UvA-DARE (Digital Academic Repository)
Endemism in Sardinia: Evolution, ecology, and conservation in the butterfly Maniola nurag
Grill, A.
Publication date 2003
Link to publication
Citation for published version (APA): Grill, A. (2003). Endemism in Sardinia: Evolution, ecology, and conservation in the butterfly Maniola nurag. IBED, Universiteit van Amsterdam.
General rights It is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s) and/or copyright holder(s), other than for strictly personal, individual use, unless the work is under an open content license (like Creative Commons).
Disclaimer/Complaints regulations If you believe that digital publication of certain material infringes any of your rights or (privacy) interests, please let the Library know, stating your reasons. In case of a legitimate complaint, the Library will make the material inaccessible and/or remove it from the website. Please Ask the Library: https://uba.uva.nl/en/contact, or a letter to: Library of the University of Amsterdam, Secretariat, Singel 425, 1012 WP Amsterdam, The Netherlands. You will be contacted as soon as possible.
UvA-DARE is a service provided by the library of the University of Amsterdam (https://dare.uva.nl)
Download date:30 Sep 2021 B^PWMB-a^ip^pMM»!^p^fppq| |
differentiationn in the island endemic Maniola wi< <
withh Wil van Ginkeï, Gabriel Nève, and StephB,J.M«iken
13$ $
^-^ ^ Abstract t Inn butterflies, the distribution areas of widespread species and their endemic relativess are usually vicariant. In Sardinia, the ranges of an endemic and a widespreadd Maniola species overlap, and the two species possibly hybridise. In this paper,, we analyse patterns of genetic differentiation in Maniola nurag and Maniola jurtinajurtina from Sardinia by means of allozyme markers, compare them to mainland M.M. jurtina populations, and interpret the data with regard to the endemic species' evolutionaryy history. Sardinian M. nurag and M. jurtina have equally high levels off genetic variation (H = 0.141-0.270; H = 0.137-0.189) as mainland M. jurtina
(H(Hnainlandnainland== 0.141-0.236). Total genie diversity at fifteen polymorphic loci is mostly duee to within population variation (FIS). The close relationship of the two species is illustratedd by the fact, that 63 of the 76 alleles screened are shared by both species. Smalll genetic distance between them (Nei's D = 0.21) indicates that divergence initiatedd after the desiccation of the Mediterranean (+ 3 ma ago), and was possibly associatedd with the abrupt climate changes at the turn from Pliocene to Pleistocene (1.8-33 ma). Geographic patterns in allozyme allele frequencies hint at the existence off hybrizymes, and suggest the presence of hybrids in areas where M. nurag and M.M. jurtina are sympatric. Island populations of neither species show signs of loss off genetic diversity, inbreeding, or bottlenecks. We propose that M. nurag did not resultt from vicariance or dispersal, but originated under sympatric or parapatric conditions,, as a consequence of local adaptation along an environmental gradient.
Keywords:: Maniola nurag, butterfly, genetic population structure, hybridisation, endemic,, allozymes, inbreeding, Lepidoptera, Nymphalidae, divergence time
Introduction n AA species is usually genetically structured over space and in time. Historical founderr events, bottlenecks, and gene-flow are important evolutionary agents responsiblee for changes in the genetical structure of populations. Assessing genetic variationn across geographic areas can thus provide means to trace the history of thesee populations and eventually the history of species (Avise, 1994; Hewitt, 1999; Schmitt&Seitz,2002). .
Severall speciation models have been proposed, which are habitually characterized byy the level of gene-flow between diverging populations during the initial stages off speciation {Dobzhansky, 1940). Population divergence in the presence of gene- floww was often considered to be unrealistic. However, a number of theoretical
136 6 studiess have reported the plausibility of sympatric and parapatric speciation, and havee shown that spatially localized interactions along environmental gradients cann facilitate species' differentiation (e.g., Kondrashov & Kondrashov, 1999; Doebelii & Dieckmann, 2000; 2003). Despite this growing theoretical evidence that ecologicallyy driven speciation can occur, empirical studies showing examples for suchh speciation modes still remain scarce (Ogden & Thorpe, 2002; Scriber, 2002; Lushaii et al., 2003). Earlier sympatric speciation models involved ecologically drivenn reproductive isolation associated with adaptation to alternative resources (nichee shift), as was elegantly shown for the host races in the tephritid fly Rhagoletis pomonellapomonella (Bush, 1994) or the large cactus finch Geospiza consirostris (Grant & Grant, 1989).. Recent modelling advances suggest that competition for continuously distributedd resources, driven by sexual selection against intermediate phenotypes, couldd be the driving force for sympatric speciation (Doebeli & Dieckmann, 2003). Intermediatee phenotypes procure fewer resources as a consequence of density- andd frequency-dependent selection, and are selected against under disruptive selectionn (Turelli et al., 2001).
Hybridd zones form an ideal environment to study sympatric speciation (Arnold, 1997).. A 'hybrid zone' sensn Arnold (1997) is a geographical area where "two populationss of individuals that are distinguishable on the basis of one or more heritablee characters overlap spatially and temporally and cross to form viable and att least partially fertile offspring." In such a parapatric situation, gene-flow can sloww down or even inhibit differentiation by spreading favourable alleles across the hybridd zone (Kim & Rieseberg, 1999), whereas reinforcement can cause prezygotic reproductivee isolation (Turelli et al., 2001). Reinforcement intensifies mate preferencee (Dobzhansky & Pavlovsky, 1957; Butlin, 1995), and can lead to character displacement.. With character displacement, the differences between sympatric populationss of two species are accentuated as a result of reproductive or ecological interactionss between them (Futuyma, 1998). Although character displacement has generallyy been interpreted as an evolutionary response to secondary contact, it cann also evolve in situ across an environmental gradient, despite continuing gene- floww (Turelli et al., 2001). The existence of hybrid zones and steep genetic clines (Schilthuizenn et al, 1999, Lushai et al., 2003) shows that selection can dominate gene-floww over small spatial scales and therefore allow for parapatric divergence. Inn many hybrid zones, particular allozymes called 'hybrizymes' (Woodruff, 1989) havee been found, representing alleles that are not present or very rare in the parentall taxa, and reflect novel genetic variation (Schilthuizen & Gittenberger, 1994b;; Arntzen, 2001). Hybrid zones have been extensively investigated in plants, andd also in animals (e.g. Barton & Hewitt, 1985; Hewitt, 1988,1999; Schilthuizen &
137 7 Lombaerts/1995;; Arntzen, 2001; Capula, 2002) but only rarely so in Lepidoptera (Aagaard,, 2002; Scriber, 2002; Lushai et al, 2003).
Thee distribution areas of widespread species and their endemic relatives are usually disjunctt in butterflies (Dennis et al., 2000). In the genus Maniola (Lepidoptera, Nymphalidae),, however, the Sardinian endemic Maniola nurag (GHILIANI 1852) andd its widespread close relative, Maniola jurtina (L. 1758), are found in sympatry andd possibly hybridise (Grill et al, 2003b, 2003d). In order to find out whether thee present sympatric occurrence of the two species can be best explained under thee assumption of a sympatric, parapatric, or allopatric mode of speciation, we investigatee the population genetic structure in a number of island populations of bothh species, and compare these to continental populations of M. jurtina by means off allozyme markers. As we found ecological as well as morphological support suggestingg that M. nurag and Al jurtina possibly hybridise in Sardinia (Grill et al.,al., 2003b; 2003c), we further evaluate the probability of hybrid occurrence in the Sardiniann Maniola.
Allozymess are co-dominant markers and efficient to study population differentiationn in Lepidoptera because of the large number of polymorphic loci {e.g., Raijmannn & Menken, 2000; Nève, 2000; Schmitt & Seitz, 2002), and also provide uss with a straightforward tool to detect interspecific hybridisation (Menken & Ulenber,, 1987; Schilthuizen & Gittenberger, 1994b; Schilthuizen & Lombaerts 1995; Arntzen,, 2001; Capula 2002); diagnostic loci differentiate between species (Hewitt, 1988;; Grant & Grant, 1996; Schilthuizen et al., 1999), and consequently can reveal whetherr hybridisation takes place.
Iff the present sympatric occurrence of M. nurag and Al jurtina in Sardinia resulted fromm a sympatric or parapatric speciation event we would expect to find evidence forr reinforcement or disruptive selection on traits that are associated with the use off alternative niches (Mayr, 1963, Bush, 1969). In an early phase of differentiation, mostt alleles at polymorphic loci are still shared between the populations in similarr frequencies, and gene-flow between the diverging populations is large. Geneticc regions that are involved in differential adaptation, however, continue too diverge through selection. In later stages of differentiation, gene-flow will be furtherr reduced, and neutral alleles increasingly become subject to independent drift,, resulting in different frequencies of such alleles in the diverging populations (Tautz,, 2003b). If M. nurag evolved allopatrically, vicariant speciation would not havee led to a substantial loss of genetic variation, but founder speciation would, if thee number of founding individuals was small (Mayr, 1954). If M. jurtina invaded
138 8 Sardiniaa later on, we would similarly expect a lower level of genetic variability in thee Sardinian M. jurtina, compared to the continental M. jurtina as a consequence off a founder effect. If M. nurag was a palaeo-endemic instead of a neo-endemic, variationn levels can be high or low, depending on historical bottlenecks, speed off size reduction, and the level of variation before contraction of the distribution area. .
Background Background Thee genus Maniola is divided into one widespread species, M. jurtina, common andd abundant throughout the A estern i aiaearctic arm six largeiy vicanant speciesi ManiolaManiola telmessia largely replaces M. jurtina in southern and western Turkey, eastwardss from the Bosporus, Maniola halicarnassus flies in the Bodrum peninsula (Turkey)) and the Aegean island of Nissiros, Maniola megala occurs in southern Turkeyy and eastwards to Iran, Maniola chia is endemic to the Greek island of Chios, ManiolaManiola cypricola to Cyprus, and M. nurag to Sardinia.
