Climate change impacts on biodiversity : a short introduction with special emphasis on the ALARM approach for the assessment of multiple risks

1 UFZ, Helmholtz Centre for Environmental Research, Department of Community Ecology, Th eodor-LieserStr. 4, 06120 Halle, Germany 2 Vermont Energy Investment Corporation 155, S. Champlain St, Burlington VT 05401, USA 3 Finnish Environment Institute, Climate Change Programme, Box 140, 00251 Helsinki, Finland 4 L.U.P.O. GmbH, Friedhofstr. 28, 67705 Trippstadt, Germany 5 Department of Conservation Ecology and Entomology and Centre for Agricultural Biodiversity, Faculty of AgriSciences, Stellenbosch University, Private Bag X1, 7602 Matieland, South Africa 6 Sustainable Europe Research Institute SERI Germany, Vorsterstr. 97–99, 51103 Cologne, Germany 7 Department of Plant Ecology, University of Bayreuth, University Street 30, 95440 Bayreuth, Germany

on the ALARM approach for the assessment of multiple risks.In: Ott J (Ed) (2010) Monitoring Climatic Change With Dragonfl ies.How to study multiple risks: setting the scene Climate change and human modifi cation of the landscape are synergistic (Travis 2003), and aff ect biodiversity and the stability of ecosystems.Th is is because certain species will be favoured by changes, while others will not.However, as stated by Fanslow (2006), the simplicity stops here, with Samways et al. (1999) illustrating the diff erential impact of climate change events on a range of closely related species.It is relatively easy to test how increased temperature will aff ect an organism.We can isolate almost any organism, put it in a box and observe how it responds to environmental changes we can simulate in a controlled setting, such as a laboratory.We might fi nd, for example, that warming benefi ts this isolated organism.But what if warming also benefi ts a disease to this organism?What if temperatures become too warm for other organisms on which our hypothetical organism depends such as its prey when it is a predator or a pollinator of its host plant if it is a herbivore?What if warming benefi ts a competitor even more?
Once we step outside the small hypothetical box that defi nes just one organism, or some isolated parts of an ecosystem, and start to ask questions about how it will interact with other "boxes" in the environment, we are quickly inundated with uncertainty about how environmental change will reshape our world.Ecosystems are remarkably complex, which makes it exceedingly diffi cult to predict their behaviour (Walther 2010).In some cases, we may be able to understand how one species affects another species, but in a relatively simple hypothetical system of 50 species, for example, each species potentially interacts with 49 other species.Th is gives no less than 1,225 possible two-way interactions in a simple 50-species system.If then the frame of reference is expanded to larger areas with many more types of ecosystems, it is clear that even a large group of dedicated scientists could not study even a small percentage of the possible two-way interactions using traditional controlled experiments, much less the three-and four-way interactions that are often just as important.
Another challenging aspect to developing an understanding of interactions between components of a complex system is the matter of communication among scientists of diff erent disciplines.Th e diff erent scientifi c disciplines -which can be thought of as diff erent boxes in which scientists work -have traditionally been viewed as distinct and have developed strikingly diff erent languages.As a result, interdisciplinary collaboration tends to be rare because getting through language barriers with someone in a diff erent discipline requires a lot of valuable time and energy for people who generally don't have a lot to spare.
When you want to understand processes in a very large scaled system, but cannot do experiments, modelling is a useful way to synthesize information gathered independently about components of a larger system.
One approach to understand environmental risks over large areas was followed by the EU funded research project ALARM, which had a scope matched only by the ambition of its acronym: "Assessing LArge-scale environmental Risks for biodiversity with tested Methods".

How to study multiple risks: the ALARM approach
Th e objective of the ALARM project (Settele et al. 2005(Settele et al. , 2010; http://www.alarmproject.net) was to apply our best understanding of how terrestrial and freshwater organisms and ecosystems function and to use new ways to assess large scale environmental risks.Th e ultimate aim was to develop and test methods and protocols for such an assessment and provide information that can be used to reduce negative impacts on humans and, in turn, minimize negative human impacts -both direct and indirect.
Research was related to ecosystem services in the broadest sense including the relationships between society, economy and biodiversity.In particular, risks to biodiversity were assessed that arise from -climate change, -environmental chemicals, -biological invasions and -loss of pollinators in the context of current and potential future European socio-economic development options and their respective land use patterns, for which scenarios were applied.
Here, dragonfl ies are mainly and directly impacted by the fi rst 3 factors, by the forth only indirectly.
Risk assessments in ALARM were hierarchical and examined a range of organisational (genes, species, ecosystems), temporal (seasonal, annual, decadal) and spatial scales (habitat, region, continent) determined by the appropriate resolution of current case studies and databases (compare Figure 1).
