Marine Biology (2006) DOI 10.1007/s00227-006-0300-x

RESEARCH ARTICLE

Concetta Tigano · Adriana Canapa · Venera FerritoMarco Barucca · Isabella Arcidiacono · Alan DeidunPatrick J. Schembri · Ettore Olmo

A study of osteological and molecular differences in populationsof Aphanius fasciatus Nardo 1827, from the central Mediterranean (Teleostei, Cyprinodontidae)

Received: 20 January 2005 / Accepted: 2 February 2006© Springer-Verlag 2006

Abstract Nine populations of Aphanius fasciatus Nardo,1827 from the central Mediterranean were analysed byexamining the mitochondrial control region and the mor-phology of the bony elements of the skull and vertebralcolumn, to study the degree of intraspeciWc diVerentiationof A. fasciatus considering the level of isolation of thediVerent populations and the palaeogeographic history ofthe central Mediterranean area. Both the molecular andmorphological analyses diVerentiate between the popula-tions, even if the topologies of the two trees are diVerent.Molecular and osteological investigations have consis-tently demonstrated a well-supported diVerentiation of thesouth-eastern Sicilian populations both within the samegroup (Tigano et al. in Ital J Zool 71:1124–1133, 2004a;Tigano et al. in Abstract volume XI European Congress ofIchthyology, Tallin, Estonia, 2004b), and from the popula-tions from western Sicily, Tunisia and the island of Malta.The molecular results show that the nine populations arecharacterised by haplotypes that are well deWned in rela-tion to a probably limited gene Xow; while, as regards themorphological data the diVerentiation found could beexplained in terms of the geographic isolation of the vari-ous populations, although the inXuence of environmental

factors, which diVer greatly between the various siteswhere the populations live, cannot be ruled out.

Introduction

Models of genetic variation relative to the geographicdistribution of a species, correlated to changes in geneXow over time and to diversiWcation of the populations,have recently been studied using a phylogeographicapproach that has allowed a better interpretation of themicro-evolutionary processes in nature (Avise 2000);studies based on the analysis of variations in mitochon-drial DNA, have shown a considerable phylogeographicstructure in populations of marine Teleostei (Bernardi2000; Stepien et al. 2001).

Amongst the vertebrates, cyprinodontids are a group ofteleosts that is particularly suitable for the study of micro-evolutionary processes which shape patterns of geneticstructuring and geographic variation in natural popula-tions, as well as of those adaptive processes in response toecological conditions (Villwock 1976). There is a vastamount of literature on the population diVerentiation ofkilliWsh from the morphological, genetic and molecularpoints of view (e.g. Echelle et al. 1987; Garciamarin et al.1990; Fernandez-Pedrosa et al. 1995; Strecker et al. 1996;Wilde and Echelle 1997; Duhan and Minckley 1998;Duvernell and Turner 1999; Echelle et al. 2000; Perdiceset al. 2001; Torralva et al. 2001; Reichenbacher and Sien-knecht 2002; Doadrio et al. 2002; Lussen et al. 2003).

The cyprinodontid Aphanius fasciatus Nardo(Fig. 1), is currently distributed in the saline coastalwaters of the central and eastern Mediterranean, in saltXats and also occasionally in inland fresh water(Wildekamp 1993). The natural fragmentation of suchsaline habitats contributes to the non-contiguouscoastal distribution of this species, and studies of geneXow between populations suggest that the one-dimen-sional stepping-stone model proposed by Kimura andWeiss (1964) and Slatkin (1994) might best describe the

Communicated by R. Cattaneo-Vietti, Genova

C. Tigano (&) · V. Ferrito · I. ArcidiaconoDepartment of Animal Biology, University of Catania, Via Androne, 81, 95124 Catania, ItalyE-mail: tigaconc@unict.itTel.: +39-95-7306030Fax: +39-95-327990

A. Canapa · M. Barucca · E. OlmoInstitute of Biology and Genetics, Marche Polytecnical University, Via Brecce Bianche, 60131 Ancona, Italy

A. Deidun · P. J. SchembriDepartment of Biology, University of Malta, MSD06 Msida, Malta

mechanisms shaping the genetic structure of this spe-cies (Maltagliati 1998b). Studies on the populationdiVerentiation of A. fasciatus have utilised the osteol-ogy of the skull (Tigano and Ferrito 1985; Tigano andParenti 1988; Tigano 1991; Parenti and Tigano 1993;Tigano et al. 1999, 2001), allozymic variation (Compa-rini et al. 1984; Maltagliati 1998a, b, 1999, 2002; Malta-gliati et al. 2003; Cimmaruta et al. 2003), andcytogenetic variation (Vitturi et al. 1995; Ferrito et al.2000; Tigano et al. 2003). The allozymic and morpho-logical investigations have all demonstrated a notablediVerentiation of the populations; analogously, theanalysis of the highly variable D-loop tract of the mito-chondrial DNA (Tigano et al. 2004a, b) has indicated astrong genetic divergence between three Sicilian popu-lations of this species. On the other hand, the molecularanalysis carried out by Hrbek and Meyer (2003)showed that there is limited structuring of A. fasciatuspopulations; however, authors who analysed variousspecies of the genus Aphanius, considered mitochon-drial genes more useful for studies above the specieslevel. While the allozyme studies indicate that, in somecases, there is indeed genetic divergence of the popula-tions in relation to their geographic distribution

(Maltagliati 1998a, 1999), other studies both based onallozymic (Cimmaruta et al. 2003) and morphologicaldata (Tigano et al. 2001; Ferrito et al. 2003), suggestthat this divergence does not relate to the geographicdistance between the diVerent populations. The aim ofthe present study is to evaluate the degree of diVerentia-tion of nine populations of the killiWsh A. fasciatusfrom the central Mediterranean, on the basis of molec-ular and morphological analyses, with reference to thepalaeogeographic history of the Mediterranean area.

Materials and methods

Specimens from six diVerent populations of A. fasciatuswere divided into three groups (two populations in eachgroup), based on their place of origin: western Sicily,Tunisia and the island of Malta; data from a fourthgroup of three populations from south-eastern Sicilypreviously studied by Tigano et al. (2004a) (Table 1,Fig. 2) were also used in the statistical analyses of geneticstructure and osteological variables.

Meristic and morphometric analysis

A total of 150 specimens of A. fasciatus (standardlength=2.1–4.6 cm) were Wxed in 5% formalin and thenpreserved in 70% ethanol. For the study of the skeletalelements, the specimens were treated according to theprocedure described by Dingerkus and Uhler (1977)which employs alcian blue for the staining of cartilageand alizarin red S for bone. Nineteen meristic and 25morphometric characters were examined (Table 2). A dia-grammatic representation of the osteological charactersconsidered in this study has been given by Ferrito et al.(2003). The Kruskall–Wallis test was used for analysis ofvariance while comparisons between the diVerent popula-tions were made using the Mann–Whitney U test. DiVer-ences in body shape were analysed using principalcomponent analysis (PCA). To summarise inter-popula-tion diVerences, cluster analysis based on the unweightedpair-group method using arithmetic average (UPGMA)was applied to the matrix of the generalised Mahalanobisdistance. Statistical analyses were made using the soft-ware package STATISTICA (http://www.statsoft.com).

