Phylogenetic Relationships Among Cranes (Gruiformes: Gruidae) Based on Dna Hybridization

Home , Antigone (bird), Balearica, Crane (bird), Gruiformes, Grus (genus), Limpkin

PHYLOGENETIC RELATIONSHIPS AMONG CRANES (GRUIFORMES: GRUIDAE) BASED ON DNA HYBRIDIZATION

CAREY KRAJEWSKI• ZoologicalMuseum, Department of Zoology,University of Wisconsin,Madison, Wisconsin 53706 USA

ABSTRACr.--Iused a DNA-DNA hybridizationmethod to generatemore than 1,200pairwise comparisonsamong speciesof the family Gruidae (cranes)and an outgroupLimpkin (Ara- midae). A matrix of genetic distancesincluded averagedelta Tmvalues for all cells, and reciprocalsamong cranes,based on 3-10 replicateexperiments per cell. I chosedelta T• as the appropriate dissimilarity measure becausevirtually all homologoussingle-copy DNA sequencesin cranegenomes were sufficientlysimilar to form hybrid duplexesunder standard experimental conditions. The normalized percent hybridization (NPH) values approached 100 for all crane speciespairs. The delta Tmdata departed slightly from metricity as a result of experimentalvariation associatedwith the measurementof small geneticdistances. The DNA data, analyzed with a best-fit tree approachand checkedfor internal consistencyby jackknifing,support the traditional view that crowned cranes(Balearica) are the mostancient lineageof extantgruids. The enigmaticSiberian Crane (Grus leucogeranus) appears as the sister group of the remaining species,which fall into four closelyrelated groups.Bugeranus and Anthropoidesare sistergroups. The three speciesof AustralasianGrus (antigone, rubicunda, and vipio)form a clade, as do five predominantly PalearcticGrus (grus, monachus, nigricollis, japonensis,and americana).The SandhillCrane (G. canadensis),while clearlya memberof the gruine clade, is an old lineage without close relatives. Received21 March 1989, accepted18 May 1989.

THE CLASSIFICATIONSof Peters (1934) and Wet- terials that cranes diverged from a common more (1934) establishedthe traditional family- ancestorwith limpkins (Aramidae) in the late level distinction of cranes (Gruidae) within the Paleocene. Order Gruiformes. This arrangement, though Archibald (1975, 1976) performed the first not universally accepted (Cracraft 1973), has comprehensiveanalysis of cranes based on a persistedin mostrecent revisions(e.g. Wetmore coherent set of characters.His study of crane 1960, Storer 1971, Sibley et al. 1988). Opinions unison calls (stereotypedbehavior patterns in- regarding speciesaffinities within the Gruidae volved in pair-bonding) led him to identify have varied greatly. Following Peters (1934), clusters of similar species(Fig. 1). Archibald's most workers have acceptedthe existenceof 14 work verified the distinctness of crowned cranes extant speciesin four genera, though Walkin- (Balearica)and suggestedan unexpected rela- shaw's(1964) designationof a 15th specieshas tionship between Bugeranuscarunculatus (Wat- gained many adherents. The crowned cranes tled Crane) and Grus leucogeranus(Siberian (Baleatica)are commonly placed in the Balear- Crane). Archibald included leucogeranusas a icinae, apart from the remaining species(Gru- member of Bugeranus, and recommended inae; Brodkorb 1967) based on their lack of ster- "SpeciesGroup" statusfor the remaining Grus hal excavation and tracheal convolution. and Anthropoidesclusters. Wood (1979), like Ar- Balearicines,apparently the more ancient lin- chibald, found a high level of similarity be- eage, are abundantly represented by Tertiary tween Wattled and Siberian cranes, which, aside fossil materials from western Eurasia (Brodkorb from complicated multivariate resemblances, 1967). Gruine fossilsappear in the Miocene of lack the exaggerated tracheal convolutions Europe, but are best represented in North present in other gruines. American Pliocenedeposits and by a scattering Ingold et al. (1987)attempted to resolvecrane of Pleistoceneremains worldwide (Johnsgard relationships with allele-frequency analysis. 1983). Cracraft (1973) inferred from these ma- Their speciesarrangement was a major departure from traditional views, although Balearica Current address:Museum SupportCenter, Smith- appearsdistinct from gruines.Their work in- sonian Institution, Washington, DC 20560 USA. volved small sample sizes, and only approxi- 603 The Auk 106: 603-618. October 1989 604 C^R•¾KRAJEWSI

G. grus tions are given in Krajewski(1988) and Springer and G. nigricollis Kirsch (1989). DNA was recovered from erythro- cyteslysed with sodiumlauryl sulfatein TNE buffer, G. monachus purified by repeatedphenol/chloroform extractions, G. japonensis and treated with proteaseand RNase (Maniatis et al. G. americana 1982).Native DNA was fragmentedinto short strands G. canadensis (100-2,000 bp) using high-frequency sound. Frag- ment size distributions were monitored for each sam- G, vipio pie by agarosegel electrophoresisand comparison G. ontigone with commercial size markers. G. rubicunda Single-copy sequences(scDNA) were recovered from eachspecies using the reassociation-kineticand G. [eucogeronus hydroxyapatitecolumn-chromatography methods of Bugeranus Kohne and Britten (1971). Sheared DNA was boiled A. virgo and incubatedat 60øCin 0.48 M phosphatebuffer (PB) A. paradiseo to a Cotof 200, diluted to 0.12 M PB, applied to hydroxyapatite(HAP) columnsat 60øC,and single-copy Baleor/co sequenceseluted in 20 ml of 0.12M PB.Samples were Fig. 1. Crane relationshipsas indicatedby overall dialyzed for 24-48 h, then frozen and lyophilized. similarity in their unison calls (Archibald 1975;fig- The methodemployed to produceradioactive tracer ure redrawn from Wood 1979). The horizontal axis is DNA was derived from the general iodination pro- arbitrary, showing only Archibald's view of nested tocols of Commorford (1971), Davis (1973), Tereba similarity levels. and McCarthy (1973), Orosz and Wettour (1974), A1- tenberg et al. (1975), Scherbergand Refetoff (1975), mately two birds per specieswere assayedfor Chan et al. (1976),Anderson and Folk (1976),Prensky allele-frequencies. (1976),and Sibleyand Ahlquist (1981).Modifications of these proceduresare noted below. Basedon a very limited DNA hybridization Lyophilized,single-copy DNAs were rehydratedin study of the cranes, Ingold (1984) concluded a small volume (25-100 •1) of 0.2 M NaAc (pH 5.7), that substantial genetic divergence exists be- transferred to a 1.5 ml microfuge tube, vortexed for tween putative Balearicaspecies, and that G. leu- 30 s, and centrifuged at 3,000 g for 30 s. Reaction cogeranuswas not particularly "close" to Buge- mixtures were prepared by combining 100 •g of ranus,contra Archibald (1975, 1976) and Wood scDNA with enough 0.2 M NaAc (pH 5.7) to make a (1979). The latter conclusion seems difficult to solution at 0.77 •g/•l, as well as 5 •1 of 0.002 M KI defend in light of Cracraft's (1987) criticisms and 10.4•1 bromcresolgreen dye (BCG),in 1 ml stop- regarding the inadequaciesof partial genetic peredsepturn vials. Samples were adjustedto pH 4.7- distancedata for resolving relationships. 4.8 with 0.2 M NaAc at pH 4.0. Iodinationswere usually carried out for a set of I derive a phylogeny for the cranesbased on eight DNA samples.The isotope(•25I, Amersham IMS- a complete matrix of pairwise DNA hybridiza- 300) was obtained in 5 mCi amounts (ca. 10 •1), diluted tion comparisons.Previous avian work with this with 340 •1 of 0.2 M NaAc and 10 •1 of 0.001 M KI, technique (see referencesin Sibley et al. 1987) and allowed to equilibrate for 1 h. Eachtracer sample has focused on high-level taxonomy, though received 40 •1 of isotope solution (0.625 mCi), fol- Sheldon (1987a) used it to resolve intrafamilial lowed by 60 •1 of 0.0018 M thallium (III) chloride relationshipsamong herons (Ardeidae). Crane (T1C13).Samples were incubated at 60øCfor 15 rain, relationshipsare predominantly those of con- then cooled in an ice bath for 5 min. The pH of the genericspecies, and previouswork hasleft some reaction mixtures was raised above neutrality by ad- doubt as to the resolving power of DNA hy- dition of 30 •1 1.0 M Tris, sampleswere heated again at 60øC for 10 rain, and chilled on ice for 5 min. After bridization at this level. ! will show that the cooling, sampleswere dialyzed against4 1 of "iodin- method provides limited resolution among ation buffer" (Sibley and Ahlquist 1981). cranes, yet, with numerous replicate compari- DNA hybridswere preparedwith 0.5 •g tracer DNA sons,it is capable of defining clustersof phy- and 250 •g sheared,native DNA ("driver") in 0.48 M logenetically related species. PB. Hybrids were boiled and incubated at 60øC to a Cotof at least5,600, diluted to 0.12 M PB,and applied METHODS to HAP columns in an automated Thermal EIution Device. Column temperatures were raised to 60øCand DNA hybridization.--Thebiochemical protocol for three 8-ml fractions of 0.12 M PB were collected. Ad- DNA hybridization was essentiallythat describedby ditional fractions were collected at 2øC intervals from Sibley and Ahlquist (1981, 1983). Detailed descrip- 64øC to 94øC, and a final fraction was collected at 98øC. October1989] CranePhylogeny 605