Butterfliess of the genus Maniola fly in diverse, open habitats, with occasional treess and shrubs. Variation within species includes pronounced wing-pattern polymorphismm within populations, and ecological variability related to latitude andd altitude (Shreeve, 1989; Goulson, 1993). Time of emergence is well adapted to thee environmental conditions of a particular site, and in the Southern areas of the Palaearcticc females perform an imaginal diapause (aestivation) during the hottest partt of the summer, with a concomitant delayed ovarian maturation (Verity, 1953); Masettii & Scali, 1972; Garcia-Barros, 1987). Adults of M. nurag are on the wing fromm May to September depending on altitude and local weather conditions, M. jurtinajurtina flies in Sardinia from late April to June. Like many other butterflies, Maniola speciess are protandric, with males hatching at least one week earlier than females, causingg an initial disproportionate sex ratio (Grill et al. 2003b). Maniola nurag is aa mountain species, restricted to areas above 500 m (Grill et al., 2003b), with its mainn populations at 1000 m a.s.1. The main Sardinian populations of M. jurtina flyy at sea-level, but the species can occasionally be found sympatric with M. nurag att higher altitudes; single individuals of M. jurtina are found up to 900 m, but no stablee populations were observed at this altitude (Grill et al. 2003b). Both species aree univoltine and overwinter as larvae. Larvae of M. nurag have been reared with FestucaFestuca morisiana (Jutzeler et al. 1997). Their actual range of host plants has not yet beenn investigated in the field, but probably includes other Festuca species, and grasss species outside the genus Festuca. They are very evasive, and hide close to thee ground during the day, while coming out to feed only at night. Host plants of
139 9 Figuree 1. Study areas and sampling sites. White circles represent M. jurtina, grey circless M. nurag. Population numbers are explained in Table 1.
M.M. jurtina include a wide range of grass species, preferably Poa spp. (Higgins & Rileyy 1970; Hesselbarth et al. 1995), but also F. morisiana. Adults of both species can bee observed at different nectar sources including Cistus monspeliensis and Carlina corymbosacorymbosa (Grill, 2001).
Thee main phenotypic characteristics to differentiate M. nurag from the M. jurtina aree yellow-orange areas on the upper side of the male wings. Male upper fore- wingg brand marks are more conspicuous and female markings brighter and moree sharply defined in the island endemic. Contours of both species' wings are similar,, but M. nurag is smaller than M. jurtina in both sexes. Due to these slight morphologicall differences, M. nurag has been considered a close relative of M. jurtinajurtina since the time of its description (Simmons, 1930; Thomson, 1976; Jutzeler et al.al. 1997). In the locality where populations of the two species occur sympatrically, wee occasionally encountered individuals that could not be clearly attributed to
140 0 eitherr species, and showed intermediate wing patterns. Interestingly, individuals withh similar intermediate wing characteristics have been identified in older materiall {since 1970s) of the entomological collection of the Zoological Museum off Amsterdam. As a result, the Maniola specimens from Sardinia can be divided in threee rather than two phenotypic groups.
Materialss and methods
Collectionn of samples Adultt individuals were collected trom 1U M. mtrag populations and from 16 M. jurtinajurtina populations in both Sardinia and in continental Europe during the 1999- 20022 flight season (Table 1, Figure 1). Geographical distance between populations wass estimated as the crow flies, and ranged from 2 to 1930 km. A total number off 406 adult butterflies have been collected. Samples were frozen in the field in liquidd nitrogen, then brought to the laboratory in Amsterdam, and stored at -70 °CC until they were analysed electrophoretically. Sample sizes ranged from 3 to 34 perr population (Table 1). Moreover, we encountered individuals that could not bee clearly attributed to either species on the basis of wing morphology. These individualss were assigned to a third group. They resembled the intermediate form inn the collection of the Zoological Museum in Amsterdam, and are supposedly hybridss between M. nurag en M. jurtina.
AUozymee electrophoresis Fromm an initial screening of 22 enzyme systems, allozyme variation was assayed att 15 interpretable allozyme loci, which were polymorphic at the 99% level (in parentheses,, locus-abbreviation and E.C. number): Aldolase (aldo, 4.1.2.13), Aspartatee aminotransaminase, two loci (aat, 2.6.1.1), Glucose-6-phosphate dehydrogenasee (gópdh, 1.1.1.49), Glycerol-3-phosphate dehydrogenase (gpd, 1.1.1.8), Isocitratee dehydrogenase, two loci (idh, 1.1.1.42), Malate dehydrogenase, two loci {mdh,{mdh, 1.1.1.37), Malic enzyme {me, 1.1.1.40), Mannose-6-phosphate isomerase (mpi,(mpi, 5.3.1.8), Peptidase leu-ala (pep-ku-ala, 3.4.11), Phosphoglucoseisomerase (pgi,(pgi, 5.3.1.9), Phosphoglucomutase (pgm, 5.4.2.2), and 6-Phosphogluconic dehydrogenasee (6pgdh, 1.1.1.44). After sexing, head and thorax of a specimen were groundd by hand in 120 ul homogenizing buffer (0.1 M Tris, 0.05 M Malic acid, 0.001 MM EDTA, 0.001 M MgC12, 0.05 mM NADP). Homogenates were then centrifuged att 10,000 rpm at 4°C for 5 minutes. The clear supernatant was either stamped onn cellulose acetate gels (Titan III, Helena Laboratories, Beaumont, TX, USA) followingg the procedure of Hebert and Beaton (1989), or soaked onto Whatman
141 1 o»» 3 O O iTii O m T33 ,ü tc c cc ^ .=; Si Si CBB QJ )-c *s. . ++ 1 +1 +1 +1 +1 -ll +l ++ 1 +1 +->> in 73 inn "O III QJ 3 PP I CNN (N m m dodo o £-- « bO >-. . Xtt M_ JH N N mm S cc ^ J55 3 CC u° s s IIS S «« ^^+11 +ï1! 5 ü ü ü iiililii ++ 1 fcfc fc s 33 c (NN (V) T-l B.. n OO O o OO OJ DC C
LAA '—' C SS « 3 00 3^ cc .2 ss «
CC u 5J CC S3 ++ 1 -II +1 +1 +1 +1 +1 +1 ++ 1 +1 BB ™ SPSS? ? «« 's ^^ *" ^" fljflj a 0-,0-, fl
a** -C
(88 U L w w Xi Xi a a O O oo J2 c c < < 3^ ^ FÏ Ï 'Ll l rNN r4 r^( c c ^ ^ css —' CC _Q c c rr o ü, , o o 'u u 'Eb b /X X O O 55 Ë u u cfi i ë ëX X a»H ,H , 55 t3 aa 'J3 2 2 cc ns o o Uu). ). aa -o o I—1 1 CM M CM M * * [V. . m mm m in n in n in n sD D O O t-v. . sD D * * * * in n m m m me n n m m 3 3 o o o o o o O O o o 3 3 o o o o O O o o o o o-* o * O O oa oa o o ö ö o o o o ö ö o o ös ö sö ö ö ö cd d Ö Ö ö ö ö ö ö ö o o o o cd, , il l Ü Ü il l il l il l il l Ü Ü Ü Ü il l il l ++ 1 +1 1 ++ 1 +1 1 +1 1 ++ 1 o o fN N SO O sO O o o sC C Os s * * 1—t t oo o CS S t?v, , Os s Os s 00 0 ,_-tf |f | in n "* * sO O t^<*. . * 00 0 ^ ^ tv. . <*} } m m oo o m m renH H enl-e nH H n r-i i (-H H l-H H 1-H H 9 9 CM M T—< < l-H H i—i i l-H H l-H H l-H H l-H H rH H ö ö Q Q ö ö ö ö ö ö O O ö ö ö ö ö ö ö ö cd d ö ö ö ö ci i ö ö ö ö ______1 CT"j j io o L ^ ^ o o o o Cl l rj j CJ J tJ J en n en' ' Ö Ö en n ö ö en n ö ö rO O en n sD D o o o o en n P P cd d cd d en' ' |v. . oo o tv v *r r m m0 0 0 in n in n sD D * * sD D in n tv v ^t t \D D m m 5 en n en n en n fN N en n CM M m mC M M en n CM M en n en n en n (N N fN N m m ee -ï cd d ö ö ö ö Ö Ö ö ö Ö Ö o o Ö Ö Ö Ö Ö Ö Ö Ö cd d cd d Ö Ö Ö Ö cd d +J J ++ 1 +J Ji l l Ü Ü Ü Ü Ü Ü Ü Ü Ü Ü Ü Ü Ü Ü Ü Ü il l Ü Ü Ü Ü il l sO O o o in n fM M in n (N N sO O in n sD D tv. . m mC M M in n trv v Os s "* * 1-1 1 1-1 1 I-H H r—f f fN N (VI I rH H en n CM M Os s (N N I-H H ffi ffirH H tv. . CM M CM M l-H H * * CM M r-i i Ö Ö rH H Ö Ö Ö Ö Ö Ö O O O O O O O O cd d cd dcd , , cd, , Ö, , cd, , Ü Ü Ü Ü J JÜ Ü Ü Ü il l Ü Ü ij j iöl öl il l Ü Ü ü ü Ü Ü Ü Ü Ü Ü il l oo o Os s fNI I sO O CM M o o in n Os s oo o t>. . sD D Os s [v. . fN N i-H H CM M ^-H H sD D « « SO* O* in' ' r-. . Ö Ö ö ö fN N os s só ó * * 1-i i en n i-i i i-i i l-H H 0Ó Ó CM M CM M m m rH H Os s fM M i-< < r-i i m m en n oo o in n cn cn ro o CC C t t tv. . sC C CM M i—i i en n i—i i l-H H CM M CM M T3 T3 o o o o O O o o o o o o O O o o o o o o O O O O o o in n m me n n in n in n O O * * l-H H CM M CM M tv. . CJs s * * CM M CM M Os s tn n tv. . Os s rH H r>. . 1—f f os s o o o o o o o o fM M r-i r-i o o CM M g g o o o o o o o o O O O O O O 8 8 o o § § O O O O O O o o o o o o o o O O o o o o o o o o o o g g o o o o fN N ai—i i aCM M CM M CM M CM M CM M CM M eM M CM M esl l sfN N sCN N es( ( CM M CM M T3 3 C C .2 2 .3 3 C C C C re re re re ra ra ra ra O O C C ja ja _ra a '2 2 '2 2 '2 2 'ra a C C c c Ui i '2 2 « « ra ra in n t t o. . 