THE AUK A QUARTERLY JOURNAL OF ORNITHOLOGY VOL. 105 APRIL 1988 NO. 2 GENETIC VARIATION IN SYMPATRIC WILLOW FLYCATCHERS (EMPIDONAX TRAILLII) AND ALDER FLYCATCHERS (E. ALNORUM) GILLESSEUTIN 1 AND JEAN-PIERRESIMON D•partementde sciencesbiologiques, Universit• de Montreal,90 VincentD'Indy, Outremont,Quebec H2V 2S9, Canada ASSTRACT.--Weused allozyme electrophoresisto estimatethe level of genetic variability and differentiation of allopatric and sympatricpopulations of the Willow (Empidonaxtraillii) and Alder (E. alnorum)flycatchers from southeasternCanada. We reasonedthat, if hybrid- ization occurs,sympatric populations should be more variable and interspecificallyless dif- ferentiatedthan allopatricpopulations. These predictions were not supportedby our results. Therefore, we concluded that in the surveyed region the two speciesare reproductively isolated.This conclusionis weakenedby the fact that samplesizes were relatively small and that geneticdifferentiation of the two speciesis low. Theseconditions render interbreeding difficult to detect if it occursat a low frequency. Received3 June1987, accepted 16 November 1987. THE Tyrannidae form the largest and most lation of two taxa, however, cannot be inferred diversefamily of birds in the New World (Tray- from the mere presenceof differences in their lot and Fitzpatrick 1982). Many genera com- vocalizations. For instance, in Myiarchusswain- prise groups of similar species, a number of soniand in M. tuberculifersome conspecific pop- which are consideredsuperspecies (e.g. Mayr ulations have different vocalizations, and allo- and Short 1970,A.O.U. 1983). Reproductive iso- patric birds do not recognize foreign songs as lation between sympatric members of these being conspecific.Nevertheless, they hybridize speciesgroups is often achieved through hab- freely when sympatric(Lanyon 1978). itat isolation (e.g. Johnson 1963, 1980). If eco- Ornithologists usually have relied on anal- logical preferences of two speciesare similar, ysis of morphometric and plumage characters however, their isolation depends either on the to determine if potentially hybridizing taxa are existenceof a postmating isolating mechanism genetically isolated (e.g. Rising 1983). It is dif- or on the ability of individual birds to discrim- ficult to study hybridization between sibling inate between heterospecificsand conspecifics speciesusing phenotypic characteristics,how- before mating (Mayr 1963). ever. Electrophoretictechniques provide orni- In birds, song differencesare thought to be thologistswith a potentially powerful tool for of prime importanceto ensurereproductive iso- the investigation of such cases(Barrowclough lation in many pairsof sympatricsibling species 1983,Matson 1984).In a few instancesthey were (Becker1982), particularly in the tyrannids (e.g. usedto survey contactzones between taxa that Lanyon 1963, Johnson 1980). Reproductive iso- are known or suspectedto hybridize (Martin and Selander 1975, Corbin et al. 1979, Barrow- clough 1980, Braun 1981, Corbin 1981, Johnson • Presentaddress: Department of Biology, Queen's and Zink 1983, Braun et al. 1984, Braun and University, Kingston, Ontario K7L 3N6, Canada. Robbins 1986, Grudzien and Moore 1986). A1- 235 The Auk 105: 235-243. April 1988 236 SEUTINAND SIMON [Auk,Vol. 105 lozyme traits are of great interest for the anal- Braunand Parker 1985).To maximizethe reliability ysis of hybridization becausethey are discrete of our estimatesof geneticvariability and differen- characters,generally inherited in a simpleMen- tiation, a specific effort was made to detect hidden delian fashion,and are usuallythought to be alleles(Coyne et al. 1979).For mostloci assayed, a set selectivelyneutral (Kimura 1983,Barrowclough of individual samplesof both specieswas run in dif- et al. 1985).Alleles of different frequenciesin ferent gels and buffer systems.The gel-buffer com- bination that provided the clearestbanding pattern allopatricpopulations of two taxamay be good and highestvariability was selectedfor the analysis markersto assessgene flow acrossa contactzone. of the remaining individuals (Table 2). Moreover, simple models make it possibleto Staining techniqueswere similar to thosedescribed predictthe behaviorof somegenetic parameters by Harris and Hopkinson (1976) and Barrowclough through a hybrid zone (Workman 1969,Endler and Corbin (1978).GLUD wasstained by the method 1977,Corbin 1981). These predictionscan then of Brewer (1970). Two esterases(Es-1 and Es-D) were be used to detect hybridization between sym- characterizedusing 4-methylumbelliferylacetate (4- pattic sibling species. Mg) as substrate;Es-3 and Es-4 were stained using We usedthis approachto determine if Willow a-naphthyl acetate. On the basis of heterozygote (EmpidonaxtraiIlii) and Alder (E. aInorum)fly- banding patterns, Es-D and Es-4 were dimeric, and the other two were monomeric.Three globins (Gb) catchershave hybridized in areasof sympatry were stained using the benzidine test (Gordon 1969), in southeasternCanada. These are closely re- and three general proteins (Ptm) were characterized lated sibling species(Stein 1958, 1963;Zink and with CoomassieR-250 (Righetti and Drysdale 1974). Johnson 1984; Shields et al. 1987) that can be Thirty-six presumptivegenetic loci were scored.We separated almost exclusively by advertising alsoattempted to analyze GDH (E.C no. 1.1.1.47),GR songs. Their phenotypic resemblance and the (1.6.4.2),NP (2.4.2.1), HK (2.7.1.1), ACP (3.1.3.2), HEX closesimilarity of their ecologicalpreferences (3.2.1.3.),ALD (4.1.2.13),and FH (4.2.1.2.),but they in southeastern Canada (Barlow and Mc- proved to be unscorable.When more than one locus Gillivray 1983)make them good candidatesfor wasobserved for a particularenzyme, they were num- hybridization. bered sequentially,beginning with the most anodal. Alleles at variableloci were codedby letters,begin- ning with "a" for the most anodal. MATERIALS AND METHODS Presumptiveindividual genotypeswere deduced from banding patterns on stained gels. From these We analyzedthe geneticvariability of 94 specimens we derived a table of allelic frequencies.Observed from 6 Willow or Alder flycatcherpopulations. Birds heterozygosity(Hobs) was determined by directcount were collectedduring the summersof 1985and 1986. for eachspecimen and then averagedfor eachsample Geographiclocation of the populationsamples, their (Corbin 1983).The BIOSYS-1program (Swoffordand status(sympatric or allopatric), and sample sizesare Selander 1981) was used to compute the proportion givenin Table1. Mostspecimens from sympatricpop- of polymorphicloci, the mean numberof allelesper ulationswere collectedat locationswhere syntopy locus,the expectedheterozygosity (Hexp), F-statistics occurred.Specific identity wasestablished by a play- (Wright 1965,1978),and Nei's (1978)and Rogers'(1972) backexperiment before collection (Seutin 1987a).Pro- genetic distances.For each sample genotypic distri- ceduresfor collectionand storageof tissuesamples bution at eachlocus was examined for departurefrom followed those describedby Johnsonet al. (1984). Hardy-Weinberg expectations,using a X2 test (Sokal Sampleswere transportedfrom the field on dry ice and Rohlf 1981). When more than two alleles oc- and stored at -80øC in the laboratory. Extractsof curredat a locus,the actualgenotypes were grouped breastand heart muscleswere preparedseparately; into three classes,according to the procedure pro- liver and kidney sampleswere combinedto produce posedby Swoffordand Selander(1981). This reduced a third extract.Tissue samples were homogenizedwith the number of classeswith low frequencyand less- an equal volume of the buffer solution (Karig and enedthe probabilityof generatingartificially inflated Wilson 1971)and centrifugedat 23,000g for 20 min test values. Among-sampleheterogeneity in allelic at 4øC. frequencydistributions was analyzed using the G test. Electrophoreseswere carried out on horizontal 10% This test was used because it is less sensitive than the starch gels or on vertical polyacrylamideslab gels X2 test to the presence of rare classesin the distri- (Chapel et al. 1974) that varied in acrylamidecon- butions, and it usually better approximatesthe the- centration from 7% to 10% (Seutin 1987b). A discon- oretical X2 distribution (Sokal and Rohlf 1981). The tinuous buffer system(Ornstein 1964) was used for SPSS(Nie et al. 1975) and BMDP (Dixon 1981) com- electrophoresesconducted in acrylamidegels, and a puter packageswere usedto compute theseand other 25-30-mA current was applied for 4-7 h. Four buffer statistics. systemswere usedfor starchgels (Selander et al. 1971, We usednonmetric multidimensional scaling (MDS) April1988] Willowand Alder Flycatcher Genetics 237 T^BLE1. Geneticvariability at 36 loci in 6 flycatcherpopulations. Acronyms in parenthesescorrespond to those used in other tables and in Fig. 1. % of Mean poly- no. of mor- al- phic leles/ Populationa Status n locib locus Hobs+ SE Hexp+ SEc /•sa Empidonaxalnorum St.-Jean-Vianney(Ea-SJV) Allopatric 19 22.2 1.56 0.082 + 0.009 0.083 + 0.030 -0.025 Mont-Tremblant(Ea-MtT) Allopatric 16 16.7 1.44 0.063 + 0.007 0.073 + 0.030 0.048 Montreal (Ea-Mtl) Sympatric 16 19.4 1.50 0.081 + 0.010 0.084 + 0.030 -0.001 Brighton (Ea-Br) Sympatric 14 19.4 1.44 0.070 + 0.009 0.075 + 0.028 0.021 Empidonaxtraillii Montreal (Et-Mtl) Sympatric 16 27.8 1.44 0.082 + 0.013 0.090 + 0.029 0.006 Brighton(Et-Br) Sympatric 13 22.2 1.42 0.069 + 0.009 0.076 + 0.026 0.063 Coordinatesof locationsare given by Seutin (1987a). Frequencyof mostcommon allele -< 0.95. Unbiased