Content uploaded by Michele Panuccio
Author content
All content in this area was uploaded by Michele Panuccio on Jan 27, 2015
Content may be subject to copyright.
1 23
Ecological Research
ISSN 0912-3814
Volume 30
Number 1
Ecol Res (2015) 30:181-190
DOI 10.1007/s11284-014-1226-2
Motion capacity, geography and ecological
features explain the present distribution of
a migratory top predator
Michele Panuccio, Giuseppe Lucia,
Nicolantonio Agostini, Dario Ottonello
& Giuseppe Bogliani
1 23
Your article is protected by copyright and all
rights are held exclusively by The Ecological
Society of Japan. This e-offprint is for personal
use only and shall not be self-archived
in electronic repositories. If you wish to
self-archive your article, please use the
accepted manuscript version for posting on
your own website. You may further deposit
the accepted manuscript version in any
repository, provided it is only made publicly
available 12 months after official publication
or later and provided acknowledgement is
given to the original source of publication
and a link is inserted to the published article
on Springer's website. The link must be
accompanied by the following text: "The final
publication is available at link.springer.com”.
ORIGINAL ARTICLE
Michele Panuccio •Giuseppe Lucia •Nicolantonio Agostini
Dario Ottonello •Giuseppe Bogliani
Motion capacity, geography and ecological features explain the present
distribution of a migratory top predator
Received: 14 July 2014 / Accepted: 24 November 2014 / Published online: 6 December 2014
The Ecological Society of Japan 2014
Abstract Presence and distribution of ecological barriers
shapes the distribution of migratory birds as well as any
other living organism. In Italy, short-toed snake eagles
(Circaetus gallicus) breed in the northern and western
areas of the peninsula but the species is rare in the south
or the islands. The Italian population of this species
migrates across the Mediterranean at the Strait of
Gibraltar rather than crossing the large stretch of sea
between Sicily and Tunisia. This suggests that, in Italy,
fall migration is oriented south–north and spring
migration north–south. In this paper we test the
hypothesis that the accessibility of the suitable habitat
area along the Italian Peninsula is in relation to the
geographical migration pattern of the studied species.
We integrated information from the movement ecology,
the geography and the traditional ecological features in
order to provide an ecological explanation of the current
biogeographical pattern of our model species. We com-
pared statistical models with and without latitude as a
predictor. Each model was based on ecological and
geographical variables, including land use, prey avail-
ability, spatial distribution of environmental elements
(patch analysis), geomorphology, and geography. These
models predict two patterns of suitability for short-toed
snake eagles in Italy. Our results suggest that the
abundance of this species increases with latitude despite
the existence of large areas of suitable habitat in
southern Italy. We suggest that the actual distribution of
the short-toed snake eagle in Italy is influenced by the
particular migration path used by this population, sup-
porting the hypothesis that this species is still colonizing
the Italian Peninsula through an unexpected coloniza-
tion direction from north to south.
Keywords Colonization direction ÆAccessibility Æ
Migratory detour ÆEcological barriers ÆShort-toed
snake eagle ÆCircaetus gallicus ÆItaly
Introduction
The present distribution of an organism is the result of
the interaction between ecological, movement-related
and geographical features (Cumming et al. 2012). In
particular the ecological niche and the probability of
colonization affect the distributional limits of a species
(Begon et al. 2006). Ecological niche is related to several
different factors, including foraging and nesting oppor-
tunities. The probability of colonization depends on the
dispersal abilities of the species and its dispersal behav-
iour, as well as geographic features such as natural
barriers (Alerstam 1990; Tellerı
´a et al. 2009). Therefore,
the current biogeographical patterns might be explained
by hypotheses that are related to environmental, mor-
phological and historical data (Cumming et al. 2012).
Europe has been colonized by several species of birds
since the end of the last glacial age about 10,000 years
ago. Several species have colonized areas located at
higher latitudes through a northern extension of their
breeding ranges (Mayr and Meise 1930; Bruderer and
Salewski 2008). From the end of last glaciation, Palae-
arctic migrants have followed a colonization path from
south to north and were assisted by the process of dis-
persal (Alerstam 1990; Berthold 2001; Rappole and
Jones 2002).
Electronic supplementary material The online version of this article
(doi:10.1007/s11284-014-1226-2) contains supplementary material,
which is available to authorized users.
