A spatially explicit, indicator-based methodology for ...

Czúcz B, Torda G, Molnár Zs, Horváth F, Botta-Dukát Z & Kröel-Dulay Gy (2009): A spatially explicit, indicator-based methodology for quantifying the vulnerability and adaptability of natural ecosystems. In: Filho WL, Mannke F (eds): Interdisciplinary Aspects of Climate Change. Peter Lang Scientific Publishers, Frankfurt, pp. 209-227.

Chapter 10

A spatially explicit, indicator-based methodology for quantifying the vulnerability and adaptability of natural ecosystems Bálint Czúcz, Gergely Torda, Zsolt Molnár, Ferenc Horváth, Zoltán Botta-Dukát & György Kröel-Dulay Institute of Ecology and Botany of the Hungarian Academy of Sciences Alkotmány u. 2-4., H-2163 Vácrátót, HUNGARY

Abstract Ecosystems contribute inconspicuously, yet fundamentally, to human well-being by supplying vital goods and services, including genetic resources, habitat maintenance and climate and runoff regulation. The combined effects of climate change and other global change drivers may impose dramatic impacts on species and ecosystems worldwide, with potentially detrimental consequences on human society. In this chapter we present a vulnerability assessment for the natural and semi-natural ecosystems of Hungary, calculating the local exposure, sensitivity and adaptive capacity of different habitat types. Exposure was calculated using six different global climate model (GCM) outputs comprising of four different models and three emission scenarios, providing a cross-section of the climatic and socio-economic uncertainties within the projections. To estimate the sensitivity of habitats, four types of climate sensitivity were identified and estimated either quantitatively or semi-quantitatively. Adaptive capacity of habitat occurrences was assessed using landscape ecological evaluation of the quality and distribution of habitat patches. Three potential indicators of adaptive capacity were identified, describing (1) the potential resilience of the individual habitat patches, (2) the local refuge-providing ability of the landscape, and (3) the connectivity and permeability of the landscape. By combining results of exposure, sensitivity and adaptive capacity, climatic vulnerability maps of natural ecosystems were produced. This case study, prepared for the Hungarian National Climate Change Strategy, provides the first example of a methodology to give quantitative estimation of the potential climatic vulnerability and adaptive capacity of ecosystems based on a detailed habitat database.

Keywords climate change, ecosystem services, adaptation policy, vulnerability assessment, adaptive capacity, habitat, bioclimatic model, landscape index

Introduction The combined effects of climate change, associated disturbances (e.g., flooding, drought, wildfire, insect outbreaks, ocean acidification) and other global change drivers (especially land-use change, pollution and over-exploitation of resources) will seriously challenge most species and ecosystems worldwide during the course of this century (Fischlin et al., 2007). Major losses in biological diversity seem to be inevitable, resulting in a general decrease in the level of goods and services supplied by ecosystems, especially the regulating and supporting services, including genetic resources, habitat provision and climate and runoff regulation (Fischlin et al., 2007; Duraiappah et al., 2005). The erosion of local and global biodiversity prognosticates an increased likelihood of “ecological surprises” (Fischlin et al., 2007; Burkett et al., 2005), which might impose detrimental consequences on human welfare. To address the pervasive risks and uncertainties pertinent to climate change, impact adaptation and vulnerability (CCIAV) studies have surged in this decade. Most studies for estimating climate impacts on ecosystems and biological diversity can be characterized by using either a correlative, or a mechanistic approach (Fischlin et al., 2007). Mechanistic models are based on the current understanding of the dynamics of energy, water, nutrient and carbon among major species groups and their physical environment. Such models are unrivalled in representing vegetation transformations on a global scale; however, for biodiversity on lower spatial levels the application of these data-hungry models is rather limited. Correlative models, on the other hand, are generally based on more readily available species distribution data, producing projections on future distribution by assuming a statistical relationship between actual distribution and climatic needs of the species (Guisan & Zimmermann, 2000). Nonetheless, due to the limitations of the model structure and the data sources applied, there is generally no straightforward way to take processes of natural adaptation into account. Such models usually apply two extreme assumptions of non-adaptive systems (usually referred to as “no dispersal” of species) or perfectly adaptive systems (“universal dispersal”), as a generalized “interval projection”. Thus, studies employing correlative models can be regarded as simple “impact assessments” (the first stage of CCIAV assessments sensu Füssel & Klein, 2006), and according to Rothmann and Robinson (1997) the evolution towards realistically adaptive agents is inevitable as assessments become increasingly integrated. A further issue is raised by uneven data coverage between taxonomical groups, showing significant imbalance towards easily observable, popular and charismatic species groups, however, modelling studies that include large numbers of species are generally accepted to reveal general tendencies for biodiversity. The ecosystem vulnerability model, presented here, is fundamentally based upon established techniques of correlative species distribution models, but extending the framework into a “first generation vulnerability assessment” (the second stage of CCIAV assessments sensu Füssel & Klein, 2006). The principal goal of such assessments is to understand the vulnerability of the investigated systems as a function of exposure to climatic forcing, the inherent sensitivity of the exposure units, and their capacity of autonomous adaptation (Metzger el al., 2005): vulnerability = f(exposure, sensitivity, adaptive capacity) = = f(potential impact, adaptive capacity), Potential impact means the impacts that may occur without adaptation (considered here as a function of exposure and sensitivity), whereas adaptive capacity accounts for the ability of a system to adjust to climate change to moderate potential damages (IPCC, 2007, Glossary).

