Upload
others
View
0
Download
0
Embed Size (px)
Citation preview
351
q 2001 The Society for the Study of Evolution. All rights reserved.
Evolution, 55(2), 2001, pp. 351–379
THE EVOLUTIONARY HISTORY OF BROWN TROUT (SALMO TRUTTA L.) INFERREDFROM PHYLOGEOGRAPHIC, NESTED CLADE, AND MISMATCH ANALYSES OF
MITOCHONDRIAL DNA VARIATION
LOUIS BERNATCHEZDepartement de Biologie, GIROQ, Pavillon Alexandre-Vachon, Universite Laval, Sainte-Foy, Quebec, G1K 7P4, Canada
E-mail: [email protected]
Abstract. Phylogeographic, nested clade, and mismatch analyses of mitochondrial DNA (mtDNA) variation wereused to infer the temporal dynamics of distributional and demographic history of brown trout (Salmo trutta). Bothnew and previously published data were analyzed for 1794 trout from 174 populations. This combined analysis improvedour knowledge of the complex evolutionary history of brown trout throughout its native Eurasian and North Africanrange of distribution in many ways. It confirmed the existence of five major evolutionary lineages that evolved ingeographic isolation during the Pleistocene and have remained largely allopatric since then. These should be recognizedas the basic evolutionarily significant units within brown trout. Finer phylogeographic structuring was also resolvedwithin major lineages. Contrasting temporal juxtaposition of different evolutionary factors and timing of major de-mographic expansions were observed among lineages. These unique evolutionary histories have been shaped both bythe differential latitudinal impact of glaciations on habitat loss and potential for dispersal, as well as climatic impactsand landscape heterogeneity that translated in a longitudinal pattern of genetic diversity and population structuringat more southern latitudes. This study also provided evidence for the role of biological factors in addition to that ofphysical isolation in limiting introgressive hybridization among major trout lineages.
Key words. Europe, fish, mismatch, mitochondrial DNA, nested clade, phylogeography, Salmo.
Received March 7, 2000. Accepted September 18, 2000.
Elucidating the evolutionary history of extant species isan important objective of any research program that seeks tounderstand population divergence and, ultimately, speciation.This history is also directly relevant to conservation biologybecause historical contingencies have been largely respon-sible for creating the most important genetic subdivisions inmany, if not most extant taxa (e.g., Zink 1996; Avise et al.1998; Hewitt 2000). The phylogeographic approach has beenused to test the congruence between distributional historiesagainst paleo-environmental settings and determining thechronology of evolutionary diversification (Avise 1998; Mo-ritz and Bermingham 1998). Comparative phylogeographyhas also emerged as a powerful method to address broaderecological and evolutionary issues.
In northern temperate freshwater fishes, comparative phy-logeography revealed predictive trends in phylogeographicstructure, genetic diversity, and speciation rates among spe-cies inhabiting formerly glaciated and unglaciated regions ofNorth America (Bernatchez and Wilson 1998). Further gen-eralizations of the effect of Pleistocene glaciations on fresh-water fish fauna could be gained by comparing phylogeo-graphic data from other regions. Contrasts between Eurasianand North American freshwater fish phylogeographic struc-ture (e.g., Bernatchez et al. 1989) would be particularly in-formative, given the much less extensive glacial advances inEurasia and its contrasting landscape with that of NorthAmerica (Hewitt 2000). However, the phylogeographic struc-ture of the Eurasian fish fauna is still largely unknown (Ber-natchez and Dodson 1994; Durand et al. 1999a; Nesbo et al.1999; Englbrecht et al. 2000).
The brown trout (Salmo trutta L.) is the most widely dis-tributed freshwater fish native to the Palearctic region. Itsnatural range extends from northern Norway and northeastern
Russia, southward to the Atlas Mountains of North Africa.From west to east, its range spans from Iceland to the head-waters of Aral Sea affluents in Afghanistan. Salmo trutta alsoexhibits considerable morphological diversity and life-his-tory variation, including specializations for anadromous, flu-viatile, and lacustrine modes of life. Large-scale patterns ofgenetic diversity in brown trout have been studied extensivelyover the last two decades, using both allozymes (e.g., Ryman1983; Ferguson 1989; Guyomard 1989; Osinov 1990; Garcıa-Marın and Pla 1996; Largiader and Scholl 1996) and mito-chondrial (mtDNA) analyses (Bernatchez et al. 1992; Giuffraet al. 1994; Hynes et al. 1996; Osinov and Bernatchez 1996;Apostolidis et al. 1997; Hansen and Mensberg 1998; Weisset al. 2000). Yet, the analysis of brown trout phylogeographicstructure is still lacking over important geographic areas,such as North Africa and eastern Europe. Patterns of post-glacial recolonizations, finer phylogeographic structure with-in major trout lineages, and demographic history have notbeen rigorously assessed or remain controversial (e.g., Ham-ilton et al. 1989; Osinov and Bernatchez 1996; Garcıa-Marınet al. 1999). This ambiguity may be attributed to both limitedanalytical resolution and statistical treatments.
Previous phylogeographic analyses of brown trout haverelied upon the use of haplotype trees and geographic dis-tribution to make biological inference by visual inspectionof how geography overlays haplotype relationships. This maynot make full use of all historical information contained ingene genealogies. Namely, this does not allow the estimationof the dynamic structure and temporal juxtaposition of dif-ferent evolutionary factors that are most likely compatiblewith the patterns of genetic diversity observed in extant pop-ulations. The recent development of statistical nested cladeanalysis of phylogeographic data offers a potentially useful
352 LOUIS BERNATCHEZ
TABLE 1. Number of populations, sample sizes, and haplotype diversity (h) within major trout evolutionary lineages. For populations, thefirst number indicates the number of pure populations for a particular lineage and the number in parentheses indicates the number of admixedpopulations but predominated in relative abundance by the particular lineage.
LineageNumber ofpopulations
Numberof fish
Number of haplotypes
Sequence RFLP CombinedHaplotype diversity
(SE)
Atlantic (AT)Danubian (DA)Adriatic (AD)Mediterranean (ME)marmoratus (MA)Total
55 (1)48 (1)37 (5)12 (1)14 (0)
166 (8)
993191298104205
1791
81410
33
38
142811
22
57
163517
34
75
0.582 (0.011)0.931 (0.005)0.851 (0.009)0.523 (0.011)0.511 (0.007)0.679 (0.008)
framework in that respect (Templeton et al. 1995; Templeton1998). Yet, the use of this method has been limited thus far(Templeton et al. 1995; Durand et al. 1999b; Nesbo et al.1999; Turner et al. 2000). The traditional phylogeographicapproach is also limited in inferring the dynamics of de-mographic history. The analysis of mismatch distributionprovides a way to estimate the magnitude and age of pop-ulation growth by statistically comparing the distribution ofintrapopulation molecular diversity with that expected underhypotheses of equilibrium (Rogers and Harpening 1992; Rog-ers 1995; Schneider and Excoffier 1999). However, the useof mismatch distribution has almost entirely been limited tohuman studies (but see Lavery et al. 1996; Merila et al. 1997;Petit et al. 1999).
In this study, I performed a genetic diversity analysis ofmtDNA and combined phylogeographic, nested clade, andmismatch analyses to infer the temporal dynamics of distri-butional and demographic history of the brown trout through-out its native geographic range. New results and previouslypublished data were combined to document the overall geo-graphic distribution of major trout evolutionary lineages. Iinfer the temporal juxtaposition of factors most compatiblewith the differential patterns of genetic diversity observedamong these lineages and discuss these factors against paleo-environmental settings encountered by brown trout in dif-ferent parts of its range.
MATERIALS AND METHODS
Samples
This study combines new analyses and previously pub-lished results (Appendix). A total of 277 fish representing 92sampling sites had never been analyzed. Samples previouslyanalyzed for sequence variation of the mtDNA control region(D-loop) (Bernatchez et al. 1992) were reanalyzed using poly-merase chain reaction–restriction fragment length polymor-phism (PCR-RFLP) to increase the number of character statesand allow standardization with previous studies. We also in-cluded the results of PCR-RFLP and sequence analyses per-formed by Giuffra et al. (1994), Bernatchez and Osinov(1995), Osinov and Bernatchez (1996), and Apostolidis et al.(1996a, 1997). The results obtained by Hynes et al. (1996)were standardized by performing the same sequence andPCR-RFLP analyses used for the other fish on representativesof all mtDNA haplotypes identified by these authors usingRFLP analyses over the entire mtDNA genome. Sixteen fish
from the Mediterranean basin that possessed the mtDNA lin-eage of Atlantic origin were discarded because they mostlikely represented the results of stocking, thus potentiallyblurring historical signals (Guyomard 1989; Bernatchez etal. 1992). Given these exclusions and that several populationsand samples overlapped among some of these studies (Ap-pendix), this analysis was based on the genetic informationobtained for 1794 trout representing 174 populations (Table1).
Laboratory Analyses
MtDNA variation was analyzed using both sequencing andRFLP performed on PCR-amplified products. Sequencingwas done on the same 310-bp segment (now 313 due to newindels) located at the 59 end of the control region studiedpreviously in brown trout (Bernatchez et al. 1992; Apostolidiset al. 1997). The primers LN20 (59-ACC ACT AGC ACCCAA AGC TA-39) and HN20 (59-GTG TTA TGC TTT AGTTAA GC-39), which amplify the whole control region, wereused for PCR, whereas primer H2 (59-CGT TGG TCG GTTCTT AC-39) was used for sequencing (Bernatchez and Danz-mann 1993). Technical procedures of mtDNA purification,amplification, and sequencing are detailed in Bernatchez etal. (1992). RFLP analysis was performed using six restrictionenzymes (HinfI, HpaII, MboI, NciI, RsaI, and TaqI) on twoadjacent PCR-amplified segments; the complete ND-5/6 re-gion (;2.4 kb), and a second segment (;2.1 kb) comprisingthe whole cytochrome oxydase b gene and the D-loop. Prim-ers, amplifications, restriction digests, and electrophoresisprocedures were as described in Bernatchez and Osinov(1995). Distinct single endonuclease patterns were identifiedby a specific letter in order of appearance and used in com-bination with sequence variation to define sequence/RFLPcomposite genotypes.
Genetic Diversity Analyses
The combined information of three previous studies basedon a small number of individuals but higher genetic resolutionthan used here resolved five major mtDNA phylogenetic lin-eages in brown trout (Fig. 1). Based on the origin of the firsthaplotypes identified, these were named Atlantic (AT), Dan-ubian (DA), Mediterranean (ME), marmoratus (MA), andAdriatic (AD) lineages. Each of these lineages were differ-entiated by eight to 12 apomorphies and were supported bybootstrap values of .99%. Instead of performing an overall
353EVOLUTIONARY HISTORY OF BROWN TROUT
FIG. 1. Consensus tree based on maximum-parsimony analysis relating five major trout mtDNA evolutionary lineages. Relationshipsamong major lineages and their bootstrap values (as a percentage) are derived from the combined information of three studies (Bernatchezet al. 1992; Giuffra et al. 1994; Bernatchez and Osinov 1995) based on sequence analysis of 1250 nucleotides and PCR-RFLP analysesof an estimated 576 additional sites resolved with 19 restriction enzymes. The rooting position of Atlantic salmon (Salmo salar) is derivedfrom the sequence analysis of Giuffra et al. (1994). The numbers in parentheses refer to the number of apomorphies unique to eachlineage. See Bernatchez and Osinov (1995) for further details. Relationships among haplotypes within each lineage was based on thesequence analysis of a 313-bp segment located at the 59 end of the control region and PCR-RFLP analysis performed using six restrictionenzymes (HinfI, HpaII, MboI, NciI, RsaI, and TaqI) on two adjacent PCR-amplified segments: the complete ND-5/6 region (approximately2.4 kb) and a second (approximately 2.1 kb), comprising the whole cytochrome oxydase b gene and the D-loop. Composite haplotypesare defined in Tables 2 and 3. Branch lengths were scaled using number of mutations, the shortest corresponding to a single mutation.
354 LOUIS BERNATCHEZ
phylogenetic analysis in the present study, we first assignedhaplotypes (with confidence .99%) to these different groupsbased on their combined composition at 20 synapomorphicpositions (13 resolved by RFLP analyses, seven by sequenc-ing) that can unambiguously diagnose haplotypes belongingto any of the five trout lineages (detailed in Bernatchez andOsinov 1995). The phylogenetic relationships among hap-lotypes within each lineage was first assessed by parsimonyanalysis, as detailed in Bernatchez and Osinov (1995). Amore detailed analysis was then achieved by the nested cladeanalysis (see below).
Levels of genetic diversity within and among the five majortrout lineages were compared using the haplotype diversityand the maximum-likelihood estimation of the average num-ber of nucleotide substitutions per site within and amonggroups (Nei 1987) using REAP, version 4 (McElroy et al.1992). A hierarchical analysis of molecular variance (AMO-VA; Excoffier et al. 1992) was performed using Arlequin,version 2.0 (Schneider et al. 1999) to compare the componentof genetic diversity imputable to the variance among the fivetrout lineages to that observed among populations within eachof them. The significance of the variance components asso-ciated with the different levels of genetic structure were testedusing 1000 permutations.
