This report summarises the work conducted in Work Package 4.4 – BQE fish in transitional (i.e. estuarine and lagoon) waters (TW) within the project WISER under the sponsorship of the European Commission. It omits most technical details of the analyses given in the four previous Work Package reports, but still provides the necessary information to understand the rationale, approach and underlying assumptions necessary to discuss the results. The focus is therefore to discuss and integrate the results obtained within Work Package 4.4 and with this, make recommendations to improve fish-based ecological assessments in TW, principally estuaries and lagoons. In addition, and to assist with the WFD implementation which is the overarching theme of WISER, the deliverable includes, where appropriate, case studies where we have used multi-metric fish indices currently under development, or already in use for WFD compliance monitoring across Europe. Furthermore, results of the work package have been shared with relevant Geographical Intercalibration Groups (GIGs) supporting the harmonization and equalization process across transitional fish indices in Europe. Development strategies for fish indices in TW vary but generally include: (1) the calibration of metrics to anthropogenic pressures, (2) the development of reference conditions, (3) the calculation of ecological quality ratios, and (4) the designation of thresholds for Ecological Status (ES) class. New fish indices are developed for a defined geographical area, using specific sampling method and under locally relevant pressure fields. The former two factors, area and sampling methods, define the relevant reference condition in the calculation of Ecological Quality Ratios (EQR) and the latter factor, human pressures, define the index structure and especially the fish metric selection. To assess index relevance across areas, we calculated a suite of transitional fish indices on a standardized WISER dataset and then compared the agreement of the outcomes (using correlation analysis). The application of current indices to areas (or countries) different from the area in which it was originally developed leads to inconclusive or spurious results. The failure to accommodate the different indices to a standardized dataset in this work clearly demonstrates the fundamental reliance of current fish indices on the sampling methods and design of monitoring programmes used in the development of the index. Despite this, for some indices, correlations although weaker are statistically significant, also indicating the possible agreement in successful inter-calibration between these indices. Harmonization of BQE fish methodologies across Europe (common metrics) is unlikely by adapting or creating new fish indices but inter-comparison assessments are possible and valid using a common pressure index to harmonise different indices on a common scale. We found a negative response of fish quality features to pressure gradients which make BQE fish in TW suitable for greater ecological integration than other BQEs. However, successful assessment of Ecological Status (ES) require a matching combination of fish index, reference values and local dataset gathered with compatible sampling methods. Whole indices provide more consistent overall ES assessments but fish metrics considered individually may be more useful as a means to focus restoration measures. Future work is needed to identify those specific pressures affecting fish assemblages providing targets for minimising the effects of stress in mitigation and restoration plans. In order to achieve this, and although the interpretation of outcomes is still difficult, more recent transitional fish indices are leading in the use of comprehensive appraisal and validation exercises to test the responsiveness of BQEs for the assessment of ES. Here we proposed for the first time a simple sensitivity exercise under realistic scenarios of metric change to explore the expected inertia (i.e. the tendency to buffer ES change after quality alterations), dynamic range (i.e. the ratio between the largest and smallest possible ES values) and most relevant metric components (i.e. the those driving the most likely scenarios leading to ES change) from a multi-metric fish index under relevant human pressure gradients. Overall, the behaviour of multi-metric indices under manipulations of metric scores clearly indicated that metric type, number of metrics used and correlations between metrics are important in determining the index performance, with indices including more and/or uncorrelated metrics or metrics with skewed distribution being less affected by extreme metric manipulations. Results of this analysis may be used to set realistic management targets and also to identify the aspects of the indices that are more likely to affect the outcomes leading to more robust and responsive indices. Further improvements of fish indices may be attained by reducing the variability confounding biological quality metrics. This variability is undesirable noise in assessments and can be technical (i.e. linked to the method of assessment including sampling effort) or natural (physicochemical and biological). The implication for assessments is that different factors might then confound the metric-pressure correlation (the ‘signal’ in the signal-to-noise ratio in the assessments) increasing uncertainty in ES assignment. Models showed that salinity