Elution fractionswere assayedfor radioactivityin a PackardAuto-Gamma 5000 Seriesgamma counter. Two tracer DNA preparations were employed for every crane species,with replicate interspecieshy- brids arranged in experimental sets of 25. Each ex- •6f Gc perimental set was defined by one or two reference ("homologous")hybrids (i.e. tracer and driver DNAs from the same extraction of the same individual bird), with which all interspecies("heterologous") hybrids were compared.Three to 10 replicate heterologous hybrids were examinedfor each pairwise combina- tion of species(reciprocals treated separately). The end product of each thermal elution experiment is a list of radioactivities of eluate for each of

the 20 temperature intervals. Of these 20, the first 0 • • three 60øCfractions represent counts from isotope atoms which either did not bind to tracer DNA or •oo b boundatoms on tracerfragments which did not form stable duplexes during reassociation.Such unreas- • 8C Br sociatedstrands in homologoushybrids likely rep- Gc resent very small DNA fragments (<100 bp), pro- duced during sonicationand iodination, which are •= 60 Gj,Ap prohibitedby their sizefrom formingstable duplexes under standardconditions. In heterologousprepa- rations,this fractionwill alsocontain tracer sequences '• '•C which are too divergent from their driver homologs to form stableduplexes at 60øC.Thus, the "percent hybridization" of a particular hybrid preparationis (") 2C the ratio of counts eluted at and above 64øC divided by total countsmultiplied by 100. For heterologous hybrids, this value is standardized for comparison 60 70 80 90 100 (i.e. expressedas "normalized percent hybridization" Temperature (degrees Celsius) or NPH) by dividing the heterologouspercent hy- Fig. 2. Thermal elution profilesfor DNA hybrids bridization by that for its correspondinghomolog. among cranesand outgroups.Top: Stepwiseplot of Because the normalized percent hybridization percent counts eluted at indicated temperatures,il- (NPH) values among cranes were essentially 100% lustratingmodal divergence. Bottom: Cumulative plot (seebelow), I usedTm to measurehybrid thermal sta- of percent counts eluted at indicated temperatures, bility (Sheldon 1987a, b). T• is the median melting illustrating divergence in Tin.Tracer speciesis Grus point of all DNA sequenceswhich have hybridized leucogeranus(G1). Other speciesabbreviations are Ap in a particular experiment, and is given by the tem- = Anthropoidesparadisea, Gc = Gruscanadensis, Gj = peratureat which 50%of total countsabove 60øC were Grusjaponensis, Br = Balearicaregulorurn, Ag = Ararnus eluted (valuesbetween specificelution temperatures were estimated by linear interpolation). Genetic distance measurements were calculated as delta Tmvalues between homologousand heterolo- Treeconstruction.--I generated a phenogramfor a goushybrids within the sameexperimental set. Final folded matrix of distances with trimmed means (see distanceestimates were the averagesof all replicate below)using the unweightedpair-group method with deltaT• valuesfor eachpairwise comparison of species arithmeticaverages (UPGMA) implementedin Rohlf's (reciprocalstreated separately). In three cases,a single NTSYS softwarepackage for the IBM-PC. value was "trimmed" from the sample of replicate Estimatesof phylogeneticbranching order were measurements(Krajewski 1988; see below) to obtain made with the FITCH routine of Felsenstein's PHY- a more robust estimate of actual distance. LIP software package(version 2.8; for a justification Outgroup comparisonswere provided by DNA hy- see Felsenstein1982, 1986; Springer and Krajewski bridsbetween the 14cranes and the Limpkin(Ararnus 1989). Relevant parameter values for this algorithm guarauna;Gruiformes: Aramidae). Reciprocal compar- were set at P = 0.0 (least-squarescriterion of fit) and isonsinvolving Ararnuswere not performed, because n = 1 (default).Negative distances in the input matrix DNA from the Limpkin did not produce adequate (seebelow) were setequal to 0.01and negativebranch- tracer preparations.The final matrix of genetic dis- lengthswere disallowedin the output trees.To in- tances contained 15 rows (drivers) and 14 columns creasethe likelihood of globally optimum results,I (tracers). employedsix alternateinput ordersof taxa,and I took 606 c• KV.AIgWSK• [Auk,Vol. 106 October1989] CranePhylogeny 607

I 608 Oa•Y KR•IEWSm [Auk, Vol. 106

8.00 6.00 4.00 2.00 0.00 Aramus Boleatica

Bo•eori•o

• A.porodiseoA.virgo

r• G.Bugeronus conodensis

G. vipio _•G.G. ontigonerubicundo Sum-of- Squores = 7..•2774 G.gru$ Exommed 427 trees. t G.leucogero•u$ G.monochus G.omericono G.joponensis G. nigricollis • G. leucogeronus 015 i O.... i.... Ararnus Fig. 4. Best-fitleast-squares tree for a folded ma- Fig. 3. UPGMA phenogramfor DNA hybridiza- trix of delta Tmvalues. Generatedby the FITCH rou- tion distancesamong cranes.Based on a folded distance matrix of delta Tm values. tine in PHYLIP (P = 0.0).Aramus was specified as the outgroup.Six alternateinput orderswere examined. The terminal branch to G. grushas length 0. the solution of choice as that with lowest sum-of- squares value. BecauseAramus was not labeled, the genetic dis- expectedto be a consistentestimator of the ac- tancematrix was folded (i.e. reciprocalsamples were tual percent nonidentity between the se- pooled and averaged)to positionthe root of the crane quencescompared when NPH is near 100 (this tree. With Aramusspecified as the outgroup,this root assumesthat the error distribution of delta T• occurred as expected between the two traditional is symmetricalabout its expectation).For this subfamilies (see below). All subsequentanalyses were performed on the squarematrix (reciprocalsdistin- reason,I report genetic distancesbetween taxa guished)with Balearicadesignated as the outgroupto as averagedelta Tmvalues based on a specified other cranes.The best-fit tree obtained from analysis number of replicates(n). The replicate delta T• of the square matrix was tested for internal consis- valuesare tabulatedin the Appendix. Average tencyusing the JackknifeStrict Consensus Tree meth- delta T• valuesamong cranesrange from -0.2 od of Lanyon (1985a, 1987). to 4.2 (see below for a discussionof negative distances),n range from 3 to 10, and SD from RESULTS 0.07 to 1.40 (average = 0.48 + 0.66 SD) (Table 2). The average SD of 0.48 is slightly higher Normalizedpercent hybridization (NPH).--Fig- thanthat obtained for deltaTsoH (0.35) by Sibley ure 2 illustratestypical elution profilesfor crane et al. (1987) and for delta T• (0.20) by Sheldon DNA hybrids. The averageNPH among crane (1987a).Standard deviation values in this range specieswas 101.5 (SD = 20.0), and virtually all probably reflect a lower limit on the precision homologoussequences between speciesform of averagedelta Tm's obtained with the standard duplexesin hybridization experiments(Table experimental protocol. 1). Standard deviation values of the NPH sta- Phenetic relationships.--As clustered by tistic for individual cells in the matrix are char- UPGMA (Fig. 3), Aramusappears, as expected, acteristicallyhigh (range:0.9-53.9) as described as an outgroup to the cranesand links to the by Sheldon(1987a, b). The rangeof off-diagonal gruid cluster between Balearicaand the other NPH valuesfor tracercrane species is from 84.0 (gruine) species.The gruines form a relatively to 138.6, which implies considerableexperi- tight cluster,from which Grusleucogeranus de- mental error in this measurement. Neverthe- partsfirst. The remainingspecies fall into three less,no speciesdeparted consistently from the clusters.One containsfive membersof Grus(grus, averagevalue. The Limpkin (Aramus)also had monachus,americana, japonensis, and nigricollis); a high degree of similarity to cranes(average one contains Anthropoides,Bugeranus and G. NPH = 92.1, SD = 3.0, n = 55). canaden$is;and the third containsG. antigone,G. Geneticdistance matrices.--DNA hybridization vipio, and G. rubicunda.Although internodal is subjectto an inherent level of imprecision, lengthsare extremelyshort, it is significantto and distancemeasurements must be replicated note that this phenogram representsthe dis- repeatedlyto obtain high confidencein an av- tancesin the folded matrix with very little dis- eragevalue or to assessaccurately the level of tortion: the copheneticcorrelation is 0.99364. variation imposedby the technique.Delta Tmis Best-fittrea for the folded matrix.--A FITCH October1989] CranePhylogeny 609

