Content uploaded by Sara Fratini
Author content
All content in this area was uploaded by Sara Fratini
Content may be subject to copyright.
1 23
Biological Invasions
ISSN 1387-3547
Biol Invasions
DOI 10.1007/s10530-013-0487-y
Establishing the eradication unit of Molara
Island: a case of study from Sardinia, Italy
Lapo Ragionieri, Giulia Cutuli, Paolo
Sposimo, Giovanna Spano, Augusto
Navone, Dario Capizzi, Nicola Baccetti,
Marco Vannini, et al.
1 23
Your article is protected by copyright and all
rights are held exclusively by Springer Science
+Business Media Dordrecht. 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
Establishing the eradication unit of Molara Island: a case
of study from Sardinia, Italy
Lapo Ragionieri •Giulia Cutuli •Paolo Sposimo •
Giovanna Spano •Augusto Navone •Dario Capizzi •
Nicola Baccetti •Marco Vannini •Sara Fratini
Received: 8 August 2012 / Accepted: 4 May 2013
ÓSpringer Science+Business Media Dordrecht 2013
Abstract Molara is a small island belonging to the
Marine protected Area Tavolara—Punta Coda Caval-
lo, in Sardinia. During 2006–2007, a bio-monitoring
program reported a strong presence of the black rat,
Rattus rattus, on Molara island. Rat predation has
detrimentally affected the unique biodiversity of this
island, thus, in 2008 an eradication campaign was
conducted. Our eradication protocol included a pre-
eradication genetic investigation, using 8 microsatel-
lite loci, on a rat population of Molara as well as on
neighbour islands within the Marine Protected Area
(MPA). The main goal of this genetic investigation
was to establish the correct borders of the eradication
unit of Molara island. As several recent eradication
campaigns have been unsuccessful, due to incomplete
and unstable eradication, we also aimed to assess
possible hidden sources of reinvasion. Specimens
were also collected during post- eradication monitor-
ing on Molara for genetic screening to establish their
origin, and thus validate the effectiveness of our
eradication campaign. According to our genetic anal-
ysis, within the MPA there are four different eradica-
tion units, corresponding to the islands of Molara,
Tavolara, Piana and to the Sardinia mainland. Gene
flow among these four units is more or less absent. The
assignment and clustering tests performed on pre and
post-eradication samples seem to indicate that the
population of Sardinia mainland is a possible source of
re-invasion for the Piana and Molara populations.
L. Ragionieri (&)G. Cutuli M. Vannini
S. Fratini (&)
Department of Biology, University of Florence, via
Madonna del Piano 6, 50019 Sesto Fiorentino, Italy
e-mail: lapo.ragionieri@ua.pt
S. Fratini
e-mail: sarafratini@unifi.it
L. Ragionieri
RNA Biology Laboratory, Department of Biology and
CESAM, University of Aveiro, 3810-193 Aveiro,
Portugal
P. Sposimo
Nature and Environment Management Operators
srl(NEMO), Follonica, GR, Italy
G. Spano A. Navone
Consorzio di Gestione Area Marina Protetta Tavolara
Punta Coda Cavallo, Olbia, Italy
D. Capizzi
Regional Park Agency, via del Pescaccio 96, 00166
Rome, Italy
N. Baccetti
ISPRA, via Ca’ Fornacetta 9, 40064 Ozzano Emilia, BO,
Italy
123
Biol Invasions
DOI 10.1007/s10530-013-0487-y
Author's personal copy
Keywords Population genetics Eradication
campaign Rattus rattus Invasive species
Mediterranean Sea
Introduction
The black rat (Rattus rattus), the Norway rat (Rattus
norvegicus) and the Pacific rat (Rattus exulans) are
recognised as dangerous worldwide pest (Lowe et al.
2000; King et al. 2011), and during the last decades
many eradication campaigns have been conducted,
particularly in insular systems and forests, in order to
maintain and protect the biodiversity, which the
presence of rats may impact. According to these
studies, only a few years after these eradications,
incredible re-establishment of lost biodiversity has
been observed (Towns et al. 2001; Graham and Veitch
2002; Kerbiriou et al. 2004; Pascal et al. 2005; Amaral
et al. 2010; Veitch et al. 2011). However, there are also
examples of eradication campaigns that were not as
successful, and a complete and stable eradication was
not achieved (Thorsen et al. 2000; Courchamp et al.
2003; Parkes et al. 2011; Savidge et al. 2012). Various
factors may have been responsible for the failure of
eradication in these cases, such as the capability of rats
to re-invade the same environments, and the presence
of some individuals which survived the eradication
(Abdelkrim et al. 2007). In fact, the high reproduction
rate of rats, coupled with the absence of predators and/
or competitors, can counteract onerous eradication
efforts in only a few years (Abdelkrim et al. 2007;
Russell et al. 2009a; Russell et al 2009b). For example,
within island systems, rats originating from popula-
tions located on neighbor islands or mainland sites
may re-invade an eradicated island (Russell et al.
2010). For these reasons, in recent years, the ‘‘erad-
ication unit’’ concept (sensu Abdelkrim et al. 2007)
has been defined as ‘‘the interconnected populations
that must be eradicated at the same time to prevent
rapid recolonization’’ (for a more comprehensive view
of the notion of eradication unit also see: Robertson
and Gemmell 2004; Abdelkrim et al. 2005a,2007,
2010; Capizzi et al. 2010).
The use of genetic techniques for assessing the
geographic boundaries of an eradication unit has risen
considerably in recent studies (Abdelkrim et al. 2010;
Russell et al. 2010; Savidge et al. 2012). From a
genetic perspective, an eradication unit consists of a
group of populations among which the gene flow is
high enough to genetically homogenize populations.
The success of an eradication project thus relies on the
removal of all the populations belonging to that
eradication unit, to reduce the risk of further re-
invasion from interconnected populations (Abdelkrim
et al. 2007). Genetic methods are also useful in
clarifying the geographic sources of past and future
arrivals (Pinceel et al. 2005; Carvalho et al. 2009).
