The Auk 112(3):672-679, 1995 PHYLOGENETIC RELATIONSHIPS OF THE ZIGZAG HERON (ZEBRILUS UNDULATUS) AND WHITE-CRESTED BITTERN (TIGRIORNIS LEUCOLOPHUS) ESTIMATED BY DNA-DNA HYBRIDIZATION FREDERICKH. SHELDONl, KEVIN G. MCCRACKEN2, AND KEELEY D. STUEBING 1 •Museuraof Natural Science,Louisiana State University, BatonRouge, Louisiana 70803, USA; and 2LouisianaCooperative Fish and WildlifeResearch Unit, Schoolof Forestry,Wildlife, and Fisheries, LouisianaState University, Baton Rouge, Louisiana 70803, USA ASSTRACT.--Recently,we acquiredDNA of two rare speciesof heron, the NeotropicalZigzag Heron (Zebrilusundulatus) and the African White-crestedBittern (Tigriornisleucolophus). To estimatetheir phylogenetic relationshipsto other herons, we comparedthese specieswith representativesof the major heron cladesusing DNA-DNA hybridization. Even though the Zigzag Heron resemblesa tiger-heron in its barred plumage and forest habitat, it is most closelyrelated to bitterns.The White-crestedBittern is monophyleticwith the New World tiger-herons (Tigrisoma)and, thus, is better termed the White-crestedTiger-Heron. These findings accordwell with phylogeneticanalyses based on osteology.The remaining uncer- taintiesin higher-level heron phylogenyare principally: (1) the positionand composition of someenigmatic genera (e.g. Gorsachius,Agamia, Pilherodius, and Ardeola);and (2) the iden- tification of the basal heron lineage, which appearsto be either tiger-heronsor the Boat- billed Heron (Cochlearius).Received 15 June1994, accepted 27 January1995. THE HERONS(Ciconiiformes: Ardeidae) may groups, the tiger-herons. This study supple- be divided ecologicallyinto groupsthat: (1) feed mentsa previousDNA-DNA hybridization ef- largely in openareas either by day (e.g.Egretta, fort to estimatethe intergenericphylogenetic Ardea,Bubulcus, and Butorides)or night (Nyctic- relationshipsof herons(Sheldon 1987a,see also orax, Nyctanassa,and Cochlearius);(2) live and Sheldon and Kinnarney 1993). At the time of nest in marshes (bitterns); or (3) live in forested the 1987 study, DNA was available from only areas (tiger-herons and Gorsachiusnight-her- one tiger-heron species--theRufescent Tiger- ons). Of these groups, the least known are the Heron (Tigrisomalineatum)--and, as a result, lit- forest-dwellingspecies, which generallyare rare tle could be said about the relationshipsamong and difficult to observe.In comparisonto other members of this group. Also, becausethe de- herons,they are poorly representedin museum termination of the position of the tiger-heron collections,have received little ecologicaland cladewithin the heronfamily relied on a single behavioral study, and have obscurephyloge- species,the overall estimateof heron phylog- netic relationships (Payne and Risley 1976, eny potentially sufferedfrom the limited sam- Hancock and Elliott 1978, Hancock and Kushlan ple representing this important group. Since 1984). They also exemplify some intriguing the earlier study,we haveobtained DNA of two evolutionary issues and problems. Forest- more tiger-herongenera (sensu lato): the White- dwelling heronshave undergone marked adap- crestedBittern (Tigriornisleucolophus) of Africa tive changesin apparent responseto their hab- and the Zigzag Heron (Zebrilusundulatus) of itat (especiallyin terms of their plumage),and South America. With Tigrisomalineatum, these they have relictual tropical distributions--the new speciesconstitute three of the four tradi- tiger-heronsoccur in the Neotropics,Africa, and tional tiger-heron genera. Unfortunately, a New Guinea, and Gorsachius occurs in Africa sampleof the fourth tiger-heron,the ForestBit- and Asia. tern (Zonerodiusheliosylus) of New Guinea, is In this paper we present DNA-DNA hybrid- still lacking. ization evidenceof the phylogeneticrelation- The inclusionof two more tiger-herongenera ships of membersof one of these little-known not only improvesthe likely accuracyand use- 672 July 1995] HeronPhylogeny 673 TAnLEI. Speciesand samplesused in study. Name Country Preparation no. Egrettathula (Snowy Egret) USA 3705 Cochleariuscochlearius (Boat-billed Heron) Ecuador 3281 Tigrisomalineatum (Rufescent Tiger-Heron) Ecuador 3165, 4507 Tigriornisleucolophus (White-crested Bittern) Liberia 1910 Zebrilusundulatus (Zigzag Heron) Ecuador 3170 Ixobrychusexilis (Least Bittern) USA 409 Plegadisfalcinellus (Glossy Ibis) USA 852 fulness of the DNA-DNA estimate of phylog- ization estimatesof phylogeny (e.g. Sheldon 1994, eny, but it permits a more substantialcompar- Slikas et al. 1996). ison between the DNA-DNA hybridization re- With the exceptionof Tigrisoma,only one individ- ual of eachspecies was compared.For Tigriornisand suitsand other studiesof heron phylogeny (e.g. Zebrilus,only one sample was available; for the other Bock1956, Curry-Lindahl 