• Europe;
  • global warming;
  • habitats directive;
  • insect conservation;
  • range shift;
  • SDM;
  • species distribution model;
  • species range;
  • species-specific dispersal ability


  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

Abstract.  1. The effects of climate change on the distribution of species are typically inferred using bioclimatic envelope models, assuming either no or unrestricted dispersal abilities. Information on species-specific dispersal abilities, especially of animals, is rarely incorporated.

2. We analysed European records of two damselflies and four dragonflies protected by the Habitats Directive of the European Union. In addition to no or unrestricted dispersal scenarios, we considered species-specific dispersal distances based on literature information to improve realism in assessing conservation implications of climate change. The climate model HadCM3 and the emission scenario A2 were applied to project potential changes in occurrence probabilities up to 2035. As modelling algorithms, generalised linear models (GLM) and boosted regression trees (BRT) were used.

3. The species Coenagrion ornatum, Coenagrion mercuriale and Ophiogomphus cecilia are projected to lose range (up to −68%) when incorporating specific dispersal distances, while they are projected to extend their range (up to +23%) in the unrestricted dispersal scenario. Furthermore, suitable climatic conditions tend to decline for Leucorrhinia albifrons and Leucorrhina caudalis (up to −73%), whereas Leucorrhinia pectoralis is projected to gain distribution area (up to +37%) assuming either species-specific or unrestricted dispersal and subsequently successful breeding. Cross-validated model performance (AUC values) ranges between 0.77 and 0.92.

4. The integration of species-specific knowledge about dispersal distances in species distribution models promises to improve estimates of potential range changes and their implications for conservation management. Contrasting model results under different dispersal scenarios highlight the importance of research on species’ ecology including dispersal distances.


  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

Climate change is a driving factor for species range shifts (Walther et al., 2005; Hickling et al., 2006; Hitch & Leberg, 2007; Ott, 2009). Such range changes are of major concern for nature conservation, especially because endangered and/or rare species are expected to be most vulnerable to changes in climatic conditions and may be most threatened by extinction (Schlumprecht et al., 2010). To derive effective adaptation strategies in nature conservation, it is important to assess potential influences of climate change on species ranges. Realistic projections will require assessments of both future habitat suitability and species-specific dispersal restrictions.

Climate envelope models correlate species’ occurrences with environmental variables (Guisan & Thuiller, 2005; Elith & Leathwick, 2009). The resulting climate envelopes can be used as a basis for projections of suitable future habitats of a species, which form the basis for the evaluation of potential range changes (Araújo & Guisan, 2006).

The reliability of model output depends on the selection of explanatory variables, the choice of the climate model, emission scenario and modelling algorithm (Dormann et al., 2008). Nevertheless, climate envelopes are a useful first approach to estimate potential effects of climate change on species’ distributions.

Assuming the two extreme options of ‘no dispersal’ and ‘full dispersal’ is the state-of-the-art approach to model future occurrence probabilities of species (Coetzee et al., 2009; Franklin, 2010; Fitzpatrick et al., 2011). This provides lower and upper boundaries on expected future range sizes: A ‘no dispersal’ scenario will overestimate potential losses in range size, whereas a ‘full dispersal’ scenario neglects dispersal barriers and tends to overestimate species’ dispersal and thus adaptation abilities. While both dispersal scenarios provide hints on where current suitable area might be lost and where future suitable area might be found, they give limited guidance on where species ranges may realistically shift. The integration of species-specific dispersal distances may contribute to overcome this limitation (Buse & Griebeler, 2011). However, fully integrating such biological traits into modelling requires an explicit knowledge on species-specific behaviour, stress tolerance, life cycles, vitality, activity periods and dispersal capacity.

Odonata are prominent indicator species for the biological effects of climate change (Ott, 2010). They are influenced by climate change in many ways, covering aspects of life history, thermoregulation, ecology, habitat and evolution (Hassall & Thompson, 2008). Hickling et al. (2005) provide evidence for northward range shifts of several British Anisoptera and Zygoptera species as a response to climate warming. Braune et al. (2008) analysed the voltinism flexibility along a thermal gradient for Gomphus vulgatissimus. They developed a population dynamic model allowing projections for future climate change. As their field results indicate a decreasing voltinism from warm (southern Europe) to cold (northern Europe), the model projected an increased development speed in the northern part, a range expansion at the northern range margin and an extended flight period under a warming scenario. Also, climate change induced shifts in community composition and species abundance could be observed (Flenner & Sahlén, 2008). Most of the considered Odonata included in this study have a lifespan of at least 2 years as larvae (Petersen et al., 2003; Corbet et al., 2006) and therefore highly depend on habitat conditions, for example, water temperature, oxygen content and the availability of freshwater pools (e.g. Sternberg & Buchwald, 1999, 2000). On the other hand, imagines of many species are highly mobile and thus respond rather directly to a shifting climate space. However, they are active only for a few months, which limits the temporal window for dispersal processes. Furthermore, especially the endangered dragonflies and damselflies tend to show restricted mobility. Such limitations in dispersal abilities can be related to various factors such as morphological constraints or close dependence on specific habitat conditions (Thompson et al., 2003).

Here, we analysed six strictly protected odonate species to assess how climate change may influence their future distributions. We assumed that considering species-specific dispersal abilities can lead to contrasting results in projected future range changes with regard to no and full dispersal and that the integration of dispersal distances (beside climate) in species distribution modelling enhances realism of model results. Finally, we discuss potential management options for protecting these species under future climate conditions.


  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information


We selected six Odonata (out of 16 odonate species listed in the EU Habitats Directive) for which observed dispersal distances are available in the literature. All these species are listed in Annex II and/or IV of the EU Habitats Directive and in the European Red List (Kalkman et al., 2010) and are therefore under special protection. Two species belong to the Zygoptera: Coenagrion mercuriale (Charpentier, 1840) (Annex II) and Coenagrion ornatum (Sélys, 1850) (Annex II). For the Annex II species, the member states have to designate ‘Special Areas of Conservation’. Both species develop in lotic waters with a moderate or slow flow velocity (Sternberg, 1999; Sternberg et al., 1999). The habitat requirements of both species are very similar. They can be found at sunny streams and springs rich in aquatic and riparian vegetation, often with a calcareous substrate. Flowing drainage ditches may also offer suitable habitats (Sternberg, 1999; Sternberg et al., 1999). Coenagrion mercuriale is distributed in South West Europe and partly in Central Europe. Coenagrion ornatum is common in South East Europe and very local in Central Europe.

