13
ORIGINAL ARTICLE Suitable, reachable but not colonised: seasonal niche duality in an endemic mountainous songbird Jan O. Engler Dennis Ro ¨dder Darius Stiels Marc I. Fo ¨rschler Received: 29 July 2013 / Revised: 12 January 2014 / Accepted: 28 January 2014 Ó Dt. Ornithologen-Gesellschaft e.V. 2014 Abstract The realized distribution of animals is often delimited by climatic factors which define, next to the specific habitat and food availability, their species-specific potential distribution. We studied the environmental limi- tations affecting the realized breeding and wintering dis- tributions of the Citril Finch (Carduelis citrinella), one of the few endemic bird species of European mountain ranges. To assess the environmental limits that shape the seasonal distribution, we used species distribution models (SDMs) derived from macroclimate in combination with land cover information. Our data suggest a high congruence between the potential modelled breeding distribution of the Citril Finch and the currently known breeding sites, indicating a high level of niche filling. The unusual absence in several suitable breeding habitats at the eastern and northern range limit (Eastern Alps, Carpathians, Bavarian Forest, Harz Mountains, Fichtelgebirge, Krkonos ˇe Mountains) is likely linked to a combination of both missing resources and restricted physiological migration capacities from the available wintering grounds. Since the accomplished migratory distances hardly exceed more than 500 km, it seems likely that the distance to the main wintering areas is too large for exceeding eastern and northern range limits. We discuss the differences in SDM outcomes when including distal predictor variables instead of using proxi- mal predictors alone, and highlight the importance of considering a seasonal niche duality to gain more insights into complex range effects in species with seasonal ranges. Keywords Carduelis citrinella Á Ecological niche Á MaxEnt Á Species distribution model (SDM) Zusammenfassung Geeignet, erreichbar aber unbesiedelt: saisonale Nisc- hendualita ¨t bei einem endemischen Singvogel europa ¨i- scher Gebirgsregionen Die realisierte Verbreitung von Arten wird oft durch kli- matische Faktoren begrenzt, die gemeinsam mit dem cha- rakteristischen Habitat und der Nahrungsverfu ¨gbarkeit die artspezifische potentielle Verbreitung definiert. In dieser Studie untersuchten wir die umweltbedingten Faktoren, welche das Brut- und Winterareal des Zitronenzeisigs (Carduelis citrinella) limitieren. Um die begrenzenden Faktoren der saisonalen Verbreitungen zu quantifizieren, nutzten wir Artverbreitungsmodelle basierend auf biokli- matischen Variablen in Kombination mit Land- nutzungsinformationen. Die Ergebnisse zeigten eine hohe U ¨ bereinstimmung der modellierten potentiellen Verbrei- tung mit der derzeitig bekannten Verbreitung der Art, was auf einen hohen Grad an Nischenfu ¨llung (,,niche filling‘‘) schließen la ¨sst. Klimatisch geeignete, jedoch unbesiedelte Brutgebiete entlang des no ¨rdlich und o ¨stlich gelegenen Arealrandes (o ¨stliche Alpen, Karpaten, Bayerischer Wald, Electronic supplementary material The online version of this article (doi:10.1007/s10336-014-1049-5) contains supplementary material, which is available to authorized users. J. O. Engler (&) Á D. Ro ¨dder Á D. Stiels Zoological Research Museum Alexander Koenig, Adenauerallee 160, 53113 Bonn, Germany e-mail: [email protected] J. O. Engler Department of Wildlife Management, University of Go ¨ttingen, 37077 Go ¨ttingen, Germany M. I. Fo ¨rschler Department of Monitoring and Research, National Park Black Forest, Schwarzwaldhochstraße 2, 77889 Seebach, Germany 123 J Ornithol DOI 10.1007/s10336-014-1049-5

Suitable, reachable but not colonised: seasonal niche duality in an endemic mountainous songbird

  • Upload
    marc-i

  • View
    224

  • Download
    1

Embed Size (px)

Citation preview

Page 1: Suitable, reachable but not colonised: seasonal niche duality in an endemic mountainous songbird

ORIGINAL ARTICLE

Suitable, reachable but not colonised: seasonal niche dualityin an endemic mountainous songbird

Jan O. Engler • Dennis Rodder • Darius Stiels •

Marc I. Forschler

Received: 29 July 2013 / Revised: 12 January 2014 / Accepted: 28 January 2014

� Dt. Ornithologen-Gesellschaft e.V. 2014

Abstract The realized distribution of animals is often

delimited by climatic factors which define, next to the

specific habitat and food availability, their species-specific

potential distribution. We studied the environmental limi-

tations affecting the realized breeding and wintering dis-

tributions of the Citril Finch (Carduelis citrinella), one of

the few endemic bird species of European mountain ranges.

To assess the environmental limits that shape the seasonal

distribution, we used species distribution models (SDMs)

derived from macroclimate in combination with land cover

information. Our data suggest a high congruence between

the potential modelled breeding distribution of the Citril

Finch and the currently known breeding sites, indicating a

high level of niche filling. The unusual absence in several

suitable breeding habitats at the eastern and northern range

limit (Eastern Alps, Carpathians, Bavarian Forest, Harz

Mountains, Fichtelgebirge, Krkonose Mountains) is likely

linked to a combination of both missing resources and

restricted physiological migration capacities from the

available wintering grounds. Since the accomplished

migratory distances hardly exceed more than 500 km, it

seems likely that the distance to the main wintering areas is

too large for exceeding eastern and northern range limits.

We discuss the differences in SDM outcomes when

including distal predictor variables instead of using proxi-

mal predictors alone, and highlight the importance of

considering a seasonal niche duality to gain more insights

into complex range effects in species with seasonal ranges.

Keywords Carduelis citrinella � Ecological niche �MaxEnt � Species distribution model (SDM)

Zusammenfassung

Geeignet, erreichbar aber unbesiedelt: saisonale Nisc-

hendualitat bei einem endemischen Singvogel europai-

scher Gebirgsregionen

Die realisierte Verbreitung von Arten wird oft durch kli-

matische Faktoren begrenzt, die gemeinsam mit dem cha-

rakteristischen Habitat und der Nahrungsverfugbarkeit die

artspezifische potentielle Verbreitung definiert. In dieser

Studie untersuchten wir die umweltbedingten Faktoren,

welche das Brut- und Winterareal des Zitronenzeisigs

(Carduelis citrinella) limitieren. Um die begrenzenden

Faktoren der saisonalen Verbreitungen zu quantifizieren,

nutzten wir Artverbreitungsmodelle basierend auf biokli-

matischen Variablen in Kombination mit Land-

nutzungsinformationen. Die Ergebnisse zeigten eine hohe

Ubereinstimmung der modellierten potentiellen Verbrei-

tung mit der derzeitig bekannten Verbreitung der Art, was

auf einen hohen Grad an Nischenfullung (,,niche filling‘‘)

schließen lasst. Klimatisch geeignete, jedoch unbesiedelte

Brutgebiete entlang des nordlich und ostlich gelegenen

Arealrandes (ostliche Alpen, Karpaten, Bayerischer Wald,

Electronic supplementary material The online version of thisarticle (doi:10.1007/s10336-014-1049-5) contains supplementarymaterial, which is available to authorized users.

J. O. Engler (&) � D. Rodder � D. Stiels

Zoological Research Museum Alexander Koenig, Adenauerallee

160, 53113 Bonn, Germany

e-mail: [email protected]

J. O. Engler

Department of Wildlife Management, University of Gottingen,

37077 Gottingen, Germany

M. I. Forschler

Department of Monitoring and Research, National Park Black

Forest, Schwarzwaldhochstraße 2, 77889 Seebach, Germany

123

J Ornithol

DOI 10.1007/s10336-014-1049-5

Page 2: Suitable, reachable but not colonised: seasonal niche duality in an endemic mountainous songbird

Harz, Fichtel- und Riesengebirge) sind hochstwahrschein-

lich durch eine Kombination aus fehlenden Ressourcen

einerseits und begrenzter physiologischer Migrationsleis-

tung andererseits zu erklaren. Da die bekannten Zugdis-

tanzen nur selten weiter als 500 km reichen, ist es sehr

wahrscheinlich, dass die Entfernung dieser Gebiete im

Norden und im Osten zu den Hauptuberwinterungsgebieten

zu groß ist. Wir diskutieren die Unterschiede in den

Modellresultaten zwischen einer Pradiktorauswahl, die

auch distale Pradiktorvariablen beeinhaltet, und einer mit

rein proximalen Variablen. Wir unterstreichen, dass die

Berucksichtigung der saisonalen Nischendualitat zusatzli-

che Einsichten in komplexe Arealeffekte bei Arten mit

saisonalen Verbreitungen gewahrt.

