• Keine Ergebnisse gefunden

“ „ Habitat suitability model for the Redspotted Bluethroat (Luscinia svecica svecica) in Central Europe Master Thesis

N/A
N/A
Protected

Academic year: 2022

Aktie "“ „ Habitat suitability model for the Redspotted Bluethroat (Luscinia svecica svecica) in Central Europe Master Thesis"

Copied!
84
0
0

Wird geladen.... (Jetzt Volltext ansehen)

Volltext

(1)

Master Thesis

im Rahmen des

Universitätslehrganges „Geographical Information Science & Systems“

(UNIGIS MSc) am Zentrum für GeoInformatik (Z_GIS) der Paris Lodron-Universität Salzburg

zum Thema

„Habitat suitability model for the Redspotted Bluethroat (Luscinia svecica svecica) in Central Europe“

vorgelegt von

Mag. rer.nat. Ingrid Kohl

U1300, UNIGIS MSc Jahrgang 2006 Zur Erlangung des Grades

„Master of Science (Geographical Information Science & Systems) – MSc(GIS)”

Gutachter:

Ao.Univ.Prof. Dr. Josef Strobl

Wien, April 2009

(2)

Habitat suitability model for the Redspotted Bluethroat (Luscinia svecica svecica) in Central Europe

Das Rotsternige Blaukehlchen im Hundsfeldoor bei Obertauern im Bundesland Salzburg (Foto: Ingrid Kohl, 23.6.2005)

Mag. rer.nat. Ingrid Kohl Friesgasse 9/2b, A-1150 Wien

+436508338313 kohl.ingrid@gmx.at

Institutionelles Umfeld:

University of Vienna, Faculty of Life Sciences, Rennweg 14, A-1030 Wien Department of Conservation Biology, Vegetation- and Landscape Ecology

Betreuer: o.Univ.Prof. Mag. Dr. Georg Grabherr

V.I.N.C.A. – Vienna Institute for Nature Conservation and Analyses Gießergasse 6/7, A-1090 Wien

Betreuer: Mag. Christoph Plutzar MAS

(3)

Aus naturschutzfachlichen Gründen werden keine genauen Ortsangaben von Brutstandorten gemacht! Ich bitte um Verständnis!

Due to conservational reasons the exact localities of breeding sites are not publicated!

I hope you understand!

(4)

Preface

Preinvestigations to this work were carried out for my Master thesis in Ecology at the University of Vienna: „Habitat requirements and territory size in an isolated population of Redspotted Bluethroat (Luscinia svecica svecica) in the Austrian Alps”. Field work to habitat, habitat use and prey availiability was carried out continuously from May until October 2005 in the mire Hundsfeld Moor in Obertauern in the province Salzburg in Austria at an altitude of 1750 to 1900 m a.s. In this area the biggest and longest known population of the subspecies in Central Europe is located. At the same time it was the first discovered in Central Europe by Mrs Johanna Gressel in 1975. GIS based habitat modelling is based on knowledge about habitat requirements from habitat investigations of 2005 and on additional data, e.g. climate data.

Since February 2002 Mrs Johanna Gressel taught me about the study species and soon I will complete my PhD in Natural Sciences/Ecology about this subspecies at the Department of Conservation Biology, Vegetation- and Landscape Ecology of the Faculty of Life Sciences of the University of Vienna supervised by o.Univ.Prof. Mag.

Dr. Georg Grabherr. The topic is the phenomenon of the Redspotted Bluethroat in Central Europe. The main topics of my dissertation are habitat, habitat suitability, distribution, population trends, origin, colonization, migration routes, population dynamics and conservation. Aims of the PhD are a profile of habitat requirements, a distribution map, an identification of population trends over more than three decades, phylogenetic research, an identification of wintering areas with a comparison to Scandinavian populations, a reconstruction of migration routes and a synthesis of the topics with the intention of an application in conservation.

Risks and prerequisites of the Master thesis were the availability and the sufficient spatial and attributive resolution of the suitable data basis, sufficient information about the habitat requirements of the study species and communication in the ornithologists‟

network. Data for the model covered 220 GB storage. It is the first habitat suitability model for the subspecies of this extent.

Acknowledgements

I thank Christoph Plutzar from V.I.N.C.A. (Vienna Institute for Nature Conservation and Analyses) for his help in GIS matters. I also thank BirdLife Austria for providing breeding data of the Avifaunistic Archive, and further Kurt Bauer (Natural History Museum Vienna), Hans Schmid & Niklaus Zbinden (Vogelwarte Sempach), Bernhard Petersen (Leer, Germany) and Vaclav Pavel (Laboratory of Ornithology, Palacký University Olomouc, Czech Republic) for providing breeding data and maps of breeding areas. For bluethroat data and support in the survey (until the development of the model) I thank (in alphabetical order, without mentioning titles) Georg Bieringer, Martin Ellenbroek, Johanna Gressel, Carl-Heinz Gressel, Hemma Gressel, Günter Hauska, Marianne Kern, Josef Kohl, Viktoria Kohl, Bernhard Kohler, Clemens Lunczer, Hans Machart, Klaus Mieslinger, Franz Niederwolfsgruber, Jörg Oberwalder, John Parker, Vaclav Pavel, Katharina Peer, Bernhard Petersen, Jürgen Pollheimer, Hans Schmid, Christian Schulze, Franz Schurtenberger, Anne Sutter, Harald Sutter, Benedikt Thurner, Werner Weißmair, Rainer Windhager, Niklaus Zbinden and many more.

(5)

Erklärung der eigenständigen Abfassung der Arbeit

Ich versichere, diese Master Thesis ohne fremde Hilfe und ohne Verwendung anderer als der angeführten Quellen angefertigt zu haben, und dass die Arbeit in gleicher oder ähnlicher Form noch keiner anderen Prüfungsbehörde vorgelegen hat. Alle Ausführungen der Arbeit, die wörtlich oder sinngemäß übernommen wurden, sind entsprechend gekennzeichnet.

Mag. rer.nat. Ingrid Kohl Wien, April 2009

(6)

Abstract

This approach of habitat suitability modelling can be carried out for most animal and plant species – as far as habitat requirements are known – in every extent – as far as response and predictor data are available in a sufficient spatial and attributive resolution. An interesting case for the habitat suitability modelling is the Redspotted Bluethroat (Luscinia svecica svecica) that was discovered in 1975 in very local areas in Central European mountain ranges (the Alps, the Giant Mountains and the Carpathian Mountains) next to its continuous distribution range in Scandinavia. Some populations resp. breeding sites were found but it is expected that more exist. With known habitat requirements of the study species a GIS based, spatially explicit habitat suitability model was developed for Central Europe to identify potential breeding areas and to aid future surveys, and with this approach a possibility was created to aid later the specific search for new populations with smaller extents in a higher (geometric and attributive) resolution. Former observations of bluethroats in Central Europe in the breeding period were collected, summarized and localized. Sixty variables to land cover, elevation and climate were used to model spatially explicit and GIS-based the habitat suitability for the Redspotted Bluethroat in Central Europe in a resolution of 1000x1000m2. For this habitat suitability model the maximum entropy approach was used. The two distribution fractions of the Eastern and the Western Alps can also be seen in the model. Moreover the two small distribution areas in the Giant Mountains and the High Tatra were recognized suitable. The model result of habitat suitability gives a good overview of the potential distribution of the bluethroat. As support for the field work it is recommended to develop models of smaller extents in a higher geometric and attributive resolution.