class, depth, season, time of fishing (day vs. night) and year of fishing may influence the values of the fish metrics. The modelling exercise also demonstrated that unexplained variance remains generally much higher within-systems than between-systems suggesting a higher importance of sources of variability acting at the WB level. Modelling and improved standardization in monitoring campaigns should reduce uncertainty in ES assignment. One important factor that was assessed further was the effect of sampling effort. The results suggest that richness-based metrics require larger sampling efforts although a similar effort-related bias may be an issue for density-based metrics if fish distribution is very patchy (i.e. schooling fish or those aggregated in specific habitats) and insufficient replicates are taken to fully characterise the patchiness in their distribution. It is apparent that to overcome a potential large source of error, the Reference Conditions must be defined according to the level of effort used in the monitoring programme or, conversely, the monitoring must be carried out at the same level of effort used to derive the Reference Condition. The WP finally explored the use of a predictive linear modelling approach to define reference conditions for fish metrics in transitional waters. The fish response data was modelled together with Corine Land Cover (CLC)-derived pressure proxies (percentage of agricultural, urban and natural land coverage). Based on the obtained models, the expected metric score was predicted by setting pressure levels either to the lowest observed pressure in the dataset or to zero in order to define the sample and theoretical reference condition, respectively. Even when significant, the effect of pressures on fish metrics was generally very weak, probably reflecting the use of too-generic pressure indicators (such as land cover data instead of more relevant estuarine proxies such as dredging, port development, waterborne pollutants, etc). The best explanatory models included sampling factors and natural characteristics considered important discriminant features in the definition of water body types. In particular, the present work argues for considering not only estuaries and lagoons as different typologies but also other natural and design characteristic such as the gear type, the sampling season and the salinity class. Furthermore, a relevant reference needs to account for survey design bias, including rare species contribution to assessment datasets, patchiness, choice of pressure proxies or sampling gear. The modelling approach of fish metrics against the physico-chemical variables has proved useful to derive Reference Conditions. This is important for the computation of relevant EQRs in Europe where there is a general lack of pristine areas or historical data on fish BQE and it provides an alternative to best professional judgment. Taking all WP analysis and case studies together, the work conducted has highlighted the following key messages and linked research needs necessary to optimize BQE fish for the quality assessment of transitional waters: Key Message 01: Harmonization of BQE fish methodologies across Europe (common metrics) is unlikely by adapting or creating new fish indices but inter-comparison assessments are possible and valid using a common pressure index to harmonise different indices on a common scale. Research needs to be focused on more widely-applicable fish indices will require the formulation of completely new indices based on a more flexible use of fish metrics according to system typologies, relevance and, probably, an increased use of functional traits. For current indices, further research on a method of intercalibration is needed. Key Message 02: BQE Fish in TW respond consistently to human pressure gradients across transitional waters providing the means to assess Ecological Status (ES). Further work will be needed to identify those specific pressures affecting fish assemblages providing targets for minimising the effects of stress in mitigation and restoration plans. Key Message 03: Although the interpretation of outcomes is still difficult, more recent transitional fish indices are leading in the use of comprehensive appraisal and validation exercises to test the performance of BQEs in the assessment of Ecological Status (ES). Further appraisal of fish indices behaviour is needed to understand the meaning of the quality outcomes, to set realistic management targets and also to identify the aspects of the indices that are more likely to affect the outcomes leading to more robust and responsive indices Key Message 04: Uncertainty levels associated with metric variability in multi-metric fish indices can be managed to increase the confidence in Ecological Status (ES) class assignment. Further research is needed to include knowledge of habitat partition within systems, to understand metrics behaviour and precision, to test new combination rules allowing metric weighting by robustness and importantly to evaluate more robust sampling tools and methods. Key Message 05: Reference conditions for BQE fish-based quality assessments can be objectively estimated using predictive modelling. Further refinements will require the use of better pressure proxies, robust metrics amenable to modelling and to account for survey design bias (effort and choice of sampling gear) at the relevant scales used in monitoring programmes.