Evolutionary Genetics of Flycatchers. Ii

EVOLUTIONARY GENETICS OF FLYCATCHERS. II. DIFFERENTIATION IN THE EMPIDONAX DIFFICILIS COMPLEX

NEI) K. JOHNSONAND JILL A. MARTEN1 Museumof VertebrateZoology and Department of Zoology,University of California, Berkeley,California 94720 USA

ABSTR^CT.--Weused starch-gel electrophoresis to assessvariability at 41 geneticloci in 208 individualsfrom 11 breedingpopulations of the WesternFlycatcher (Empidonax difficilis) complex.Genic variability was substantialin mostpopulations and equivalentto levelsfound in other avian taxa. A sampleof E. d. insulicolafrom SantaCatalina Island, however, showed reducedheterozygosity and an unusuallylow percentageof polymorphicloci. We attribute this to a bottleneckat the time of the original colonization. Nei's genetic distancesamong populationsof onetaxon ranged from I• = 0.0003(in E. d. difficilis)to I• = 0.0033(in E. d. hellmayri).Intertaxon Nei's/• rangedfrom 0.009 (E. d. insulicolavs. E. d. difficilis)and 0.0149 (E. d. difficilisvs. E. d. hellmayri)to 0.0228(E. d. insulicolavs. E. d. hellmayri).Fst statistics revealed significantpopulation subdivision within the complex.With Slatkin'srare-allele method we estimatedthe gene-flowparameter, Nm. Mainland populationsexperience moderately high gene flow (9.62 immigrants/generation).In contrast,Santa Catalina Island receivesan estimated 0.093 immigrants/generation,pointing to very low gene flow and essentialgenetic isolation. Geneticdistances yielded phenogramsand distanceWagner treesthat provide hypotheses for the relationshipsand phylogenesisof populationsin western North America.The lineage leading to modernE. d. difficilissplit from that leadingto E. fiavescensin the mid-Pleistocene at 866,800yr BP;the ancestorsof modernE. d. difficilisdiverged from thoseof present-dayE. d. hellmayriat 248,700yr BP;and the stockleading to modernE. d. insulicolabudded from the lineagethat becameE. d. difficilisin the late Pleistocene,approximately 187,000 yr BP. Empidonaxd. difficilisand E. d. hellmayrinest sympatricallyand mate assortativelyin the Siskiyou region of northern California. Interbreeding has not been demonstratedconclu- sively,and we regardthese taxa as biologic species. In the absenceof a testof sympatry,the well-differentiatedform E. d. insulicolaof the California Channel Islandscannot be proved to be a biologicspecies. It is clearlya phylogeneticspecies, however, in the senseojf Cracraft. Received14 May 1987, accepted2 October1987.

THEAmerican Ornithologists' Union (A.O.U. netic variationin restrictedlineages comprised 1983) includes the various forms of the Empi- of relatedclusters of near-speciesand full species donaxdifficilis complex under one species,the providea valuableperspective on phylogenetic Western Flycatcher. Insular, coastal, and inte- history, one that is not always clear from study rior populationsof the western United States, of morphology(Wake 1981).Furthermore, be- however,are stronglydifferentiated in size,col- causeof their great morphologicresemblance, or, voice, and preferred habitat (Johnson1980). thesetyrannids are aptly termedsibling species In this paper we analyze allozyme data as re- (Dobzhansky1951), a categoryoften suspected lated to the evolutionary statusand taxonomic in the past of being more similar genetically rank of E. difficilis. than "ordinary"species (Dobzhansky 1978: 101). It is alsoof interestto compareallozyme vari- Although the evidence available for sibling ation in the Western Flycatchercomplex with speciesof flycatchers(in Empidonaxand Mio- that shown by its more distantly related con- nectes)does not support this view (Zink and geners(Zink and Johnson1984). Levels of ge- Johnson1984, Capparella and Lanyon 1985), the degree to which molecularand morphologic featuresare either congruentor decoupledin ' Presentaddress: Department of Pathology,Pacific both sibling speciesand non-sibling species PresbyterianMedical Center, Clay at BuchananStreet, continues to merit attention (Lewin 1985). San Francisco, California 94120 USA. Other generalgoals encourage additional avi-

177 The Auk 105:177-191. January 1988 178 JohNsoNAND MARTEN [Auk,Vol. 105

TABLE1. Geneticvariability measuresfor 11 samplesrepresenting 3 taxaof the Empidonaxdifficilis complex.

Percent- age poly- Mean morphic no. Taxon n Sample area namea Hobs.ñ SE H,st.ñ SE locib allelesc E. d. insulicola 27 Santa Catalina Isl. 0.037 +- 0.005 0.034 _+ 0.018 9.76 1.12 E. d. difficilis 16 Monterey 0.049 ñ 0.009 0.053 +_0.021 19.51 1.24 E. d. difficilis 15 Lake Co. 0.054 ñ 0.007 0.057 ñ 0.021 24.39 1.29 E. d. difficilis 18 Shasta 0.051 ñ 0.007 0.054 ñ 0.021 26.83 1.34 E. d. difffi'cilis 10 Rogue River 0.061 ñ 0.013 0.058 +_0.023 19.51 1.29 E. d. hellmayri+ E. d. difficilis 17 Siskiyou 0.066 ñ 0.004 0.065 -+ 0.025 19.51 1.32 E. d. hellmayri 23 Warner 0.053 _+0.007 0.057 ñ 0.021 24.39 1.39 E. d. hellmayri 20 Snake 0.052 ñ 0.006 0.055 ñ 0.021 21.95 1.34 E. d. hellmayri 20 Black Hills 0.045 ñ 0.005 0.039 --- 0.017 17.07 1.22 E. d. hellmayri 28 Wet Mtns. 0.063 ñ 0.006 0.066 ñ 0.021 29.27 1.49 E. d. hellmayri 14 SE Arizona 0.064 ñ 0.014 0.065 ___0.021 26.83 1.34 Total 208 Mean 0.054 0.055 21.73 1.31

Breeding specimensonly. Exactlocalities are available from authors. Frequencyof most common allele -< 0.99. Unweighted by samplesize.