Introduction

One of the principles of biogeography states that most

animals and plants show species-specific distribution pat-

terns (Huggett 2004). Thereby, on a macro-scale, the

realized distribution of a species is often delimited by

climatic factors which, beside specific habitat and food

availability, define the potential distribution of a species

(e.g. Grinnell 1917; Mackey and Lindemayer 2001).

However, commonly, not every environmentally suitable

habitat patch is also colonised due to limited accessibility

or biotic interactions (Soberon 2007; Soberon and Na-

kamura 2009), often also shaping distributional borders

(e.g. Gaston 2003). Analysing the causes of range limits is

of great interest in ecological und biogeographical research

(Gaston 2003; Holt and Keitt 2005). Even though general

patterns of range limits of multiple species are known for

some regions (McInnes et al. 2009), which frequently fall

together with climatic thresholds (Gaston 2003), the

explicit factors acting along range boundaries are multi-

faceted and vary species-specifically (Gaston 2003; Case

et al. 2005; Fortin et al. 2005). Furthermore, the degree of

range filling strongly depends on species-specific dispersal

abilities affecting the colonisation potential of isolated

suitable habitat patches (Araujo and Pearson 2005, Laube

et al. 2013).

Unlike other vertebrate groups, most birds show

extraordinary high capabilities to reach such distant areas

(Gorman 1979; Newton 2003; Araujo and Pearson 2005).

This is most evident in migratory birds distributed in the

Western Palaearctic, where a high degree of niche and

range filling becomes evident, at least comprising a large

portion of suitable habitats being in close proximity to the

species’ core distribution (Araujo and Pearson 2005).

Following the terminology of Soberon and Peterson

(2005), the potential distribution of most birds should be

close to their realized distribution, as accessibility of suit-

able but isolated habitat patches are of lesser importance

compared to non-flying vertebrates. In particular, range

filling becomes evident for almost all bird species with

mountainous breeding ranges in the Western Palaearctic.

Passerine birds typical for mountainous areas such as

Water Pipit (Anthus spinoletta), Alpine Accentor (Prunella

collaris), Ring Ouzel (Turdus torquatus), Rock Thrush

(Monticola saxatilis), Wallcreeper (Tichodroma muraria),

Alpine Chough (Phyrrhocorax graculus), Snow Finch

(Montifrigilla nivalis) or the Common Crossbill (Loxia

curvirostra) are characterised by broad distributions, cov-

ering mountainous areas from the west (Iberian Peninsula)

to the east (Carpathians) and beyond, irrespective of their

migration behaviour (e.g. Vaurie 1959; Svensson et al.

2009). One remarkable exception is the Citril Finch

(Carduelis citrinella). It is one of the few true European

endemic passerines and is restricted to some Central and

Southwest European mountain ranges (Cramp and Perrins

1994), whereas a sharp range edge separates the Eastern

Alps and Carpathian mountains from being colonised

(Marki 1976). Being still quite frequent in Vorarlberg, the

species abruptly becomes very rare in the limestone Alps of

Tirol and Salzburg only ca. 200 km eastward (Dvorak

et al.1993; Moritz and Bachler 2001), although apparently

suitable breeding sites do occur. Currently, only a few and

small populations exist in the Carnic Alps (Dvorak et al.

1993; Feldner and Rass 1999; Probst 2012) and the Julic

Alps (Gregori 1977; Matvejev 1981; Geister 1983, 1995;

Tomaz Mihelic and Borut Stumberger, personal commu-

nication). The exact reasons for this sharp range edge

remain speculative (Glutz von Blotzheim and Bauer 1997).

As spatio-temporal climate dynamics (e.g. Brambilla

et al. 2012), as well as specific migration dynamics (e.g.

Bensch 1999; Thorup 2006), may affect distribution pat-

terns and the shape of realized distributions, we evaluated

the environmental limits affecting the realized breeding

and wintering distributions of the Citril Finch taking into

account its known migratory behaviour. We assess the

environmental limits that shape its seasonal distribution

using species distribution models (SDMs) derived from

macroclimate in combination with land cover information.

For the breeding distribution, two SDMs with different sets

of climatic variables were developed: one based on year-

round climatic conditions and a second based on summer

climate only. Based on these models, and in combination

with a multivariate assessment of the climatic conditions

between settled and unsettles mountain ranges, we

hypothesise in particular that (1) the Carpathian Mountains

markedly differ in climatic conditions from the populated

habitats in Central and Southwest Europe, and (2) that the

sharp range edge in the Eastern Alps is mainly driven by a

climatic cline. Furthermore, we computed an additional

J Ornithol

123

Page 3: Suitable, reachable but not colonised: seasonal niche duality in an endemic mountainous songbird

SDM based on the winter distribution for direct comparison

with the breeding distribution to get precise insights into

the seasonal migratory dynamics of the species. Here, we

hypothesise that environmental conditions equalling those

from the known winter range are lacking in the potential

wintering range in the Balkans, which consequently could

have impeded the successful colonization in the past.

Methods

Study species

The largest part of the breeding population of the Citril

Finch, about 80 % of the world population, inhabits the

Spanish mountains, especially the Pyrenees (Baccetti and

Marki 1997), where the highest population densities of the

species were also recorded (Forschler 2006). Other smaller

populations occur in the lower mountain ranges of the

Massif Central, Cevennes, Mont Ventoux, Jura, Black

Forest and Vosges (Cramp and Perrins 1994; Glutz von

Blotzheim and Bauer 1997). A second stronghold of the

species is situated in the Alps (Baccetti and Marki 1997).

In all breeding areas, the species is linked to a combination

of flower-rich mountain meadows and some key conifers,

such as Mountain Pine Pinus mugo, Black Pine Pinus

nigra, Scots Pine Pinus sylvestris and Spruce Picea abies

(Forschler and Kalko 2006a, b).

The winter range of the Citril Finch is even smaller than

its breeding range, wherein the average distance to the next

wintering areas is about 400–500 km (Cramp and Perrins

1994). Only a very few recoveries show larger migration

distances than 600 km (Bernis and Bernis 1963; Borras

et al. 2010) with an observed maximum of 946 km (Spina

and Volponi 2008). Birds from northern and eastern pop-

ulations overwinter in the Massif Central, the Cevennes

and in the southern and western Alps, where they conduct

short-distance altitudinal movements according to the

predominant weather conditions (De Crousaz and Lebreton

1963; Marki 1976; Dejonghe 1991; Zink and Bairlein

1995; Marki and Adamek 2013). In Spain, populations

often stay closer to their breeding sites at lower altitudes

(e.g. Borras et al. 2010; Borras and Senar 2013), but some

populations are also known to conduct considerable

migratory movements, e.g. from the Western Pyrenees to

Catalonia (Aymı and Tomas 2003; Alonso and Arizaga

2004). Some birds from northern populations even winter

in Spain (Benoit and Marki 2004; Borras et al. 2005).

Occurrence records

Since Citril Finches inhabit different geographic areas

during different times of the year due to seasonal migration,

we analysed seasonal ranges separately, distinguishing

between breeding and wintering areas. Their breeding

season starts in April and ends in August. In the subsequent

wintering period, from November to February, the species

conducts short- to medium-distance migration often com-

bined with altitudinal movements.

Species records from the breeding season were collected

directly at nest sites during different range-wide field sur-

veys conducted between 1999 and 2006 (Forschler and

Kalko 2006a; Forschler et al. 2009; Forschler, unpublished

data). Additional samples were obtained from compre-

hensive literature studies (Moltoni 1969; Marki 1976;

Genard and Lescourret 1987; Geister 1995; Feldner and

Rass 1999; Aymı and Tomas 2003; Alonso and Arizaga

2004; Berlit 2005; Borras et al. 2005), internet platforms

(Global Biodiversity Information Facility GBIF, http://

www.gbif.org; BirdLife Schweiz, http://www.birdlife.ch;

http://www.birdinggermany.de) and observations of local

field ornithologists (J.M. Alonso Pulmar, T.P. Aparisi, J.

