Mapping bundles of ecosystem services reveals distinct types of multifunctionality within a Swedish landscape

Ecosystem services (ES) are the benefits provided by ecosystems to humans and are often distinguished as provisioning, regulating, cultural, and supporting services (MA 2005). These services are often co-produced by humans and nature having therefore an intrinsically social–ecological character (Andersson et al. 2007; Reyers et al. 2013). As ES have direct and indirect impacts on human well-being, they are a valuable tool to be used in landscape planning and management, and when communicating the value of nature to different stakeholder groups and decision makers (Kareiva et al. 2011).

While ES research has increased exponentially over the past decade, the assessment and measurement of these services are still challenging (Tallis and Polasky 2011). So far, most research in this area has been developed for data-rich situations or required investment in substantial data collection. Furthermore, with the exception of provisioning services, most ES are difficult to quantify (Turner and Daily 2008; Norris 2012). The flow and delivery of individual services depend on both social factors, such as policies, management practices, or human demand, and biophysical processes (de Groot et al. 2002). This complexity confounds ES assessments and has led to calls for new empirically based research in this field (e.g., Carpenter et al. 2006; Daily and Matson 2008; Johnson et al. 2012). Specifically, there is a need for social–ecological integration and lower cost assessments of ES.

In addition to the social–ecological complexity inherent to the generation of individual ES, these also interact with each other in often unpredictable ways (Rodríguez et al. 2006). Some services “come together” in interdependent bundles, while others occur as trade-offs (Bennett et al. 2009; Raudsepp-Hearne et al. 2010). Trade-offs can also occur across space and time, meaning that the increase in the provision of a particular service may negatively affect ES elsewhere or in the future (Rodríguez et al. 2006; Wang and Fu 2013). Additionally, observed trade-offs between two services can be due to the effect of a common driver or a real interaction between those services. For example, the use of fertilizers and pesticides for improving crop yields is a driver that leads to trade-offs between crop production and water quality. In contrast, the trade-off between carbon sequestration and water provisioning is the result of a real interaction between services, as the increase in evapotranspiration caused by tree growth decreases water availability (Bennett et al. 2009). Thus, depending on the type of trade-offs, different measures will be required for mitigating negative outcomes.

A growing number of studies have conducted empirical assessments of multiple ES and their interactions (e.g., Lavorel et al. 2010; Raudsepp-Hearne et al. 2010; Andersson et al. 2012; Maes et al. 2012; Martín-López et al. 2012; Plieninger et al. 2013; Qiu and Turner 2013; Turner et al. 2014). However, not surprisingly, most of this research varies in scale and methodology, which makes cross-study comparisons difficult. Comparisons across studies are important in ES research, because they allow for the generalization of local or regional findings and help to disentangle the effect of context-dependent drivers from real interactions between services.

This paper aims to contribute with a case-based empirical approach that can easily be comparable to other case studies. We follow the approach developed by Raudsepp-Hearne et al. (2010) and replicated by Turner et al. (2014) to examine the existence of bundles of ES across the Norrström drainage basin in the Stockholm region. Numerous studies have focused on the social processes and practices behind the generation of ES in this region (e.g., Andersson et al. 2007; Ernstson et al. 2008; Barthel et al. 2010; Ernstson et al. 2010; Andersson et al. 2015), and have highlighted how urban spaces can provide multiple services. However, most of these studies have focused explicitly on the city of Stockholm and its neighboring municipalities. Additionally, with one exception (Andersson et al. 2015), none of them specifically aimed to assess and quantify bundles of multiple ES, as we propose to do in this paper. The article by Andersson et al. (2015) differs from our study as it explores the farm to landscape scale, meaning also that the set of services and their indicators were different from the ones in this paper.

Our paper has four main goals as follows: (a) to identify patterns of multiple ES across a diverse landscape, (b) to characterize the type and strength of interactions among both individual services and bundles of ES, (c) to investigate how the bundling of these services compares to that found in other regions, and (d) to discuss the implications of the different bundles of ES for landscape management in the Norrström region, and other regions alike.

We consider that the results originated by our study can provide valuable comparative insights that can contribute to the better understanding of local and global patterns of interactions between services and be useful for local and regional stakeholders, managers, and decision makers in the Norrström region.

We found significant differences on the patterns of ES distribution across the study area, with cultural services presenting the highest spatial aggregation.