an electrophoretic studies. Although beyond during the spring and summer from 1977 through infancy, avian allozyme analysis has not com- 1982.These specimens were segregatedinto 11 breeding populations(Table 1). Liver, muscle,kidney, and pleted adolescence(Barrowclough 1983, Corbin heart tissue was taken from each individual in the 1983, Matson 1984, Barrowcloughet al. 1985). field and placed in liquid nitrogen (-196øC) until Taxonomicrepresentation and geographiccov- transportto Berkeley,California, for permanent stor- erage of investigationsare still spotty. For the age. Tissuefrom 10 individuals of the Acadian Fly- huge family Tyrannidae, only two allozyme catcher(E. virescens) from Oklahomawas analyzed for studieshave appeared(Zink and Johnson1984, outgroup comparisons.Tissue homogenates (a com- Capparella and Lanyon 1985); together these bination of a single tissuetype and an equal volume cover comparatively limited samplesof only 5% of de-ionizedwater) were centrifugedat 4øCand 15,000 of the existing species.Geographic analyses of rpm for 40 min. The aqueousprotein extractswere breeding populationsof single speciesare also storedat -76øC for later electrophoreticanalysis. scarceand often based on few widely spaced, Forty-one presumptive genetic loci were examined by horizontal starch-gel electrophoresisusing stan- regionalsamples. A notableexception is Zink's dard proceduresdescribed by Selander et al. (1971) (1986) study of 31 populationsof the Fox Spar- and Yang and Patton (1981), with the slight modifi- row (Passerellailiaca). Becausesevere difficulties cationsof Johnsonet al. (1984). Protein assayswere accompany collection of even common bird prepared accordingto Harris and Hopkinson (1976) species (Johnson 1982), we can never expect and Selander et al. (1971). Alleles (electromorphs)at many comprehensiveavian studiesof intraspe- eachlocus were designatedalphabetically in decreas- cific genetic variation acrossgeography. ing order of mobility. The most frequent allele at a Finally, good techniques for estimating rel- locuswas designated100; other alleles were assigned ative levels of gene flow are now available.One higher or lower numbers,on a percentagebasis, de- of these, Slatkin's (1981, 1985a) rare-allele meth- pending on their position on the gel relative to the most frequent allele. For example, at the malic en- od, can be applied directly to electrophoretic zyme locus(ME), allele "a" (105%)moved 5% farther results, and we make such an attempt in this from the origin than did the mostcommon allele, "b" paper.Few estimatesof geneflow in birdshave (100%). For multiple isozymes of proteins, the most been offered previously (Barrowclough 1980, anodal locuswas designated"1," and more cathodal Zink 1986, Zink et al. 1987). loci were indicatedby progressivelyhigher numbers. Hemoglobin was the only cathodalprotein detected METHODS AND MATERIALS on LiOH gels. Levelsof heterozygositywere determinedby direct A total of 208 specimensof the Western Flycatcher count (Hob•) and by estimation based on the expec- complex was collected in the western United States tationof Hardy-Weinbergequilibrium (Nei 1975).The January1988] Geneticsof Flycatchers 179 program BIOSYS-1(Swofford and Selander 1981) was to 29.27% (Wet Mountains, Colorado), with a usedto calculateexpected heterozygosity per sample mean of 21.73%. The mean number of alleles (He•p)and heterogeneityChi-square statistics. Allelic per locus ranged from 1.12 in E. d. insulicolaon frequencieswere convertedto genetic distancesusing Santa Catalina Island to 1.49 (E. d. hellrnayriof the methodsof Nei (1978) and Rogers(1972). To com- Wet Mountains, Colorado),with an average of pare patternsof population relatedness,phenograms 1.31. (UPGMA and WPGMA; Sheath and Sokal 1973) and a phylogenetic tree (distanceWagner network [Farris In simple regressions,Hobs. showed no rela- 1972], optimized according to Swofford 1981) were tionship to sample size (r = 0.391), a weak re- constructed from Rogers' distance values. Wright's lationship to percentage of polymorphic loci (1965) Fst,with the modificationsof Wright (1978) for (r = 0.527), and a weak relationship to mean small sample size and of Nei (1975) for multiple al- number of alleles per locus (r = 0.647). As ex- leles, was computed overall for the 11 populations in pected, percentage of polymorphic loci corre- the complexand for various subsetsof populations. lated stronglywith mean number of allelesper To estimatelevels of gene flow we used Slatkin's locus (r = 0.893). No significant relationship (1985a)formula, ln[p(1)] = a ln(Nm) + b, where p(1) was found between samplesize and either per- is the averagefrequency of private alleles, Nm is the product of the population size and immigration rate, centage of polymorphic loci (r = -0.106) or a is a constant, -0.505, and b is a constant, -2.44. mean number of alleles per locus (r = 0.149). This formula is suitable for the analysis of samples Furthermore, in a multiple regressionanalysis, of 25 or greater. Becauseour sample sizes differed with samplesize as the dependent variable and among populations, we applied Slatkin's (1985a) rec- (1) total number of alleles at polymorphic loci, ommendedcorrection in which Nm is divided by the (2) observedheterozygosity, and (3) percentage ratio of the actual average sample size to 25. Thus, of polymorphic loci as independent variables, Nrnc = Nrn(•,/25)%where •, = (• N,/11),the average R2 = 0.0124, an insignificant value (P = 0.744). samplesize. N, is the number of specimensin a given The lack of dependenceof mean levels of H (or sample.For our averagesample size of 18.9, the cor- rection factor is therefore 0.756. of number of alleles) on sample size, when the number of loci surveyed is relatively large, RESULTS agreeswith both the theoretical predictionsof Nei (1978) and the empirical findings of Gor- Variationat lociand heterozygosity.--Ofthe 41 man and Renzi (1979). loci scored, 21 (51%) were variable either within Geographictrends in allelicfrequency.--Five loci or between membersof the E. difficiliscomplex exhibited clear geographic patterns in the fre- and E. virescens(Table 2). Of these variable loci, quency of alleles. Patterns in four of these loci 17 (41.4% of total) showed at least a single het- are illustrated (Fig. 1). At ICD-1 (Fig. 1A) all erozygotein the complex;the remaining 4 loci mainland populations were monomorphic for were monomorphic. The outgroup, E. virescens, the M allele. A unique fast allele (F) occurred was fixed at 2 loci (aGPD and GPT) at alleles only on Santa Catalina Island, where it was that did not occurin any form of the E. difficilis found with fairly high frequency (33.3%). A complex and was polymorphic at 3 other loci slightly more complicated pattern was shown (EST-4, SDH, ADH) at which all members of at the EST-2locus (Fig. lB) at which two alleles the complex were fixed. The 24 monomorphic (F, M) were widespreadand showedweak clines lociin the E. difficiliscomplex were: ACON, ALD, in their geographicfrequency. The F allele was ICD-2, EST-1, EST-4, CK-3, SOD-l, SOD-2, PT- commonin the interior of the southeasternpor- 1, PT-2, PT-3, HG, MDH-1, MDH-2, GOT-1, GOT- tion of the region but declined in frequency 2, GLUD, GDA, GPT, LAP, LDH-2, ACP, ADH, toward the north and northwest. Two other, and SDH. Three loci, GLO, GAPDH, and rare alleles(F+, S) occurredonly in the interior, G-6-PDH, were unscorable. at low frequencies(2.5%), in the populationsof Levels of genetic variation among popula- the Snake Range and Black Hills, respectively. tionsof the E. difficiliscomplex are listedin Table Five alleles were recorded at 6-PGD (Fig. 1C). 1. Hobsranged from 0.037 in E. d. insulicolato Again, only two of these turned up in coastal 0.066 in the sympatricSiskiyou sample,which populations, one of which (M) was sharedwith is comprisedabout equally of E. d. hellmayriand all other populations;the other (S) occurredonly E. d. difficilis.The averageH over all populations at low levels in the northern three coastalpop- was 0.054. The percentageof polymorphic loci ulations. The Monterey and Santa Catalina Is- ranged from 9.76% (on Santa Catalina Island) land populations were monomorphic for M. In 180 JOHNSON^ND M^•TEN [Auk,Vol. 105