M. Panuccio (&)ÆG. Bogliani
DSTA-Department of Earth and Environmental Sciences,
University of Pavia, Via Ferrata 9, 27100 Pavia, Italy
E-mail: panucciomichele@gmail.com
Tel.: 0039-349-2340719
M. Panuccio ÆG. Lucia ÆN. Agostini
MEDRAPTORS (Mediterranean Raptor Migration Network),
Via Mario Fioretti 18, 00152 Rome, Italy
D. Ottonello
DAIS-Department of Environmental Sciences, Informatics and
Statistics, Ca
`Foscari University of Venice, Dorsoduro 2137,
30123 Venice, Italy
Ecol Res (2015) 30: 181–190
DOI 10.1007/s11284-014-1226-2
Author's personal copy
The short-toed snake eagle (Circaetus gallicus)isa
widespread summer breeder in Europe and in the
Mediterranean basin and overwinters in a wide area of
Africa south of Sahara (Cramp and Simmons 1980;
Ferguson-Lees and Christie 2001). It is a highly spe-
cialized raptor, preying predominantly on snakes (up to
100 %; Ferguson-Lees and Christie 2001); therefore, as
a specialized predator, its distribution should coincide
with the distribution of its prey (Newton 1998). A study
carried out in Spain (Moreno-Rueda and Pizarro 2007)
showed that short-toed snake eagles are found mainly in
areas with shrub-lands, where they can easily locate and
capture their prey, and in forests, which provide suitable
nesting habitat. Eagle presence also correlates with
snake species richness (Moreno-Rueda and Pizarro
2007). Research in Greece also showed that the presence
of short-toed snake eagle is linked to forests for nest
building and to open areas like pasturelands and culti-
vation for foraging (Bakaloudis et al. 1998,2005). The
European population of short-toed snake eagle is esti-
mated to be less than 10,000 pairs and its status is Rare;
its low reproductive rate and high diet specialization
makes it vulnerable to extinction (Birdlife International
2004). The population in Italy is estimated to 350–400
breeding pairs (Birdlife International 2004), but recent
counts at some migration hotspots have shown that their
numbers are increasing (Baghino and Premuda 2007;
Panuccio et al. 2012,2013). Most of the breeding pairs in
Italy are distributed in northern Italy and along the
western slope of central Italy (Campora and Cattaneo
2006). In Southern Italy few tens of pairs do breed but it
seems that the species has been expanding in recent years
(Mellone et al. 2011). The short-toed snake eagle is a
broad-winged raptor that uses soaring-gliding flight ra-
ther than powered flight; as a result this species tends to
migrate over land and to avoid water crossing where
thermal currents are absent and it would be forced to use
active flight (Kerlinger 1989). Individuals belonging to
the Italian population migrate both during spring and
autumn across the Strait of Gibraltar (14 km of water
crossing) rather than taking the Central Mediterranean
flyway that is shorter, although the water crossing is
longer (at least 130 km; Agostini et al. 2002a,b,2004;
Mellone et al. 2011). Similarly, individuals breeding in
Greece migrate through Turkey and cross the sea at
Bosphorus (Panuccio et al. 2012). Since most breeding
pairs of this species in Italy are located along its
migration route, Agostini and Mellone (2008) suggested
that the short-toed snake eagle may still be colonizing
Italy from the western part of its European breeding
range and that, therefore, the population has not yet
reached its carrying capacity.
The distribution of living organisms is a widely
studied subject in ecology but, on the other hand, the
interaction between distribution and migration routes is
a poorly investigated topic. In this paper we test the
hypothesis that the actual distribution of short-toed
snake eagle in Italy reflects the colonization history and
is strongly influenced by the migration route more than
other ecological parameters such as prey availability and
the spatial distribution of vegetation and morphology.
For this purpose in the analysis we considered several
ecological (land use and prey richness), topographic
(altitude), and latitudinal parameters in order to deter-
mine which affect the breeding distribution of short-toed
snake eagle in Italy and to determine whether some areas
of Southern Italy are not suitable for short-toed snake
eagles or, although if suitable, they have not been col-
onized or re-colonized yet.
Methods
Study area
The study was performed in Italy (301,302 km
2
)
including the large islands of Sicily and Sardinia. On the
01-Jan-2010 the human population was 60,340,328
(ISTAT, website: http://demo.istat.it/pop2010/index.
html). The climate varies considerably with latitude. In
the south it is warm, with almost no rain in summer,
while in the Northern part of the Peninsula temperature
is cool with rainfall more evenly distributed throughout
the year. The mean altitude in Italy is 337 m above sea
level. According to Corine Land Cover 2000, Italy is
composed of 152,435 km
2
of cultivated lands (50.6 %),
18,835 km
2
of permanent pasture lands (6.3 %),
79,425 km
2
of woodlands (26.4 %), 22,555 km
2
of
shrubs (7.5 %), 3,827 km
2
of wetlands (1.3 %) and
14,340 km
2
of urbanized areas (4.8 %). In the lowlands
agriculture is very intensive and devoted mainly to
monoculture, while on the hills and mountains tradi-
tional and less intensive agriculture is still practiced al-
though land abandonment is spreading. Mainland Italy
may be divided into four major geographical and vege-
tational areas: the Alps and the Padana plain in the
north, the Apennines and the coastal slopes (7,456 km
of coastlines) in the center and the south. The Alpine
area is relatively unspoiled, with deciduous, mixed
coniferous forests, alpine pastures above the timberline,
and snowfields and glaciers on higher peaks. The Po
valley is the largest lowland area in Italy, mainly covered
by intensive agriculture. Large areas of the Apennine
mountains are still covered by semi-natural deciduous
forests, mostly oak (Quercus spp.), beech (Fagus sylvat-
ica) and sweet chestnut (Castanea sativa), although large
areas have been cleared for agriculture and pasture.
Coastal areas in Italy are dominated by shrub and in
particular by typical Mediterranean scrub. The pro-
tected areas in Italy cover a surface of 13,000 km
2
(4.3 %; Brandmayr 2002).
Sampling units
The geographical grid we used to divide the Italian
peninsula in several sampling units is the same used from
182
Author's personal copy
Military Geographical Institute (IGM) to draw the
1:25,000 maps of Italy in UTM projections; we consid-
ered 3,541 square cells of 10 ·10 km. Because the Ital-
ian peninsula is located over three ‘‘zones’’ of 6 degrees
in latitude, at the transition from one zone to another
the grid is compressed and rotated, so some squares (less
than 5 % of the total) have an area less than 100 km
2
.A
similar limitation of extension involves also coastal cells.
To avoid effects due to the difference in size among cells,
we included variables that used percent coverage instead
of absolute cell surface. To determine prey availability
(reptile species richness), we used the grid and data set
used by Sindaco et al. (2006) for the atlas of the reptiles
and amphibians of Italy made available by SHI (Soci-
etas Herpetologica Italica).
