We searched for publications in the main scientific literature databases, namely Scopus, ISI Web of Knowledge, ScienceDirect, JSTOR, Blackwell Synergy, and Ingenta Connect. To identify relevant publications, we used the search string “(marine OR coast* OR sea OR ocean) AND (“cultural ecosystem service*” OR “cultural service*”)” on article title, abstract or keywords, or only on abstracts, depending on the searching options available on the bibliographic databases. Searches included all articles published until our cut-off date of 21 March 2016.

We established four criteria to include publications in our quantitative synthesis. Namely, the publication had (1) to conceptualise, map, quantify and/or value CES; (2) to assess marine or coastal ecosystems; (3) to be peer-reviewed; and (4) to be written in the English language. We limited the scope of our analysis to publications using the concept of ecosystem services because this concept is increasingly employed in the understanding, management and governance of natural environments (Hansen et al. 2015). Despite persistent critique, it has several strong features, such as revealing and structuring human benefits from nature; creating objects that can be assessed regarding trade-offs and synergies occurring in social-ecological interactions; supporting biodiversity conservation; reconnecting society with natural environments; and linking natural, social sciences, and humanities (Schröter et al. 2014).

Our systematic review followed the Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA) statement (Moher 2009). The PRISMA provides a checklist of items (Suppl. material 1) that is reported to enhance the transparency and robustness of the review process. Following the PRISMA, we divided the systematic review in four phases (Fig. 1). First, we identified 206 potentially relevant records. Among these, we obtained 187 publications from bibliographic databases, and identified further 19 potentially relevant publications in these articles’ reference lists. Second, after removing duplicated entries, we screened titles and abstracts of 120 publications that resulted in the rejection of 23 publications because they did not meet one or more of our inclusion criteria. Third, we read through full-texts to examine whether publications were eligible for our final quantitative synthesis. At this phase, we assessed 97 publications of which 25 failed to meet our inclusion criteria. Finally, we carried out a quantitative synthesis of 72 publications.

We undertook a second round of reviews of a random 25% sample of the initial 120 publications to validate their eligibility for the quantitative synthesis (Pita et al. 2011). Twelve of the co-authors read through the full-texts of the sampled publications and confirmed the decisions made in the first round of reviews, except for two disagreements. The co-authors involved in the first and second rounds of reviews of these two publications deliberated over the eligibility of the studies to be included in the quantitative synthesis and arrived at a common decision. The number of eligible publications of the first round of reviews did not change. Conducting two rounds of reviews enhanced the validity and robustness of the entire review process.

Before starting the data collection, we carried out a ‘calibration’ exercise to attain a uniform data collection procedure among co-authors. This consisted of reviewing a randomly selected publication that had been previously identified for quantitative synthesis. Each co-author assessed this publication individually and subsequently the results were compared against each other. The outcome of the exercise resulted in our template for the data collection process.

The systematic review data included 20 variables and their corresponding response categories (Table 1). Data variables included CES classes assessed; quantification, valuation, and/or mapping methodologies; synergies and trade-offs identified; links between habitats, CES, and human wellbeing; countries of case studies; drivers of change affecting ecosystems and their CES; and others.

We used descriptive statistics to calculate the number of publications assessing CES classes, classifications used, type of publication, spatial scale, type of assessment, and whether publications assessed relationships between human wellbeing, ecological integrity (Müller 2005) and CES. We have searched for authors' main discipline(s) in their institutional and personal webpages to know the most prevalent disciplines in marine and coastal CES research and to know whether research teams were interdisciplinary. Additionally, we compared the number of publications assessing marine and coastal ecosystem service categories, i.e., provisioning, regulating and cultural, with the number of publications only assessing marine and coastal CES. To this purpose, we compiled data from Scopus between 1995 and 2015, which we obtained using a similar search string as the one described before, but excluding the search term ‘cultural’. We did not screen these records because this analysis did not intend to be exhaustive, but instead served as an indicator for comparison between the assessment in the literature of marine and coastal CES and other ecosystem service categories.