Genetic analysis conducted on individuals collected
during post-eradication monitoring, for instance,
recorded a discrepancy between the low rates of gene
flow estimated among rat populations from different
islets and the source of reinvasion in eradicated
populations (Abdelkrim et al. 2007). The most plau-
sible explanation for this phenomenon arises from a
tracking experiment conducted on rats released on
islands already populated by conspecifics (Granjon
and Ceylan 1989). In this study, all the individuals
released on islands, where a well established rat
colony was formally present, died of injuries within a
few days, suggesting strong role of intra-specific
competition. Thus, such behavior of rats suggests
caution in the delimitation of the true geographic
boundaries of an eradication unit.
The Marine Protected Area Tavolara Punta Coda
Cavallo is located in Sardinia, near Olbia, and
comprises two main islands (Tavolara and Molara),
other islets and 76,09 km of the Sardinian coastline. In
2006–2007, a bio-monitoring program performed on
Molara island reported a strong presence of the black
rat (R. rattus), which preys on chicks and eggs of
nesting seabirds. Consequently, an eradication cam-
paign was conducted on this island in 2008. The
eradication protocol included a pre-eradication
genetic investigation of rat populations belonging to
the islands of Molara, Tavolara, Piana and in the area
of Capo Coda Cavallo, on mainland Sardinia. This
study, through genotyping 8 microsatellite loci in
approximately 30 individuals per population, aimed to
investigate the extent of the Molara eradication unit
and to establish the level and direction of gene flow
among rats populations within the MPA Tavolara
Punta Coda Cavallo. Genetic analysis, using the same
microsatellite loci, was also performed on two new
individuals captured 21 months after the eradication
campaign, during the post-eradication monitoring on
Molara island. We aimed to assess whether they were
L. Ragionieri et al.
123
Author's personal copy
part of the eradicated population (if the eradication
campaign had not been completely successful) or
whether they were new arrivals from an unknown
source population.
Materials and methods
Study area and eradication project
Tavolara Punta Coda Cavallo was established as a
Marine Protected Area (MPA) in 1997; the MPA
comprises 15,000 ha of sea and 40 km of coastal land,
near Olbia, Sardinia (Italy, Fig. 1). The largest islands
are Tavolara (600 ha) and Molara (340 ha), with
several islets (Piana is the largest, with a surface of
12 ha). The island of Tavolara hosts a NATO military
post, and in summer the human presence on this island
is quite high, with people travelling daily to the island
by a ferry leaving from the village of San Paolo
(Sardinia mainland). Conversely, the human presence
on Molara island is significantly less, and is mainly
tourism based along the coast.
The faunal composition of these islands is typically
Mediterranean, with a high species richness of reptiles
and invertebrates. Moreover, these islands are a
preferred area for sea bird nesting. In 2005 and 2006
a monitoring program was conducted for three species
of marine bird: the European shag (Phalacrocorax
aristotelis desmarestii), Audouin’s gull (Larus
audouinii) and the largest global population of
Yelkouan shearwater (Puffinus yelkouan) (Zenatello
et al. 2011). The island of Molara hosts around
300–600 pairs of Yelkouan shearwaters, that nest
among fallen boulders of granite (Baccetti et al.
2009a). Unfortunately, the reproductive success of this
species during the monitoring program was estimated
to be equal to zero (Sposimo et al. 2012), due to the
strong presence of the black rat, which preys heavily
on the eggs and chicks of Yelkouan shearwaters. After
a preliminary census of rat population density, in
2008, an eradication campaign was initiated on Molara
Island. Rodenticide pellets of brodifacoum were
spread over the island using a bucket suspended from
a helicopter. The effect of the eradication project on
Molara was immediately evident as the reproductive
success of Yelkouan shearwater in 2009 and 2010
increased (Sposimo et al. 2012). In order to maintain
the success of this eradication project, the MPA is
planning new eradication projects on neighbor islands.
Sampling collection and DNA extraction
In order to establish the eradication unit of Molara
Island, specimens were collected using bait stations
with corns from three islands within the MPA using
bait stations with corns along five transects in two
different capture session of five consecutive nights:
Tavolara Island (N =30); Molara (N =30); Piana
(N =30); and one additional population from Sardi-
nia mainland, Capo Coda Cavallo (N =24). We also
collected two specimens in Molara after eradication.
For each captured individual, 10–50 g of tail muscle
tissue were preserved in pure ethanol. Then, DNA was
isolated using the Puregene Kit (Gentra System),
resuspended in TE buffer and then preserved at
-20 °C for further analysis.
Gene amplification
Currently, many loci described for Rattus norvegicus
and related species (Jacob et al. 1995) are available,
and from these we selected eight microsatellite loci
used in similar studies (Abdelkrim et al. 2005b,2009).
For detection of polymorphisms, six out of eight
primer combinations were divided into two different
sets based on similar annealing temperatures and
different fragments length (Set R1: D10Rat20,
D5Rat83, D7Rat13; Set R2: D9Rat13, D11Mgh5,
Fig. 1 Map indicating the collection localities of the four
population of R. rattus
A case of study from Sardinia, Italy
123
Author's personal copy
D16Rat8) for multiple PCRs. The two sets of loci were
amplified in a Perkin Elmer 9,600 thermal cycler using
master mix (Quiagen) by PCR mixture in 15 lLof
final volume containing: 1 lL of DNA, 3.5 lLof
master mix and 0.3–0.8 lL of primers 10 lM; the PCR
cycling conditions were: 35 cycles with 30 s for
denaturation at 95 °C, 90 s for annealing at 57 °C and
60 s for extension at 72 °C, preceded by 15 min of
initial denaturation at 95 °C, and followed by 10 min
of final extension at 72 °C. The remaining two loci
(D19Mit2 and D10Mit5) were separately amplified by
PCR mixture in 20 lL of final volume containing: 1 lL
of DNA, 2 lL of buffer 10X (Invitrogen), 2 mM of
MgCl2, 0.5 lL of primers 10 lm, 200 lm of each
dNTPs and 0.4 U of Taq (Invitrogen); the PCR cycling
conditions were: 35 cycles with 30 s for denaturation
at 94 °C, 45 s for annealing at 57 °C and 60 s for
extension at 72 °C, preceded by 10 min of initial
denaturation at 94 °C and followed by 30 min of final
extension at 72 °C. For each locus, the forward primer
was 50-labeled with a fluorescent dye of the three
different fluorophores (6-Fam, Hex and Ned).
