Arts & Architecture

Contrasting levels of genetic differentiation among putative neutral microsatellite loci in Atlantic herring Clupea harengus populations and the implications for assessing stock structure

Contrasting levels of genetic differentiation among putative neutral microsatellite loci in Atlantic herring Clupea harengus populations and the implications for assessing stock structure
of 7
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
  CHALLENGES TO MARINE ECOSYSTEMS Contrasting levels of genetic differentiation among putativeneutral microsatellite loci in Atlantic herring  Clupea harengus  populations and the implications for assessingstock structure Phillip C. Watts   David O’Leary   Mary C. Cross   Jamie Coughlan   Eileen Dillane   Suzanne M. Kay   Suzanne Wylde   Rene´ Stet   Richard D. M. Nash   Emma M. C. Hatfield   Thomas F. Cross   Springer Science+Business Media B.V. 2008 Abstract  Microsatellite DNA loci, when used inpopulation genetic studies, are usually assumed to beneutral (unaffected by natural selection, eitherdirectly or as a result of tight linkage), but thisassumption is rarely tested. Here, the assumption of neutrality is examined using established methods,principally that based on the expected relationshipbetween  F  ST  and heterozygosity, at 12 putativeneutral microsatellite loci utilised in a study of Atlantic herring  Clupea harengus  in the north eastAtlantic (west of Great Britain and around Ireland)and in the Baltic Sea. All but two of these locidemonstrate relationships that suggest that they maybe regarded as neutral genetic markers. Of the othertwo loci, however, one shows a relationship sugges-tive of the action of directional selection and the otherof balancing natural selection, though other locus-specific effects may operate. Thus, the latter two locimay provide inaccurate inference if used in phylog-eographic studies and also demonstrate the danger of assuming neutrality at all microsatellite loci withoutexplicit testing. However, such loci, particularlythose affected by directional as opposed to balancingselection, may be of great use in stock discriminationstudies, and selected loci in general, have consider-ably potential in studies of adaptation. Keywords  F  ST    Microsatellite   Neutrality    Selection Introduction Molecular genetic markers offer a convenient solu-tion to the problem of inferring connectivity amongpopulations of species that are difficult or impossibleto track directly. Accordingly, genetic techniqueshave been used widely to study dispersal in aquaticorganisms, with particular emphasis on defining stock  Guest editors: J. Davenport, G. Burnell, T. Cross,M. Emmerson, R. McAllen, R. Ramsay & E. RoganChallenges to Marine EcosystemsP. C. Watts ( & )    S. M. KayMarine and Freshwater Biology Research Group,The Biosciences Building, School of Biological Sciences,University of Liverpool, Crown Street,Liverpool L69 7ZB, UK e-mail: D. O’Leary    M. C. Cross    J. Coughlan   E. Dillane    S. Wylde    T. F. CrossDepartment of Zoology, Ecology & Plant Science,Aquaculture & Fisheries Development Centre, NationalUniversity of Ireland, Cork, IrelandR. StetUniversity of Aberdeen, Aberdeen, UK R. D. M. NashInstitute of Marine Research, PO Box 1870, Nordnes,5817 Bergen, NorwayE. M. C. HatfieldFRS Marine Laboratory, Aberdeen, UK   1 3 Hydrobiologia (2008) 606:27–33DOI 10.1007/s10750-008-9350-z  boundaries for fisheries management. Although notable to provide a migration rate per se (see Whitlock & McCauley, 1999), Wright’s (1951) estimator of  genetic differentiation among populations ( F  ST )is broadly correlated with dispersal capability(Bohonak, 1999; Kinlan & Gaines, 2003; Watts & Thorpe, 2006) and is the most widely employedmeasure of population connectivity. The basic pre-mise behind the use of this (and also many othergenetic statistics) is that the spatial distribution of allele frequencies is a consequence of the effects of migration, genetic drift, mutation and selection. Foran estimate of migration, it is necessary to make