Step 1: projections of future climatic suitability for butterflies
Projections using three different calibrated BEMs under five climate change scenarios show clear spatial differences in the projected locations of the climatically most suitable areas for different butterfly species, though some areas overlap. An illustration comparing four species is provided in SM, section S3. However, there are also several sources of uncertainty in projections of suitability. Figure 2 explores two of these, showing a comparison of the impacts of alternative climate change scenarios and of different modelling methods for one butterfly species (Parnassius mnemosyne). Three BEM techniques can be compared for the same climate change scenario (1MMH—see Table S2, SM) in panels A, B and C, and show large differences in projected suitability. Similarly, there are even greater differences in suitability across five climate change scenarios applied to a single modelling technique (GAM) when comparing panels A, D, E, F and G.
Figure 2 emphasizes the large uncertainties involved in projecting future species suitability, presenting potential challenges for adaptation planning. On the other hand, a focus on those areas where model projections agree across scenarios and models, such as in panels A, C, D and E, could offer valuable information for designing adaptation strategies that are more robust under uncertainty.
For one example species, Hesperia comma, the BEM results were analysed in detail and employed to investigate two hypothetical adaptation responses: (1) the construction of dispersal corridors between present-day species populations and future suitable areas, and (2) translocations, involving the capture, transport and release of species into future suitable sites (the corridor endpoints). For this particular species, the analysis identified thirty 10-km grid cells that showed the highest climatic suitability under the climatic conditions of 2051–2080 and which already contain grasslands. To provide an illustration of the impacts of varying the modelling method and climate change scenario, this process was repeated for three scenarios (2LLL, 1MMH and 4HMM) and two modelling methods (GAM and GLM). The results show that there are clear spatial differences between the locations of the thirty future climatically most suitable grid cells, including the ten most suitable cells in the results based on different models or scenarios (Fig. 3).
Step 2: the policy environment for adaptation measures
In 2006, the agri-environmental schemes (AES) implemented by farmers covered an area of about 24,500 hectares, which is almost ten times more than the area managed by all other landowners combined. Metsähallitus, the public enterprise that manages state-owned land and water areas, is second in importance, managing ca. 2,700 ha of semi-natural habitats annually. A number of NGOs, such as the Finnish Association for Nature Conservation and the World Wildlife Fund (WWF), have an additional, minor role in grassland conservation.
The utilized agricultural area (UAA) in Finland is ca. 2.3 million ha, representing about 8 % of the total land area (TIKE 2011). Natural meadows and pastures constitute approximately 1 % of the UAA. Public funding for the management of semi-natural grasslands comes almost exclusively from the national AES directed to farmers, with some additional funding from the regional authorities directed to local-level NGOs. Although there has been sufficient funding for biodiversity-related AES measures in recent years, their uptake has been rather low, slowing down but not reversing the decline of traditional rural biotopes. This has been attributed to low payment levels and high transaction costs (Schulman et al. 2006).
The future agricultural policy landscape over the period of analysis here (to 2080) is difficult to anticipate. Pyykkönen et al. (2010) have presented forecasts of structural changes in Finnish agriculture from 2010 to 2020, which is particularly relevant in the context of short-term adaptation responses. According to their estimates, the continuing decline seen over recent decades in the number of livestock farms is expected to increase by an additional 50 % by 2020. The largest decreases are projected for dairy farms, which are currently responsible for maintaining most of the semi-natural grasslands. The projected increases in other livestock are unlikely to compensate these losses due to their relatively small extent. There is also an ongoing drive to increase livestock units per farm, which is likely to increase pressures to abandon the less productive semi-natural grasslands and replace them with cultivated pastures. These underlying socio-economic trends are likely to exacerbate the effects of climate change, through the loss of suitable future sites and poorer connectivity. However, it is possible that currently unforeseeable changes could arise in either EU- or Finnish agri-environmental policies, which would provide greater incentives for semi-natural grassland management through AES or some other funding mechanism.
Step 3: identification and appraisal of options
Following the identification of future risks, potential adaptation options to enhance the resilience of grassland butterflies to climate change were identified and appraised. An important element was to compare alternative options in terms of their relative effectiveness and their efficiency (costs), as measured through a CEA. However, it was also necessary to assess whether these options were sufficient to ensure the survival of grassland species, recognizing that thresholds to ecosystem viability might be exceeded as the limits to adaptation for these natural ecosystems are approached.