For each set, 3.5 lL of each PCR product obtained
with master mix was mixed with 1.5 lL of PCR
product from single locus PCR (Set R1 ?D19Mit2;
Set R2 ?D10Mit5) and combined with water in a
final volume of 10 lL for successive dimensional
analysis. Sizing was performed in an ABI Prism 310
Genetic Analyzer (Applied Biosystems) with refer-
ence to an internal size standard (ROX400) using
GENOTYPER ver. 3.7 (Applied Biosystems).
Genetic diversity, population genetic structure
and bottlenecks
The number of alleles and the allelic richness for each
locus and population were calculated using FSTAT
ver. 2.9.3 (Goudet 1995). We estimated the Nei’s
standard genetic distance (Nei 1978) using Microsat-
ellites analyzer 4.05 (Dieringer and Schlo
¨tterer 2002).
Linkage Equilibrium among loci and Hardy–Wein-
berg equilibrium (HWE) were assessed for each
population using GENEPOP ver. 3.4 (Raymond and
Rousset 1995). We used the software MICRO-
CHECKER 2.2.3 (van Oosterhout et al. 2004)to
evaluate whether heterozygote deficiencies could be
explained by the existence of null alleles.
We estimated the genetic differentiation among
populations using the Exact test of population
differentiation (Raymond and Rousset 1995), as
implemented in GENEPOP. This test verifies the
existence of differences in allele frequencies at each
locus and for each population. Single locus pvalues
were calculated using a Markov chain with 1,000
batches and 1,000 iterations per batch, combined over
loci using the Fisher method.
The existence of population genetic structure was
also assessed by one level AMOVA (Excoffier et al.
1992), using ARLEQUIN ver. 3.11 (Excoffier et al.
2005). Significance of the fixation indices, under the
null hypothesis of no differentiation among popula-
tions, was tested using a non-parametric permutation
approach with 10,000 permutations.
In addition, the spatial analysis of molecular
variance (SAMOVA) was used to define groups of
populations that are geographically homogenous and
maximally differentiated from each other as imple-
mented in SAMOVA 1.1 (Dupanloup et al. 2002). The
aim of this approach is to define groups of populations
(K), which maximize the proportion of total genetic
variance due to differences among groups of popula-
tions (Fct), by means of an annealing procedure.
We used STRUCTURE version 2.3 (Pritchard et al.
2000) to infer population genetic structure. This
Bayesian cluster method takes a sample of genotypes
and uses the assumption of HWE and linkage
disequilibrium within sub-populations to find the
number of populations (K) that fits the data best and
the individual assignments which minimize HWE and
linkage disequilibrium in those populations. We used
the admixture model which is the most appropriate for
populations that may have recent ancestors from more
than one population (Pritchard et al. 2000). Likelihood
of model was assessed by the number of possible
clusters (K) ranging between 1 and 4. A further
analysis was performed including the two samples
collected on MOL during the post eradication mon-
itoring, with K ranging between 1 and 5. We
performed five independent runs using an admixture
model with allele frequencies correlated. Each run
consisted of 1,000,000 iterations (conducted three
times for each K value) with the first 100,000
iteractions discarded as burn-in.
In order to determine the most probable origin of
the individuals captured in different locations and
during the post-eradication monitoring, we used an
assignment method as implemented in GENECLASS
2.0 software (Piry et al. 2004). This method calculates
L. Ragionieri et al.
123
Author's personal copy
the likelihood of the multi-locus genotype of a given
individual in a set of pre-determined populations. We
chose an assignment threshold of 0.05 and obtained
the rejection probability by simulating 10,000 indi-
viduals from allelic frequencies. We used the Bayesian
method proposed by Rannala and Mountain (1997)
and simulation algorithm of Paetkau et al. (2004). The
mean probability values were estimated for each
individual and population.
We used the software Bottleneck 1.2.02 (Cornuet
and Luikart 1997) to determine if the four populations
of R. rattus underwent a recent bottleneck. This
software is based on the principle that populations
which have recently experienced a reduction in their
effective population size exhibit a corresponding
reduction of the allele numbers (k) and gene diversity
at polymorphic loci. Usually the number of alleles is
reduced faster than the gene diversity. Thus, in a
recently bottlenecked population, the observed gene
diversity is higher than the gene diversity expected at
equilibrium (Heq) which is computed from the
observed number of alleles (k), under the assumption
of a constant-size (equilibrium) population (Luikart
et al. 1998). For detecting if populations underwent
genetic bottlenecks we applied the heterozygosity
excess method of Luikart et al. (1998) using the Two-
phased model (TPM), with 70 % of single-step
mutations and 30 % of multi-step mutations, and the
Stepwise Mutation Model (SMM) as mutation models
with 10,000 iterations. We ran Bottleneck using two
different statistical tests, the Sign test and the Wilco-
xon sign-rank test (Cornuet and Luikart 1996; Luikart
and Cornuet 1997; Luikart 1997), both based on
10,000 replications. Bottlenecked populations are also
expected to exhibit a characteristic ‘mode shift’ in the
frequency distribution of alleles away from the
L-shaped distribution expected under mutation-drift
equilibrium (Luikart et al. 1998). Consequently,
BOTTLENECK was also used to generate a qualita-
tive descriptor of whether the observed allele frequen-
cies at each locus deviate from such a distribution.
