Speciation, Heterochrony, and Genetic Variation in Hispaniolan Palm-Tanagers

SPECIATION, HETEROCHRONY, AND GENETIC VARIATION IN HISPANIOLAN PALM-TANAGERS

MARA A. MCDONALD TM AND MICHAEL H. SMITH 1'3 •SavannahRiver Ecology Laboratory, Drawer E, Aiken,South Carolina 29802 USA, 2Departmentof Zoology,University of Florida,Gainesville, Florida 32611 USA, and 3Departmentsof Geneticsand Zoology, School of ForestResources, and Institute of Ecology, Universityof Georgia,Athens, Georgia 30602 USA

ABSTP,•CT.--Wedocumented levels of genetic variability for two speciesof Hispaniolan palm-tanagers.Significant differences between age classesin levels of geneticheterozygosity were concordantwith an age dimorphismin foragingbehavior and morphologyin Phaeni- cophiluspalmarum; juveniles (H = 0.121) were almosttwice as heterozygousas adults (H = 0.074).Phaenicophilus poliocephalus (H = 0.104)was not characterizedby a distinctage dimorphism in any characterexamined. Although P. poliocephalusresembled juvenile P. palmarum in morphology and behavior, it was not significantly different from either adult or juvenile P. palmarumin levels of genetic variability. Both speciesof Phaenicophiluspossess levels of geneticvariability (9-10%) that are high for birds, and they differ in allele frequenciesand presenceof private alleles, although they are not characterized by fixed allelic differences. Phaenicophiluspoliocephalus was probably derived from smallfounding populations (ca. 50,000- 260,000yr BP),composed mostly of juvenile P. palmarumthat colonizedthe southisland of Hispaniola formed during the Pleistocene.Rapid divergencebetween speciesis consistent with predictionsfrom modelsof heterochronyby paedomorphosisand speciationby a founder event. Received25 August1989, accepted 28 April 1990.

HETEROCHRONY(i.e. shifts in developmental heterochronymay explain this paradoxif spe- patterns)provides a framework for understand- ciation occurs primarily by regulatory gene ing how population processeslead to evolu- changes,as proposed by Gould(1977), although tionarychange (Gould 1977). Alterations in body he did not believe heterochrony to be signifi- size and shape, fecundity, age-structure,and cant in avian evolution. significantchanges in social structurewithin Heterochrony can be achieved in two ways: populationscan result by changingindividual peramorphosis(i.e. terminal deletion) and pae- growth rates or age of sexual maturity (Geist domorphosis(i.e. terminal addition) (Kluge and 1971, Gould 1977, Lawton and Lawton 1986). Strauss 1985). Whereas peramorphosis results Life history characteristics,such as fecundity in the acquisition of novel characters,paedo- and growth rates,are correlatedto levels of ge- morphosisresults in the retention of juvenile neticvariability (Cothran et al. 1983,Mitton and charactersin reproductively capableindividu- Grant 1984),although there is no unifying the- als (Gould 1977, Lawton and Lawton 1986) and ory to predict the geneticconsequences of het- it may be achievedby small changesin regu- erochrony. Differential selectionon various life latorygenes early in development(Larson 1980). history stagesdue to heterochrony can lead to Paedomorphosismay occurby delaying somatic genetic changeswithin and between species, maturation (i.e. neoteny)or acceleratingsexual and it canprovide a basisfrom which to develop maturation (i.e. progenesis).Gould (1977) char- a general model predicting the genetic conse- acterized neoteny as a responseto limiting re- quencesof heterochrony.Furthermore, avian sources(i.e. K-selectedstrategy) and progenesis groups are known to be morphologicallydi- as a responseto plentiful resources(i.e. r-se- verse (Wyles et al. 1983) but genetically con- lected). Only recently has neoteny been equat- servative(Avise and Aquadro 1982)when com- ed with delayed maturation in birds (Lawton pared with other vertebrates. A model of and Lawton 1986, Foster 1987), although delayed maturationhas already been documented * Present address:Division of Biological Sciences, for numerousbird species(Selander 1965,Roh- University of Missouri, Columbia, Missouri 65211 wer et al. 1980, Flood 1984, Hamerstrom 1986). USA. Avian groupsthat are paedomorphicmay rep- 707 The Auk 107: 707-717. October 1990 708 McDONALDAND SMITH [Auk,Vol. 107

