The Baltic Health Index (BHI): assessing the socialecological status of the Baltic Sea

Thorsten Blenckner1 | Christian Möllmann2 | Julia Stewart Lowndes3 | Jennifer R. Griffiths1,4 | Eleanore Campbell1 | Andrea De Cervo1 | Andrea Belgrano5,6 | Christoffer Boström7 | Vivi Fleming8 | Melanie Frazier3 | Stefan Neuenfeldt9 | Susa Niiranen1 | Annika Nilsson10 | Henn Ojaveer9,11 | Jens Olsson12 | Christine S. Palmlöv13 | Martin Quaas14 | Wilfried Rickels15 | Anna Sobek13 | Markku Viitasalo8 | Sofia A. Wikström16 | Benjamin S. Halpern3,17


| INTRODUC TI ON
The health of the oceans and especially of their coastal areas is inextricably linked to human well-being and societal development, as marine ecosystems generate a large share of services needed and used by humans (Franke et al., 2020;Neumann et al., 2017).
Unfortunately, human activities often have negative impacts on marine resources  and utilization of ecosystem services (or benefits, Díaz et al., 2015) has caused rapid changes in coastal seas world-wide (Cloern et al., 2016;Duarte et al., 2020;Jouffray et al., 2020). Improving the health of coastal and open sea marine ecosystems, that is, sustainably delivering a range of benefits to people now and in the future (Halpern et al., 2012), hence represents a substantial challenge for marine resource management since it requires balancing human benefits and impacts on the ocean. This challenge is exacerbated by often limited capabilities to enact and enforce effective governance due to limited knowledge about the cumulative effects caused by the multiple pressures marine ecosystems presently face. Reaching sustainability goals through ecosystem-based management (EBM) of the oceans thus requires an understanding of interactions between nature, society and the economy (Crowder & Norse, 2008;Long et al., 2015;Merkel, 1998). This is especially relevant following commitment of the global community to the 2030 Agenda for Sustainable Development, where in particular the Sustainable Development Goal (SDG) 14 (Life below water) seeks a balance between environmental, economic and social sustainability in relation to oceans and coastal development (UN, 2015). Besides that, many other SDGs, such as SDG 3 (Good health and well-being), 8 (Decent work and economic growth), 9 (Industry, innovation and infrastructure), 11 (Sustainable cities and communities), 12 (Responsible consumption and production), 13 (Climate Action) and 17 (Partnerships for the goals) are relevant for developing the sustainable use and management of the Baltic Sea.
Consequently, there is an urgent need for adequate metrics and tools that quantitatively and comprehensively measure ocean and coastal ecosystem health for better monitoring of progress towards predefined management targets.
The Ocean Health Index (OHI) is a well-tested and widely applied approach to capture the human benefits and the interdependence between humans and nature (Halpern et al., 2012). It defines a healthy ocean as sustainably delivering a range of benefits to people now and in the future (Halpern et al., 2012). The OHI scores quantitatively a suite of socio-ecological benefits and ecosystem services (called goals in OHI) the ocean provides to humans (e.g. food provision, natural products extraction, and tourism and recreation) including conservation objectives (e.g. clean waters and biodiversity). These scores are calculated by measuring the status relative to their defined targets as well as the pressures and resilience measures that most influence that aspect of ocean health (Halpern et al., 2012).
Globally, the OHI framework has been used to annually assess 220 coastal nations and territories from 2012 to 2020 (Halpern et al., 2012. OHI assessments use open data science tools and best practices to ensure that methods are transparent, collaborative and repeatable. As each assessment can build directly on previous work rather than starting from scratch , it makes a valuable integrated evaluation tool that can inform EBM (Longo et al., 2017) and tracks the progress to reach the SDG targets . Furthermore, the OHI is a scalable approach which can be modified to match regionally or locally relevant questions and management targets that are framed by area-specific conservation objectives and data availability. Consequently, OHI assessments have been tailored to smaller areas at finer spatial scales in almost 20 places, from countries to smaller regions within countries . objectives and associated targets. Subregionally, the lowest BHI scores were measured for carbon storage, contaminants and lasting special places (i.e. marine protected areas), albeit with large spatial variation. 4. Overall, the likely future status of all goals in the BHI averaged for the entire Baltic Sea is better than the present status, indicating a positive trend towards a healthier Baltic Sea. However, in some Baltic Sea basins, the trend for specific goals was decreasing, highlighting locations and issues that should be the focus of management priorities.
5. The BHI outcomes can be used to identify both pan-Baltic and subregional scale management priorities and to illustrate the interconnectedness between goals linked by cumulative pressures. Hence, the information provided by the BHI tool and its further development will contribute towards the fulfilment of the UN Agenda 2030 and its Sustainability Development Goals.