twoimportant assumptions: (1) mutation has a negligibleeffect during the contemporary timescale of the studyand (2) the loci used for genetic investigations are(relatively) unaffected by selection, so-called neutralmarkers.Typically, mutation is sufficiently rare to beignored when quantifying dispersal at scales relevantto interacting populations, whereas the general pre-sumption of marker neutrality is more contentious.For example, many studies of marine species presentevidence for a possible influence of selective pro-cesses operating at certain genetic loci (e.g. Hilbish &Koehn, 1985; Pogson et al., 1995; Lemaire et al., 2000; Karlsson & Mork, 2003; Case et al., 2005). Notably, the assumption of neutrality is seldomquestioned at putative non-coding markers such asmicrosatellites because they are expected to respondto selective processes only rarely, for instance whenthey are linked to a functional gene, i.e. ‘‘hitch-hiking’’ selection. A crucial point is that differentevolutionary processes leave characteristic signatureson the genome: migration and genetic drift areexpected to affect all loci more or less equally whileselective processes should operate on a specificsubset of gene regions. This suggests a framework to test for selective neutrality among a panel of genetic markers, with neutral loci demonstratingbroadly similar levels of   F  ST  and balancing ordirectional (spatial) selection at a locus indicated byatypically low or high values of genetic differentia-tion respectively. Failure to recognise and identifythese effects can lead to misinterpretation of theevolutionary processes that generate spatial geneticstructure.Lewontin & Krakauer (1973) proposed that direc-tional selection could be detected by significantlylarge observed variance in values of   F  ST  ( s 2 F  )compared with a theoretically expected variance of  F  ST  among loci ( r 2 F  ) that is equal to  kF  ST 2  /( n - 1),where  k   is a constant specific to the underlyingdistribution of allele frequencies among populations, F  ST  is an average (over all loci) and  n  is the numberof populations sampled. This test has been criticised,particularly because it is sensitive to the model of population structure and also correlated allele fre-quencies among populations (Robertson, 1975; Nei &Maruyama, 1975), and, as a consequence, is consid-ered unreliable and rarely used. Beaumont & Nichols(1996) developed an alternative method to detectoutlying loci, which is based on an expected distri-bution of   F  ST  conditional on heterozygosity (  H  e ), andseveral alternative statistical tests for selection havebeen proposed subsequently (reviewed by Guinandet al., 2004). It is apparent, however, that there is ageneral lack of an explicit inspection for locus-specific effects, with heterogeneity among lociexplained as a stochastic (sampling) effect (underthe  untested   assumption of locus neutrality) oroverlooked when the data set is presented as multi-locus averages. With this in mind, Baer (1999) usedthe Lewontin-Krakauer test to examine 102 publishedallozyme data sets on fish and concluded that ‘locus-specific’ processes were not sufficiently strongenough to mask real population structures.Given the widespread use of genetic markers forfisheries research (e.g. see EU-funded projects: CODTRACE ,;  HERGEN , ;  WESTHER , westher/ ), it is of substantial importance that outlyingloci are identified during a genetic analysis of popu-lation genetic structure or history. In this paper weexamine the extent of variability among a panel of 12microsatellite loci that were used to characterise thelevel of genetic differentiation among putative stocksof Atlantic herring  Clupea harengus . Atlantic herringstocksintheNorthandBalticSeasarecharacterisedbysubstantial differences between the North Sea and theBaltic (Ruzzante et al., 2005), but weak if any popu-lation structure within the North Sea (Mariani et al.,2005) or Irish Sea, Celtic Sea and west of Ireland(WESTHER EU consortium, unpublished data).Here,wequantifythelevelofheterogeneityinallelefrequencies