The main initiative objective of this modelling study was to describe imminent regional impacts of climate change on biodiversity for the Hungarian National Climate Change Strategy1. The spatial scope of the study was set to the country of Hungary (~93 000 km2), situated in Central Europe. On a temporal scale, the study focuses on the relatively near future (time horizon set to 2025, with an outlook for 2050) according to the basic interest of stakeholders reflected in the design principles of the National Climate Change Strategy. These circumstances determined three important underlying concerns of our analysis design. Accordingly, we intended to implement a methodology, that is (1) policy relevant, (2) realistic and (3) appropriate for the addressed spatial and temporal scales. Most models that account for ecological impacts of climate change work either with species or with major biomes as exposure units. However, neither of these approaches seems to be appropriate for regional or national level ecosystem vulnerability assessments; there are too few biomes and too many species which lack reliable data on life traits and distribution. Accordingly, we selected an ecosystem classification with a high thematic resolution and local relevance with respect to the study area: the major habitat types of Hungary, building upon the Hungarian National Habitat Classification System (Bölöni et al., 2007). In this sense, habitats (habitat types) are more than just the spatial locations of species occurrences – habitats can rather be defined as communities of species living together, determined by trophic linkages and/or similar environmental (e.g. climatic) needs. Consequently, while species react individually to the external changes, climate impacts on species sharing similar climatic needs will presumably be similar. Owing to the completeness and the local ecological relevance of the classification system used, we considered habitats superior to species for estimating overall ecological vulnerabilities on a regional level. The main source of habitat distribution data in Hungary is the MÉTA database, a country-level, vegetation database, relying on recent field survey results (Molnár et al., 2007; Horváth et al. 2008). Habitat data were collected on a network of 35 ha hexagonal grid cells. Data for each grid cell comprise a list of the occurring habitat types, along with their estimated area and habitat quality (naturalness), as well as several other attributes. Altogether 86 different types of natural and semi-natural habitats were distinguished, with a detailed Habitat Guide (Bölöni et al., 2007) to assist the more than 200 surveyors, and standardize the mapping procedure, performed between 2003 and 2006. With respect to the policy goals, the available data sources and the characteristics of first generation vulnerability assessments, two main objectives were set: o to give a comprehensive evaluation of the vulnerability of all habitat types on a national level, and o to provide a spatially explicit evaluation of the potential impacts and adaptation capacity for the most vulnerable habitats. Consulting with the national nature conservation authorities, a further, third goal was also added: o to compile a synthesis of practical policy recommendations tailored for Hungary. To meet these objectives, we developed a new, spatially explicit, indicator-based methodology for quantifying the vulnerability of natural and semi-natural ecosystems, based on the MÉTA database. The remainder of this chapter is organized according to the construction of this methodology, following the outlined structure of first generation vulnerability assessments. First we summarize the changes in the climatic drivers to which

1

The National Climate Change Strategy, accepted unanimously by the Hungarian Parliament on 17 March 2008, can be downloaded from http://klima.kvvm.hu/documents/14/nes_080219.pdf (in Hungarian)

ecological systems are exposed (climatic exposure). Next we analyze the changes that climatic exposure might cause to ecosystems (potential impacts), identifying and quantifying all the relevant sensitivities of the different habitat types. Further, we introduce a conceptual framework to calculate landscape ecological indicators of the adaptive capacity, and give general policy recommendations based on the lessons learned and literature survey. Finally we conclude this chapter by placing this methodology in the context of current developments in CCIAV research.

Exposure to climate change In the CCIAV jargon, exposure generally refers to “the nature and degree to which a system is exposed to significant climatic variations” (IPCC, 2001, Glossary). Understanding the character and magnitude of climatic changes that the investigated system will probably be exposed to is the starting point of most climate impact studies. To get a reliable picture on the direction and uncertainties of the climatic changes that Hungary is currently facing, we resorted to climatic scenarios from the third phase of the Coupled Model Intercomparison Project, run by the World Climate Research Programme, available from the IPCC Data Distribution Centre (Meehl et al., 2007)2. In order to allow for both climatic and socioeconomical uncertainties of the plausible futures, a range of six climate change scenarios were selected, representing a combination of four different global climate models (GCM) and three emission scenarios. The main characteristics of the projected monthly temperature and precipitation values for the region of Hungary are summarized in Table I. The low resolution GCM outputs were downscaled spatially using the simple pattern scaling approach proposed by Parry and Carter (1998), using high resolution present climate surfaces (temperature from the WORLDCLIM database and precipitation from the Hungarian Meteorological Service – see Hijmans et al., 2005 and HMS, 2001) adjusted by the difference (or ratio) between future projections and control runs with the respective GCMs. We calculated with 30-year climatic means for the periods 1960-1990 (the baseline), 2010-2040 (for 2025) and 2035-2065 (for 2050). To get more biologically meaningful variables, from the downscaled means we calculated twelve simple bioclimatic indices, which are frequently used in ecological applications (bio1, 4 and 10-19 of the WORLDCLIM database). The habitat distribution models introduced in the following section relies on this set of bioclimatic variables.

Habitat sensitivity and potential impacts According to the conventions of the climate impact research community, the sensitivity of a system designates the degree to which it is affected either adversely or beneficially by climate change (IPCC, 2007, Glossary). In the case of habitat types, climatic sensitivity can be interpreted as the aggregated sensitivity of the specialist species (species with specific habitat requirements) to changing macroclimate conditions. Based on the assumption that the present day distribution of habitats reflects the fundamental climatic requirements of the adherent organisms, the direct sensitivity of the ecosystems can be modelled with correlative bioclimatic distribution models. However, this approach has several limitations (see e.g. Pearson & Dawson, 2003 for a detailed discussion), from which we highlight the following four: o specific dispersal strategies and limitations are hardly possible to take into account in correlative models;