K E Y W O R D S
ecosystem-based management, health, management targets, social-ecological system, sustainability We here present a first assessment of Baltic Sea health using the OHI approach. The semi-enclosed Baltic Sea is a classic example of a brackish ecosystem impacted by multiple anthropogenic pressures comprising eutrophication, elevated levels of hazardous substances, introduction of non-native species and habitat degradation as well as unsustainable fishing pressure (Elmgren et al., 2015;Reusch et al., 2018;Rickels et al., 2019). The Baltic Sea is also one of the fastest warming large marine ecosystems on the globe (Rutgersson et al., 2014). Cumulative effects of these multiple pressures have impaired the resilience of the Baltic Sea ecosystem (Korpinen et al., 2012) and substantially changed ecosystem structure and function (Casini et al., 2008;Lindegren et al., 2012;Möllmann, 2019;Möllmann et al., 2009).
Nine countries border the Baltic Sea and its catchment area has a total population of ~90 million people (Elmgren et al., 2015).
The most prominent examples are the two holistic environment assessments by HELCOM (Helsinki Convention, HELCOM, 2010, 2018b. However, while these assessments are strong in evaluating the impacts of human activities on the ecosystem, they have not been designed to include the benefits provided to humans. This lack of a human dimension may demotivate decision-makers to allocate sufficient resources for remediation or restoration of the Baltic Sea, despite the increasingly strong scientific evidence of the anthropogenic degradation of its status. Public policy needs to serve multiple goals and interests (e.g. species conservation, food production, aesthetic values, recreation, economic growth) and to objectively consider costs and benefits for restoration.
Hence, additional to ecological state, an assessment of ecosystem health through the human lens of meeting societal goals and delivering desired benefits and ecosystem services is needed (Halpern et al., 2014).
We introduce the Baltic Health Index (BHI) that tailors the OHI approach to the unique needs of environmental management of the Baltic Sea. Our BHI assessment presents the first transboundary application of the OHI framework in a region governed by a multitude of comprehensive national and international policies, and which can thus serve as an example for areas with similar policy landscapes in Europe and beyond. The BHI complements existing, more ecological-oriented assessments (e.g. HELCOM, 2018b) by providing a human dimension on the status of the Baltic Sea. Using the best available local data, we assessed the health of the Baltic Sea and its spatial variation. Here, we discuss our process and the implications of the results for local (e.g. bays and basins) and regional (e.g. Baltic Sea) management as well as future research.

| MATERIAL S AND ME THODS
We developed the BHI following the standard methodology of the OHI (Halpern et al., 2012, and tailored this assessment approach to best represent the social-ecological system of the Baltic Sea. In the process of developing the BHI, we followed four best practices : 1. incorporation of key characteristics and priorities of the study area into the OHI framework design before gathering necessary information; 2. a priori definition of spatial boundaries to achieve a balance between availability of data and operational management areas; 3. development of the goal models to provide a fuller picture concerning key characteristics and priorities outlined in (1); and 4. documenting and sharing data, methods and tools openly throughout the assessment process (GitHub, 2016;RStudio Team, 2016).
Overall, we developed the BHI based on openly accessible data, and conducted data preparation, combination and modelling in a transparent and repeatable way . Full details on the BHI calculation can be found at https://github.com/OHI-Scien ce/bhi, while data and code are available at https://github.com/ OHI-Scien ce/bhi-prep.