Two models were calculated. In the first model breeding data from BidLife Österreich were used to calculate a habitat suitability model for Central Europe, in the second one all known breeding data of Central Europe were used that could be localized in the necessary accuracy. The predictor data were ident in both models. The quality of the second model with breeding data of whole Central Europe was better than the quality of the first model that was calculated only with Austrian breeding data.

(7)

Kurzfassung

Dieser Ansatz der Habitateignungsmodellierung kann für die allermeisten Tier- und Pflanzenarten, sofern Habitatansprüche ausreichend bekannt sind, in jedem beliebigen Extent, sofern Response- und Prädiktordaten in einer ausreichenden Genauigkeit vorhanden sind, durchgeführt werden. Ein interessanter Fall für die Habitateignungsmodellierung ist das Rotsternige Blaukehlchen (Luscinia svecica svecica), das neben seiner kontinuierlichen Verbreitung in Skandinavien im Jahr 1975 an sehr lokalen Standorten in mitteleuropäischen Gebirgen (Alpen, Riesengebirge und Karpaten) entdeckt wurde. Einige Populationen bzw. Brutstandorte wurden entdeckt, aber es wird erwartet, dass es weitere gibt. Anhand bekannter Habitatansprüche der untersuchten Vogelart wurde ein GIS-gestütztes, räumlich explizites Habitateignungsmodell für Mitteleuropa erstellt, um potentielle Brutgebiete zu identifizieren, und mit diesem Lösungsansatz eine Möglichkeit geschaffen, um später in geringeren Extents mit höherer (geometrischer und attributiver) Auflösung die Suche nach Populationen gezielt zu unterstützen. Beobachtungen des Blaukehlchens in Mitteleuropa innerhalb der Brutzeit wurden gesammelt, zusammengefasst und lokalisiert. Sechzig Variablen zu Landbedeckung, Höhe und Klima wurden verwendet, um die Habitateignung für das Rotsternige Blaukehlchen in Mitteleuropa räumlich explizit und GIS-basiert in einer Auflösung von 1000x1000m2 zu modellieren. Der Maximum Entropy Ansatz wurde für die Modellierung verwendet. Die zwei Verbreitungsschwerpunkte in den Ost- und den Westalpen zeigen sich im Modell wieder. Ebenso wurden die zwei kleinen Verbreitungsgebiete im Riesengebirge und in der Hohen Tatra als geeignet erkannt. Das Modellergebnis der Habitateignung gibt einen guten Überblick über die potentielle Verbreitung des Blaukehlchens. Zur Unterstützung für Freilanduntersuchungen werden kleinräumigere Modelle in höherer geometrischer und attributiver Auflösung empfohlen. Es wurden zwei Modelle erstellt.

Im ersten Modell wurden die Brutdaten von BirdLife Österreich verwendet, um ein Habitateignungsmodell für Mitteleuropa zu erstellen, im zweiten wurden alle bekannten Brutdaten Mitteleuropas, die sich in der benötigten Genauigkeit verorten ließen, verwendet. Die Prädiktordaten waren in beiden Modellen ident. Die Qualität des zweiten Modelles mit den Brutdaten aus ganz Mitteleuropa ist höher als die des ersten Modelles, das nur mit österreichischen Brutdaten gerechnet wurde.

(8)

Table of Contents

Page

1. Introduction……….……….………...1

1.1 Species‟ geographic distributions.……...………...……….…….………….1

1.2 Study species………..……….1

1.2.1 History………...………...………….…….………….…...2

1.2.2 Biology………..………..…...4

1.2.3 Habitat………..……….…...4

1.2.4 Action demand………..……...…….….…..………...5

1.3 Motivation………..………….…….………...6

1.4 Objectives…………...………..………..………..7

1.5 Approach...…………....………..………7

1.5.1 Theory and Methods………..………..7

1.5.2 Tool………...……….………...8

1.5.3 Tested area / tested data set………..………….………8

1.6 Expected results………...8

1.7 Topics not treated...………..……....…………....9

1.8 Structure of thesis and intended readership…..…………..………....9

2. Overview of literature……….…………..11

2.1 Breeding data………..………...………...…..11

2.2 Biology of study species………..………...….………11

2.3 Habitat………..………...………11

2.4 Conservation………..………...………...12

2.5 Modelling....………..………...………12

2.6 Maxent………..………...………12

3. Approach...………...13

3.1 Theory………13

3.1.1 Central terms……….………...…13

3.1.2 Existing modelling approaches.……….………….………14

3.2 Methods………….……….………...15

4. Description of the project………...……….…17

4.1 Concept...………...………..…17

4.2 Implementation………...……….…17

4.2.1 Tool: Maxent……….…17

4.2.2 Tested species: The Redspotted Bluethroat...19

(9)

4.2.3 Tested area: Central Europe.…… ………..………...19

4.2.4 Tested data set...22

4.2.4.1 Continuous and categorial data………23

4.2.4.2 Land cover………...23

4.2.4.3 Elevation………..………25

4.2.4.4 Climate……….26

4.2.5 Tested variables...28

5. Results………31

5.1 Habitat parameters………31

5.2 Model quality...………...………….33

5.3 Habitat suitability...………...…………..37

6. Analysis of results……...……….39

7. Summary, Discussion, Prospects.……….43

7.1 Summary………43

7.2 Discussion………..…43

7.3 Prospects………...45

References………...………....47

Appendix A: Environmental data...52 Appendix B: MAXENT model output file...59-71

(10)

List of figures

Page Fig. 1: The Redspotted Bluethroat in Central Europe (picture: Ingrid Kohl, 23.6.2005,

Hundsfeld Moor, Obertauern, province of Salzburg, Austria). ...2 Fig. 2: A coarse overview of the breeding areas of the study species in Central

Europe (source: Cereda et Posse 2003). ...3

Fig. 3: An overview of the European distribution of the Bluethroat: the redspotted (L.sv.sv.) and the whitespotted subspecies (L.sv.cyan.) and further subspecies in France and Spain (L.sv.nam. and L.sv.ssp) (source: Müller 1982; legend added by author). In this early publication only the two Central European

breeding areas Hundsfeld Moor and Giant Mountains were mentioned. ...3 Fig. 4: Typical breeding area: Hundsfeld Moor, Salzburg, Austria (picture: Ingrid

