The sources of relevant articles for this literature study were tripartite: First, peer-reviewed scientific literature, published in English language, was screened for publications on ES in ORs and OCTs. We selected Scopus (Elsevier) and Web of Science (Thomson Reuters), the two largest scientific databases for environmental and social sciences, for our search. Second, relevant scientific literature (published in English) designed to identify the current status of ES mapping and assessment was obtained from Google Scholar, including to some extent also grey literature. However, grey literature that was not publicly available or not published in English was not included in this review. As a third source, the review included policy documents and empirical studies that were presented during the first MAES OR&OCT expert workshop within the scope of the ESMERALDA EU Project in Ponta Delgada, Azores, from 28.02.-03.03.2017.

The publications were reviewed in four steps (see Fig. 1). An initial screening of articles took place from April to July 2017 by searching the scientific databases for key words ("ecosystem services" AND "mapping") and their relevant synonyms (Table 1) for each of the ORs and OCTs individually. Publications were selected based on the search terms in title, abstract or key words, using Boolean operators. Search terms were kept broad to obtain the full range of studies addressing biodiversity, ecosystems and their services.

Figure 1.  

Flow diagram of the methodology and selection processes used in this systematic review, following the rules and templates of PRISMA (Preferred Reporting Items for Systematic Review (Moher 2009)).

The screening provided 1030 papers divided over the global regions of Amazonia, Macaronesia, the Caribbean, Indian Ocean, Pacific, Polar and Subpolar regions as well as the EU regions and territories in the South Atlantic (Fig. 1).

In a second step, articles were catalogued and categorised, using an adjusted version of the ES mapping and modelling blueprint (Crossman et al. 2013) and the ESMERALDA project's ES method database (Brown et al. 2018). Publications were classified according to the following characteristics:

• Year of publication
• Global and geographical region of mapping and assessment per publication
• Dimension of study (biophysical, social, economic, socio-economic)
• Type of Study (Mapping, assessment, review, indicator development, biodiversity, other)
• Spatial scale of mapping (global, multi-national, national, subnational, local, other)
• Ecosystem type considered (urban, cropland, grassland, woodland and forest, heath and shrub, sparsely vegetated, wetlands, rivers and lakes, transitional waters, coastal or marine/coral reefs)
• Information on the role of invasive species, if mentioned
• Ecosystem classification schemes (CICES, TEEB, MEA, not defined)
• Ecosystem services cascade classification per ES described
• Ecosystem service described
• Input data (GIS, remote sensing, statistical data, expert opinion, field data, literature, others)

In addition, all publications were categorised according to the ES Cascade conceptual model (Haines-Young and Potschin 2010). Afterwards, all screened papers were removed that i) were not related to EU territories, leaving a dataset of 562 publications. Further on, we removed publications that ii) focused solely on biodiversity, iii) did not mention the ES concept or iv) were descriptive in nature rather than focusing on ES mapping and assessment. This reduced the number of included papers to 161 (Suppl. material 8). As the fourth and last step, an in-depth analysis of these remaining papers was conducted, looking at the type of mapping and assessment done in the individual territories, its extent and the current shortcomings and knowledge gaps that need to be addressed in future research.

The 161 ES mapping and assessment publications were spread over all 7 regions (Fig. 3). Twenty five (25) out of the 34 ORs and OCTs were represented in the publications. Some publications both mapped and assessed ES in multiple regions, thus were counted for each OR and OCT separately. Sixty 60 publications covered the 14 EU-related islands in the Caribbean region (31%). The European territories in the Pacific were covered by 39 publications (21%). Macaronesia and the Indian Ocean were covered by a total of 29 and 32 studies each, followed by the four territories in the South Atlantic (12 publications). The eight polar- and subpolar territories were following with 10 publications, of which 8 cover ES and 2 ES benefits or valuation studies. Amazonia, with French Guiana as only country in this region, was least represented with 7 ES mapping and assessment publications, of which 4 focussed on mapping ES and 3 studies handled benefits or monetary assessments. This trend of ES studies outnumbering benefit studies could be observed for all regions (regional overviews can be found in the Suppl. materials 1, 2, 3, 4, 5, 6, 7).

