Dispersal and connectivity effects at different altitudes in the Euphydryas aurinia complex

Abstract

Across its European range, the Euphydryas aurinia complex (Annex II of the Habitats Directive) includes a series of distinct populations. At least 3 taxa occur in Italy, each showing slight morphological differences and distinct eco-ethological features. For the first time, we compared metapopulation dynamics of E. (a.) glaciegenita inhabiting a site in the NW Alps (2,100–2,300 m) with E. (a.) provincialis occurring in the Mediterranean biogeographical region in hilly dry grasslands (700 m). To describe patterns of dispersal, we applied the virtual migration model (VMM) to data collected using Mark-Release-Recapture (MRR). We used parameters of survival and migration to explore metapopulation characteristics. In particular we investigated the relative role of connectivity and patch quality in affecting migration rates. We observed differences between the two metapopulation systems, with the “Alpine” population occurring at higher altitude and in more open habitats, showing lower dispersal propensity. In contrast, even though the “Mediterranean” population is more prone to disperse, migration appears to have higher costs. Dispersal abilities affect metapopulation dynamics, which are at the basis of long-term perspectives of survival for butterfly populations. We discuss our results in the framework of conservation and management options for habitats occupied by these Italian taxa of the E.
aurinia complex.

Full text

1 23
Journal of Insect Conservation
An international journal devoted to
the conservation of insects and related
invertebrates
ISSN 1366-638X
J Insect Conserv
DOI 10.1007/s10841-014-9715-8
Dispersal and connectivity effects at
different altitudes in the Euphydryas
aurinia complex
L.P.Casacci, C.Cerrato, F.Barbero,
L.Bosso, S.Ghidotti, M.Paveto,
M.Pesce, E.Plazio, G.Panizza,
E.Balletto, R.Viterbi, et al.

1 23
Your article is protected by copyright and
all rights are held exclusively by Springer
International Publishing Switzerland. 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 PAPER
Dispersal and connectivity effects at different altitudes
in the Euphydryas aurinia complex
L. P. Casacci •C. Cerrato •F. Barbero •L. Bosso •S. Ghidotti •
M. Paveto •M. Pesce •E. Plazio •G. Panizza •E. Balletto •
R. Viterbi •S. Bonelli
Received: 6 June 2014 / Accepted: 17 October 2014
ÓSpringer International Publishing Switzerland 2014
Abstract Across its European range, the Euphydryas
aurinia complex (Annex II of the Habitats Directive)
includes a series of distinct populations. At least 3 taxa
occur in Italy, each showing slight morphological differ-
ences and distinct eco-ethological features. For the first
time, we compared metapopulation dynamics of E. (a.)
glaciegenita inhabiting a site in the NW Alps
(2,100–2,300 m) with E. (a.) provincialis occurring in the
Mediterranean biogeographical region in hilly dry grass-
lands (700 m). To describe patterns of dispersal, we
applied the virtual migration model (VMM) to data col-
lected using Mark-Release-Recapture (MRR). We used
parameters of survival and migration to explore metapop-
ulation characteristics. In particular we investigated the
relative role of connectivity and patch quality in affecting
migration rates. We observed differences between the two
metapopulation systems, with the ‘‘Alpine’’ population
occurring at higher altitude and in more open habitats,
showing lower dispersal propensity. In contrast, even
though the ‘‘Mediterranean’’ population is more prone to
disperse, migration appears to have higher costs. Dispersal
abilities affect metapopulation dynamics, which are at the
basis of long-term perspectives of survival for butterfly
populations. We discuss our results in the framework of
conservation and management options for habitats occu-
pied by these Italian taxa of the E. aurinia complex.
Keywords Butterfly conservation Connectivity 
Emigration propensity Mark-recapture Species
complex Virtual migration model
Introduction
Populations of the Euphydryas aurinia complex range from
all across Europe and temperate Asia to as far East as
Yakutia. Because of this very broad range, E. aurinia is not
at risk of becoming extinct (Least Concern, IUCN 2013
regional assessment—van Swaay et al. 2010), but this
taxon is known for having declined in most European
countries, and is reported as extinct in the Netherlands.
Declines of over 30 % in range or population size have
been reported from Germany, Latvia, Luxembourg, the
Republic of Ireland, Slovakia and Ukraine while smaller
declines of 6–30 % are known for having occurred in fif-
teen other European countries (Asher et al. 2001). E. au-
rinia is, accordingly, listed in Annex II of the Habitats
Directive (HD).
In butterflies, the study of the flight behaviour and use of
the environmental matrix is crucial for management
L. P. Casacci F. Barbero L. Bosso E. Plazio E. Balletto 
S. Bonelli
DBIOS Department of Life Sciences and Systems Biology,
University of Turin, Turin, Italy
C. Cerrato S. Ghidotti R. Viterbi
Gran Paradiso National Park, Turin, Italy
C. Cerrato (&)
Institute of Atmospheric Sciences and Climate, National
Research Council, Turin, Italy
e-mail: cri.entessa@virgilio.it
S. Ghidotti
DISAT, Department of Environmental Sciences and Territory,
University of Milano Bicocca, Milan, Italy
M. Paveto M. Pesce
DISTAV, Department of Earth, Environmental and Life
Sciences, Genoa University, Genoa, Italy
G. Panizza
Capanne di Marcarolo Natural Park, Alessandria, Italy
123
J Insect Conserv
DOI 10.1007/s10841-014-9715-8
Author's personal copy

