Zonotrichia, Melospiza, Junco , and Passerella

PATTERNS OF GENIC AND MORPHOLOGIC VARIATION AMONG SPARROWS IN THE GENERA ZONOTRICHIA, MELOSPIZA, JUNCO , AND PASSERELLA

ROBERT M. ZtNK Museumof VertebrateZoology and Departmentof Zoology,University of California, Berkeley,California 94720 USA

ABSTRACT.--Analysesof genic variation and skeletal morphology were used to examine relationshipsamong 10 speciesin the generaZonotrichia, Junco, Melospiza, and Passerella. Cluster and principal componentsanalyses showed that morphologic differentiation in- volves size and aspectsof the skull, synsacrum,and hind limb. The genera have basically different skeletalmorphologies; there is someoverlap, however. The pheneticgroupings of taxa, based on skeletal morphology, were congruent with neither traditional taxonomic limits nor the geneticanalysis. The pheneticgroupings seem biased by nondivergenceand convergence.I conclude that the analysisof the genicdata produces a betterestimate of the phylogenetichistory of the group than the analysisof morphology. The genic data show that the generaare distinctlineages, although the affinitiesof Z. capensisneed clarification.The branchingorder of Zonotrichia,J. hyemalis,and Melospizais unresolved,while P. iliacais probablya sistertaxon to thesethree genera.Melospiza melodia and P. iliaca are not dosely related, contraryto previous suggestions.The Lincoln'sSparrow and Swamp Sparrow are more similar to each other than either is to the Song Sparrow. Within Zonotrichia,the phylogenetichypothesis suggeststhat the White-crowned Sparrow and Golden-crownedSparrow are the most similar species,with the remaining taxa joining them in the following order:Harris' Sparrow,White-throated Sparrow, and Rufous-collared Sparrow. This phylogeneticscheme contradicts previous opinions, which, I suggest,were based on inconclusive evidence. The Harris' Sparrow, which is divergent relative to its congenersin both plumageand skeletalcharacters, is genicallysimilar to its congeners.The Rufous-collaredSparrow is morphologicallyand genicallydifferentiated from its north-tem- peratecongeners, and, if it is a valid member of Zonotrichia,it must be an early derivative. Heterozygosity(• = 0.039), percentagepolymorphic loci (,• = 20.3), and mean number of allelesper polymorphiclocus (• = 2.4) were similar to valuesreported for otherbirds and support the observationthat birds possesslevels of within-species geneticvariation typical of other vertebrates.However, geneticdistances averaged 0.06 between congenericspecies and 0.26 for noncongeneric(but confamilial)interspecific comparisons and are lower than those observedat comparabletaxonomic levels in other vertebrates.Rates of genic change seemhomogeneous. Converting genetic distances into estimatesof divergencedates suggeststhat the generaoriginated from 1.3 to 6.6 MYBP. Speciationwithin Zonotrichiaand Melospizaprobably occurred in the Pleistocenebut beforeapproximately 140,000 yr ago. ! suggestthat Passerella,Melospiza, Junco, and Zonotrichiabe retainedas distinctgenera until comparablegenetic data from other emberizinesare available.Received 11 November1981, accepted26 March 1982.

THE methodsof data gatheringand analysis (seeNevo 1978)and has provided, for example, used in systematicbiology have greatly ex- data on levels of genetic variation in natural panded in recent years. Biochemical assess- populations, on the genetic structure of pop- ments of genetic variation, quantitative anal- ulations and species,and on phylogenetic re- yses of genetic and morphologic data, and lationships among taxa. However, there have progressin the theoreticalaspects of system- been few quantitative multi-locus studies of aticshave provided powerful meansfor inves- proteinvariation at the specificand genericlevel tigating evolutionary relationshipsamong or- in birds (e.g. Barrowcloughand Corbin 1978; ganisms. The use of electrophoresisto study Avise et al. 1980a, b; Yang and Patton 1981) protein variationhas been especiallyprevalent relative to other vertebrate groups (Nevo 1978,

632 The Auk 99: 632-649. October 1982 October1982] EvolutionaryRelationships of Sparrows 633

Barrowclough 1980b). Similarly, there have ipers. The mean of the five measurementsfor each been relativelyfew phenetic(e.g. Schnell1970, characterwas computed for each taxon. While five Robins and Schnell 1971, Hellack and Schnell individuals cannot encompass within-taxon varia- 1977,Wood 1979)and cladistic(e.g. Payne and tion, it has been apparentlyassumed in other, sim- Risley 1976, Raikow 1976, Simpson and Cra- ilar studies(e.g. Schnell1970, Wood 1979)that such sample sizes are sufficient, because within-taxon craft 1981)appraisals of morphologicvariation variation is much less than among-taxonvariation. in birds. In few studies (Handford and Not- This assumption,in regardto geographicvariation, tebohm 1976) were protein and morphologic will be discussed below. variation examinedin concert.The increasing Principal componentsanalysis (PCA) and cluster use of thesetechniques and analyses,and more analyses(UPGMA and WPGMA) were usedto study than one data set, will be important to (1) eval- patternsof morphologicalvariation (seeSneath and uate evolutionary and systematicrelationships Sokal 1973). PCA is appropriatefor elucidating the among avian taxa, (2) compare avian evolution majorphenetic groupings of taxa.The principal com- at different levels(e.g. morphologicand genic), ponents, orthogonal and unrotated, were extracted and (3) generatedata that canbe comparedwith from a covariancematrix using the program PNCOMP (Duncan and Phillips 1980).The correlationsof char- resultsfrom other vertebrate groups. acterswith the first four componentswere examined The systematicrelationships among species to determine which characters best "summarized" in the generaZonotrichia, Junco, Melospiza, and the variation. Passerellahave been addressedby severalau- Becauseduster analysis is mostaccurate at the level thors(Linsdale 1928; Dickerman 1961; Paynter of "branchtips" (i.e. higher levelsof similarity), it 1964, 1970; Short and Simon 1965; Mayr and is not suitablefor definingbroader groupings, the Short1970), but therehas been little recentsys- latter being sensitiveto different clusteringalgo- tematicwork on these taxa. In this paper I ana- rithms and distancecoefficients (Presch 1979; see be- lyze patterns of genic (= allozymic)and mor- yond). Therefore,ordination (e.g. PCA) and cluster- phologic(skeletal) variation in order to produce ing techniques are complementary and are both employedhere to evaluatethe morphologicdata set. independent,quantitative estimates of the sys- The raw character means were variance standard- tematic relationshipsand evolutionaryhistory ized, and the Taxonomic Distance (TD, Sneath and of these emberizine sparrows. Genic versus Soka11973)measure was used to constructan OTU x morphologicdifferentiation, rates and levels of OTU distance matrix (available from author). The genic divergence,and taxonomicrelationships program CLUST, written by W. W. Moss (Duncan are discussed. and Phillips 1980), was used to construct phenograms(for morphologicand geneticdata) and a PRIM METHODS AND MATERIALS (or minimum spanning)network. The clusteringand ordinationanalyses were used Morphology.--Taxa examined were: Rufous-col- on the 13-taxon x 40-character matrix. In addition, I lared Sparrow (Zonotrichiacapensis), White-crowned constructedphenograms from subsetsof the 13 taxa. Sparrow(Z. leucophrys),White-throated Sparrow (Z. This procedure involved the removal of one or two albicollis),Golden-crowned Sparrow (Z. atricapilla), subspeciesof P. iliaca in order to test whether mor- Harris' Sparrow (Z. querula), Song Sparrow (Melo- phologicallydifferent subspecies,when used in var- spizamelodia), Lincoln's Sparrow (M. lincolnii),Swamp ious combinations,resulted in different among- Sparrow(M. georgiana),Fox Sparrow (Passerella iliaca speciespattems. stephensi,P. i. canescens,P. i. townsendi),Dark-eyed Electrophoresis.--Samplesizes and generallocali- Junco (Juncohyemalis), and Bachman's Sparrow ties for specimensare given in Table 1. Samplesof (Aimophilaaestivalis); the latter taxonwas used as an liver, heart, and pectoralmuscle were storedat -76øC. outgroup. The subspeciesof P. iliaca were used as For preparation of tissue extracts, samples were separateOperational Taxonomic Units (OTU's) (see thawed, minced with a razor blade, and combined below). Specimensused in this studyare housedat with an equal volume of deionized water. Samples the Museum of Vertebrate Zoology, University of were then centrifuged at 16,000 rpm for 40 min at California, Berkeley;further informationconcerning 4øC. The supematant was frozen at -76øC for later these specimensis available from the author. Five electrophoreticanalysis. skeletonsper taxon were measuredfor 40 characters ! examined39 presumptivegenetic loci (Appendix (described in Robins and Schnell 1971) from most 2) using standardstarch gel electrophoreticproce- bodyregions (see Appendix 1). Only adultmales were dures (Selander et al. 1971, Avise et al. 1980a, Bar- used exceptfor one female eachof Z. leucophrysand rowclough1980b, Yang and Patton1981). Protein as- Z. albicollisand two of M. georgiana.Measurements sayswere preparedfollowing Selander et al. (1971), were recorded to the nearest 0.05 mm with dial cal- Harris and Hopkinson (1976), and Yang and Patton 634 ROBERTM. Z•NK [Auk,Vol. 99