resent ideal systemsfrom which to develop a AmericanMuseum of Natural History. The remaining coordinated theory relating ecological con- specimensare deposited in the Florida Museum of straints,life history traits,and geneticvariabil- Natural History. Tissuesamples are depositedin the ity to rapid speciation. LouisianaState University Museum of Zoology. Field identifications were based on crown and chin char- The Neotropicaltanagers are goodcandidates acters(McDonald and Smith in press).Juveniles of for the study of heterochronybecause of their both speciescould be identified in hand by the pres- adaptiveradiation into a number of ecological ence of a yellow wash in the plumage.This plumage niches in the tropics. Moreover, in at least 39 characterwas correlatedwith a gray--instead of red- species,first-year birds can be distinguished brown--iris observedfor adults, gray crown, small from adultsby plumage.Because molt often does gonads,and rictal flangesin six of eight specimens. not occuruntil after the first potentialbreeding Becausedry iceor liquid nitrogenwas not available season,many tanagerscan breed in juvenile (or in Haiti, tissueswere storedin 1.5%buffered 2-phen- subadult) plumage (Isler and Isler 1987). We oxyethanol solution 0.5-4.0 h after collection, and vials were stored in a conventional freezer at -20øC chose to examine the evolution and speciation within a week after collection until September 1985. of Hispaniolanpalm-tanagers using the predic- Tissues were stored at -60øC thereafter. Little or no tions from the model of heterochrony,because degradation,as indicated by mobility changesand the similaritiesbetween the two speciessuggest subbanding,occurred in mostenzymes after 2 weeks one is paedomorphicto the other and may have storage of tissuesin the preservativeat room tem- undergone rapid speciation on a small island peratureand after 4 monthsstorage in a conventional (McDonald 1988). freezer (McDonald MS). Gray-crownedPalm-Tanagers (Phaenicophilus Liver and muscleextracts were groundin the phen- poliocephalus)resemble juvenile Black-crowned oxyethanolsolution. Sampleswere centrifuged for 5 Palm-Tanagers(P. palmarum)in foraging be- s to clearthe supernatantof particulatematter. Tissue havior and morphology.They are also smaller extractswere analyzed using horizontal starch gel in body size, are more social relative to Black- electrophoresis,according to the modified methods of Selander et al. (1971) and Harris and Hopkinson crownedPalm-Tanagers, and are mostlikely de- (1976). rived from them (McDonald and Smith in press). We assayedfor 39 presumptiveloci. Locusdesig- We proposethat age-relatedgenetic differences nations,abbreviations, and buffer conditionsare giv- observedfor P. palmarummay be related to the en in Table 1, except where listed below. Loci were other observedphenotypic differencesin this numberedaccording to the mobility of the products species,and may have relevance to the evolu- (with the most anodal as 1) when two or more iso- tion of P. poliocephalus. zymesappeared on the samegel. Monomorphicloci Our purpose is to describe the patterns of included AAT-2 (analyzed using Amine Citrate 6.2 geneticvariation within and betweentwo palm- and Tris Maleate 7.4 buffers), CK-1, FH-2, aGPD-1, lactate dehydrogenase(LDH)-I and -2, realate de- tanager speciesand to relate these patterns to hydrogenase(MDH)-I and -2, malic enzyme (ME), their possiblemode of speciation. Our specific and SOD-2(all on Amine Citrate 6.2buffer), and gen- objectivesare to describeage-specific genetic eral proteins (GP)-I and -2 (Lithium Hydroxide 8.2 variation within these speciesand relate it to buffer). Alleles were designatedalphabetically, with differencesin behavior and morphology, and A correspondingto the allele with the fastest mi- to quantify the degree of genetic differentia- gratingproduct. No locushad morethan threealleles. tion between these speciesand estimate diver- All loci were scored independently from the fresh gence time. We discussspeciation of these spe- gelsby one or both authors,and then rescoredfrom cies as it relates to isolation, founder effect, and the pictures by both authors. Photographs are ar- genetic variability, and test whether the ob- chivedwith the dataat the SavannahRiver Ecology Laboratory. served genetic variation is consistentwith a General statistical tests were conducted with the paedomorphicderivation of one speciesfrom StatisticalAnalysis System (SAS 1985) or Statistical the other. Packagefor the SocialSciences (SPSS, Norusis 1985). Significancewas set at P -< 0.05; highly significant

METHODS rejectionof null hypothesesoccurred when P -< 0.01. Acceptancelevels for multiple comparisonsinvolv- Collectionswere made of two speciesof Phaenico- ing the samedata set were adjustedto an experi- philusand their hybrids in Haiti from May through mentwide error of P• _< 0.05 (Harris 1975). Tests are September,1985. Hybrid specimensand representa- reportedas two-tailed exceptwhere noted. tives of both parental speciesare depositedin the Allele frequenciesand genetic variability were October1990] GeneticConsequences of Heterochrony 709

measuredby the proportionof heterozygousloci de- migrants per generation was 0.963 (SE 0.011) termined by direct count per individual averaged based on the rare allele method. There were 9 across39 loci (H), the averagenumber of allelesper private allelesin juvenile (n = 8) and 6 private locus,and the percentloci polymorphicwith the cri- alleles in adult (n = 14) P. palmarumat I1 of the terionof a secondaryallele frequency of no morethan 0.01 (Table 1). The averagenumber of loci observed 39 loci. Privatealleles may be expectedto occur per individualvaried becausesome of the samples in the largeradult sample,but not in the smaller weretaken in the absenceof iceor cooltemperatures. juvenile sample. Geneticdata were analyzedwith BIOSYS-1(Swofford Differences in heterozygosity were signifi- andSelander 1981). Using a t-test,we performedsta- cant between juvenile (H = 0.121) and adult P. tisticalanalyses of individual heterozygositiesthat palmarum(H = 0.074) (t = -3.04, df = 19, P• -< were transformedby taking the square-rootof the 0.007). We found no significant differencesin arcsine of the value (Archie 1985). Slatkin's (1985) heterozygosity between age classeswithin P. rareallele model was used to estimategene flow across poliocephalus(H = 0.104) (t = -1.73, df = 18, P• palm-tanagers.Hybrids were excludedfrom this lat- = 0.10) or within the hybrids (t = -1.40, df = ter analysis.Standard error for the geneflow estimate was evaluatedby jackknifing (Lanyon 1987) each of 12, P• = 0.18), although adult P. poliocephalus(H the 31 uniquealleles found in one or the otherspe- = 0.115)were more heterozygousthan juveniles cies. (H = 0.069). Only five juvenile P. poliocephalus To estimatethe relativesize of the foundingpop- and three juvenile hybrids were available for ulations required for colonization and maintenance this test. Juvenile or adult P. palmarumdid not of currentheterozygosity levels in P. poliocephalus,we differ significantly from either P. poliocephalus calculatedexpected heterozygosity (He) for the col- or the hybrids in genetic heterozygosity. onists as follows (Baker and Moeed 1987): Samplingbiases due to taking juvenilesfrom He = (1 - 1/2No)Ho, the same nest are not likely to accountfor the observedheterozygosity difference between age whereNo is the sizeof the foundingpopulation and classesbecause 5 of the 8 juvenile P. palmarum Hois heterozygosityof the founders.Founding pop- were collected with another adult in the same ulationsderived from ancestralP. palmarumwere as- habitat, and these habitats were distributed sumed for the calculation to consist of one of three acrossHaiti. Habitats included cloud forest, dis- groups:all adults,all juveniles, or a mixture of adults and juveniles collected at random. turbed pines, woodland, farms, rural areas,and deserts.One juvenile was collectedalone. Two juvenileswere collectedin the samehabitat and RESULTS couldbe potentialnest mates. Of the 5 juvenile- adult pairs,4 juvenileshad higher levelsof het- Genetic distance(Nei 1978) between P. pal- erozygositythan the adults. Of the 2 juveniles marumand P. poliocephaluswas D = 0.01. No collectedin the samehabitat, 1 ranked highest significantdifferences in geneticheterozygos- for individual heterozygosity(H = 0.143); the ity existedbetween the two parental species(t other ranked secondto the lowest for juveniles = -0.81, df = 40), between P. palmarumand (H = 0.094). hybrids (t = -0.17, df = 34), or between P. Of the 17variable loci in P. palmarum(c•GPD-2 poliocephalusand hybrids (t = 0.54, df = 32) (Ta- was omitted becauseof low samplesizes), only ble I). No fixed allelic differences were found 4 did not have higher levels of heterozygosity betweenthe speciesalthough there were highly in juvenilesthan in adults.Data from the 4 loci significant shifts in five loci between the two (GDH, PGM-I, PepAI, and XDH) that did not species(i.e. AAT-I, DIA-2, FH-I, IDH, and PGD- follow this trend were excludedfrom the jack- I). knife procedurebecause removal of these data We found "private" alleles,defined as those would result in higher juvenile heterozygosi- observedin only one group (Slatkin 1985),at ties and bias the following tests.We jackknifed 21 of the 27 polymorphicloci. Phaenicophilus(Lanyon 1987) the remaining data to evaluate poliocephalushad 15 unique allelesin 63 alleles, single-locuseffects on heterozygositydiffer- and P. palmarumhad 16 in 64. Samplesize for encesbetween age classes.The data for one lo- each specieswas sufficient to detect 56.2% and cus at a time were removed for the 13 variable 74.1%,respectively, of the private alleles in at loci, and heterozygositieswere computedfor leastone individual at the observedfrequencies each age classwith the data for the remaining in the other species.The estimatednumber of 38 loci. Average heterozygosityfor each age 710 MCDONALDAND SMITH [Auk, Vol. 107