Kohl, 12.8.2005). ...5

Fig. 5: Structure of the Master thesis (arranged by the author). ...10 Fig. 6: Structure of the dissertation with the habitat suitability model as one topic

(arranged by the author). ...10

Fig. 7: Used for MODEL 1: Bluethroat occurrences (red dots) in Central Europe that could be localised in the resolution of 1000mx1000m (sources: breeding data

by the author; elevation data from the SRTM 90m Digital Elevation Model). ...19 Fig. 8: Central European mountain range definition that was finally not used for the

model (defined by the author and the supervisor C. Plutzar). ...21 Fig. 9: CORINE landcover 2000 data set (source: European Environmental Agency

EEA). ...24

Fig. 10: CORINE 2000 (100x100m) and CORINE 1990 (250x250m) cells resampled

to 1000x1000m grid cells (sketches by the author). ...25 Fig. 11: The Global Land Cover 2000 (GLC 2000) data set was not used for

modelling (source: Institute for Environment and Sustainability IES). ...25 Fig. 12: One example for the WORLDCLIM data set for Europe: Mean Temperature

of the Wettest Quarter (0,1°C) (“bio_8”) (source: WORLDCLIM; see also

Hijmans et al. 2005). ...27

Fig. 13: Another example for the WORLDCLIM data set for Austria: Amount of Precipitation in July (“prec_7_eur”) (source: WORLDCLIM; see also

Hijmans et al. 2005). ...27

Fig. 14: The meaningfulness of grid cell size investigating the importance of

environmental data for the study species (source: Kohl 2006a). ...30 Fig. 15: MODEL 1: ROC plot of the model with all breeding data samples (source:

output from MAXENT software; modelled by the author). ...34

(11)

Fig. 16: MODEL 2: ROC plot of the model only with BirdLife data base data

(source. Output from MXENT software; modelled by the author). ...35 Fig. 17: MODEL 1: Omission versus predicted area of the model with all breeding

data (source: output from MAXENT software; modelled by the author). ...35 Fig. 18: MODEL 2: Omission versus predicted area of the model only with BirdLife

data base data (source: output from MAXENT software; modelled by the

author). ...36

Fig. 19: MODEL 1: Habitat suitability for the Redspotted Bluethroat in Central Europe from suitable (red) to unsuitable (grey). The whole map displays the extent of the modelled area (except white areas) (source: output from

MAXENT software; modelled by the author). ...37

Fig. 20: MODEL 2: Habitat suitability for the Redspotted Bluethroat in Central Europe from suitable (red) to unsuitable (grey). The whole map displays the extent of the modelled area (except white areas) (source: output from

MAXENT software; modelled by the author). ...38

Fig. 21: Habitat suitability for the area of Hundsfeld Moor, Obertauern (province of Salzburg, Austria), and surroundings. For the legend see figure 19. (Source:

output from MAXENT software; modelled by the author). ...39 Fig. 22: Habitat suitability for the area of Giant Mountains (Czech Republic and

Poland) and surroundings. For the legend see figure 19. (Source: output from

MAXENT software; modelled by the author). ...40

Fig. 23: Habitat suitability for the area of High Tatra (Slovak Republic and Poland) and surroundings. For the legend see figure 19. (Source: output from

MAXENT software; modelled by the author). ...40

Fig. 24: MODEL 1: response curves of the eight most important variables (source:

output from MAXENT software; modelled by the author). ...41 Fig. 25: MODEL 2: Response curves of the eight most important variables (source:

output from MAXENT software; modelled by the author). ...42

(12)

List of tables

Page

Table 1: Habitat requirements and tested data sets (arranged by the author). ...22

Table 2: Tested data sets (arranged by the author and the supervisor C. Plutzar). ...22

Table 3: CORINE 2000 legend of the for the bluethroat interesting land cover classes (sources: German legend from the European Environmental Agency EEA; English translation by the author). ...24

Table 4a: Tested variables: climate (source: WORLDCLIM; table arranged by the author). ...28

Table 4b: Tested variables: land cover (source: CORINE landcover 2000; table arranged by the author). ...29

Table 4c: Tested variables: elevation (source: SRTM 90m Digital Elevation Data, CGIAR Consortium for Spatial Information; table arranged by the author). ...29

Table 5a: MODEL 1: Variable contributions (model with all breeding data) (source: output from MAXENT software; modelled by the author). ...32

Table 5b: MODEL 2: Variable contributions (model only with BirdLife breeding data) (source: output from MAXENT software; modelled by the author). ...32

Table 6: The quality of models evaluated by AUC values (see Swets 1988 and Hosmer & Lemeshow 2000). ...33

List of Appedix A tables Page Table A: CORINE landcover 2000 legend in German (source: European Environmental Agency EEA). ...52

Table B: CORINE landcover 2000 legend in English (source: European Environmental Agency EEA). ...52

Table C: CORINE landcover 2000 legend in English, extract (source: European Environmental Agency EEA). ...53

Table D: Gridnames of environmental data (prepared for modelling). ...53

Table E: List of ascii files of environmental data (prepared for modelling). ...54

Table F: Data derived from SRTM 90m DEM (prepared for modelling). ...55

Table G: Data derived from WORLDCLIM (prepared for modelling). ...57

List of abbreviations

CORINE Coordinated Information on the Environment

GLC Global Land Cover

(13)

SRTM Shuttle Radar Topographic Mission MAXENT Maximum entropy (software)

(14)

1. Introduction

1.1 Species‘ geographic distributions

Breeding bird atlases asthe EBCC atlas of European breeding birds (Hagemeijer et Blair 1997) show known present distributions, whereas other atlases as Huntley et al. (2007) show the known present distribution, the simulated present distribution for present climate and the potential breeding distribution for the late 21st century. Also Schuster (1996) models potential breeding distributions for all breeding birds of the National park Berchtesgaden (Bayern, Germany).

Gallaun et al. (2005) developed very small scale habitat suitability models for 13 bird species, including the Redspotted Bluethroat, in the Lower Towern with high emphasize on the remote sensing technique. The result for the Redspotted Bluethroat was not published.

Niederberger (2008) models habitat suitability for the Swiss endemic snail species Trochulus biconicus and states that knowledge about rare or cryptic species is often incomplete.

Beard et al. (1999) state that prediction is the more accurate the more frequent a species is. Vice versa prediction is the more inaccurate the rarer a species is.

Guisan et al. (2007) state that the quality of a model is strongly dependent on the modelled species.

The distribution of the Redspotted Bluethroat in Central Europe is very local and it is difficult to survey its occurrence completely. B. Petersen (pers. comm.) carried out meticulous studies of maps of the whole Alps, and after years of searching for the bluethroat, the call for a spatially explicit habitat suitability model became louder.

1.2 Study species

The Redspotted Bluethroat (Luscinia svecica svecica) (see figure 1) is an interesting species for the habitat suitability modelling approach because it colonized Central Europe only a few decades ago and it offers an opportunity to investigate a global phenomenon of changes in distribution and colonization of new areas by fauna and flora

(15)

due to climate resp. global change.