The four other species are Anisoptera: Leucorrhinia albifrons (Burmeister, 1839) (Annex IV), Leucorrhinia caudalis (Charpentier, 1840) (Annex IV), Leucorrhinia pectoralis (Charpentier, 1825) (Annex II and IV) and Ophiogomphus cecilia (Geoffrey in Fourcroy, 1785) (Annex II and IV). For the species listed in Annex IV, a strict protection is required. Except for O. cecilia, the larvae of all these Anisoptera species inhabit lentic waters (e.g. Corbet et al., 2006). The three Leucorrhinia species prefer oligotrophic to mesotrophic lakes and shallow waters, often located in forests (Dijkstra & Lewington, 2006). Their distribution ranges over Eastern and Central Europe. Ophiogomphus cecilia is the only representative of this genus in Europe and widespread in most of Eastern Europe up to Germany, with isolated populations in France and Italy. Preferred habitats are large to mid lowland and small highland rivers with a sandy substrate (Sternberg et al., 2000).

Species and climate data

Information on current species distributions was taken from the European Environment Information and Observation Network (EIONET) Central Data Repository server (EIONET, 2009). The data originate from the European reporting of the year 2007 pursuant to Article 17 of the Habitats Directive. They are available for 25 EU countries in different spatial resolutions. All of those member states are committed to report the current occurrences and the conservation status of the listed species in a 6-year interval. Data from non-EU countries (Switzerland, Balkan region, Norway) were added; For Switzerland, we used the database of the Swiss Biological Records Center (; public access). Balkan data were taken from Boudot et al. (2009) and data for Norway from Olsvik and Dolmen (1992). These data were digitised and geo-referenced in ArcGIS 9.3.1 (ESRI, Redlands, CA, USA).

Current and projected future European climate was quantified on a 10′ (arcminutes) grid from interpolated observed and future simulated climate data (Mitchell et al., 2004). Future projections were based on the intermediate BAMBU (‘Business As Might Be Usual’, A2) scenario (Spangenberg, 2007), developed for the European project ALARM (Settele et al., 2005). The future projection is driven by the global HadCM3 climate model (Hadley Centre Coupled Model, version 3, Hadley Centre, UK) and covers the period 2021–50 (2035). The emission scenario A2 assumes a temperature increase of 3.4 °C up to 2100 based on a high global population growth, and a slow economic development and technological change (IPCC, 2007).

The following climatic variables were used in the modelling process, each with monthly, mean, minimum and maximum values: cloudiness (CLD, %), equilibrium evapotranspiration (EET, mm), precipitation (PRE, mm), temperature (TMP, °C), diurnal temperature range (DTR, °C), minimum temperature (TMN, °C), maximum temperature (TMX, °C) and growing degree days above 5 °C (GDD, degree days).

In this study, we decided to exclusively use climatic variables, leaving out other potentially relevant factors such as elevation and land cover. Initially, we carried out an analysis (with hierarchical partitioning) testing which factor (climate, elevation, land cover) explains most of the current distribution for each of the six species. For all tested species, elevation plays only a minor role. For three of the six species, climate is most important. For the other three species (L. albifrons, L. caudalis and O. cecilia), land cover (in these cases forest) is the most or the secondary important factor, followed by climate. However, the problem with land cover is its coarse resolution and its constraint predictability. On a European scale, we cannot yet distinguish between, for example, types of forests and have to work with classifications like ‘urban’, ‘crop’, ‘grassland’ and ‘forest’, which are difficult to interpret in the present context. Another problem is the future projection of land cover. Although some scenarios exist, the future development is not only influenced by climate but also by political developments, making these scenarios highly uncertain. For these reasons, we left land cover out of the analysis.

Dispersal scenarios

Although species distribution models assume that species’ range margins are in equilibrium with environmental variables, current ranges are in a state of flux. To counter this problem, a consideration of realistic dispersal abilities is required. We applied three dispersal scenarios: the conventional ‘no dispersal’ and ‘full dispersal’ scenarios to detect sources of potential extinction and to identify future climatically suitable areas, and a scenario which accounts for the species-specific dispersal distances.

To account for species-specific dispersal abilities, observed maximum dispersal distances of the six Odonata were used. Dispersal lags caused by larval development were considered by allowing dispersal only after the completion of the development cycle. The dispersal distances were taken from the literature (Table 1).

Table 1. Applied maximum dispersal distances until 2035 derived from literature information for each species. The maximum reachable distances take into account the duration of larval development by allowing dispersal every 2 or 3 years (after completion of the life cycle). For species with time spans in larval development (e.g. 2–3 years), we used the more likely value given by the literature.
SpeciesObserved dispersal distances (km/a)Used dispersal distances (km/a)Used larval development (years)ReferencesMaximum distance 2007–2035 (km)
Coenagrion mercuriale Up to 112 Thompson et al. (2003), Corbet et al. (2006)14
Coenagrion ornatum 200 m up to several12 Burbach et al. (1996) 14
Leucorrhinia albifrons Up to 18182 Mauersberger (2003a) 252
Leucorrhinia caudalis Up to 772 Mauersberger (2003b), Corbet et al. (2006)98
Leucorrhinia pectoralis Up to 27272 Wildermuth (1993), Corbet et al. (2006)378
Ophiogomphus cecilia Up to 10103 Suhling et al. (2003), Corbet et al. (2006)93

To take species-specific dispersal abilities throughout Europe into account, we used the Euclidean Distance, calculated through the maximum dispersal distance divided by developmental time and multiplied by the number of considered years. Based on this, a buffer zone around each current occurrence point was calculated. This allows restricting the potential distance of movement in a given time frame. By clipping the projected future full dispersal distribution and the calculated buffer zone, we got the projected suitable and accessible ranges for the six species. This was implemented with ArcGIS using the ‘Euclidean Distance’ function of the ‘Spatial Analyst Tools’.

Species distribution modelling

We used two different modelling algorithms, namely generalised linear models (GLM) and boosted regression trees (BRT) (see Elith et al., 2008 for details) to assess the uncertainty in these model decisions relative to other uncertainties in the modelling process (Dormann et al., 2008). For both model algorithms, we first dealt with collinearity in the predictors by selecting a variable set where pairwise Pearson correlations are < 0.7. In pairs of correlated variables, we retain that variable with higher univariate predictive ability (assessed by GLM with a quadratic term) of the species’ distribution. Subsequently, a stepwise selection in the GLM model of the retaining variables was based on Bayesian information criterion (BIC), no variable selection was performed for BRT models.