ao •'.,,i

c5c5 o

ao •",,I

o c• o',,, c5 • o January1988] GeneticsofFlycatchers 181

o. o

o. o. o

o. o 182 JOHNSONAND MARTEN [Auk, Vol. 105

Fig.1. Geographicoccurrence and percentage frequency of allelesat fourloci in theEmpidonax difficilis complex.The 11 samples represent three taxa: E. d. insulicola (1population), E.d. difficilis (4), and E. d. ,•½llmayri (6).The key to allelespresent is locatedin thelower center of thequadrant for each locus. Sample sizes are given within the symbolfor eachpopulation in quadrantA. the interior, a fairly common(5.9-21.4%) allele fixation in the interior samplesbut is disap- (F) was found that was missingalong the west pearing in the coastalsamples. Finally, at the coast.Furthermore, two scarce(2.9-4.3%) alleles glutathione reductase (GR) locus (not illustrat- (F+, F-) were sharedby two populationseach ed), a clear clinal pattern of allelic frequencies of the interior group.The moststriking pattern occurred(Table 2). Allele b was very common occurredat the malic enzyme(ME) locus(Fig. (94.4%) and allele c was scarce(5.6%) on Santa 1D). There, coastal and interior populations CatalinaIsland. On the mainland,allele b grad- contrasteddramatically in the proportionsof M ually diminished as allele c increasedin fre- and S- alleles.In E. d. difficilisthe M allelewas quency from along the coast northward, and found at levels of zero (Monterey) to 17.6% into the interior, until allele c was more com- (Shasta).In E. d. insulicolathis allele occurred at mon than allele b. The populationfrom the Black 18.5%.In contrast,in all populationsof "pure" Hills was slightly deviantfrom this trend; there, E. d. hellmayri(those populations exclusive of b was at 62.5% and c at 37.5%. the mixedsample from Siskiyou),the M allele A heterogeneityChi-square analysis at each was very common, occurring at 73.9% (Warner) locus(Swofford and Selander 1981) demonstrat- to 85.7%(SE Arizona). In the Siskiyoupopula- ed significantgeographic variation among pop- tion the M allele occurred at 50%. At the ME ulations. locus,the M alleleapparently is headingtoward Geneticdistances.--Nei's genetic distances January1988] GeneticsofFlycatchers 183

T^I•I•E3. Matrix of geneticdistances between 11 samplesof 3 taxaof the Empidonaxdifficilis complex and E. virescens.Nei's (1978)D-values are abovethe diagonal,and Rogers'(1972) D-values are below the diagonal.

1 2 3 4 5 6 7 8 9 10 11 12

1. Santa Catalina Isl. -- 0.007 0.009 0.011 0.011 0.015 0.017 0.028 0.013 0.034 0.021 0.110 2. Monterey 0.032 -- 0.0 0.001 0.001 0.006 0.012 0.018 0.012 0.027 0.018 0.099 3. Lake Co. 0.037 0.016 -- 0.0 0.0 0.003 0.009 0.014 0.011 0.021 0.015 0.097 4. Shasta 0.037 0.022 0.018 -- 0.0 0.002 0.008 0.012 0.010 0.022 0.015 0.097 5. Rogue River 0.038 0.021 0.017 0.019 -- 0.002 0.009 0.012 0.012 0.019 0.014 0.092 6. Siskiyou 0.045 0.030 0.022 0.025 0.021 -- 0.001 0.003 0.004 0.009 0.005 0.085 7. Warner 0.049 0.036 0.027 0.030 0.031 0.019 -- 0.002 0.0 0.006 0.001 0.084 8. Snake 0.057 0.041 0.036 0.037 0.033 0.025 0.020 -- 0.006 0.002 0.003 0.078 9. Black Hills 0.039 0.033 0.030 0.029 0.034 0.025 0.016 0.025 -- 0.010 0.002 0.089 10. Wet Mtns. 0.069 0.056 0.050 0.057 0.049 0.042 0.031 0.023 0.038 -- 0.001 0.074 11. SE Arizona 0.057 0.047 0.043 0.049 0.040 0.032 0.024 0.028 0.027 0.020 -- 0.077 12. E. virescens 0.142 0.134 0.133 0.136 0.129 0.121 0.119 0.112 0.122 0.111 0.118 --

(Table 3) betweenpairs of samplesrepresenting allelesamong populations,whereas a value of the samesubspecies were low:/• = 0.0003in E. 0 indicatestotal panmixis.In 11 populationsof d. difficilis(excluding Rogue River, a samplethat the WesternFlycatcher complex, the averageFst is near the boundarywith E. d. hellmayri),and acrossloci was 0.153, indicating definite genetic /• = 0.0033in E.d. hellmayri (excluding the mixed subdivisionamong populations.Loci that con- Siskiyousample). The lesservalue for E. d. dif- tributed heavily to Fstvalues were ME (total ficilisvs. E. d. hellmayrisuggests that the geo- Fst= 0.324), GR (Fst = 0.173), and ICD-1 (Fst = graphically more far-flung populations of the 0.313). Among the loci surveyed these three, latter are more strongly differentiated geneti- therefore,have significantlydifferentiated geo- cally. graphically. Computation of F•t for different Empidonaxd. insulicolais clearly allied to E. d. subsetsof populationsis alsoinstructive, for in difficilis(/• = 0.009for SantaCatalina Island vs. this manner geographicunevenness in amount Shasta,Lake County, and Monterey)rather than of differentiation can be identified. For exam- to E. d. hellmayri(/• = 0.0228for SantaCatalina ple, F•t for E. d. insulicolaplus E. d. difficiliswas Island vs. Warner, Snake, SE Arizona, Wet 0.079. But when E. d. insulicola was excluded and Mountains, and Black Hills). The value 0.009 is the Fstcalculated for the four populationsof E. similar to those commonly seen between sub- d. difficilisalone, a value of 0.026 resulted.Sim- species;the value 0.0228 is a level of D com- ilarly, for the six populationsof E. d. hellmayri, monly seen between avian species (Barrow- F•tequaled 0.069. Removalof Siskiyoufrom the clough 1980: 661). latter analysis of E. d. hellmayrigave an Fstof Comparisonof all populationsof E. d. difficilis 0.062. From these analyseswe conclude that E. (exceptRogue River) with all populationsof E. d. difficilisis more homogeneousgenetically than d.hellmayri (except Siskiyou) gave a/• of 0.0149 is E. d. hellmayri. for the 15 values. This value is intermediate Slatkin (1981) introduced a method for qual- between those given by Barrowclough(1980: itatively estimating gene flow in subdivided 661) for subspecies(0.0048) and species(0.0440) populations. His approach was based on the of other birds. On the basisof genetic distance averagefrequency of alleles in relation to the alone, E. d. difficilisand E. d. hellmayriqualify as numberof populationsin which they occur(the megasubspecies(Amadon and Short 1976), a "conditionalaverage frequency"). Figure 2 pre- conclusion reached earlier on the basis of mor- sents the results of such an analysis for the phology and song (Johnson1980: 111-113). Western Flycatcher complex. The shape of the Analysis of genetic population structure.-- curve indicatesthat gene flow is relatively high Wright's (1951)F•t statistic measures the degree among the populationsanalyzed. That is, in of geneticfragmentation of populationsshow- general,alleles found only in one or a few pop- ing varying degreesof interbreeding(Barrow- ulations occur with very low frequencies;al- clough 1980:657, Corbin 1983).At a given locus, leleswith high frequenciesoccur in all or near- an F• value of 1 points to fixation of alternative ly all populations. 184 JOHNSONAND MARTEN [Auk, Vol. 105

