Figure 3 shows the area and number of vulnerable people (i.e. Fig. 1 classes C and D) with respect to each of the ES vulnerability indices for the 40 combined climate and socio-economic scenarios. The relative levels of vulnerability reproduced are consistent with the socio-economic and climate scenarios. In general terms, the dystopian scenarios show greater vulnerability in terms of both the number of vulnerable people and the area vulnerable than the utopian ones for the majority of sectors, particularly in terms of the number of people affected. This reflects the lower significant potential impacts due to the more optimal IAP input settings in the scenarios where innovation is successful, and higher coping capacities where higher capital stocks are available in the utopian scenarios. Similarly in most cases, the more moderate climate scenarios (B1 emissions, low climate sensitivity) have lower vulnerability than the extreme climates (A1 emissions, high climate sensitivity). In addition to these general trends, Fig. 3 shows nuances that reflect the exact combination of climate model, level of climate sensitivity/emissions scenario, socio-economic scenario and sector.
Irrespective of climate scenario, VAFOOD is considerably lower in the SoG scenario compared to the other socio-economic scenarios (Fig. 3a). In SoG, agricultural yields are low, GDP and irrigation efficiency have decreased, and dietary habits remain focussed on space-intensive meat production. Furthermore, the population grows rapidly (+23 % compared to −9 % in Icarus) so there is much greater demand for food. Consequently, as in SoG there are very few grid cells that do not produce food, grid-cell self-sufficiency is high and VAFOOD is low. In contrast, in WRW, little to no food is produced in Norway and a belt from southern France across the Alps to Hungary and, thus, there are more cells classified as being not self-sufficient in terms of food provision, and thus vulnerable. However, in most climate scenarios, VPFOOD for SoG (Fig. 3b) is comparable with levels of vulnerability in the WRW and Icarus scenarios.
The agricultural model in the IAP does not take food trade between cells into consideration, hence, a grid-cell measure of food provision is used as the best available indicator of food security. As such, this difference between VP and VA in SoG reflects that, although food is being grown wherever possible, urban areas, which do not grow food, remain vulnerable unless they have capital available to access food from areas that do. As an index of vulnerability in terms of the food provisioning ES (rather than food availability) this is reasonable given that urban areas will always be dependent on food producing areas, and those with low levels of available capital will be most vulnerable. Furthermore, with an increasing population in SoG these urban areas become even more dependent on the areas that supply food: and thus more vulnerable in terms of food provision.
There is relatively little difference between the climate scenarios in terms of vulnerability, CSMK3 and MPEH5 show relatively greater vulnerability and HadGEM and GFCM21 show relatively less, but in general the patterns are similar across climate scenarios, and there is little difference between the high and moderate climate scenarios. Again, the differences identified are driven by the extra stress put on the system by climate change: where there is greater stress, such as in the hotter, drier GFCM21 scenario the vulnerability to food provision is projected to be less because food is being grown wherever possible, at the expense of other sectors.
In general, the WEI shows increasing vulnerability through the socio-economic scenarios in the order Riders < WRW < SoG < Icarus both in terms of VA and VP. Moderateclimate scenarios show less vulnerability than the high-end scenarios for each GCM. Milder and wetter GCMs (e.g. CSMK3) show less vulnerability than hotter and drier ones (e.g. GFCM21). In some scenarios, such as MPEH5 (and IPCM4) the difference between socio-economic scenarios is much less notable, than in GFCM21 (Fig. 4). Spatial patterns also reflect the complexities of socio-economic interactions. Under MPEH5, WRW does not show vulnerability in the UK, Belgium or the Netherlands that is present in SoG due to improved coping capacity in the former scenario (Fig. 4). However, Greece and the south coast of France are actually less vulnerable in the dystopian SoG, than in the utopian WRW. This is due to WRW’s improving GDP (+94 %) leading to lifestyle changes and the use of more water-intensive appliances; a factor most notable in areas with low baseline GDP. Additionally, SoG’s reduced water efficiency leads to irrigation becoming less profitable, compounding the pressures on the agriculture sector and leading to greater areal expansion of agriculture (and concurrent reduction in VAFOOD). However, this leaves more water available for other purposes and can reduce water exploitation in some areas. These cross-sectoral interactions help to reveal potential synergies and trade-offs within and between sectors in both the utopian and dystopian scenarios.
Vulnerability to biodiversity loss is hard to conceptualise, particularly in terms of how it would manifest itself in practice. However, it is useful to have an index to consider in relation to the changes in other indicators. Biodiversity vulnerability is absent in all areas where the number of species present at baseline remain the same in the scenario. It increases with the number of species lost due to change in either climate- or habitat-space and when there is not sufficient coping capacity available to address this through active conservation.
