ArticlePDF Available

Abstract

Genetic diversity and differentiation were assessed in 12 populations of roe deer, Capreolus capreolus, from Italy, through examination of restriction fragment length polymorphism of two segments in the mitochondrial genome, the D‐loop and NADH dehydrogenase 1, and analysis of 13 microsatellite loci. Both methods yielded concordant results and provided evidence for the existence of two genetically distinct lineages of roe deer in the Italian peninsula. One lineage occurs in populations of the Alpine arc, whilst the other is found in those of central‐southern Italy, where the existence of subspecies C. c. italicus had been previously proposed. This southern lineage could have a more ancient origin or, alternatively, diverged as a consequence of movements of populations southwards during the Late Pleistocene. The unexpected rediscovery of dense populations of C. c. italicus in southern Tuscany marks them as the most suitable source of roe deer for reintroductions into southern Italy, a very large area where presently the Italian roe deer is nearly extinct. A Bayesian approach to microsatellite data allowed a finer resolution of population structure, indicating that some populations in central Italy, as well as in the western Alps, are admixed, and share ancestry partly in non Italian gene pools, suggesting that human manipulation has greatly affected the natural genetic structure of populations. A palaeontological perspective of the former presence of roe deer in Italy and implications for the management and conservation of C. c. italicus are provided.
This article was downloaded by: [203.86.16.249]
On: 25 March 2014, At: 08:40
Publisher: Taylor & Francis
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer
House, 37-41 Mortimer Street, London W1T 3JH, UK
Italian Journal of Zoology
Publication details, including instructions for authors and subscription information:
http://www.tandfonline.com/loi/tizo20
The rediscovery of the Italian roe deer: Genetic
differentiation and management implications
Rita Lorenzini
a
, Sandro Lovari
b
& Marco Masseti
c
a
Istituto Zooprofilattico Sperimentale dell'Abruzzo e del Molise “G. Caporale” ,
Campo Boario, Teramo, I64100, Italy E-mail:
b
Sezione di Etologia, Ecologia Comportamentale e Gestione della Fauna,
Dipartimento di Scienze Ambientali , Università di Siena , Via P. A. Mattioli 4, Siena,
I53100, Italy E-mail:
c
Dipartimento di Biologia Animale e Genetica, Laboratori di Antropologia ,
Università di Firenze , Via del Proconsolo 12, Firenze, I50122, Italy E-mail:
Published online: 28 Jan 2009.
To cite this article: Rita Lorenzini , Sandro Lovari & Marco Masseti (2002) The rediscovery of the Italian roe
deer: Genetic differentiation and management implications, Italian Journal of Zoology, 69:4, 367-379, DOI:
10.1080/11250000209356482
To link to this article: http://dx.doi.org/10.1080/11250000209356482
PLEASE SCROLL DOWN FOR ARTICLE
Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”)
contained in the publications on our platform. However, Taylor & Francis, our agents, and our licensors
make no representations or warranties whatsoever as to the accuracy, completeness, or suitability
for any purpose of the Content. Any opinions and views expressed in this publication are the opinions
and views of the authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of
the Content should not be relied upon and should be independently verified with primary sources of
information. Taylor and Francis shall not be liable for any losses, actions, claims, proceedings, demands,
costs, expenses, damages, and other liabilities whatsoever or howsoever caused arising directly or
indirectly in connection with, in relation to or arising out of the use of the Content.
This article may be used for research, teaching, and private study purposes. Any substantial or
systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution
in any form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at
http://www.tandfonline.com/page/terms-and-conditions
Ital.
J. Zool., 69- 367-379 (2002)
The rediscovery of the Italian roe deer:
genetic differentiation and management
implications
RITA LORENZINI
Istituto Zooprofilattico Sperimentale dell'Abruzzo e del Molise
"G.
Caporale", Campo Boario, I-64100 Teramo (Italy)
E-mail: r.lorenzini@izs.it
SANDRO LOVARI
Sezione di Etologia, Ecologia Comportamentale e Gestione della Fauna,
Dipartimento di Scienze Ambientali, Università di Siena,
Via P. A. Mattioli 4, I-53100 Siena (Italy)
E-mail: lovari@unisi.it
MARCO MASSETI
Dipartimento di Biologia Animale e Genetica, Laboratori di Antropologia,
Università di Firenze, Via del Proconsolo 12, I-50122 Firenze (Italy)
E-mail: marco.masseti@unifi.it
ABSTRACT
Genetic diversity and differentiation were assessed in 12 popu-
lations of roe deer, Capreolus capreolus, from Italy, through ex-
amination of restriction fragment length polymorphism of two
segments in the mitochondrial genome, the D-loop and NADH
dehydrogenase 1, and analysis of 13 microsatellite loci. Both
methods yielded concordant results and provided evidence for
the existence of two genetically distinct lineages of roe deer in
the Italian peninsula. One lineage occurs in populations of the
Alpine arc, whilst the other is found in those of central-southern
Italy, where the existence of subspecies C. c. italicus had been
previously proposed. This southern lineage could have a more
ancient origin or, alternatively, diverged as a consequence of
movements of populations southwards during the Late Pleis-
tocene. The unexpected rediscovery of dense populations of C. c.
italicus in southern Tuscany marks them as the most suitable
source of roe deer for reintroductions into southern Italy, a very
large area where presently the Italian roe deer is nearly extinct. A
Bayesian approach to microsatellite data allowed a finer resolu-
tion of population structure, indicating that some populations in
central Italy, as well as in the western Alps, are admixed, and
share ancestry partly in non Italian gene pools, suggesting that
human manipulation has greatly affected the natural genetic struc-
ture of populations. A palaeontological perspective of the former
presence of roe deer in Italy and implications for the manage-
ment and conservation of C. c. italicus are provided.
KEY WORDS: Capreolus capreolus italicus - Mitochondrial DNA
- RFLPs - Microsatellites - Restocking - Fossil records.