| Expert and stakeholder process
We designed an expert elicitation process involving a diversity of scientists and environmental managers to allow for an objective, transparent and well-informed BHI development and assessment. In this process, we engaged scientists and representatives from non-governmental organizations as well as from management authorities from the entire Baltic region in four BHI workshops. However, small-scale fisheries and tourist sectors were not included here as no representative was found. The goal of the process was better alignment of the global OHI-assessment framework to existing management targets for the Baltic Sea. At the first workshop (in 2014), potential BHI goals and data availability for these as well as pressure and resilience were discussed. In the second workshop (in 2015), the final BHI structure and data sources to be used were agreed upon. Subsequently, the BHI core team gathered the data in a continuous dialogue with 'goalkeepers' (see below for more information) to assure the quality and proper interpretation. Preliminary BHI calculations were discussed in the third workshop (in 2016) and critically evaluated. Some goals were subsequently recalculated, and the revised results were presented for collective agreement and support from experts.
At the beginning of the process we assigned a 'goalkeeper' to each BHI goal (see below), that is, an expert in a particular field/subject, to ensure the scientific quality of each goal in the assessment.
Goalkeepers supervised the whole BHI process, especially decisions on data use and treatment for goal calculations, as well as decisions on management targets. We repeated the expert elicitation process several times, both using remote communication tools and expert meetings, assuring consistent implementation of newly available information and data. During the three expert workshops, every goalkeeper was also part of all other goal discussions, which opened up some general discussions, such as 'do we use a similar approach for setting management targets across goals'. The entire expert and stakeholder process strongly facilitated a close cooperation between goalkeepers and the BHI team and helped in integrating diverse knowledge and data in one comprehensive assessment product.

| Assessment regions
We divided the Baltic Sea into spatial units that account for the large heterogeneity in climate, hydrography and biodiversity as well as geographical and social gradients. We initially used the 17 Baltic Sea sub-basins (in line with the second holistic assessment of HELCOM, 2018b) and subsequently intersected these with the boundaries of the nine nations bordering the Baltic Sea (territorial waters and exclusive economic zones, EEZ) using the geographic information system ArcGIS (ESRI, 2016). For the resulting 42 BHI regions (see Figure S1 in Supporting Information), goal scores were computed and then aggregated into (a) region-specific scores and (b) scores for the entire Baltic Sea as a whole (using area-weighted averaging).
For some of the goals, only one value existed for the whole Baltic Sea, for example, for the sprat biomass (NP), and in these cases, all smaller BHI units were assigned the same score value.

| Baltic Health Index goals
The BHI assesses nine of the 10 goals initially outlined in the OHI (Halpern et al., 2012, ohi-science.org/ohi-global). We excluded the coastal protection goal, since coastal erosion is a minor issue in large parts of the Baltic Sea due to the shallowness of the coasts and sheltering archipelagos. However, due to future climate change and potential sea level rise, this goal will likely need to be included in future assessments. The definition of the goals and their reference points are tailored to best address critical management and policy objectives for the Baltic Sea (see Table 1, and for all goal-specific models and more detailed information see Supporting Information).
Each goal score is calculated along four dimensions ( Figure 1): 1. Present status x is a goal's current value compared to its reference point, that is, the management target.
2. Trend T is the average percentage change in a goal's status over the most recent 5 years. Each of these dimensions incorporates both ecological and social data as the focus of the assessment is on the human benefits derived from the ecosystem.

Pressures
The overall goal scores are calculated as the average of present (x) and likely future status. Likely future status is calculated as current status modified by the recent trend (T), cumulative pressures and resilience (r) associated with the goal. Each goal status and trend are calculated individually by goal and region (see Figure 1 and Halpern et al., 2012; for more details). Below we describe in detail the development and assessment of the various goals within the BHI (see also Table 1). The maximum score for each goal and the entire BHI is 100, where 100 does not represent pristine conditions, but instead represents if the reference points (shown in Table 1) are achieved. The flowerplots in Figure 2 were produced using the circlize tool (Gu et al., 2014).

| Artisanal fishing opportunity (AO)
The AO goal assesses the opportunities to engage with coastal non-recreational fishing. For the BHI, we focused on coastal fish stocks as a proxy for fishing opportunities and used abundance data for coastal piscivores, cyprinids and other mesopredator (i.e. mid-trophic level fish) species (see HELCOM, 2018b). The AO model assesses the health of these fish stocks, represented by the mean of two HELCOM Core Indicators for stock abundance (HELCOM, 2018b) and we used the good environmental status (HELCOM, 2018b) as the reference point for the AO goal.

