Music & Video

Italy as a major Ice Age refuge area for the bat Myotis myotis (Chiroptera: Vespertilionidae) in Europe

Description
Italy as a major Ice Age refuge area for the bat Myotis myotis (Chiroptera: Vespertilionidae) in Europe
Categories
Published
of 14
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Related Documents
Share
Transcript
  Molecular Ecology (2008) 17 , 1801–1814 doi: 10.1111/j.1365-294X.2008.03702.x© 2008 The Authors Journal compilation © 2008 Blackwell Publishing Ltd BlackwellPublishingLtd Italy as a major Ice Age refuge area for the bat  Myotis myotis  (Chiroptera: Vespertilionidae) in Europe MANUEL RUEDI,* SIMON WALTER,† MARTIN C. FISCHER,† DINO SCARAVELLI,‡ LAURENT EXCOFFIER† and GERALD HECKEL†* Natural History Museum of Geneva, PO Box, 1211 Genève 6, Switzerland, † Computational and Molecular Population Genetics Laboratory, Zoological Institute, University of Berne, Baltzersrasse 6, 3012 Berne, Switzerland, ‡ Dino Scaravelli, IUCN SSC Bat Specialist Group, Associazione Chiroptera Italica, via Veclezio 10 A, 47100 Forlì, Italy Abstract The distribution of biota from the temperate regions changed considerably during theclimatic fluctuations of the Quaternary. This is especially true for many bat species thatdepend on warm roosts to install their nursery colonies. Surveys of genetic variation amongEuropean bats have shown that the southern peninsulas (Iberia and the Balkans) harbourendemic diversity, but to date, no such surveys have been conducted in the third potentialglacial refuge area, the Apennine peninsula. We report here the phylogeographical analysisof 115 greater mouse-eared bats (  Myotis myotis ) sampled throughout Italy, and show that15 of the 18 different haplotypes found in the mitochondrial control region of these batswere unique to the Apennine peninsula. Colonies within this region also showed substantialgenetic structure at both mitochondrial ( F ST =0.47, P  <0.001) and nuclear markers(  F  ST =0.039, P  <0.001). Based on a comprehensive survey of 575 bats from Europe, thesegenetic markers further indicate that central Italian populations of  M. myotis  are moreclosely related to Greek samples from across the Adriatic Sea, than to other Italian bats.Mouse-eared bat populations from the Apennine peninsula thus represent a complex mixtureof several endemic lineages, which evolved in situ  , with others that colonized this regionmore recently along an Adriatic route. Our broad survey also confirms that the Alps representa relatively impermeable barrier to gene flow for Apennine lineages, even for vagile animalssuch as bats. These results underline the conservation value of bats from this region and theneed to include the Apennine peninsula in phylogeographical surveys in order to providea more accurate view of the evolution of bats in Europe. Keywords :bats, Chiroptera, glacial refugium, mitochondrial DNA, population structure , STR Received 17 August 2007; revision received 26 November 2007; accepted 11 January 2008 Introduction The climatic fluctuations during the Pleistocene had majorimpacts on the distribution range of biota, especially inthe temperate regions (see, e.g. Hewitt 2000). Contraryto the North American continent, which has a considerablelatitudinal extension of continuous lands, the Europeancontinent is bound to the south by several transversemountain ranges (the Alps, the Pyrenees, the Balkans), and by the Mediterranean Sea. The retreat of temperate biotaduring ice ages was thus limited to three main refuge areas,the Iberian, the Italian and the Balkan peninsulas. Patternsof gene diversity assayed in a variety of organisms supportthis paradigm as most phylogeographical analyses identifiedone or several of these refuge areas as the source of recolonization for more northerly regions (Taberlet etal .1998; Hewitt 1999). Bats were severely impacted by theseclimatic fluctuations because of their dependence onrelatively mesic habitats to feed and reproduce (Altringham1996), and therefore their distribution range must havechanged dramatically during the glacial cycles. This isespecially true for the greater mouse-eared bat (  Myotismyotis ), which needs forested habitats to find its main prey,carabid beetles (Audet 1990), and warm caves to establishits nursery colonies (Güttinger etal . 