Arizaga, S. Blache, J. Canadas, R. Kilzer, P. Perret, J.J.

Pfeffer). After visual inspection for possible errors, i.e.

georeferencing inaccuracies, using DIVA-GIS 5.4 (Hij-

mans et al. 2001; available through http://www.diva-gis.

org), a total of 81 records remained for further analysis.

Likewise, wintering records (n = 64) were obtained

from own field surveys (Forschler, unpublished data), lit-

erature studies (Moltoni 1969; Marki 1976; Bocca and

Maffei 1984; Mingozzi et al. 1988; Maestri et al. 1989;

Fornasari et al. 1998; Aymı and Tomas 2003; Alonso and

Arizaga 2004; Benoit and Marki 2004; Borras et al. 2005),

internet platforms (http://fr.groups.yahoo.com/group/

obsmedit, http://www.oiseaux.net, http://www.atlasdeaves.

org) and observations of local field ornithologists (J.M.

Alonso Pulmar, T.P. Aparisi, J. Canadas).

Environmental data

Climate information relevant for Citril Finch seasonal

distributions with a spatial resolution of 2.5 arc-minutes

(i.e. about 3.16 km in E–W direction and 4.62 km in N–S

direction in the Alps) was obtained from WorldClim v.1.4

(Hijmans et al. 2005; http://www.worldclim.org). Based on

the temporal definition of seasonal ranges given above,

climate layers from monthly temperature and precipitation

data were extracted to calculate their range and the mean

(Tmean, Trange, Pmean, Prange) within the breeding and win-

tering seasons. As the Citril Finch is a mountainous species

avoiding continental winter climate, additional environ-

mental information that might shape the general altitudinal

and non-continental breeding distribution is necessary. We

considered the minimum temperature of the coldest month

(BIO6) as a distal variable that describes extreme climatic

conditions and thus could characterise the general

J Ornithol

123

Page 4: Suitable, reachable but not colonised: seasonal niche duality in an endemic mountainous songbird

altitudinal environment, especially in the eastern parts of

Europe in a climatic manner. We checked correlations

between variables using a correlation matrix based on

Spearman rank correlations and set the correlation coeffi-

cient of [0.9 or \-0.9 as previously recommended

(Fourcade et al. 2013). As no strong intercorrelations were

detected, all variables were kept for modelling.

Land use data as overlay for the presence/absence maps

from the models were obtained from the global land cover

facility (http://www.landcover.org/index.shtml). Subse-

quently to modelling, they were overlaid with the resulting

presence/absence maps in order to restrict the potential

distribution to those areas where adequate microhabitats

are available. In the breeding and wintering ranges of the

Citril Finch, the species is restricted to open and semi-open

conifer forests consisting mainly of spruce (Picea) and pine

(Pinus) species. Beside their function as breeding and

resting habitats, these conifers also contribute largely to the

species’ diet (see Forschler and Kalko 2006a, b; Forschler

2007 for details). Consequently, we selected ‘needle leaf

forests’ and ‘mixed forest’ as relevant habitat classes for

the species.

Species distribution modelling

Both datasets of species records were cleaned for unequal

sampling efforts in environmental space using a cluster

analysis approach based on Euclidean distances as sug-

gested by Rodder et al. (2009a). This approach allows a

reduction of a possible spatial bias causing autocorrelation

that would negatively affect SDM predictions (Dormann

et al. 2007; Phillips 2008; Phillips et al. 2009). As a result

of this analysis, 50 records remained for each of the initial

sets of occurrence records for subsequent processing.

For SDM computation, MaxEnt v.3.3.3a (Phillips et al.

2004, 2006; Phillips and Dudık 2008), a machine-learning

algorithm following the principle of maximum entropy

(Jaynes 1957; Elith et al. 2011), was used. In comparative

studies, MaxEnt has frequently outperformed various other

SDM algorithms (e.g. Heikkinen et al. 2006) even when the

sample sizes were comparatively small (e.g. Wisz et al.

2008). Recently, SDMs have been successfully applied for

spatial conservation planning (Araujo et al. 2004; Guisan

and Thuiller 2005; Kremen et al. 2008; Rodder et al. 2010;

Jiguet et al. 2011), invasive species assessments (Peterson

and Vieglais 2001; Ficetola et al. 2007; Stiels et al. 2011),

evolutionary biology (Jakob et al. 2010; Kozak et al. 2008;

Kozak and Wiens 2007; Smith and Donoghue 2010),

exploring biodiversity patterns (Carnaval and Moritz 2008;

Marini et al. 2010; Schidelko et al. 2011), to estimate

reproductive parameters (Brambilla and Ficetola 2012), to

predict species overlap (Brambilla et al. 2013) and as a

complementary tool in phylogeographic studies (Chan

et al. 2011; Kozak and Wiens 2007; Rodder et al. 2013).

Following Phillips et al. (2009), the potential Citril Finch

distribution was computed based on a set of 10,000 back-

ground points that were drawn from a rectangular polygon

comprising the potential colonisable range for the species

in Europe. Citril Finches are highly mobile, and it is well

known that they conduct seasonal movements outside their

breeding range (Marki 1976; Zink and Bairlein 1995;

Fornasari et al. 1998; Aymı and Tomas 2003; Alonso and

Arizaga 2004; Benoit and Marki 2004; Borras et al. 2010)

and occur as vagrants outside regular breeding and win-

tering sites (Hyndman 2008; Forschler et al. 2011). Thus,

almost the entire mountainous areas in the European con-

tinent north to 60.7�N and east to 11�E was considered to

be potentially colonisable.

For model evaluation, the ‘area under the receiver

operation characteristic curve’ (AUC) statistics was cal-

culated (Swets 1988; Fielding and Bell 1997). In 100

model runs, a bootstrap approach was applied in which

iteratively 30 % of the species records were omitted from

model training and used for testing each of the models.

Subsequently, the mean AUC score of the 100 models was

calculated to eliminate possible record omission effects.

We acknowledge recent criticisms on AUC scores (e.g. by

Lobo et al. 2008), but since they are still widely used and

given the general problems of other model evaluation

approaches (e.g. Baldwin 2009), we prefer to use it here,

too. However, AUC scores should always be interpreted

with caution when using them as stand-alone validation

procedure. Therefore, commission and omission errors are

also provided. Furthermore, as the present distribution of

this species is very well known, a visual inspection of the

model output maps might help to assess the quality of the

models as well.

In MaxEnt, the logistic model output format was chosen,

and the resulting map shows the likelihood of environ-

mental suitability in each grid cell in values ranging

between 0 and 1, wherein more suitable climate conditions

are indicated by higher values. The species was assumed to

be present in those grid cells with probabilities above the

10 percentile training presence logistic threshold. This is a

non-fixed threshold as recommended by Liu et al. (2005)

which cuts off the areas with poor model support assuming

an error rate of 10 %. In the comparison with other non-

fixed thresholds commonly used in SDM studies, we

decided to give preference to this single threshold as it was

especially sensitive in describing the known altitudinal

limitation of the species’ realized range (data not shown).

For the breeding season, two models were constructed.

The first was developed using only the environmental layers

(proximal factors) characterizing the breeding season

(Model 1) and the second model emphasizing stronger the

altitudinal extent of the potential distribution (i.e. including

J Ornithol

123

Page 5: Suitable, reachable but not colonised: seasonal niche duality in an endemic mountainous songbird

also BIO6 as an additional distal factor; Model 2). Finally,

the species’ winter distribution was modelled using envi-

ronmental variables of the winter season (Model 3).

Multivariate assessment of realized versus potential

distributions

In order to compare climatic differences between geo-

graphical regions, 100,000 random points were generated

across the whole rectangular prediction area. Subsequent to

the removal of duplicate points in single grid cells, the

remaining points were clipped within the potential distri-

bution suggested by Model 1 comprising all areas with

probabilities above the 10 percentile logistic threshold.