To compare assessments, we adapted the CES classes found in the literature (Millennium Ecosystem Assessment 2005, TEEB 2010, Beaumont et al. 2007, Böhnke-Henrichs et al. 2013, Liquete et al. 2013, UK National Ecosystem Assessment 2011) to the Common International Classification for Ecosystem Services (CICES) (Haines-Young and Potschin 2013). CICES is arguably the most comprehensive and adaptable ecosystem services typology available to date (Turkelboom et al. 2013). However, we modified this typology and expanded it to cover the diversity of CES found in the literature (Suppl. material 2). Specifically, we added recreation and leisure; cultural heritage and identity; spiritual, sacred and/or religious services; inspiration for culture, art and design; sense of place; social relations; and services of humans to ecosystems – a new class proposed by Comberti et al. (2015). A few CES did not fit well into any of the previous or the CICES classes, and thus we classified them under the CICES group level (instead of class level as above) of ‘intellectual and representative interactions’.

To analyse the relationships between drivers of change, habitats and CES, we created matrices linking drivers to habitats, and habitats to CES. Each matrix cell contained the sum of the number of links identified by the empirical studies of our database. We created a third matrix linking drivers to CES by multiplying the two previous matrices. We then applied hierarchical clustering on the Euclidean distance between the rows and columns of the matrices to group drivers affecting CES (Rocha et al. 2014), according to the number of links found in the literature review. For this analysis, we used the heatmap.2 function from gplots package (Warnes et al. 2016) of R version 3.2.2 (R Core Team 2015).

Synergies and trade-offs (Howe et al. 2014) were also a focus in our systematic review. We examined whether publications specified synergies and/or trade-offs within CES, or between CES and other ecosystem service categories, or between CES and human-related activities. Moreover, we investigated whether publications linked explicitly the benefits of supplied CES to human wellbeing, and whether they specified changes in wellbeing or proposed wellbeing indicators (Breslow et al. 2016). In a similar way, we analysed whether publications related CES to the ecological integrity or state of marine and coastal ecosystems, and whether they indicated positive or negative effects of ecological integrity variables on CES (Müller 2005, Kandziora et al. 2013).

Finally, we mapped the geographical location of case studies found in the literature (n=60), using QGIS (Quantum GIS Development Team 2016). Besides location, the maps also depict the number, spatial scale and type of assessments per country, the CES classes evaluated, and the habitats assessed.

Our literature synthesis includes 72 publications (Suppl. material 3) where the most assessed marine and coastal CES classes were recreation and leisure (60), aesthetic services (44), and cultural heritage and identity (39) (Fig. 3). Bequest (10), symbolic (2), and services to ecosystems (1) were the least represented CES classes. On average, four CES classes were assessed per publication, but 12 studies assessed just one class.

The 72 publications synthesised include 46 empirical studies, 19 literature reviews and 4 conceptual papers (Fig. 4a). Three publications are literature reviews but present new case studies (Everard et al. 2010, Jordan et al. 2012, Satterfield et al. 2013), thus we classified them as both review and empirical papers. To classify CES in all conceptual, empirical and review papers, authors mostly used the MA classification (46) (Fig. 4b), followed by adapted, combined or new classifications (13) (Chan et al. 2012a, Chan et al. 2012b, Beaumont et al. 2007, Böhnke-Henrichs et al. 2013, Liquete et al. 2013, Kandziora et al. 2013, Farber et al. 2006, Groot 2011, Loomis and Paterson 2014, Fletcher et al. 2014, Papathanasopoulou et al. 2015, Comberti et al. 2015, Chung et al. 2015). Several publications did not specify any classification (12), and one publication (Jobstvogt et al. 2014) used the UK National Ecosystem Assessment classification. Remarkably, no publications used the CICES alone (Haines-Young and Potschin 2013), although it was adapted and combined with other classifications in two publications (Papathanasopoulou et al. 2015, Chung et al. 2015). Among the 46 empirical publications, there were 60 case studies with a diverse range of spatial scales (Fig. 4c). Regional assessments (22) are the most common in the literature, followed by local (14) and national (8) studies. Three studies combined different spatial scales. We found one continental (Everard et al. 2010) and one global study for one CES class (Coscieme 2015).