occurred at all, from 0.022 (Wet Mountains) to 0.050 (RogueRiver). Two populations (Siskiyou and Warner) lacked private alleles. In contrast, the single private allele in the samplefrom San- ta Catalina Island, ICD-1, allele a, occurred at a substantialfrequency, 0.333. These allelic frequency values translateinto moderateor high

•(i) estimatesof Nmc for continental populations (3.97-20.22) but into an extremely low Nmc, 0.093, for Santa Catalina Island (Table 4, left column). As Slatkin (1985a) suggested,additional information on population structure can be ob- tained by computingp(1) for different subsam- pies of the overall data set. Among other possibilities,such an approachcan identify the influence of a single, isolated, or otherwise de- •ig. 2. Relationshipbetween the conditional av- viant population on the estimate of Nmc. The eragefrequency of a protein variant, p(i} (the average right column of Table 4 shows the estimated frequencyof a protein variant in the populationsin valuesof Nmcin 10 of the 11 total samplesafter which it occurs), and its incidence (the number of successiveremoval of different single samples. populations in which the variant is observed [i] di- Considering all locations,Nmc = 4.14, a value vided by the total number of populations sampled similar to that seen when any continental sam- [d]). The plotted points represent the averagevalue ple was excluded (Nmc= 2.86-4.69). When the of p(i) for all variants having the same value of i/d. This curveis roughly congruentwith an island model isolatedsample from SantaCatalina Island was simulation when Nm = 0.25 (Slatkin 1981). Sample removed, however, the estimate of Nmc = 9.62, calculationsof the coordinatesfor a single data point a value 2-3 times greater. Again, this result (e.g. the third data point from the left) are asfollows: points to a substantialprobable difference in Allele e of the 6-PGD locus occursin three popula- the magnitude of gene flow, as estimated by tions (at a frequency of 0.067 in Lake County, 0.025 Nmc,among continental populations vs. that be- in Shasta,and 0.050 in Rogue River; Table 2). The tween the mainland and insular populations. other allele occurring only in three populations, al- Inheritanceof electromorphs.--Sevenpartial or lele a of the PGM locus,was found at a frequencyof complete family groups totaling 32 specimens 0.022 in Warner, 0.025 in Snake, and 0.015 in the Wet were collected. These ranged in composition Mountains.Thus, these two allelesoccur at an average frequencyof 0.048and 0.022,respectively. Therefore, from one family consisting of 1 parent and 3 p(3) = (0.048 + 0.022)/2 = 0.035, and i/d = 3/11 = nestlings to an example of both parents and 4 0.273. nestlings.In everyexample the parentsand their offspring had electromorphicpatterns at all polymorphic loci that were expectedaccording Recently, Slatkin (1985a) refined this tech- to a systemof Mendelian inheritance. nique to enablequantitative estimation of gene Phenogramsand phylogenetictrees.--To allow flow using the average frequency of alleles visual assessmentof relationships of popula- found only in single populations(private al- tions,three schemeswere applied. UPGMA and leles). Slatkin's new method derives from his WPGMA clusteringalgorithms (Sneath and So- discoverythat the logarithmof the averagefre- kal 1973; Fig. 3) are phenetic approachesthat quencyof private alleles,p(1), is approximately assumeconstant rates of allozymic change in linearly related to the logarithm of the average accordancewith a molecular-clock hypothesis number of migrantsexchanged between local (Wilson et al. 1977, Thorpe 1982). The third dis- populations,Nm. The methodassumes that the tanceapproach, a Wagner network (Fig. 4), pro- populationsare in genetic and demographic vided a phylogenetic perspective;it does not equilibrium, which is usually uncertain.The assumehomogeneity of rates of allelic substi- averagefrequency of privatealleles (Table 4) in tution. The three approachesin essenceoffer the Western Flycatchercomplex was very low different hypotheseson patterns of relation- in all continental populations in which they ship, patternsdeveloped from a somewhatdif- January1988] GeneticsofFlycatchers 185

TABLE4. Number and averagefrequency of private alleles and estimatesof gene flow in populationsof the Etnpidonaxdifficilis complex. See Table 1 for samplesize of each population.Average samplesize was 18.9.

Combined populations Single populations (one population excludeda) No. private No. private Sample location alleles p(1) Nmcb alleles p(1) Nmc Santa Catalina Isl. 1 0.333 0.093 17 0.032 9.62 Monterey 1 0.031 10.25 18 0.049 4.14 Lake Co. 1 0.033 9.05 18 0.048 4.31 Shasta 2 0.028 12.54 20 0.048 4.31 Rogue River 4 0.050 3.97 15 0.048 4.31 Siskiyou 0 0 0 20 0.047 4.50 Warner 0 0 0 21 c 0.046 4.69 Snake 1 0.025 15.69 18 0.050 3.98 Black Hills 1 0.025 15.69 18 0.049 4.14 Wet Mtns. 5 0.022 20.22 16 0.059 2.86 SE Arizona 2 0.036 7.62 18 0.050 3.98 All samples 18 0.049 4.14 • For example, 17 private alleles occurat an averagefrequency of 0.032in the 10 populationsexcluding Santa Catalina. bNm• = Nm correctedfor sample size. • More privatealleles can occur in a subsample(e.g. 21 in all populationsexclusive of Warner)than in all samplescombined (18) because alleles found in two differentpopulations will be privatealleles in somesubsamples.

