Future changes in the supply of goods and services from natural ecosystems : prospects for the European north

Humans depend on services provided by ecosystems, and how services are affected by climate change is increasingly studied. Few studies, however, address changes likely to affect services from seminatural ecosystems. We analyzed ecosystem goods and services in natural and seminatural systems, specifically how they are expected to change as a result of projected climate change during the 21st century. We selected terrestrial and freshwater systems in northernmost Europe, where climate is anticipated to change more than the global average, and identified likely changes in ecosystem services and their societal consequences. We did this by assembling experts from ecology, social science, and cultural geography in workshops, and we also performed a literature review. Results show that most ecosystem services are affected by multiple factors, often acting in opposite directions. Out of 14 services considered, 8 are expected to increase or remain relatively unchanged in supply, and 6 are expected to decrease. Although we do not predict collapse or disappearance of any of the investigated services, the effects of climate change in conjunction with potential economical and societal changes may exceed the adaptive capacity of societies. This may result in societal reorganization and changes in ways that ecosystems are used. Significant uncertainties and knowledge gaps in the forecast make specific conclusions about societal responses to safeguard human well-being questionable. Adapting to changes in ecosystem services will therefore require consideration of uncertainties and complexities in both social and ecological responses. The scenarios presented here provide a framework for future studies exploring such issues.


INTRODUCTION
Northern areas of the globe and in particular the Arctic have warmed about twice as fast as the global average over the last century (ACIA 2005, IPCC 2013, and models indicate continued warming. The changing climate affects ecosystems as well as human livelihood and well-being (ACIA 2005, Hovelsrud andSmit 2010). Ecosystem services, defined as the benefits humans obtain from ecosystems (MA 2005a) and use for their well-being (Fisher et al. 2009), encapsulate human dependence on ecosystems and are likely to change along with climate (Mooney et al. 2009).
The Millennium Ecosystem Assessment (MA 2003) provided a framework for assessing changes in ecosystem services, and a few studies have since evaluated climate-change effects on ecosystem services at regional scales (MA 2005b, Schröter et al. 2005. Most such studies focus on human-dominated ecosystems, such as agricultural and urban areas, analyzing effects on services such as crop production and pollination. Less is known about how services provided by natural or seminatural ecosystems, where most of Earth's biodiversity resides, will be affected by climate change. Most such efforts have focused on carbon balance and sequestration. Given that both communities and large resource production networks such as forestry rely on resources from seminatural ecosystems, such knowledge is important. Forecasts of ecosystem services are hampered by uncertainty about drivers, unknown responses to climate and ecosystem change, stochasticity and nonlinearities in interactions among species, ecosystem components, and ecosystems, in addition to uncertainty related to climate change itself, and those inherent to modeling and scale of projections. Successful forecasts must identify ecosystem attributes for which uncertainty can be reduced sufficiently to provide information useful in decision making (Clark et al. 2001). Methods to synthesize information and provide predictions relevant to society at regional scales are still largely lacking. Previous attempts have either focused on a diversity of drivers (MA 2005b, Arctic Council 2013, single ecosystem types (Hermy et al. 2008), small areas (Peterson et al. 2003, Ford andSmit 2004), or large-scale quantitative projections necessitating simplifying assumptions (Schröter et al. 2005).
We discuss climate-driven changes in ecosystem services in the Barents Region, which comprises the arctic and subarctic regions of mainland Europe, encompassing the northern parts of Norway, Sweden, and Finland, and northwestern Russia (Fig. 1). The region is already in a period of rapid climate change, and also future warming is expected to be pronounced Källén 2005, Overland et al. 2011). With over 6 million people, the Barents Region is the most densely populated Arctic Region with clusters of heavy industry and developed infrastructure. Natural-the greatest warming in the east (Rummukainen et al. 2004, Benestad 2008, Overland et al. 2011. Winters will warm more (Dec-Feb; 4-7°C) than summers (Jun-Aug; 1-4°C). The temperature increase is expected to result in longer growing seasons (30-60 more days with a mean temperature > 5°C). The number of days with snow cover will decrease by 20-90 days, with the largest decreases in the southwestern lowland areas and the smallest in the east. The number of freeze-thaw events (zero crossings) during winter is projected to increase in areas with continental climate, going from 0-6 to 15 in 2071-2100. In northern Sweden, annual runoff will increase by 20-30% between 1961-1990and 2071-2100(Andréasson et al. 2004. Seasonal variation in runoff and streamflow will change as less precipitation falls as snow, leading to higher winter flows and lower spring flood peaks, except for the northernmost Russian rivers where most winter precipitation will still come as snow (Woo et al. 2008).