Mismatch Analysis
A mismatch analysis was performed using Arlequin tocompare the demographic history of the five major trout lin-eages. Because the main objective was to compare the his-torical demography of the five major trout lineages and pop-ulation structure has a limited effect on the mismatch dis-tribution (Rogers 1995), pure samples were pooled withineach lineage. Following the method of Schneider and Ex-coffier (1999), we quantified the moment estimators of timeto the expansion t, the mutation parameter before (u0 5 2mN0)and following (u1 5 2mN1) expansion, expressed in units ofmutational time, where N0 and N1 are the female effectivepopulation sizes before and following an expansion that oc-curred t generations ago. These estimators were used to plotthe expected distribution of probabilities of observing S dif-ferences between two randomly chosen mtDNA haplotypes(Watterson 1975). We also computed the raggedness indexof Harpening (1994). The comparison of the sum of squaredeviation (SSD) between the observed and estimated mis-match distribution was used as a test statistic for the estimatedstepwise expansion models (Schneider and Excoffier 1999).Confidence intervals for those parameters were obtained byMonte Carlo simulations of 1000 random samples using thecoalescent algorithm of Hudson (1990) as modified bySchneider and Excoffier (1999). We finally used the rela-tionships t 5 mt (where m is the mutation rate and t is theexpansion time in generations) to compare the timing of pos-sible demographic expansion among trout lineages.
Nested Clade Analysis
The probability of a parsimonious relationship among hap-lotypes within each lineage was assessed as described in Tem-pleton et al. (1992), using the program ProbPars kindly pro-vided by A. R. Templeton (Washington University, St. Louis,
MO). MtDNA haplotypes differing by up to four mutationshad a probability .95% of being connected in a parsimoniousmanner. Because the number of mutations among adjacenthaplotypes never exceeded four mutational steps, we thenconstructed a minimum spanning network for each trout lin-eage using Minspnet (Excoffier and Smouse 1994). The re-sulting networks were then converted into a nested designfollowing the rules of Templeton et al. (1987) and Templetonand Sing (1993). The clade distance, Dc (which measures thegeographical range of a particular clade), the nested cladedistance, Dn (which measures how a particular clade is geo-graphically distributed relative to other clades in the samehigher-level nesting category), and contrasts of these mea-sures between tip and interior clades (Itc [y], Itn [y]) werequantified by first calculating the geographical centers of eachclade resolved at all hierarchical levels (Templeton et al.1995). Secondly, the geographical distances (in kilometer)between individuals and clade centers were calculated asgreat circle distances. The various distances were recalculatedafter each of 1000 random permutations of clades and/orhaplotypes against sampling locality to statiscally test at a5 0.05 for significantly large and small distances and interior-tip contrasts for each clade with respect to the null hypothesisof no geographic association within each of the nested clades.The model of population structure and historical events thatwas most suitable with the pattern of genetic structure ob-served within each clade showing statistically geographicalassociations was identified using the inference key providedas an appendix to Templeton (1998). To reduce bias in pa-rameters estimation and permutations procedures due to largediscrepancies among sample sizes (Appendix), those werelimited to a maximum of 10 by respecting the observed hap-lotype frequencies in the total sample (but with a minimumabsolute haplotype frequency of n 5 1). All calculations wereperformed using Excell (Microsoft, vers. 97) spreadsheetsdeveloped by J. E. Stacy (University of Oslo, Norway), asused in Nesbo et al. (1999).
RESULTS
Relationships among Major Trout Lineages and OverallPatterns of Genetic Diversity
Figure 1 illustrates the branching topology among the fivemajor trout lineages resolved from the combined informationof previous studies. The salient feature of this tree was therooting position of the outgroup, Atlantic salmon (Salmo sa-lar), which suggested the ancestral divergence of the Atlanticlineage from all others (Giuffra et al. 1994). The combinedsequence and RFLP analyses resolved 75 composite haplo-types (Tables 2, 3, Figure 2). These could unambiguously beassigned to either one of the five major trout lineages, basedon their composition at 13 RFLP and seven sequence syn-apomorphic positions, as detailed in Bernatchez and Osinov(1995).
The overall level of haplotype diversity was high (h 50.856), but the extent of nucleotide diversity (p 5 0.0133)was comparable to that observed in other northern temperatefreshwater fishes of similar latitudinal distribution (Bernatch-ez and Wilson 1998). The level of genetic diversity, however,was highly variable among the five trout lineages (Tables 1,
355EVOLUTIONARY HISTORY OF BROWN TROUT
TA
BL
E2.
Var
iabl
esi
tepo
siti
ons
inth
eco
ntro
lre
gion
amon
g38
Salm
otr
utta
mtD
NA
geno
type
sde
fine
dby
sequ
ence
anal
ysis
.D
ots
refe
rto
nucl
eoti
deid
enti
tyw
ith
AT-
s1se
quen
ceho
mol
ogy
and
dash
esin
dica
tein
dels
.Num
bers
refe
rto
nucl
eoti
depo
siti
ons
inF
igur
e2.
Ast
eris
kin
dica
tes
sequ
ence
sre
port
edin
Apo
stol
idis
etal
.(19
97)
wit
hth
eor
igin
alno
men
clat
ure
give
nin
pare
nthe
ses.
Gen
otyp
e
Var
iabl
esi
tes
713
1416
2237
3941
5969
8695
122
126
128
150
157
162
194
199
212
242
250
251
252
259
278
279
280
310
Atl
anti
c(A
T)
AT-
s1A
T-s2
AT-
s3A
T-s4
T · · ·
G · · ·
— · · ·
T · · ·
A · · ·
— · · ·
A · · ·
T · · ·
C · · ·
A · · ·
T · C ·
T · · ·
A · · ·
T · · ·
— · · ·
T · · ·
— · · ·
G · · ·
T · · ·
G · · ·
A · · ·
C · · ·
G A · ·
A G · ·
T · · ·
T · · ·
G · · ·
C · · ·
T · · ·
T · · CA
T-s5
AT-
s6A
T-s7
AT-
s8
· · · ·
· · · ·
· · · ·
· · · ·
· · · ·
· · · ·
· · · ·
· · · ·
· · · ·
· · · ·
· · · ·
· · · C
· · · ·
C C · ·
· · · ·
· · · ·
· · · ·
A · A ·
· · · ·
· · · ·
· · · ·
· · · ·
· · · A
· · · G
· · · ·
· · · ·
· · · ·
· · · ·
· · · ·
· · · ·D
anub
ian
(DA
)D
A-s
1D
A-s
2D
A-s
3D
A-s
4
· · · ·
· · · A
· · · ·
C C C C
· · · ·
· · · ·
· · · ·
A A A A
· · · ·
· · · ·
· · · ·
· · · ·
· · · ·
· · · ·
· · · ·
· · · ·
· · · ·
· · · ·
· · · ·
· · · ·
· · · ·
· · · ·
· · A A
· G · G
G G G G
· · · ·
· · · ·
· · · ·
· · · ·
· · · ·D
A-s
5D
A-s
6D
A-s
7D
A-s
8D
A-s
9
· · · · ·
· · · · ·
· · · · ·
C C C C C
· · · · ·
· · · · ·
· · G · ·
A G A A A
· · · · ·
· · · · ·
· · · · ·
· · · · ·
G · · · ·
· · · · ·
· · · · ·
· · · — ·
· · · · ·
· · · · ·
· · · · ·
· · · · ·
· · · · ·
· · · · ·
A · · · A
G · · · G
G G G G G
· · · · ·
· · · · ·
· · · · ·
· · · · ·
· · · · ·D
A-s
10D
A-s
11D
A-s
12D
A-s
13D
A-s
14*
(J)
· · · · ·
· · · · ·
· · · G ·
C C C C C
· · · · ·
· · · A ·
· · · · ·
A A A A A
· · · · ·
· · · · ·
· · · · ·
· · · · ·
· · · · ·
· · · · ·
· · · · ·
· · · · ·
A · · · ·
· · · · ·
· · · · ·
· · · · A
· · · · ·
· T T T ·
· · · · ·
· · G G ·
G G G G G
· · · · ·
· · · · ·
· · · · ·
· · · · ·
· · · · ·A
dria
tic
(AD
)A
D-s
1A
D-s
2A
D-s
3A
D-s
4A
D-s
5
· · · · ·
· · · · ·
· · · · ·
· · · · ·
· · · · ·
· · · · ·
· · · · ·
C C C C C
· · T · ·
· · · · ·
· · · · ·
T · · · ·
· · · · ·
· · · · ·
· · · · ·
· · · · ·
· · · · ·
· · · · ·
· · · · C
· · · · ·
· · · · ·
· · · · ·
· · · · ·
· · · · ·
· · · · ·
· C · · ·
C C C C C
· · · · ·
· · · C ·
· · · · ·A
D-s
6A
D-s
7*
(C)
AD
-s8
*(D
)A
D-s
9*
(E)
AD
-s10
*(F
)
· · · · ·
· · · · ·
· · · · ·
· · · · ·
G · · · ·
· · · · ·
· · · · ·
C C C — C
· · · · ·
· · · · G
· · · · ·
· C · · ·
· · · · ·
· · · · ·
· · · · ·
· · · · ·
· · · · ·
· · A · A
· · · · ·
· · · · ·
· · · · ·
· · · · ·
· · · · ·
· · · · ·
· · · · ·
· · · · ·
C C C C C
· · · · ·
· · · · ·
· · · · ·M
edit
erra
nean
(ME
)M
E-s
1M
E-s
2M
E-s
3*
(I)
· · C
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
C T C
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
A A A
· · ·
· · ·
C C C
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·m
arm
orat
us(M
A)
MA
-s1
MA
-s2
MA
-s3
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
C C C
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
A A A
· · ·
· · ·
· · ·
· · C
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
A A A
· T T
· · ·
· · ·
356 LOUIS BERNATCHEZ
TABLE 3. Definitions of 75 composite sequence/RFLP (sn/rn) mtDNA haplotypes in Salmo trutta: sn refers to sequence definition in Table 1,and rn refers to the composite fragment patterns resolved for restriction enzymes HinfI, HpaII, MboI, NciI, RsaI, and TaqI.
Haplotype RFLP Haplotype RFLP Haplotype RFLP
Atlantic1
AT-s1/r1AT-s1/r2AT-s1/r3AT-s6/r4
AAAAAABAAAAFBHAEAFQAAAAA
DanubianDA-s1/r1DA-s2/r2DA-s3/r3DA-s4/r4
CBABBCCBCCBCGAAAHCCEAAHC
MediterraneanME-s1/r1ME-s2/r2ME-s3/r1
DAADCD
AT-s7/r5AT-s1/r6AT-s1/r7AT-s1/r8AT-s1/r9
KAAAAABAAGAFBAAAAAAAAAJABAAAJF
DA-s5/r5DA-s6/r6DA-s9/r7Da-s10/r8Da-s7/r9
CEAFBCHEAGHCJEAAHCCEAAGCCEAABC
MarmoratusMA-s1/r1MA-s2/r1MA-s3/r1MA-s1/r2
EACADDEACADDEACADDEKCADD
AT-s1/r10AT-s1/r11AT-s5/r12AT-s1/r13AT-s8/r14
OAAAAFLAAAAFAAAALABHAAAFBIABAF
DA-s1/r10DA-s1/r11DA-s1/r12DA-s9/r13DA-s1/r143
JEAGHCIAAAHCJAAAICCAAAHCNA---C
AT-s3/r3AT-s4/r2
AdriaticAD-s1/r1AD-s3/r3
CCBACECDBECE
DA-s1/r15DA-s1/r16DA-s1/r15DA-s1/r16DA-s2/r17DA-s1/r18
CEAGHCMEAFBCCEAGHCMEAFBCFBABBCCEAABD
AD-s5/r4AD-s1/r5AD-s1/r6AD-s3/r7AD-s4/r8AD-s1/r92
CDBEEECDBACEPCBAMECDBECGCCBAMECC---E
DA-s12/r19DA-s2/r20DA-s2/r21DA-s12/r22DA-s13/r23DA-s1/r24
CBACBDCBACBCCBABKCCBABBDCADABDCAAABC
AD-s1/r102
AD-s1/r112
AD-s1/r3AD-s2/r1AD-s6/r1AD-s7/r9
CC---ECC---E
DA-s2/r25DA-s1/r26DA-s2/r27DA-s1/r28DA-s7/r5DA-s8/r3
CFACFCHJAGHCCBABOCCEAAPE
AD-s8/r1AD-s9/r1AD-s10/r11
DA-s14/r1DA-s1/r4Da-s1/r3DA-s1/r6DA-s11/r1
1 Correspondance to haplotype definition of Hynes et al. (1996a) : AT-s1/r1 (IX, X), AT-s2/r2 (I, III, VI, VIII, XIII, XIV), AT-s1/r6 (V), AT-s1/r10(IV), AT-s1/r11 (XI), AT-s1/r14 (XX), AT-s4/r2 (II), AT-s3/r3 (XII).
2 RFLP patterns described in Apostolidis et al. (1996a): r9, r10, and r11 corresponds to Type 3, 2, 4 of these authors, respectively. Data for MboI,NciI, RsaI were missing, but these haplotypes had new mutations at other enzymes (either AluI, AvaII, HaeIII) not seen previously.
3 r14 corresponds to Type 9 of Apostolidis et al. (1996a).
4). The DA lineage was the most genetically diverse, fol-lowed by the AD and AT lineages. Both ME and MA lineagesexhibited extremely reduced levels of diversity. Assumingthat the rooting position of S. salar indicates the ancientdivergence between two ancestral lineages (AT vs. one fromwhich diverged all others), it is noteworthy that the geneticdiversity within the AT lineage one was highly reduced com-pared to the other. The net nucleotide divergence among thefive trout lineages varied between 1.21% and 2.19% (Table4).
Geographic Distribution of Major Trout Lineages
Contrasting patterns of geographic distribution were ob-served among the five major trout lineages (Fig. 3A–E), asonly eight populations out of 174 (4.6%) with more than onelineage were observed (Appendix). All populations from theAtlantic basin and Morocco were fixed for the AT lineage.The DA lineage was almost exclusively associated with
drainages of the Black, Caspian, and Aral Sea basins, as wellas the Persian Gulf. The other three lineages (AD, ME, MA)showed little overlap in distribution with the other two andexhibited a differential pattern in their overall geographicdistribution within the Mediterranean. The MA lineage wasalmost strictly associated with the Adriatic basin. The MElineage was predominantly found in tributaries draining inthe western basin, whereas the AD lineage predominated inthe eastern part of the Mediterranean basin (Appendix).