planning, particularly in more or less fragmented land-
scapes, and butterflies are currently among the most well-
known model organisms for dispersal studies (Stevens
et al. 2010a,b).
Dispersal abilities are assumed to be fixed traits in most
metapopulation studies and especially by the classic for-
mulation of this theory (Travis and French 2000; Goodwin
2003; Bowler and Benton 2005), where individual move-
ments between populations are modelled as a function of
these fixed dispersal traits, on the one hand, and of extre-
mely variable spatial structures, on the other.
The configuration of habitat patches can also positively
or negatively act as a selective pressure on dispersal abil-
ities. Habitat fragmentation, for instance, is known to affect
dispersal within systems of populations, by reducing both
the emigration rate as well as the length of movements
(Schtickzelle et al. 2006). Various studies have pointed out
differences among species and populations, as well as
between sexes (e.g. Fric et al. 2010; Stevens et al. 2010b;
Bonelli et al. 2013). Other differences are known to occur
in butterflies’ propensity to move, as well as in more
general flight patterns.
E. aurinia is a complex of generally allopatric taxa
characterised by more or less distinct external morpholo-
gies and strongly different ecological requirements. As is
the case for American E. editha, many subspecies have
been described (see Catalogue of Life, under Eurodryas),
but at the moment no genetic evidence is available to
validate them (e.g. Zimmermann et al. 1999,2000;
Descimon and Mallet 2009; Sinama et al. 2011; Mikheyev
et al. 2013) and few studies are currently under way at
European level to investigate these differences. As con-
cerns Italy, we have at least three Evolutionarily (and
Ecologically) Significant Units (ESUs, see Casacci et al.
2013) which occupy all three of the Italian biogeographical
regions (Balletto et al. 2007; Balletto et al. 2014b) and
have been listed as separate species in the Check List of the
Italian butterflies (Balletto and Cassulo 1995; Balletto et al.
2014a). E. (a.) aurinia (Rottemburg, 1775) occurs in the
wet meadows in the Po Plains, within the ‘Continental
Region’, E. (a.) glaciegenita (Verity, 1928) is restricted to
the ‘Alpine Region’ around and above 2,000 m, while E.
(a.) provincialis (Boisduval, 1828) typically occurs in the
dry calcareous grasslands and maquis of the ‘Mediterra-
nean Region’ (Fig. 1).
According to the most recent assessment (under HD Art.
17) the conservation status of the Continental E. aurinia
populations is as ‘‘Bad’’ as in central and northern Europe,
since suitable habitat has become strongly reduced and
fragmented. In Italy, at least 12 populations have become
extinct because of habitat destruction (Bonelli et al. 2011).
On the contrary, the conservation status of the Mediterra-
nean and Alpine populations is ‘‘Favourable’’ and we have
no records of any population having become lost.
In this work we focused on two of the three Italian taxa,
i.e. the Alpine and the Mediterranean populations, since
they are the only ones which may potentially guarantee the
long-term survival of the E. aurinia complex in Italy. Both
are threatened by land abandonment and/or land use
intensification, two threats which can still be reversed if
and when more appropriate forms of land use are
implemented.
At a local level, recent conservation schemes, such as
the ‘‘Environmental Sensitive Areas Agro-environmental
Schemes, Rural Development Practices’’ promoted by the
EU agricultural policy, have become crucial for the long
term conservation of grassland butterflies. Working within
this protocol, farmers are expected to adopt environmen-
tally friendly agricultural practices, such as low intensity
grazing regimes. Conservationists, however, will be able to
advise on how these opportunities may be translated into
practice only after we are able to understand how distances
are perceived by individual butterflies, as well as how
barriers are identified, and how all this affects adults’
Fig. 1 a Distribution of the three Euphydryas aurinia subspecies and
location of the two study sites in Italy. White E. (a.) glaciegenita;
black E. (a.) aurinia;grey E. (a.) provincialis.bNetwork of habitat
patches for the Mediterranean (Marcarolo site) and cthe Alpine
(Bardoney site) E. aurinia populations. Black areas habitat patches;
dark grey bushes; grey woodland; light grey grassland; white rock and
screes
J Insect Conserv
123
Author's personal copy

movements within the environmental matrix. In other
words, we need to understand the effect of fragmentation
and connectivity on movements within the patch systems.
In general we wish to contribute to investigations into
whether in a metapopulation system emigration propensity
is species-specific or may be variable depending on dif-
ferent environmental pressures.
In synthesis, we focused our work on adult dispersal, in
order to obtain crucial information on butterflies’ behav-
iour, in the expectation that our results could be directly
converted into conservation practice, at various scales.
Methods
Study system
The three Italian taxa of the E. aurinia complex have
similar life cycles and all are single brooded. Females lay
eggs in large batches, while larvae are gregarious and
overwinter in smaller or larger webs, close to the ground.
E. (a.) aurinia’s food plant is Succisa pratensis, that of E.
(a.) provincialis (Mediterranean) is either Cephalaria leu-
cantha or Knautia arvensis (all three are Dipsacaceae),
while larvae of E. (a.) glaciegenita (Alpine) feed on
Gentiana acaulis (Gentianaceae). We will only mention for
completeness that Spanish E. (a.) beckeri (Herrich-Scha
¨f-
fer, [1851]) feeds on shrubs or bushes of yet another family
such as Lonicera implexa (Loniceraceae see Singer and
Wee 2005; Pen
˜uelas et al. 2006; Stefanescu et al. 2006), a
plant that is never consumed by Italian populations, even
though it is generally abundant in many habitats where E.
(a.) provincialis is common. Matters, however, are not
always as clear-cut, since for instance in Lithuania some
populations of E. (a.) aurinia live on G. cruciata (S
ˇvitra
and Sielezniew 2010), while in Belgium larvae feed on S.
pratensis, in the wet meadows, whereas on chalky grass-
lands they exploit other food plants, such as K. arvensis or
Scabiosa columbaria (Schtickzelle et al. 2005).
Interestingly enough, each group of the Italian E. auri-
nia complex is also locally linked to a specific habitat
included in Annex I of the Habitats Directive, i.e. E. (a.)
aurinia to Habitat 6410, Molinia meadows on calcareous,
peaty or clayey-silt-laden soils—Molinion caeruleae;E.
(a.) provincialis to Habitat 6210, Semi-natural dry grass-
lands and scrubland facies on calcareous substrates—Fe-
stuco-Brometalia;E. (a.) glaciegenita, although partially,
to Habitat 4060, Alpine and Boreal heaths. In particular,
the first two taxa are good candidates for becoming ‘‘typ-
ical species’’ (according to the definition in the Habitats
Directive).
Study sites
During 2013, two populations of E. aurinia, one from the
Alps and one located in the Mediterranean biogeographical
region, in the northern Apennine (Fig. 1; Table 1), were
studied with a shared protocol. Both populations were
relatively numerous and inhabited large semi-natural
grasslands in fairly good ecological conditions. The two
areas are protected within the NATURA 2000 network and
the two Parks where they occur (see below) are actively
Table 1 Characteristics of the investigated metapopulations of E. aurinia
E. (a.) provincialis E. (a.) glaciegenita
Site Marcarolo Bardoney
Biogeographical Region Mediterranean Alpine
Altitude 650–750 m 2,100–2,300 m
Location (WGS84) 8°4601600 E–44°3304900 N7°2504200 E–45°3405600 N
Protected area ‘‘Capanne di Marcarolo’’ Natural Park ‘‘Gran Paradiso’’ National Park
SCI
a
code IT1180026 IT1201000
Number of patches 15 16
Total area of patches (ha) 20.34 28.15
Habitat patch areas (min–max) (ha) 0.62–2.44 0.49–5.78
Habitat patch areas (mean–CV) (ha) 1.35–0.39 1.76–0.84
Inter-patch distances (min–max) (m) 105–1,984 88–2,014
Habitat type Dry grassland Alpine grassland
Marked individuals 1,499 (509 $; 990 #) 1,684 (532 $; 1,151 #)
Recaptured individuals 764 (229 $; 535 #) 483 (108 $; 375 #)
Intra-patch recapture events 870 (252 $; 618 #) 523 (108 $; 415 #)
Inter-patches recapture events 571 (132 $; 439 #) 161 (20 $; 141 #)
a
SCI Site of Community Importance (EU Habitat Directive 92/43/EEC)
J Insect Conserv
123
Author's personal copy