at 12 putative neutral microsatellite lociusing several methods that were developed to identifyoutlying loci that may be responding to selective 28 Hydrobiologia (2008) 606:27–33  1 3  processes. Under a shared, neutral demographichistory, all genetic markers should engender equiva-lent values of   F  ST . However, we uncover significantheterogeneity in the level of spatial genetic differen-tiation among loci, suggesting that selective neutralitycannotbeassumedwithinapanelofmicrosatelliteloci. Methods Sample collection and genotypingSpawning herring were sampled between 2003 and2005 from 10 locations in the North Atlantic: six sitestothewestofGreatBritainandIreland,twositesofftheNorwegian Coast, one area in the Baltic Sea and onefrom Hudson Canyon in the western Atlantic (Fig. 1).Liver tissue or fin clips were taken from the fish andstored in absolute ethanol until sample processing.Total genomic DNA was extracted using either Che-lex-100 (Walsh et al., 1991) or high-salt (Aljanabi &Martinez, 1997) standard protocols. PCR was per-formed using standard reagents and thermal cyclingconditionsfor12polymorphic,unlinkedmicrosatelliteloci ( Cpa 101,  Cpa 111,  Cpa 112,  Cpa 114,  Cha 1020, Cha 1202,  Cpa 4,  Cpa 6,  Cha 1017,  Cha 1059,  Cpa 107and  Cpa 113) described by McPherson et al. (2001a),Miller et al. (2001) and Olsen et al. (2002): these loci were selected from a larger group of microsatellites asapanel thatwere unlinked anddidnothave nullalleles(P. C. Watts & D. O’Leary, unpublished). Microsat-ellite alleles were separated on either a LiCor4200 oran ABI3100, with fragment sizes quantified againstcustom-made size ladders or by comparing mobilitieswith a  GENESCAN LIZ -500 size-standard (Applied Bio-systems) and scored using  GENEMAPPER  (AppliedBiosystems) software.Statistical analysesBasic measures of diversity (numbers of alleles,  N  a ;gene diversity,  H  e ) were calculated by  FSTAT  v.2.9.3(Goudet, 1995). Every locus was tested for departurefrom expected Hardy–Weinberg equilibrium (HWE)conditions for each sample using the randomisationtests (5,000 permutations) implemented by  FSTAT .Individual locus values of   F  ST  over all samples werecalculated using Weir & Cockerham’s (1984) unbi-ased estimator of   F  ST  using  FSTAT , with standarderrors for the single-locus estimates generated by jackknifing over populations and 95% confidenceintervals about the multi-locus estimates made bybootstrapping over loci.We performed Lewontin-Krakauer tests for twovalues of   k   (= 2 and 7.6) by calculating the theoret-ically expected variance of   F  ST  ( r 2 F  ) using theequation described in the  Introduction . After simu-lating several allele frequency distributions,Lewontin & Krakaeur (1973) suggested that  k   B  2for loci that are governed by neutral processes. Toaccount for some of the criticisms of this test, Baer(1999) employed a more stringent threshold value of  k   =  7.6, which is based on a skewed allele frequencydistribution (see Baer, 1999 for further details). Thesignificances of any differences between  r 2 F   and  s 2 F  were tested using a standard variance-ratio test.Outlying values of genetic differentiation at spe-cific loci were detected using the procedure described -5-1052545658 S01S02S03S04S05S06S10AS10B Fig. 1  Sample locations of spawning herring west of GreatBritain and around IrelandHydrobiologia (2008) 606:27–33 29  1 3  by Beaumont & Nichols (1996). Briefly,  FDIST 2(  * mab/software.html)runs coalescent simulations to generate an expecteddistribution of   F  ST  with heterozygosity (  H  e ), that isbased on the average genetic differentiation amongsamples (over all samples and loci) for the test dataset. The distribution of   F  ST  as a function of   H  e  isthen described, with 0.025, 0.5 and 0.975 quantiles(i.e. 95% confidence limits and median) againstwhich the locus-specific values were plotted todetect anomalous loci. Results As typical for this species, most  C. harengus microsatellite loci investigated here were highlypolymorphic. Average gene