3 3 3 3 Ui i l-H H C/1 1 ,£ £ re re re re < << <' S S UJ J in ini n n c c z z c c Z Z ra ra T3 T3 U-j j V V 5" " 3 3 h^ ^ -^H H in n ïï ^r^ ^ Threee buffer systems were used: tris citrate pH 8.0 (Selander et ah, 1971) for gpd, aconitase,aconitase, mpi, me, and mdh; tris glycine pH 8.4 (Hebert and Beaton, 1989) for pgm, pep-leu-ala,pep-leu-ala, and gpi; and amine citrate pH 6.1 (Clayton and Tretiak, 1972) for aai, 6pgdh,6pgdh, idh, and gBpdh. Alleles were named in alphabetical order of increasing (anodal)) migration except for mdh-1, which migrated cathodally. All loci are locatedd in the nuclear DNA, and probably inherited as autosomal genes in a Mendeliann fashion. Because sample sizes were sometimes small, it was necessary too analyse a large number of (polymorphic) loci to accurately estimate genetic variabilityy (Nei, 1978). Dataa analysis Locii were tested for linkage disequilibrium with GENEPOP, Version 3.3 (Raymond && Rousset, 2001), and checked for sex-linkage by hand. GeneticGenetic differentiation. Too estimate genetic variation, the following parameters were calculated per population:: proportion of polymorphic loci, mean effective number of alleles, and observedd and expected heterozygosity. Exact tests for Hardy-Weinberg equilibrium weree performed with the GENEPOP computer package, Version 3.3 (Raymond & Rousset,, 2001). Probability values without bias using a Markov chain method were usedd following the algorithm of Guo & Thompson (1992). We further used default parameterss in GENEPOP for dememorisation number, batches, and permutations forr all Markov chain tests performed. Sequential Bonferroni adjustments (Rice, 1989)) were applied to judge significance levels for all simultaneous tests with ann initial a level of 0.05. GENEPOP was also used to estimate three summary statistics:: Fit, the correlation of genes within individuals in the entire population; F]S,, the correlation of genes within individuals within (sub)-population; and Fst, thee correlation of genes of different individuals in the same (sub)-population (Wright,, 1951; definitions following Weir & Cockerham (1984). Fst values were testedd for departure from zero by the x2 method of Workman & Niswander (1970), andd inbreeding coefficients (F|S) were tested for departure from zero by the test developedd by Li & Horvitz (1953) using the computer program 'Theta' (Ellis 1994). F|T,, F|S, and FST estimates were calculated according to Weir & Cockerham's (1984) 144 4 ass F, f, and 0 respectively, because their method is not based on assumptions concerningg numbers of populations, sample sizes, and heterozygote frequencies. A bootstrapp procedure with 10000 repeats was employed to estimate the variance of thee F-statistics (Van Dongen, 1995). Following Slatkin & Barton (1989) negative FST- valuess must be considered as 0, and indicate that mathematically, that the amount off gene-flow between the respective populations is infinite, i.e., the populations functionn as one panmictic unit. Clusterr analyses were conducted on Nei's genetic distances (1978) with (1) the unweightedd pair-group method with arithmetic averaging (UPGMA procedure inn BIOSYS 2, Swofford & Selander, 1997), and (2) the neighbour joining method (Saitouu & Nei, 1987) in PHYLIP (Felsenstein, 2002). For the cluster analyses, sampless with less than five individuals were left out. The data was tested for recent reductionss in effective population size with the computer program BOTTLENECK (Cornuett & Luikart, 1996). This program computes for each population sample and forr each locus the distribution of the gene diversity expected from the observed numberr of alleles, at a given sample size under the assumption of mutation-drift equilibrium.. This distribution is obtained through simulating the coalescent processs of n genes under three possible mutation models, viz. the infinite allele model,, the two-phased model of mutation, and the stepwise mutation model. Thiss enables the computation of the average expected heterozygosity which is comparedd to the observed gene diversity to establish whether there is a gene diversityy excess or deficit at a locus. Once all loci available in a population sample havee been processed, three statistical tests are performed for each mutation model {Cornuett and Luikart, 1996), and the allele frequency distribution is established in orderr to see whether it is approximately L-shaped (as expected under mutation- driftt equilibrium) or not (due to recent bottlenecks which provoke a mode shift). Gene-flow Gene-flow Gene-floww between populations was indirectly estimated with both Wright's FST (Wrightt 1931) and the private allele method [a private allele is an allele found inn only one subpopulation (Slatkin 1985; Barton & Slatkin 1986)]. Wright (1951) -1 showedd that FSTis a useful estimator of gene-flow if Nm *= 1/4[(1/ FST) ] (Wright 1951).. Frequencies of alleles averaged over the number of populations in which theyy occur are useful to measure the spatial distribution of alleles in subdivided populations,, and are relatively independent of selection and mutation rates (Slatkinn 1981). The conditional average frequency of alleles found in only one (sub)populationn (i.e., a private allele), can be used as a quantitative estimator of 145 5 Tablee 2. Allele frequencies at 15 polymorphic loci in populations of M. nurag (n) and M.M. jurtina (j), and intermediates (f). Alleles coded alphabetically according to their rela- tivee mobility. Loci not in Hardy Weinberg equilibrium printed in bold. Populations are presentedd per year, study area and species and coded according to the abbreviations givenn in Table 1. 1999 9 20000 2000 2000 2000 2000 2000 2000 2000 2000 2000 2000 2000 RO O SFF SF MN MN PO MF KF CO SA DÜ RO OR ) ) nn j is j a n j j j j j i i idh-1idh-1 i 0.000o.oo 0 o o.(Bi o.ooo o.ooo 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 b b 0.091 1 0.0311 0.000 0.021 0.000 0.000 0 071 0.065 0.000 0.063 0.000 0.094 0.071 c c 0.909 9 0.9388 1000 0.979 1000 0.500 0.857 0.935 1.000 0.875 0 955 0.S91 0.929 d d 0.000 0 0.0000 0000 0.000 0.000 0.500 0.071 0,000 0.000 0.063 0.045 0.016 0.000 e e 0.000 0 00000 0,000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 O.OOO 0.000 0.000 geness 23 64 14 48 6 4 14 46 6 76 22 64 42 tdh-2tdh-2 a 0.000 0.091 0.000 0.060 0.000 0.000 0.000 0.022 0.000 0.125 0.000 0.000 0.048 bb 0.136 0.152 0.071 0.100 0167 0 250 0.063 0,022 0.000 0.063 0 000 0.061 0.048 cc 0.864 0 758 0 929 0,840 0.833 0.750 0.938 0.957 1.000 0.813 0.955 0.939 0.905 dd 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.045 0.000 O.OOO geness 22 66 14 50 6 4 16 46 6 16 22 66 42 pep-leu-ala pep-leu-alaa a 0.318 8 0.5477 0.214 0.661 0.167 0.250 0.500 0.196 0.500 0,125 0.182 0.258 0.190 b b 0.682 2 0.3599 0.500 0.339 0667 0.000 0.500 0.783 0.500 0.625 0.500 0.742 0.786 c c 0.000 0 0.0000 0,000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0,227 0,000 0.000 d d 0.000 0 0.0944 0.000 0.000 0.167 0.000 0.000 0.022 0.000 0.250 0.091 0,000 0.000 e e 0.000 0 0.0000 0.286 0.000 0.000 0.5O0 0.000 0.000 0.000 0.000 0.000 0.000 0.024 f f 0.000 0 0.0000 0.000 0,000 0.000 0.250 0000 0.000 0.000 0.000 0.000 0.000 0,000 genes s 22 2 644 14 566 4 18 466 16 226642 pgm pgm3 3 0.000 0 0.0766 0.000 0.052 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 b b 0.091 1 0.0611 0.071 0.259 0,000 0.500 0.188 0.063 0.000 0.000 0.045 0.061 0.000 c c 0091 1 0,2422 0.071 0.086 0.167 0.250 0.250 0.042 0333 0.250 0.045 0.045 0048 d d 0.727 7 0.4700 0.357 0,448 0.500 0,000 0.438 0 542 0.333 0.500 0.636 0.742 0,810 e e 0.000 0 0.0455 0.429 0.086 0333 0.250 0.125 0.229 0.167 0.250 0 273 0.152 0.048 f f 0.045 5 0.0455 0.071 0.000 0.000 0.000 0.000 0.125 0.000 0.000 0.000 0.000 0.095 g g 0.000 0 0.0300 0.000 0.069 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 h h 0.045 5 0.0300 0.000 0.000 0.000 0 000 0 000 0.000 O.OOO 0.000 0.000 0.000 0.000 i i 0.000 0 0.0000 0.000 0.000 0.000 0.000 0.000 0.000 0.167 0000 0.000 0.000 0.000 i i 0.000 0 0.0000 0.000 0.000 0000 0.000 0.000 0.000 0000 0000 0.000 0.000 0.000 genes s 22 2 666 14 58 6 4 16 48 6 16 22 66 42 Tablee 2 continued. 20011 2001 2001 20011 2001 2001 20011 2001 2001 20022 2002 2002 20022 2002 2002 SFF SF EC ECC FO GI KFF VE SF SFF SF MN ECC AM SF nn j n ;; " j jj i ' nn ƒ n "" ƒ ' 0.0000 0.000 0,000 0,0833 0.000 0.000 0.0455 0.000 0.000 0.0000 0.000 0.024 0.0255 0.000 0.000 0.0422 0.125 0.059 0.0833 0.023 0.029 0.0455 0.333 O.000 00000 0.071 0.024 0,0500 0,115 0.000 0.9388 0.875 0.941 0.8333 0.977 0.971 0.9099 0.667 1.000 0.8644 0.929 0.905 0.9255 0 885 1000 0.0211 0.000 0.000 0.0000 0.000 0.000 0.0000 0.000 0.000 0.1366 0.000 0.024 0.0000 0.000 0.000 0.0000 0.000 0,000 0.0000 0.000 0.000 0.0000 0.000 0.000 0,0000 0.000 0.024 0,0000 0,000 0.000 488 S 34 122 44 34 444 6 16 222 14 42 400 26 16 0.0000 0.000 0.059 0,0833 0.068 0.000 0.0000 0.000 0.000 0.0833 0.000 0.000 0.0000 0.000 0.000 0.2500 0.125 0.176 0.0833 0068 0118 0.0455 0.000 0125 00422 0071 0.100 0,2500 0.038 0,000 0.7299 0.875 0.735 0.8333 0.864 0.882 0.9555 1.000 0.875 0.8755 0.929 0.900 0.7500 0.962 1,000 0,0211 0.000 0.029 0.0000 0.000 0.000 0.0000 0.000 0.000 O.OOOO O.OOO 0.000 0.0000 0.000 0.000 488 8 34 122 44 34 444 6 16 244 14 40 400 26 16 0.6044 0.625 0.206 0.4177 0.477 0.316 0.3188 0.167 0.688 0.4588 0.571 0.500 0.6000 0.167 0.389 03966 0.375 0.735 0,5833 0.2O5 0.684 0.6822 0.667 0.313 0.5422 0.429 0.500 0.4000 0.800 0.556 0.0000 0.000 0.000 0.0000 0.000 0000 0.0000 0.167 0.000 0.0000 0.000 0,000 0.0000 0.000 0,056 0.0000 0.000 0.059 0.0000 0.182 0000 0.0000 0.000 0.000 0.0000 O.OOO 0.000 0.0000 0.033 O.OOO 0.0000 0.000 0.000 0.0000 0.023 0.000 0.0000 0.000 0.000 O.OOOO O.OOO 0.000 O.OOOO 0.000 0.000 0.0000 0.000 0.000 0.0000 0.114 0.000 0.0000 0.000 0.000 O.OOOO O.OOO 0.000 0.0000 0.000 0.000 488 8 34 122 44 38 444 6 16 244 14 42 400 30 18 0.0833 0000 0.029 0.0833 0.114 0.000 0.0000 0.000 0.125 0.O422 O.OOO 0.071 D.0000 0.000 0.056 0.2088 0.125 0,441 0.4177 0.591 0.026 0.0455 0.667 0.188 0.5833 0.143 0.333 0.3500 0.20O 0.333 0.4388 0.375 0,412 0.1677 0.182 0105 0.0911 0.000 0.313 0.0422 0.071 0.095 0.1000 0.000 0.111 0.2711 0500 0.088 0.3333 0.000 0 842 0.8411 0.000 0.375 0,2922 0.571 0.429 0.4500 0.667 0.389 0,0000 0.000 0.000 0.0000 0.000 0.000 0.0000 0.333 0,000 0.0422 0.214 0.048 0.1000 0.133 0.111 0.024 4 0.0000 0 000 0.000 0.0000 0 000 0 026 00233 0000 0000 0.0000 0.000 0.000 O.OOOO 0.000 0 0.0000 0 000 0.029 0.0000 0.068 0.000 0.0000 0.000 0.000 0.0000 0.000 0.000 0.0000 0.000 0.000 0.0000 0.000 0.000 0.0000 0.000 0.000 0.0000 0.000 0.000 0.0000 0.000 0.000 0.0000 0.000 0.000 0.0000 0.000 0.000 0.0000 0.000 0.000 0.0000 0.000 0.000 0.0000 0.000 0.000 0.0000 0.000 0.000 0.0000 O.OOO 0.000 00000 0.045 0,000 0.0000 0.000 0.000 0.0000 0.000 0.000 0.0000 0.000 0.000 488 8 34 122 44 38 444 6 16 400 30 18 244 14 42 Tablee 2 continued. 