ACKNOWLEDGEMENTS
We express our gratitude to M. Apollonio, E. Bruno, M. Cal-
darella, C. M. Caló, L. Cimino, N. Franconi, M. Gioiosa, A. Giu-
liani, V. Mazzarone, R. Mazzoni della Stella, P. G. Meneguz, and E.
Randi for the provision of samples and information on the origin
of each animal. We thank A. Polci for assistance in the lab work,
and F. Frati for his comments on our manuscript. G. Bianucci, P.
Boscato, W. Landini and P. Mazza greatly helped with information
on palaeontological aspects. Our work has been endorsed and/or
economically supported by the Italian Ministry of the Environ-
ment (Conservation of Nature Bureau).
(Received 2 July 2002 - Accepted 24 July 2002)
INTRODUCTION
The roe deer Capreolus capreolus (L., 1758) is the
most abundant and widespread deer in Europe, with
one presumably monotypic species (Danilkin, 1996).
Conversely, two subspecies have been proposed, differ-
ent from the north-central European one C. c capreo-
lus:
C. c garganta Meunier, 1983, in central Spain (Meu-
nier, 1983), and C. c italicus Festa, 1925, in southern
Italy (Festa, 1925). However, the differences in morpho-
logic traits observed in C. c. garganta and C. c. italicus
are small and fall within the range of variation of the
species (Paaver, 1965). On the other hand, no genetic
data supporting the existence of subspecies are present-
ly available. Previous genetic studies based on classical
biochemical markers provided evidence for significant
divergence between roe deer of the Alpine range and
some populations of central Italy (Lorenzini et al.,
1993),
but failed to uncover fixed differences, thus not
allowing sound conclusions to be made about the sys-
tematic status of the populations studied. Randi et al.
(1998) found only minor differences in the sequence of
the mitochondrial DNA control region between the sup-
posed C. c italicus from the fenced Castelporziano Re-
serve (Rome), the same population studied by E. Festa
(loc.
cit.), and roe deer from some areas in the Italian
Alps.
The lack of variability at mitochondrial level re-
vealed for the Castelporziano nucleus was ascribed to
demographic events, such as bottlenecks, population
fragmentation and prolonged genetic isolation.
The central-southern areas of Italy harbour some
species of medium/large mammals which differ sub-
stantially from their northern counterparts, showing
high genetic divergence, i.e., the Italian hare Lepus cor-
sicanus (Pierpaoli et al., 1999), the Apennine chamois
Rupicapra pyrenaica (Nascetti et al., 1985; Lorenzini &
Fico,
unpubl. data), the red fox Vulpes vulpes, along
with the Roman mole Talpa romana and the snow vole
Microtus nivalis (e.g. Frati et al., 1998), as well as other
small mammals (Bilton et al., 1998). Therefore, the pos-
sibility exists that a southern roe deer could have been
originally present in the south of the peninsula as a
relict metapopulation, divergent from northern roe deer
and possibly of more ancient origin.
In terms of morphometric characteristics, forms very
close to those of the extant one have been documented
in western Europe from as far back as the early part of
the Middle Pleistocene (Danilkin, 1996). Kurten (1968)
noted that all roe deer of the Middle Pleistocene are al-
most identical to C. capreolus, differing from the exist-
ing form predominantly in their larger size. The dimen-
sions of the largest fossil remains in Europe (Boessneck,
1956) fall within the morphologic variability accepted
for C. capreolus (Paaver, 1965), significantly smaller
than those of the fossils of the Asiatic roe deer
C.
pygar-
gus Pallas 1771 (Kosintsev, 1981). During the Late
Glacial the roe deer was frequently reported throughout
the Italian distribution range, but almost always poorly
Downloaded by [203.86.16.249] at 08:40 25 March 2014
368
R. LORENZINI, S. LOVARI, M. MASSETI
represented (Masseti et al., 1995). The limited presence
of roe deer in many anthropozoic sites could be as-
cribed not so much to environmental reasons, but to
the casual manner by which it was hunted because of
aspects mainly related to the behaviour and ecology of
this cervid (Tagliacozzo et al., 1989; Villari, 1995). On
the contrary, there is a significant quantity of roe deer
remains in the fossil context of Vado all'Arancio, south-
ern Tuscany. In this site, the roe deer figures as the sec-
ond most frequently found ungulate, revealing dimen-
sions consistently greater than those of the existing roe
deer from various areas of Tuscany, which were used
for comparison (Boscato, 1996).
The mitochondrial DNA (mtDNA) molecule is an ex-
cellent marker to uncover genetic divergence between
populations, detect population-specific markers, and in-
fer phylogeographic structure (Avise et al., 1987). On
the other hand, the study of mtDNA polymorphism
yields a female-biased description of population struc-
ture.
Data on nuclear biparental markers, such as the
highly variable microsatellites, provide a more complete
picture of population subdivision, and can be used on a
finer scale to assess genetic relationships among closely
related populations, and identify the individuals by mul-
tilocus genotyping (see Goldstein & Schlotterer, 2000,
for a review).
In the present survey, we have analysed variation at
both mtDNA and microsatellite loci in roe deer from
twelve localities in Italy, to examine the extent of genet-
ic divergence among populations. In particular, we test-
ed the hypothesis that roe deer in the Alpine region are
genetically distinct from those of central-southern Italy,
where the existence of C. c. italiens was suggested by
Festa (1925) for Castelporziano (Rome) and by von
Lehmann (1973) for northern Calabria, on the basis of
morphological, qualitative characters. Results on current
population structure will be discussed, taking into ac-
count the palaeontological information on the presence
of roe deer in Italy. Considerations and suggestions on
management and conservation strategies for this cervid
in Italy will also be provided.