ferent mathematicaltreatment of Rogers'D. We (a(b,c)),each produced by a differentbranching assumedthat topologicsimilarity among trees algorithm, is "resolved"by the consensusmeth- servedas evidencefor true relationshipsof the od, the resulting tree, (a,b,c),is an unresolved populationsinvolved. Conversely,when dif- trichotomy. In this exercise we have lost two fering clusteringapproaches yielded ambigu- potentiallyvalid hypotheses,one of which may ous or conflicting patterns of branchwork, we be true, and gainedthe unlikely hypothesisthat judged relationshipsamong the populationsto all three taxa sprangsimultaneously from the be unresolvedby the genetic distancedata. same ancestor. For all of the above reas,•s we We are aware that some current workers ad- did not apply the consensusapproach in our vocate the general use of "consensustrees" to analysisof branching patterns. resolveconflicting branching topologies. How- The resultsof the phenetic and phylozenetic ever appropriate such trees may be for the rec- analyseswere consistentin all important re- onciliationof a seriesof equally parsimonious spects.Resolution was especially clear at the but different trees,when each is derivedby the level of fine branch tips. For example, the samemethod (see Swofford 1985 for examples), UPGMA and WPGMA phenograms each re- the construction of a consensus tree is unwar- vealed four clusters,two of interior samples, ranted when the trees on which it is based were one of coastalsamples, and one of the single derived through different methodologies. If insularsample. The coastalseries, Shasta, Rogue eachapproach (e.g. UPGMA vs.Wagner) is based River, Lake County, and Monterey, formed one on different assumptions,then the trees that tight clade; the interior seriesof populations, result cannotbe blended logically without de- Warner,Black Hills, Siskiyou,and Snake, formed stroying the integrity of either set of underly- another. In the UPGMA a third clade, that link- ing assumptions.Furthermore, significant in- ing SE Arizona with Wet Mountains, formed a formation on branch lengths can be lost in sister group with the previous eight popula- consensustrees because they identify only the tions. In the WPGMA, SE Arizona and Wet positionof nodesand not the distancesbetween Mountains joined as a sistergroup to the clade nodes.Finally, we feel that one or more plau- consistingof the remaining four interior pop- siblehypotheses of relationshipthat areby def- ulations. The fourth "cluster," Santa Catalina inition uncertainare better than either no hy- Island, linked with the four geographically pothesisor an illogicalhypothesis. For example, proximal coastalpopulations in the WPGMA. if the conflictin the relationshipof three taxa In the UPGMA, in contrast,the island sample as expressedby two different trees, ((a,b)c) and formed a sistergroup to all the remaining 10 186 JOHNSONAND MARTEN [Auk, Vol. 105