Sum-of-Squares= ! 4.60406 Boleorico Examined :560 trees G. leucogeronus Boleorico G. ontigone G leucogeronus G. ontigone G. rubicundo G. vipio G. vipio G. rubicundo G. conodensis G. conodensis Bugeronus Bugeronus A. virgo A. porodiseo A. virgo G. nigricollis A. porodiseo G. [oponensis G. grus G.nigricollis G. monochus I I G. joponensis 0.5! • G. omericonO Fig. 5. Best-fitleast-squares tree for the squarema- G.grus trix of distances(Table 2). Balearicawas specified as G. monochus the outgroup,and the root positioned(Fig. 4). Gen- eratedby the FITCH routine of PHYLIP (P = 0.0). Six G. omericana alternate input orders were examined. The terminal branch to G. grushas length 0. Fig. 6. Jackknifestrict consensus tree for the square matrix of distances(Table 2). Pseudoreplicatetrees generatedby the FITCH routineof PHYLIP (P = 0.0). best-fit tree for the folded matrix of genetic distances(Fig. 4) placed Aramusoutside the crane G. vipio.Another input order moved Bugeranus group (d'= 5.2 from Aramusto the node linking adjacentto the G. grusgroup, but this tree had Balearica).Balearica is isolated on a long branch a noticeablypoorer fit than the others.A final (d'= 2.2) and a relatively long internode (d '= shuffling of the matrix gave a tree which sep- 0.7), supporting the traditional view that these arated G. canadensisfrom the G. antigonegroup, birds lack an extant close relative. The most but produceda muchpoorer fit than the optimal important conclusion from the folded-matrix solution. analysisis that Balearicais suitable as a desig- The speciesgroups indicated in Figure 5 are nated outgroup with which to root square ma- almostidentical to thosein Figures3 and 4. Bale- trix trees.The positionof Balearicain the folded- aricaroots the tree betweenG. leucogeranusand matrix tree remained consistent when the data the remaining species.Among the latter taxa, were challengedby jackknifing. G. grus,G. monachus,G. americana,G. nigricollis, The folded-matrixtopology is virtually iden- andG. japonensis form a cladewhose sister group tical to the DNA phenogram (Fig. 3) with re- is composedof Bugeranusand Anthropoides.G. spect to the composition of species groups, canadensisappears as an early branchfrom the though the suggestedrelationships among them G. antigone,G. vipio,and G. rubicundalineage. are altered. The short internodes which sepa- The only substantivedifference between this rated most of the speciesclusters in the phe- tree and that for the folded matrix is in the nogram are found again in best-fit trees (Fig. position of G. canadensis.Relationships among 4). This is the first hint of lowered confidence groupsshown are basedon relatively small fit- in someof the branchingsabove G. leucogeranus.ted branch lengths (Fig. 5). While the inter- Best-fittree for the squarematrix.--The opti- nodes which separateBalearica and G. leucoge- mum tree solution was found for the distance ranusfrom the main tree appear substantial (d' data (Table 2) with Balearicadesignated as the = 0.74 and 0.20, respectively),that which places outgroup (Fig. 5). Three alternate input orders canadensisin the antigonegroup is only d' = 0.03, gave identical topological results and differed that which unites Bugeranuswith Anthropoides from the branching order in Figure 6 only by is only d'= 0.08, and that which unites the G. interchanging the positionsof G. rubicundaand grusand Bugeranus/Anthropoidesgroups is only 610 CAREYKRAJEWSKI [Auk, Vol. 106 d' = 0.07. With the exceptionof the Bugeranus/ the samelevel of repeatability and implied vir- Anthropoidesclade, most internodes within tually identical rank-orderingsof drivers with- speciesgroups are lessthan 0.10. in tracer-sets.I emphasizethe T• values simply Jackknifestrict consensus tree.--Lanyon (1985a) to avoid the necessityof computercurve-fitting, united jackknifing with systematicconsensus- which may introducean additionallevel of error tree methods to deal with problems of estimat- into distance calculations. A reassociation-ki- ing the confidencethat can be placed in a tree netic survey of crane DNAs (Krajewski 1989) derived from distancedata. Lanyon (1985a)pre- revealed no grossdifferences in the structure sentedthe details of his algorithm, the product of crane genomes which would complicate in- of which is a "jackknife strict consensustree" terpretation of either distance measure. (JSCtree). Jackknifinghas been applied to DNA The T50H measure has received much critical hybridization studies by Lanyon (1985b) and attention (Sarich et al. 1989). This index takes Sheldon (1987a). "unhybridized sequences"into account in the The JSCtree (Fig. 6) for the squarematrix of calculation of a median melting point (Sibley crane distances(Table 2) agrees with the best- and Ahlquist 1983). Given that such sequences fit tree (Fig. 5) in that Balearicaand G. leucoge- are rare in crane hybrids (as indicated by NPH ranus are the earliest branches, but resolution values near 100), one would expectTs0H and T• breaks down at the next interior node. The com- valuesto be equivalent. However, the inherent position of terminal groups is preserved, but variability of NPH measurementis carried over their internal structure is mostly lost. Once into the estimation of T•0H, and delta T•0H is a again,G. antigone,G. vipio,and G. rubicundaform very imprecise measure of genetic distance a clade, but the association between the first (Sheldon 1987a,b). This problem is especially two is obscured. G. canadensis loses its associa- acute for closelyrelated speciessuch as cranes, tion with the antigonegroup, but the branching where high variance in distancemeasures will order of the Bugeranus/Anthropoidesclade is re- result in low phylogeneticresolving power un- tained. Resolution virtually disappearsfor the less a very large number of replicate experi- G. gruscluster, except an unexpectedlystable ments are performed.In any case,NPH levels association between G. americana and G. mona- among cranesare such that calculationof T5oH chus. values adds nothing to the robustnessof the distances.