The results were validated with a 32-fold geographically stratified cross-validation, separating Europe in 32 equally sized parts. We used the AUC (area under the receiver operating characteristic curve) as model performance criterion to measure overall model discrimination (Swets, 1988), that is, the model’s ability to differentiate between locations where the species occurs from those were it is absent. In addition, we used the slope of the calibration curve to measure model calibration, that is, the correspondence of predicted occurrence probabilities to observed occurrence frequencies (Reineking & Schröder, 2006). The cut-off point for occurrence and non-occurrence projections was selected such that the resulting prevalence (i.e. fraction of occupied sites) equalled the mean predicted occurrence probability.

All analyses were performed with R 2.10.0 (R Development Core Team, 2010). In addition to the standard R packages, we used the PresenceAbsence package version 1.1.4 (Freeman, 2007). Model performance was quantified with val.prob from the Design package version 2.3-0 (Harrell, 2009). The gbm package version 1.6-3.1. (Ridgeway, 2010) was used for the BRT. Spatial climate and species distribution data were processed with ArcGIS 9.3.1.


  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

Climate change is projected to strongly affect the investigated Odonata. All modelled species are projected to lose more than 50% of their climatically suitable area with both modelling algorithms under the assumption of no dispersal (Table 2). Leucorrhinia albifrons and Leucorrhinia caudalis are also projected to lose at least 30% of their current distribution by 2035, both with the species-specific and the full dispersal scenario. In contrast, L. pectoralis is projected to gain distribution area independent of modelling algorithm and dispersal scenario (with the exception of no dispersal). The modelling results of the two Coenagrion species (Fig. 1) and O. cecilia (Fig. 3) differ considerably between the species-specific and full dispersal scenario. These three species are projected to gain distribution area under a full dispersal scenario, but to lose distribution area under a species-specific dispersal scenario.

Table 2. Projected change in the distribution of six odonate species in Europe for 2035, using boosted regression trees (BRT) and generalised linear models (GLM) as modelling algorithms. The percentage change until 2035 is given for the three dispersal scenarios: no, species-specific, and full dispersal.
SpeciesHadCM3, A2, 2035
No dispersalSpecific dispersalFull dispersal
BRT (%)GLM (%)BRT (%)GLM (%)BRT (%)GLM (%)
Coenagrion mercuriale −71−65−52−48+7+5
Coenagrion ornatum −73−77−65−68+23+17
Leucorrhinia albifrons −64−65−39−38−35−30
Leucorrhinia caudalis −73−71−59−57−35−34
Leucorrhinia pectoralis −67−67+34+7+37+7
Ophiogomphus cecilia −58−60−16−31+8+9

Figure 1.  Current and future projected distribution of Coenagrion mercuriale and Coenagrion ornatum in Europe. Both damselflies are projected to gain distribution area under the full dispersal scenario, but to lose distribution area under the species-specific dispersal scenario (1 km/a). Occurrence thresholds: BRT: 0.32 (C. mercuriale), 0.17 (C. ornatum), GLM: 0.27 (C. mercuriale), 0.24 (C. ornatum); modelling algorithms: BRT and GLM; climate model: HadCM3; scenario: A2; AUC (test data): BRT: 0.89 (C. mercuriale), 0.79 (C. ornatum), GLM: 0.88 (C. mercuriale), 0.77 (C. ornatum).

Download figure to PowerPoint

Species show geographically differentiated responses to projected climate change (Figs 1–3). Both Coenagrion species are projected to lose most of their occurrences in Central Europe and in parts of Southern Europe, leading to a range contraction to France and Northern Spain (C. mercuriale) and to the Balkans and parts of Turkey (C. ornatum), probably caused by a higher temperature and lower PRE in the current distribution areas. Coenagrion mercuriale is mainly distributed in the Atlantic biogeographical region with further occurrences in the Continental and Mediterranean biogeographical regions. The projected decline in the Continental and Mediterranean regions can be related to the projected increasing aridity in the future in these regions. For C. mercuriale, the full dispersal scenario projects new climatically suitable area in the Czech Republic, Austria and the Balkans, causing a potential eastward shift in distribution. With the application of the specific dispersal scenario, these projected new occurrences are excluded if they are not in reach within the considered time period and the given dispersal distance, leading to a smaller expansion in the surrounding of the current occurrence. In contrast, the full dispersal scenario for C. ornatum projects new climatically suitable area in Spain and Portugal provoking a westward shift in the potential future distribution.


Figure 2.  Current and future projected distribution of Leucorrhinia albifrons and Leucorrhinia caudalis in Europe. Both dragonflies are projected to lose distribution area with all dispersal scenarios. Occurrence thresholds: BRT: 0.33 (L. albifrons), 0.27 (L. caudalis), GLM: 0.40 (L. albifrons), 0.37 (L. caudalis); modelling algorithms: BRT and GLM; climate model: HadCM3; scenario: A2; AUC (test data): BRT: 0.92 (L. albifrons), 0.84 (L. caudalis), GLM: 0.92 (L. albifrons), 0.88 (L. caudalis).

Download figure to PowerPoint


Figure 3.  Current and future projected distribution of Leucorrhinia pectoralis and Ophiogomphus cecilia in Europe. Both dragonflies are projected to gain distribution area under the full dispersal scenario, but O. cecilia is projected to lose distribution area with the species-specific dispersal scenario. Occurrence thresholds: BRT: 0.42 (L. pectoralis), 0.30 (O. cecilia), GLM: 0.19 (L. pectoralis), 0.31 (O. cecilia); modelling algorithms: BRT and GLM; climate model: HadCM3; scenario: A2; AUC (test data): BRT: 0.83 (L. pectoralis), 0.81 (O. cecilia), GLM: 0.80 (L. pectoralis), 0.77 (O. cecilia).

Download figure to PowerPoint

Two of the three Leucorrhinia species, L. albifrons and L. caudalis, are projected to lose almost all locations within their current distribution in Central Europe, Western France and in the Baltic states. The full dispersal scenario projects new suitable area in Finland, Sweden and Norway, leading to a slight range shift towards the north-east. Although these two species may be good dispersers, the climatically suitable area is projected to be reduced to such a large extent that their dispersal ability has no influence on their potential future distribution. For L. pectoralis, the full dispersal scenario projects a tendency to a range expansion towards the north-east, similar to the two other Leucorrhinia species. Projected range loses in the western (France) and southern (Turkey) parts of the current distribution could lead to a range shift. With the specific dispersal distances, L. pectoralis is projected to reach almost all of the future suitable climate area.