Figure 3.  

Geographical location of ES mapping and assessment studies in ORs and OCTs. All publications (n=161) obtained between April and June 2017 are included and divided by global regions. The map shows the number of case studies per region (number in brackets), their study type (top pie charts), scale of publication (middle pie charts) and group of ES assessed (bottom pie charts). Assessments at national or larger spatial scales are included in the map.

A higher number of studies was expected in the ORs, because of their closer legislative and administrative bonds with their respective EU Member States. However, this expected trend was not confirmed by the data. There were no clear differences between numbers of publications on ORs and OCTs or between northern or southern hemisphere. The numbers ranged from 4 publications in Guadeloupe to 18 on the Azores, which was the overall highest number on ES mapping and assessment-related studies in one of the regions considered in this review.

The majority of publications explicitly mentioned the ecosystem type in which the mapping activities took place (Fig. 4). Within the 161 ES and benefit publications, an average of 2 ecosystems were mapped or assessed per publication. Most frequently, coastal and marine ecosystems were addressed (77, 74). Forty publications did not define the ecosystem type in which mapping and assessment took place. With 37 studies, woodlands and forests were the fourth most dominant land use types in the study regions, closely followed by cropland and agriculture (37). With only 17 publications, urban ecosystems received the least attention concerning mapping and assessment of ES. Habitat types such as deserts, tundra or ice and rock were included under type category “other”, as in Kubiszewski et al. 2016.

Figure 4.  

Overview of ecosystem types described in the 161 ES mapping and assessment publications and the spatial extent at which they performed (global, multi-national, within the EU context, island group, island level or local). Publications that did not define any spatial scale of analysis were classified as 'undefined'.

After presenting the results of the systematic review of the mapping and assessment studies, a short overview of the information included in the publications is given. This overview elaborates the ES classification, spatial coverage as well as the enormous impacts of invasive species on ecosystems in ORs and OCTs. The 161 publications included 72 ES mapping studies, 85 ES assessment papers and 28 review papers (Fig. 5). Six papers proposed new ES indicators, for example for mapping of impacts of gold mining on soils (see, for example, Schimann et al. 2012) or indicators to classify marine habitats for mapping (e.g. Mumby and Harborne 1999) or for monitoring of Caribbean ecosystems (e.g. Verweij et al. 2015). Two publications were classified in more than one category as they combined indicator development with ES mapping or assessments (e.g.Chape 2006).

Figure 5.  

Overview of the individual literature review data variables for Europes ORs and OCTs. Data variables include: (a) type of publication (192 counts); (b) spatial scale (161); (c) type of assessment (202); (d) ES Classifications used (168); (e) ES described in literature (233) and (f) input data used in publications (264). Multiple counts per study were possible.

The ES classification applied in the study was left undefined in 82 publications (48%). For the publications that defined the ES classification, authors mostly used the TEEB*9 (14%) and the Millennium Ecosystem Assessment classification system (16%). 17% of all studies do not clearly define the classification of services described. Only 1% of all reviewed papers (i.e. 2 papers) applied the CICES classification (Picanço et al. 2017a, Pereira and Navarro 2015). Looking at the ES typology, we found that almost 30% of all reviewed papers mapped and/or assessed provisioning ES. 24% referred to regulating ES and, independently of the classification system used to categorise the ES, 20% of all publications did not explicitly define any specific ES. These non-defined ES can partly be explained by the fact that monetary valuation of ecosystem types or aspects was often uncoupled from the individual ES provided in a specific ecosystem. With 17%, cultural ES was the least covered ES category in ORs and OCTs (Fig. 5).