Hierarchical Genetic Diversity Analysis
Most of the mtDNA molecular variance observed in nativetrout populations was imputable to differences among line-ages (Table 5). The amount of genetic variance imputable tothe among-populations within-lineage component was ap-proximately one order of magnitude less than the among-lineage components (8.942 vs. 0.873). Yet, populations werehighly structured within lineages, with an overall FST of
357EVOLUTIONARY HISTORY OF BROWN TROUT
FIG. 2. Sequence of the 59-end, 313-bp segment of the mtDNA control region; type AT-s1 for Salmo trutta. The sequence shown isfor the light strand, and includes 10 nucleotides of the proline tRNA gene. Asterisks and numbers above sequence indicate the 30variable positions among the 38 resolved sequences. The AT-s1 sequence is entered into GeneBank under accession number U18198.
TABLE 4. Average number of nucleotide substitutions per site within(main diagonal) and net nucleotide divergence (below main diagonal)among the five major evolutionary lineages of brown trout.
Lineage AT DA AD ME MA
Atlantic (AT)Danubian (DA)Adriatic (AD)Mediterranean (ME)marmoratus (MA)
0.00200.01620.01850.01750.0188
0.00450.02040.02030.0219
0.00180.01210.0142
0.00050.0146 0.0005
0.776. The amount of molecular variance among populationswas highest for the DA and AD lineages, two to three timeslower for the AT lineage, and extremely reduced for both theME and the MA lineages. This also translated into variableFST values depending on lineages.
Mismatch Distribution and Demographic History
The overall mismatch distribution was clearly bimodal, onemode corresponding to the number of differences among ma-jor lineages (;20) and the other to differences among in-dividuals within lineages (;2; Fig. 4). The mismatch distri-
bution, however, differed substantially among trout lineages,with the mean number of differences being the lowest in ME(0.53) and MA (0.54) lineages, followed by AT (1.3), AD(2.1), and the highest value observed in DA (4.7) lineage.The mismatch distribution within AT, AD, and DA lineagesfitted well the predicted distribution under a model of suddenexpansion, but this was refuted for both ME and MA lineages(Table 6). Considering the 95% confidence interval generatedby the simulations, the observed number of polymorphic sitesvalues were consistent with simulated sites, meaning that theparameters estimated by the model were sufficiently accurateto account for the observed polymorphism in AT, AD, andDA lineages. The observed values of the age expansion pa-rameter (t) differed substantially among these three lineages.Confidence intervals on u1 values were too large to alloweven rough estimates of the magnitude of expansions.
Nested Clade Analysis
Contrasting patterns of nested clade design, both in num-bers of haplotypes and levels of clades was observed amongthe five trout evolutionary lineages (Figs. 5–7). The null hy-pothesis of no association between the position of haplotypes
358 LOUIS BERNATCHEZ
→
FIG. 3. Geographic distribution of the five major trout mtDNA evolutionary lineages: Atlantic (AT), Danubian (DA), marmoratus (MA),Mediterranean (ME), and Adriatic (AD). Symbols were positioned with the exact latitudinal and longitudinal coordinates of each sampleusing the software MapInfo, and symbols of geographically proximate samples may overlap. A detailed description of lineage distributionamong all populations surveyed in provided in the Appendix.
in a cladogram with geographic position was rejected for atleast several clades at any nesting level for AT, DA, or DAlineages (Tables 7–9). Thus, we identified the likely causesfor the geographic association using the inference key ofTempleton (1998).
A temporal juxtaposition of past fragmentation at all cladelevels best explained the overall genetic structure observedwithin the AD lineage (Table 7). This pattern was mainlydriven by more easterly distributed populations. One-stepclades were highly structured geographically in the easternpart of the distribution range (from southern Italy and east-ward), whereas a more homogeneous pattern was observedamong western populations, where only two clades, largelyoverlapping in distribution, were represented (Fig. 5B).
The geographic pattern of genetic diversity observed with-in the AT lineage was best explained by the temporal jux-taposition of past fragmentation and range expansion (Table8). This was mainly driven by the strong geographic struc-turing observed among more southern populations (Fig. 6C).However, clade AT1–4 (corresponding to single haplotypeATs1/r1) was widely distributed among northern populations.Although clades AT1–1 and AT1–2 broadly overlapped indistribution among northern populations, a significantly smallDc values was observed for both (Table 8). Namely, therewas a tendency for clade AT1–1 to be mainly found in west-central populations, being almost absent from more northernpopulations (Iceland, northern Scandinavia, northern Russia).A pattern of either contiguous range expansion or dispersalimplying gene flow with isolation by distance best explainedthe geographic pattern observed among single haplotypeswithin AT1–1 and AT1–2 clades. In contrast, a signal of pastfragmentation best explained the geographic distribution ofhaplotypes within clade AT1–5 found in southern France andnorthern Spain.
The geographic pattern of genetic diversity observed with-in the DA lineage was best explained by a complex patternof temporal juxtaposition of various historical events (Table9). Past fragmentation best explained the geographic distri-bution of three-step clades. A strong geographic discontinuitywas particularly clear in more easterly distributed popula-tions, where those from the Caspian and Aral Sea basins weretypically characterized by clade DA3–3 and those surround-ing the Black Sea area by clade DA3–2 (Fig. 7C). CladeDA3–1 generally had a more western distribution, where itoverlapped extensively with DA3–2. At the two-step level,different historical events best explained the clade distribu-tion, depending on geographic areas. The distribution ofclades within DA3–1 (DA2–1 and DA2–2), which were foundin the most western part of the range, was best explained bya history of either range expansion and/or restricted dispersalwith gene flow (Table 9). In contrast, the distribution ofclades within either DA3–2 or DA3–3 clades better fit a his-
tory of fragmentation. Within DA3–2, clade DA2–3 mainly(but not exclusively) characterized populations from theBlack Sea area, whereas DA2–4 was confined to the upperreaches of the Danube drainage (Austria and Slovenia, Fig.7B). Within DA3–3, populations from the Caspian Sea basinwere characterized by clade DA2–8, whereas DA2–5, DA2–6, and DA2–7 were largely confined to the Aral Sea basin.A geographic east-west dichotomy between a history of frag-mentation and dispersal was also apparent at the one-step andzero-step clade levels (Table 9).
Nested clade analysis was not warranted for either ME orMA lineages, which were characterized by the predominanceof single haplotypes with a relatively broad geographic dis-tribution within both lineages (Appendix).
DISCUSSION
The combined use of traditional phylogeography, nestedclade, and mismatch analyses of mtDNA diversity revealedcomplex patterns in the distribution and demography ofbrown trout. The following discussion treats the species’ evo-lutionary history in hierarchical order to infer a temporaljuxtaposition of different paleo-environmental settingsthroughout its distribution range that is most compatible withthe differential patterns of genetic diversity observed amongand within major evolutionary lineages.
Choice of a Molecular Clock
Clearly, there are many ambiguities in the application ofa molecular clock; consequently, this clock was used in con-junction with estimated phylogenetic and demographic pat-terns only to provide an approximate time frame to evaluatephylogeographic hypotheses. Estimates in the range of 1–2%sequence divergence per million years were chosen for thefollowing reasons. A previous indirect calibration derivedfrom fossil records led to an average estimate of mtDNAmutation rate of 1% per million years for salmonids (Smith1992). Correlations reported between phylogeographic pat-terns and successions of Pleistocene glaciation events havepreviously been reported in other salmonids when using suchclock calibration (e.g., Bernatchez and Dodson 1991; Wilsonet al. 1996). In a more formal attempt to calibrate a fishmtDNA molecular clock using physically isolated geminatesby the Isthmus of Panama, Donaldson and Wilson (1999)estimated a divergence rate of 1% per million years in theND 5/6 region and 3.6% per million years for the controlregion, suggesting an average mutation rate for the wholemitochondrial genome intermediate of these values (the con-trol region represented approximately 40% of the subsamplednucleotides in our study). Finally, the oldest fossils reportedfor S. trutta date from the early Pleistocene (2 million yearsago; discussed in Osivov and Bernatchez 1996). Consequent-
359EVOLUTIONARY HISTORY OF BROWN TROUT
FIG. 3. Continued.
360 LOUIS BERNATCHEZ
FIG. 3. Continued.
361EVOLUTIONARY HISTORY OF BROWN TROUT
FIG. 3. Continued.
TABLE 5. Analysis of molecular variance in brown trout using thefive major evolutionary lineages as the top level of grouping. Eightpopulations admixed for different lineages, and populations with sam-ple size ,5 were omitted for this analysis. The molecular varianceand related F-statistics imputable to the among-populations compo-nent is provided first for the overall system and then within eachlineage. All F-statistics values are significant at a 5 0.0001.
Source ofvariation df
Variancecomponents
%variation f-statistics
Among phylogeneticgroups
Among populationswithin groups
ATDAADMEMA
4
1254333271012
8.942
0.8730.6761.87311.2030.2370.251
88.83
8.6762.4278.4893.7375.0790.42
0.8883
0.77660.62420.78480.93730.75070.9042
ly, the extent of divergence among the deepest clades inbrown trout should fit within this time frame, which is thecase if one considers the extent of divergence among the fivemajor trout lineages (1.4–2.0%) and a molecular clock of 1–2% per million years.
Pleistocene Origins of Five Major Evolutionary Lineages inBrown Trout
Extending analyses to its whole distributional range con-firmed that the S. trutta complex is composed of five majorevolutionary lineages and basically ruled out the possibilityof additional subdivisions of similarly deep divergence.Theselineages most likely identify ancestral populations of troutthat evolved independently as a result of ancient allopatricfragmentation. Results of previous studies of nuclear genevariation are also congruent with these lineages. The samepopulations characterized here by different major mtDNAlineages or with the same geographic distribution were alsohighly differentiated either by diagnostic alleles or strongallele frequency differences (Guyomard 1989; Bernatchezand Osinov 1995; Giuffra et al. 1996; Apostolidis et al.1996b). A broad-scale geographic survey of microsatelliteloci variation recovered four of the five major mtDNA lin-eages (Presa 1995). Finally, Berrebi (1995) showed that Cor-sican trout populations originated from two distinct lineages,most likely corresponding to the ME and AD lineages.
Given their overall extent of divergence, the time frameinvolved in the lineages’ separation would span between 0.5million and 2 million years. This indicates that major phy-logeographic subdivisions in brown trout are associated withthe climatic and environmental changes that occurred duringPleistocene glaciations events. The most ancient separationwould have involved allopatric fragmentation between thethree major drainage subdivisions: the Atlantic (AT lineage),
362 LOUIS BERNATCHEZ
FIG. 4. Frequency distributions of pairwise number of mutationaldifferences between trout individuals observed in all trout mtDNAevolutionary lineages combined (top panel) and separately in thefive major trout lineages defined by mitochondrial analysis: Atlantic(AT), Adriatic (AD), Danubian (DA), Mediterranean (ME), andmarmoratus (MA). Diamonds represent the observed data, the boldcurve is the model fitted to the data, and dashed lines delineate the2.5 and 97.5 percentile values of 1000 simulated samples.
TABLE 6. Observed (S) and 95% confidence intervals (CI) of simulated number of polymorphic sites, demographic parameters (95% CI),among major trout lineages. t, u0, and u1 are the age of expansion and population size before and following expansion, expressed in units ofmutational time. P(SSDobs) is the probability of observing by chance a less good fit between the observed and the mismatch distribution for ademographic history of the lineage defined by the estimated t, u0, and u1 parameters. P(Ragobs) is the probability of observing by chance ahigher value of the raggedness index than the observed one, under the hypothesis of population expansion.
Lineage S 95% CI t u0 u1 P(SSDobs) Raggedness P(Ragobs)
Atlantic (AT)Adriatic (AD)Danubian (DA)Mediterranean (ME)marmoratus (MA)
191937
23
4–2118–5134–9722–4251–80
0.5 (0.0–5.6)2.4 (0.5–5.7)5.5 (2.0–8.2)0.7 (0.3–1.0)0.7 (0.4–0.89)
1.36 (0–2.6)0.001 (0–1.9)0.004 (0–3.5)0 (0–0.7)0 (0–0.5)
3.8 (0.5–2681)9.7 (1.5–6792)
25.2 (10.4–6491)807 (1–3645)555 (1.6–3444)
0.3930.7690.3950.000010.00001
0.1200.0220.0150.2350.233
0.5900.8370.4380.0020.00001
Ponto-Caspian (DA), and Mediterranean followed by sub-sequent and possibly simultaneous fragmentation within theMediterranean basin, which led to the divergence of the ME,MA, and AD lineages. The most drastic climatic changesover the last 3 million years occurred approximately 700,000years ago (Webb and Bartlein 1992; Andersen and Borns1994); it is also likely that the first large isolation of severallarge river systems now draining into different basins (e.g.,Rhone, Rhine, and Danube) occurred during that period (Vil-linger 1986). Consequently, it is plausible that the most im-portant genetic subdivisions within the brown trout complexare associated with major climatic changes and basin isola-
tion that occurred in Europe between the early to the uppermid-Pleistocene.
Considering the paleo-environmental settings during thePleistocene and the overall geographic pattern of distributionand genetic diversity within each of them, it is possible toinfer hypothetical centers of origins for the five major troutevolutionary lineages. Given its clear association with theAtlantic basin, the center of origin of AT lineage is obviouslyassociated with drainages of this system. Trout from the At-lantic basin could survive in refuges marginal to ice sheetsduring the most recent glaciation events (Ferguson and Flem-ing 1983; Hamilton et al. 1989; Osinov and Bernatchez 1996).Given the much more severe climate, however, and the moreimportant glacial expansion both on land and on the oceanduring earlier glacial cycles of the Pleistocene (Webb andBartlein 1992; Andersen and Borns 1994), it is more likelythat the ancestral center of origin of the AT lineage wasfurther south, perhaps in coastal tributaries of the IberianPeninsula or even North Africa. This is corroborated by themore pronounced pattern of clade diversity and divergenceamong southern populations (see also Weiss et al. 2000).