Thereafter, 2,228 random points remained for further

analysis. Geographical subsets were defined based on the

current breeding ranges of the species (Group 1) and on

regions that were predicted as environmentally suitable by

Model 1 but are not colonised today (Group 2). Subse-

quently, the climatic information used for Model 2 was

extracted at the corresponding records. In a principal

component analysis, the multi-collinearity among the

environmental data was eliminated. Three principal com-

ponents (PC) explaining 88 % of the total variance of the

raw data were extracted and rotated with the varimax

method (Kaiser 1958) for a better explanation of the

respective contributions. Subsequently, the a priori defined

groups (Group 1 and 2) were analysed applying a linear

discriminant analysis (LDA) which assumes equal prior

probabilities in both groups. Therein, the climate profile

was represented by the first three PC. LDA was trained

with 70 % of the data randomly chosen from the whole

dataset and tested against the remaining data. In order to

account for uncertainties, this procedure was repeated

1,000 times and the results were used to calculate the mean

group classification as well as the probability by which

each data point was assigned to the other group (as per-

centage of all runs). All multivariate analyses were con-

ducted using R 2.11.1 (R development core team 2010),

wherein the LDA was performed using the MASS package

for R (Venables and Ripley 2002).

Results

Climatic differences in the extent of the potential

breeding range

In both models describing the potential distribution of the

Citril Finch during the breeding season (Model 1 and

Model 2), AUC values were ‘excellent’ (Model 1:

AUC = 0.958 ± 0.014; Model 2: AUC = 0.973 ± 0.022),

wherein both the inhabited mountainous regions of central

and south-western Europe are predicted to be climatically

suitable (Fig. 1). The lowest 10 percentile training pre-

sence value was 0.160 (±0.068) in Model 1 and 0.282

(±0.064) in Model 2. Mean precipitation and temperature

during the breeding season were most important in Model 1

(Pmean = 57.78 ± 13.14 %; Tmean = 31.65 ± 12.52 %),

whereas the combined precipitation and temperature ranges

only contributed about 10 % to the model (Prange =

3.18 ± 1.26 %; Trange = 7.39 ± 3.51 %). An analogous

picture became evident in Model 2, where mean precipi-

tation and temperature remained the most important factors

in the final model (Pmean = 45.53 ± 12.97 %; Tmean =

29.89 ± 9.49 %). However, when adding the BIO6 as a

distal variable reflecting fine-scale altitudinal patterns as

expected especially for the more continental regions in

eastern Europe, it contributed equally high as the temper-

ature range (BIO6 = 12.71 ± 2.88 %; Trange = 9.50 ±

3.60 %). The precipitation range was rather unimportant

in Model 2 (Prange = 2.36 ± 1.56 %). In both breeding

range models, Jackknife tests show that Tmean was most

important when used in isolation, as well as when it

became omitted from the model, showing that this pre-

dictor appears to have the most useful information by

itself and further contains information that is not present

in the other variables. Pmean ranked second best in the

Jackknife evaluation whereby all other predictors ranked

equally in the Jackknife test as in the above mentioned

variable contributions.

Although the AUC scores of both models were excel-

lent, suggesting high discrimination ability between suit-

able and unsuitable habitat, regionally large differences

became evident by visual inspection of the output maps.

Here, the mountainous character of the species’ realized

distribution is better visualised in Model 2 compared to

Model 1. Furthermore, the Carpathian region was sug-

gested to provide potentially suitable habitats in Model 1

(Fig. 1a) but not in Model 2 (Fig. 1b), where this region is

classified as being unsuitable. Both breeding season models

indicate climatically suitable regions in Scotland and

south-western Scandinavia that are not colonised (Supple-

mentary Fig. A1). However, an overlay of relevant habitat

types (‘needle leaf forests’ and ‘mixed forests’) shows that

these northern regions largely lack suitable habitat (Fig. 1a,

b). In contrast, the Carpathian region contains potentially

suitable habitat types (Fig. 1a). Additionally, major dis-

crepancies between the potential and realized distribution

are found in the Eastern Alps, the Swabian Mountains, the

Harz Mountains and the Dinaric Alps.

Potential distribution of wintering range

The SDM developed for the winter distribution of the Citril

Finch (Model 3) also performed ‘excellent’

J Ornithol

123

Page 6: Suitable, reachable but not colonised: seasonal niche duality in an endemic mountainous songbird

(AUC = 0.916 ± 0.019). The lower 10 percentile training

presence value was 0.256 (± 0.068). Variable importance

in Model 3 differed from the former models, wherein the

temperature range was followed by the BIO6 in terms of

variable importance (Trange = 41.24 ± 6.85 %; BIO6 =

31.21 ± 10.96 %). Mean precipitation and temperature

contributed intermediately to Model 3 (Pmean = 11.86 ±

9.63 %; Tmean = 10.85 ± 6.62 %), whereas the precipita-

tion range contributed \5 % (Prange = 4.85 ± 3.11 %).

According to Jackknife tests, Trange appears to contains the

most useful information by itself (using alone) and contains

information that is not present in the other variables (using

without this predictor). All the other predictors ranked

equally in the Jackknife test as shown in the above men-

tioned variable contributions.

According to Model 3, most suitable wintering regions

can be found around the Massif Central, in the lower

altitudes of the Pyrenees and Southern Alps as well as in

northern Spain. Most interestingly, suitable wintering sites

are absent close to the Carpathian region. Overlaying the

potential distribution with relevant habitat data in terms of

land cover information as stated above, suitable areas are

very scattered over large parts of Central and Southern

Europe (Fig. 2). In this model, a much larger discrepancy

between the potential and realized distribution becomes

apparent, indicating many areas north and east of the cur-

rent range as suitable winter areas, where the species is not

actually found wintering.

Climatic structure of geographic regions

Three PC with Eigenvalues[1 were extracted representing

88.3 % of the total variance of the five input variables

(Table 1). All three components represent well explainable

Fig. 1 Potential breeding

distribution of the Citril Finch

(Carduelis citrinella) using two

different sets of environmental

variables after clipping with

adequate habitat classes (needle

leaf forest and mixed forest;

Model 1—upper half a, Model

2—lower half b). White dots

represent presence localities

used for modelling

J Ornithol

123

Page 7: Suitable, reachable but not colonised: seasonal niche duality in an endemic mountainous songbird

gradients. PC1 (explained variance = 38.9 %) represents

an aridity gradient, where increasing values of the com-

ponent imply a decrease in mean precipitation and an

increase in mean temperature and its range. PC2 (explained

variance = 28.1 %) is strongly related to BIO6 as well as

to the mean temperature. It can be interpreted as an ‘alti-

tudinality/continentality’ variable, where lower values

represent colder winter climate. Finally, PC3 explains the

rain seasonality, in which higher values represent larger

seasonality. The variable contributions to each component

as well as their ecological explanations are presented in

Table 1.

Our results indicate that the Carpathian region was the

only larger region that was represented in Model 1 but not

in Model 2. In contrast, other areas with differences

between both models showed either only minor differences

in range sizes, i.e. range contractions in one model com-

pared to the other, or were generally too small for a sig-

nificant interpretation. Thus, we decided to define the

Carpathian region as Group 2 and the remaining potential

distribution as Group 1 before conducting the LDA

(Fig. 3a). It is important to note that the northernmost

regions in Scandinavia and Scotland were also assigned to

Group 1 because both regions were suggested as being

suitable in Model 1 as well as in Model 2, even though

these regions are excluded from the realized distribution of

the Citril Finch due to lacking microhabitats (Supplemen-

tary Fig. A1).

The LDA shows a high distinction of the Carpathian

region from Group 1, where 100 % of all cases (n = 468)

were correctly classified. In contrast, there are several

regions in the potential suitable range defined in Group 1

that were sorted into Group 2 (18.4 %, n = 321 out of

n = 1,741 in total; Fig. 3b). These regions were mainly

located in the Austrian part of the Alps (eastern to north-

eastern Alps) as well as in the Iberian region west of the

Pyrenees, plus a remote area in Algeria.

Discussion

Investigating patterns and processes that shape species

range boundaries are a central goal of biogeography (e.g.