(1981). Further information on electrophoreticcon- the genera generally occupy different regions ditions is available from the author. of the 40-character space. The superimposed Acrossall species,the most frequent allele at a lo- PRIM network, however, showsthat Z. querula cus was designatedas 100; higher numbers imply a is the nearest neighbor of P. i. townsendiand more anodal migration. Allelic mobilities, relative to not of Z. atricapilla. Also, Z. capensisis nearer 100, were estimatedto the nearest5 units. Isozymes to A. aestivalis than to other Zonotrichia. The (e.g. GOT-1 and GOT-2) are designatedby a 1 for the mostanodal and sequentiallyhigher numbersfor PCA reveals that species of Zonotrichia differ more cathodal forms. Estimates of within-species substantiallyin size (PC I). The subspeciesof variability includeaverage heterozygosity per locus, P. iliaca and the speciesof Zonotrichia are rel- percentageof loci polymorphic (most frequent allele atively dispersedin the principal component •0.99), and meannumber of allelesper polymorphic space, whereas species of Melospiza are more locus (per species).Heterozygosity was determined tightly grouped. The sample of ]. hyemalis by direct countfor eachspecimen and then averaged standsapart and is nearest to Melospiza. (+SD) for each species.The methodsof Nei (1978) The separation of taxa on PC II (Fig. 1) is and Rogers(1972) were used to calculategenetic dis- mostly attributable to differencesin the sizes tancesbetween taxa. Phenograms(Sneath and Sokal 1973) and phylogenetictrees, constructedusing the and shapesof skulls (Appendix 1). Passerellai. methods of Fitch and Margoliash (1967;F-M trees) stephensi,which has the largest skull of any and Farris (1972; Wagner trees), were used to esti- subspeciesof P. iliaca (Linsdale 1928, Zink un- mate the branching pattern and amount of diver- publ.), and Z. querula represent the extremes gence among the taxa. Phenogramsimply homoge- of variation on this "skull" axis, with the re- neity of ratesof genic changealong branches, while maining taxa having intermediatepositions. the phylogenetictrees do not. The F-M procedure The taxa are arrayedalong PC III from ]. hye- constructsa number of treesby alteringthe branch- malis to A. aestivalis.Characters contributing ing patternand branch lengths. Alternative trees were to PC III (Appendix 1) include: skull (POW), evaluated by the percentageStandard Deviation pectoralregion (FPL), leg (FMW, TTW, TMW), (%SD;Fitch and Margoliash 1967) and the magnitude and number of negativebranches (the fewer the bet- and wing (ULL, CML). The highest loadings ter). A lower %SD implies a better fit of branching on PC IV are from the synsacrum(PSL, SMW), diagram to original distance data. The programs leg (FEL, FPE, FDE, TTL, TML, TDE), and wing EVOLVE (written by W. M. Fitch) and WAGNER (ULW, CMD). (written by J. S. Farris) were used for the F-M and The plot of PC II-IV (Fig. lb), which ac- Wagner trees, respectively (Duncan and Phillips counts for 14.5% of the variation, can be in- 1980). A cladistic (sensuHennig 1966) analysis, terpretedas portraying"shape" rather than size using alleles as character states(see Wake 1981), was because of the deletion of PC I. Because of the partially useful at the generic level; however, few low amount of variation explained, however, synapomorphic allelic states were discovered (see and the fact that some nonsize-related varia- Appendix2) that defined cladesat lower levels,e.g. tion is present on PC I, the results should be groups of species. Alleles in P. chloruruswere con- sidered plesiomorphic(ancestral). interpreted with some caution. The following groupings appear to occupy

RESULTS different areas of the plot of PC II-IV: species of Zonotrichia(excluding capen$i$), A. aestivalis SKELETALMoRPHOLOGY and Z. capensis,1. byemalls,and two subspecies of P. iliaca (canescensand stephensi).The Principalcomponents analysis.--The plot of the three subspeciesof P. iliaca are still dispersed, taxa on Principal Components (PC) I-III is suggestingconsiderable shape heterogeneity. shown in Fig. la and on PC II-IV in Fig. lb. The first three axes account for 94% of the vari- In Fig. lb, the speciesof Zonotrichia(excluding ation in skeletal characters, and PC 1V accounts capensis)are more tightly groupedthan in Fig. for an additional 3.1%. Along PC I, the taxa la, again suggestingthat size differencescon- are distributed accordingto size, from left to tribute in an important way to their separation right.The correlation of mX/•-•• with scoreson in Fig. la. The overlap of Melospizaand P. i. PC I, for each species, is 0.847 (P < 0.01, df = townsendiin Fig. lb suggestseither conver- 11). Most charactersload highly and positively gence in "shape" or retention of the ancestral on PC I (Appendix 1), further supportingthe morphology. conclusionthat PC I is a "size axis." In Fig. la, Thus, the two plots (Fig. 1) show that there October1982] EvolutionaryRelationships of Sparrows 635

o. Pcm ...step..I' (4. [% ) /'fi•co' .I -•\ •

• _•.••.•59- I• I - •

(z•%)

Fig. 1. Three-dimensionalrepresentations of principalcomponents analysis of 40 skeletalcharacters (see Appendix 1). a. Plot of taxa on PC I-III. Becausethe original 40-dimensionalspace is reducedhere to 3 dimensionsby the PCA, a PRIM ne•ork is superimposedto indicatea taxon'snearest neighbor in "skeletal morpholo• space"and to show disto,ions in this space-reducingprocedure. Branch leng•s are in units of TaxonomicDistance. The percentageof variancein the data explainedby ea• componentis indicated in parentheses.The dashed lines enclosetaxa traditionally assignedto Melospiza,Zonotrichia, and Passerella iliaca. b. Plot of taxa on PC II-W. 636 ROSERTM. Zz•q•c [Auk,Vol. 99

P conestens There are two basicclusters in Fig. 2a, one con- sisting of Zonotrichiaand Passerellaand one UPGMA P stephens/ rcc = •700 including ]unco and Melospiza.The similarity Z querula of Z. capensisand A. aestivalis(the "outgroup") P to•vnsend/ was unexpected, as was the occurrenceof Z.