Socio-economics was a cross-cutting theme that contributed to the integration of driver-specifi c risk assessment methods, developed instruments to communicate risks to biodiversity end users, and indicated policy options to mitigate such risks.So as to have a platform for practical interdisciplinary research, a fi eld site network (FSN) was established within ALARM, where the diff erent ALARM modules conducted joint research.All sites included freshwater as well as terrestrial habitats, including both lotic and lentic environments.Th e FSN covered most of European climates and biogeographic regions, from Mediterranean environments through central European and boreal zones to the subarctic (see Hammen et al. 2010aHammen et al. , 2010b, for further details), for further details).
Th e feature of ALARM that has set it apart from overly complex model exercises is that it made use of scientifi c narratives (or storylines) based on scientists' best understanding of the environmental systems they study (Spangenberg 2007a, Spangenberg et al. 2010, in press).ALARM puts these narratives together to paint a larger picture of how something as large and complex as the environment of a continent will react to diff erent environmental and -interacting with them -socio-economic driving forces.
Just as challenging as reaching an understanding of how environmental change will play out, is translating that understanding into language that policymakers and the general public can understand.
To illustrate what we were trying to do through the ALARM project, we may contrast diff erent forms of environmental storytelling: scientists tend to be reluctant to let a good story to distract attention from the facts, while journalists or activists can often be accused of ignoring facts for the sake of a good story.Th e goal of ALARM is to fi nd a compromise between these ways of telling environmental stories and treat stories as the envelopes to carry facts, bearing in mind that facts are often the basis for a good story.For dragonfl ies the example of the expansion of the Scarlet Darter (Crocothemis erythraea, see Ott 2001Ott , 2007bOtt , 2010aOtt , 2010b) ) is meanwhile well known and besides scientifi c papers also many popular articles or presentations start with this "success story".
After expanding its geographic reach in early 2007 by adding scientists and institutions particularly from outside the European Union, ALARM encompassed a total of more than 250 scientists from 68 institutions from 35 countries, with a total budget of more than 20 Mio.Euro (slightly more than 50% funded by the EU).Th e consortium was co-ordinated by the German Helmholtz-Centre for Environmental Research -UFZ.ALARM started in February 2004, and the EU funding lasted for 5 years until early 2009.It was an Integrated Project (IP) within the 6 th Framework Programme of the European Commission (EC) within the sub-priority 6.3 -Sustainable Development, Global Change and Ecosystems.
In the following chapters we will detail some of the more climate related aspects of biodiversity conservation in general and of the ALARM approaches which are also relevant to dragon-and damselfl ies in particular.

Scenarios, narratives, models, and strategy development
Biodiversity is infl uenced by a combination of natural processes (e.g.evolution, succession, catastrophes) and anthropogenic pressures (e.g.land use, nitrogen deposition, climate change, alien species invasions).From a policy point of view, this situation constitutes an urgent need to identify the human drivers causing pressures on biodiversity, and to develop strategies and policies to mitigate the resulting impacts in order to minimise biodiversity losses (Spangenberg et al. in press).
Given that just like the environment, society and the economy are complex, developing (i.e.neither deterministic nor stochastic) systems, their future interaction and thus the development of biodiversity pressures cannot be predicted or expressed as quantifi ed risks.However, since the system development is path dependant, such pathways can be evaluated by scenario techniques, with each pathway represented by a scenario narrative or story line, and some aspects of each illustrated by computer modelling.In turn, the modelling results have to be interpreted in the context of the narrative to integrate the qualitative elements into the scenarios and the strategy proposals derived from them.Such scenarios are means for the evaluation of potential risks, and policy strategies are the search for or the creation of bifurcation points in the trajectories.
Developing eff ective strategies for biodiversity conservation and management requires the transdisciplinary combination of capabilities, concepts, insights and tools of several disciplines (e.g.ecology, chemistry, economics, and political science) with nonscientifi c knowledge, and so does scenario development.A major challenge is to ensure that the assumptions used in the various modelling exercises are consistent (or at least their interpretation is), that the issues addressed are relevant and the assumptions made are plausible.Th e latter is the contribution of non-scientifi c knowledge and experience, realised in the case of ALARM by establishing a multi-stakeholder Consultative Forum, which had signifi cant infl uence on the scenario formulation.
