Despite the high number of marine biodiversity indicators available, there were important gaps in terms of the MSFD requirements (Table 2; Berg et al., 2015). Additionally, there is insufficient information regarding the quality and confidence of the indicators (Queiros et al., 2016). Most indicators lack regional targets or quality threshold values, and few have a measure of confidence and a demonstrated link to pressures (Teixeira et al., 2014). Thus, although the indicators were operational in the sense that they were used in previous marine assessments, their applicability to fulfill the criteria of MSFD is less evident (Berg et al., 2015).

Major gaps observed for the Descriptor 1 were indicators to assess the biodiversity at ecosystem level and the genetic composition of populations, indicators for microbes, pelagic invertebrates (Cephalopods), and reptiles (Table 2). Abyssal and bathyal zones totally lack indicators addressing those depths (Hummel et al., 2015, Teixeira et al., 2014). Moreover, habitats with restricted distribution in the regional seas (like ice-associated species and communities) had a low overall number of indicators (Teixeira et al., 2014). Indicators for the MSFD biodiversity criterion 1.7 “Ecosystem structure” and the associated processes and functions are relatively scarce (Berg et al., 2015), and those few addressing this criterion focused essentially on communities, often in isolated components. Only few biodiversity indicators related to ecosystem processes and function were reported (Table 1), all of them for the North-Eastern Atlantic (Teixeira et al., 2014).

MSFD Descriptor 2 addresses the threat on natural biological diversity caused by the non-indigenous species (NIS), which are the taxa introduced outside their natural range and natural dispersal potential as result of human activities. Particularly the invasive NIS are recognized as a global threat to biodiversity (Olenin et al., 2010). Despite the requirements of MSFD Descriptor 2, there were no reported indicators in Europe specifically assessing the “Impacts of non-indigenous invasive species at the level of species, habitats and ecosystem, where feasible” (MSFD criteria 2.2., indicator 2.2.2; Tables 1, 2; Teixeira et al., 2014). In most cases, the MSFD initial assessments of the Member States did not include reporting of the adverse effects in biodiversity or the magnitude of impacts caused by NIS (Palialexis et al., 2014; Berg et al., 2015). RSCs (e.g., Barcelona and Bucharest Conventions, HELCOM, and OSPAR) have not included indicators for NIS impacts within their agreed set of indicators. Only recently there have been research activities to develop practical proposals to help managers assess these requirements in standardized ways (e.g., Zaiko et al., 2011; Katsanevakis et al., 2016; Nentwig et al., 2016; Rabitsch et al., 2016).

Most of the MSFD food web indicators (Descriptor 4) are related to “Abundance or distribution of key trophic groups or species” (MSFD criteria 4.3), and are thus simultaneously D1 (MSFD criteria 1.1) indicators (Berg et al., 2015). There is a need to develop indicators related to MSFD criterion 4.1 (“Productivity (production per unit biomass) of key species on trophic groups”) especially focusing on primary and secondary producers, as well as for the criterion 4.2 (“Proportion of selected species at the top of food-webs”; Table 2).

Finally, the remaining biological descriptor of the MSFD (Descriptor 6 on sea-floor integrity) is rather well developed in terms of indicators, having a high number of them (≥20 indicators), both for criterion 6.1. (“Substrate characteristics—physical damage”) and 6.2. (“Condition of benthic community”) for angiosperms, macroalgae, and benthic fauna. However, there is a need for developing indicators especially for cephalopods, but also for other biological components. Also there is a gap in the development of targets at the level of benthic habitats and ecosystems.

In order to address the identified gaps in the indicator availability for marine assessments, 16 new indicators were developed and 13 indicators were refined (Table 3) to better fulfill the MSFD requirements (Berg et al., 2016). The indicators were scored according to the eight indicator quality criteria listed by Queiros et al. (2016) in order to evaluate their fitness as potential indicators for the MSFD assessments (Figure 1). The quality criteria are: (1) Scientific basis, (2) Ecosystem relevance, (3) Responsiveness to pressure, (4) Possibility to set targets, (5) precautionary capacity, (6) Quality of sampling method, (7) Cost-effectiveness, and (8) Existing and ongoing monitoring data. In general, the newly developed indicators covered both pelagic and benthic ecosystems and addressed several biological components and habitats that were identified as lacking indicators or as being under-represented in several marine regions (Teixeira et al., 2014). In addition, 13 pre-existing indicators were further refined in order to improve their performance and confidence. For example, they were tested for responsiveness to pressures, or using new data sets to validate their applicability in wider sea regions.