Z a)bico/hg querulawithin the P. iliaca duster. [D. Scott Wood (pers. comm.) alsofound the latter result Z. /eucophrys in a similarstudy.] The remainingthree species Z atricapi//a of Zonotrichiaduster together fairly tightly, with

Z capensis Z. leucophrysand Z. albicollisbeing more similar to eachother than either is to Z. atricapilla. A aestiva//s The subspeciesof P. iliacaare relativelydiver- M Hnco/nH gent and more similar to Zonotrichia(excluding M georgiano perhaps capens/s)than to Melospizaor Junco. Within Melospiza,lincolnii and georgianaare M. rne/odia more similar to each other than either is to me- d h.vema/is lodia. The J. hyemalissample is doser to Me- lospizathan to other taxa. The recvalues for 1.7 .81 .21 these phenograms, 0.74 for the WPGMA and 0.70 for the UPGMA, suggestthat these dia- P / canescens grams are only fair representationsof the TD WPGMA Z a/bica///s matrix. rcc =,757 Cluster analysis of various subsets consis- Z. /eucophr.vs tently placedZ. querulawith subspeciesof P. Z. atricapi//a iliaca, although the relationshipwas ambigu- JE /. tawnsend/ ous.For example,in Fig. 2b, Z. querulais more similar to P. i. townsend/ than to other Zono- Z. queru/a trichia, while P. i. canescensis more similar to M. /incalnii Z. atricapilla,leucophrys, and albicollis(TD = M •1ear•1iana 0.80), than to P. i. townsend/(TD = 1.0).

M me/ad/a By varying either the taxa induded and/or the clusteringalgorithm of the analysis,the d. h.vema//s phenetic placementof Z. capensisoscillated be- Z. capensis tween joining either Melospiza-]uncoor other

1,55 0.95 0;50 Zonotrichia(excluding querula) as a sistergroup. Also, in a few instances, the higher-level Taxonorn ic Distance branchingstructure was altered. The relation- Fiõ. 2. Phenoõramsbased on t_hesame J.0 skeletal ships of Z. albicollis,Z. leucophrysand Z. atri- charactersas in Fiõ. 1. a. LTPGMAaz•alysis of alttaxa. capilla,the speciesof Melospiza,and J. hyemalis b. WPGMA anatysisof sametaxa exctudinõ.4. aes- were consistentwith Fig. 2 in all analysesof tivalis and P. i. stephens/.The resultsof analysesof subsets; therefore, their phenetic positions other subsets of taxa are summarized in the text. seem stable.

GENETIC VARIATION are size and shapedifferences among the genera but that there is also some overlap. Varia- Levelsof intraspecificgenetic variation and dif- tion in overall skeletal morphology includes ferentiation.--Allelic frequenciesare given in most body regionsbut particularlythe skull, Appendix 2. Heterozygoteswere observedat pelvic region, and hind limb. 18 of 39 loci (46.2%), while 6 loci showedonly Clusteranalysis.--The UPGMA phenogram fixed differencesacross species. The remaining of the 13 taxa is shown in Fig. 2a. A WPGMA 15 loci were monomorphicand fixed for the analysis(not shown)of the same13 taxa dif- same allele for all taxa. Thus, a total of 24 of 39 fered only in that the Z. capensis-A.aestivalis loci (61.5%) were variable for these 11 species. cluster was most similar to Melospiza-Junco. A partial correlationanalysis (BMDP6R, Frane October1982] EvolutionaryRelationships of Sparrows 637

04 04 04 C• • 04

• o o • o • o • D o

+1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 •

o o • • o o o o • o o 638 ROItERTM. Ztm< [Auk, Vol. 99

1977) showedthat samplesize (i.e. number of animalsper species)was not a significantpre- dictor (determined by two-tailed t-tests, P > 0.05) of heterozygosity,percentage of loci polymorphic, or number of allelesper polymorphic locus (Table 1). Therefore, the values can be compareddirectly between species.The Har- ris' Sparrow has lower levels of genetic variability than the other species,which have similar levels of variation among themselves.The levelsof geneticvariation shown in Table 1 are similar to those reported for other birds (Bar- rowclough and Corbin 1978; Barrowclough 1980b; Avise et al. 1980a, b; Yang and Patton 1981). Even with broad geographicrepresentation for somespecies (Table 1), within-speciesgen- ic differentiation was much less than among- specieslevels. Nei's (1978)D among breeding M. melodia from California and Minnesota, and migrantsfrom Illinois, was lessthan 0.(J07.The sampleof P. iliacais part of a major study(Zink unpubl.) of 619 breeding individuals from a total of 31 localitiesin California, Oregon, and Nevada; 7 subspeciesincluding canescensand stephensihave been examined. The largest D-value for any pair of localitiesis 0.004 (• = 0.001);the t) betweencanescens and stephensi is 0.0004. Fox Sparrowsfrom Illinois (migrants of the iliaca subspecies group) and Cali- fornia (migrantsof the unalaschcensissubspe- cies group) are very similar to the breeding birds I examined (discussed above) from the western United States. The sample of J. hyemalis is part of a genic analysisby George F. Barrowclough (unpubl.), which includes breeding birds from populationsdistributed from South Dakota to southern California. The largestD-value between populationsis 0.004. These levels of within-species (i.e. population or subspecies)genic divergence are typical for birds (Barrowclough1980b) and indicate that further samplingof additionalindividuals and geographicregions would probably not alter the relationshipsoffered below (with the possible exceptionof Z. capensis). Interspecificgenetic distance.--The genetic distances(Nei's and Rogers')between species are given in Table 2. The averagegenetic distance(t) + SD; Nei 1978)among five species of Zonotrichia is 0.118 + 0.073 (0.063 + 0.024 excludingcapensis) and amongthree speciesof Melospiza,0.059 + 0.026.These congeneric, in- October1982] EvolutionaryRelationships of Sparrows 639 terspecificlevels are an order of magnitude leles at GPT (120) and G-6-PDH (110); differ- greaterthan the intra-specificlevels discussed entiation within Melospizais due mostly to al- above.The/• amongspecies of Zonotrichia and leles at ACON, EAP, ACP, and ADA. Melospizais 0.262+ 0.033(15 comparisons); P. Representative branching diagrams sum- iliaca and Zonotrichia, 0.335 + 0.027 (n = 5); P. marizing the pairwisematrix of Rogers'D-val- iliaca and Melospiza,0.254 + 0.048 (n = 3); ]. ues (Table 2) are shown in Fig. 3. A total of 17 hyemalisand Zonotrichia,0.205 + 0.031 (n = 5); branching diagrams(14 F-M trees, 2 pheno- ]. hyemalisand Melospiza,0.185 + 0.008(n = 3). grams,Wagner tree) was examined and the fol- The/• amongcongeneric species in Zonotricia lowing conclusionswere consistentlysupport- (excluding capensis)and Melospiza, 0.063 + ed: (1) P. chlorurusis an appropriateoutgroup; 0.024,is clearlyless than the/• for nonconge- (2) P. iliaca,Melospiza, ]. hyemalis,north-tem- neric comparisons(0.262 + 0.033). The impli- perate speciesof Zonotrichia,and possiblyZ. cationsof therelatively high/• betweenZ. ca- capensisare distinct lineages,and P. iliaca is pensisand its congeners,0.199 + 0.010,will be probablya sistergroup to thesegenera; (3) the consideredbelow. Theselevels of geneticdif- relationshipsof speciesin Zonotrichiaand Me- ferentiation are similar but somewhat larger Iospizaare as shownin Fig. 3; and (4) Z. ca- than those reported by Barrowclough(1980b) pensisis not closelyallied to its north-temper- for other passerinecongeneric (,• = 0.044) and' ate congeners,and it is of uncertainaffinities. noncongeneric(,• = 0.214) interspecificcom- Disagreementamong branchingdiagrams of parisons. similar efficiency (i.e. similar r,.c or %SD) Genicrelationships among taxa.--Loci that best showedthat the branchingorder of ]. hyemalis, discriminate among taxa were analyzed da- Melospiza,and Zonotrichiais unresolved.For distically (seeWake 1981 for discussion);only example, the UPGMA phenogram and the allelesin high frequencywere used.Assuming "best" F-M tree had the same branching se- that the allelesin P. chlorurusrepresent the an- quence(Fig. 3a), while the Wagner parsimony cestralcondition, the four generaconstitute a tree (Fig. 3b) places]. hyemaliscloser to Zo- monophyleticgroup based on the possession notrichiathan to Melospiza.A WPGMA phe- of the same shared-derived alleles (synapo- nogram (rcc= 0.94) and an F-M tree (%SD = morphies) at GOT-1, GAPDH, and GLUD. 8.45, two negativebranches) (neither shown) Passerellailiaca has unique alleles(autapomor- resembledFig. 3b; the positionsof J. hyemalis phies) at NP, SOD-l, G-6-PDH, and crGPD-1 and Z. capensiswere exchanged,however, re- and sharesa synapomorphywith Melospizaat flectingthe conflictingdata discussed above for LDH-1 (175). The genic monotypy of J. hye- NP, LDH-1, and G-6-PDH. Overall, Z. capensis mallsis supportedby autapomorphiesat LDH- is moresimilar to its congeners(b = 0.199+ I and ACON, but J. hyemalisshares a synap- 0.010)and J. hyemalis(D = 0.227)than to either omorphy with Melospiza at NP (120), with Melospiza(•) = 0.277 + 0.020)or to P. iliaca north-temperate Zonotrichiaat G-6-PDH (100) (D = 0.322). and an uncommonallele in high frequencywith Ratesof genicdifferentiation.--The F-M and Z. querulaat c•GPD-1 (110). Speciesof Zono- Wagner proceduresestimate branch lengths, trichia form a monophyleticgroup based on which can be interpretedas ratesof genic di- synapomorphiesat LDH-1 (100) and NP (85, vergence. There are no confidencelimits on 100),and the four north-temperatespecies share these "rates," and they cannot be evaluated distinctive alleles at ICD-1. Alleles at 6-PGDH statistically.Also, "rates"will be dependenton and ADA also serve to identify Zonotrichiaas the loci included in a survey, i.e. the rates are a genic clade,and allelesat SDH, ACON, GPI, averagesacross loci. crGPD-1, and LGG contribute to differentiation Within Zonotrichia(Fig. 3a), branch lengths within the north-temperatespecies. Zonotrich- in units of Rogers'D (x10) from the "common ia capensishas unique alleles at PGM-2, G-6- ancestor"range from 0.85 for Z. albicollisto PDH, and crGPD-1 and a unique allele at NP 1.16 for Z. atricapilla (mean + SD = 0.98 + (85) in high frequency.The distributionof syn- 0.14). Branchlengths (Fig. 3a) to speciesof Me- apomorphiesat G-6-PDH and NP suggestsdif- lospizaare 0.39, 0.42, and 0.67 for M. melodia, ferent cladisticrelationships for ]. hyemalis,Z. M. lincolnii,and M. georgiana,respectively (• + capensis, and north-temperate Zonotrichia. SD = 0.49 + 0.15). If Z. capensisis excludedbe- Speciesof Melospizahave shared-derivedal- causeof its uncertain affinities, the comparable 640 ROBERTM. ZZNK [Auk,Vol. 99