involved in ensuring the long-term conservation of open
habitats within the framework of the Rural Development
Programme (Measure 323—Conservation and upgrading of
the rural heritage).
The Mediterranean population is located within the
‘‘Capanne di Marcarolo’’ Natural Park, in the northern
(Ligurian) Apennines, at around 700 m in altitude. It
occurs in a fragmented semi-natural grassland (Habitat
6210—HD Annex I), of the so-called Arrhenatherion
elatioris and Brometalia erecti type, where vegetation is
dominated by a mixture of mesophilous and xerophilous
plants, such as Knautia arvensis (E. a. provincialis’s food
plant), Poa pratensis,Sanguisorba officinalis,Achillea
millefolium,Silene vulgaris,Trisetum flavescens, Daucus
carota, Leucanthemum vulgare, Lolium perenne, Lotus
corniculatus, Taraxacum officinale and Centaurea nigres-
cens. In this area, the vegetation is particularly rich because
of the geographical position, climate and geology, which
results in a mixture of Mediterranean and Alpine species.
Most of the meadows occur on south facing slopes and are
surrounded by forest and shrubs, while northern slopes are
covered in woods or sparse trees. The site is partially
fragmented because of the abandonment of traditional hay
cutting and is locally invaded by shrubs and small trees.
The Alpine population flies in the Cogne Valley
(Bardoney area, ca. 2,200 m), within the Gran Paradiso
National Park, in the north-western Italian Alps. The
population occurs around and above the timberline in an
open and apparently continuous Alpine grassland. At its
lower altitudinal boundary the site is limited by a conif-
erous woodland (mainly with Larix decidua and Pinus
cembra), and is irregularly fragmented by rocks and screes,
as well as by Alpine heaths (Habitat 4060—HD Annex I).
This Alpine grassland is characterised by a mixture of
mesophilous, meso-xerophilous, and hygrophilous habitats.
In particular, the central part of the site is composed by wet
meadows and buried marshes growing on acidic soils. The
larval food plant, Gentiana acaulis, is widespread and
abundant, mainly at the border between the meso-xeroph-
ilous and the more hygrophilous habitats. The area is
currently managed by extensive grazing, which maintains
floristic heterogeneity and diversity.
For both sites, potential habitat patches around the areas
where butterflies were known to occur were mapped on the
basis of a rough vegetational survey and of aerial pictures.
Patches were defined as distinct when separated by obvious
barriers (e.g. woods, roads), or when land cover features
changed consistently. The patch configuration was con-
firmed at the end of the study (Hanski et al. 2000)by
superimposing the GPS fixes of captures on the initially
selected patches (see sampling section for details). Infor-
mation on the investigated populations is summarised in
Table 1.
Population sampling
We performed a mark-release-recapture (MRR) study,
designed to analyse dispersal ability using the Virtual
Migration Model. Sampling of the Mediterranean popula-
tion was carried out from 27 May to 22 June, 2013 while
the study of the Alpine population was conducted from 14
July to 14 August, 2013 by following the same protocol.
All patches were visited daily, weather permitting, and all
observed butterflies were captured and permanently
marked with an individual number on the ventral surface of
the hind wings. For each (re)capture, we recorded the
individual code, date, patch location and sex along with the
GPS position. After marking, butterflies were immediately
released at the location of their capture.
The virtual migration model
We analysed MRR data by means of the virtual migration
model (program VM2, see http://www.helsinki.fi/science/
metapop/), to investigate the movement patterns of E. au-
rinia metapopulations and estimate their parameters of
survival and migration. The Virtual Migration Model has
been described in detail by Hanski et al. (2000), Petit et al.
(2001), Wahlberg et al. (2002), and we only briefly sum-
marize its main assumptions. The model assumes that
individuals staying in habitat patches experience a certain
dispersal-independent and constant mortality l
p
. Emigra-
tion rate from a natal patch (e
j
) depends on area (A
j
):
ej¼gAfem
j
where gdefines emigration propensity (here expressed as
daily emigration rate from a 1-ha patch), while f
em
repre-
sents the emigration scaling with patch area. Survival of
dispersing individuals (u
mj
) is a sigmoid function of their
natal patch connectivity (S
j
):
umj ¼S2
j
kþS2
j
The square root of the scaling parameter krepresents the
connectivity level up to which half of the dispersers suc-
cessfully reach other patches. Patch connectivity is mea-
sured as:
Sj¼X
k6¼j
exp adjk

Afim
k
with d
jk
being the Euclidean distance between patches, A
k
refers to the target patch area and finally aand f
im
are
respectively scaling distance-dependence of dispersal and
immigration probability. Successful dispersers are distrib-
uted among target patches proportionally to their contri-
butions to the natal patch connectivity. The VM model
J Insect Conserv
123
Author's personal copy

allows to estimate all six parameters (l
p
,g,f
em
,k,a,f
im
)
as well as their 95 % confidence intervals. Parameter
estimates for the two populations were produced separately
for each sex. We considered VM model parameter esti-
mates to differ significantly if the confidence interval of
one parameter did not include the parameter estimate of the
other one, and vice versa (as suggested by Wahlberg et al.
2002).
Patch quality
The quality of each habitat patch was estimated at 86
randomly distributed 5 95 m quadrats, in the case of the
Alpine site, and at 30 5 95 m quadrats for the Mediter-
ranean site. For each quadrat we measured the number and
height of food plants (either Gentiana acaulis,orKnautia
arvensis) and the height of the surrounding vegetation.
Using Braun-Blanquet’s method (Braun-Blanquet 1932),
we measured the percent cover of the vegetational and
physical components of the environment, i.e. of nectar
sources, graminaceous plants, grass, litter or moss, rocks,
bushes or larch trees, and bare soil. For each parameter, the
mean values were calculated and used for further analysis.
To evaluate the effect of host plant and nectar plant
density (taken as indicators of patch quality), as well as
their influence on E. aurinia male and female density, on
scaling of immigration and emigration, we replaced patch
area with an ‘‘effective’’ area, by following the approach
described by Rabasa et al. (2007). ‘‘Effective’’ areas were
calculated by multiplying the area of each patch by the
number of host or nectar plants, or by the number of female
or male butterflies, and dividing this value by the median
value of each parameter calculated on all the patches.
Factors affecting dispersal were also analysed by gen-
eralized linear models (GLM) with binomial response and
logit link function, by the software R 2.15.0 (R Core Team
2012). The total fraction of individuals that moved to a
‘‘recipient’’ patch (immigrants) and the total fraction of
individuals moving from a ‘‘source’’ patch (emigrants)
were set as the dependent variables. The models were
performed separately for males and females of both pop-
ulations, to test for the effect of average male and female
densities, food plant cover and height and their interaction,
cover of the other vegetational components, patch area and
connectivity (S
j
).
We used the dredge function of the R package MuMIn
(Barton 2013) to automatically construct all possible
models based on the set of explanatory variables in the full
model, including the null model, as well as to identify
minimum-adequate models by using the Akaike Informa-
tion Criterion (AIC
c
) for model evaluation (Burnham and
Anderson 2004). Finally, only models with D(AIC
c
)\2
were retained.
Results
Population dynamics
In total, 1,684 individuals (532 females, 1,151 males, 1
undetermined) of the Alpine population (hereafter ALP)
and 1,499 (509 females and 990 males) individuals of the
Mediterranean population (hereafter MED) were marked.
483 (28.7 %) butterflies were recaptured at least once (108
females, 20.3 %, and 375 males, 32.6 %) in the Alps, while
from the Mediterranean population 764 (51.0 %) butterflies
were recaptured (229 females, 45.0 %, and 535 males,
54.0 %). The sex ratio was around 2:1 in favour of males in
both populations (MED: 1.94 males per female; ALP: 2.16
males per female). In both populations we recaptured a
higher proportion of males than of females (ALP:
v
2
=26.21, df =1, pvalue \0.001, n =1,683; MED:
v
2
=10.66, df =1, pvalue =0.001, n =1,499).
The capture probability estimated in the VM model was
higher for the Mediterranean population (0.214 ±0.001
males; 0.207 ±0.003 females) than for the Alpine one
(0.125 ±0.004 males; 0.084 ±0.001 females).
Dispersal behaviour
Figure 2shows the estimated values of parameters from
the VM model and their 95 % confidence intervals. The
probability of daily mortality within a patch (l) ranged
from 0.079 to 0.161. The Mediterranean population had
lower within-patch daily mortality (12 % for females and
8 % for males) than the Alpine population (16 % for
females and 14 % for males) but estimates were signifi-
cantly different only for males as confidence intervals did
not overlap (Fig. 2a). This parameter allows the calculation
of the average life span for both sexes and both popula-
tions. In the Alpine population, lifespan values were sim-
ilar for males (mean from VMM, 6.2 days; maximum from
field data, 19 days) and females (mean, 5.7 days; max,
18 days). In the Mediterranean population, values were
higher, in particular for males, with an average survival of
12.2 days (max, 23 days), while females survived on
average 7.7 days (max, 20 days).
Adults’ propensity to move was significantly different
between populations, as indicated by the (g) parameter
(Fig. 2b). The Mediterranean population was more prone to
disperse (around 21 % for males and 18 % for females),
while butterflies of the Alpine population showed a
reduced tendency to migrate. The males had half the
migration rate values per 1 hectare of habitat patch (10 %)
and females showed an even lower value of migration
propensity (5 %) with respect to the Mediterranean
individuals.
J Insect Conserv
123
Author's personal copy

