Abstract

Conservation of migratory species faces the challenge of understanding the ecological requirements of individuals living in two geographically separated regions. In some cases, the entire population of widely distributed species congregates at relatively small wintering areas and hence, these areas become a priority for the species’ conservation. Satellite telemetry allows fine tracking of animal movements and distribution in those less known, often remote areas. Through integrating satellite and GPS data from five separated populations comprising most of the breeding range, we created a wide habitat suitability model for the Eleonora’s falcon on its wintering grounds in Madagascar. On this basis, we further investigated, for the first time, the impact of climate change on the future suitability of the species’ wintering areas. Eleonora’s falcons are mainly distributed in the north and along the east of Madagascar, exhibiting strong site fidelity over years. The current species’ distribution pattern is associated with climatic factors, which are likely related to food availability. The extent of suitable areas for Eleonora’s falcon is expected to increase in the future. The integration of habitat use information and climatic projections may provide insights on the consequences of global environmental changes for the long-term persistence of migratory species populations.

Introduction

Understanding how organisms interact with their environment and, especially, how they cope with environmental variability is a fundamental question in ecology and of critical importance to the conservation of migratory animals. Migratory species occur in geographically distinct areas during their annual cycle and exploit a variety of habitats1. However, because of the inaccessibility and remoteness of certain areas, the volatility of some political regimes and/or the high cost of large-scale field expeditions, researchers have often been restricted to the study of just a small subset of migratory species’ life stages, usually the breeding period. On top of this, the varying ecological needs of these species constitute a fascinating, yet challenging field of study, which has recently become even more complex in the context of rapid changes occurring at a global level.

The impact of aberrant weather conditions on the abundance and distribution of species has been recognized early during the past century2,3,4. More recently, the analysis of long-term data has provided sound evidence on the impact of climate change on living organisms. For instance, many birds have shifted the timing of breeding and/or migration in response to climate change5, 6. These changes may have cascading effects, often affecting food availability7, thus impacting breeding output and survival rates8,9,10,11, as well as triggering range shifts12,13,14. Even though climate will undoubtedly continue to be a key driver of species’ abundance and distribution in the future, anthropogenic habitat changes can incur equally severe effects on species’ ranges and fitness. For example, intensification of agricultural practices in Mediterranean countries have negatively impacted farmland birds’ populations and have been linked to lower adult survival- and population growth rates e.g. ref. 15 as well as to alterations in the species’ breeding distributions e.g. ref. 16.

Due to the strong endogenous control of migratory behavior, long-distance migrants are expected to be less resilient to the effects of global change17, 18, especially those species depending on seasonal habitats19, 20. Taking into account the velocity at which these changes occur as well as the complexity of developing appropriate management policies, the conservation of long-distance migrant species calls for prompt, high quality databased decision making processes to improve their future prospects e.g. ref. 20. Telemetry has recently allowed almost continuous remote tracking of animal movements providing opportunities to investigate the effect of natural and human-induced factors, including climate change, throughout their life cycle. Telemetry data may be coupled with future climate predictions and ecological niche modeling, thus allowing identifying key limiting environmental factors21, potential wintering areas22,23,24,25 as well as aiding the development of management guidelines26.

The Eleonora’s falcon (Falco eleonorae) is a long-distance migratory raptor that breeds in the Mediterranean basin and winters primarily in Madagascar27, 28, one of the eight ‘hottest’ biodiversity hotspots in the world29. The species was successfully tracked to Madagascar for the first time in 200328. Since then, further studies analyzed the species’ flight characteristics and strategies during migration30,31,32,33, and its wintering habitat distribution34,35,36. During winter, Eleonora’s falcons are mainly insectivorous37,38,39 and food availability strongly depends on the landscape structure and environmental conditions40,41,42,43,44.

As any long-distance migratory species, Eleonora’s falcon may be affected by selective pressures and factors operating in widely separated geographically breeding and non-breeding areas, as well as during migration. This calls for an integrated inter-continent conservation program45, whose effectiveness largely depends on the availability of accurate information about the limiting factors throughout the range of habitats used by the species. Knowledge about the main non-natural factors threatening the persistence of the Eleonora’s falcon breeding populations is accumulating (e.g. persecution46, human disturbance associated with tourism development47, 48, and introduction of mammal predators49, 50. However, the identification of the main threats at its wintering grounds is still difficult to account for, due to the lack of precise information about its distribution and habitat use in Madagascar. Malagasy economy and subsistence relies heavily on the exploitation of natural resources, mainly through agricultural practices, which often result in severe habitat transformations. Consequently, the description of the Eleonora’s falcon habitat requirements and the timely prediction of the impact of global change on the suitability of its wintering grounds are considered essential tools for policy makers, who could then prioritize conservation measures at a global level.

Here, we determined habitat use by Eleonora’s falcons in Madagascar based on all available satellite and GPS tracking data from individuals breeding on the Canaries, Columbretes and Balearic islands (all in Spain), as well as in Italy, Croatia, Greece, and Cyprus32, 34,35,36, 51 (Gangoso et al. unpublished data, Hadjikyriakou et al. unpublished data). In spite of recent large-scale, multipopulation studies52, 53, to our knowledge this is the first time to study a highly mobile raptor species with data from individuals of the entire breeding range. We first explored whether habitat use patterns were consistent across years, as well as among individuals of different breeding origin. Based on these results, we proceeded to develop Species Distribution Models (SDM), which accounted for the effect of individual variability and annual heterogeneities in environmental factors. Finally, we assessed how the present habitat suitability for Eleonora’s falcon in Madagascar is likely to change based on future climate scenarios. Our ultimate goal is to provide a baseline framework for the Eleonora’s falcon conservation for the wintering area, where almost the entire population spends the majority of its annual cycle27.

Results

Exploratory space-use analyses

Space-use analyses demonstrated a high within-individual home range overlap between consecutive years (UDOI = 1.145 ± 0.614, N = 6 adults), suggesting strong winter site-fidelity (Fig. 1). This result allowed us to compare individuals tracked in different years in order to detect differences in home range overlap linked to breeding origin. The falcons’ breeding population did not have a significant effect on the overlap of the estimated home ranges (Table 1). In fact the inter-colony overlap was equal or even exceeded in some cases the intra-colony overlap (Supplementary Table S3), suggesting that falcons from different colonies may aggregate during the wintering period and exploit the same areas.