| Biodiversity (BD)
Contrary to the global OHI, we did not separate the BD goal into species and habitats but instead combined both together. We used the already available assessment results from HELCOM (2018b), which consist of five components: benthic and pelagic habitats, fish, mammals and seabirds and has been evaluated using the biological quality ratios and seabird abundance, derived in the integrated biodiversity assessments from HELCOM (the HELCOM assessment tool: https://github.com/NIVA-Denma rk/ Balti cBOOST).
These are based on core indicators for key species and species groups, including abundance, distribution, productivity, physiological and demographic characteristics. Statuses of these five biodiversity components are aggregated first within each component, combining coastal area values with area-weighted averages, then combining the values for coastal and offshore areas of each BHI TA B L E 1 Goal description, its definition and reference points used

| Carbon storage (CS)
The CS goal assesses the potential of coastal vegetation to capture and store carbon and uses data on spatial coverage of eelgrass Zostera marina from the HELCOM HOLAS assessment (HELCOM, 2010).
Carbon stocks in coastal sediments and ecosystems are substantial compared to the open ocean (Regnier et al., 2013), but often data are very limited (Testa et al., 2017). As reference point we used the spatial extent of the presence of eelgrass before and after 1995, with the exception of low-saline areas such as the Gulf of Bothnia where eelgrass does not naturally occur (Kindeberg et al., 2019;Röhr et al., 2016). We are aware that the confidence of this goal might be low as many other coastal vegetation could not be included, but we wanted to include the goal as the potential to capture carbon is an important ecosystem service.

| Clean water (CW)
In contrast to the global OHI , we as- and #180), and perfluorooctanesulfonic acid (PFOS) in fish. These indicators were selected because the substances are hazardous (i.e. they are persistent, bioaccumulative and toxic) and pose a risk to organisms living in the Baltic Sea and to humans, and there is monitoring data available. As reference levels, we used targets agreed on internationally (see Supporting Information). In addition to the three contaminant indicators, we calculated the monitored proportion of persistent, bioaccumulative and toxic substances of very high concern (SVHC) on the F I G U R E 1 Illustration of how the BHI score is calculated based on the Ocean Health Index framework. Note that the status will rise relative to current trend trajectory if resilience exceeds pressures, though this could still mean a decline in status (if the trend is strongly negative); and similar (but opposite) for when pressures exceed resilience, that is, status will fall relative to trend trajectory F I G U R E 2 Spatial patterns in BHI scores. The large flowerplot indicates overall BHI score (index, centre number), with petal lengths indicating relative values (0-100) for each goal and subgoal. The lengths of the bars transecting each goal and subgoal petal represent the spatial variability of the score values of the particular goal or subgoal. Small flowerplots indicate basin-specific score (centre number) and goal values (petal lengths) of the major basins. Colours correspond to goals and subgoals, and the goal petals on small plots correspond with goals indicated on the large flowerplot. The width of each petal represents the contribution to the Index score. The results for the smaller basins can be found at Table S2 in Supporting Information and at: https://balti c-ohi.shiny apps.io/dashb oard Candidate List (Annex XV) of the EU chemical regulation REACH (also part of the BHI Resilience assessment, see below).
The reference point is having all contamination levels of the three pollutants/pollutant groups fall below their respective thresholds, and all persistent, bioaccumulative and toxic SVHC monitored.
The trash subgoal assesses the ability to prevent litter from entering the sea and harming the coastal and marine environment.
Marine litter is a global concern, impacting all marine environments. For the Baltic Sea, no comparable long-term trash datasets exist. We, therefore, used model data on the countries' amount of mismanaged plastic litter that has the potential to enter the ocean (Jambeck et al., 2015 approach did not account for the drastic reduction in growth and body condition for the Eastern Baltic cod stock over the past two decades (Casini et al., 2016). To account for this reduced body size, we penalized the Eastern Baltic cod score using Fulton's K condition index, a proxy for the cod condition (Casini et al., 2016, see also Supporting Information).

| Livelihoods and economies (LE)
The LE goal contains the two subgoals: livelihoods and economies.
While the subgoal livelihoods aims to assess employment in maritime sectors, data on employment in specific marine-related sectors in the Baltic Sea coastal areas were not available at a regional level.
Hence, we used the finest-available regional data (Eurostat NUTS2 regions) on employment rates for the age group 15-64 (Eurostat, see Supporting Information), assuming these to reflect a similar employment situation in the marine sectors. As a reference level we used maximum region-to-country employment ratios of the past 5 years, and highest country employment rate in the last 15 years.
The region-to-country ratio puts the value into local context, then adjusting with respect to highest country employment rate in the last 15 years from around the Baltic Sea situates the ratio in broader geographic context.
We computed the economies subgoal using sector-specific values (gross value added) associated with maritime-related industries and a 1.5% annual growth rate as the reference level (EC, 2013).