Spatial scales used to describe the individual ES in the publications were heterogeneous. 53 studies referred to global and multi-national scales. That means that more than half of all publications that describe ecosystems and their services used maps with comparably less detail and lower resolution. 12 publications mapped or valued ES for island groups and 40 publications referred to specific ORs and OCTs. Only 3 publications compared the ORs/OCTs and their respective EU Member States (e.g. Quintas-Soriano et al. 2016, Russi et al. 2016, Murillas-Maza et al. 2011). Notably, the majority of studies covers coastal and marine ecosystems mapped on global and international scale.

In EU comparison, most ORs and OCTs still show slightly skewed publication numbers in a) geographical coverage and b) spatial scale.

a) In terms of geographical coverage, some regions have been mapped more than others. With 60 out of 161 publications, the Caribbean region is covered most extensively. However, as the Caribbean hosts 13 EU OCTs and ORs, this results in an average of 4.5 publications per region. In comparison, the Macaronesian region is only covered by 30 publications spread over 3 ORs, which results in a higher density of published studies with an average of 10 studies per OR. Furthermore, publications are not equally spread over the territories within each region. With a maximum of 18 mapping and assessment publications, this review found the Azores Archipelago leading in ES mapping and assessment efforts, a number that is average compared to continental Europe (Kopperoinen et al. 2016). On the contrary, some territories are not covered by any mapping or assessment study, such as Saint-Pierre and Miquelon, the British Antarctic Territories, Scattered Islands, Madeira, Tristan da Cunha, Ascension Island, Anguilla, Saint Martin or Aruba.

There are several explanations for the overall low number of publications mapping EU overseas. Firstly, the legal complexity of EU law in ORs and OCTs obscures the necessity to implement the EU Biodiversity Strategy overseas. Whereas ORs are obliged to implement all EU laws and legislation, the legal status of the OCTs can differ for each territory (Kochenov 2013). Though most EU Member States with bonds to ORs and OCTs explicitly mention ES in their overseas territories nature conservation plans (Ministerie van Economische Zaken 2013, UK Government 2014), the number of peer-reviewed publications is astonishingly low. This can be explained by the fact that none of the policy plans refers to the EU Biodiversity Strategy or define assessment targets or MAES strategies within their policies. In addition, the voluntary schemes fostering MAES in overseas have not yet focused on ES specifically. Their latest output, regional ecosystem profiles*4, have been focused on biodiversity, listing species, defining Key Biodiversity Areas (KBA) and exploring Marine Protected Areas (MPA), listing existing projects and their assessment efforts. ES maps, however, are generally lacking.

The lack of ES maps can be explained by three factors. Firstly by the fact that such efforts usually take place on local scales and refer more to policy than to scientific levels. Therefore, peer-reviewed publications are often lacking for ORs and OCTs. Even though this review considered grey literature, existing efforts that are of internal nature and not publicly available could not be included in this review. Secondly, the remoteness and relatively small size of most regions encroach upon the importance of ORs and OCTs, because their contribution to EU Member States is often small in terms of economic impact or benefits (ten Brink et al. 2017). Therefore, coordinated MAES efforts are still rare, with no coherent databases in which ES mapping or assessment studies have been published. Thirdly, data availability and research efforts are still scarce in ORs and OCTs. Whereas continental Europe provides detailed ecosystem maps and land use and land cover data, the EU overseas areas are still much less explored and data collection on an island scale is expensive. Our results show that only where universities have become involved in ES mapping, assessment, data collection and ground truthing, good efforts can be seen, such as on the Azores or Bonaire and Sint Eustatius in the Dutch Caribbean.