diversity (  H  e ) over allsamples varied from 0.36 up to 0.92 and the meannumber of alleles (  N  a ) ranged between seven and 30,with only one locus ( Cpa 107) having an average of less than 10 alleles (Table 1). Just four out of the 156tests for departure from expected HWE conditionswere significant ( P \ 0.05, Bonferroni correction,Rice, 1989, applied for  k   =  13 tests per sample) (datanot shown). While this is indicative of non-randommating within samples, it should be noted that onlyfour loci ( Cpa 112,  Cha 1017,  Cha 1020,  Cha 1059)demonstrated significant ( P \ 0.05,  k   =  13) devia-tions from expected HWE conditions over allsamples.Average genetic differentiation ( F  ST  =  0.007) inthe study area was weak but significantly differentfrom zero (95% confidence intervals: 0.003, 0.012).Individual locus estimates of   F  ST  varied consider-ably, from zero ( Cha 1202) up to  * 0.028 ( Cpa 107, Cpa 112). Rather than displaying a random distribu-tion of values of   F  ST  among loci there is a markeddichotomy to the locus-specific pattern of differenti-ation: (1) most ( n  =  9) markers are characterised byvery weak differentiation over all samples ( F  ST \ 0.01), while (2) three loci ( Cpa 6,  Cpa 107, Cpa 112) possess mean values of   F  ST  greater than orequal to 0.02 (Table 1). Given this heterogeneityamong loci it is not surprising, therefore, that theLewontin-Krakaeur tests are significant for  k   =  2( r 2 F  = s 2 F   =  79.62, d.f.  =  12,  P \\ 0.001), althoughthe test is not significant at the more stringent valueof   k   =  7.6 ( r 2 F  = s 2 F   =  20.95, d.f.  =  12,  P * 0.06).When values of   F  ST  were plotted against hetero-zygosity (Fig. 2) we note that  F  ST  is depressed withhigher values of this statistic, which is an expectedconsequence of high polymorphism (see Hedrick,1999). With respect to the loci themselves, twofeatures are evident: (1) most loci fall within the 95%envelope of neutral expectations of   F  ST  conditionalon heterozygosity and (2) only two loci are identifiedas outliers,  Cpa 112 that lies above the upper 95%confidence interval and  Cpa 114 that is less differen-tiated than expected if it were a neutrally evolvinglocus. Discussion In this paper we uncovered significant heterogeneityto the overall level of population differentiationwithin a panel of supposedly neutral microsatelliteloci. While we do not characterise the populationstructure of Atlantic herring in detail—these data willbe published in more detail elsewhere—it is evidentthat this species is weakly, but significantly, Table 1  Locus-specific and average estimates of genetic dif-ferentiation and genetic variability at 12 microsatellite locifrom samples of Atlantic herring,  C. harengus .  F  ST , Wright’s(1951) estimator of genetic differentiationLocus  F  ST  SE  H  e  N  a Cpa 101 0.006 0.003 0.913 29 Cpa 111 0.007 0.005 0.414 10 Cpa 112 0.027 0.016 0.880 30 Cpa 114 0.000 0.001 0.910 23 Cha 1020 0.004 0.003 0.916 23 Cha 1202 0.001 0.004 0.752 13Cpa4 0.003 0.002 0.873 24Cpa6 0.020 0.014 0.400 10 Cha 1017 0.003 0.003 0.826 14 Cha 1059 0.004 0.004 0.715 15 Cpa 107 0.028 0.009 0.363 7 Cpa 113 0.006 0.003 0.925 20 F  ST  + 95% CI  - 95% CIAll loci 0.008 0.013 0.0049 loci 0.004 0.005 0.0023 loci 0.025 - -SE, standard error of   F  ST ;  H  e , expected heterozygosity;  N  a ,number of alleles; 95% CI, 95% confidence intervals30 Hydrobiologia (2008) 606:27–33  1 3  differentiated at microsatellite loci throughout a largeportion of its distribution. This is concordant withprevious research (Shaw et al., 1999; Mariani et al.,2005; Ruzzante et al., 2005). However, the results presented here highlight an important, but oftenneglected, finding that all loci are not necessarilyindependent replicates of the same evolutionaryprocesses. Thus, the level of spatial genetic structureuncovered can depend on the choice of geneticmarkers used.Much of the statistical framework to assess thelevel of genetic differentiation among loci is pre-sented as an appraisal of selective processes operatingat one