2

http://www.ipcc-data.org/

o there are many indirect feedback mechanisms involved in shaping species and community responses (e.g. biotic interactions between species; but also complex physical or socio-economical feedback mechanisms); o communities never reach a real climatic equilibrium due to the complexity of interactions, large-scale environmental changes (i.a. biotic succession and climatic alterations) and evolution; o extrapolation to future climatic conditions not occurring in the calibration region of the model can result in biologically meaningless results. If the objects of analysis are not individual species but communities, one further issue of concern arises. Since different species respond idiosyncratically, present communities might increasingly disassemble as climate change proceeds (Thuiller, 2004). Disintegration starts when the changes are too fast for the species to follow, or the available habitat patches are too small or degraded. Accordingly, at least for the starting phase of the changes, it is reasonable to assume, that climatic conditions that have for long supported the existence of a habitat, will still remain available for the majority of its species, providing ground for the community to survive. This problem is closely related to the first limitation mentioned above (the uncertainty pertinent to species dispersal). One of the main goals of the adaptive capacity component of our vulnerability model is to address this issue, and provide indicators of habitat availability and connectivity. Traditional species distribution models excel in grasping direct relationships between the species and their environment (direct sensitivity). Indirect linkages, especially when spatially or temporally remote impacts are involved, are hardly possible to reckon with in correlative models. Unfortunately, this is typical for climate change driven physical mechanisms (e.g. changes in fire or flood regimes, termed as “associated disturbances” by Fischlin et al., 2007), leading to significant changes in the physical environment with important impacts on ecosystems (indirect sensitivities). Indirect sensitivities are very difficult to model, but it is possible to evaluate them qualitatively based on expert knowledge on a national level, which is generally a desirable practice in advanced CCIAV assessments (Rothmann & Robinson, 1997). Keeping in mind these considerations and our main policy goals, we proceeded in the following way: o we first evaluated the (direct) climatic sensitivity of all habitat types using correlative distribution models; o we identified 12 common habitat types with high direct sensitivity, and gave spatially detailed evaluation of the potential impacts (changes in occurrence probabilities) for these habitat types; and o we gave a comprehensive qualitative evaluation (with the indirect sensitivities included) on a national level for all habitat types, aggregated from model results and qualitative expert analysis. To estimate the direct sensitivity of the habitats and the potential impacts of climate change thereon, we used conditional inference-based decision tree models (Hothorn et. al. 2006). Apart from the bioclimatic indices, we also considered non-climatic variables (soil and distance from water bodies) as model predictors. Climatic sensitivity of the habitat types was interpreted on the basis of the predictive power of different models including either only climatic, non-climatic or both types of predictors.3 Spatially explicit potential impact maps for the most climate sensitive habitat types were calculated on basis of the models including all 3

the list of the predictors and the evaluation metrics for climate sensitivity can be found in Czúcz et al, 2007.

predictors. All the calculations were performed in an ensemble-like framework, based on twelve random bootstrap model runs for each of the aforementioned six climate change scenarios, using the means as consensus predictions, and the range of predictions indicating uncertainties. Figure 1. shows the mean projection for mixed oak-hornbeam woodlands, the most widespread forest type of Hungary. To accomplish the comprehensive qualitative evaluation, we first enumerated the most important mechanisms originating sensitivities, which were hidden from the statistical model: o Warming and drying can lead to increasing frequency of wildfires, which may affect many habitat types, especially forests already at the edge of their climatic tolerance. o Changing river flow regimes are projected to decreasing flood frequency with increasing flood levels – both of these can adversely affect riverine habitats. o Habitats attached to springs and upwelling waters from aquifers (fens and partly other types of wetlands) can be seriously affected by decreasing precipitation sums. This relationship works primarily on large spatial and temporal scales (Ruprecht & BottaDukát, 2000). o Upwelling waters play a moderately known, but important role in the formation and sustenance of alkaline soils and the related habitat types. Changing hydro-geological structures might on the long term harm salt steppes harbouring high portion of the remaining ecological capital in Hungary. The latter two processes are not likely to exert considerable impact on a limited time scale as ours. Still, we included these, since due to the slow feedback cycles and the large inertia of the systems, it is important to be conscious about these feedbacks well before they start to evoke problems. Besides the physical mechanisms of climatic sensitivity, there are several habitat types, which will definitely be affected by the direct impacts of climate change, however, due to the limitations of the study area and habitat database, these relationships could not be recognized by the model: o Weak relationships can only be detected for prevalent habitats by the statistical model. For extremely rare habitat types even strong relationships may remain unnoticed. o Habitats occurring in the warmest / most arid regions of the country require extrapolation from the model. (This problem could theoretically be handled by including data from a wider geographical range.) All natural and semi-natural habitat types were evaluated qualitatively with respect to each hidden sensitivity mechanism by a group of five independent experts on a five grade scale. In order to bring expert-based and quantitative results to a common platform, we derived ordinal-scale potential impact estimates from the modelled direct impacts, aggregated on the national level. In Table II we present the complete list of projected climate impacts on the natural and semi-natural habitat types of Hungary.

Indicators of adaptive capacity Estimating the adaptive capacity of the exposure units is generally the most important, but also the most problematic part of traditional climate change vulnerability assessments (Grothmann & Patt, 2005). For most CCIAV studies adaptation and adaptive capacity mostly refers to planned adaptation based on deliberate policy decisions, whereas the spontaneous adjustment (= autonomous adaptation) of the agents in reaction to the persistent changes is generally considered under the scope of sensitivity (Carter et al., 2007). Notwithstanding the common application, in this chapter adaptive capacity refers to autonomous adaptation caused by organism, species and ecosystem level responses to the changing external conditions.