On the basis of an initial literature review and scoping analysis, three different types of adaptation options were identified: (1) AES measures, (2) dispersal corridors and (3) species translocations. It should be noted that the first two options have inter-linkages, because constructing dispersal corridors for species dispersal would be likely to use AES measures along the planned corridor areas. Species translocation is a quite different alternative.
Determining the area of grassland to support viable populations
The CEA in this paper is based on analysis of the most cost-effective way to achieve pre-defined target levels, in this case a scientific target level for the amount of grassland habitat required to support viable grassland butterfly species populations. This is an absolute target resilience level for ecosystem sustainability and follows from a high-level policy goal to preserve these grassland ecosystems (Salminen and Kekäläinen 2000). The analysis of the viability threshold for the CEA was based on existing survey analysis, with analysis in a GIS database. This mapped three types of grassland habitats at a resolution of 25 m × 25 m in Finland:
grasslands managed with funding from the AES (Arponen et al. 2013);
grasslands identified as valuable for agro-biodiversity in a previous nation-wide survey (Vainio et al. 2001), but which currently do not have an AES contract; and
common grasslands that are of low or moderate value for agro-biodiversity (Kuussaari et al. 2007).
The grassland cover information was integrated with data from more than 170 butterfly monitoring transects. The data from the 30 transect areas with most abundant grasslands and records of one or more grassland specialist butterfly species (usually over a number of years) were analysed separately, and the median of the summed cover of the three grassland types in the 2-km grid cells with the species transects was calculated. The results suggested that ca. 2.5 % of the landscape should be covered by grassland habitats to support viable grassland butterfly populations. A viable population was broadly defined here as a local population of a given species which had persisted over several years, based on transect monitoring data or local butterfly recorders’ knowledge. Many of the chosen transects have been established on sites known a priori to have important grasslands or to have historical records of demanding grassland species. The analysis of each of the three main adaptation options is presented in the following sections.
AES measures are currently the main tool for conservation of grassland biodiversity in Finland, implemented through the use of financial incentives (payments) to farmers. A detailed description of the Finnish AES measures can be found in MAF (2007). Based on the earlier expert evaluation by Grönroos et al. (2007), we selected those AES measures which were known to have at least some significance for butterflies, i.e. that the area where the measure takes place can serve as habitat for some of the more common grassland butterfly species. The potential AES options included one obligatory basic measure: buffer strips of >3 m along waterways, and three voluntary special measures: management of traditional biotopes, buffer zones of >15 m along waterways and environmental fallow. For all AES measures, both obligatory and voluntary, the farmer is entitled to monetary compensation (€ per hectare) based on generally agreed support levels. The current extent of implementation of these measures (in hectares) was obtained from Finnish agricultural statistics.
The relative effectiveness of these AES measures for enhancing butterfly diversity was assessed using the results from previous Finnish case studies. These data and the valuation process are described in more detail in SM, section S4. In this approach, the actual species richness estimates for each AES measure were replaced with relative estimates. This facilitated quantitative comparison and ranking of individual AES measures according to their significance for butterflies (Fig. 4).
The results revealed that the voluntary special measure for the management of traditional biotopes, usually grazing, is by far the most efficient AES measure promoting suitable habitats for grassland butterflies that are currently in decline (Fig. 4). All other AES measures were of relatively low value for grassland butterflies.
The total areas of agricultural land under each AES option are currently rather low (Table 1). This hampers their significance for overall species conservation and adaptation. A national target level for the management of traditional biotopes has been set at 60,000 ha (Salminen and Kekäläinen 2000), but there is still some way to go to achieve this goal, as the area under AES contracts has remained rather stable over the past decade.
Most of the measures in the Finnish AES are targeted primarily to reduce nutrient run-off from arable areas (Grönroos et al. 2007; Aakkula et al. 2012). These measures are typically applied on species-poor habitats of little or no value for butterflies. It is therefore evident that the Finnish AES is not optimally designed for benefiting grassland butterflies. Additional work on targeted new AES options for enhancing biodiversity is therefore considered a priority.
Dispersal corridors—also known as ecological corridors (Öckinger and Smith 2008) or movement corridors (Simberloff et al. 1992)—offer an explicit option for enhancing species adaptation to climate change (Huntley et al. 2006; Heller and Zavaleta 2009) and have been highlighted in European adaptation policy (CEC 2009). These are designed to enhance species’ range expansion to new, currently unoccupied areas by linking present-day populations to alternative locations that are projected to become climatically suitable in the future.
The analysis of the potential effectiveness of dispersal corridors for grassland butterflies under future climate change is laborious. Thus, the analysis here focused on one example species, the Hesperia comma butterfly, which provides a useful indicator for valuable grasslands. This species was also used in a subsequent analysis of translocation.
