Monophyly of the Piciformes: a Reply to Olson

MONOPHYLY OF THE PICIFORMES: A REPLY TO OLSON

ROBERTJ. RAIKOW1 AND JOELCRACRAFT 2 •Departmentof BiologicalSciences, University of Pittsburgh,Pittsburgh, Pennsylvania 15260 USA and Carnegie Museumof Natural History, Pittsburgh,Pennsylvania 15213 USA, and 2Departmentof Anatomy,University of Illinoisat the MedicalCenter, Chicago, illinois 60680 USA and FieldMuseum of Natural History,Chicago, illinois 60605 USA

ABSTP•CT.--Olson(1983) questioned the hypothesisthat the order Piciformesis monophyletic and suggestedinstead that each piciform suborder is allied with a nonpiciform group. His attempt to discreditthe synapomorphiesjoining the Galbulae and Pici is refuted by correctionsof his interpretationsof previous work. The Piciformes share a complex derived hindlimb morphologyinvolving zygodactyly,a tripartite flexorhallucis longus, and Type VI flexortendons. Olson's argument for polyphylycombines inadequate data with the inappropriatetechnique of equatinggeneral overall similarity with affinity.His hypothesis is thereforerejected. Problems concerning fossil taxa are also discussed.Based on current information,we believethat a monophyleticorigin of the Piciformesremains the hypothesis of choice.Received 18 August1982, accepted18 October1982.

SWIERCZEWSKIand Raikow (1981)and Simp- more primitive than that of the Pici, which is son and Cracraft (1981) studied the phyloge- more "specialized," and thereforethat mono- netic relationshipsof the Piciformes.Both con- phyly requiresthe conditionin the Pici to have cluded that the order is monophyleticand that evolved from that in the Galbulae. Here, he it contains two monophyletic suborders, the confusessister-group relationship with ances- Galbulae (Bucconidae and Galbulidae) and Pici tor-descendentrelationship. We did not pro- (Capitonidae, Ramphastidae, Indicatoridae, pose that the Pici evolved from the Galbulae and Picidae). Olson (1983) criticizesthe hy- (higher taxa cannotbe ancestors).Olson asks pothesisof monophyly. First, he questionsthe "... why would such a transformationtake argumentssupporting this view. He then pre- place?Once a group of birds has becomeper- sents an alternative hypothesis that the Gal- manently zygodactyl,is it possibleto become bulae are most closely related to the Coracii more zygodactyl?"This question manifestsa and the Pici to the Passeriformes.Finally, he confusion. One group is not "more zygodac- criticizes Simpson and Cracraft'sdiscussion of tyl" than the other;both are describedby this certain fossil birds. After carefullyexamining term. It may be expected,however, that dif- Olson's critique, we believe that monophyly ferenceswill accumulatein separatelyevolving remains the more strongly supportedhypoth- lineagesafter they have split; it is this process esis. The possiblevalue of these discussions of evolutionary divergence or charactertrans- goesbeyond the questionof a singlebranching formation that gives rise to the hierarchical point in a phylogeny,as it providesan oppor- structureof organicdiversity. We suggestthat tunity to compareand contrasttwo profoundly the zygodactylconditions of the Galbulae and different approachesto systematicanalysis. Pici are homologousbecause other characters (see below) corroboratethe unity of the Pici- formes. PICIFORM MONOPHYLY M. flexorhallucis longus.--Swierczewski and Z!/godact!/l!/.--Olsonsuggests that zygodac- Raikow (1981)reported that the flexor hallucis tyly aroseindependently in the Pici and Gal- longusmuscle (FHL) shows a derived condi- bulae becauseSteinbacher (1935) found differ- tion in its origin by a certainpattern of three ences in their tarsometatarsi. He does not heads. In most birds there are one or some- consider the possibility that the differences times two femoral heads, but the Piciformes aroseafter the two groupsdiverged from a zy- have as well an extensiveorigin from the fib- godactylouscommon ancestor. Olson consid- ula. Olson dissectedone capitonid (Trachy- ers that the tarsometatarsus of the Galbulae is phonus;Pici) and one bucconid(Hypnelus; Gai- 134 The Auk I00: 134-138.January 1983 January1983] Monophylyof thePiciformes 135