Fig. 1 Aphanius fasciatus Nardol

Table 1 List of the populations examined: sampling areas with geographic co-ordinates, population identiWcation code and number ofspecimens examined

Groups Sampling areas Co-ordinates Code Specimens

Sicily south-eastern Pantano Longarini 36°40�N 15°02�E LO n=25: 8 ## 17 $$ From Tigano et al. (2004a) Foce Marcellino 37°15�N 15°12�E FM n=25: 4 ## 21 $$

Pantano Viruca 36°39�N 15°03�E PV n=24: 8 ## 16 $$ Sicily western Salina Chiusa Trapani 38°01�N 12°28�E TP n=25: 5 ## 20 $$

Salina Curto Marsala 37°47�N 12°27�E SMA n=25: 6 ## 19 $$ Malta Ghadira Bird Sanctuary 35°27�N 14°21�E GHA n=25: 14 ## 11 $$

Salina 35°56�N 14°24�E SAL n=25: 11 ## 14 $$ Tunisia Lake Tunisi South 36°49�N 10°14�E LTS n=25: 6 ## 19 $$

Ghar El Milh 37°09�N 10°09�E GEM n=25: 11 ## 14 $$

To remove the size component from the shape measure-ments (Thorpe 1976), all the measurements of the individ-ual morphometric variables were standardised accordingto the formula: MS=MO (LS/LO)b, where MS is the stan-dardised measurement, MO the measured characterlength, LS the overall (arithmetic) mean standard lengthfor all Wsh from all samples in each analysis, LO the stan-dard length of specimen; b was estimated for each charac-ter from the observed data by the allometric growthequation M=aLb. Parameter b was estimated as the slopeof the regression of log MO on log LO, using all the speci-mens from all groups, but allowing the intercept to diVerbetween groups (Elliott et al. 1995).

Mitochondrial control region sequence analysis

Total DNA was extracted according to the method ofJeVreys and Flavell (1977) from 53 specimens of A. fascia-tus. About ten specimens from each population were analy-sed. For PCR ampliWcation of the mtDNA control region,the forward primer (5�-ACTATTCTTTGCCGGATTCTG-3�) (Tigano et al. 2004a) and the reverse primerdesigned by Meyer et al. (1990) were used. The ampliWca-tion conditions were as follows (30 cycles): 94°C, 1 min;55°C, 1 min; 72°C, 2 min. The ampliWed DNA was directlysequenced using an automated DNA sequencer (ABIPRISM 310, from PE Biosystems). Alignments were per-formed with the CLUSTAL W program (Thompson et al.1994) set at default parameters. CLUSTAL W is available

from the Internet web site http://www.ftp.ebi.ac.uk. Thetrees were produced using neighbour-joining (NJ) andmaximum parsimony (MP) with the PAUP 4.0 beta ver-sion (SwoVord 1998). The NJ tree was constructed usingpairwise distances calculated following the application ofKimura’s (1980) two-parameter correction for multiplesubstitutions. The MP tree was produced using branch andbound search with equal character weighting and randomstepwise addition with ten replications, with only minimaltrees being retained. Bootstrap values, indicating robust-ness of nodes, refer to 1,000 replications. The alignment ofhaplotype sequences (accession numbers from AM184186to AM184201) with another eight sequences of three popu-lations of A. fasciatus from south-eastern Sicily obtainedfrom GenBank (AJ605322-AJ605329) and the sequence ofA. danfordii used as outgroup (U06062) was 381 nucleo-tides long. To complement these results, Nested CladeAnalysis was performed. The algorithm developed by Tem-pleton et al. (1992) was implemented using the computerprogram TCS 1.13 (Clement et al. 2000). The permutationtests were carried out using the GEODIS 2.0 program(Posada et al. 2000) and the results were analysed with theinference key provided in Templeton and Sing (1993).

In order to investigate the correlation between molec-ular and morphological distance matrices, Mantel’s testwas carried out on matrices of Mahalanobis and D-loopK80 molecular pairwise distances. Probabilities wereread directly from the distribution of 5,000 randomisedmatrices computed by permutation.

Fig. 2 Sampling areas in Sicily, Malta and Tunisia. The asterisk indicates the three south-eastern Sicilian populations (Tigano et al. 2004a). South-eastern Sic-ily: LO Pantano Longarini, FM Foce Marcellino, PV Pant-ano Viruca; Western Sicily: TP Salina Chiusa Trapani, SMA Salina Curto Marsala; Malta: GHA Ghadira Bird Sanctuary, SAL Salina; Tunisia: LTS Lake Tunisi South, GEM Ghar El Milh

Results

Meristic and morphometric analyses

The Kruskall–Wallis analysis of variance showed signiW-cant diVerences between all the variables taken into con-sideration. Results of the Mann–Whitney U test, carryingout a pairwise comparison within each of the four groupsof populations, indicated that the south-eastern Sicilianpopulations are well diVerentiated within the group; onthe other hand, the two western Sicilian populations arethe least diVerentiated.

The pairwise comparison between the four groups ofthe killiWsh populations analysed in this study, show thatthe south-eastern Sicilian populations are very much

diVerentiated from all the other groups, while the west-ern Sicilian populations are well diVerentiated fromthose of Malta and, to a lesser extent, from those ofTunisia; Wnally, the Maltese and the Tunisian popula-tions are the least diVerentiated (Fig. 3).

Multivariate analysis

A PCA on morphometric measurements normally resultsin a Wrst principal component characterised by consis-tently positive loadings, where size is the primary sourceof variance in the data. However, because we analysedstandardised measurements to eliminate the size eVect, nosuch eVect is expected in PC1. Principal components 1–5account for 34.7, 9.3, 7.4, 6.7 and 5.0% of the totalvariance, respectively; Fig. 4 shows PC1 and PC2. PC1

Table 2 List of the osteologicalvariables and the codes used inthe text

*Morphometric characters**Meristic characters

Code Osteological variable

1 BB1 b1* Length of basibranchial 12 BB2 b2* Length of basibranchial 23 BS bc* Length of cartilagonous basihyal4 BS bo* Length of ossiWed basihyal5 BS n* Width of the area between cartilage and bone of basihyal6 CB5 a* Width of ceratobranchials 57 CB5 dt** Total number of teeth of ceratobranchials 58 CB5 x* Tooth antero-posterior diameter of ceratobranchials 59 CB5 y* Tooth medio-lateral diameter of ceratobranchials 510 DE dt** Total number of teeth of dental11 MX b* Length of maxillary2 MX h* Width of maxillary13 MX R* Ratio length/width of maxillary14 PA br** Number of branched rays of anal Wn15 PA rt** Total number of rays of anal Wn16 PA ur** Number of unbranched rays of anal Wn17 PB2 df** Number of tooth rows of pharyngobranchials 218 PB2 dt** Total number of teeth of pharyngobranchials 219 PB2 bs** Number of bicuspid teeth of pharyngobranchials 220 PB2 ts** Number of tricuspid teeth of pharyngobranchials 221 PB3 dt** Total number of teeth of pharyngobranchials 322 PB4 tp** Total number of teeth of pharyngobranchials 4 toothplate23 PB4bs tp** Number of bicuspid teeth of pharyngobranchials 4 toothplate24 PB4ts tp** Number of tricuspid teeth of pharyngobranchials 4 toothplate25 PD br** Number of branched rays of dorsal Wn26 PD rt** Total number of rays of dorsal Wn27 PD ur** Number of unbranched rays of dorsal Wn28 PMX b1* Length of premaxillary29 PMX df** Number of tooth rows of premaxillary30 PMX dt** Total number of teeth of premaxillary31 PMX h1* Width of premaxillary32 PMX R* Ratio length/width of premaxillary33 PRS a* Width of parasphenoid34 PRS b* Length of parasphenoid35 SPC a* Width of the anterior bony end of supraoccipital36 SPC b* Length of the antero-median cartilagineous strip of supraoccipital37 SPC e* Width of the lateral processes of supraoccipital38 SPC v* Length of the lateral processes of supraoccipital39 UR b* Length of uroyal40 UR c* Minimum width of urohyal41 UR h* Heigth of hook of urohyal42 VR n** Total number of vertebrae43 VR x* Length of vertebrae44 VR y* Width of vertebrae