| Natural products (NP)
The assessment of the NP goal was restricted to the small pelagic fish sprat (Sprattus sprattus) which is mainly used for fish meal production or animal food (see Supporting Information). The goal was assessed using spawning stock biomass and fishing mortality data as well as related MSY reference points from ICES (2020). No data for other natural products were readily available at the time of the assessment.

| Sense of place (SP)
The SP goal contains two subgoals, namely iconic species and lasting special places. We derived a list of 15 iconic species: cod, flounder, herring, sprat, perch, pike, salmon, trout, white-tailed sea eagle, common eider, grey seal, harbour seal, ringed seal, harbour porpoise and European otter, from a survey sent to 89 experts (36 responses) from Baltic Sea countries. These species were then assigned a threat category (ranging from 'extinct' to 'least concern') based on International Union for Conservation of Nature assessments (IUCN, 2015), and assigned a numeric weight based on that category.
We calculated the goal score as the average weight of all species assessed (see biodiversity goal for more information) with a reference level at which all species are in the 'least concern' category.
For the subgoal lasting special places, the designation and man-

| Likely future state
Our strong focus on sustainability in the index calculation requires that both the current status and the likely direction of change in this status influence the score of each goal. We explicitly focus on the near-term future (future trends are calculated over 5 years) rather than longer term sustainability because the near-term future is most relevant to policymakers and long-term future states of many of the subgoals are very difficult to project. To improve our understanding of the likely near-term future condition, resilience and pressure dimensions are included to provide additional information beyond the recent trend. The OHI approach identifies those factors that negatively affect a goal as pressures and those that positively affect a goal as resilience (see Section 2). The expectation of a likely future condition suggested by the trend will become more or less optimistic depending on the effects of pressures and resilience ( Figure 1). Note that the likely future status does not predict the future, but only estimates what the status score is likely to be in approximately 5 years hence, given what is known today about recent trends and the counterbalance of pressure versus resilience metrics.

| Pressures
We used readily available data at consistent spatial scales for the (score = 2) or 'low' (score = 1; see Supporting Information, Table S3 for pressure rankings for all goals). Subsequently, we summed the weighted intensities of each stressor within a pressure category and divided the value by the maximum weighted intensity that could be achieved by the worst stressor across all categories (Halpern et al., 2012).

| Resilience
In both the OHI and BHI, resilience contains three components: ecological integrity, goal-specific regulations and social integrity. The ecological integrity is measured as biodiversity (same as in the BD goal). Social integrity is measured by using the World Governance Index, same as in the global OHI assessment (Halpern et al., 2012 compliance with the legal text. A fail score was given when the country had an obligation to report, but it did not obey, when it did not follow the legal instructions respecting thresholds or minimum legal standards, or when it failed to take action when it was required to. The not applicable score represents either a lack of information in the report regarding a specific obligation or indicates that compliance was not assessed at all, such as in the case of Russia where EU legislation does not apply. We decided to weight the different directives according to their specific assessment quality and ability to assess compliance (see Table S5 in Supporting Information).

| BHI goal scores and its regional variability
Overall, the regional BHI scored 76 out of a possible maximum of 100 ( Figure 2) indicating that the health of the Baltic Sea is suboptimal, and that substantial efforts are required to reach the management objectives and associated targets. Subregionally, the lowest BHI scores were observed for the Gdansk Bay (55), Kiel Bay (65) The Sound, score = 96), in most subregions of the Baltic Sea, the carbon storage potential is assessed to be very or extremely low. These results are mainly due to the use of eelgrass, a marine seagrass with high carbon sequestration capacity (Boström et al., 2014;Kindeberg et al., 2019;Röhr et al., 2016Röhr et al., , 2018