Gene flow among populations
In order to estimate the gene flow among the four
populations, we used MIGRATE 3.2.14 (Beerli and
Felsenstein 1999). Input data were converted to
Migrate format using Microsatellites analyzer 4.05
(Dieringer and Schlo
¨tterer 2002). Migrate estimates
the mutation-scaled effective population size Theta
(h=xNel, where x is a multiplier depending on the
ployd phase x =4 for nuclear data, Ne is the effective
population size and mu is the mutation rate per site per
generation l) and the mutation-scaled migration rate
M(m/l, where m is the immigration rate and lthe
mutation rate), which is a measure of the importance
of immigration in bringing new variants into the
population. The effective number of immigrants per
generation was estimated by multiplying h9M (as the
equation N
e
m
ji
=h
i
9M
ji
). This analysis produces
values of h9M (4Nem for microsatellites) estimated
in each direction among the four populations with their
approximate 95 % confidence intervals (Beerli and
Felsenstein 2001). We ran MIGRATE three times
using a Singlestep Model with mutation rates esti-
mated for each locus, uniform prior distribution (h
distribution: minimum =0.0, maximum =20.0,
mean =10.0; M distribution: minimum =0.0, max-
imum =100.0, mean =50.0), starting parameters
based on Fst calculations, burn-in equal to 10,000
trees, and 10 replicates. Finally the overall number of
migrants per generation (Nem) was estimated by
summing hM in each direction and dividing by four for
microsatellites (Wright et al. 2005).
Results
Genetic diversity, population genetic structure
and bottlenecks
All loci, except one (D10Rat20), were informative,
and presented a relatively high level of polymorphism.
The locus D10Rat20 had a high number of null alleles
in all populations, and an excess of homozygotes,
according to MICROCHECKER; therefore, this locus
was removed from subsequent analysis and the overall
analyses were conducted using seven out of eight loci.
Similar problems for this locus were reported in other
studies (Abdelkrim al. 2010; King et al. 2011).
No significant linkage disequilibrium was recorded
across all populations. This was expected since the
microsatellite loci employed in this study are located
on different chromosomes (Jacob et al. 1995).
The mean number of alleles per locus was 10; the
populations of MOL and TAV had a similar number of
alleles, while the population of PIA had the lowest
number of alleles and the CCC population was the
A case of study from Sardinia, Italy
123
Author's personal copy
most polymorphic. In addition, the three populations
of MOL, TAV and CCC showed quite a high number
of private alleles, those alleles that are present in just
one population (8, 7 and 19 respectively), while the
population of PIA was the only population without any
private alleles.
The values of expected and observed heterozygosis
was higher in the populations of MOL, TAV and CCC,
and comparatively lower in the population of PIA
(Table 1). The populations of TAV, CCC and PIA
deviated from HWE, while the population of MOL
was the only one in HWE (Table 1). In order to test if
an excess of homozygotes was due to the presence of
null alleles at different loci, we employed the software
MICROCHECKER. An excess of homozygotes was
recorded for the population of PIA due to the locus
D19Mit2 and the presence of two loci monomorphic,
D7Rat13 and D11Mgh5. The population of TAV
presented three loci out of the HWE, D19Mit2 and
D11Mgh5 with an excess of homozygotes, and
D10Mit5 with an excess of heterozygotes. Similar
results were recorded with Fis index for the popula-
tions of PIA (in locus D19Mit2) and TAV (for locus
D19Mit2 and D11Mgh5), while no significant values
were recorded in the populations of MOL and CCC.
The Nei’s standard genetic distance, corrected for a
small sample size, produced similar values of pairwise
genetic divergence for the comparisons involving the
three populations of TAV, MOL and CCC (Table 2).
The smallest value of genetic divergence was recorded
between the populations of PIA and TAV, compared
to the other pairwise comparisons.
The AMOVA test recorded a high value of
population partitioning (Fst =0.328, P\0.001).
This was also evident from the pairwise Fst values
among the four populations (Table 2); all of the
populations were strongly differentiated from each
other. Similar results were also recorded with the exact
test of population differentiation (data not shown).
The results of the SAMOVA were in agreement
with those of the AMOVA; the number of population
groups that maximised the distribution of genetic
variation was K =4 (data not shown). The cluster
analysis conducted with the program STRUCTURE
recorded the presence of four groups of populations
(K =4), each corresponding to one of the four
analysed populations. Furthermore, in this analysis
we also included the genotypes of the two samples of
R. rattus collected on MOL during the post-
eradication monitoring. These two samples clustered
within the CCC population and not with the MOL
samples collected before the eradication campaign
(Fig. 2).
The results of the assignment analysis also sup-
ported the presence of four independent groups. All
the individuals collected from the populations of TAV,
MOL and CCC were assigned to their own population
(Fig. 3). A small percentage of PIA individuals were
assigned to the CCC population, and notably the two
rats collected on MOL during the post-eradication
monitoring were unambiguously assigned to the
population of CCC (Fig. 3).
The analysis conducted with Bottleneck recorded an
heterozygosity excess in five out of the seven loci for
the population of PIA (Sign test under TPM,
P=0.036; Wilcoxon test under TPM, P=0.015),
while the remaining two loci were monomorphic. In the
populations of TAV and MOL, six loci showed a
heterozygosity deficiency, and one locus a heterozy-
gosity excess (TAV: Sign test under SMM, P=0.02;
Wilcoxon test under the SMM, P=0.039. MOL: Sign
test under SMM, P=0.021; Wilcoxon test under the
SMM, P=0.039). Finally, the population of CCC
showed a heterozygosity deficiency (Wilcoxon TEST
under the SMM, P=0.019). These results accord with
the allele frequency distribution, which was a normal
L-shaped distribution in the three populations of TAV,
MOL and CCC, while a shifted distribution of allele
frequencies, typical of populations which experienced a
bottleneck, was recorded for the population of PIA.
Gene flow
The four populations of R. rattus had low values of
effective population size, with the PIA population and
CCC population having the smallest and largest values
respectively (Table 3). The gene flow among the four
populations was very weak, without any evidence of
asymmetric gene flow. Regarding the overall migra-
tion, all the values were weak and without clear
evidence of any preferential migration channel for
gene flow between pairs of populations (Table 3).
Discussion
The four populations of R. rattus collected in the
Marine Protected Area Tavolara Punta Coda Cavallo
L. Ragionieri et al.
123
Author's personal copy
were genetically differentiated, based on summary
statistics (AMOVA and SAMOVA) and clustering and
assignment methods. Based on these genetic analyses,
we thus recorded four independent eradication units
with extremely reduced or absent gene flows.