divides the south-eastern Sicilian populations from theothers (Fig. 4a). PC1 scores from the two groups of Sicil-ian populations (eastern and western) diVer signiWcantlyfrom those of Malta and Tunisia. Moreover, PC1 scoresfor individuals from eastern Sicily diVer signiWcantly fromthose of western Sicily, while PC1 scores for individualsfrom Malta and Tunisia do not diVer signiWcantly(Table 3a). Six measures correlate highly (>0.7) with PC1(Fig. 4b), of which Wve are measures of the oral and vis-ceral bony elements: premaxillary and maxillary length,basibranchial 1 and 2 length, and uroyal length; the othermeasure is the width of the vertebrae.

PC2 distinguishes the bulk of the specimens of thePantano Viruca population. The parasphenoid lengthand width are the only measures that are highly corre-lated with PC2. PRSb is the only measure negatively cor-related with PC2 (Table 3b). PC2 scores for individualsfrom the Pantano Viruca population diVer signiWcantlyfrom those of all other populations. Box plots of thevariables highly correlated with PC1 and PC2 are shownin Fig. 5.

The Wrst node of the UPGMA cluster analysis, basedon the generalised Mahalanobis distance matrix, clearlyseparates Pantano Viruca from all the other populations,while the second node separates the remaining twosouth-eastern Sicilian populations (Longarini and FoceMarcellino) from the rest. The western Sicilian groupclusters together with the Maltese and the Tunisiangroups and shows greatest aYnity with the Tunisianpopulations (Fig. 6).

Molecular analysis

For the molecular study, the sequences of a portion ofthe D-loop between the tRNA proline and the CSBDzone (length between 376 and 378 nucleotides) wereanalysed and showed a total of 32 variable sites whichdeWned 24 haplotypes from the 82 specimens examined.The number of haplotypes identiWed for each populationwere: two for Pantano Viruca, Salina, Ghar El Milh andSalina Chiusa Trapani; three for Pantano Longarini,Foce Marcellino, and the Ghadira Bird Sanctuary; Wvefor Salina Curto Marsala; and six for Lake Tunis South.The number of segregating sites for diVerent haplotypes

Fig. 3 Diagrammatic representation of the mean percentages of sig-niWcantly diVerent variables from paired comparisons, by Mann–Whitney U test, in each group of populations (a) and betweengroups of populations (b). Malta (Mal), South-eastern Sicily (SESic), Tunisia (Tun), Western Sicily (W Sic)

a)

9 % W Sic

27 %

Mal

43 %29 %

SE SicTun

b)

71 %

58 %

67 %SE Sic/Tun

SE Sic/Mal

W Sic/Mal62 %

48 %W Sic/Tun

39 %

Mal/TunSE Sic/W Sic

Fig. 4 Results of PCA: a PC-scores of the nine populations; b cor-relations of the diVerent morphometric characters with PC1 andPC2; the variables highly correlated with PC1 and PC2 are in bold.South-eastern Sicily: Foce Marcellino (FM), Pantano Longarini(LO), Pantano Viruca (PV); Western Sicily: Salina Chiusa Trapani(TP), Salina Curto Marsala (SMA); Malta: Salina (SAL), Ghadira(GHA); Tunisia: Ghar El Milh (GEM), Lake Tunisi South (LTS).For the code of the variables see Table 2

TP TPTP

TPTP TPTP

TP

TP

TP

TP

TPTPTP

TP

TPTPTP

TPSMASMASMA

SMASMA SMA

SMA

SMA

SMASMASMA SMA

SMA

SMASMA

SMA

LO

LO

LO

LO LO

LOLO

LO

LO

LO

LOLO

LO

LO

LO LO

LO

LO

FM

FMFM

FM

FM

FM

FMFM

FM

FM

FMFM

FM

FMFM

FMPV

PV

PV

PV

PVPV

PVPV

PV

PVPV

PV

PV

PV PV

PV

PV

PV

PVPV

PV

GHAGHA

GHAGHAGHA

GHAGHAGHAGHAGHA

GHAGHA

GHA

GHAGHA

GHA

GHAGHA

SALSAL

SALSAL

SAL

SAL

SALSALSAL

SALSALSAL

SALSAL

SALSAL

SALSALSALGEM

GEM

GEM

GEM

GEMGEMGEM

GEM

GEM

GEMGEMGEMGEMGEMGEM

GEM

GEMGEM

GEM

GEMGEMLTS

LTS

LTSLTS LTSLTSLTS

LTSLTS

LTSLTS

LTSLTS

LTSLTSLTS

LTS LTS

-14,0 -12,0 -10,0 -8,0 -6,0 -4,0 -2,0 0,0 2,0 4,0 6,0 8,0PC1

-8,00

-6,00

-4,00

-2,00

0,00

2,00

4,00

6,00

8,00

10,00

PC2

PV

SMA+TPGHA+SALGEM+LTS

LO+FM

-1,0 -0,5 0,0 0,5 1,0PC1

-1,0

-0,5

0,0

0,5

1,0

PC

2

PMXb1

MXb

PRSb

MXR

SPCv

URcSPCe

SPCa

PRSa

BSbc

PMXR

MXh

BB1b1BB2b2

URb VRy

BSn

CB5aBSbo

PMXh1

URhVRx

Ls

SPCb CB5y CB5x

a)

b)

ranges from 3 (Pantano Viruca vs. Pantano Longarinia) to 20 (Salina Curto Marsala vs. Pantano Longarini a)(Table 4). As regards the percentage of divergence, theanalysis within the four groups indicates that the south-eastern Sicily populations are the most diVerentiated(percentage of divergence: minimum 0.5% Pantano Vir-uca and Pantano Longarini, maximum 5% Foce Marcel-lino and Pantano Viruca, and Foce Marcellino andPantano Longarini); within the other three groups, thevalues of divergence range from 0 to 0.7% among thewestern Sicilian and the Tunisian populations, and from0 to 0.5% for Malta. Among the groups, the smallestdiVerentiation was for western Sicily and Malta (percent-age of divergence between 0 and 0.5%), while the greatestdivergence was for the south-eastern Sicilian group andTunisia (from 1.8 to 5.3%) (Table 5). The NJ and MPmethods gave similar phylogenetic trees (Fig. 7). There isa clear separation of the two Sicilian populations fromPantano Longarini and Pantano Viruca from all theother populations. The second node isolates the thirdpopulation from south-eastern Sicily, that of Foce Mar-cellino, which is completely detached from the clusterformed by the western Sicily/Malta/Tunisia populations;within this last group there is a separation of the Tuni-sian populations, while the populations of western Sicilyand Malta are not diVerentiated.