The methodologies used by empirical studies to quantify, value and/or map marine and coastal CES are heterogeneous (Suppl. material 4). Regarding the type of assessment (Fig. 4d), sociocultural assessments were the most applied type to evaluate CES in marine ecosystems. Twenty empirical studies assessed CES by applying sociocultural assessment methods such as perceptions surveys, Q-methodology, and participatory mapping. Ten publications used economic assessment methods such as contingent valuation, choice experiments, and benefit transfer. Only five publications applied ecological assessment methods such as habitat frequency analysis and ecological surveys. The remaining nine publications used mixed methods, applying sociocultural, economic, and/or ecological methods. Among these publications, only one conducted an integrated assessment by applying the three assessment types (Outeiro et al. 2015a).

Most publications (55) were authored by an interdisciplinary research team (Fig. 4e), i.e., more than one discipline was represented. Most of the authors of the publications included in our synthesis are ecologists (Fig. 5). Economics and biology are, respectively, the second and third-most represented disciplines and, besides economists, only a few social scientists and humanities’ experts participated in these publications.

We evaluated whether publications assessed how marine and coastal CES affect human wellbeing and how these services depend on the ecological integrity of ecosystems. Among the 72 publications synthesised, 41 linked CES and human wellbeing in their text. Yet none measured the contributions of CES to human wellbeing, and only two publications – a review (Liquete et al. 2013) and a conceptual paper (Loomis and Paterson 2014) – presented indicators to measure human wellbeing. The links between ecological integrity and CES were referred to in 25 papers. Most of these publications (16) highlighted the positive effect of marine biodiversity in the supply of 28 CES. Some examples presented were the bequest and existence values that people derive from a diverse marine life (Beaumont et al. 2008), the larger number of recreational activities and higher aesthetic quality of more diverse and conservation-dominated coastal areas (Chung et al. 2015), and the level of marine biodiversity as a determining factor of diving locations (Ruiz-Frau et al. 2013). Eight publications mentioned the importance of habitat heterogeneity for CES provision, such as the dynamic spatial extent of sea-ice that provides a place for instruction and mentoring for Arctic communities and possibilities for tourism activities (Eicken et al. 2009), and the variety of deep-sea habitats such as hydrothermal vents and trenches that provide opportunities for education and scientific discovery (Thurber et al. 2014). One publication denoted the importance of natural/environmental flow regimes in an estuary for the provisioning of recreation and leisure and aesthetic amenities (Davis and Kidd 2012). On the other hand, one publication reported the negative effect of ctenophore and jellyfish blooms on the supply of recreation and leisure and aesthetic services (Baulcomb et al. 2015). Another publication denoted the negative effect of dune systems on the generation of cultural heritage and identity, and spiritual, sacred and/or religious services (Everard et al. 2010).

CES are often generated synergistically within bundles or sets of ecosystem services or, in some cases, other services are provided at the expense of CES, generating trade-offs (Fig. 6). Our literature synthesis revealed that synergies are common among marine and coastal CES bundles. For example, areas with conditions for recreation and leisure are often valuable due to their aesthetic and cultural heritage and identity qualities (e.g., Chan et al. (2012b), Comberti et al. (2015), Chung et al. (2015), Jobstvogt et al. (2014), Gee and Burkhard (2010)). Likewise, areas with scenic beauty often provide inspiration and opportunities for education, holding significant existence values (e.g., Fletcher et al. (2014), Outeiro et al. (2015a), de Oliveira and Berkes (2014), Hashimoto et al. (2014), Martínez-Pastur et al. (2015)). Synergies with other ecosystem service categories are also identified in the literature, namely cultural heritage and identity, and the creation of social relations that often occur together with seafood provisioning ecosystem services (Fletcher et al. 2014).