Gaston 2003). Here, we aim to balance possible reasons for

the absence of Citril Finches in suitable mountainous areas

in the east of Europe. We were able to show that the eastern

mountain ranges like the Carpathians and Eastern Alps

possess suitable climate conditions during the breeding

Fig. 2 Potential wintering

distribution of the Citril Finch

(Carduelis citrinella) after

overlaying with adequate

habitat classes (needle leaf

forest and mixed forest). White

dots represent presence

localities used for modelling

Table 1 Variable contributions to each principal component as well

as Eigenvalues and explained variance under the rotated solution

(varimax rotation)

Climatic

variables

Principal component

1 2 3

BIO6 0.084 0.930 -0.270

Pmean -0.743 -0.376 0.203

Prange -0.034 -0.186 0.972

Tmean 0.696 0.600 0.055

Trange 0.949 -0.074 0.038

Eigenvalue 1.945 1.406 1.064

Explained

variance

38.9 % 28.1 % 21.3 %

Ecological

meaning

Aridity Altitudinality/

continentality

Rain

seasonality

Total explained variance represented through principal components is

88.3 %. Pearson correlation coefficients [0.5 and \-0.5 are indi-

cated in bold as they represent notable support with the principal

component

J Ornithol

123

Page 8: Suitable, reachable but not colonised: seasonal niche duality in an endemic mountainous songbird

period. However, distal factors, represented by the BIO6

variable, might influence the environmental conditions

indirectly and exclude these areas from being potentially

suitable. Thus, the incorporation of this factor provides a

more realistic picture of the realized distribution. In the

following, we discuss these findings for the realized

breeding and winter distribution of the Citril Finch and

question the importance of the seasonal niche duality as

discussed in the final section.

Breeding range limits

Our results suggest a high congruence between the poten-

tial breeding distribution of the species and known breed-

ing sites, indicating a high level of niche filling. Obviously,

Citril Finches have occupied almost their entire potential

breeding niche in Europe apart from huge areas in the

Carpathians and some rather small areas at the northern and

eastern range limit in the Eastern Alps, the Harz Moun-

tains, the Swabian Mountains and the Dinaric Alps. The

observed pattern raises the question why the species does

not occur in these areas although apparently very suitable

breeding conditions are available.

One could speculate whether this pattern is due to

human activities like past or present persecution or specific

anthropogenic habitat alterations in areas of absence (see,

e.g., Heuck et al. 2013). However, we are not aware of any

such geographically restricted processes that are in broad

congruence with our results. One possibly more likely

explanation for the observed pattern is an incomplete

spread of the species over its full potential range after the

last glacial maximum. This might be due to restricted

Fig. 3 Suitable areas from both

models represented through

random points (see text for full

information) (a). The

Carpathian region was

suggested to be potential

suitable in Model 1 but not in

Model 2. So it was a priori

defined as group 2 (black

points) and compared with the

rest of suitable space (group 1,

white points in upper panel)

with linear discriminant analysis

(LDA). Classification results

after LDA are shown in the

lower panel (b)

J Ornithol

123

Page 9: Suitable, reachable but not colonised: seasonal niche duality in an endemic mountainous songbird

dispersal out of potential refuge sites in France and Spain.

In line with this assumption, Citril Finch populations north

of the Pyrenees were found to show fewer haplotypes and a

considerable lower nucleotide and gene diversity in their

mitochondrial DNA than the Iberian populations (Forschler

et al. 2011). Furthermore, the generally low genetic vari-

ability indicates a strong and relatively recent bottleneck

event in the species population history, potentially

reflecting a sudden decrease of crucial resources during the

Mid-Holocene and a subsequent breakdown of the popu-

lation (see Forschler et al. 2011 for details). It seems rea-

sonable that the species is still recovering from this event

and is currently expanding slowly eastwards as indicated

by small satellite populations in the Dobratsch area, Austria

(Feldner and Rass 1999; Probst 2012) and the Triglav

region, Slovenia (Geister 1995). However, even though

some fluctuations in range size exist in these populations,

we cannot exclude the possibility that the species has a

longer population history in the Carnic and Julian Alps

(Keller 1890; Dvorak et al. 1993; Feldner and Rass 1999).

This hypothesis is supported by the fact that, during the last

glacial phase, a small refugial range existed in this region

for the Scots pine Pinus sylvestris (Cheddadi et al. 2006),

one of the host plants of the Citril Finch. Further analysis

of the genetic structure of these populations in comparison

to the main populations in the Alps and Pyrenees could

help to shed light into the refugial state of these

populations.

Another possibility for the observed pattern could be

linked to the physiological movement capacities of the

species. Citril Finches from populations north of the Py-

renees are short- to medium-distance migrants which

overwinter mainly in the Massif Central, the Cevennes and

in the Southern and Western Alps (De Grousaz and Lebr-

eton 1963; Marki 1976; Cramp and Perrins 1994; Zink and

Bairlein 1995; Marki and Adamek 2013). The migration

distance of the species exceeds more than 500 km in only a

small proportion of the population (Cramp and Perrins

1994). Therefore, it might be possible that the main current

wintering areas are too far away and the distance exceeds

the physiological capacities of the species, thus making the

areas inaccessible for breeding habitat given the current

extent of the main wintering range. On a larger geographic

extent, our finding is in concordance with the results of

Bensch (1999) who showed that migration limitations are

responsible for range size limitations in migratory bird

species breeding in the higher latitudes of the Palaearctic

(see also Thorup 2006 for other biogeographic realms).

This would also explain the abrupt range limit in the

Eastern Alps (Marki 1976), despite good breeding areas,

but with a distance to the wintering areas of around

650 km, and the absence of the species in more distant

mountain ranges such as the Dinaric Alps, the Harz

Mountains and the Carpathians. The lack of short-term

refuge sites during cold spells in spring in these areas could

play a major role, as well as increasing continental climatic

conditions towards the eastern parts. Interestingly, the

additional discriminant analysis reflects the current distri-

bution pattern of the species to a large extent. According to

the climatic conditions, the widely unpopulated Eastern

Alps were sorted together with the Carpathians where the

species is completely absent (Fig. 3b). This might also

explain the lack of any stable populations in the Eastern

Alps even though Model 2 shows potentially suitable areas.

In addition, the potential distribution on the Iberian Pen-

insula, and the predicted but not realized range in the

Maghreb, cluster with the Carpathian range. The Mediter-

ranean area shares with continental areas in Eastern Europe

a general more arid climate in comparison to the more

humid Atlantic areas, and are thus bordering each other

along the first principal component axis. However, the

Mediterranean ranges were still more isolated from the

Carpathian and eastern alpine regions along the PC1 axis

(Supplementary Fig. A2), and are thus characterised by

lower precipitation, higher temperatures and a larger tem-

perature range during the breeding season (Table 1).

Wintering range limits

The potential wintering range of the Citril Finch includes

some of the most important current wintering areas in the

Massif Central, the Cevennes, Southern and Western Alps

and Eastern Pyrenees (De Crousaz and Lebreton 1963;

Marki 1976; Dejonghe 1991; Cramp and Perrins 1994;

Borras et al. 2005, 2010; Borras and Senar 2013; Marki and

Adamek 2013). However, our model suggests some addi-

tional potential wintering areas north of the current range

of the species in low mountains of Eastern France (Vosges,

Jura), Switzerland (Jura) and Germany (Black Forest),

which are currently not regularly used by the species dur-

ing winter, perhaps due to a higher risk of lethal sudden

cold spells with snow and ice (see Cramp and Perrins

1994). Consequently, the realized distribution is much

smaller than the potential wintering distribution. However,

in concordance with the model, several of the larger

potential wintering areas have been occasionally recorded

as wintering sites of the species, e.g. Forschler (1997) and

Bauer et al. (1995) for the Black Forest, De Grousaz and

Lebreton (1963) for the Jura, and Praz and Oggier (1973)

for the Valais. Astonishingly, there is evidence for large

numbers of wintering Citril Finches in the 1800s (Land-

beck 1834; von Kettner 1849) which supports model results

but points to a change in migratory behaviour in the last

century. Another supporting factor for the narrower win-

tering range of the species—at least in the populations

north of the Pyrenees—refers to one of its preferred food

J Ornithol

123

Page 10: Suitable, reachable but not colonised: seasonal niche duality in an endemic mountainous songbird

plants during winter, the Wood Sage Teucrium scordonia.

Besides pine seeds, the seeds of this plant species are one

of the main food resources, especially during cold spells

(Marki 1976; Holzinger and Dorka 1997; Forschler 2001,

2007; Marki and Adamek 2013). Interestingly, the main

distribution area of Wood Sage (Anderberg and Anderberg

1997) overlaps widely with the winter distribution of the

Citril Finch and might explain the absence of the species in

apparently suitable areas without occurrences of this plant

species. Furthermore, in the Black Forest, Wood Sage

abundance has decreased significantly in the last decades

(Forschler, personal observation) which might at least

partially explain the disappearance of wintering Citril

Finches in the Black Forest. However, this holds only for

populations north of the Pyrenees, since birds on the Ibe-

rian Peninsula depend on other food plants during winter

(Borras et al. 2010).