19999 2000 2000 2000 2000 2000 2000 2000 2000 2000 2000 2000 2000 ROO SF SF MN MN PO MF RP CO SA DÜ RO OR ;; njnjnnjiiiij mdh-1mdh-1 a o.ooo 0.000 0.000 0.000 0.000 0.000 0.071 0.000 0.000 0.000 0.000 0.010 0.000 bb 1.000 0.963 0.93S 0.732 0.667 0.750 0 500 0.979 1.000 0.786 0.688 0.953 1000 cc 0.000 0.037 0.063 0.196 0.333 0.000 0.143 0.021 0.000 0.071 0.125 0031 0 000 dd 0000 0.000 0000 0.071 0000 0 250 0286 0.000 0000 0.071 0.188 0000 OOOO ee OOOO 0.000 O.OOO 0.000 0.000 0.000 0 000 0.000 0.000 0.071 0.000 0000 0000 geness 22 54 16 56 6 I 14 48 6 14 16 64 26 mdh-2mdh-2 a oooo 0.000 0000 0.000 0.000 0000 0000 0.000 0000 0000 0.000 0.045 0,095 bb 0 000 0.015 0 000 0.054 0.000 0.000 0 071 0.021 0.000 0.000 0.000 0.030 0,095 cc 0 864 0.941 0 938 0.893 1000 0 500 0 643 0 958 1000 0 786 0 773 0.909 0.762 dd 0 000 0.029 0 000 0.000 0.000 0.000 0.286 0.021 0.000 0.143 0.227 0.000 0.000 ee 0.136 0.015 0063 0.054 0.000 0.500 0.000 0.000 0.000 0.071 0.000 0.015 0.048 geness 22 68 16 56 6 4 14 48 6 14 22 66 42 gpigpi a 0.000 0.034 0.000 0.000 0,000 0.000 0.056 0.000 0.000 0.000 0.000 0,045 0,024 bb 0 000 0.190 0143 0.069 0,167 O.OOO 0.056 0.071 0.000 0 000 0.050 0.030 0.000 cc 0909 0.690 0643 0.862 0.667 0 750 0.611 0.786 1000 0 875 0.650 0.758 0.929 dd 0045 0.052 0.143 0.069 0.167 0 250 0.111 0.119 0.000 0.063 0.250 0.061 0.000 ee 0.000 0.034 0.071 0.000 0.000 0.000 0.167 0.024 0.000 0.063 0.050 0.106 0.048 ff 0.045 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 geness 22 58 14 58 6 4 18 42 4 16 20 66 42 g6pdhg6pdh a 0.091 0.063 0.714 0.017 0.000 0,000 0 000 0.022 0.500 0.214 0.000 0.121 0.048 bb 0.864 0.484 0.286 0.417 0.750 0.250 0.375 0.848 0.5OO 0.500 0 850 0.803 0.952 cc 0.045 0.422 0.000 0.383 0.000 0.750 0.625 0.130 0.000 0.286 0100 0.061 0.000 dd O.OOO 0.031 0.000 0.183 0.250 0 000 0.000 0.000 0 000 O.OOO 0 050 0.015 0.000 geness 22 64 14 60 4 4 16 46 4 14 20 66 42 6pgdh 6pgdha a 0.091 1 0.0000 0.143 0019 0.000 0.000 0.000 0.114 OOOO 0,125 0,000 0.076 0.071 b b 0.909 9 0.4677 0.286 0.596 0.000 0.000 0.429 0.364 0.500 0.750 0.773 0.848 0.905 c c 0.000 0 0.2000 0.000 0.173 0.000 1.000 0 214 0.068 0.250 0.000 0.000 0.000 0.000 d d 0.000 0 0.2333 0.571 0.135 1.000 0.000 0 357 0.432 0.000 0.125 0.227 0.076 0.000 e e 0.000 0 01000 0.000 0077 0.000 0 000 OOOO 0000 0250 0 000 0.000 0.000 0.024 I I 0.000 0 0.0000 0.000 0.000 0.000 0.000 0.000 0.023 0 000 0.000 0.000 0.000 0.000 genes s 22 2 600 14 52 4 2 14 44 4 16 22 66 42 Tablee 2 continued. 20011 2001 2001 2001 2001 20011 2001 2001 2001 2002 20022 200Z 2002 2002 2002 SFF SF EC EC PO GII RP VE SF SFF MN EC AM nn ƒ n ƒ n ƒƒ J ) i " jj n n ) i 0.0000 0.000 0 000 0,000 0.000 0.0000 0.000 0.000 0.000 0.000 0.0000 0,111 O.OOO 0.000 0,000 0,8100 1,000 0.969 1.000 0.905 0.9588 0.975 1.000 0.875 1.000 1,0000 0.889 1.000 1.000 1 O00 0.1900 0.000 0.031 0.000 0.095 0.0422 0.025 0.000 0.125 0.000 0.0000 0.000 0.000 0.000 0.000 O.0000 0.000 0.000 0.000 0.000 0.0000 0.000 0.000 0.000 0.000 0.0000 0.000 0.000 0.000 0.000 0.0000 0.000 0,000 O.OOO 0.000 0.0000 0.000 0.000 0.000 0.000 0.0000 0.000 0.000 0.000 0.000 422 8 32 12 42 244 40 6 16 16 66 18 10 28 10 0.0000 0.000 0.000 0.000 0.000 0.0000 0.000 0.000 0.000 0.000 0.1677 0,056 0.000 0.036 0.000 0.0000 0.000 0,033 0.000 0.024 0.0000 0.000 0.000 0.000 0.000 0.0000 0.000 0.000 0,000 0.000 1.0000 1.000 0.900 1.000 0.976 0.9177 1.000 1.000 1000 1.000 0.8333 0.944 1,000 0929 0.900 0,0000 O.OOO 0.O67 0.000 O.OOO 0.0288 0,000 0 000 0.000 0.000 0.0000 0.000 0,000 0000 0.100 0,0000 0.000 0.000 0.000 0.000 0.0566 0OO0 0.000 0.000 O.OOO 0.0000 0,000 O.OOO 0.036 0.000 444 8 30 12 42 366 40 6 16 16 66 18 8 28 10 0.0000 0.000 0.000 0.000 0.000 0.0288 0.000 0.000 0.000 0.000 0.0000 0.000 0.025 0.000 0.000 0.1099 0.167 0.088 0.OS3 0.000 0.0000 0.O25 0.167 0.063 0.042 0.0000 0,000 0.050 0.067 0.056 0.7611 0,833 0.765 0.750 0.932 0,8066 0.825 0.833 0.875 0.833 0.7866 0.952 0.800 0.833 0.944 0.1300 0.000 0.147 0,167 0.068 0.1399 0.150 0,000 0.063 0.042 0.2144 0.048 0.075 0.100 0.000 0.0000 0.000 O.OOO 0,000 0.000 0.0288 0,000 0 000 0.000 0.000 0.0000 0.000 0,025 0.000 0.000 0.0000 0.000 0.000 0.000 0.000 0,0000 0.000 0.000 0.000 0.083 0.0000 0.000 0,025 0.000 0,000 466 6 34 12 44 366 40 6 16 24 144 42 40 30 18 0.0000 0.000 0.000 0,000 0.000 0.0000 0.000 0,000 0.000 0.000 00000 0.000 0.025 O.OOO 0.125 0.5911 1.000 0.529 0.417 0.477 1.0000 0.977 0.500 0.813 0.455 1.0000 0.548 0.625 0 767 0.438 0.3644 0.000 0.471 0.5S3 0.523 0.0000 0.023 0.500 0.188 0.545 0.0000 0.452 0,350 0.233 0.438 0.0455 0,000 O.OOO 0.000 0.000 0.0000 0.000 0.000 0.000 0.000 0.0000 O.OOO 0.000 0.000 O.OOO 444 8 34 12 44 366 44 6 16 22 122 42 40 30 16 0.0000 0.000 0.071 0.000 0167 0.0000 0.068 0 000 0.000 0.083 0.0000 0.000 0.050 0 000 0.000 0,9588 1.000 0.857 1.000 0 714 0.9744 0.886 0 833 0.875 0.875 1.0000 1.000 0.950 1.000 0.125 0,0422 0.000 0.071 0.000 0.119 0.0266 0.CW5 0.167 0.125 0.042 0.0000 0.000 0.000 0.000 0.498 0-0000 0.000 0.000 0,000 0.000 0.0000 0.000 0.000 0.000 0 000 0.0000 O.OOO 0.000 0.000 0.498 0.0000 0.000 0.000 0.000 0.000 0.0000 0.000 0.000 0.000 0.000 0.0000 0.000 0.000 0.000 O.OOO 0.0000 0.000 0.000 0.000 0.000 0,0000 0.000 0.000 0.000 0.000 0.0000 0.000 0.000 0.000 0,000 488 8 28 8 42 388 44 6 16 24 144 42 40 30 18 Tablee 2 continued. 19999 2000 2000 2000 2000 2000 2000 2000 2000 2000 2000 2000 2000 ROO SF SF MN MN FO MF RP CO SA DÜ RO OR 11 " ) n ! n » i ) j i j j aat-1aat-1 3 0.909 0.982 l.OOO 0.957 1,000 l.OOO l OOO 1,000 0.833 0.900 1.000 0.984 0.944 bb 0.000 0.018 0.000 0.022 0.000 0000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 cc 0.091 0.000 0000 0.022 0.000 0.000 0.000 0.000 0.167 0.000 0.000 0.016 0.056 dd 0000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.100 0.000 0.000 0,000 geness 22 56 14 466 4 6406 10 20 6436 aat-2aat-2 a 0.000 o.isi 0.000 0.000 0.000 0.000 0.000 0.025 0.000 0.000 0.000 0.015 0.000 bb 0.000 0.016 0.000 0.000 0.000 0.000 0.000 0.075 0000 0.000 0.000 0.121 0.050 cc 1.000 0.823 0,929 1.000 1000 1.000 1.000 0.900 1.000 1.000 1.000 0.86* 0.950 AA 0.000 0.000 0.071 0.000 0.000 (1000 0.000 0.000 O.OOO 0.000 0.000 0.000 0.000 geness 20 62 14 46 6 4 14 40 6 16 20 66 40 meme a 0.909 0.979 0929 0.972 1.000 0.250 1.000 0.958 0.750 1.000 1.000 0.960 1000 bb 0091 0.021 0.071 1)000 0.000 0.750 0.000 0,042 0.250 0.000 0,000 0.040 0.000 cc 0.000 0.000 0.000 0.028 0,000 0.000 0000 0.000 0.000 0.000 0.000 0.000 0.000 geness 22 48 14 366 4 44844 25022 mpimpi a 0.000 0.000 0,000 0.000 0000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0,000 bb 0,045 0.063 0.000 0,000 0.000 0.000 0.000 0.000 0.OO0 0,000 0.000 0.171 0.029 cc 0,136 0.000 0.000 0,000 0.000 0 000 0.000 0.000 0.000 0,000 0.000 0.056 0.118 dd 0.091 0.000 0000 0.000 0.000 0.000 0,000 0.000 0.000 0.500 0.500 0.130 0.059 ee 0.591 0.938 0.625 0.917 1.000 1.000 0.500 0.750 0.000 0.5OO 0.500 0.593 0 618 ff 0.136 0.000 0.375 0.O28 0.000 0.000 0.500 0.250 0,000 O.OOO 0.000 0.111 0.176 gg 0.000 O.OOO 0.000 0.056 0.000 0.000 0000 0.000 0.000 0.000 0.000 0.000 0.000 geness 22 16 8364 2 42804 25434 gpdgpd a 0.000 0.000 0.000 O.OOO 0 000 0.000 0.000 0.000 0.000 0.250 0.000 0.000 0.000 bb 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 0.750 1.000 1.000 1,000 cc 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 geness 22 44 16 406 4 6466 42250 22 aldoaldo a 0.000 0000 0.000 0.000 0.000 0.000 0.01X1 0.000 0.000 0.000 0.000 0.000 0,000 bb 0.045 0.000 0.000 0.083 0,000 O.OOO 0.000 0.000 0.000 0 000 O.OOO O.OOO 0,111 cc 0091 0.000 0000 0.000 0.000 0000 0.000 0083 0.000 0.000 0.000 0.063 0.278 dd 0.864 1.000 1.000 0.917 1.000 1.000 1.000 0.917 1.000 1.000 1.000 0.938 0.611 geness 22 30 12 24 6 2 2 36 6 4 22 32 18 Tablee 2 continued. 