Presence data of short-toed snake eagle
Short-toed snake eagle presence data was obtained by
consulting several atlas of bird distribution in Italy; we
assumed that the coverage effort for each local atlas was
similar since all atlas were realized following similar
guidelines (Fig. 1; Mingozzi et al. 1988; Fraissinet and
Kalby 1989; Brichetti and Fasola 1990; Meschini and
Frugis 1993; Gruppo Vicentino Studi Ornitologici
‘‘NISORIA’’ 1994; Boano et al. 1995; Ravasini 1995;
Tellini Florenzano et al. 1997; Bon et al. 1999; Gellini
and Ceccarelli 2000; Fracasso et al. 2003; Bordignon
2004; Pedrini et al. 2005; Bionda and Bordignon 2006;
Giacchini 2007; Mezzavilla and Bettiol 2007; Ientile and
Massa 2008; La Gioia 2009) and also by using personal
observations of nest locations. In the atlas, the presence
of species in each sample unit is listed in three ways:
certain, probable and possible, following the indications
of EBCC (Hagemeijer and Blair 1997). In this study, we
considered only the first two categories to classify pre-
sence. In most cases we used the same grid as that used
in the atlas; when the atlas grid was smaller than ours,
we assigned a presence value to our cell if at least one of
the atlas cells were included in our sampling units. At the
end of data collecting we had 94 cells (10 ·10 km) with
presence data.
Predictive variables
To model the relationship between habitat features and
short-toed snake eagle presence, we first considered 30
environmental variables (Table S1) grouped into 5 cat-
egories: land use, prey availability, spatial distribution of
environmental elements (patch analysis), geomorphol-
ogy, and geography (latitude and longitude). Variables
representing land use were obtained by Corine Land
Cover 2000 to the third level. We first grouped land use
categories that had similar ecological significance for the
study species into seven new variables: forests, shrubs,
crops and orchards, meadows and pastures, populated
areas, bare rock, and wetlands. For each cell we calcu-
lated the percentage coverage of land use variables. Prey
availability was measured as the number of reptile spe-
cies (snakes and lizards, excluding Gekkonidae for lack
of daytime activities of this family) present in each cell.
Data on the presence of reptiles were derived from the
atlas of Sindaco et al. (2006). Analysis of the spatial
distribution of land use patches was carried out with the
Esri ArcMap 9.2 Patch Analyst extension (Rempel and
Carr 1999). This analysis investigates the geometry of
selected patches (here, land use polygons) and perime-
ters (for details on these measures see Elkie et al. 1999).
Data on the geomorphology were derived from digital
elevation model (DEM) of Italy with a spatial resolution
of 250 m. In each cell six measures of altitude were
calculated: minimum, maximum, average, median,
standard deviation, and the coefficient of variation.
Geographical variables were represented by values of
latitude and longitude of the cell centroid. In particular
we used the latitude as a proxy of the distance from the
main migration route of the studied species. Finally we
squared the following variables to evaluate if the values
of forest cover, mean altitude and mean patch fractal
dimension were non-monotonically related to the suit-
ability of the study area for the short-toed snake eagle.
Statistical models and procedure
We compared models with and without latitude as a
predictor to test the hypothesis that our studied species
Fig. 1 Distribution map of short-toed snake eagle in Italy,
according to the data compilation from different local atlases.
Each dot denotes presence in a 10 ·10 km
2
183
Author's personal copy
is still colonizing Italy along north–south gradient. Here
following are the details of the procedure we followed to
build the different models.
We first tested for collinearity of variables within the
categories inside the sampling units in order to select
variables to use in the models. To do this we tested the
normality of variables with Kolmogorov–Smirnov test,
and if any variables were not normally distributed we
transformed the data (square root, arcsine or natural
logarithm depending on the variable). We used a logistic
regression to test which predictive variables affect the
distribution of the species in Italy. In order to verify
which variables influenced the presence of short-toed
snake eagles we compared variables in squares with
presence data (n= 94) with the same number of cells
without presence (n= 94); these cells were randomly
selected among those in which the species was not
present according to the distribution reported in the
atlas we consulted. We also generated new variables by
squaring those variables that were non-monotonically
related to the presence of the species. We retained a
priori the latitude as predictor since it is in our starting
hypothesis, but also the three variables that have been
previously considered significant in explaining short-
toed snake eagles presence. These variables are: prey
availability, proportion of forests, proportion of shrubs
(Sa
´nchez-Zapata and Calvo 1999; Bustamante and
Seoane 2004; Bakaloudis et al. 2005; Moreno-Rueda
and Pizarro 2007;Lo
´pez-Iborra et al. 2011). After that
we tested 23 random combination of models (all con-
taining those four retained variables) excluding those
variables who were auto-correlated. The starting set of
variables for the logistic regression is shown in Table 1.
Model selection was made choosing the model with the
smallest value of Akaike information criterion (AIC;
Manly et al. 2002). As model B we run the same selected
model without the latitude as predictor. In the models
we forced all of the chosen variables in the same run
with the Enter method in SPSS software.
We then obtained a value of probability of presence
between 0 and 1 for each cell that varies depending on
the model and we used this difference to compare each
model including ‘‘Y_CENTR’’ and the corresponding
model without this variable.
Moreover, to test the results of logistic regression
models we generated a further model based on boosted
regression trees analysis (BRT). Unlike the logistic
regression, BRT fits a large number of relatively simple
models which are then combined for prediction. BRT
uses two algorithms: regression trees and boosting builds
and combines a collection of models (De’ath and Fab-
ricius 2000; Friedman et al. 2000; Schapire 2003;
Leathwick et al. 2006; Elith et al. 2006,2008). For the
BRT model we used the same variables used in the
regression model and we chose the best model selected
according to lower deviance value and the area under
curve (AUC) of its receiver operating characteristic
(ROC).
Results
Variables used to run the analysis are shown in Table 1.
