566 Commentaries [Auk,Vol. 104

Responseto A. H. Bledsoeand J. E. Ahlquist et al.

PETER HOUDE •

Bledsoe's(1987) commentary, while basicallysound, sourcescurrently available.My purposehere, as be- does not addressthe methodsthat have been gen- fore (Houde 1987), is to distinguishbetween the part erally used in DNA hybridization studiesof avian of their work that reports results, which is almost relationships.Nonuniformity of ratesof phenotypic certainly correct,and the part of their work that is evolutionnever precluded the useof morphologyin couchedin assumptionsand thereforesusceptible to phylogeneticanalysis, so, obviously,differences in ambiguity in interpretation, as other methods. The rates of DNA evolution do not precludethe use of following criticisms apply: (1) Ahlquist et al. dis- DNA distancedata from phylogeneticanalysis either cussedthe congruence of DNA distancesfor closely (Houde 1987:25; contraBledsoe's account of my po- related and noncontroversialtaxa, i.e. "a duck always sition). The relativeadvantage of DNA distancedata clusterswith the other waterfowl .... "Sibley et al. over phenotypicdata in phenetic analysesis, how- (1987) admitted that they rarely have sufficientdata ever, directly dependenton the extent to which mo- to constructcomplete matricesof nominal taxa, but lecular evolution is clocklike (Ahlquist et al. 1987, they claimed to have "virtually completerepresen- Houde 1987). No one can deny the central role that tation at the level of categoricalhierarchy" (p. 119). the "uniform averagerate" (UAR) of nucleotidesub- Thus,different membersof a taxonmay be substituted stitution has played in the interpretationand pro- to fill the emptycells of incompletedistance matrices. mulgation of avian DNA hybridization data, until The dubiousnessof substitutingspecies in a matrix Sheldon (1987a) and I (Houde 1987) demonstrated is best exemplifed by the procellariiform data set. In evolutionaryrate differences in .We haveyet to 1983Sibley and Ahlquistsupported UAR by claiming see in the publishedrecord how the new "average that Procellariiformesyielded similar distancesto an genomic rates" (AGR) differ from UAR. outgroup,despite different generationlengths. In 1987 Differencesin ratesof evolutionalone will produce they interpreted the samedata set as supportingdif- ambiguity in the clusteringof taxa when phenetic ferent rates of evolution--the result of differences in algorithmsare employed (Sourdis and Krimbas1987). generation length. Are these values the same or are The algorithmsused in all but 3 (Sibleyand Ahlquist they not? If not, then can one speciesbe legitimately 1980; Sheldon 1987a, b) of the roughly 26 studies substitutedfor another in data matricesof categorical using DNA hybridization to estimateavian phylog- hierarchy?If an average value is used instead, then eny were of this type. Bledsoe'sability to correctly how doesthe proportionof speciesof long and short reconstructsix "true" phylogenieswith a nonphe- generations skew this AGR from UAR? netic algorithm inspiresconfidence, but it fails sub- (2) Ahlquist et al. did not addressthe most objec- stantiallyshort of demonstratingstatistically the com- tionabledates of divergences(the origin of passerine plete rate independenceof the least-squaresmethod. families), which are probably25-50% too great. The Nonphenetic algorithms,such as those in the PHYLIP "cuckoo" (Cuculidae) to which they refer is package,do not produceunambiguously unique so- known from only a half of a bone--preciselythe ques- lutions(Lanyon 1985, Hobish 1986), and "treetopol- tionablepractice I warned against(Houde 1987);yet, ogiesthat differ by only a few stepsare not signifi- Ahlquist et al. are of the opinionthat the series of cantly different" (Felsenstein1985: 157). Bledsoewas completeskeletons on which I reported(Houde 1986) able to recognize his correct reconstructionsbecause is "unconvincing." This "cuckoo"does not date the he already knew the "true" phylogenies,but this is divergencesof cuculiform families, becausethe mor- never the case in nature. As it is also clear that no phology of basalCuculiformes is unknown. On the singlemethod of analysisis mostappropriate in all other hand, the fossil limpkin, Badistornis,does tell instances(Hobish 1986),when is leastsquares appro- about the divergenceof finfoots() and priate and when is it not? ConsideringBledsoe's re- limpkins (Aramidae), if Sibley and Ahlquist (1985) suits,least squares is a promisingmethod of analysis are correctthat Old World finfootsare a sistergroup that deservesconsiderably more attention and test- of the neotropicalSungrebe (Heliornis) and limpkin ing. (Aramus).The finfoot morphologyis primitive and I enthusiasticallyendorse Ahlquist et al.'s (1987) the limpkin morphologyis derivedif they are correct, commentaryand their DNA hybridization studiesin but their Aramus-Heliornisdivergence (20 MYA + 10%) general, but I do not think that these authors have is 30% younger than Badistornis,which has been un- fully exploited the analyticaland presentationalre- ambiguouslyinterpreted as a limpkin (Wetmore 1940, Brodkorb 1967, Cracraft 1973, Olson 1985). Becausedifferent groups of birds (and mammals) ' Division of Birds, National Museum of Natural have evolved at different rates, no single temporal History, Smithsonian Institution, Washington, D.C. calibration or single correctioncoefficient will apply 20560 USA. to more than one group of organismsexcept by co- July1987] Commentaries 567