However, autonomous adaptation is mostly influenced by the distribution and quality of the habitats and the level of additional anthropogenic pressures (e.g. pollution), which are largely determined by the forms and intensity of human land use. Thus, autonomous ecological adaptation acts, in fact, as a feedback mechanism from land management and natural resource allocation towards keystone ecosystem services, such as climate and runoff regulation, pest control, and especially the maintenance of genetic resources. In this sense understanding autonomous ecological adaptation is an important step in developing comprehensive socioeconomic adaptation strategies. As it is common practice in the evaluation of complex systems (e.g. Haddad, 2005; Yohe & Tol, 2002), we proposed a framework of adaptive capacity indicators that account for the most important coping options or mechanisms of the system. The proposed indicators aim at estimating the chances that the species inhabiting the studied habitat types can avoid degradation (and the resulting erosion of genetic diversity) caused by climate alteration. To this end, we distinguished three basic components (mechanisms) of ecological resilience to climate change: o ecosystem resilience: habitat occurrences in more favourable (more natural) condition are more likely to be resilient to climatic stresses; o refuge-based adaptation: withdrawing of species into local microclimatic refuges in a generally unfavourable environment; o migration-based adaptation: large-scale movement of species following macroclimatic changes (the shift of climatic zones) by migration. There are no sharp boundaries between the three conceptual mechanisms of adaptation (e.g. adapting by migration is equal to finding a distant zonal refuge). In effect, all three forms of adaptation are the manifestations of the structure, ordination and coordination of living systems, on different levels and spatial scales. Although occurring on different spatial scales, all the mentioned adjustment processes rely upon the resilience resulting from the spatially structured diversity of the systems, making it possible to dampen external changes by internal restructuring. Naturally, different spatial scales imply different temporal scales. Accordingly, there will be a presumable shift in the availability and importance of adaptation mechanisms from local resilience towards large-scale migration, which will remain the only option for more and more species to avoid extinction as climate change becomes increasingly severe. Although migration can be accompanied with local extinctions, on larger scales it can moderate biodiversity loss, if it helps to avoid global extinctions. Migration can also help to sustain local ecosystem services by letting transformed habitats be colonized by new arrivals. Migration can, naturally, have negative aspects as well (e.g. the migration of invasive species introduced from distant localities), but climate induced distribution changes of native species should still be accepted as an important natural adjustment mechanism of ecosystems. Unfortunately, the dispersal capacity of most species is poorly known; several studies suggest it to be inadequate to track climatic changes in a landscape that is heavily fragmented by humans (e.g. Opdam & Wascher, 2004; Skov & Svenning, 2004). However, there is some paleoclimatic evidence suggesting that in extreme situations, species are able to migrate substantially faster, governed by rare, long distance dispersal events and subsequent local colonization (Bacles et al., 2006). Early Pleistocene-Holocene vegetation history suggests that both refuge- and migration-based adaptation played an important role in that species were capable of surviving significant changes in climate. The proposed indicators for the outlined three mechanisms are based on the basic limiting factors of the processes. Local ecosystem resilience (determined by the compositional, structural and functional diversity of the habitat – see. e.g. Noss, 1990) is limited primarily by patch size and local land use intensity. According to this, we estimated local resilience for

each habitat patch in each grid cell with the “natural capital” of the habitat (the product of patch area and habitat quality from the MÉTA database – for a justification of this approach see e.g. ten Brink, 2000 or Parkes et al., 2003). Concerning the other two mechanisms, while migration-based adaptation requires large connected habitat networks, refuge-based adaptation is limited mainly by the local/regional availability of appropriate recesses. Based on this, to estimate local refuge-based adaptive capacity we used a habitat diversity index (the naturalness-weighted Shannon diversity of the habitat patches in the grid cell)4, whereas local migration potential was assessed by a landscape connectivity index (ESLI patch connectivity, see Vos et al., 2001 and Swihart & Verboom, 2003). Landscape connectivity was calculated for each cell considering all similar habitat types in the neighbourhood of the focal grid cell, assuming an effective dispersal distance of ~1-8 km. Results of adaptive capacity calculations are shown in Figure 2 for two important and typical habitat types. To ease interpretation, we classified the calculated indicator values into five ordinal categories based on the habitat type specific distribution of the values (for habitat resilience), or the analysis of model landscapes (for local diversity and landscape connectivity).

Vulnerability and policy recommendations Although the individual indicators are often more meaningful for practical purposes than aggregated ones (Patt et al., 2005), composite indicators can also be informative for policy decisions, for example by distinguishing especially “hopeful” and “hopeless” situations. Areas highly exposed (i.e. high potential impact) to climate change with low adaptive capacity can be identified as particularly vulnerable sites, deserving urgent policy attention. Areas of high exposure and high adaptive capacity can be seen as potential targets for monitoring. Such an analysis is given for oak-hornbeam forests in Figure 3, highlighting “high vulnerability areas” (high exposure and generally low adaptive capacity), and “high adaptivity areas” (high exposure, and either generally high or varying adaptive capacity), which are potential targets for monitoring of climate impacts and autonomous adaptation processes. The identification of vulnerable situations can be of particular interest to authorities and organizations engaged in nature conservation. Such analyses may form the basis for: o objective evaluation of current reserve network, defining spatial priorities, o identification of potential migration networks, and o determination of unsustainable conservation targets in the face of climate change. Current conservation practices are generally considered to be poorly adapted to the challenges caused by climate change (Fischlin et al., 2007). To provide some information on the relationship between the vulnerability of oak-hornbeam woodlands, and the current protection status of the different areas, we placed the outlines of the Natura 2000 ecological network on Figure 3. Interestingly, examining the pattern of high vulnerability and high adaptivity areas, we can see, that the former ones seem to avoid, while the latter ones seem to fit within the Natura 2000 areas. This has definitely a lot to do with the designation priorities of Natura 2000 areas, providing justification for the delineation of the network, at least for oakhornbeam woodlands. Although the illustration is informative, just one vulnerability map of a single habitat type cannot tell much about the general conservation potential of the landscape. Nonetheless, mass evaluation of similar maps has the potential to provide significant assistance to the development of conservation and land-use strategies. 4

the details can be found in Czúcz et al., 2007, p. 224-225.