Figure 1
Figure 1

Strong winter site-fidelity for 6 Eleonora’s falcons as indicated by the overlap of their home ranges in consecutive wintering events and the corresponding value of the Utilization Distribution Overlap Index. Map was created with ArcGIS v10.1 (www.arcgis.com).

Table 1 Home range overlap analysis results based on 95% Kernel Density Estimation (KDE) of 23 Eleonora’s falcons overwintering in Madagascar analyzed using the Utilization Distribution Overlap Index (UDOI).

The subsequent comparisons indicated a strong age effect; the juvenile overlap was significantly lower than adult overlap (Table 1). Accordingly, the distribution patterns of the 6 juveniles differed considerably among them, as well as compared to the adults. While adults were mainly distributed in the north and central part of Madagascar, some juveniles roamed also in the southern and western part of the country (Fig. 2). Nonetheless, considering the small sample size of juveniles (N = 6), as well as the fact that first calendar-year falcons are expected to be more prospective and have less defined habitat preferences than adults, we used only adult data (N = 17) to model habitat suitability for Eleonora’s falcon in Madagascar.

Figure 2
Figure 2

Home ranges of 23 Eleonora’s falcons marked in colonies spanning from the westernmost (Canaries) to the easternmost (Cyprus) breeding range of the species. The existing breeding colonies are highlighted in black (data from87). Map was created with ArcGIS v10.1 (www.arcgis.com).

Present habitat suitability

According to the Maxent model results, minimum precipitation and mean maximum temperature were the main factors related to Eleonora’s falcon wintering habitat distribution (Fig. 3). Given the fact that mean maximum temperature is highly correlated with elevation (r = 0.84, p < 0.001) as well as the drop in training gain when elevation was omitted from the model (Fig. 3), we also consider the latter an important determinant of the species’ habitat suitability. Therefore, according to the corresponding response curves (Fig. 4), habitat suitability peaks in areas around 2100 m, receiving in total at least 100–120 mm of rainfall during the wintering period, where maximum temperature does not exceed 20–22 °C. The areas fulfilling these criteria are mainly located in the northern part of Madagascar, the Tsaratana Massif region, as well as along the eastern part of the country and central highlands (Fig. 5).

Figure 3
Figure 3

Training gain results (average of 100 models based on adult data only) for the explanatory variables used in model building. According to these, the most significant predictors of habitat suitability during the wintering period are minimum precipitation (minprec) and mean maximum temperature (meantemp). Elevation (elev) is also considered to be an important predictor of habitat suitability, since it was found highly correlated with mean maximum temperature and thus its effect might have been overshadowed by the latter. Variable abbreviations as in Supplementary Table S2.

Figure 4
Figure 4

Response curves (average values of 100 models based on adult data only) of the three most significant predictors of habitat suitability for Eleonora’s falcon during the wintering period.

Figure 5
Figure 5

Predicted habitat suitability (average of 100 models) according to current environmental conditions for Eleonora’s falcon in Madagascar, based on satellite telemetry data of 17 adult falcons originating from colonies spanning from the westernmost (Canaries) to the easternmost (Cyprus) breeding range. Map was created with ArcGIS v10.1 (www.arcgis.com).

The models fitted the training datasets well (training AUC = 0.960 ± 0.004) and their predictive power was high (test AUC = 0.953 ± 0.006). In addition, the fit of the models was statistically significant compared to null models (training AUC of null models = 0.540 ± 0.017, 95% confidence limits of null models = 0.538, 0.541).

Future habitat suitability

Compared to present conditions, the extent of suitable areas for Eleonora’s falcon is expected to increase in the future. For instance, considering those areas with a habitat suitability score of at least 0.75 as most suitable, then under present climatic conditions such appropriate areas cover a surface of 28,346 sq. km. In the future, however, they will extend over an area between 38,697 and 44,710 sq. km., according to the projected year and climate scenario (Table S4). In addition, our results suggest a southward shift in the location of the most suitable habitat, a reduction of highly suitable areas in the North and an increase along the center and East of Madagascar extending further southwards (Fig. 6), which is dominated by evergreen humid forests54 and croplands.

Figure 6
Figure 6

Predicted habitat suitability according to future climate scenarios (RCP2.6 above and RCP8.5 below) for 2050 (left) and 2070 (right) for Eleonora’s falcon at its main wintering area, Madagascar, based on satellite telemetry data of 17 adult falcons. Map was created with ArcGIS v10.1 (www.arcgis.com).

Discussion

The Eleonora’s falcon is one of 22 raptor species in the world that are complete migrants at a global level55. Owed to its large breeding range and (apparently) increasing population, estimated in 29,200–29,600 mature individuals, the Eleonora’s falcon is classified as Least Concern56. Nonetheless, it is unevenly distributed within its breeding range, with Greece holding more than 85% of the global population57, while the increase in population numbers observed in late years is attributed to more extensive field surveys57 rather than to an actual increase. The Eleonora’s falcon is well-known for delaying its breeding season to late summer to take advantage of the autumn flux of small migratory birds27. Nonetheless, this strong linkage to this particular food source during the breeding period could seriously challenge its persistence, given the already evidenced impacts of climate change on the phenology of passerine migration58. In the present study, we showed that adult Eleonora’s falcons are faithful to a common, relatively small wintering area, apparently forming a single wintering population regardless of their breeding origin. Therefore, the selective pressures, such as those resulting from climate change, faced in this restricted area are expected to have a more severe impact on the species’ persistence, given that they would affect the whole Eleonora’s falcon population, compared to the unequal influence of local factors acting across the species’ breeding range46,47,48,49,50.