The production of an ES is a combination of the social–ecological potential (ecological and biophysical conditions, management practices) of a landscape to produce it, and the human demand for that same service (Ernstson et al. 2010; Reyers et al. 2013). Cultural services were assessed through indicators that expressed the demand/use of a particular service. Therefore, the clustering of cultural services around Stockholm reflects a higher demand for this type of services in this region. Cultural services are often locally produced, as they result from the direct experience of nature by people. Thus, a greater expression of these services in higher population density areas, such as urban and peri-urban municipalities, is not surprising. Nevertheless, high population densities can also constrain ES, as the case of outdoor recreation and moose hunting, which were higher in lower populated forested areas elsewhere. This is due to both ecological factors, such as the lack of large connected semi-wild ecosystems in cities, as well as restrictions on hunting on urban areas.

This urban demand for cultural ES was found in previous work in Spain (Martín-López et al. 2012), and Danish peri-urban landscapes have been shown to be important for services such as recreation, sense of place, and nature appreciation (Turner et al. 2014). These findings are particularly interesting in the Swedish context, as the connection with nature and access to outdoor life is very imbedded in the Swedish culture and often expressed by people as essential to their well-being (Gidlöf-Gunnarsson and Öhrström 2007). For example, the possession of a summer cottage, is not only restricted to high-income people rather being spread at all society levels, and often meaning more than a holiday residence inserted in a resort area. Summer cottages give sense of place and open access to surrounding outdoor areas (often forest or water), where a wide range of popular activities can be experienced (forest walks and jogging, mushroom and berry picking, bird watching, swimming, boat riding, fishing, biodiversity appreciation, aesthetical beauty).

We found a variety of strong correlations among ES, and these reflected both similarities and dissimilarities with patterns found in other studies. The positive correlations found between forest products and regulating services (water quality and nutrient retention) reflect the synergies between forest areas and regulating services described in other work (MA 2005; Mäler et al. 2013). In contrast, the absence of strong negative trade-offs between provisioning agricultural services and other regulating and cultural services differed strikingly from studies using similar approaches in Montreal (Raudsepp-Hearne et al. 2010) and Denmark (Turner et al. 2014), as well as other studies (e.g., Maes et al. 2012). This might be explained by the relatively low intensity of agricultural practices in the Stockholm region, where agricultural areas hold a diverse number of habitats mixed in a mosaic-type landscape (Sporrong 2008). This differs from the Montreal region, dominated by high input intensive agriculture (Raudsepp-Hearne et al. 2010), and Denmark, which has the highest agricultural land use pressure in Europe (Turner et al. 2014).

Nevertheless, hotspots of provisioning services were cold spots of regulating services and vice versa, suggesting that trade-offs appear whenever a particular type of service is maximized. Still, the high ES diversity across the study area points for high multifunctionality at the municipal scale. Multifunctional landscapes have been correlated with higher levels of regulating services (Raudsepp-Hearne et al. 2010). However, we only found weak evidence for such a relationship. One explanation might be that most municipalities in our study had a high diversity of regulating services, which weakens the potential to detect such a relationship.

We found that bundles of services were associated to different types of social–ecological systems spatially aggregated across the basin. These patterns could be explained by social and ecological gradients in the landscape, and similar findings were found for other empirical analyses of multiple ES (Raudsepp-Hearne et al. 2010; Maes et al. 2012; Martín-López et al. 2012; Plieninger et al. 2013; Andersson et al. 2015). Distance from Stockholm strongly predicted the patterns found for most ES, while proximity to inland and coastal water was determinant for wheat production and summer cottages, respectively. This demonstrates how patterns of ES are shaped by the historical social–ecological interactions among people and nature.

Multifunctionality was the highest on the mosaic cropland livestock and the mosaic cropland horse bundle groups, but municipalities of the urban group could also provide multiple services. Swedish urban areas are relatively green, and there has been substantial historic public investment in urban green infrastructure. Therefore, we expect Stockholm to have relatively high levels of ES compared to other urban regions in other countries. Still, high multifunctionality and provision of cultural services in urban and peri-urban areas have also been found in Denmark (Turner et al. 2014). While both associated with forestry-dominated areas, and thus similar in land cover, the remote forest group and the forest and towns groups substantially differed on their capacity for nutrient retention. This might be explained by different factors in inland and coastal municipalities in remote forests that negatively affect nutrient retention. Due to their proximity to the Baltic, coastal municipalities have a short traveling distance of nutrients to the sea, which is the cause of low retention levels. Additionally, remote forest’s inland municipalities are crossed, from west to east, by the Dalälven River that quickly transports nutrients from inland to the Baltic Sea, not allowing for high nutrient retention (Fölster et al. 2012). In this case, biophysical and topographical characteristics of the landscape are key elements for understanding the interactions between services. This is a typical example where the influence of other factors can easily be interpreted as a real trade-off between services. Without knowing the geography of this region, the remote forest bundle group could easily be interpreted as a trade-off between water quality, production of forest goods, and nutrient retention when in reality the low nutrient retention level in these municipalities is not the result of a real negative interaction between these services.