The fourth dragonfly, O. cecilia, is as well projected to lose range in the western parts (France) but additionally also in Denmark. The full dispersal scenario projects a range shift towards the north-east, especially to Belarus. With the application of the specific dispersal distances, a great part of these projected new occurrences would be in reach because of the species’ high dispersal ability. However, for the most northerly projected suitable areas, the species dispersal ability is insufficient.

The statistically selected climatic variables differ between the species as the applied method incorporates the current occurrence in the decision process (Figs S1–S6). Therefore, all selected variables describe the current distribution best considering a correlation of the remaining variables of < 0.7. For all modelled species, PRE is an important factor (Table 3). In the models, PRE amounts in spring and summer as well as minimum and maximum values were selected, reflecting the dependence on water availability for reproduction. All other selected climatic variables, that is, minimum temperature, GDD, DTR and cloudiness, are related to temperature. Diurnal temperature range and cloudiness are important for five of the six species. All these temperature-related variables consider the cold period of the year and therefore the diapause, which is especially relevant for the survival of the larvae.

Table 3. Model performance of the two modelling algorithms (GLM, BRT) showing the AUC and slope of the calibration curve for the test data (32-fold cross-validation), and importance of the climatic variables selected by the model.
SpeciesCalibrationAUCVariable importance
  1. TMN, minimum temperature (°C); PRE, precipitation (mm); CLD, cloudiness (%); DTR, diurnal temperature range (°C); GDD, growing degree days (°C).

Coenagrion mercuriale 0.660.640.890.88TMN minimum 26.6%PRE minimum 22.4%PRE in May 22%CLD in January 17%DTR in January 12%
Coenagrion ornatum 0.590.550.790.77DTR in October 64.2%PRE in April 35.8%   
Leucorrhinia albifrons 0.600.810.920.92CLD in November 36.5%PRE in April 24.4%PRE in August 19.7%DTR in November 19.4% 
Leucorrhinia caudalis 0.480.820.840.88CLD in November 43.5%PRE in April 29.4%PRE in August 27.1%  
Leucorrhinia pectoralis 0.540.730.830.80CLD in January 26.2%DTR in January 20.6%GDD in December 20.1%DTR in May 16.7%PRE in February 16.3%
Ophiogomphus cecilia 0.500.530.810.77CLD in November 32.8%PRE in February 18%DTR in November 17.3%PRE maximum 16.5%DTR in May 15.4%

Both modelling algorithms, GLM and BRT, perform well in predicting the current distribution of all six species (Table 3). All cross-validated AUC values are between 0.77 und 0.92, with BRT showing slightly better discriminatory performance (mean AUC values: 0.85 BRT, 0.84 GLM). Both algorithms tend to be overconfident in modelling the current occurrence (BRT more so than GLM), as indicated by the slope of the calibration curve (mean values: 0.57 BRT, 0.68 GLM; values of 1 correspond to well-calibrated models).


  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

Dispersal scenarios

Most recent modelling studies only apply two dispersal scenarios: no dispersal and unlimited dispersal (e.g. Schweiger et al., 2008; Lawler et al., 2009; Carvalho et al., 2010). In the case of nature conservation and its adaptation needs in times of climate change, these projected extremes in dispersal are informative but insufficient.

The species C. ornatum, C. mercuriale and O. cecilia illustrate the limitations of the full dispersal approach. Both modelling algorithms project more suitable space in the near future considering unrestricted dispersal ability. In contrast, when considering species-specific dispersal distances, the model projects a large loss of suitable climate space. This is because of the species’ limited dispersal ability relative to the distance to the projected future suitable climate space. For C. mercuriale, it seems that the no dispersal scenario is more likely at least for parts of Europe. For Great Britain, the applied distance of one kilometre per generation might be rather optimistic. However, maximum dispersal distances of one kilometre can be also observed there (Purse et al., 2003).

We have shown that available ecological knowledge such as observed dispersal distances can be integrated into the modelling process for animal species. We believe that this gives a more realistic projection of the potential future distribution of the studied species. A particular challenge of this approach lies in the definition of suitable dispersal scenarios.

In the few studies that account for specific dispersal in species distribution modelling, especially in plants (e.g. Dullinger et al., 2004; Brooker et al., 2007; Smolik et al., 2010), long-distance dispersal is one of the most widely considered processes. Long-distance dispersal is a rare event, but plays an important role in plant species dispersal (Nathan, 2006). It is also relevant in animal dispersal, especially for small species that can be blown away by wind over large distances (Bonte et al., 2009). A recent study on L. caudalis by Keller et al. (2010) investigated the spread of this species over the last 20 years in Switzerland. They could demonstrate long-distance colonisation at distances of 30–50 km. Such long-distance dispersal is also conceivable for all other species in our study, adding uncertainty to the observed dispersal distances and their application in species distribution modelling. However, we did not include long-distance dispersal to represent a realistic conservative instead of a realistic optimistic scenario.

Climatic suitability of a site alone and the organismic potential to reach these novel habitats are not sufficient to project in a realistic way whether species might adapt to climate change by range shifts. Keller et al. (2010) trace the observed spread of L. caudalis in Switzerland back to the recreation and restoration of ponds. Coenagrion mercuriale is also highly dependent on the habitat (Rouquette & Thompson, 2007). Beside climate and dispersal ability, the habitat requirements are limiting for all investigated species. Habitats may not be available at the new climatically suitable area and may not develop in the short term thus preventing successful breeding and colonisation. Furthermore, colonisation success depends on propagule size (e.g. Ahlroth et al., 2003). Nevertheless, for all the studied species, climate change effects are reported regarding trends in range, area and/or population [European Topic Centre on Biological Diversity (ETC/BD), 2008] pointing to a sensitivity of these to climate change. The relative importance of habitat versus climate has not been investigated yet for these species. However, indirect effects of climate change, such as desiccation of water bodies or reduced prey abundance combining habitat characteristics and climate change, play also a major role in assessing the impacts of environmental change on Odonata.

Further, other abiotic factors like elevation and land cover determine the current distribution and the future spreading potential. For example, the size of fragmented patches of suitable habitats can influence the dispersal distance of a species (Ahlroth et al., 2010). However, the problem with land cover is its presently coarse classification, especially on a continental scale, and its constrained predictability.