The parameter a, describing the effect of distance on
daily movements, was similar for males and females, but
varied between populations (Fig. 2c). Values obtained for
the Alpine population (a&6.5) were almost double of
those of the Mediterranean population (a&3.5). The
average range of daily dispersal distances (calculated using
the parameter a) were about 150 m for the Alpine popu-
lation and 270 m for the Mediterranean form. Male and
female mortalities during dispersal of both populations
were not found to be significant (confidence intervals
overlap) (Fig. 2d).
Movements in the patch networks
Intra-patch movement distances were significantly different
between populations and sexes (Kruskal–Wallis, v
2
=9.118,
df =3, pvalue =0.028) and their mean values ranged from
17 m for the Mediterranean females to 32 m for the Alpine
males (Fig. 3a). Males and females did not differ in their daily
intra-patch distances (MED: Mann–Whitney, W =889,
pvalue =0.476; ALP: Mann–Whitney, W =10,375,
pvalue =0.119). Distances travelled by males within the
same patch were slightly, but significantly, higher for the
Alpine population (MW, W =10,048, pvalue =0.045), but
no difference was observed for females (MW, W =1,536,
pvalue =0.183). Also, inter-patch movements were signifi-
cantly different between populations and sexes (KW,
v
2
=13.658, df =3, pvalue =0.003; Fig. 3). Males and
females of the Mediterranean population moved similar dis-
tances when dispersing to another patch (MW, W =1,157,
pvalue =0.862). Also in the Alps, movements were not
significantly different between sexes (MW, W =22,
pvalue =0.471), but this may be due to the few events of
inter-patch dispersal recorded for females. We found a
significant difference between males of the two populations.
Males of the Mediterranean population tended to move for
longer distances than those of the Alpine one (ALP: 132.8 m;
MED: 309.0 m; MW, W =316, pvalue =0.0005) but the
same was not true for females (ALP: 208.9 m; MED:
300.5 m; MW, W =15, pvalue =0.381).
Factors influencing dispersal
As concerns patch area, scaling parameters for emigration
(f
em
) and immigration (f
im
) differed between populations
Fig. 2 Parameters estimated by
Virtual Migration Model plotted
with their confidence intervals
(95 %) for the Alpine (squares)
and the Mediterranean (circles)
populations of E. aurinia.
Values were calculated
separately for females (white)
and males (black). Bars
represent 95 % confidence
intervals. aMortality within
patch (l). bMigration rate for 1
hectare patch (g). cDistance
dependence of migration (a).
dMigration mortality (k)
Fig. 3 Distance of daily movements within (a) and between
(b) patches recorded in the Alpine (ALP) and Mediterranean
(MED) populations, calculated separately for females (white) and
males (black). The boxes show the median, first and third quartile, the
whiskers the minimum and maximum values. Outliers are plotted as
open circles. Comparisons were significant only for males. *p\0.05,
**p\0.01
J Insect Conserv
123
Author's personal copy

(Fig. 4). In the Mediterranean population they equalled
zero for both sexes, thereby indicating that there was no
effect of patch area on emigration and/or immigration
(Fig. 4a, b).
In the Alpine system, on the contrary, patch area
affected both emigration and immigration, although in the
case of females, the observed values presented large con-
fidence intervals (Fig. 4c, d). The combination of negative
effect of patch area on emigration (-0.40) and its positive
effect on immigration (0.80) indicates that large patches
discourage emigration and promote immigration, in par-
ticular in the case of males (Fig. 4c, d).
When patch areas were modified to account for patch
quality, calculated separately as cover in food plant and
nectar sources, the emigration scaling coefficient estimated
for females of the Mediterranean population differed from
zero when area was corrected for nectar sources. Thus, a
higher abundance of nectar sources can dampen the
females’ tendency to emigrate (Fig. 4a). On the other hand,
it was the cover in food plants (gentians) that reduced
females’ emigration, in the Alpine population (Fig. 4c).
Interestingly, the immigration scaling coefficient had a
positive value when area was corrected for the number of
specimens, both of the same and of opposite sex (Fig. 4b,
d). This is true in all the cases, but in particular for males of
the Mediterranean population, when area is corrected for
female numbers (f
im
reached the highest value, 1.2—
Fig. 4b), thereby indicating that high female densities
strongly promoted male immigration.
In Fig. 5, the daily probability of emigration was plotted
against patch area. The values of daily probability of
emigration were higher for the Mediterranean population
and males were more likely to emigrate than females. In
the Mediterranean population the probability of emigration
was not affected by patch area in any of the two sexes. On
the contrary, in the Alpine system, males and females were
more likely to emigrate from small patches, while migra-
tion probability decreased in the case of larger patches.
Comparing patches of similar size, however, the daily
emigration rate was around 10 % for the Alpine males and
Fig. 4 Virtual Migration Model estimated values of Emigration and
Immigration scaling coefficients and their confidence limits (95 %)
calculated on the basis of real area (area), and on ‘‘effective’’ areas.
The latter was calculated by considering either the cover of (i) food
plant or (ii) nectar sources, and the number of specimens of either (iii)
the same sex or the (iv) opposite sex. Each coefficient was estimated
separately for females (white) and for males (black). a,bshow data
for the Mediterranean population; c,dfor the Alpine population
Fig. 5 Daily probability of emigration from a given patch, based on
the estimates provided by VM model for females (white) and males
(black) of the Alpine (squares) and Mediterranean (circles) popula-
tion of E. aurinia
J Insect Conserv
123
Author's personal copy