Otherwise, a small fraction of individuals captured
on PIA were assigned to the population of CCC,
supporting the theory of a recent invasion of this island
by a few individuals from the Sardinian mainland.
Although these two populations are separated by a
considerable distance (*1.1 km), there are many
small islets between the Sardinian mainland and PIA,
such as Isola dei Cavalli, which may have acted as a
bridge for sporadic migration events. In Isola dei
Cavalli, for instance, rats and mice were present, and
they were eradicated at the same time of PIA in a
successive eradication campaign conducted between
December 2009 and January 2010. Isola dei Cavalli is
distant around 200 m from CCC and around 300 m
from PIA and these distances can be potentially
covered by black rats (Abdelkrim et al. 2009; Russell
et al. 2009a;2010; Savidge et al. 2012). Moreover, the
Table 1 Rattus rattus collection localities and summary statistics
Locus GPS Piana (PIA) Tavolara (TAV)
40°53017.1700N; 9°3907.0800E40°53036.6400N; 9°4102.5600E
Range Na Ar Ho He Fis Na Ar Ho He Fis
D10Rat20 114–128 – – – – – – – – – –
D5Rat83 169–195 3 3 0.6 0.559 -0.08 4 3.97 0.567 0.555 -0.02
D7Rat13 153–189 1 1 0 0 – 8 6.83 0.833 0.727 -0.15
D9Rat13 112–130 2 2 0.3333 0.2825 -0.18 2 1.57 0.033 0.033 0.00
D11Mgh5 234–286 1 1 0 0 – 4 3.58 0.172 0.533 0.68
D16Rat81 146–174 3 3 0.767 0.658 -0.17 5 4.56 0.766 0.673 -0.14
D19Mit2 195–235 5 4.54 0.5 0.704 0.29 7 6.05 0.466 0.74 0.37
D10Mit5 185–195 2 2 0.2667 0.2825 0.06 3 3 0.733 0.616 20.19
Mean 2.4 2.4 0.352 0.355 0.01 4.7 4.22 0.51 0.554 0.08
Locus GPS Molara (MOL) Capo Coda Cavallo (CCC)
40°52010.4900N; 9°42054.7800E40°52045.1400N; 9°38059.6300E
Range Na Ar Ho He Fis Na Ar Ho He Fis
D10Rat20 114–128 – – – – – – – – – –
D5Rat83 169–195 4 3.58 0.552 0.558 0.01 7 6.9 0.833 0.804 -0.04
D7Rat13 153–189 6 5.58 0.69 0.796 0.14 8 7.38 0.75 0.702 -0.07
D9Rat13 112–130 4 3.25 0.172 0.165 -0.05 7 6.34 0.875 0.761 -0.15
D11Mgh5 234–286 4 3.56 0.483 0.477 -0.01 10 8.94 0.667 0.824 0.19
D16Rat81 146–174 4 3.97 0.552 0.554 0 8 7.33 0.833 0.7332 -0.14
D19Mit2 195–235 8 7.43 0.62 0.783 0.21 7 6.82 0.75 0.812 0.08
D10Mit5 185–195 4 3.58 0.345 0.482 0.29 3 3 0.294 0.4332 0.33
Mean 4.9 4.42 0.488 0.545 0.11 7.1 6.67 0.715 0.724 0.01
For each population: GPS coordinates, the size range in base pairs of each locus (range), total number of alleles for each locus (Na),
allelic richness per locus (Ar), observed heterozygosity (Ho), expected unbiased heterozygosity (He), within population inbreeding
coefficient (F
is
). In bold significant of Pvalue of departure from the Hardy–Weinberg equilibrium
Table 2 Pairwise comparisons of genetic differentiation,
estimated from the pairwise Fst values (under the diagonals;
significant values are in bold; Pthreshold \0.05), and Nei’s
standard genetic distance of the four populations of Rattus
rattus (over the diagonal)
PIA TAV MOL CAV
PIA – 0.398 0.572 0.495
TAV 0.34 – 0.514 0.553
MOL 0.44 0.32 – 0.483
CAV 0.34 0.26 0.25 –
A case of study from Sardinia, Italy
123
Author's personal copy
fact that PIA and CCC populations were genetically
independent could be a consequence of the strong
genetic reduction usually associated with recent
founding events, which may increase the genetic
divergence from the source population (Abdelkrim
et al. 2005b). This hypothesis is also supported by the
relatively small heterozygosity value, from the
reduced number of alleles (two out of seven loci are
monomorphic and no private alleles) and the departure
from the HWE recorded in PIA population, typical of
recent founder events.
The allelic richness, heterozygosity levels and the
number of private alleles recorded in the two popu-
lations of MOL and TAV were quite similar. The only
difference recorded being that the population of MOL
was at the HWE, while the population of TAV was out
of the equilibrium, due to an excess of homozygotes at
two loci. This deviation from HWE could be ascribed
to the collection of individuals which were not
representative of the overall population. The popula-
tion of TAV was, in fact, exclusively collected in an
area close to the touristic harbour, because access on
the island is restricted to this area due to the presence
of a NATO military base.
The population of CCC has comparatively higher
levels of genetic variability than the other populations,
twice the number of private alleles and a greater
population size; however, the CCC population was not
in the HWE, with an overall excess of homozygotes.
This could be due to the presence in the mainland
population of CCC of sub (family) groups (Wahlund
effect) or inbreeding effect.
An insular colonization generally involves few
individuals and produces effects similar to a genetic
bottleneck. After a genetic bottleneck, the observed
heterozygosity may exceed the expected heterozygos-
ity as a consequence of a faster reduction of allelic
diversity than of heterozygosity (Cornuet and Luikart
1996). At the same time, a modal shift in the
distribution of alleles is generally observed together
with a relative deficit of rare alleles (Luikart et al.
1998). This is essentially what we observed in the
population of PIA. An heterozygosity excess was
observed in five out of seven loci, in addition to a shift
in the distribution of alleles, typical of populations
which have recently undergone a bottleneck or a
founding event. On the contrary, in the other three
sampled populations, no clear signal of a recent
founder or bottleneck event was evident, based on the
heterozygosity indexes and on the allele frequency
distribution. The three populations of MOL, TAV and
CCC appear to have been founded sufficiently long
Fig. 2 Cluster analysis: each individual is represented by a
vertical bar, with K colours, where K is the number of
predefined populations and the length of the segments corre-
spond to the individual membership to each population. The run
with the highest posterior probability corresponds to K =4.