| Clean water and fisheries-The major concerns
Traditionally, the major ecological concerns in the Baltic Sea include the availability of clean water and the sustainability of fisheries (Elmgren et al., 2015). The BHI goal of clean water comprises the levels of contamination with various chemical substances, nutrient inputs from multiple sources (mainly agriculture but also waste water treatment plants, industries, managed forestry, storm overflows and natural background sources; Heiskanen et al., 2019), and the recently developing concern of trash polluting the marine  (Sobek et al., 2016). Hence, we added to the original contaminants score, the proportion of monitored persistent, bioaccumulative and toxic SVHC (see Section 2) to account for lack of data and knowledge on currently used and emerging hazardous substances. This modification of the goal resulted in the overall low score because only a small number (spatial average is 40%, ranging from 0 to maximum of 63%) of these new hazardous substances are currently monitored in the entire Baltic Sea. The low score can thus be seen as a result of lacking data and knowledge (see Section 2), rather than an assessment of the known impact of contaminants. We hope that future management and monitoring will broaden the scope with less focus on legacy contaminants and more emphasis on the challenges and potential risk caused by new and emerging SVHC contaminants as well as combined effects caused by mixtures of chemicals.  (64)) as well as Bornholm Basin (67). However, the most northern basin, Bothnian Bay (83) and a few areas in the south-western Baltic Sea, for example, Great Belt (99) and Kiel Bay (99) are characterized by relatively high scores.
The impact of fishing has also been one of the major concerns in the Baltic Sea, particularly concerning the major commercial targets cod and herring. Stock assessments for both species were the basis for our fisheries subgoal (see Section 2), which scored relatively high for the entire Baltic Sea (average 82, lowest 49, highest 96).
However, there is a large discrepancy in the status of the herring and cod stocks, the latter (Western and Eastern Baltic cod) being below sustainable MSY reference points due to overfishing and environmental change (Casini et al., 2016;ICES, 2020;Möllmann, 2019;Orio et al., 2019). Excluding herring stocks from the score, which are in a better state (see Supporting Information) decreased the overall regional score to 73.
Artisanal fishing opportunities (AO) scored high but varied across the Baltic Sea region (average 93, lowest 66, highest 100).
The lowest score was calculated for the coastal areas in The Sound (66) and the highest for the coastal areas in Eastern Gotland Basin (99), Bornholm Basin (100), Kattegat (100), Bothnian Bay (100) and Northern Baltic Proper (99). Note that the level of confidence in the AO assessment differs throughout the Baltic Sea, but is higher in those areas having the longest data series. Coastal fish communities are local in their appearance (Olsson et al., 2011;Östman et al., 2017) and the current monitoring programs do not cover all coastal areas. As such, the reference points are locally derived and varies between areas and coastal fish communities (HELCOM, 2018b;Olsson, 2019). In several areas, the data available represent shorter time series (<10 years of data) which also limits the confidence of the status assessment. Furthermore, we currently lack a comprehensive compilation of data on access to the fishery, which limits the applicability of the AO goal in assessing the provisioning of ecosystem services by coastal fisheries. However, it is essential to include this goal as it represents the coastal fishing opportunity, which is an important use of the Baltic Sea environmental resources distinct from the commercial fishing represented in the food provision: fisheries subgoal, and not captured elsewhere in the BHI.

| Likely future status
An important asset of the BHI compared to other assessments of the Baltic Sea such as HOLAS II (HELCOM, 2018b) is the consideration of the likely future change of the socio-ecological system, and that this new approach accounts for the relative effects of human pressures on and resilience of the ecosystem centrally in the assessment (Halpern et al., 2012). However, the BHI approach embeds the challenge of anticipating likely future direction of change in status. We hence deliberately focused on the near-term future, that is, 5 years only, rather than long-term sustainability, since long-term future states are difficult to project and are associated with high uncertainty. Furthermore, the short-term future is also most relevant to policymakers, which makes the results useful for deciding upon urgent measures needed to remediate the state of the sea.
To further illustrate the interactions between the different BHI components that are involved in the likely future calculation (i.e. resilience, pressures and trend, Figure 1), we demonstrate the various effects of these components in computing the eutrophication subgoal across three Baltic Sea basins (Figure 3). In the Bornholm Basin, the likely future status was higher (71) than the current status (64), due to a positive trend (+0.1) of the eutrophication indicators during the last 5 years combined with a high resilience (66) in relation to pressures (44, Figure 3). In the Bothnian Sea, the likely future status (79) was higher than the current status (67) even so the trend was negative (−0.14), but note that here the resilience score (84) is very high in relation to the non-existing pressure (0, Figure 3), as the maximum allowable input of the nutrient loads of both nitrogen and phosphorus (both are the pressure component) is below the thresholds (set by the HELCOM BSAP process).
In contrast, in the Gulf of Riga, the likely future status was lower (48) than the current status (52) due to a strongly negative trend (worsening of the status, −0.4), even though, the resilience (73) was higher than the pressure (21, Figure 3). The negative trend indicates that urgent management actions are needed for improving the eutrophication status in the Gulf of Riga, in particular as also long-term projections indicate a slow recovery from eutrophication (Murray et al., 2019).
Overall, the likely future status of all goals averaged for the whole Baltic Sea is higher than the present status (Figure 4) Table S6). to understand and estimate likely future changes, which can serve as indicators for management priorities.  (Giakoumi et al., 2015;Hunsicker et al., 2015). In general, reference points should be science informed, but optimally would be to develop these reference points in a co-design process with diverse stakeholders and scientists in order to define goals of restorative and active intervention and implement appropriate management measures (Franke et al., 2020). Learning exercises are needed to successfully operationalize and implement ocean management strategies that integrate environmental, social, cultural, health at local and regional scales.