O9 Z leucophrys

Z atricapilla

31 .44 Z querula

.33 41 i$•.36 Z. albicoHis I09 Z capensis

.28 24 M. Iincolnii

.54 M. georgiana

39 •L•ß.49 M. melodia .86 d. hyemalis

1.23 Passerella iHaca

2.07 Pipilo chlorurus

3.0 2.0 1.0

Rogers' D (xlO)

08 Z. leucophrys .31 36 Z. atricapilla 47 32 Z. querula 38 .26 Z. a/b/col/is

38[ .61 120J.Z. hyemaliscapens/s

50 53 M.georg/ana .30 M. melodia i.07 P. /1iaca • 2.22 '[•.19• M.p. ch/oruruSI/ncoln/i Fig. 3. Branchingdiagrams summarizing the matrixof Rogers'D-values (x 10).a. Branchingdiagram depictingresults of a UPGMAcluster analysis (rcc = 0.95)and an F~Mtree. Branch lengths are from the F-M tree,which has a %SDof 7.85. b. Wagnertree, rooted at P. chlorurus,with a %SDof 13.8.This branching diagramis an approximationof the mostparsimonious, or minimumlength, tree (Farris 1972). In boththe F-M and Wagnertrees, the distancefrom P. chlorurusto the remainingtaxa is not partitionedinto two branchlengths, because this wouldrequire an additionaloutgroup to P. chlorurus. valuefor the otherspecies of Zonotrichia,0.51 + ia (distancesaveraged), are 1.23, 1.14, 1.31 + 0.14, is equivalent to that for speciesof Melo- 0.15, and 1.59 + 0.14, respectively,for the F- spiza.I concludethat within eachgenus there is M tree. The samecomparisons for the Wagner no compellingevidence to rejecta hypothesisof tree are 1.07, 1.30, 1.34 + 0.26, and 2.15 + 0.22, homogeneity of rates. The distancesfrom the respectively, and suggesta faster relative rate joining of P. chlorurusto P. iliaca,J. hyemalis, for Zonotrichia.Because the Wagnertree has a Melospiza(distances averaged), and Zonotrich- higher %SD than the F-M tree, however, I note October1982] EvolutionaryRelationships of Sparrows 641 only that genicchange in Zonotrichiamay be attributable to incorrecttaxonomy, nondiver- rapid, relative to the other genera,but it is not gence(i. e. plesiomorphy)at the level of overall proven. The suggestionof rate heterogeneity skeletal morphology, homoplasticevolution among genera does warrant the use of tree- (e.g. convergence),or methodologicalfactors constructingmethods, such as F-M and Wag- (e.g. small samplesizes used here; seeabove). ner, whichtake into accountthis possibility. Thus, it is difficultto evaluatethe significance of the phenetic groupingsproposed by this DISCUSSION analysis. Comparisonof morphologicand genicpatterns Evaluationof morphologicanalyses.--Several of variation.--Here, I comparethe phenetic aspectsof the pheneticanalysis deserve atten- groupings(Figs. 1, 2) to the independentgenic tion. The subspeciesof P. iliaca are as diver- estimates(Fig. 3) of relationshipsand analyze gent as Z. atricapilla,Z. leucophrys,and Z. al- specificcases of noncongruence.The phenetic bicollis,and speciesin Melospiza,are from each analysisof skeletalmorphology is designedto other(Fig. 2a). Althoughthese subspecies con- identify groups of taxa based on measuresof sistentlycluster with Z. querula,the analysis overall similarity. In such analyses,contribu- suggeststhat geographicvariation should be tions of individual characters are obscured due evaluatedbefore addressing among-species re- to their combinationinto a singlemeasure, such lationships.While P. iliaca and M. melodiaare as Taxonomic Distance (e.g. Fig. 2), or into both highly polytypic, the remaining species principalcomponents; in PCA, examinationof examined here are less variable. Therefore, I character"loadings" can help recoversome in- assumethat the samples(number of characters formation about individual characters. Thus, and individuals and geographicrepresenta- patternsof changein divergentcharacters may tion) used here are sufficientfor documenting be swampedout by inclusionof nondivergent patternsof pheneticvariation in all forms ex- characters(i.e. plesiomorphicor ancestral)or cept M. melodia(discussed below). charactersthat exhibit convergence,parallel- The stability of pheneticgroupings can be ism, or reversals. In cluster analysis, for ex- examinedby alteringclustering algorithms (see ample,the degreeto which pheneticgroupings also Presch 1979, Wood 1979, Duncan et al. reflectphylogeny will dependon how well the 1980). In Fig. 2 for example,a comparisonof TD measuretracks divergent evolution, which the UPGMA and WPGMA phenograms(each is a point of contention among systematists with similar rcc'S)shows that the phenetic (Mickevich 1978, Presch1979, Rohlf and Sokal placement of Z. capensisis inconsistent.Con- 1980). Independent analysesare necessaryto clusionsdrawn from the phenogramsshould establishwhether or not pheneticgroups of taxa mainly involve the "stable" phenetic group- representphylogenetic groupings. ings. Previous avian phenetic studies (e.g. It is useful to documentpatterns of phenetic Schnell 1970, Wood 1979) have examined alter- variation, becausethey may provide general native phenograms,but assessmentsof geo- indicationsof evolutionarytrends in morphol- graphic variation and independentphyloge- ogy or identify instancesof ecologicalconver- netic analysesare rarely included (see Payne gencebetween species. The/J) among Z. albi- and Risley 1976). collis,Z. atricapilla, and Z. leucophrys,0.06 + The pheneticgroupings of P. iliaca and Zo- 0.03, is two ordersof magnitudehigher than notrichia(especially Z. querula),Melospiza and the/J)between P. i. canescensand P. i. stephensi J. hyemalis,Z. capensisand A. aestivalis,and (0.0004, Zink unpubl.), yet the subspeciesof the relationshipsof Z. leucophrys,Z. albicollis, P. iliaca show greaterlevels of morphologicdi- and Z. atricapilla(Fig. 2) contradicttraditional vergence(Fig. 2). These speciesof Zonotrichia taxonomicarrangements (Mayr and Short 1970, breed in similar habitats (Godfrey 1966), but Paynter 1970).The clusteringof P. i. townsendi the subspeciesof P. iliaca breed in quite dif- with Z. querularather than with its conspecif- ferent habitats (Linsdale 1928). It seems clear, ics does not reflect biologicalspecies limits. in this example, that genic divergence in Zo- Only the phenetic relationshipsof speciesof notrichiahas proceededwithout concomitant Melospizain Fig. 2 resembleprevious taxo- morphologicchange. nomic opinions(Mayr and Short1970). The in- It would be temptingto suggestthat the sim- stancesof disagreementcited above could be ilarity of P. iliaca and Z. querula(Fig. 2) rep- 642 ROBERTM. ZINK [Auk, Vol. 99