DISCUSSION Trimmingmeans.--If the level of experimental error describedhere and in Sibley and Ahl- Distancemeasures and normalizedpercent hy- quist'swork (1983)represents a lower threshold bridization(NPH).--Melting points of DNA hy- of precision with this experimental protocol, brids are often expressedas the temperatureat few analytical optionsare available which will which 50% of duplexed tracer DNA has disso- further improve the "signal to noise" ratio of ciated (Kohne 1970; Sibley and Ahlquist 1981, DNA hybridization data. Sibley and Ahlquist 1983). This value is the one which Kohne (1970) (1981, 1983) describea simple statisticalpro- and othersrelated to sequencedivergence. Sib- cedure to minimize the effect of a few aberrant ley and Ahlquist (1981, 1983) also calculated data points on the averagevalue of a larger set modal and Ts0Hvalues for thermal elution pro- of replicates.Such aberrant values may exert a files after fitting those profiles to one of four significantinfluence when the technique is re- distributionfunctions by nonlinearleast-squares quired to measuredistances at the low end of regression.In somecases (e.g. for hybridswhich its precisionrange. Sibley and Ahlquist omitted display a "low temperature component"), the the largestand smallestdistances from eachset mode may be a more precisemeasure of thermal of replicate measures, and averaged the re- stability than Tm(Sarich et al. 1989). The crane maining values to obtain a "trimmed mean." profiles, however, are nearly symmetrical and Sibleyand Ahlquist claimedthat this procedure one would expectmodes and T•'s to be roughly improved the robustnessof their distancemea- equivalent. Analysis of a subset of the crane sures.Indeed, this follows from the assumption data revealed that delta-mode and delta Tmval- of a symmetric error distribution. ues are almostidentical, with averagevalues of The omission of both extremes of a range of the former within a single standard deviation distanceshas the disadvantageof reducing the of the latter. Both measureshad approximately samplesizes for each cell and assumesthat the October1989] CranePhylogeny 611 sample of scores,however small, is symmetric per measurementsof such small intraspecific about its average.Both thesedifficulties can be divergences. mitigated by omitting only a single extreme Intraspecificvariation and the potential prob- score (i.e. that with the greatest magnitude of lems it may causefor phylogeneticreconstruc- departure from the sample mean). A problem tion are discussedby Springer and Krajewski related to trimming is the assumptionthat aber- (1989) and, briefly, by Sheldon (1987a). Aside rant values occur in every sample (i.e. no cri- from the effects of unusual bottlenecks in the teria are employed to identify suchvalues, and speciationhistory of a clade, genetic distances the effectsof trimming may vary from sample among specieswill be phylogenetically infor- to sample). An obvious method to screen sam- mative when the amount of detectable intra- piesfor aberrantscores is suggestedby our prior specific differentiation is less than the amount knowledge of the variation expected of DNA of detectable interspecific differentiation. In hybridization measurements.From this infor- cranes, the precise amount of intraspecific dis- mation and the conclusionsof Sibley et al. (1987), tance may not be detectable over experimental we expect each sample of scoresfor which n > "noise" and, in any case,appears comparable 5 to have a standard deviation of about 0.5 (the in magnitude to only the smallestdistances be- sample size constraint is significant, because tween species. smaller samples will generally have higher Reciprocity.--Somereciprocal delta T,,'s (Ta- variance).Samples with substantiallyhigher SDs ble 2) showed substantial departures from the are likely to contain at least one aberrant score. symmetry expected of metric distances. Al- I chosea conservativelyhigh SD value of 1.0 though some level of asymmetry is expected to insure that only aberrant values (not merely from experimental "noise" (iraprecision), this marginal ones) were omitted. phenomenonmay alsosignal a biasin the mea- This modified trimming procedurewas used surements.Groups of delta Tmvalues derived for the matrix (Table 2) where only three cells from different tracer preparations within each were found to meet the trimming criteria. These column covary(Felsenstein 1987). The mostob- were d(A. paradisea,Bugeranus), d(G. canadensis, vious source of such bias is the "compression A. paradisea),and d(G. monachus,G. nigricollis). effect" on distances,introduced when a partic- In all three cases,the aberrant value was a large ular tracer preparation gives a low (e.g. <82øC) delta Tin,consistent with the notion that poorly homologousTm as a result of biochemicaldeg- reassociatedhybrid preparationswere respon- radation. Several such tracerswere of necessity sible for the error. employed here (seeAppendix), and it is likely Levelsof geneticand experimentalvariation.--I that their influence accounts for much of the usedspecies as the "operational taxonomicunit," "internal inconsistency"detected by jackknif- but it is of interest to examine intraspecificvari- ing. ation to attempt to estimate levels of experi- Negative distances.--Two cells in the crane mental error and genetic differentiation. One matrix contain average delta Tmvalues of less may consider, for example, variation among than 0: d(G. vipio,G. antigone)= d(G. grus,G. DNA extracts from the same individual. For monachus)= -0.2. Whether these average val- multiple hybrids formed between crane tracer ues are significantly lessthan zero is irrelevant and driver DNAs of the same extract (i.e. ho- to the nature of measurement error. Both values mologoushybrid preparations),the averageSD are associatedwith speciesfor which one tracer among Tmvalues is 0.47 (SD = 0.51, n = 25; see preparation gave low-melting (T,, < 82øC) ho- Appendix). The average difference in homolo- mologoushybrids and with driver speciesthat gous Tmvalues between different extractsof the were extremely similar to the tracers.This com- same individual crane is 0.38 (SD = 0.86, n = 7; bination of biased experimental error and in- see Appendix), which approximatedthe mag- creaseddemand for precisionis the causeof the nitude of experimental error (though the sam- negative averages.The negative average delta ple size is relatively small). The DNA hybrid- Tmdoes not imply a negative genetic distance ization data for cranes had almost no measurable (this makes no sense),but rather indicates mea- variation, on average, between conspecificin- surement variation in delta Tm around a small dividuals (averagedelta Tm= 0.07, SD = 0.62, actual distance.These negative distancesoccur n = 19; see Appendix). I believe that experi- within speciesgroups whose internal structure mental error is of sufficient magnitude to ham- cannot be fully resolved (Figs. 5 and 6). 612 C^RE¾KRAJEWSKI [Auk, VoL 106

Triangle inequality.--Although Sibley and of species-groupsI detected (Fig. 6) is precisely Ahlquist (1983) have claimed that DN^ hy- that suggestedby Archibald (1976).Perhaps the bridization distancesusually satisfythe triangle most basic agreement among all systematic inequality, it is unclear if measurementerror treatmentsof cranes,including this one, is that associated with small distances causes viola- the Balearicarepresent a distinct clade. tions of the inequality. Such violations do oc- The sole discrepancy between Archibald's cur: the squarematrix (including Aramusin the (1976)arrangement and the DNA hybridization lower half) contains 85 triplets which fail the consensustree lies in the placement of Grus triangle test, <4% of the total (there are 2,302 leucogeranus.My results do not support a close nonredundant triplets in all: 1,080 in the upper relationshipbetween G. leucogeranusand Buge- half, 1,222 in the lower half). Most of the tri- ranus,but rather they indicate that leucogeranus angle violations involve one or more of the is the first branch from the gruine line. The speciesG. grus, G. monachus,and, in the upper DNA data agree with Archibald's linkage of half-matrix, G. vipio.Many of the violationsthat Bugeranusand Anthropoides.It appears that the involve grusand vipioalso involve the two neg- highly variable and, in many respects,unique ative distances discussed above. These two features of the leucogeranusunison call render speciesalso gave one low-melting tracer and, its similarity to Bugeranusphylogenetically mis- consequently,may show somewhatcompressed leading. The primary aspectsof unison-call re- column values. As noted above, all three species semblancebetween the two speciesare more belong to closelyrelated clusterswhich cannot likely due to convergentevolution than to the be resolved consistently by the DNA data. retention of primitive characters.These features Jackknifingand internal inconsistency.--The are not found in other cranes(the tracheal/ster- jackknifing procedure doesnot constitutea sta- nal similarities are more problematic, in light tistical test of the best-fit tree or assignlevels of their absence in balearicines). of confidenceto particularnodes. Rather, it rep- Though most often associatedwith Grus resents an attempt to evaluate conservatively grus (Johnsgard1983), G. canadensisappears as the effect of "internal inconsistency" in the a relatively isolated species,perhaps an ancient original data.The relationshipbetween the best- branch from the antigoneline. The Anthropoides fit tree and the jackknife strict consensus(JSC) speciesare sistergroups (Figs. 5 and 6), which, tree is not, in fact, perfectly clear. Given the though now expected,has not alwaysbeen rec- distance matrix and the tree-building algo- ognized. The antigoneand grus clusters are es- rithm, our method of estimation directs us to sentially those recognized by earlier system- choosethe best-fit solution. This solution may atists(Krajewski 1988). correctly report the evidential meaning of the Within the antigoneand grusclusters, the DNA data, even though the data set containssome resultsprovide little resolution.In the antigone internal inconsistency.Jackknifing does not an- cluster, the best-fit tree associatesantigone with swer this question about estimation, it merely vipiorather than rubicunda.Taxonomic tradition detectsthe points on which the data are equiv- has favored a sister-grouprelationship for an- ocal. To this extent, a JSC tree representsa con- tigoneand rubicunda,since Sharpe (1894) united servative interpretation of the distancedata. the two in the genus"Antigone" on the basisof Cranephylogeny.--The phylogenetic hypoth- plumage similarities. The behavioral (Archi- eses(Figs. 5 and 6) are self-explanatory,but the bald 1975,1976) and anatomical-phenetic(Wood associationsof species merit some additional 1979) studies support this associationbased on comments.The degree to which the DNA phy- similarity. Moreover, natural hybridization has logeny concurswith previousopinions of re- been documentedbetween antigonesharpei and lationships is striking. The limited DNA hy- rubicundawhere the two are sympatricin north- bridization data (Sibley and Ahlquist 1985) easternAustralia (Johnsgard1983). produce a phenetic branching order for four The G. gruscluster (Fig. 6) is even more enig- speciesconsistent with Figures 6 and 7. The matic.The proposalArchibald (pers.comm.) de- fitted distancesin Sibley and Ahlquist (1985, veloped on the basisof behavior and anatomy fig. 3) are only slightly differentfrom thoseI is noncladistic but presents a hypothesis for found, although their measurementsare in units subsequent investigation. Archibald considers of delta T50H. G. grus as the earliest derivative within the With one notable exception,the compositions group, a position now reflected by its wide- October1989] CranePhylogeny 613 spread distribution (the eastern end of its Pale- genetic divergence among gruine species, it arctic range brings it into close geographic seemsmost reasonable to endorseIngold's (1984) proximity to all of the other specieswithin the recommendationthat Anthropoidesand Bugera- cluster).G. japonensis,G. monachus,and G. nigri- nusbe mergedwith Grus(see Sibley et al. 1988, coiliseach display relatively specializedbehav- for typical geneticdivergences at variouslevels ioral repertoiresand habitat requirements,and among avian taxa). Until further phylogenetic they have restrictedranges in the easternPale- resolution within Gruscan be obtained, ! sug- arctic. The relationships among them are un- gest that the informal designationof "Species clear (Johnsgard[1983] speculates that monachus Group" be retained for the terminal clusters is the "nearestliving relative" of grus,but pre- (Fig. 6). sents no argument to support the claim). G. The DNA data support the traditional americana,the only Nearctic representativeof subfamilial distinction of balearicines from the gruscluster, most closely resembles japonen- gruines, though they do not addressthe cor- sisin anatomy,and grusin behavior. Resolution rectness of current partitions of extinct taxa of relationshipsat this level will requirefurther amongthe subfamilies.A classificationof extant study. species,consistent with the DNA hybridization Lanyon (1985a) warned that a breakdown of resultsand entailing minimal departuresfrom resolution among groups upon jackknifing traditional views, is given below. should not be viewed to imply that they orig- Family Gruidae inated simultaneously.In general,this caution is well-taken. However, the small internodal SubfamilyBalearicinae Genus Balearica distanceson the best-fitcrane tree suggestthat B. pavonina the ancestorsof the antigone,Bugeranus/Anthro- B. regulorum poides,and grusclusters became isolated over a SubfamilyGruinae short span of time. This assumesthat genetic Genus Grus distancesassayed by DNA hybridization be- have more or less like a "molecular clock," a SpeciesGroup Leucogeranus G. leucogeranus notion endorsed cautiously by Sibley et al. SpeciesGroup Antigone (1987), and which is a reasonablyaccurate in- G. antigone terpretation of the crane data (Krajewski 1988, G. rubicunda in prep.). G. vipio Rates of molecularevolution.--ANOVA and SpeciesGroup Canadensis Felsenstein's (1986) F-ratio tests revealed that G. canadensis rate variation implied by DNA hybridization SpeciesGroup Anthropoides distancesamong the cranesis significant(Kra- G. [="Anthropoides"]virgo jewski 1988,in prep.). The rate disparityseems G. [="Anthropoides"]paradisea to be localizedin the Balearicalineage, which G. [="Bugeranus"]carunculatus evolvessome 1.3-1.6 times fasterthan gruines. SpeciesGroup Grus Although G. leucogeranusappears to have a rel- G. grus atively shortroot-to-tip pathlength on the best- G. monachus fit tree (Fig. 6), rate variationamong gruines is G. americana not statisticallysignificant. G. nigricollis G. japonensis. NOMENCLATURAL AND CLASSIFICATORY RECOMMENDATIONS The compositionsof speciesgroups within Grusare similar to those given by Archibald Becausethe jackknife strict consensus tree (Fig. (1976), exceptthat group namesare chosenby 6) leavesrelationships among speciesclusters priority (i.e. Group Grus replaces Archibald's unresolved, it is not possible,on the basis of Group Americana). Authorities are easily re- DNA hybridizationdata alone, to proposea for- coveredfrom Peters(1934) or Sharpe(1894). mal revision of the Gruidae in which sister- ACKNOWLEDGMENTS groups are given coordinate rank. The genus Grus,as currentlyconceived, is probablypoly- I thank Jon Ahlquist, A. H. Bledsoe, Robert Blei- phyletic(Figs. 5 and6). In light of the very small weiss,Allan Dickerman,Rober• Gallagher,John A. 614 CAREYKRAJEWSlCI [Auk, Vol. 106