Besides providing factual risk indicators, the whole process of vulnerability analysis can also enhance general understanding of the linkages and mechanisms of the complex systems. This understanding can help us to find appropriate points for monitoring and intervention, to enhance adaptation and mitigate adverse side effects. Since undisturbed ecosystems are dynamically stable, self-regulating systems, the best available adaptation policy approach for nature conservation is enhancing the autonomous adaptive capacity of ecosystems (Fischlin et al., 2007). In accordance with the conceptual model that was used in the “adaptive capacity” part of the vulnerability analysis discussed above, this can be done on three levels: o enhancing the local natural status of the habitats (compositional, structural and functional diversity) by reducing additional stresses and degrading land use practices (drainage, over-harvesting, over-grazing, etc.); o preserving and increasing local/regional landscape diversity; o preserving and increasing the connectivity and permeability of the landscape for the species of natural habitats on a regional, national or preferably international level. Under stable environmental conditions, many species and habitats can be preserved by simply creating reserves of appropriate dimensions. In a changing climate where species need to track the changes, the state of the agricultural-urban matrix that surrounds natural ecosystems will have crucial impact on the success of adaptation. Thus, under advancing climatic changes, biodiversity in densely populated areas, such as Central Europe, will not be retainable without integrating the aspects of conservation into the policies of several sectors (e.g. agriculture or water management). There is already growing evidence that environmental impacts may result from perverse or unintended effects of policies from other sectors (Fischlin et al. 2007, based on Chopra et al. 2005). Accordingly, conservation policies should increasingly focus on managing areas outside the protected reserves (von Maltitz et al., 2006). Fortunately, the existence of policy side-effects is getting increasingly recognized and addressed in several sectoral policies (e.g. EU Water Framework Directive, Agroenvironmental schemes, Pro Silva forest management, etc.), but further advancements are still desirable (Table III.).

Conclusions At first sight the survival of the species might seem to be of marginal relevance for mankind, compared to economic and social challenges and conflicts shaping our everyday life. Nevertheless, ecosystems contribute inconspicuously yet fundamentally to human well-being and significant impairment of ecosystems may lead to ecological surprises with detrimental side-effects for human society. Ecological changes induced by anthropogenic climate change have the potential to impose major negative feedback on society, which makes it important to reckon with the ecological impacts of climate change in adaptation (or mitigation) strategies. Accordingly, ecological systems should be regarded as an important subsystem of socioeconomic systems (coupled by ecosystem services) and general CCIAV models should incorporate appropriate sub-models for estimating climate change impacts on ecosystems. Recognizing the complexity of interactions between biophysical and socio-economical systems, climate change impact, adaptation and vulnerability studies are increasingly shifting from linear to more complex chains of analysis, capable of addressing policy relevant questions (Rothman & Robinson, 1997; Füssel & Klein, 2006). Both Chapter 2 and Chapter 4 of the Fourth Assessment Report of the IPCC underline the importance of cross-sectoral, integrative studies and set these as research priorities (Carter et al., 2007, p. 162; Fischlin et al., 2007, p. 249). However, even the most complex models consist of relatively simple constituent sub-models, and the development of new methods to address specific climate

change problems is also considered a research priority by IPCC (Carter et al., 2007, p. 161). We believe that the simple methodology to perform regional or national level vulnerability assessment for natural ecosystems facing climate change presented here constitutes such a relevant sectoral improvement, and we hope that it might also contribute to national or regional integrative adaptation strategy assessments, by improved representation of local level biodiversity feedback on human society.

References Bacles, C.F.E., Lowe, A.J. and Ennos, R.A. (2006), “Effective seed dispersal across a fragmented landscape”, Science, Vol. 311, pp 628. Bölöni, J., Molnár, Zs., Illyés, E. and Kun, A (2007), “A new habitat classification and manual for standardized habitat mapping”, Annali di Botanica, Vol. 7, pp. 105-126. ten Brink, B.J.E. (2000) “Biodiversity indicators for the OECD Environmental Outlook and strategy - A feasibility study”, RIVM. Available: http://www.rivm.nl/bibliotheek/rapporten/402001014.pdf Carter, T.R., Jones, R.N., Lu, X., Bhadwal, S., Conde, C., Mearns, L.O., O’Neill, B.C., Rounsevell, M.D.A. and Zurek, M.B. (2007), “New assessment methods and the characterisation of future conditions”, in Parry, M.L., Canziani, O.F., Palutikof, J.P., van der Linden, P.J. and Hanson, C.E. (Eds.), Climate Change 2007: Impacts, Adaptation and Vulnerability. Contribution of Working Group II to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change, Cambridge University Press, Cambridge, pp. 133-171. Chopra, K., Leemans, R., Kumar, P. and Simons, H. (2005), Ecosystems and Human Wellbeing: Volume 3: Policy Responses, Island Press, Washington, DC. Czúcz, B., Kröel-Dulay, Gy., Rédei, T., Botta-Dukát, Z., Molnár, Zs. (2007), “Climate change and biological diversity – explorative analysis for a more effective adaptation strategy in Hungary” (in Hungarian with English summary), Institute of Ecology and Botany of the Hungarian Academy of Sciences, Vácrátót, Hungary, Available: http://www.botanika.hu/download-01/NES Duraiappah, A., Naeem, S., Agardi, T., Ash, N., Cooper, D., Díaz, S., Faith, D.P., Mace, G., McNeilly, J.A., Mooney, H.A., Oteng-Yeboah, A.A., Pereira, H.M., Polasky, S., Prip, C., Reid, W.V., Samper, C., Schei, P.J., Scholes, R., Schutyser, F. and van Jaarsveld, A. (2005), Ecosystems and Human Well-being: Biodiversity Synthesis. Island Press, Washington, DC. Fischlin, A., Midgley, G.F., Price, J.T., Leemans, R., Gopal, B., Turley, C., Rounsevell, M.D.A., Dube, O.P., Tarazona, J. and Velichko, A.A. (2007), “Ecosystems, their properties, goods, and services”, in Parry, M.L., Canziani, O.F., Palutikof, J.P., van der Linden, P.J. and Hanson, C.E. (Eds.), Climate Change 2007: Impacts, Adaptation and Vulnerability. Contribution of Working Group II to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change, Cambridge University Press, Cambridge, pp. 211-272. Füssel, H.M. and Klein, R.J.T. (2006), “Climate change vulnerability assessments: An evolution of conceptual thinking”, Climatic Change, Vol. 75, No. 3, pp. 301-329. Grothmann, T. and Patt, A. (2005), “Adaptive capacity and human cognition: The process of individual adaptation to climate change”, Global Environmental Change, Vol. 15, No. 3, pp. 199-213.