The best model (Model A) with the lowest value of AIC
included nine variables (Table 2). This model achieved
an overall value of correct percentage of 77.7 % (84.1 %
for absence and 64.9 % for presence). The AUC value of
the ROC curve was 0.848. We tested the same model
without the latitudinal variable; this second model
(Model B) showed a higher AIC value and included
eight variables (Table 2). This model achieved an overall
value of correct percentage of 74.5 % (81.5 % for ab-
Table 2 Logistic regression statistics (Wald), parameter estimates (B) and significance (P) of the variables included in the short-toed snake
eagle distribution Model A and B, in Italy
Model A Model B
B Wald PB Wald P
Latitude 0.000 20.655 0.000 – – –
Mean altitude 0.001 1.943 0.163 0.002 3.453 0.063
(Mean altitude)
2
0.000 3.280 0.070 0.000 8.125 0.004
Proportion of forest 10.547 22.306 0.000 10.960 24.842 0.000
(Proportion of forest)
2
8.895 13.975 0.000 8.390 12.763 0.000
Number of reptile species 0.185 19.102 0.000 0.212 28.727 0.000
Proportion of shrubs 1.937 2.785 0.095 488.816 0.061 0.805
Mean patch fractal dimension 534.33 0.766 0.381 187.219 0.710 0.399
(Mean patch fractal dimension)
2
201.555 0.738 0.390 0.308 0.704 0.402
Table 1 Variables used to perform logistic regression
Category Variables
Land use Proportion of meadows and pastures
Proportion of crops and orchards
Proportion of forests
(Proportion of forests)
2
Proportion of shrubs
Prey availability Number of reptile species
(snakes and lizards)
Spatial distribution of
environmental elements
(land use—patch analysis)
Mean patch edge
Mean patch fractal dimension
(Mean patch fractal dimension)
2
Geomorphology Mean altitude
(Mean altitude)
2
Geography Latitude
184
Author's personal copy
sence and 60.4 % for presence); the AUC value of the
ROC curve was 0.824. Using the results of these two
models we created two different distribution maps of
probability. Model A, which includes latitude as a var-
iable, predicts that the potential distribution of short-
toed snake eagles is mainly restricted to central and
northern Italy (Fig. 2a). By contrast, Model B, which
does not include latitude as a variable, predicts a wider
potential distribution that includes southern Italy
(Fig. 2b).
Similarly, plotting the probability of occurrence of
short-toed snake eagles in the sample units in relation to
latitude, a clear difference between the two models ap-
pears. Model B predicted a more homogeneous proba-
bility of species distribution along the latitudinal
gradient (Fig. 3b). On the other hand, according to the
same model tested with latitude as predictor, the prob-
ability of occurrence of the studied species is restricted to
the northern and the central part of the Italian penin-
sula (Fig. 3a).
The model obtained using boosted regression trees
(deviance = 0.244, deviance Cross-validation esti-
mate = 0.161; SE = 0.006) confirmed that the three
most important predictive variables for the presence of
short-toed snake eagle are latitude, mean altitude and
the number of prey species (Fig. 4). For all these
parameters, increasing values increased the fitted func-
tion of the model. Plotting the interaction between the
mean altitude and the number of prey species we ob-
tained a maximum fitted value of 0.57 (Fig. 5) in cor-
respondence of the large number of species and a
medium–high value at mean altitude. The increase of
value of the forest variable was also linked to an increase
in model fitting despite its low contribution in the model.
Discussion
Ecological parameters
This study shows that, of the variables that potentially
affect the breeding distribution of the short-toed snake
eagle, the strongest predictors are prey richness, forests,
and the mean altitude of the territory. This result was
obtained using both models A and B as well as the
boosted regression tree model. Forests are important for
short-toed snake eagles because the species nests in these
habitats; this is in agreement with previous research
(Sa
´nchez-Zapata and Calvo 1999; Bustamante and
Seoane 2004; Bakaloudis et al. 2005;Lo
´pez-Iborra et al.
2011). As shown by the high Wald value of the variable
‘‘square root forest’’ (Table 2), the increase of forested
areas is important for the presence of short-toed snake
eagles; however, there is a threshold value beyond which
the extension of forests becomes limiting for this species,
since snake eagles hunt in open areas where reptilian
prey are probably abundant and most accessible (see
Moreno-Rueda and Pizarro 2007; Bakaloudis 2009;
Lo
´pez-Iborra et al. 2011). Mean altitude may be
important in the BRT model because, in Italy, forests
are almost absent from lowlands, such as the Po valley,
and because agriculture on the Italian plains is inten-
sive (monoculture). Therefore, lower altitudes do not
Fig. 2 Probability classes of the short-toed snake eagle distribution
according to Model A (a) and Model B (b) in Italy
185
Author's personal copy
provide suitable nesting habitat for short-toed snake
eagles in Italy. Prey richness was the other variable that
explained eagle distribution in all models. A previous
study of short-toed snake eagles in southern Spain found
a relationship between the presence of eagles and species
richness of reptiles (Moreno-Rueda and Pizarro 2007).
Despite its high trophic specialization, the short-toed
snake eagle behaves as a taxonomic generalist within this
prey type, and selects optimal prey sizes regardless of the
species (Gil and Pleguezuelos 2001).
What factors, then, explain the correlation between
prey species richness and eagle distribution? Different
hypotheses have been proposed to explain this correla-
tion. One possibility is that the abundance of reptile
individuals correlates with reptile species richness and
the abundance of individuals determines the presence of
raptors (Evans et al. 2005; Moreno-Rueda and Pizarro
2007). Intra- and inter-species interactions have also
been proposed. It has also been hypothesized that the
presence of short-toed snake eagles increases reptile
species richness by controlling the most abundant spe-
cies and reducing the intensity of competition in the
reptile community through a top-down process (Mo-
reno-Rueda and Pizarro 2007).
Differently from previous research, our model did not
reveal shrubland to be an important variable. This may
be because other variables have a higher importance in
explaining the probability of the occurrence of the
studied species in the study area (i.e. number of reptile
species and proportion of forests).
The role of the latitude
In our first Model (A) latitude also had a significant
effect on the breeding distribution of short-toed snake
eagles in Italy. Its effect is the opposite of that docu-
mented in Spain (Lo
´pez-Iborra et al. 2011), where lati-
tudinal increase was shown to have a negative effect on
the distribution of the species. By contrast, we found
that eagle presence seems to increase with latitude.