Locating the center of origin for the DA lineage is muchmore problematic, given the less severe direct impacts ofhabitat loss during glacial advances in the Ponto-Caspianbasin and the very complex pattern of expansion, contraction,and interconnection of the Black, Caspian, and Aral Sea ba-sins (Arkhipov et al. 1995). Nevertheless, the localization ofearly trout fossils in the Caucasus area and the central po-sition of clade 3–2, both geographically and in the DA clad-ogram (intermediate between 3–1 and 3–3), suggest that theancestral populations from which all extant populations ofthe DA lineage originated inhabited from drainages associ-ated with the Black Sea.
The differential pattern of geographic distribution of thethree other lineages (ME, MA, and AD) broadly corroboratetraditionally recognized Mediterranean refugial areas: asouthwest (Ibero-Mediteranean), central (Adriato-Mediter-ranean or Italian), and an eastern (Balkans/Anatolia) refuge(Keith 1998). The ME lineage was predominantly associatedwith tributaries draining in the western basin of the Medi-terranean Sea, suggesting that it originated from this region.Persat and Berrebi (1990) proposed that a restricted area ofsouthern France (Rousillon region) that was isolated by thePyrenees to the south and by severe environmental conditionsprevailing during glacial advances to the east may haveserved as a Pleistocene refuge for other fishes. The MA lin-eage, typical of the phenotypically and ecologically distinct
363EVOLUTIONARY HISTORY OF BROWN TROUT
TABLE 7. Nested geographical analysis of the Adriatic (AD) lineage following the inference key of Templeton (1998). Nested design, haplotypeand clade designations are given in Figure 5A. Geographic distribution of clades is provided in Figures 5B and 5C. Following the name ofhaloptypes or clade number are the clade (Dc) and nested clade (Dn) distances. For those clades containing both interior (bold) and tip nestedclades, the average difference between interior versus tip clades for both distance measures is given in the row labelled I-T. A superscript Smeans that the distance measure was significantly small at the 5% level, and a superscript L means that the distance measure was significantlylarge at a 5 0.05. At the bottom of a nested set of clades in which one or more sets of the distance measures was significant is a line indicatingthe biological inference using the same number and letter codes as those provided in Templeton (1998). Abbreviation: past frag.; past frag-mentation.
Haplotypes
No Dc Dn
One-step clades
No Dc Dn
Two-step clades
No Dc Dn
ADs1r1ADs6r1ADs2r1ADs8r1ADs9r1
I-T
6200S
00S
0S
620L
6751996L
705761916
2418S
1-2-3-5-15 No: past frag. 1-1 757L 790L
ADs1r9ADs7r9ADs1r10
I-T
0S
0S
0S
0
17S
176172
2158S
1-2-3-4-9 No: past frag. 1-2 17S 598ADs1r6ADs4r8ADs1r11ADs10r11
0S
14900
0000
1-31-41-5
I-T
0S
149S
0S
594L
524235S
419S
312L
1-2-3-4-9 No: past frag. 2-1 665S 670ADs1r3ADs1r5ADs3r3ADs3r7
I-T
5630
50S
0546L
1152L
1200445S
150616
1-2-3-5-15 No: past frag. 1-6 676 702 2-2 742 746Ads5r4 0 0 1-7 0S 1575L I-T 277.6 276.8
I-T 676L 2873S 1-2-3-4-9 No: past frag.1-2-3-5-15 No: past frag.
marble trout, was mainly confined to the Po River basin, butincluded drainages from Croatia and Slovenia, which couldall interconnect during phases of maximal interglacials (Bian-co 1990). Previous allozyme studies also support the hy-pothesis of a North Adriatic origin for the marble trout (Giuf-fra et al. 1996). Finally, the species’ predominance in trib-utaries of the eastern Mediterranean basin relative to bothME and MA lineages, along with the higher level of cladediversity observed in populations from the Balkans, indicatedthat the AD lineage most likely originated from the Balkans/Anatolia refuge.
Physical Versus Ecological Constraints to Dispersal andSecondary Intergradation
Given S. trutta’s dispersal potential, the variety of habitatsit occupies, and the evidence for interconnections among ma-jor basins during the Pleistocene, it is remarkable that majorevolutionary lineages remained largely allopatric in distri-bution for hundreds of thousands of generations. For instance,interconnections between the Atlantic and Ponto-Caspian ba-sins have been possible during different glacial phases dueto the development of either intermittent fluvial connectionsor large ice-dam lakes at the ice margins that dischargedsouthward into the Ponto-Caspian basin (Grosswald 1980;
Gibbard 1988; Arkhipov et al. 1995). Recent phylogeograph-ic (Durand et al. 1999a; Nesbo et al. 1999a) and traditionalbiogeographic studies (Banarescu 1990) confirmed that otherfishes originating from Ponto-Caspian refuges have indeedrecolonized the Atlantic basin following the most recent gla-ciations. Similarly, fish dispersal between the Ponto-Caspianand the Mediterranean basins was also possible via connec-tions between the Danube and rivers draining to the Medi-terranean, increased water discharge from the north thatdrained into the Black Sea and overflowed to the Mediter-ranean, or sea level lowering and reduced salinity that al-lowed fish movements across the Agean Sea basin (Bianco1990; Economidis and Banarescu 1991; Arkhipov et al.1995). The identification of a few trout populations charac-terized by mtDNA lineages not representative of the basinin which they were found confirms that trout could also usesuch connections. Yet, introgressive hybridization (excludingthat due to stocking or contemporary human-induced habitatdisturbance) remained relatively limited, as exemplified instudies at nuclear genes (Bernatchez and Osinov 1995; Giuf-fra et al. 1996).
This indicates that, in addition to physical isolation, bio-logical factors have contributed to limiting dispersal and in-trogressive hybridization among major trout lineages. One of
364 LOUIS BERNATCHEZ
FIG. 5. (A) Minimum spanning network for 17 mitochondrial composite haplotypes resolved in the trout Adriatic lineage. Each line inthe network represents a single mutational change. A zero indicates an interior node in the network that was not present in the sample.Composite haplotypes are defined in Tables 2 and 3. Dotted-lined polygons indicate one-step clades nested together into two-step clades,and lined-shaded polygons indicate haplotypes grouped together into one-step clades. (B) Geographic distribution of two-step AD clades.
365EVOLUTIONARY HISTORY OF BROWN TROUT
FIG. 5. Continued.
←
(C) Geographic distribution of one-step AD clades. Other lineages refer to haplotypes not belonging to the Adriatic lineage. Symbolswere positioned with the exact latitudinal and longitudinal coordinates of each sample using the software MapInfo, and geographicallyproximate samples may overlap. A detailed description of single haplotype distribution is provided in the Appendix.
these factors could be the differential dynamics of demo-graphic expansion between pioneer dispersers (presumablythose issued from the closest refuge) and later migrants innewly colonized habitats. Simulation studies showed that pi-oneers would be able to establish and expand rapidly in newlyavailable habitats (Ibrahim et al. 1996). Later migrants wouldcontribute little because they would be entering populationsat or near carrying capacity with only limited replacementdynamics. According to this scenario, trout already presentin a refuge located in a given basin or geographic area wouldbe the first to expand and occupy available ecological niches,leaving only little probability of establishment for popula-tions that survived in a more remote refugial area. This pro-cess would be enhanced if time of geographic isolation and/or differential selection in different environments (in termsof climate, physical environments, and species communities)have been sufficiently important for the accumulation of ge-netic and ecological differences. For salmonids, it has beenshown in lake whitefish (Coregonus clupeafomis) that geo-
graphic isolation of evolutionary lineages during the last gla-ciations led to the development of their partial genetic in-compatibility, resulting in a much higher embryonic mortalityrate of hybrid compared to pure progeny (Lu and Bernatchez1998). This, along with the potential for occupying distincttrophic niches, apparently explains the maintenance of re-productive isolation between populations of different evo-lutionary lineages when found in sympatry (Bernatchez et al.1999; Lu and Bernatchez 1999). Ecological and/or geneticconstraints to gene flow among trout lineages have been sug-gested by studies in which different lineages were found inparapatry (Giuffra et al. 1996; Largiader and Scholl 1996;Poteaux et al. 1998).
Mismatch Analysis: Differential Timing of MajorDemographic Expansions among Lineages
A model of sudden demographic expansion was statisti-cally supported for the AT, AD, and DA lineages. Because
366 LOUIS BERNATCHEZ
367EVOLUTIONARY HISTORY OF BROWN TROUT
FIG. 6. Continued.
←
FIG. 6. (A) Minimum spanning network for 16 mitochondrial composite haplotypes resolved in the trout Atlantic lineage. Each line inthe network represents a single mutational change. A zero indicates an interior node in the network that was not present in the sample.Composite haplotypes are defined in Tables 2 and 3. Heavier-lined polygons indicate two-step clades nested together into three-stepclades, dotted-lined polygons indicate one-step clades nested together into two-step clades, and lined-shaded polygons indicate haplotypesgrouped together into one-step clades. (B) Geographic distribution of two-step AT clades. (C) Geographic distribution of one-step ATclades. Other lineages refer to haplotypes not belonging to the Atlantic lineage. A detailed description of single haplotype distributionis provided in the Appendix.
the effects of less important demographic changes may bemasked by that of the most important one (Rogers 1995;Lavery et al. 1996), the variable estimates of the age expan-sion parameter should be interpreted as indicating that themost important demographic expansion of each lineage oc-curred at different evolutionary times. The most recent de-mographic expansion was detected within the AT lineage.Because the Atlantic basin was the most directly affected byglaciations in terms of habitat loss, one would predict moreimportant reduction of population abundance during glacialadvances in this area, as is generally reported in north tem-perate fishes differentially affected by glaciations (Bernatch-ez and Wilson 1998). This was also supported by the morereduced mtDNA diversity in AT relative to both the AD andDA lineages. Although they should be interpreted cautiouslygiven their large 95% CI, absolute estimates obtained by
considering the two different mutation rates (1% and 2% permillion years) suggested that the time of the most importantdemographic expansion of the AT lineage (13,400 years agoat 2% per million years, 26,800 years ago BP at 1% permillion years) roughly coincided with the onset of the lastglacial retreat (approximately 18,000 years ago).
In contrast, the most important demographic expansion thatoccurred in the DA lineage was much older (mean values 5154,500–309,000 years ago). This is congruent with geolog-ical evidence that the geographic range occupied by this lin-eage was much less directly affected by recent glacial ad-vances compared to the Atlantic basin. This is also reflectedby its much higher mtDNA genetic diversity. Several pos-sibilities of large demographic expansion occurred in thisregion, namely through interconnections that developedamong expanding Black, Caspian, and Aral Sea basins. How-
368 LOUIS BERNATCHEZ
369EVOLUTIONARY HISTORY OF BROWN TROUT
FIG. 7. Continued
←
FIG. 7. (A) Minimum spanning network for 35 mitochondrial composite haplotypes resolved in the trout Danubian evolutionary lineage.Each line in the network represents a single mutational change. A zero indicates an interior node in the network that was not present inthe sample. Composite haplotypes are defined in Tables 2 and 3. Heavier-lined polygons indicate two-step clades nested together intothree-step clades, dotted-lined polygons indicate one-step clades nested together into two-step clades, and thinned-lined polygons indicatehaplotypes grouped together into one-step clades. (B) Geographic distribution of three-step DA clades. (C) Geographic distribution oftwo-step DA clades. Other lineages refer to haplotypes not belonging to the Atlantic lineage. A detailed description of single haplotypedistribution is provided in the Appendix.
ever, the most important interconnections and sea expansionmost likely developed approximately 270,000–290,000 yearsago (Arkhipov et al. 1995). It is thus plausible that the majordemographic expansion within the DA lineage was associatedwith the opportunities of large-scale dispersal that occurredat that time.
The timing of demographic expansion within the AD lin-eage (67,300–134,600 years ago) was intermediate betweenthat of the AT and DA lineages. Trout populations fromdifferent parts of the Mediterranean basin were likely dif-ferentially affected by glacial advances, with habitats of thosefrom the western part of the range being more compresseddue to more severe climatic changes and the sea barrier, as
compared to more eastern populations (Hewitt 2000). Thisis reflected by the more important genetic diversity amongeastern than western populations. The most important de-mographic expansion within the AD lineage could thus beassociated with westward dispersal that occurred during theearly Wurm or late Riss glacial period.
A model of sudden demographic expansion was rejectedfor both the ME and MA lineages. This could reflect a tem-porally more stable demography and near mutation-drift equi-librium conditions for these lineages (Rogers and Harpening1992). In such a case, however, an overall high level ofgenetic diversity, as well as pronounced population structureamong physically isolated populations would be expected.
370 LOUIS BERNATCHEZ
TABLE 8. Nested clade analysis of the Atlantic (AT) lineage following the inference key of Templeton (1998). Nested design, haplotype andclade designations are given in Figure 6A. The geographic distribution of clades is provided in Figures 6B and 6C. See Table 7 for furtherdetails. Abbreviations: rest. gene flow; restricted gene flow, IBD; isolated by distance., cont. range ext.; contiguous range expansion.