Research gaps and uncertainties

The estimation of dispersal distances contains several uncertainties. First, observed maximum distances can be highly unrepresentative. In addition, observed dispersal distances of populations (e.g. assessed by mark-release-recapture studies) do not necessarily represent the dispersal ability of the species, but may reflect regional characteristics or methodological constraints, and therefore underestimate the real dispersal ability. Next, dispersal abilities and dispersal distances may change over time because of climate change. Alterations in environmental conditions can force adaptation processes leading directly to higher mobility (Hill et al., 1999) and increased dispersal distances (Hill et al., 2011), or indirectly by improving a species’ fitness and thereby its ability to spread. Hill et al. (1999) studied morphological traits of a butterfly from newly colonised sites. They observed individuals with larger adult live mass, larger thoraxes and lower wing aspect ratios compared to reference sites with established populations. Similarly, morphological changes over short periods have been observed for Odonata, in the form of changes in wing–abdomen length ratio and aspect ratio (Hassall et al., 2009). Alternatively, range expansions can decrease the predator or parasite pressure (Menéndez et al., 2008) and thereby increase realised dispersal distances. However, there is some evidence that infection by, for example, parasites may increase dispersal distance in damselflies (see Suhonen et al., 2010), so that release of parasite pressure can have differential effects. Finally, climate change can also lead to dispersal inhibition, as shown for the common lizard (Massot et al., 2008).

Similar problems can be suggested for the larval development time. Depending on latitude, the larval development can be longer or shorter. This is hardly to cover in species distribution models as the climatic information on large scales often provides only monthly values even though daily values are needed. It can be further suggested that climate change will influence the larval development time (Richter et al., 2008).

Another source of uncertainty relates to effects of winter warming on the diapause. Winter warming passing certain diapause-inducing temperature thresholds can prevent the beginning of the diapause (Hassall & Thompson, 2008) or increase the metabolic rate during this stage (Irwin & Lee, 2000). This and the fact that there are not enough food resources to compensate the energy deficit can lead to higher winter mortality and an increased extinction risk.

Although observations on dispersal distances already exist for some species, for most species, the real dispersal ability is unknown, limiting the applicability of species-specific dispersal approaches. Allouche et al. (2008) provide alternative methods to incorporate distance constraints in species distribution models beside observed dispersal distances. These methods calculate the occurrence likelihood at a site based on the geographical locations of known occurrences. However, this approach is just another estimation of dispersal distance with its corresponding uncertainties. Hence, the improvement of existing and the development of new methods to estimate dispersal distances is required. Testing for correlations combining geographic range sizes with species-specific traits such as morphology and dispersal abilities is one option (Boehning-Gaese et al., 2006). Field studies, like mark-release-recapture, can also provide information on dispersal distances. Further, other factors like Allee effects should be considered as these can influence the dispersal ability of a species (e.g. Veit & Lewis, 1996). In addition, modelling studies can be helpful tools to estimate dispersal distances. Recent work by Cabral and Schurr (2010) applies process-based modelling approaches to estimate plant species wind dispersal. While estimates of dispersal abilities will remain uncertain, the more we know about a species’ ecology the better we can interpret model estimates of potential range changes. Such models with integrated species-specific dispersal abilities can help identifying species that may not keep up with rapid climate change. A further step to take species-specific dispersal abilities into account is to apply a cost grid (e.g. Foltête et al., 2008). Such kind of ecological filters are enabling to consider a more realistic measure of the accessibility of suitable area than merely geographic distances, based on resistance values that are assigned to specific spatial parameters, such as landscape units.

All species distribution modelling approaches are influenced by the quantity and quality of occurrence data (Bittner et al., 2011). The spatial resolution of the distribution data of the 25 EU states (Article 17 Habitats Directive) differs between countries and provides only data for EU member states at the time of the reporting obligation in 2007. Non-EU countries, such as Switzerland, Norway and the Balkan States, are not represented in the Habitats Directive but hold a certain part of the European distribution of listed species (especially the Balkan States). Leaving these occurrences out of consideration may distort the species distribution model, but the availability of such data (if they exist at all) is often limited. Therefore, the database of the species listed in the Habitats Directive, covering the European Union, provides a substantial and valuable source of distribution data in Europe. Nevertheless, a higher resolution of occurrence and distribution data (Seo et al., 2009) as well as homogenous reporting of all countries in the next reporting obligation in 2013 would improve the basis for estimating effects of environmental change on species distributions.

Implications for nature conservation

An analysis of species and their habitats concerning their vulnerability to climate change is a first step. Such an assessment gives insights into potential future threats and highlights future conservation needs. In spite of model uncertainties, nature conservation practice needs more specific information on expected impacts of climate change on protected species and habitats for developing adaptation strategies. More ‘realistic’ model projections of future occurrences integrating species-specific traits, like dispersal abilities, can provide decision support for nature conservation (Franklin, 2010). These projections can be used to derive targeted management measures.

For species that cannot keep up with climate change, management measures have to be initiated. One opportunity would be the much-criticised assisted migration (Davidson & Simkanin, 2008; Hoegh-Guldberg et al., 2008; Ricciardi & Simberloff, 2009). Kreyling et al. (2011) are discussing the pros and cons of this technique. The pros of this concept are a reduced risk of extinction for the focal unit, a conservation of genetic diversity and its pragmatic and cost-effective implementation. On the other hand, there is a high risk of adverse effects on native species compositions; it can lead to biological homogenisation or a biased fauna and flora and poses the problem of identifying recipient localities with imperfect knowledge on ecology and climate change. For these reasons, assisted migration cannot be proposed as a suitable method without restrictions. It is an option in times of climate change, worth of consideration, needing a carefully weighting of pros and cons and the expected effectiveness.

However, not only such novel methods may be considered in times of climate change. Well-established nature conservation approaches, such as monitoring, habitat preservation, creation/extension of protected areas, retaining viable population sizes and the increase of landscape permeability (Opdam et al., 2006; Bissonette & Adair, 2008), are important instruments to support species range changes and to improve the vitality of populations. Ott (2010) emphasises the increasing need for monitoring programs that allow the detection and contemporary quantification of changes in distribution and population size.

Concluding remarks

The present study highlights the need of explicit knowledge on species dispersal ability for the purpose of modelling potential impacts of climate change. Simple modelling approaches under the assumption of no and full dispersal may indicate where to find future suitable space and where it may potentially be lost. However, the integration of specific dispersal distances in the modelling process may substantially improve assessments of expected range shifts. This is needed for the development of targeted and efficient adaptation strategies for the conservation of endangered species.


  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

We thank the Federal Agency for Nature Conservation Germany for funding this work (FKZ 3508 85 0600). We also thank two anonymous reviewers for their helpful comments on an earlier version of this manuscript.