Acadian Flycatcher

' oL,•' o',2' o',o '0'08' 0'06' o'o.' o'o2' o'oo

Block Hdls

Snake Shasta

Lake CO,Ca,if FJ•. •. Optimized distanceWa•e• •etwo•k •oot- UPGMA• " WetSanta Mtn$Catalina lsled at •he ou•oup, E•pi•on• •Jrescens.This a•aJ•sJs p•oduced•o •e•a•Jve b•a•ches. ]• addJtJo•a[dis- ranceWa•e• a•aJyses,•umblJn• the o•de• TaxonomJcDistance (based on DR) datawere e•e•ed p•oducedt•ees esse•Qa[[ 7 identical Fig. 3. Phenogramsbased on Rogers'D-values and •o the o•e showm derived by the WPGMA and UPGMA methods.The high copheneticcorrelation coefficients(rcc) indicate Siskiyoupopulation was assumedto represent excellentagreement between the distancesshown in either a hybrid swarmor a situationof second- each phenogramand the original data matrix. ary sympatrywith either limited or no hybridization (Johnson1980). Reinterpretation of the populations of the complex. The shortnessof earlier sampleand analysisof 17 new specimens the branch involved in this linkage, however, takenon their breedingterritories in 1981clear- indicates that this implied relationship is not ly supportedthe secondalternative. The new definitive. sampleconsisted entirely of pure parentaltypes: Although the distanceWagner network re- 8 certainor probableE. d. hellmayriand 9 defi- vealed somewhat less clumping, the funda- nite or probableE. d. difficilis.The discriminant mental pattern expressedwas similar to that of scoresof adult specimensfrom both the old and the two phenetic analyses.In the Wagner re- new samplesare plotted in Fig. 5. Two speci- suits, the series of populationsbegan with a mens from the older sample, a male at 0.5146 distantly located interior sample, Wet Moun- and a female at 2.739,had low scoresfor typical tains, and ended with the most remote western E. d. difficilis(half-shaded squares in Fig. 5). Al- sample, Santa Catalina Island. Between these though identified as E. d. difficilisby the dis- two extremes,the five interior samplesand, in criminant analysis,because of their apparently turn, the four coastal samples, split off in or- intermediate scoresthe possibility cannot be derly progression in a sequence that rather ruled out that they representhybrids. Another closely paralleled their geographic arrange- individual, from the recent sample and hence ment. Appropriately,the mixed Siskiyousam- with geneticdata (NKJ No. 4616),although pos- ple was placedbetween the interior seriesand sessinga discriminantscore of 0.7025(typical the coastalgroup of populations.Finally, we of male E. d. difficilis),had an "interior" geno- note that the great length of the branch leading type (FM) at the 6-PGD locus.The F allele at from the E. d. difficilisclade to the singlesample this locus is unknown from other examplesof of E. d. insulicola underscores the basic distinc- coastalE. d. difficilis.This bird may have been tivehess of the latter taxon. of recentbackcross origin. Alternatively, the F Sympatryin the Siskiyouregion.--Coastal and allele at 6-PGD may occurin the Siskiyoupop- interior representativesof the complexcontact ulation at low frequency even in typical E. d. and interact in the Siskiyou region of north- difficilisbecause of infrequent, ancient inter- central California (Johnson 1980: 87-94). Be- breedingwith E. d. hellmayri. causeof the enormouslyexpanded range of dis- In 1981three matedpairs were taken, two of cr[minant function scoresshown by a sample E. d. difficilisand one of E. d. hellmayri.A single from that region taken in 1964 and 1968, the pair, both matesrepresenting E. d. difficilis,had January1988] GeneticsofFlycatchers 187 been reported earlier from Siskiyou (Johnson 1980).Thus, the four knownbreeding pairs were each mated in a positively assortativemanner. Except for the male mentioned above, individuals with either interior genotypesor vocal- Shasta izations, or both, were consistentlylarge and representedE. d. hellmayri.The discriminant analysisshowed clearly the sympatriccharacter of the Siskiyou population. Sympatry of significant populations, with little or no inter- breedingand assortativemating of pure parental types, arguesfor the biologic speciesstatus of E. d. difficilisand E. d. hellmayri. Fig. 5. Discriminantscores of adult specimensof DISCUSSION WesternFlycatchers, with Shastaserving as a refer- ence standard for E. d. difficilisand Warner repre- These genetic results add qualitatively dif- sentingthe standardfor E. d. hellmayri.Birds from ferent information to that already publishedon Siskiyourepresent unknowns, as described by John- geographicvariation in the Empidonaxdifficilis son(1980: 87-92). Empidonaxd. difficilisand E. d. hell- mayriare clearly sympatricin the Siskiyou region. complex (Johnson1980). Such analysesof pro- The two half-shadedsymbols represent either hy- tein-coding loci elucidate patterns of genetic brids or extreme values for either form. The scores isolation,gene flow, and dispersalwith respect are basedon lengthsof primaries10, 7, and 4; length to environmental barriers. In turn, these pat- of tail; length, depth, and width of bill; length of terns permit inferences on important historical tarsusplus middle toe;and body mass.Because color events such as founding (colonization),bottle- measurementswere unavailable for the 1981 sample necks, and population fluctuations(Lewontin from Siskiyou,color characterswere excluded from 1974, Nei et al. 1975, Barton and Charlesworth the presentanalysis. Thus, the discriminantscores of 1984, Brussard 1984, Carson and Templeton referencespecimens from Warner and Shastadiffer from those presentedby Johnson(1980), although 1984). Knowledge of patterns of population they are computedfrom the samespecimens. Note structure,considered in concertwith geograph- that the scale for males is in tenths whereas that for ic variation in morphology,behavior, and hab- females is in whole numbers. itat selection,can illuminate the processesre- sponsiblefor speciationin particulargroups and can provide evidence for rational decisionson ulation of a specieshas passedthrough a bot- their taxonomic status. Finally, because these tleneck or existscurrently at low densities(Nei allozyme data were obtainedfrom moderately et al. 1975). Becausethe speciesis common on large breeding samples,they contribute gen- Santa Catalina Island, we assume that the di- erally to our asyet superficialunderstanding of minished geneticvariability cannot be related genetic structuring of natural populationsof to low presentdensity. Instead, the low genetic birds (Barrowclough1980, 1983). variability of the Insular Flycatcher,compared Levelsof genetic variation.--Although levels of with mostcontinental populations, probably re- H ranged from 3.7 to 6.6% and are therefore suitedfrom a bottleneckat the time the original typicalof otherbirds (/2 = 5.3%;Barrowclough group of founders colonized the California 1983), it is noteworthy that the two lowest val- Channel Islands. The population of California ues of observedheterozygosity were from iso- Quail (Callipeplacalifornica catalinensis) on Santa lated populations, Santa Catalina Island (H = Catalina Island also has the lowest value for 3.7%) and Black Hills (H = 4.5%). The value for observedheterozygosity when comparedwith the insular sampleis especiallyinteresting be- six mainland populations(Zink et al. 1987).Low causeon the basisof another important measure observedheterozygosities have been reported of geneticvariation, percentageof polymorphic previously for other insular forms (Selander loci, the Santa Catalina Island birds, at 9.8%, also 1976). fell far below the range (17.1-29.3%)and mean Genetic population structure.--As Barrow- (21.7%)of the other 10 samplesof the complex. clough (1983: 231) described,"If most of the Low values of H are expectedif the total pop- genic variability present in natural populations 188 JOHNSONAND MARTEN [Auk, Vol. 105 is neutral or near-neutral ... it could then be in his simulations,"patterns in both rare alleles usedas an indicatorof populationstructure and and Fs•when there is no gene flow are too sen- relative divergence times." An explicit test of sitive to weak selection to be of use in under- the infinite allele-constant mutation rate mod- standingthe history of populations."We thus el, using the mostthorough studiesavailable of assumethat the geographicdistribution of rare genic variation within avian populations(Bar- alleles servesbest to illuminate current, not past, rowclough et al. 1985), found good agreement levels of gene flow. This is not to say that these with the neutral theory. Becauseone of the pop- estimates of present, "average" rates of gene ulations examined in that study was from the flow cannot be misleading. As Slatkin (1985b) Empidonaxdifficilis complex, we feel justified in emphasized,because of changingdemographic using Fs•statistics (Wright 1951,1978) as a mea- or environmental conditions,gene flow may be sure of relative population fragmentation. unpredictable and episodic, resulting irregu- The averageFs• of 0.153 over the entire com- larly in levelsfar higher than are averagefor a plex indicates that 15.3% of the total genetic given species.Therefore, even for E. d. insulicola, variance was distributed among populations, a which at present seemsto be characterizedby clear departure from panmixia. Computation of very low gene flow, we should not rule out F• for subsetsof the 11 total populations was infrequent periods of higher gene flow, per- alsorevealing. For example,although Fs• = 0.079 haps resulting from a dramatic flush in main- among all populationsof E. d. difficilisplus E. d. land populations.Episodic gene flow in the re- insulicola, when the latter was removed Fst verse direction could also occur. dropped to 0.026. This suggestssubstantial in- Broadcongruence of geneticand otherpatterns terchangeof birds among the coastalmainland acrossgeography.--Gene flow is assumedto in- populations but much less dispersal between fluence all loci in the same manner. In contrast, the continent and Santa Catalina Island. Fst = because both selection and mutation affect each 0.069 for the six populationsof E. d. hellmayri, genetic locus and each phenotypic character a value almost three times larger than in E. d. separately,loci can vary greatly in their degree difficilis.This reflectsthe known greater geo- of geographicvariation (Slatkin 1985b). Such graphic interruption of interior vs. coastal variability in genetic differentiation is shown mainland populations. clearly by allozymesin the E. difficiliscomplex Geneflow.--Although applicationof Slatkin's (Table 2, Fig. 1). The geographic patterns of (1981, 1985a)technique to the allozyme datafor allelic frequencyat severalloci are not chaotic; all populationssuggests moderately high gene rather, they illustrate differentiation in insular flow generally, the analysesof subsetsof pop- vs. mainland populations (at ICD-1), coastalvs. ulations showed that it is far from uniform geo- interior populations(EST-2, 6-PGD, ME), or, in graphically. In particular, an estimated0.093 one instance (GR), a north-south cline. Fur- immigrants/generation enter the Santa Catali- thermore, definite geographic concordanceis na Island population. In contrast, mainland seenbetween the geneticpatterns and features populationsreceive an estimatedaverage of 9.62 of morphology,voice, coloration,and habitat immigrants/generation.The island value is ex- describedpreviously (Johnson 1980). Such in- ceptionallylow, whereasthe mainlandaverage tegrationof diversecharacter suites into three falls at the upper end of estimatesof Nm (range, regionalgroups of populations,representing the 1.8-9.5) for the few other speciesof birds that California Channel Islands, the Pacific coast, and have been examined (Zink and Reinsen 1986: the interior of the western United States, re- 28). We conclude that the miniscule value of spectively,strongly supports the view that these Nmc for the Santa Catalina Island sample pro- groups comprisedistinctive but cohesiveevo- vides strong evidence for its essential genetic lutionary units. independence.We hope these indirect esti- Phylogenesisand historical biogeography.--In ad- matesof gene flow will be tested directly in the dition to portraying the genetic relationships future through recapturestudies of marked in- of populations, the distance Wagner network dividuals. (Fig. 4) can serve as a hypothesisof phylogen- Larsonet al. (1984),working with salamander esisin western North America. Earlier, Johnson data, expressedconcern that Slatkin's method (1980:120) speculated on the origin, spread,and cannotdistinguish present from pastgene flow. diversificationof the E. difficiliscomplex as an Slatkin (1985b:424) respondedby arguing that, exampleof a general model for speciationin January1988] GeneticsofFlycatchers 189 the genus. That model envisioned the stepwise this complexof the genusEmpidonax diversified budding of pioneer populations from the pe- from the middle to late Pleistocene. riphery of the breeding rangesof existing taxa. Taxonomicstatus of populations.--Ifsignificant These pioneers would crossbarriers, colonize, populationsof two forms can maintain their expand their distributions, speciate,and repeat integrity in local sympatry,they are reproduc- the process.For the E. difficiliscomplex specifi- tively isolatedand typically are regardedas bi- cally, Johnsonproposed that the lineage arose ologic species.Empidonax d. diffficilisand E. d. somewhere within the range of modern E. d. hellmayrisatisfy this criterion in the Siskiyou occidentalisin central or southern Mexico. Very region of northern California. Although the early, the ancestorsof E. fiavescenssplit off in possibility of infrequent hybridization cannot Central America. Sometime later, the lineage be excluded,most individuals in the region of leading to modern E. d. hellmayrispread from overlap clearly represent pure parental types Mexico northward into the southern Rocky and, insofar as is known, mating is positively Mountains and Great Basin. Eventually, pi- assortative.Thus, we do not hesitate to regard oneersfrom the stockthat becamepresent-day E. d. difficilisand E. "d." hellmayrias full species E. d. hellmayricrossed the CascadeMountains and formally recommendthat they be so con- and evolved into E. d. diffficilisalong the Pacific sidered by the Committee on Classificationand coast.After the lineage that developed into E. Nomenclature of the American Ornithologists' d. diffficilisexpanded into southern California, Union. If this proposal is accepted,we would colonistsspread to the California Channel Is- recommend further that the vernacular names lands.These individuals eventually evolved into "CoastalFlycatcher" and "Interior Flycatcher" E. d. insulicola.Thus, speciationby the founder would be appropriatefor these taxa. effect (Mayr 1942: 237, Bush 1975) accounted Although originally describedas a species, for the present diversity of the complex(John- the taxonomic status of E. d. insulicola is less son 1980). easily•determined. This taxon breeds only on The genetic distancedata completely support five of the California Channel Islands and has the above scenario.Furthermore, by using Mar- never reinvaded the adjacent mainland coast, ten and Johnson's (1986) modification of the as have two other island endemics, to permit a calibration of Gutierrez et al. (1983), we can naturaltest of sympatrywith E.d. diffficilis. None- offer estimated datesfor three important clado- theless, insular and coastal birds differ substan- geneticevents. We apply the formula t = 19.7 x tially in morphology, coloration, voice, and 106D,where t is the time since divergence and habitat preference (Johnson1980). The genetic D is Nei's (1978) genetic distance.This exercise differences and extremely reduced gene ex- assumes a molecular clock (Wilson et al. 1977, change described here further argue for the Thorpe 1982), in which allelic substitutions evolutionary independence of these forms. De- among populations accumulate more or less spite the fact that its biologic speciesstatus can- steadilyover time, and neutrality of patternsof not be proved in the absenceof a test of sym- geneticdivergence (Kimura 1979,1982; Barrow- patry, we concludethat E. d. insulicolais at least clough et al. 1985). Becauseof problems asso- a "basal evolutionary unit" or phylogenetic ciated with the calibration figure of 19.7, we speciesin the senseof Cracraft (1983). cautionthat divergencetimes calculatedby this formula are grossapproximations at best (Mar- ACKNOWLEDGMENTS ten and Johnson 1986). Basedon a Nei's D of 0.044 (Zink and Johnson We arevery gratefulto the late JohnDavis, Pamela 1984), the lineage that eventually led to pres- Williams, D. Scott Wood, and Robert M. Zink for as- ent-day E. d. difficilisdiverged from that which sistancein obtaining specimens.A. Douglas?ropst becamemodern E. f. fiavescensat 866,800yr BP and his rangersof the SantaCatalina Island Conser- (yearsbefore present). At 248,700 yr BP (based vancy kindly arranged for our fieldwork on Santa Catalina Island. Monica M. Frelow and Carla Cicero on D = 0.0126), the ancestorsof modern E. d. willingly provided expert advice on computeranal- diffficilissplit from the stock that becameE. d. yses.George F. Barrowclough,Scott M. Lanyon,and hellmayri.In turn, the lineage leading to pres- Montgomery Slatkin offered important suggestions ent-day E. d. insulicoladivided from that which for improving an earlier version of the manuscript. eventually became modern E. d. difficilisat Gene M. Christman and Karen Klitz drafted the final 187,150yr BP (basedon D = 0.0095).Therefore, versionsof the figures. This researchwas supported 190 JOaNSONAND MARTEN [Auk,Vol. 105 by the National ScienceFoundation (grants DEB-79- lution (F. J. Ayala, Ed.). Sunderland, Massachu- 20694 and DEB-79-09807to N.K.J.) and in part by setts, Sinauer Assoc. grants from the Annie M. Alexander Fund of the FARRIS,J.S. 1972. Estimatingphylogenetic trees from Museum of VertebrateZoology and from the Com- distance matrices. Am. Nat. 106: 645-668. mittee on Research,University of California,Berke- GORMAN,G. C., & J. RENZIJR. 1979. Genetic distance ley. and heterozygosityestimates in electrophoretic studies:effects of samplesize. Copeia 1979:242- 249.