As for individual services, the five bundles had both similarities and dissimilarities with the Montreal region (Raudsepp-Hearne et al. 2010) and Denmark (Turner et al. 2014). These contained municipalities dominated by intensive agriculture, destination tourism, and county homes, none of which were present in the in the Stockholm region. Nevertheless, similar bundle groups were identified in Montreal, despite the Swedish bundle group urban had higher levels of cultural services. Overall, our results suggest that the Montreal and Danish regions are more intensively used and less diverse within municipalities than the Stockholm region. Still, there are enough similarities among bundle types to suggest that the extension of the approach used in this paper to other places of the world could be useful to identify general patterns in which ES are co-produced.

We aimed to identify patterns of distribution of multiple ES and characterize the strength and type of interactions among services by using publicly available data at a scale that was relevant for governance. Therefore, we did not conduct our own field assessment and rather focused on already existing data, relevant for our study. While this methodology has clear advantages, allowing a cheap and rapid ES assessment in a given area, it also presents limitations. Firstly, the use of single indicators for individual services does not capture the full complexity of the interaction between the ecological potential, management practices and demand/use that makes the co-production of ES. Thus, a methodology that would combine these types of different indicators would be desirable, being an open field for future research. Secondly, some data used in this study are fairly crude and it might miss to capture more fine-grain aspects particularly important for certain services. This, for example, is the case of pollination services that in our study were based on relatively crude land-cover data (see Electronic Supplementary Material) as well as a broad classification of suitable habitat. These downsides of the approach should be taken into account when drawing conclusions from our results. That said, and as this approach is increasingly being applied in multiple regions and across different scales, we consider that this methodology has a substantial potential for providing comparable insights on the distribution and interaction of multiple ES, especially when the capacity for conducting more exhaustive assessments is limited.

Our study demonstrated that a high number and abundance of ES could be found in human-dominated landscapes, which should be considered by management options. In particular, we showed that most cultural services were found in, or near, densely populated urban municipalities. Thus, enhancing the green infrastructure that provides these services in urban areas is likely to provide substantial further benefits, relative to increasing such infrastructure far from the city. Despite the high competition for housing space in a fast growing city as Stockholm, ensuring that housing can co-exist with natural areas would enable the production of ES utilized by large numbers of people.

Identifying bundles of ES can bring numerous benefits for managers and policy makers, when managing complex landscapes (Bennett et al. 2009; Deal et al. 2012). Distinct bundles are likely produced by different sets of social–ecological interactions, and implementing policies targeted for bundles of services, instead of individual services, takes advantage of those interactions. This approach has three clear benefits: First, by highlighting interactions among services, it discourages interventions to enhance single ES, such as carbon sequestration, without thinking about its impact on other services. Second, it helps identifying interventions that can have simultaneously desired effects on multiple ES. Third, by identifying types of ES bundles, decision makers can focus on potential shifts between these groups, and identify where, when, and how such shifts are possible or desirable. In this study, municipalities less typical from each bundle group had likely higher potential for the implementation of management strategies that can lead to a change in the balance of ES.

Our study shows that in the Norrström region in Sweden, there is high potential for multifunctionality of ES at the municipality scale, in particular, in the municipalities around lakes Mälaren and Hjälmaren. Although these are very positive and encouraging findings for regional and local managers and politicians, it is important to notice that trade-offs among services might happen at other spatial scales. Furthermore, multifunctionality might not always be desirable at this scale for all cases (Pasari et al. 2013). For example, the protection of some key iconic species needing large areas of native habitat might imply the existence of contiguous areas of forest, at the cost of some provisioning and cultural services. Such decisions need to be taken based on a broader scale analysis, even knowing they will compromise heterogeneity and multifunctionality at smaller scales. A lot of this knowledge required for management decisions at multiple scales is still lacking and we encourage future research that would replicate the methodology used in this study at other spatial and administrative scales.