incidence. Thus, Ahlquist et al.'s examplesof the con- ods,"especially in the "categoricallevels at and above gruenceof their data to selectedfossil evidence have families." Sheldon (1987a, b), however, pointed out no bearing on the larger picture of dating avian di- that only delta T•oH,not delta mode and delta Tin,can vergences. be used to measure the divergence of disparate taxa (3) There is no evidence that trogons (Trogonidae), (e.g. families and orders);yet, delta T5oHvalues clear- parrots (Psittacidae),and pigeons (Co]umbidae) are ly violate one of the assumptionsrequired for the either more vagile than passetines(Passeriœormes), clustering of interordinal DNA distance data, viz. barbets (Capitonidae), and cuckoos or had a past homologsexist between all sequencesof heterologous distribution that would have better facilitated their DNA. Homologsof all sequencesclearly do not exist dispersal except that their delta T5oH values of in interordinalhybrids, because the slopeof the melt- trans-At]antic sister taxa are lower. Evolutionary rate ing curve must be extrapolated to the point at which differencesare anotherexplanation, or possiblya con- 50%of the DNA would hypotheticallybe in hetero~ tributing factor, for the low values. duplex form to calculate TsoH.Both molecular and (4) Sib]ey and Ah]quist (1981) first calibratedtheir traditional analysesare subject to uncertaintiesat molecular clock by the hypothesized(Cracraft 1974) higher taxonomic levels. divergenceof ratites,and subsequentlystrengthened (8) Do DNA distance data provide any advantage their calibrationby the addition of other divergence over traditional phenotypic data? Yes; if DNA hy- dates,including apes(Sib]ey and Ah]quist 1984),that bridization were not such a powerful tool, then it seemed to agree. I (Houde 1987) may have been in- would not deservethe severecritical scrutiny I have correct semantical]y,but not in essence,to say that given it. Reproducibility,in both data collectionand they calibrated the avian clock on a primate diver- analysis,is one clear advantageof DNA hybridiza- gence.They advocateda singleDNA molecularclock, tion. Although all the coding information that makes partly basedon primates,which they did not specify up an organism is contained in DNA, counterintui- as avian but applied universally to birds (Sibley and tively, this is not necessarilyanother advantage of Ahlquist 1985). DNA hybridization. Phenotypicexpression is influ- (5) Ahlquist et al. missedmy point when I said, "Is enced by the number of copies of a gene, its pro- ßit DNA or organismsthat is to be classified?"The moters, enhancers, and location in relation to these taxonomiclevels indicated by genetic distance may and other genes(Borst and Greaves 1987).Traditional be so different from that indicatedby morphologyas studies incorporate this information in the form of to obscurediversity. For example,in an unpublished phenotypic characters;such potentially useful infor- poster presentation (19th Intern. Ornithol. Congr., mationcan be invisibleto DNA hybridization,which Ottawa) Sibley and Ahlquist included 9 or 10 of the usessingle copiesof shearedDNA. traditionally recognized avian orders (Mayr and Amadon 1951, Wetmore 1960) in a single order, Ci- LITERATURE CITED coniiformes. (6) It was not my intention to expresslyrefute any AHLQUIST, J. E., A. H. BLEDSOE,F. H. SHELDON, & C. of Ahlquist et al.'s phylogeneticconclusions, only to G. SI•3LE¾.1987. DNA hybridization and avian show what constitutesthe best criteria for drawing systematics.Auk 104: 556-563. such conclusions.Templeton's (1985) complaint that BLEDSOE,A.H. 1987. Estimationof phylogenyfrom Sibley and Ahlquist (1984) did not statisticallyweigh molecular distance data: the issue of variable rates. alternative hypothesesdoes not depend on the ve- Auk 104: 563-565. racity of his own phylogenetichypothesis and statis- BONNER,T. I., R. HEINEMANN, & G. J. TODARO. 1981. tical test. I am, furthermore, baffled by Ahlquist et A geographicfactor involved in the single copy al.'s rationale for excluding Cladorhynchusin their DNA sequenceof primates.Pp. 293-300 in Evo- treatmentof flamingos(Phoenicopteridae; Sibley and lution today (G. G. E. Scudderand J. L. Reveal, Ahlquist 1985) on the grounds that this is not an Eds.).Proc. 2nd Intern. Congr. Syst.Evol. Biol. African , when in the samepaper they included BORST,P., & D. R. GREAVES.1987. Programmed gene several neotropical and Australian taxa (e.g. Aramus, rearrangements altering gene expression. Sci- Heliornis, Pedionomus). ence 235: 658-667. (7) Ahlquist et al. (1987) adopted an adversarial BRODKOR•3,P. 1967. Catalogueof fossilbirds. Part 3, position,broadly criticizing traditional systematics as Ralliformes, Ichthyornithiformes, Charadri- "subjective"and "mysterious;... it couldonly be done iformes. Bull. State Mus. 2: 99-200. by thoseblessed with the power of intuition coupled BROWNELL,E. 1983. DNA/DNA hybridizationstud- with long experience"(Sibley et al. 1987: 113). Yet, iesof muroid rodents:symmetry and ratesof mo- Ahlquist et al. castsimilar aspersionson the relative lecular evolution. Evolution 37: 1034-1051. abilities of other molecular biologists (e.g. Bonner et CRACRAFT,J. 1973. Systematicsand evolution of the al. 1981,criticized by Sibley 1981;Brownell 1983,crit- (class Aves), 3. Phylogenyof the sub- icized by Sibley et al. 1987). Ahlquist et al. further order Grues. Bull. Amer. Mus. Nat. Hist. 151: 1- criticized the "limited resolution of traditional meth- 127. 568 Commentaries [Auk,Vol. 104