The seasonal niche duality

SDMs have become increasingly important in biogeo-

graphical and macroecological research over the past dec-

ade. According to Soberon (2007) or Godsoe (2010), the

theoretical basis for interpreting SDM results follows

Hutchinson’s niche concept (Hutchinson 1957, 1978).

According to that, a species physically responds only to

conditions to which it is actually exposed to and which

determine its realized niche. Therefore, variable selection is

a crucial step for the application of SDMs (Rodder et al.

2009b). In species with seasonal ranges like migratory birds,

predictor variables span different seasons and might thus tell

different stories. When strictly following Hutchinson’s niche

concept, the variables entering an SDM for modelling a

seasonal range should only cover the period in which the bird

is physically present. We followed this assumption in Model

1 which covers only climatic parameters of the breeding

season. The result was a potential distribution also extending

into the eastern mountainous regions where many other

mountainous breeding finches, which typically occur in

sympatry with the Citril Finch like Siskin (Carduelis spinus)

or Common Crossbill, regularly breed. This indicates that the

bioclimatic tolerances of the Citril Finch potentially enable

the species to breed there. However, in a model which

includes extreme climatic variables from the winter season,

the potential distribution equals the realized distribution of

the species. The inclusion of additional information might

add further biologically relevant information, although this

conflicts with the assumptions of the Hutchinsonian niche

concept, wherein the niche is shaped only by conditions to

which a species is actually exposed. It also contradicts the

conclusions of Austin (2002), who emphasised the impor-

tance of proximal factors in the prediction of species distri-

butions. Furthermore, the inclusion of an additional

parameter increases the risk of overfitting the model which

might lead into a statistical artefact instead of a biological

signal (Heikkinen et al. 2006). On the other hand, the justi-

fied addition of climatic information which acts as a distal

factor has to be considered. In particular, the additional

parameter BIO6 in Model 2 might reflect physiological

limitations influencing the occurrence of important plant

species needed for foraging or breeding. The importance of

this predictor is further underlined in the multivariate

assessment in which the non-inhabited parts of the Eastern

Alps cluster with the Carpathian Mountains, forming its

western border in close concordance with the true eastern

range edge of the species. In consequence, a comparative

evaluation of SDMs following different assumptions can

shed additional light on the causal foundation of the position

of range boundaries of species in general. To our knowledge,

literature accounting for this type of niche duality in sea-

sonally occurring species is so far lacking, even if the

importance of the use of seasonal variables in modelling the

breeding ranges of birds was recently mentioned (Brambilla

et al. 2012). However, we call for a cautious acknowledge-

ment of this issue, especially for the application of SDMs of

migratory bird species.

Acknowledgments We are very grateful to D. Alonso, J.M. Alonso,

T.P. Aparisi, J. Arizaga, P. Bergier, S. Blache, T. Borras, J. Cabrera,

T. Cabrera, J. Canadas, J. Calleja, C. de Jaime, E. del Val, A. Godino,

R. Hevia, R. Kilzer, J.J. Lorite, G. Lopez, H. Marki, T. Mihelic, M.

Quintana, S. Peregrina, P. Perret, J.J. Pfeffer, R. Probst, J. Rivas, J.C.

Senar and B. Stumberger for providing data on breeding and win-

tering areas. T. Gottschalk, K. Schidelko and two anonymous

reviewers gave valuable comments on an earlier draft of this

manuscript.

Conflict of interest The authors declare that they have no conflict

of interest.

References

Alonso D, Arizaga J (2004) El verderon serrano (Serinus citrinella)

en Navarra: parametros fenologicos y movimientos migratorios.

Munibe 55:95–112

Anderberg A, Anderberg AL (1997) Den virtuella floran. Naturhis-

toriska riksmuseet. http://linnaeus.nrm.se/flora/di/lamia/teucr/

teucscav.jpg

Araujo MB, Pearson RG (2005) Equilibrium of species’ distributions

with climate. Ecography 28:693–695

Araujo MB, Cabeza M, Thullier W, Hannah L, Williams PH (2004)

Would climate change drive species out of reserves? An

assessment of existing reserve-selection methods. Glob Change

Biol 10:1618–1626

Austin MP (2002) Spatial prediction of species distribution: an

interface between ecological theory and statistical modelling.

Ecol Mod 157:101–118

Aymı R, Tomas X (2003) Balanc de les activitats d’anellament

cientific d’ocells realitazades per L’insitut Catala d’Ornitologia

durant er periode 2000–2002. Rev Catalana Ornitol 20:108

J Ornithol

123

Page 11: Suitable, reachable but not colonised: seasonal niche duality in an endemic mountainous songbird

Baccetti N, Marki H (1997) Citril finch. In: Hagemeijer WJM, Blair

MJ (eds) The EBCC atlas of European breeding birds: their

distribution and abundance. Poyser, London

Baldwin RA (2009) Use of maximum entropy modeling in wildlife

research. Entropy 11:854–866

Bauer HG, Boschert M, Holzinger J (1995) Atlas der Winterverbrei-

tung der Vogel Baden-Wurttembergs. In: Holzinger J (ed) Die

Avifauna Baden-Wurttembergs, vol 5. Ulmer, Stuttgart

Benoit F, Marki H (2004) Premieres donnees sur l’aire de reproduc-

tion et la distribution hivernale du Venturon montagnard Serinus

citrinella au nord des Pyrenees. Nos Oiseaux 33:322–323

Bensch S (1999) Is the range size of birds constrained by their

migratory program? J Biogeogr 26:1225–1236

Berlit T (2005) Brutkartierung des Zitronengirlitz (Serinus citrinella)

in den Gebirgswaldern des Oberengadin und des oberen

Puschlav (Schweiz). Diploma thesis, Westfalische Wilhelms-

Universitat Munster

Bernis F, Bernis C (1963) Breve comentario sobre la invernada de

aves en la Cuenca del Ebro (enero 1962). Ardeola 8:228–231

Bocca M, Maffei G (1984) Gli uccelli della valle d’Aosta. Tipografia

la Vallee, Aosta

Borras A, Senar JC (2013) Verderon Serrano Serinus citrinella. In:

Martı R, Del Moral JC (eds) Atlas De Las Aves En Invierno En

Espana 2007–2010. Direccion General de Conservacion de la

Naturaleza-Sociedad Espanola de Ornitologıa, Madrid

Borras A, Blache S, Cabrera J, Cabrera T, Senar JC (2005) Citril

Finch (Serinus citrinella) populations at the north of the

Pyrenees may winter in the northeast of the Iberian Peninsula.

Aves 42:261–265

Borras A, Cabrera J, Colome X, Cabrera T, Senar JC (2010) Citril

Finches during the winter: patterns of distribution, the role of

pines and implications for the conservation of the species. Anim

Biodivers Conserv 33:89–115

Brambilla M, Ficetola GF (2012) Species distribution models as a tool

to estimate reproductive parameters: a case study with a

passerine bird species. J Anim Ecol 81:781–787

Brambilla M, Falco R, Negri I (2012) A spatially explicit assessment of

within-season changes in environmental suitability for farmland

birds along an altitudinal gradient. Anim Conserv 15:638–647

Brambilla M, Bassi E, Bergero V, Casale F, Chemollo M, Falco R,

Longoni V, Saporetti F, Vigano E, Vitulano S (2013) Modelling

distribution and potential overlap between Boreal Owl Aegolius

funereus and Black Woodpecker Dryocopus martius: implica-

tions for management and monitoring plans. Bird Conserv Int.

doi:10.1017/S0959270913000117

Carnaval AC, Moritz C (2008) Historical climate modelling predicts

patterns of current biodiversity in the Brazilian Atlantic forest.

J Biogeogr 25:1187–1201

Case TJ, Holt RD, McPeek MA, Keitt TH (2005) The community

context of species’ borders: ecological and evolutionary per-

spectives. Oikos 108:28–46

Chan LM, Brown JL, Yoder AD (2011) Integrating statistical genetic

and geospatial methods bring new power to phylogeography.