However, the distribution of the short-toed snake eagle
in Europe does not lead to the conclusion that latitude
could have, at a continental scale, the effect showed in
the model A. In fact, this eagle is widely distributed in
southern Europe at the same latitude of southern Italy in
Spain, Greece and Turkey (Handrinos and Akriotis
1997; Man
˜osa 2003; Kirwan et al. 2008). In addition,
our model B clearly showed that large areas, especially
in the southernmost part of the peninsula, are suitable
for this species although the known distribution, showed
by data available for Italy, is well represented by the
map obtained with the Model A. The discrepancy be-
tween model B and the actual distribution of the short-
toed snake eagle in Italy (well reflected by Model A)
could result from the interaction between the motion
capacity of the species and the geography of the Italian
Peninsula. The Central Mediterranean is a dispersal
barrier for this species, since short-toed snake eagles
avoid crossing large bodies of water during migration by
using soaring-gliding flight over land (Agostini et al.
2002a,b; Mellone et al. 2011; Panuccio et al. 2012). The
majority of birds breeding in Italy migrate to and from
Africa through the Strait of Gibraltar; in spring they
arrive in Italy about 1,000 km NW of southern regions
of the Peninsula via southern France. As suggested by
previous research (Agostini and Mellone 2008), the ac-
tual distribution in Italy may reflect the particular
migration path used by this population. Considering the
increasing number of breeding pairs and of the occupied
territories in central and southern Italy (Baghino and
Premuda 2007; Mellone et al. 2011), our results support
the hypothesis that short-toed snake eagles are still
colonizing the country moving from north to south of
the peninsula primarily along its western slope. How-
ever, we cannot exclude that the increasing number of
the short-toed snake eagle in Italy reflects a recovery
from a previous reduction of the population (i.e. con-
servationist effort, urbanization) but also in this case our
results indicate that the latitudinal gradient could ex-
plain the actual distribution of the species.
Does the distribution reflect the colonization process?
All the above considerations lead us to further
questions: from where is the short-toed snake eagle
Fig. 3 The probability of occurrence of short-toed snake eagles at
different latitudes in the two different models with (a) and without
(b) latitude variable
186
Author's personal copy
colonizing the Italian peninsula? What is the original
range distribution of short-toed snake eagle? These are
topics linked to the origin of the Palearctic-African
migration system and of migration itself. Most authors
agree that birds started to migrate from tropical to
temperate areas (Alerstam 1990; Rappole 1995; Safriel
1995; Berthold 2001; Rappole and Jones 2002;Bo
¨hning-
Gaese and Oberrath 2003; Jahn et al. 2004). According
to this hypothesis, migration is an adaptation of tropical
birds to use seasonally abundant resources in temperate
regions in order to optimize breeding success or to avoid
seasonal resource depression (Alerstam 1990; Rappole
1995; Rappole and Jones 2002; Alerstam et al. 2003).
Alerstam and Enckell (1979) indicate that competi-
tion in Africa savannas during the breeding season was
a prerequisite for the evolution of the Palearctic-Afri-
can migration system. Glaciations were also events of
paramount importance in the evolution of migrations.
During cold glacial phases, the earth surface available
for many bird species drastically decreased (Moreau
1972). Since the northern polar front shifted south,
temperate forests and bush vegetation were reduced;
furthermore the range extension of tundra, deserts and
steppes reduced the Mediterranean and the tropical
vegetation to a minimum (Hooghiemstra et al. 2006;
Bruderer and Salewski 2008). During interglacial
Fig. 4 Variation in probability occurrence of short-toed snake eagle predicted by BRT model (Ylatitude, ALTM mean altitude, NSP
number of reptile species, CVALT standard deviation of altitude, FOR2 square root forests, FOR forests, COLINT crops and orchards,
ARB shrubs, PRAPAS meadows and pastures, ED urbanized areas)
Fig. 5 Prediction surface of the decision tree realized combining
the two main predictor variables of the model: mean altitude
(ALTM) and the prey species richness (NSP)
187
Author's personal copy
periods, including the current one, ancestral species
diverged into what are known as ‘‘twin’’ species, dis-
tinctly sedentary, partially migrants and completely
migrants (Berthold 2001; Rappole, 2005). Rappole and
Jones (2002) provide evidence that most species that
currently migrate descended from tropical residents.
For the Palearctic-African migration system, 42 of 185
species of migrants (23 %) have conspecific populations
that breed in the Tropics while 139 (75 %) have
Tropical-breeding congeners (Rappole and Jones 2002).
Of the snake eagles (Circaetus spp.), only the short-
toed snake eagle is a complete migrant while the other
five species (Beaudouin’s snake eagle, C. beaudouini,
black-chested snake eagle, C. pectoralis, brown snake
eagle, C. cinereus, east African snake eagle, C. fasciol-
atus, banded snake eagle, C. cinerascens) live in tropical
Africa and are mainly resident (Ferguson-Lees and
Christie 2001). However, it is hard to generalize a
model like the southern-home theory since birds con-
tinuously adapt their migratory behavior according to
changes in resource availability (Bruderer and Salewski
2008). In fact, the current migratory system began
evolving about 15,000 years ago at the end of the last
glaciation and is still evolving; the consequence of this
process is not fully predictable (Berthold 2001). Rap-
pole and Jones (2002) proposed a mechanism for the
evolution of long distance migration, via gradual col-
onization at the edge of the breeding range furthest
from the wintering area, with populations on the edges
more exposed to extinction risk. Consistently with this
hypothesis, we suggest that the Italian population of
the short-toed snake eagles should be considered as
part of a metapopulation comprising those of Western
Europe (France, Spain). Small and isolated populations
of southern Italy, despite the increasing number, could
be considered as small patches of this metapopulation
system cut off from the bulk of the population of
Western Europe. As a matter of fact, distribution and
abundance of migrants could be influenced by the cost
of reaching areas located far away from the main
migratory pathways (Henningsson and Alerstam 2005a,
b; Tellerı
´a et al. 2009) and several studies show that
isolation could lead small and periphery patches to
local extinction (Lomolino et al. 1989; Peltonen and
Hanski 1991; Hanski et al. 1995; Whitcomb et al. 1996;
Smith and Gilpin 1997; Thomas and Hanski 1997;
Hanski 1998; Dunham and Rieman 1999; Clinchy et al.