resents a "hidden" link between the genera. ogy is a better index to ecologythan to phy- However, Nei's D betweenthese species, 0.314, logeny," and, although I doubt the generality is greater than most comparisonsbetween of this claim (see Simpsonand Cracraft1981), noncongeners.Passerella iliaca is a separate Voous' idea may pertain here. Phenetic anal- cladein Fig. 3, and Z. querulais clearlysimilar ysis of species,based on skeletalmorphology, to its congeners.Therefore, the phenetic re- probablymeasures, at leastin part, phenotypic semblanceis attributableto convergence,ples- responsesto ecologicalconditions. That is, the iomorphy, or the fact that both are "large." A genetic basis of these groupingsis unknown. phylogeneticanalysis (Hennig 1966) of mor- At the allozymiclevel, almostcertainly genetic, phology is required to prove convergencevs. the data, which are bands on gels, have a sim- plesiomorphy. Researchon comparativeas- ple and better understoodrelationship to the pectsof the ecologyof P. iliacaand Z. querula genotype. might show whether or not adaptation to sim- In summary, based on 39 presumptive ge- ilar environments or niches could explain the neticloci, the D's betweenconspecific popu- apparentlysimilar morphologies. lations, congenericspecies, and noncongeneric The branching sequenceamong the other speciesare 0.002(Zink unpubl.), 0.06, and 0.26, species of Zonotrichia is not congruent be- respectively,and show a monotonic increase tween the data sets. Zonotrichia albicollis and of/J)as the taxonomic(and probably biologi- Z. leucophrysare more similar to each other cal) unit examined becomes more inclusive. than either is to Z. atricapilla,based on skeletal Thus, I propose,based on the hierarchicallevels morphology,but the genicestimate substitutes of genicvariation found here, that the analysis atricapillafor albicollis.The genicand morpho- of genic data (Fig. 3) best depictsphylogenetic logic data setsagree, however, that Z. capensis relationshipsamong these emberizine taxa. The is divergent relative to its congeners. morphologicgroupings in this study include Melospizalincolnii and M. georgianacluster instancesof nondivergence(some Zonotrichia) togetherin analysesof both data sets. Wood and probably convergence (P. iliaca and Z. (pers.comm.) found that, by accountingfor the querula). morphologicvariation among subspecies of M. Evolutionaryhistory and levelsof genicdiffer- melodia,the relationships suggestedby Fig. 2 entiation.•Genetic distances can be used to es- were substantially altered. Hence, concor- timate dates of divergence among taxa (Nei dancebetween morphologic and genic data sets 1975,Sarich 1977,Yang and Patton 1981).There for speciesin Melospizamay be tenuous.The are few independent estimatesof divergence genic data (Fig. 3) show that the relationship times against which to calibrate the avian elec- between J. hyernalisand Melospiza is unre- trophoreticclock; such calibrationsare badly solved. This conflicts with the closer relation- needed. Also, geneticdistances are assumedto ship of these taxa shown in Fig. 2. be linear functions of time since divergence Thus,the morphologic(Fig. 2) and genic(Fig. (Wilson et al. 1977, Vawter et al. 1980; but see 3) estimatesof relationshipsamong taxa are not Lesslos1981), and this assumptionneeds ver- congruent.The correlationbetween the genetic ification for avian systems,especially at low distance and taxonomic distance matrices is 0.47 levels of D. (excludingoutgroups and P. i. townsendi).The Estimates presented here are offered as genic data are congruentwith traditionalge- "ballpark" figures. Values were derived using neric limits, however, (A.O.U. 1957), with the the formula t = c x 106 D, where D is Nei's possibleexception of thesefor Z. capensis.There (1978) D-value; the value of c was varied from is no a priori reasonto expectconcordance be- 5 (Nei 1975) to 26.3. The latter value was esti- tween the genic and morphologic data sets mated by Gutierrez et al. (in press)based on (Schnell et al. 1978, Templeton 1981, Wake a calibration of /) and the fossil record of 1981). Genic evolution can occur without con- someNew World Quail (Odontophorinae). comitant morphologic change (Gorman and The genetic distinctnessof the genera (ex- Kim 1977, Highton and Larson 1979), and or- cludingPipilo) examined here, b = 0.25, con- ganismswith different morphologiesmay be verts to ages of from 1.3 to 6.6 million years genicallysimilar (Turner 1974, Avise et al. 1975, beforepresent (MYBP)--a wide range,but im- King and Wilson1975, Yang and Patton1981). portant in suggestingthat late Pleistocenegla- Voous(1980: 1232) proposed that "morphol- ciationswere probably not responsiblefor the October1982] EvolutionaryRelationships ofSparrows 643 origin of theselineages. At the genericlevel, is moreclosely related to otherSouth American given the unresolvedbranching order (Fig. 3) emberizines. and the wide overlap in breeding ranges Passerinebirds are less differentiated geni- (A.O.U. 1957),a biogeographicreconstruction cally, at comparabletaxonomic levels, than is not possible.That is, the data do not allow other vertebrates (Barrowclough1980b, Avise predictionof the time and placeof the origin et al. 1980b,this study).Heterozygosity levels of theselineages. in passerinesincluding those studied here, ap- Congenericspecies in Zonotrichia(excluding proximately 4%, are similar to other verte- capensis)and Melospiza (/• = 0.06)are about brates (Nevo 1978); therefore, birds are not 300,000-1,600,000yr old, suggestingthat spe- lackingthe "stuff" of genicdivergence. ciation in thesegenera occurred in the Pleis- Large effective population sizes and high tocenebut probably not more recentlythan ratesof gene flow typify many avian popula- 140,000yr ago (basedon the lowestobserved tions (Barrowclough1980a), and these factors D, 0.028, between Z. leucophrysand Z. atri- may inhibit genicdivergence at this level.Avi- capilla,and c = 5). Thus,it is unlikelythat gla- an specieslimits are often clearly defined, ciations of WisconsinJanand possiblyIllinoian however, because of the numerous "tests of agesinfluenced speciation in thesegenera, in sympatry"among breeding species. Therefore, contrastto the suggestionby Hubbard (1973) the low interspecificgenic differences (/• = that speciationin other passerinegroups was 0.04; Barrowclough1980b) among passerines affectedby suchglacial events. seem to be a real result. Recencyof common Within north-temperate species of Zono- ancestry(e.g. Baker1981) or slowrates of pro- trichia,Z. albicollisoriginated the earliest(Fig. tein evolutionand rapid ratesof morphological 3). The "primitiveness" of Z. albicollisis pos- changeare possibleexplanations. Tests among sibly corroboratedby the breaststreakings in such alternatives should be pursued at the the first winter plumage, reminiscentof the specieslevel; comparisonsat other levels in condition in M. lincolnii but different from the vertebratesmay be biased, becausetaxoho- remaining three speciesof Zonotrichia.Zono- mists partition variation in different ways in trichiaquerula, which presentlybreeds in Mac- different groups. kenzie, Keewatin, and northern Manitoba Summaryand critiqueof previoustaxonomic (Godfrey1966), differentiated next. The diver- opinions.--Evidencecited for close relation- gent plumageand skeletalmorphology of Z. ships among the taxa discussedhere has not querulahas probablydeveloped in a relatively been criticallyevaluated; a brief review of pre- short span of time. Last, a sisterlineage to Z. vious ideas follows. Mayr and Short (1970) querula, possibly widely distributed across agreedwith Linsdale's(1928) merger of Melo- northern North America, fragmented into Z. spiza into Passerellaand suggestedthat these atricapillain the west and Z. leucophryspre- generawere closelyallied to Juncoand Zono- sumably in the east; subsequent dispersal trichia.Mayr and Shortconsidered Z. albicollis would explain the occurrenceof Z. leucophrys and Z. atricapillato be componentsof a super- across all of North America. speciesand includedthese with Z. leucophrys The/• of Z. capensisto its congeners,0.19, as a speciesgroup. Paynter (1964, 1970) sug- representsfrom 1.0 (c = 5) to 5.0 (c = 26.3) gestedthat Passerellaand Melospizashould be MYBP. This speciesis distributedfrom M•xico mergedwith Zonotrichia,and Shortand Simon to southern South America (Edwards 1974). If (1965)recommended the mergerof Zonotrichia (sensuPaynter) into Junco.Morony et al. (1975) Z. capensisis indeeda memberof a monophy- use the taxonomic scheme of Paynter (1970), letic Zonotrichiaclade, it is an early offshoot and their treatment is used in some recent lit- that has undergone considerable anagenetic erature(e.g. Avise et al. 1980b).Edwards (1974) evolution,as evidencedby the branchlengths and the A.O.U. Check-list (1957)maintain sep- in Fig. 3. Zonotrichia(and Junco) possibly orig- arategenera; see also Parkes (1954). inated in SouthAmerica and subsequentlydis- Paynter(1964) stated that Melospizaand Pas- persed northward. Intraspecificgenic differ- serellahave similar morphologies,song, nests, entiation, however, must be evaluated in Z. eggs,and habits. He statedthat "... after an capensisbefore its systematicposition can be elaboratecomparison of skullosteology, as well resolved.The possibility existsthat Z. capensis [as] externalmorphology and habits, Linsdale 644 ROnERTM. ZI•c [Auk, Vol. 99