Trade-offs are more common between CES and other ecosystem service categories, and human-related activities, than within CES themselves. For instance, aquaculture and commercial fishing areas often provide seafood at the expense of aesthetic, and recreation and leisure opportunities (Chan et al. 2012b, Chung et al. 2015, Outeiro and Villasante 2013, Ruiz-Frau et al. 2013, Thurber et al. 2014, Outeiro et al. 2015b). Renewable and non-renewable energy production areas affect the supply of CES such as existence and bequest values, aesthetic amenities, symbolic, spiritual, sacred and/or religious experiences, cultural heritage and identity, and recreation and leisure (Papathanasopoulou et al. 2015, Gee and Burkhard 2010, Burkhard and Gee 2012). Similarly, human-related activities such as seabed mining, eutrophication, coastal development, and navigation, affect negatively the provision of CES by marine and coastal areas (Fletcher et al. 2014, Jobstvogt et al. 2014, Thurber et al. 2014, Rönnbäck et al. 2007, Tengberg et al. 2012).

Most CES were assessed in the coastal zone (119), coastal and marine areas (40) and only a few in the open-ocean (12) (Table 2). These results correspond to publications that did not specify the habitat under assessment. For the empirical studies that specified the habitat, we identified that CES are mainly assessed in coastal lagoons and in intertidal areas. Subtidal areas are represented mostly by assessments in tropical coral reefs and seagrass meadows. Excluding coastal lagoons, transitional waters are among the least studied areas. Reflecting the importance of non-natural capital in the co-production of ecosystem services, a few marine and coastal CES were assessed in artificial habitats, i.e., human-made structures.

Our literature synthesis revealed that economic drivers negatively affect the provision of practically all CES classes (Fig. 7). Several publications listed this type of drivers in a general and aggregate manner, but others specified the type of economic drivers affecting CES, namely economic growth, damming, tourism, land reclamation, to name a few. Also, two main clusters of drivers primarily affect a set of CES composed by recreation and leisure, aesthetic amenities, cultural heritage and identity, spiritual, sacred and/or religious experiences, and educational services. The main cluster of drivers includes depopulation, aging of local community, human perception of environmental threat (public opinion about areas potentially under environmental threat such as pollution, toxins, debris, etc.), urbanization, industrial fishing, and sociopolitical and economic drivers. A second cluster of drivers affects primarily the same set of CES plus the group 'intellectual and representative interactions'. This cluster is composed by traditional activities abandonment, local communities’ isolation, identity loss, invasive species, habitat degradation, and biodiversity loss. A common characteristic in the composition of these two clusters is a set of economic, demographic, sociocultural and ecological drivers that together negatively affect marine and coastal CES.

The global spatial distribution of case studies assessing marine and coastal CES is markedly skewed (Fig. 8). Most marine and coastal CES were assessed in the Global North, mainly on the coasts of Western Europe, USA, Canada, Australia, Japan, and South Korea. Consequently, the Global South is underrepresented, with only a few studies in South America, Madagascar, and China (the Global North/South Divide groups countries according to their socioeconomic, political, and historical characteristics, thus southern hemisphere countries such as Australia and New Zealand belong to the Global North, while some northern hemisphere countries such as China and India belong to the Global South). Remarkably, no marine and coastal CES were assessed in any country of the African continent so far, at least in publications of our systematic review. The countries where most case studies were carried out are the UK (8), USA (7), Canada (6), and Germany (4). One cross-national assessment was undertaken for the coasts of the countries surrounding the Baltic Sea. But in most of these countries, no individual CES studies have been published. Unsurprisingly, for the countries with more case studies there has been reported a wider range of assessment types to evaluate more CES classes in a more diverse set of habitats. Nonetheless, countries with only one case study such as Brazil, France, Portugal, and Singapore assessed more than seven CES classes in each study.