Our previous research highlighted the dominant role of climate, topography and habitat type in the wintering habitat distribution of a few individuals from breeding colonies in Spain35, Italy34 and Greece36. In the present study we were able to generate a species-wide wintering habitat suitability map for the Eleonora’s falcon, by analyzing 23 wintering events spanning a decade and pertaining to 17 adults from 7 breeding colonies. To our knowledge, this is the first time that a habitat suitability model has been generated for a species based on data from individuals originating from colonies across its entire breeding distribution. Therefore, our habitat modeling approach provides a solid benchmark for the species’ conservation and will help answer critical questions about their resilience to global environmental changes.

Present habitat suitability

Our results indicated that habitat suitability for the Eleonora’s falcon in Madagascar is mainly shaped by regional variations in climate regime, in particular precipitation and temperature, both likely related to insect prey availability37,38,39. Beyond the variety of factors influencing insect abundance at various spatial and temporal scales, climatic variables, such as temperature and humidity, are considered among the key drivers of insect abundance and phenology41, 43, 44. Tropical insect species tend to peak in numbers during the wet and transitional seasons59 (and references therein), given the time lag in vegetation phenology42, 60. Insects abound in areas dominated by humid forest and cultivations characterized by moderate to high precipitation levels42, such as the northern and central highlands of Madagascar. At the same time, woodlands with openings attract a high number of insects61, 62. Such landscape mosaics in Madagascar are the result of extensive land use changes since the 1970s, where a large part of forests (both humid and dry spiny forest) have been cleared out for cultivations63, 64. Compared to our previous findings35, 36, the lack of a direct effect of the percent cover of humid and/or degraded forests (namely, veg 14 and 16) on habitat suitability does not undermine the ecological significance of these habitat types for Eleonora’s falcon, but is rather linked to the fact that they occur in relatively fragmented and isolated patches36. Thus, although the ecological drivers of Eleonora’s falcon habitat suitability in Madagascar, namely precipitation, temperature and elevation, are consistent with the species’ food requirements, it is foreseeable that the ongoing forest felling and increased use of biocides can limit the availability of food for Eleonora’s falcons and ecologically related species. This raises an interesting conservation debate; while the increase in degraded land may favor Eleonora’s falcons in the short term, since they often use cultivated areas and degraded forests to forage35, 36, the massive habitat loss undoubtedly poses a serious threat for the entire biodiversity of Madagascar and, in the long term, for Eleonora’s falcons too.

Adult Eleonora’s falcons showed a high degree of site fidelity among years, yet they use multiple core areas likely associated with temporal and regional variations in rainfall patterns. On the contrary, juveniles were more prospective than adults, exploring a larger part of the country and, thus exhibiting a more diffuse distribution pattern (Fig. 2). Still, although the reduced sample size precluded further analyses, the distribution pattern of the 6 juveniles indicated that the most frequented regions were wetlands and areas with minimum precipitation levels above 100 mm, similar to our findings for adult falcons, occurring at areas with at least 100–120 mm of rainfall.

Future habitat suitability and implications for conservation

Our predictions indicate that the extent of suitable habitat for Eleonora’s falcon in Madagascar will increase over the next 50–60 years, even under the worst climate scenario (RCP8.5: high emissions scenario). More specifically, current suitable habitats in the north will be reduced, while habitats located along the eastern and central parts of the country, dominated by humid forests, cultivated areas and grasslands, will become more favorable for the species. However, the most vulnerable regions to climate change in the East are considered to be the ones lying from 1200 to 1600 m altitude65, which partially coincide with highly suitable areas for the species under the worst climate change scenario (namely, RCP 8.5 2070).

Agroecosystem responses to global environmental changes, including the insect community, are harder to predict, given the intensive human-related habitat transformations. Considering that deforestation is still ongoing in Madagascar, further habitat loss may counterbalance the potential positive effects of more favorable climatic conditions. While the northern highlands have remained fairly intact due to their inaccessibility, the forests lying in the central and southern parts of the country, which are expected to be rendered suitable for Eleonora’s falcon in the future, can be largely impacted by the expansion of agricultural practices. Thus, the predicted southwards shift of the species’ habitat suitability and its expansion into cultivated areas in the central part of Madagascar could result in an increased exposure of the species to anthropogenic habitats and to the biocides used in agriculture, reducing prey availability and increasing the risk of secondary poisoning66.

Consequently, although our predictive models may suggest optimism with regards to future availability of suitable wintering habitat, our results should be considered as a possible outcome all other things being equal. Moreover, the species is an island dweller both during its breeding and wintering period and a highly specialized raptor in terms of food requirements37. In this regard, specialist predators are predicted to have a lower potential to buffer environmental variations than generalist species67, 68. In addition, Eleonora’s falcon population gathers at a relatively small wintering area (present and predicted) relative to the size of the country. Therefore, strong perturbations in Madagascar associated with extreme weather events, habitat loss and decline in habitat quality (mainly as a result of human intervention), may induce carry-over effects, affecting species’ long-term fitness69. Other factors fuelled by global environmental change, such as the use of biocides, should be considered an issue of major concern for the species’ conservation. For example, following the locust outbreak in 2012 in Madagascar, the widespread use of insecticides (more than 42,000 km2 were sprayed within four years) caused serious declines in populations of non-target invertebrates and vertebrates, leading to food-chain perturbations70. Such outbreaks are likely to increase as a consequence of climate change65, raising serious concerns about mitigation measures-derived hazards to wildlife.

Moreover, the highly suitable areas (i.e. those with a habitat suitability score over 0.75) under current conditions overlap spatially with the corresponding areas under future conditions at just 59.54% (SD = 6.07%), suggesting that the Eleonora’s falcon current distribution will change substantially in the following years. On top of this, the existing protected areas network in Madagascar overlaps only a fraction (less than 20%) of the highly suitable habitat for the Eleonora’s falcon both under present and future climatic conditions (see Supplementary Table S5 and Fig. S2). Therefore, our model predictions may on one hand help to identify the key areas for this falcon in the future in order to develop appropriate conservation strategies. On the other hand, our results can contribute to the reinforcement of the protection status of the habitats Eleonora’s falcon depends upon, especially primary evergreen forests, and thus the biodiversity these habitats host as a whole.