The GIS analysis first identified six 10-km grid cells containing grassland habitats and projected to be climatically very suitable under future climate in 2051–2080 by the BEMs developed for Hesperia comma (see Fig. 3 above). The six target 10-km grid cells were selected based on model outputs for combinations of three climate change scenarios and two modelling methods (Fig. 5). They were used both as tentative destination points in the corridor constructions and as target areas for species translocation. For characteristics of the corridor target cells, see SM, section S5.1.
A GIS environment was used to construct six example corridors between one of the 10-km grid cells with current known records of Hesperia comma, and each of the six 10-km grid cells deemed climatically very suitable under the 2051–2080 climate conditions. The corridor building was done by manually constructing a pathway made of 2 km grid cells linking the present-day occurrence area and each of the target areas (for details, see SM, section S5.2).
Interestingly, the six example corridors varied considerably with respect to their length and the spatial location of the target area (Fig. 5), suggesting that conservation planning for dispersal corridors based on BEMs may be strongly affected by the methodological choices made in the process. In a clear majority of the six corridors, the present-day grassland habitat cover along the corridor is less than the target level of 2.5 %, or 10 ha in a 2-km cell (Table 1). This suggests that very large additional adaptation efforts, such as land cover conversion or AES-based restoration, would be needed to ensure the effectiveness of these dispersal corridors, which implies high costs (see below).
These six example corridors are only a very small selection of the hundreds of potential dispersal corridors. Nevertheless, these examples illustrate that there are considerable uncertainties in selecting suitable corridors attributable to the different climate change scenarios and BEM modelling methods applied. Without doubt, differences in the suitability of corridor pathways would also emerge between different species. Thus, increasing the number of hypothetical corridors to cover more examples for one species and also to cover other species would increase the robustness of the estimates. However, given the low present-day levels of suitable habitats, it is also apparent that additional AES would be needed along tens or even hundreds of kilometres to ensure that these could function effectively as dispersal corridors for grassland species.
The final major option considered is the translocation of species (also referred to as assisted colonization or assisted migration). This option is increasingly advocated as a potentially useful measure to reduce the harmful impacts of climate change on species populations (Hoegh-Guldberg et al. 2008; Thomas 2011). However, the success of species translocation to date has been variable and also somewhat controversial (Ricciardi and Simberloff 2009; Lawler and Olden 2011). Nonetheless, one recent UK study employing a wealth of empirical data shows that assisted colonization could be a useful and cost-effective means to assist butterfly species in tracking climatic shifts more effectively (Willis et al. 2009). In this study, the economic potential for translocation was investigated with a detailed analysis of Hesperia comma, using the identified six example target end points analysed above.
Economic analysis of adaptation options
The preceding analysis provides valuable information on potential options. However, these need to be complemented with other criteria, notably in relation to the costs of the measures, to help prioritize (or rank) the options. CEA is the tool applied in this study (“Step 3: cost-effectiveness of adaptation options” section, above).
CEA is first applied to the AES options. Cost information on the establishment and management of different AES measures was collected from a stakeholder workshop in 2011, a farmer survey 2012 and interviews with key farmers. This was supplemented with data and the literature on AES measures and costs. A summary of the costs and the overall present value costs are presented in Table 2. The overall present value costs are the costs discounted over the time period of the AES agreement, which is a 3-year lifetime for most measures (see Table S6, SM for more detail).
Costs are uncertain because different farms have varying unit costs, depending on a number of context-specific factors, including topography and scale at which the scheme is applied. For example, a survey of nine traditional biotope AESs shows a range of annual establishment costs of €55/ha–€1,214/ha and a range of annual management costs of €174/ha–€1,083/ha. In order to provide an initial indication of this cost uncertainty, ranges are adopted that reflect the variance around the mean in this small survey of traditional biotope AES costs. Thus, low and high cost estimates are shown in Table 2, along with medium estimates. It should be noted that this analysis is undertaken from the perspective of maximizing social welfare (i.e. the net economic benefits that accrue to both consumers and producers). It does not include the effects of subsidies that have an influence on farmers’ decisions regarding AES uptake, because subsidies simply represent a transfer of resources between parties and have no effect on total net economic benefits (see Mishan and Quah 2007).