The Nested Clade Analysis (Templeton et al. 1995;Templeton 1998, 2004) (data not shown) conWrms theresults of the phylogenetic analysis. In fact, the PantanoLongarini and the Pantano Viruca haplotypes were notinserted in the network due to their high degree of

divergence from the other haplotypes. From the nestedcladogram constructed using data from the remainingkilliWsh populations, only two clades were signiWcantlyseparated from the rest (permutational chi-square proba-bility value less than 0.05), revealing continuous rangeexpansion ‘for the haplotypes from the Maltese, Tunisianand western Sicilian populations and allopatric fragmen-tation’ of these with respect to the Foce Marcellino popu-lation. These Wndings should be further substantiated bya more detailed study of killiWsh populations along thewhole Sicilian coast, although the high genetic divergenceof the Pantano Longarini and Pantano Viruca populations,located geographically close together, is further proof ofthe eVective genetic isolation of the Foce Marcellino popu-lation. Mantel’s test indicates a signiWcant correlationbetween the molecular variability and the morphology ofthe populations examined (G=2.423, P=0.01).

Discussion

Molecular and osteological studies of populations ofA. fasciatus from the central Mediterranean have dem-onstrated a high degree of diVerentiation in the south-eastern Sicilian populations both within this group ofpopulations (Tigano et al. 2004a), and from three othergroups of populations from western Sicily, Tunisia andMalta (this paper). Comparing the dendrograms result-ing from the molecular and morphological analyses, itcan be observed that both analyses clearly diVerentiatethe south-eastern Sicilian populations from the others,even if the topologies of the two dendrograms are diVer-ent. This discrepancy, however, does not appear to besigniWcant, and Mantel’s test indicates a strong correla-tion between the two distance matrices based on molecu-lar and osteological characters.

The observed diVerentiation of the killiWsh popula-tions studied can be partly explained in terms of the pal-aeogeographic history of the Mediterranean basin,especially in relation to the evolution of the coastal envi-ronments where the species examined here lives. Thegenus Aphanius seems to be of Miocenic origin and stud-ies by Hrbek and Meyer (2003) support a predominantlyvicariance-based speciation model for this genus, associ-ated with the closing of the Tethys Sea. During the Mes-sinian A. crassicaudus lived in the Mediterranean basin(Gaudant 2002); however, the exact period when A. fas-ciatus Wrst appeared is unknown, nor is it known if anancestral A. pliocenii existed (Gaudant in litteris). More-over, Hrbek and Meyer (2003) attest that A. fasciatus isthe only species of the genus that does not Wt the hypoth-esis of Messinian vicariant diVerentiation. During theearly Pleistocene, the Mediterranean did not form asingle marine unit but was fragmented into smaller east-ern and western basins due to desiccation (Hsu 1972,1983; Krijgsman et al. 1999); the ongoing westwarddiVerentiation of Aphanius was most likely driven by thisgeographic isolation. It is probable that A. fasciatus

Table 3 Statistical tests on PC scores: (a) test for diVerences in PC1scores of individuals from nine populations of A. fasciatus; (b) testfor diVerences in PC2 scores (statistical signiWcance Mann–WhitneyU test)

For the abbreviations of the populations, see Table 1n.s. not signiWcant*P<0.05**P< 0.01***P< 0.001

TP SMA LO FM PV GHA SAL GEM LTS

(a)TP –SMA n.s.LO *** *** –FM *** *** *PV *** *** n.s. n.s. –GHA *** ** *** *** *** –SAL *** *** *** *** ** * –GEM *** *** *** *** ** n.s. n.s. –LTS *** *** *** *** ** * n.s. n.s. –(b)TP –SMA n.s. –LO n.s. n.s. –FM n.s. n.s. n.s. –PV *** *** *** *** –GHA * ** ** n.s. *** –SAL n.s. * ** n.s. *** n.s. –GEM n.s. n.s. n.s. n.s. *** * ** –LTS n.s. n.s. * n.s. *** * n.s. n.s. –

already became separated from the main ancestral popu-lation before the division of the Mediterranean Sea intotwo basins was complete. During the Pliocene (from 5 to1.7 million years ago) Sicily consisted of two islands sep-arated from the nearby land masses: one, to the north,that included northern Sicily (equivalent to today’s Sicil-ian provinces of Trapani, Palermo and Messina) and asmaller one that comprised the present Hyblean region(equivalent to today’s Sicilian provinces of Ragusa andSyracuse) (Blanc 1942; Pasa 1953; La Greca 1957); theisolation of the Hyblean region was only interrupted forbrief periods through a connection with the island ofMalta during marine regressions in the Pleistocene, fol-lowing the last of which, all connections to the landmasses around Sicily were deWnitively interrupted.

Therefore, the pronounced diVerentiation of the threekilliWsh populations from south-eastern Sicily from theother populations studied here, can be explained in termsof the relatively long period of isolation of the Hybleanregion. Moreover, the data presented here and thatderived from the molecular substitution rate generallyaccepted for the mitochondrial control region of Wsh(Stepien et al. 2001) suggest that the Pantano Viruca andPantano Longarini populations from south-eastern Sic-ily could be of Pliocenic origin. This marked diVerentia-tion of the south-eastern Sicilian group of killiWshpopulations is indicated by the high percentage ofsequence divergence (ranging from 1.5 to 5.3%) withrespect to the other three groups of killiWsh populations(Table 5). The range of molecular clock rates generally

Fig. 5 Box plots showing the median (Wlled square), the 25th and 75th percentiles ( ) and the data range ( ) of the variables highly correlated wiith PC1 and PC2. For the abbreviation of the populations see Table 1 and for the code of the variables see Table 2