VABIODIVERSITY follows a pattern that reflects decreasing coping capacity across the socio-economic scenarios (Riders < WRW < Icarus < SoG). The same pattern is clear for VPBIODIVERSITY in the two utopian scenarios; however, dystopian SoG has greater VPBIODIVERSITY than Icarus due to SoG’s higher population.
Based on a bioclimatic envelope model, the Biodiversity Index is one of the most climatically sensitive indices: moderate climate scenarios show considerably less VP and VA than their high-end counterparts, irrespective of socio-economic scenario. The level of vulnerability in the CSMK3 high-end climate scenario is about a third less the high-end climate scenarios of the other GCMs (Fig. 3). This is most likely because this scenario leads to the least changes relative to baseline with regard to both climate and land-use, causing fewer negative implications for species.
In all climate scenarios VAFLOOD and VPFLOOD increase in the order WRW < Riders < Icarus < SoG. Although climatic changes do influence the levels of vulnerability, the differences across climate models are very small (VA/VPFLOOD range < 0.5 %). This results from the small rises in sea-level, the primary driver of coastal flooding, between moderate and high-end climate scenarios (0.12 m and 0.3 m respectively). Fluvial flooding increases in the wetter climate scenarios, but there are few cells which change class between scenarios due to the strong influence of topography, i.e. the flood prone zones. This contrasts with the water and biodiversity indices where there are significant shifts in spatial pattern. Differences between the socio-economic scenarios are more evident reflecting lower coping capacity in the more dystopian scenarios. Changes in population drive the differences in VPFLOOD within the utopian/dystopian scenario pairs; coping capacity drives the differences between the two sets of scenarios.
SoG has the greatest VAINTENSITY (≈20 %) followed by Icarus (VAINTENSITY ≈ 8 %) with neither of the utopian scenarios showing significant vulnerability (VAINTENSITY < 3 %). Changes in food production (see 3.2.1) drive the majority of this vulnerability. In SoG all available land area is converted to agriculture, at the expense of less intensive land uses and only those areas with little agricultural development (i.e. eastern Sweden) or higher coping capacity (France) are not vulnerable. In Icarus, the declining population reduces the stress on the system and so, despite similarly low levels of coping capacity and failed technological innovations, less area is vulnerable. In the utopian scenarios, there is both a lower need for extreme agricultural intensification and greater coping capacity, which reduces vulnerability significantly. There is very little difference between the climate scenarios in terms of intensity. This is consistent with other studies which show that socio-economic changes are likely to have a greater influence on land use intensity than climate change (Rounsevell et al. 2006).
As agricultural expansion into new areas is a key driver of the landscape diversity index it shows a similar order of scenarios to food provision and an equally weak response to climate drivers. However, whilst the diversity index is closely linked to the food indicator for the standard settings of the socio-economic scenarios, an analysis of uncertainty associated with the parameter values for the amount of set-aside, agricultural yields and dietary preferences shows that the index is considerably more sensitive to these values than the food index. This is because these parameter values shift the land use distribution between classes within the food-producing land uses (i.e. between arable and the different grassland types) and between food-producing and abandoned land. As such the index has the potential to demonstrate very different responses to the land use intensity and the food provision indices.
Multi-sectoral aggregate vulnerability
Figure 5 highlights the spatial patterns of cross-sectoral vulnerability for a low and high vulnerability combination of socio-economic and climate scenarios. In the low vulnerability case (moderate climate scenario with CSMK3, WRW, 2050s) there are a few key areas of vulnerability linked mostly to single indicators – for example, southern Spain (water exploitation) and Estonia (food) along with some coastal areas, particularly in northeast Italy (flood). There are very few areas that are vulnerable according to multiple indicators, the most notable being Fennoscandia and the Alps (food and diversity) and pockets of France, Austria and Hungary (food, biodiversity and diversity). This is reflected by the low proportion of Europe vulnerable for at least one indicator both in terms of people (46 %) and area (36 %).
In the high vulnerability case (high-end climate scenario with GFCM21, Icarus, 2050s) European vulnerability is considerably greater with 81 % of the area and 88 % of the baseline population (443,004,000 people) vulnerable in at least one sector. Furthermore, significant areas of Fennoscandia, France, Spain, Italy, Lithuania, Romania, Bulgaria and Greece are vulnerable for more than one indicator. The types of vulnerability differ with geographical area: in Fennoscandia, the vulnerability is due to changes of food and diversity, whilst in southern and eastern Europe, and the areas around Prague and Paris, the vulnerability is to changes of biodiversity and water exploitation. Some areas are vulnerable for three indicators mostly along the coast where they are exposed to floods, but also in areas of Germany, the Czech Republic and Romania where vulnerability to changes of land use intensity is identified.