Ecosystem services
To forecast changes in ecosystem service provision for the Barents Region, we adopted a strategy whereby experts selected services from natural ecosystems considered to be typical and characteristic to the region, as well as the most important for local communities in the Barents Region. Given this distinction, a focus was placed on renewable resource-utilizing practices. The expert assessments were made by a team of expert ecologists, social scientists, and cultural geographers, who first convened in a workshop and selected ecosystem services. We analyzed changes in provisioning and cultural ecosystem services and their consequences in interdependent ecological and social systems (Fig. 2;MA 2005a). We selected the provisioning services wood products, animal products, berries and mushrooms, along with the cultural services tourism, recreation, and cultural ties to the land. We considered biodiversity, defined as the diversity of species, communities, and ecosystems, as supporting other services (Fig. 2). With higher species richness, species providing specific ecological functions are more likely to be present, and rare species have been found to often support unique functions (Mouillot et al. 2013). In addition, both species additions and losses have societal consequences.
For each ecosystem service identified, cause-effect relationships were constructed, going from climate drivers to societal consequences and possible adaptation strategies, based on expert workshops and literature surveys. At the first workshop, we had experts on climate change effects on ecosystem services, societal consequences, as well as societal adaptation strategies. At the meeting, cause-effect chains were drafted and discussed until consensus was reached. This was followed by literature review, checking the available evidence for the suggested model, which was then modified if needed, so that the final model was supported by empirical evidence and minimizing subjectivity stemming from the opinions of workshop participants. Literature searches were made in ISI-Thompson Web of Science combining keywords describing services with keywords for climate change. Following the workshop and review, a manuscript was drafted, which was circulated for revision, and further revised after two additional workshops, where focus was put on developing projections of societal consequences and adaptation strategies. Also, additional experts were consulted and included in the manuscript writing process. The scenarios resulting from this process are narratives Conceptual model used in the analyses, consisting of an ecological system of natural or seminatural ecosystems providing ecosystem goods and services to human actors, made possible by the biodiversity at the level of species and ecosystems, and a social system, represented by determinants of human well-being and societal adaptive capacity, affecting the vulnerability or resilience of human actors. The determinants of adaptive capacity are relevant at a hierarchy of scales nested within each other, going from global to local communities and individual households. Human actors manage ecosystems by harvest, control measures and conservation strategies (solid arrows), having side-effects on adaptive capacity (dashed arrows), and develop or affect societal adaptive capacity by various activities and decisions, having side-effects on ecosystems. Adapted from Millennium Ecosystem Assessment (2003), Chapin et al. (2006), and Eakin and Lemos (2006). and hypothesized cause-and-effect relationships (Greeuw et al. 2000). An example flow chart displaying the cause-and-effect relationship from one ecosystem service, wood production, is displayed in Figure 3. The societal consequences of climate change include both effects on human well-being and effects on the society's adaptive capacity. Adaptive capacity can be defined as actors' or communities' management of current and past stresses, their ability to anticipate and plan for future change, and resilience to perturbations (Plummer and Armitage 2010). Although a focus is placed on how changes impact the natural environment, we do not a priori distinguish between the two social subsystems, human well-being and adaptive capacity, but rather emphasize the intersection between the ecological and social systems in the selected literature.

RESULTS
Of the 14 ecosystem services analyzed, we project decreases until 2100 in the provision of six services, and increases or no change in the remaining eight services (Table 1). We do not predict collapse or a dramatic increase in any service, although drawing conclusions about the magnitude of change was generally difficult. For most services, the direct or immediate responses are expected to be different from the indirect or delayed ones, making linear temporal trajectories of change unlikely. Ignorance of the strength of species interactions and the unpredictability of extreme events are the greatest cause for uncertainty in the predictions. Substantial societal consequences are expected as a result of changes in ecosystem services, calling for strategies for adaptation (Table 2). To put emphasis on the projected chain of events, we have in essence "linearized" processes that in reality consist of feedback loops. These feedbacks are represented by "indirect or delayed effects" in Table 1 and "secondary ecosystem and societal responses" in