Black vertical bars delineate predefined populations (Group 1,
PIA in green; Group 2, Tavolara, in red; Group 3 MOL, in
yellow; Group 4, two MOL individuals collected after the
eradication campaign, in blue; Group 5, CCC, in blue)
Fig. 3 Assignment test: mean probability values of individuals
assigned per population (PIA, Piana; TAV, Tavolara; MOL,
Molara; Capo Coda Cavallo, CCC)
L. Ragionieri et al.
123
Author's personal copy
ago to become genetically independent, and without
any further gene flow connecting these three
populations.
In a recent study conducted on a western Mediter-
ranean insular system, Lavezzi island and its sur-
rounding islets, Abdelkrim et al. (2009) reported
similar levels of genetic diversity in populations
resulting from an ancient colonization event to those
recorded in the two rat populations of MOL and TAV,
supporting the idea that these two rat populations were
well established in the respective islands. Anyway is
not possible to evaluate how old were the colonization
events on MOL and TAV as far rats seems to be able to
established populations with demography and popu-
lation genetic structure similar to longer established
populations, without short-time genetic consequence
(Russell et al. 2009b). The population of CCC, indeed,
had considerably higher values of genetic diversity,
suggesting that this population is considerably larger
than any other rat population collected in the Medi-
terranean Sea, or at least is part of a larger population
present on Sardinia mainland. Finally, the genetic
diversity recorded in PIA is much less than in all the
other populations, with values similar to those
recorded from Abdelkrim et al. (2009) on the small
islets surrounding Lavezzi Island. In addition, all the
alleles present in PIA were also found in the other
populations without any private alleles. Even here, all
these evidences strongly support a recent origin of PIA
probably from the surrounding populations such as
CCC and TAV.
Based on our analysis, it is not possible to assess if
the population of CCC is the source population of the
MOL and TAV populations, or if the rat populations
founded on TAV and MOL islands are the product of a
single or multiple invasion events. Rats may have
reached the islands swimming or, more likely, through
secondary vectors. However, it is also possible that the
strong intra-specific behavioral competition, typical of
rats, may have limited the survival and establishment
of new settlers (Granjon and Ceylan 1989). Moreover,
the two rats collected during the post-eradication
monitoring on MOL clustered with the Sardinian
mainland population (CCC), and not with the pre-
eradication MOL population. Although rats are capa-
ble swimmers (e.g. R. norvegicus, Russell et al. 2010),
and in some areas are considered one of the most likely
invaders of offshore islands (e.g. New Zealand,
Russell et al. 2005), the minimum geographic distance
between the Sardinia mainland and the island of
Molara (*1.5 km) exceeds the known swimming
capability of the black rat (Russell et al. 2010; Calmet
et al. 2001). Thus, according to our genetic analyses
the eradication campaign on MOL island appears to
have been successful, and it is highly probable that the
rapid re-invasion of R. rattus of MOL was driven by
tourist or private boats arriving from Sardinia.
Although the populations of R. rattus analyzed in
this study appear to be genetically independent, these
results must be interpreted with caution for two
reasons. Firstly, in the past rats were able to invade
and establish permanent populations on all of the
major neighbor islands to MOL. Secondly, according
to our data, the Sardinia population is a possible source
of re-invasion for the islands of MOL (rats were
collected during post-eradication monitoring) and PIA
(cluster and assignment analysis), and consequently
for the entire study area. These evidences suggest that
there could be a rat exchange between the Marine
Protected Area Tavolara Capo Coda Cavallo and the
Table 3 Effective population size and gene flow among the four populations of R rattus, in the Marine Protected Area of Capo Coda
Cavallo, using MIGRATE
Theta Pop 1 Pop 2 N
e
m
12
N
e
m
21
N
e
m
PIA 0.14 (0.00–0.37) PIA TAV 0.25 (0.00–1.26) 0.23 (0.00-1.39) 0.12
TAV 0.23 (0.00–0.52) PIA MOL 0.29 (0.00–1.60) 0.30 (0.00–1.34) 0.15
MOL 0.21 (0.00–0.49) PIA CCC 0.25 (0.00–1.81) 0.06 (0.00–0.53) 0.07
CCC 0.30 (0.00–0.61) TAV MOL 0.33 (0.00–1.77) 0.30 (0.00–1.60) 0.16
TAV CCC 0.30 (0.00–1.34) 0.33 (0.00–1.77) 0.19
MOL CCC 0.30 (0.00–1.45) 0.18 (0.00–0.93) 0.12
Theta effective population size (2.5–97.5 % confidence intervals), N
e
m
12
mean of migrants from population 1 (Pop 1) to population 2
(Pop 2), and corresponding confidence interval (2.5–97.5 %), N
e
m
21
are mean migrants in the opposite direction from Pop 2 to Pop 1
and N
e
m is the effective number of migrants exchanged per generation
A case of study from Sardinia, Italy
123
Author's personal copy
Sardinia mainland, probably driven by humans. The
eradication campaign of MOL appears to have been
successful thus far; an increase in the reproductive
success of Yelkouan shearwater has been observed
(Sposimo et al. 2012). Therefore, to maintain the
success of this campaign, post eradication monitoring
should be coupled with more strict bio-security
measures (see Russell et al. 2008), and further
eradication campaigns on other islands within the
MPA.
Outlook
In the last decade the use of genetic information
acquired from hypervariable autosomal markers, such
as microsatellites, became a fundamental prerequisite
of many eradication campaigns of pest species.
Notwithstanding numerous studies successfully estab-
lished eradication units, these studies also highlighted
some weaknesses of this approach. In first instance, as
already reported from Savidge et al. (2012), genetic
methods may not always be valid if the populations did
not reach the equilibrium conditions for such analysis.
The identification of source population, as in this
study, remains a difficult task as well as the ability of
discriminating between eradication failure and recol-
onization events (Russell et al. 2009a; Savidge et al.
