structure of the stocks configuration and to investigate the relationship of these stock structures with respect to the main environmental drivers. Fuzzy clustering was used in a first step. Different numbers of potential management units emerged. For European hake, three stocks were identified on the combination of the genetic and otolith shape data, one in the Western Mediterranean, one in the Adriatic-Ionian-Tyrrhenian basins and one Eastern Mediterranean. For Norway lobster four stocks were identified; one was characterizing the Adriatic Sea and the subareas of the Ionian and Aegean Sea, while the other three were distributed in the Western Mediterranean. For deep-water rose shrimp two stocks were identified and the discontinuity between the eastern and the western stocks was positioned in the Ionian Sea south of Italy. No stock was clearly identified for either Aristaeomorpha foliacea or Aristeus antennatus as the relation with the spatial and environmental variables was low and non-significant. For red mullet the conclusion is similar, even though there were some rather inconclusive evidences for the existence of 2 to 3 sub-populations in the Mediterranean. A denser sampling design, perhaps restricted to smaller areas, might help in confirming or rejecting this hypothesis. The most reliable and comprehensive stock configurations by species were selected and a series of interrelated maps prepared to illustrate a synthesis on the spatial correspondence between stocks, management areas, and fishing grounds. Finally, we tested the potential development of an adaptive spatial fisheries management through a simulation approach using three case studies. The new stock configurations of giant red shrimp and blue and red shrimp were examined in the Western Mediterranean, giant red shrimp in the Central Mediterranean and deep-water rose shrimp in the Adriatic-Ionian region. These species are covered by a Multiannual Management Plans. Management scenarios were run using bio-economic simulation models (BEMTOOL and SMART), considering the reduction of fishing effort, the improvement of the gear selectivity and spatial closures in areas critical for biological cycles of the targeted species (Essential Fish Habitats). Despite BEMTOOL and SMART assessed the same fisheries under the same scenarios, they were quite different in terms of results. These differences are mainly due to dissimilarities in modelled processes and assumptions. However, both models evaluated as the best management strategy, both in terms of gain in Spawning Stock Biomass (SSB) and improvement or light decrease of the current yield, the scenario characterized by a 10% reduction of the current fishing effort, coupled with an improvement of trawl net selectivity and protection of Essential Fish Habitats (EFHs). Finally, an “ideal” roadmap for developing adaptive spatial fishery management aimed to reduce uncertainty in assessment and management procedures was explored, considering the development of operating models, the simulation of alternative management strategies and evaluation of their performance (against biological, economic, and social objectives). The need for accurate knowledge of populations units and their connectivity was highlighted. Furthermore, an adaptive management aimed at protecting the spatial structure of the stocks should be considered as a main tool for the identification and protection of nurseries and spawning areas, in order to mitigate both growth and recruitment overfishing