Guisan, A. and Zimmermann, N.E. (2000), “Predictive habitat distribution models in ecology”, Ecological Modelling, Vol. 135, No. 2-3, pp. 147-186. Haddad, B.M. (2005), “Ranking the adaptive capacity of nations to climate change when socio-political goals are explicit”, Global Environmental Change, Vol. 15, No. 2, pp. 165-176. Hijmans, R.J., Cameron, S.E., Parra, J.L., Jones, P.G. and Jarvis, A. (2005), “Very high resolution interpolated climate surfaces for global land areas”, International Journal of Climatology, Vol. 25, No. 15, pp. 1965-1978. HMS (2001), Magyarország Éghajlati Atlasza [Climate Atlas of Hungary] (in Hungarian), Hungarian Meteorological Service, Budapest. Horváth, F., Molnár, Zs., Bölöni, J., Pataki, Zs., Polgár, L., Révész, A., Krasser, D. and Illyés, E. (2008), “Fact sheet of the MÉTA Database 1.2.”, Acta Botanica Hungarica (in print) Hothorn, T., Hornik, K. and Zeileis, A. (2006), “Unbiased recursive partitioning: A conditional inference framework”, Journal of Computational and Graphical Statistics, Vol. 15, No 3, pp. 651–674. IPCC (2001), Climate Change 2001: Impacts, Adaptation, and Vulnerability. Contribution of Working Group II to the Third Assessment Report of the Intergovernmental Panel on Climate Change. McCarthy, J.J., Canziani, O.F., Leary, N.A., Dokken, D.J., White, K.S. (Eds.), Cambridge University Press, Cambridge. IPCC (2007), Climate Change 2007: Impacts, Adaptation and Vulnerability. Contribution of Working Group II to the Third Assessment Report of the Intergovernmental Panel on Climate Change. Parry, M.L., Canziani, O.F., Palutikof, J.P., van der Linden, P.J. and Hanson, C.E. (Eds.), Cambridge University Press, Cambridge. von Maltitz, G.P., Scholes, R.J., Erasmus, B. and Letsoalo, A. (2006), “Adapting conservation strategies to accommodate impacts of climate change in southern Africa”, Working paper No. 35, Assessments of Impacts and Adaptations to Climate Change (AIACC), Washington DC. Meehl, G.A., Stocker, T.F., Collins, W.D., Friedlingstein, P., Gaye, A.T., Gregory, J.M., Kitoh, A., Knutti, R., Murphy, J.M., Noda, A., Raper, S.C.B., Watterson, I.G., Weaver, A.J. and Zhao, Z.-C. (2007), “Global climate projections”, in: Solomon, S., Qin, D., Manning, M., Chen, Z., Marquis, M., Averyt, K.B., Tignor, M. and Miller, H.L. (Eds.) Climate Change 2007: The Physical Science Basis.Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change, Cambridge University Press, Cambridge Metzger, M.J., Leemans, R. and Schröter, D. (2005), “A multidisciplinary multi-scale framework for assessing vulnerabilities to global change”, International Journal of Applied Earth Observations and Geoinformation, Vol. 7, No. 4, pp. 253-267. Molnár, Zs., Bartha, S., Seregélyes, T., Illyés, E., Botta-Dukát, Z., Tímár, G., Horváth, F., Révész, A., Kun, A. and Bölöni, J. (2007), “A grid-based, satellite-image supported, multi-attributed vegetation mapping method (MÉTA)”, Folia Geobotanica, Vol. 42, pp. 225-247. Noss, R.F. (1990), “Indicators for monitoring biodiversity: a hierarchical approach”, Conservation Biology, Vol. 4, No. 4, pp. 355-364. Opdam, P. and Wascher, D. (2004), “Climate change meets habitat fragmentation: linking landscape and biogeographical scale levels in research and conservation”, Biological Conservation, Vol. 117, No. 3, pp. 285-297.