Changes in provisioning services and projected impacts
Wood production Warmer and longer growing seasons will stimulate tree growth, facilitate regeneration, and increase the forest area (Fig. 3, Table  1; Wolf et al. 2008a). These responses are expected to increase productivity in managed forests (Kellomäki et al. 2008) and expand forestry to new areas (Koca et al. 2006, Devi et al. 2008. Opposing effects could counteract these beneficial effects of warming (Fig. 3). For example, warmer weather may increase forest pests such as pathogens, diseases, invasive herbivorous insects, and plants (Niemelä et al. 2001, Wolf et al. 2008b, Valtonen et al. 2011. In northern Fennoscandia, northward expanding geometrid moth outbreaks have caused massive devastation of subarctic birch forests (Jepsen et al. 2011), most likely owing to warmer winters and springs (Jepsen et al. 2008(Jepsen et al. , 2011. Climate change is also expected to lead to more severe winter storms and icing events (Chapin et al. 2007). Freeze-thaw events, which kill growing plant tissues (meristems) and cause irregular tree growth, are also expected to increase in frequency (Lindner et al. 2010). In areas with more continental climate, such as Russia, higher temperatures could cause summer drought and more fires. Lindner et al. (2010) also suggest that climate change will reduce access to forest land outside the frost period because of wet soil. Finally, tree species composition may change as deciduous tree species expand northward (Lindner et al. 2010), although tree species composition in managed forests are mostly determined by planting and seeding.  Impact on forestry A higher wood production will increase profits from wood and paper products (Solberg et al. 2003, Trømborg et al. 2008. It may also result in intensified forest management with shorter rotations, more fertilizers, use of more southern provenances, and new tree species (Table 2), leading to increased forest fragmentation (Kellomäki et al. 2008) and local species losses. Shorter periods of ground frost may require new techniques to avoid or repair damage from heavy machinery. The risks for forestry associated with climate change may be reduced by promoting pluralism in forest management, so that each forestry action, e.g., fertilization, planting of new species, is implemented where best suited. However, these adaptive strategies might conflict with biodiversity conservation goals and require adjustments in legislation and training (Table 2). More intense forestry may be balanced by enhancing actions to conserve http://www.ecologyandsociety.org/vol20/iss3/art32/ species, such as managing for increased abundance of dead wood and leaving buffer strips along ecological boundaries ( Table 2). The regional effects of increased forestry profits are uncertain for at least two major reasons: (1) It is uncertain how regulations and institutions will influence the regional retention of profits. Transnational corporations may reap most of them, resulting in minimal stimulation of the regional or local economy.
(2) Many effects of globalization, including changes in demand for wood products, expanded plantation forestry in developing nations, and import/export regulations, all influence the economic benefits of increased wood harvest. Improved access to the Northern Sea Route in response to melting sea ice may lead to new seaports (Ragner 2008), providing access for timber harvest in new areas of remaining old-growth forest in Russia (Yaroshenko et al. 2001).

Meat and fish production
Moose (Alces alces), ptarmigan (Lagopus spp.), and salmonid fishes play significant cultural and economic roles in rural communities in the Barents Region. Climate factors regulate growth-rate variation in moose, probably through impacts on predator success and foraging (Vucetich and Peterson 2004), with warming being expected to lead to higher survival and reproduction of moose (Table 1). Milder winters with less snow should speed up population growth rates, although density regulation may buffer population increases (Post and Stenseth 1998). Long-term expansion of Salix shrub and forest communities to the north and toward higher elevations is likely to allow geographical expansion of moose similar to that during earlier warming periods of the Quarternary (Schmölcke and Zachos 2005). Species distribution modeling predicts future range expansion of moose (Hof et al. 2012), and observations indicate recent population increases (Hörnberg 2001).
Willow Ptarmigan (Lagopus lagopus) and to some extent Rock Ptarmigan (L. mutus) are economically valuable game species in alpine and subalpine parts of the Barents Region. Ptarmigans may respond positively to warmer starts of growing seasons (Watson et al. 1998), but predator-prey dynamics make outcomes uncertain, and they may ultimately lose habitat as tundra is replaced by forest. Their populations vary cyclically, often synchronously with other species over large areas (Watson et al. 2000), indicating climate control that may interact with predatorprey dynamics. Population cycles of Willow Ptarmigan follow those of voles and lemmings through predation (Moss and Watson 2001), but it is unclear how the fading of vole and lemming fluctuations over large areas may affect ptarmigan dynamics. Cycles of Willow Ptarmigan and rodents are no longer synchronous (Henden et al. 2011), and recent population dynamics appear more irregular with less frequent good years and long troughs with low numbers (Holmstad et al. 2005). Willow Ptarmigan density is indirectly controlled by reindeer through their grazing on Salix shrubs , den Herder et al. 2008, Henden et al. 2011 (Reist et al. 2006). Thus, even within the same species, responses to climate change may vary. Many populations are expected to experience more frequent warm spells that reduce survival of eggs and juvenile fish (Reist et al. 2006) and restrict habitats for cold-water fish such as Atlantic salmon, Arctic charr, and whitefish. Framstad et al. (2006) add that ice cover on rivers protects young salmon during winter, and a reduction of ice cover may thus be expected to be negative. Temperature variation during winter may also increase the formation of frazil ice and anchor ice that can kill juveniles during winter (Weber et al. 2013).
Increased interspecific competition and changing parasite loads may also play a role. For example, Arctic charr have disappeared from subarctic lakes following invasion by pike (Esox lucius), presumably as a result of recent warming (Byström et al. 2007). Arctic charr and whitefish will probably disappear from the south of the Barents Region (Wrona et al. 2005), whereas range expansions to the north are limited by the lack of landmasses (Reist et al. 2006). Arctic charr is predicted to lose 73% of its range in Sweden by 2100 because of increased temperatures and projected pike invasions (Hein et al. 2012). In the Russian north, local residents have observed expansion in the distribution of perch (Perca fluviatilis), replacing salmonid fishes (Vlassova 2006). Climate change is also expected to change fish migration patterns. Higher productivity in freshwater ecosystems following increased temperature may reduce charr migration from freshwater to marine habitats after spawning (Wrona et al. 2005). This will probably reduce yields because charr grow faster in the productive marine environments. Thus, responses to climate change should vary among populations in all salmonids in the Barents Region, making it difficult to make regional generalizations.