The new developments and refined indicators by DEVOTES addressed some of the gaps identified with respect to the MSFD criteria and indicators in different marine regions where possible. In addition, DEVOTES indicator development focused on those biological components and habitats where monitoring data and the expertise in the DEVOTES research consortia were available.

Some of the indicators developed and refined were already established indicators, meaning that those are applied in the management frameworks such as EcoQO indicators (i.e., the OSPAR System of Ecological Quality Objectives—EcoQO—for the North Sea), or in the national MSFD monitoring of the EU member states (Table 3). However, many of the new indicators (such as new genetic indicators for microbes) are not yet used in the marine assessments meaning that those need to be tested in different marine regions and approved by the national and regional managers to be part of the MSFD monitoring and assessment programmes. Many of the NIS indicators were evaluated as relevant for MSFD, but those are not yet included in the marine monitoring programmes. Most of the fish and macrozoobenthic indicators developed and refined are used for food web and benthic integrity and biodiversity assessments being more mature for assessment purposes, while many of the phytoplankton and zooplankton indicators were judged not to be particularly useful for MSFD purposes, besides the new indicator on Phytoplankton food quality traits, that is currently considered as a candidate for a HELCOM core indicator.

Two of the identified gaps were the absence of indicators for biodiversity on a genetic level (Descriptor 1, criteria 1.3. Genetic structure of populations) as well as the lack of indicators for microbial communities. A DNA metabarcoding technique was applied to develop indicators to assess both microbial communities (Berg et al., 2016) and benthic invertebrate biodiversity (Aylagas et al., 2014, 2016). Genetic methods such as nucleic acid microarrays were considered a suitable methodology to quickly determine diversity and abundance of microbial communities (DeSantis et al., 2007). Marine prokaryotes respond rapidly to environmental changes and anthropogenic pressures and are thus considered as useful components for the assessment of microbial biodiversity, and impacts of eutrophication and toxic substances (Caruso et al., 2015). When the genetic methods were applied to identify microbial and benthic organisms in already established and tested indicator methodologies such AMBI (Borja et al., 2000), they appeared to provide a cost-effective and robust methodology for biodiversity assessment. Development of genetic tools is expanding rapidly and more effort on benchmarking and standardization will be needed to enable the use of genetic tools in biodiversity assessments in the future (Aylagas et al., 2016).

We also developed indicators to assess habitats of key species as indicators of biodiversity on ecosystem functions. Keystone species, such as species of seagrasses, kelps, and intertidal algae are recognized as effective ecosystem engineers forming vegetated habitats with multiple ecosystem functions including the stimulation of biodiversity (Gutiérrez et al., 2011) and the mitigation of climate change (Duarte et al., 2013). As increasing human pressures on coastal ecosystems threaten the continued supply of essential functions and services, the protection of marine vegetated habitats should be a management priority (Duarte et al., 2013). The high number of indicators available for vegetated habitats (Table 1) suggests that this is already well recognized and that these indicators (such as lower depth limit of macrophytes) are used in practice in marine assessments for policy purposes. In a recent review on seagrass indicators globally, Roca et al. (2016) carried out a meta-analysis and compared the applicability of different types of seagrass indicators to detect environmental improvement. They concluded that physiological and biochemical indicators are more suitable due their fast response to changes of environmental stressors than structural indicators (e.g., shoot density or biomass). Moreover, there is global and local evidence that biodiversity and top–down control strongly influences the functioning of threatened vegetated ecosystems, and indication that biodiversity is comparably important to global change stressors (Duffy et al., 2015; Matheson et al., 2016). Therefore key stone species prove well suited as operational indicators of biodiversity at the levels of species, habitat and ecosystems.