20011 2001 2001 2001 2001 20011 2001 2001 2001 2002 20022 2002 2002 2002 2002 SFF SF EC EC PO Gll RP VE SF SFF MN EC AM nn i n ) n !! ! i ' " 11 " " ] ' 1.0000 1.000 1.000 1000 1000 0.9477 0.977 0.833 1.000 1.000 0.9299 1.000 1.000 1.000 1.000 00000 0.000 0.000 0000 0.000 0,0533 O.OOO 0.000 0.000 0.000 O.OOOO 0.000 0.000 0.000 0000 0.0000 0.000 0.000 0.000 0.000 0.0000 0.000 O.OOO 0.000 0.000 0.0000 0.000 0,000 0 000 0.000 0.0000 0.000 0.000 0.000 0.000 0.0000 0.023 0.167 0.000 0.000 0.0711 0.000 0 000 0.000 0.000 488 8 34 12 44 388 44 6 16 24 144 42 40 30 18 0.0000 0.000 0.000 0.000 0.023 0.0799 0.000 0.000 1.000 0.000 0.0000 0.000 O.OOO 0.000 1.000 Ü.Ö633 Ö.ÖÜO Ü.Ü59 Ü.ÜÜÜ Ö.Ü2J U.U/yy 0.114 U.UUIJ U.UUU U.U56 u.uuuu o.u,i4 u.uuu u.06/ u.uuu 0.9388 1.000 0.912 1.000 0.955 0.8422 0.864 1.000 0.000 0.944 1.0000 0.976 0.975 0.900 0.000 0.0000 0.000 0.029 0.000 0.000 0,0000 0.023 0.000 0.000 0.000 0.0000 0.000 0,025 0033 0.000 488 8 34 12 44 388 44 6 16 18 144 42 40 30 16 1.0000 1000 0.893 1.000 1.000 0.8955 0.955 1.000 1.000 0.958 1.0000 0.976 0.950 1.000 0.611 00 000 0.000 0.107 0.000 0.000 0.1055 0.045 0.000 0.000 0.042 0.0000 0.024 0.050 0.000 0.389 0.0000 0.000 0.000 0.000 0.000 0.0000 0.000 0.000 0.000 0.000 0.0000 0.000 0.000 0.000 0.000 466 8 28 10 42 388 44 4 14 24 144 42 40 30 18 0,0000 0.000 0.000 0.000 0,000 0.0000 0.114 0.000 0.000 0.042 0.0000 0.048 0,000 0.033 0.000 0,0211 0.000 0.000 0,000 0.050 0.1055 0.000 0,000 0.000 0.000 0,0000 0 024 0.026 0.133 0.100 00633 0.000 0.031 0.000 0.000 0.0266 0.045 0.000 0.000 0.000 0,0000 0.000 0.026 0.100 0.100 0.0422 0.000 0,031 0.000 0.075 0,0533 0.114 0.167 0.000 0.125 02866 0.000 0.026 0.100 0.100 0.8133 1.000 0906 0.917 0.875 00 816 0*14 0.833 1.000 0.667 0.7144 0.857 0.868 0.633 0.700 0.0633 0,000 0 031 0.083 0.000 0.0000 0.11* 0,000 0.000 0.167 0.0000 0,071 0.053 0 000 0.000 0,0000 0000 0.000 0.000 O.OOO 0.0000 0.000 0.000 0.000 0.000 0,0000 0,000 0 000 0.000 0.000 488 8 32 12 40 388 44 6 16 24 144 42 38 30 10 0.0000 0.000 0,033 0.000 0.024 0,0000 0.023 0.000 0.250 O.OOO 0.0000 0.000 0.000 0.000 0.000 1.0000 1.000 0867 1.000 0.976 1,0000 0.977 1.000 0.750 1.000 1.0000 1.000 1.OO0 1.000 1.000 0.0000 0.000 0100 0.000 0.000 0.0000 0.000 0.000 0.000 0.000 0.0000 0.000 0,000 0.000 0.000 488 8 30 12 42 388 44 4 16 24 144 42 40 30 18 O.OOOO 0 000 0.000 0.000 0.000 0.0533 0.000 0000 0 000 0.000 0.0000 0ÜO0 0.000 0.000 0.000 0.0799 0.250 0.000 0.000 0.029 0.0000 0.000 0 000 0 000 0.000 0.0000 O.OOO 0.000 0.000 0.000 0.1Ï22 0.250 0.000 0.125 0.353 0.1055 0.13ft 0.000 0.000 0.045 0.2144 0.154 0.133 0.038 0.000 0.7899 0.500 1.000 0.875 0.618 0.8422 0.864 1.000 1.000 0.955 0.7866 0.846 0.867 0.962 1000 388 4 12 8 34 388 44 6 14 22 144 26 30 26 12 Njn.Njn. Results obtained from the private allele method are reliable if 0.1 < N m < 10, andd if the number of private alleles is >20 (Slatkin 1985). The relationship between gene-floww and geographical distance was tested by regressing FST/ (1- FST) estimates too the natural logarithm of geographic distances (Rousset, 1997). The association betweenn generic and geographic distance was tested with a Mantel test using 1000 randomm permutations to test for independence between genotype counts and locationn (Mantel 1967 in Raymond & Rousset 2001). Significance was evaluated withh Spearman Rank correlation using GENEPOP {Raymond & Rousset 2001). InterspecificInterspecific genetic differentiation Thee total allozyme dataset was examined for loci with alternatively fixed alleles or non-overlappingg variation patterns between M. nurag and M. jurtina. To quantify spatiall and temporal genetic differences, genetic similarity of individuals at the sixx loci that appeared most distinctive between the two species (i.e., idh-1, idh-2, pep-leu-ala,pep-leu-ala, pgtn, gpi, gSpdh) was analysed by a principal coordinate analysis (PCA) followingg the procedure in Arntzen (2001), in which the presence or absence of eachh allele at each locus was defined as a separate character state and was assumed too be independent (although in reality it is limited to a maximum of two scores off 1 per locus). Jaccard's coefficient of association was chosen to represent the geneticc similarity between individuals because it considers joint absences to be uninformativee (Sneath & Sokal, 1973). This similarity matrix was transformed into aa scalar product and subsequently factored. All these calculations were conducted withh NTSYS 1.80 (Rohlf 1993). Results s Thee results are presented in five sections describing in order, general aspects of bothh species' genetic variation, temporal variability, spatial variability, genetic differentiationn between the two species, and divergence time estimates. Generall aspects of genetic variation Inn total, 76 different alleles were detected at the 15 allozyme loci studied, 63 of whichh were shared by both species. None of the loci was found to be alternatively fixedd between the two species. Neither linkage disequilibrium between loci, norr sex-linkage was detected. A number of private and low-frequency alleles weree found across island and continental M. nurag and M. jurtina populations. Sevenn alleles were restricted to M. nurag, among which five were private alleles (frequenciess ranged from 0.022-0.250), and six alleles were restricted to M. jurtina, 152 2 amongg which three were private alleles (0.023-0.071); in M. jurtina one private allelee belonged to the Corsican population (0.167), and two to Sardinian (0.023- 0.053)) populations. Testss for Hardy-Weinberg equilibrium showed significant deviations in 21 out of 3033 tests (r><0.01) after the significance level was corrected for experiment-wise errorr using the Bonferroni procedure; deviations were found at ten loci in one or moree populations (Table 2). FST-values departed significantly from zero in 13 out off 72 tests for M nurag, and in 11 out of 73 tests for M. jurtina. F[C.-values departed significantlyy from zero in 19 of 72 tests for M. nurag, and in 19 of 73 tests for M. jurtinajurtina (Table 3). There was no pattern to the heterozygote deficit, neither with respectt to population nor to locus (compare also Table 1). Generally,, levels of genetic variation in Sardinian populations (M jurtina, HH =0.137-0.189; M. nurag, H =0.141-0.270) were comparable to the mainland M. jurtinajurtina (H=0.141-0.236). The intermediate individuals contained similar levels off genetic variation (H=0.167-0.178) (Table 1). Mean number of alleles per locus andd percentages of polymorphic loci in M. nurag (A=l.7-3.1, P=60-80%) were also comparablee to M. jurtina (A=1.5-3.0, P=40-80). No significant effect of genetic bottleneckingg was detected in the Sardinian populations of M. nurag or M. jurtina (BOTTLENECK,, p > 0.05). Temporall variability in genetic variation Temporall variation in heterozygosity (H0), percentage polymorphic loci (P), and numberr of alleles per locus (A) is small in both species (Table 1). For all three seasons,, total genie diversity, Weir and Cockerham's (1984) F([ (M. nurag: mean Fn == 0.170 - 0.376; M. jurtina: Fn = 0.213 - 0.343) in both species is mostly due to within populationn variation (F[S= 0.163 - 0.325) and to a lesser extent to among population variationn (FST) (Table 3). In M. nurag mean FST values (=0) summarized over the 15 polymorphicc loci were similar in the 2000 (0.052 + 0.060) and 2001 season (0.061 + 0.031),, but much lower in the 2002 season (0.008 + 0.022). In M. jurtina mean FST valuess calculated across Europe (season 2000: 0.065 + 0.039), were similar as when calculatedd within Sardinia (season 2000: 0.065 + 0.090, and season 2001: 0.050 + 0.048).. For the 2002 season, FST values could not be calculated for M. jurtina, as onlyy one Sardinian population was analysed. FST values differed significantly from zeroo in 13 instances for M nurag, and in 11 instances for M. jurtina. None of these deviationss was consistent over loci or years (Table 3). 