CES contribute to the multiple dimensions of societal and individual wellbeing with myriad benefits such as health, knowledge, inspiration, spirituality, tranquillity, and discovery, among others (Plieninger et al. 2013, Russell et al. 2013). To capture the diversity and complexity of human wellbeing and what provides it, multiple values of marine ecosystem services need to be considered (Martinez-Alier 2002, O'Neill and Spash 2000, Max-Neef et al. 1992). Sociocultural, ecological, and economic values are all crucial to understand the importance, meaning, and worth of CES benefits to human societies. Hence, they need to be considered together to support environmental management and decision-making (Gómez-Baggethun and Barton 2013, Dendoncker et al. 2013, Martín-López et al. 2012, Martín-López et al. 2014). Most assessments of marine and coastal CES in our literature synthesis have a single valuation perspective, either sociocultural, economic, or ecological (Suppl. material 4 and Fig. 4d). And sociocultural valuation is the most used valuation perspective. These findings are in line with previous reviews showing that CES assessments tend to take a sociocultural valuation perspective (Scholte et al. 2015) and mainly assess the social benefits provided by CES (Schmidt et al. 2016), although CES can also be associated with economic and ecological values. Among the 72 publications synthesised in our study, only nine combined more than one valuation type, and just one considered the three broad value types (Outeiro et al. 2015a). A future challenge for marine and coastal CES research is to engage more often in integrated valuation assessments that ensure value pluralism which, in the best case, would be elicited through participatory processes (Palomo et al. 2011, Oteros-Rozas et al. 2015). To capture the complexity inherent to integrated valuation assessments there is a need for interdisciplinary research teams. Indeed, as observed in our literature synthesis, most publications were authored by interdisciplinary research teams mostly composed by ecologists and economists. Having interdisciplinary research teams that include experts not only on ecology and economics, but also on other social sciences and humanities, is crucial to tackle complex integrated valuation assessments. Moreover, marine and coastal CES assessments can potentially be more relevant if they go beyond intrinsic and instrumental values, and focus also on relational values (Chan et al. 2016) that are behind peoples’ kin and stewardship relationships with nature (Chapin et al. 2010). Such integrated valuation assessments can be part of, e.g., ecosystem-based management and marine spatial planning approaches (Lillebø et al. 2017), and thus inform policies that address the sustainability challenges we face today (Haberl et al. 2009).

Diverse worldviews and knowledge systems are necessary to understand the worldwide range of interactions between cultural practices and the natural environment, i.e., CES (Norgaard 1994, Fish et al. 2016, Berkes et al. 2000). Yet the geographical distribution of marine and coastal CES case studies is skewed. Assessments in our literature synthesis were mostly conducted in the Global North, with only a few assessments undertaken in the Global South (Fig. 8), and even fewer incorporating indigenous and local knowledge (Oleson et al. 2015, Comberti et al. 2015, Outeiro et al. 2015a, Butler et al. 2012, Sousa et al. 2013). Thus, there is mainly a Western European and North American (Canada and USA) perspective on the non-material contributions of natural systems to human societies. Such bias narrows the understanding of social-ecological interactions to a particular cultural setting, undermining the role of other worldviews in the understanding of the full spectrum of interactions between cultural practices and ecosystems worldwide. Examples of alternative worldviews are Sumak Kawsay from indigenous traditions of South America (Gudynas 2011), Ubuntu from Sub-Saharan countries (Shumba 2011), or Ecological Swaraj from grassroot communities in India (Kothari 2014).

Our literature synthesis indicates that synergies are common among marine and coastal CES (Fig. 6). This suggests that CES often occur in bundles (Raudsepp-Hearne et al. 2010), and that often people cannot clearly separate the benefits provided by one CES class from another (Plieninger et al. 2013). Synergies with other ecosystem service categories are also identified in the literature, namely cultural heritage and identity, and the creation of social relations that often co-occur with seafood provision (Fletcher et al. 2014). Our results suggest that trade-offs are more common between CES and other ecosystem service categories, and human-related activities, than within CES. Provisioning ecosystem services related to commercial fisheries, aquaculture, and abiotic outputs such as renewable and non-renewable energy production are identified as occurring at the expense of marine and coastal CES. These trade-offs are noticeable when such activities compete for space with maritime recreational uses and when they are perceived to degrade seascapes. In fact, the use of maritime space for aquaculture and energy production implies building and assembling infrastructure that can alter the perceived visual, peaceful and remoteness qualities of seascapes, from natural to more industrial environments. Artificial elements in seascapes are likely to proliferate in the future given that the world human population is expected to increase to 11.2 billion people by 2100 (United Nations 2015), and so will the demand for seafood and energy.