Regarding the contribution of quantifi ed modelling, so far no comprehensive model has been developed integrating the diverse relevant ecological, economic, individual and societal processes (and even if it existed, it would not be too helpful).Instead, socio-economic, climate and biodiversity models exhibit a wide range of assumptions concerning population development, economic growth and the resulting pressures on biodiversity, and they deal with signifi cantly diff erent time scales and spatial framings. 1 Th erefore it is necessary to derive consistent assumptions and scenario interpretations from a comparative analysis of models and scenarios from several disciplines, 1 Within ALARM, socio-economic, land use and nitrogen deposition models are run, one using the output of the other as input parameter, but land use and biodiversity models must be reconciled by interpretation on the basis of the storyline.assessing their overlaps and the possible contradictions between the results of one and the assumptions of other scenarios.Within the ALARM scenario development process, this is done by interpreting the modelling results against the backdrop of a joint narrative.A complementary, cross-disciplinary knowledge base needs to be developed in order to support eff ective policy decisions and provide a basis for future modelling exercises on all levels.Th is requires the close cooperation within an interdisciplinary team of economists, climatologists, land use experts, biologists, modellers and policy experts.Th e three internally coherent but amongst them contrasting scenarios developed are one of liberal policies (GRAS), one of a continuation of rather mixed EU policies (BAMBU) and one of consequential sustainability policies (SEDG) (see chapter "Th ree basic scenarios.." below, for further details).
Usually, scenarios are based on rather linear extrapolations of past trends, which is a rather unrealistic assumption given the uncertainty inherent to the dynamics of evolving systems.Consequently, the eff ects of non-linear developments need to be taken into account (Walther 2010).Th us, complementing the rather linear scenarios underpinned by simulation runs, shock scenarios have been developed.Th ey serve as sensitivity analysis for the basic scenarios, and although their probability of occurrence cannot be quantifi ed, they illustrate how diff erent future developments can and most probably will be from an extrapolation of past trends.
Given the interaction of the economic, social and natural systems, one illustrative shock to each of the systems is taken into account.Th e climate shock (collapse of the thermo-haline circulation, vulgo: the Gulf Stream) is conceptualised as a modifi cation of the liberal GRAS scenario (as it provides the highest probability for such a shock occurring).Th e economic shock (Peak Oil: oil price quadrupling) and the societal shock (a pandemic) are applied to BAMBU, the current politics scenario, as they are not dependant on the policy changes assumed under both variants.Economic and social aspects, environmental impacts of the shocks and of the most plausible reaction of the political system to them are developed in the scenario narratives.
Th e scenarios and their interpretations have been presented to decision makers to support refl exive policy development eff orts.Th ey identify the most important drivers, show how they need to be modifi ed, changed or abandoned in order to achieve a signifi cant reduction of biodiversity loss, contributing to the new EU policy goals for 2020 and beyond.In this context, ALARM provides an up-to-date information base for decision makers which intend to ex ante evaluate policy strategies before implementing them.Th us, feedback circles, rebound eff ects and other system characteristics can be taken into account, supporting policies for eff ective protection of biodiversity Th ree basic scenarios and three deviations (shocks) GRAS (GRowth Applied Strategy): Deregulation, free trade, growth and globalisation are policy objectives actively pursued by governments.Environmental policies focus on damage repair and limited prevention based on cost-benefi t-calculations.No emphasis is put on biodiversity.
BAMBU (Business-As-Might-Be-Usual): Policy decisions already made in the EU are implemented and enforced.At the national level, deregulation and privatisation continue except in "strategic areas".Internationally, there is free trade.Environmental policy is perceived as another technological challenge.
SEDG (Sustainable European Development Goal): Th e sustainability of societal development is enhanced by integrated social, environmental and economic policy.Policy aims for a competitive economy in a healthy environment, gender equity and international cooperation.SEDG is a normative scenario with stabilisation of GHG emissions.
GRAS-CUT (Cooling Under Th ermohaline collapse): Deregulation, free trade, growth and globalisation are policy objectives (as for GRAS) before a climate shock  (the collapse of the thermohaline circulation) in 2050.Alternative economic and environmental policies are then introduced in reaction to this shock.
BAMBU-SEL (Shock in Energy price Level): High prices for energy and high price volatility is to be expected, and absolute scarcities may occur in the near future -2015.Alternative economic and environmental policies are then introduced in reaction to this shock.
BAMBU-CANE (ContAgious Natural Epidemic): A global pandemic in the near future -2015, causes changes in population numbers, distribution and behaviour, with subsequent social and political implications.Alternative economic and environmental policies are introduced in reaction to this shock.
For all scenarios the impacts on biodiversity (species groups and ecosystems in diff erent biomes) have been explored by deliberation methods, with the ALARM scientists serving as the expert base (Marion et al. 2010)

Observed and projected climate change in Europe
Th e Intergovernmental Panel on Climate Change (IPCC) estimates a 0.07±0.02°Cper decade increase in global surface temperatures over the last 100 years (IPCC 2007).Temperature reconstructions present strong evidence that this magnitude has been the largest over the last 1000 years (Folland et al. 2001).Furthermore, the 1990s are likely to have been the warmest decade of the last millennium (Folland et al. 2001) with a continuation of the trend until 2009 (Jones and Moberg 2003 and updates2 ).20 th century annual mean temperature rise in Europe was 0.08±0.03°C,thus slightly larger than the global mean with the warming being more pronounced during winter than summer (Luterbacher et al. 2004).