ß 1974. Phylogeny and evolution of the ratite tionshipsof the ratite birdsas indicated by DNA- birds. 116: 494-521. DNA hybridization. Pp. 301-335 in Evolution to- FELSENSTEII•,J. 1985. Confidencelimits on phylog- day (G. G. E. Scudderand J. L. RevealßEds.). Proc. enieswith a molecularclock. Syst. Zool. 34: 152- 2nd Intern. Congr. Syst. Evol. Biol. 161. --, & . 1983. Phylogenyand classification HOBISI-I,M. K. 1986. The role of the computer in of birds basedon the data of DNA-DNA hybrid- estimatesof DNA sequencedivergence. Pp. 1-18 ization. Pp. 245-292 in Current ornithologyßvol. in DNA systematics(S. K. Dutta, Ed.). BocaRaton, 1 (R. F. Johnston, Ed.). New York, Plenum Press. CRC Press. ß& . 1984. The phylogeny of the hom- I-IOUr•E, P. 1986. Ancestors of ostriches found in the inoid primates, as indicated by DNA-DNA hy- Northern Hemisphere suggestnew hypothesis bridization. J. Mol. Evol. 20: 2-15. for origin of ratites.Nature 324: 563-565. --, & --. 1985. The relationshipsof some 1987. Critical evaluationof DNA hybridiza- Africanbirds, based on the geneticmaterial, DNA. tion in avian systematics.Auk 104: 17-32. Pp. 115-161 in Proceedingsof the international LA•¾Oi•, S. M. 1985. Molecular perspective on symposiumon African vertebrates:systematics, higher-level relationships in the Tyrannoidea phylogeny, and evolutionary ecology (K. L. (Aves). Syst.Zool. 34: 404-418. Schuchmann, Ed.). Bonnß Zool. Forsch. Mus. A. MAYR, E., & D. AMADOI•. 1951. A classification of Koenig. recent birds. Amer. Mus. Novitates 1496: 1-42. , --, & F. H. S•ELr•O•. 1987. DNA hy- OLso•,I,S.L. 1985. The fossil record of birds. Pp. 79- bridization and avian phylogenetics:reply to 252 in Avian biology, vol. 8 (D. S. Farher, J. R. Cracraft.Pp. 97-125 in Evolutionarybiology, vol. King, and K. C. Parkes, Eds.). New Yorkß Aca- 21 (M. K. Hecht, B. Wallaceß and G. T. Pranceß demic Press. Eds.). New York, Plenum Publ. S•ELr•Oi•, F. H. 1987a. Rates of single-copy DNA SOURDISßJ., & C. KRIMBAS.1987. Accuracyof phy- evolution in herons. Mol. Biol. Evol. 4: 56-69. logenetic trees estimated from DNA sequence 1987b. Phylogeny of heronsestimated from data. Mol. Biol. Evol. 4: 159-166. DNA-DNA hybridization data. Auk 104:97-108. TEMPLETONßA.R. 1985. The phylogeny of the hom- SIB•E¾,C.G. 1981. Comparativestudy of the genetic inoid primates:a statisticalanalysis of the DNA- material, DNA. Pp. 291-292 in Evolution today DNA hybridization data. Mol. Biol. Evol. 2: 420- (G. G. E. Scudder and J. L. Reveal, Eds.). Proc. 433. 2nd Intern. Congr. Syst. Evol. Biol. WETMORE,A. 1940. Fossil bird remains from the Ter- --, & J.E. Ahlquist. 1980. The relationships of tiary deposits in the United States.J. Morphol. the "primitive eaters" (Aves: Passeri- 66: 25-37. formes) as indicated by DNA-DNA hybridiza- 1960. A classification for the birds of the tion. Proc. 17th Intern. Ornithol. Congr.: 1215- world. Smithsonian Misc. Coil. 139: 1-37. 1220. --, & --. 1981. The phylogeny and rela- Received24 April 1987, accepted24 April 1987.