Haplotypes
No Dc Dn
One-step clades
No Dc Dn
Two-step clades
No Dc Dn
Three-step clades
No Dc Dn
ATs1r2ATs1r10ATs4r2ATs1r6ATs1r7
I-T
1084S
00S
0S
01084L
1083942942942
1678243
1-2-3-4 No: restr. gene flow, IBD 1-1 1081S 1085S
ATs1r3ATs3r3ATs1r13
I-T
834S
0738465S
834S
86777811
1-2-11-12 No: cont. range exp. 1-2 832S 830S
ATs8r4 0 1-3I-T
0S
658L
529S
397L
1-2-3-5-15 No: past frag. 2-1 1008 1008 3-1 1008S 1082S
ATs1r1ATs1r8ATs1r9
I-T
10940
256S
2256
1094133S
106S
26.4
1-4 1094S 1091S
1-2-3-5-15 No: past frag. 1-5 109S 1258I-T 984L 21661-2-3-5-15 No: past frag. 2-2 1110S 1103S
ATs6r4ATs7r5ATs5r12
4400
000
1-61-71-8I-T
44S
0S
0S
44
59S
43S
29S
231-2-3-5-15 No: past frag. 2-3 47S 2017L
I-T 1063L 2913S
1-2-3-5-15 No: past frag. 3-2 1240 1231I-T 232 2491-2-3-5-15 No: past frag.
Clearly, this was not the case, as illustrated by the lineages’extremely reduced diversity and molecular variance imput-able to population structuring. Instead, the predominance ofsingle haplotypes found over a wide geographic area is morecompatible with nonequilibrium conditions caused by a se-vere and relatively recent bottleneck. This is also congruentwith the evidence for more reduced habitat availability inareas occupied by refugial populations of the ME and MAlineages. Extremely reduced diversity may in turn hamperthe detection of demographic signals in the genetic signatureof populations (e.g., Lavery et al. 1996). As such, geneticdiversity observed for the ME and MA lineages do not followa typical north-south pattern of glaciation effects on popu-lation demography and illustrate that conditions favorable totrout survival may have been more severe in those southernareas than further north during the last glacial advance.
Nested Clade Analysis: Differential Dynamics of PopulationStructuring among Lineages
The nested clade analysis provided new insights into theresolution of finer evolutionary lineages within major line-ages, from which new hypotheses of brown trout evolutionary
history can be inferred and contrasted with previous inter-pretations.
Atlantic lineage: latitudinal contrasts in historicalpopulation structuring
The more southern distribution of clade AT3–2, along withits more complex pattern of clade diversity suggest that itrepresents the ancestral lineage from which lineage AT3–1derived following isolation in more northern latitudes. Itslong history of fragmentation also reflects a more temporallystable population structuring relative to more northern pop-ulations, as would be predicted from the glacial history ofthe different regions. In contrast, the nested clade analysisrevealed that northern trout populations were historically sub-divided into two ancestral lineages that intermixed exten-sively since their recent range expansion throughout most ofthe north Atlantic region following the last glacier retreat.Given their differential pattern of geographic distribution, itis most likely that AT1–1 characterizes a trout lineage thatsurvived in a refuge located in the more northwestern partof the Atlantic range of distribution, whereas lineage AT1–2 identifies a lineage that survived in a northeastern refuge.
371EVOLUTIONARY HISTORY OF BROWN TROUT
In summary, three trout lineages, all from AT origins, wereinvolved in the recolonization process of the North Atlanticbasin; one from a southern refuge and more ancient origin(characterized by haplotype ATs1r1 belonging to lineageAT3–2) that first intergraded with an northern ancestral lin-eage (AT3–1 or AT2–1), then two lineages originating fromthis admixed group that later evolved in isolation in a west-central and a northeastern refuge.
The origin and postglacial history of brown trout in theNorth Atlantic basin has been the subject of divergent in-terpretations. Most researchers agreed on the fact that theNorth Atlantic was recolonized by different trout evolution-ary lineages. However, interpretations varied substantially asto their number, center of origin, and timing of dispersal.The broader geographic coverage in this study, along withthe ability of the nested-clade analysis to interpret the tem-poral juxtaposition of historical processes, partly reconcilesprevious interpretations. First, these results support the ex-istence of a northwestern refuge, as first proposed by Fer-guson and Fleming (1983), then reiterated by Hamilton et al.(1989), Osinov and Bernatchez (1996), and Garcıa-Marın etal. (1999). Second, they also support the existence of a north-eastern refuge first proposed by Osinov and Bernatchez(1996) and provide evidence for the contribution of a south-ern refuge, as originally proposed by Hamilton et al. (1989)and later by Garcıa-Marın et al. (1999). Unlike, these inter-pretations, however, the present results implied that northerncolonization by this southern group occurred prior to the lastglaciation. They also refute the view of a contribution of aPonto-Caspian lineage (Garcıa-Marın et al. 1999) becausethis would imply the complete disappearance of DA haplo-types in northern Europe, along with all known allozymealleles private to the DA lineage (see also Weiss et al. 2000).Finally, as previously hypothesized, the present results sta-tistically confirmed that these evolutionary lineages inter-graded extensively following the last glacier retreat.
Mediterranean lineages: longitudinal contrasts in historicalpopulation structuring
The overall geographic pattern of genetic diversity ob-served among Mediterranean lineages revealed a longitudinalpattern of reduced population structuring from east to west.The predominance of single haplotypes with a relativelybroad geographic distribution within both ME and MA lin-eages is most compatible with nonequilibrium conditions re-sulting from recent range expansion, which eventually re-sulted in partial geographic overlap between these and theAD lineage. An east-west pattern of reduced population struc-turing was also illustrated within the AD lineage. Thus, thepattern of population structuring among western populationsof the AD lineage was more similar to that observed withinME and MA lineages than among eastern populations of thissame lineage. This indicated that, in contrast to highly frag-mented structure among eastern populations, the observedpopulation structuring among western populations was morecompatible with a history of recent range expansion. It isnoteworthy that the Balkans, despite its relatively small size,is the area harboring the most diverse phenotypic diversityamong trout populations, which may be the outcome of their
long-term isolation in different environments compared toother parts of the species range of distribution (Kottelat1997).
Danubian lineage: complex juxtapositions of varioushistorical events
The geographic area occupied by the DA lineage has beenthe least investigated previously. The present study providednew insights into the existence of finer phylogeographic sub-divisions within this lineage and allowed us to propose hy-potheses regarding its evolutionary history. Previous allo-zyme studies did not show any pattern of major geographicdiscontinuity in the distribution of genetic diversity (Ber-natchez and Osinov 1995). The mtDNA anaylsis of Bernatch-ez and Osinov (1995) provided slight, but inconclusive ev-idence for genetic discontinuity among sea basins. The nestedclade analysis, however, statistically supported such a struc-ture. These results support the long-standing hypothesis pro-posed on the basis of morphological variation that popula-tions of each sea basin should be recognized as distinct evo-lutionary lineages (Berg 1948).
Conclusions
The combined use of traditional phylogeographic, nested-clade, and mismatch analyses of mtDNA diversity proved tobe very efficient in improving our knowledge of the complexevolutionary history of brown trout throughout its nativerange of distribution. This confirmed the existence of fiveevolutionary lineages that evolved in geographic isolationduring the Pleistocene and have remained largely allopatricsince then. These should be recognized as the basic evolu-tionary significant units within brown trout (Bernatchez1995). In addition to physical isolation, biological factorsmust have contributed to limiting their dispersal and intro-gressive hybridization among them. The unique evolutionaryhistories of each lineage have been shaped both by the thedifferential latitudinal impact of glaciations on habitat lossand potential for dispersal, as well as climatic impacts andlandscape heterogeneity that translated in a longitudinal pat-tern of genetic diversity and population structuring at moresouthern latitudes.
ACKNOWLEDGMENTS
I am indebted to numerous individuals that made this studypossible. I am particularly grateful to J. Schoffman for pro-viding unique tissue samples from throughout the distributionrange of brown trout. Thank you also to A. Ferguson, P.Berrebi, and R. Guyomard for sharing samples; J.E. Stacyand C.L. Nesbo for providing the Excell sheets to performthe nested clade analysis; A.R. Templeton for the programProbPars; T. Gosselin for producing the illustrations; and N.Tessier, G. Lu, S. Martin, and M. Parent for their technicalassistance. This paper was improved by the constructing com-ments of J. Turgeon and two anonymous reviewers. The re-search program of LB on the ecological and genetic deter-minism of northern fish evolution is supported by NSERC(Canada) research grants.
372 LOUIS BERNATCHEZ
TA
BL
E9.
Nes
ted
clad
ean
alys
isof
the
Dan
ubia
n(D
A)
line
age
foll
owin
gth
ein
fere
nce
key
ofTe
mpl
eton
(199
8).
Nes
ted
desi
gn,h
aplo
type
and
clad
ede
sign
atio
nsar
egi
ven
inF
igur
e.7A
.T
hege
ogra
phic
dist
ribu
tion
ofcl
ades
ispr
ovid
edin
Fig
ures
7Ban
d7C
.A
bbre
viat
ions
:in
conc
l.;
inco
nclu
sive
,re
st.
disp
-gen
efl
.;re
stri
cted
disp
ersa
lw
ith
gene
flow
.
Hap
loty
pes
No
Dc
Dn
One
-ste
pcl
ades
No
Dc
Dn
Tw
o-st
epcl
ades
No
Dc
Dn
Thr
ee-s
tep
clad
es
No
Dc
Dn
DA
s1r1
DA
s14r
1I-
T
448S
044
8S
527S
389S
138
1-2-
11-1
2N
o:co
nt.
rang
eex
p.1-
148
10
2-1
480S
506S
DA
s2r2
0D
As1
2r19
DA
s12r
22I-
T
0 016
9S
216
9S
015
4S
166S
212
1-2
0S29
9S
1-2-
3-4
No:
rest
.ge
nefl
ow/I
BD
1-3
164S
490S
DA
s2r1
7D
As2
r21
DA
s2r1
7I-
T
0 0 862
43
230S
783
892
260
71-
2-11
-17
No:
inco
ncl.
1.4
635S
1054
DA
s2r2
DA
s2r2
513
1 00 0
1-5
1-6
I-T
131S
02
232
763
741
246
31-
2-3-
5-6-
7-8
Yes
:re
st.
gene
flow
/dis
p.2-
262
0S62
5S
I-T
213
92
118
1-2-
3-5-
6to
ofe
wcl
ades
:ra
nge
exp/
rest
r.di
sp-g
ene
fl.
3-1
583S
1326
DA
s7r9
DA
s7r5
I-T
900S
0S
900
939
1215
227
51-
2-3-
5-15
No:
past
frag
.1-
798
0S10
15S
DA
s1r1
6D
As5
r50
506
0 01-
81-
9I-
T
232S
506S
611L
684S
727S
310
1-2-
3-5-
6-7-
8Y
es:
res.
gene
flow
/dis
p.2-
387
4S99
6S
DA
s1r1
8D
As1
1r18
I-T
25S
381S
235
6
171S
359S
218
81-
2-3-
5-15
No:
past
frag
.1-
1017
1S15
0S
DA
s1r2
4D
As1
3r23
0 00 0
1-11
1-12
I-T
0 017
1S
141S
237S
238
1-2-
3-5-
6-7-
8Y
es:
rest
.ge
nelo
g/di
sp.
2-4
279S
824S
I-T
595L
172
1-2-
3-5-
15N
o:pa
stfr
ag.
3-2
940S
1047
S
DA
s1r4
DA
s1r2
8I-
T
141S
014
1S
137S
013
7S
1-2-
3-6-
7-8
No:
inad
.S
ampl
ing
1-13
132S
187S
DA
s1r1
4D
As1
0r8
0 00 0
1-14
1-15
I-T
0S 013
2S
238S
276S
270
373EVOLUTIONARY HISTORY OF BROWN TROUT
TA
BL
E9.
Con
tinu
ed.
Hap
loty
pes
No
Dc
Dn
One
-ste
pcl
ades
No
Dc
Dn
Tw
o-st
epcl
ades
No
Dc
Dn
Thr
ee-s
tep
clad
es
No
Dc
Dn
1-2-
3-5-
6-7-
8Y
es:
rest
.ge
nefl
ow/d
isp
2-5
203S
1165
DA
s4r4
DA
s9r7
DA
s9r1
3D
As1
r3D
As3
r3D
As8
r3D
As1
r11
DA
s1r1
2
0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0
1-16
1-17
1-18
1-19
I-T
0 0 0S 0S 0
0 0
698S
1351
265
3
2-6
0S12
94
1-2-
3-5-
15N
o:pa
stfr
ag.
2-7
915
1067
DA
s1r1
5D
As1
r10
I-T
812S
0S
812L
913S
624S
289
1-2-
3-5-
15N
o:pa
stfr
ag.
1-20
812S
774S
DA
s1r6
DA
s6r6
DA
s1r2
6I-
T
0S 067
8S
233
9S
330S
330
607S
213
91-
2-3-
5-15
No:
past
frag
.1-
2192
7S12
77I-
T2
114
250
2S
1-2-
3-5-
15N
o:pa
stfr
ag.
2-8
947S
1266
I-T
241
7S2
441-
2-3-
5-15
No:
past
frag
.3-
313
2118
91L
I-T
212
214
31-
2-3-
5-15
No:
past
frag
.
374 LOUIS BERNATCHEZ
LITERATURE CITED
Andersen, B. G., and H. W. Borns Jr. 1994. The Ice Age world.Scandinavian Univ. Press, Oslo, Norway.
Apostolidis, A., Y. Karakousis, and C. Triantaphyllidis. 1996a. Ge-netic differentiation and phylogenetic relationships amongGreek Salmo trutta L. (brown trout) populations as revealed byRFLP analysis of PCR amplified mitochondrial DNA segments.Heredity 77:608–618.
———. 1996b. Genetic divergence and phylogenetic relationshipsamong Salmo trutta L. (brown trout) populations from Greeceand other Europeans countries. Heredity 76:551–560.
Apostolidis, A. P., C. Triantaphyllidis, A. Kouvatsi, and P. S. Econ-omidis. 1997. Mitochondrial DNA sequence variation and phy-logeography among Salmo trutta L. (Greek brown trout) popu-lations. Mol. Ecol. 6:531–542.