Precipitation patterns are spatially and temporally more heterogeneous than temperature with some regions experiencing dryer conditions while others have become wetter.Due to the large variation, signifi cant trends are generally more diffi cult to detect.Increases in temperature lead to increased water-holding capacity of the atmosphere, altering the hydrological cycle and thus also precipitation events (Treydte et al. 2006).Globally, observed annual precipitation records indicate a twentieth-century increase of about 9 mm over land areas (excluding Antarctica), although this trend is relatively small compared to the century-long variability (New et al. 2001).European trends in annual precipitation reveal a wettening in northern Europe while large parts of southern Europe show little change or drying (IPCC 2007).
Climate extremes are rare events that fall in the tails of the distribution of e.g.daily temperature or precipitation.In order to statistically detect any trends in the frequency and magnitude of extreme weather situations, longer observation time-series are required compared to changes in the mean climate.A global analysis with a large set of Insuffi cient studies for assessment indices of daily climate extremes such as a warm spell duration index, the number of frost days or the occurrence of very wet days was conducted by Alexander et al. (2006) for the period 1951-2003.Th ey found signifi cant increases in daily minimum and maximum temperatures throughout the globe and increases in precipitation extremes over many areas, although much less spatially coherent.Table 1 gives an overview of observed and projected changes in extremes and the level of confi dence.Observed changes in the frequency of temperature-related extremes show generally increases in heat events and decreases in cold events.
Th e projected warming until the end of the 21 st century is 1.1 -6.4°C in global mean annual temperature (IPCC 2007)3 .Global annual precipitation is projected to increase by 1.3 -6.8% until the period 2071-2100 according to simulations under the SRES A2 scenario (IPCC 2001).Projections for Europe show wetter conditions in northern Europe mainly during winter and drier conditions in southern Europe for the summer (Ruosteenoja et al. 2003).Still, observed increase in atmospheric CO 2 over the past decades was mostly above the mean (but within the 95% confi dence intervals) of the extreme A1FI scenario (Le Quere et al. 2009).
One of the main objectives of the ALARM project was to study the risks of climate change to biodiversity in Europe.Both historic information about climate as well as climate scenarios projecting changes into the future are needed for this.Historic climate datasets on a regular grid system developed by the Climatic Research Unit (CRU) at the University of East Anglia, UK, provide high-resolution information for key climate variables in monthly time steps throughout the 20 th century (New et al. 2002, Mitchell et al. 2003).Th e datasets consist of six variables: mean surface temperature, diurnal temperature range, precipitation, vapour pressure and cloudiness.
Coupled atmosphere-ocean global circulation models (AOGCMs) are the most sophisticated tools currently available for simulating responses of the climate system to increases in greenhouse gas concentrations.Projected changes in climate variables from AOGCMs were used to construct the core set of ALARM climate scenarios for Europe that continues the historic dataset into the 21 st century (Fronzek et al. 2010).For the scenarios, labelled GRAS, BAMBU and SEDG, narrative storylines have been developed (see previous chapter) and also other drivers of biodiversity change were quantifi ed to allow a multi-pressure assessment.
For the BAMBU basic scenario, simulations from three diff erent AOGCMs were selected covering a wide range of climate model uncertainties to represent the scenario.Simulations from one AOGCM, the HadCM3, have been used to represent climate changes in all three ALARM scenarios.Th e range of temperature and precipitation changes is summarized in Table 2.
Th e spatial pattern of simulated changes in temperature and precipitation are shown below for the GRAS scenario with the HadCM3 AOGCM.For the fi ve ALARM climate scenarios described here, this scenario gives the strongest warming by the end of the 21 st century.In this scenario, winter warming until the period 2071-2100 (relative to  shows a gradient from south-western to north-eastern Europe with the smallest increases of c. 3°C over the Iberian peninsula and the largest increases of more than 10°C in northern Finland (Figure 3, left).Summer warming is strongest in the Mediterranean countries (Figure 3, right).Th e CSIRO2 model expects a similar temperature increase by 2100 (4.6° as compared to 5.0°), whereas the NCAR-PCM model results in a lower increase (3.0°) for the same scenario.Th e pattern of stronger warming in winter in North-East Europe and in summer in Southern Europe is consistent among all fi ve climate scenarios (Fronzek et al. 2010).