1971,Payne and Ris- species,samples from more individuals were avail- ley 1976). Payne and Risley's (1976) study is able, but degrees of individual variation were ex- particularlyuseful as a soundingboard, because pectedto be well below genetic differencesamong it is a thorough cladisticand phenetic analysis species(Sheldon 1987a,Bleiweiss and Kirsch 1993, of heron osteologyand, thus,may be compared Sheldon and Winkler 1993). We comparedtwo in- to the DNA-DNA hybridization resultsvia tax- dividuals of EcuadorJanTigrisoma: one from the east- onomic congruenceanalysis (e.g. Cracraft and ern (Amazonian) lowlands (tissue no. 3165; ANSP Mindell 1989, Bledsoe and Raikow 1990). catalog183558); and one from 1,500 m elevation on the easternslopes of the Andes(tissue no. 4507; ANSP METHODS catalogno. 185105).DNA of the latter was received late in the study and was included becausewe be- The taxa and samplesused in this study are listed lieved it to representa separatespecies, the Fasciated in Table 1. Becauseof the n2 problem noted by Bar- Tiger-Heron (T. fasciaturn).Now, we are not certain rowclough (1992), in which the required number of about the speciesof the specimen,but suspectit to pairwise DNA-DNA hybridization comparisonsin- be lineatum.This individual is a juvenile bird, whose creasesgeometrically with the number of taxa (n), we powderdown pattern suggestslineatum, but was col- limited the number of speciesin this study to save lected at an altitude more typical of fasciaturn.We money and time. Our selectionof speciesfor com- include its data in this paper becausethe specimen parison (in addition to Tigrisomalineatum, Tigriornis ultimately will be identified. leucolophus,and Zebrilusundulatus) was based on the Methods of DNA preparationand hybridization following observations. Sheldon (1987a) identified were basedon thoseof Sibley and Ahlquist (1990), three fundamentallineages of herons:"typical" her- with the modificationsof Sheldon and Winklet (1993) ons (including day-herons and night-herons); bit- and Slikaset al. (1996). Hybrids were fractionatedin terns; and tiger-herons. Therefore, we decided to in- a 35-column machine from 60ø-95øC in 2.5øC incre- clude representativesof each of theseclades. Egretta ments.All DNA sampleswere radiolabeledand com- thulawas selected because it is a commontypical her- pared as drivers (targets),except for Tigrisomasample on. Similarly, Ixobrychusexilis is a common bittern. 4507, which was radiolabeled but not used as a driver Sheldon (1987a) also found that Cochleariuscochlearius in reciprocalcomparisons. Data are availablefrom the was geneticallyremote from other heronsand pos- authors. sibly monophyleticwith tiger-herons.Thus, Coch- The indexesof hybrid stability(Tin, Tmoa,, ATe, and leariuswas includedin the study.Night-herons were ATmoa,)and normalized percent reassociation(NPR) excludedbecause they were found to be unambigu- were computedby the methodsof Sheldonand Bled- ouslymonophyletic with day-heronsand distantfrom soe (1989). Individual hybrids were excluded from all other lineages,including Cochlearius(contra Bock further analysisif they exhibited technical problems 1956,Cracraft 1967,Payne and Risley 1976).Finally, or less than 60% NPR (rationale discussedin Sheldon the GlossyIbis (Plegadisfalcinellus) was selected as an and Winklet 1993).Trees were built using the Fitch, outgroup; DNA-DNA hybridization studiesof vari- Kitch, and neighbor-joiningoptions of PHYLIP 3.4 ous ciconiiform birds suggestthat ibisesare as close (Felsenstein 1989). Becausemeasurement error is not or closerto heronsthan any group of birds (Sibley correlatedwith geneticdistance (Fig. IA), the fitting and Ahlquist 1990, Sheldon and Kinnarney 1993). option was set to unweighted least squares(Cavalli- Moreover,within reason,outgroup choice appears to Sforza and Edwards 1967). Branch robustness was have remarkablylittle effect on DNA-DNA hybrid- testedby bootstrappingwith the programof A. Dick- 674 SHELDON,McCRAc•cEN, AND STUEBINC [Auk, Vol. 112 I I III July1995] HeronPhylogeny 675 0.50 12 0.45 lO 0.40 0.35 ø 8 0.30 0.25 o 6 o•_ ø 0.20 4 0.15 0 00 0 0 O.lO 0 00 0 0 0 2 0.05 o 0.00 [ I o I I I 0 • I I I I 0 2 4 6 6 10 12 0 2 4 6 6 10 12 /I Tmod, /I Tmod, Fig. 1. (A) ATmoa,versus standard deviation for each ATmod,value (from Table 2). (B) AT•oa,versus ATe, showinghigh correlationbetween the two distancemeasures (R = 0.995,n = 49). erman(pers. comm.; Krajewski and Dickerman1990) building methods produced a single, fully re- and jackknifing(Lanyon 1985). solvedbranching pattern (Fig. 2). This tree de- pictstwo majorheron groups--tiger-herons and RESULTS other herons. The tiger-heron clade includes Tigrisomaand Tigriornis.The other herons are The phylogeneticanalysis was basedon 467 divided