Production of edible berries and mushrooms
Berries and mushrooms are harvested commercially across the entire region, but are also important for subsistence in rural areas such as northwest Russia (Keskitalo and Kulyasova 2009). Berry production may be stimulated by warmer climate, but habitat loss and species interactions may result in decline, making predictions uncertain ( Table 1). Production of berries varies among years, and depends on the weather in the preceding years (Selås 2000, Krebs et al. 2009). Also, some of the berry-producing dwarf shrubs are particularly sensitive to warm events in winter (Bokhorst et al. 2009) but recovery from damage is common (Aerts 2010). The complex response of berry production to climate reflects the multiple steps of climate influence, such as flowering response to spring frosts and temperature, pollination to temperature and precipitation, and berry formation and growth in response to moisture and heat. Changing snow cover may also promote pathogen attacks, reducing growth of berry-producing plants (Olofsson et al. 2011). Increased extents of insect outbreaks in oligotrophic birch forest zones induce understory vegetation state shifts from dominance of berry-producing shrubs to grasses (Karlsen et al. 2013). In contrast, dampened lemming cycles resulting from warmer winters are likely to favor increased http://www.ecologyandsociety.org/vol20/iss3/art32/ biomass of berry-producing shrubs (Olofsson et al. 2012). Thus, any projections of future berry production are speculative.
Predictions for mushroom production are also uncertain (Table  1). Mushroom production is highly variable but tends to increase in wet years. Cues for mushroom fruiting are poorly known, but a warmer climate might lead to earlier resource acquisition necessary for fruiting, whereas cues for fruiting might occur over a longer period. Gange et al. (2007) reported extended fruiting periods for a set of species in the UK, starting earlier and ending later. In contrast, Kauserud et al. (2008) documented fruiting dates of mushrooms being delayed by on average 13 days since 1980 in Norway, with longer delays in earlier-fruiting (> 30 days) than later-fruiting species, making the net response to climate change uncertain.

Effects on hunting/fishing/gathering
The predicted climate changes should affect the provision of meat, berries, and mushrooms by modifying harvest rates and species composition (Table 1). These changes may be accommodated by using existing harvest methods but may also require new methods (Table 2). With changing harvest methods, local communities, and often the entire society, run the risk of losing local and traditional knowledge, which is already threatened by depopulation (Heleniak 1999) and globalization. On the other hand, large environmental and climatic changes may make certain types of knowledge less relevant. However, the introduction of new harvest techniques may also provide an opportunity to exploit new knowledge and foster adaptive and sustainable resource use. The options to mitigate effects of changing climate on production of salmonids may be taken as an example. One potential action used previously to counter population reductions caused by fragmentation by dams is introduction of fish, for example salmon from hatcheries. Spread of parasites (Heggberget et al. 1993) and mixing of escapees with wild populations, leading to genetic deterioration of wild populations (Hindar et al. 2006) make this alternative unsustainable. In rivers where recreational fishing is important, bag and size limits could be further constrained, and catch and release fishing, although not ideal (Wilkie et al. 1996), could be made mandatory. However, another option is to restore degraded salmonid habitats by removing artificial migration barriers, restore spawning beds, or improve winter low-flow refuges to reduce mortality (Battin et al. 2007). Changes in the timing of reproduction associated with climate change may be met by altering harvest regulations, including timing. Changes in the geographic distribution of productivity of for example ptarmigan and salmonids call for development of sustainable harvest models, to avoid over-harvest in areas and time periods of low productivity (Table 2). Such strategies may encounter obstacles in the form of inadequate legislation and cause land-use conflicts.

Reindeer populations
Reindeer provide subsistence for husbandry in most of the Barents Region. In many areas, grazing by primarily sheep may be important, potentially competing with reindeer. Reindeer population sizes are largely regulated by winter food availability , Moen 2008, whereas meat production is largely related to reindeer density on summer pastures ). Effects of climate change on reindeer populations are uncertain, but may be mostly negative in the long run (Table 1).