153 3 Spatiall variability in genetic variation FSTT values indicate a low degree of spatial differentiation in both species. Correlationss between genetic differences and geographic distance were not significantt for either species at any spatial scale (Figure 2). Mantel tests show that inn both species gene exchange among populations is not dependent upon their geographicc location (M. jurtina: p = 0.234; M. mirag: p = 0.762). Gene-floiv Gene-floiv Levelss of gene-flow were estimatedd using two different 88 o approaches:: (1) the average 88 °^ numberr of migrants exchanged perr generation among A A Inn (distance) populationss (Nm) on basis of Weirr and Cockerham's (1984) FS|,, (2) Slatkin's (1985) private allelee method to estimate Nm (correctedd for sample size). UÜÜ 0.05 Estimatess were conducted acrosss different geographic B B Inn (distance) scales,, i.e., within Sardinia for M.M. nurag and M. jurtina, and Figuree 2. Relationship between geographical acrosss Europe for M. jurtina. FST distancee and gene-flow plotted as FSTAI-^ST) basedd gene-flow (Nm) estimates againstt the natural logarithm of geographic reachedd similar level in both distancess (in kilometers) between populations, to a+b fittingg FST/(1-FST) ln(distance), (A) M. species,, regardless if they were jurtinajurtina a=0.083, b=0.001, Mantel test associcalculate- dd across Sardinia (Nm atedd P=0.234; (B) M. nurag a=0.090, b=0.004, == 3.6 - 4.6) or across Europe P=0.762. . (Nm(Nm = 3.6) (Table 3). Private M.M. nurag and from 0.88 - 1.64 for allelee based estimates ranged from 0.88 - 3.13 for M.M. jurtina (Table 4). GeneticGenetic differentiation between the two species Populationss clustered into two main groups when plotted in a UPGMA phenogram,, based on Nei's (1978) unbiased genetic distances (D) (Figure 3a). Thee average genetic distance between M. nurag and M. jurtina is D=0.21. The intermediatee individuals split into a separate group on the basis of the first two groups.. The M. jurtina samples 'Sette Fratelli 2000' appeared separated, and the M. 154 4 Tablee 3. Estimates of Weir & Cockerham's F-statistics at 15 polymorphic loci sum- marizedd over populations of M. nurag and M. jurtina sampled per year, and mean estimatedd gene-flow level (Nem). Means are based on a bootstrap of 10 000 repeats andd given with standard deviation (SD). Asterisks indicate significant departure fromm zero. M.. nurag 2000 MiurtkaMiurtka 2000 Frs s FF si F„ „ N,m m (acTPSSS ïiinïpe) FL, , Fs. . fl, fl, N,m m i£>) i£>) 0.36SS *" 0.1355 *** 0.453 3 krtr-1 1 0.220' ' -0026 6 0.200 0 <.lh-2 <.lh-2 0.110 0 •0.(110 0 0.10! ! iitfr-2 2 -0.052 2 -0.0O2 2 -0.054 4 rvji-knrvji-kn tthi 0.229 9 o.ono o 0.252 2 pep-ltu-uLi pep-ltu-uLi0.20 00 • 0.212 2 0.2» » r^ffl l 0,21» » 0.019 9 0.233 3 fX"! fX"! 0,2633 " 0.036 6 0.290 0 nUh-1 nUh-1 0.4** *** 0.1255 *** 0515 5 Mfitfl-1 1 0.084 4 (1.0999 *" 0.175 5 rmfll-2 2 0.3055 * 0.1388 "* 0401 1 mJli-2 mJli-20.38 77 " 0.029 9 0.405 5 XP XP 0.3644 *" Ü.022 2 0.377 7 xpi xpi 0.180 0 0.012 2 0.190 0 x6vtEi x6vtEi 0.32)) ' 0.002 2 0.321 1 ïSr* * 0.6333 *•*- 0.1744 — 0697 7 épxiït épxiït 0.4»» •** 0009 9 0435 5 6p$ili 6p$ili 0.5%% *** 0.2222 *** 0.686 6 aat-l aat-l 0.009 9 •0,036 6 -0.027 7 rfj(-I rfj(-I -0.043 3 0.031 1 •0011 1 Mt-2 Mt-2 amam -** 0.072* * 0905 5 ilur-2 2 0409" " -0.O13 3 0.401 1 lit-lit- -0.0(5 5 0.5511 *** 0.523 3 TO' TO'-0.03 2 2 -aixx x •0.036 6 mp mp 0.6266 "** (1.117* * 0.669 9 ttipi ttipi 0.1.144 *** -0.012 2 0.326 6 vW W W W -00 2B9 * 0.2633 *** 0.050 0 aUe aUe 1.00)) "• •0.115 5 1.000 0 Mo Mo 0.005 5 0.1199 " 0.524 4 Mean n 0.3C C 0.052 2 0.376 6 4.592 2 Mean n UL298 8 U.U65 5 0.343 3 3.625 5 SD D o.ost t 0.060 0 0101 1 SD D 0081 1 0,039 9 Ü.084 4 M.M. our ax 2001 M.M. jurtina 2000 FK K Fst t F„ „ N,m m (onHH Sardinia) F.i i Fi, , F., , S^m m !,*-•( ( 0.3288 * •0.029 9 0307 7 Jift-I I -0.028 8 -0.028 8 -0.049 9 idii-2 idii-2 0.3555 " 0.022 2 0.369 9 sSi-2 sSi-2-0.36 5 5 -00 002 41.039 9 jirpAeujirpAeu tj/tf 0.4755 "* 0.1.188 *** 0,547 7 p*ip-Lp*ip-Lffti-ftLi 0.13ti-ftLi9 9 0.076* * 00 204 Wi«" " 0.11 70 0.11 in **" 0.256 6 ligiii ligiii -0.039 9 -0039 9 0053 3 wifll-1 1 0.039 9 0.034 4 0.072 2 rmfft-I I -0.188 8 00 219 *« 0.072 2 nnfli-2 2 -0.03) ) Ü.IG5 5 43.013 3 ti«ih-2 2 0.001 1 -0.034 4 -0.033 3 Xf< Xf< 0.199 9 01115 5 0.212 2 w> > 0.273 3 -0.060 0 0.229 9 $6pÓll $6pÓll 0.2CO O 0.015 5 O.ISS S «bjiJïi i 0.7155 — 0.468*** * 0,848 8 trx M.iwrttgM.iwrttg 2002 M.M. /wrtsM 2001 F» » Fjr r FM M N,m m (wilyy SstJinii) Fis s h, h, f.i i Hm m WJi-I I 0.165 5 •6.009 9 0.157 7 hBi-l hBi-l 00 361 * -0035 5 0,339 9 Lih-2 Lih-2 -COS S 0030 0 41.021 1 Lih-2 Lih-2 0.043 3 -0D26 6 0.018 8 pfp-JrfMito o 0.2* * -0.011 1 0.228 8 j\*p~lnifiln j\*p~lnifiln00 106 0.014 4 0.118 8 rxm rxm0.25 3 3 0.009 9 0.260 0 P*nt P*nt 0182 2 0.1411 "* 0.298 8 tiidb-tiidb- 1 0.6722 •** •0.O38 8 0.659 9 miBt-1 miBt-10.01 3 3 0.016 6 -0.027 7 iraft-2 2 turn turn -o.irw w -0010 0 mJfc-1 mJfc-1001 9 9 0.008 8 -0.011 1 XP XP -0.0H) ) 0,009 9 -0.ÜSU U W' ' O.U74 4 -41,034 4 -0.111 1 tfH',n, tfH',n, -0,097 7 0.029 9 -0.065 5 •jÉnilr r 0.303' ' 0.4599 *** 0.613 3 fipfifli i 0.3Off * 0.018 8 0.317 7 « ** ƒ><00 5 5 0.001 1 '*~~"" j*0.000pp <1 1< jurtinajurtina samples 'Monte Eccas 2001' clustered with the M. nurag group; both come fromm sites where the two species are sympatric. The neighbour-joining diagram hadd similar topologies as the UPGMA phenogram, and showed the same splitt into aa M. nurag, a M. jurtina group, and an intermediate group (Figure 3b). 155 5 Tablee 4. Nem values (corrected for small sample size) based on the private allele methodd (Barton & Slatkin, 1986). Mean sample size (n) and estimates (Nem) are presentedd for each year at different geographic scales: within Sardinia, Sardinia & Corsica,, and Europe. year r n n Nm Nm region n M.. nurag 2000 0 14.98 8 0.88 8 Sardinia a 2001 1 15.85 5 2.47 7 Sardinia a 2002 2 12.9 9 3.13 3 Sardinia a M.. jurtina 2000 0 10.46 6 1.64 4 Sardinia a 2000 0 8.6 6 1.27 7 Sardinia+Corsica a 2000 0 12.3 3 0.88 8 Europe e 2001 1 9.32 2 1.42 2 Europe e 2002 2 7.85 5 1.33 3 Europe e GeographicGeographic patterns in allele distribution Att some loci there was a geographic pattern to the frequency distribution of alleles (Figuree 4): Pgm-a was limited to M. nurag (0.029-0.083), the intermediate group (0.056-0.125),, and the M. jurtina 'Monte Eccas 2001' population (0.083) (Figure 4a);; pgm-g was restricted to four M. nurag populations (0.029 - 0.069); pep-leu-ala- cc was found in two M. jurtina populations (0.167-0.227) and some intermediate individualss (Figure 4b); aat-l-d was restricted to four M. jurtina populations (0.071- 0.167);; idh-l-a, aat-l-b were limited to Sardinian populations and did not occur onn the mainland. G6pdh~b was fixed or almost fixed in most Sardinian M. jurtina populationss (0.977 - 1.000) and overall common in all 'pure' populations of this speciess (0.500 - 0.952), whereas it occurred in lower frequencies in M. nurag (0.250 -- 0.625) but was more common in the intermediate form (0.438-0.813); g6pdh-c was raree in M. jurtina (0.045 - 0.286), more frequent in M. nurag (0.350 - 0.625), and the intermediatee form had intermediate levels of this allele (0.188 - 0.438). Hybrizymes Hybrizymes Otherr alleles showed clines in frequency from the mainland M. jurtina, over the Sardiniann M. jurtina to M. nurag, with intermediate individuals at intermediate frequencyy levels (Figure 5; only populations with H>5 were included in Figure 5),, and we consider them potential hybrizymes. Me-b was absent or rare in the 156 6 JJ Roquebrussanne 1999 JJ La Giara 2001 JJ Rio di Pula 2001 ]] Roquebrussanne 2000 JJ Amsterdam 2002 JJ Orgiva 2000 JJ Sette Fratelli 2002 JJ Rio di Pula 2000 ]] Sette Fratelli 2001 JJ Salzburg 2000 JJ Diirnberg 2000 NN Sette Fratelli 2000 NN Monte Novo 2000 NN Sette Fratelli 2001 NN Monte Novo 2002 NN Monte Eccas 2002 NN Sette Fratelli 2002 NN Monte Eccas 2001 NN Pir'e Onni 2001 NN Monte Fumai 2000 JJ Sette Fratelli 2000 Intermediatee 2001 Intermediatee 2002 0.200 0.00 A A Distancee (D) Figuree 3. (A) UPGMA diagram of Nei's genetic distance D (Nei, 1978) for M. nurag ('N')) and M. jurtina ('J') samples from different geographic areas and years with n >> 5 individuals. (Cophenetic correlation = 0.88). Population's names as in Table 1. (B)) Neighbour-joining tree with the same set of samples. 157 7 4A A 20022 2001 2000 20000 2001 2002 pep-leu-ala pep-leu-ala 4B B 20022 2001 2000 20000 2001 2002 158 8 4CC 2002 2001 2000 2000 2001 2002 Figuree 4 . Allele frequencies for M. nurag and M.jurtina in Sardinian samples from 2000-20022 at three allozyme loci; (A) pgm, (B) pep-Ieu-ala, (C) g6pdii. Numbers refer too populations listed in Table 1. White circles represent M. jurtina, grey circles M. nurag.nurag. On the map, areas where only M. jurtina was found are white, areas where onlyy M. nurag was found are shaded dark-grey, areas of the suggested hybrid zone aree shaded in light-grey. continentall M. jurtina (0 - 0.091), slightly more frequent in the Sardinian M. jurtina (0.125),, and most frequent in the intermediate group (0.389). Pgm-b was verv rare orr absent in all mainland populations of M. jurtina (0 - 0.091) except population 'Amsterdam',, but considerably more common in M. nurag populations (0.188 -- 0.591), with intermediate levels in the intermediate group (0.188 - 0.333). 6pgdh- cc was absent or rare in M. jurtina (0 - 0.068), higher frequent in M. nurag (0.119 -- 0.200) and still more frequent in the intermediates (0.125 - 0.498). 6pgdh-d was lowerr frequent on the continent (0.076 - 0.227) than in the Sardinian populations (0.1355 - 0.571) and the intermediate group (0.498). G6pdh-a was absent in 'pure' M.M. nurag populations, but more frequent at the sympatric sites (0.714), and low frequentt in the continental M. jurtina (0 - 0.214). SimilaritySimilarity matrix of allozymes Characterr loadings of the first- and second axis of a PCA analysis derived from individuall genetic profiles at the six allozyme loci, that appeared to be most distinctivee between the two species, and contained a high amount of variation (viz., 159 9 it»» v.. 50% % " " ü% % 1000 i 50% % _ _ 0% % -- ~ _ r r OO I KI 6pgdh 6pgdh •• < Dii i -- -- L Ob b i,, -:• -i sst,t,rr,!h ,!h 50% % " " bb . IÜ1ÖIE11IB8HIII: : Figuree 5. Allele frequencies for M. nurag and M. jurtina samples with n > 5 from Sardiniaa and Europe in the years 2000-2002 at four allozyme loci: we, pgm, bpdh, andd gSpdh. Figuree 6. Bivariate scatterplot of the character loadings from the first and the secondd PCA-axes derived from Jaccard's coefficient of association of individual allozymee profiles of the 6 most differentiating loci (idlt-1, idli-2, pep-leu-ala, pgm, gpi, gbpdli);gbpdli); included only individuals where all loci were analysed. Species member- shipp is indicated by outline polygones: dashed line for M. nurag, solid line for M. jurtina.jurtina. Individuals of the intermediate type are indicated with open circles. 