The cost-effectiveness is next derived by combining these estimates with the earlier analysis of relative species richness (Table 2). The table shows, under each part of the cost range, that all three options are fairly similar in terms of their cost-effectiveness at the order of magnitude level. The use of CEA demonstrates that the ranking changes from that which would result from reliance on costs or effectiveness on their own: the environmental fallow option has the lowest cost per hectare, but the higher species richness of buffer zones more than compensates for their higher costs and they are therefore slightly more cost-effective within a given part of the unit cost range. Conversely, while the traditional biotope has the highest species richness of the three, its relatively high cost indicates it is slightly less cost-effective.
However, Table 2 also shows that comparison across different parts of the cost range can yield multiple rankings. Thus, if traditional biotopes are found to have unit costs at the low end of their range, while the buffer zone has unit costs at the high end of its range, then traditional biotopes might be more cost-effective. The instability in the rankings is likely to be further exacerbated if the uncertainties in the measures of effectiveness are included in the analysis. This highlights how CEA results are not likely to be easily transferable; ranking is more reliable when the precise context is defined, and data collected accordingly. It should also be recalled that AESs are not primarily designed for butterfly conservation, and a broader view of the multiple environmental benefits of these measures is needed when considering their current application—and potentially their ranking for future adaptation options.
The principle of applying the concept of CEA to dispersal corridors is illustrated by building on the analysis in the “Dispersal corridors” section for Hesperia comma. In this case, the costs are associated with the additional interventions needed to achieve the target levels for corridor connectivity, i.e. to achieve the target habitat level of 2.5 % for each of the six indicative corridors, shown in Table 1. This is achieved through the introduction of AES contracts and non-productive investment subsidies for forest habitats to convert cultivated fields and forested land on a half-and-half basis, so that 2.5 % of the cover of the 2-km corridor cells would become suitable habitat for grassland butterflies. In this exercise, it was assumed that the converted habitat patches throughout the corridors would be managed based on AES contracts or non-productive investment subsidies for at least 20 years. Such a time period is generally required for converted cultivated field and forest sites to develop into suitable habitats for grassland butterflies. Here, of the six corridors evaluated, Corridor 2 appears to be most cost-effective based on the establishment and management costs.
Here, it should be noted that the costs of conserving this one butterfly species are estimated separately for one climate scenario/model combination. A more robust adaptation strategy, that considered uncertainty, would possibly need multiple corridors, to connect to a portfolio of target climatically suitable locations.
Finally, cost-effectiveness of the translocation option has been analysed, again building on the results for Hesperia comma. The estimated costs ranged from €5,400 to €6,800 per intervention, with some variation between the six target locations identified from the model outputs from GAM, GLM and the three climate scenarios (Table 3) due to differences in travel costs. The total costs of translocation were a fraction of the costs of stepping-stone corridor construction (cf. Tables 1, 3). However, these costs may not fully capture the activities needed for successful translocation of grassland butterflies, such as potential need for selective cutting of trees or rotational (e.g. five yearly) clearance of bushes from the translocation target site, which cautions against the direct comparison of the options. Furthermore, these costs only relate to the translocation of one species, while the dispersal corridors potentially offer dispersal routes for multiple species and also provide wider environmental benefits. Translocation of multiple species, possibly even all key species elements of a given grassland ecosystem, might also be considered, though cost comparisons with constructing dispersal corridors could become quite complicated. This is because the costs of transferring multiple species are not directly additive—there can be differences between species in translocation costs for a given target location, depending on the future suitability and species’ ecology at the location. Moreover, the success rate of translocations varies from case to case and is often low due to variation in site-specific habitats. Overall, there is a need for a portfolio of translocation sites to take account of climate and model uncertainties in the selection of suitable future climatic locations.
Step 4: farmer survey
Out of the 385 respondents who answered the farmer survey, 54 % stated that the current state of grassland biodiversity conservation was adequate, while a third thought that conservation should be increased. One-tenth of the farmers responded that too much effort had already gone into biodiversity conservation, but three quarters of farmers viewed the AES measures as an effective way to protect biodiversity and increase recreational opportunities, while less than 10 % disagreed with this view. A large majority of respondents also considered beautiful scenery and the strengthening of emotional ties to farmland nature to be important to them.
Finally, the farmers were asked which of the available biodiversity conservation AES measures they would be willing to implement at their farm. Their preferences, in descending order, were the establishment and management of environmental fallows (68 % indicated a willingness to implement), buffer zones (57 %) and traditional biotopes (42 %). Forty-two per cent would also be willing to manage other biodiversity habitats and 35 % wetlands. The relative willingness of farmers may reflect a strategic interest in these measures or may simply reflect which are the easiest to implement, require less work than other measures or are most profitable, i.e. where the subsidy is large in comparison to the realized costs.