The bluethroat has a transpalearctic distribution and is distinguished into ten subspecies.

The Redspotted Bluethroat (ssp. svevica) has its continuous distribution in Scandinavia, northern Siberia and western Alaska. Highly isolated populations exist in Central Europe, whereas it is discussed if it has recently colonized Central Europe or if it is a glacial relic. The author claims that the truth lies in between.

Fig. 1: The Redspotted Bluethroat in Central Europe (picture: Ingrid Kohl, 23.6.2005, Hundsfeld Moor, Obertauern, province of Salzburg, Austria).

1.2.1 History

The Redspotted Bluethroat (Luscinia svecica svecica) was first discovered as breeding bird in Central Europe in 1975 by Mrs Johanna Gressel (Gressel 1991) and more discoveries followed (e.g. Flore 2000 & 2001, Miles & Formánek 1988). Next to its continuous distribution in Scandinavia, isolated breeding occurrences can be observed since the 1970ies in Central European mountain ranges – the Alps, the Giant Mountains and the High Tatra – with a non inhabited gap of 1000 km between the continuous distribution range and the local breeding occurrences. It is expected that more populations exist (Gressel 2001). Due to the roughness and steepness of some mountains, it could be that some of these mountains were maybe not reached by many ornithologists, and if it would have been so, maybe not at the right time. The month of June is the time of display when males are singing – almost the only time a population can be discovered. Moreover, a potential breeding area has to be controlled more often, because population dynamics are high. That means, if the species is not breeding in an area one year, it can be breeding there a few years later.

(16)

Cereda & Posse (2003) tried to give a coarse overview of the breeding areas of the study species in Central Europe (see figure 2).

Fig. 2: A coarse overview of the breeding areas of the study species in Central Europe (source: Cereda et Posse 2003).

In figure 3 (Müller 1982) it is shown that the next continuous breeding occurrence of the species is located in Scandinavia (marked with L.sv.sv.). The breeding areas in Central Europe are very local and confined to small areas. The whitespotted subspecies can be found in lowland areas (L.sv.cyan.) and further European subspecies in France and Spain (L.sv.nam. and L.sv.ssp.).

Fig. 3: An overview of the European distribution of the Bluethroat: the redspotted (L.sv.sv.) and the whitespotted subspecies (L.sv.cyan.) and further subspecies in France and Spain (L.sv.nam. and L.sv.ssp) (source: Müller 1982; legend added by author). In this early publication only the two Central European breeding areas Hundsfeld Moor and Giant Mountains were mentioned.

(17)

1.2.2 Biology

The Bluethroat is a transpalaearctic faunal element (that means it is distributed over Eurasia). The redspotted subspecies - one of 10 subspecies – is the nominate subspecies of the bluethroat and has its distribution range over the Northern Palaearctic - Scandinavia, Siberia and a small part of Alaska. The population estimation for Norway is on average 450.000 (200.000-1 Mio) individuals, in Sweden 180.000 (100.000- 300.000) and in Finland 150.000 (100.000-200.000) (Meijer & Štastn 1997). At least since the Seventies it also breeds in Central European Mountain ranges. In the 20th century and end of the 19th it was observed and ringed regularly as migrant (often the subspecies are not distinguished) – also in areas were the Bluethroat could be found breeding in the last few decades. Frühauf (2005) estimates the Austrian part of the Central European population at much more than 50 percent.

The Bluethroat is a thrush and closely related to the nightingale (Luscinia megarhynchos). The systematic position of the bluethroat shows difficulties since a long time (Franz 1998) – it lies in the closer relational area with thrushes, flycatchers and warblers. The Bluethroat is insectivore, feeds additionally on berries, breeds on the ground and is long-term migrant. Franz (1998) explains the difficulty of distinguishing the winter distribution of the subspecies. In India four subspecies occur together in their wintering area and in Israel five subspecies are described as migrants (Franz 1998).

Because in Europe only the redspotted subspecies is a long term migrant, it is supposed that only the redspotted subspecies (Central European and Scandinavian populations) winters south of the Sahara. Bluethroats also winter at the Persian Gulf and the Gulf of Oman; it is speculated that also here European bluethroats winter.

1.2.3 Habitat

Essential for the breeding occurrence of the Redspotted Bluethroat is structured scrub vegetation in a peat bog or swamp: open areas (meadows) as foraging grounds and dense vegetated areas (scrubs) for breeding and to hide. In Central Europe only breeding areas in silicious/crystalline areas are known. Tree patches are avoided. Only areas with single trees are occupied; trees can be used as songposts. Wet meadows with low grass are used as foraging grounds. In my Master thesis in Ecology at the University of

(18)

Vienna (Kohl 2006a & 2006b & 2008, Kohl & Schulze 2006a & 2006b) I could collect and analyse several habitat parameters of the Hundsfeld Moor in Obertauern (province Salzburg, Austria) where a population of 10 to 20 territories can be found. It is the prime example for a breeding area (see figure 4). Parameters like the wetness, the shrub vegetation structure and special habitat elements turned out to be important as expected.

Järvinen & Pietiäinen (1983) describe habitat of the bluethroat in Finland where the habitat is partly different but the same in the structure.

Fig. 4: Typical breeding area: Hundsfeld Moor, Salzburg, Austria (picture: Ingrid Kohl, 12.8.2005).

1.2.4 Action demand

In the Red List of Threatened Species of Austria Frühauf (2005) claims monitoring, the survey of unknown breeding areas, the estimation of population size and the conservation of the potential breeding areas of the Redspotted Bluethroat (wet areas in alpine regions). Primarily the habitat has to be protected. Regular monitoring of habitat should show changes and make an early management possible.

For species like the bluethroat (e.g. the Curlew Numenius arquata) that always occupy the same territory every year, it is possible that one generation occupies one territory also in suboptimal habitat ever and ever again, has negative breeding success because of the negative changes of habitat and in the next generation the territory is not occupied

(19)

any more, or the adults have breeding success but the Juveniles or other adults do not occupy the territory anymore in the next breeding season.

The topic is not only relevant for Austria (the Bluethroat was declared as “the bird of Austria” by the European Union, Ranner 2005), moreover the subspecies offers an opportunity to investigate a global phenomenon of changes in distribution and colonization of new areas by fauna and flora.

1.3 Motivation

I wanted to know if it is possible to create a habitat suitability map for the Redspotted Bluethroat for Central Europe with known breeding locations and environmental data like land cover, climate and elevation. The questions were:

 Is it possible to develop a habitat suitability model for the very local occurring subspecies?

 Which data are available?

 In which resolution are they available?

 For which extent a model can be developed?

 Has a Central European model a too weak resolution?

 etc.

