The Auk A QuarterlyJournal of Ornithology Vol. 111 No. 4 October 1994 The Auk 111(4):787-805, 1994 MONOPHYLY OF THE FALCONIFORMESBASED ON SYRINGEAL MORPHOLOGY CAROLE S. GRIFFITHS Departmentof Ornithology,American Museum of NaturalHistory, Central Park West at 79thSt., New York, New York 10024, USA; and Departmentof Biology,City College of CityUniversity of New York, ConventAvenue and 138thStreet, New York,New York 10037,USA ASSTRACT.--Thesystematic relationships of the diurnalbirds of prey (Falconiformes)are unresolved.The monophylyof the orderhas not beenestablished, and the relationshipsof the familieswithin the orderand of generawithin the threepolytypic families are unclear. To derivea phylogenyfor the orderand to assessthe usefulnessof thesyrinx for resolving the systematicsof nonpasserines, I analyzed variation in syringealmorphology of genera within eachof the currentlyrecognized families in the orderas well asamong four orders of outgroups.The phylogeny derived from these syringeal data supports the monophyly of the Falconiformes.In addition, syringeal data provide strong support for themonophyly of threeclades within the Falconiformes:the Cathartidae;the Falconidae;and an Accipitrinae- Sagittariidae-Pandioninaecluster. The Cathartidaeare positionedas basalto the other two clades.Overall results indicate that syringealmorphology is conservative,with mostof the informativevariation occurring at highersystematic levels. Received 10May 1993,accepted 24 October 1993. OF THEAPPROXIMATELY 30 orders of birds, the previously(Jollie 1976, Sibley and Ahlquist order Falconiformes offers some of the more 1990),and only a summarywill be presented interestingsystematic questions; the monophy- here.The orderhistorically has been united by ly of the order is in question,as are relation- several external morphological characters shipsof familieswithin the order,and of genera (hookedbeak and curvedtalons) and by several within each of the families. Falconiformes cur- internal characters(biceps slip absent,ambiens rently consistsof 76 generaand 290 speciesdi- present, caeca rudimentary; Beddard 1898). vided into four families (Stresemann and Area- However, four current classificationsdisagree don 1979):Accipitridae, including Accipitrinae on the monophyly of the order, and on sub- (hawksand eagles,59 genera,217 species)and ordinal and familial relationships(Fig. 1). Two Pandioninae(Osprey, monotypic); Falconidae considerthe order monophyletic(Stresemann (falcons, 10 genera, 62 species);Cathartidae and Amadon 1979,Storer 1971),while the other (New World vultures,5 genera,7 species);and two propose either removal of the Cathartidae Sagittariidae(Secretarybird, monotypic). from the order (Sibley and Ahlquist 1990) or Comprehensivereviews of the historyof fal- inclusionof Strigiformes(owls) within the or- coniform classificationhave been published der (Cracraft 1981). 787 788 CAROLE S. GRIFFITHS [Auk, Vol. III Cathartidae Cathartidae Accipitrinae Pandioninae F AccipitrinaePandioninae Falconidae Falconidae Sagittariidae Sagittariidae Storer 1971 Stresemann and Amadon 1979 Cathartidae Cathartidae -- Accipitrinae Ciconiidae Falconinae -- Accipitridae -- Pandionidae Pandionidae Strigidae Sagittariidae Sagittariidae Falconidae Cracraft 1981 Sibley and Ahlquist1990 Fig. I. Four recent, conflicting classificationsof the Falconiformes. The idea that the Cathartidaemay not belong 1898, Brown and Amadon 1968) have been sug- in the orderhas been proposed previously based gestedas links between falconiformsand stri- on morphologicaland behavioraldifferences of giforms. this family from othersin the Falconiformes.In Relationships of the Falconiformes to other addition, similarities of cathartids to other ordershave alsobeen proposed,including the groupshave been noted, including the Cicon- orders Pelecaniformes, Ciconiiformes and Psit- iidae (Garrod 1873, Beddard 1898, Ligon 1967, taciformes(reviewed in Sharpe 1891, Shufeldt Rea 1983), Pelecaniformes (Beddard 1898, Jollie 1909), Gruiformes (Shufeldt 1909), Cuculi- 1976), Procellariiformes (Jollie 1976), and Grui- formes and Columbiformes (Verheyen 1950, formes (Pycraft 1902). Friedmann (1950:6)stat- reviewed in Jollie 1977). The AOU (1983) Check- ed that the Cathartidaewere a primitive group list places the Falconiformesbetween Anseri- not differentiatedfrom the "primitive stockfrom formes and Galliformes, whereas Sibley and which the Ciconiiformes, Pelecaniformes and Ahlquist (1990) position the order (with Ca- Procellariformeshave been developed." thartidae removed) as sister taxon to a group Hypothesesof a falconiform-strigiformrela- includingthe Podicipedidae,Sulidae, Phalacro- tionship have been advancedat various times, coracidae, and Phaethontidae. basedon similaritiesof palatal