MATERIALS AND METHODS
Sampling and DNA isolation
A total of 189 roe deer from 12 sampling sites (hereafter indi-
cated as populations) across Italy were gathered between 1999
and 2001 (Fig. 1). Tissue samples of roe deer from the eastern
Alps (Val Rendena), western Alps (Val di Susa), and the provinces
of Siena, Grosseto (Tuscany) and Pesaro (Marches) were collected
during the hunting season. Samples from Gargano and Pollino
(Orsomarso Mountains) National Parks, Maremma Regional Park
and the Castelporziano Presidential Estate (province of Rome)
were obtained from roe deer that had died accidentally or were
collected from live-captured animals. Total genomic DNA was iso-
lated from either skeletal muscle preserved in 75% ethanol, or
blood collected in EDTA-coated tubes, by standard methods
(Sambrook & Rüssel, 2001), or alternatively, by using the FastPrep
apparatus and FastDNA extraction kit for tissues (BIO 101). Only
hair samples for one and five specimens from Gargano and
Fig. 1 - Collection localities of roe deer samples: 1, REN (Val Ren-
dena, eastern Alps; n = 13); 2, SUS (Val di Susa, western Alps; n =
11);
3, CHI (Chianti, northern province of Siena; n = 11); 4, ARE
(province of Arezzo; n = 15); 5, SIE (southeastern province of
Siena; n = 11); 6, AMI (Mount Amiata, provinces of Siena and Gros-
seto;
n = 23); 7, PES (province of Pesaro; n = 12); 8, GRO (north-
ern province of Grosseto and Maremma Regional Park; n = 47); 9,
CAP (Capalbio, southern province of Grosseto; n = 26); 10, POR
(Castelporziano, province of Rome; n = 10); 11, GAR (Gargano Na-
tional Park, province of Foggia; n = 6); 12, POL (Pollino National
Park, Orsomarso Mountains, province of Cosenza; n = 4).
Maremma Regional Park, respectively, were available, and DNA
was isolated from hair roots using the Qiagen DNeasy Tissue kit.
DNA quality and amount of these samples allowed only mi-
crosatellite analysis.
RFLP
of mitochondrial DNA
The polymerase chain reaction (PCR) was used to amplify two
segments in the mitochondrial genome: the D-loop (D-loop, 1
kilobase in length) and NADH dehydrogenase 1 (ND1, 1.7 kilo-
base in length). Each PCR reaction was carried out in 50 pi total
volume, using 10-100 ng of DNA, 5 pi of 10X buffer Gold (Perkin-
-Elmer), 200 pM of each dNTP, 2.5-3.0 mM MgCl
2
, 20 pm of each
primer, and 1U of AmpliTaq Gold polymerase (Perkin-Elmer).
Amplifications were performed in a Perkin-Elmer Cetus thermal
cycler 9700 or 2400, and consisted of an initial 3 min denaturation
step at 94 °C, 32 cycles of 30 s at 94 °C, 30 s at the annealing
temperature (60 °C for the D-loop and 54° C for ND1), 2 min ex-
tension step at 72 °C, and was followed by 5 min at 72 °C,
Primers were based on Jäger et al. (1992) (5'-CTG CAG TCT CAC
CAT CAA CCC CCA AAG C; 5'-GGG AGA CTC ATC TAG GCA TTT
TCA GTG) and Cronin et al. (1994) (5'-ACC CCG CCT GTT TAC
CAA AAA CAT; 5'-GGT ATG AGC CCG ATA GCT TA) for the am-
plification of the D-loop and ND1 segments, respectively. En-
donucleases with tetranucleotide recognition sites {Alul,
BstUI,
Downloaded by [203.86.16.249] at 08:40 25 March 2014
GENETIC DIFFERENTIATION AND MANAGEMENT OF ROE DEER IN ITALY
369
Ddel, Haelll, Hhal, Hsp92II, Mbol, Mspl, Rsal,
ScrFI,
TaqI,
Tni9l,
Tsp509I, Bfal, Acil) and pentanucleotide sites (Hinfl) were used
to produce restriction digests for each segment. Digestion frag-
ments were run on 5% or 10% vertical precast polyacrylamide
gels (BIORAD) and visualized under UV light after ethidium bro-
mide staining. Fragment lengths were determined by comparison
to a Hinfl-digested <KX174 DNA marker (Promega). Each restric-
tion morph was designated by a capital letter. Individuals were
identified by a composite mtDNA haplotype created by the re-
striction profiles across segments.
Microsatellite analysis
All specimens were genotyped at 13 microsatellite loci. Species-
-specific primers
ROE01,
ROE05, ROE06, ROE08, and ROE09
were characterised by Fickel & Reinsch (2000), primers NVHRT16,
NVHRT21,
NVHRT24, NVHRT30, NVHRT48 and NVHRT71 were
described for reindeer by Roed & Midthjell (1998), while primer
pairs RT1 and RT7, originally isolated in reindeer as well, were
published by Wilson et al. (1997). Polymerase chain reactions
contained 50-100 ng of DNA, 2.5 pi of 10X buffer Gold (Perkin-
-Elmer), 200 pM of each dNTP, 15 pm of each primer (with the
forward primer fluorescently labelled), 2.5 mM MgCl
2
, and 1 U of
AmpliTaq Gold polymerase (Perkin-Elmer) in 25 pi total volume.
Loci ROE08 and ROE09 were co-amplified in a single reaction, as
well as loci RT1 and NVHRT48. Samples were heated to 94 °C for
3 min, followed by thirty-five cycles of 94 °C for 30 s, the anneal-
ing temperature for 30 s and 72
C
C for 30 s. The final synthesis
step was extended to 30 min at 72 °C to minimize the "plus A
artefact", i.e. the tendency of Taq polymerase to add a non-tem-
plate nucleotide (usually an A) to the 3' end of double stranded
DNA (Smith et al, 1995). PCR products were multiplex pooled in
individual tubes into groups of 6 and 7 loci, depending on prod-
uct size and fluorescent dye label. Pools were heated at 95 °C for
2 min, placed on ice and loaded onto an ABI Prism 310 Genet-
ic Analyzer. Alíele sizes were calibrated using a Genescan-500
TAMRA size standard run along with each sample. Fragment siz-
ing was performed by the Genescan Software ver 3.1.