(1928)proposed their genericmerger" (Paynter tial, phylogeneticallydistant birds may have 1964:278). Linsdale'sstudy was not elaborate, similar karyotypes. Similarity of karyotypes however. He (1928)presented three skull mea- between Juncoand Zonotrichiais not necessar- surementsfor six subspecieseach of the Fox ily conclusiveof a closephylogenetic relation- and Songsparrows and showedthat they were ship. Karyotypicanalysis, especially with newer basicallysimilar, by inspection.Although Pas- banding techniques(Shields 1980) should be serella and Melospizaoverlap in multivariate exploredin avian phylogeneticstudies. space(Fig. lb), largelybecause of the influence The genic data do not resolvewhether or not of skull characters(Appendix 1), this does not J. hyemalisand Zonotrichiaare nearest rela- a priori point to a closerelationship (also see tives, a hypothesisoffered by previousauthors Figs. la, 2). Similarities in song between P. (citedabove). The b fromJ. hyemalisto Zo- iliaca and M. melodiaare discussedby Martin notrichia(0.205) is similarto the b betweenJ. (1977), and they are not as clear-cutas Paynter hyemalisand Melospiza(0.185); further work is suggested.In any event, the value of songas needed to darify the cladisticrelationships of a "generic character"is uncertain.For exam- these and the other genera studied here. ple, the song of Pipilo chlorurus is more An electrophoretic study by Avise et al. similar to that of Passerella iliaca than the latter (1980b) of 21-22 loci revealed lower levels of is to M. melodia.I doubt that many system- genic divergenceamong the Song Sparrow, atists would favor the genericmerger of Pipilo SwampSparrow, White-throated Sparrow, and and Passerellaon the basisof this similarity in Dark-eyed Juncothan thosereported here. To song. Passerellaand Melospizaare genically compareour studies,I first recalculatedD-val- distinct(/• =. 0.254),and the apparentpheno- ues from the allelicfrequencies given by Avise typic similaritiescited aboveare not indicative et al. (1980b)using Nei's (1978)correction for of a closephylogenetic relationship. sampling error. The major differencebetween The result (Fig. 3) that Z. atricapillaand Z. the originaland correctedD-values of Avise et albicollisare the most divergent members of al. (1980b) was that their D-value for M. geor- Zonotrichia conflicts with Mayr and Short's giana-M. melodia,0.028, was reducedto 0.0007. (1970) daim that these two speciescomprise a Other correctedvalues were lower in general. superspecies.Mayr and Short apparentlyrea- Basedon either original or correctedD-values soned that the essentiallyallopatric breeding of Avise et al. (1980b), the speciesin Melospiza distributions of Z. atricapillaand Z. albicollis and Zonotrichiaappear closelyrelated [D-val- meant that they were sistertaxa (i.e. superspe- ues of 0.086 (corrected 0.053) for M. melodia cies). The genic data, however, show that Z. and Z. albicollisand 0.057 (corrected0.055) for leucophrysis most similar to Z. atricapilla(Fig. M. georgianaand 'Z..albicollis]. In contrast,the 3). values obtained in the present study for the Dickerman(1961) and Short and Simon (1965) same comparisons,0.192 and 0.218, suggesta suggestedthat the apparenthigh frequencyof more distantrelationship. The allelicfrequen- hybridizationbetween Z. albicollisand J. hye- cies reported by Avise et al. (1980b) are very malis indicated a close evolutionary relation- similar to mine (Appendix 2) for the 21 loci in ship. This implies a direct relationshipbe- common between the two studies. The addi- tween frequencyof hybridization and genomic tional 18 loci examined here, however, account similarity (Short 1969) and predictsthat hy- for the differences. The results of Avise et al. brids shouldbe more frequent between (sym- (1980b)seem not to portrayadequately the level patric)congeners, a hypothesisnot supported of divergencebetween Junco, Zonotrichia, and here. Hybridization may occurbetween species Melospiza;they did not examineP. iliaca. pairs of birds that are well separatedphylo- In conclusion,I suggestthat the evidencecit- genetically(Prager and Wilson 1975).There- ed by previous workers in support of various fore, the phyletic significanceof hybrids be- taxonomicrearrangements of Zonotrichia,Jun- tween Z. albicollisand J. hyemalisis unknown. co, Passerella,and Melospizais inconclusiveand Shields (1973) and Shields and Straus (1975) that the proposedgeneric mergers cited above showed that the karyotypesof Juncoand Zo- would obscure significant genetic differentia- notrichiawere similar, althoughShields (1980) tion. Thus, I propose that the four genera be noted that avian karyotypic evolution is con- maintained separately,at least until evidence servative.Therefore, like hybridization poten- is advancedthat corroboratestheir monophyly October1982] EvolutionaryRelationships of Sparrows 645 and degree of differentiation relative to other DUNCAN,T., & R. PHILLIPS(Eds.). 1980. A guide 13asserines. to computingfacilities and applicationsfor the life sciences,vol. 2. Dept. Botany,Berkeley, Cal- ifornia, Univ. Cfilif. ACKNOWLEDGMENTS , --, & W. H. WAGNER, JR. 1980. A com- I am gratefulto the followingpeople for providing parison of branchingdiagrams derived by var- specimens:L. Baptista, G. Butcher,J. Fitzpatrick, S. ious phenetic and dadistic methods. Syst. Bot. Rowher, D. Watt, and D. Willard, and especiallyN. 5: 264-293. K. Johnsonfor collectingthe sampleof Z. capensis EDWARDS,E. P. 1974. A coded list of birds of the from Paraguay.I thank G. F. Barrowcloughand D. world. Sweet Briar, Virginia, Publishedby the S. Wood for permission to cite their unpublished author. data. G. F. Barrowclough,J. E. Cadle,M. M. Frelow, FARRIS,J. S. 1972. Estimating phylogenetic trees N. K. Johnson, A. Larson, J. K. Liebherr, J. A. Mar- from distance matrices. Amer. Nat. 106: 645-668. ten, J. L. Patton, R. D. Sage, and M. F. Smith pro- FITCH, W. M., & E. MARGOLIASH. 1967. Construc- vided helpful advice and assistance.G. F. Barrow- tion of phylogenetictrees. Science155: 279-284. dough, J. E. Cadle, K. W. Corbin, N. K. Johnson,A. FP,XNE, J. 1977. Partial correlation and multivariate Larson, J. A. Marten, J. L. Patton, P. B. Samollow, regression.Pp. 697-710 in BMDP-77 Biomedical and an anonymousreviewer made useful comments Computer Programs P-Series (W. J. Dixon and on the manuscript, and I am grateful to them for M. B. Brown, Eds.). Berkeley, California, Univ. sharingtheir time. Gene Christmandrafted Figs. 1 Calif. Press. and 2b. JeanneDawson typed the manuscript. Fi- GODF•Y, W. E. 1966. The birds of Canada. Natl. nancial support came from the Department of Zool- Mus. Canada Bull. 203: 1-428. ogy and Museum of VertebrateZoology, University GORMAN,G. C., & Y. J. KIM. 1977. Genotypicevo- of California, Berkeleyand from NSF grant DEB- lution in the face of phenotypic conservative- 7920694 to N. K. Johnson. ness:Abudefduf (Pomacentridae) from the Atlan- This paperis dedicatedto the memoryof my friend tic and Pacific sides of Panama. Copeia 1977: and colleague,Laurence M. Conroy. 694-697. GUTI/•RREZ,R. J., ROBERTM. ZINK, & SUH Y. YANG. In press. genic variation, systematic,and bio- LITERATURE CITED geographicrelationships of somegalliform birds. Auk. AMERICAN ORNITHOLOGISTS'UNION. 1957. Check- HANDFORm,P., & F. NoTTEBOHM.1976. Allozymic list of North American birds, fifth ed. Baltimore, and morphologicalvariation in populationsam- Maryland, Amer. Ornithol. Union. ples of Rufous-collaredSparrow, Zonotrichiaca- ArisE, J. C., J. J. SMITH, & F. J. AYALA. 1975. Fensis,in relation to vocal dialects. Evolution 30: Adaptivedifferentiation with littlegenic change 802-817. between two native California minnows. Evo- HARRIS, H., & D. A. HOPKINSON. 1976. Handbook lution 29: 411-426. of enzyme electrophoresisin human genetics. --, J. C. PATTON, & C. F. AQUADRO. 1980a. Amsterdam, North-Holland Publ. Co. Evolutionarygenetics of birds. I. Relationships HELLACK, J. J., & G. D. SCHNELL. 1977. Phenetic among North American thrushesand allies. Auk analysisof the subfamilyCardinalinae using ex- 97: 135-147. ternal and skeletal characters. Wilson Bull. 89: 1980b. Evolutionarygenetics of birds. II. 130-148. Conservativeprotein evolution in North Amer- HENNIG,W. 1966. Phylogeneticsystematics. Chi- ican sparrowsand relatives.Syst. Zool. 29: 323- cago, Illinois, Univ. Illinois Press. 334. HIGHTON,R., & A. LARSON.1979. The geneticre- BAKER,M. C. 1981. Effectivepopulation size in a lationships of the salamandersof the genus songbird:some possible implications. Heredity Plethodon.Syst. Zool. 28: 579-599. 46: 209-218. HUBBARD,J.P. 1973. Arian evolution in the arid- BARROWCLOUGH,G. F. 1980a. Gene flow, effective landsof North America.Living Bird 12: 155-196. populationsizes, and geneticvariance compo- KING, M.-C., & A. C. WILSON. 1975. Evolution at nents in birds. Evolution 34: 789-798. two levelsin humansand chimpanzees.Science 1980b. Geneticand phenotypicdifferentia- 188: 107-116. tion in a woodwarbler (genus Dendroica) hybrid LESSIOS,H.A. 1981. Divergencein allopatry:mo- zone. Auk 97: 655-668. lecular and morphologicaldifferentiation be- , & K. W. Corbin. 1978. Genetic variation tween sea urchins separatedby the Isthmus of and differentiation in the Parulidae. Auk 95: 691- Panama. Evolution 35: 618-634. 702. LINSDALE,J. M. 1928. Variationsin the Fox Spar- DICKERMAN,R. W. 1961. Hybridsamong the frin- row (Passerellailiaca) with reference to natural gillid genera]unco-Zonotrichia and Melospiza. history and osteology.Univ. Calif. Publ. Zool. Auk 78: 627-632. 30: 251-392. 646 ROBERTM. Z•NK [Auk, Vol. 99