W. Kirsch,and Mark Springerfor their assistancein FELSENSTEIN,J. 1982. Numerical methods for infer- variousphases of thisproject. George Archibald, Claire ring evolutionarytrees. Q. Rev. Biol. 57: 379-404. Mirande, Scott Swengel, and the staff of the Inter- 1986. Distance methods: a reply to Farris. national Crane Foundationprovided the raw mate- Cladistics 2: 130-143. rial, as well as enthusiasm, for the work. I also thank 1987. Estimationof hominoid phylogeny David Stock of Stetson University for the Limpkin from a DNA hybridization data set. J. Mol. Evol. tissues,and GeorgeGee of the PatuxentWildlife Re- 26: 123-131. searchCenter for additional bloodsamples of Whoop- INGOLD,J.L. 1984. Systematicsand evolution of the ing Cranes.Gary Graves(National Museum of Nat- cranes (Aves: Gruidae). Ph.D. dissertation, Ox- ural History)and PattyMcGill-Harelstead (Brookfield ford, Ohio, Miami Univ. Zoo) provided trumpeter tissues,which, though not --, S. I. GIJTTMAN,& D. O. OSBORNE.1987. Bio- included, provided valuable perspectiveson data chemicalsystematics and evolution of the cranes analysis.Fred Sheldonand JoelCracraft reviewed the (Aves: Gruidae). Pp. 575-584 in Proceedingsof manuscript. This study was supported by NSF re- the 1983 international crane workshop (G. W. searchgrants BSR-8320514 and BSR-8503687to John Archibald and R. F. Pasquier,Eds.). Baraboo, Wis- Kirsch, by the University of Wisconsin Graduate consin, Int. Crane Foundation. School, the University of Wisconsin Zoological Mu- JOHNSGARD,P.A. 1983. Cranes of the world. Bloo- seum,and the Department of Zoology, University of mington, Indiana Univ. Press. Wisconsin-Madison. Contribution No. 7 from the KOHNE,D. E. 1970. Evolution of higher-organism University of Wisconsin Zoological Museum Labo- DNA. Q. Rev. Biophys.33: 327-375. ratory of Molecular Systematics. --, & R. J. BRITTEN.1971. Hydroxyapatite techniques for nucleic acid reassociation.Pp. 500-512 in Procedures in nucleic acid research (G. L. Can- LITERATURE CITED toni and D. R. Davies, Eds.). New York, Harper and Row. ALTENBERG,L. C., M. J. GETZ,& G. F. SAUNDERS.1975. KRAIEWSKI,C. 1988. Phylogenetic relationships '2•I in molecular hybridization experiments.Pp. amongcranes (Aves: Gruidae) based on DNA hy- 325-342 in Methods in cell biology (D. M. Pres- bridization. Ph.D. dissertation, Madison, Univ. cott, Ed.). New York, Academic Press. Wisconsin. ANDERSON,D. g., & W. R. FOLK. 1976. Iodination of 1989. Comparative DNA reassociationki- DNA. Studies of the structure and iodination of netics of cranes. Biochem. Gen. 27: 131-136. papovavirusDNA. Biochemistry15: 1022-1030. LANYON,S. g. 1985a. Detecting internal inconsis- ARCHIBALD,G.W. 1975. The unison call of cranes as tencies in distance data. Syst. Zool. 34: 397-403. a useful taxonomic tool. Ph.D. dissertation, Ith- --. 1985b. Molecularperspective on higher-level aca, New York, Cornell Univ. relationships in the Tyrannoidea (Aves). Syst. 1976. Crane taxonomy as revealed by the Zool. 34: 404-418. unison call. Pp. 225-251 in Crane researcharound 1987. Jackknifing and bootstrapping:im- the world (J. C. Lewis and H. Masatomi, Eds.). portant "new" statistical techniques for orni- Baraboo, Wisconsin, Int. Crane Foundation. thologists.Auk 104: 144-146. BRODKORB,P. 1967. Catalogueof fossilbirds, part 3 MANIATIS, T., E. F. FRITSCH,& J. SAMBROOK. 1982. (Ralliformes, Ichthyornithiformes, Charadri- Molecular cloning. Cold Spring Harbor, Cold iformes). Bull. Florida State Mus., Biol. Sci. 2: 99- Spring Harbor Lab. 220. OROSZ,J. M., & J. G. WETMVR. 1974. In vitro iodin- CHAN, H.-C., W. T. RUYECHAN,& J. G. WETMUR. 1976. ation of DNA. Maximizing iodinationwhile min- In vitro iodinationof low complexitynucleic acids imizingdegradation; use of buoyantdensity shifts without chain scission.Biochemistry 15: 5487- for DNA-DNA hybrid isolation.Biochemistry 13: 5490. 5467-5473. COMMORFORD,S.L. 1971. Iodination of nucleic acids PETERS,J. L. 1934. Check-list of birds of the world, in vitro. Biochemistry10: 1993-2000. vol. 2. Cambridge,Harvard Univ. Press. CRACRAFT,J. 1973. Systematicsand evolution of the PRENSKY,W. 1976. The radioiodination of RNA and Gruiformes (Class Aves). 3. Phylogeny of the DNA to high specificactivities. Pp. 121-152 in Suborder Grues. Am. Mus. Nat. Hist. Bull. 151: Methodsin cell biology(D. M. Prescott,Ed.). New 1-127. York, Academic Press. 1987. DNA hybridization and avian phy- SARXCH,V. M., C. W. SCHM•D,& J. MARKS. 1989. DNA logenetics.Pp. 47-96 in Evolutionarybiology, vol. hybridizationas a guide to phylogenies:a critical 21 (M. K. Hecht, B. Wallace, and G. T. Prance, analysis.Cladistics 5: 3-32. Eds.). New York, Plenum. SCHERBERG,N.H., & S. REFETOFF.1975. Radioiodine DAVIS,M. B. 1973. Labeling of DNA with •251.Car- labelingof ribopolymersfor specialapplications negie Inst. WashingtonYear Book72: 217-221. in biology. Pp. 343-359 in Methods in cell biol- October1989] CranePhylogeny 615