Shorter winters reduce the risk of starvation and mortality.
Winter forage is however expected to become more limited because of ice-crust formation following frequent zero transitions (Hansen et al. 2013, Stien et al. 2012, and because of lichens, the preferred winter food of reindeers, being outcompeted (Moen 2008, Rees et al. 2008. The net outcome is uncertain and may vary regionally, for example between oceanic and continental areas ). In addition, harassment by insects is expected to increase, reducing fecundity and growth rate of calves (Moen 2008). At the same time, higher plant productivity and longer growing seasons will make more food available to reindeer during summer, although management of reindeer population density may also affect plant productivity  and forage composition (Ravolainen et al. 2011). In the long run, the observed (Callaghan et al. 2013) and expected (Wolf et al. 2008a) expansion of trees and shrubs into alpine areas may reduce grazing areas (Rees et al. 2008), although reindeer may themselves curb this process , den Herder et al. 2008, Ravolainen et al. 2014). In the most extreme scenarios of climate change, most of the Barents Region will be warmer than the present climate envelope for reindeer.

Impact on reindeer husbandry
Scenarios for the future of reindeer husbandry are mixed (Rees et al. 2008). Reindeer husbandry already operates under marginal economic conditions. In addition, its flexibility has been compromised by habitat loss and changes in governance (Tyler et al. 2007. Options for reorganization include geographic range contractions, stationary herding, or even farming. Range contraction may result in traditional grazing rights being questioned if herds are no longer present (Table 2).
Longer growing seasons would advance migration to and increase the residence time in summer grazing areas. Moving herds by foot may become more difficult because of unreliable ice and snow conditions, which are of special concern in Russia, where reindeer herds are closely monitored by herders (Rees et al. 2008). Such problems may foster modified herding techniques with less transportation and increased use of supplemental forage during winter. Adaptive capacity of reindeer husbandry for a crosscompliance of conservation and development is however dependent on well-functioning governance . Accordingly, Brännlund and Axelsson (2011) found that nonclimatic factors, such as border closures, forestry, and reindeer herding laws pose a greater threat to future reindeer herding than climate, at least in the short term. Thus, management changes may compensate for a climate-related decline in reindeer husbandry (Table 2). From an ecological point of view, an important strategy to enhance the adaptive capacities of reindeer husbandry is to increase or maintain a choice of grazing sites (Moen 2008). This capacity is already limited as a result of forestry, leading to up to 50% losses of winter grazing in the Nordic countries (Sandström et al. 2006, Horstkotte et al. 2011. Reservoirs, roads, and petroleum industries also constrain grazing areas (Rees et al. 2008), and increasing infrastructural expansion and ecosystem degradation may further impact reindeer husbandry (Forbes et al. 2011).

Changes in cultural services and projected impacts
Some cultural services are closely connected to the abovementioned products that people obtain from ecosystems, whereas Ecology and Society 20(3): 32 http://www.ecologyandsociety.org/vol20/iss3/art32/ others relate to the experience or aesthetics of nature. We consider tourism, recreation, and cultural ties to the land.

Changing conditions for, and responses from, tourism and recreation
Because choice of destination and activity are highly seasonal, we distinguish between summer and winter recreation. Climate conditions for summer tourism are projected to improve in the Barents Region (Perch-Nielsen et al. 2010), whereas winter tourism may decline (Table 1). Simulations developed by Hamilton et al. (2005) and Lindner et al. (2010) predict that a warmer climate would both increase the number of tourists travelling to the Barents Region countries in the summer and make people in the region choose destinations closer to home. Summer outdoor recreation will benefit from an increased number of warm days balanced by an increase in the number of days with rainfall, although occasional rain is tolerated by visitors to the area (Førland et al. 2013). The effect of climate change on winter recreation will depend on trends in snow-and ice-related recreation. Shorter duration of snow cover, thinner snowpack, more wet and icy snow, more thaw periods, and more frequent rain and storm events will make conditions less predictable and favorable for skiing and other activities requiring snow or ice. The fate of commercial outdoor recreation activities depends on tourist behavior, the majority being visitors from outside the region. The Barents Region will be the only European region with a long snow season except for mountainous areas. Moen and Fredman (2007) suggest that shorter winters and lower exposure to winter-like conditions in most of Europe would reduce interest in snow-and ice-related recreation, reducing numbers of winter visitors. Alternatively, snow and ice could be exploited as exotic attractions. Changes in opportunities for tourism and recreation can be met by diversifying the type of activities offered and by targeting a different, foreign public (Table 2).
Ties to the land Ecosystems contribute to human well-being by providing culturally important species and scenic landscapes (MA 2005a), and ecosystem changes may result in loss of cultural ties to the land (Table 1). Ties to the land are partly manifested through harvesting of products provided by ecosystems, allowing local, including indigenous, peoples to continue traditional land-use patterns. Such ties to the land may be affected by changes in geographic distribution of species in response to climate change, as well as by ongoing urbanization and globalization. Because of climate change, important species may be displaced geographically or disappear, and landscapes as they are featured in cultural traditions may no longer exist. Examples of such landscape changes are forestation of alpine and tundra areas and loss or significant reduction in duration of winters with continuous cover of snow and ice. The aesthetic and cultural values of the Barents Region, with some of the last relatively intact ecosystems in Europe, are important to local residents and the broader European community both as a motivation to spend time on the land and for their contribution to the existence value of the land. However, these values are likely to differ among stakeholder groups. For many local residents, any ecological change, such as changes in species composition and diversity, or landscape structure, such as location of treeline, may reduce their ties to the landscape (e.g., Forbes et al. 2011), thus potentially increasing its vulnerability to other social and economic changes.
When environmental cues traditionally used by residents no longer work, the predictive ability of residents is reduced, making traditional knowledge less relevant and ultimately reducing selfconfidence in local use. These challenges can be met to some extent by research, education, and training (although we have low predictive power for some of the changes), and by increasing the awareness of how previous challenges have been met by innovative as well as traditional adaptation measures (Table 2).

