Phylogenetic Patterns in the Trochilidae
Total Page：16
File Type：pdf, Size：1020Kb
The Auk 115(1):105-118, 1998 PHYLOGENETIC PATTERNS IN THE TROCHILIDAE JOHN A. GERWIN•'2 AND ROBERTM. ZINK L3 •Museumof NaturalScience and Department of Zoologyand Physiology, LouisianaState University, Baton Rouge, Louisiana 70803, USA ABSTRACT.--Althoughmany aspectsof hummingbirdbiology have been studied, few re- centanalyses of higher-levelsystematic relationships exist. Based on morphology, it hasbeen hypothesizedthat the Trochilidaeincludes six majorclades. We used starch-gel electropho- resisto constructand testphylogenetic hypotheses for representativesof the six clades,us- ing two speciesof swifts(Apodidae) as outgroups. Of 45 loci scored,38 werepolymorphic. The averageNei's geneticdistance (D) among14 hummingbirdtaxa was 0.625;D averaged 1.61between the swiftsand hummingbirds.These distances are large,and are consistent with othernonpasserine groups, suggesting that hummingbird taxa are relativelyold. Phy- logeneticanalyses generally were consistentwith the hypothesisthat hermitsare a sister groupto all othertrochilines. Within the Trochilinae, two broad groups are recognized, here calledtrochiline-A and B, whichcorrespond to the morphologicallydetermined "primitive" and"advanced" trochiline groups of Zusiand Bentz (1982). Androdon aequatorialis is genet- ically distinctbut generallyaligns with the trochiline-Agroup. Within the trochiline-B group,four radiationshypothesized by Zusi (pers.comm.), here called Bee, Amazilia ("Em- eralds"),Andean, and High Andean,were corroboratedby our analyses.Our distanceanal- ysissuggests a phylogeneticpattern consistent with that derivedfrom Sibley and Ahlquists' (1990)and Bleiweisset al.'s(1997) DNA-DNA hybridizationstudies. Received 31 October1996, accepted20 June1997. THE HUMMINGBIRDS(TROCHILIDAE) form one DNA-DNA hybridization(Sibley and Ahlquist of thelargest bird families, with approximately 1990;Bleiweiss et al. 1994,1997). Most of these 325 species.Although many aspectsof hum- studieswere not comprehensiveat the levelof mingbirdbiology have been studied, few high- the family.However, the studyby Bleiweisset er-levelstudies of systematicrelationships ex- al. (1997),which involved 26 species,provided ist. Early taxonomists(Gould 1861, Boucard a molecularphylogenetic hypothesis for the 1895, Simon 1921, Peters 1945, Zimmer 1950- majorclades in the family.We report an allo- 1953) providedspecies descriptions and gen- zymic surveyof major groupsof humming- eraldetails of geographicvariation in mostspe- birdsdesigned to testphylogenetic hypotheses cies,but they did not explicitlyidentify system- derived from previous morphologicaland aticrelationships. Recently, several approaches DNA-DNA hybridizationanalyses. havebeen used to identifyphylogenetic pat- Our study was based in part on Zusi and ternswithin parts of this family: comparative Bentz's (1982) hypothesis of higher-level analysisof vocalizationsand matingbehavior groupsin the Trochilidae.They identified four (Schuchmannunpubl. data), external morphol- major groupings:hermits (Phaethorninae), ogy and ecology (Graves 1980, 1986; Stiles "primitive"trochilines, and two groupsof "ad- 1983, 1996; Hinklemann 1989), comparative vanced" trochilines.The terms primitive and myology (Zusi and Bentz 1982), protein elec- advanced are reserved for discussion of char- trophoresis(Gerwin and Zink 1989, Gill and actersand are misleadingwithout an in-depth Gerwin 1989), mitochondrialDNA sequences phylogeneticanalysis; we refer to the two (Hernandez-Banoset al. unpubl. data), and groupsas trochiline-A (Zusi and Bentz's [1982] primitive group) and trochiline-B(their ad- 2 Present address: North Carolina State Museum of vancedgroup). The widespread notion that the NaturalSciences, P.O. Box 29555, Raleigh, North Car- hermitsrepresent the "primitive" humming- olina 27626,USA. E-mail:[email protected] birds is incorrect in one sense: if these are sister 3Address correspondence to this author.Present address:James Ford Bell Museum, University of clades,each is by definitionthe sameage. Phy- Minnesota, 100 Ecology Building, St. Paul, Minne- logeneticanalysis is requiredto showthat the sota 55108, USA. E-mail: [email protected] hermitclade retains a disproportionatenumber 105 106 GERWINAND ZINK [Auk, Vol. 115 of plesiomorphiccharacters relative to the sis- (1997)refer to the Beegroup and the hermits.Blei- ter group of hummingbirdsbefore it would be weiss et al. (1997) further referred to a Mountain "primitive" in any sense.Also, within trochi- Gemgroup, of whichwe had no representatives.For lines,if thereare onlytwo groups,neither can simplicity,we useZusi's names, except for usingEm- erald in place Amazilia, becauseconfusion results be primitive in a phylogenetictree. From his fromnaming a