b) In terms of spatial scale, the review showed that mapping and assessment of ecosystems and their services takes place on multiple scales, including local (e.g. provisioning ES of fishing capacity and effort (Thiault et al. 2017)), regional (e.g. cultural ES of experiencing marine wildlife (Roberts et al. 2017)) to global (e.g. regulating ES of hazard risk reduction provided by coral reefs (Ferrario et al. 2014)). In contradiction to the findings of Malinga et al. 2015, who found that almost all their reviewed ES mapping studies referred to municipal, regional and national levels, this review found 97 publications (60%) on MAES in ORs and OCTs on a global or multi-national scale. Furthermore, these 97 studies covered approximately 70% of all ecosystem types and their services described in all 161 publications in this review. Such global and multinational studies represented the only sources of information on ES mapping and assessment for territories such as Saint Helena, the Sandwich Islands or the Kerguelen Islands. Nonetheless, in most of these global and multinational publications, the smaller ORs and OCTs are invisible on the maps, such as by Ghermandi and Nunes 2013 study on coastal recreation values, by Li and Fang 2014 on global ES values and their relation to GDP or Sutton et al. 2012 on the human ecological footprint.

We need to acknowledge that such global and multinational publications are important to provide a good overview and raise awareness on biodiversity and ES globally and multinationally. However, at policy level, it is important to understand where and what services are provided by an ecosystem, island, region or globally in order to monitor the achievement of policy goals (Crossman et al. 2013), such as the EU Biodiversity Strategy's target 2. This review found only 64 publications (40%) that describe ecosystems and their services on national or local level (46; 18 publications, respectively). However, maps on small to intermediate spatial scales are needed if data should feed into land use or land management plans for the 34 overseas regions. Hence, such global or multinational studies have limited value for the individual EU territories in communicating data and information locally, as their resolution is too coarse. Therefore, such studies would be ill-suited to feed into local land use planning and decision-making concerning natural resources management.

Hence, both geographical and spatial coverage show fragmented ES mapping and assessment efforts in EU ORs and OCTs. Based on this fact and the general outcomes of the review, the following knowledge gaps were identifietd.

Several knowledge gaps related to dealing with diverging conceptual frameworks, lacking knowledge on mapping and assessing island ES, the role of key biodiversity areas and marine protected areas, applied methods as well as information sharing, were identified for the ORs and OCTs:

a) The lack of common conceptual frameworks to analyse ES seems to be obvious in ORs and OCTs. Rather, a broad variety of concepts was found within the publications. The European Commission has set up a common conceptual ES framework within the EU Biodiversity Strategy (Maes et al. 2016, Maes et al. 2013). This framework builds upon the Common International Classification of Ecosystem Services (CICES*3) and tries to ensure consistent approaches (Maes et al. 2016). EU wide, CICES is applied broadly. However, in the ORs and OCTs, only two studies using CICES were found (in the Azores; Cerqueira et al. 2015, Picanço et al. 2017a). This low application rate of CICES was surprising and, considering the efforts to implement MAES in ORs and OCTs, it gives rise to the following questions: Is CICES sufficiently known outside of mainland Europe? Is it too complex or too new to be applied? Or does it just not fit the ORs and OCTs, be it due to the geographic characteristics (e.g. specific land use types and related ES; small area effects), the remoteness of the territories or their special situation, leading to a shift in focus out of EU efforts? Although the MEA classification has found much more application in ES mapping and assessment publications, still different conceptual frameworks have been applied. Hardly any mapping publication in this review used the same methodologies or used the same indicators to map ES. Only amongst the economic assessment publications, congruencies in methodologies was found. An example of consistent ES valuation throughout a whole region is shown by the extensive works assessing the economic value of ES in the Dutch Caribbean (Slootweg and Van Beukering 2008, Schep et al. 2012, van der Lely et al. 2013, van Zanten et al. 2014, Tieskens et al. 2014). These studies mainly applied the TEEB classification and used economic valuation methods such as Contingent Valuation, Willingness to Pay or Total Economic Value.