Th e pattern of winter precipitation changes for the same scenario, again with the HadCM3 AOGCM, shows wetter conditions over nearly all of central and northern Europe and dryer conditions in southern Europe (Figure 4, left).Summer precipitation in this scenario decreases over large part of Europe with the only exceptions being Fennoscandia and parts of the Baltic countries (Figure 4, right).Averages for Europe shown in Table 2 do not convey these regional diff erences.Precipitation changes projected by the NCAR-PCM and CSIRO2 models show wetter conditions compared to the Had-CM3 scenario (3.8% resp.5.8% as compared to 0.1% increase averaged for Europe).A further climate scenario (labelled GRAS-CUT) explores the impacts of a sudden collapse of the North-Atlantic thermohaline circulation (THC) that would cause a major cooling over north-western Europe.

Impacts of climate change on plants
A changing climate modifi es the conditions which shape the physiological behaviour, the productivity and the ranges of many plants and thus, is expected to induce manifold reactions of climate sensitive species and ecosystems (e.g.Huntley et al. 1995, Sykes et al. 1996, Kappelle et al. 1999, Th eurillat and Guisan 2001, Th uiller et al. 2005, Pompe et al. 2008).In recent years, an increasing number of ecological "fi ngerprints" of climate change impacts (Walther et al. 2001, Parmesan and Yohe 2003, Root et al. 2003) provide ground-truth data of observed changes in the behaviour and distribution of plant species (Walther et al. 2010).While a European scale analysis yielded strong impacts especially on Mediterranean and high mountain plant species (Th uiller et al. 2005, Rickebusch et al. 2008), a regional analysis from Germany showed specifi c vulnerability of plant species in the North-East and South-West of Germany due to increasing droughts (Pompe et al. 2008).Using this data and assigning the species to their habitat specifi c species pool (Pompe et al. 2010) found that the species pools of tall herb communities, bushes, and turfs near or above the treeline were most sensitive, followed by dwarf shrub communities below alpine areas.Th e species assigned to forb communities, forest grassland ecotones and tall herb slopes outside fl oodplains, plant cultures, and urban, commercial, and industrial areas were least negatively impacted by climate change.
On the basis of long-term phenological records, trends in the response of living organisms to climatic changes can be tracked.Evidence that events in spring have been happening earlier in recent decades arises from a wide range of species and across a wide range of geographic locations.Despite some inconsistencies in the numeric values of the data, an overall trend of 2.3 days per decade towards an earlier onset of spring has been documented (Parmesan and Yohe 2003).Fewer phenological data are available for the fall season.However, the few data sets that include phenophases in both spring and autumn reveal a trend towards a prolongation at both ends of the season and thus, an extension of the growing season (Walther 2004).Th e observed lengthening of the growing season is based on terrestrial phenological data records with satellite observations of leaf area index anomalies over the past two decades (e.g.Lucht et al. 2002).
In addition to phenological changes, climate warming is also expected to shift the margins of species ranges or boundaries of biomes (e.g.Huntley et al. 1995, Sykes et al. 1996).Evidence for species range shifts has been reported from various habitats (Walther et al. 2010).Th e period of milder winter conditions since the 1970s for example is in temporal synchrony with a major phase of spread and establishment of thermophilous evergreen broad-leaved species on sites with former deciduous forest vegetation south of the Alps (Walther et al. 2002, Walther andBerger 2010).
In analogy, the partial replacement of neighbouring altitudinal belts is reported by Penuelas and Boada (2003) from north-eastern Spain.Th is upward shift of vegetation belts is ascribed to the rising annual temperature of 1.2-1.4°C during the last 50 years with the main increase in the last 30 years.In the Arctic, Cornelissen et al. (2001) suggest a climate-induced change in species composition of arctic plant communities with declining macrolichen abundance as a consequence of the increased abundance of vascular plants.An analogue process of increasing species number and frequency is found at the altitudinal margin of plant life.In the Alps, e.g.Hofer (1992) and Grabherr et al. (1994) provide data on increasing species abundance and richness of plants on high mountain tops showing the overall trend of an upward shift of the alpine-nival fl ora, which is attributed to the observed warming in climate in these areas.A recent update of the fl ora of high mountain peaks in the Swiss Alps based on the Hofer (1992) revealed that the trend of increasing species numbers in the summit areas continues and might even have been accelerated in the last decade (Walther et al. 2005b).
Th e biotic response to thirty years of enhanced global warming has become perceptible and substantial.An overwhelming number of studies provide evidence for climate change impacts on species, communities and ecosystems (Hughes 2000, Mc-Carty 2001, Walther et al. 2002, Root et al. 2003, Parmesan 2006, Walther 2010; for plants see also Walther 2004).In the long-term perspective, the biotic implications of climate change and its evolutionary consequences depend on both the magnitude and rate of global warming as well as on the development of other human infl uences on biological systems such as habitat conversion, overexploitation and pollution (e.g.Leemans 2001, Travis 2003).It is the combination of these infl uences that also determines the full extent of the impact of climate change on plants.