20 % for the Mediterranean ones (patch area &0.5 ha). In
the largest patch of the Alpine system (5–6 ha), the daily
emigration rate went down to as far as 5.6 % for males and
to 2.6 % in females.
We obtained from 1 to 3 appropriate models (DAIC \2;
Table 2) for males of both populations. In the case of Alpine
females, however, the fraction of immigrant and emigrant
individuals was too low to allow for any statistical com-
parison. In both systems the fraction of immigrating males
was positively affected by the density of males in the
‘‘recipient’’ patch, but it was also influenced by the patch
area in the Mediterranean system, and by the grass height in
the Alps. Female density was a significant predictor of
emigrating males in both populations, indicating that low
numbers of females at the ‘‘source’’ patch encourage males
to disperse. Patch connectivity also entered the best appro-
priate models, suggesting that males tend to disperse more
successfully from ‘‘source’’ to ‘‘recipient’’ patches having
high connectivity values (Table 2).
Connectivity (Sj), together with patch area, also influ-
ences female migration in the Mediterranean system,
although parameter estimates were not significant. The
fraction of immigrating females was positively affected by
the density of nectar sources and by the grass height at the
‘‘recipient’’ patches (Table 2).
Estimated daily mortalities during migration are shown
as a function of connectivity (ln Sj) in Fig. 6, for both
populations. For the Alpine system, patch connectivity
ranged from 1.18 to -2.88 and the estimated daily mortality
rate during migration was less than 10 %. In the case of the
most isolated patch (connectivity value \-2), however,
migration mortality drastically increased beyond 60 %. This
pattern was evident for males, but not for females, whose
migration mortality values were equal to zero across the
whole patch system, probably because of too few recapture
data. When grouping together 3 habitat patches where the
lowest number of adult females was recaptured, the values of
migration mortality differed from zero (dotted squares and
curves—Fig. 6). Although the connectivity values for the
Mediterranean patch system ranged from 0.21 to 0.98, the
slope of the curve describing the relationship between
migration mortality and connectivity was steeper than the
one observed for the Alpine system, for both males and
females. At this connectivity interval (0 \ln Sj\1) the
Mediterranean population showed estimated daily mortality
ranging from 13 to 35 % while the estimates for the Alpine
system hardly exceeded 1 %.
Discussion
The aim of this work was to study the metapopulation
dynamics of two E. aurinia populations occurring in two
Table 2 Best generalized linear models, explaining the fraction of immigrating and emigrating males and females in our two populations
Sex Model
number
Intercept N of individuals Patch features Vegetation AIC
c
Males Females Area (m
2
) Connectivity (S
j
) Nectar sources (%) Grass cover (%) Grass height (cm)
E. a. provincialis
Fraction of Immigrants #1-11.820 ±1.649** 0.646 ±0.090** 0.600 ±0.160** 98.6
Fraction of Emigrants #1-0.782 ±0.478 -0.245 ±0.133 0.861 ±0.392* 89.2
2-1.530 ±0.250** 0.558 ±0.345 89.4
3-0.691 ±0.475 -0.114 ±0.120 91.2
Fraction of Immigrants $1-19.227 ±5.052** 0.488 ±0.217* 3.954 ±1.344** 70.3
2-15.430 ±4.890** 3.431 ±1.362 72.2
Fraction of Emigrants $1-0.232 ±0.483** 0.558 ±0.563 76.4
2-4.131 ±2.628 0.274 ±0.280 76.4
E. a.glaciegentia
Fraction of Immigrants #1-7.495 ±1.074** 0.293 ±0.103** 1.029 ±0.421* 108.1
Fraction of Emigrants #1 16.079 ±3.577** -0.884 ±0.163** -0.607 ±0.206** -1.933 ±0.538** 81.8
2 3.263 ±1.660* -0.223 ±0.109* -0.492 ±0.191* 0.474 ±0.155** 83.8
The predictor variables entering the initial models were males and females of both populations, to test for the effect of average male and female densities, cover and height of food plants and their interaction; cover of the vegetational
components; patch area and connectivity (Sj). Coefficient estimates ±standard errors and AICc values of each best (DAIC \2) model are shown
*p\0.05, **p\0.01
J Insect Conserv
123
Author's personal copy