2001). The colonies Correspondence: Manuel Ruedi, Fax: +41 22418 63 01; E-mail: manuel.ruedi@ville-ge.ch  1802 M. RUEDI ETAL. © 2008 The Authors Journal compilation © 2008 Blackwell Publishing Ltd typically contain adult females and their offspring, butonly rarely adult males. The current distribution range of this sedentary species (see inset of Fig.1) is essentiallylimited to continental Europe and few Mediterraneanislands; its oriental limit of distribution reaches Crimea andthe Near East (Arlettaz etal . 1997; Castella etal . 2000).Despite their ability to fly and disperse over large areas, bats often show strong population structure as evidenced by variation of maternally inherited mitochondrial DNA(Burland etal . 1999; Kerth etal . 2000; Burland & Worthington-Wilmer 2001; Ruedi & Castella 2003). Bat dispersal intemperate species is usually highly biased towards males(Castella etal . 2001; Petit etal . 2001; Kerth etal . 2002; Rivers etal . 2005; Senior etal . 2005); thus, population structure isapparently much lower when measured at nuclear markers, but signs of range expansion from restricted refuge areasare still detectable. In a previous study based on 24 popu-lations of greater mouse-eared bats surveyed across Europe(Ruedi & Castella 2003), we have shown that the vastmajority of individuals sampled in Spain, France, Switzer-land and Poland bear mtDNA haplotypes representative of a single major clade (haplogroup A) of presumed Iberiansrcin. Five other major clades (B to F) were identified infew admixed areas in the Alps and in Greece (Fig.1), buttheir exact srcins could not be identified because of in-adequate geographical sampling. One important feature of the admixed populations in the Alps was the presence of two major, well-differentiated clades (B and C) in the onlysample of 20  M. myotis  from northern Italy. The mixedancestry of this population was detectable with nuclearmarkers as well, which clearly supports the existence of several unidentified source populations for this Italiannursery colony (Castella etal . 2001). Furthermore, theApennine peninsula was never sampled adequately to testwhether this potential refuge area harbours endemic geneticdiversity, as is found in many other indigenous organisms(e.g. Pierpaoli etal . 1999; Randi etal . 2003; Brito 2005; Fritz etal . 2005; Heckel etal . 2005; Hofman etal . 2007), or whetherthis area is inhabited by widespread European lineages.Given the current need to define practical units of conser-vation in the management of wild populations (Frankham etal . 2002), the detection of major genetic components is animportant issue that was investigated in the present study.In particular, we used both phylogenetic and populationgenetic methods to determine the structure in nuclear andmitochondrial DNA, and estimate more precisely the com-position and srcins of  M. myotis  bats from the Apenninepeninsula and Sicily. By pooling these new data with resultsfrom previous surveys of genetic variation, we analysehere one of the largest data set on a wild species of mammalin Europe to infer phylogeographical patterns. This is espe-cially relevant because bats were never transported ormanaged by men, unlike many other well-studied gamespecies such as roe deer (Lorenzini etal . 2002) or hare(Pierpaoli etal . 1999). Because  M. myotis  has been recentlyshown to be an important reservoir for bat rabies virus(Serra-Cobo etal . 2002; Amengual etal . 