The Auk 117(1):22-32, 2000 MOLECULAR PHYLOGENY OF JACANAS AND ITS IMPLICATIONS FOR MORPHOLOGIC AND BIOGEOGRAPHIC EVOLUTION LINDA A. WHITTINGHAM, TMFREDERICK H. SHELDON,2 AND STEPHENT. EMLEN3 1Departrnentof BiologicalSciences, University of Wisconsin-Milwaukee,Milwaukee, Wisconsin 53201, USA; 2Museumof Natural Science,119 FosterHall, LouisianaState University, Baton Rouge, Louisiana 70803, USA;and 3Sectionof Neurobiologyand Behavior, Division of BiologicalSciences, Cornell University, Ithaca, New York 14853, USA ABSTRACT.--Wecompared sequences of mitochondrialcytochrome-b and ND5 genesin a phylogeneticanalysis of sevenspecies of jacanas,representing all six generaand including the GreaterPainted-snipe (Rostratula benghalensis) asan outgroup.When analyzedseparately by parsimonyand maximum-likelihoodbootstrapping, the two genesproduced consistent trees,although the ND5 tree was better resolvedthan the cytochrome-btree. When com- bined,the data from the two genesproduced a fully resolvedtree that was identicalto the ND5 tree.This tree had the followingform: ((((Irediparra,Microparra), Metopidius), Actophi- lornis),((Jacana jacana, J. spinosa), Hydrophasianus) ), Rostratula. The phylogenyconsists of two majorclades that wereknown to traditionaland phylogenetictaxonomists. It alsocontains sistertaxa that are geographicallydisjunct: the New WorldJacana and Asian Hydrophasianus, and the AfricanMicroparra and AustralianIrediparra. We postulatethat this biogeographic pattern resultsfrom the extinctionof interveningAfrican and Asian taxa, respectively.Re- ceived31 May 1998,accepted 26 April 1999. JACANASare worldwide inhabitantsof trop- Sibleyand Ahlquist1990, Chu 1995).Strauch's ical and subtropicalopen wetlands.Eight ex- (1978)character-compatibility (clique) analysis tant speciesin six generaare recognizedin the of 63 skeletal and seven muscular characters of family Jacanidae.Four genera are monotypic shorebirdsdefined two groupsof jacanagen- and occuron three continents:Microparra (Af- era: (1) Hydrophasianusand Jacana,and (2) Me- rica), Irediparra (Australia), Hydrophasianus topidius,Actophilornis, Irediparra, and Micropar- (Asia), andMetopidius (Asia). Two othergenera ra. Following a critical appraisalof Strauch's consist of two species:Actophilornis africanus shorebirdstudy (Michevichand Parenti1980), (Africa) and A. albinucha(Madagascar), and Ja- Chu (1995)reanalyzed Strauch's data using cla- canajacana (South America) and J. spinosa (Cen- distic parsimony.His reanalysisindicated the tral America).The generaof jacanasare mainly monophylyof jacanasbut was not designedto allopatric, but some co-occur in portions of resolverelationships within the family. Sibley Asia (Hydrophasianusand Metopidius)and Af- and Ahlquist's(1990) DNA-DNA hybridization rica (Actophilornisand Microparra).The two Ja- study indicatedthat Actophilorniswas much canaspecies co-occur and hybridizeonly in a closerto Irediparrathan to Jacana.Unfortunate- small area of western Panama (Wetmore 1965). ly, Sibleyand Ahlquistdid not includeother ja- Until recently,relationships of the jacanasto canataxa in their comparisons. other groupsof birds were poorly known. It is To shedmore light on intergenericrelation- now clearthat they are membersof the Char- shipsin jacanas,we comparedDNA sequences adriiformesand probablyare most closelyre- of portionsof the mitochondrialcytochrome-b lated to painted-snipes,family Rostratulidae geneand the fifth subunitof nicotinamidead- (Kitto and Wilson 1966, Strauch1978, Sibley eninedinucleotide dehydrogenase gene (ND5) and Ahlquist1990). Within the Jacanidae,there of all sixgenera. These comparisons resolve the havebeen no thoroughmodern studies of in- jacanaphylogeny and providean initial assess- tergenericrelationships. The only systematic ment of the utility of ND5 sequencesfor study- studieshave been tangentsof larger efforts to ing avian phylogeny.Knowledge of the phy- reconstructshorebird phylogeny (Strauch 1978, logenyof jacanaspermits us to speculateon the evolutionaryforces that shapedtheir morphol- 4E-mail: [email protected] ogy and distribution. 22 January2000] JacanaPhylogeny 23 METHODS proportion of transversions(Moore and DeFillipis 1997); ti:tv was also estimated via the maximum- One individual from eachof sevenspecies repre- likelihoodoption in PAUP*(Swofford 1998). Phylo- sentingall six generaof the Jacanidaewere sampled genetic relationshipswere estimated using maxi- (Table 1): African Jacana (Actophilornisafricanus), mum-likelihood (randomized heuristic search) and LesserJacana (Microparra capensis), Comb-crested Ja- maximum-parsimony(branch and bound search) cana (Irediparragallinacea), Pheasant-tailed Jacana methodswith PAUP*.Bootstrapping values (100 rep- (Hydrophasianuschirurgus), Bronze-winged Jacana licates) were determined for both maximum-likeli- (Metopidiusindicus), Wattled Jacana(Jacana jacana), hood and parsimonymethods to assesssupport for and NorthernJacana (J. spinosa). The GreaterPainted- internalbranches. Cytochrome-b and ND5 datawere snipe (Rostratulabenghalensis) was used as an out- combinedand analyzedseparately to observethe be- group becauseof its apparentclose relationship to haviorof the genes. the jacanas(Sibley and Ahlquist 1990). Laboratorymethods.