Of particular relevance for dragon-and damselfl ies are studies on climate change impacts on aquatic plants, as these are the core resources for oviposition.Heikkinen et al. (2009) present the northern spread of the invasive aquatic plant Elodea canadensis in Europe, which is fostered by climate change eff ects.Th e expansion of this plant may favour the expansion of damsel-and dragonfl ies showing endophytic oviposition, and in fact in Scandinavian countries Calopteryx species show an expansion to the north (Ott 2010b).
On the other hand, the decrease of the water soldier (Stratiotes aloides) in northern Germany in recent years, which seems to be a combined eff ect of climate change and eutrophication, has an immediate impact on the populations of the endangered and protected Green Darter (Aeshna viridis), a dragonfl y which lays its eggs only into this plant.

Impacts of climate change on animals
As for plants there is already strong scientifi c evidence of the impact of climate change on animals in Europe.During the 21st century rapidly shifting climate zones and rising sea levels will put increasing pressure on species already under threat for other reasons.Among the many examples of climate change eff ects are: i) phenological changes such as earlier fi rst appearances of British butterfl ies in the summer (Roy and Sparks 2000) or general changes in the phenology for dragonfl ies (Ott 2001), ii) northward expansion of many species (Parmesan and Yohe 2003), in particular for Mediterranean dragonfl ies (Ott 2001(Ott , 2010a(Ott , 2010b)), iii) spreading of sea shell animals (e.g. the barnacle Balanus perforatus ) from warmer seas around SW England 100 km eastwards up the Channel (Hiscock et al. 2004), iv) overwintering of migratory water birds from the Arctic along the North Sea coast rather than the milder western seaboard of Britain (Robinson et al., 2005), v) microevolutionary adaptations such as diet expansion of a butterfl y in response to climate change (Th omas et al. 2001), vi) local extinction of low elevation butterfl ies in the southern parts of their geographical range (Hill et al. 2002), and (vii) changes in the structure of local bird and butterfl y communities (Devictor et al. 2008, Van Swaay et al. 2010), as well as for dragonfl y communities (Ott 2007b).
One of several studies from within the ALARM project (Araujo et al 2006) has shown that projected climate change could trigger massive range contractions among amphibian and reptile species in the southwest of Europe.Th e authors projected distributions of 42 amphibian and 66 reptile species 20-50 years into the future under 4 emission scenarios proposed by the Intergovernmental Panel on Climate Change and 3 diff erent climate models (HadCM3, CGCM2, and CSIRO2).Th e researchers found that increases in temperature are not likely to constitute a major threat to amphibian and reptile species in Europe.Indeed, a global cooling scenario would be much worse.However, increases in aridity could trigger contractions in the distributions of nearly all species occurring in the southwest of Europe, including Portugal, Spain and France.Impacts in these three countries are not trivial because, together, they hold 62% of the amphibian and reptile species present in Europe.Th e high proportion of amphibian and reptile species occurring in these three countries is due to the key role played by the Iberian Peninsula as refugia against extinctions during past glacial periods.With projected climate changes these hotpots of persistence might be at risk of becoming hotspots of extinction (see Araujo et al. 2006 for further details).
Just as amphibians, also dragonfl ies will of course show similar reactions and serve as an "indicator", as well as a "victim": higher temperatures during summertime and warmer winters (see chapter on climate change scenarios) will lead to many biological eff ects (e.g.change in the phenology, trend to an increased expansion, invasion of southern species, elimination of cold stenotherm species; see Ott 2001), weather extremes at a local or regional scale will lead to droughts in certain biotopes and eliminate all species of a water body or region or alter the communities (see Ott 2010b) and extreme storms can lead to long distance drifts of individuals and new areas may be readily occupied by a species.
In general all changes of the distribution of amphibians will alter also the dragonfl y communities, as amphibians and dragonfl ies are prey and predators at the same time: dragonfl y larvae prey on tadpoles and adult frogs prey on damsel-and dragonfl ies (Ott 2001).
In contrast, impacts of climate change on butterfl ies are projected to be more severe.Under an extreme GRAS scenario (climate corresponds to IPCC SRES A1FI) over 95 per cent of the present land occupied by 70 diff erent butterfl ies would become too warm for continued survival.Th e best case SEDG scenario (A2) sees 50 per cent of the land occupied by 147 diff erent butterfl ies would become too warm for them to continue to exist there.Many butterfl ies will largely disappear from where they are regularly seen now (Settele et al. 2008(Settele et al. , 2009)).