MARTIN,D. J. 1977. Songsof the Fox Sparrow. I. --, T. L. BEST, & M. L. KENNEDY. 1978. Inter- Structureof songand its comparisonwith song specificmorphologic variation in kangaroorats in other emberizidae. Condor 79: 209-221. (Dipodomys):degree of concordancewith genic MAYR, E., & L. L. SHORT,JR. 1970. Speciestaxa of variation. Syst. Zool. 27: 34-48. North American birds, a contribution to com- SELANDER,R. K., M. H. SMITH, S. Y. YANG, W. E. parativesystematics. Publ. Nuttall Ornithol. Club JOHNSON,& J. B. GENTRY. 1971. Biochemical 9: 1-127. polymorphismand systematicsin the genus MtCKEVICH,M. F. 1978. Taxonomic congruence. Perornyscus.I. Variation in the old-field mouse Syst. Zool. 27: 143-158. (Perornyscuspolionotus). Univ. TexasPubl. 7103: MORONY,J. J., W. J. BocK, & J. FARRAND,JR. 1975. 49-90. Reference list of the birds of the world. New SHIELDS,G.F. 1973. Chromosomalpolymorphism York, Spec. Publ. Amer. Mus. Nat. Hist. commonto severalspecies of Junco(Aves). Can. NEI, M. 1975. Molecular population geneticsand J. Genet. Cytogenet. 15: 461-471. evolution. Amsterdam, North-Holland Publ. Co. 1980. Avian cytogenetics:new methodol- --. 1978. Estimationof averageheterozygosity ogy and comparativeresults. Acta. 17th Congr. and geneticdistance from a smallnumber of in- Intern. Omithol.: 1226-1231. dividuals. Genetics 89: 583-590. , & N. A. STRAUS.1975. DNA-DNA hybrid- NEvo, E. 1978. Genetic variation in natural pop- ization studies of birds. Evolution 29: 159-166. ulations:patterns and theory. Theor. Pop. Biol. SHORT,L. L., JR. 1969. Taxonomicimplications of 13: 121-177. avian hybridization. Auk 86: 84-105. PARKES,K. C. 1954. Notes on some birds of the , & S. W. SIMON. 1965. Additional hybrids Adirondack and Catskill mountains, New York. of the Slate-coloredJunco and the White-throat- Ann. Carnegie Mus. 33: 149-178. ed Sparrow. Condor 67: 438-442. PAYNE,R. B., & C. J. RISLEY. 1976. Systematicand SIMPSON,S. F., & J. CRACRAFT.1981. The phylo- evolutionary relationshipsamong the herons genetic relationships of the Piciformes (Class (Ardeidae). Misc. Publ. Mus. Zool., Univ. Mich- Aves). Auk 98: 481-494. igan, No. 150. SNEATH,P. H. A., & R. R. SOKAL. 1973. Numerical PAYNTER,R. A., JR. 1964. Generic limits of Zono- taxonomy. San Francisco,W. H. Freeman and trichia. Condor 66: 277-281. Co. 1970. Checklist of birds of the world. Cam- TEMPLETON,A. R. 1981. Mechanismsof specia- bridge, Massachusetts,Mus. Comp. Zool. tion--a populationgenetic approach. Ann. Rev. PRAGER, E. M., & A. C. WILSON. 1975. Slow evo- Ecol. Syst. 12: 23-48. lutionary loss of the potential for interspecific TURNER,B. J. 1974. Genetic divergenceof Death hybridization in birds: a manifestationof slow Valley pupfishspecies: biochemical versus mor- regulatory evolution. Proc. Natl. Acad. Sci. 72: phologicalevidence. Evolution 28: 281-294. 200-204. VAWTER, A. T., R. ROSENBLATT,& G. C. GORMAN. PRESCH,W. 1979. Pheneticanalysis of a singledata 1980. Geneticdivergence among fishes of the set: phylogeneticimplications. Syst. Zool. 28: easternPacific and the Caribbean: support for 366-371. the molecular clock. Evolution 34: 705-711. RA•KOW,R. J. 1976. The origin and evolution of Voous, K. H. 1980. New developments in avian the Hawaiian honeycreepers (Drepanididae). systematics:A summary of results. Acta 17th Living Bird 15: 95-117. Congr. Intern. Ornithol.: 1232-1234. ROBINS, J. D., & G. D. SCHNELL. 1971. Skeletal WAKE,D.B. 1981. The applicationof allozymeevi- analysisof the Ammodramus-Ammospizagrass- denceto problemsin the evolutionof morphol- land sparrow complex: a numerical taxonomic ogy. Pp. 257-270 in Evolution today. Proc. 2nd study. Auk 88: 567-590. Intern. Congr. Syst.Evol. Biol. (G. G. E. Scudder ROGERS,J. S. 1972. Measuresof geneticsimilarity and J. L. Reveal, Eds.). Pittsburgh, Pennsylva- and genetic distance. Univ. Texas Publ. 7213: nia, Hunt Inst. Bot. Documents. 145-153. WILSON, A. C., S.S. CARLSON,& T. J. WHITE. 1977. ROHLF, F. J., & R. R. SOKAL. 1980. Comment on Biochemical evolution. Ann. Rev. Blochem. 46: taxonomiccongruence. Syst. Zool. 29: 97-101. 573639. SARICH,V. M. 1977. Rates, sample sizes, and the WOOD, D. S. 1979. Phenetic relationships within neutrality hypothesisfor electrophoresisin evo- the family Gruidae. Wilson Bull. 91: 384-399. lutionary studies. Nature 265: 24-28. YANG, S. Y., & J. L. PATTON. 1981. Genic variabil- SCHNELL,G. D. 1970. A pheneticstudy of the sub- ity and differentiationin the GalapagosFinches. order Lari (Aves). I. Methods and resultsof prin- Auk 98: 230-242. cipal componentsanalyses. Syst. Zool. 19: 35- 57. October1982] EvolutionaryRelationships of Sparrows 647