2012): this is especially true for mainland populations,
with genetically continuous populations, where few
local variations may be due to reduced gene flows or to
the presence of family groups (Abdelkrim et al. 2010;
King et al. 2011). In addition if low local variation is
recorded, clustering methods completely fail to iden-
tify possible groups of populations (Abdelkrim et al.
2010).
Another critical point is that the true level of gene
flow may be underestimated owing to behavioural
mechanisms of rats, such as strong intra-specific
competition and migrant exclusion (Granjon and
Cheylan 1989), in such a situation genetic methods
may not be able to record the true gene flow and once
the eradicated population is removed, new colonists
from interconnected populations may spread rapidly
(Abdelkrim et al. 2007).
At the light of the above-mentioned shortcomings,
we auspicate that future eradication campaigns will
integrate pre-eradication genetic investigations with
eco-ethological studies in order to shed new light on
the population structure of pests as well as on
behavioural mechanisms which can limit the reliabil-
ity of the genetic tool.
Acknowledgments We are grateful to Massimo Putzu for
help with rat sampling. We also thank Jenny Booth for the
accurate linguistic revision of the manuscript. We thank two
anonymous reviewers for their helpful comments. This research
was partially supported by Fondi d’Ateneo to M. Vannini (ex
60 % University of Florence).
References
Abdelkrim J, Pascal M, Samadi S (2005a) Island colonization
and founder effects: the invasion of the Guadeloupe islands
by ship rats (Rattus rattus). Mol Ecol 14:2923–2931
Abdelkrim J, Pascal M, Calmet C, Samadi S (2005b) Importance
of assessing population genetic structure prior to eradica-
tion of invasive species: examples from insular Rattus
norvegicus populations. Cons Bio 19:1509–1518
Abdelkrim J, Pascal M, Samadi S (2007) Establishing causes of
eradication failure based on genetics: case study of ship rat
eradication in ste. Anne Archipel Cons Bio 3:719–730
Abdelkrim J, Pascal M, Samadi S (2009) Genetic structure and
functioning of alien ship rat populations from a corsican
micro-insular complex. Biol Invasion 11:473–482
Abdelkrim J, Byrom AE, Gemmell NJ (2010) Fine-scale genetic
structure of mainland invasive Rattus rattus populations:
implications for restoration of forested conservation areas
in New Zealand. Conserv Genet 11:1953–1964
Amaral J, Almeida S, Sequeira M, Neves V (2010) Black rat
Rattus rattus eradication by trapping allows recovery of
breeding roseate tern Sterna dougallii and common tern
S.hirundo populations on Feno Islet, the Azores. Portugal.
Cons Evidence 7:16–20
Baccetti N, Capizzi D, Corbi F, Massa B, Nissardi S, Spano G,
Sposimo P (2009) Breeding shearwater on Italian islands:
population size, island selection and co-existence with their
main alien predator. Riv Ital Ornitol 78:83–99
Beerli P, Felsenstein J (1999) Maximum-Likelihood estimation
of migration rates and effective population numbers in two
populations using a coalescent approach. Genetics
152:763–773
Beerli P, Felsenstein J (2001) Maximum likelihood estimation
of a migration matrix and effective population sizes in n
subpopulations by using a coalescent approach. Proc Natl
Acad Sci USA 98:4563–4568
Calmet C, Pascal M, Samadi S (2001) Is it worth eradicating the
invasive pest Rattus norvegicus from Molene archipelago?
Genetic structure as a decision-making tool. Biodivers
Conserv 10:911–928
Capizzi D, Baccetti N, Sposimo P (2010) Prioritizing rat erad-
ication on islands by cost and effectiveness to protect
nesting seabirds. Biol Conserv 143:1716–1727
Carvalho DC, Oliveira DAA, Santos JE, Teske P, Beheregaray
LB, Schneider H, Sampaio I (2009) Genetic characteriza-
tion of native and introduced populations of the neotropical
cichlid genus Cichla in Brazil. Genet Mol Biol 32:601–607
L. Ragionieri et al.
123
Author's personal copy
Cornuet JM, Luikart G (1996) Description and power analysis of
two tests for detecting recent population bottlenecks from
allele frequency data. Genetics 144:2001–2014
Cornuet JM, Luikart G (1997) Description and power analysis of
two tests for detecting recent population bottlenecks from
allele frequency data. Genetics 144:2001–2014
Courchamp F, Chapuis JL, Pascal M (2003) Mammal invaders
on islands: impact, control and control impact. Biol Rev
78:347–383
Dieringer D, Schlo
¨tterer C (2002) Microsatellite analyser
(MSA): a platform independent analysis tool for large data
set. Mol Ecol Notes 3:167–169
Dupanloup I, Schneider S, Excoffier L (2002) A simulated
annealing approach to define the genetic structure of pop-
ulations. Mol Ecol 11:2571–2581
Excoffier L, Smouse PE, Quattro JM (1992) Analysis of
molecular variance inferred from metric distances among
DNA haplotypes: application to human mitochondrial
DNA restriction data. Genetics 131:479–491
Excoffier L, Laval LG, Schneider S (2005) Arlequin, version 3:
an integrated software package for population genetics data
analysis. Evol Bioinform 1:47–50
Goudet J (1995) FSTAT, version 1.2. A computer program to
calculate F-statistics. J Hered 86:485–486
Graham MF, Veitch CR (2002) Changes in birds numbers on
Tiritiri Matangi Island, New Zealand, over the period of rat
eradication. In Veitch CR, Clout MN (eds) Turning the
tide: the eradication of invasive species. Occas Pap of the
IUCN Species Surviv Comm 27:120–123
Granjon L, Ceylan G(1989) Le sort de rats noirs (Rattus rattus)
introduits sur une ı
ˆle, re
´ve
´le
´par radio-tracking. C R Acad
Sci Paris 571–575
Jacob HJ, Brown DM, Bunker RK et al (1995) A genetic linkage
map of the laboratory rat, Rattus norvegicus. Nat Genet
9:63–69
Kerbiriou C, Pascal M, Le Viol I, Garoche J (2004) Conse-
quence sur l’avifaune terrestre de l’Ile de Trielen (reserve
naturelle d’Iroise; Bretagne) de l’eradication du surmulot
(Rattus norvegicus). Revue d’Ecologie (Terre and Vie)
59:319–329 (in French)
King CM, Innes JG, Gleeson D, Fitzgerald N, Winstanley T,
O’Brien B, Bridgman L, Cox N (2011) Reinvasion by ship
rats (Rattus rattus) of forest fragments after eradication.