TABLE1. Allele frequencies,direct count heterozygosity (H), percentpolymorphic loci (P00•)(common allele --0.99),and meannumber of alleles(A) across27 variablesot 39 enzymeloci. BPA = Adult Phaenicophilus palmarum;BPI = juvenileP. palmarum;GPT = P. poliocephalus;HYB = hybrids;sample sizes are in parentheses in Samplecolumns.

Buffer Locus/ Samplec Name (EC No.)ß pH b Alleles BPA BPI GPT HYB Aspartate aminotransferase(2.6.1.1) AC 6.2 AAT-1 (14) (8) (20) (14) TM 7.4 A 0.143 0.188 0.000 0.036 B 0.857 0.813 0.850 0.929 C 0.000 0.000 0.150 0.036 Aconitate hydratase(4.2.1.3) AC 6.2 ICON-1 (13) (8) (17) (13) A 0.000 0.000 0.029 0.000 B !.000 1.000 0.971 1.000 ICON-2 (12) (6) (16) (12) A 0.000 0.083 0.063 0.042 B 1.000 0.917 0.938 0.958 Catalase (1.11.1.6) TC 8.0 CAT (14) (8) (20) (14) A 0.000 0.000 0.100 0.214 B 0.857 0.875 0.750 0.750 C 0.143 0.125 0.150 0.036 Creatine kinase (2.7.3.2) AC 6.2 CK-2 (14) (8) (16) (13) A 0.000 0.000 0.063 0.038 B 1.000 1.000 0.938 0.962 Diaphorase TC 8.0 DIA-2 (12) (7) (12) (12) (1.6.2.2) A 0.167 0.143 0.000 0.000 B 0.792 0.643 1.000 1.000 C 0.042 0.214 0.000 0.000 (1.6.4.3) DIA-3 (14) (8) (19) (13) A 0.036 0.000 0.000 0.000 B 0.929 0.875 0.921 0.923 C 0.036 0.125 0.079 0.077 Fructosediphosphate aldolase (4.1.2.13) AC 6.2 FDA-1 (8) (6) (15) (8) A 0.000 0.000 0.000 0.063 B 1.000 1.000 1.000 0.938 Fumaratehydratase (4.2.1.2) AC 6.2 FH-1 (13) (8) (16) (13) A 0.885 0.750 1.000 0.962 B 0.115 0.250 0.000 0.038 Glucosedehydrogenase (1.1.1.47) TM 7.4 GDH (12) (6) (17) (12) A 0.042 0.000 0.000 0.042 B 0.958 1.000 0.912 0.917 C 0.000 0.000 0.088 0.042 Alpha glycerophosphatedehydrogenase AC 6.2 aGPD2 (1) (3) (20) (9) (1.1.1.8) A 0.000 0.167 0.000 0.000 B 1.000 0.333 1.000 0.833 C 0.000 0.500 0.000 0.167 Glucosephosphate isomerase (5.3.1.9) AC 6.2 GPI (14) (8) (20) (14) A 0.214 0.188 0.200 0.107 B 0.000 0.000 0.000 0.036 C 0.786 0.813 0.800 0.857 Beta-glucuronidase(3.2. !.31) TC 8.0 •GUS (14) (8) (20) (13) A 0.000 0.188 0.050 0.192 B 1.000 0.750 0.925 0.808 C 0.000 0.063 0.025 0.000 Hexokinase (2.7.1.1) AC 6.2 HK (9) (6) (17) (12) A 0.000 0.000 0.000 0.125 B 1.000 1.000 1.000 0.833 C 0.000 0.000 0.000 0.042 Isocitratedehydrogenase (1.1.1.42) JRP7.1 ICD-1 (14) (8) (17) (14) TC 8.0 A 0.000 0.000 0.118 0.000 B 1.000 1.000 0.853 1.000 C 0.000 0.000 0.029 0.000 October 1990] GeneticConsequences of Heterochrony 711

TABLE 1. Continued.