  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information
  • Ahlroth, P., Alatalo, R.V., Holopainen, A., Kumpulainen, T. & Suhonen, J. (2003) Founder population size and number of source populations enhance colonization success in waterstriders. Oecologia, 137, 617620.
  • Ahlroth, P., Alatalo, R.V. & Suhonen, J. (2010) Reduced dispersal propensity in the wingless waterstrider Aquarius najas in a highly fragmented landscape. Oecologia, 162, 323330.
  • Allouche, O., Steinitz, O., Rotem, D., Rosenfeld, A. & Kadmon, R. (2008) Incorporating distance constraints into species distribution models. Journal of Applied Ecology, 45, 599609.
  • Araújo, M.B. & Guisan, A. (2006) Five (or so) challenges for species distribution modelling. Journal of Biogeography, 33, 16771688.
  • Bissonette, J.A. & Adair, W. (2008) Restoring habitat permeability to roaded landscapes with isometrically-scaled wildlife crossings. Biological Conservation, 141, 482488.
  • Bittner, T., Jaeschke, A., Reineking, B. & Beierkuhnlein, C. (2011) Comparing modelling approaches at two levels of biological organisation – Climate change impacts on selected Natura 2000 habitats. Journal of Vegetation Science, 22, 699710.
  • Boehning-Gaese, K., Caprano, T., van Ewijk, K. & Veith, M. (2006) Range size: disentangling current traits and phylogenetic and biogeographic factors. American Naturalist, 167, 555567.
  • Bonte, D., Clercq, N., de Zwertvaegher, I. & Lens, L. (2009) Repeatability of dispersal behaviour in a common dwarf spider: evidence for different mechanisms behind short- and long-distance dispersal. Ecological Entomology, 34, 271276.
  • Boudot, J.-P., Kalkman, V.J., Amorin, M.A., Bogdanovic, T., Rivera, A.C., Degabriele, G., Dommanget, J.-L., Ferreira, S., Garrigós, B., Jovic, M., Kotarac, M., Lopau, W., Marinov, M., Mihokovic, N., Riservato, E., Samraoui, B. & Schneider, W. (eds.) (2009) Atlas of the Odonata of the Mediterranean and North Africa. GdO, Börnsen, Germany.
  • Braune, E., Richter, O., Söndgerath, D. & Suhling, F. (2008) Voltinism flexibility of a riverine dragonfly along thermal gradients. Global Change Biology, 14, 470482.
  • Brooker, R.W., Travis, J.M.J., Clark, E.J. & Dytham, C. (2007) Modelling species’ range shifts in a changing climate: the impacts of biotic interactions, dispersal distance and the rate of climate change. Journal of Theoretical Biology, 245, 5965.
  • Burbach, K., Faltin, I., Königsdorfer, M., Krach, E. & Winterholler, M. (1996) Coenagrion ornatum (SELYS) in Bayern (Zygoptera: Coenagrionidae). Libellula, 15, 131168.
  • Buse, J. & Griebeler, E.M. (2011) Incorporating classified dispersal assumptions in predictive distribution models – A case study with grasshoppers and bush-crickets. Ecological Modelling, 222, 21302141.
  • Cabral, J.S. & Schurr, F.M. (2010) Estimating demographic models for the range dynamics of plant species. Global Ecology and Biogeography, 19, 8597.
  • Carvalho, S.B., Brito, J.C., Crespo, E.J. & Possingham, H.P. (2010) From climate change predictions to actions – conserving vulnerable animal groups in hotspots at a regional scale. Global Change Biology, 16, 32573270.
  • Coetzee, B.W.T., Robertson, M.P., Erasmus, B.F.N., van Rensburg, B.J. & Thuiller, W. (2009) Ensemble models predict Important Bird Areas in southern Africa will become less effective for conserving endemic birds under climate change. Global Ecology and Biogeography, 18, 701710.
  • Corbet, P.S., Suhling, F. & Soendgerath, D. (2006) Voltinism of Odonata: a review. International Journal of Odonatology, 9, 144.
  • Davidson, I. & Simkanin, C. (2008) Skeptical of assisted colonization. Science, 322, 10481049.
  • Dijkstra, K.-D.B. & Lewington, R. (2006) Field Guide to the Dragonflies of Britain and Europe: Including Western Turkey and North-Western Africa. British Wildlife Publishing, Dorset, UK.
  • Dormann, C.F., Purschke, O., Márquez, J.R.G., Lautenbach, S. & Schröder, B. (2008) Components of uncertainty in species distribution analysis: a case study of the Great Grey Shrike. Ecology, 89, 33713386.
  • Dullinger, S., Dirnbock, T. & Grabherr, G. (2004) Modelling climate change-driven treeline shifts: relative effects of temperature increase, dispersal and invasibility. Journal of Ecology, 92, 241252.
  • EIONET. (2009) Central Data Repository (CDR). <> 2nd December 2009.
  • Elith, J. & Leathwick, J.R. (2009) Species distribution models: ecological explanation and prediction across space and time. Annual Review of Ecology Evolution and Systematics, 40, 677697.
  • Elith, J., Leathwick, J.R. & Hastie, T. (2008) A working guide to boosted regression trees. Journal of Animal Ecology, 77, 802813.
  • European Topic Centre on Biological Diversity (ETC/BD). (2008) Article 17 Technical Report (2001–2006). Annex J - Some specific analysis on conservation status.
  • Fitzpatrick, M.C., Sanders, N.J., Ferrier, S., Longino, J.T., Weiser, M.D. & Dunn, R. (2011) Forecasting the future of biodiversity: a test of single- and multi-species models for ants in North America. Ecography, 34, 836847.
  • Flenner, I. & Sahlén, G. (2008) Dragonfly community re-organisation in boreal forest lakes: rapid species turnover driven by climate change?Insect Conservation and Diversity, 1, 169179.
  • Foltête, J.C., Berthier, K. & Cosson, J.F. (2008) Cost distance defined by a topological function of landscape. Ecological Modelling, 210, 104114.
  • Franklin, J. (2010) Moving beyond static species distribution models in support of conservation biogeography. Diversity and Distributions, 16, 321330.
  • Freeman, E. (2007) PresenceAbsence: An R Package for Presence-Absence Model Evaluation, Ogden, Utah.