LITERATURE CITED GUTIgRREZ,R. J., R. M. ZINK, & S. Y. YANG. 1983. Genic variationßsystematic, and biogeographic AMADON, D., & L. L. SHORT. 1976. Treatment of sub- relationshipsof somegalliform birds. Auk 100: speciesapproaching species status. Syst. Zool. 25: 33-47. 161-167. HARRISßH., & D. A. HOPKINSON. 1976. Handbook of AMERICAN ORNITHOLOGISTS' UNION. 1983. Check-list enzymeelectrophoresis in human genetics.Am- of North American birds, 6th ed. Washington, sterdam, North Holland Publ. Co. D.C., Am. Ornithol. Union. JOHNSON,N. K. 1980. Character variation and evo- BARROWCLOUGH,G.F. 1980. Geneticand phenotypic lution of siblingspecies in the Empidonaxdifficilis- differentiationin a wood warbler (genus Den- fiavescenscomplex (Aves: Tyrannidae). Univ. Cal- droica)hybrid zone. Auk 97: 655-668. ifornia Publ. Zool. 112: 1-151. 1983. Biochemical studies of microevolu- 1982. Retain subspecies--atleast for the time tionary processes.Pp. 223-261 in Perspectivesin being. Auk 99: 605-606. ornithology (A. H. Brush and G. A. Clark Jr., ßR. M. ZINK, G. F. BARROWCLOUGH,& J. A. MAR- Eds.). Cambridge, England, Cambridge Univ. TEN. 1984. Suggestedtechniques for modern Press. avian systematics.Wilson Bull. 96: 543-560. ßN. K. JOHNSON,& R. M. ZINK. 1985. On the KIMURA,M. 1979. The neutral theory of molecular nature of genic variation in birds. Pp. 135-154 in evolution. Sci. Am. 241: 98-126. Current ornithologyßvol. 2 (R. F. Johnston,Ed.). 1982. The neutral theory as a basisfor un- New Yorkß Plenum Press. derstandingthe mechanismof evolution and BARTON, N.H., & g. CHARLESWORTH. 1984. Genetic variation at the molecular level. Pp. 3-56 in Mo- revolutions,founder effectsßand speciation.Annu. lecular evolution, protein polymorphismand the Rev. Ecol. Syst. 15: 133-164. neutral theory (M. Kimura, Ed.). Tokyo, Japan BRUSSARD,]•. F. 1984. Geographicpatterns and en- Scientific Societies Press. vironmental gradients: the central-marginal LARSON,A., D. g. WAKE, & K. P. YANEV. 1984. Mea- model in Drosophilarevisited. Annu. Rev. Ecol. suringgene flow amongpopulations having high Syst. 15: 25-64. levelsof geneticfragmentation. Genetics 106: 293- Bt•SH,G.L. 1975. Modesof animal speciation.Annu. 308. Rev. Ecol. Syst. 6: 339-364. LEWIN,R. 1985. Moleculesvs. morphology:of mice CAPPARELLA,A. P., & S. M. LANYON. 1985. Biochem- and men. Science 229: 743-745. ical and morphometricanalyses of the sympatric, LEWONTIN,R. C. 1974. The genetic basisof evolu- neotropical,sibling species,Mionectes macconnelli tionary change.New York, ColumbiaUniv. Press. and M. oleagineus.Pp. 347-355 in Neotropical or- MARTENßJ. A., & N. K. JOHNSON. 1986. Genetic re- nithology(P. A. Buckley,M. S. Foster,E. S. Mor- lationshipsof North American carduelinefinch- ton, R. S. Ridgely, and F. G. Buckley, Eds.). Or- es. Condor 88: 409-420. nithol. Monogr. 36. MATSON,R.H. 1984. Applications of electrophoretic CARSON, H. L., & A. g. TEMPLETON. 1984. Genetic data in avian systematics.Auk 101:717-729. revolutions in relation to speciationphenomena: MAYR,E. 1942. Systematicsand the origin of species. the founding of new populations. Annu. Rev. New York, Columbia Univ. Press. Ecol. Syst. 15: 97-131. Nr•, M. 1975. Molecular population genetics and CORBINßK. W. 1983. Genetic structure and avian evolution. Amsterdamß North Holland Publ. Co. systematics.Pp. 211-244 in Current ornithology, 1978. Estimationof averageheterozygosity vol. 1 (R. F. JohnstonßEd.). New York, Plenum and genetic distancefrom a small number of in- Press. dividuals. Genetics 89: 583-590. CRACRAFT,J. 1983. Speciesconcepts and speciation --, T. MARUYAMA, & R. CHAKRABORTY. 1975. The analysis.Pp. 159-187 in Current ornithology,vol. bottleneckeffect and genetic variability in pop- 1 (R. F. JohnstonßEd.). New Yorkß Plenum Press. ulations. Evolution 29: 1-10. DOBZHANSKY,T. 1951. Genetics and the origin of ROGERS,J. S. 1972. Measuresof genetic similarity species,3rd ed. New York, ColumbiaUniv. Press. and geneticdistance. Univ. TexasStud. Genet. 7: 1978. Organismicand molecular aspectsof 145-153. speciesformation. Pp. 95-105 in Molecular evo- SELANDER, g. K. 1976. Genic variation in natural January1988] GeneticsofFlycatchers 191