Furthermore, suitable breeding areas of bluethroats located in basins in alpine regions are supposed to be threatened, because these areas are also preferred by winter tourism, expanding since the last few decades. Flagship species like the Redspotted Bluethroat can help to protect these endangered habitats. One example is the Hundsfeld Moor (province Salzburg, Austria) where a very special area with many protected and endangered species, different mire types and habitat structures, got its conservation status because of the bluethroat. Research and the enlargement of knowledge can help to find arguments for the protection of habitat. That is why the further intention is the application of the outcomes in conservation and to build a basis for a following better understanding in phenomena happening due to global changes.

(20)

1.4 Objectives

The main research question is, how the situation of the habitat suitability for the

Redspotted Bluethroat is in Central Europe, and how the modelled habiat suitability and the actual distribution can be compared.

The second research question is, how it could be possible to realize a Central European habitat suitability model resp. potential distribution model – spatially explicit and with satisfying geometric and attributive accuracy, with available information and data basis, no budget and the limited time for the Master thesis. And if it is not possible to model habitat suitability for the whole extent of Central Europe, it has to be proofed in which extent or for which countries a model can be developed.

1.5 Approach

1.5.1 Theory and Methods

Breeding data samples were collected, environmental data obtained and resampled to the same resolution. Maxent recognizes the most important variables for the occurrence of the study species and creates dependent on the importance of the variables, suitability values for each raster cell.

- Maxent was developed especially for species distribution modelling. In one modelling step the potential distribution for one or even more species can be modelled.

- The chosen resolution is 1000m (raster cells of 1000x1000m).

- Land cover data were obtained from CORINE landcover 2000 and land cover data of Switzerland from CORINE landcover 1990.

- CORINE 2000 was chosen to gain a better resolution than GLC2000 (Global Land Cover 2000).

- SRTM 90m Digital Elevation Model was chosen for elevation data and the derived wetness index.

- WORLDCLIM seemed to be the ideal source for climate data.

The work with MODIS data would have been too time consuming.

For more information see table 2, page 22.

(21)

1.5.2 Tool

The software MAXENT turned out to be the ideal solution for my modelling demands.

With the Maximum entropy approach (García Márquez 2006, Phillips et al. 2004, Phillips et al. 2006, Phillips 2006) the habitat suitability is modelled for the study species in Central Europe. Maxent version 3.0 beta was used. GIS software is ArcView 3.2, ArcGIS 9.2 and SAGA GIS.

1.5.3 Tested area / tested data set

For Central Europe a grid with a resolution of 1000 m x 1000 m was created and all input data sets were adapted to this reference raster. Data sets of land cover, elevation and climate were obtained to develop a habitat suitability model by means of all known habitat requirements of the study species.

1.6 Expected results

Within the phenomenon of the Redspotted Bluethroat in Central Europe that was discovered more than three decades ago, several questions arise:

Which habitat requirements does the study species have in Central Europe?

How well does the model represent reality?

How useful is the model for a survey?

Which shortcomings does the model have?

Which areas in Central Europe are especially suitable?

One aim of developing the spatially explicit habitat suitability model is to identify potential breeding areas of the Redspotted Bluethroat and to deliver basic information for providing help for decision-making for the protection of alpine habitats. The model could also provide a basis for an alp-wide survey of an ornithologists‟ network (Kohl 2007a, Kohl 2007b) that could find further populations, confirm old ones and estimate the alpine population. Further protected areas could be expelled for the protection of the population and the habitat and the quality of habitat models as predictors in avifauna could be clarified.

(22)

1.7 Topics not treated

My intention in this thesis is to describe how I proceeded on the way from known populations of the bird to a habitat suitability model for Central Europe. Models in higher resolutions need breeding data as well as environmental data in a higher resolution. For example an Austrian model in a resolution of 100x100m2 requires breeding data that can be located in this resolution as well as land cover, climate, elevation and in the best case also geologic data. These data have to be prepared and modelled. Efficient and fast hardware would be a benefit. And time and money are limiting variables for this work. For this reasons I decided to confine this work to the Central European model. The model is to aid the survey that is not part of the thesis.

Also it is not an aim of this work to give prognoses for climate change resp. global change.

1.8 Structure of thesis and intended readership

The used diction is intended for a circle of readers with biological background. The requirements of the Redspotted Bluethroat are described, as well as environmental data (like land cover, climate and elevation data as well as the derived variables such as the wetness index from the elevation model), the software Maxent and the results of the model.

The survey is not part of the Master thesis but it is shown that the feedback of the survey should enlarge knowledge about habitat requirements of the study species and test and improve the quality of the habitat model.

In the future the survey and the model could give permanent feedback and improve each other step by step (figure 5). The Master thesis is embedded in my dissertation that should clarify questions about the phenomenon of the Redspotted Bluethroat in Central Europe (figure 6).

(23)

Fig. 5: Structure of the Master thesis (arranged by the author).

Fig. 6: Structure of the dissertation with the habitat suitability model as one topic (arranged by the author).

(24)

2. Overview of literature

For more literature see the citations in the other chapters. Only a few examples are given here.

2.1 Breeding data

Cereda & Posse (2003) give a coarse overview of the breeding areas in Central Europe.

Gressel (1991) writes about the discovery of the breeding occurrence, Flore (2000 &

2001) gives insights in the second largest alpine population, and Miles & Formánek (1988) and Pavel & Chutny (2007) write about the study species in the Krkonoše National Park in the Giant Mountains.

2.2 Biology of the study species

Gressel (2001) published in the journal Monticola reports of migrating bluethroats in Austria of a famous ornithologist at the end of the 19th century. He claimed that he would not be surprised if a breeding occurrence would be found in the future.

Franz (1998) describes the general biology of the bluethroat and its population increase.

Meijer & Stastny (1997) give a summary the European situation of distribution and abundance of the bluethroat in the EBBC Atlas of European Breeding Birds.

2.3 Habitat

In my Master thesis in Ecology at the University of Vienna (Kohl 2006a & 2006b &

2008, Kohl & Schulze 2006a & 2006b) I could collect and analyse several habitat parameters.

Järvinen & Pietiäinen (1983) describe habitat of the bluethroat in Finland.

(25)

2.4 Conservation

Frühauf (2005) describes in the Red List of Threatened Species of Austria (in the chapter Red List of Breeding Birds (Aves) of Austria) the action demand for the Redspotted Bluethroat.

Ranner (2005) describes characteristics of the Redspotted Bluethroat in the Report on the Birds Directive of the European Commission where the bluethroat was nominated as

“The bird of Austria”.

2.5 Modelling

Predictive modelling is described in Elith et al. (2006), Guisan & Thuiller (2005) and in Guisan & Zimmermann (2000). In the national park Berchtesgaden in Germany accurate predictive models were developed for all occurring bird species (Schuster 1996). Gallaun et al. (2005) planned to develop an accurate predictive model for the Redspotted Bluethroat in a very small area in the Niedere Tauern region.

2.6 Maxent

Maxent is described by its developers (Phillips et al. 2004, Phillips et al. 2006, Phillips 2006) that give insights and instructions for the use of the software, and by a user (García Márquez 2006) that models potential distributions of herpetofaunal species for Italy and Europe.