FIlL MED each of two species.In contrast,Swierczewski INT LAT dissectedbetween one and four specimenseach of 45 species, and concluded unequivocally (1977: 58) that "Three heads of origin were foundin allspecies studied herein." Nevertheless, one of us (R.J.R.) has, in re- sponseto Olson'scomments, dissected the limb of Colaptes auratus (Picidae), Bucco tamiata (Bucconidae), and Galbula dea (Galbulidae). Like Olson, Raikow confirmed in Colaptesthe three heads previously described(Fig. 1). In Hypnelus, Olson "... could not detect any separateheads of origin . . ." and questioned whether "... the nature of the origin of this muscle is homologousbetween the Galbulae and the Pici or even that it can really be said to have three heads in the Galbulae." The dissection of Galbula and Bucco has fully con- firmedour previousreport that the threeheads, including the large fibular head, are presentin the pattern described (Fig. 1). This corrobo- ratesthe hypothesisof piciformmonophyly and refutes Olson's claim that the condition is not found in both piciform suborders. Type VI flexor tendons.--Did this condition Fig. 1. Lateralviews of the regionof the knee and upper crus in three speciesof piciform birds to show arise oncein the Piciformesor separatelyin the Galbulae and Pici? Olson states that it is the modeof originof M. flexorhallucis longus (FHL). In all Piciformesthe musclearises by three separate "probably a convergentsimilarity," but offers heads, medial (MED), intermediate (INT), and lateral no evidence. He suggeststhat the Galbulae (LAT), all of which lie medial to the tendon of M. would be difficult to identify with their feet cut iliofibularis (IL. FIB.). The representativeforms il- off, but fortunately all of our specimenspos- lustrated are Colaptesauratus (Picidae; Pici), Bucco sessedfeet. (Actually, the relevant structures tamiata (Bucconidae; Galbulae), and Galbula dea are legs, not feet, as the FHL belly lies in the (Galbulidae;Galbulae). The drawingswere madewith crus.) Olson feels that such characters do not a camera lucida and dissectingmicroscope and are not to scale. justify ordinalrank, but we do not agree that taxonomic rank is related to the corporeal lo- cation of the relevant characters. Here, Olson bulae). In Trachyphonushe confirmedthe report confusesphylogeny with taxonomy;the ques- of Swierczewski and Raikow (1981), but in tion at issue is whether the group is mono- Hypnelus he did not. Olson quotes Swier- phyletic, not what rank it shouldhave in a clas- czewski (1977: 57) as saying that in the Gal- sification. Perhaps the FHL should rank the bulae the headsare "somewhatdifficult to sep- group at the generic level, becauseits origin arate from each other." Swierczewski's full lies at the knee joint (Juncturagenus). statement, however, conveysa different meaning: "In the Galbulidae and Bucconidae, the ANALYSIS OF OLSON'S HYPOTHESIS medial head arises semitendinous from the in- tercondyloidregion of the femur, and the in- Inadequatedata.--Olson comparescharacters termedlate head arises fleshy from the proxi- in only some of the relevant groups, so that mocaudal surface of the external femoral comprehensivecomparisons cannot be made. condyle;the headsbeing somewhatdifficult to For example, he notes that "in plumage pat- separatefrom each other." This statement re- tern, the ground roller Brachypteraciaslepto- fers to only two of the three heads;it doesnot somusis quite similar to certain of the Buccon- refer to the fibular head, which is the signifi- idae such as Malacoptila." But what are the cant structure. Olson dissected one specimen patterns in the other Coracii, the other Buc- 136 RAIKOWA•r) CRACRAFX [Auk, Vol. 100 conidae, and the Galbulidae?Why are they not ic similarity. He claimsthat we acceptthe idea discussed?Convergence in plumagepatterns is that "... differences between taxa [are] evi- common, which is why comprehensivecom- dence of nonrelationship .... "but this is in- parisons are needed to distinguish between correct.We mean only that the absenceof sim- synapomorphiesand spuriousresemblances. ilarities fails to corroboratea hypothesis of Again, in discussingthe skull Olson argues monophyly, not that the presenceof differ- for similarities between the Galbulae and Cor- encesrefutes such a hypothesis."Nonrelation- acii, but restricts his comparison almost en- ship" has no intrinsic meaning;we are search- tirely to the Bucconidaeand Coraciaswhen list- ing for patternsof commonancestry identified ing severalfeatures. "In all of thesecharacters," by synapomorphy. he notes, "the Bucconidaeare consistentlydif- The problem