Arkhipov, S. A., J. Ehlers, R. G. Johnson, and H. E. Wright Jr.1995. Glacial drainages towards the Mediterranean during mid-dle and late Pleistocene. Boreas 24:196–206.
Avise, J. C. 1998. The history and purview of phylogeography:concepts and applications. Mol. Ecol. 7:371–380.
Avise, J. C., D. Walker, and G. C. Johns. 1998. Speciation durationsand Pleistocene effects on vertebrate phylogeography. Proc.Roy. Soc. Lond. B. 265:1707–1712.
Banarescu, P. 1990. Zoogeography of freshwater: general distri-bution and dispersal of freshwater animals. Aula Verlag ed.Wiesbaden 1:1–511.
Berg, L. S. 1948. Freshwater fishes of the Soviet Union and adjacentcountries. Vol. 1. Akademy Nauk, Moscow.
Bernatchez, L. 1995. A role for molecular systematics in definingevolutionarily significant units in fishes. Pp. 114–132 in J. L.Nielsen, ed. Evolution and aquatic ecosystem: defining units inpopulation conservation. American Fisheries Society Sympo-sium no.17, Bethesda, Maryland.
Bernatchez, L., and R. G. Danzmann. 1993. Congruence in control-region sequence and restriction-site variation in mitochondrialDNA of brook charr (Salvelinus fontinalis Mitchill). Mol. Biol.Evol. 10:1002–1014.
Bernatchez, L., and J. J. Dodson. 1991. Phylogeographic structurein mitochondrial DNA of the lake whitefish (Coregonus clupea-formis) and its relation to Pleistocene glaciations. Evolution 45:1016–1035.
———. 1994. Phylogenetic relationships among Palearctic and Ne-arctic whitefish (Coregonus sp.) populations as revealed by mi-tochondrial DNA variation. Can. J. Fish. Aquat. Sci. 51:240–251.
Bernatchez, L., and A. Osinov. 1995. Genetic diversity of trout(genus Salmo) from its most eastern native range based on mi-tochondrial DNA and nuclear gene variation. Mol. Ecol. 4:285–297.
Bernatchez, L., and C. C. Wilson. 1998. Comparative phylogeog-raphy of Nearctic and Palearctic freshwater fishes. Mol. Ecol.7:431–452.
Bernatchez, L., J. J. Dodson, and S. Boivin. 1989. Population bot-tlenecks: influence on mitochondrial DNA diversity and its effectin coregonine stock discrimination. J. Fish Biol. 35:233–244.
Bernatchez, L., R. Guyomard, and F. Bonhomme. 1992. DNA se-quence variation of the mitochondrial control region among geo-graphically and morphologically remote European brown troutSalmo trutta populations. Mol. Ecol. 1:161–173.
Bernatchez, L., A. Chouinard, and G. Lu. 1999. Integrating mo-lecular genetics and ecology in studies of adaptive radiation:whitefish, Coregonus, as a case study. Biol. J. Linn. Soc. 68:173–194
Berrebi, P. 1995. Etude genetique des truites de Corse. LaboratoireGenome et Populations, Universite de Montpellier pour ParcNaturel Regional de Corse.
Bianco, P. G. 1990. Potential role of the paleohistiory of the Med-iterranean and Paratethys basins on early dispersal of Euro-Med-iterranean freshwater fishes. Ichthyol. Explor. Fresh. 1:167–184.
Donaldson, K. A., and R. R. Wilson Jr. 1999. Amphi-Panamic gem-inates of snook (Percoidei: Centropomidae) provide a calibration
of the divergence rate in the mitochondrial DNA control regionof fishes. Mol. Phyl. Evol. 13:208–213.
Durand, J. J., H. Persat, and Y. Bouvet. 1999a. Phylogeographyand postglacial dispersion of the chub (Leuciscus cephalus) inEurope. Mol. Ecol. 8:989–998.
Durand, J. D., A. R. Templeton, B. Guinand, A. Imsiridou, and Y.Bouvet. 1999b. Nested clade and phylogeographic analyses ofchub, Leuciscus cephalus (Teleostei, Cyprinidae), in Greece: im-plications for the Balkan Peninsula biogeography. Mol. Phyl.Evol. 13:566–580.
Economidis, P. S., and P. M. Banarescu. 1991. The distribution andorigins of freshwater fishes in the Balkan Peninsula, especiallyin Greece. Intern. Rev. Der Gesam. Hydrobiol. 76:257–283.
Englbrecht, C. C., J. Freyhof, A. Nolte, K. Rassmann, U. Schliewen,and D. Tautz. 2000. Phylogeography of the bullhead Cottus gobio(Pisces: Teleostei: Cottidae) suggests a pre-Pleistocene originof the major central European populations. Mol. Ecol. 9:709–722.
Excoffier, L., and P. E. Smouse. 1994. Using allele frequency andgeographic subdivision to reconstruct gene trees within a spe-cies: molecular variance parsimony. Genetics 136:343–359.
Excoffier, L., P. E. Smouse, and J. M. Quattro. 1992. Analysis ofmolecular variance inferred from metric distances among DNAhaplotypes: application to human mitochondrial DNA restrictiondata. Genetics 131:479–491.
Ferguson, A. 1989. Genetic differences among brown trout, Salmotrutta, stocks and their importance for the conservation and man-agement of the species. Fresh. Biol. 21:35–46.
Ferguson, A., and C. C. Fleming. 1983. Evolutionary and taxonomicsignificance of protein variation in the brown trout (Salmo truttaL.). Pp. 85–99 in G. S. Oxford and D. Rollinson, eds. Proteinpolymorphisms: adaptive and taxonomic significance. AcademicPress, London.
Garcıa-Marın, J.-L., and C. Pla. 1996. Origins and relationships ofnative populations of Salmo trutta (brown trout) in Spain. He-redity 77:313–323.
Garcıa-Marın, J.-L., F. M. Utter, and C. Pla. 1999. Postglacial col-onization of brown trout in Europe based on distribution of al-lozyme variants. Heredity 82:46–56.
Gibbard, P. L. 1988. The history of the great northwestern Europeanrivers during the past three million years. Phil. Trans. Roy. Soc.Lond. 318:559–602.
Giuffra, E., L. Bernatchez, and R. Guyomard. 1994. Mitochondrialcontrol region and protein coding genes sequence variationamong phenotypic forms of brown trout Salmo trutta from north-ern Italy. Mol. Ecol. 3:1161–1171.
Giuffra, E., R. Guyomard, and G. Forneris. 1996. Phylogeneticrelationships and introgression patterns between incipient par-apatric species of Italian brown trout (Salmo trutta L. complex).Mol. Ecol. 5:207–220.
Grosswald, M. G. 1980. Late weichselian ice sheet of northernEurasia. Quat. Res. 13:1–32.
Guyomard, R. 1989. Diversite genetique de la truite commune. Bull.Fr. Peche Piscic. 314:118–135.
Hamilton, K. E., A. Ferguson, J. B. Taggart, and T. Tomasson.1989. Post-glacial colonization of brown trout, Salmo trutta L.:Ldh-5 as a phylogeographic marker locus. J. Fish Biol. 35:651–664.
Hansen, M. M., and V. Loeschcke. 1996. Genetic differentiationamong Danish brown trout populations, as detected by RFLPanalysis of PCR amplified mitochondrial DNA segments. J. FishBiol. 48:422–436.
Hansen, M. M., and K.-L. D. Mensberg. 1998. Genetic differenti-ation and relationship between genetic and geographical distancein Danish sea trout (Salmo trutta L.) populations. Heredity 81:493–504.
Harpening, H. 1994. Signature of ancient population growth in alow resolution mitochondrial DNA mismatch distribution. Hum.Biol. 66:591–600.
Hewitt, H. G. M. 2000. The genetic legacy of the Quaternary iceages. Nature. 405:907–913.
Hudson, R. R. 1990. Gene genealogies and the coalescent process.
375EVOLUTIONARY HISTORY OF BROWN TROUT
Pp. 1–44 in D. Futuyama and J. Antonovics, eds. Oxford surveysin Evolutionary Biology, Vol.7. Oxford Univ. Press, Oxford.
Hynes, R. A., A. Ferguson, and M. A. McCann. 1996. Variation inmitochondrial DNA and post-glacial colonization of north west-ern Europe by brown trout. J. Fish Biol. 48:54–67.
Ibrahim, K. M., R. A. Nichols, and G. M. Hewitt. 1996. Spatialpatterns of genetic variation by different forms of dispersal dur-ing range expansion. Heredity 77:282–291.
Keith, P. 1998. Evolution des peuplements ichtyologiques de Franceet strategies de conservation. Ph.D. diss., Universite de RennesI, France.
Kottelat, M. 1997. European freshwater fishes. Biol. Brat. 52(suppl.5):1–271.
Largiader, C. R., and A. Scholl. 1996. Genetic introgression be-tween native and introduced brown trout Salmo trutta L. pop-ulations in the Rhone River basin. Mol. Ecol. 5:417–426.
Lavery, S., C. Moritz, and D. R. Fielder. 1996. Genetic patternssuggest exponential population growth in a declining species.Mol. Biol. Evol. 13:1106–1113.
Lu, G., and L. Bernatchez. 1998. Experimental evidence for reducedhybrid viability between dwarf and normal ecotypes of lakewhitefish (Coregonus clupeaformis Mitchill). Proc. Roy. Soc.Lond. Series B 265:1025–1030.
———. 1999. Correlated trophic specialization and genetic diver-gence in sympatric lake whitefish ecotypes (Coregonus clupea-formis): support for the ecological speciation hypothesis. Evo-lution 53:1491–1505.
McElroy, D. P., P. Moran, E. Bermingham, and I. Kornfield. 1992.REAP: an integrated environment for the manipulation and phy-logenetic analysis of restriction data. J. Hered. 83:157–158.
Merila, J., M. Bjorklund, and A. J. Baker. 1997. Historical de-mography and present day population structure of the greenfinch,Carduelis chloris. Evolution 51:946–956.
Moritz, C., and E. Bermingham. 1998. Comparative phylogeogra-phy: concepts and applications. Mol. Ecol. 7:367–370.
Nei, M. 1987. Molecular evolutionary genetics. Columbia Univ.Press, New York.
Nesbo, C. L., T. Fossheim, L. A. Vollestad, and K. S. Jakobsen.1999. Genetic divergence and phylogeographic relationshipsamong European perch (Perca fluviatilis) populations reflect gla-cial refugia and postglcial colonization. Mol. Ecol. 8:1387–1404.
Osinov, A. G. 1990. The levels of genetic variation and differen-tiation of the brown trout (Salmo trutta L.) in Tadjikistan. Biol.Sci. Bull. 45:37–41.
Osinov, A., and L. Bernatchez. 1996. Atlantic and Danubean phy-logenetic groupings of brown trout (Salmo trutta L.) complex:genetic divergence, evolution, and conservation. J. Ichthyol. 36:762–786.
Persat, H., and P. Berrebi. 1990. Relative ages of present popula-tions of Barbus barbus and Barbus meridionalis (Cyprinidae) insouthern France: preliminary considerations. Aquat. Living Re-sour. 3:253–263.
Petit, E., L. Excoffier, and F. Mayer. 1999. No evidence of bottle-neck in the postglacial recolonization of Europe by the noctulebat (Nyctalus noctula). Evolution 53:1247–1258.
Poteaux, C., D. Beaudou, and P. Berrebi. 1998. Temporal variationsof genetic introgression in stocked brown trout population. J.Fish Biol. 53:701–713.
Presa, P. M. 1995. Determinisme et polymorphisme genetiques dessequences microsatellites de Salmo trutta et d’autres salmonides.Comparaison avec le polymorphisme des locus enzymatiques.Ph.D. diss., Universite de Paris-Sud, Orsay, France.
Rogers, A. R. 1995. Genetic evidence for a Pleistocene populationexplosion. Evolution 49:608–615.
Rogers, A. R., and H. Harpending. 1992. Population growth makeswaves in the distribution of pairwise genetic differences. Mol.Biol. Evol. 9:552–569.
Ryman, N. 1983. Patterns of distribution of biochemical geneticvariation in salmonids: differences between species. Aquaculture33:1–21.
Schneider, S., and L. Excoffier. 1999. Estimation of past demo-graphic parameters from the distribution of pairwise differenceswhen the mutation rates vary among sites: application to humanmitochondrial DNA. Genetics 152:1079–1089.
Schneider, S., D. Roessli, and L. Excoffier. 1999. Arlequin, version2.0: a software for genetic data analysis. Genetics and BiometryLaboratory, Univ. of Geneva.
Smith, G. R. 1992. Introgression in fishes: significance for pale-ontology, cladistics, and evolutionary rates. Syst. Biol. 41:41–57.
Templeton, A. R. 1998. Nested clade analyses of phylogeographicdata: testing hypotheses about gene flow and population history.Mol. Ecol. 7:381–398.
Templeton, A. R., and C. F. Sing. 1993. A cladistic analysis ofphenotypic associations with haplotypes inferred from restric-tion endonucleases mapping. IV. Nested analyses with clado-gram uncertainty and recombination. Genetics 134:659–669.
Templeton, A. R., E. Boerwinkle, and C. F. Sing. 1987. A cladisticanalysis of phenotypic associations with haplotypes inferredfrom restriction endonuclease mapping. I. Basic theory and ananalysis of alcohol dehydrogenase activity in Drosophila. Ge-netics 117:343–351.
Templeton, A. R., K. A. Crandall, and C. F. Sing. 1992. A cladisticanalysis of phenotypic associations with haplotypes inferredfrom restriction endonuclease mapping and DNA sequence data.III. Cladogram estimation. Genetics 132:619–633.