Statistical methods
Mitochondrial DNA
The proportion of shared restriction fragments was used to esti-
mate genetic distance (Nei, 1987) in terms of nucleotide sequence
divergence (p, base substitutions per nucleotide) between all
pairs of haplotypes and populations following the fragment
method of Nei & Miller (1990) as implemented in the program
RESTSITE (Miller, 1991). For population comparisons, the values
of genetic distance were corrected for within-population diversity.
All fragments, variant and invariant, were used in the calculations.
Populations and haplotypes were clustered on unrooted neigh-
bour-joining trees (NJ) (Saitou & Nei, 1987) using the matrix of
the p-values derived by RESTSITE. Robustness of tree topologies
was assessed through 1000 bootstrap replications. Frequencies of
haplotypes were used to compute the value of F
ST
(Weir & Cock-
erham, 1984) as an estimator of genetic subdivision of popula-
tions,
using the updated version 2.9.3 of FSTAT (Goudet, 1995).
FSTAT was also used to perform a G-statistic per pair of samples
to test for population differentiation. Significance of pairwise val-
ues at the 0.1% nominal level was tested by 1000 permutations af-
ter standard Bonferroni correction for multiple testing (Rice,
1989).
Genetic diversity within populations was assessed by esti-
mating nucleotide diversity (p) and haplotype or gene diversity
(h) according to Nei (1987).
Microsatellite data
Genetic variability at each locus was evaluated by computing
the expected proportion of hétérozygotes (He) and unbiased
gene diversity (Ht) (Nei, 1987) across populations. Fisher's exact
test was performed to test for genotypic linkage disequilibrium
for all pairs of loci by employing the Markov chain method. Ge-
netic polymorphism for each population was measured as the
mean number of alíeles per locus (A), mean observed heterozy-
gosity (Ho) and heterozygosity expected (He) from Hardy-Wein-
berg (HW) assumptions corrected for small sample size. Because
of the presence of rare alíeles, deviations from HW expectations
were tested using the Markov chain method of Guo & Thompson
(1992).
Standard Bonferroni correction for multiple tests was em-
ployed to adjust the significance level.
Two approaches were used to investigate genetic differentia-
tion of populations. First, the null hypothesis of identical allelic
distribution was tested for all the roe deer specimens considered
as a whole population, and for pairs of samples, using the genie
differentiation test, as described in Raymond & Rousset (1995a).
For each locus, the test was performed on a contingency table,
and an unbiased estimate of the P-value was obtained using a
Markov chain. A combination of all test results was provided
across loci (Fisher's method) assuming a statistical independence
of loci. As a second approach to investigate population diver-
gence, estimates of F
ST
were computed as an overall value and
for all pairs of populations. In addition, we also calculated Rho,
an unbiased estimator of Slatkin's (1995) R
S
x> an analogue to F
ST
adapted to microsatellite loci, which accounts for differences in
alíele sizes under a stepwise mutation model. Confidence inter-
vals for both the overall value of F
ST
and Rho were obtained
through bootstrapping over loci (1000 replicates). Analyses were
conducted using the programs FSTAT (Goudet, 1995), GENEPOP
(Raymond & Rousset, 1995b), and RST CALC (Goodman, 1997).
To test for congruence between pairwise values of F
ST
and Rho, a
Mantel correlation test was applied with 1000 iterations using the
program MANTEL version 2.0
(Liedloff,
1999).
A drift-based measure of genetic distance, the proportion of
shared alíele (Dps) (Bowcock et al, 1994) was computed among
roe deer of different populations and among single individuals,
using the program MICROSAT (Minch et al, 1995). This distance
makes use of the sum of alíele frequencies shared between sam-
ples without the assumption of any mutation model, and has
been shown to reconstruct microsatellite-based relationships be-
tween closely related taxa better than other measures (Cornuet et
al, 1999). For microsatellite alíele frequency, drift and not muta-
tion seems to play a major role in genetic differentiation among
closely related populations (Perez-Lezaun et al, 1997). Estimates
of Nei's standard genetic distance D (Nei, 1972), chord distance
D
c
(Cavalli-Sforza & Edwards, 1967) and delta mu (8p)
2
(Gold-
stein et al, 1995) were also computed. The program NEIGHBOR
in PHYLIP version 3.6a (Felsenstein, 1993) was used to construct
NJ trees using as distance measure the negative logarithm of the
proportion of shared alíeles (-log Dps), and one minus the pro-
portion of shared alíeles (1 - Dps), respectively for the tree of
populations and tree of individuals. Statistical support at the in-
ternodes on the trees was assessed by 1000 bootstrap replications.
In order to investigate further the pattern of population struc-
ture,
we applied a Bayesian method, using the model-based clus-
tering algorithm as implemented by the program STRUCTURE
(Pritchard et al, 2000). This method does not assume a particular
mutation process, and uses multilocus genotypes to identify clus-
ters of genetically similar individuals without prior knowledge of
their population affinities (option POPINFO = 0). The individual,
rather than the population, is considered as the operational unit,
and alíele frequencies are replaced by individual multilocus geno-
types.
All individuals are initially considered as belonging to a
single population. K theoretical populations that are in Hardy-
-Weinberg equilibrium and show linkage equilibrium between
loci are inferred. The probability of the number of populations
equalling K was calculated from the estimated posterior log likeli-
hood of the data. If more than one subpopulation contributes to
the gene pool of the sample population as a whole, the log likeli-
hood of the data assumes the highest estimate with the most
probable value of K. The proportional membership (q®) of the
genotype of individual i in each of the K theoretical populations
was also derived. We ran the program for 100,000 iterations after
Downloaded by [203.86.16.249] at 08:40 25 March 2014
370
R. LORENZINI, S. LOVARI, M. MASSETI
a burn-in period of 10,000, setting the number of theoretical clus-
ters K from one up to twelve.