To sum up, the natural landscape of the Eleonora’s falcons’ wintering area has been altered by drastic human intervention in a very short period of time and is expected to be largely affected by climate change in the future37, 71, 72. Regardless of the uncertainty surrounding future climate scenarios, as well as the species’ ability to disperse, colonize and survive in the newly climatically suitable areas, our results provide a compass for future studies furnishing valuable information for the species’ conservation in Madagascar. More specifically, we explored habitat use patterns and winter site-fidelity through home range analysis and created a species-wide habitat suitability model, providing the first evidence of climate change effects on habitat suitability. Given the increased pace of climate change, and in alignment with the current trend in animal ecology and conservation, we prompt future studies to investigate to what extent genetic variation, plasticity and population dynamics jointly determine how migratory species living in seasonal environments may respond to the ongoing and future environmental changes. This multidisciplinary approach will contribute to the re-evaluation of conservation priorities at a global level, a task that should reconcile trade-offs between biodiversity conservation and socioeconomic development in a framework of unprecedented global change.

Methods

Eleonora’s falcon present and future habitat suitability was modeled using MaxEnt, one of the most powerful SDM methods73, 74. MaxEnt is a niche-modeling technique for presence-only data, which approximates habitat suitability by identifying areas that share the same environmental conditions (i.e. combination of ecological variables) as those areas where the target species has been recorded with a given set of environmental variables. In order to identify which environmental conditions are mostly preferred by the target species, MaxEnt contrasts the values of the candidate environmental variables between the area occupied by the species and the available geographical space, known as “background”.

Presence-background models, like MaxEnt, that are based on satellite telemetry data, especially when derived from a large number of individuals, can provide robust predictions on the geographic distribution of more suitable environmental conditions at the species-level75. In this context, our methodological approach is intended to produce a reliable species-wide habitat suitability map under present and future environmental conditions.

Data preparation was implemented in ArcGIS v10.176 statistical analyses were performed in R v3.2.277, while SDMs were carried out in MaxEnt v3.3.3k78. Details on capture methods and animal handling have been described in our previous papers34,35,36. The methods described here have been adapted from34 and36. A detailed description of methods used is provided in the Supplementary material. Mean values and standard deviations are reported. Statistical significance level was set to α = 0.05. All spatial data used for the subsequent analyses, as well as the maps presented herein have been projected to WGS 84/UTM zone 38 S (EPSG: 29738).


Occurrences

We compiled all available telemetry data to date (i.e., from 2003 to 2014) for the wintering period (November–April) of Eleonora’s falcon in Madagascar. These telemetry data derived from 23 individuals (adults and first calendar year juveniles, hereafter “juveniles”, Table 2) originating from colonies located across the entire species’ breeding range and adding to 30 wintering events. Following appropriate filtering (see Supplementary Appendix 1) c.a. 5,000 data points constituted the data pool for subsequent analyses (Supplementary Table S1).

Table 2 Summary of satellite telemetry data for Eleonora’s falcon at its Malagassy wintering areas used in this study.

Environmental correlates

According to the species’ ecology and the results of our previous research34,35,36, we considered topography (elevation, topographic roughness), proximity to fresh water bodies, vegetation composition (percentage of vegetation classes), vegetation phenology (NDVI) and climate regime (mean maximum temperature, maximum and minimum precipitation) as candidate predictors of habitat suitability for the occurrence of Eleonora’s falcon at its wintering quarters (Supplementary Table S2).

Exploratory space-use analyses

Prior to model building we assessed whether patterns of habitat use differ during the wintering season among years, populations and age classes. In particular, we estimated home range (95%-KDE) utilization distribution (UD) for each individual and wintering event based on the aforementioned occurrence data pool (see Supplementary Appendix 1). Firstly, we assessed wintering site-fidelity by calculating the UD Overlap Index (UDOI79), for the same individual between consecutive wintering seasons (N = 6 individuals). We then calculated UDOI between age classes and populations. To avoid pseudo-replication, we averaged the UDOI values derived from multiple wintering events referring to the same individual.

Modeling present habitat suitability

According to the results of the space-use analyses, we generated habitat suitability models considering adults separately following the procedure outlined below (for a detailed description please see Supplementary Appendix 1).

In order to achieve equal representation of all individuals during model building80, we created 10 random subsamples for each individual that equaled the number of data points of the falcon with the smallest sample size rounded to the nearest integer (Supplementary Table S1). Taking into account the number of explanatory variables81, we trained the MaxEnt model with 80% of each presence subsample and evaluated the model predictive performance with the remaining 20%. Data splitting was conducted 10 times at random per subsample. Overall, we created 100 models (10 subsamples for calibration × 10 subsamples for evaluation). Model results are reported as averages of these 100 MaxEnt models (hereafter, “present model”). The predicted habitat suitability scores range from 0 to 1 (i.e. logistic output, which approximates the probability of occurrence).

As a measure of the overall model’s predictive power, MaxEnt uses by default the Area Under the Curve (AUC score82). However, considering the criticism against its use as a metric of model accuracy, especially in large study areas as in our case83, we assessed the statistical significance of the AUC scores of the resulting 100 models by contrasting them with the resulting values derived from null models following the recommended approach by84.

In order to assess the predictive power of the candidate explanatory variables we compared the variance explained by each variable when used in isolation (i.e. resembling a univariate model) and the information lost when that variable was omitted from a model containing the remaining variables.

Modeling future habitat suitability

We predicted future habitat suitability based on different climate change scenarios, using the latest Global Climate Models (GCMs) of the fifth phase of the Coupled Model Intercomparison Project (CMIP5; http://cmip-pcmdi.llnl.gov/cmip5/). GCMs are generated based on scenarios concerning emissions of pollutants, future climatic and environmental conditions, as well as socioeconomic changes. CMIP5 in particular considers four scenarios, known as Representative Concentration Pathways (RCPs), of which we chose the two extremes, the low emissions scenario (RCP 2.6) and the high emissions scenario (RCP 8.5) to model future habitat suitability (for an overview of the RCPs see85). Among the available GCMs we used future climate data produced by the HadGEM2-ES model86 for 2050 and 2070. We generated four future habitat suitability projections (2 scenarios × 2 years; hereafter, “RCP2.6_2050 model”, “RCP2.6_2070 model”, “RCP8.5_2050 model”, “RCP8.5_2070 model”) by repeating the methodological steps described above for the present model.