MXb

2,4

2,8

3,2

3,6

4,0

4,4

4,8

5,2

TP SMA LO FM PV GHA SAL GEM LTS

PMXb1

2,2

2,6

3,0

3,4

3,8

4,2

4,6

TP SMA LO FM PV GHA SAL GEM LTS

PRSa

0

1

2

3

4

5

6

7

8

9

TP SMA LO FM PV GHA SAL GEM LTS

PRSb

1

2

3

4

5

6

7

8

9

TP SMA LO FM PV GHA SAL GEM LTS

BB1b1

0,4

0,5

0,6

0,7

0,8

0,9

1,0

1,1

1,2

1,3

TP SMA LO FM PV GHA SAL GEM LTS

BB2b2

0,6

0,8

1,0

1,2

1,4

1,6

1,8

TP SMA LO FM PV GHA SAL GEM LTS

URb

2,0

2,4

2,8

3,2

3,6

4,0

4,4

4,8

TP SMA LO FM PV GHA SAL GEM LTS

VRy

0,2

0,6

1,0

1,4

1,8

2,2

2,6

TP SMA LO FM PV GHA SAL GEM LTS

accepted for the control region of Wsh (from 2 to 10%sequence divergence per million years; Stepien et al.2001) suggests that the period of isolation of the Hybleanpopulations took place between 150,000 (fast clock cali-bration) and 2,650,000 years (slow clock calibration) ago.The Pantano Viruca and Pantano Longarini populationshave the highest sequence divergence (between 3.4 and5.3%) while the third south-eastern Sicilian population(Foce Marcellino) has a sequence divergence rangingbetween 1.5 and 3.1% with respect to all the other popu-lations examined; furthermore, within the south-easternSicilian group, a high divergence between the PantanoViruca and Pantano Longarini populations and thatfrom Foce Marcellino (5% sequence divergence with apopulation separation time between 500,000 and2,500,000 years) is observed; this can be interpreted inthe light of palaeogeographic events that took place inthe Hyblean region in the Lower Pleistocene (about 1.7million years ago) (Tigano et al. 2004a). Analogously, theonly diVerentiated population of A. fasciatus included inthe phylogenetic study by Hrbek and Meyer (2003) wasthat from Lake Bafa in Turkey, with an average 6.86%sequence divergence from other A. fasciatus populations,corresponding to a 3.99 million year separation.

The use of haplotype networks allows a Wne discrimi-nation of biological patterns as this technique examinesspatial/temporal patterns of genetic variation (Temple-ton 1998). The values of permutational chi-square prob-ability obtained in our study allow us to reject the nullhypothesis of no association between haplotype varia-tion and geography for the two clades that show signiW-cant geographical structure caused by “ContiguousRange Expansion” for the haplotypes belonging to thepopulations from Malta, Tunisia and western Sicily andby “Allopatric Fragmentation” with respect to thesouth-eastern Sicilian population from Foce Marcellino(the Nested Clade Analysis did not allow us to includethe haplotypes from Pantano Viruca and Pantano Lon-garini within the network because of their high diver-gence). The bulk of the specimens from Malta andWestern Sicily exhbit the h1 and h2 haplotypes that

would be considered ancestral for their internal positionin the nested cladogram and several Tunisian specimensexhibit the k1 and k2 haplotype (tip position) that seemto be derived from the former. This haplotype patterncould be related to past Pleistocenic climatic events thatcaused notable changes in coastlines as a consequence ofmarine regression and transgression. One interestinghypothesis put forward by some authors (e.g. Aprile1998) is the possible existence during the last Ice Age ofan archipelago between Sicily and the North Africancoasts which could have promoted the range expansionof an ancestral population of A. fasciatus. Bathymetricdata indicate a continuous submarine platform up to200 m deep running from northern Tunisia to westernSicily across the Adventure Bank, parallel to the south-ern Sicilian coast, which borders the westernmost marginof the Hyblean–Maltese platform and which extendssouth to the Medina Bank; this whole area experiencedperiodic, alternate emersion–immersion events duringthe latest Quaternary tectonic movements (Finetti andMorelli 1973; Ogniben et al. 1975; Ben-Avraham andGrasso 1990). The low sequence divergence values (0–0.7%) between the populations of western Sicily andMalta are also reported by Hrbek and Meyer (2003),who employed other mitochondrial genes.

With regards to morphological diVerences in the killi-Wsh populations studied, it is known that environmentalfactors may play a decisive role in intraspeciWc diVerentia-tion. Studies by Murta (2000) showed that morphometriccharacters are more susceptible to environmental varia-tions than meristic characters. IntraspeciWc variation ofmorphometric and meristic characters was assessed inrelation to the salinity of the environment in A. anatoliae(Leidenfrost) by Grimm (1979) and in A. fasciatus byBoumaiza et al. (1981); in addition, analysis of morpho-metric data discriminated between the genetically diver-gent Atlantic and Mediterranean populations of A. iberus(Valenciennes) now generally regarded as two separatespecies: A. iberus in the Mediterranean and A. baeticus inthe Atlantic (Doadrio et al. 2002). Moreover, intraspeciWcmorphological variation of several species of teleosts wasfound to be related to the exploitation of food resourcesand habitat, and to predation pressure (Huysseune et al.1994; Milano et al. 2002; Maltagliati et al. 2003).Although ecological studies of the diVerent habitats sam-pled in this study were not made, it is evident that at leasttwo environmental factors—salinity (hyperhaline waters,marine water with various contributions of freshwater)and the size of the water body (small coastal ponds, salts-works and a large coastal lake)—diVered greatly betweenall the sites sampled.

The observed osteological diVerences in the popula-tions of A. fasciatus studied here may be explained interms of the geographic isolation of the various popula-tions, although the inXuence of environmental factors,which diVer greatly between the various sites, cannot beruled out; therefore, further studies are necessary to testhow much of the observed osteological diVerentiationcould be due to phenotypic plasticity. The slight discrep-

Fig. 6 UPGMA dendogram of the generalised distance (X axis).For the abbreviation of the populations, see Table 1

D

PV

FM

LO

LTS

SAL

GHA

GEM

SMA

TP

0 100 200 300 400 500 600 700

Tab

le4

Sequ

ence

s re

lati

ve to

the

regi

ons

of m

itoc

hond

rial

con

trol

of t

he in

divi

dual

s of

the

pop

ulat

ions

exa

min

ed

For

the

abbr

evia

tion

s of

the

popu

lati

ons,

see

Tab

le1

a T

he p

osit

ion

refe

rs to

the

sequ

ence

of t

he h

aplo

type

a

Pop

ulat

ion

Hap

loty

pe

(Num

ber

of

indi

vidu

als)

Nuc

leot

ide

posi

tion

in th

e m

itoc

hond

rial

con

trol

reg

iona

1213

1729

3096

107

109

113

127

137

138

142

143

166

177

184

188

190

191

218

223

224

227

255

256

257

259

260

286

342

369

LO

a (5

)T

TT

AA

GC

AG

TA

TA

TC

GA

TA

TC

AG

TA

TT

TC

AG

Tb

(4)

CA

GC

Cb1

(1)

–C

AG

CC

PV

c (9

)C

GC

d (1

)G

CF

Me

(7)

AC

CA

TC

CC

AA

TA

GC

AC

f (1

)A

CC

AT

AC

CC

AA

TA

GC

AC

g (1

)A

CC

TC

CC

AA

TA

GC

AC

SAL

h1 (6

)A

––

AT

CC

AA

TA

GA

CC

Ai (

2)A

––

AT

CC

AA

TA

GA

CC

CA

GH

Ah1

(3)

A–

–A

TC

CA

AT

AG

AC

CA

h2 (2

)A

AT

CC

AA

TA

GA

CC

Aj (

2)A

AT

CC

GA

AT

AG

AC

CA

LT

Sk1

(4)

AA

TT

CC

AA

TA

GA

CC

Ak2

(2)

A–

AT

TC

CA

AT

AG

AC

CA

l (1)

AA

TT

CC

AA

TA

GA

CC

CC

TA

m (1

)A

AT

TC

CG

AT

AG

AC

CA

n (1

)A

AT

TC

CA

AT

AG

AC

–C

Ao

(1)

AA

TT

CC

AA

TA

GA

CA

GE

Mk1

(7)

AA

TT

CC

AA

TA

GA

CC

Ap

(2)

AA

TT

CC

AA

TA

GA

CC

CC

AT

Ph2

(7)

AA

TC

CA

AT

AG

AC

CA

q1 (3

)A

AT

CC

TA

AT

AG

AC

CA

SMA

h2 (5

)A

AT

CC

AA

TA

GA

CC

Ar

(1)

AA

TC

CA

AT

AT

GA

CC

As

(1)

AA

TC

CA

AT

AG

CC

CA

t (1

)A

AT

CC

AA

TA

GA

CA

q2 (1

)A

–A

TC

CT

AA

TA

GA

CC

A

ancy between the two dendrograms resulting from themolecular and osteological analyses may be attributed tothe fact that morphological variation is occasionally alocal response to diVerent environments, reXecting onlyin part genetic diVerences (Jerry and Cairns 1998).