RESULTS
Genetic variability within populations
Nine out of 32 enzyme-segment combinations pro-
duced variable patterns of restriction fragments, which
defined 12 composite haplotypes for the roe deer popu-
lations sampled (Table I). Different RFLP profiles were
detected with 7 and 2 out of 16 restriction enzymes in
the D-loop and ND1 segments, respectively. The D-loop
was highly polymorphic, producing 12 different restric-
tion patterns in the 183 individuals surveyed. ND1
proved to be less informative, with only 3 RFLP profiles
identified. At the intrapopulation level, five populations
TABLE I - Mitochondrial DNA composite haplotypes observed in
roe deer from Italy.
Haplotype
G
E
PP
P
H
L
I
M
Q
o
R
N
Segment
a
D-loop
AAABABA
AAAAAAB
AAABADA
AAAADAB
ABABAAB
BAABBBA
ABCCACA
AAABAAB
ABAAACA
ABBCACA
ABABADB
AABCABA
ND1
AA
AA
AA
AA
BB
AA
AA
AA
AA
AA
AB
AA
a
, from left to right letters refer to: D-loop: Ddel, Sau3A, Hinfl,
Hsp92H,
Tru9I,
Tsp509I, Adi; ND1: Hinfl, Hsp92II.
were monomorphic, while seven showed more than
one haplotype, with values of haplotype diversity rang-
ing from h = 0.040 in roe deer from Grosseto to h =
0.758 in roe deer from Val Rendena (Table II). Consi-
dering all samples as a whole population, an estimate
of haplotypic diversity was calculated as h = 0.724,
which is indicative of a high heterogeneity in the mito-
chondrial genome of the Italian roe deer, regardless of
sequence divergence of haplotypes. In samples where
haplotypic variation occurred, values of nucleotide di-
versity of haplotypes within populations ranged from n
= 0.004% in roe deer from Grosseto to n = 0.147% in
roe deer from Val Rendena (Table II), which suggests
that, on average, the Alpine populations harbour the
highest level of mitochondrial variability.
A total of 100 distinct alíeles were observed at 13 mi-
crosatellite loci over the complete data set. Estimates of
gene diversity (Ht) and the proportion of expected hét-
érozygotes (He) for each marker calculated across pop-
ulations revealed on average high levels of nuclear vari-
ation for the roe deer in Italy (Table III). All loci were
polymorphic, showing from two (ROE01) up to twelve
(ROE06, RT1) alíeles. Values of Ht ranged from 0.157
(ROE09) to 0.902 (RT1), with an overall value of 0.662,
and values of He ranged from 0.146 (ROE09) to 0.787
(RT1),
with an overall value of 0.549- Alíele frequencies
are available from the first Author. No evidence of link-
age disequilibrium (P > 0.20 for all pairs of loci) was
found, and therefore each microsatellite marker was
considered as providing independent genetic informa-
tion. Measures of genetic variation for each population
(Table IV) were computed from the observed distribu-
tions of alíele frequencies. Populations showed from
moderate to high levels of variability, both in terms of
expected heterozygosity He, which ranged from 0.174
in Gargano to 0.578 in Val di Susa, and the mean num-
ber of alíeles per locus A, which ranged from 1.6 in
Gargano to 4.8 in Grosseto. The Monte Carlo approxi-
mation of Fisher's exact test for all loci combined
showed significant deviations from HW expectations (P
< 0.05, after Bonferroni correction) in Grosseto, Siena,
TABLE II - Haplotype diversity (h) and nucleotide diversity (n, in per cent) in the roe deer populations
studied.
Acronyms as in Figure 1.
h*
GRO
(42)
0.040
0.004
SIE
01)
0.0
0.0
CHI
(11)
0.413
0.075
POL
(4)
0.428
0.021
POR
(10)
0.0
0.0
Populations)
CAP
(26)
0.395
0.094
AMI
(23)
0.288
0.031
GAR
(5)
0.0
0.0
ARE
(15)
0.0
0.0
PES
(12)
0.0
0.0
REN
(13)
0.758
0.147
SUS
(11)
0.455
0.072
*, h was calculated as 2« (1 - 2f¡
2
) / (2w - 1), where f¡ is the frequency of the ith haplotype and n is the number of individuals sampled;
jt was calculated as Xx¡ Xj
K
{
¡,
where x¡ is the population frequency of the ith sequence, x¡ is the population frequency of the jth se-
quence, and rtjj is the number of nucleotide differences per nucleotide site between the ith and jth sequences
Downloaded by [203.86.16.249] at 08:40 25 March 2014
GENETIC DIFFERENTIATION
AND
MANAGEMENT
OF ROE
DEER
IN
ITALY
371
TABLE III -Variability per microsatellite locus across populations.
He, expected proportion of hétérozygotes; Ht, total unbiased gene
diversity, No., number of alíeles; F^, fixation index.
ROE09
NVHRT71
NVHRT30
NVHRT48
ROE05
ROE01
NVHRT21
RT7
NVHRT24
ROE08
NVHRT16
RT1
ROE06
Overall
He
0.146
0.609
0.366
0.470
0.406
0.327
0.643
0.690
0.636
0.657
0.701
0.787
0.699
0.549
Ht
0.157
0.816
0.419
0.574
0.530
0.350
0.760
0.858
0.803
0.805
0.796
0.902
0.835
0.662
No.
3
9
7
3
5
2
7
11
10
11
8
12
12
100
Size range
172-176
94-120
152-168
85-89
118-138
134-136
155-179
207-231
125-151
58-90
155-175
216-238
84-108
FST
0.077
0.225
0.146
0.181
0.219
0.042
0.155
0.159
0.169
0.141
0.100
0.106
0.134
0.149
Casteporziano, Amiata, Arezzo and Chianti, where a de-
ficiency of hétérozygotes was recorded at 5, 2, 2, 2, 1,
and 3 loci, respectively. For the loci not in equilibrium
(NVHRT71, NVHRT21, NVHRTló, RT1, RT7), departures
were due largely to the presence of homozygotes for
rare alíeles. The bi-allelic locus ROE01 was evaluated
with the x
2
goodness-of-fit test, and no departures were
revealed for any of the populations.