Additional information

Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

References

  1. 1.

    Newton, I. The Migration Ecology of Birds (Academic Press, 2008).

  2. 2.

    Taylor, W. P. Significance of extreme or intermittent conditions in distribution of species and management of natural resources, with a restatement of Liebig’s law of minimum. Ecology 15, 374–379 (1934).

  3. 3.

    Webster, R. L. Effect of low temperature on the oyster-shell scale. J. Econ. Entomol. 8, 371–375 (1915).

  4. 4.

    Larsen, E. B. The influence of the severe winters of 1939–42 on the soil fauna of Tipperne. Oikos 1, 184–207 (1949).

  5. 5.

    Gordo, O. Why are bird migration dates shifting? A review of weather and climate effects on avian migratory phenology. Clim. Res. 35, 37–58 (2007).

  6. 6.

    Jaffré, M. et al. Long-term phenological shifts in raptor migration and climate. PLoS ONE 8, e79112 (2013).

  7. 7.

    Terraube, J. et al. Coping with fast climate change in northern ecosystems: mechanisms underlying the population-level response of a specialist avian predator. Ecography 38, 690–699 (2015).

  8. 8.

    Lehikoinen, A. et al. Impact of climate change and prey abundance on nesting success of a top predator, the goshawk. Oecologia 171, 283–293 (2013).

  9. 9.

    Fisher, R. J. et al. Extreme precipitation reduces reproductive output of an endangered raptor. J. Appl. Ecol. 52, 1500–1508 (2015).

  10. 10.

    Anctill, A., Franke, A. & Bêty, J. Heavy rainfall increases nestling mortality of an arctic top predator: experimental evidence and long-term trend in peregrine falcons. Oecologia 174, 1033–1043 (2014).

  11. 11.

    Dybala, K. E., Eadie, J. M., Gardali, T., Seavy, N. E. & Herzog, M. Projecting demographic responses to climate change: adult and juvenile survival respond differently to direct and indirect effects of weather in a passerine population. Glob. Change Biol. 19, 2688–2697 (2013).

  12. 12.

    Parmesan, C. & Yohe, G. A globally coherent fingerprint of climate change impacts across natural systems. Nature 421, 37–42 (2003).

  13. 13.

    Crick, H. Q. P. The impact of climate change on birds. Ibis 146, 48–56 (2004).

  14. 14.

    Robinson, R. A. et al. Travelling through a warming world: climate change and migratory species. Endangered Spec. Res. 7, 87–99 (2009).

  15. 15.

    Di Maggio, R., Campobello, D., Tavecchia, G. & Sarà, M. Habitat- and density-dependent demography of a colonial raptor in Mediterranean agro-ecosystems. Biol. Conserv. 193, 116–123 (2016).

  16. 16.

    Sarà, M. Spatial analysis of lanner falcon habitat preferences: Implications for agro-ecosystems management at landscape scale and raptor conservation. Biol. Conserv. 178, 173–184 (2014).

  17. 17.

    Gwinner, E. Circannual clocks in avian reproduction and migration. Ibis 138, 47–63 (1996).

  18. 18.

    Both, C. & Visser, M. E. Adjustment to climate change is constrained by arrival date in a long-distance migrant bird. Nature 411, 296–298 (2001).

  19. 19.

    Both, C. et al. Avian population consequences of climate change are most severe for long-distance migrants in seasonal habitats. Proc. R. Soc. London Ser. B 277, 1259–1266 (2010).

  20. 20.

    Culp, L. A., Cohen, E. B., Scarpignato, A. L., Thogmartin, W. E. & Marra, P. P. Full annual cycle climate change vulnerability assessment for migratory birds. Ecosphere 8, e01565, doi:10.1002/ecs2.1565 (2017).

  21. 21.

    Oberhauser, K. & Petterson, T. Modeling current and future potential wintering distributions of eastern North American monarch butterflies. P. Natl. Acad. Sci. USA 100, 14063–14068 (2003).

  22. 22.

    Limiñana, R., Soutullo, A., Arroyo, B. & Urios, V. Protected areas do not fulfill the wintering habitat needs of the trans-Saharan migratory Montagu’s harrier. Biol. Conserv. 145, 62–69 (2012).

  23. 23.

    Limiñana, R., Arroyo, B., Terraube, J., McGrady, M. & Mougeot, F. Using satellite telemetry and environmental niche modelling to inform conservation targets for a long-distance migratory raptor in its wintering grounds. Oryx 49, 329–337 (2015).

  24. 24.

    Tegetmeyer, C., Frick, A. & Seifert, N. Modelling habitat suitability in the Aquatic Warbler wintering ground Djoudj National Park area in Senegal. Ostrich 85, 57–66 (2014).

  25. 25.

    Zeng, Q. et al. Using Species Distribution Model to Estimate the Wintering Population Size of the Endangered Scaly-Sided Merganser in China. PLoS ONE 10, e0117307 (2015).

  26. 26.

    Franco, A. M. A., Brito, J. C. & Almeida, J. Modeling habitat selection of Common Cranes Grus grus wintering in Portugal using multiple logistic regression. Ibis 142, 351–358 (2000).

  27. 27.

    Walter, H. Eleonora’s falcon: adaptations to prey and habitat in a social raptor (University of Chicago Press, 1979).

  28. 28.

    Gschweng, M., Kalko, E. K. V., Querner, U., Fielder, W. & Berthold, P. All across Africa: highly individual migration routes of Eleonoras falcon. Proc. R. Soc. London Ser. B 275, 2887–2897 (2008).

  29. 29.

    Meyers, N., Mittermeier, R. A., Mittermeier, C. G., da Fonseca, G. A. B. & Kent, J. Biodiversity hotspots for conservation priorities. Nature 403, 853–858 (2000).