different environmental frameworks. In choosing the two
populations we ensured that (i) both of them included a
considerable number of individuals, (ii) both were spatially
organised in a system of linked patches and (iii) the habi-
tats occupied by each population was in relatively suitable
conservation conditions.
Population dynamics
In both populations the sex-ratio is strongly male-biased, a
fact that can negatively influence their effective population
size (Brook 2008). This pattern, however, is commonly
observed in MRR studies carried out on Euphydryas au-
rinia in many European countries (e.g. Belgium: Schtick-
zelle et al. 2005; the Czech Republic: Fric et al. 2010;
Finland: Wahlberg et al. 2002; N. China: Wang et al.
2004). As shown by VMM estimates, expressed as a higher
catchability, our results may be partially explained by the
stronger detectability of males, which makes them easier to
net than females, especially in the Alpine population.
Males are also longer-lived than females, particularly at
our Mediterranean site, which may increase the probability
that each of them is caught at least once in its lifetime.
Another explanation for this unbalanced sex ratio may be a
consequence of males’ mate-searching behaviour, which in
this case, as in most Melitaeini, is described as an alter-
nating mixture of perching and patrolling (see Wahlberg
2000). In both populations we observed patrolling, but in
the Alpine site this behaviour is apparently mixed with a
lek-like assembly strategy (see Wickman and Rutowski
1999), where male aggregations allow for their higher
probability of detection and capture. At the same time,
however, a highly biased sex-ratio towards males is
apparent and cannot be justified by different catchability
alone.
Dispersal behaviour
Virtual Migration parameters have shown strong differ-
ences between populations as well as between sexes. In our
study system males are more prone to move than females
and we observed a higher emigration rate (g) for the
Mediterranean population. In contrast, adults of the Alpine
population are more sedentary and consequently show
higher within patch mortality (l). In our study, differences
between populations are stronger then between sexes, a
finding that contrasts with results obtained on several other
Melitaeini, where differences between males and females
are generally stronger than between species (see Fric et al.
2010).
Other works have observed that dispersal parameters
differ between metapopulations located in landscapes
characterised by different networks of patches and different
matrix composition, in a variety of butterflies (e.g. Sch-
tickzelle et al. 2006; Turlure et al. 2011; Bonelli et al.
2013; Nowicki et al. 2014), which clearly shows that dis-
persal propensity is highly context-dependent (Baguette
and Van Dyck 2007).
Fragmented landscapes usually imply lower emigration
rates (e.g. Schtickzelle et al. 2006; Nowicki et al. 2014),
and a stronger and inverse influence of patch area on
emigration propensity (Mennechez et al. 2004; Schtickzelle
et al. 2006). This is apparently in contrast with our results,
since we observed higher propensity to move in the Med-
iterranean system where, although the open areas are
fragmented in a wooded matrix, dispersal parameters
suggest a higher degree of landscape connectivity. Con-
sequently, as already suggested by other authors (e.g.
Baguette and Van Dyck 2007) it is necessary to consider
landscape connectivity not only in its structural component
(spatial configuration of habitat patches), but mainly in its
functional counterpart (how the behaviour of organisms is
actually affected by landscape structure).
Dispersal rate, in butterflies and more in general in
insects, can be positively influenced by temperature. Warm
weather promotes dispersal (e.g., Dennis and Bardell 1996;
Walters et al. 2006; Mitikka et al. 2008) and dispersal can
be influenced by weather fluctuation, with hotter years
characterised by higher dispersal rates (e.g. Franze
´n and
Nilsson 2012). In any case, the relationship between tem-
perature, flight and dispersal propensity is not always
straightforward and clear (e.g., Matter et al. 2011), because
many other factors, both climate related or not, are also
involved.
Differences between our two populations could be par-
tially explained in the light of different climatic conditions,
Fig. 6 Relationship between daily mortality during emigration and
connectivity, based on the estimated values of parameters generated
by VM model, for females (white) and males (black) of the Alpine
(squares) and Mediterranean (circles) populations of E. aurinia.
Dotted curve and squares were obtained for females of the Alpine
population, grouping together 3 habitat patches where few recaptures
of adult females occurred
J Insect Conserv
123
Author's personal copy

but not directly by temperature. Individuals of the Alpine
population are adapted to low temperature and can fly also
in days with suboptimal weather conditions (own obser-
vations). Similarly to that observed by Junker et al. (2010),
our Alpine population is characterised by low mobility,
which may be seen as an adaptation to the alpine envi-
ronment and climate. As these authors underline, such
behaviour could prevent accidental drift events due to
strong wind. Indeed, many high-altitude species living in
open environment show lower dispersal capacity than
species living in forest clearings (see Junker et al. 2010 and
references therein).
In our Alpine system, patches are more isolated than in
the Mediterranean one, where emigration is more risky in
terms of daily mortality. The degree of connectivity of the
system needed to ensure the long term survival of a
metapopulation should, therefore, be different in the two
systems.
According to Travis et al. (2012), selection should
operate at two parallel levels and should have favoured
both the ability to reach a new patch and the propensity to
leave the old patch, even though movement towards a new
patch is necessarily more risky and energetically expen-
sive. This is certainly true in a hostile matrix where only
highly skilled or ‘lucky’ individuals are able to find new
suitable sites. In our Mediterranean population, adults have
occupied until recently a vast and very well preserved area,
where patches were not very fragmented and were always
rich in nectar sources as well as in larval food plants. With
the abandonment of traditional agricultural practices, the
system became progressively more fragmented because of
forest expansion, and movements between patches have
become more risky. In other words, the recent emergence
of wooded barriers may have caught the adults of our
Mediterranean population unprepared. Despite occurring in
a still relatively good habitat, populations having these
characteristics are therefore particularly at risk if the pro-
cess of recolonisation of grasslands by forest continues.
Even if males are more prone to move between patches,
particularly if these are well connected or close to each
other, the mean distance that they can cover is not signif-
icantly larger than the distance covered by females, and is
in general small. The Mediterranean population not only
shows higher propensity to move, but adults fly longer
distances than those from the Alpine one (calculated daily)
when moving to a different patch (aparameter of VM). On
the other hand, adults of the Alpine population move
longer distances in intra-patch movements but this differ-
ence may be partly explained by the larger patch sizes in
the Alpine system.
According to Hanski et al. (2000), patch area strongly
influences movements. In our Alpine system where patches
present stronger differences in size, it is clear that
butterflies are more prone to move from the small patches
to the bigger patches, and small patches receive fewer
immigrating adults. However, in the Mediterranean system,
where the landscape is a matrix of more similarly sized
patches in a homogeneous context created by an anthropic
tradition of land use, immigration and emigration are not
driven by patch size. Recalculating the VM parameters
according to Rabasa et al. (2007), we have been able to
evidence that in both systems movements are driven by a
density factor. In the Mediterranean population this factor
actually is the only one that explains the immigration and
emigration scaling coefficient.
In particular, the immigration scaling coefficient (f
im
)is
positively influenced by higher numbers of specimens both
of the same and opposite sex. This type of immigration
may be considered a kind of Allee effect (Greene and
Stamps 2001; Altwegg et al. 2013), which occurs when
species show conspecific attraction. High densities of
conspecifics may suggest good habitat quality, as well as an
opportunity for finding possible mates (Greene and Stamps
2001; Altwegg et al. 2013).
Comparing European Euphydryas aurinia populations
Since Hanski et al. (2000) proposed the modelling
approach known as Virtual Migration Model to estimate
crucial parameters in metapopulation systems, measures of
migration became comparable. In the last decade these
parameters were estimated in different butterfly popula-
tions, some of which belong to the Euphydryas aurinia
complex (Wahlberg et al. 2002; Wang et al. 2004; Sch-
tickzelle et al. 2005; Fric et al. 2010). Comparing pro-
pensity to migrate, as expressed by the gparameter, we
observe both a general pattern as well as visible differences
among populations (Fig. 7). In general, females are more
sedentary than males, with strong differences among
populations.
The E. aurinia populations studied in N China, in the
Czech Republic and at our two Italian sites show higher
propensity to move than those from Finland and Belgium.
Observing the system of patches where our E. aurinia
populations occur, we notice that differences in emigration
propensities cannot be explained by distances between
patches.
All the analyzed metapopulations are organized in
groups of patches quite close to each other (see legend of
Fig. 7for patch system descriptions of each E. aurinia
population). The only exception is the population studied in
Finland by Wahlberg et al. (2002), which occupies a net-
work of more isolated patches, constituted by 12 patches,
within 1.5 92 km. In general we can observe that emi-
gration propensity (g) is stronger in bigger metapopula-
tions occupying a system of more numerous patches, and
J Insect Conserv
123
Author's personal copy