(26)

3. Approach

3.1 Theory

3.1.1 Central terms

- Presence and absence of species

Presence-only approaches are recommended by Hirzel et al. (2002) in situations where absence data are not available (many data banks), unreliable (most cryptic or rare species) or meaningless (invaders). Hirzel et al. use a multivariate approach for modelling the potential distribution of the reintroduced alpine ibex in Switzerland that has not yet recolonized its entire range. The situation of distribution of the alpine ibex is comparable to the bluethroat that is supposed to breed only in a minority of suitable areas.

- Entropy approach

The software MAXENT creates probability distributions from incomplete information.

The origin of MAXENT lies in linguistic science and was established in ecology by Philips et al. (2004). MAXENT tries to find the probability distribution of the maximum entropy, that means the distribution that is most continuous distributed over the study area – only restricted by the environmental conditions found at the actual distribution areas of the presence data (Philips et al. 2004). MAXENT models with presence data and does not need absence data. Pseudo-absences are created in the modelling procedure. Model result is a map with values that show a relative habitat suitability.

Elith et al. (2006) state that MAXENT shows very good results compared to other modelling approaches.

- Commission und Omission Errors

Miller et al. (2007) describe two kinds of modelling errors that can occur: omission errors in underestimated areas and commission errors in overestimated areas. But that

(27)

phenomenon also occurs in the breeding distribution of species. Species can breed in less suitable areas and be absent in areas suitable to their habitat requirements. This phenomenon is discribed by Jenkins et al. (2003) with the example of populations of the endangered American Cape Sable seaside sparrow (Ammodramus maritimus mirabilis), a subspecies of the seaside sparrow (Ammodramus maritimus). Some suitable habitat does not contain birds, what is described as commission errors. In this case the model predicts birds where they are absent. The other case are omission errors where some birds remain in unsuitable habitat that was formerly suitable resulting from the species„

high site fidelity – not contributing tot he species„ survival. Here the model omits birds from places where they actually occur. Dependent on the biology and the physiology of a bird species, it does not occur in a suitable area e.g. because of its history of distribution, whereas it occurs in less suitable areas or breeds there as an exception, because it came there through circumstances, and an occurrence and a brood is maybe possible dependent on its physiology.

3.1.2 Existing modelling approaches

Different approaches were possible to solve the problem.

The first step is the summary and localisation of available breeding data and the sighting of relevant and available GIS data bases.

Data that were available were:

 CORINE 2000 (100m) – Coordinate Information on the Environment

 Global Land Cover 2000

 MODIS - Moderate Resolution Imaging Spectroradiometer

 SRTM 90m Digital Elevation Data (DED)

 WORLDCLIM

(All except the GLC 2000 and the MODIS data set were used.)

For models with higher resolution also special data like Austrian or Swiss mires could have been used.

(28)

More approaches were possible:

 Top-down, e.g. expert model

 Bottom-up, e.g. logistic regression, GLM Generalized Linear Models, GAM Generalized Additive Models, Environmental Envelopes

(The used approach was kind of a mixture. It was not only a bottom-up model but also by the choice of the environmental data a top-down model.)

Possible study areas were (see also figure 5, page 10):

- Central Europe

- Central European mountain ranges (Alps, Giant Mountains, Carpathian Mountains)

- The Alp countries - The Alps

- Austria and Switzerland - Austrian and Swiss Alps - Austria

- The Austrian Alps

(For the thesis the maximum possible extent of Central Europe was chosen and could be realized.)

Spatially explicit habitat suitability modelling was carried out with the maximum entropy approach.

3.2 Methods

Former observations of bluethroats in the Alps in the breeding period were collected and summarized. Sixty variables were used to model spatially explicit and GIS-based habitat suitability for the Redspotted Bluethroat in Central Europe. The model is supposed to identify potential breeding areas and aid the survey in the breeding season 2008. For this habitat suitability model the maximum entropy approach was used.

Bluethroat breeding sites that could be located in an accuracy of 1000x1000m2 were mapped with the software ArcGIS. Dealing with data of the different countries with different projections and data sources was a logistic challenge.

(29)

Environmental data were obtained and derived from the SRTM 90 m Digital Elevation Model, CORINE landcover 2000, CORINE landcover 1990 and WORLDCLIM (see chapter 4).

(30)

4. Description of the project

4.1 Concept

In many cases of potential distribution modelling presence as well as absence data are required. In this case habitat suitability resp. potential distribution is modelled for the Redspotted bluethroat in an approach that only requires presence data.

Two modelling steps were carried out – both with the same set of environmental data to enable comparability between the two models based on different sets of breeding data samples.

MODEL 1: For MODEL 1 all breeding data samples that could be localized in the required accuracy were used.

MODEL 2: For MODEL 2 only breeding data from the BirdLife breeding bird data base (data from 1989 onwards) were used.

4.2 Implementation

4.2.1 Tool: Maxent

The software Maxent 3.0, Version 3.0-beta, July 2007, (Maximum Entropy Modeling of Species Geographic Distributions) was used.

Recent modelling techniques to predict species‟ distributions are discussed in Elith et al.

(2006). To compute a habitat suitability map for the Redspotted Bluethroat the maximum entropy approach (MAXENT, Phillips et al. 2004, Phillips et al. 2006, Phillips 2006, García Márquez 2006, http://www.cs.princeton.edu/~schapire/maxent/, 2.9.2008) is used.

The concept of entropy in information theory was described by Shannon (1948) as a

“measure of how much „choice‟ is involved in the selection of an event.” To approximate an unknown probability distribution the maximum principle “ensures that

(31)

the approximation satisfies any constraint on the unknown distribution, that we are aware of, and that subject to those constraints, the distribution should have maximum entropy” (Jaynes 1957). One major advantage of Maxent is that it uses presence-only data sets, absence data are not required.

Niederberger (2008) carried out four levels of modelling with MAXENT. From step to step he reduced variables from 18 to three to exclude possible correlations. The model with three environmental parameters had the worst quality due to lack of information, whereas the model after the first reduction of variables has the best quality. In his study MAXENT and BRT (Boosted Regression Trees) show better prediction results than the GLM approach.

For this model I applied the Maxent default settings and used 10 percent as random test percentage to validate the result of the model.

Settings

Random test percentage 10

Regularization multiplier 1

Maximum iterations 500

Convergence threshold 0,00001

Max number of background points 10.000

Already Jaynes (1957) describes the creation of probability distributions on basis of incomplete knowledge and the statistic inference of the “maximum entropy estimate”:

“Information theory provides a constructive criterion for setting up probability distributions on the basis of partial knowledge, and leads to a type of statistical inference which is called the maximum-entropy estimate. It is the least biased estimate possible on the given information; i.e., it is maximally noncommittal with regard to missing information. If one considers statistical mechanics as a form of statistical inference rather than as a physical theory, it is found that the usual computational rules, starting with the determination of the partition function, are an immediate consequence of the maximum-entropy principle. In the resulting "subjective statistical mechanics," the usual rules are thus justified independently of any physical argument, and in particular independently of experimental verification; whether or not the results agree with experiment, they still represent the best estimates that could have been made on the basis of the information available.