  30. 30.

    López-López, P., Limiñana, R., Mellone, U. & Urios, V. From the Mediterranean Sea to Madagascar: are there ecological barriers for the long-distance migrant Eleonora’s falcon? Landscape Ecol. 25, 803–813 (2010).

  31. 31.

    Mellone, U., López-López, P., Limiñana, R. & Urios, V. Weather conditions promote route flexibility during open ocean crossing in a long-distance migratory raptor. Int. J. Biometeorol. 55, 463–468 (2011).

  32. 32.

    Mellone, U., López-López, P., Limiñana, R., Piasevoli, G. & Urios, V. The trans-equatorial loop migration system of Eleonora’s falcon: differences in migration patterns between age classes, regions and seasons. J. Avian Biol. 44, 417–426 (2013).

  33. 33.

    Kassara, C., Fric, J., Gschweng, M. & Sfenthourakis, S. Complementing the puzzle of Eleonora’s Falcon (Falco eleonorae) migration: new evidence from an eastern colony in the Aegean Sea. J. Ornithol. 153, 839–848 (2012).

  34. 34.

    Gschweng, M., Kalko, E. K. V., Berthold, P., Fielder, W. & Fahr, J. Multi-temporal distribution modelling with satellite tracking data: predicting responses of a long-distance migrant to changing environmental conditions. J. Appl. Ecol. 49, 803–813 (2012).

  35. 35.

    Mellone, U., López-López, P., Limiñana, R. & Urios, V. Wintering habitats of Eleonora’s Falcons Falco eleonorae in Madagascar. Bird Study 59, 29–36 (2012).

  36. 36.

    Kassara, C., Fric, J. & Sfenthourakis, S. Distribution modeling of Eleonora’s Falcon Falco eleonorae Géné, 1839 occurrence in its wintering grounds: a niche-based approach with satellite telemetry data. Bird Conserv. Int. 24, 100–113 (2014).

  37. 37.

    Zefania, S. Observation of Sooty and Eleonora’s falcons in Madagascar In: Wings over Africa, Proceedings of an International Seminar on Bird Migration: Research, Conservation, Education and Flight Safety, Israel, pp. 151–159 (2001).

  38. 38.

    Ristow, D. On the insect diet of Eleonora’s Falcon Falco eleonorae and its importance for coloniality in Raptors Worldwide (eds Chancellor, R. D. & Meyburg, B.-U.) 705–712 (WWGBP/MME, 2004).

  39. 39.

    Gschweng, M. Eleonora’s falcon Falco eleonorae in The Birds of Africa: Volume VIII: The Malagasy Region: Madagascar, Seychelles, Comoros, Mascarenes (eds Safford, R. & Hawkins, F.) 229–233 (McMillan, 2013).

  40. 40.

    Whitney, K. S. et al. Explicit modeling of abiotic and landscape factors reveals precipitation and forests associated with aphid abundance. Ecol. Appl. 26, 2600–2610 (2016).

  41. 41.

    Willmer, P. G. Microclimate and the environmental physiology of insects. Adv. Insect Physiol 16, 1–57 (1982).

  42. 42.

    Wolda, H. Insect seasonality: Why? Annu. Rev. Ecol. Syst. 19, 1–18 (1988).

  43. 43.

    Bale, J. S. et al. Herbivory in global climate change research: direct effects of rising temperature on insect herbivores. Global Change Biol. 8, 1–16 (2002).

  44. 44.

    Savopoulou-Soultani, M., Papadopoulos, N.T., Milonas, P. & Moyal, P. Abiotic factors and insect abundance. Psyche 2012, doi:10.1155/2012/167420 (2012).

  45. 45.

    McInnes, R. J. African-Eurasian Waterbird Agreement (AEWA) and Wetland Management In The Wetland Book I: Structure and Function, Management and Methods (eds Finlayson, C.M et al.) 1–5 (Springer Netherlands, 2016).

  46. 46.

    Clark, A. L. The population and reproduction of the Eleonora’s falcon in Morocco. Bull. Soc. Sci.Nat. Maroc 54, 61–69 (1974).

  47. 47.

    Martínez-Abrain, A., Oro, D., Ferrís, V. & Belenguer, R. Is growing tourist activity affecting the distribution or number of pairs in a small colony of the Eleonora´s Falcon? Anim. Biodivers. Conserv. 25, 47–51 (2002).

  48. 48.

    Touati, L. et al. On the brink: Status and breeding ecology of Eleonora’s Falcon Falco eleonorae in Algeria. Bird Conserv. Int., doi:10.1017/S0959270916000484 (2017).

  49. 49.

    Bonnín, J. Recompte i paràmetres reproductors de la població balear del Falcó marí Falco eleonorae a l’any 2004. Anuari ornitològic de les Balears: revista d’ observació estudi i conservació dels aucells 19, 1–9 (2004).

  50. 50.

    Ristow, D. & Wink, M. Breeding success and conservation management of Eleonora’s falcon in Conservation Studies on Raptors, ICBP Technical Publication No. 5 (eds Newton, I. & Chancellor, R. D.) 147–152 (Cambridge, 1985).

  51. 51.

    Viana, D. S., Gangoso, L., Bouten, W. & Figuerola, J. Overseas seed dispersal by migratory birds. Proc. R. Soc. London Ser. B 283, 20152406 (2016).

  52. 52.

    Trierweiler, C. et al. Migratory connectivity and population-specific migration routes in a long-distance migratory bird. Proc. R. Soc. London Ser. B 281, 20132897 (2014).

  53. 53.

    Finch, T. et al. A pan-European, multipopulation assessment of migratory connectivity in a near-threatened migrant bird. Divers. Distrib. 21, 1051–1062 (2015).

  54. 54.

    Du Puy, D.J. & Moat, J. Using geological aubstrate to identify and map primary vegetation types in Madagascar and the implications for planning biodiversity conservation in The Natural History of Madagascar (eds Goodman, S. M. & Benstead, J. P.) 51–67 (The University of Chicago Press, 2013).

  55. 55.