One of the major global threats to marine biodiversity is the spreading of invasive NIS (Costello et al., 2010). There are a number of regulations for controlling NIS, including the MSFD that aim to restrict the spreading of new NIS (Ojaveer et al., 2014). However, the intrinsic complexity of NIS for being detected on time implies the need to provide early warning indicators for tracking potential vectors of invasions as well as to assess the impacts of NIS on native species, habitats, and ecosystems. Therefore, DEVOTES developed and refined indicators to assess the impacts of invasive species which can be used to identify vulnerable areas, environmental targets and to prioritize management actions.

The MSFD also calls for indicators that address the deviations from the normal structure and functioning of the marine food webs. Such indicators would reflect distortions in the top-down control or bottom-up regulation of the food webs. The cascading impacts of top-down control reach into all trophic levels of food webs. As an example, removal of large predatory species can cause a relief on lower trophic levels causing increase on zooplanktivorous fish and thus, change composition and biomass zooplankton communities. Likewise, the bottom-up regulation based on the changes in the phytoplankton composition and quality of algae as a food source for zooplankton can change productivity and species composition of zooplankton impacting the fish communities. D4, food web, was identified to have relatively few indicators (Table 1), and DEVOTES developed and refined indicators that can be used to assess different aspects of food web structure and functioning, such as the abundance and distribution of fish indicators, productivity of key trophic groups (phytoplankton primary production) and phytoplankton composition as food web indicator providing an indication of the palatability of phytoplankton for zooplankton.

Some of the indicators developed or refined in the DEVOTES project were evaluated to show poor responsiveness to pressures (Table 3, Figure 1). Nevertheless, we suggest maintaining these ecologically relevant indicators in the assessment suites, particularly if they are collected on the side of other monitoring programmes and the monitoring would not be very costly. They could then be used as surveillance indicators and applied to complement the indicators with clear pressure-state links (ICES, 2014). Some factors used in the assessment of marine ecosystem health do not have a very clear pressure-state relationship (ICES, 2014). Many foodweb components are simultaneously affected by multiple pressures and processes, showing response to cumulative and synergistic effects, and singling out the effect of any one of those may be impossible. These surveillance indicators would be supplementary and provide information of the overall “health status” of the ecosystem even in the absence of clear pressure-state relationships.

Despite the current work to supplement the indicator suite already included in the DEVOTool catalog, important gaps still remain; many of the biological components and habitats are not adequately monitored to allow development and testing of potential indicators. Reptiles, such as the European sea turtles, Caretta caretta, Chelonia mydas, and Lepidochelys kempii, are all either endangered or vulnerable species (IUCN Red List), and thus require conservation measures through marine protected areas. There are specific regional or local programmes for sea turtle protection (such as Sea Turtle Protection Society of Greece ARCHELON) that provide information of the status of local populations and advocate conservation measures. Likewise, Cephalopods represent a group of species with only few existing indicators included in the DEVOTool Indicator Catalog. A recent ICES report on Cephalopod biology and fisheries stated: “[Despite the importance of several species for European fisheries, there is limited management of the fisheries and no routine assessment; data collection is often either not part of routine fishery data collection or the data are inadequate for assessment. Increasingly, however, cephalopods are seen as alternative target species to replace overexploited finfish stocks, and the growing fishing effort means that management will almost certainly be needed within the next few years. Also on the horizon is the development of commercial aquaculture].” (Jereb et al., 2015).