Finally, the results relative to the Pantano Virucapopulation from south-eastern Sicily are very interestingin that both morphological and molecular data are inconcordance that this population is unique; this popula-tion is diVerentiated by the shape of the parasphenoid as

well as being characterised by a haplotype present in nineout of ten specimens examined, which could be the con-sequence of the founder eVect.

In conclusion, the wide range of the sequence diver-gence (ranging between 0 and 5.3%), indicates that, insome cases, there is a high degree of diVerentiationbetween the killiWsh populations of south-eastern Sicilyand the other populations studied; moreover, diVerentia-tion is also evident on a more local scale (e.g. between thepopulations from Pantano Longarini and Pantano Vir-uca and that from Foce Marcellino). In other cases(Malta and western Sicily), the molecular analysis indi-cates an absence of diVerentiation. The allozyme datareported by Maltagliati (1998a, 1999, 2002), Maltagliatiet al. (2003), and Cimmaruta et al. (2003) show a highdegree of genetic divergence between thirty Italian popu-lations of A. fasciatus, while, on a more local scale, theauthors demonstrate the existence of a genetic structureof these killiWsh populations resulting from occasionalgene Xow, and from evolutionary forces that are stochas-tic (drift) or deterministic (natural selection), which acttogether to shape the gene pool of the species. Analo-gously, allozymic and morphological studies by Ferritoet al. (2003) indicate well-supported diVerentiationbetween two Adriatic and two Sicilian populations ofA. fasciatus. On the other hand, the molecular data ofHrbek and Meyer (2003) revealed only a limited geneticdiVerentiation in seven populations of A. fasciatus and,consequently, a limited genetic structuring of this species.Therefore, the genetic structuring of A. fasciatus stillappears to be controversial. Data from the present studyand from the analysis of the mitochondrial DNA byCostagliola et al. (2003) and by Tigano et al. (2004b) indi-cate that at least 41% of the studied populations are welldiVerentiated; taking into consideration the ample distri-bution of this species, it is probable that this percentagecould be higher. Therefore, the results reported here,while interesting, indicate the need for more extensivesampling and comparison of populations. Nevertheless,these results, together with those already reported in theliterature, provide a valid knowledge base for theunderstanding of the micro-evolutionary processes actingin A. fasciatus populations and for the implementation ofmanagement plans aimed at conserving the adaptive

Table 5 Riassuntive table of matrices of the pairwise distances for the mitochondrial control regions (the values of diVerence are inpercentages)

The maximum and minimum values are shown. For the abbreviations of the populations see Table 1

LO PV FM TP SMA LTS GEM GHA SAL

Sicily south-east LOPV 0.5–0.7FM 3.9–5.0 4.2–5.0

Sicily west TP 3.7–4.5 3.9–4.4 1.8–2.3SMA 3.4–4.5 3.7–4.5 1.5–2.3 0.0–0.7

Tunisia LTS 3.7–5.3 3.9–5.0 1.8–3.1 0.2–1.3 0.2–1.5GEM 3.9–5.0 4.2–5.0 2.1–2.9 0.2–1.0 0.2–1.3 0.0–0.7

Malta GHA 3.7–4.5 3.9–4.4 1.8–2.3 0.0–0.2 0.0–0.5 0.2–1.3 0.2–1.0SAL 3.7–4.5 3.9–4.5 1.8–2.3 0.0–0.5 0.0–0.5 0.2–1.3 0.2–1.0 0.0–0.5

Fig. 7 Tree obtained both with the neighbour-joining and maxi-mum parsimony methods. Values above branches represent NJbootstrap estimates, those below branches represent MP estimates,based on 1,000 replications in both cases (>50%). For the abbrevia-tion of the populations, see Table 1

b (LO 4)

b1 (LO 1)

c (PV 9)

d (PV 1)

a (LO 5)

h1 (SAL 6; GHA 3)

h2 (TP 7; SMA 5; GHA 2)

q2 (SMA 1)

q1 (TP 3)

j (GHA 2)

r (SMA 1)

i (SAL 2)

m (LTS 1)

p (GEM 2)

l (LTS 1)

o (LTS 1)

k1 (GEM 7; LTS 4)

n (LTS 1)

k2 (LTS 2)

s (SMA 1)

t (SMA 1)

e (FM 7)

f (FM 1)

g (FM 1)

A. danfordii

51

52

52

64

64

100100

7175

5857 80

91

96

6174

5363

82

95

variability in this species, some populations of which arein danger of extinction.

Acknowledgements We wish to thank Prof. G. Lunetta and Dr. A.Mazza from the Dipartimento di Economia e Territorio, Universityof Catania, Italy, for their valuable suggestions about statisticalanalysis. Prof. M. Grasso from the Dipartimento di Geologia, Uni-versity of Catania, for his valuable suggestions about the interpreta-tion of the palaeogeographic events in the Central Mediterranean,and Prof. S. Nouira and Dr. A. Santulli for their help in the collectionof samples in Tunis and in Trapani, respectively. We thank the Envi-ronment Protection Directorate of the Malta Environment andPlanning Authority for permission to work on this protected speciesin Malta and for access to the Ghadira Bird Sanctuary; we are grate-ful to Mr Charles Gauci, the warden in charge of the sanctuary, forfacilitating our work.

References

Aprile LAA (1998) La diVusione del genere Homo in Eurasia e il piùantico popolamento extra africano: vie di accesso al continenteEuropeo, con particolare riferimento a un ipotetico “arcipelago”tunisino-siculo. Degree thesis, University of Catania (unpub-lished)

Avise JC (2000) Phylogeography: the history and formation of spe-cies. Harvard University Press, Cambridge, MA

Ben-Avraham Z, Grasso M (1990) Collisional zone segmentation inSicily and surrounding areas in the Central Mediterranean. AnnTectonicae 4(2):131–139

Bernardi G (2000) Barriers to gene Xow in Embiotoca jacksoni, a ma-rine Wsh lacking a pelagic larval stage. Evolution 54:226–237

Blanc AC (1942) Variazioni climatiche ed oscillazioni della linea diriva nel Mediterraneo centrale durante l’Era Glaciale. Geol DerMere Binnengewasset 5:137–219

Boumaiza M, Ktari MH, Quignard JP (1981) Etude de la variabilitédu nombre de vertébres et d’écailles sur la ligne latérale chezAphanius fasciatus Nardo, 1827 (Poisson Cyprinodontidae).Rapp Comm Int Explor Sc Mer Médit 27(5):115–117