Samplec Buffer Locus/ Name(EC No.)' pH b Alleles BPA BPI GPT HYB

Mannose phosphateisomerase (5.3.1.8) TC 8.0 MPI (9) (5) (14) (10) EDTA 8.6 A 0.000 0.100 0.000 0.000 B 1.000 0.800 1.000 1.000 c o.ooo O.lOO o.ooo o.ooo Purine nucleosidephosphorylase EDTA 8.6 NP (14) (7) (19) (12) (2.4.2.1) A 0.143 0.214 0.316 0.083 B 0.107 0.214 0.158 0.833 C 0.750 0.571 0.526 0.083 Phosphoglucomutase(2.7.5.1) AC 6.2 PGM-1 (12) (7) (17) (14) A 0.042 0.000 0.000 0.036 B 0.917 1.000 1.000 0.821 C 0.042 0.000 0.000 0.143 PGM-2 (5) (6) (14) (9) A 1.000 1.000 0.929 1.000 B 0.000 0.000 0.071 0.000 PGM-3 (4) (3) (13) (13) A 0.000 0.000 0.038 0.000 B 1.000 1.000 0.962 0.962 C 0.000 0.000 0.000 0.038 Peptidasea (3.4.11) TM 7.4 PEP-A1 (9) (6) (15) (14) A 0.000 0.000 0.000 0.036 B 1.000 1.000 0.967 0.893 C 0.000 0.000 0.033 0.071 PEP-G (13) (8) (16) (14) A 0.038 0.063 0.000 0.143 B 0.962 0.938 1.000 0.857 PEP-A2 (14) (7) (18) (14) A 0.000 0.071 0.000 0.036 B 1.000 0.929 0.889 0.964 C 0.000 0.000 0.111 0.000 PEP-L (14) (7) (16) (14) A 0.071 0.000 0.000 0.000 B 0.929 1.000 0.906 1.000 C 0.000 0.000 0.094 0.000 Phosphogluconatedehydrogenase TM 7.4 6PGD-1 (14) (S) (16) (12) (1.1.1.44) A 0.000 0.000 0.219 0.000 B 1.000 1.000 0.781 1.000 Superoxide dismutase(1.15.1.1) AC 6.2 SOD-1 (7) (6) (lS) (7) A 0.000 0.000 0.167 0.429 B 1.000 0.917 0.833 0.571 C 0.000 0.083 0.000 0.000 Xanthine oxidase (1.2.3.2) TC 8.0 XDH (10) (7) (19) (14) A 0.100 0.000 0.079 0.000 B 0.900 1.000 0.895 0.964 C 0.000 0.000 0.026 0.036 Direct count heterozygositye 0.074 0.121 0.104 0.095 $E 0.011 0.009 0.012 0.011 Mean number of alleles 1.4 1.5 1.6 1.7 % polymorphic loci 30.8 35.9 48.7 51.3 • Enzyme namesand numbersrecommended by the Commissionon BiologicalNomenclature (1973). bAbbreviations for buffersare: AC = Amine Citrate (Claytonand Tretiak 1972);JRP = Tris Citrate 7.0 (Ayalaet al. 1972);TM • Tris Maleate; TC = Tris Citrate 8.0; EDTA = EthylenediamineTetraacetic Acid (Selanderet al. 1971,Harris and Hopkinson 1976).When more than one buffer condition is specified,both were routinely used to assayfor multiple loci. ' Samplesizes of adultsand juvenilesfor P. poliocephaluswere 15 and 5, respectively,and for hybridswere 9 and 3. aPeptidase substrates used were PEP-A1and PEP-A2= L-leucyl-L-alanine;PEP-G = DL-leucylglycylglycine;PEP-L = L-leucyl-L-leucine. ßGenetic variability pooled for adultand juvenile Black-crowned Palrn-Tanagers: H = 0.091;SE = 0.009;,• = 1.6;Poo• = 46.2. 712 McDONALDAND SMITH [Auk,Vol. 107

TABLE2. Mean % heterozygosityfor adult (AD) and juvenile (JUV) Phaenicophiluspalmarum after jack- knife simulation(Lanyon 1987)for 13 variable loci. A t-testwas performedto test the null hypothesis that the average difference acrossall 13 loci between age classesis not significant.Differences were highly significant(t = 24.3, df = 24, P < 0.001).

Locus % hetero- % heterozygosity zygositya,b before after removal removal Samplesize Locus c 20 40 60 80 100