with phenetic comparisonsis ferent from the Pici." But what about the Gal- that charactersare not analyzed so as to iden- bulidae and other Coracii? tify the level within the phylogenetichierarchy This patternof selective,almost casual choice at which they define taxa. To illustrate this of data characterizes Olson's entire presenta- point, we will consider one example from a tion. It is also difficult to assess his data, be- recentstudy of the relationshipsof Pedionomus cause they are presented in a narrative form torquatus.This speciesis placed in the mono- within the text but are not tabulated in a way typic family Pedionomidae, and the problem that would allow one to determine the state of is to find its closest relatives. Olson and Stead- each character in all the relevant taxa. A tab- man (1981) rejected the previous hypotheses ulation of this sort would have revealed that that it is related to the Turnicidae (Gruiformes) broad conclusions are often based on little data; or the Galliformes and concluded that it be- e.g. Olson dissectedonly one barbet to deter- longs in the Charadriiformes. Olson and mine the form of the postorbital ligament in Steadman(1981: 3) note that a hallux is present the large and diverse suborderPici. in Pedionomus and most Charadriiformes but Incorrectdetermination of polarity.--Although is absent in the Turnicidae. They consider (p. generallyeschewing cladistic methods, Olson 21) that this supportsthe removal of Pediono- does make one attempt to show the derived mus "from the vicinity of the Turnicidae and nature of a condition, the complexof the post- its placement in the Charadriiformes." Actual- orbital process,postorbital ligament, and ad- ly, it does nothing of the kind. The hallux ductor mandibulae. He considers this derived evolved in vertebrates as part of the transfor- becauseit "... does not occurin Archaeopter- mation of the pelvic fin into a limb. As such, yx, in presumablyprimitive land birds suchas it is a derived character at the level of the Te- Opisthocomusand the Cuculiformes,or else- trapoda. Inasmuch as birds form a tetrapod where in the higher groups of land birds." subgroup, the presenceof a hallux is a primi- Several problems invalidate this conclusion. tive state and reveals nothing about the rela- First, there is no reason to assume that a char- tionships of any avian speciesto any other. It acter is primitive becauseit occursin "primi- tells us onething only about Pedionomus,name- tive" taxa. All taxa, including fossil forms, are ly that it is a member of the Tetrapoda. It def- mosaicsof primitive and derived characters. initely refutes any hypothesisthat Pedionomus Second,while we concedethe general primi- is afish, but that is all it does. tivenessof Archaeopteryx,we questionthe ex- This example illustrates the principle that tent to which the ligament and musclecan be charactercomparisons convey maximum infor- reconstructed in it. Furthermore, on what basis mation about phylogenetic relationships only are the Cuculiformes and Opisthocomus"pre- when one determines the specific points at sumablyprimitive"? What are "higher groups" which they are relevant within the nested sys- of land birds? Why is the comparisonlimited tem of cladesforming a phylogeny.Simplistic to the nonmonophyleticnontaxon of landbirds? phenetic comparisonsfail to provide this in- Olson's analysisfails to show that the condi- formation. Olson'sassessments of "similarity" tion is derived, or even that it is generallychar- cannot be interpreted, becausewe do not know acteristicof the groupsinvolved. which of them represent shared primitive Use of pheneticsimilarity.--The basic prob- charactersand which derived for any group of lem with Olson's study is that he tries to de- species.The necessityfor suchanalysis may be termine phylogeneticrelationships by phenet- emphasized by pointing out the distinction January1983] Monophylyof thePiciformes 137