even when patches are reciprocally not very close. E. au-
rinia populations from Finland and Belgium show the
lowest gvalues (i.e. the parameter describing propensity to
move). Both populations are rather (Finland) or very small
(Belgium) and have been studied in a system of 11–12
patches. In contrast, the highest values of migration pro-
pensity are shown in the large populations from China and
the Czech Republic, which were studied in a system of
more than 30 well connected patches. The two Italian
populations, both quite numerous and each occupying a
system of more than 15 patches, show intermediate values.
One can speculate that a bad conservation status and a
small number of individuals may depress populations and
make individuals slightly more mobile, even in a system of
closed patches. This might be the condition for many
European populations of E. aurinia.
In this case, management should be aimed at improving
the size of suitable habitat and consequentially also popu-
lation size (see Anthes et al. 2003). On the other hand,
strong habitat fragmentation may depress propensity to
move, as in the case of the Finnish population where dis-
tances among patches are more than triple those in the
other sites. In this case the probability to reach a new
suitable patch is low and mortality during migration (k)is
above zero (Wahlberg et al. 2002).
Conclusion
As is the case with several other systems (e.g. Maculinea
arion, see Bonelli et al. 2013), dispersal propensity varies
depending on environmental pressures. Not only the pro-
pensity to move differs between species, populations and
sexes, but consequences of individual movement, in terms
of mortality during migration, differ depending on context.
This is not only interesting per se but fundamental while
making concrete management decisions, always keeping in
mind necessities of ‘‘actual functional connectivity’’ (see
Calabrese and Fagan 2004).
Of course we cannot always simplify habitats in order to
destroy or avoid barriers. What we can do is to try and
create or maintain permeable barriers that allow grassland
specialist species to penetrate.
Where habitat restoration is planned and resources for
long term species conservation are available we suggest
that a population-centred approach rather than a species
centred approach is pursued (see Casacci et al. 2013).
In this work we observed that not only are butterflies
sensitive to fragmentation and can be differently prone to
leave their native patch, but the same level of isolation
among patches can play different roles in different land-
scape contexts. Consequences of these findings need to be
addressed in any management program.
Acknowledgments We would like to thank authorities of the Gran
Paradiso National Park and Capanne di Marcarolo Natural Park for
financial support and for giving permissions to perform this research.
The work was carried out with the authorization (prot. 0039115/PNM,
20/06/2013) by the Italian Ministry of the Environment MATTM and
within the project ‘A multitaxa approach to study the impact of cli-
mate change on the biodiversity of Italian ecosystems’ of the Italian
Ministry of Education, University and Research (MIUR). F. Barbero
was partially supported by Italian Ministry of Education, University
and Research (MIUR). Authors are also very grateful to Emanuel
Rocchia and Eleonora Rossi, for their fundamental help in the
fieldwork.
References
Altwegg R, Collingham YC, Erni B, Huntley B (2013) Density-
dependent dispersal and the speed of range expansions. Divers
Distrib 19:60–68
Anthes N, Fartmann T, Hermann G, Kaule G (2003) Combining
larval habitat quality and metapopulation structure—the key for
successfully management of pre-alpine Euphydryas aurinia
colonies. J Insect Conserv 7:175–185
Asher J, Warren M, Fox R, Harding P, Jeffcoate G, Jeffcoate S (2001)
The millennium atlas of butterflies in Britain and Ireland. Oxford
University Press, Oxford
Baguette M, Van Dyck H (2007) Landscape connectivity and animal
behavior: functional grain as a key determinant for dispersal.
Land Ecol 22:1117–1129
Balletto E and Cassulo L (1995) Lepidoptera Hesperioidea, Papilio-
noidea. In: Minelli A, Ruffo S, La Posta S (eds) Checklist delle
specie della fauna italiana. Calderini, Bologna and Ministero per
l’Ambiente, Roma. pp 11
Balletto E, Bonelli S, Cassulo L (2007) Insecta Lepidoptera
Papilionoidea (Rhopalocera). In: Ruffo S, Stoch F (eds) Check-
list and distribution of the Italian fauna. Ministero
Fig. 7 Estimated values of VM model emigration propensity (g), as
reported in literature for E. aurinia populations of different Countries.
Belgium (Schtickzelle et al. 2005): 11 patches within 0.3 91 km;
Finland (Wahlberg et al. 2002): 12 patches, within 1.5 92 km;
Czech Republic (Fric et al. 2010): 30 patches within 2 91.5 km;
China (Wang et al. 2004): 38 patches, within 3 93 km. The Italian
data referred to our research in the Alpine (ALP) 16 patches, within
391.5 km and in the Mediterranean (MED) sites, 15 patches, within
290.6 km. Values of females (white) and males (black) are reported
separately
J Insect Conserv
123
Author's personal copy