removed AD JUV AD JUV AD JUV SIZE OF FOUNDING POPULATION ACON-2 7.6 12.0 0.0 16.7 12 6 CAT 7.3 11.6 14.3 25.0 14 8 Fig. 1. Expectedheterozygosity (He) of colonists DIA-2 7.0 11.4 25.0 42.9 12 7 basedon founding population size (No) and observed DIA-3 7.2 11.7 14.3 25.0 14 8 heterozygosity(Ho) in the founders.The lower curve FH-1 7.0 10.9 23.1 50.0 13 8 was generatedbased on the assumptionthat Ho was AAT-2 6.6 11.3 28.6 37.5 14 8 contributed by colonistsconsisting only of adult GPI 6.9 11.3 28.6 37.5 14 8 Phaenicophiluspalmarum; the middle curve assumeda /•GUS 7.7 10.9 0.0 50.0 14 8 mixture of adultsand juveniles;the upper curve as- MPI 7.6 11.9 0.0 20.0 9 5 sumed colonistswere all juveniles. The solid hori- NP 6.4 11.0 35.7 57.1 14 7 zontal straight line is the current heterozygosityob- PEP-C 7.7 12.0 0.0 14.3 14 7 PEP-B 7.4 12.1 7.7 12.5 13 8 served in Gray-crowned Palm-Tanagers, P. SOD-1 7.7 12.1 0.0 16.7 7 6 poliocephalus.The dotted straight lines are one and two standarderrors below this, respectively. a Standard errors ranged from 0.010 to 0.013 for adults and 0.007 to 0.013 for juveniles. bProbabilities, after removal, ranged from 0.004 to 0.037 for single locus t-tests. of dispersersto maintain certain levels of het- c Abbreviations for loci in Table 1. erozygositymay be so high as to make the dis- persalmodel unreasonable.Therefore, we used a modification of Crow and Kimura's (1970) classwas then computed across13 combinations equation to estimatethe size of founding pop- of 38 loci. Heterozygositydifferences between ulations for given levels of heterozygosity(Ba- the averagesfor each age classwere testedun- ker and Moeed 1987).Three curveswere gen- der the hypothesisthat the differenceswere not erated and compared with the current levels of due to significantsingle-locus effects. Levels of heterozygosityin P. poliocephalus(H = 0.104) heterozygositywere still highly significantly (Fig. 1). The curve basedon a founding popu- different between age classes(t = -24.33, df = lation of all juvenile P. palmarumhad an ex- 24). Therefore,differences in heterozygositybe- pected heterozygosity(He) = 0.106 at a found- tween age classescould not be attributed to the ing populationsize (No) = 4, where the curve effectsof any single variablelocus in the study. basedon a mixed group of adult and juvenile Moreover, the direction of the difference be- founders had H, = 0.090 at No = 40. This latter tween age classesremained unchanged.Juve- value is more than one standard error below nile P. palmarumwere always more heterozy- the current level of heterozygosityof P. polio- gousthen adultswere, regardlessof which data cephalus.The curve basedon a founding pop- were removed (Table 2). ulation of all adultsreached an asymptoteat H, The manner in which the paedomorphicform = 0.061 and No = 10, more than two standard (i.e.P. poliocephalus)was derived could be im- errors below the current heterozygositylevel portant in determiningits level of geneticvari- of P. poliocephalus. ability. Equalor reducedheterozygosity in the derived form relative to that in the antecedent DISCUSSION would be consistent with a vicariance model, but equal levels in both forms would not nor- Paedomorphosisin palm-tanagers.--Beforewe mally be expectedgiven limited dispersalto a can discussthe evolutionary patterns observed relatively isolated area. The required number in Phaenicophilusunder a model of hetero- October!990] GeneticConsequences ofHeterochrony 713 chrony, the polarity of evolutionary relation- experiencedindividuals that breedat an earlier ships(i.e. which speciesis derivedfrom which) agewould be offsetby fitnessdecreased because must be established.The problem is to detect of intense intraspecificcompetition for scarce whether a terminal addition (peramorphosis)or resources.As a result, inexperiencedindivid- terminal deletion (paedomorphosis)has been ualsin a resource-limitedenvironment may de- made.This may be resolvedby the use of out- lay somaticmaturation and benefit from cryptic group analysis(Kluge and Strauss1985). Out- or deceptive morphology by a decreasein in- group analysisinvolves the choiceof a taxon traspecificcompetition (Rohwer 1978, Lawton (i.e. sistergroup) that is closelyrelated to the and Lawton 1986, Foster 1987). When resources taxa being comparedto determinewhether a are not severelylimiting, theseindividuals may character is common or unique to one or more breed.For example,subadult Northern Harriers of the taxa examined. If a character is shared (Circuscyaneus) breed only when Microtusden- with the sistergroup by one but not the other sitiesare high, but refrainfrom breedingat oth- taxa, the first taxon is most likely antecedentto er times (Hamerstrom 1986). Paedomorphosis the second. can be achievedby two different reproductive Three criteriawere significantto demonstrate strategiesin responseto differential resource the derivation of P. poliocephalusfrom P. pal- availability:neoteny and progenesis(Gould marum.First, neoteny occursin P. palmarumbut 1977).Neotenic species can be characterizedby not in P. poliocephalus.Neoteny is observedin an agedimorphism in morphologyand possibly Piranga,the most likely sistergroup to Phaeni- in behavior. cophilus,based on DNA-DNA hybridizationdata In contrast,progenesis occurs when resources (Sibley in litt0 and on biochemicalevidence may be relatively more abundant.Species col- (McDonald 1988). Second, Phaenicophiluspolio- onizing new habitatrepresent potential for pro- cephalushas five allelesnot sharedwith Piranga, genesisto occur. Under these circumstances, whereasP. palmarumhas only three. Third, the early sexual maturation is no longer con- presenceof a distinctwhite chin againsta gray strained, nor is there strong competition be- throat is unique to P. poliocephalus.In addition, tween age classesfor resources.With attain- P. poliocephalusis smaller, resembles juvenile P. ment of sexualmaturity, somaticdevelopment palmarumin foragingbehavior, and is relatively slowsdown or stops(Gould 1977),which results more socialthan is P. palmarum(McDonald and in individuals with smaller body size, juvenile Smith in press),a tendencypredicted by the morphology,juvenile behaviors associated with retentionof juvenilemorphology (Lawton and juvenile morphology, and a concomitant in- Lawton 1986).The absenceof neoteny,the pres- creasein group behaviors (Geist 1971, Gould enceof a uniqueplumage character, the number 1977, Lawton and Lawton 1986). If the coloniz- of unsharedalleles with the sistergroup, and ing population becomesisolated from the main the similarityin behaviorand morphologyof stock, divergence can occur, with genetic P. poliocephalusto juvenile P. palmarumargue changesprimarily expectedat regulatory loci. stronglyin favorof the derivationof the former Age-relateddifferences in heterozygosity.--Age- from the latter species.Once this polarity is related geneticdifferences may be an expected accepted,the absenceof a blackcrown in adult consequenceof neoteny, asare changesin mor- P. poliocephalus(i.e. a terminal deletion) sup- phologyand behavior,particularly becauseage- portsthe hypothesisthat it is paedomorphicto related differencesare most likely selectedin P. palmarum. environments with limited resources. We found Understanding the importance of ecological age-relateddifferences in P. palmarumbut not constraintsto the model of paedomorphosisfor in the derived paedomorphicform, P. polioceph- palm-tanagersis critical to evaluating their alus (Table 1). Age-classdifferences in single- probablemode of speciation.There are several and multi-locus heterozygosityhave also been assumptionsrelative to the model. Early sexual observedfor grouse,lizards, deer, toads,and maturation is favored when all other factors are mosquitofish(Redfield 1973,Tinkle and Selan- equal. However, limiting resourcesmay impose der 1973, Cothran et al. 1983, Samollow and strongintraspecific competition on individuals Soule 1983,Smith et al. 1989), although higher that attempt to breed. Under theseconditions, geneticvariability is not always found for ju- the relative increase in fitness accrued by in- veniles as in P. palmarum.Increased heterozy- 714 MCDONALDAND SMITH [Auk, Vol. 107 gosity seemsto be favored in age classesunder species(Cothran et al. 1983, Smith et al. 1989), intense differential mortality or selection (Sa- even for Phaenicophilus.Although significantdif- toollow and Soule 1983). Juvenile P. palmarum ferences in heterozygositydid not exist be- (H = 0.121)are significantlymore heterozygous tween age classeswithin P. poliocephalus(P = than adults (H = 0.074) of their own species. 