It is concluded that statistical mechanics need not be regarded as a physical theory dependent for its validity on the truth of additional assumptions not contained in the laws of mechanics (such as ergodicity, metric transitivity, equal a priori probabilities, etc.). Furthermore, it is possible to maintain a sharp distinction between its physical and statistical aspects. The former consists only of the correct enumeration of the states of a system and their properties; the latter is a straightforward example of statistical inference.”

(32)

4.2.2 Tested species: The Redspotted Bluethroat (Luscinia svecica svecica)

All breeding data I could collect from 2002 to 2007 I localized if possible in an accuracy of 1000x1000m2 (figure 7). Different projections were necessary to do so: For Austria BMN 28, BMN 31 and the geographic projection were used, and for Switzerland and the Czech Republic the Lambert Equal Area Azimuthal Projection. In the end all samples were reprojected to the last mentioned Lambert projection, unified to one shapefile and transformed to an ascii file.

Fig. 7: Used for MODEL 1: Bluethroat occurrences (red dots) in Central Europe that could be localised in the resolution of 1000mx1000m (sources: breeding data by the author; elevation data from the SRTM 90m Digital Elevation Model).

4.2.3 Tested area: Central Europe

At the first working step the extent of the study area was discussed (see chapter 3.1.2, page 14).

No political border or general definition for Central Europe exists, that is why it was quite difficult to define a study area.

(33)

Because computational effort increases exponentially with growing extent, it was discussed to lay a mask over Central European mountain ranges. Chosen were areas of an altitude of 900 m a.s. or higher of Central European countries or mountain ranges as the Alps, the Giant Mountains and the Carpathian Mountains. Island polygons were removed, the area was buffered to minimize the border effect. Through the moving window method also cells outside the study area are investigated, and if the neighbouring cells include e.g. „no data‟ it would influence the cells inside negatively.

The establishment of core and border zones exclude the border effect completely. Island polygons smaller than 800sqkm were removed, so that only ten polygons remained.

Gaps resp. islands inside the polygons were also removed. (Polygons covering gaps were digitized and with the union function one uniform polygon was produced. Another possibility is to create a large buffer to both sides – to the inside and the outside and to delete the buffer to the outside and keep the buffer to the inside that covers islands inside the polygon.) Contours of polygons – pixelated because of producing process – were smoothed.

Seven polygons remained:

(1) The Alps and Bacher mountains (2) Carpathian mountains

(3)Siebenbürgener Erzgebirge

(4)Tatra (multipart polygon in 3 parts) (5) Gesenke = Altvater mountains (6) Giant mountains

(7) Erzgebirge

In the beginning of the considerations for the model I thought of defining Central European mountain ranges (figure 8) to have lower computational effort. Mountains were defined as areas with an altitude of 900 m a.s. or higher. A buffer was laid around the resulting areas.

It would eventually not have been necessary to model the two most Eastern polygons of the Eastern Carpathians (in Southeastern Poland, Eastern Slovakia, Ukraine and Romania; including the Ukrainian Wooded Carpathians), the Southern Carpathian mountains (in Romania and Serbia) (see

(34)

http://en.wikipedia.org/wiki/Carpathian_Mountains) and the Transylvanian Ore Mountains, with no known occurrence of the study species. Land cover data of Ukraine (see figure xx) are missing, but the most important area (of the Alps, the Giant mountains and the High Tatra) are covered.

The Western Carpathians are covering the Czech Republic, Poland, Slovakia and Hungary (see http://en.wikipedia.org/wiki/Carpathian_Mountains).

Fig. 8: Central European mountain range definition that was finally not used for the model (defined by the author and the supervisor C. Plutzar).

However, at least I decided to try to model the whole area of Central Europe. The extent can be seen in figures 19 and 20 (page 37 and 38). For Central Europe a raster with a resolution of 1000x1000m2 was created and all input data sets were adapted to this reference raster.

With the study area extent covering wide parts of Central Europe, clear definitions of mountain ranges (altitude and size) and Central Europe (political borders etc.) were not necessary anymore.

(35)

4.2.4 Tested data sets

In my Master thesis in Ecology at the University of Vienna (Kohl 2006a & 2006b &

2008, Kohl & Schulze 2006a & 2006b) I collected and analysed several parameters to habitat of the study species in Hundsfeld Moor in Obertauern (province Salzburg, Austria; Kohl & Gressel (in press)). Järvinen & Pietiäinen (1983) describes habitat of the bluethroat in Finland that is partly similar to the alpine habitat and partly different.

Breeding data of the bluethroat were collected from different sources (BirdLife Österreich, maps provided by Vogelwarte Sempach, B. Petersen and V. Pavel, pers.

comm., pers. obs.) leading to a total number of 98 observation points covering 92 raster cells.

To cover the supposed and known habitat requirements of the bird following data sets are used (table 1 and 2).

Table 1: Habitat requirements and tested data sets (arranged by the author).

Habitat requirements Required data sets Altitude around 1600-2300 m a.s. Digital Elevation data

Loose scrub vegetation Land cover data

Wetland Wetness data (derived from Elevation Model)

Forage, wetness... Climate data

Table 2: Tested data sets (arranged by the author and the supervisor C. Plutzar).

Data set Source Resolution Coverage Type

CORINE Landcover 2000

Coordinated Information on the Environment

European

Environment Agency

100 meters Study area except

Switzerland and Ukraine

Land Cover

CORINE Landcover 1990

Switzerland

European

Environment Agency

250 meters Switzerland Land Cover

SRTM

Shuttle Radar Topographic Mission

CGIAR-CSI 3 arc seconds Study area Digital Elevation Model WORLDCLIM

(Hijmans et al. 2005)

University of California, Berkeley

30 arc seconds Study area Climate

All data sets were resampled to the same resolution of 1000x1000m2.

(36)

4.2.4.1 Continuous and categorial data

All used environmental data were continuous data. The MAXENT approach allows continuous as well as categorical data, such as land cover data. However, for the bluethroat model all categorial land cover data were changed to continuous data. 100 CORINE land cover grid cells (100x100m2) fit into one model grid cell (1000x1000m2).

For each land cover class the percent number for each model grid cell was established;

e.g. if coniferous forest covered a half of the model grid cell, the land cover class

“coniferous forest” got the value 50.

4.2.4.2 Land cover

For most countries of the study area the CORINE landcover 2000 data set (European Environment Agency; see figure 9 and table 3), with a spatial resolution of 100x100m2 and distinguishing 37 landcover classes, is available

(http://dataservice.eea.europa.eu/dataservice/metadetails.asp?id=950, 2.9.2008).