Parkes, D., Newell, G. and Cheal, D. (2003), “Assessing the quality of native vegetation: the ‘habitat hectares’ approach”, Ecological Management and Restoration, Vol. 4, Supp. 1, pp. S29-S38. Parry, M.L. and Carter, T. (1998), Climate Impact and Adaptation Assessment: A Guide to the IPCC Approach, Earthscan, London. Patt, A., Klein, R.J. and de la Vega-Leinert, A. (2005), “Taking the uncertainty in climatechange vulnerability assessment seriously”, Comptes Rendus Geosciences, Vol. 337, No. 4, pp. 411-424. Pearson, R.G. and Dawson, T.P. (2003), “Predicting the impacts of climate change on the distribution of species: are bioclimate envelope models useful?” Global Ecology and Biogeography, Vol. 12, No. 5, pp. 361-371. Rothman, D.S. and Robinson, J.B. (1997), “Growing pains: A conceptual framework for considering integrated assessments”, Environmental Monitoring and Assessment, Vol. 46, No. 1, pp. 23-43. Ruprecht, E. and Botta-Dukát, Z. (2000), “Long-term vegetation textural changes of three fen communities near Cluj-Napoca (Romania)”, Acta Botanica Hungarica, Vol. 42, pp. 265-284. Skov, F. and Svenning, J.C. (2004), “Potential impact of climatic change on the distribution of forest herbs in Europe”, Ecography, Vol. 27, No. 3, pp. 366-380. Swihart, R.K. and Verboom, J. (2003), “Using ecologically scaled landscape indices to assess biodiversity consequences of landuse decisions”, in Swihart, R.K. and Moore, J.E. (Eds.) Conserving Biodiversity in Agricultural Landscapes: Model-based Planning Tools, Purdue University Press, West Lafayette, IN, USA. Thuiller, W. (2004), “Patterns and uncertainties of species' range shifts under climate change”, Global Change Biology, Vol. 10, No. 12, pp. 2020-2027. Vos C.C., Verboom J., Opdam P.F.M. and Braak C. (2001), “Toward ecologically scaled landscape indices”, The American Naturalist, Vol. 157, No. 1, pp. 24-41. Yohe, G. and Tol, R.S.J. (2002), “Indicators for social and economic coping capacity: Moving toward a working definition of adaptive capacity”, Global Environmental Change, Vol. 12, No. 1, pp. 25-40.

Table I: Main characteristics of the climate change scenarios used (annual, summer and winter temperature [°C] and precipitation changes [%]) for the region of Hungary, grouped by GCMs (a) and emission scenarios (b). 2025 a)

2050

2025

2050

Ta

Ts

Tw

Ta

Ts

Tw

Pa

Ps

Pw

Pa

Ps

HADCM3 A2

1.4

1.6

1.2

2.6

2.9

2.3

-2.4

-8.2

3.1

-3.6

-13.4

7.0

CNCM3 A2

1.3

1.6

1.1

2.3

2.4

2.1

-4.0

-8.3

0.6

-5.6

-9.6

-0.8

CSMK3 A2

1.2

1.2

1.2

1.7

1.8

1.5

4.6

4.7

4.1

-1.5

0.4

-3.3

GFCM21 A2

0.9

1.1

0.7

1.8

2.1

1.6

0.7

-1.9

4.5

-5.2

-14.2

6.2

Ta

Ts

Tw

Ta

Ts

Tw

Pa

Ps

Pw

Pa

Ps

Pw

HADCM3 A2

1.4

1.6

1.2

2.6

2.9

2.3

-2.4

-8.2

3.1

-3.6

-13.4

7.0

HADCM3 A1B

1.6

1.7

1.6

2.9

3.3

2.6

1.4

-3.3

5.6

-1.1

-10.9

9.4

HADCM3 B1

1.4

1.6

1.1

2.4

2.6

2.3

-0.2

-1.5

0.4

-4.7

-12.4

3.5

b)

Pw

Table II: Estimated level of climate change potential impacts (PI) on the natural and seminatural habitat types in Hungary (according to the Hungarian National Habitat Classification System, Bölöni et al. 2007; 5: most vulnerable
1: least concern) PI

Habitat types (MÉTA code, Natura2000 code*)

5

Euhydrophyte communities of fens (A4, 3160)f, Transition mires and raised bogs (C23, 7140)f, Rich fens (D1, 7230)f, Calluna heaths (E5, 4030)a, Alder and ash swamp woodlands (J2, 91E0)f, a, Acid beech woodlands (K7a, 9110)a, Closed lowland steppe oak woodlands (L5, 91F0)d, b, c, Open sand steppe oak woodlands with openings (M4, 91I0)d, b, c, Acid coniferous woodlands (N13)a

4

Festuca rubra hay meadows and related communities (E2, 6510)a, Salt meadows (F2, 1530)a, f, Closed rock grasslands, species rich Bromus pannonicus grasslands (H1, 6190)b, Birch mire woodlands (J1b)f, Lowland oakhornbeam woodlands (K1a, 91F0)a, Limestone beech woodlands (LY3, 9150)b, Open loess steppe oak woodlands with openings (M2, 91I0)b, c, d, Open salt steppe oak woodlands with openings (M3, 91I0)b, c, d

3

Oligotrophic reed and Typha beds of fens, floating fens (B1b)f, Tussock sedge communities (B4, 7230)f, Soft and hard water flushes (C1)f, Molinia meadows (D2, 6410)f, Mesotrophic meadows (D34, 6440)f, Water-fringing and fen tall herb communities (D5, 6430)f, a, Cynosurion grasslands and Nardus swards (E34, 6520)c, a, Bromus erectusBrachypodium pinnatum xero-mesophilous grasslands, dry tall herb communities and forest steppe meadows (H4, 6210)a, Open vegetation of shadowed rocks (I4)a, Willow mire woodlands (J1a, 91E0)f, Riverine oak-elm-ash woodlands (J6, 91F0)e, Oak-hornbeam woodlands (K2, 91G0)a, Beech woodlands (K5, 9130)a, Closed and mixed steppe oak woodlands on foothills (L2x, 91I0)d, Ravine woodlands (mesic rock woodlands rich in Acer pseudoplatanus) (LY1, 9180)b, Poplar-juniper steppe woodlands (M5, 91N0)d, b, c