One of the major oceanic ecosystems lacking indicators is the deep-sea habitats and their respective communities (Teixeira et al., 2014). The deep pelagic ocean and deep seabed ecosystems represent the largest biomes of the global biosphere, but still the knowledge of their biodiversity, habitats and processes is scarce (Webb et al., 2010; Danovaro et al., 2014). The deep-sea physical, biogeochemical and ecological processes present distinctive features from other marine ecosystems (Danovaro et al., 2014; Thurber et al., 2014), and the specific habitats or ecosystems host very specific communities (Danovaro et al., 2010). Also many deep-sea invertebrates are exceptionally long-lived and grow extremely slowly (Clark et al., 2016). Current evidence indicates that cumulative stressors from e.g., fishing and resource exploitation will cause important and largely unpredictable ecological changes in these biotopes (Gramling, 2014; Clark et al., 2016). Removal of habitat-forming species, decline in diversity, change in abundance and biomass, reduction in distribution, change in community structure, namely its composition and food web structure are a few of the expected impacts (Clark et al., 2016). Climate change will further expose these already vulnerable ecosystems to combined stresses of warming, ocean acidification, deoxygenation, and effects of altered food inputs (Levin and Le Bris, 2015; Rogers, 2015). Due to the attributes mentioned, the recovery capacity of these deep-sea ecosystems is highly limited and predicted to take much more time to recover after pressures have ceased (Clark et al., 2016). Specific metrics of sensitivity of deep-sea fauna and habitats are therefore urgently needed for assessing the risk stemming from impacts and for identifying vulnerable ecological units (Clark et al., 2016). Baselines need to be established for diversity, abundance, and biomass of deep-sea ecosystems, particularly for the less studied pelagic realm and an understanding of ecological processes needs to be developed (Danovaro et al., 2014; Rogers, 2015). Such indicators will allow prioritizing areas for protection and designing more efficient monitoring programmes for the deep-sea realm.

There were not any indicators for the sea ice habitats, as these represent quite marginal habitat in the European regional seas. In the Polar Regions, sea ice habitats are important for the productivity of the sea and harbor rich biological communities and food webs associated with those (see Thomas and Dieckmann, 2010, for further references). Due to the climate change and warming of the Polar Regions, the extension of the seasonal, and permanent ice cover is shrinking with an alarming pace (Dieckmann and Hellmer, 2010). In the Baltic Sea, the seasonal ice cover is mostly restricted to the northern parts, and similarly as in the Arctic regions, it is an important habitat with rich community of ice-associated algae and micro-organisms as well as migrating birds, and as a primary breeding ground for the two seal species (Granskog et al., 2006). Climate change is projected to change further the biology and ecology of the Baltic Sea, including the diminishing duration and extent of the sea ice cover with its consequences to the ice associated biota (Viitasalo, 2012). However, due to the lack of monitoring and indicators for the sea ice habitat and biota, it is not possible to assess their impacts on biodiversity status in the ice-covered sub-basins of the Baltic Sea.

The conservation initiatives worldwide often share common assessment elements and make use of similar baseline information (e.g., Duffy et al., 2013; Pereira et al., 2013). Versatile use of indictors across environmental policies, geographical regions, and spatial scales is apparent as many of methods and biodiversity indicators are applied in several assessment or monitoring programmes (Teixeira et al., 2014). Approximately 30% of the indicators suggested to be applied for MSFD were already used for assessment needs of the other EU Directives or regulations (e.g., Birds and Habitat Directives, Water Framework Directive or Common Fisheries Policy). The need to economize marine monitoring, but yet maintain and enhance operational monitoring networks (Borja and Elliott, 2013) can benefit from careful planning of interoperable monitoring and assessment where the same indicators could be used for several purposes, and combined depending on the needs of each respective assessment purpose.

There are indicators used for MSFD assessments targeting threatened marine species included in the IUCN Red List, as well as several MSFD indicators miming the candidate metrics to Essential Biodiversity Variables proposed by Pereira et al. (2013, e.g., Abundances and distributions, Taxonomic diversity, Habitat structure, Allelic diversity). Also the nine Essential Ocean Variables (EOVs) on Biology and Ecosystem health of marine ecosystems under discussion can benefit from operational indicators included in the DEVOTool Catalog (Teixeira et al., 2014) or those refined and developed by DEVOTES (GOOS, 2016). These relate to the “Status of functional groups” and the “Health of living ecosystems” (Phytoplankton biomass and productivity, Incidence of harmful algal blooms, Zooplankton diversity, Fish distribution and abundance, Apex predator distribution and abundance, Seagrass cover, Macroalgal cover, Live coral cover, Mangrove cover). The wider use of environmental indicators applicable for various marine EU legislation and international agreements such as the RSC or the CBD (Zampoukas et al., 2012; Pereira et al., 2013) promotes harmonization between the different assessment systems and allows effective use of monitoring data for different reporting purposes, provided that the indicators satisfy specific quality criteria (Tittensor et al., 2014; Queiros et al., 2016) that should be common for all programmes.