  • Guisan, A. & Thuiller, W. (2005) Predicting species distribution: offering more than simple habitat models. Ecology Letters, 8, 9931009.
  • Harrell, F.E. Jr. (2009) Design: Design Package. <>
  • Hassall, C. & Thompson, D.J. (2008) The effects of environmental warming on Odonata: a review. International Journal of Odonatology, 11, 131153.
  • Hassall, C., Thompson, D. & Harvey, I. (2009) Variation in morphology between core and marginal populations of three British damselflies. Aquatic Insects, 31, 187197.
  • Hickling, R., Roy, D.B., Hill, J.K., Fox, R. & Thomas, C.D. (2006) The distributions of a wide range of taxonomic groups are expanding polewards. Global Change Biology, 12, 450455.
  • Hickling, R., Roy, D.B., Hill, J.K. & Thomas, C.D. (2005) A northward shift of range margins in British Odonata. Global Change Biology, 11, 502506.
  • Hill, J.K., Griffiths, H.M. & Thomas, C.D. (2011) Climate change and evolutionary adaptations at species’ range margins. Annual Review of Entomology, 56, 143159.
  • Hill, J.K., Thomas, C.D. & Blakeley, D.S. (1999) Evolution of flight morphology in a butterfly that has recently expanded its geographic range. Oecologia, 121, 165170.
  • Hitch, A.T. & Leberg, P.L. (2007) Breeding distributions of North American bird species moving north as a result of climate change. Conservation Biology, 21, 534539.
  • Hoegh-Guldberg, O., Hughes, L., McIntyre, S., Lindenmayer, D.B., Parmesan, C., Possingham, H.P. & Thomas, C.D. (2008) Assisted colonization and rapid climate change. Science, 321, 345346.
  • IPCC. (2007) Climate Change 2007: Synthesis Report. Contribution of Working Group I, II and III to the fourth Assessment Report of the Intergovernmental Panel on Climate Change. IPCC, Geneva, Switzerland.
  • Irwin, J.T. & Lee, R.E. Jr. (2000) Mild winter temperatures reduce survival and potential fecundity of the goldenrod gall fly, Eurosta solidaginis (Diptera: Tephritidae). Journal of Insect Physiology, 46, 655661.
  • Kalkman, V.J., Boudot, J.-P., Bernard, R., Conze, K.-J., de Knijf, G., Dyatlova, E., Ferreira, S., Jović, M., Ott, J., Riservato, E. & Sahlén, G. (2010) European Red List of Dragonflies. Publications Office of the European Union, Luxembourg.
  • Keller, D., Brodbeck, S., Flöss, I., Vonwil, G. & Holderegger, R. (2010) Ecological and genetic measurements of dispersal in a threatened dragonfly. Biological Conservation, 143, 26582663.
  • Kreyling, J., Bittner, T., Jaeschke, A., Jentsch, A., Steinbauer, M.J., Thiel, D. & Beierkuhnlein, C. (2011) Assisted colonization: a question of focal units and recipient localities. Restoration Ecology, 19, 433440.
  • Lawler, J.J., Shafer, S.L., White, D., Kareiva, P., Maurer, E.P., Blaustein, A.R. & Bartlein, P.J. (2009) Projected climate-induced faunal change in the Western Hemisphere. Ecology, 90, 588597.
  • Massot, M., Clobert, J. & Ferrière, R. (2008) Climate warming, dispersal inhibition and extinction risk. Global Change Biology, 14, 461469.
  • Mauersberger, R. (2003a) Leucorrhinia albifrons (Burmeister, 1839). Das europäische Schutzgebietssystem Natura 2000: Ökologie und Verbreitung von Arten der FFH-Richtlinie in Deutschland. Band 1: Pflanzen und Wirbellose (ed. by B. Petersen, G. Ellwanger, G. Biewald, U. Hauke, G. Ludwig, P. Pretscher, E. Schröder and A. Ssymank), pp. 574579. Landwirtschaftsverlag, Bonn - Bad Godesberg.
  • Mauersberger, R. (2003b) Leucorrhinia caudalis (Charpentier, 1840). Das europäische Schutzgebietssystem Natura 2000: Ökologie und Verbreitung von Arten der FFH-Richtlinie in Deutschland. Band 1: Pflanzen und Wirbellose (ed. by B. Petersen, G. Ellwanger, G. Biewald, U. Hauke, G. Ludwig, P. Pretscher, E. Schröder and A. Ssymank), pp. 580585. Landwirtschaftsverlag, Bonn - Bad Godesberg.
  • Menéndez, R., González-Megías, A., Lewis, O., Shaw, M.R. & Thomas, C.D. (2008) Escape from natural enemies during climate-driven range expansion: a case study. Ecological Entomology, 33, 413421.
  • Mitchell, T.D., Carter, T.R., Jones, P.D., Hulme, M. & New, M. (2004) A comprehensive set of high-resolution grids of monthly climate for Europe and the globe: the observed record (1901-2000) and 16 scenarios (2001-2100): working Paper 55.
  • Nathan, R. (2006) Long-distance dispersal of plants. Science, 313, 786788.
  • Olsvik, H. & Dolmen, D. (1992) Distribution, habitat, and conservation status of threatened Odonata in Norway. Fauna Norvegica, Series B, 39, 112.
  • Opdam, P., Steingröver, E. & van Rooij, S. (2006) Ecological networks: a spatial concept for multi-actor planning of sustainable landscapes. Landscape and Urban Planning, 75, 322332.
  • Ott, J. (2009) The Big Trek Northwards: recent Changes in the European Dragonfly Fauna. In Atlas of Biodiversity Risk (ed. by J. Settele, L. Penev, T. Georgiev, R. Grabaum, V. Grobelnik, V. Hammen, S. Klotz, M. Kotarac and I. Kühn), pp. 7879. Pensoft Publishers, Sofia, Bulgaria.
  • Ott, J. (2010) Dragonflies and climatic change - recent trends in Germany and Europe. BioRisk, 5, 253286.
  • Petersen, B., Ellwanger, G., Biewald, G., Hauke, U., Ludwig, G., Pretscher, P., Schröder, E. & Ssymank, A. (eds.) (2003) Das europäische Schutzgebietssystem Natura 2000: Ökologie und Verbreitung von Arten der FFH-Richtlinie in Deutschland. Band 1: Pflanzen und Wirbellose. Landwirtschaftsverlag, Bonn - Bad Godesberg, Germany.
  • Purse, B.V., Hopkins, G.W., Day, K.J. & Thompson, D.J. (2003) Dispersal characteristics and management of a rare damselfly. Journal of Applied Ecology, 40, 716728.
  • R Development Core Team. (2010) R: A Language and Environment for Statistical Computing. R Development Core Team, Vienna, Austria <>.