2007), a betterunderstanding of population structure and gene flow inthis species has also important bearings for public health. Materials and methods Sampling One hundred fifteen  Myotis myotis  bats from six nurserycolonies were sampled every ~300km along a transect of 1330km across Italy. The geographical coverage of thissurvey includes the island of Sicily, separated from theApennine peninsula by the Strait of Messine (3.4km of open sea) and the Apennine Range (Fig.1). These physicalfeatures might represent barriers to gene flow for Italian bats. Most nursery colonies were located in underground Fig.1 Map showing the location of 29sampled  Myotis myotis  nursery colonies.Mitochondrial DNA haplogroups corres-pond to those presented on the phylogeneticreconstruction in Fig.2 and pie charts showtheir frequencies in each colony. Numberson the pie charts correspond to the colonynumbers of Table1. The inset illustrates thecurrent distribution of  M. myotis .  ITALIAN BAT PHYLOGEOGRAPHY 1803 © 2008 The Authors Journal compilation © 2008 Blackwell Publishing Ltd roosts (caves, lava tunnels or mines), that are typicalrefuges for  M. myotis  living in Mediterranean areas(Güttinger etal . 2001). Exceptions are the two colonies fromthe Alpine Piemont (Bolzano and Aglié; Fig.1) that occupiedhuman buildings. The colony of Aglié was sampled in 1999(Castella etal . 2001), whereas the remaining ones from theApennine peninsula and Sicily were sampled during thesummer 2002.For the general survey of genetic variation acrossEurope, we used these 115 Italian and 460 European batssampled in a further 23 colonies (Ruedi & Castella 2003).Thus, a total of 575 bats were analysed from 29 nurserycolonies that cover most of the European range of  M. myotis (Fig.1). A mean of 20 bats per roost was used for the geneticanalyses (Table1). Nondestructive tissue samples wereobtained from a sterile biopsy punch of the wing mem- brane (Worthington-Wilmer & Barratt 1996) preserved inabsolute ethanol until processing. Animals were measured,identified and released within 1 hour upon capture,according to the appropriate capture licenses. Specialattention was paid to differentiate  M. myotis  from  Myotisblythii  and possible hybrids (Berthier etal . 2006). GenomicDNA was extracted from half wing punches following asalt/chloroform procedure modified from Miller etal . (1988) by adding one step of chloroform/isoamyl alcohol (24:1)extraction to the srcinal protocol. DNA amplification and sequencing We sequenced the second hypervariable domain (HVII)of the mtDNA control region, as described in Castella etal .(2001). The primers L16517 (5 ′ -CATCTGGTTCTTACT-TCAGG-3 ′  Fumagalli etal . 1996) and sH651 (5 ′ -AAGGCT-AGGACCAAACCT-3 ′  Castella etal . 2001) were used and Table1 Analysis of molecular variation of 29  Myotis myotis  nursery colonies genotyped at 307bp of hypervariable mtDNA and 11 STR loci.The following parameters were estimated: number of individuals genotyped per colony ( n ), total number of haplotypes (  A ), haplotypediversity ( h ), nucleotide diversity ( π ), allelic richness ( k  ), observed (  H  0 ) and expected (  H  E ) heterozygosities. F, France; GR, Greece; I, Italy;PL, Poland; E, Spain; CH, SwitzerlandNbColony nA mtDNASTRs h  π  (%) nkH  0  H  E 1Catania (I)2040.43±0.130.26±0.22216.550.590.622Crotone (I)2140.62±0.090.99±0.60216.550.610.633Foggia (I)2140.61±0.091.50±0.86217.550.600.614Rimini (I)1640.64±0.083.66±1.97237.360.670.685Bolzano (I)1740.33±0.140.46±0.33208.360.740.736Aglié (I)2040.57±0.091.82±1.02207.910.720.707Eysins (CH)2040.55±0.112.08±1.15218.820.780.788Fully (CH)2020.34±0.111.10±0.66207.450.760.749Raron (CH)2040.65±0.071.73±0.97208.180.770.7410Naters (CH)2030.63±0.081.27±0.74207.450.710.7311Corbières (CH)2030.56±0.060.20±0.18209.640.730.7612Courtetelle (CH)2020.10±0.090.03±0.06208.450.760.7613Perreux (CH)2030.47±0.100.16±0.16209.270.770.7814Meiringen (CH)2020.19±0.110.06±0.09208.550.770.7715St-Hippolyte (F)2020.10±0.090.03±0.06209.550.790.7816Balme d’Epy (F)2050.76±0.060.33±0.26209.090.810.7817Annonay (F)2020.34±0.110.22±0.20209.180.750.7918Ougney (F)2050.57±0.120.21±0.19209.090.730.7519Chabanne (F)2030.61±0.060.34±0.262010.000.780.7620Cádiz (E)2050.51±0.130.35±0.272010.000.740.7721Malaga (E)2060.81±0.050.58±0.39209.500.750.7822Granada (E)2040.78±0.040.66±0.43209.400.780.7723Alicante (E)2040.50±0.120.31±0.24209.100.740.7524Castellone (E)2030.35±0.120.28±0.232010.000.800.7925Mallorca (E)2030.62±0.060.23±0.20206.800.660.7026Lesna (PL)2030.19±0.110.07±0.09209.000.760.7527Kastria (GR)2040.43±0.130.31±0.24206.500.640.6728Serres (GR)2060.57±0.121.14±0.68208.200.730.7429Maronia (GR)2030.59±0.071.42±0.82208.500.730.75Mean for Europe3.620.470.658.480.730.74  1804 M. RUEDI ETAL. © 2008 The Authors Journal compilation © 2008 Blackwell Publishing Ltd amplifications were carried out in 50  µ L reaction volumesunder the following conditions: 10  µ L of template DNA,2.5m m  MgCl 2 , 0.2  µ L of each primer, 200 µ L of each dNTP,2 U of Taq  DNA Polymerase (QIAGEN) with its buffer andthe Q-solution. Thermal profiles started with an initialdenaturation at 95  ° C for 3min, followed by 35 cyclesconsisting of 45s at 94  ° C, 45s at 50  ° C and 1min at 72  ° C.The cycles ended with one final extension of 7min at 72  ° C.Polymerase chain reaction (PCR) products were purifiedwith the GenElute PCR Clean-Up Kit (SIGMA). Fourmicrolitre of cleaned PCR products were then mixed with4  µ L of BigDye Terminator version 3.0 Cycle SequencingKit (Applied Biosystems) and 0.5  µ L of primer L16517 in atotal volume of 10  µ L. Thermal profiles of the sequencingreaction consisted of 25 cycles of 30s at 96  ° C, 20s at 50  ° Cand 4min at 60  ° C. After purification, the reactions wererun on an ABI 3100 sequencer (Applied Biosystems).All sampled bats were also genotyped at 11 short tandemrepeat (STR) loci (B11, B22, C113, D9, E24, F19, G9, G25,H19, H29 and G30) with primers developed by Castella& Ruedi (2000). Forward primers were labelled with thefollowing dyes: 6-FAM (locus C113, D9, E24, F19, G9 andH19), HEX (locus G25) and NED (locus B11, B22, H29 andG30). These loci were amplified as two multiplex reac-tions including B11, C113, D9, E24, G25, H29, and B22, F19,G9, G30, H19, respectively. Amplifications were carriedout in 10 µ L reaction volumes, following the manufacturer’sprotocols (Multiplex PCR master kit of QIAGEN) andusing an annealing temperature of 57  ° C. Single loci thatdid not amplify properly in a multiplex reaction wererepeated singly. PCR products were run along with theGenScan 500–ROX size standard on an ABI 3100 sequencerand genotypes determined with genemapper  version 3.0(Applied Biosystems). Statistical methods  Mitochondrial DNA analyses. The 307 base pairs (bp) of theHVII control region were aligned and edited manuallyusing bioedit  version 5.0.6 (Hall 2001). The typical R2tandem repeats located immediately after this uniquesegment (see Castella etal . 2001) were excluded. Haplotypesand their frequencies were determined using arlequin  3.1(Excoffier etal . 2005). Intracolonial haplotypic variabilitywas quantified as the number of haplotypes (  A ), gene ( h )and nucleotide ( π ) diversity. Phylogenetic relationshipsamong haplotypes were reconstructed with Bayesianand distance methods, using mrbayes  3.1 (Ronquist &Huelsenbeck 2003) and mega  3.1 (Kumar etal . 2004),respectively. The Bayesian tree was obtained by runningtwo Markov chains of one million generation each and alldefault settings. The general time reversible model of DNAevolution with gamma-shaped rate of variation and aproportion of invariable sites (Yang 1993) was selected asthe evolutionary model. We used the same model of evolution to calculate the matrix of pairwise distances between haplotypes and the neighbour-joining (NJ) tree(Saitou & Nei 1987). Three bats of the species  M. blythii from Kyrgyzstan were chosen as outgroups to root thephylogenetic trees. Supports of nodes were assessed eitheras posterior probabilities (Bayesian tree, with the initial2000,000 generations considered as burn-in), or as bootstraps(NJ tree, 10000 pseudoreplicates).The level of mitochondrial genetic differentiation of colonies within Italy and within Europe was assessedwith a hierarchical analysis of molecular variance ( amova ;Excoffier etal . 