160 0 idh-1,idh-1, idh-2, pep-hii-ala, pgm, gpi, and gópdh), described no clear clustering of data pointss (Figure 6). The horizontal axis (PCA-2) provides a separation into a M. nurag andd a M. jurtina group, with a zone of overlap. The 'intermediate' individuals are positionedd partly in this zone of overlap, partly outside the two main groups. Divergencee times Assumingg a clockwise behaviour of evolutionary rates for proteins (Thorpe 1982, andd references therein, Marchi et ah, 1996), we estimated species divergence times fromm genetic distance values as approximately 3.6 - 1 million years ago (ma), dependingg on the calibration used, with Nei's D = 18 ma (Thorpe 1982), or D = 5 ma(Nei,1972). . Discussion n ManiolaManiola nurag and M. jurtina are closely related species. This is reflected in the few species-specificc alleles, and the absence of diagnostic loci. Sixty-three of the alleles wee detected in our analysis were shared by both species. This genetic similarity betweenn the two species was also underpinned by the results obtained through the clusterr and PCA analyses (Figures 5 and 6). In both clustering methods, UPGMA andd Neighbour-joining, the samples were separated into a M. jurtina, a M. nurag group,, with the intermediate individuals apart. The separation of the 'Sette Fratelli 2000'' sample from their respective conspecifics could be an artefact due to missing dataa at certain loci (Swofford et al., 1996), as another sample from 'Sette Fratelli' fromm a different year clustered well within the M. jurtina group. Thee island populations of both species have equally high genetic variation as thee continental populations of M. jurtina. This finding is surprising, as island populationss (Frankham, 1997), and isolated populations in general (Cassel &z&z Tammaru, 2003) frequently have lower genetic variation in comparison to mainlandd populations (for review see: Frankham, 1997). In this review covering aa variety of taxa, including mammals, birds, plants, and arthropods, Frankham (1997)) found that a significant majority of island species had lower levels of geneticc variation than related mainland species (80%), and that insular endemic speciess showed less genetic variation than related non-endemic species (89%). Suchh a pattern was also found in a study on another endemic Sardinian butterfly, PolyommatusPolyommatus coridon gennargenti (Lycaenidae) (Marchi et ah, 1996). These authors detectedd severe inbreeding in the endemic lycaenid, low levels of genetic variation andd a low proportion of polymorphic loci (P = 17.6 %, H = 0.024) in comparison to itss continental relatives (P = 58.8 %, H = 0.185). In contrast to these findings, for the 161 1 endemicc Maniola nurag we found equally high levels of genetic variation (H=0.141- 0.270)) as in M. jurtina from the mainland (H=0.141-0.236). Why does this endemic butterflyy have such a high genetic variability? Dispersal ability is an important factorr increasing variation in island species (Frankham, 1997), as it increases the effectivee size of the population, and therefore counteracts drift and thus slows downn the march to homozygosity. In a three-year field study on M. nurag (Grill et al.,al., 2003b), it was concluded that butterflies disperse regularly over distances up too 2 km, and that there is a substantial amount of dispersal between neighbouring populations.. The findings from mark-release-recapture experiments are also well inn agreement with our estimates of gene-flow. Iff we calculate the heterozygosity values expected for neutral alleles (Kimura & Crow,, 1964) following H=4N^/1+47V,^), assuming that the species has a total populationn size of 50 000 individuals on the whole island (derived from an estimatedd number of 300 -1000 individuals per population as inferred in Grill et al., 2003b),, we obtain a similar heterozygosity level (H=0.09), like we found in our data (0.141-0.270),, considering that we only used polymorphic loci for the analysis. Sevenn percent of the tests for Hardy-Weinberg equilibrium revealed significant deviationss from equilibrium. All deviations were caused by a heterozygote deficit. Deviationss were detected at several loci in different populations, but there was no apparentt pattern to them, i.e., they were neither locus specific, nor did they occur consistentlyy at all loci of a certain population. Deviations from Hardy-Weinberg equilibriumm are often found in butterfly populations (Nève et al., 2000, E. Meglécz pers.. comm.), and do not always have a straightforward explanation. In the studiedd Maniola populations, they might be due to the Wahlund-effect, i.e., pooling off different populations, which are by themselves in Hardy-Weinberg equilibrium (Wahlundd 1928). Apart from the Wahlund-effect, other possible reasons for deviatingg from Hardy Weinberg equilibrium are recent immigration having nott yet reached equilibrium, inbreeding, underdominance, sex-linkage or null alleles.. Sex-linkage was not present; inbreeding can be excluded because it would influencee all loci at the same time, whereas underdominance and null alleles are locuss specific, and can thus also be excluded. Thee intermediate form - a hybrid? Thee individuals that we a priori defined as 'intermediates' according to wing patterns,, appear in the PCA scatterplot (Figure 6) partly as a non-overlapping group outsidee the overlapping clusters of the 'pure' M. jurtina and M. nurag, and partly inn the zone where both species overlap. The intermediate individuals also appear 162 2 ass a separate branch in the phenograms obtained by UPGMA or Neighbour-joined clustering.. Results of both analyses reflect that the allele frequency distribution in thee intermediate group is distinct from M. nurag as well as M. jurtina. In the zone off overlap, together with the phenotypic 'intermediates', cluster a majority of the individualss from the 'Sette Fratelli' and 'Monte Eccas' populations situated in the Southernn part of Sardinia. Thee 'Sette fratelli' population has been identified as one of the sites in Sardinia, wheree both species occur sympatrically in stable populations, i.e., are present inn similar abundances every year (Grill et al., 2003b). Ecological similarity of thesee sympatric populations suggests that hybridisation oï the two species is conceivable.. Genital preparations revealed no apparent anatomical pre-mating barrierss prohibiting hybridisation (Grill et al., 2003c). Interestingly, two individuals withh intermediate wing patterns, that were found to have also intermediate genitaliaa characteristics (Grill et al, 2003c), had a pgm-a allele, which was normally onlyy found in M. nurag. Thus, they can not be 'pure' M. jurtina; what is more, they containedd a me-b allele, one of the potential hybrizymes. At four other loci, we findd a cline in frequency towards the intermediate individuals (Figure 5). Me-b, for example,, is absent in most continental M. jurtina populations, very low frequent inn the French and the Sardinian lowland populations, but very well represented inn the intermediate individuals (0.389), and it is also present at the locality, where bothh species occur sympatrically. In pgm-b we observe a cline from 'pure' M. jurtina (00 - 0.091) towards Sardinian M. jurtina, intermediates (0.333), and finally M. nurag, withh highest frequencies in the last mentioned (0.591). G6pdh-c and g6pdh-d show aa similar cline towards intermediate individuals and the sympatric populations; andd likewise does g6pdh-a (Figure 5). These patterns look very similar to what has beenn observed by Schilthuizen & Gittenberger (1994) in snails, where a so called 'hybrizyme'' allele had a very low frequency outside the hybridzone, and peaked inn its centre. Low frequency Ft hybrids between an endemic and a widespread butterflyy species paralleled with first generation backcrosses have been detected betweenn the Sardo-Corsican endemic Papilio hospiton and the holarctic P. machaon (Cianchii et ah, 2003). Based on the data discussed here, we would anticipate a similarr scenario for M. nurag and M. jurtina. 