--In an attempt to avoid ampli- ficationof nuclear pseudogenes,we extractedDNA RESULTS from tissue,which has a relativelyhigh mitochon- drial DNA contentcompared with blood (Sorenson Characteristicsofsequence data.--We examined and Fleischer 1996, Quinn 1997). DNA was extracted a total of 705 nucleotides(345 of cytochromeb from 0.1 g of heart, liver, or muscletissue following and 360 of ND5) for sevenspecies of jacanas the standardphenol/chloroform extraction method and the outgroup(Rostratula). Average nucle- (Hillis et al. 1990).A portionof the mitochondrialcy- otide compositionsfor both genesand all po- tochrome-band ND5 geneswas isolatedand ampli- sitions were as follows: G = 11.3%, A = 30.2%, fied via the polymerasechain reaction (PCR). We used the following primers for cytochrome b T = 24.3% and C = 34.2%. Nucleotide compo- (H16065: GGAGTCTTCAGTCTCTGGTTTACAAGAC sitionalbias was highestat third-codonposi- and L15656: AACCTACTAGGAGACCCAGA; Helm- tions, intermediate at second positions, and Bychowskiand Cracraft 1993) and ND5 (H14149: lowestat first positions(Table 2). First positions CCTATTTTGCGGATGTCTTGTTC and L 13753: CAG- were biasedslightly, being G-poor and C-rich GAAAATCCGCTCAATTCGG; Garcia-Moreno et al. for cytochromeb andslightly T-poor and A-rich 1999). Primer sequencesare listed 5' to 3' and num- for ND5 (Table2). Secondpositions were G- bersrefer to the 3' baseof the primer referencedto poor and T-rich for both cytochromeb and the completemtDNA sequenceof the chicken(Gallus ND5. Third positionsof bothgenes were most gallus;Desjardins and Morais 1990).H and L refer to primerslocated on the heavyand light strandsof the heavily biasedtoward G- and T-poor and A- and C-rich. The total number of variable and mitochondrialgenome, respectively. PCR reactionswere done in 50-•L volumes using potentiallyphylogenetically informative sites 0.5 •M of eachprimer, 2.5 mM MgCI2,10 mM of each was similar for both cytochrome-band ND5 dNTP, and 3 U of Taqpolymerase. Thermal cycling fragments(Table 3). Althoughthe numberof began with 3.0 min at 94øCfor initial denaturation, variableamino acidsappeared lower for ND5 followedby 35 cyclesof denaturation(94øC, 1 min), than for cytochromeb, the numberof poten- annealing (46 to 50øC,1.5 min) and extension(72øC, tially informative amino acidswas similar (Ta- 1 min), and a final extensionof 10 min at 72øC.PCR ble 3). The numbersof variableand informative productswere run on a 2% low-meltagarose gel that was stained with ethidium bromide, allowing the siteswere highest at third- codonpositions, in- amplifiedDNA fragmentto be visualized under UV termediateat first positions,and lowest at sec- illumination and excisedfrom the gel. PCR products ond positions(Table 2). Thesetrends were ev- were purified and concentratedwith Wizard PCR identwhether or not the outgroupwas includ- PrepsKit (PromegaA7170) and then sequenceddi- ed with the jacanas. rectly on an ABI 373 Stretchautomated sequencer. The percentdivergence (using the Tamura Sequencesare deposited in GenBank under acces- and Nei [1993] correction)among taxa for cy- sion numbers AF146616 to AF146631. tochrome b and ND5 is summarized in Table 4 Dataanalysis.--Sequences were alignedby eye rel- and Figure 1. For cytochrome-bsequences, in- ativeto the chickensequence (Desjardins and Morals group divergenceranged from 2% (Jacanaja- 1990)using the editor and translatorin MEGA (Ku- mar et al. 1993).Base frequencies, variation, pairwise canato ]. spinosa)to 17.2%(Irediparra to ]acana transition and transversion values, and distances jacana). Similarly, ND5 ingroup divergence were determined with MEGA. Overall transition to ranged from 1.7% (Jacanajacana to J.spinosa) to transversionratio (ti:tv) was estimatedby plotting 17.4% (Microparrato Actophilornisand Micro- the pairwise proportion of transitionsversus the parra to Hydrophasianus).For the combined 24 WHITTINGHAM,SHELDON, AND EMLEN [Auk,Vol. 117 TABLE1. Distributionsand collectionlocalities of jacanaspecies sampled plus the outgroup. Speciesa Distribution Samplenumber b Collectionlocality African Jacana Central and southern Africa LSUMNS B19187 Dallas Zoo (Actophilornisafricanus) Lesser Jacana Central and southeastern Africa LSUMNS B23806 South Africa (Microparracapensis) Comb-crested Jacana Malaysia, northeasternAustralia