Our literature synthesis also highlights that marine and coastal CES classes are not assessed by equal numbers of studies (Fig. 3). More tangible and commensurable CES classes such as recreation and leisure, and aesthetic services, receive comparatively more attention in the literature than more subjective and intangible CES such as existence, bequest, and symbolic services. One reason for this bias could be related with the high economic importance of recreation and leisure (mainly related to tourism) and aesthetic services, to the global economy (Costanza et al. 1997, Groot et al. 2012), stimulating more research and valuation assessments in these CES classes. In addition, the higher level of complexity inherent to more intangible CES classes, coupled with a fuzzy distinction between services, benefits and values associated to e.g., existence and bequest values, could also hinder more research on the topic. Moreover, it is challenging to define the ecosystem unit to which societies and individuals assign value (Gee and Burkhard 2010). This might indicate the limits of the CES concept and the notion of discrete services. Nonetheless, when spatial scales and system boundaries are well-defined, it is easier to relate CES with particular ecosystems and spatial properties (Plieninger et al. 2013). On the other hand, intangible CES such as spiritual and inspirational services are not usually associated with a specific landscape or seascape attribute (Brown 2004). It remains poorly understood what the effects of marine ecosystem change are on these intangible CES, and thus to what extent relevant stakeholders are affected by such changes. Consequently, ecosystem service assessments provide incomplete information to decision-makers, underlying a need to acknowledge this issue in CES research.

Another way to approach the difficulties in assessing CES is by explicitly linking their benefits to the multiple dimensions of human wellbeing (Plieninger et al. 2013, Klain and Chan 2012, Busch et al. 2011, Bryce et al. 2016, Daw et al. 2016). This is an underdeveloped research topic within marine and coastal CES. Connections between CES and human wellbeing are often mentioned in the literature, but they are hardly measured, and indicators are lacking. Similarly, conditions for the supply of marine and coastal CES remains largely unexplored, although a few publications from our literature synthesis highlight the positive effect of marine biodiversity and uninterrupted view out to seascapes in CES provision (Butler et al. 2012, Busch et al. 2011, Gee and Burkhard 2010, Ruiz-Frau et al. 2013, Fletcher et al. 2014).

Coastal areas and their seascapes constitute the physical limit for most of the world population to interact with marine systems. They are among the most populated areas on the planet (Brown et al. 2006), which makes them highly relevant for CES assessments. Beaches, intertidal mudflats, and dunes are well-assessed coastal areas as regards their economic importance in providing opportunities for recreation and leisure (Garcia Rodrigues 2015, Everard et al. 2010, de Oliveira and Berkes 2014). Other coastal habitats such as estuaries, fjords, reefs, seagrass meadows, mangroves, and wetlands are poorly assessed, particularly regarding those CES that arise from cognitive interactions with the environment such as spiritual, sacred and/or religious experiences, inspiration for culture, art, and design, sense of place, existence and bequest values, and symbolic services. Practical aspects such as accessibility or proximity to the study area and the expertise of authors may explain this lack of assessments (Liquete et al. 2013). Another possible reason is that experts who could assess these CES are critical to the concept of ecosystem services, refusing to measure or place value on the benefits from nature (Silvertown 2015).

Open-ocean and deep-sea habitats are inaccessible to the great majority of the world population and could be understood as irrelevant for CES research. Yet a few assessments exist for the open-ocean (e.g., Graham et al. (2014)) indicating that CES are not generated solely by human interaction with physical space and its natural processes, but also by people’s mental space with its symbolic and interpretative interactions, formed by perceptions, emotions, attitudes, values and understandings of a given natural setting (Chan et al. 2012a, Chan et al. 2012b, Fish et al. 2016, Gee and Burkhard 2010, Kumar and Kumar 2008). In this sense, even if one has never visited or interacted with open-ocean or deep-sea habitats, such areas could be valuable and meaningful to people – having existence and/or bequest value – generating CES benefits to wellbeing such as inspiration, knowledge, or spirituality through photographs, books, documentaries, paintings, exhibitions, or even scientific publications. Knowing the extent and contents of these marine ecosystem depictions in the media, cinema, literature, and music (Coscieme 2015), can provide insights about the importance that the wider public assign to those ecosystems (Jobstvogt et al. 2014b, Börger et al. 2014). This notion points out to the difference between the presence of marine and coastal CES, i.e., their availability, and their valuation by different people at different times, depending on the perceived benefits they provide. This raises important questions about the relative value assessment of CES, namely on which CES are valuable to whom and why. CES are not static entities meaning that their availability and importance to society and individuals change over time. These are relevant knowledge gaps that still need to be addressed in future CES research. Overcoming such knowledge gaps has the potential to help framing conservation policies aimed at the largest ecosystems on the planet, i.e., open-ocean and deep-sea areas.