Butterfl ies are a typical prey for damsel-and dragonfl ies, but a decrease of butterfl ies will probably have only little eff ect on them, as butterfl ies will not in general be eliminated and dragonfl ies also can easily switch to other prey (e.g.Diptera).
In general, looking to the future, wild plants and animals will go extinct in some places unless they can keep pace with the rapidly changing climate.While some mobile species can do this, other, less mobile and stenoecious species, will fi nd it much more diffi cult.Th ere is also a concern that biodiversity may be aff ected in multiple ways because of other responses to climate change, such as increased demand for water, leading to drying out of rivers and wetlands.
Th e reactions of single species can have severe cascading eff ects on higher organisational levels of biodiversity.Since single species react individualistically to climate change, this will ultimately lead to the generation of novel communities (Walther 2010).Th ese novel communities will be characterised by the disruption of currently existing species interactions and the potential of new interactions (Schweiger et al. 2010a).During the ALARM project Schweiger et al. (2008) showed that the future overlaps of climatically suitable areas of the monophagous butterfl y Boloria titania and its larval host plant Polygonum bistorta may virtually disappear from where they presently co-exist and allow co-occurrence only in distantly located areas, whereby the ability to colonise these areas is questionable for both plant and butterfl y.In a follow up study Schweiger et al. (2010b) showed that such mismatches are not the case for every species.In fact most butterfl y species are supposed to be not limited by the distribution of their host plants and thus future mismatches are not an issue for them.However, there are several species which are to a certain extent limited by their host plants and for them future mismatches are indeed a serious issue.Of particular concern in this context are species that utilise range limited host plants.
Such mismatches of interacting species are not restricted to pairwise interactions but can expand to whole interaction networks as has been reviewed by Schweiger et al. (2010a).Although, the architecture of such networks and their redundancy and fl exibility might impede cascading extinctions (Hegland et al. 2009, Memmot et al. 2004, Vilà et al. 2009), such buff er capacities are not unlimited and will not necessarily circumvent severe changes in species interactions and the consequent species extinctions (Memmott et al. 2004, Fortuna andBascompte 2006).Further, changes in community composition and species interactions can, especially in combination with additional pressures (see below) lead to severe consequences for ecosystem services (Potts et al. 2010).Th is is in particular true for wetland ecosystems, where dragonfl ies are excellent indicators for the eff ects of climatic changes (Ott 2001(Ott , 2008b(Ott , 2010b)).
Dragonfl ies only have a limited dependency on plants, but very much on waters, its quality and quantity.Alterations of the water level lead to changes of the dragonfl y community (e.g.favour eurycious species and eliminate mooreland species, see Ott 2007bOtt , 2010b) ) or could lead to invasions by Mediterranean species which previously did not inhabit these water bodies (e.g. by Lestes barbarus or Ischnura pumilio, see: Ott 2006Ott , 2008a)).If the water level recovers due to an increased precipitation the old dragonfl y communities may recover as well and the new species may leave again, but it also could lead to an irreversible change and new aquatic communities (Ott 2010b, unpubl. data).

Multiple risks for biodiversity: Climate change in interaction with other pressures
In addition to climate change, global change creates many drivers that aff ect biodiversity (e.g., Potts et al. 2010, Schweiger et al. 2010).Among the most important drivers are land-use change with the consequent loss and fragmentation of habitats (Westphal et al. 2003, Tscharntke et al. 2005, Schweiger et al. 2007, Öckinger et al. 2010); increasing pesticide application and environmental pollution (Rortais et al. 2005, Dormann et al. 2007); alien species (Stout and Morales 2009, Walther et al. 2009, Vilá et al. 2010); and the spread of pathogens (e.g., Cox-Foster et al. 2007).Th ese drivers are often in confl ict with desired ecosystem services.Sustaining pollination services, for instance, is for sure highly desired by both conservationist and farmers, but it is often decreased as a consequence of other demands such as increasing agricultural production.Habitat loss and fragmentation are generally thought to be the most important factors driving pollinator declines (Brown and Paxton 2009).In addition, increased use of insecticides can cause pollinator mortality by direct intoxication (Alston et al. 2007).Increased herbicide and fertiliser use can aff ect pollinators indirectly by decreasing fl oral resource availability (Gabriel andTscharntke 2007, Holzschuh et al. 2008).
All these drivers act simultaneously and very likely synergistically on local communities (Tylianakis et al., 2008).So far, most studies have analysed specifi c drivers in isolation, and therefore evidence of interactive eff ects is scant.However, in a recent review within ALARM Schweiger et al. (2010) show that the eff ects of multiple interacting pressures can be contrasting.In the face of climate change, alien species can serve as additional pollen and nectar sources (Stout and Morales 2009) or pollinators (Goulson 2003).Such species can thus substitute otherwise lost functions.On the other hand, alien species can lead to reduced reproductive success and population declines of native pollinators by competitive displacement of native plants (Traveset and Richardson 2006) or by high levels of resource competition among native and alien pollinators (Matsumura et al. 2004, Th omson 2006).