0.10), the trends are reversed from P. palmarum, Selection could account for this difference, but with adult P. poliocephalus(H = 0.115)relatively other explanationsmust be consideredalso. more heterozygousthan juveniles (H = 0.069). Age-related differencesin heterozygosityin Furthermore,levels of heterozygosityare sim- P. palmarummight be spuriousbecause of small ilar in adult P. poliocephalus(H = 0.115) and samplesizes. Because of tl'•esmall samplesize juvenile P. palmarum(H = 0.121). The similarity in this study, the magnitudeof the differences in termsof geneticvariability betweenadult P. in heterozygositybetween age classes had to be poliocephalusand juvenile P. palmarummay be large to yield a probability of <0.007 for rejec- serendipitous, but the pattern across several tion of the null hypothesis.It seemsunlikely character sets (i.e. behavior and morphology; that stochasticprocesses that were due to small McDonald and Smith in press)suggests there samplesize producedsuch large differences.It may be a single underlying causerelated to how is also possiblethat one or a few loci are re- and when Phaenicophilusdiverged. sponsible for the age-related differences, but The genetic divergence between the two higher levels of heterozygositywere observed Phaenicophilustaxa was low (D = 0.01), even for for juvenile P. palmarumfor 13 out of 17 variable avian specieswhere the averageD = 0.044(Bar- loci (Table 2). In addition, removal of the data rowclough 1980). However, Johnsonand Zink for any single locus did not alter the overall (1983)reported D = 0.004for two closelyrelated difference(Table 2). Becausejuveniles were col- sympatricspecies of sapsuckers(Sphyrapicus) that lected acrosssites, a sampling bias due to their were undergoing character displacement and coming from a limited number of nests is not assortativemating. Genetic distanceshould not likely. Age-relateddifference in heterozygosity be useda priori asthe only criterion for species seemsto be a multilocus phenomenon and not recognitionin birds (Hepp et al. 1988).The low due to small-samplebiases. However, this result D for Phaenicophilusis likely an indication of doesnot provide a basisfor understandingthe two recentlydiverged but distinctspecies. There mechanism(s)that generatesthe differences. were no fixed allelic differences between the The differencesin genetic variability across two taxa (Table 1), but there were a significant life historystages may be due to developmental number of private alleles (31) not shared by changesin isozyme patterns, negative assorta- them and significantfrequency shifts in alleles tive mating, or nonrandom dispersalof indi- for 5 loci. It takesat leastone migrant per gen- viduals with different genotypes.These pro- eration to maintain allelic equivalencebetween cessesare not likely to producethe consistent taxa (Crow and Kimura 1970). The significant decreasein heterozygosityfor adults observed difference in frequencies for 5 of 17 variable for 13 loci. The decrease in the number of rare loci and the largenumber of private allelessup- alleles in juveniles comparedwith adults in- portsour contentionthat gene flow is severely dicatesselection is probably operating(Samol- restricted between the two taxa. These charac- low and Soule 1983). Private alleles observed teristicsare consistentwith the specificdesig- only in juvenilesare probablynot completely nation of the two recently diverged Phaenicoph- lost in adults, but may occur at much lower ilus taxa. frequenciesthan in juveniles. Our sample size The estimatedtime of divergence (Nei 1975, was not sufficient to detect these alleles. Selec- Gutierrez et al. 1983, Marten and Johnson 1986) tion may act on both life history stages--first is from 5.0 x 104to 2.6 x 105yr before present increasing, then decreasing genetic variabil- (BP). This range in the estimate corresponds ity-but this hypothesisremains to be tested. remarkablywell to the time of the mostrecent The importanceof age-relateddifferences in interglacial period, when sea levels rose 8-10 genetic variability'extendsbeyond its concor- m some 6.5 x 104yr BP, and multiple times dancewith behavior and morphology in P. pal- throughout the Pliocene and Pleistocene(Pre- marumand maybe relatedto the processof rapid gill and Olson 1981). The rise in sealevel would divergence in P. poliocephalus.Lower genetic have inundated the Cul-de-Sac Plain, which variability is not characteristicof adults in all runs from west to east acrossHispaniola, and October1990] GeneticConsequences ofHeterochrony 715 cleaved it into north and south islands. This subsequentdivergence and with Templeton's plain is presently below sea level; during in- (1980) model of speciationby founder effect. terglacialperiods when glaciersmelted and sea The size of the founder population should be levels roseit would have formed an open water justsmall enough to causea rapid accumulation barrier to gene flow (Pregill and Olson 1981). of inbreeding without a severereduction in ge- Model of speciation.--Wepropose that Phaeni- netic variability (Templeton 1980). These con- cophilusspeciated in allopatry on the two is- ditionsenhance the probabilityof the reorgani- lands. Current distributionsof the two species zation of the genome,primarily for regulatory suggestthat P. poliocephalusarose on or dis- genes(Templeton 1980). Therefore, fixed allelic persed to the south island. If the divergence differencesin palm-tanagersmay not be an ex- were a result of a vicariance event, then P. po- pectedconsequence of speciation.Changes in liocephaluswould be expectedto be similar to P. regulatorygenes are consistentwith a modelof palmarumin having a distinct age dimorphism heterochrony(Gould 1977).Even small changes in behavior, morphology, and genetics,unless in regulatorygenes can effectsignificant•phe- neoteny was derived in P. palmarumafter di- notypic changes(Larson 1980) and provide for vergenceoccurred. A vicariancemodel might the establishmentof isolatingmechanisms. Few explain similaritiesin overall genetic variabil- differences in structural loci and radical shifts ity between the two species,but is inadequate in ecological niches are expected in either to explain the pattern of similarity of P. polio- Templeton's(1980) or Gould's (1977) models. cephalusto juvenile P. palmarumin behavior and Large or multiple founding populationsthat morphology.In contrast,a model of hetero- consistedmostly of adult P. palmarumwould not chrony can explain the lack of an age dimor- have achievedthe high levels of heterozygos- phism in the derived species,the retention of ity, regardlessof the numbersof colonizers(Fig. juvenile behaviorand morphologyin that spe- 1). A small mixed group (<40) of ancestralP. cies,its smaller body size and increasedsociality palmarumadult and juvenile colonizerswould (McDonald and Smith in press),and the reten- have retained sufficientlevels of heterozygosity tion of high levels of genetic variability (Fig. currently observed in P. poliocephalus(Fig. 1). 1). The absenceof a striking age dimorphism Juveniledispersal is commonin birds (Green- in P. poliocephaluscould have resultedif it were wood and Harvey 1982). Flocks of neotenic derived from neotenicP. palmarumafter colo- Brown Jays(Cyanocorax morio) that colonize re- nization to the south island. Low competition cently cleared habitats have lower mean ages and relatively abundant resourceson the south than flocks in the main populations (Lawton island would favor the evolution of progenesis, and Lawton 1985).Geographic isolation, small which is characterizedby earlier sexual matu- numbersof founders,and few founding events ration often at smallerbody size than normally set the stage for rapid speciation(Templeton expected(Gould 1977).Phaenicophilus polioceph- 1980),and paedomorphosisprovides the basis alusis smaller than P. palmarum,as expectedfor for explaining the resemblanceof P. polioceph- progeneticspecies. If the colonizerswere few alusto juvenile P. palmarumacross several char- in number, then genetic variability might be actersets and a mechanism(via progenesis)for expectedto have declinedbecause of founder rapid populationgrowth after founding.An al- effect (Crow and Kimura 1970). Both species ternative explanation based on a vicariance have surprisinglyhigh levelsof heterozygosity modeldoes not explainadequately similar lev- (H = 9-10%) for islandspecies (Nevo et al. 1984, els of genetic variability in both species,the but see Yang and Patton 1981). A decline in absenceof an age dimorphism in P. polioceph- heterozygositycould be avoidedby rapid pop- alus,few structuralgene differences,and the ulation growth after colonization (likely for resemblanceof adult P. poliocephalusto juvenile progeneticspecies), a largenumber of founders, P. palmarum. and multiple invasions. We conclude that neoteny in P. palmarumis The simplestexplanation for the current sit- expressedboth behaviorally and morphologi- uation is that heterozygositywas not reduced cally, and the age dimorphism in these char- on the south island in ancestralP. poliocephalus. actersis congruentwith observedheterozygos- Multiple invasionsor large populationsof col- ity differencesbetween age classes. I-Iispaniolan onizerscould explain the retention of high lev- palm-tanagershave recently diverged, most els of heterozygosity,but are inconsistentwith likely during the Pleistocene,when higher sea 716 MCDONALDAND SMITH [Auk,Vol. 107