b) This literature review showed a general lack of knowledge on mapping island ecosystems and the services they provide. As the current MAES efforts mainly focus on specific ecosystems, adjusted for continental Europe, comprehensive island ES maps are lacking. This review shows that trends observed in ES research in ORs and OCTs differ from European MAES efforts. Most comparable (global) literature reviews have found a strong focus on regulating ES (e.g. Malinga et al. 2015, Egoh et al. 2012, Martínez-Harms and Balvanera 2012). Our review, looking solely at ORs and OCTs, however, highlights that, in most studies, provisioning ES were mapped or assessed (74), followed by regulating ES (58). Cultural ES were mapped the least, with only 42 publications. This can probably be explained through the strong reliance of ORs and OCTs on marine and coastal ES related to tourism or fisheries. The diverging conceptual frameworks adopted in the reviewed publications suggest a need to adapt international ES classification systems to regional and local (island) needs.

c) Mapping in ORs and OCTs has often been reduced to key biodiversity areas (KBA) or marine protected areas (MPA) (Knowles et al. 2015, Fonseca et al. 2014). However, the relationship between protected areas and the enhanced provision of ES is highly debated (Boersma and Parrish 1999, Halpern et al. 2010), arguing that the establishment of MPAs is not necessarily leading to increased ES supply. Their efficacy on

local scale is acknowledged, however, the flexible nature of resources (fish stock, larvae), the pressure of invasive species and chemical and biological pollutants affecting protected coastal ecosystems and the often inadequate size of the parks are amongst the main criticisms (Boersma and Parrish 1999). Other scholars argue that, if managed properly, MPAs strongly contribute to species diversity, ecosystem health and tourism (Leenhardt et al. 2015). Even though the number of MPAs is still growing exponentially since the last decade*10, this review found few comprehensive examples from practice, proving such a positive relation in EU ORs and OCTs. Few publications in the Caribbean and the Azores link MPAs with ES (Leenhardt et al. 2015, Waite et al. 2015, Beaumont et al. 2007), focussing on cultural ES (key species, recreational activities in MPAs (e.g. Green and Donnelly 2003, Roberts et al. 2017). Only one study investigated the complex interplay of multiple ES (Leenhardt et al. 2015) but ES supply dynamics related to MPA installations remain unstudied. Not surprisingly, there are few maps of ES provided by OR- or OCT-related MPAs (Green and Donnelly 2003). Exploring this link will be key in optimising effective management of European MPAs to safeguard the provision of multiple ES simultaneously.

d) Most of the reviewed studies were applying economic and biophysical/ecological mapping and assessment methods. ES supply and demand are, however, embedded in the complex human-environmental system, requiring transdisciplinary research approaches that integrate methods and value domains from ecological, economic and social sciences. Our review found that researchers often apply methods that they are familiar with. Moreover, data sets used for research might contain gaps. As already described by Eppink et al. 2012, data sets used for ES mapping and assessment might have been collected using inconsistent methods, indicators or leaving out processes crucial for the understanding of ES supply (Eppink et al. 2012). Therefore, there is a need to apply and interlink methods from different scientific disciplines and to communicate the reliability and uncertainties of resulting ES maps.

e) In addition, ways to store information on ES such as related data, metadata, maps and the like differ. Existing EU platforms to share case studies and publications are largely unused for ORs and OCTs, as the screening of databases like BISE*6, ESP Visualization Tool*7 or OPPLA*8 showed. Here, there is great potential for knowledge, data and information exchange for the future and in support of MAES implementation in EU ORs and OCTs.

The role of invasive species in ORs and OCTs

Biological invasions are a shared problem in all regions – about 25% of all reviewed studies mentioned invasive species and ecosystem disservices linked to alien flora and fauna. With typically limited biodiversity on islands, invasive species can alter ecosystem structures and processes and hence impact the provision of ES (e.g. Vila and Hulme 2017), often with yet unknown consequences. Invasive species such as termites have been mapped or assessed for the Caribbean region (Evans et al. 2013) and the West Indian Drywood Termite in the Azores (Borges et al. 2014, Guerreiro et al. 2014). Such invasive species flourish as trade and tourism increase and remoteness is no longer a geographical barrier (Vila and Hulme 2017). These changes in ecosystems and their services can also be observed in the marine realm, where invasive species are not restricted by any borders and have spread over all oceans, as shown in publications on mapping salmonids in the French Antarctic Territories, amphipods or lionfish in the Caribbean (Lecomte et al. 2013, Rogan-Finnemore 2008; van Beukering et al. 2014). Other examples can be found by the Government of the Falkland Islands, one of the few OCTs that systematically analysed threats and opportunities linked to invasive species (Upson et al. 2016). In Macaronesian islands, an exhaustive systematic study was performed to list the TOP100 most invasive species (Silva et al. 2008).