APPœI•IDIX1. Correlation of characterswith first four principal components(PC).

Character PC I PC II PC III PC IV

Premaxillalength (PML) 0.8737 0.4598 -0.1193 0.0727 Premaxillalength from narial opening(PNO) 0.7780 0.4669 -0.2078 0.1610 Premaxilladepth (PMD) 0.8549 0.4848 0.0753 0.0243 Nasal bone width (NBW) 0.9708 0.0505 0.0311 0.0181 Postorbitalwidth (POW) 0.5274 -0.7403 -0.3034 -0.1444 Skull width (SKW) 0.9548 0.2064 0.0701 -0.0218 Skull depth (SKD) 0.9559 0.2385 -0.0132 -0.0679 Skull length (SKL) 0.9501 0.2181 -0.1434 -0.1110 Mandible length(MAL) 0.9201 0.3433 -0.1118 0.0389 Minimum mandible length (MML) 0.7269 0.6445 0.0037 0.1339 Mandible depth (MAD) 0.7539 0.6154 0.1252 0.0835 Coracoldlength (COL) 0.9670 -0.1926 0.0789 0.0020 Scapulalength (SCL) 0.9626 -0.1374 0.2128 -0.0224 Scapulawidth (SCW) 0.9754 -0.1243 0.1547 0.0291 Furcularprocess length (FPL) 0.9195 -0.0880 0.3468 0.1263 Sternumlength (STL) 0.9650 -0.1700 0.1402 0.0490 Keel length(KEL) 0.9365 -0.1692 0.2150 0.0644 Sternum width (STW) 0.9660 -0.0618 0.1309 -0.1211 Keel depth (KED) 0.9371 -0.2226 0.2040 0.0858 Posteriorsynsacrum length (PSL) 0.9271 -0.0996 -0.1150 -0.2400 Anterior synsacrumlength (ASL) 0.9241 - 0.1835 - 0.2385 - 0.1813 Synsacrumwidth (SYW) 0.9648 -0.0108 0.0502 -0.2158 Synsacrumminimum width (SMW) 0.9545 -0.1425 -0.0023 0.0240 Femur proximal end width (FPE) 0.9741 -0.0025 -0.0643 -0.2106 Femur minimum width (FMW) 0.9039 -0.1161 -0.3900 0.0954 Femur distal end width (FDE) 0.9577 0.1239 -0.0768 -0.2146 Femur length (FEL) 0.9486 -0.1024 -0.1639 -0.2177 Tibiotarsus width (TTW) 0.8485 -0.2261 -0.4398 0.1509 Tibiotarsuslength (TTL) 0.9241 -0.0159 0.0420 -0.3413 Tarsometatarsuslength (TML) 0.8998 -0.0343 0.0563 -0.3337 Tarsometatarsuswidth (TMW) 0.7567 0.1371 -0.5816 0.1838 Tarsometatarsusdistal end width (TDE) 0.9262 0.1571 -0.1086 -0.2484 Humerus trochanterlength (HTL) 0.9636 -0.0964 0.1290 0.1510 Deltoid crest depth (DCD) 0.9218 -0.2530 -0.0514 0.1973 Humerus distal end width (HDE) 0.9529 -0.0108 0.2546 0.1110 Humerus length (HUL) 0.9806 -0.1524 0.0498 0.0511 Ulna length(ULL) 0.9304 -0.0809 0.2985 0.1678 Ulna width (ULW) 0.8242 -0.2868 -0.1201 0.4536 Carpometacarpuslength (CML) 0.9258 -0.1665 0.3208 -0.0091 Carpometacarpusdepth (CMD) 0.8847 -0.2190 -0.0913 0.3715 Percentageof varianceexplained 82.6% 7.3% 4.1% 3.1% 648 Ro•az M. ZI•c [Auk, Vol. 99 October1982] EvolutionaryRelationships of Sparrows 649