levelsseparated Hispaniola into north and south populationsof Common Mynas (Acridotherestris- islands. Phaenicophiluspoliocephalus is distinct tis). Evolution 41: 525-538. from P. palmarumbased on the distribution of BARROWCLOUGH,G.F. 1980. Geneticand phenotypic private alleles. Gene flow between speciesis differentiationin a wood warbler (genusDen- low. Phaenicophiluspoliocephalus resembles ju- droica)hybrid zone. Auk 97: 655-668. CLAYTON,J. W., & D. N. TRETIAK. 1972. Amine-citrate venile P. palmarumin behavior and morpholo- buffersfor pH control in starchgel electropho- gy. Higher juvenilegenetic variability in P. pal- resis. J. Fish. Res. Bd. Can. 29: 1169-1172. marum, combined with a greater amount of COMMISSION ON BIOLOGICAL NOMENCLATURE. 1973. juvenile dispersal and the high genetic vari- Enzymenomenclature. Amsterdam, Elsevier. ability observedin P. poliocephalus,is consistent COTHRAN, E.G., R. K. CHESSER,M. H. SMITH, &: P. E. with the derivationof P. poliocephalusfrom small JOHNS.1983. Influencesof genetic variability ' founding populationsthat consistedmostly of and maternalfactors on fetal growth in white- juvenile P. palmarumon the south island of His- tailed deer. Evolution 37: 282-291. panioladuring the Pleistocene.The generality CROW,J. F., & M. KIMURA. 1970. An introduction to of this model of speciationshould be investi- population genetics theory. New York, Harper and Row. gated for other groups,particularly the tana- FLOOD,N.J. 1984. Adaptive significanceof delayed gers. plumage maturation in male Northern Orioles. Evolution 38: 267-279. ACKNOWLEDGMENTS FOSTER,M. S. 1987. Delayed maturation, neoteny, and socialsystem differences in two manakinsof We thank Paul Leberg, Margaret Mulvey, James the genus Chiroxiphia.Evolution 41: 547-558. Novak, Margaret Ptacek,Richard Sage, and Robert GEIST,V. 1971. Mountain sheep:a study in behavior Zink for commentson this manuscript;Jan Hinton, and evolution.Chicago, Univ. of ChicagoPress. Pat Davis, and JoAn Lowery for help with the manu- GOULD,S.J. 1977. Ontogenyand phylogeny.Cam- script; and the Organization of American States, bridge, Massachusetts,Belknap Press. Chapman Memorial Fund of the American Museum GREENWOOD,P. J., & P. H. HARVEY. 1982. The natal of Natural History, and SigmaXi Grants-in-Aid.We and breedingdispersal of birds.Annu. Rev. EcoL are indebted to JosephFelix, Jean-EdnerFrancois, Ed- Syst. 13: 1-21. mond Magny, Florence Sergile, Robert Cassagnol, GUTIf:RREZ,R. J., R. M. ZINK, & S. Y. YANG. 1983. Santha Faiia, Scott Faiia, Jean-Phillipe Audinet, Ste- Genic variation, systematic,and biogeographic phan Dix, Glen Koehler, SusanaMolnar, JoyceFlores, relationshipsof somegalliform birds. Auk 100: Jim Keith, Jim Talbot, Ragnat Arneson, Roland Roi, 33-47. and the governmentof Haiti for their supportin Hai- HAMERSTROM,F. 1986. Harrier: hawk of the marshes. ti. This publication is based on work performed in Washington,D.C., SmithsonianInst. Press. the LaboratoryGraduate Participation Program at the HARRIS, H., & D. A. HOPKINSON. 1976. Handbook of Savannah River Ecology Laboratory under contract enzymeelectrophoresis in human genetics.Am- numbers DE-AC05-76OR00033 and DE-AC09- sterdam, North Holland Publ. Co. 76SROO-819between the U.S. Departmentof Energy HARRIS,R.J. 1975. A primer of multivariate statistics. and Oak Ridge AssociatedUniversities and the Uni- New York, Academic Press. versity of Georgia'sInstitute of Ecology,respectively. HEPp, G. R., J. g. NOVAK, K. T. SCRIBNER,& P. W. STANGEL.1988. Geneticdistance and hybridiza- LITERATURE CITED tion of Black Ducksand Mallards:a morph of a different color? Auk 105: 804-807. ARCHIE,J. W. 1985. Statisticalanalysis of heterozy- ISLER,M., & P. ISLER.1987. The tanagers:natural gosity data: independent sample comparisons. history, distribution and identification. Wash- Evolution 39: 623-637. ington, D.C., Smithsonian Inst. Press. AvISE,J. C., & C. F. AQUADRO.1982. A comparative JOHNSON,N. K., & R. M. ZINK. 1983. Speciationin summaryof genetic distancesin the vertebrates. sapsuckers(Sphyrapicus): I. Genetic differentia- Evol. Biol. 15: 151-185. tion. Auk 100: 871-884. AYALa, F. J., J. R. POWELL,M. L. TRACEY,C. A. MouRAo, KLUGE,A. G., & R. E. STRAUSS.1985. Ontogeny and & S. PEREZ-SALAS.1972. Enzyme variability in systematics.Annu. Rev. Ecol. Syst. 16: 247-268. the Drosophilawillistoni group. IV. Geneticvaria- LANYON,S.M. 1987. Jackknifingand bootstrapping: tion in natural populationsof Drosophilawillistoni. important "new" statisticaltechniques for orni- Genetics 70: 113-139. thologists.Auk 104: 144-146. BAKER,A. J., & A. MOEED. 1987. Rapid genetic dif- LARSON,A. 1980. Paedomorphosisin relation to rates ferentiation and founder effect in colonizing of morphologicaland molecularevolution in the October1990] GeneticConsequences of Heterochrony 717