ogy, vol. 10 (D. M. Prescott,Ed.). New York, Ac- --, & F. H. SHEEDON.1987. DNA hy- ademic Press. bridization and avian phylogenetics: reply to SHARPE,R. g. 1894. Catalogueof birds in the British Cracraft.Pp. 97-125 in Evolutionarybiology, vol. museum, vol. 23. London, Brit. Mus. (Nat. Hist.). 21 (M. K. Hecht, B. Wallace, and G. T. Pranceß SHELDON,F. H. 1987a. Phylogeny of herons esti- Eds.). New Yorkß Plenum. mated from DNA-DNA hybridization data. Auk SPRINGER,M. S., & J. A. W. KIRSCH. 1989. Rates of 104: 97-108. single-copyDNA evolution in phalangeriform 1987b. Ratesof single-copyDNA evolution marsupials.Mol. Biol. Evol. 6. In press. in herons. Mol. Biol. Evol. 4: 56-69. -, & C. Krajewski. 1989. DNA hybridizationin SIBLE¾,C. G., & J. E. AHLQWS?.1981. The phylogeny animal taxonomy:a critique from first principles. and relationships of the ratite birds as indicated Quart. Rev. Biol. In press. by DNA-DNA hybridization. Pp. 301-335 in Evo- $TORER,R.W. 1971. Classificationof birds. Pp. 1-18 lution today (G. G. E. Scudderand J. L. Revealß in Avian biology, vol. 1 (D. S. Farher and J. R. Eds.).Pittsburghß Carnegie-Melon Univ. King, Eds.).New York, AcademicPress. ß & •. 1983. The phylogeny and classi- TEREBA,A., • B. J. MCCARTHY. 1973. Hybridization fication of birds based on the data of DNA-DNA of '2•I-labeledribonucleic acid. Biochemistry12: hybridization. Pp. 245-292 in Current ornithol- 4675-4679. ogy (R. F. JohnstonßEd.). New YorkßPlenum. WALKINSHAW,L. I-•. 1964. The African Crowned , & --. 1985. The relationshipsof some Cranes. Wilson Bull. 76: 355-377. groups of African birds, based on comparisons WETMORE,A. 1934. A systematicclassification of the of the geneticmaterialß DNA. Pp. 115-161 in Pro- birds of the world, revised and amended. Smith- ceedingsof the internationalsymposium on Af- sonian Misc. Coil. 89: 1-11. rican vertebrates (K.-L. Schuchmann, Ed.). Bonn, 1960. A classification of the birds of the Zool. Forschungsinst.Mus. Alexander Koenig. world. Smithsonian Misc. Coil. 117(4). ß --, & B. L. MUNROE JR. 1988. A classi- WooD, D.S. 1979. Phenetic relationshipswithin the fication of living birds of the world based on family Gruidae. Wilson Bull. 91: 384-399. DNA-DNA hybridization. Auk 105: 409-423. 616 c,•v KRAIEWSKI [Auk,Vol. 106

APPENDIX. Tabulation of experimental results. These data represent the results of individual pairwise DNA hybridization comparisonsand are recordedas the delta Tmvalues for each experiment. Raw data (i.e. lists of radioactivecounts) for the constructionof melting profiles or for the calculationof distancestatistics are available from the author and are filed permanently with J. A. W. Kirsch at the Department of Zoology, University of Wisconsin-Madison. What follows is a summary,by tracerspecies, of all measureddistances among cranes. For eachcombination of tracer/driver species,I listed calculateddelta T• values.Averages, standard deviations, and sample sizes are given in Table 2. Two tracer preparationswere employed for each species.Each is identified by an extractionnumber at the top of the record (in parentheses),along with its averagehomologous Tm (in degrees Centigrade). For each driver species,results involving different tracer preparationsare offsetby colons (:). Resultsof replicate drivers involving the sameDNA extractionare offsetby commas(,); replicatesinvolving different driver DNA extractionsare offset by semicolons(;). Trimmed values are shown in parentheses. Asterisks(*) denote distancesmeasured between different conspecificindividuals; other conspecificdistances involve different driver preparationsfrom the individual usedas tracer. Delta T,, valuesbetween cranesand the American Coot (Fulicaamericana) are given to establisha frame of reference,but they were not included in the phylogeneticanalyses discussed in the text. EachDNA extractionmay be traced to an individual bird. Detailed information on individuals, including pedigrees,is on file at the InternationalCrane Foundation in Baraboo,Wisconsin. Inquiries should be addressed to the author.