  • Reineking, B. & Schröder, B. (2006) Constrain to perform: regularization of habitat models. Ecological Modelling, 193, 675690.
  • Ricciardi, A. & Simberloff, D. (2009) Assisted colonization is not a viable conservation strategy. Trends in Ecology & Evolution, 24, 248253.
  • Richter, O., Suhling, F., Müller, O. & Kern, D. (2008) A model for predicting the emergence of dragonflies in a changing climate. Freshwater Biology, 53, 18681880.
  • Ridgeway, G. (2010) gbm: Generalized Boosted Regression Models <>.
  • Rouquette, J.R. & Thompson, D.J. (2007) Patterns of movement and dispersal in an endangered damselfly and the consequences for its management. Journal of Applied Ecology, 44, 692701.
  • Schlumprecht, H., Bittner, T., Jaeschke, A., Jentsch, A., Reinking, B. & Beierkuhnlein, C. (2010) Gefährdungsdisposition von FFH-Tierarten Deutschlands angesichts des Klimawandels: eine vergleichende Sensitivitätsanalyse. Naturschutz und Landschaftsplanung, 42, 293303.
  • Schweiger, O., Settele, J., Kudrna, O., Klotz, S. & Kühn, I. (2008) Climate change can cause spatial mismatch of trophically interacting species. Ecology, 89, 34723479.
  • Seo, C., Thorne, J.H., Hannah, L. & Thuiller, W. (2009) Scale effects in species distribution models: implications for conservation planning under climate change. Biology Letters, 5, 3943.
  • Settele, J., Hammen, V., Hulme, P., Karlson, U., Klotz, S., Kotarac, M., Kunin, W., Marion, G., O’Connor, M., Petanidou, T., Peterson, K., Potts, S., Pritchard, H., Pysek, P., Rounsevell, M., Spangenberg, J., Steffan-Dewenter, I., Sykes, M., Vighi, M., Zobel, M. & Kuhn, I. (2005) Alarm: assessing Large-scale environmental Risks for biodiversity with tested Methods. GAIA - Ecological Perspectives for Science and Society, 14, 6972.
  • Smolik, M., Dullinger, S., Essl, F., Kleinbauer, I., Leitner, M., Peterseil, J., Stadler, L.-M. & Vogl, G. (2010) Integrating species distribution models and interacting particle systems to predict the spread of an invasive alien plant. Journal of Biogeography, 37, 411422.
  • Spangenberg, J.H. (2007) Integrated scenarios for assessing biodiversity risks. Sustainable Development, 15, 343356.
  • Sternberg, K. (1999) Coenagrion ornatum. Die Libellen Baden-Württembergs. Bd. 1 (ed. by K. Sternberg and R. Buchwald), pp. 270277. Ulmer, Stuttgart, Germany.
  • Sternberg, K. & Buchwald, R. (eds.) (1999) Die Libellen Baden-Württembergs. Bd. 1. Ulmer, Stuttgart, Germany.
  • Sternberg, K. & Buchwald, R. (eds.) (2000) Die Libellen Baden-Württembergs. Bd. 2. Ulmer, Stuttgart, Germany.
  • Sternberg, K., Buchwald, R. & Röske, W. (1999) Coenagrion mercuriale. Die Libellen Baden-Württembergs. Bd. 1 (ed. by K. Sternberg and R. Buchwald), pp. 255270. Ulmer, Stuttgart, Germany.
  • Sternberg, K., Höppner, B., Heitz, A. & Heitz, S. (2000) Ophiogomphus cecilia. Die Libellen Baden-Württembergs. Bd. 2 (ed. by K. Sternberg and R. Buchwald), pp. 358373. Ulmer, Stuttgart, Germany.
  • Suhling, F., Werzinger, J. & Müller, O. (2003) Ophiogomphus cecilia (Fourcroy, 1785). Das europäische Schutzgebietssystem Natura 2000: Ökologie und Verbreitung von Arten der FFH-Richtlinie in Deutschland. Band 1: Pflanzen und Wirbellose (ed. by B. Petersen, G. Ellwanger, G. Biewald, U. Hauke, G. Ludwig, P. Pretscher, E. Schröder and A. Ssymank), pp. 593601. Landwirtschaftsverlag, Bonn - Bad Godesberg, Germany.
  • Suhonen, J., Honkavaara, J. & Rantala, M.J. (2010) Activation of the immune system promotes insect dispersal in the wild. Oecologia, 162, 541547.
  • Swets, J.A. (1988) Measuring the accuracy of diagnostic systems. Science, 240, 12851293.
  • Thompson, D.J., Rouquette, J.R. & Purse, B.V. (2003) Ecology of the Southern Damselfly: Conserving Natura 2000 Rivers Ecology Series No. 8. English Nature, Peterborough, UK.
  • Veit, R.R. & Lewis, M.A. (1996) Dispersal, population growth, and the Allee effect: dynamics of the house finch invasion of eastern North America. American Naturalist, 148, 255274.
  • Walther, G.-R., Beißner, S. & Burga, C.A. (2005) Trends in the upward shift of alpine plants. Journal of Vegetation Science, 16, 541548.
  • Wildermuth, H. (1993) Populationsbiologie von Leucorrhinia pectoralis (Charpentier) (Anisoptera: Libellulidae). Libellula, 12, 269275.

Supporting Information

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

Figure S1. Histograms of selected climaticvariables for Coenagrion mercuriale at occurrence and absence locations in Europe.

Figure S2. Histograms of selected climaticvariables for Coenagrion ornatum at occurrence and absence locations in Europe.

Figure S3. Histograms of selected climaticvariables for Leucorrhinia albifrons at occurrence and absence locations in Europe.

Figure S4. Histograms of selected climaticvariables for Leucorrhinia caudalis at occurrence and absence locations in Europe.

Figure S5. Histograms of selected climaticvariables for Leucorrhinia pectoralis at occurrence and absence locations in Europe.

Figure S6. Histograms of selected climaticvariables for Ophiogomphus cecilia at occurrence and absence locations in Europe.

Please note: Neither the Editors nor Wiley-Blackwell are responsible for the content or functionality of any supporting materials supplied by the authors. Any queries (other than missing material) should be directed to the corresponding author for the article.

icad194_sm_FigureS1-S6.pdf782KSupporting info item

Please note: Neither the Editors nor Wiley Blackwell are responsible for the content or functionality of any supporting materials supplied by the authors. Any queries (other than missing material) should be directed to the corresponding author for the article.