European status assessments of marine biodiversity have chosen an ambitious path, where data-driven indicators with numeric thresholds should be used to depict definite status classifications. Moreover, the purpose is to link indicators to anthropogenic pressures and further to the human activities. Comparison with the assessment approaches in the marine and coastal areas of the United States (U.S.) shows that there is a conceptual difference, which is mainly due to the different understanding of the indicator concept and approaches for setting the assessment thresholds. A general trend in the U.S. assessments is to give scores for different indicators (or assessment questions) and the scoring is based on descriptive definitions. For instance, the National Oceanographic and Atmospheric Administration (NOAA) Marine Sanctuary Programme¹ makes status assessments based on expert analysis. The experts make the status interpretation based on 17 questions and the descriptions of status classes which are elaborated on the basis of monitoring data. Similarly, the NOAA fish stock assessments² and the Sea Turtle Assessment (National Marine Fisheries Service, 2013) are based on four criteria scored by experts and guided by information from the monitoring data. The National Estuarine Eutrophication Assessment (Bricker et al., 2007) uses quantitative thresholds for some eutrophication symptoms (e.g., percentage change in vegetation coverage) but the class boundaries are defined qualitatively using expert knowledge. The NOAA Marine Mammal Assessment (Carretta et al., 2015) is based on monitoring data but the assessment is descriptive, no specific status class is given, and the state of the populations is determined based on the viability analysis of the population. The U.S. National Park Service³ assesses the country's intertidal zone within protected areas based on expert interpretation of data variability and trends. The indicators in the U.S. Environmental Protection Agency (EPA) National Coastal Condition Report (US EPA, 2012) are closest to the European indicator concept, containing some biodiversity aspects, similar to EU MSFD, and using numerical indicator thresholds to define status classes. Due to the relatively strong European consensus on the indicator concept, the US indicators would not likely be applicable in the MSFD context. Though being used successfully in the US-wide assessments such as the National Coastal Condition Report, the US indicators do not have similar marine assessment framework as in the EU which aims at covering all marine elements, regions and pressures in a coherent and coordinated way.

The EU MSFD depicts a cyclical implementation and the next assessment of the environmental status of the marine environment is planned to be completed in 2018. Thus, the EU member states and RSCs are currently on the way in planning this assessment. Based on the experiences from the previous MSFD initial assessment (completed in 2012), the EU Commission is on its way to revise the earlier Commission Decision (2010/477/EU) to advice the on-going initial (MSFD Article 8) assessment that is due to 2018. Different aspects of biodiversity will be in the focus of the assessment, as the ecosystem services provided by the living part of the marine ecosystems are strongly dependent on structural and functional status of biodiversity. Monitoring and managing the health of the seas and oceans is highly relevant for the sustainable use of the marine resources particularly in the light of the recent Blue Growth initiatives is Europe (EU's Blue Growth Strategy⁴) and worldwide (e.g., FAO's Blue Growth Initiative⁵) that emphasize the sustainable use of marine resources, but with the expectation that more seafood, energy, and other living and non-living resources can be extracted from the seas. Full ecosystem approach, with concise cover of the marine ecosystem components, is needed for the managers to evaluate that Blue Growth is carried out sustainably, i.e., not threatening the future potential of delivering marine living resources and ecosystem services.

There is a need to have a comprehensive set of indicators, in order to cover all important multifaceted components of marine biodiversity assessments; ideally including both surveillance indicators and those with a clear pressure-state link. Some of the remaining gaps and further specific research needs are presented in Table 2. Further development and validation of marine biodiversity indicators requires improved data with better spatial and temporal coverage based on novel monitoring methods. In addition to the tests carried out in the DEVOTES project, more experimental indicator testing is needed to ensure their ecological relevance, robustness, and responsiveness to pressures, and also to enable incorporation into models in order to extrapolate marine assessments for into larger spatial regimes and temporal scales. With the help of combination of different tools, indicators covering both early warming and long-term assessment needs across different spatial scales can be combined into a holistic assessment of marine environment.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest. The reviewer IM declared a shared affiliation, though no other collaboration, with the authors MY and AB to the handling Editor, who ensured that the process nevertheless met the standards of a fair and objective review.