Mol Phyl Evol 59:523–537

Cheddadi R, Vendramin GG, Litt T, Francois L, Kageyama M,

Lorentz S, Laurent J-M, de Beaulieu J-L, Sadori L, Jost A, Lunt

D (2006) Imprints of glacial refugia in the modern genetic

diversity of Pinus sylvestris. Glob Ecol Biogeogr 15:271–282

Cramp S, Perrins CM (1994) The birds of the Western Palearctic, vol

VIII, Crows to finches. Oxford University Press, Oxford

De Grousaz G, Lebreton P (1963) Notes sur la migration du Venturon

montagnard (Carduelis citrinella L.) aux cols de Cou-Bretolet, et

sur son hivernage en Suisee et en France. Nos Oiseaux 27:46–61

Dejonghe JF (1991) Venturon montagnard Serinus citrinella. In:

Yeatman-Berthelot D (ed) Atlas des oiseaux de France en hiver.

Societe Ornithologique de France, Paris, pp 462–463

Dormann CF, McPherson J, Araujo MB, Bivand R, Bollinger J, Carl

G, Davies RG, Hirzel A, Jetz W, Kissling WD, Kuhn I,

Ohlemuller R, Peres-Neto PR, Reineking B, Schroder B, Schurr

FM, Wilson R (2007) Methods to account for spatial autocor-

relation in the analysis of species distributional data: a review.

Ecography 30:609–628

Dvorak M, Ranner A, Berg HM (1993) Atlas der Brutvogel

Osterreichs. Ergebnisse der Brutvogelkartierung 1981–1985 der

Osterreichischen Gesellschaft fur Vogelkunde.

Umweltbundesamt

Elith J, Phillips SJ, Hastie T, Dudık M, Chee YE, Yates CJ (2011) A

statistical explaination of MaxEnt for ecologists. Divers Distrib

17:43–57

Feldner J, Rass P (1999) Zwei neue Brutvogelarten fur Karnten:

Zwergschnapper (Ficedula parva) und Zitronengirlitz (Serinus

citrinella). Carinthia II 189(109):241–246

Ficetola GF, Thuiller W, Miaud C (2007) Prediction and validation of

the potential global distribution of a problematic alien invasive

species: the American bullfrog. Divers Distrib 13:476–485

Fielding AH, Bell JF (1997) A review of methods for the assessment

of prediction errors in conservation presence/absence models.

Environ Conserv 24:38–49

Fornasari L, Carabela M, Corti W, Pianezza F (1998) Autumn

movements of Citril Finches Serinus citrinella in the southern

Alps. Ring Migr 19:23–29

Forschler MI (1997) Zum Wintervorkommen 1995/1996 des Zitro-

nengirlitzes Serinus citrinella in den Hochlagen des Nordsch-

warzwaldes. Naturkundl Beob Kreis Freudenstadt 2:24

Forschler MI (2001) Witterungsbedingte Ausweichbewegungen des

Zitronengirlitzes Serinus citrinella im Nordschwarzwald. Orni-

thol Beob 98:209–214

Forschler MI (2006) Absence of insular density inflation in Corsican

Finches Carduelis [citrinella] corsicanus. Acta Ornithol 41:171–174

Forschler MI (2007) Seasonal variation in the diet of Citril Finches

Carduelis citrinella: are they specialist or generalists? Eur J

Wildl Res 53:190–194

Forschler MI, Kalko EKV (2006a) Macrogeographic variations in

food choice of mainland Citril Finches Carduelis [citrinella]

citrinella versus insular Corsican (Citril) Finches Carduelis

[citrinella] corsicanus. J Ornithol 147:441–447

Forschler MI, Kalko EKV (2006b) Breeding ecology and nest site

selection in allopatric mainland Citril Finches Carduelis [citri-

nella] citrinella and insular Corsican Finches Carduelis [citri-

nella] corsicanus. J Ornithol 147:553–564

Forschler MI, Senar JC, Perret P, Bjorklund M (2009) The species

status of the Corsican finch Carduelis corsicana assessed by

three genetic markers with different rates of evolution. Mol Phyl

Evol 52:234–240

Forschler MI, Shaw DN, Bairlein F (2011) Deuterium analysis reveals

potential origin of the Fair Isle Citril Finch Carduelis citrinella.

Bull BOC 131:189–191

Fortin MJ, Keitt TH, Maurer BA, Taper ML, Kaufmann DM, Blackburn

TM (2005) Species’ geographic ranges and distributional limits:

pattern analysis and statistical issues. Oikos 108:7–17

Fourcade Y, Engler JO, Besnard AG, Rodder D, Secondi J (2013)

Confronting expert-based and modelled distributions for species

with uncertain conservation status: a case study from the

corncrake (Crex crex). Biol Conserv 167:161–171

Gaston KJ (2003) The structure and dynamics of geographic ranges.

Oxford University Press, Oxford

Geister I (1983) European news. Brit Birds 76:276

Geister I (1995) Ornitoloski atlas Slovenije. Razsirjenost gnezdilk.

DZS

Genard M, Lescourret F (1987) Organisation du peuplement avien

d’une foret des Pyrenees orientales francaises. Le Gerfaut

77:463–476

J Ornithol

123

Page 12: Suitable, reachable but not colonised: seasonal niche duality in an endemic mountainous songbird

Glutz von Blotzheim UN, Bauer KM (1997) Handbuch der Vogel

Mitteleuropas Band 14. Aula, Wiebelsheim, pp 501–532

Godsoe W (2010) I can’t define the niche but I know it when I see it: a

formal link between statistical theory and the ecological niche.

Oikos 119:53–60

Gorman ML (1979) Island ecology. Chapman and Hall, London

Gregori J (1977) Ekoloski in favnisticni pregeld pticev severozahodne

Slovenije. Larus 29–30:70

Grinnell J (1917) Field tests of theories concerning distributional

control. Am Nat 51:115–128

Guisan A, Thuiller W (2005) Predicting species distributions: offering

more than simple habitat models. Ecol Lett 8:993–1003

Heikkinen RK, Luoto M, Araujo MB, Virkkala R, Thullier W, Sykes

MT (2006) Methods and uncertainties in bioclimatic envelope

modelling under climate change. Prog Phys Geogr 30:751–777

Heuck C, Brandl R, Albrecht J, Gottschalk T (2013) The potential

distribution of the red kite in Germany. J Ornithol 154:911–921

Hijmans RJ, Cruz JM, Rojas E, Guarino L (2001) DIVA–GIS, version

1.4. A geographic information system for the management and

analysis of genetic resources data. Manual. International Potato

Center and International Plant Genetic Resources Institute

Hijmans RJ, Cameron SE, Parra JL, Jones PG, Jarvis A (2005) Very

high resolution interpolated climate surfaces for global land

areas. Int J Climatol 25:1965–1978

Holt RD, Keitt TH (2005) Species’ borders: a unifying theme in

ecology. Oikos 108:3–6

Holzinger J, Dorka V (1997) Zitronengirlitz. In: Holzinger J (ed) Die

Vogel Baden-Wurttembergs. Band 3.2. Eugen Ulmer, Stuttgart,

pp 584–603

Huggett RJ (2004) Fundamentals of Biogeography, 2nd edn. Routl-

edge, London

Hutchinson GE (1957) Concluding remarks. Cold Spring Harb Symp

Quant Biol 22:415–427

Hutchinson GE (1978) An introduction to population ecology. Yale

University Press, New Haven

Hyndman T (2008) The Citril Finch on Fair Isle: a new British bird.

Bird World 21:243–249

Jakob SS, Heibl C, Rodder D, Blattner FR (2010) Population demog-

raphy influences climatic niche evolution: evidence from diploid

American Hordeum species (Poaceae). Mol Ecol 19:1423–1438

Jaynes ET (1957) Information theory and statistical mechanics. Phys

Rev 106:620–630

Jiguet F, Barbet-Massin M, Chevallier D (2011) Predictive distribu-

tion models applied to satellite tracks: modelling the western

African winter range of European migrant Black Storks Ciconia

nigra. J Ornithol 152:111–118

Kaiser HF (1958) The varimax criterion for analytic rotation in factor

analysis. Psychometrika 23:187–200

Keller FC (1890) Ornis Carinthiae. Kleinmayr, Klagenfurt

Kozak KH, Wiens JJ (2007) Climatic zonation drives latitudinal

variation in speciation mechanisms. Proc R Soc Lond B

274:2995–3003

Kozak KH, Graham CH, Wiens JJ (2008) Integrating GIS–based

environmental data into evolutionary biology. Trends Ecol Evol

23:141–148

Kremen C, Cameron A, Moilanen A, Phillips SJ, Thomas CD,

Beentje H, Dransfield J, Fisher BL, Glaw F, Good TC, Harper

GJ, Hijmans RJ, Lees DC, Louis E, Nussbaum RA, Raxworthy

CJ, Razafimpahanana A, Schatz GE, Vences M, Vieites DR,

Wright PC, Zjhra ML (2008) Aligning conservation priorities

across taxa in Madagascar with high-resolution planning tools.