163 3 Speciation n VicarianceVicariance and subsequent dispersal Inn his study of British populations of M. jurtina, Goulson (1993) found little geneticc variation, with only 4 out of 12 allozyme loci to be polymorphic; at the samee four polymorphic loci, we found a considerably greater amount of variation forr our samples. He further observed that in Great Britain geographically distant populationss differed very little in gene frequencies. Similarly to Sardinian and Centrall and Western European populations, also British M. jurtina population structuree did not follow an isolation by distance pattern. As island size generally correlatess with genetic diversity (Malone ct al. 2003), and England is considerably largerr than Sardinia, and also closer to the continent, this result is contrary to expectation.. An explanation might be, that in Britain, M. jurtina approaches the endd of its range, and the reduced genetic variability results hereof. Thiss comparison with the British M. jurtina once more illustrates, the rather high geneticc variability in Sardinian Maniola, which is also the main evidence against ann allopatric, vicariant speciation event of M. nurag in Sardinia, followed by a later invasionn of continental M. jurtina. If M. jurtina had colonized the island relatively recently,, we would expect a sign of a founder-event (genetic bottleneck) in our data.. As our genotype data from M. jurtina did not show traces of bottlenecks, wee conclude that it is unlikely that the widespread species has colonized Sardinia onlyy recently, unless the colonizing population was large, would have experienced rapidd growth directly after the founder-event, or continuous immigration from the mainlandd has taken place. Another,, theoretical consideration against the hypothesis that M. nurag has differentiatedd in allopatry, arises from a comparison with the distribution areas off the two other island endemics in this genus: the Greek endemic ill. chia and the Cypriotee endemic M. cyprkola have no distributional overlap with other Maniola species,, although their respective islands are much closer to the mainland than Sardiniaa is to Italy. Phenotypically, these species are even more similar to M. jurtinajurtina and M. telmessia respectively, than M. nurag is to M. jurtina. Consequently, inn these two endemics are likely to be the result of vicariance or dispersal. In M. nuragnurag on the other hand, the distributional overlap with M. jurtina, leaves room for speculationss on a sympatric mode of speciation. 164 4 SympatricSympatric differentiation Recentlyy diverged taxa often differ very conspicuously in secondary sexual characteristics,, which indicate the action of sexual selection on phenotypes involved inn mate recognition (Darwin 1879, Shaw & Parsons 2002). This is probably also the casee in M. nurag. However, the extent of morphological diversification does not necessarilyy reflect the extent of ecological radiation (Strauss 1984). In the case of M. jurtinajurtina and M, nurag, evidence from field data suggests that the two species respond too slightly different ecological gradients (Grill et al., 2003c). We assume that similar smalll ecological diversification could originally have initiated differentiation alongg an environmental cline, a scenario increasingly receives attention (Ogden & Thorpee 2002, Doebeli & Dieckmann 2003). Maniola nurag, which today is restricted too areas above 500 meters a.s.1., would have evolved into a mountain species, veryy well adapted to the particular conditions on Mediterranean mountains, with extremee aridity in summer, relatively cold winters, and extreme temperature shifts betweenn day and night in spring and autumn. Maniola jurtina on the other hand remainedd predominantly in the lowlands. The occasional hybridisation we suspect too happen between the two species, could be either the result of secondary contact andd have arisen in situ, similar to what has been suggested for P. hospiton (Cianchi etet al., 2003), or it indicates that the speciation process between M. nurag and M. jurtinajurtina is still going on. RelictRelict species Duringg the last glacial maximum, many plant and animal species that were adaptedd to warmer climates retracted to Mediterranean refugia in southern Europe.. At the end of the glaciation period, strong climatic fluctuations took place, withh rapid switches between warm and cold periods, leading to major changes in thee distribution and composition of the European flora and fauna {Hewitt 1999). Somee species with a formerly larger distribution area continued to persist only as aa relict species in a relatively small area. Boioria aquilionaris (Nymphalidae) is an arctic-alpinee example for such a relict (Baguette & Schtickzelle, 2003). Populationss isolated in refugia during the last glaciation usually exhibit reductions inn heterozygosity and genetic variation (Broders et al. 1999, Malone et al. 2003) as aa consequence of the bottleneck they passed when populations became isolated. Iff M. nurag also was a relict species, we again would expect a reduction of geneticc diversity, as a rapid severe reduction in distribution area and numbers of individualss due to climatic changes, is analogous to a severe bottleneck. In such a scenario,, we might also expect low numbers of private alleles. Usually, mostly the commonn alleles survive in bottlenecked populations, as happened in the endemic 165 5 PolyommaiusPolyommaius coridon gennargenti (Marchi et al, 1996). Maniola nurag on the contrary possessess a similar number of private alleles as the widespread species. Another argumentt against the relict-hypothesis is, that Tyrrhenian relicts are usually scatteredd over several islands (Kleinekuhle, 1999; Médail & Quézal, 1999), while M.. nurag is restricted to Sardinia. Divergencee time estimates Despitee the continuing controversy about molecular clocks (Arbogast et al., 2002), wee used a 'sloppy' protein-based molecular clock to obtain some rough estimates off divergence times. For the split of M. nurag from M. jurtina from a common ancestor,, we estimate a divergence time, of 3.6 - 1 million years, and thus falls into thee late Pliocene. These estimates happen to coincide with the major biotic turnover inferredd for this period (3 - 1.8 Ma ago) (Vrba 1985, Vrba 1992); significant faunal changee and increased speciation events appear to have happened 1.8 - 2.5 ma agoo as a result of major climate changes toward a cooler, drier, and more variable climate.. This massive species turnover has been viewed as part of the 'turnover pulse'' hypothesis, which postulates that climate change results in brief periods of significantt evolutionary change (Vrba 1985, Vrba 1992, Behrensmeyer et al., 1997). Forr example, the late Pliocene radiation of hominid species, and ultimately even the emergencee of the genus Homo have been attributed to such global climatic events (dee Menocal 1995). In the Mediterranean region, this climate change forced warm andd humid subtropical forests to gradually change into a savanna-like vegetation {LaGreca{LaGreca 1998). Caccone & Sbordoni (2001) attribute the differentiation of endemic Sardiniann cave beetles in the genus Patriziella to Pliocene climate changes. Our ownn divergence time estimates for M. nurag and M. jurtina are congruent with thee estimates presented by Marchi et al (1996) for the Sardinian endemic lycaenid butterflyy Polyommatus coridon gennargenti. As Marchi et al. (1996) used the same calibrationss we used (i.e., Nei's D = 18 ma, or D = 5 ma), our estimates are directly comparablee with theirs. Calculations of divergence times for other Sardinian taxaa (references in Grill et al, 2003a), show that the split of the Sardinian lineage hass occurred after the Messinian crisis (+ 5 ma), when marine regressions led to aa desiccation of the Mediterranean sea (Steininger & Ro'gl, 1984), and after the continuingg marine regressions from late Miocene to Pliocene (Arias et al. 1980; Citaa 1976) that brought Sardinia in contact with continental Italy and southern Francee (5.7 - 2 Ma ago) (Grill et al., 2003a). During the last glacial maximum, 20 000 yearss ago, Sardinia was connected with Corsica for the last time. The last glacial maximumm is the very period, when M. jurtina differentiated into a western and an 166 6 easternn group from two disjunct glacial refugia (Schmitt, 1999), analogous to what hass been shown for the bear, Ursus arctos (Taberlet & Bouvet, 1994). Notably, g6pdh, wheree we found a potential hybrizyme, was also one of the two loci indicating ice- agee induced genetic differentiation in Schmitt's study. In Schmitt's (1999) scenario, thee Sardinian M.jurtina would belong to the western lineage. Obviously, allozymes cann only provide very approximative estimations, but could be useful for relative comparisonss with other Sardinian taxa. A more precise dating of the speciation eventt of M. nurag awaits a phylogenetic analysis based on a combination of DNA sequencess and wing pattern characteristics, and would entail the verification of the sympatricc speciation hypothesis we proposed here. Acknowledgments s Wee thank Léon E.L. Raijmann for constant support, statistical and technical advice, andd many insightful discussions; Steven Weiss for linguistic improvements; Pirn Arntzenn and Peter Roessingh for critical reading of an earlier version of this manuscript. . 167 7