| Reference levels
These different 'management experiments', that is, adjusting different solutions in different regions, can help to potentially overcome conflicting societal interests and to identify common values (Franke et al., 2020). Therewith, different regions could learn from different management practices.
The purpose of this Perspective is to highlight the need to (a) provide a conceptual and simultaneously operational ocean health framework that integrates the links between ocean and human health and (b) address potential solutions and obstacles to sustain and restore a healthy and productive ocean for future generations through advancing approaches for a broad transdisciplinary integration of marine sciences.

| CON CLUS IONS
We conducted the first assessment of Baltic Sea health following the internationally applied OHI framework. Such a quantitative and comprehensive assessment requires robust and continuous monitoring F I G U R E 4 Difference between likely future and current status for each goal and subgoal for the whole Baltic (point) and the subregional variability (horizontal lines, with shaded density curves). The horizontal lines show min-max ranges and the shaded areas show the distribution of subregions' associated values across the ranges, that is, the thicker the shaded area, the more subregions with values in that vicinity. Values greater than zero on the x-axis (likely future minus status) indicate the likely future status is greater than the current status, while values less than zero indicate expected decline in status going into the future data and needs to be flexible to include new studies, indicators and knowledge, especially for target setting (Borja et al., 2016).
One objective of this initial BHI assessment was to identify lacking or deficient data ( There are no comprehensive data on areas providing cultural services, therefore marine protected areas and their management plan implementation are used as proxy. This assumes that the existing MPAs are chosen to represent important cultural values

Tourism
No Baltic Sea-wide ecotourism data and sustainable reference levels exist macrophytes than eelgrass. The livelihoods and economies goal was also difficult to calculate because no indicators exist on the level of sustainability of economic activities. Further, in some cases, reference points are either unclear or missing. Future BHI assessments will hopefully benefit from recent sampling and monitoring activities.
Our first BHI assessment is not the final word on the health state of the Baltic Sea. It however provides a robust platform for a constructive dialogue on strengths and weaknesses as well as the required next steps to improve the assessment. In this first assessment, the BHI added new dimensions beyond previous Baltic Sea assessments by integrating new goals such as livelihoods and economies, natural products, carbon storage and tourism, along with advancements of resilience metrics based on countries compliance assessment to regulations, future trend assessments and transparent, reproducible methods with openly available code (https://github.com/OHI-Scien ce/bhi-prep). Furthermore, BHI takes the likely near-term future of the goals into account when assessing their state, which helps focus on mitigation actions on the most severe pressures.
The BHI outcomes can be used to identify both pan-Baltic and subregional scale management priorities (focusing on goals with low scores and likely future status) and to illustrate the intercon-

CO N FLI C T O F I NTE R E S T S
Andrea Belgrano is an Associate Editor for People and Nature, but was not involved in the peer review and decision-making process.
There are no other conflicts of interest. All the authors contributed to the design of the study, the development of the goals and the writing of the manuscript. All the authors approved the manuscript for submission.

DATA AVA I L A B I L I T Y S TAT E M E N T
Data and the code can be found at https://balti c-ohi.shiny apps.io/ dashb oard.