Human-made structures such as waterfronts, harbours, and artificial reefs, were identified as CES-providing areas in our literature review (Ruiz-Frau et al. 2013, Faggi et al. 2013). However, it remains unclear what the ecosystem contribution to these services would be beyond the fact that an ecosystem or habitat has been transformed to carry these structures, while the previous natural habitat is not involved or it has a reduced capacity of delivering ecosystem services. This indicates that a common understanding of what constitutes an ecosystem service is still lacking. Nonetheless, artificial elements in association with the natural environment can facilitate interactions between people and ecosystems, enabling cognitive and physical benefits from CES. For example, one may need a boat and diving gear (non-natural capital) to access a coral reef (natural capital) for being able to enjoy (benefit) the beauty of the reef (CES). This highlights the role of social-ecological processes in the co-production of CES (Palomo et al. 2016, Reyers et al. 2013) by the joint contribution of natural and non-natural capital (e.g., social, human, manufactured).

Although ecosystems structures, processes and functions are needed to provide ecosystem services and their benefits (Haines-Young and Potschin 2009), they are also mediated by perceptions, values, institutions, and power relations among stakeholders, that can change ecosystems’ state and integrity. Substituting natural capital with forms of non-natural capital usually imply trade-offs and equity issues (Palomo et al. 2016). For example, building a harbour in a sensitive coastal area may increase recreation and leisure for tourists who can afford renting boats, while lowering the benefits that the local population receive from ecosystem services such as mitigation of extreme events, lifecycle maintenance (habitat for fish larvae) for seafood provisioning, or recreation and leisure for those who do not own boats. In this sense, the notion of co-production increases the complexity of social-ecological interactions. Such notion needs to be further understood to inform environmental decision-making to address trade-offs and equity issues associated with the substitution of natural capital with forms of non-natural capital.

Marine and coastal CES are affected by combined sets of drivers of change including economic, demographic, sociocultural, climate change, and other ecological drivers (Fig. 7). By contributing to marine and coastal CES decline, such drivers negatively affect human wellbeing causing societal impacts. Some economic drivers play a particularly important role, namely land reclamation and habitat change in coastal wetlands for the purpose of economic growth (Chung et al. 2015), coastal infrastructure development (Tengberg et al. 2012, Hynes et al. 2012), urbanization (Faggi et al. 2013, Soy-Massoni et al. 2016, Thiagarajah et al. 2015), tourism (Wang et al. 2016), industrial fishing (Ruiz-Frau et al. 2013, Moberg and Folke 1999, Jobstvogt et al. 2014, Rocha et al. 2014), aquaculture (Outeiro and Villasante 2013), and maritime shipping (Jobstvogt et al. 2014). Demographic drivers also affect marine and coastal CES supply and are related with depopulation (Hashimoto et al. 2014), local communities’ isolation (Sousa et al. 2013), and aging of local communities (Hashimoto et al. 2014). These demographic drivers erode cultural values through the loss of indigenous and local knowledge, and are linked to sociocultural drivers, namely traditional activities abandonment and identity loss. Lastly, but not least, we identified ecological drivers contributing to marine and coastal CES decline, namely biodiversity loss, habitat degradation, and the spread of invasive species (Sousa et al. 2013). To some extent, these drivers of change act synergistically upon marine ecosystems (Fig. 7) and can disrupt the supply of cultural and other ecosystem services (Rocha et al. 2014). Thus, environmental management strategies that target single drivers of change are likely to fail. Understanding the co-occurring nature of all drivers affecting ecosystems and the appropriate spatial and temporal scales for management are crucial to halt the decline of CES in marine and coastal areas (Rocha et al. 2014). Also, an important aspect to bear in mind is the intrinsic dynamic nature of both cultural practices and the marine environment, which might contribute to the production of new CES and their appreciation.