Science 320:222–226

Landbeck CL (1834) Systematische Aufzahlung der Vogel Baden-

Wurttembergs mit Angabe ihrer Aufenthaltsorter und ihrer

Strichzeit. Cotta, Tubingen

Laube I, Graham CH, Bohning-Gaese K (2013) Intra-generic species

richness and dispersal ability interact to determine geographic

ranges of birds. Glob Ecol Biogeogr 22:223–232

Liu C, Berry PM, Dawson TP, Pearson RG (2005) Selecting

thresholds of occurrence in the prediction of species distribu-

tions. Ecography 28:385–393

Lobo JM, Jimenez-Valverde A, Real R (2008) AUC: a misleading

measure of the performance of predictive distribution models.

Glob Ecol Biogeogr 17:145–151

Mackay BG, Lindemayer DB (2001) Towards a hierarchical frame-

work for modelling the spatial distribution of animals. J Biogeogr

28:1147–1166

Maestri F, Voltolini L, Lo Valvo F (1989) Biologia riproduttiva di

una comnuita’ di fringillidi in un mugeto dell Alpe Retiche

(Sondrio). Riv Ital Ornitol 59:159–171

Marini MA, Barbet-Massin M, Lopes LE, Jiguet F (2010) Predicting

the occurrence of rare Brazilian birds with species distribution

models. J Ornithol 151:857–866

Marki H (1976) Brutverbreitung und Winterquartier des Zitronenzei-

sigs Serinus citrinella nordlich der Pyrenaen. Ornithol Beob

73:67–88

Marki H, Adamek G (2013) Nahrungsbedingt wechselnde Winter-

habitate des Zitronengirlitzes Serinus citrinella in Sudfrankreich.

Ornithol Beob 110:437–452

Matvejev SD (1981) Laska konopeljscica Serinus citrinella. Acro-

cephalus 2:59

McInnes L, Purvis A, Orme CDL (2009) Where do species’ geographic

ranges stop and why? Landscape impermeability and the Afro-

tropical avifauna. Proc R Soc Lond B 276:3063–3070

Mingozzi T, Boano G, Pulcher C (1988) Atlante degli uccelli

nidificanti in Piemonte e Val d’Aosta 1980–1984. Monografie

VIII, Museo Regionale di Scienze Naturali di Torino

Moltoni E (1969) Gli uccelli del Parco nazionale dello Stelvio.

Tipografia, Sondrio

Moritz D, Bachler A (2001) Die Brutvogel Osttirols. Ein kommen-

tierter Verbreitungsatlas. Author’s edition

Newton I (2003) The speciation and biogeography of birds.

Academic, Waltham

Peterson AT, Vieglais DA (2001) Predicting species invasions using

ecological niche modeling: new approaches from bioinformatics

attack a pressing problem. Bioscience 51:363–371

Phillips SJ (2008) Transferability, sample selection bias and back-

ground data in presence-only modelling: a response to Peterson

et al. (2007). Ecography 31:272–278

Phillips SJ, Dudık M (2008) Modeling of species distributions with

MaxEnt: new extensions and comprehensive evaluation. Eco-

graphy 31:161–175

Phillips SJ, Dudık M, Schapire RE, (2004) A maximum entropy

approach to species distribution modeling. In: Proceedings of the

21st international conference on machine learning, Banff

Phillips SJ, Anderson RP, Schapire RE (2006) Maximum entropy

modeling of species geographic distributions. Ecol Model

190:231–259

Phillips SJ, Dudik M, Elith J, Graham CH, Lehmann A, Leathwick J,

Ferrier S (2009) Sample selection bias and presence-only

distribution models: implications for background and pseudo-

absence data. Ecol Appl 19:181–197

Praz JC, Oggier PA (1973) Sur l’hivernage due Venturon montagnard

en Valais. Nos Oiseaux 32:109–112

Probst R (2012) Warum brutet der Zitronenzeisig (Carduelis

citrinella) in Karnten geanu am Dobratsch. Carinthia II

122:493–504

R Development Core Team (2010) R: a language and environment for

statistical computing. R Foundation for Statistical Computing,

Vienna. ISBN 3–900051–07–0. http://www.R-project.org

J Ornithol

123

Page 13: Suitable, reachable but not colonised: seasonal niche duality in an endemic mountainous songbird

Rodder D, Kielgast J, Bielby J, Schmidtlein S, Bosch J, Garner TWJ,

Veith M, Walker S, Fisher MC, Lotters S (2009a) Global

amphibian extinction risk assessment for the panzootic chytrid

fungus. Diversity 1:52–66

Rodder D, Schmidtlein S, Veith M, Lotters S (2009b) Alien invasive

slider turtle in unpredicted habitat: a matter of niche shift or

predictors studied? PLoS ONE 4:e7843

Rodder D, Engler JO, Bonke R, Weinsheimer F, Pertel W (2010)

Fading of the last giants: an assessment of habitat availability of

the Sunda gharial Tomistoma schlegelii and coverage with

protected areas. Aquat Conserv 20:678–684

Rodder D, Lawing AM, Flecks M, Ahmadzadeh F, Dambach J,

Engler JO, Habel J-C, Hartmann T, Hornes D, Ihlow F,

Schidelko K, Stiels D, Polly PD (2013) Evaluating the signif-

icance of paleophylogeographic species distribution models in

reconstructing quaternary range-shifts of Nearctic chelonians.

PLoS ONE 8:e72855

Schidelko K, Stiels D, Rodder D (2011) Historical stability of

diversity patterns in African estrildid finches (Estrildidae). Biol J

Linn Soc 102:455–470

Smith SA, Donoghue MJ (2010) Combining Historical Biogeography

with Niche Modeling in the Caprifolium Clade of Lonicera

(Caprifoliaceae, Dipsacales). Syst Biol 59:322–341

Soberon J (2007) Grinnellian and Eltonian niches and geographic

distributions of species. Ecol Lett 10:1115–1121

Soberon J, Nakamura M (2009) Niches and distributional areas:

concepts, methods and assumptions. Proc Natl Acad Sci USA

106:19644–19650

Soberon J, Peterson AT (2005) Interpretation of models of funda-

mental ecological niches and species’ distributional areas.

Biodivers Inf 2:1–10

Spina F, Volponi S (2008) Atlante Della Migrazione Degli Uccelli in

Italia. 2. Passeriformi. Roma: Ministero dell’ Ambiente e della

Tutela del Territorio e del Mare, Instituto Superiore per la

Protezione e la Ricerca Ambientale (ISPRA)

Stiels D, Schidelko K, Engler JO, van den Elzen R, Rodder D (2011)

Predicting the potential distribution of the invasive common

waxbill Estrilda astrild (Passeriformes: estrildidae). J Ornithol

152:769–780

Svensson L, Grant PJ, Mullarney K (2009) Collins bird guide. Harper

Collins, New York

Swets K (1988) Measuring the accuracy of diagnostic systems.

Science 240:1285–1293

Thorup K (2006) Does the migration programme constrain dispersal

and range sizes of migratory birds? J Biogeogr 33:1166–1171

Vaurie C (1959) The birds of the palearctic fauna. Passeriformes.

Witherby, London

Venables WN, Ripley BD (2002) Modern applied statistics with S,

4th edn. Springer, New York

von Kettner WF (1849) Darstellung der ornithologischen Verhaltnisse

des Großherzogtums Baden. Beitr Rheinischer Naturgesch

1:39–100

Wisz MS, Hijmans RJ, Peterson AT, Graham CH, Guisan A, NPSDW

Group (2008) Effects of sample size on the performance of

species distribution models. Divers Distrib 14:763–773

Zink G, Bairlein F (1995) Zug europaischer Singvogel. Band 3. Aula,

Wiebelsheim

J Ornithol

123