Genetic divergence and relationships between populations
The distribution of mitochondrial haplotypes among
populations showed a clear segregation according to
geographic distribution (Table V). Haplotype G was
confined to the roe deer of central-southern Italy
(northern Grosseto and Maremma Regional Park, south-
ern Siena, Chianti, Pollino National Park, Castel-
porziano, Capalbio, Amiata), while haplotype H was
common only in those from the Alpine areas (Val Ren-
dena, Val di Susa), where it occurred at the highest fre-
quency, and in roe deer of presumed Alpine origin
TABLE IV - Measures of genetic variability in roe deer populations as estimated by 13 microsatellite loci. He (Ho), mean expected (ob-
served) heterozygosity; A, mean number of alíeles per locus. Standard error in parentheses.
GRO
SIE CHI POL POR CAP
AMI
GAR
ARE
PES REN
SUS
He 0.505 0.388 0.499 0.468 0.404 0.523 0.436
(0.062) (0.057) (0.058) (0.091) (0.115) (0.052) (0.073)
Ho 0.601 0.557 0.669 0.466 0.454 0.6l6 0.604
(0.057) (0.060) (0.067) (0.067) (0.091) (0.052) (0.070)
(0.7)
0.174 0.437 0.437 0.553 0.578
(0.063) (0.080) (0.068) (0.062) (0.068)
0.218 O.552 0.558 0.584 0.594
(0.063) (0.067) (0.066) (O.O6D (0.068)
3.1
(0.3)
4.6
(0.6)
2.1
(0.2)
2.5
(0.3)
4.5
(0.5)
4.0
(0.6)
1.6
(0.2)
3.9
(0.5)
3.3
(0.4)
3.5
(0.4)
4.2
(0.5)
TABLE
V - Distribution ofmtDNA haplotypes in roe deer populations.
PP H M O
R N
GRO
SIE
CHI
POL
POR
CAP
AMI
GAR
ARE
PES
REN
SUS
Total
41
11
8
3
10
3
4
80
1
19
2
22
1
1
3
3
3
15
12
4
8
42
20
20
Downloaded by [203.86.16.249] at 08:40 25 March 2014
372
R. LORENZINI, S. LOVARI, M. MASSETI
presently dwelling in central Italy (Arezzo, Pesaro).
There were no shared haplotypes between populations
of Alpine regions and those of central-southern Italy.
Roe deer in Mount Amiata (on the border between the
provinces of Siena and Grosseto) are known to derive
from restocked individuals of eastern European origin
(Mazzoni della Stella, 1990). Not surprisingly, haplotype
E, found in this area at high frequency, was also pre-
sent in roe deer from the eastern Alps (Val Rendena)
(Table V) and central-eastern Europe (Lorenzini et al.,
unpubl. data). The same holds for roe deer in Chianti,
where the "Alpine" haplotype H was also found. In this
area, restocking has been carried out recently, presum-
ably using Alpine individuals. Haplotype N was restrict-
ed to roe deer from Gargano, while haplotypes I, M,
and L, the last being highly frequent, occurred as pri-
vate haplotypes in roe deer sampled in Capalbio. Apart
from haplotype PP in the Pollino sample, all other hap-
lotypes were found in roe deer from the Alpine range.
Haplotypes showed different levels of differentiation,
divergence of sequences ranging from p = 0.04% be-
tween haplotypes P and E, to p = 0.61% between haplo-
types H and N (Table VI). No evidence was given of any
relation between sequence divergence and the distribu-
tion of haplotypes within and between populations, e.g.,
the haplotypes within populations in Val Rendena (E, P,
H, O) or Val di Susa (H, Q, R) showed p-values as high
as or greater than those found for haplotypes occurring
between different populations. In a neighbour-joining
tree based on pairwise nucleotide divergence (Table VI),
two main clusters of haplotypes are detectable (Fig. 2).
However, no evidence for significant structure or cluster-
ing of haplotypes according to populations was provid-
ed. Haplotypes found in different populations grouped
together in the same cluster and the degree of sequence
divergence was equally distributed within clusters (p
ranging from 0.11% to 0.53%) and between clusters (p
ranging from 0.22% to 0.61%), with the only exception of
haplotypes P and E in one cluster, showing a sequence
divergence an order of magnitude smaller (p = 0.04%).
Fig. 2 - Neighbour-joining tree based on pairwise nucleotide di-
vergence of mtDNA haplotypes. Branch lengths reflect sequence
divergence according to scale, and numbers at nodes show sup-
port from 1000 bootstrap replicates when values are over 50%.
The value of F
ST
= 0.783, calculated for all populations
combined, indicated that 78% of total mitochondrial vari-
ation was due to differences among populations. Fur-
thermore, significant levels of pairwise divergence were
obtained for 53 out of 66 comparisons (G- test, P <
0.001,
after standard Bonferroni correction) (Table VII).
Roe deer from different sampling locations were clus-
tered on a neighbour-joining tree (Fig. 3A) derived from
a matrix of pairwise nucleotide divergence between
populations (Table VII). In the tree, roe deer of the
Alpine range (Val Rendena, Val di Susa) are separated
in western and eastern populations, and come together
with roe deer of well-known or supposed Alpine origin
introduced into central Italy (Amiata, Arezzo, Pesaro). A
cluster is formed by roe deer of central-southern Italy
(southeastern Siena, northern Grosseto and Maremma
Regional Park, Castelporziano, Pollino National Park,
Chianti hills) and is well separated from the "Alpine"
group. Roe deer from Gargano and Capalbio are quite
distinct and occur on separate branches within the
TABLE
VI - Matrix of pairwise nucleotide divergence, p (in per cent), of
mtDNA
haplotypes.