2002; Wahlberg et al. 2002). In support of this
hypothesis, historical data confirm that the short-toed
snake eagle was breeding in areas of southern conti-
nental Italy and Sicily at least until the end of the 19th
century (Salvadori 1872; Lucifero 1898; Arrigoni degli
Oddi 1929) while current data show that it is no longer
found in those areas. Its failure to re-colonize the
southernmost areas of Italy may, therefore, be due to
the distance from the center of its current metapopu-
lation range. In this scenario, we predict that this trend
will lead to increasing occupancy in the future in
southern Italy but, being at the edge of population
distribution, it is likely that density remains lower than
in Northwestern Italy, despite the availability of suit-
able areas for this species.
Conclusions
The direction of migration is the result of natural
selection; for this reason it is likely that only migration
routes leading to survival areas and avoiding ecological
barriers are favored by natural selection through an
increasing of survival rates and fitness. Therefore, it is
expected that migratory populations may follow their
historical expansion route (Salewski and Bruderer 2007),
and the main migration flyways of short-toed snake
eagles breeding in Europe may restrict accessibility to
and colonization of other potential breeding regions
(Henningsson and Alerstam 2005a,b). The southerly
direction of colonization of Italy performed by short-
toed snake eagle suggests that birds facing a natural
barrier show behavioural plasticity leading to unex-
pected migratory and colonization directions.
Acknowledgments We wish to thank Massimo Campora, Guido
Cattaneo, Ugo Mellone and Paolo Forconi for their help in pro-
viding data of short-toed snake eagles presence and the Societas
Herpetologica Italica (SHI) for allowing to use the data of prey
species richness. We are grateful to Simona Tarantino for her
assistance in setting out the figures. We appreciate the improve-
ments in English usage made by Christina Riehl through the
Association of Field Ornithologists’ program of editorial assis-
tance. MEDRAPTORS (www.raptormigration.org) is a network of
ornithologists that works to improve research and conservation of
migrating birds of prey.
References
Agostini N, Baghino L, Coleiro C, Corbi F, Premuda G (2002a)
Circuitous autumn migration in the short-toed eagle (Circaetus
gallicus). J Raptor Res 36:111–114
Agostini N, Baghino L, Panuccio M, Premuda G (2002b) A con-
servative strategy in migrating short-toed eagles (Circaetus
gallicus). Ardeola 49:287–291
Agostini N, Baghino L, Panuccio M, Premuda G, Provenza N
(2004) The autumn migration strategies of juvenile and adult
short-toed eagles (Circaetus gallicus) in the Central Mediterra-
nean. Avocetta 28:37–40
Agostini N, Mellone U (2008) Does migration flyway of short-toed
snake-eagles breeding in central Italy reflect the colonization
history? J Raptor Res 42:158–159
Alerstam T (1990) Bird migration. Cambridge University Press,
Cambridge
Alerstam T, Enckell PH (1979) Unpredictable habitats and evolu-
tion of bird migration. Oikos 33:228–232
Alerstam T, Hedenstro
¨mA,A
˚kesson S (2003) Long-distance
migration: evolution and determinants. Oikos 103:247–260
Arrigoni degli Oddi E (1929) Ornitologia italiana. Hoepli, Milano
Baghino L, Premuda G (2007) Nuovi dati sulla migrazione pri-
maverile ‘‘a circuito’’ del biancone, Circaetus gallicus, in Italia.
Avocetta 31:70–72
Bakaloudis DE (2009) Implications for conservation of foraging
sites selected by short-toed eagles (Circaetus gallicus) in Greece.
Ornis Fennica 86:89–96
188
Author's personal copy
Bakaloudis DE, Vlachos CG, Holloway GJ (1998) Habitat use by
short-toed eagle Circaetus gallicus and reptilian prey during the
breeding season in Dadia Forest (north-eastern Greece). J Appl
Ecol 35:821–828
Bakaloudis DE, Vlachos CG, Holloway GJ (2005) Nest spacing
and breeding performance in short-toed eagle Circaetus gallicus
in northeast Greece. Bird Study 52:330–338
Begon M, Townsend CR, Harper JL (2006) Ecology: from indi-
viduals to ecosystem, 4th edn. Blackwell publishing, Oxford
Berthold P (2001) Bird migration. A general survey. Oxford Uni-
versity Press, Oxford
Bionda R, Bordignon L (2006) Atlante degli uccelli nidificanti del
Verbano Cusio Ossola. Quaderni Natura e Paesaggio del VCO
Birdlife International (2004) Birds in Europe: population estimates,
trends and conservation status. Birdlife Conservation series No.
12, Cambridge
Boano A, Brunelli M, Bulgarini A, Montemaggiori A, Sarrocco S,
Visentin M (1995) Atlante degli Uccelli nidificanti nel Lazio.