Tracer Balearicaregulorum (281:84.13, 290:83.28) G. antigone 1.56, 1.61; 2.27, 0.79: 1.06, 1.76 Driver spp. Delta T• G. rubicunda 1.88, 1.68; 0.49: 1.10, 0.56 G. vipio 0.90, 1.00; 1.72, 1.21, 1.29: 1.92, 1.70 B. pavonina 1.09, 0.63 G. grus 1.22, 0.93; 0.80:1.57 B. regulorum 0.08 G. monachus 1.38, 1.09:0.85 A. virgo 3.61: 3.82, 3.86;4.11 G. americana 1.62, 1.66; 1.06 A. paradisea 2.98:3.48, 3.60; 4.55, 4.75 G. nigricollis 2.34, 1.23;0.64:0.27 Bugeranus 4.73,5.71; 3.75, 3.46: 4.08, 3.91;4.08, 3.90 G. japonensis 1.56, 1.64; 1.60: 1.99, 1.71 G. leucogeranus 3.80, 2.31: 3.01;2.98, 3.57, 2.97; 3.53, 3.63 Aramus 6.52, 6.98, 5.90 G. canadensis 3.34, 4.39: 3.61, 4.33; 3.96, 3.86; 3.98, 3.86 F. americana 12.46:12.61 G. antigone 3.11, 3.12: 3.91, 4.19; 4.21, 3.88; 3.87, 3.73 G. rubicunda 3.04, 3.21: 4.08, 4.18; 3.54 Tracer Bugeranuscarunculatus (392:85.20, 392:83.90) G. vipio 3.37, 3.16: 4.77, 4.55; 4.54, 4.20; 4.22, 4.42 Driver spp. Delta T• G. grus 3.22,3.71: 3.30, 2.93; 3.69 G. monachus 3.72: 3.92, 4.52; 3.27 B. pavonina G. americana 3.83, 3.89:3.61 B. regulorum 4.42, 4.03: 3.83, 3.39;4.12 G. nigricollis 3.28:3.99, 3.84;3.92 A. virgo 0.99, 1.23:1.01 G. japonensis 2.61,3.31: 3.62, 3.83; 3.58, 3.90; 4.13, 3.43 A. paradisea 1.45, 1.36; 1.04: 0.60, 1.07; 1.34, 3.23 Aramus 7.33: 7.56, 7.33 Bugeranus 1.93:0.67 F. americana 12.22, 12.62 G. leucogeranus 1.92, 1.29; 1.35, 2.44: 1.24; 1.60, 1.98, 0.56 G. canadensis 1.68, 1.54; 1.14, 1.22: 1.09, 0.78; 1.38, 1.06 TracerAnthropoides virgo (390:81.65, 534:84.41) G. antigone 2.33, 1.56; 1.19,0.93: 1.84, 1.55; 1.90, 1.53 G. rubicunda 1.26, 1.81; 0.89: 1.96, 1.57; 1.59 Driver spp. Delta Tm G. vipio 2.07, 2.02; 1.56, 1.37: 1.27, 2.64; 2.52, 2.19 B. pavonina 4.81 G. grus 1.40, 1.57; 1.29, 0.13 B. regulorum 3.63, 3.62: 3.67, 3.79 G. monachus 1.10, 1.47; 1.16: 2.05, 1.69; 1.29 A. virgo -0.44*: -0.70* G. americana 1.69, 1.65; 0.93, 1.72 A. paradisea 0.92,0.18; 0.54, 1.17:0.45, 0.60; 0.47 G. nigricollis 1.24, 1.41; 1.30: 1.26, 1.23 Bugeranus 0.85,0.79; 0.45, 0.90: 1.45, 1.35; 1.21, 1.45 G. japonensis 1.31, 1.17;0.94, 1.23:0.19, 0.28; 1.11, 1.45 G. leucogeranus 1.63,1.43; 1.92, 2.51: 1.88, 1.69; 1.10, 0.71 Aramus 6.94, 6.94: 7.20, 6.09 G. canadensis 1.86, 1.25; 0.28, 0.43: 1.13, 1.09; 2.06, 1.79 F. americana 12.36 G. antigone 0.64, 1.34;0.82, 0.95: 1.13, 0.98; 2.07, 2.00 G. rubicunda 1.69, 0.54; 0.55: 1.31, 1.09; 1.07, 1.11 TracerGrus leucogeranus (136:85.57, 137:85.06) G. vipio 1.59, 1.89;0.82, 0.95: 1.64, 1.46; 1.35 Driver spp. Delta T• G. grus 1.30,0.10; 0.81: 1.14, 1.20; 1.65 G. monachus 0.16, 0.16: 1.24, 1.16; 2.03 B. pavonina G. americana 0.81, 1.11: 2.56, 1.31 B. regulorum 4.97; 3.95, 4.18: 3.66, 3.53; 3.54, 3.35 G. nigricollis 0.74,0.59: 2.15, 1.74;1.79 A. virgo 3.02, 3.01: 1.12, 1.08; 1.85 G.japonensis 0.85,0.74; 0.59, 0.81: 1.94, 1.93; 2.08, 2.06 A. paradisea 3.15, 3.45: 1.17, 1.18; 2.04, 2.50 Aramus 7.01, 6.21: 6.70, 6.58 Bugeranus 2.19, 2.66: 1.12, 0.91; 1.38, 1.40; 1.04, 1.05 F. americana 11.46:11.49 G. leucogeranus 0.41'; 0.31' G. canadensis 1.03, 1.26: 2.00, 1.15; 1.27, 1.26; 1.78, 1.79 TracerAnthropoides paradisea (282:84.50, 391:84.50) G. antigone 1.18, 1.09: 1.39, 1.49; 1.38, 1.29; 1.32, 1.40 G. rubicunda 1.35, 1.25: 1.43, 1.48; 1.55, 1.62; 1.88, 2.88 Driver spp. Delta Tm 1.50, 1.37: 1.68, 1.54; 1.74, 1.95; 1.99, 2.10 B. pavonina G. grus 1.56, 1.36: 0.82, 0.81; 1.43, 1.03 B. regulorum 3.98,4.28; 3.52, 3.65: 3.35, 3.94 G. monachus 1.22, 2.92: 1.45, 1.41; 1.43, 1.14 A. virgo 0.64, 1.54; -0.08, -0.04: -0.05, 0.18 G. americana 1.71, 2.16: 1.36; 0.79 A. paradisea 0.48*; -0.56* G. nigricollis 1.57, 1.68: 1.30, 1.79; 1.11 Bugeranus 2.20,(4.24); 1.15, 1.12: 0.75, 0.46 G. japonensis 1.11, 1.23: 0.61, 1.19; 1.48, 1.60; 1.29, 2.09 G. leucogeranus 1.18,1.40; 1.61, 1.25: 1.72, 2.39 Aramus 6.43, 7.21: 6.78, 6.90 G. canadensls 1.56, 2.46; 0.94, 0.66: 1.63, 1.89 F. americana 12.40, 12.67 October1989] CranePhylogeny 617

APPENDIX. Continued.