    Bildstein, K.L. Migrating raptors of the world: Their ecology and conservation (Cornell University Press, 2006).

  56. 56.

    BirdLife International. Falco eleonorae. The IUCN Red List of Threatened Species 2015: e.T22696442A80385285 (2015).

  57. 57.

    Dimalexis, A. et al. The status of Eleonora’s falcon (Falco eleonorae) in Greece. J. Ornithol. 149, 23–30 (2008).

  58. 58.

    Gangoso, L. et al. Ecological specialization to fluctuating resources prevents long-distance migratory raptors from becoming sedentary on islands. PLoS ONE 8, e61615 (2013).

  59. 59.

    Pinheiro, F., Diniz, I. R., Coelho, D. & Bandeira, M. P. S. Seasonal pattern of insect abundance in the Brazilian cerrado. Austral Ecol. 27, 132–136 (2002).

  60. 60.

    Liebhold, A., Koenig, W. D. & Brønstad, O. N. Spatial synchrony in population dynamics. Annu. Rev. Ecol. Evo. Syst. 35, 467–490 (2004).

  61. 61.

    Fuentes-Montemayor, E., Goulsom, D., Cavin, L., Wallace, J. M. & Park, K. J. Fragmented woodlands in agricultural landscapes: The influence of woodland character and landscape context on bats and their insect prey. Agric. Ecosyst. Environ. 172, 6–15 (2013).

  62. 62.

    Hahn, P. G. & Orrock, J. L. Land-use history alters contemporary insect herbivore community composition and decouples plant-herbivore relationships. J. Anim. Ecol. 84, 745–754 (2015).

  63. 63.

    Jarosz, L. Defining and explaining tropical deforestation: shifting cultivation and population growth in colonial Madagascar (1896–1940). Econ. Geogr. 69, 366–379 (1993).

  64. 64.

    Harper, G. J., Steininger, M. K., Tucker, C. J., Juhn, D. & Hawkins, F. Fifty years of deforestation and forest fragmentation in Madagascar. Environ. Conserv. 34, 325–333 (2007).

  65. 65.

    Conservation International & WWF Madagascar. Assessing the Impact of Climate Change on Madagascar’s Livelihoods and Biodiversity. Workshop report. Antananarivo, Madagascar, 28 January 2008 (2008).

  66. 66.

    Paprocki, N., Heath, J. A. & Novak, S. J. Regional distribution shifts help explain local changes in wintering raptor abundance: implications for interpreting population trends. PLoS ONE 9, e86814 (2014).

  67. 67.

    Ferrer, M. & Negro, J.J. The near extinction of two large European predators: super specialists pay a price. Conserv. Biol. 18, 344–349 (2004).

  68. 68.

    Kerbiriou, C., Le Viol, I., Jiguet, F. & DeVictor, V. More species, fewer specialists: 100 years of changes in community composition in an island biogeographical study. Diversity Distrib. 15, 641–648 (2009).

  69. 69.

    Senner, N. R., Hochachka, W. M., Fox, J. W. & Afanasyev, V. An Exception to the Rule: Carry-Over Effects Do Not Accumulate in a Long-Distance Migratory Bird. PLoS ONE 9, e86588 (2014).

  70. 70.

    Peveling, R. et al. Impact of locust control on harvester termites and endemic vertebrate predators in Madagascar. J. Appl. Ecol. 40, 729–741 (2003).

  71. 71.

    Hannah, L. et al. Climate change adaptation for conservation in Madagascar. Biol. Lett. 4, 590–594 (2008).

  72. 72.

    Tadross, M., Randriamarolaza, L., Rabefitia, Z. & Zheng, K.Y. Climate change in Madagascar; recent past and future (World Bank, 2008).

  73. 73.

    Elith, J. et al. Novel methods improve prediction of species’ distributions from occurrence data. Ecography 29, 129–151 (2006).

  74. 74.

    Searcy, C. A. & Shaffer, H. B. Do ecological niche models accurately identify climate determinants of species ranges? Am. Nat. 187, 423–435 (2016).

  75. 75.

    Jiguet, F., Barbet-Massin, M. & Chevalier, D. Predictive distribution models applied to satellite tracks: modeling the western African winter range of European migrant Black Storks. Ciconia nigra. J. Ornith. 152, 111–118 (2011).

  76. 76.

    ESRI. ArcGIS Desktop for Windows. Version 10.1. Redlands, CA: ESRI (2012).

  77. 77.

    R Core Team. R: a language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria www.R-project.org (2015).

  78. 78.

    Phillips, S. J., Anderson, R. P. & Schapire, R. E. Maximum entropy modeling of species geographic distributions. Ecol. Modell. 190, 231–259 (2006).

  79. 79.

    Fieberg, J. & Kochanny, C. O. Quantifying home-range overlap: the importance of the utilization distribution. J. Wildl. Manage. 69, 1346–1359 (2005).

  80. 80.

    Edrén, S. M. C., Wisz, M. S., Teilmann, J., Dietz, R. & Söderkvist, J. Modelling spatial patterns in harbour porpoise satellite telemetry data using maximum entropy. Ecography 33, 698–708 (2010).

  81. 81.

    Fielding, A. H. & Bell, J. F. A review of methods for the assessment of prediction errors in conservation presence/absence models. Environ. Conserv. 24, 38–49 (1997).

  82. 82.

    Hanley, J. A. & McNeil, B. J. The meaning and use of the Area under a Receiver Operating Characteristic (ROC) Curve. Radiology 143, 29–36 (1982).

  83. 83.

    Lobo, J. M., Jiménez-Valverde, A. & Real, R. AUC: a misleading measure of the performance of predictive distribution models. Glob. Ecol. Biogeogr. 17, 145–151 (2008).

  84. 84.

    Raes, N. & ter Steege, H. A null-model for significance testing of presence-only species distribution models. Ecography 30, 727–736 (2007).

  85. 85.

    Van Vuuren, D. P. et al. The representative concentration pathways: an overview. Clim. Change 109, 5–31 (2011).

  86. 86.