Alula 1–2: volume speciale
Bo
¨hning-Gaese K, Oberrath R (2003) Macroecology of habitat
choice in long-distance migratory birds. Oecologia 137:296–303
Bon M, Cherubini G, Semenzato M, Stival E (1999) Atlante Degli
uccelli nidificanti in Provincia di Venezia. SGE, Padova
Bordignon L (2004) Gli Uccelli della Provincia di Novara. Pro-
vincia di Novara
Brandmayr P (2002) Ambienti e fauna nel paesaggio italiano. In:
Minelli A et al (eds) La fauna in Italia. Touring editions, Tor-
ino, pp 30–45
Brichetti P, Fasola M (1990) Atlante degli uccelli nidificanti in
Lombardia 1983–1987. Editoriale Ramperto, Brescia
Bruderer B, Salewski V (2008) Evolution of bird migration in a
biogeographical context. J Biogeogr 35:1951–1959
Bustamante J, Seoane J (2004) Predicting the distribution of four
species of raptors (Aves: Accipitridae) in southern Spain: sta-
tistical models work better than existing maps. J Biogeogr
31:295–306
Campora M, Cattaneo G (2006) II Biancone Circaetus gallicus in
Italia. Riv Ital Orn 76:1–46
Clinchy M, Haydon DT, Smith AT (2002) Pattern does not equal
process: what does patch occupancy really tell us about meta-
population dynamics? Am Nature 159:351–362
Cramp S, Simmons KEL (1980) The birds of the Western Pale-
arctic, vol II. Oxford University Press, Oxford
Cumming GS, Gaidet N, Ndlovu M (2012) Towards a unification
of movement ecology and biogeography: conceptual framework
and a case study on Afrotropical ducks. J Biogeogr
39:1401–1411
De’ath G, Fabricius KE (2000) Classification and regression trees:
a powerful yet simple technique for ecological data analysis.
Ecology 81:3178–3192
Dunham JB, Rieman BE (1999) Metapopulation structure of bull
trout: influence of physical, biotic, and geometrical landscape
characteristics. Ecol Appl 9:642–655
Elith J, Graham CH, Anderson RP, NCEAS Modeling Group
(2006) Novel methods improve prediction of species’ distribu-
tions from occurrence data. Ecography 29:129–151
Elith J, Leathwick JR, Hastie T (2008) A working guide to boosted
regression trees. J Anim Ecol 77:802–813
Elkie PC, Rempel RS, Car AP (1999) Patch Analyst User’s Man-
ual. NWST Technical Manual. Ont Min Nat Resour Northwest
Sci Technol. Thunder Bay, Ontario TM–002
Evans KL, Warren PH, Gaston KJ (2005) Species energy rela-
tionships at the macroecological scale: a review of the mecha-
nisms. Biol Rev 80:1–25
Ferguson-Lees J, Christie DA (2001) Raptors of the World. Helm,
London
Fracasso G, Verza E, Boschetti E (2003) Atlante degli Uccelli ni-
dificanti in provincia di Rovigo. Provincia di Rovigo
Fraissinet M, Kalby M (1989) Atlante degli uccelli nidificanti in
Campania (1983–1987), Monografia n. 1. ASOIM, Salerno
Friedman JH, Hastie T, Tibshirani R (2000) Additive logistic
regression: a statistical view of boosting. Ann Stat 28:337–407
Gellini S, Ceccarelli PP (2000) Atlante degli uccelli nidificanti
nelle province di Forlı
`-Cesena e Ravenna (1995–1997).
S.T.E.R.N.A., Amministrazioni Provinciali di Forlı
`-Cesena e
Ravenna
Giacchini P (2007) Atlante degli uccelli nidificanti nella provincia di
Ancona. Provincia di Ancona
Gil JM, Pleguezuelos JM (2001) Prey and prey-size selection by the
short-toed eagle (Circaetus gallicus) during the breeding season
in Granada (south-eastern Spain). J Zool 255:131–137
Gruppo Vicentino Studi Ornitologici ‘‘NISORIA’’ (1994) Atlante
degli uccelli nidificanti nella Provincia di Vicenza. Gilberto
Padovan Editore, Vicenza
Hagemeijer WJM, Blair MJ (1997) The EBCC atlas of European
breeding birds: their distribution and abundance. T. and A.D.
Poyser, London
Handrinos G, Akriotis T (1997) The birds of Greece. Helm, Lon-
don
Hanski I (1998) Metapopulation dynamics. Nature 396:41–49
Hanski I, Pakkala T, Kuussaari M, Lei G (1995) Metapopulation
persistence of an endangered butterfly in a fragmented land-
scape. Oikos 72:21–28
Henningsson SS, Alerstam T (2005a) Barriers and distances as
determinants for the evolution of bird migration links: the
arctic shorebird system. Proc R Soc B 272:2251–2258
Henningsson SS, Alerstam T (2005b) Patterns and determinants of
shorebird species richness in the circumpolar Arctic. J Biogeogr
32:383–396
Hooghiemstra H, Le
´zine AM, Leroy SAG, Dupont L, Marret F
(2006) Late Quaternary palynology in marine sediments: a
synthesis of the understanding of pollen distribution patterns in
the NW African setting. Quat Int 148:29–44
Ientile R, Massa B (2008) Atlante della biodiversita
`della Sicilia.
Vertebrati Terrestri. Cap. 2 Uccelli (Aves). Collana Studi e
Ricerche dell’ARPA Sicilia, Palermo
Jahn AE, Levey DJ, Smith KG (2004) Reflections across hemi-
spheres: a system-wide approach to new world bird migration.
Auk 121:1005–1013
Kerlinger P (1989) Flight strategies of migrating hawks. Chicago
University Press, Chicago
Kirwan GM, Boyla KA, Castell P, Demirci B, O
¨zen M, Welch H,
Marlow T (2008) The birds of Turkey. Helm, London
La Gioia G (2009) Atlante degli uccelli nidificanti in provincia di
Lecce (2000–2007). Edizioni del Grifo, Lecce
Leathwick JR, Elith J, Francis MP, Hastie T, Taylor P (2006)
Variation in demersal fish species richness in the oceans sur-
rounding New Zealand: an analysis using boosted regression
trees. Mar Ecol Prog Ser 321:267–281
Lomolino MV, Brown JH, Davis R (1989) Island biogeography of
montane forest mammals in the American southwest. Ecology
70:180–194
Lo
´pez-Iborra G, Limin
˜ana R, Pavon D, Martinez-Perez J (2011)
Modelling the distribution of short-toed eagle (Circaetus galli-
cus) in semi-arid Mediterranean landscapes: identifying
important explanatory variables and their implications for its
conservation. Eur J Wildlife Res 57:83–93
Lucifero A (1898) Avifauna Calabra. Elenco delle specie di uccelli
sedentari o di passaggio in Calabria. Avicula, Siena
Manly BFJ, McDonald LL, Thomas DL, McDonald TL, Erickson
WP (2002) Resource selection by animals: statistical design and
analysis for field studies, 2nd edn. Kluwer Academic Publisher,
Dordrecht
Man
˜osa S (2003) Aguila culebrera, Circaetus gallicus. In: Martı
´R,
Del Moral JC (eds) Atlas de las Aves Reproductoras de Espan
˜a.