Range shifts, exotic and range-restricted species
Longer, warmer summers and shorter, milder winters will favor invasion of southern species to the Barents Region provided they can reach the region (Henningsson and Alerstam 2005, Huntley et al. 2008, Hof et al. 2012, whereas cold-adapted species may experience contracting geographic ranges because there are no appropriate areas to move to (Hof et al. 2012; Table 1). Given that the distribution of many species is limited by temperature (Normand et al. 2009), and that species richness generally decreases with latitude, more species may be gained than lost (Table 1), assuming that species can track their preferred climate and expand their distributions (Saetersdal et al. 1998, Bakkenes et al. 2002, Hof et al. 2012. Northern Europe is dominated by natural and seminatural vegetation, implying that anthropogenic barriers are less constraining than in many regions further south (Saetersdal et al. 1998). However, many species in Europe have not filled their potential climatic ranges Skov 2004, Araújo and, suggesting that they were unable to respond to past climatic shifts and may not fully exploit this expanded climate envelope. Also, long-term observations show contrasting responses even in the same plant species during recent past warming, because of variation in barriers to dispersal and species interactions (Pellissier et al. 2010, le Roux et al. 2014) and because climate warming affects many aspects of plant growth, reproduction, and survival (Callaghan et al. 2013). Thus, many species cannot be expected to track future climate change (Svenning and Sandel 2013).
Although past shifts in climate have been as rapid as the ones expected for the 21st century (Overpeck et al. 2005), dispersal rates vary among species, with good dispersers, such as winged insects, responding rapidly (Coope and Wilkins 1994), whereas trees require centuries or longer to attain equilibrium with climate (Bennett et al. 1986, Svenning andSandel 2013). Easily dispersed and common species are most likely to move in synchrony with climate (Alsos et al. 2012), because production of many propagules with good dispersal ability allows rapid colonization of areas that become climatically suitable. The Barents Region harbors rather few exotic species (Dynesius et al. 2004), the majority of which have invaded from warmer areas. A warmer climate will pave the way for an expansion of these species (Sykes 2001) and an invasion of new ones that are already adapted to warmer conditions (Hof et al. 2012). Thus, although species numbers are expected to increase, the new species are likely to be widespread ones with good dispersal ability or species introduced by humans. Local species richness may increase throughout the region but may be highly variable depending on propagule arrival rates and the resistance to invasion by local ecosystems.
The species most likely to go extinct or decline substantially are specialists such as those dependent on snow and ice (Callaghan Ecology and Society 20(3): 32 http://www.ecologyandsociety.org/vol20/iss3/art32/ et al. 2005) and those with small population sizes in shrinking northern and high altitude habitats, such as high-Arctic plant species (Thuiller et al. 2005) and alpine snow bed habitats (Björk and Molau 2007). Most of those species are circumpolar and are only locally threatened, although many of them have subspecies or local varieties endemic to the Barents Region. Some threatened species may be of central importance for the ecosystem, and if these disappear others will likely follow. Such a chain of events has been implicated in the fading of distinct population cycles of voles and lemmings from many areas (Ims and Fuglei 2005). This is potentially the result of a shorter snow period, thinner snow cover, and more freeze-thaw events leading to ice crusts and reduced food availability (Callaghan et al. 2005, Hansen et al. 2013. As a result, predator populations dependent on voles and lemmings have decreased (Hörnfeldt et al. 2005, Hellstedt et al. 2006, Schmidt et al. 2012.