groupafter one of thegenera in it. We work on comparativemyology, using the swifts includedtwo speciesof swifts (Apodidae;Reinarda (Apodidae) as the outgroup, Zusi (pers. squamata,Chaetura cinereiventris) that servedas out- comm.)suggested as a workinghypothesis: (1) groups(see Appendix 1). Althoughinferences based that the trochiline-Agroup included Androdon, on low numbersof specimenshave been criticized Schistes,Colibri, Doryfera,and close relatives; (Archieet al. 1989),in our study geneticdistances and (2) that four major cladesexist within the weresufficiently high that our patterns likely are ro- trochiline-Bgroup: Bee, High Andean,Andean, bust to small samplesizes of individuals.Most spec- and Amazilia ("Emeralds").By selectingat imens were collectedover severalyears during ex- leasttwo representativesof eachmajor group peditionsto variousregions of the Neotropics.Sam- ples of tissueswere preservedin liquid nitrogenin to reflect hummingbird diversity,our phylo- the field until transferred to the Louisiana State Uni- geneticanalysis of allozymevariation provides versity Museum of Natural Sciences(LSUMNS), a higher-levelphylogenetic hypothesis that can wherethey were storedat -70øC (seeJohnson et al. be comparedwith thosegenerated from other [1984] for details of collection and preservation data sets. methods).Voucher specimens (study skinsand tis- Most modern molecularstudies of phyloge- suesamples) are housedat the LSUMNS. ny rely on DNA sequences(Hillis et al. 1996). Samplesof pectoraland heart muscleand liver (to- However,phylogenetic hypotheses based on al- tal volumeof tissuewas approximately0.5 cc)were lozymesand mitochondrialDNA oftenare to- placedin 0.8 mL of grinding buffer consistingof 10 pologically consistent(Zink and Avise 1990, rng NADP and 100 }xLof 2-mercaptoethanolin 100 Zink and Dittmann 1991, Zink et al. 1991, Zink mL distilled water (Richardsonet al. 1986) and ground for 10 to 15 s using a Tekmar Tissuemizer. and Blackwell1996). Thus, althoughthere are Thesecrude homogenates were then centrifugedat reasonsfor preferringDNA data for phyloge- 36,000 x g for 30 min, and the supernatantwas netic inferences(Hillis et al. 1996), allozyme- storedat -70øC.Six aliquots of 20 }xLof eachsample basedphylogenetic inferences are valuablein werestored separately and usedfor the first sixgels, studies of congruenceof tree topologiesin- and the rest of the homogenatewas storedin indi- ferred from different data sets(Swofford et al. vidual vials. Electrophoreticprocedures followed 1996).In our study,we comparetree topologies Selanderet al. (1971),Harris and Hopkinson(1976), from allozymes and DNA-DNA hybridization Johnsonet al. (1984) and Richardsonet al. (1986) (Bleiweisset al. 1997).Although both are based with slight modifications(available from author). on informationfrom the nucleargenome, each Variousgel-buffer combinations were used to opti- mize the resolutionof bandingpatterns (Appendix is likely an independentestimate of phyloge- 2). netic relationships.Hence, congruence of tree Forty-five presumptivegenetic loci were scored topologiescan be used as a measureof confi- (Table1), with allelescoded in referenceto their mo- dencein the phylogenetichypothesis. bility from the origin.The most cathodal alleles were coded"a,"with subsequentlyfaster alleles coded as MATERIALS AND METHODS "b," "c," and so on. Multiple isozymesat a locus were alsocoded by mobility.The mostanodally ap- We used standardhorizontal starch-gel electro- pearingisozymes were codedas "1," with the more phoresisto investigatepatterns of geneticvariation cathodalisozymes coded as "2," "3," and so on. Ac- among 14 taxa (Zusi pers. comm.):Glaucis and Phae- ronymsfor loci follow Harris and Hopkinson(1976). thornisrepresent the hermit, or phaethornineline; We enteredindividual genotypes into the computer Androdon,Colibri, Schistes, and Doryferathe trochi- program BIOSYS-1 (Swofford and Selander 1981), line-A group; Acestruraand Selasphorusthe "Bee" which produceda table of allelic frequencies,Nei's line; Amazilia and Campylopterusthe "Amazilia" (1972, 1978) and Rogers' (1972) geneticdistances, (Emerald) group; Aglaeactisand Coeligenathe "An- Cavalli-Sforzaand Edwards' (1967) chord distance, dean" clade; and Metallura, and Oreotrochilusthe and four UPGMA phenograms(Sheath and Sokal "high Andean"clade. Bleiweiss et al. (1997)renamed 1973) derived using thesefour distancemeasures. these clades Mangoes (Trochiline-A), Emeralds Becausethere is no consensuson tree-buildingmeth- (Amazilia group), Brilliants (Andean), and Co- ods, we usedboth distanceand discrete-locusap- quettes(high Andean);both Zusi and Bleiweisset al. proachesand comparedresults. Distance-Wagner January1998] TrochilidaePhylogeny 107 TABLE1. Allelicfrequencies for variableloci. Numbers in parenthesesare frequencies