2

Salt marshes (B6, 1530)f, Arrhenatherum hay meadows (E1, 6510)a, Artemisia salt steppes (F1a, 1530)f, Tall herb salt meadows and salt meadow steppes (F3, 1530)f, Dense and tall Puccinellia swards (F4, 1530)f, Annual salt pioneer swards of steppes and lakes (F5, 1530)f, Open sand steppes (G1, 6260)c, Closed sand steppes (H5b, 6260)c, Riverine ash-alder woodlands (J5, 91E0)a, Turkey oak - sessile oak woodlands (L2a, 91M0)a, Closed acid oak woodlands (L4a, 91M0)a, Continental deciduous steppe thickets (M6, 40A0)a

1

Standing water communities with Trapa, Lemna, Salvinia and Ceratophyllum (A1, 3150), Euhydrophyte communities with Nymphaea, Nuphar, Utricularia and Stratiotes (A23, 3160), Slowly running water communities with Potamogeton and Nymphoides (A3a, 3150), Athalassal saline euhydrophyte communities (A5, 1530), Eu- and mesotrophic reed and Typha beds (B1a), Glyceria, Sparganium and Schoenoplectus beds (B2), Water-fringing helophyte beds with Butomus, Eleocharis and Alisma (B3), Non-tussock beds of large sedges (B5), Tall herb communities of floodplans and marshes (D6, 6430), Achillea salt steppes on meadow solonetz (F1b, 1530), Calcareous open rock grasslands (G2, 6190), Acid open rock grasslands (G3, 6190), Calcareous rock steppes (H2, 6240), Slope steppes on stony ground (H3a, 6240), Closed steppes on loess, clay, tufa (H5a, 6240), Amphibious communities on river gravel and sand banks (I1), Semi-desert vegetation on loess cliffs (I2), Riverine willow scrub (J3, 91E0), Riverine willow-poplar woodlands (J4, 91E0), Acid oak-hornbeam woodlands (K7b, 91G0), Closed termophilous oak woodlands (L1, 91H0), Turkey oak - pedunculate oak woodlands (L2b, 91M0), Open acid oak woodlands (L4b, 91M0), Slope woodlands (LY2, 9180), Mixed relic oak woodlands on rocks (LY4, 9150), White oak scrub woodlands (M1, 91H0), Continental deciduous rock thickets (M7, 40A0), Thermophilous woodland fringes (M8, 6210), Calcareous Scots pine woodlands (N2)

a: direct sensitivity modelled – b: direct sensitivity unmodelled (extremely rare habitat) – c: direct sensitivity unmodelled (too close to southern/xeric distributional limit) – d: indirect sensitivity (wildfires) – e: indirect sensitivity (river flow regimes) – f: indirect sensitivity (upwelling water). *Natura2000 codes are only provided for unambiguous matches

Table III: A list of cross-sectoral policy recommendations prepared for the Hungarian National Climate Change Strategy to minimize climate change impacts on biodiversity (Czúcz et al. 2007) Sectors

Policy recommendations

Nature conservation

o o o o

Water management

o develop / focus on retention oriented water management policies (instead of current drainage oriented policies) o introduce ecological aspects in the management regime of reservoirs and floodplain areas o follow the prescriptions and recommendations of the EU Water Framework Directive

Forestry

o emphasize natural forest management techniques providing continuous forest cover o introduce different regulations for forests with semi-natural structure and composition and timber / biofuel plantations o acknowledge forests of low canopy closure (forest–steppe mosaics) as valid management targets

Agriculture

o maintain / reintroduce elements of traditional landscape management (mowing, grazing) o provide buffer zones around areas of high conservational value o promote low intensity agricultural techniques o increase the heterogeneity of agricultural landscapes with networks of tree lines and hedges

Transportation

o increase the number of green bridges, ecoducts and other types of wildlife crossings on national motorways o introduce / maintain strips of semi-natural vegetation (hedges, forests) on the margin of major roads

prepare priority lists of climate sensitive habitats and species design and perform necessary restoration activities improve water retention of reserves wherever possible preserve / enhance mosaicity of habitats and successional states wherever possible o introduce / increase network concept in reserve selection o elaborate a concept / measures for protecting networks (of hedges, tree rows, roadsides) in agricultural / industrial / urban landscapes o improve monitoring activities

Figure 1. Actual distribution, and modelled occurrence probabilities of oak-hornbeam woodlands for the Southern-Transdanubia (Dél-Dunántúl) region of Hungary. a) actual distribution according to the MÉTA database; b) projected occurrence probabilities for the present (1960–1990) climate; c) projected occurrence probabilities for the 2010–2030 period; projected impact (decrease in occurrence probability) of climate change for the 2010–2030 period

Figure 2. Adaptive capacity indicators for two major habitat types: oak-hornbeam woodlands (a, c, e) and mesotrophic meadows (b, d, f) in the Southern-Transdanubia (Dél-Dunántúl) region of Hungary. a-b) level of habitat resilience (estimated by the product of patch area and habitat quality); c-d) level of local diversity (estimated by the naturalness-weighted Shannon diversity of the habitat patches); e-f) level of landscape connectivity (estimated by the ESLI patch connectivity index – see Vos et al., 2001).

Figure 3. Highlighted areas of beech-hornbeam woodlands, and the Natura 2000 ecological network in Hungary. a) “high vulnerability areas” with high exposure (the projected decrease in occurrence is more than 10%) and a generally low adaptive capacity (all three adaptive capacity indicators have low scores); b) “high adaptivity areas” with high exposure, and a generally high adaptive capacity (recommended for monitoring)