Cimmaruta R, Scialanca F, Luccioli F, Nascetti G (2003) Geneticdiversity and environmental stress in Italian populations of theCyprinodont Wsh Aphanius fasciatus. Oceanol Acta 26:101–110

Clement MD, Posada D, Crandall KA (2000) TCS: a computer pro-gram to estimate gene genealogies. Mol Ecol 9:1656–1660

Comparini A, Scattolin N, Rodinò E (1984) Genetic diVerentiationamong some populations of the Cyprinodont Aphanius fasciatusNardo. Nova Thalassia 6:261–268

Costagliola D, Ferrito V, Rocco L, Marsilio A, Pappalardo AM,Tigano C (2003) Analisi dell’mtDNA e RAPD in popolazioniitaliane di Lebias fasciata (Teleostei, Cyprinodontiformes). Attidel 64° Congresso U.Z.I., Varese 21–25 Settembre 2003, p 105

Dingerkus G, Uhler LD (1977) Enzyme clearing of alcian bluestained whole small vertebrates for demonstration of cartilage.Stain Technol 52:229–232

Doadrio I, Carmona JA, Fernandez-Delgado C (2002) Morphomet-ric study of the Iberian Aphanius (Actinopterygii, Cyprinodonti-formes), with description of a new species. Folia Zool 51:67–79

Duhan JB, Minckley WL (1998) Allozymic variation in desert pup-Wsh from natural and artiWcial habitats: genetic conservation inXuctuating populations. Biol Conserv 84:7–15

Duvernell DD, Turner BJ (1999) Variation and divergence of DeathValley pupWsh populations at retrotransposon-deWned loci. MolBiol Evol 16:363–371

Echelle AA, Echelle AF, Edds DR (1987) Population structure offour pupWsh species (Cyprinodontidae: Cyprinodon) from theChihuahuan Desert region of New Mexico and Texas: allozymicvariation. Copeia 3:668–681

Echelle AA, Van Den Bussche RA, Malloy TP Jr, Haynie ML,Minckley CO (2000) Mitochondrial DNA variation in pupWshesassigned to the species Cyprinodon macularius (Atherinomorpha:

Cyprinodontidae): taxonomic implications and conservationgenetics. Copeia 2:353–364

Elliott NG, Haskard K, Koslow JA (1995) Morphometric analysisof orange roughy (Hoplostethus atlanticus) oV the continentalslope of southern Australia. J Fish Biol 46:202–220

Fernandez-Pedrosa V, Gonzalez A, Planelles M, Moya A, Latorre A(1995) Mitochondrial DNA variability in three Mediterraneanpopulations of Aphanius iberus. Biol Conserv 72:251–256

Ferrito V, La Paglia L, Sgarlata MT, Mauceri A, Tigano C (2000)Preliminary remarks on NOR location in a Sicilian populationof Lebias fasciata (Teleostei, Cyprinodontidae). Ital J Zool67:269–272

Ferrito V, Maltagliati F, Mauceri A, Adorno A, Tigano C (2003)Morphological and genetic variation in four Italian populationsof Lebias fasciata (Teleostei, Cyprinodontidae). Ital J Zool70:115–121

Finetti I, Morelli C (1973) Geophisical exploration of the Mediter-ranean Sea. Boll GeoWs Teor Appl 15:263–341

Garciamarin JL, Vila A, Pla C (1990) Genetic-variation in the Ibe-rian toothcarp, Aphanius iberus (Cuvier and Valenciennes). JFish Biol 37:233–234

Gaudant J (2002) La crise messinienne et ses eVets sur l’ichthyofaunenéogène de la Meéditerranée: le témoignage des squelettes enconnexion de poissons téléostéens. Geodiversitas 24:691–710

Grimm VH (1979) Veränderungen in der variabilität von populatio-nen des zahnkarpfens Aphanius anatoliae (Leidenfrost, 1912)während 30 jahren: 1943–1974. Sonderdruck aus Z.f.zool. Sys-tematik u. Evolutionsforschung 17(4):272–280

Hsu KJ (1972) When the Mediterranean dried up. Sci Am 227(6):27–36

Hsu KJ (1983) The Mediterranean was a desert. Princeton Univer-sity Press, Princeton, NJ

Hrbek T, Meyer A (2003) Closing of the Tethys sea and the phylog-eny of Eurasian killiWshes (Cyprinodontiformes: Cyprinodonti-dae). J Evol Biol 16:17–36

Huysseune A, Sire JY, Meunier FJ (1994) Comparative study oflower pharyngeal jaw structure in two phenotypes of Astatore-ochromis alluaudi (Teleostei: Cichlidae). J Morph 221:25–43

JeVreys AJ, Flavell RA (1977) A physical map of the DNA Xankingthe rabbit ß-globin gene. Cell 12:429–439

Jerry DR, Cairns SC (1998) Morphological variation in the catadro-mous Australian bass, from seven geographically distinct river-ine drainages. J Fish Biol 52:829–843

Kimura M (1980) A simple method for estimating rate of base sub-stitution through comparative studies of nucleotide sequences. JMol Evol 16:111–120

Kimura M, Weiss GH (1964) The stepping stone model of popula-tion structure and the decrease of genetic correlation with dis-tance. Genetics 49:561–576

Krijgsman W, Hilgen FJ, RaY I, Sierro FJ, Wilson DS (1999) Chro-nology, causes and progression of the Messinian salinity crisis.Nature 400:652–655

La Greca M (1957) Considerazioni sull’origine della fauna siciliana.Boll Zool 24:593–631

Lussen A, Falk TM, Villwock W (2003) Phylogenetic patterns inpopulations of Chilean species of the genus Orestias (Teleostei:Cyprinodontidae): results of mitochondrial DNA analysis. MolPhylogenet Evol 29:151–160

Maltagliati F (1998a) A preliminary investigation of allozyme genet-ic variation and population geographical structure in Aphaniusfasciatus from Italian brackish-water habitats. J Fish Biol52:1130–1140

Maltagliati F (1998b) Does the Mediterranean killiWsh Aphaniusfasciatus (Teleostei, Cyprinodontidae) Wt the one-dimensionalstepping stone model of gene Xow? Environ Biol Fishes 53:385–392

Maltagliati F (1999) Genetic divergence in natural populations ofthe Mediterranean brackish-water killiWsh Aphanius fasciatus.Mar Ecol Prog Ser 179:155–162

Maltagliati F (2002) Genetic monitoring of brackish-water popula-tions. The mediterranean toothcarp Aphanius fasciatus (Cypri-nodontidae) as a model. Mar Ecol Prog Ser 235:257–262

Maltagliati F, Domenici P, Fosch CF, Cossu P, Casu M, Castelli A(2003) Small-scale morphological and genetic diVerentiation inthe Mediterranean killiWsh Aphanius fasciatus (Cyprinodonti-dae) from a coastal brackish-water pond and an adjacent pool innorthern Sardinia. Oceanol Acta 26:111–119

Milano D, Cussac VE, Macchi PJ, Ruzzante DE, Alonso MF, Vigli-ano PH, Denegri MA (2002). Predator associated morphology inGalaxias platei in Pantagonian lakes. J Fish Biol 61:138–156