dell’Ambiente. Memorie del Museo civico di Storia naturale di
Verona, 2nd Serie, Sezione Scienze della Vita, pp 259–263
Balletto E, Cassulo L, Bonelli S (2014a) Annotated Checklist of the
Italian Butterflies and Skippers (Papilionoidea, Hesperiioidea).
Zootaxa 1:1–114
Balletto E, Bonelli S, Zilli A (2014b) Lepidotteri. In: Genovesi P et al.
(eds) Specie e habitat di interesse comunitario in Italia:
distribuzione, stato di conservazione e trend. Rapporti ISPRA
Barton K (2013) MuMIn: Multi-model inference. R package version
1.9.13. http://CRAN.R-project.org/package=MuMIn
Bonelli S, Cerrato C, Loglisci N, Balletto E (2011) Population
extinctions in the Italian diurnal Lepidoptera: an analysis of
possible causes. J Insect Conserv 15:879–890
Bonelli S, Vrabec V, Witek M, Barbero F, Patricelli D, Nowicki P
(2013) Selection against dispersal in isolated butterfly metapop-
ulations. Popul Ecol 55:469–478
Bowler DE, Benton TG (2005) Causes and consequences of animal
dispersal strategies: relating individual behaviour to spatial
dynamics. Biol Rev Camb Philos Soc 80:205–225
Braun-Blanquet J (1932) Plant Sociology. New York
Brook BW (2008) Demographic versus genetics in conservation
biology. In: Carroll SP, Fox CW (eds) Conservation biology.
Oxford University Press, pp 35–49
Burnham KP, Anderson DR (2004) Multimodel inference: under-
standing AIC and BIC in model selection. Sociol Methods Res
33:261–304
Calabrese JM, Fagan WF (2004) A comparison-shopper’s guide to
connectivity metrics. Front Ecol Environ 2:529–536
Casacci LP, Barbero F, Balletto E (2013) The ‘‘Evolutionarily
Significant Unit’’ concept and its applicability in biological
conservation. It J Zool. doi:10.1080/11250003.2013.870240
Dennis RLH, Bardell P (1996) The impact of extreme weather events
on local populations of Hipparchia semele (L.) (Nymphalidae)
and Plebejus argus (L.) (Lycaenidae): hindsight, inference and
lost opportunities. Ent Gaz 47:211–225
Descimon H, Mallet J (2009) Bad species. In: Settele J, Shreeve TG,
Konvicka M, Van Dyck H (eds) Ecology of butterflies in Europe.
Cambridge University Press, Cambridge, pp 219–249
Franze
´n M, Nilsson SG (2012) Climate-dependent dispersal rates
in metapopulations of burnet moths. J Insect Conserv
16:941–947
Fric Z, Hula V, Klimova M, Zimmermann K, Konvicka M (2010)
Dispersal of four fritillary butterflies within identical landscape.
Ecol Res 25:543–552
Goodwin BJ (2003) Is landscape connectivity a dependent or
independent variable? Landsc Ecol 18:687–699
Greene CM, Stamps JA (2001) Habitat selection at low population
densities. Ecology 82:2091–2100
Hanski I, Alho J, Moilanen A (2000) Estimating the parameters of
survival and migration of individuals in metapopulations.
Ecology 81:239–251
IUCN (2013) IUCN Red List of Threatened Species. Version 2013.2.
www.iucnredlist.org
Junker M, Wagner S, Gros P, Schmitt T (2010) Changing demog-
raphy and dispersal behaviour: ecological adaptations in an
alpine butterfly. Oecologia 164:971–980
Matter SF, Doyle A, Illerbrun K, Wheeler J, Roland J (2011) An
assessment of direct and indirect effects of climate change for
populations of the Rocky Mountain Apollo butterfly (Parnassius
smintheus Doubleday). Insect Sci 18:385–392
Mennechez G, Petit S, Schtickzelle N, Baguette M (2004) Modelling
mortality and dispersal: consequences of parameter generalisa-
tion on metapopulation dynamics. Oikos 106:243–252
Mikheyev AS, Mcbride CS, Muelleret UG et al (2013) Host-
associated genomic differentiation in congeneric butterflies: now
you see it, now you do not. Mol Ecol 22:4753–4766
Mitikka V, Heikkinen RK, Luoto M, Araujo MB, Saarinen K, Poyry
J, Fronzek S (2008) Predicting range expansion of the map
butterfly in Northern Europe using bioclimatic models. Biodivers
Conserv 17:623–641
Nowicki P, Vrabec V, Binzenho
¨fer B, Feil J, Zaks
ˇek B, Hovestadt T,
Settele J (2014) Butterfly dispersal in inhospitable matrix: rare,
risky, but long-distance. Landsc Ecol 29:401–412
Pen
˜uelas J, Sardans J, Stefanescu C, Parella T, Filella I (2006)
Lonicera implexa leaves bearing naturally laid eggs of the
specialist herbivore Euphydryas aurinia have dramatically
greater concentrations of iridoid glycosides than other leaves.
J Chem Ecol 32:1925–1933
Petit S, Moilanen A, Hanski I, Baguette M (2001) Metapopulation
dynamics of the bog fritillary butterfly: movements between
habitat patches. Oikos 92:491–500
R Core Team (2012) R: a language and environment for statistical
computing. R Foundation for Statistical Computing, Vienna,
Austria. http://www.R-project.org/
Rabasa SG, Gutie
´rrez D, Escudero A (2007) Metapopulation structure
and habitat quality in modelling dispersal in the butterfly Iolana
iolas. Oikos 116:793–806
Schtickzelle N, Choutt J, Goffart P, Fichefet V, Baguette M (2005)
Metapopulation dynamics and conservation of the marsh fritil-
lary butterfly: population viability analysis and management
options for a critically endangered species in Western Europe.
Biol Conserv 126:569–581
Schtickzelle N, Mennechez G, Baguette M (2006) Dispersal depres-
sion with habitat fragmentation in the bog fritillary butterfly.
Ecology 87:1057–1065
Sinama M, Dubut V, Costedoat C et al (2011) Challenges of
microsatellite development in Lepidoptera: Euphydryas aurinia
(Nymphalidae) as a case study. Eur J Entomol 108:261–266
Singer MC, Wee B (2005) Spatial pattern in checkerspot butterfly-
host plant association at local, metapopulation and regional
scales. Ann Zool Fenn 42:347–361
Stefanescu C, Penuelas J, Sardan J, Filella I (2006) Females of the
specialist butterfly Euphydryas aurinia (Lepidoptera: Nymphal-
idae: Melitaeini) select the greenest leaves of Lonicera implexa
(Caprifoliaceae) for oviposition. Eur J Entomol 103:569–574
Stevens VM, Turlure C, Baguette M (2010a) A meta-analysis of
dispersal in butterflies. Biol Rev Camb Phil Soc 85:625–642
Stevens VM, Pavoine S, Baguette M (2010b) Variation within and
between closely related species uncovers high intra-specific
variability in dispersal. PLoS ONE 5:e11123
S
ˇvitra G, Sielezniew M (2010) The first observation of Euphydryas
aurinia (Lepidoptera, Nymphalidae) developing on Gentiana
cruciata in Lithuania. Polish J Entomol 79:195–201
Travis JM, French DR (2000) Dispersal functions and spatial models:
expanding our dispersal toolbox. Ecol Lett 3:163–165
Travis JM, Mustin K, Barton
´KA et al (2012) Modelling dispersal: an
eco-evolutionary framework incorporating emigration, move-
ment, settlement behaviour and the multiple costs involved. Met
Ecol Evol 3:628–641
Turlure C, Baguette M, Stevens VM, Maes D (2011) Species-and sex-
specific adjustments of movement behavior to landscape heter-
ogeneity in butterflies. Behav Ecol 22:967–975
van Swaay C, Wynhoff I, Verovnik R, Wiemers M, Lo
´pez Munguira
M, Maes D, Sasic M, Verstrael T, Warren M, Settele J (2010)
Euphydryas aurinia. In: IUCN 2013. IUCN red list of threatened
species. Version 2013.2. www.iucnredlist.org
Wahlberg N (2000) Comparative descriptions of the immature stages
and ecology of five Finnish melitaeine butterfly species (Lep-
idoptera: Nymphalidae). Entomol Fenn 11:167–174
Wahlberg N, Klemetti T, Selonen V, Hanski I (2002) Metapopulation
structure and movements in five species of checkerspot butter-
flies. Oecologia 130:33–43
J Insect Conserv
123
Author's personal copy

Walters RJ, Hassall M, Telfer MG, Hewitt GM, Palutikof JP (2006)
Modelling dispersal of a temperate insect in a changing climate.
Proc R Soc B 273:2017–2023
Wang R, Wang Y, Chen J, Lei G, Xu R (2004) Contrasting movement
patterns in two species of chequerspot butterflies, Euphydryas
aurinia and Melitaea phoebe, in the same patch network. Ecol
Entomol 29:367–374
Wickman PO, Rutowski RL (1999) The evolution of mating
dispersion in insects. Oikos 84:463–472
Zimmermann M, Aubert J, Descimon H (1999) Syste
´matique
mole
´culaire des Me
´lite
´es. Cr Accad Sci Paris (Sciences de la
vie) 322:429–439
Zimmermann M, Wahlberg N, Descimon H (2000) Phylogeny of
Euphydryas Checkerspot Butterflies (Lepidoptera: Nymphali-
dae) based on mitochondrial DNA sequence data. Ann ent Soc
Amer 93:347–355
J Insect Conserv
123
Author's personal copy