E
PP
P
H
L
I
M
O
Q
R
N
G
0.223
0.112
0.264
0.382
0.149
0.416
0.150
0.416
0.264
0.341
0.224
E
0.259
0.037
0.319
0.373
0.409
0.074
0.409
0.259
0.345
0.296
PP
0.301
0.419
0.262
0.375
0.188
0.375
0.225
0.225
0.338
P
0.341
0.416
0.453
0.112
0.453
0.301
0.379
0.338
H
0.535
0.573
0.227
0.573
0.419
0.188
0.615
L
0.568
0.301
0.568
0.417
0.494
0.375
I
0.494
0.146
0.148
0.453
0.335
M
0.494
0.341
0.264
0.379
O
0.148
0.453
0.184
Q
0.301
0.338
R
0.573
Downloaded by [203.86.16.249] at 08:40 25 March 2014
GENETIC DIFFERENTIATION AND MANAGEMENT OF ROE DEER IN ITALY
373
TABLE VII - Below the diagonal: nucleotide divergence, p (in per cent), of mitochondrial haplotypes between pairs of Italian roe deer
populations. Values are corrected for within-population diversity. Above the diagonal: significance of unbiased P values of
G-test
for pair-
wise comparisons
(**,
P < 0.01; "*, P < 0.001; NS).
GRO
AMI
REN
CAP
ARE
SUS
SIE
CHI
PES
POR
GAR
POL
GRO
0.138
0.176
0.119
0.375
0.271
0.0
0.018
0.375
0.0
0.221
0.001
AMI
***
0.055
0.216
0.281
0.194
0.149
0.103
0.281
0.149
0.251
0.134
REN
**
***
0.234
0.143
0.064
0.184
0.091
0.143
0.184
0.211
0.160
CAP
***
***
***
0.443
0.334
0.092
0.104
0.443
0.092
0.288
0.088
ARE
***
***
***
***
0.013
0.382
0.194
0.0
0.382
0.616
0.364
SUS
**
***
NS
***
NS
0.277
0.122
0.013
0.277
0.481
0.252
SIE
NS
**
***
***
***
*#*
0.020
0.382
0.0
0.225
0.0
CHI
NS
***
#**
***
***
***
***
0.194
0.020
0.247
0.015
PES
***
***
*•*
***
NS
NS
***
*•*
0.382
0.616
0.363
POR
NS
».«
**
***
***
**
NS
NS
***
0.225
0.0
GAR
*#*
***
**#
**
**
*•*
••
*•*
••*
***
0.225
POL
NS
NS
**
***
***
***
**
NS
NS
••
B
PES
ARE
SUS
REN
AMI
CAP
SIE
GAR
POR
Fig. 3 - Neighbour-joining trees of populations. A, topology based on
pairwise nucleotide divergence of mtDNA haplotypes. B, microsatel-
lite topology based on the negative logarithm of the proportion of
shared alíeles between populations. Bootstrap support at the nodes
(1000 replicates) are indicated only when values are over i
"central-southern" cluster. Between the "Alpine" and
"central-southern" groups, the p-values ranged from
0.09% between Chianti and Val Rendena, to 0.62% be-
tween Gargano and Arezzo/Pesaro, while comparable
values of within-clusters divergence were found in both
groups (p values < 0.29, Table VII). The tree was quite
stable, the nodes all having a bootstrap support equal
to or greater than 50%.
Populations differed substantially in alíele frequencies
at microsatellite loci, as suggested by the genie differen-
tiation test (P < 0.0001). In pairwise comparisons across
loci (Fisher's method), the probability test yielded sig-
nificant values in all but 2 of 66 pairs of populations
(northern Grosseto-Maremma Regional Park/Siena and
Arezzo/Pesaro) at P level less than 0.05 (Bonferroni cor-
rected). Genetic differentiation between populations
was also apparent when values of F
ST
per locus (Table
III) were considered, with all loci contributing almost
equally to differences in the distribution of alíele fre-
quencies among populations. The overall value of F
ST
was 0.149, and 99% confidence interval obtained via
bootstrapping over loci was 0.117-0.180. The estimate of
Rho,
as the analog to F
ST
adapted to microsatellite loci,
was 0.219 (99% confidence interval 0.161-0.283) across
the whole data set. These values suggest that about 15-
-22%
total variation at microsatellite loci is distributed
among populations. Mantel correlation analysis between
estimates of pairwise F
ST
and Rho (values not shown)
revealed the two matrices to be significantly correlated
(g = 2.78, 2 = 0.2889, r = 0.72, P < 0.001), indicating
that both the distribution of alíele frequencies and alíele
size account for distinction among populations.
The overall pattern of population divergence was ex-
amined by deriving a NJ tree based on one minus the
proportion of shared alíeles (1-Dps) between popula-
tions.
The topology from microsatellites (Fig. 3B) was
congruent with that from mtDNA (Fig. 3A) in showing
Downloaded by [203.86.16.249] at 08:40 25 March 2014
374
R. LORENZINI, S. LOVARI, M. MASSETI
one cluster, formed by Alpine populations and those of
northern origin reintroduced into central Italy, well sep-
arated from the cluster formed by central-southern pop-
ulations. In the microsatellite tree, however, the roe
deer of Gargano and Pollino National Park proved to be
the most distant ones. Use of other distance measures
(Nei's D, F
ST
, D
c
) gave the same results or performed
worse (5u)
2
(not shown). The analog inter-individual
clustering method allowed no grouping of individuals
according to pre-defined population membership. Most
specimens from the same sampling sites were scattered
throughout the tree (not shown), which showed a low
statistical support at the internodes (p < 40%) and,
therefore, the genetic relationships between population
clusters were difficult to establish.
A different approach was also used to infer population
structure from microsatellite data. A likelihood-based
method of clustering, the Bayesian method, was applied
using the program STRUCTURE. The highest log proba-
bility of our data was obtained for K = 8 (-3376.5), while
lower estimates resulted for smaller or greater values of
K (not shown), indicating that eight genetically distinct
populations are most appropriate for interpreting the
data. According to the values of q
(i)
(Table VIH), roe
deer from the pre-defined populations of Pollino,
Gargano and Val Rendena fell into clusters VIII, VII, and
V, respectively, with a proportion of membership rang-
ing from 0.92 to 0.94. Roe deer of Capalbio grouped
mainly in cluster VI (q = 0.74), with a small portion of
their ancestry falling in cluster III. Individuals from two
pre-defined populations, Siena and Grosseto, were
grouped together into cluster III, with q = 0.74. The
same holds for roe deer of Arezzo and Pesaro
provinces, both included in cluster IV, with q = 0.82.