Tracer Gruscanadensis (143:84.75, 393:84.20) G. canadensis 0.98, 0.82, 1.47: 0.10, 0.37; 0.80 -0.20, 0.68, -0.56: -0.09, 0.01; 0.54 Driverspp. DeltaT• G. antigone G. rubicunda 0.40, 0.02, 0.89; 0.54, 0.58, 0.36 B. pavonina G. vipio -0.36', 0.60*: -0.78* B. regulorum 3.94, 3.85: 3.81, 4.12 G. grus 0.32, 0.29, 1.55: 0.98, 0.76 A. virgo 1.62,1.86: 1.19, 1.55; 1.22 G. monachus 0.13, 0.01, 1.72: 0.70, 0.88 A. paradisea 1.09,1.29; (3.64): 1.19, 1.21; 0.79 G. americana 0.66, 0.47, 1.77: 1.22, 0.72 Bugeranu$ 1.21,1.26; 1.32: 1.39, 1.12; 1.05, 0.70 G. nigricollis 1.69, 0.40, 0.80: 0.92, 1.22 G. leucogeranus 1.82,1.42; 1.34, 1.41: 1.94; 1.36, 0.86 G. japonensis 1.38, 1.10, 0.35: 0.21, 0.33, 0.82 G. canadensis 0.03*: 0.13' Aramus 6.14, 5.62, 6.63: 5.90, 5.53 G. antigone 1.44,1.90; 1.64, 1.28: 1.71, 1.22; 1.51, 1.76 F. americana 10.94 G. rubicunda 0.69, 1.62; 1.62, 0.60: 0.96, 1.25; 2.08, 1.15 G. vipio 1.61,1.37; 1.10, 1.48: 1.69, 1.33; 1.66, 1.02 TracerGrus grus (285:83.16, 293:84.36) G. grus 1.28,1.16; 1.42: 1.34, 1.09; 1.08 Driver spp. Delta T• G. monachus 1.18, 1.14; 1.14: 2.46, 1.30; 1.11 G. americana 1.32, 1.26; 1.28: 1.46, 1.13 B. pavonina 2.63, 2.78 G. nigricollis 2.22, 2.96; 0.82: 1.43, 1.81 B. regulorum 3.39, 3.34: 3.85, 4.34 G.japonensis 1.24,1.72; 1.46, 1.28; 0.74, 1.07; 1.55, 1.23 A. virgo 0.56, 0.37: !.56, 1.34 Aramus 7.24, 7.29: 7.46, 7.11 A. paradisea 0.72, 0.40: 1.04, 1.20; 1.49 F. americana 12.69:12.50 Bugeranus 1.10, 1.17: 1.15, 1.94; 0.88 G. leucogeranus 1.78, 1.06; 0.65, 1.54: 1.43, 1.46; 1.10, 1.11 TracerGrus antigone (138:84.98, 395:85.02) G. canadensis 0.67, 1.49; 0.28, 0.09: 1.55, 1.82; 0.97, 1.31 G. antigone 1.25, 0.64; 0.36, 0.41: 2.40, 1.21; 1.07, 1.19 Driver spp. DeltaT• G. rubicunda 0.60, 1.06; 0.39, 1.27: 1.17; 1.33 B. pavonina G. vipio 1.31; 0.65, 0.43; 1.85: 1.63, 1.79; 1.36, 1.76 B. regulorum 3.08, 3.31: 4.12, 4.03 G. grus 0.77', -0.24 A. virgo 2.97,2.27: 1.40, 0.93; 1.38 G. monachus -0.95, -0.42: 0.08; -0.04, 0.38 A. paradisea 2.22,2.91: 1.44, 1.58;1.73 G. americana 0.03, 0.18: 0.26, 0.12 Bugeranus 2.07,1.55: 1.49, 0.79; 1.53, 0.87 G. nigricollis 0.29: 1.08, 1.13 G. leucogeranus 0.78,0.64: 1.10, 1.05; 1.04, 1.20 G. japonensis 1.11, 0.38; 0.45, 0.57: 0.09, 0.23; 0.13, 0.38 G. canadensis 0.58, 0.66: 1.25, 1.23; 1.34, 1.95 Aramus 5.58, 5.83: 6.96, 8.42 G. antigone 0.43*, 1.13'; 0.89* F. americana 10.64, 10.40:13.04 G. rubicunda 0.50, 0.20: 0.67, 0.77; 0.65, 0.57 G. vipio 0.48,0.46: 0.99, 1.22; 1.43, 0.97 Tracer Grus monachus(286:84.29, 401:85.08) G. grus 1.39,0.91: 1.19, 1.12; 1.40 Driver spp. Delta T• G. monachus 3.07, 1.07: 1.78, 1.34; 1.73 G. americana 1.36, 3.34: 1.74, 1.22 B. pavonina G. nigricollis 1.29,2.79: 1.42, 1.61; 1.52 B. regulorum 4.39, 3.64: 4.48, 4.47 G. japonensis 1.28,1.52: 0.90, 1.37; 1.41, 1.33 A. virgo 1.04, 1.07: 1.56; 1.11, 0.78 Aramus 5.67, 6.11: 6.79, 6.92 A. paradisea 0.61, 0.22: 2.07, 2.12 F. americana 12.57 Bugeranus 2.80, 0.87: 1.33, 1.17; 1.90, 1.98 G. leucogeranus 1.17, 3.70; 1.58, 1.59; 0.81, 1.43: 1.30, 1.84; 0.95, Tracer Grusrubicunda (396:84.25, 397:84.82) 0.59 G. canadensls 1.64, 1.50; 1.13, 1.16: 1.94, 1.81; 1.43, 1.20 Driver spp. DeltaT,• G. antigone 2.71, 2.07; 2.32, 1.20; 1.30, 1.67: 1.72, 1.71; 1.58, B.pavonina 4.42,4.39 1.30 B. regulorum 3.56, 3.33: 3.78, 3.79 G. rubicunda 1.60, 1.75; 1.40: 1.47, 1.57; 1.35, 1.26 A. virgo 1.00,1.02: 1.83, 1.14; 1.48 G. vipio 0.70, 0.90; 1.60, 1.38; 1.06: 1.89, 2.00, 1.90; 1.55 A. paradisea 0.70, 0.75: 1.28, 0.88; 1.11 G. grus -0.07, 1.02; 0.37: 0.30, 0.16; 0.48 Bugeranus 1.23, 1.51;1.14, 1.32: 1.07, 0.86; 1.93, 1.49 G. monachus -0.01': 0.65* G. leucogeranus 1.03,1.53; 1.10, 1.21:1.99, 1.85; 0.74, 0.72 G. americana 0.52, 0.27; 1.20: 0.54; 0.32. 0.42 G. canadensis 0.91, 0.99; 1.19, 0.88: 1.81, 1.70; 1.34, 1.02 G. nigricollis (3.34), 1.81;0.18: 0.23, 0.21; 0.35 G. antigone 0.60, 1.57;0.94, 0.78: 0.28, 0.18; 0.20 G. japonensis 0.42, 0.66; 0.57:1.11, -0.23: 0.05, 0.40; 0.35, 0.59 G. rubicunda 1.07', 0.06* Aramus 6.81, 7.36: 7.41, 7.23 G. vipio 1.46,1.18; 1.45, 1.11: 1.07, 1.03; 0.93, 0.85 F. americana 14.89:12.76 G. grus 0.68, 1.95: 1.14, 1.13; 1.14 G. monachus 1.28, 0.81: 1.30, 1.08; 1.99 Tracer Grusamericana (287:83.58, 402:84.62) G. americana 2.13, 1.75: 1.25, 1.12 Driver spp. Delta T,, G. nigricollis 2.07: 1.16, 1.19; 1.42 G. japonensis 1.05,0.97; 2.83, 2.11: 0.89, 0.93; 1.19, 1.29 B. pavonina 4.35 Aramus 7.41, 7.01: 6.98, 6.54 B. regulorum 3.10, 2.88: 4.23, 3.80 F. americana 11.72, 12.66:12.75 A. virgo 0.70, 0.85: 1.64, 1.47; 1.06 A. paradisea 0.70, 0.76; 0.69: 1.66, 1.74; 1.42 TracerGrus vipio (41:82.94, 284:81.20) Bugeranus 0.41, 0.67; 0.52: 1.60, 1.68; 1.27, 1.27 G. leucogeranus 1.84; 1.33, 1.03: 2.38; 1.96 Driver spp. DeltaT,, G. canadensis 1.52, 1.48;0.62, 0.70: 1.96, 2.01; 1.86, 1.79 B. pavonina G. antigone 0.44, 0.34; 0.57, 0.71: 1.59, 1.20; 1.30, 1.91 B. regulorum 2.35, 2.17, 3.83:3.52 G. rubicunda 0.66, 0.60; 0.47, 0.81: 2.05, 2.33; 2.16, 1.90 A. virgo 0.53,0.13, 1.64:0.70, 0.70 G. vipio 1.08, 1.06; 1.97, 1.28: 2.84, 2.16; 1.37, 1.84 A. paradisea 0.24,0.37, 1.20: 0.78, 0.80; 2.10 G. grus 0.44, -0.52; 0.32: 0.60, 1.03; 0.27 Bugeranus 0.80,0.79, 1.79: 0.46,0.08; 1.00 G. monachus 0.52, 0.20; 0.39: 0.64, 0.78; 1.14 G. leucogeranus 1.04,0.76, 1.48:2.09, 1.00,2.68 G. americana 0.54 618 CAREYKRAJEWS• [Auk, Vol. 106

APPENDIX. Continued.

G. nigricollis 0.50; 0.77, 0.74: 0.63, 0.45; 0.69 Aramus 6.22, 6.27: 7.11, 6.03 G. japonensis 0.72, 0.06; 1.19, 0.78: 0.89, 0.89; 0.66, 0.52 F. americana 10.78:11.84 Aramus 6.48, 5.94: 8.12, 7.52 F. americana 11.78:13.16 Tracer Grusjaponensis (139:85.44, 404:84.73) Driver spp. Delta T,, Tracer Grusnigricollis (403:81.56, 403:81.58) B. pavonina 4.63 Driver spp. Delta T• B. regulorum 4.21, 3.34: 4.19, 4.09 A. virgo 2.21; 1.74: 1.64, 1.17 B. pavonina A. paradisea 2.37: 1.80, 1.76; 1.42 B. regulorum 3.33, 3.34: 3.36, 3.01 Bugeranus 1.92, 1.81: 1.18, 1.50; 1.39, 1.12 A. virgo 1.53, 2.41: 1.79, 0.68 G. leucogeranus 1.00, 2.79; 2.63, 2.80: 1.78, 1.57; 1.16 A. paradisea 0.80, 0.98; 1.88: 1.19, 0.85; 0.94 G. canadensis 2.32, 1.21; 1.63, 1.96: 1.68, 1.53; 1.76, 1.88 Bugeranus 1.10, 1.70; 1.40: 1.30, 0.68; 1.23, 0.97 G. antigone 1.74, 1.09; 1.11, 1.54: 1.59, 1.37; 1.63, 1.63 G. leucogeranus 3.14, 1.50; 2.02, 1.00: 0.88; 0.94, 0.77 G. rubicunda 0.58, 1.15: 1.44, 1.60; 1.31, 1.54 G. canadensis 1.28, 2.41; 1.35, 2.53: 1.56, 0.97; 1.74 G. vipio 0.65, 0.85; 2.36, 1.64: 1.39, 1.43, 1.70 G. antigone 2.11, 0.84; 1.35, 1.53: 1.42, 1.39; 1.58, 1.53 G. grus 0.33: 0.53, 0.51; 0.54 G. rubicunda 1.52, 1.41; 1.51: 2.54, 1.62; 2.79, 2.60 G. monachus 0.97: 0.51, 0.62; 1.22 1.43, 1.84; 1.45, 1.98: 1.65, 1.39; 1.70, 1.29 G. americana 0.89, 0.88: 0.45, 0.72; 1.08 G. grus 0.79, 1.04; 0.13: 0.19, -0.42; 0.35 G. nigricollis 0.57: 0.41, 0.65; 0.80 G. monachus 0.13, 1.47; 0.01: 0.24, 0.98; 0.71 G. japonensis 0.65:0.11 G. americana 0.63, 0.34; 0.51: 2.23, 0.47 Aramus 7.10, 1.15: 7.17, 7.09 G. nigricollis 0.50: 0.82 F. americana 13.24, 12.27 G. japonensis 1.46, 0.34; 0.74, 0.47: 0.11, 0.07; 1.51, 1.01