Biol Invasions 13:2391–2408
Lowe S, Browne M, Boudjelas S (2000) 100 of the world’s worst
invasive alien species. A selection from the global invasive
species database, Invasive Species Specialist Group
Luikart G (1997) Usefulness of molecular markers for detecting
population bottlenecks and monitoring genetic change.
University of Montana, Dissertation
Luikart G, Cornuet JM (1997) Empirical evaluation of a test for
identifying recently bottlenecked populations from allele
frequency data. Conserv Biol 12(1):228–237
Luikart G, Allendorf FW, Cornuet JM, Sherwin WB (1998)
Distortion of allele frequency distributions provides a test
for recent population bottlenecks. J Hered 89:238–247
Nei M (1978) Estimation of average heterozygosity and genetic
distance from a small number of individuals. Genetics
89(3):583–590
Paetkau D, Slade R, Burden M, Estoup A (2004) Genetic
assignment methods for the direct, real-time estimation of
migration rate: a simulation-based exploration of accuracy
and power. Mol Ecol 13(1):55–65
Parkes J, Fisher P, Forrester G (2011) Diagnosing the cause of
failure to eradicate introduced rodents on islands: bro-
difacoum versus diphacinone and method of bait delivery.
Cons Evidence 8:100–106
Pascal M, Siorat F, Lorvelec O, Ye
´sou P, Simberloff D (2005) A
pleasing consequence of Norway rat eradication: two
shrew species recover. Divers Distrib 11:193–198
Pinceel J, Jordaens K, Van Houtte N, Bernon G, Backeljau T
(2005) Population genetics and identity of an introduced
terrestrial slug: Arion subfuscus s.l. in the North-East USA
(Gastropoda, Pulmonata, Arionidae). Genetica 125:
155–171
Piry S, Alapetite A, Cornuet JM, Paetkau D, Baudouin L, Estoup
A (2004) GeneClass2: a software for genetic assignment
and first generation migrants detection. J Hered
95:536–539
Pritchard JK, Stephens M, Donnelly P (2000) Inference of
population structure using multilocus genotype data.
Genetics 155:945–959
Rannala B, Mountain JL (1997) Detecting immigration by using
multilocus genotypes. Proc Natl Acad Sci USA 94:
9197–9201
Raymond M, Rousset F (1995) GENEPOP ver. 1.2: population
genetics software for exact tests and ecumenicism. J Hered
4:248–249
Robertson BC, Gemmell NJ (2004) Defining eradication units in
pest control programmes. J Appl Ecol 41:1032–1041
Russell, JC, Towns DR, Clout MN (2008): Review of rat
invasion biology: implications for island biosecurity. Sci-
ence for Conservation 286. Department of Conservation,
Wellington 53 p
Russell JC, Towns DR, Anderson SH, Clout MN (2005) Inter-
cepting the first rat ashore. Nature 437:1107
Russell JC, Mackay JWB, Abdelkrim J (2009a) Insular pest
control within a metapopulation context. Biol Conserv
142:1404–1410
Russell JC, Abdelkrim J, Fewster RM (2009b) Early colonisa-
tion population structure of a Norway rat island invasion.
Biol Invasions 11:1557–1567
Russell JC, Miller SD, Harper GA, MacInnes HE, Wylie MJ,
Fewster RM (2010) Survivors or reinvaders? Using genetic
assignment to identify invasive pests following eradica-
tion. Biol Invasions 12:1747–1757
Savidge JA, Hopken MW, Witmer GW, Jojola SM, Pierce JJ,
Burke PW, Piaggio AJ (2012) Genetic evaluation of an
attempted Rattus rattus eradication on Congo Cay, U.S.
Virgin Islands, identifies importance of eradication units.
Biol Invasions 14:2343–2354
Sposimo P, Spano G, Navone A, Fratini S, Ragionieri L, Putzu
M, Capizzi D, Baccetti N (2012) Rodent eradication on
Molara Island and surrounding islets (NE Sardinia): from
success to the riddle of reinvasion. Aliens 32:33–38
Thorsen M, Shorten R, Lucking P, Lucking V (2000) Norway
rats (Rattus norvegicus) on Fregate Island, Seychelles: the
invasion; subsequent eradication attempts and implications
for the island’s fauna. Biol Conserv 96:133–138
Towns DR, Daugherty CH, Cree A (2001) Raising the prospect
for a forgotten fauna: a review of 10 years of conservation
effort for New Zealand reptiles. Biol Conserv 99:3–16
A case of study from Sardinia, Italy
123
Author's personal copy
Van Oosterhout C, Hutchinson WF, Wills DPM, Shipley P
(2004) MICROCHECKER: software for identifying and
correcting genotyping errors in microsatellite data. Mol
Ecol Notes 4:535–538
Veitch CR, Clout MN, Towns DR (2011) Island Invasives:
eradication and management. In Proceedings of the inter-
national conference on island invasives. Gland, Switzer-
land, IUCN and Auckland, New Zealand, CBB.
xii ?542pp
Wright T, Rodriguez A, Fleischer R (2005) Vocal dialects, sex-
biased dispersal, and microsatellite population structure in
the parrot Amazona auropalliata. Mol Ecol 14:1197–1205
Zenatello M, Spano G, Baccetti N, Zucca C, Navone A, Putzu
M, Azara C, Trainito E, Ugo M, Phillips R (2011) Move-
ments and moving population estimates of Yelkouan
shearwaters at Tavolara, Sardinia. In 13th MEDMARAVIS
Pan-Mediterranean symposium, Alghero (Sardinia) 14–17
(in press)
L. Ragionieri et al.
123
Author's personal copy