Roe deer from the Maremma Regional Park grouped to-
gether with those from Castelporziano in cluster I, with
q = 0.86. By contrast, populations of Amiata and Chianti
proved to be admixed, falling partially in clusters II, III,
VI,
VII, and I, II, IV, respectively, which is congruent
with their supposed reintroduced origin. Roe deer in
Val di Susa were also admixed, and were associated
partially to clusters II, IV and V. Thus, cluster II seems
not to be associated with any of the Italian populations
studied.
DISCUSSION
A palaeontologic perspective
In the fossiliferous horizons of continental Italy, re-
mains of C. capreolus have been recorded from many
anthropozoic deposits of the Middle and Upper Pleis-
tocene (Masseti et al, 1995; Gliozzi et al, 1997; Sardella
et al, 1998). Although mostly the examination of fossil
materials does not provide substantial information on
the Late Pleistocene populations at the phenotypic lev-
el,
it helps to indicate their distribution on the basis of
the location of fossil deposits providing their remains.
To assess the range of the original distribution of the
species, earlier chronologies prior to the Neolithisation
(6th millennium BC) of the Italian territories should be
considered, after which improved human sea-faring
skills and the established commercial networks between
Mediterranean countries allowed the exportation of
game animals, together with those already involved in
the process of domestication (Masseti, 1998). In palaeo-
environmental reconstructions, the roe deer almost al-
ways features in association with the red deer Cervus
elaphus and the wild boar Sus scrofa (cf. Cassoli &
Tagliacozzo, 1994). The late Pleistocene dispersion of
roe deer in Italy is documented from the Alpine region
to the southernmost parts of the peninsula (Caloi &
Palombo, 1994). Fossil remains of roe deer have also
been provided from the Riss-Early Wurm glacial
episode in several islands of the Tuscan archipelago
(Azzaroli et al, 1990; Rustioni & Mazza, 1993). Howev-
er, the occurrence of roe deer in the Pleistocene-Early
Holocene of Sicily has yet to be confirmed (Burgio,
1997;
Burgio et al, 1998). Evidence for the distribution
of roe deer in Upper Pleistocene in the Italian peninsu-
la has been reported, besides at other prehistoric sites,
from several of those geographic areas where the extant
populations show special genetic patterns (central and
southern Tuscany, West Latium, the Gargano peninsula,
and northern Calabria) (Tassi, 1976), possibly support-
ing the hypothesis of local maintenance of native popu-
lations. Remains of roe deer have been provided by the
exploration of several prehistoric sites from northern
and western Tuscany (Biagi et al, 1980; Castelletti et al,
1994;
Sammartino & Tozzi, 1994; Boscato, 1996), Latium
(Blanc, 1953; Radmilli, 1974; Sala, 1983), the Gargano
peninsula (Palma di Cesnola, 1984; 1992), and northern
Calabria (Massed & Rustioni, 1988). In several cases, the
occurrence of roe deer is also documented in relatively
recent archaeological contexts from Apulia, where they
have been reported among the Mesolithic cultural as-
semblages (Bon & Boscato, 1993), both in Neolithic and
Bronze Age deposits (Siracusano, 1990/199D-
Genetic diversity
A population genetic assessment of contemporary roe
deer sampled throughout its current range in Italy has re-
vealed that their genetic variability and differentiation are
distributed in a complex pattern. DNA data from RFLP of
two mitochondrial gene segments, D-loop and ND1, and
composite microsatellite genotypes for 13 loci are consis-
tent and provide evidence for two genetically distinct
groups of roe deer populations currently present in the
Italian peninsula (Fig. 3A, B). One group includes roe
deer from the Alpine arc, which share mtDNA haplo-
types and allelic variants at microsatellite loci with roe
deer of central-northern Europe (Lorenzini et al., unpubl.
data),
whilst the other includes populations dwelling in
central-southern Italy, where the existence of subspecies
C. c. italiens was previously proposed (Festa, 1925).
Downloaded by [203.86.16.249] at 08:40 25 March 2014
GENETIC DIFFERENTIATION AND MANAGEMENT OF ROE DEER IN ITALY
375
The occurrence of mtDNA haplotype G, shared with
neither the Alpine populations nor any other population
in Europe, along with highly significant differences in al-
lele frequencies observed at microsatellite loci, as well
as the presence of private alíeles, suggests a different ge-
netic lineage for roe deer in central-southern Italy. Thus,
according to our results, the existence of C. c. italiens
may be supported. However, in contrast to a generally-
-accepted opinion, this "southern" lineage of populations
seems not to be confined to Castelporziano, the Orso-
marso Mountains (Pollino National Park) and Gargano,
but it may still be present in some areas of central Italy.
In a recent study by Vernesi et al. (2002) on some Italian
populations of roe deer, sample sizes seem not adequate
to draw "safe conclusions" about the non-existence of C.
c. italiens, nor the number of sampling localities are at
all representative of the complex genetic pattern of nat-
ural and restocked populations of roe deer in central-
-southern Italy. On the other hand, a low genetic varia-
tion at the population level, as stated by the Authors, is
not supported by our data, based on larger samples and
different genetic markers. Our results indicate that nuclei
of "native" roe deer have survived in some areas of the
provinces of Siena and Grosseto, in spite of the strong
hunting pressure and habitat reduction during the last
centuries (e.g. Ghigi, 1911; Lovari, 1993).
In Europe, the Zoogeographie pattern of many species
has been influenced mainly by the movements of ani-
mals following the Quaternary climatic events (Hewitt,
1996).
In this scenario, a primary role for population dis-
tribution and differentiation was probably played by the
refuge areas of the Mediterranean b