populations.Pp. 21-45 in Molecularevolution (F. the SecondInternational Congressof Systematic J. AyalaßEd.). Sunderland, Massachusetts,Sin- and EvolutionaryBiology. Pittsburghß Carnegie- auer Assocß Mellon Univ. Press. --, M. H. SMITH, S. Y. YANG, W. E. JOHNSON,& J. WILSON, g. C., S.S. CARLSON,& t. J. WHITE. 1977. B.GENTRY. 1971. Biochemicalpolymorphismand Biochemical evolution. Annu. Rev. Biochem. 46: systematicsin the genusPeromyscus. I. Variation 573-639. in the old-field mouse (Peromyscuspolionotus). WRIGHT,S. 1951. The genetical structure of popu- Univ. Texas Stud. Genet. 6: 49-90. lations. Ann. Eugen. 15: 323-354. SLATKIN,M. 1981. Estimatinglevels of gene flow in ß 1965. Theinterpretationofpopulationstruc- natural populations.Genetics 99: 323-335. ture by F-statisticswith special regard to systems --. 1985a. Rarealleles as indicators ofgene flow. of mating. Evolution 19: 395-420. Evolution 39: 53-65. ß 1978. Evolution and the geneticsof natural 1985b. Gene flow in natural populations. populations.Vol. 4, Variability within and among Annu. Rev. Ecol. Syst. 16: 393-430. natural populations. ChicagoßUniv. Chicago SNF•TH, P. H. A., & R. R. SOKAL. 1973. Numerical Press. taxonomy. San FranciscoßW. H. Freeman. YANG,S. Y., & J. L. PATTON.1981. Genic variability SWOFFORD,D.L. 1981. On the utility of the distance and differentiationin the Galapagosfinches. Auk Wagner procedure.Pp. 25-43 in Advancesin cla- 98: 230-242. distics (V. A. Funk and D. R. BrooksßEds.). New ZINK, R. M. 1986. Patternsand evolutionary signif- York, Proceedingsof the Willi Hennig Societyß icance of geographicvariation in the schistacea New York Botanical Garden. group of the Fox Sparrow (Passerellailiaca). Or- 1985. Phylogenetic analysis using parsi- nithol. Monogr. 40. mony, version 2.4. Champaign, Illinois Nat. Hist. ß & N. K. JOHNSON.1984. Evolutionary ge- Surv. neticsof flycatchers.I. Siblingspecies in the gen- --, & R. B. SELANDER.1981. A computerprogram era Empidonaxand Contopus.Syst. Zool. 33: 205- for the analysisof allelic variation in genetics.J. 216. Hered. 72: 281-283. ß, D. F. LOTT, & D. W. ANDERSON. 1987. Genetic THORPE,Jß P. 1982. The molecularclock hypothesis: variationßpopulation structure, and evolutionof biochemicalevolution, genetic differentiation and California Quail. Condor 89: 395-405. systematics.Annu. Rev. Ecol. Syst. 13: 139'2168. ß & J. V. REMSENJR. 1986. Evolutionary pro- W•CE, D.B. 1981. The applicationof allozyme evi- cessesand patterns of geographicvariation in dence to problemsin the evolution of morphol- birds.Pp. 1-69 in Current ornithologyßvol. 4 (R. ogy. Pp. 257-270 in Evolution today (G. G. E. F. Johnston, Ed.). New Yorkß Plenum Press. Scudderand J. L. Reveal,Eds.). Proceedings of