Meyer A, Kocher TD, Basasibwaki P, Wilson AC (1990) Monophy-letic origin of Lake Victoria Wshes suggested by mitochondrialDNA sequences. Nature 347:550–553

Murta AG (2000) Morphological variation of horse mackerel (Tra-churus trachurus) in the Iberian and North African Atlantic:implications for stock identiWcation. ICES J Mar Sci 57:1240–1248

Ogniben L, Parotto M, Praturlon A (eds) (1975) Structural model ofItaly. Quaderni La Ricerca ScientiWca 90:502

Parenti LR, Tigano C (1993) Polymorphic skeletal characters inAphanius fasciatus (Teleostei, Cyprinodontiformes). Copeia4:1132–1137

Pasa A (1953) Appunti geologici per la paleogeograWa delle Puglie.Mem Biogeogr Adr 2:175–286

Perdices A, Carmona JA, Fernandez-Delgado C, Doadrio I (2001)Nuclear and mitochondrial data reveal high genetic divergenceamong Atlantic and Mediterranean populations of the IberiankilliWsh Aphanius iberus (Teleostei: Cyprinodontidae). Heredity87:314–324

Posada D, Crandall KA, Templeton AR (2000) GeoDis: a programfor the cladistic nested analysis of the geographical distributionof genetic haplotypes. Mol Ecol 9:487–488

Reichenbacher B, Sienknecht U (2002) Allopatric divergence andgenetic diversity of recent Aphanius iberus and fossil Prolebiasmeyeri (Teleostei, Cyprinodontidae) from southwest and west-ern Europe, as indicated by otoliths. Geobios 34:69–83

Slatkin M (1994) Gene Xow and population structure. In: Real LA(ed) Ecological Genetics. Princeton University Press, Princeton,NJ, pp 3–17

Stepien CA, Rosenblatt RH, Bargmeyer BA (2001) Phylogeographyof the spotted sand bass, Paralabrax maculatofasciatus: diver-gence of gulf of California and PaciWc coasts populations. Evo-lution 55:1852–1862

Strecker U, Meyer CG, Sturmabauer C, Wilkens H (1996) Geneticdivergence and speciation in an extremely young species Xock inMexico formed by the genus Cyprinodon (Cyprinodontidae, Te-leostei). Mol Phylogenet Evol 6:143–149

SwoVord DL (1998) PAUP*. Phylogenetic Analysis Using Parsi-mony (* and other methods.) version 4. Sinauer Associates, Sun-derland, MA

Templeton AR (1998) Nested clade analyses of phylogeographic da-ta: Testing hypotheses about gene Xow and population history.Mol Ecol 7:381–397

Templeton AR (2004) Using haplotype trees for phylogeographicand species inference in Wsh populations. Envir Biol Fishes69:7–20

Templeton AR, Sing CF (1993) A cladistic analysis of phenotypicassociations with haplotypes inferred from restriction endonu-clease mapping. IV. Nested analyses with cladogram uncertaintyand recombination. Genetics 134:659–669

Templeton AR, Crandall KA, Sing CF (1992) A cladistic analysis ofphenotypic associations with haplotypes inferred from restric-tion endonuclease mapping and DNA sequence data. III clado-gram estimation. Genetics 132:619–633

Templeton AR, Routman EJ, Phillips CA (1995) Separating popu-lation structure from population history: A cladistic analysis ofthe geographical distribution of mitochondrial DNA haplotypesin the tiger salamander, Ambystoma tigrinum. Genetics 140:767–782

Thompson JD, Higgins DG, Gibson TG (1994) CLUSTAL W:improving the sensitivity of progressive multiple sequence align-ment through sequence weighting position-speciWc gap penaltiesand weight matrix choice. Nucleic Acids Res 22:4673–4680

Thorpe RS (1976) Biometric analysis of geographic variation andracial aYnities. Biol Rev 51:407–452

Tigano C (1991) Il polimorWsmo della mascella superiore in Apha-nius fasciatus Nardo (Pisces Cyprinodontidae). Animalia 18:61–70

Tigano C, Ferrito V (1985) Studio osteologico comparato del craniodi popolazioni di Aphanius fasciatus (Nardo 1827) (Pisces: Cyp-rinodontidae) dell’Adriatico e di Wumi di Sicilia. Animalia 12:13–57

Tigano C, Parenti LR (1988) Homology of the median ethmoid ossi-Wcations of Aphanius fasciatus and other Atherinomorph Wshes.Copeia 4:866–870

Tigano C, Ferrito V, Nicosia R (1999) Morphological analysis of thepharyngeal jaws in two populations of Lebias fasciata Valenci-ennes, 1821 (Teleostei, Cyprinodontidae). J Morphol 241:107–114

Tigano C, Ferrito V, Adorno A, Mannino MC, Mauceri A (2001)Pharyngeal and oral jaw diVerentiation in Wve populations of Le-bias fasciata (Teleostei: Cyprinodontidae). Ital J Zool 68:201–206

Tigano C, Rocco L, Ferrito V, Costagliola D, Pappalardo AM, Stin-go V (2003) Chromosomal mapping and nucleotide sequences ofribosomal genes in Lebias fasciata (Teleostei, Cyprinodontidae).Genetica 121:95–100

Tigano C, Canapa A, Ferrito V, Barucca M, Arcidiacono I, Olmo E(2004a) Morphological and molecular analysis of three Sicilianpopulations of Lebias fasciata (Teleostei, Cyprinodontidae). ItalJ Zool 71:1125–1133

Tigano C, Pappalardo AM, Ferrito V, Messina A, De Pinto V(2004b) IntraspeciWc variation in killiWsh Aphanius fasciatusNardo (Teleostei, Cyprinodontidae): evidence from moleculesand morphology. In: Saat T (ed) Abstract Volume XI EuropeanCongress of Ichthyology, Tallin, Estonia. Estonia Marine Insti-tute Report series 12:128

Torralva M, Oliva Paterna FJ, Andreu A, Garcia Mellado A, Mi-nano PA, Cardozo V, Garcia Alonso J, Fernandez Delgado C(2001) Distribution and conservation status of Aphanius iberus(Valenciennes, 1846) in Murcia region (S.E. of Iberian Peninsula.An Biol 23:63–83

Villwock W (1976) A contribution to the understanding of the evo-lution of the meristic characters, with special reference to OldWorld Cyprinodontids. Abh NatwissVer Hambg 18/19:11–27

Vitturi R, Catalano E, Colomba MS, Montanino L, Pellerito L(1995) Karyotype analysis of Aphanius fasciatus (Pisces, Cyprin-odontiformes): Ag-NORs and C-band polymorphism in fourpopulations from Siciliy. Biol Zentralbl 114:329–402

Wildekamp RH (1993) The genus Aphanius Nardo. In: Watters BR(ed) A World of killies, atlas of the oviparous cyprinodontiformWshes of the world, vol I. American KilliWsh Association, Hisha-waka Indiana, pp 19–67

Wilde GR, Echelle AA (1997) Morphological variation in intergradepupWsh populations from the Pecos River, Texas, USA. J FishBiol 50:523–539