Community and ecosystem changes
The Barents Region has condensed climatic gradients with shorter distances between temperate, boreal, and tundra biomes than anywhere in the Arctic, suggesting that local ecosystems may be displaced in a large portion of the region (Callaghan et al. 2004), with some cold-adapted ecosystems disappearing, and new ones appearing (Table 1). Species generally respond individually to climate change, implying that present communities will disappear and new ones will form. Some ecosystem types associated with cold climate are likely to disappear. For example, lowland permafrost areas, such as palsa mires, are predicted to melt, and their communities will be lost although individual species may persist (Luoto et al. 2004). Indeed, observations have recorded loss of permafrost mires in northern Sweden in the past 30 years (Åkerman and Johansson 2008) The greatest expected change in terrestrial habitat is replacement of alpine and tundra areas by forests and shrublands (Moen et al. 2004, Wolf et al. 2008a, although the extent of forest expansion may be constrained by soil conditions (Sutinen et al. 2005) and ecosystem processes, such as grazing (Wookey et al. 2009). In the southernmost parts of the Barents Region, colonization by new species may lead to development of new types of communities and ecosystems. The southwestern parts are expected to become climatically suitable for deciduous hardwood species such as Quercus robur and Tilia cordata (Koca et al. 2006), although they are unlikely to establish before 2100 because of lags in species responses to climate change (Svenning and Sandel 2013). The total area of tundra is anticipated to remain stable in the region as losses to expanding forests are compensated by northward expansions of tundra into polar desert habitats (Wolf et al. 2008a).

Societal impact from biodiversity change
Changes in species distributions may have profound consequences for conservation management (Table 2) and governments' ability to fulfill international environmental conventions. Protected areas may no longer adequately conserve the flora and fauna of the region, as species move out of, or invade reserves , Kujala et al. 2011, calling for evaluation of forecasted biodiversity changes in protected areas together with more proactive (adaptive) planning (Araújo et al. 2004, Pressey et al. 2007. Biodiversity protection traditionally controls introduction and spread of invasive species to avoid extinctions of native ones. The challenge will be to enhance the spread of native or accepted species, and to control the spread of unwanted, exotic species that are potential pests (Chapin et al. 2007). Successful conservation strategies need to facilitate species migration so that native species are allowed to adjust their distributions to match new climatic conditions, for example by minimizing the distance between reserves and fostering land use that favors dispersal between reserves and biodiversity hotspots , Kujala et al. 2011. This may be done, for example, by managing natural corridors such as streams and rivers to remove barriers and increase their continuity (Williams et al. 2005). For some species, assisted migration (McLachlan et al. 2007), a highly debated alternative (Ricciardi and Simberloff 2009), or captive breeding/ cultivation may be necessary. All these actions need proper risk assessments and require accommodation by legal frameworks. Spread of new species implies new potential services, whereas the (local) disappearance of native species means that social adjustments to the loss of certain services may be needed. Stimulation of innovation and entrepreneurship to exploit such new services may be needed.

Interactions among ecosystem services
So far, we have mostly ignored potential interactions among ecosystem services to facilitate identifying the direct cause-andeffect relationships of climate change associated with each ecosystem service considered. In reality, such interactions are, however, numerous and important. In Figure 4, we have indicated the most significant ones. Changes in all provisioning services except for wood production are expected to influence summer outdoor recreation (summer taken to be equivalent of the entire bare-ground season), because harvest of these resources may change. Likewise, we envision that changes in biodiversity-related services affect cultural ties to the land, as would changes in forestry and reindeer herding practices. Changes in forestry associated with changes in wood production might also affect the abundance of moose, ptarmigan, edible berries, and conditions for reindeer grazing. In addition, planting of new tree species may result in new ecosystem types and facilitate invasions. In contrast, larger moose populations might negatively affect wood production because of sapling herbivory. Making projections for ecosystem services being influenced by many other services, such as summer outdoor recreation, is more difficult than for those lacking such interactions. If interactions among the societal consequences and adaptation strategies are considered, the situation becomes even more complex.