Whereas non-native invasive species have long been recognised as the largest cause of species loss in island ecosystems, the link between alien species and ecosystem (dis)services has just recently been established (Yam et al. 2015, Vila and Hulme 2017) and will need to be addressed urgently across EU overseas territories in the near future. However, exotic and/or invasive species may also provide ES (e.g. alien plants can increase microbial activity) and introduced natural enemies can control pests (Vilà et al. 2011). Also relevant was the result obtained by Rigal et al. 2018 on the impact of land-use changes on the functional role of exotic and native arthropods in the Azores. The main finding was that there is a potential functional complementarity between indigenous and exotic species, which means that, in agricultural landscapes, exotic species might provide and maintain key ecosystem functions (Rigal et al. 2018). However, pollinator native species may also contribute to pollination services in human-made habitats and may promote the spread of invasive plants, as seen in the Azores (Picanço et al. 2017b). Therefore, the link between invasive species and ES or disservices needs further exploration.

The high number of publications on biodiversity and ecosystem conditions in ORs and OCTs clearly indicates the importance of flora, fauna and habitats of the ORs and OCTs. As the numbers of publications on invasive species showed, there is an urgent need for nature conservation action. Moreover, there is the need to understand the role of exotic and invasive species on both ES and disservices in island native and human-made habitats. It is also important to identify the role of native insect pollinators or native birds as promoters of disservices, namely on helping plant invasions through, for instance, seed dispersal.

This review has found no studies that were officially carried out under the umbrella of MAES. Only a few studies have effectively identified or defined the goods and services derived from ecosystems and only a fraction of these publications mapped or assessed the respective ES. Past and current efforts are still skewed in terms of geographical and spatial coverage and conceptual frameworks diverge. In addition, data and the process of translating biodiversity and environmental studies into ES maps have still not clearly been communicated. Combined, these three areas of concern lead to limited comparability of studies across the EU, the overseas regions, landscapes and amongst ES maps. Experts criticise that such a broad variability, in technical and contextual aspects, would make it difficult to assess the individual and political value of such studies (Eppink et al. 2012). Such comparability, however, is needed - at least to some extent - to include the ORs and OCTs in MAES comparisons, within their related EU Member States as well as EU wide.

To enhance comparability between ES mapping and assessment studies, developing a standardised approach or one-size-fits-all solution would be unfavourable for the ORs and OCTs. Due to their immense diversity and special needs, such standardised approaches or lessons learned from continental EU might fail to obtain the data on ES mapping and assessment needed to enhance local decision-making and natural resources management. Rather, a flexible, guidance-based approach based on the vast knowledge on ES mapping and assessment from around the world needs to be developed, taking into consideration the individual ORs' or OCTs' needs. We strongly believe that developing such a flexible guidance for mapping and assessing ES that focuses on smaller spatial scales and finer resolutions can enhance local resources management for multiple, interacting ES bundles and a sustainable land management and biodiversity conservation in ORs and OCTs.

To achieve the targets of the EU Biodiversity Strategy 2020, a stronger emphasis on ES in the governance of ORs' and OCTs' natural resources and policy is needed. The MAES initiative needs much more exposure outside of continental Europe. Rather than solely placing the ES concept in the overseas policy plans of related EU Member States, the ES concept needs anchoring in national nature conservation and decision-making efforts as well as in local societies and cultures of the overseas regions, countries and territories. Voluntary programmes, such as the BEST initiative, are a step in the right direction. Capacity building amongst relevant stakeholders is required to transfer the ES concept from science into practice.