Templeton, A. R., E. Routman, and C. Phillips. 1995. Separatingpopulation structure from population history: a cladistic analysisof the geographical distribution of mitochondrial DNA haplo-tytpes in the tiger salamander, Ambystoma tigrinum. Genetics140:767–782.
Turner, T. F., J. C. Trexler, J. L. Harris, and J. L. Haynes. 2000.Nested cladistic analysis indicates population fragmentationshapes genetic diversity in a freshwater mussel. Genetics 154:777–785.
Villinger, E. 1986. Untersuchungen zur Flussgeschichte von Aare-Donau/Alpenrhein und zur Entwicklung des Malm-Karts in Sud-westdeutschland. Germany.
Watterson, G. 1975. On the number of segregating sites in geneticalmodels without recombination. Theor. Popul. Biol. 7:256–276.
Webb, T., III, and P. J. Bartlein. 1992. Global changes during thelast 3 millions years: climatic controls and biotic responses.Annu. Rev. Ecol. Syst. 23:141–173.
Weiss, S., A. Antunes, C. Schlotterer, and P. Alexandrino. 2000.Mitochondrial haplotype diversity of brown trout Salmo truttaL. in Portuguese rivers supports a simple broad-scale model ofthe Pleistocene recolonization of northern Europe. Mol. Ecol.9:691–698.
Wilson, C. C., P. D. N. Hebert, J. D. Reist, and J. B. Dempson.1996. Phylogeography and postglacial dispersal of arctic charrSalvelinus alpinus in North America. Mol. Ecol. 5:187–197.
Zink, R. M. 1996. Comparative phylogeography in North Americanbirds. Evolution 50:308–317.
Corresponding Editor: R. Burton
376 LOUIS BERNATCHEZ
AP
PE
ND
IX.
Sam
ple
size
,lo
cati
ons,
and
dist
ribu
tion
ofco
mpo
site
hapl
otyp
esam
ong
nati
vepo
pula
tion
sof
brow
ntr
out.
Ref
eren
ces:
1.A
post
olid
iset
al.
(199
6a,
1997
);2.
Ber
natc
hez
and
Osi
nov
(199
5);
3.B
erna
tche
zet
al.
(199
2);
4.G
iuff
raet
al.
(199
4);
5.H
ynes
etal
.(1
996)
;6.
Osi
nov
and
Ber
natc
hez
(199
6);
7.th
isst
udy.
No.
Lin
eage
Pop
ulat
ion
N
Coo
rdin
ates
Lat
itud
eL
ongi
tude
Bas
inS
ampl
ing
Com
posi
teha
plot
ypes
(num
bers
)R
efer
-en
ce
1 2 3 4 5
AD
AD
AD
AD
AD
Zrm
anja
R.
Mus
kovc
iR
.K
rupa
R.
Tohm
aR
.Z
aman
tiR
.
4 3 3 3 3
4480
89N
4481
29N
4481
09N
3884
79N
3884
59N
1681
09E
1584
59E
1585
29E
3685
59E
3683
09E
Adr
iati
cS
eaA
dria
tic
Sea
Adr
iati
cS
eaP
ersi
anG
ulf
Med
iter
rane
anS
ea
1996
1996
1996
1993
1993
AD
s4r8
AD
s4r8
AD
s4r8
AD
s6r1
AD
s1r3
7 7 7 7 76 7 8 9 10
AD
AD
AD
AD
AD
Cey
han
R.
Ana
mur
R.
Gok
suR
.K
opru
R.
Pre
psa
L.
1 1 2 3 26
3784
59N
3681
59N
3684
09N
3781
49N
4085
09N
3684
09E
3285
09E
3283
89E
3181
09E
2181
59E
Med
iter
rane
anS
eaM
edit
erra
nean
Sea
Med
iter
rane
anS
eaM
edit
erra
nean
Sea
Adr
iati
cS
ea
1993
1993
1993
1993
1993
AD
s1r1
AD
s1r1
AD
s5r4
AD
s1r3
AD
s1r1
0
7 7 7 7 1,7
11 12 13 14 15
AD
AD
AD
AD
AD
Tara
voR
.O
hrid
L.
Dra
nce
R.
San
gro
R.
Fib
reno
L.
8 38 8 1 6
4185
29N
4180
09N
4682
49N
4184
59N
4184
29N
0980
79E
2084
09E
0683
09E
1385
09E
1384
09E
Med
iter
rane
anS
eaA
dria
tic
Sea
Med
iter
rane
anS
eaA
dria
tic
Sea
Tyr
rhen
ian
Sea
(Med
.)
1991
1991
1990
1996
1996
–199
7
AD
s3r3
AD
s1r1
1(32
)-A
Ds1
0r11
(6)
AD
s1r1
(7)-
AD
s2r1
AD
s4r8
AD
s1r1
3,7
1,3,
73,
77 7
16 17 18 19 20
AD
AD
AD
AD
AD
Sin
niR
.P
ardu
R.
Osu
R.
Lia
mon
eR
.F
angu
R.
2 2 1 1 1
4080
59N
3985
09N
4183
59N
4281
59N
4283
09N
1585
59E
0983
09E
0982
09E
0884
59E
0883
89E
Ioni
anS
ea(M
ed.)
Sar
dini
aC
orsi
caC
orsi
caC
orsi
ca
1996
1996
1996
1996
1996
AD
s4r8
AD
s3r3
AD
s1r1
AD
s1r1
AD
s3r3
7 7 7 7 721 22 23 24 25
AD
AD
AD
AD
AD
Asc
uR
.L
ouro
sR
.D
roso
pigi
(Lou
dhia
sR
.)S
kopo
sR
.(L
oudh
ias
R.)
Val
bona
R.
(Alb
ania
)
2 1 5 8 5
4283
39N
3982
09N
4080
09N
4085
09N
4282
09N
0980
59E
2085
09E
2182
59E
2184
09E
2080
59E
Cor
sica
Adr
iati
cS
eaA
egea
nS
eaA
egea
nS
eaA
dria
tic
Sea
1996
1993
1994
1994
1993
AD
s3r3
AD
s1r3
AD
s1r6
AD
s1r3
AD
s1r1
7 7 7 7 726 27 28 29 30
AD
AD
AD
AD
AD
Ese
Ver
acul
ungu
Gol
oC
astr
ila
Krk
icR
.S
tura
diD
emon
te
4 5 4 4 15
4185
79N
4185
29N
4282
29N
4480
59N
4482
59N
0980
39E
0980
99E
0980
89E
1681
29E
0781
09E
Med
iter
rane
anS
eaM
edit
erra
nean
Sea
Med
iter
rane
anS
eaA
dria
tic
Sea
Adr
iati
cS
ea
1996
1996
1996
1997
1991
AD
s3r3
AD
s3r3
AD
s3r3
(3)-
AD
s3r7
AD
s1r1
AD
s1r1
7 7 7 7 431 32 33 34 35
AD
AD
AD
AD
AD
Alfi
osE
vino
sT
hyam
isT
ripo
tam
osD
imca
yR
.
5 6 25 26 1
3783
79N
3882
79N
3985
09N
4083
59N
3683
49N
2182
79E
2184
09E
2083
09E
2281
59E
3281
79E
Ioni
anS
eaIo
nian
Sea
Ioni
anS
eaA
egea
nS
eaM
edit
erra
nean
Sea
— — — — 1998
AD
s7r9
AD
s1r1
AD
s1r9
AD
s1r1
(4)-
Ads
8r1(
22)
AD
s2r1
1 1 1 1 736 37 38 39 40
AD
AD
-AT
AD
-MA
AD
AD
-MA
-ME
Ado
ur(G
aube
)A
rtes
iaga
R.
Krk
aR
.N
esto
sR
.G
arda
L.
5 9 4 25 23
4285
09N
4380
39N
4385
09N
4182
09N
4585
09N
0080
99W
0183
39W
1585
89E
2484
09E
1084
09E
Atl
anti
cA
tlan
tic
Adr
iati
cS
eaA
egea
nS
eaA
dria
tic
Sea
1996
1992
1996
1993
1990
–199
1
AD
s1r1
(4)-
AD
s1r5
AD
s1r1
(5)-
AT
s1r1
(3)-
AT
s1r8
AD
s1r1
-AD
s4r8
-MA
s1r1
(2)
AD
s9r1
(11)
AD
s1r1
(15)
-MA
s2r1
(5)-
MA
s3r1
-Mes
2r1(
2)
7 7 7 1,7
3,4,
7
41 42 43 44 45 46 47
AD
-MA
-ME
AD
-ME
AT
AT
AT
AT
AT
Chi
sone
R.
Vec
chio
R.
Aub
onne
R.
Red
onR
.T
sna
R.
Bre
sle
R.
Saa
rR
.
23 6 5 3 6 8 8
4485
09N
4281
39N
4682
79N
4684
09N
5782
09N
4685
09N
4980
09N
0781
09E
0981
29E
0682
39E
0780
09E
3480
09E
0183
59E
0780
09E
Adr
iati
cS
ea
Med
iter
rane
anS
eaM
edit
erra
nean
Sea
Med
iter
rane
anS
eaC
aspi
anS
eaA
tlan
tic
Atl
anti
c
1990
–199
1
1991
1990
1990
1994
1990
1990
AD
s1r1
(12)
-MA
s2r1
(8)-
ME
s2r1
(3)
AD
s3r3
(5)-
ME
s2r1
AT
s1r1
AT
s1r1
AT
s1r1
(3)-
AT
s1r2
(2)-
AT
s1r7
AT
s1r1
(4)-
AT
s1r2
(4)
AT
s1r1
(6)-
AT
s1r2
(2)
3,4,
7
3,7
3 3 6 3,7
3,7
377EVOLUTIONARY HISTORY OF BROWN TROUTA
PP
EN
DIX
.C
onti
nued
.
No.
Lin
eage
Pop
ulat
ion
N
Coo
rdin
ates
Lat
itud
eL
ongi
tude
Bas
inS
ampl
ing
Com
posi
teha
plot
ypes
(num
bers
)R
efer
-en
ce
48 49 50 51 52 53
AT
AT
AT
AT
AT
AT
Vit
ava
R.
Slu
psk
R.
Sw
ibno
R.
Tri
ashe
noL
.N
il’m
aR
.M
edja
R.
8 4 2 14 7 8
4982
09N
5484
09N
5484
59N
6980
09N
6683
09N
5681
59N
1481
09E
1780
09E
1885
09E
3684
09E
3381
59E
3284
09E
Atl
anti
cA
tlan
tic
Atl
anti
cA
tlan
tic
Bar
ents
Sea
Whi
teS
ea
1991
1990
1990
1993
1993
1993
AT
s1r3
AT
s1r2
(2)-
AT
s1r1
-AT
s1r1
3A
Ts1
r1A
Ts1
r2A
Ts1
r1(6
)-A
Ts1
r2A
Ts1
r2(5
)-A
Ts1
r3(3
)
3,7
3,7
3,7
2,6
2,6
254 55 56 57 58 59
AT
AT
AT
AT
AT
AT
Via
shen
oL
.V
orob
’yev
R.
Luv
en’g
aR
.Je
gess
lerg
-dra
mm
erD
alal
ven
R.
Lek
saR
.
2 3 6 7 8 10
6980
09N
6682
59N
6782
09N
6081
09N
6083
09N
6382
09N
3687
09E
3382
09E
3281
09E
1182
09E
1581
09E
1085
59E
Bal
tic
Sea
Bar
ents
Sea
Whi
teS
eaW
hite
Sea
Atl
anti
cA
tlan
tic
1994
1994
1994
1992
1992
1994
AT
s1r2
AT
s1r1
AT
s1r2
(4)-
AT
s1r1
(2)
AT
s1r1
AT
s1r3
(6)-
AT
s1r2
(2)
AT
s1r1
(2)-
AT
s1r2
(7)-
AT
s1r3
6 6 6 3,7
3,7
760 61 62 63 64 65
AT
AT
AT
AT
AT
AT
Tavl
aaR
.Jy
llan
d(D
enm
ark)
Kal
lsjo
nF
alpf
jall
t-ja
mar
naB
lank
tjar
nL
eba
10 30 10 10 5 18
6482
79N
5681
09N
6384
09N
5983
59N
5983
59N
5880
09N
1183
69E
0980
09E
1380
09E
1485
09E
1482
59E
1980
09E
Atl
anti
cA
tlan
tic
Atl
anti
cA
tlan
tic
Atl
anti
cA
tlan
tic
1994 — — — — —
AT
s1r1
(4)-
AT
s1r2
(6)
AT
s1r1
AT
s1r2
AT
s1r2
AT
s1r1
(2)-
AT
s1r3
(3)
AT
s1r1
(5)-
AT
s1r2
(11)
-A
Ts1
r3(2
)
7 5 5 5 5 5
66 67 68 69 70
AT
AT
AT
AT
AT
Cru
mli
n
Oue
dB
erre
mB
ouH
amed
Oue
del
Kan
arO
ued
Ziz
56 4 2 2 5
5382
59N
3284
09N
3581
39N
3581
49N
3282
59N
1080
59E
0484
79W
0580
19W
0580
59W
0484
59W
Atl
anti
c
Med
iter
rane
anS
eaM
edit
erra
nean
Sea
Med
iter
rane
anS
eaA
tlan
tic
— 1995
1995
1995
1992
AT
s1r2
(3)-
AT
s1r1
(6)-
Ats
8r14
(17)
AT
s5r1
2A
Ts1
r1A
Ts1
r1A
Ts6
r4(2
)-A
Ts7
r5(3
)
5 7 7 7 771 72 73 74 75 76
AT
AT
AT
AT
AT
AT
Oue
dO
umer
Rbi
aA
riza
kum
R.
Tru
bia
R.