Yet, knowledge about the relative contribution and the importance of interactive pressures is an indispensible precondition to understand current and to predict future changes in biodiversity and resulting ecosystem services.

New developments in relation to dragon-and damselflies
In the course of the project, ALARM was enlarged and partners from other continents have been included.With this expansion the ALARM approach had to be tailored for the respective regions, which in the context of dragon-and damselfl ies was particularly the case for research in Asia and Africa.
For Asian rice-growing systems a close collaboration with IRRI was started to analyse long-term trends in the biodiversity of natural enemies of rice pests, with a particular focus on parasitoids (by applying and adjusting the ALARM fi eld site network approach, compare Grabaum et al. 2006).A further aim was to look into the options to develop appropriate sustainability indicators for rice growing systems, where in particular dragon-and damselfl ies might play a key role and where we have a direct fi eld for the further application of the research results (Heong et al. 2010).
In southern Africa, there are many narrow range endemics that are at risk from the eff ects of global climate change.Among these are Colophon beetles (Samways 2005) and certain dragonfl ies.One species of dragonfl y, only discovered in 2003, is Syncordulia serendipitor, which lives in primary high elevation streams (Samways 2008).With global change, it appears to have nowhere to go.However, genetic work has indicated that this species diverged 60 million years ago (Ware et al. 2009), and it has seemingly survived climate changes in the past, some of which have been very rapid.Th is leads to the speculation that perhaps even some narrow range endemics are physiologically adapted to climate change.But then perhaps in the past the overall population size and hence genetic variation, was greater, enabling the species to survive climatically diffi cult times.Th ere is some evidence for this among some other, more widespread, odonate species in southern Africa.Th ere is remarkable elevational tolerance among some species, enabling them to survive at higher or lower elevations according to the prevailing climatic conditions (Niba and Samways 2006), even in the case of some endemic species (Samways and Niba 2010).Other species show great plasticity in their ability to expand their geographic ranges and colonize water bodies during wet phases of El Niño cycles, then shrinking back to predictably wet refugia in the dry phases (Samways 2010).As this is a common phenomenon, there appears to have been strong selection pressure on a whole range of species to survive stressful climatic conditions.Th ese shifts in population presence can be so extreme, that in the case of one species, Aciagrion congoense, it appeared as a new national record to South Africa in 2000, apparently driven south by fl oods in Mozambique.It then became the dominant damselfl y at iSimangaliso Wetland Park in 2001.However, a few years later, after an extended dry spell, it again disappeared from South Africa.Th e point here is that at least in this part of Africa, there is some evidence that odonate species are to some extent already honed to tolerate some anthropogenic climate change.
Within ALARM the eff ects on dragonfl ies were mainly studied in Germany and Europe: here the waters in the Palatinate (Germany) and the Gran Sasso area (Italy) were studied.It could be shown that the general trends for dragonfl ies (e.g.range expansion of Mediterranean species, alteration in the phenology, changes in the aquatic communities, decrease of stenoecious species, see Ott 2001Ott , 2008b)), which have already shown earlier, still continued and the expansion is an ongoing and European wide process (Ott 2010a(Ott , 2010b)).In general biodioversity will increase, but in the medium term there will probably be a decrease, as stenoecious species, such as alpine and mooreland species, will decrease due to the negative eff ects on their biotopes.In the Mediterranean the lack of water, in particular in intensively used areas, will lead to the decrease and eventual extinction of many species (e.g.species of running waters).

T able 2 .Fi gure 3 .
Simulated changes in annual mean temperature (°C) and annual precipitation by 2071-2100 relative to 1961-1990 averaged over Europe for the ALARM scenarios.Change in air temperature (in °C) between the periods 1961-1990 and 2071-2100 in winter (December-February, left) and summer (June-August, right) for the GRAS scenario using the HadCM3 AOGCM with the A1FI emission scenario (taken fromFronzek et al. 2010).

Figure 4 .
Figure 4. Relative change in precipitation (in %) between the periods 1961-1990 and 2071-2100 in winter (December-February, left) and summer (June-August, right) for the GRAS scenario using the HadCM3 AOGCM with the A1FI emission scenario (taken from Fronzek et al. 2010).

Table 1 .
Change in extremes for meteorological phenomena over the specifi ed region and period, with the level of confi dence (Source: IPCC, 2007).In the IPCC terminology, "very likely" expresses a 90−99% chance and "likely" a 66−90% chance.