1992) by computing overall and pairwise Φ ST  with arlequin . The program samova  (Dupanloup etal . 2002) was also used to find groups of populationswith maximum extent of genetic differentiation, as sum-marized by the F CT  statistic (Excoffier etal . 1992). Thisanalysis also delineates genetic barriers between theinferred groups. samova  was run assuming up to six groupsfor Italy and up to 12 groups for Europe.To check for a pattern of isolation by distance (IBD)within Italy, we plotted Φ ST values linearized as Φ ST / (1– Φ ST )(Slatkin 1995) against the geographical distance expressedin kilometres. As localities were arranged along a relativelylinear transect (i.e. close to a one-dimensional case), straightdistance was preferred over log-distance, following Rousset(1997). The significance of the correlation was tested withthe Mantel test implemented in arlequin . From the samelinearized genetic distances, a population tree was recon-structed with the NJ algorithm to visualize relationshipsamong colonies for the mtDNA marker. STR analyses. The hypothesis of random mating and thepotential occurrence of null alleles were tested by computinginbreeding coefficients ( F IS ) and by performing exact testsof Hardy–Weinberg equilibrium (HWE) using arlequin .Tests of linkage disequilibrium (LD) between pairs of loci were performed with the same program. The nucleargenetic variability within colonies was then quantified bythe mean allelic richness across loci ( k  ) calculated with fstat  2.9.3.2 (Goudet 2002), and the observed (  H  O ) andexpected (  H  E ) heterozygosities calculated with arlequin .Nuclear population structure within Italy was quantifiedinitially with both F ST  (Weir & Cockerham 1984) and R ST (Slatkin 1995) values. However, as both values producedessentially similar results, we only report here those basedon F ST . Significance of the F ST  values was tested by 10000permutations. The genetic relationships between colonieswere visualized on an NJ tree based on a matrix of pairwise F ST values linearized as F ST /(1– F ST ) (Slatkin 1995). Thisgenetic distance was plotted against the straight geograph-ical distance (in kilometres) to test the hypothesis thatpopulations follow a pattern of IBD, using the Mantel testimplemented in arlequin . As for the mtDNA data set, we  ITALIAN BAT PHYLOGEOGRAPHY 1805 © 2008 The Authors Journal compilation © 2008 Blackwell Publishing Ltd then performed amova  and samova  tests to find groups of populations that are maximally differentiated, investigatingup to six groups for Italy and up to 12 for Europe. Results  Mitochondrial DNA variability The second hypervariable segment (HVII) of the controlregion was newly sequenced in 95 Italian bats, in additionto the 20 bats from Aglié already sequenced in a previousstudy (Castella etal . 2001). Of the 307 aligned nucleotidepositions, 36 were variable among the Italian data set(AppendixI). Most inferred substitutions (31 out of 36)were transitions, with only five transversions and noalignment gap, a typical pattern of substitution amongclosely related mtDNA sequences (Avise 2000). Thesevariable sites defined 18 distinct haplotypes among the sixItalian colonies (AppendixI). Five of these haplotypes(H01 to H16) were already described in Castella etal .(2001), but the other 13 haplotypes have not been describedyet in other European samples. The most common haplotypein Italy, H12, characterizes 61 out of 115 individuals (53% of  bats) and is found in all Italian colonies, but nowhere elsein Europe (AppendixII). H51 is found in the adjacentcolonies of Crotone and Foggia, but all other Italianhaplotypes are private to a single colony. A single batsampled in the Alpine colony of Bolzano in northern Italy bears the widespread haplotype H01, which is otherwisefound throughout Europe (Ruedi & Castella 2003). The twoother haplotypes shared with non-Italian bats (H13 andH14) are exclusively detected in the colony of Aglié, and infour colonies in nearby Switzerland (Castella etal . 