Direccio
´n General de Conservacio
´n de la Naturaleza, Sociedad
Espan
˜ola de Ornitologı
´a, pp 172–173
Mayr E, Meise W (1930) Theoretisches zur Geschichte des Vo-
gelzuges. Vogelzug 1:149–172
Mellone U, Limin
˜ana R, Mallı
`a E, Urios V (2011) Extremely de-
toured migration in an inexperienced bird: interplay of trans-
port costs and social interactions. J Avian Biol 42:468–472
Meschini E, Frugis S (1993) Atlante degli uccelli nidificanti in
Italia. Suppl. Ricerche Biologia Selvaggina, INFS
189
Author's personal copy
Mezzavilla F, Bettiol K (2007) Nuovo Atlante degli Uccelli nidifi-
canti in provincia di Treviso (2003–2006). Associazione Fau-
nisti Veneti, Treviso
Mingozzi T, Boano G, Pulcher C (1988) Atlante degli uccelli ni-
dificanti in Piemonte e Val d’Aosta 1980–1984. Monografie
VIII. Museo Regionale di Scienze Naturali di Torino, Torino
Moreau R (1972) The Palaearctic-African bird migration system.
Academic Press, London
Moreno-Rueda G, Pizarro M (2007) Snake species richness and
shrubland correlate with the short-toed eagle (Circaetus galli-
cus) distribution in south-eastern Spain. Ann Zool Fennici
44:314–320
Newton I (1998) Population limitation in birds. Academic Press,
London
Panuccio M, Agostini N, Premuda G (2012) Ecological barriers
promote risk minimization and social learning in migrating
short-toed snake eagles. Ethol Ecol Evol 24:74–80
Panuccio M, Agostini N, Baghino L, Bogliani G (2013) Visible
migration of short-toed snake-eagles: interplay of weather and
topographical features. J Raptor Res 47:60–68
Pedrini P, Caldonazzi M, Zanghellini S (2005) Atlante degli Uccelli
nidificanti e svernanti in provincia di Trento. Museo Tridentino
di Scienze Naturali, Trento. Studi Trentini di Scienze Naturali,
Acta Biologica, vol 80, p 2
Peltonen A, Hanski I (1991) Patterns of island occupancy explained
by colonization and extinction rates in shrews. Ecology
72:1698–1708
Rappole JH (1995) The ecology of migrant birds: a Neotropical
perspective. Smithsonian Institution Press
Rappole JH (2005) Evolution of old and new world migration
systems: a response to Bell. Ardea 93:125–131
Rappole JH, Jones P (2002) Evolution of old and new world
migration systems. Ardea 90:525–537
Ravasini M (1995) L’Avifauna nidificante nella provincia di Par-
ma. Tipolitotecnica, Parma
Rempel EPR, Carr A (1999) Patch Analyst User’s Manual. Ontario
Ministry of Natural Resources. Northwest Science and Tech-
nologies, Thunder Bay TM-002, Ontario
Safriel UN (1995) The evolution of Palaearctic migration—the case
for southern ancestry. Isr J Zool 41:417–431
Salewski V, Bruderer B (2007) The evolution of bird migration—a
synthesis. Naturwissenschaften 94:268–279
Salvadori T (1872) Fauna d’Italia. Uccelli. Vallardi editore, Milano
Sa
´nchez-Zapata JA, Calvo JF (1999) Raptor distribution in rela-
tion to landscape composition in semiarid Mediterranean
habitats. J Appl Ecol 36:254–262
Schapire R (2003) The boosting approach to machine learning, an
overview. In: Denison DD et al. (eds) MSRI Workshop on
Nonlinear Estimation and Classification, Springer
Sindaco R, Doria G, Razzetti E, Bernini F (2006) Atlante degli
Anfibi e dei Rettili d’Italia. Societas Herpetologica Italica.
Edizioni Polistampa, Firenze
Smith AT, Gilpin ME (1997) Spatially correlated dynamics in a
pika metapopulation. In: Hanski IA, Gilpin ME (eds) Meta-
population biology: ecology, genetics, and evolution. Academic
Press, London, pp 401–428
Tellerı
´a JL, Ramı
´rez A, Galarza A, Carbonell R, Pe
´rez-Tris J,
Santos T (2009) Do migratory pathways affect the regional
abundance of wintering birds? A test in northern Spain. J Bio-
geogr 36:220–229
Tellini Florenzano G, Baccetti N, Arcamone E, Meschini E, Spo-
simo P (1997) Atlante degli uccelli nidificanti e svernanti in
Toscana (1982–1992). Quaderni Museo Provinciale di Storia
Naturale, Monografie n.1, Livorno
Thomas CD, Hanski I (1997) Butterfly populations. In: Hanski IA,
Gilpin ME (eds) Metapopulation biology: ecology, genetics,
and evolution. Academic Press, London, pp 359–386
Wahlberg N, Klemetti T, Hanski I (2002) Dynamic populations in
a dynamic landscape: the metapopulation structure of the
marsh fritillary butterfly. Ecography 25:224–232
Whitcomb SD, Servello FA, O’Connell AF Jr (1996) Patch occu-
pancy and dispersal of spruce grouse on the edge of its range in
Maine. Can J Zool 74:1951–1955
190
Author's personal copy