Elo
rnR
.R
iabh
aich
Icel
and
(13
pop)
6 7 7 8 113
3
3380
59N
4381
59N
4381
59N
4882
79N
5083
09N
6482
09N
0581
29W
0284
09W
0680
09W
0481
69W
0484
09W
1883
09W
Atl
anti
cA
tlan
tic
Atl
anti
cA
tlan
tic
Atl
anti
cA
tlan
tic
1992
1992
1992
1990
1994
1994
AT
s6r4
AT
s1r9
AT
s1r1
(5)-
AT
s1r9
(2)
AT
s1r2
AT
s1r3
AT
s1r2
7 7 7 3,7
7 577 78 79
AT
AT
AT
Mel
vin
L
Gly
nnR
.B
asqu
ere
gion
140
102 5
5483
09N
5484
59N
4381
79N
0881
09W
0585
09W
0183
39W
Atl
anti
c
Atl
anti
cA
tlan
tic
1994
1994
1996
AT
s1r2
(108
)-A
Ts1
r10(
3)-
AT
s4r2
(16)
-AT
s1r6
(9)-
AT
s1r1
(4)
AT
s1r2
(101
)-A
Ts1
r3A
Ts1
r1(4
)-A
Ts1
r9
5 5 780 81 82 83 84
AT
AT
AT
AT
AT
Gle
nari
ffR
.
Car
nlou
ghG
lenc
loy
Sut
herl
and
Loc
hN
ess
151 54 15 32 15
5580
59N
5580
09N
5485
59N
5883
09N
5783
09N
0680
59W
0680
09W
0680
09W
0485
09W
0485
09W
Atl
anti
c
Atl
anti
cA
tlan
tic
Atl
anti
cA
tlan
tic
1994 — — — —
AT
s1r3
(107
)-A
Ts1
r2(3
9)-
AT
s1r1
(3)-
AT
s1r1
3(2)
AT
s1r2
(50)
-AT
s1r3
(4)
AT
s1r2
(9)-
AT
s1r3
(6)
AT
s1r3
AT
s1r1
(14)
-AT
s1r2
5 5 5 5 585 86 87 88 89 90
AT
AT-
DA
DA
DA
DA
DA
Ern
eE
ulen
back
R.
Cet
ina
R.
Kok
raR
.G
urk
R.
Boh
ing
L.
50 6 4 3 5 1
5482
59N
4882
09N
4384
29N
4682
39N
4685
09N
4681
79N
0785
09W
1180
09E
1684
49E
1482
99E
1480
39E
1385
09E
Atl
anti
cB
lack
Sea
Adr
iati
cS
eaB
lack
Sea
Bla
ckS
eaB
lack
Sea
— 1990
1996
1996
1991
1991
AT
s1r2
AT
s1r1
(4)-
DA
s2r2
(2)
DA
s12r
22D
As1
r1D
As7
r9D
As2
r2
5 3,7
7 7 7 391 92 93
DA
DA
DA
Aba
ntL
.C
oruh
R.
Fyr
atR
.(E
uphr
ates
)
2 2 3
4083
89N
4083
09N
3983
59N
3181
69E
4183
09E
3884
09E
Bla
ckS
eaB
lack
Sea
Per
sian
Gul
f
1993
1992
1992
DA
s1r1
7D
As7
r9D
As5
r5
7 7 7
378 LOUIS BERNATCHEZ
AP
PE
ND
IX.
Con
tinu
ed.
No.
Lin
eage
Pop
ulat
ion
N
Coo
rdin
ates
Lat
itud
eL
ongi
tude
Bas
inS
ampl
ing
Com
posi
teha
plot
ypes
(num
bers
)R
efer
-en
ce
94 95 96
DA
DA
DA
Sav
aR
.L
avan
tB
r.V
ella
chS
trea
m
1 1 6
4682
69N
4780
49N
4683
59N
1385
39E
1483
69E
1483
59E
Bla
ckS
eaB
lack
Sea
Bla
ckS
ea
1993
1993
1994
1DA
s11r
18D
As7
r9D
As1
r1-D
as7r
9(2)
-DA
s1r1
8(3)
7 7 797 98 99 10
010
1
DA
DA
DA
DA
DA
Sav
aR
.
Rad
ovna
R.
Vil
san
R.
Doa
mne
iR
.To
polo
gR
.
6 2 5 4 3
4682
89N
4682
39N
4581
79N
4582
19N
4582
89N
1384
69E
1482
09E
2484
69E
2484
99E
2483
09E
Bla
ckS
ea
Bla
ckS
eaB
lack
Sea
Bla
ckS
eaB
lack
Sea
1994
1995
1995
1995
1995
DA
s12r
19-D
As1
1r18
-D
As1
2r22
(3)-
DA
s13r
23D
As1
1r18
DA
s2r2
0D
As1
r18
DA
s2r2
1(2)
-DA
s1r1
7 7 7 7 710
210
310
410
510
610
7
DA
DA
DA
DA
DA
DA
Gor
enR
.M
ende
res
R.
Kol
paR
.Z
avrs
nica
R.
Beg
unje
R.
Arp
aR
.
1 1 1 1 2 8
3984
59N
3984
59N
4583
09N
4682
49N
4682
39N
3984
09N
2781
09E
2685
09E
1580
09E
1481
09E
1481
49E
4583
59E
Aeg
ean
Sea
Aeg
ean
Sea
Bla
ckS
eaB
lack
Sea
Bla
ckS
eaC
aspi
anS
ea
1993
1993
1996
1996
1996
1982
DA
s5r5
DA
s2r2
5D
As1
r24
DA
s1r1
8D
As1
r18
DA
s1r1
5
7 7 7 7 7 210
810
911
011
111
2
DA
DA
DA
DA
DA
Tere
kR
.A
psha
kR
.K
uter
ziR
.(U
ral)
Sev
anL
.K
odor
iR
.
10 14 7 6 9
4384
59N
5383
09N
5384
09N
4083
59N
4381
09N
4482
89E
5883
09E
5883
09E
4580
09E
4183
09E
Cas
pian
Sea
Cas
pian
Sea
Cas
pian
Sea
Cas
pian
Sea
Bla
ckS
ea
1982
–199
119
9419
9419
79–1
982
1982
DA
s1r1
0(8)
-DA
s1r1
1-D
As1
r12
DA
s1r1
5D
As1
r15
DA
s1r6
(5)-
DA
s6r6
DA
s5r5
(3)-
DA
s7r5
(3)-
DA
s1r1
6(3)
2 6 6 2 2
113
114
115
116
DA
DA
DA
DA
Cri
mea
(dif
fere
nttr
ibut
arie
s)S
arda
mia
naR
.
Sofi
daro
nR
.
Ner
etva
R.
5 7 6 2
4583
09N
3980
89N
3885
09N
4384
09N
3482
09E
6881
59E
6985
59E
1880
09E
Bla
ckS
eaA
ral
Sea
Ara
lS
ea
Adr
iati
cS
ea
1992
1987
1987
1991
DA
s7r9
(4)-
DA
s1r1
6D
As3
r3-D
As8
r3-D
As1
r3(2
)-D
As1
r14(
3)D
As4
r4-D
As9
r7-D
As1
r8-
DA
s9r1
3-D
As1
r4(2
)D
As1
r1
2 2 2 3,7
117
118
119
120
121
122
DA
DA
DA
DA
DA
DA
Dra
vaR
.L
udra
naR
.B
istr
aR
.To
pla
R.
Dra
vaR
.L
udra
naR
.
1 2 2 4 1 2
4683
79N
4682
59N
4682
39N
4683
09N
4683
79N
4682
59N
1580
59E
1485
59E
1485
09E
1484
59E
1580
59E
1485
59E
Bla
ckS
eaB
lack
Sea
Bla
ckS
eaB
lack
Sea
Bla
ckS
eaB
lack
Sea
1997
1997
1997
1997
1997
1997
DA
s1r1
8D
As7
r9D
As1
2r22
DA
s7r9
-DA
s12r
22(3
)D
As1
r18
DA
s7r9
7 7 7 7 7 712
312
412
512
612
7
DA
DA
DA
DA
DA
Bis
tra
R.
Bal
ikG
olu
L.
Gol
bele
nR
.F
irti
naR
.B
alkl
iR
.
2 2 3 4 1
4682
39N
3985
09N
4180
59N
4180
09N
3985
59N
1485
09E
4383
09E
4380
59E
4180
09E
4080
09E
Bla
ckS
eaC
aspi
anS
eaC
aspi
anS
eaB
lack
Sea
Per
sian
Gul
f
1997
1997
1997
1997
1997
DA
s12r
22D
As2
r27
DA
s1r2
6D
As5
r5D
As7
r9
7 7 7 7 712
812
913
013
113
2
DA
DA
DA
DA
DA
Gor
enR
.Z
eyti
nili
R.
Man
asti
rR
.To
pkha
naR
.K
apuz
basi
R.
1 2 2 9 1
3984
59N
3984
09N
3984
09N
3681
19N
3784
69N
2781
09E
2685
99E
2685
09E
7085
79E
3582
59E
Aeg
ian
Sea
Aeg
ian
Sea
Aeg
ian
Sea
Ara
lM
edit
erra
nean
Sea
1997
1997
1997
1998
DA
s5r5
DA
s5r5
DA
s1r2
6D
As1
r28-
DA
s1r4
(8)
DA
s1r1
7 7 7 7 713
313
3b13
413
513
613
7
DA
DA
MA
MA
MA
MA
Mez
aR
.K
yzyl
suS
ocha
R.
Pel
lice
R.
Toce
R.
Hud
daG
rapp
a
2 3 4 27 18 3
4682
79N
3984
09N
4682
09N
4485
09N
4681
09N
4681
09N
1484
39E
7380
09E
1384
09E
0783
89E
0881
09E
1480
09E
Bla
ckS
eaA
ral
Adr
iati
cS
eaA
dria
tic
Sea
Adr
iati
cS
eaM
edit
erra
nean
Sea
1998
1999
1994
1990
–199
119
9019
96
DA
s1r1
Das
1r4
MA
s1r1
MA
s2r1
MA
s1r1
(9)-
MA
s2r1
(9)
MA
s1r1
7 7 7 3,4,
73,
4,7
713
813
9M
AM
AZ
adla
scic
aS
tura
diL
anzo
5 1946
8109
N45
8069
N13
8509
E07
8449
EA
dria
tic
Sea
Adr
iati
cS
ea19
9619
91M
As1
r1M
As2
r17 4
379EVOLUTIONARY HISTORY OF BROWN TROUT
AP
PE
ND
IX.
Con
tinu
ed.
No.
Lin
eage
Pop
ulat
ion
N
Coo
rdin
ates
Lat
itud
eL
ongi
tude
Bas
inS
ampl
ing
Com
posi
teha
plot
ypes
(num
bers
)R
efer
-en
ce
140
141
142
143
MA
MA
MA
MA
Bre
nta
R.
Ges
soR
.S
arca
R.
Stu
radi
Dem
onte
(Vin
adio
)
8 16 8 15
4585
09N
4482
49N
4585
29N
4482
09N
1184
09E
0783
39E
1085
29E
0782
09E
Adr
iati
cS
eaA
dria
tic
Sea
Adr
iati
cS
eaA
dria
tic
Sea
1991
1991
1991
1991
MA
s2r1
MA
s2r1
MA
s2r1
MA
s2r1
4 4 4 414
414
514
614
714
814
9
MA
MA
MA
MA
ME
ME
Ach
eloo
sM
orno
sId
rija
R.
Zal
aR
.V
oido
mat
isR
.R
ioL
lobr
egat
48 15 1 2 27 2
3882
09N
3882
59N
4680
09N
4585
59N
3985
59N
4281
59N
2180
69E
2185
09E
1385
59E
1480
59E
2084
09E
0185
79E
Ioni
anS
eaIo
nian
Sea
Adr
iati
cS
eaA
dria
tic
Sea
Adr
iati
cS
eaM
edit
erra
nean
Sea
— — 1998
1998
1993
1995
AD
s1r1
(5)-
MA
s1r1
(43)
MA
s1r1
MA
s1r1
MA
s1r1
ME
s2r1
(25)
-ME
s3r1
(2)
ME
s1r1
1 1 7 7 1,7
715
015
115
215
315
415
5
ME
ME
ME
ME
ME
ME
Rev
erot
teR
.Te
sR
.C
alor
eR
.C
aren
caA
ude
Sor
gue
(Vau
clus
e)
6 6 1 5 4 3
4781
59N
4385
09N
4084
69N
4282
69N
4284
59N
4385
59N
0684
09E
0381
59E
1580
09E
0281
39E
0280
89E
0581
09E
Med
iter
rane
anS
eaM
edit
erra
nean
Sea
Tyr
rhen
ian
Sea
(Med
.)M
edit
erra
nean
Sea
Med
iter
rane
anS
eaM
edit
erra
nean
Sea
1986
1990
1996
1996
1996
1996
ME
s2r1
ME
s1r1
(4)-
ME
s2r1
(2)
ME
s1r1
ME
s1r1
ME
s1r1
ME
s1r1
3,7
3,7
7 7 7 715
615
715
815
916
016
1
ME
ME
ME
ME
ME
-AD
ME
-DA
Hau
tG
olo
Fon
tana
ccia
R.
Tagl
iole
R.
Rio
Ara
Rip
aR
.V
enet
ikos
4 8 8 2 6 22
4281
89N
4481
29N
4481
29N
4283
09N
4485
79N
4080
39N
0885
39E
1084
59E
1083
79E
0081
09W
0684
79E
2183
39E
Med
iter
rane
anS
eaA
dria
tic
Sea
Adr
iati
cS
eaM
edit
erra
nean
Sea
Adr
iati
cS
eaA
egea
nS
ea
1996
1991
1991
1992
1991 —
ME
s1r1
ME
s1r1
ME
s1r1
ME
s1r1
ME
s2r1
(5)-
AD
s1r1
ME
s2r1
(2)-
DA
s14r
1(20
)
7 4 4 7 4 1