DISCUSSION AND CONCLUSION
This expert assessment and literature review of major ecosystem services relevant to the Barents Region project changes in many of the ecosystem services derived from natural ecosystems in the region. Although substantial changes in many ecosystem services were projected, the literature drawn upon here does not predict collapse or disappearance of any service (Table 1). It might be that the present literature does not take into account the impact of the more extreme climate change scenarios (IPCC 2013), which imply that the climate in most of the Barents Region by 2100 will be much warmer and wetter than what is able to maintain Arctic and subecosystems . Indeed, the exact nature and magnitude of the changes in ecosystem services are difficult to assess, given the limited predictive power for many services. This Ecology and Society 20(3): 32 http://www.ecologyandsociety.org/vol20/iss3/art32/ uncertainty partly reflects the dependence of many services on multiple factors, often working in opposite directions (Schmitz et al. 2003) and that we lack empirical analogues in the case of combinations of novel climates and ecosystems ). This problem is greatest for services influenced by changes in other services ( Figure 4). Large-scale social, economic, and political changes will influence the extent of climate change as well as the impact on different resource use interests under varying market conditions and population-change scenarios. The ability to forecast climate effects on ecosystem functioning also depends on approaches that include links and feedbacks between biotic components of the ecosystem (Schmitz et al. 2003). Given that the net outcome may often be determined by the strength of species interactions, which vary spatially, quantitative predictions are often likely to be specific for a certain place and time. Also, because empirical studies addressing such spatial variation are few , Callahan et al. 2013, generalizations for a large and variable region such as the Barents are inherently difficult to make. However, given that we constructed chains of cause-and-effect relationships for each ecosystem service, these can serve as hypotheses, for example, for predictive modeling in situations when quantitative projections are feasible and necessary, and for guiding adaptive monitoring systems for observing and acting upon future changes (Lindenmayer and Likens 2009) .   Fig. 4. Potential interactions among the ecosystem services studied. Arrows indicate that change in one ecosystem service is expected to have repercussions on or lead to changes in another service. In cases where the interactions go both ways, the arrows are double-headed. Only the interlinkages judged to be most relevant are depicted by arrows.
Alternatives for adapting societies to changes in service provision were identified for each ecosystem service (Table 2), although many obstacles to successful adaptation were foreseen, such as inadequate legislation, land-use conflicts, and mutually incompatible management actions to improve ecosystem services (Table 2). Even if such obstacles are overcome, there might be limits to adaptation if the changes in ecosystem services alter cultural practices and lead to loss of places important to communities (Adger et al. 2009). However, this study views change in the targeted region only, and does not take into account changes in market or resource use conditions that may impact population or ecosystems as a result of, for instance, change in other regions or in global product supply chains. In addition, there are many interdependencies and potential linkages among adaptation strategies to changes in different ecosystem services, but exploring these in greater depth was beyond the scope of the paper, because it necessitates modeling of these interactions. Major reorganization of human use of ecosystems may occur as a result of the combined effects of climate change and globalization (O'Brien and Leichenko 2000). Even in the absence of climate change, changes in social-ecological systems during the 21st century may be large. All the studied ecosystem services have either experienced large changes in their provision or how they have been utilized by humans during the 20th century. Forestry and reindeer husbandry have undergone rationalization and mechanization (Sandström et al. 2006), fundamentally changing for example working conditions (Hjelm 1991). The game species analyzed have changed in population size, as has the hunting legislation. For example, moose have gone from being very rare to abundant, with moose hunting becoming culturally and economically important in rural areas (Hörnberg 2001). Furthermore, mushrooms only started to be utilized widely for food after the Second World War in Sweden, and berry picking has shifted from being a source of extra income for local communities to being organized by labor hire companies with labor from, for example, Thailand (Sténs and Sandström 2013). Likewise, tourism and recreation has grown to become a major part of both local and global economies. Taken together, these changes demonstrate both substantial adaptive capacity of local communities, as well as the large role of reorganization and internationalization of production in resource use.
Societies and sectors will vary in their adaptive capacity. Reindeer husbandry as a largely traditional indigenous practice is not only dependent on local provision of ecosystem services but is also impacted by all other resource use in the areas, as it constitutes a very small sector with few practitioners and little economic national importance in comparison with, for instance, the large role of forestry. Given this situation, socioeconomic, political, and climate change impacts will necessarily need to be assessed together (Keskitalo 2008, Rees at al. 2008. Moreover, socially-oriented observations (SOO) based on local people's perceptions and statistics, help to identify the main issues and targets for life quality, human capital, and environmental improvement, thus setting targets for further monitoring. SOO in the Russian European North show that human-induced disturbances, such as uncontrolled forests cuttings and poaching, are increasing while observed changes in climate and biota have become additional factors influencing land use and overall sustainability (Vlasova and Volkov 2013).
The predicted changes in ecosystem services are likely to occur gradually, although rapidly emerging surprises with associated state changes are almost inevitable (Lindenmayer et al. 2010). Species distributions and vegetation composition are unlikely to be in equilibrium with climate at the end of the 21st century (Svenning and Sandel 2013). Although this means that some of the changes likely to occur in response to climate change during Ecology and Society 20(3): 32 http://www.ecologyandsociety.org/vol20/iss3/art32/ the 21st century will not be realized until later, it also means that nonstationarity is to be expected (Milly et al. 2008), and that continuous change needs to be taken into account in resource management. In addition, surprisingly slow ecosystem changes or lack of expected changes (Callaghan et al. 2013) also need to be better understood. Apart from the risks, a warmer climate may further provide new opportunities and possibilities for society, for example in making resources more accessible and resulting in potentially higher general productivity (although large-scale extreme events may be an obstacle). Taking advantage of the opportunities that may arise will require investments, knowledge, and material resources that are readily available to only a few actors, such as transnational corporations, and may thus impact the distribution of resource use rights in practice (cf. Keskitalo 2008). The policy decisions taken to mitigate and adapt to climate change or to respond to other long-term changes could substantially modify the direct responses of ecosystem services to climate. Building capacity for different and diverging communities to cope with, or adapt to, changes in ecosystem services, including beyond the local or regional level, and especially under relatively quick economic change and urbanization, is a crucial issue for future studies.
Responses to this article can be read online at: http://www.ecologyandsociety.org/issues/responses. php/7607