between similarity and phylogeneticrelation- ical unit. Presentmorphological evidence sug- ship, namely that they do not necessarilyco- geststhat the Primobucconidaeare a grade,not incide. When an evolving lineage splits, one a clade.If so, then the family hasno ontological daughter lineage may evolve faster than the status,and Olson should insteadbe defending other, so that a form may be less similar to a the usefulnessof discussingthe evolutionof a genealogicallycloser relative than to a more fictitious taxon. distant one. Lungfishes are more similar to Simpson and Cracraft (1981:491) were very goldfishesthan they areto goldfinches,but they specific in their criticism; none of the taxa of are more closelyrelated to goldfinchesthan to the Primobucconidae is known to have an en- goldfishes.Crocodiles are more similar to tur- largedsehnenhalter. For that reason,they con- tles than to turtle doves, but they are more cludedthat there is no justificationfor saying closely related to turtle doves than to turtles. that the primobucconidsare piciforms as cur- Such statements are based on the idea that re- rently accepted,let alone closelyrelated to the lationship is defined by recency of common Bucconidae. That conclusion, if true, also fal- ancestry and are correct within the context of sifies Olson'sunsupported speculations about specificphylogenetic hypotheses, in this case the independentorigin of zygodactylyin the those of Wiley (1979). Genealogicalrelation- Galbulae.From our standpoint,the latter group ships are postulatedby the recognitionof pat- was already zygodactylouswhen it arose, be- ternsof monophyly,which are definedby syn- causethe origin of zygodactylytook place at a apomorphy. For extended discussions of this higher hierarchicallevel. This is why Simpson principle, see Eldredge and Cracraft (1980), and Cracraft (1981:491) suggestedthat one or Wiley (1981), and Raikow (1982). more taxa currently included in the Primo- bucconidaemight be the sister group of the DISCUSSION Or Fossil TAXA Piciformes. Olson'scomments about the Zygodactylidae The sectionof Olson's critique "Comments containfurther inaccuracies.Simpson and Cra~ on Fossil Taxa" adds nothing new about the craft (1981: 492) did provide evidence for in- systematicsof the taxa placed in the Primo- cluding this fossiltaxon in the Pici, namely an bucconidaeand Zygodactylidaebut doesillus- trate the limitations of his systematicmeth- advancedform of the sehnenhalter,but they also stressedthe very tentative nature of this odology. Simpson and Cracraft (1981) pointed placement.Furthermore, Simpson and Cracraft out that no evidenceexists to support the mon- did not voice any major disagreementwith ophyly of the Primobucconidae;Olson in contrast, believes that this observation is "irrele- Ballman's(1969a, b) interpretations,as implied by Olson. Their disagreementwas with the vant." But inasmuch as the "primobucconids" placement of the Zygodactylidaein the Gal- have played a pivotal role in the speculations bulae (Brodkorb1971). Simpson and Cracraft of Olson and his colleaguesabout the history of the North American bird fauna, it does not specificallystated that Ballmanmay be correct seem "irrelevant" to us to ask whether the Pri- in believing that Zygodactylusis not a piciform. mobucconidaehave any objective reality as a natural group. Indeed, if the primobucconids CONCLUSIONS are not monophyletic, then Olson and his col- leaguesare constructingevolutionary scenarios Rather than reply to every individual point based upon an imaginary taxon. As a conse- raised by Olson, we have instead addressed quence, one might expect that Olson would the basic differencesin systematicphilosophy want to demonstratethat monophyly. His cri- that exemplifycurrent controversy in biology. tique, however, is a weak attempt to defend Olson's attempt to question the synapomor- the conclusions of Feduccia and Martin (1976) phies linking the Galbulae and Pici stands re- and lacksany relevant empirical evidence. futed; the Piciformes are characterizedby a Olson's attempt to shift the burden of proof complexderived morphology of the hind limb onto Simpson and Cracraft is altogether spu- involving zygodactyly,tripartite flexor hallucis rious. We suggestthat the burden of proof lies longus, and Type VI tendons.Olson supports with workers who erect a taxon without posi- his hypothesiswith a potpourriof casualphe- tive evidence for its existenceas a geneaolog- netic similarities analyzed by inappropriate 138 RAIKOWAND CR•CR•VT [Auk, Vol. 100 methods, and his argument is little more than OLSON,S. L. 1983. Evidence for a polyphyletic or- an opinion. On the basis of present under- igin of the Piciformes.Auk 100:126-133. standing,we concludethat piciform monophy- ß & D. W. STEADMAN. 