    Collins, W.J. et al. Evaluation of the HadGEM2 model. Met Office Hadley Centre, Exeter, UK. http://www.metoffice.gov.uk/media/pdf/8/7/HCTN_74.pdf (2008).

  87. 87.

    BirdLife International & NatureServe. Bird Species Distribution Maps of the World. BirdLife International, Cambridge, UK and Nature Serve, Arlington, USA (2015).

Download references

Acknowledgements

We would like to thank two anonymous reviewers for thoughtful comments and suggestions for improving a previous version of the article. The satellite telemetry project in Greece work was funded by the A.G. Leventis Foundation through project “Survey and Conservation of Seabirds in Greece, while all experiments and observations made for this project comply with current laws of Greece. The Regione Della Sardegna and the Istituto Nazionale della Fauna Selvatica, Italy, kindly granted permission for fieldwork on Sardinia. The ‘Direction des Eaux et Forets’ and the ANGAP (Association National pour la Gestion des Aires Protégés) granted permits to M.G. to conduct faunal surveys in Madagascar. The German Academic Exchange Programme (DAAD) funded fieldwork to M.G, the Madagascar BirdLife Programme assisted fieldwork in Madagascar. The Terra Natura Foundation and the “Servicio de Biodiversidad” (Generalitat Valenciana) funded the tagging in Columbretes and Balearic islands (Spain). The Cabildo de Lanzarote partially funded the tagging in the Canary Islands (Spain) and research permits were granted by the Spanish regional administration (permits no. ES-000844/2012 and ES-000642/2013) and the transmitters were funded by the Project “Adaptación y mejora de la internacionalización de la e-infraestructura ICTS-RBD para la ESFRI-Lifewatch“ funded by Spanish Ministry of Economy and Competitiveness and the European Social Fund. The transmitters used in Cyprus were funded by the Peregrine Fund. Sovereign Bases Areas (SBAs) Administration in Cyprus granted license to attach the transmitters on individuals nesting within the SBAs. C.K. was supported by the LIFΕ13 NAT/GR/000909 project, with the financial support of the European Union LIFE Instrument and the Green Fund. L.G. was supported by a Marie Curie Fellowship of the European Commission. P.L.-L. was supported by a “Juan de la Cierva-incorporación” postdoctoral grant of the Spanish Ministry of Economy and Competitiveness (reference IJCI-2014-19190). R.S. was supported by a PhD grant of the Ministerio de Educación, Cultura y Deporte of Spain (FPU13-03484).

Author information

Affiliations

  1. Department of Biology, University of Patras, GR-26500, Patras, Greece

    • Christina Kassara
    •  & Sinos Giokas
  2. Department of Wetland Ecology, Estación Biológica de Doñana, CSIC, 41092, Seville, Spain

    • Jordi Figuerola
    •  & Rafa Silva
  3. Vertebrates Zoology Research Group, CIBIO Research Inst., University of Alicante, ES-03690, San Vicente del Raspeig, Alicante, Spain

    • Ugo Mellone
    •  & Vicente Urios
  4. Public Institute for the Protected Natural Values Management in the County of Split and Dalmatia, Prilaz braće Kaliterna 10, HR-21000, Split, Croatia

    • Gvido Piasevoli
  5. Department of Biological Sciences, University of Cyprus, 1678, Nicosia, Cyprus

    • Thomas G. Hadjikyriakou
    •  & Alexander N. G. Kirschel
  6. Hellenic Ornithological Society, Themistokleous str. 80, 10681, Athens, Greece

    • Nikos Tsiopelas
  7. Cavanilles Institute of Biodiversity and Evolutionary Biology, University of Valencia, C/Catedrático José Beltrán 2, ES-46980, Paterna, Valencia, Spain

    • Pascual López-López
  8. Computational Geo-Ecology Institute for Biodiversity and Ecosystem Dynamics, University of Amsterdam, 1098 XH, Amsterdam, The Netherlands

    • Laura Gangoso
    •  & Willem Bouten
  9. The Peregrine Fund, 5668 West Flying Hawk Lane, Boise, Idaho, 83709, USA

    • Munir Z. Virani
  10. Max Planck Institute for Ornithology, Am Obstberg 1, D-78315, Radolfzell, Germany

    • Wolfgang Fiedler
    • , Peter Berthold
    •  & Marion Gschweng
  11. Institute of Experimental Ecology, University of Ulm, Albert-Einstein-Allee 11, D-89069, Ulm, Germany

    • Marion Gschweng
  12. Concepts for Conservation, Schäferweg 6, 89143, Blaubeuren, Germany

    • Marion Gschweng

Authors

  1. Search for Christina Kassara in:

  2. Search for Laura Gangoso in:

  3. Search for Ugo Mellone in:

  4. Search for Gvido Piasevoli in:

  5. Search for Thomas G. Hadjikyriakou in:

  6. Search for Nikos Tsiopelas in:

  7. Search for Sinos Giokas in:

  8. Search for Pascual López-López in:

  9. Search for Vicente Urios in:

  10. Search for Jordi Figuerola in:

  11. Search for Rafa Silva in:

  12. Search for Willem Bouten in:

  13. Search for Alexander N. G. Kirschel in:

  14. Search for Munir Z. Virani in:

  15. Search for Wolfgang Fiedler in:

  16. Search for Peter Berthold in:

  17. Search for Marion Gschweng in:

Contributions

C.K., U.M., T.G.H. and M.G. conceived the ideas. C.K. and M.G. l ed the analysis. C.K. analysed the data (data preparation, habitat suitability modeling) and wrote the paper. L.G. and R.S. analysed the data (home range analyses). All authors (C.K., L.G., U.M., G.P., T.G.H., N.T., S.G., P.L.-L., V.U., J.F., R.S., W.B., A.N.G.K., M.Z.V., W.F., P.B., M.G.) discussed and commented on drafts of the manuscript.

Competing Interests

The authors declare that they have no competing interests.

Corresponding author

Correspondence to Christina Kassara.

Electronic supplementary material

About this article

Publication history

Received

Accepted

Published

DOI

https://doi.org/10.1038/s41598-017-08753-w

Comments

By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.