or few loci. However, interpreting significantinter-locus variation as evidence for selection itself using the Lewontin-Krakauer test is problematicbecause this method is sensitive to other demographicprocesses, such as isolation by distance, geneticstructure and variation in temporal separation. Theseprocesses produce correlated values of   F  ST  amongpopulations (Nei & Maruyama, 1975; Robertson,1975). The potential confounding effect of suchprocesses on our data is not known at present,although, for example, evidence for significant iso-lation by distance in European  C. harengus populations is equivocal (Mariani et al., 2005).Perhaps more relevant, Baer (1999) raised the pointthat even without the action of selection, certain otherlocus-specific processes, including differences inmutation rate and the presence of null alleles, cangenerate significant inter-locus heterogeneity.Accordingly loci used in the present study wereselected from a larger panel of microsatellitesbecause, after preliminary data analysis, they didnot appear to have null alleles, deviate from expectedHWE conditions, etc. Nonetheless, substantial differ-ences in mutation rates do seem plausible given theheterogeneity in variability among loci (Table 1),although the level of polymorphism per se is notdirectly associated with the value of   F  ST  (Table 1,Fig. 2). From a pragmatic standpoint, it may not be asimportant to identify the specific processes (i.e.selection versus mutation) that are operating, asmuch as actually uncovering the presence of signif-icant inter-locus heterogeneity in the first place. Thelatter observation affords a more informed perspec-tive on whether the panel of loci can be interpreted interms of neutral evolutionary processes. As such, theLewontin-Krakauer test presents a convenient, pre-liminary indicator of the presence of locus-specificeffects among a panel of loci, though this should beevident from the pattern of variation in single-locusestimates of   F  ST  (Table 1), and significant resultscan be further explored using other computationallymore intensive methods, such as that proposed byBeaumont & Nichols (1996).In contrast with previous studies that failed to findheterogeneity among microsatellite loci (e.g. de Leonet al., 1997), two ‘neutral’ microsatellite loci felloutside (one above and one below) the 95% limits of an expected neutral distribution of   F  ST  againstheterozygosity (Fig. 2). One interpretation is thatthese loci are, or are at least linked to regions that are,subject to directional and balancing selection respec-tively. Addressing the major criticisms laid at theLewontin-Krakaeur test, Beaumont & Nichols (1996)tested the effects of a variety of factors that mayconfound the distribution of   F  ST  for neutral markers.They noted that while some of these effects canbroaden the distribution of   F  ST  among loci, the modelis surprisingly robust and under most circumstancesthe model should be valid. However, asymmetricalcolonisation can have a particularly substantial effect,so analyses using this test alone should be interpretedwith caution. Moreover, since Beaumont & Nichols(1996) recommend that more than 20 loci be used fortheir method, additional loci should be incorporatedinto this analysis to either confirm or refute therobustness of this result. Despite these concerns, thereis substantial variation in values of   F  ST  among loci - 0.2 0.4 0.6 0.8 1.0 Cpa 112Heterozygosity       F       S      T Cpa 114 Fig. 2  Values of average  F  ST  for 12 microsatellite loci inAtlantic herring  C. harengus  against heterozygosity. Linesindicate estimated quantiles (solid  =  95% confidence inter-vals, dashed  =  median) for the expected distribution of   F  ST (= 0.008) with heterozygosityHydrobiologia (2008) 606:27–33 31  1 3
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

We need your sign to support Project to invent "SMART AND CONTROLLABLE REFLECTIVE BALLOONS" to cover the Sun and Save Our Earth.

More details...

Sign Now!

We are very appreciated for your Prompt Action!