1981. The relation- ly remains the preferredhypothesis. ships of the Pedionomidae(Aves: Charadri- iformes). Smithsonian Contri. Zool. 337. ACKNOWLEDGMENTS RAIKOW,R. J. 1982. Monophyly of the Passeri- formes:test of a phylogenetichypothesis. Auk We thank Kenneth C. Parkes and D. Scott Wood, 99: 431-445. CarnegieMuseum of Natural History, for providing SIMPSON,S. F., & J. CI•ACI•AFT. 1981. The phylo- specimensused in dissection.Parkes, Wood, and genetic relationships of the Piciformes (Class Mary C. McKitrick providedhelpful criticismsof the Aves). Auk 98: 481-494. manuscript. Our research is supported by N.S.F. STEINBACFIERßG. 1935. Functionell-anatomische grants DEB-8010898(R.J.R.) and DEB-7921492(J.C.). Untersuchungenan Vogelftissenmit Wendzeh-

LITERATURE CITED en und Rticksehen.J. Omithol. 83: 214-282. SWIERCZEWSKI,E. V. 1977. The hindlimb myology BALLMAN, P. 1969a. Les oiseaux mioc•nes de La and phylogeneticrelationships of the avian or- Grive-Saint-Alban (Isbre). Geobios 2: 157-204. der Piciformes.Unpublished Ph.D. dissertation, 1969b. Die V6gel aus der altburdigalen Univ. Pittsburgh,Pittsburghß Pennsylvania. Spaltenffillungvon Wintershof (West) bei Eich- ß & R. J. RAIKOW. 1981. Hind limb mor- st•ittin Bayera. Zitteliana 1: 5-60. phology, phylogeny, and classificationof the BRODKORB,P. 1971. Catalogueof fossil birds: part Piciformes. Auk 98: 466-480. 4 (Columbiformes through Piciformes). Bull. WILEY, E. O. 1979. Ventral gill arch musclesand Florida State Mus. 15: 163-266. the interrelationshipsof Gnathostomes,with a ELDREDGEßN., & J. CRACRAFT.1980. Phylogenetic new classificationof the vertebrata. Zool. J. Lin- patternsand the evolutionaryprocess. New York, naean Soc. 67: 149-179. Columbia Univ. Press. 1981. Phylogenetics:the theoryand prac- FEDUCCIA,A., & L. D. MARTIN. 1976. The Eocene tice of phylogeneticsystematics. New YorkßJohn zygodactylbirds of North America (Aves: Pici- Wiley & Sons. formes). Smithsonian Contri. Zool. 27: 101-110.

(continuedfrom p. 125) tal variability; Gary Richard Hepp, Cost-benefit Ani (Crotophagaani); Michael P. Lombardo,Auxiliary analysisof pair-bond formationin wintering Green- birds at Tree Swallow (Iridoprocnebicolor) nests; winged Teal (Anas creccacarolinensis); Marc Herre- David B. McDonaldß Role of resource distribution mans, Intra- and inter-island variation in the Bul- and behavioralinteractions in a lek systemwith male- buls ( Microscelismadagascariensis/ M. crassirostris) on male cooperation;G. M. O. Maloiy, Environmental Grand Comoroand Moheli; Wendy L. Hillß Behav- physiology of two East African land birds: Kori ioralflexibility of reproductivebehavior in theAmer- Bustardand SecretaryBird; JeffreyS. Marks, Natal ican Coot; Mark Alan Holmgren, Geographicand philoparryand breeding-areafidelity in Long-eared songvariation in westernField Sparrows;William G. Owls; John L. Maron, Intraspecific competition Hoppes,Avian frugivoryand seeddispersal in north amongwintering Sanderlings;John Marzluff, Corre- temperateforests; Anne E. Houde, Nest site selec- latesof reproductivesuccess in the Pition Jay, Gym- tion by Common and Roseateterns; Laurie Ann norhinuscyanocephalus; Donald B. Miles, Patternsof Hunter, Cooperativebreeding behavior of Purple morphologyand ecologyin a SonoranDesert avian Gallinules; Kristine Johnsonß Mate choice in the community;Patrick J. MockßComparative breeding Pition Jay, Gymnorhinuscyanocephalus; Nancy Joste, ecologyand energeticsof storm-petrelspecies in Baja Electrophoreticanalysis of paternity in the Acorn CaliforniaßMexico; Michael L. Morrison, Geographic Woodpecker;Robert S. KennedyßSystematic review variation in morphologyand singingbehavior of the of Philippine birds for "A Checklist of Philippine Black-throatedGray Warbler in Oregon; CharlesA. Birds"; RichardLee and SusanKay KnightßValue of Munnß Socialorganization and ecologyof canopy flockingby Bald Eaglesin relation to foragingand flocks in Amazonian Peru; Gerald J. Niemi, Ecomor- predation: an experimentalstudy; RachelN. Levinß phologyof breeding birds in peatlandhabitats of Adaptive significanceof antiphonalsong in Thryo- North Americaand northernEurope; Manuel Nores, thorus nigricapillus;Stewart T. Levinson, Bird/fruit Studyof specimensof a new speciesof antbirdfrom interactions in northern Florida; Robert K. Loftin, the east of Brazil; StephenNowicki, Physiologyand Communalnesting behavior of the Smooth-billed physicsof harmonically-complexbird calls;Allan R. (continuedon p. 148)