salamanderAneides fiavipunctatus (Amphibia, ROHWE•,S. 1978. Passerinesubadult plumage and Plethodontidae). Evolution 34: 1-17. the deceptive acquisition of resources:test of a LAWTON,M. F.,& R. O. LAWTON.1985. The breeding critical assumption.Condor 80: 173-179. biologyof the Brown Jayin Monteverde,Costa ßS. D. F•E•VELL,& D. M. NILES. 1980. Delayed Rica. Condor 87: 192-204. maturation in passerineplumages and the de- --, & -- 1986. Heterochrony, deferred ceptive acquisition of resources.Am. Nat. 115: breeding,and avian sociality.Curr. Ornithol. 3: 400-437. 187-222. SAMOLLOW, P. B., & M. E. SOULE. 1983. A case of MARTENßJ. A., & N. K. JOHNSON. 1986. Genetic re- stressrelated heterozygote superiority in nature. lationshipsof North Americancardueline finch- Evolution 37: 646-649. es. Condor 88: 409-420. SASINSTITUTE, INC. 1985. SASuser's guide: statistics McDoN^Lr•,M.A. 1988. The significanceof hetero- version 5 edition. Cary, North Carolina, SAS In- chrony to the evolution of Hispaniolanpalm- stitute, Inc. tanagers, genus Phaenicophilus: behavioral, $ELANDER,g.K. 1965. On mating systemsand sexual morphologicaland geneticcorrelates. Ph.D. dis- selection. Am. Nat. 99: 129-141. sertation. Gainesville, Univ. Florida. , g. H. SMITH, $. Y. YANG, W. E. JOHNSONß&: J. , & M. H. SMITH. In press. Behavioral and B. GENT•¾.1971. Biochemicalpolymorphism and morphologicalcorrelates of heterochrony.Auk. systematicsin the genusPeromyscus. I. Variation MITTON, J. B., & M. C. GRANT. 1984. Associations in the oldfield mouse (Peromyscuspolionotus). amongprotein heterozygosity,growth rate, and Pp. 49-90 in Studiesin genetics,vol. 6 (Marshall developmental homeostasis.Annu. Rev. Ecol. R. Wheeler, Ed.). Univ. Texas Publ. 7103. Syst. 15: 479-499. SLATKIN,M. 1985. Rarealleles as indicatorsof gene N•, M. 1975. Molecular populationgenetics and flow. Evolution 39: 53-65. evolution.Amsterdam, No. HollandPublishing SMITH, M. H., K. T. SCRIBNER,J. D. HERNANDEZ,& M. Co. C. WOOTEN. 1989. Demographicßspatial and ß 1978. Estimationof averageheterozygosity temporalgenetic variation in Gambusia.Pp. 235- and geneticdistance from a small number of in- 287 in Ecologyand evolutionof livebearerfishes dividuals. Genetics 89: 583-590. (Poecilidae) (G. K. Meffe and F. F. Snelson Jr., NEvo, E., A. BEILES,& R. BEN-SCHLOMO. 1984. The Eds.).Edglewood Cliffs, New Jersey,Prentice Hall. evolutionary significanceof genetic diversity: SwoF•o•r•, D. L., & R. B. SELANr•ER. 1981. BIOSYS- ecological,demographic, and life historycorre- 1: a computerprogram for the analysisof allelic lates. Pp. 13-213 in Lecture notes in biomathe- variation in genetics.Urbana, Univ. Illinois. matics53 (G. S. Mani, Ed). New York, Springer- TEMPLETON,A.R. 1980. The theory of speciationvia Verlag Publ. the founder principle. Genetics94:1011-1038. NORthsis,M. J. 1985. Advancedstatistics guide to TINKLE,D. W., & R. K. SELANr•E•.1973. Age-depen- SPSS-X. SPSS, Inc. New York, McGraw-Hill. dent allozymicvariation in a natural population PRECILL,G. K., & S. L. OLSON.1981. Zoogeography of lizards. Blochem. Genet. 8: 231-237. of West Indian vertebrates in relation to Pleis- WYLES, J. F., J. G. KVNKEL, & A. C. WILSON. 1983. toceneclimatic cycles. Annu. Rev.Ecol. Syst. 12: Birds, behavior, and anatomical evolution. Proc. 75-98. Natl. Acad. Sci. USA 80: 4394-4397. REr•FIELr•,J. A. 1973. Demographyand geneticsin YANG,$. Y., & J. L. PATTON.1981. Geneticvariability colonizingpopulations of Blue Grouse(Dendra- and differentiationin the Galapagosfinches. Auk gapusobscurus). Evolution 27: 576-592. 98: 230-242.