and myological Syringeal morphology had been used in the characters(reviewed in Sharpe 1891, Cracraft classificationof the major subdivisionsof the 1981, McKitrick 1991). Pandioninae (Sharpe Passeriformes at the end of the nineteenth cen- 1891, Pycraft 1902) and Falconidae (Beddard tury (Ames 1971). Within the last 20 years, sy- October1994] FalconiformMonophyly 789 ringeal data have again becomeimportant in cussedin the character descriptions(Appendix 2). the systematicsof oscinesand suboscines(e.g. Multistate characters were coded as unordered if states were alternative variations of a character. Ames 1971, Warner 1972, Lanyon 1984, Prum 1990, 1992), but have not strongly influenced I used outgroup information to polarize characters (Maddison et al. 1984).However, the relationshipsof the systematicsof other orders.There havebeen other orders to the Falconiformes are unclear, and the no detailed reports on falconiform syringeal choiceof an appropriateoutgroup is not readily ap- anatomysince Beddard (1903), and no system- parent. Species from four orders of birds were in- atic analysisof that anatomy. cluded in the analysis.A comprehensivesubset of I examined patternsof variation in syringeal generain the Ciconiiformesand Strigiformeswas ex- morphologywithin the Falconiformesto assess amined becausethere are phylogenetichypotheses of the usefulnessof the syrinx for systematicsof relationshipsof theseorders to the Falconiformes.In nonpasserines(i.e. whether there are phylo- addition, species within the Pelecaniformeswere genetically informative syringeal characters) added becauseseveral authors suggesteda relation- ship of this order to the Falconiformes(Beddard 1898, and to answer two primary systematic ques- Shufeldt 1909, Friedmann 1950, Jollie 1976). Because tions: (1) Is the order monophyletic?(2) What monophyly of the Falconiformescould not be as- are the relationshipsof the majorclades within sumed,species from theseother orderswere not treat- the order? ed as outgroupsin the analysis,nor was there a con- straint on monophyly imposed in the analysis (Maddisonet al. 1984).Species within the Galliformes MATERIALS AND METHODS were used to root the cladogram.I used PAUP 3.0s (Swofford 1991)to derive the most-parsimoniousres- Specimens.--Iexamined syringesfrom collectionsat the American Museum of Natural History (AMNH), olution of the data.The sizeof this dataset precluded the use of exact algorithms for resolving the data; the National Museum of Natural History (USNM), therefore,the heuristicalgorithm wasused. However, the Royal Ontario Museum (ROM), the University of this option does not guarantee optimality and may KansasMuseum of Natural History (KUMNH), the Museum of Vertebrate Zoology of the University of identify a solution that is only locally optimal. To increasethe probabilityof finding solutionsthat were Californiaat Berkeley(MVZ), and the LouisianaState globallyoptimal, analyses were repeatedvarying both University Museum of Natural Science(LSUMNS), the branch-swappingand taxa-additionoptions. someof which I dissectedfrom fresh or alcohol-pre- Three indices were used to assessthe congruence served specimens.These were cleared and double- of the charactershypothesized as synapomorphies stainedto distinguishcartilaginous and ossifiedtissue (Farris 1989):(1) consistencyindex, which is the min- (Cannell 1988).Observations were made using a Wild M5A dissectingmicroscope, and drawingsmade with imum amountof changeof a characterdivided by the amount observedon the tree; (2) rescaledconsistency a camera lucida. Drawings were scanned into a Mac- intosh computerand final illustrations prepared us- index, which is a linear rescalingof the consistency index sothat valuesvary from 0 to 1; and (3) retention ing Aldus Freehand 2.0. index, which is the proportion of original characters I analyzed 124falconiform syringes and 66 syringes remaining as synapomorphies.Consensus methods from purportedoutgroups (Appendix 1). Three of the were used to summarize information from the set of five cathartid genera and all falconid genera were most-parsimonioustrees. I used strict-consensustrees, included. Within the Accipitridae, genera were cho- which include groups found in each of the most- sento representeach of the previouslyproposed sub- groups;these totaled 60% of the currentlyrecognized parsimonious cladograms,and majority-rule trees, genera.In general,sampling within the ingroup was which include groupsfound in a definedproportion of cladograms.Consensus trees must be interpreted constrainedby the availability of specimens.Two or more individuals from 27 specieswere analyzed to with care as they may not be parsimoniousrecon- structionsof the original data (Swofford 1992). Nev- assessvariation at the intraspecificlevel.