Unfortunately this data set does not cover Switzerland and the Ukraine. For Switzerland the European Environment Agency offers a CORINE data set for 1990 with 14 landcover classes and a resolution of 250x250m2

(http://dataservice.eea.europa.eu/dataservice/metadetails.asp?id=808, 2.9.2008; Nippel

& Klingl 1998). For the Ukraine no land cover data comparable to CORINE landcover could be obtained, so land cover data for Ukraine are missing in the model.

In order to harmonize the two CORINE data sets the number of landcover classes were reduced to 9. The Swiss data set was resampled to a resolution of 100 meters and both data sets were reclassified to the reclassification scheme. Then the percentage per 1km² raster cell for each landcover class was calculated.

(37)

Fig. 9: CORINE landcover 2000 data set (source: European Environmental Agency EEA).

Table 3: CORINE 2000 legend of the for the bluethroat interesting land cover classes (sources: German legend from the European Environmental Agency EEA; English translation by the author).

German legend English translation

Forest and semi natural areas

Scrub and/or herbaceous vegetation associations 321 Natural grassland

322 Moors and heathland 324 Transitional woodland-shrub Wetlands

Inland wetlands 411 Inland marshes 412 Peat bogs No data/Unclassified

The whole legend of the CORINE landcover 2000 data set you can find in the Appendix A (tables A, B, C, pages 52-53).

For the Redspotted Bluethroat five categories could be interesting:

(321) Natural Grassland (322) Moors and heathland

(323) Transitional woodland-shrub (411) Inland marshes

(412) Peat bogs

(38)

Fig. 10: CORINE 2000 (100x100m) and CORINE 1990 (250x250m) cells resampled to 1000x1000m grid cells (sketches by the author).

The Global Land Cover GLC 2000 data set was not used for modelling, the CORINE landcover 2000 data set was preferred for modelling to gain a better resolution.

Fig. 11: The Global Land Cover 2000 (GLC 2000) data set was not used for modelling (source: Institute for Environment and Sustainability IES).

4.2.4.3 Elevation

From the SRTM 90 meter Digital Elevation Data from the CGIAR Consortium for Spatial Information (http://srtm.csi.cgiar.org/, 2.9.2008) was used as a basis to calculate means and standard deviations of several derived variables (e.g. wetness index) for each 1000x1000m² raster cell. For the preparation of elevation data also the software SAGA GIS was used (Olaya 2004).

(39)

These scenes that were extracted were 37.4, 38.3, 38.4, 39.3, 39.4, 38.3, 39.2, 39.3, 40.3, 40.2, 40.3, 40.4, 41.3, 42.3, 42.4 and 41.4.

Variables derived were morphological parameters such as curvature and the convergence index.

A curvature of a zero value represents an even surface. The more positive the value is, the more hunchbacked is the surface.

The convergence index is also a relief parameter and another measure for the curvature of the surface.

4.2.4.4 Climate

Climate data were obtained from the WORLDCLIM data set provided by the University of California, Berkeley (Hijmans et al. 2005, http://www.worldclim.org/, 2.9.2008).

Hijmans et al. (2005) describe their very high resolution interpolated climate surface for global land areas. This data set has a spatial resolution of 30 arc seconds and was resampled to fit the 1km² raster of the study area.

WORLDCLIM considers 19 bioclimatic variables (see chapter 4.2.5) and for all months precipitation, minimum and maximum temperature (e.g. figures 12 and 13).

Europe was cut out of the WORLDCLIM data with a mask with Avenue.

WORLDCLIM data had a geographic projection.

(40)

Fig. 12: One example for the WORLDCLIM data set for Europe: Mean Temperature of the Wettest Quarter (0,1°C) (“bio_8”) (source: WORLDCLIM; see also Hijmans et al. 2005).

Fig. 13: Another example for the WORLDCLIM data set for Austria: Amount of Precipitation in July (“prec_7_eur”) (source: WORLDCLIM; see also Hijmans et al. 2005).

(41)

4.2.5 Tested variables

Variables that could be of interest for the bluethroat could be vegetation structure as open shrub vegetation, humidity resp. wetness, temperature, snow cover and precipitation, and variables that can influence and cause the primary variables of interest such as altitude, morphology of surface resp. slope and geology.

Next to the base variables, meaningful variables were derived from these base data sets, e.g. aspect, slope, curvature from the Digital Elevation Model. Sixty variables (tables 4a, 4b and 4c) were used to model - GIS-based and spatially explicit - habitat suitability for the Redspotted Bluethroat in Central Europe.

Table 4a: Tested variables: climate (source: WORLDCLIM; table arranged by the author).

Climate

prepared (55) tested (31) Precipitation Jan-

Dec

12 Precipitation May- Aug

4 Minimum

Temperature Jan- Dec

12 Minimum

Temperature May- Aug

4

Maximum Temperature Jan- Dec

12 Maximum

Temperature May- Aug

4

bio 1 Annual Mean Temperature bio 2 Men Diurnal Range

bio 3 Isothermality

bio 4 Temperature Seasonality

bio 5 Maximum Temperature of Warmest Month bio 6 Minimum Temperature of Coldest Month bio 7 Temperature Annual Range

bio 8 Mean Temperature of Wettest Quarter bio 9 Mean Temperature of Driest Quarter bio 10 Mean Temperature of Warmest Quarter bio 11 Mean Temperature of Coldest Quarter bio 12 Annual Precipitation

bio 13 Precipitation of Wettest Month bio 14 Precipitation of Driest Month bio 15 Precipitation Seasonality

bio 16 Precipitation of Wettest Quarter bio 17 Precipitation of Driest Quarter bio 18 Precipitation of Warmest Quarter bio 19 Precipitation of Coldest Quarter

Referenzen

ÄHNLICHE DOKUMENTE

Quo pra:viia, Dux > tanquam Princeps prudentisfimus ,f}>c tamen frctus, S vecos fidcm datam, ftipulationes, m imim & ligilla ^non fccus ac tempore Regis

The electricity impact submodel calculates a set of envi- ronmental impacts associated with model power plants of five types: coal, pressurized water reactor (PWR) ,?. boiling

Algorithm, Hadoop, Spark, framework, MapReduce, classification, parallel, k-nearest neighbor’s, naïve Bayesian, Clara, cluster, Tartu University... Spark raamistiku

Motivation Data lifecycle Metadata Publish Data Summary.!. Science

The main concept of the project is to create a Linked Open Data (SOD) infrastructure (including software tools and data sets) fed by public and freely

(3) Factors influencing the reliability of the majority opinion: When very high resolution images were available on Google Earth, the reliability of the majority opinion

The results obtained with the conversion are shown in Figures 2 and 3, respectively for the Paris and Milan areas. Results for the Paris study area. On the left results obtained

Juan Carlos Laso Bayas, Linda See, Steffen Fritz, Tobias Sturn, Mathias Karner, Christoph Perger, Martina Duerauer, Thomas Mondel, Dahlia Domian, Inian Moorthy, Ian McCallum,