2001).At the population level, each Italian colony containedfour distinct haplotypes (Table1), which is similar to thehaplotype diversity found in the rest of Europe (mean  A= 3.62, range 2–6). Gene diversity was also relativelyuniformly distributed among Italian colonies (mean h =0.535) and compared well with colonies in other Euro-pean regions (mean 0.47, range 0.10–0.81). However, whentaking into account the degree of haplotype divergence,variation is more heterogeneous among Italian colonies,with nucleotide diversities ( π ) ranging from 0.26 to 3.66%(mean 1.45%; Table1). The highest value is found in thecolony of Rimini (central Italy) where the nucleotidediversity (3.66%) is considerably higher than that foundelsewhere in Europe (mean 0.65%, range 0.03–2.08%). Phylogenetic relationships The NJ tree of haplotypes that includes both Italian andother European samples is presented in Fig.2. Bayesianreconstructions produced a very similar haplotype tree,with only minor variations within clades A and C (resultsnot shown). Both phylogenetic reconstructions support theexistence of seven major clades within Europe, five of whichare present in Italy. Interestingly, three divergent haplotypesfound in the colony of Rimini (H54, H55 and H56) clusterwith those of the Greek colony of Kastria (haplotypesH40 to H43) in haplogroup F (Fig.2). This unexpectedrelationship was highly supported by both bootstrap (98%)and posterior probability (1.0) values. A new haplogroup Gis also highly supported in these phylogenetic recon-structions and is found exclusively in southern Italy. Theother Italian haplogroups are also distributed in the Alpineregion of Switzerland (haplogroups B and C) or over mostof Europe (haplogroup A). The mean uncorrected geneticdivergence of haplotypes ranges from 0.4% (withinhaplogroup D or E) to 1.5% (within haplogroup F), whileeach haplogroup differs from the others by a mean distanceof 4.4% (range 2.8–7%, see AppendixIII).Contrary to the general pattern found elsewhere inEurope, where most colonies are composed of bats from asingle haplogroup, each colony of  Myotis myotis  from theApennine peninsula is admixed with several major haplo-groups (Fig.1 and AppendixII). This suggests that thesecolonies were founded by very different matrilines, and/orthat divergent matrilines persisted over considerableevolutionary periods. Population genetic structure The overall high fixation index ( Φ ST =0.47, P <0.001)suggests a very strong genetic structure among the Italiancolonies at the mitochondrial level. The only nonsignificantvalues are found among the three colonies of Catania,Crotone and Bolzano (values of Φ ST smaller than 0.07, whichare not significant at the 5% level; Table2). Surprisingly,Bolzano and Catania are geographically the most distantlocalities along the Italian transect (>1330km). Contras-tingly, the colonies of Aglié and Bolzano are separated byonly 330km straight distance, yet they show a remarkablyhigh level of genetic differentiation ( Φ ST =0.58, P <0.001).Thus, there is no significant correlation between geneticand geographical distance ( r 2 =0.025, P =0.44), indicatingthat the mtDNA gene flow among Italian populations of   M. myotis  does not follow a simple pattern of IBD.The program samova  was run assuming up to six groupsto search for the strongest substructure within the Italianpopulations for the mtDNA data set. The highest percentageof variation among groups (52.8%) was explained whenthe number of K   groups was set to four. Not surprisingly,the three colonies of Catania, Crotone, and Bolzano, whichare not significantly differentiated from each other (Table2)form a single group in this samova  test. The other threeItalian colonies form each a group on their own. The sameanalysis applied to the complete European mtDNA dataset (i.e. 29 colonies) suggested the existence of nine groups
Search
Similar documents
View more...
Related Search
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks