The Auk 117(4):869-891, 2000

SYSTEMATIC REVISION AND BIOGEOGRAPHY OF THE HERPSILOCHMUS PILEATUS COMPLEX, WITH DESCRIPTION OF A NEW SPECIES FROM NORTHEASTERN BRAZIL

BRETM. WHITNEY,L5 Jose FERNANDOPACHECO, 2 DANTE R. C. BUZZETTI,3 AND RICARDO PARRINI 4 •Museumof NaturalScience, 119 FosterHall, LouisianaState University, Baton Rouge, Louisiana 70803, USA; 2Departamentode Zoologia, Instituto de Biologia,Universidade Federal do Rio deJaneiro, 21944-970 Rio de Janeiro,R J, Brazil; 3Centrode EstudosOrnito16gicos, Rua DomManuel 173, 04602-050St•o Paulo, SP, Brazil; and 4RuaDesembargador Isidro 126, BlocoC, Apartamento801, 20521-160Rio deJaneiro, R J, Brazil

ABSIRACT.--Studyof vocalizationsof Herpsilochmusantwrens in northeasternBrazil re- vealedthat the traditionalH. pileatuscomplex is comprisedof not two but threespecies-level taxa.We showthat the systematicsand biogeographyof the grouphave been based on an unnamedspecies that has been called H. pileatusfor mostof thiscentury. Here, we namethis "new" speciesand concludethat its severalmorphological and vocaldistinctions clearly set it apart from othermembers of the complex.True H. pileatus,as determinedthrough com- parisonof the lectotypewith recentlycollected specimens and from the distributionof re- cordings,is confinedto southerncoastal Bahia. Morphological and vocalanalyses suggest that H. pileatusand H. atricapillusare eachother's closest relatives. We providea simplekey to diagnosticplumage and mensuralcharacters to permit identificationof museumspeci- mens.An extensiveseries of specimensdemonstrated that H. atricapillusincludes two male morphs.Geographic overlap of pure morphsin southernGoias, and the existenceof some apparentintermediates in this generalarea, suggestsecondary contact of two weakly dif- ferentiatedpopulations, perhaps along complex habitat gradients. Analysis of recordingsof H. atricapillusdisaggregated geographically and by male morphrevealed no diagnostic(by criteriaof Isleret al. [1998])variation within that species,although average measures of over- all paceand paceof three sectionsof loudsongsdiffered between the morphsby approxi- mately20%, which we interpretas anotherindication of incipientdifferentiation. Revision of the Herpsilochmuspileatus complex provides a casein point for the fundamentalrole in conservationof both field and taxonomicresearch and demonstratesthe importanceof ex- aminationof type specimensand inclusionof topotypicalsamples (whether anatomical, bio- chemical,or tape recordings)in studiesof taxonomyand systematics.Received 7 September 1999, accepted11 January2000.

RESUMO.--Oestudo de vocalizag•esdos Herpsilochmusdo nordestedo Brasilrevelou que o tradicional complexoH. pileatusnesta regiao compreendenao dois mastr6s taxonsao n/vel deesp6cie. 12demonstrado que tanto a sistematicaquanto a biogeografiadogrupo tem sido baseadanuma esp6cie inominada chamada de H. pileatusna maior parte deste s6culo. 12dado nomea esta"nova" esp6ciee conclui-seque seusvarios atributos distintivos de morfologia e vocalizag6esa colocamclaramente a parte dos outrosmembros do complexo.O verdadeiro H. pileatus,como determinado atrav6s da comparagaodo lect6tipo com esp6cimesrecente- mente coletados,e da distribuigaodas gravag6es,6 confinadoa regiao costeirado sul da Bahia.Analises morfo16gicas e vocais indicam um relacionamentomuito pr6ximoentre H. pileatuse H. atricapillus.Uma chavesimples de caracteresdiagn6sticos da plumageme re- lativos as medidas6 fornecidapara permitir a identificagaode esp•cimesem museus.Uma ampla s6riede esp6cimesd.emonstrou que H. atricapillusinclui dois morfosde macho.A so- breposigaogeografica dos morfospuros no sul de Goiase a exist6nciade algunsaparentes intermediariosnessa mesma regiao, sugereum contatosecundario de duas fracamentedi- ferenciadaspopulag6es, talvez ao Iongodo complexogradiente de habitats.Analise de gra- vag6esde H. atricapillusdissociada geograficamente, por morfosdos machos e por sexo,nao reveloudiferengas diagn6sticas (segundo crit6rios de Isler et al. [1998]) dentro da esp6cie.

s E-rnail: [email protected]

869 870 WHITNEYET AL. [Auk,Vol. 117

No entanto,as m6diasdo andamentototal e das tr•s se•es do cantoforam aproxidamente 20% diferenteentre os morfos,as quaiss•o interpretadascomo sendo outra indica•o de diferencia•o incipiente.A revis•odo complexoHerpsilochmus pileatus chama a aten•o para o papel fundamentalda pesquisataxonbmica e de camponas quest•es conservacionistas, e demonstraa import•nciado examedos esp•cimes tipo e a inclus•ode amostrastopotipicas (sejamestas anatbmicas, bioquimicas ou grava•es) em estudostaxonbmicos ou sistem•ti- COS.

WITH ITS current contingentof nearly 200 of Boa Nova, Bahia, Brazil (14ø22'S,40ø10'W), species,the Thamnophilidae(typical antbirds) at 760m; collected1 September1992 by J.E Pa- ranks amongthe numericallymost important checo,prepared by L. P.Gonzaga. Tape record- familiesof Neotropicalbirds. The genusHerp- ed by J. E Pacheco,Library of Natural Sounds silochmuscomprises a conservativeminimum (LNS), Cornell Laboratory of Ornithology of 16 closelyrelated species-leveltaxa. Species 79060;Arquivo SonoroElias P. Coelho (ASEC), diversityin the genusis highestin easternBra- Universidade Federal do Rio de Janeiro JFP zil south of the Amazon River, but most taxa 054/10; and Isler inventoryJFP 6:02 to 07. from thisregion are representedby few speci- Diagnosis:Morphology.--A lightly built tham- mens that are scattered in various museums. nophilidmost clearly allied with Herpsilochmus Thiscircumstance, together with theclose mor- asthe genuswas characterized by Whitneyand phologicalsimilarity of sometaxa (especially Alvarez (1998).Males distinguishedfrom all in adult males),has hamperedunderstanding congenersby a combinationof bill width at an- of species-levelrelationships. As is the case terioredge of naresless than 3 mm, wing chord with numerousother problems in Neotropical 43 to 50 mm, tail 44 to 51 mm, and narrow rec- ornithology,careful attentionto vocalizations trices with central ones less than 5 mm wide has permittedthe only significantrecent ad- andentirely black (lacking white lateral edging vance in resolutionof specieslimits and rela- or spotting,although pale apical fringe may be tionships. present; see Table 1, cover). Females distin- Our new specimensand tape recordings have revealedthe identity and distributionof guishedby the abovecharacters and by unique the typespecies of the genus,Herpsilochmus pi- crown pattern (see below). Further distin- leatus,and an unnamedparapatric species that guishedfrom parapatricH. pileatusand sym- is widespreadin the interiorof the region.This patric (locally syntopic)H. atricapillusby nar- "new" specieshas been treated in all pertinent rowerand shorterbill andby short,relatively literature under the name pileatus.Further- indistinct postocular streak and absence of more,its morphologicalsimilarity to and geo- conspicuouspreocular spot (bothwell defined graphicoverlap with H. atricapillushas resulted and blackishin H. pileatusand H. atricapillus; in acceptanceof a closerelationship (either as thisusually hard to judgein specimens,but see a subspeciesor allospecies)to that taxon(e.g. cover).The sympatric(locally syntopic) H. pec- Davis and O'Neill 1986, Sibley and Monroe toralis(Pectoral Antwren) is readily identified 1990). by theblack pectoral patch in malesand the or- Here,we describethis unnamed species, pro- angehead of femalesand immatures.Herpsi- vide a detailed descriptionof H. pileatus(Lich- lochmusrufimarginatus (Rufous-winged Ant- tenstein1823), map the distributionsof these wren), alsolocally syntopic, is distinguishedin taxa from documentedpoints of occurrence, all plumages by the conspicuouslyrufous- and discussrelationships based on interpreta- edged remiges. tion of morphologicalvocal, and ecological Diagnosis:Voice.--Loudsong (following Wil- data.We proposeto namethe new species: lis 1967)immediately separable from thoseof all congenersby uniformlyrapid pacewithout Herpsilochmussellowi sp. nov. changein duration of individual notesor to- Whitneyand Pacheco nality (althoughthere is a steadyrise and fall CaatingaAntwren (Chorozinho-da-caatinga in frequencyand amplitude,which are shared in Portuguese) by severalother species).Loudsong of H. axi- Holotype.--Museu Paraense Em'lio Goeldi Ilaris(Yellow-breasted Antwren) of subtropical (MPEG) 54039, adult male from about 2 km east forestson the eastslope of the Andesis per- October2000] Herpsilochmuspileatus Complex 871

TABLE1. Morphometrics(see text for detailsof measurements)of three speciesof Herpsilochmusantwrens, includingtwo morphs(and intermediatesbetween them) of male H. atricapillus.Values are œ,with range and n in parentheses.

Sex Bill width (mm) Culmen (ram)a Wing chord (mm)b Tail (mm) Body mass(g) H. sellowi Male 2.8 (2.7-2.9, 5) 14.3 (13.6-14.9, 7) 47.8 (45.3-50.0, 8) 48.3 (44.1-50.8, 6) 8.0 (1) Female 2.7 (2.6-2.9, 5) 14.1 (13.4-14.7, 5) 47.3 (43.8-49.8, 5) 49.9 (47.7-50.8, 4) 7.3 (7.0-7.5, 2) Sexunknown 2.7 (1) 14.6 (1) 45.5 (1) 48.0 (1) -- H. pileatus Male 3.7 (2) 16.9 (16.5-17.2, 2) 48.7 (48.2-49.2, 2) (41.8-45.3, 2) c 9.3 (1) Female 3.5 (1) 16.2 (1) 47.4 (1) 42.7 (1)c 8.5 (1) H. atricapillus Male d 3.5 (3.1-3.8, 71) 15.8 (14.2-17.8, 61) 52.1 (48.5-55.9, 75) 56.2 (51.9-60.0, 57) 9.1 (7.0-12.0, 40) Female 3.4 (3.0-3.9, 46) 15.6 (14.7-17.3, 44) 51.0 (48.2-55.1, 45) 56.3 (52.0-60.2, 30) 8.9 (7.0-12.0, 35) H. atricapillus (gray morph) Male 3.5 (3.2-3.8, 26) 15.3 (14.4-17.8, 22) 51.6 (48.5-55.7, 29) 55.4 (51.9-60.0, 22) 10 (7.5-12.0, 8) H. atricapillus (white morph) Male 3.4 (3.1-3.8, 37) 15.5 (14.2-16.5, 34) 52.2 (49.0-55.9, 37) 56.3 (52.9-59.8, 27) 8.9 (7.0-11.0, 26) H. atricapillus (intermediate) Male 3.4 (3.1-3.7, 8) 15.9 (15.1-16.8, 7) 53.6 (52.0-56.6, 9) 58.0 (55.2-59.5, 8) 9.2 (8.2-11.0, 6)

From base at skull. Bothwings usually measured, but longermeasurement included here. Central rectricesestimated to be within 4 mm of fully grown of smallestindividual. Entiresample of males(including both morphs and intermediates).

haps the most similar, but it risesabruptly in by conspicuoussuperciliary stripes, whitish in frequencyand amplitude,is of higherfrequen- loral and supraloralregions posterior to slight- cy overall, and has a different tonal quality. ly behind eyes,becoming subtly grayer and Somecalls also are highly distinctive(vocali- blendinginto medium-gray(Gley chart 2: 5/ zations describedin detail below). 5PB) sides of neck. Narrow postocular streak Distribution.--Patchilyfrom southerncoast- (ca.2 mm wide and 4 mm long) blackish.Sub- al Rio Grande do Norte, southern Ceari, and orbital and auricularregions same whitish as central Maranhio south through the interior anterior superciliary and similarly becoming of Bahia east to near Jeremoabo (10ø02'S, grayer posteriorly.Mantle and lateral areasof 38ø11'W), Lamario (11ø45'S,38ø55'W), Jagua- backsame medium-gray as sidesof neck.Scap- quara (13ø41'S, 39ø30'W), and Boa Nova ulars medium-gray,distal webs mostly black- (14ø22'S,40ø10'W), thencewest to northwest- ish with a whitish fringe,widest just anterior ern Minas Gerais (known as far south as Mo- to tip. Middle back dominatedby a semi-con- cambinho[15ø05'S, 44ø00'W]); a disjunct (?) cealed,loosely integrated, interscapular patch: populationin the serra do Cachimboof Pari anterior feathers with white bases and blackish (Fig. 1). Occursmostly between 500 and 900 m tips on inner webs, outer webs medium-gray; elevation and also near sea level in Rio Grande posteriorfeathers longer (to 20+ mm) and sil- do Norte. very-whitewith broad(ca. 7 mm), sharplycon- Descriptionof holotype.--Seecover. Alphanu- trasting black tips. Thus, middle back mixed meric color designationsdetermined through gray,black, and white, appearingirregularly direct comparisonwith Munsell soil color blotched,pattern varying with handlingof the charts (1994). Plumage fresh and unworn. specimen.Rump feathersmedium-gray tinged Crown semiglossyblack, this color extending ochraceousat tips; uppertailcoverts medium- posteriorlyto include centerof nape; somelat- gray. Chin and throat silvery-whitish(a few eral feathers above orbits with whitish distal grayishfeather bases showing through). Breast webs.Nasal tufts and a few tiny feathersnear and belly whitish along midline, becoming baseof bill with whitish tips. Crown bordered subtly grayer laterally. Flank feathersloosely 872 WHITNEY ET AL. [Auk, Vol. 117

20

25-

30ø5

65 50 45 40 35ø• ! I

FIc. 1. Distributionof Herpsilochmussellowl from documented(specimen or tape recordingexamined) pointsof occurrence.Star marks the type localityof BoaNova, Bahia. Dotted lines mark Brazilian state boundaries.Numbers are cities in the state of Bahia: (1) Salvador,and (2) Lamar•o. An "s" beside a point on thisand other maps indicates a sightrecord. Localities and sources for pointson all mapsare available from the senior author.

integrated and greatly lengthened(to 25+ and, from below,mostly white with blackish mm), whitishor grayish,the distalseveral mm center line and subterminal blotches. Outer- tingeddull ochraceous(10YR 7/6), this color mostpair of rectricesalmost entirely white (a intensifyingslightly posteriorly and including thin strip of blackishon margin of proximal the lower belly betweenthe legs and the un- web at base);next pair entirelywhite on distal dertail coverts. webs and black on basal one-half of proximal Tail and wing not in molt.Tail stronglygrad- webs(extent and distributionof blackdiffering uated (outer rectricesca.34 mm vs. centralrec- appreciablyon the two feathers).Amount of trices50.0 mm), appearingin dorsalaspect en- white on successivelylonger rectrices decreas- tirely blackishor blackishwith white edges, ing andconcentrated toward feather tips, such October2000] Herpsilochmuspileatus Complex 873

t; -

"• • Herpsilochmuspileatus._ IIerpsilochmussp.nov. B•SIL: Bahia;Tvancoso (holotipo) MPEG5•2 BRASIL: Bahia: Boa Nova MPEG54039 •

FIc. 2. Ventral tail patterns of some Herpsilochmusantwrens from northeasternBrazil. Left, H. sellowi. Center,H. pileatus.Both tail lengthand the graduationmaximum (see Appendix 1) aremuch shorter than in the othertaxa. Right, H. atricapillus. that blackbases (most extensive on proximal Upperwing covertssemiglossy black, each webs)are largely concealedby white areasof feathermarked with a whitetip (onlesser wing overlyingfeathers. Central rectricesentirely coverts,white usually concentratedon distal blackishwith minutegrayish or brownishtips web), forming conspicuouswing bars on the (seeFig. 2 for exactextent and patternof black median and greater coverts.Alula black with and white on rectrices). white distal marginand tip. Primarycoverts blackwith minutewhite apicalmargins. Rem- igesblackish-brown, primaries bearing narrow white fringeson distalwebs, increasing in ex- tent from outermost(which is essentiallyun- marked) to innermost; secondarieswith weak grayishmargins. Innermost secondaries with fairly conspicuouswhite distalmargins (short- est innermostones, concealed by longerfeath- ers of back, are entirely dull bownish).Irides brown, maxilla black,mandible gray, tarsi and feet light gray,soles yellow. Measurementsof holotype.--Billwidth at an- terior edge of nares2.7 mm; bill depth at an- terior edge of nares 3.4 mm; culmen from an- terior edge of nares 8.3 mm; culmen from base (at skull) 14.3 mm; wing chord46.7 mm; tail 50.0 mm; tarsus 17.8 mm; body mass8.0 g. Descriptionof allotype (female).--MPEG 54040. FIG. 3. Female specimensof H. pileatus(left) and See coverand Figure 3 (right side).Plumage H. sellowi(right) for comparisonof crownpatterns fresh and unabraded. Crown feathers black- and bill size. Herpsilochmussellowi lacks discrete white markson the crown(buffy feather tips impart centeredwith dull-buff apical margins,dark a weakly mottledpattern) and hasa relativelydeli- center invading buffy margin slightly along catebill. The crownand bill of femaleH. atricapillus shaft,imparting an overallweakly dappled ef- are closelysimilar to thoseof H. pileatus. fect. This pattern more concentratedon fore- 874 WHITNEYET AL. [Auk, Vol. 117 head to midcrown,becoming faint posteriorly, H. atricapillus, and intermediates between such that hindcrown is mostly blackish. A them, are listed separately;all femalesare list- weakly contrastingline of tiny lateral crown ed together A list of H. atricapillusspecimen feathers bordering pale supercilium mostly numbersmay be obtainedfrom the seniorau- dull-buffy.Loral regionmixed buffy and whit- thor on written request. ish, merging posteriorly with superciliary Herpsilochmussellowi, Brazil: Maranhao, two stripe. Superciliumwhitish in orbital region, males (Field Museum of Natural History returningto buffy a few mm posteriorto orbit. [FMNH] 63533 and 63534); Ceara, female Periorbital area whitish, contrastingweakly (FMNH 63517); Pernambuco, sex unknown with suborbitaland auricularregion and poor- (American Museum of Natural History ly defined dusky postocular stripe. Nape [AMNH] 243024);Bahia, four males and four mixed black and medium-gray, individual females(MPEG 54039 [holotype] and 54040 [al- featherswith black centersand weakly con- lotype]; AMNH 243020, 243021, 243023, trastingmedium-gray margins.Mantle dull ol- 243026,243042, 490646). These 12 specimens ive-gray(5Y 5/1-2), this colorcarrying weakly comprisethe type series.Three specimens (Mu- throughthe sidesof the neckand mid-back,be- seu de Zoologia da Universidadede Sao Paulo coming subtly grayer posteriorly. Scapulars [MZUSP] 38503 [male], 38504 [female], and a brownish,margined dull-buffy. Interscapular female in alcohol collectedby M. Bornschein patchas in holotypebut overallless-well de- but not yet accessioned)are not to be consid- fined, especially posteriorly; at least a few ered paratypes. small, white patchesvisible without raising Herpsilochmuspileatus, Brazil: Bahia, two anyfeathers. Rump feathers medium-gray with males and one female (MPEG 54042, 54043; distinct dull-buffy tips, which, however,be- AMNH 5381).We examinedcolor photographs causeof the weak structureand looseintegra- of the lectotypeof H. pileatus(Museum ftir Na- tion of these feathers,does not produce any turkunde der Humboldt-Universit•it zu Berlin striking pattern. Uppertail coverts medium- [MNHB] 2993). gray. Hepsilochmusatricapillus (white morph, all Chin and throat whitish,merging indistinct- males),Brazil: Bahia (11), Tocantins(1), Minas ly with weakly buffy breast.Remainder of un- Gerais (1), Sao Paulo (4), Distrito Federal (1), derpartswashed dull-buff or dull-ochraceous, Goias (11), Mato Grosso (2); Bolivia: Santa this colorintensifying slightly posteriorly and Cruz (8). becomingpurest and brightest(10 YR 7/6) in Herpsilochmusatricapillus (gray morph, all the lowerflanks and lower belly (butremaining males),Brazil: Maranhao (15), Piauf (1), Ceara a decidedly pale hue throughout).Tail and (6), Alagoas(2), Bahia(3), Goi•s (2). wing not in molt.Tail asdescribed for holotype Herpsilochmusatricapillus (intermediate be- exceptoutermost rectrices with slightlymore tween gray and white morphs,all males),Bra- black at basesof proximal webs,and pale tips zil: Bahia (1), Minas Gerais (2), Sao Paulo (1), on centralrectrices barely perceptible. Wing as Goias (3), Mato Grosso do Sul (2). One addi- describedfor holotype,but blackand blackish- tional adult male from Goias was measured but brown areas more brownish, and all white not scoredto morph. feathertips and marginsduller. Patternon in- Herpsilochmusatricapillus (females), Brazil: ner secondariesmirrors that of holotype,but Maranhao(10), Piauf (1), Ceara(1), Bahia(18), marginationwider on distalwebs and tips. Soft Minas Gerais (3), Sao Paulo (1), Goias (13), parts as describedfor holotype.Bill width at Mato Grosso (1); Bolivia: Santa Cruz (3). anterioredge of nares2.7 mm;bill depthat an- SpecimensExamined: Tape recordings.--See Ta- terior edge of nares 3.2 mm; culmenfrom an- ble2. All recordingsare in thelibrary of Phyllis terior edge of nares8.0 mm; culmenfrom base and Morton Isler; only five recordings(all of H. (at skull) 14.1mm; unflattened wing chord48.2 atricapillus)are institutionallyarchived, but al- mm; tail 50.5 mm; tarsus17.6 mm; body mass most all will be archivedby the individual re- 7.5 g. cordists.Many recordingsfeature more than Specimensexamined: Skins.--Measurements of one individual and morethan one type of vo- listed birds are in Table 1 (all specimenswere calization.Other recordingswere of sufficient- measuredby B. Whitney).Two morphsof male ly good quality to confirmtaxon identification October2000] Herpsilochmuspileatus Complex 875

III

+1 +1 +l +1 876 WHITNEYET AL. [Auk, Vol. 117 but were not measuredfor the vocal analysis ples (mostof the aboveinformation from Stre- (becausespectrograms had poor resolutionor semann [1948]). were unavailable when vocalizations were The Englishname CaatingaAntwren refers measured)and thus are not included in this to the fact that the new speciesis largely re- list. Restrictedsamples of two morphsof H. stricted to this biome. With clarification and atricapillusare listed separately. Recordists oth- modification of the nomenclature,taxonomy, er than the authors are listed in the Acknowl- and distributionof the membersof the Herpsi- edgments;a list of recordingsby recordistand lochmuspileatus complex (see below), it seems localityis availablefrom the seniorauthor least confusingto give the new name Bahia Herpsilochmussellowi, Brazil: Rio Grande do Antwren to the endemic H. pileatus.Black- Norte (2), Ceara (13), Bahia (30), Minas Gerais cappedAntwren, the long-standingname for (1). H. atricapillus,should be maintained. Herpsilochmuspileatus, Brazil: Bahia (20). Herpsilochmusatricapillus (excluding restrict- REMARKS ed samplesof morphs),Brazil: Rio Grandedo Norte (2), Ceara (26), Pernambuco(1), Alagoas Variationin the typeseries.--The type series (6), Tocantins(1), Minas Gerais (11), Sao Paulo comprises12 specimens(listed above):seven (2), Mato Grossodo Sul (1), no location(1). Par- males, four females, and one sex unknown. aguay, Amambay (3); Bolivia: Santa Cruz (9), Among males,the only appreciabledifference Chuquisaca(1); Argentina:Jujuy (1). is in the amountof dull ochraceousin the pos- Herpsilochmusatricapillus (white morph), Bra- terior underparts.This colorreaches it greatest zil: Distrito Federal (7), Goias (15). extent in the holotype and anotherspecimen Herpsilochmusatricapillus (gray morph),Bra- (which is much older, and somewhatfaded) zil: Bahia (29). from the samelocality (BoaNova, in southern Etymo logy.--Friedrich Sellow (1789-1831 ) was Bahia).It is slightlyless noticeable in onefrom a German who came to Brazil in 1814 to collect Lamarao, and there is no trace of the color in naturalhistory specimens. In the latter half of the other males. Whether this variation is in- 1815, Sellow met Prince Maximilian Wied-Neu- dividual or has a geographicalrestriction or wied at G. H. Baronvon Langsdorff'shome in tendencywe cannotjudge from thesmall series Rio de Janeiro,and in Augustof that yearhe set at hand.It alsoseems possible that featherwear off with Wied to collectspecimens. After four and oxidation from the intense sunshine in the monthsin the companyof Wied, and nearly a regionoccupied by H. sellowicould cause this year alone,mostly in Espfrito Santo collecting weak hue to fadeto the point of obsolescence. specimensfor Langsdorff, Sellow continued We have noted in the field, however, that the north alongthe coastinto Bahia.Between De- dull ochraceousof the posteri9r underparts cember1816 and May 1818,he shippedsome seemsmost conspicuousin the southernbirds 1,604 specimensof birds from Salvador(the (particularlyaround Boa Nova, where the color city was at that time known asBahia) to Berlin, can appeardarker than in the holotype).Birds wherethey werereceived by M. H. K. Lichten- in northern and western Bahia and southern stein, then director of the MNHB. Lichtenstein Ceara look essentiallywhitish (concolorwith and othersdescribed a number of new species upperbelly and throat)in the field duringthe from Sellow'smaterial, among them Myiothera samemonths of the year Comparedwith the (= Herpsilochmus)pileata, and suchspectacular holotype,AMNH 243042has a slightlymore birds as Campephilusrobustus (Robust Wood- extensiveblack area on the proximalweb of the pecker) and Aratinga auricapilla (Golden- outer rectrix (only one is present). cappedParakeet). In 1831, Sellow drowned in The femalespecimens vary little from the al- the Rio Doce of Minas Geraisat the age of 42. lotype,although AMNH 243026,which also is We have chosen to name this new antwren in in fresh plumage,has a blackercrown with recognitionof FriedrichSellow's important col- fewer and grayer feather edges (foreheadis lectionsof naturalhistory specimens from east- similarly buffy). The single specimenof un- ern Brazil, which numbered approximately known sex (AMNH 243024) is in worn plum- 5,457 birds, 263 mammals, 110,000 inverte- age. The crown is blackish with indistinct brates,12,500 plants, and 2,000 geologicalsam- (probably abraded)buff feather edges. The October2000] Herpsilochmuspileatus Complex 877 back showsmore gray (lessolivaceous tinge) near-perch maneuvers such as reaches and than any of the otherfemales, approaching the gleans,and we haveobserved almost no hangs males.Perhaps it is a subadultmale. Morpho- (see Remsen and Robinson 1990:146-148). metrics of males and females overlap exten- They frequentlyhitch (seeWhitney and Pache- sively (Table1). co 1994) upward in vine tangles and other Habitat.--Herpsilochmussellowi was listed as a dense vegetation, investigating foliage and caatingaendemic by Parkeret al. (1996;appear- bark for arthropodsand not paying specialat- ing in their databasesas H. pileatus),a desig- tention to dead foliage.Members of pairs usu- nationwith whichwe generallyagree. The caa- ally foragewithin about5 m of eachother but tingawas adequately characterized, for thepur- occasionallyare separated by several trees. posesof this paper,by Rizzini et al. (1988:162) Shallowand irregularmovements of the wings as "a mosaicof both xerophyticand deciduous and tail are almost constant, but these do not vegetation... composed predominantly of appear to be distinctive; the tail is rarely woody speciesmixed with a large numberof cockedhigher than about30 ø abovethe plane prickly, succulentcacti and spiny,rigid-leaved of the back unlessbirds are scoldingor mob- bromeliads,along with many annual species bing, involved in interspecific interactions ... plant componentsvary greatly in relative (chases,displacement behaviors), or otherwise number of individuals dependingon substrate excited. and local climate,thus affording the caatinga Schubartet al. (1965)analyzed stomach con- many facies."Above about 500 m, woodlands tents of the two above-mentionedspecimens are semideciduousor evergreen(known as bre- from Cachimboin southernPar•, identifying jo in northeasternBrazil) owing to greaterand fragmentsof Orthoptera,Hemiptera, and Co- more regular precipitationand slightly cooler leoptera(Curculionidae), among other insects. temperatures(Andrade-Lima 1982). Nothing is known of nestinghabits. Under his Another woodland habitat, rich in vines and account for H. p. pileatus(i.e.H. sellowi),H. hugeterrestrial bromeliads and known as mata- Snethlage (1928:705) described two young de-cip6("vine" forest;Maack 1962), occursin birds as near fledging in February(apparently southern Bahia between about 700 and 1,000 m. near V•rzea Formosa,Cear•) but did not men- Herpsilochmussellowi is concentratedin the en- tion having collectedadults or young,or hav- claves of semideciduous woodland and scrub ing seena nest.An adult femalecollected near of the caatinga,being absent from the mosthu- Mocambinho, Minas Gerais, on 2 February mid forests and the lowest, driest sectors. It 1997 by M. Bornschein(preserved in alcohol alsooccurs along ecotones with cerradoin west- and not yet depositedin a museum)had un- ern Bahia, and taller deciduous forest in north- developedgonads and was not in molt. ern Minas Gerais. At Tibau do Sul in southern Herpsilochmussellowi usually forages in coastal Rio Grande do Norte, H. sellowi occurs mixed-speciesflocks of insectivores(flock as- in restingawoodland (so classifiedby Salgado sociatesvary with locality). In some areas, et al. [1981]). The habitat in the serra do Ca- these include other speciesof Herpsilochmus. chimbo of southern Par• where H. sellowi has Syntopyis greatestwith H. atricapillusand has been collectedis not definitelyknown. Among been documented with direct field observa- the most abundant trees and shrubs in all hab- tions and tape recordingsat Tibau do Sul, Rio itats in which H. sellowiis common,especially Grande do Norte; in the Chapadado Araripe in the lower elevationareas, are variousspecies of southernCear•; at Jequi•(K. Zimmer record- in the family Leguminosae.We have frequently ings)and BoaNova in southeasternBahia; and observedH. sellowiforaging in Leguminosae, near Mocambinho, Minas Gerais (M. Borns- and it may be that the presenceof somespecies cheinpers. comm.). These two species(FMNH of treesin this family is an importantaspect of 63534 [sellowi] and 63535 [atricapillus])also the habitat. were collected at Barra do Corda, Maranhao, Behaviorand ecology.--Herpsilochmussellowi on 22 and 27 September1927, respectively, but foragesin the middle and upper strataof the apparently in different habitats (Snethlage woodland and scrub it inhabits, but where 1928:705). Herpsilochmussellowi also occurs scrubbygrowth is dense,it may occurin veg- with H. pectoralisnear Tibau do Sul (the only etation lessthan 2 m tall. Typical attacksare locality we know of where syntopy of three 878 WHITNEYET AL. [Auk,Vol. 117 speciesof Herpsilochmushas been document- Hellmayrstated to be "near the city of Bahia" ed), near Morro do Chap•u, and about25 km (Bahiais now knownas Salvador, on thecoast). westof Jeremoabo,Bahia. Finally, we foundH. Lamaraoactually is about140 km northwestof sellowiwith H. rufimarginatusin mata-de-cipdSalvador (Paynter and Traylor 1991) and is near Jaguaquaraand at the interfaceof drier about the same distance inland from the coast. and more humid woodland near Ibicoara, Ba- A few years later, after examinationof addi- hia (althoughnot yet in thesame mixed-species tionalspecimens, Hellmayr (1929:373) conclud- flocks).Herpsilochmus rufimarginatus is absentat ed that nominatepileatus ranged "from Bahia BoaNova where H. sellowiand H. atricapillusco- north to Cear• and thencethrough northern exist,but it occursjust a few km to the eastin Piauhy west to central Maranhao." He sum- more humid AtlanticForest not inhabitedby marized the distributionof H. p. atricapillus the otherspecies. Along the coastof southern (Cory and Hellmayr 1924:374)by saying:"re- Bahia, H. pileatusand H. rufimarginatusare placesthe preceding race [nominate pileatus] in syntopic in forest on sandy soil about 18 km centralBrazil, its range extendingnorthward west of Porto Seguroand near Trancoso.Mi- to northwesternBahia (Rio Preto) and the ex- crohabitatpreferences and nichepartitioning tremesouth of Piauhyand Maranhao." among syntopicantwrens are not obviousand Zimmer (1932) acceptedHellmayr's treat- are worthyof study. ment of atricapillusas a subspeciesof pileatus, but referred to specimensfrom easternBrazil TAXONOMIC HISTORY AND GEOGRAPHIC as "puzzling." Naumburg (1939), diagnosing DISTRIBUTION OF THE specimensof Herpsilochmuscollected in eastern HERPSILOCHMUS PILEATUS COMPLEX Brazil by Emil Kaempfer,adopted without res- ervationHellmayr's identificationof the La- Sinceerection of the genusHerpsilochmus Ca- marao specimenas typical of H. p.pileatus, and, banis, 1847 and publicationof Sclater's(1858) in comparisonwith this specimen,assigned Synopsisof theAmerican Ant Birds(Part II), the three Kaempferskins (one male and two fe- unity of Herpsilochmushas been recognized males) from Barra do Rio Grande (some560 km universally.The type speciesof Herpsilochmuswest-northwest of Salvador at the confluence of is Myiothera(= Herpsilochmus)pileata (Lichten- the riosGrande and SaoFrancisco) to pileatus. stein1823), from "Bahia." The lectotype(des- She also acceptedas pileatusspecimens from ignatedby E. Stresemannon one of the speci- Villa Nova (= Bonfim; one male and two fe- menlabels) is housedin the MNHB; we hereby males;MZUSP 7237,7271, 7272) cited by Pinto formallyrecognize this specimenas the lecto- (1932). Naumburg's(1939) recommendationof type. Lack of a more preciselocality of collec- species-levelstatus for H. pileatusand H. atri- tion for the type,coupled with the factthat this capilluswas not adoptedby Peters(1951) or important specimenapparently has not been Meyer de Schauensee(1966, 1970), who contin- examinedwith specimensof presumablyrelat- ued to treat them assubspecies, but it was res- ed speciesof Herpsilochmus,has contributedto urrected by Davis and O'Neill (1986) and fol- the polemical nomenclaturaland taxonomic lowedby Sibleyand Monroe(1990), Sick (1993, historyof H. pileatusand its presumedclosest 1997),Ridgely and Tudor (1994),and Parkeret relativesand to the misinterpretationof the al. (1996). geographicdistribution of thesetaxa. Throughstudy of extensivetape recordings Hellmayr (1908) consideredthe described of vocalizationsof Herpsilochmusin northeast- speciespileatus, atricapillus, and motacilloidesto ern Brazil,we suspectedthat threespecies-lev- be geographicrepresentatives "more properly el taxa occurredin the traditional "pileatus designatedby trinomials."Based on five males complex."We examinedphotographs of the (which did not includethe type) and one fe- lectotypeof Herpsilochmuspileatus kindly sent male (only one of the six with locality more from the MNHB. The originallabel bears the precisethan "Bahia"),Hellmayr (in Cory and name"Sello" (= Sellow;this spellingvariation Hellmayr 1924)restricted the distributionof H. was notedby Papavero[1971:56] in the follow- p. pileatusto "Eastern Brazil, coast district of ing passage:"...Friedrich Sellow(or Sello)was state of Bahia." The single definitelocality born on 12 March 1789..."). Papavero's(1971) amongthis (1924)series was Lamarao,which map (fold-out number 8), based on Sellow's October2000] Herpsilochmuspileatus Complex 879

_ I I I V 12os-

_ • 15- ,s ß BAHIA ß 14-

BoaNov•aa ß Ilh•us OceQ;ll, 15-

16-

MINAS 17- GERAIS

o 50 1 oo ' Approx. Scale (km) 1 8-

I2 IJ [II I II FIG.4. SoutheasternBahia showing area of sympatryof H. sellowi(dots) with H. atricapillus(squares) and parapatryof thesespecies with H. pileatus.Triangles indicate three localities where syntopy of H. sellowiand H. atricapillushas been confirmed. Entire known range of H. pileatus(diamonds) is shownon map;star marks Trancoso,the proposedtype localityfor H. pileatus. itinerary (Stresemann1948), shows that Se- tantly in havinga largebill; very shorttail; and 11ow's travels in Bahia were restricted to the im- gray chest,sides, and flanks),and we are con- mediate coast south of the city of Salvador. fidentthat thesethree specimens represent true Thus, the type of Herpsilochmuspileatus, col- H. pileatus.A good sampleof tape recordings lectedby Sellowand describedby Lichtenstein, of H. pileatusand other Herpsilochmustaxa in must have been taken somewhere in southern northeasternBrazil establishesthat H. pileatus coastal Bahia. is restrictedto the coastalregion of Bahiafrom A male (and a female)Herpsilochmus species, the vicinity of Baia de Todosos Santos(near tape recorded and collectedby Buzzetti near Salvador) south at least to Barra do CaK(ca. Trancoso(on the southerncoast of Bahia about 17øS;Fig. 4). We suggest Trancoso,Bahia 400 km south of Salvador)on 21 September (16ø35'S,39ø06'W), as a precisetype localityfor 1997closely matches the lectotypein plumage Herpsilochmuspileatus. and overallmorphological aspect (most impor- Having establishedthe identity and geo- 880 WHITNEYET AL. [Auk, Vol. 117 graphic distribution of H. pileatus,we exam- musbased on morphologyand inferencesfrom ined most of the specimenslisted by Cory geographicdistributions. They focusedespe- and Hellmayr (1924) and Hellmayr (1929). It cially on systematicsof the H. pileatuscomplex, is clear that Hellmayr's seriesfor H. p. pileatus supportingNaumburg's (1939) elevationof H. included mostly H. sellowiand a few H. atri- atricapillusto speciesrank and concludingthat capillus.The Lamarao specimenproved to be nominate pileatus, atricapillus, motacilloides a male H. sellowi.It was principally this mis- (Creamy-belliedAntwren), and parkeri(Ash- identificationthat set the stage for the long- throated Antwren) are the closestrelatives of standing confusion of sellowias pileatus.Al- eachother AlthoughDavis and O'Neill (1986) though Cory and Hellmayr (1924) examined stoppedshort of suggestingthat these four taxa the type of pileatusin Berlin,it is unlikely that compriseda superspecies,they were cited as it was directly compared with the rest of the sourcein the classificationby Sibley and Hellmayr's series (he did not include mea- Monroe (1990) of the four taxa as allospecies. surementsof it in the 1924paper), which was Davis and O'Neill (1986:figure 3) mappedthe scattered mostly in several European muse- distributionsof membersof their H. pileatus ums. We suspectthat someof the skinsfrom complexbut showedno membersreaching the "Bahia" he examined in the Naturhistoris- coastof easternBrazil, and apparentlythey in- chesMuseum (Vienna) and the Natural His- cludedonly the localitiesBonfim and Barrado tory Museum (formerly British Museum Rio Grande for H. pileatus. [Natural History]) represent true H. pileatus, Morphologicalanalysis alone could be inter- but we have not seenthese specimens. Aside preted as supportingthe "pileatuscomplex" of from the lectotypeand the recentMPEG pair Davis and O'Neill (1986),although their anal- mentionedabove, the only other specimenof ysis includedH. sellowibut not true H. pileatus. H. pileatuswe havelocated is a male collected We propose,however, that the significantly by Wied (locality "Brasilien" = Brazil). This narrower and shorter bill (one-wayANOVA, specimen(AMNH 5381) was inexplicablyla- sexes combined; P < 0.0005 for H. sellowivs. H. beled the "type of Formicivorapileata Wied" pileatusand H. atricapillus;see Table 1); the nar- (corrected in August 1998; M. LeCroy pers. row rectrices,lacking whitish lateral edging (or comm.). We have confirmedthat all three of spots) on the central pair; and the unique the Kaempfer specimensfrom Barra do Rio crownpattern of females(together with other Grande included in Naumburg's (1939) series less-diagnosticdifferences) serve to set H. se- of H. pileatusare H. sellowi.Similarly, two llowi distinctly apart from other membersof specimensfrom Cachimboin southernPara this aggregate.Male plumagesof true H. pilea- (MZUSP 38503 and 38504), identified as H. tus and H. atricapillusare remarkablysimilar; pileatus atricapillusby Pinto and Camargo females differ principally in extent and satu- (1957) and as H. pileatusby Stotz et al. (1997), ration of buffy wash on the underparts.The also are H. sellowi.The three specimensfrom distributionof white on the rectrices(Fig. 2), Bonfim proved to be H. atricapillus. tail length (Table1) and degreeof graduation Appendix1 is a simplekey to the specimen of the tail (Appendix1) of the two speciesare identificationof the three similar speciesof dramaticallydifferent. Herpsilochmusdiscussed in this paper:sellowi, Morphologicalvariation within H. atricapi- pileatus,and atricapillus.Their distributionsare 11us.--Examinationof approximately120 spec- mapped in Figures1, 4, and 5 from confirmed imensof H. atricapillusrevealed significant, ap- (specimensor tape recordings) points of oc- parently non-clinalplumage variation in the currence. Detailed plumage descriptionsof extentof grayin the underpartsof males.Each male and femaleH. pileatusare providedin Ap- specimenwas scoredas "gray,.... white," or pendix 2. "intermediate"without noting the locality of collection. When the distributions of these SYSTEMATICS OF THE morphswere mapped (Fig. 5), it becameap- HERPSILOCHMUS ?ILEATUS COMPLEX parentthat malesfrom the northernand east- ern parts of the range south to central and Morphology.--Davisand O'Neill (1986) dis- southeastern Bahia and eastern Minas Gerais cussedtaxonomy and systematicsof Herpsiloch- were grayishon the throat, breast,and sides October2000] Herpsilochmuspileatus Complex 881

2O

25

65 50 45 40 I F•cg.5. Distributionof H. atricapillusshowing approximate ranges of two male morphs. Black dots denote wheregray morph has been confirmed; black diamonds denote where white morph has been confirmed and, in farwestern and southern parts of range, is also presumed; white diamonds denote morph unknown (either becauseno adult male examined or nospecimen examined), but considered more likely white morph; and "x" denoteswhere specimens have not been examined and species are undetermined. Because ofwidespread sympatryof H. atricapilluswith H. sellowiin northeasternBrazil, all unexaminedspecimens and undocu- mentedreports of any members ofthe complex from this region are considered unidentified. Specimens and recordingsfrom Rond6nia, Brazil, and La Paz,Bolivia, which are under further study (see text), also are mappedas species undetermined. The single black triangle in southernGoifis marks "Ner6polis," where bothgray and white morphs have been collected (see text for discussion of some apparent intermediates fromthis region). A starmarks the type locality of H. atricapillus.Numbers indicate the cities of (1) Salvador, (2) Brasilia, and (3) Goifinia.

(often appearingweakly mottledor streaked) southern Tocantins, Distrito Federal, Goifis, and whitishon the belly. Malesfrom western Mato Grosso, Mato Grosso do Sul, and Bolivia and southern Bahia and northern Minas Gerais west to the foothills of the Andes, were white (westof the serrado Espinhaqo)south at least below and lacked the grayish wash on the throughnorthern S•o Pauloand west through throat and breast(see cover). White birds also 882 WHITNEYET AL. [Auk, Vol. 117 tendedto haveexpanded areas of white in the Secondarycontact of thetwo morphsimplied tail and a light buffy washon the lowerflanks by overlap of collectinglocalities for "pure" and undertailcoverts (rare in graybirds); these morphs(i.e. no cline)and the localconcentra- charactersmay be individuallyvariable. Like- tion of apparentintermediates might be ex- wise, female plumage variation (most evident plainedby the significanthabitat heterogeneity in the intensityof buff in the underparts,with in this region. SouthernGoias encompasses white-morphbirds generallythe whitest be- rather sharp ecologicalgradients between se- low) seemsbest explainedby individual vari- mideciduouswoodland and cerrado(see Mag- ability. One-way ANOVA of males of both nagoet al. 1983:figure 4.7) and is furthercom- morphsof H. atricapillus(see Table 1) revealed plicatedby many linear breaksof galleryforest no significantdifferences in morphometricsor and the westernmost influence of the more-hu- body mass.There were no significantdiffer- mid Atlantic Forest.White morphsare wide- encesin morphometricsor body massbetween spreadin semideciduouswoodland (B. Whit- the sexesof H. atricapillusexcept for wing ney pers.obs.). We havenot locatedgray birds chord,which was longer for malesthan females in the field in southernGoias, but this morph (P < 0.005, n = 75 males, 45 females). occursalmost exclusively in evergreen(i.e. rel- Among 76 male specimensfrom five mu- ativelyfew treesdrop all their leavesannually) seums,nine werejudged intermediate.These forest and woodland farther north and east. were like the white morph but with a slight Secondarycontact of two populationsof H. atri- grayishtinge on the sidesof the breast.Seven capillusmay havebeen permitted through hab- were collected from the rather restricted area itat shiftscontrolled by suchabiotic factors as of southwestern Bahia, northwestern Minas long-term variation and stability of mean an- Gerais, south-central Goias, and northern nual rainfall, mean annual temperature, and Sao Paulo (the other two were from Bonito, edaphicconditions. The two morphsalso occur Mato Grossodo Sul). Given sympatryof the fairly closetogether in theTocantins / Piaui bor- morphsin part of this area (seebelow), this der region,which is poorlyknown ornitholog- is the region where intermediateswould be ically. most expected.However, 24 other specimens Vocalizations.--Measurementand analysisof from most of the same localities were scored vocalizationsof H. sellowi,H. pileatus,and H. unambiguously;all but two (seebelow) were atricapillusfollowed the terminologyand meth- white morphs.No intermediateswere found odology of Isler et al. (1998) and were per- among the many specimensfrom north or formedby Phyllis and Morton Isler.Measure- east of the serra do Espinha•o.These obser- ments were taken of 461 loudsongsof 165 in- vationsare consistentwith the hypothesisof dividuals from 107 recordings. Measures of secondarycontact of two weakly differenti- multiple loudsongsof individuals were aver- ated populationsof H. atricapillusin east-cen- aged to provide means for individuals before tral Brazil. inclusionin the analyses.The sampleof H. atri- The situation in southern Goias invites a capillusloudsong recordings initially was dis- closerfocus. Of 16 specimensfrom this area,11 aggregatedby geographyand morph (ex- were white morphs,two were gray morphs, plainedabove). To reduceambiguity of morph and three were amongthe intermediates.The identification,samples were restricted to south- two gray birds (MPEG 44585, 44586) were col- easternBahia (Jequi• and Boa Nova; gray birds) lected near Ner6polis ("Fazenda Dois Ir- and southernGoias (all birds identifiedvisu- maos"), about 30 km north of Goiania. This ally in the field) and the Distrito Federal(white penetratesthe southernperimeter of thewhite- birds). The sampleof H. atricapilluswas taken morph specimendistribution in Goias, and as a whole for the interspecificanalysis. Al- white birds have been collected within 50 km thoughfemale songs of all three speciesoften to the east and west, near Trindade and Goia- soundslightly higher in frequencythan those nia. Furthermore,one of the specimensscored of males, no diagnosable differences were as intermediate(FMNH 344506;judged prob- found betweenthe sexes,so they were com- ably white with the annotation"not bright; bined for the vocal analysis. featherslightly soiled?")also was collectedat In the followingdescriptions of eachspecies' "Ner6polis." vocalizations,we define "rattles" as compris- October2000] Herpsilochmuspileatus Complex 883 ing severalstructurally similar notesdelivered at irregular intervalsin a pair-contactcontext in burstsof variableduration; they are usually (Fig. 7G). delivered in repetitive sequences.Herpsiloch- The loudsongof H. atricapillusbegins with musrattles are analogousto rattlesin the Tham- one or two (rarely three) introductorynotes of nophiluspunctatus complex, as definedby Isler individually variable frequencyand duration et al. (1997). Functionsof someother vocaliza- distinctly separated from and of different tions have not been determined with a reason- structurethan the restof the notesin the song. able degreeof certainty. The main seriesof closelyspaced notes usually The male loudsongof H. sellowiis a steady lastsless than 2 s (œ= 27 notesin 1.8 s, n = 101) seriesof closelyspaced notes of uniform struc- and maintains a frequencyof slightly more ture and tonality (œ= 56 notesin 2.2 s, n = 32). than 2 kHz beforefalling off at the end. Com- The seriesrises and falls gently in frequency paring first and secondand first and third sec- and amplitude, maintaining a frequency of tions (not includingthe introductorynotes in about3 kHz (mean peak frequency= 2.7 kHz, the first section),the loudsongdecelerates in n = 32) for most of its duration (Table2, Fig. pace (Table2, Fig. 6C). We noted two types of 6A). Thereare two basictypes of rattlesor rat- rattles.The oneillustrated in Figure7E is heard tle-like calls. The one used to mob or scold commonlyand functionsprimarily (but per- predatorssuch as snakes and FerruginousPyg- hapsnot exclusively)to scoldor mobpotential my-Owls (Glaucidiumbrasilianum; vocal imita- predators.It is the only rattle amongthe taxa tion of this speciesis sufficientto elicitthe call) studied that shows multiple, clearly visible consistsof pairs or repetitions of up to about overtonesin spectrograms.The otherrattle or sixunmodulated notes sounding like iip or wiip rattle-likevocalization (Fig. 7F) is represented (Fig. 7A). Proximity of a quiet observer(mild in our sampleby onlythree recordings. The call threat) does not elicit this call. Another rattle- heard most frequentlyis a single,modulated like call comprisesa rapidly paced seriesof greeprepeated at irregular intervals in a pair- sharp, hairpin-shaped notes usually intro- contactcontext (Fig. 7H). An unmodulatedsin- gle chupor chipcall (Fig. 7I), althoughheard ducedby a different,slightly longer note, the fairly commonlyin the field, is representedin whole call lastingabout 1 s and oftenrising in the sampleby only 10 recordings.It is some- frequency(Fig. 7B).An individual may deliver several variations of this vocalization in se- times delivered repetitively and interspersed with modulatedcalls, being givenin an excit- quence. ed, heightenedawareness context, and also is The loudsongof H. pileatusbegins with four given regularly as individuals fly between to sevennotes separated by successivelyshort- trees. er intervals merging into a regularly and more Table 2 provides an overview of loudsong closelyspaced series (œ = 25 notesin 2.2 s, n = charactersmeasured and resultsfor eachspe- 22), the wholesong characterized by essential- ciesand the morphsof H. atricapillus(complete ly uniform note structure.The songrises and data setsavailable from P. and M. Isler). Follow- falls gentlyin frequencyand amplitude,main- ing Isler et al. (1998),in the followingcompar- taining a frequencyof about2 kHz (meanpeak isons,"diagnosable" means that differencesin frequency2.1 kHz, n = 22) for most of its du- pairwise speciescomparisons were significant ration. Comparing first and secondand first at the 97.5 percentileand would remainso even and third sections,the loudsongaccelerates in if n were equal to 30 or more (samplesizes for pace(Table 2, Fig. 6B). We have 10 recordings eachspecies provided in Table2). Theloudsong of rattles;two variationsare illustratedin Fig- of H. sellowiis highly distinctiveand is diag- ures 7C and D. The one shownin Figure 7C is nosablefrom thoseof H. pileatusand H. atrica- givenfrequently; it is closelysimilar to and ap- pillusin at leastfive characters:number of notes parentlygiven in the samecontext as the mob- (morein sellowi);pace (overall pace and paceof bing/scolding calls of H. gentryi (Ancient all sectionsslower in pileatus;pace of section3 Antwren) and H. stictocephalus(Todd's Ant- slower in atricapillus);change of pace (nearly wren) shown in Whitney and Alvarez (1998: constantpace in sellowi);note length;and struc- figures5C and D). The mostfrequently heard ture of individual notes (essentiallynarrow callis a single,modulated greep that is repeated mirror imagesof the otherspecies' notes). This 884 WHITNEYET AL. [Auk,Vol. 117

0 0.5 1.0 1.5 2.0 2.5 3.0 51H. pileatus • • 41 ' • " ' • • ' '•'•' • '•'

O/ , • , , 0 0.5 1.0 1.5 2.0 2.5 3.0 • 5• H.atricaœillus, whitemorph _/ ."• • ,•, • • • i ,• '• • -• ,• •:.i • :- • • ,• . :• •<, "•

• • ...... •...... •-•;'•••.• , •:•, '••'•_, --• • ... •, ...... •---.• ...... • O/ , 0 0.5 1.0 1.'5 210 215 3

H. atricapillus 4 3 ...... •.- '•.:--,• • * '•"-:' - ;• .... • ...... :" ' - ...... o

Time (seconds)

FIG. 6. Loudsongsof someHerpsilochmus antwrens from northeasternBrazil selectedto illustrate mean differences.Quantitative, diagnostic, interspecific differences are describedin the text from measurements presentedin Table2. (A) H. sellowi;Isler BMW 122:09; Chapada do Araripe, Cear•, 25 February 1996; recordist B.Whitney. (B) H. pileatus;Isler MISC 3:11; near Trancoso, Bahia, 20 September 1997; recordist D. R. C. Buz- zetti.(C) H. atricapillus,white morph; Isler BMW 159:19; Parque Eco16gico Ulysses Guimar•es, Goi•s, 21 June 1998;recordist B. Whitney. (D) H. atricapillus,gray morph; Isler JFP 6:09; near Boa Nova, Bahia, 30 August 1992;recordist J.E Pacheco.(E) H. atricapillus,aberrant loudsong (see text); Isler KJZ 46:13; near Jequi6, Bahia, January1996; recordist K. J. Zimmer. October2000] Herpsilochmuspileatus Complex 885

H. sellowi H. sellowi 6-

5-

4-

3

2

1

0 i i 0 O.•5 I .0 1.•5 210 2. 5 3.0

• o 0.5 1.0 1.5 2.0 2.5 3.0

6-

5-

4-

3-

2

1

0 0 0.5• I .0• 1. •5 2.0• 2. •5 3.0

H. pileatus H. atricapillus H. atricapillus

...... :......

o I 0 015 1.0 1.•5 2.0 2.5 3.0 Time (seconds) FIG.7. Rattlesand other frequently delivered calls of someHerpsilochmus antwrens from northeastern Brazil. Behavioralcontexts of thesevocalizations are discussed in thetext. (A) H. sellowirattle; Isler JFP 6:07; near Boa Nova,Bahia, 30 August1992; recordist J. E Pacheco.(B) H. sellowirattle-like call; Isler BMW 91:05;40 km North Manga,Minas Gerais, 11 November 1994; recordist B. Whitney.(C) H. pileatusrattle; Isler MISC 3:11;near Tran- coso,Bahia, 20 September1997; recordist D. R. C. Buzzetti.(D) H. pileatusrattle-like call; Isler MISC 3:26; near Canavieiras,Bahia; recordist R. Parrini.(E) H. atricapillusrattle, given by bothmorphs; Isler BMW 123:20; near BoaNova, Bahia, 7 March1996; recordist B. Whitney.(F) H. atricapillusrattle-like call; same data as (E). (G) H. pileatusmodulated greep call; Isler MISC 3:27; near Canavieiras, Bahia; recordist R. Parrini.(H) H. atricapillus modulatedgreep call; this call is indistinguishablefrom that of H. pileatus;Isler BMW 67:12;near BoaNova, Bahia,10 November 1993; recordist B. Whitney. (I) H. atricapillusunmodulated chup or chipcall; Isler BMW 68: 11;near Boa Nova, Bahia, 10 November 1993; recordist B. Whitney. 886 WHITNEYET AL. [Auk, Vol. 117 is easyto understandwhen the songsare heard morphs,was not recorded in the whitemorphs; together;H. seIIowisounds nothing like the oth- rattles appear to be indistinguishable.Diag- ers.The rattlesof H. seIIowialso are highly dis- nosabledifferences in pace of loudsongswere tinctive,although the individualunmodulated demonstratedby Isler et al. (1998)to be among notes of H. selIowi(Fig. 7A) and of H. piIeatus the mostimportant vocal characters separating (Fig. 7D) bear resemblance.The loudsongsof thamnophilidtaxa at the specieslevel. We sus- H. pileatusand H. atricapiIIussound much more pect that the averagepace differencesof gray alike, primarily owing to close similarity in and white morphsof H. atricapiIIusreflect in- structureand tonality of individual notes,and cipient speciation. A less-rigorousstatistical muchmore similar pace. analysisand interpretationof the abovedata In the analysis,loudsongs of H. piIeatuswere couldresult in judgementof subspeciesor even consideredto lack introductorynotes and, be- speciesstatus for the H. atricapiIIusmorphs, but causethe introductorynotes of H. atricapiIIus we believe that the Isler et al. (1997, 1998) cri- were individually variableas mentioned above, teria allow a more conservative and evolution- loudsongsof that specieswere measured with- arily accurateperspective on differentiationat out the introductorynotes. Loudsongs of the the specieslevel. The gray morph may merit two specieswere diagnosable,then, by lack of recognitionas a subspecies,but we hesitateto introductorynotes in H. piIeatus;pace (section name it based on a brief overview and so few 1 wasslower in piIeatus);change of pace(as de- characters. scribedabove, piIeatus accelerated and atricapi- Finally,the aberrantloudsong illustrated in IIusdecelerated); and notelength (second note Figure 6E, which was delivered consistently of piIeatuswas longer, although of similarstruc- by an individual H. atricapiIIusat Jequi6,Bahia, ture). Diagnostic differenceswould be altered is unique in the sample of 165 individuals. in importantways if the loudsongof H. pileatus This exampleemphasizes that vocalizationsof were consideredto have introductorynotes a singleindividual shouldnot form the basis and the introductorynotes of H. atricapilIus for judgementsof taxonomicdivergence. were includedin the analysis.For example, the Systematicssummary and origins.--Substitu- H. piIeatusloudsong shown in Figure6B could tion in the traditional H. piIeatuscomplex of be interpreted as having five introductory true H. pileatusfor H. seIIowihas the effect of notes ahead of the main series. Inclusion of the maintainingthe nomenclaturalstatus quo of introductory notes of H. atricapiIIuswould the recenttaxonomic literature, although geo- probablyshift mostor all differencesbetween graphicdistributions and relationshipsof the that speciesand H. piIeatusto the first third of revisedmembership have required significant the loudsong,the latter two-thirds of their clarification.HerpsiIochmus seIIowi is a highly songsshowing close similarities in theseand distinctivespecies, both morphologicallyand other measures.Rattles of H. piIeatusand H. vocally,and we recognizethat a reasonablear- atricapiIIusare quite different,but the modulat- gument could be made for its genericsepara- ed, single-notecall of the two speciesis re- tion. However,pending biochemical analysis markably similar both aurally and spectro- of the Thamnophilidae,we prefer that it be graphically. maintained in HerpsiIochmus.It is sympatric We found no diagnosablevariation between with four speciesof Herpsilochmus(including the samplesof loudsongsof gray (n = 25) and H. Iongirostris[Large-billed Antwren] near white (n = 23) morphsof H. atricapiIIus.How- Correntina, Bahia [B. Reinert pers. comm.], ever, we documented mean differences in the and probablyin northernMinas Gerais,pers. overall pace and pace of each sectionof the comm.). loudsongs:gray birds averaged 21% faster than Morphologicaland vocal analysesof all white birds overall, and 22%, 22%, and 17% membersof HerpsiIochmus(Whitney and Alva- fasterthrough each of the threesections (intro- rez 1998, this paper) indicatethat H. piIeatus ductory notes excludedfrom analysis).Aver- and H. atricapiIIusare eachother's closest rela- ageloudsongs of white andgray morphs are il- tives.Although differentiation of thesetwo es- lustratedin Figures6C andD, respectively.The sentially parapatric speciesmay be relatively unmodulated, single-notechup call recorded recent,we are confidentthat their evolutionary from severalindividuals in the sampleof gray destiniesare independent.A more recent(dif- October2000] Herpsilochmuspileatus Complex 887 ferentiationonly incipient)genetic rift in inte- Bahiaare endemicto thisregion (Thomas et al. rior easternand centralBrazil, perhaps across 1998). Accordingto the quarterly newsletter the serra do Espinhago,seems to have given "Bahia Invest:Tourism" (year 4, number14; rise to the gray and white morphsof H. atrica- publishedby the Departmentof Culture and pillus.The evolutionaryrecord in this caseis Tourismof Bahia),the first of a seriesof large- obscuredby apparent secondarycontact and scaleresort developments was initiated in Sep- lackof diagnosticvocal markers, which we sus- tember1998 near Trancoso,with a budgetof pectis anotherindication of relativelyincipient 150 million dollars. These resorts and the as- differentiation. We concur with Davis and sociatedairstrips, roads, and other infrastruc- O'Neill (1986) that the H. pileatuscomplex ture will take a heavy toll of the environment shouldembrace H. motacilloidesand H. parkeri. in the comingdecade. Whitneyand Alvarez (1998) suggested that the allospeciesH. stictocephalusand H. gentryi are ACKNOWLEDGMENTS the closestrelatives of the H. pileatuscomplex We probably would not have written this paper north of the Amazon, Solim6es, and Marafion without the photographsof the lectotypeof Herp- rivers, another judgement with which we silochmuspileatus kindly sent by B. Stephan and V. agree.Indeed, the H. pileatuslineage has had a Heinrich of the MNHB; we thank them very much long and geneticallyfragmented evolutionary for their cooperation. Morton Isler helped track history. down specimenlocalities and prepared the maps. Preliminary examinationof specimensand Phyllis Isler interpreted and measuredthe large tape recordingsof the Herpsilochmuspopula- sample of vocalizations presented in Table 2 and, tion in northeasternRond6nia, Brazil (Stotz et togetherwith Mort, analyzed it statistically;time- ly completionof the manuscriptwas greatly facil- al. 1997),suggests that it is distinctive.It is un- itated by their efforts, which, in our opinion, cer- der further study by its discoverers.Another tainly merited coauthorship.We are grateful to D. poorly known populationconspecific with or C. Oren of MPEG; M. LeCroy, C. Blake, and D. closelyallied to H. atricapillusinhabits dry for- Sloss of AMNH; D. Willard of FMNH; K. Garrett est at about900 m elevationin the valleyof the of the Los AngelesCounty Museum;J. V. Remsen, Rio Tuichi in La Paz, Bolivia (Pearman 1993, Jr. and S. Cardiff of the LouisianaState University Perryet al. 1997).It alsois underfurther study Museum of Natural Science;and J. L. de Figuei- (Pearman 1993). redo of MZUSP for allowing us to examine speci- Conservation.--The"arboreal" caatingain- mens and for loan of specimensin their care. We greatlyappreciate the hard work of collectors,past habitedby H. sellowiand many other endemic and present,that went into securingand prepar- organismshas suffered recent and widespread ing these specimens.In this regard, we acknowl- clearingand burning. Furthermore, grazing by edge especially M. S. Brigida, J. Hidasi, D.C. Pi- goatseffectively inhibits regeneration of most mentel Neto, and J. M. C. da Silva. In the same woody caatingaplants. The mata-de-cip6is in- spirit, we are thankful to R. Behrstock,W. Belton, habitedby a highproportion of globallythreat- N. Gardner, S. Herzog, T. A. Parker III, T. Schulen- ened species(Gonzaga et al. 1995, Whitney berg, and E. Willis for contributingrecordings for 1996). We know of no patchesof undisturbed our analysis. Foremost among contributing re- mata-de-cip6,and none of it is officially protect- cordistswas K. J. Zimmer, whose data were espe- cially helpful. A small number of recordingswas ed. Fortunately, H. sellowiseems reasonably obtained from the British Library of Sounds, the adaptableto habitatdisturbance, but overallit Florida State Museum Sound Archive, and the Li- must be considereda rare species.Even worse brary of Natural Soundsat the Cornell Laboratory off is H. pectoralis,which is moredependent on of Ornithology. During several years of attention the presenceof intact woodlandand has be- to the Herpsilochmuspileatus complex we have ben- comeextirpated or quite rare in most of its for- efitted from field companionship of several col- mer range. leagues,to whom we extend our appreciation:C. Althoughit is presentlycommon in manylo- Bauer,C. E. S. Carvalho, G. D. A. Castiglioni,M. calities,H. pileatusis restrictedto the narrow Cohn-Haft, P.S. M. da Fonseca,L. P. Gonzaga,J. Minns, R. Otoch, H. R. Rajfio,R. Ribon, J. L. Row- band of restinga(low stature,sandy soil) wood- lett, R. A. Rowlett, D. J. Stejskal,and R. E. Webster. land and coastal forest of southern Bahia, oc- A. Gitz, D. Hassett, and J. Silbert helped DRCB curring only a short distanceinland. At least with field work in Rio Grande do Norte and Bahia. 40% of coastalforest plant speciesin southern We received valuable personal communicationsof 888 WHITNEYET AL. [Auk, Vol. 117 unpublishedobservations from J.M. Barnett,M. R. Minist6rio das Minas e Energia,Rio de Janeiro, Bornschein, B. L. Reinert, and A. Whittaker. M. Brazil. LeCroyhelped clarify the statusof the H. pileatus MEYERDE SCHAUENSEE, R. 1966. The speciesof birds specimenlabeled the type at the AMNH, D. Stotz of SouthAmerica and their distribution.Living- and T. Schulenbergkindly helped us interpret sta- ston PublishingCompany, Narberth, Pennsyl- tus of the Herpsilochmuspopulation in Rond6nia, vania. and D. Altshuler kindly helpedus perform an ap- MEYERDE SCHAUENSEE, R. 1970. A guideto thebirds propriate statisticalanalysis of the morphometric of SouthAmerica. Livingston Publishing Com- data. Specialthanks to Dan Lane for his fine wa- pany,Narberth, Pennsylvania. tercolor painting illustrating the Herpsilochmus MUNSELL. 1994. Soil color charts, revised edition. taxa treated in this paper. Macbeth Division of KollmorganInstruments Corporation,New Windsor,New York. LITERATURE CITED NAUMBURG,E. M. B. 1939. Studies of birds from east- ern Brazil and Paraguay,based on a collection ANDRADE-LIMA,D. 1982.Present-day forest refuges made by Emil Kaempfer.Bulletin of the Ameri- in northeasternBrazil. Pages245-251 in Biolog- can Museum of Natural History 76:231-276. ical diversificationin the tropics(G. T. Prance, PAPAVERO,N. 1971.Essays on the history of Neo- Ed.). ColumbiaUniversity Press, New York. tropical dipterology,with specialreference to BALDWIN, S. P., H. C. OBERHOLSER,AND L. G. WOR- collectors(1750-1905). Museu de ZoologiaUni- t•E¾. 1931. Measurements of birds. Scientific Pub- versidade de Sao Paulo, Sao Paulo, Brazil. lication of the Cleveland Museum of Natural PARKER,t. g. III, D. E STOTZ,AND J. W. FITZPATRICK. History 2:1-165. 1996. Ecologicaland distributionaldatabases. CORY,C. B., AND C. E. HELLMAYR.1924. Catalogue Pages118-436 in Neotropicalbirds: Ecology and of birds of the Americasand the adjacentis- conservation(D. E Stotz, J. W. Fitzpatrick,T. A. lands.Field Museum of NaturalHistory Zoolog- ParkerIII, and D. K. Moskovits,Eds.). University ical Series 13, Part 3. of ChicagoPress, Chicago. DAVIS, t. J., AND J.P. O'NEILL. 1986. A new species PAYNTER,R. g., JR.,AND M. g. TRAYLOR,JR. 1991. Or- of antwren (Formicariidae:Herpsilochmus) from nithological gazetteer of Brazil. Museum of Peru,with commentson the systematicsof other ComparativeZoology, Cambridge, Massachu- membersof the genus.Wilson Bulletin 98:337- setts. 352. PEARMAN,M. 1993. The avifauna of the Rio Macha- GONZAGA, L. P., J. E PACHECO,C. BAUER,AND G. D. riapo dry forest,northern La Paz department g. CASTIGLIONI.1995. An avifaunalsurvey of the Bolivia: A preliminary investigation.Bird Con- vanishingmontane Atlantic forest of southern servation International 3:105-117. Bahia, Brazil. Bird Conservation International 5: PERRY,A., M. KESSLER,AND N. EELME. 1997. Birds of 279-290. the centralRio TuichiValley, with emphasison HELLMAYR, C. E. 1908. An account of birds collected dry forest,Parque Nacional Madidi, Depto. La by Mons.G. A. Baerin thestate of Goyaz,Brazil. Paz, Bolivia. Pages557-576 in Studiesin Neo- NovitatesZoologicae 15:13-102. tropical ornithologyhonoring Ted Parker(J. V. HELLMAYR,C. E. 1929. A contribution to the orni- Reinsen,Jr., Ed.). OrnithologicalMonographs thologyof northeasternBrazil. Field Museum of No. 48. Natural History ZoologicalSeries 12, No. 18. PETERS,J. L. 1951. Check-list of birds of the world, ISLER, M., P. ISLER, AND B. M. WHITNEY. 1997. Bio- volume 7. Museum of Comparative Zoology, geographyand systematicsof the Thamnophilus Cambridge,Massachusetts. punctatuscomplex. Pages 355-381 in Studiesin PINTO,O. M. O. 1932.Resultados ornithologicos de Neotropicalornithology honoring Ted Parker (J. uma excursaopelo oestede Sao Paulo e sul de V. Remsen,Jr., Ed.). Ornithological Monographs Matto Grosso. Revista do Museu Paulista 17: No. 48. 689-826. ISLER,M., P. ISLER,AND B. M. WHITNEY. 1998. Use of PINTO, O. M. O., AND E. A. CAMARGO. 1957. S6bre vocalizationsto establishspecies limits in ant- umacolegao de avesda regiaode Cachimbo(sul birds (Passeriformes:Thamnophilidae). Auk do Estado do Para). Pap6is Avulsos do Depar- 115:577-590. tamentode ZoologiaSao Paulo 13:51-69. MAACK,R. 1962.Geologia e geografiafisica da bacia REMSEN, J. V., JR., AND S. K. ROBINSON. 1990. A hidrogrfficado rio dasContas no estadoda Ba- classificationscheme for foragingbehavior of hia. Universidade do Parana, Curitiba, Brazil. birds in terrestrial habits. Pages144-160 in MAGNAGO, H., M. t. MARTINS DA SILVA, AND B.C. Avian foraging:Theory, methodology, and ap- FONZAR.1983. Vegetagao,4.1. Estudo Fitogeo- plications (M. L. Morrison, C. J. Ralph, J. Ver- graffco.Pages 581-617 in ProjetoRadambrasil, ner, and J. R. Jehl, Jr., Eds.). Studies in Avian Levantamento de Recursos Naturais, volume 31. Biology No. 13. October2000] Herpsilochmuspileatus Complex 889

RIDGELY,R. S., AND G. TUDOR. 1994. The birds of and vocalizationsof Gyalophylaxand Megaxenops SouthAmerica, vol. 2. Universityof TexasPress, (Furnariidae), two little-known generaendemic Austin. to northeastern Brazil. Condor 96:559-565. RIZZINI, C. T., A. E COIMBRA-FILHO, AND A. HOUAISS. WILLIS, E. O. 1967. Behavior of Bicolored Antbirds. 1988. Ecossistemas Brasileiros. Editora Index, Universityof California Publicationsin Zoology Rio de Janeiro, Brazil. No. 79. SALGADO,O. A., S. JORDYFILHO, AND L. g. C. GON- ZIMMER,J. T. 1932. Studies of Peruvian birds, 5. The CALVES.1981. Vegeta•ao: As regi6es fitoeco16- genera Herpsilochmus,Microrhopias, Formicivora, gicas,sua natureza e seusrecursos econ6micos, Hypocnemis,Hypocnemoides, and Myrmochanes. estudofitogeogrfifico. Pages 485-544 in Projeto American Museum Novitates No. 538. Radambrasil, Levantamento de Recursos Natu- rais,volume 23. Minist•rio dasMinas e Energia, Associate Editor: R. M. Zink Rio de Janeiro, Brazil. SCHUBART, O., A. C. AGUIRRE, AND H. SICK. 1965. APPENDIX1. Key to specimen identification of Contribuiqaopara o conhecimentoda alimen- Herpsilochmuspileatus, H. atricapillus,and H. sello- taqao das aves brasileiras. Arquivos de Zoolo- wi. Bill widths measuredat the anterior edge of gia Departamento de Zoologia Sao Paulo 12: the nares. Tail measurementsapply to fully 95-249. grown (at most slightly abraded), normally SCLATER,P. L. 1858. Synopsisof the Americanant closedtails. Measuretail length from the point birds (Formicariidae).Proceedings of the Zoo- where centralrectrices enter skin to the tip. It is logical Society of London 1858:204-224, 232- 254, 272-289. necessaryto carefully part the feathersof the rump and uppertail covertsto obtain a clear view SIBLEY, C. G., AND B. L. MONROE. 1990. Distribution of this point. The distance from the tip of the and taxonomyof birds of the world. Yale Uni- shortest(outer) pair of rectricesto the tip of the versity Press,New Haven, Connecticut. longestpair (central, sometimesmatched by one SICK,H. 1993. Birds in Brazil, a natural history. Princeton University Press, Princeton, New Jer- or more adjacent pairs) of rectriceswe define as the "graduation maximum," modified from sey. SICK,H. 1997. Ornitologia Brasileira,ediqao revista "graduation" of Baldwin et al. (1931: figure 127). e ampliada. Editora Nova Fronteira, Rio de Ja- One rectrix of a pair may be slightly longerthan neiro, Brazil. its counterpart(especially with outer rectrices); SNETHLAGE,H. 1928. Meine Reise durch Nordost- in this case,measure from/to the tip of the longer brasilien, 3. Bausteinezur Biologie der ange- one to obtain a value maximum as closeas pos- troffenenArten. Journalffir Ornithologie76: sibleto the fully grown condition(i.e. it is not al- 668-738. wayspossible to determinewhether rectrices that STOTZ,D. E, S. M. LANYON, t. S. SCHULENBERG,D. E. appearto be completelygrown truly are). Width WILLARD, A. T. PETERSON,AND J. W. FITZPAT- at the approximatemidpoint of the central rec- RICK.1997. An avifaunal survey of two tropical tricesis obtainedfrom naturally arrangedfeath- forestlocalities on themiddle RioJiparanfi, Ron- ers (i.e. barbs regularly interlocking, vanes not d6nia, Brazil. Pages763-781 in Studiesin Neo- sleekedor flattened).Note that juvenilebirds not tropicalornithology honoring Ted Parker(J. V. fully developedcan show measurementsbelow Remsen,Jr., Ed.). OrnithologicalMonographs the thresholdsin this key. The coverand Figures No. 48. 2 (tail patterns)and 3 (femaleheads) may aid in STRESEMANN,E. 1948. Der Naturforscher Friedrich identification of individual specimensand pro- Sellow(1831) und seinBeltrag zur KenntnisBra- vide referencefor the appearanceof "discrete, siliens.Zoologischen Jahrbficher 77:401-425. whitish marks [crown]" of females. Mensural THOMAS, W. W., A.M. V. CARVALHO, A.M. A. AMO- thresholdsfor H. pileatuswere determinedfrom a RIM, J. GARRISON, AND A. L. ARBEL•EZ. 1998. sample of two males and one female (female Plant endemism in two forests on southern Ba- crown and breast from one specimenaugmented hia, Brazil. Biodiversity and Conservation7: by field observations);a larger sample could re- 311-322. sult in slight adjustmentsof someof the elements WHITNEY,B. M. 1996. Sites to save:Boa Nova, Bahia, of this key but probablywould not alter its di- Brazil. World Birdwatch18(3):9-10. agnosticutility. SeeTable 1 for selectedmeasure- WHITNEY, B. M., AND J. ALVAREZ ALONSO. 1998. A mentsfor eachspecies. new Herpsilochmusantwren (Aves:Thamnophi- lidae) from northern Amazonian Peru and ad- la Bill width less than 3.0 mm; wing chord 50 jacentEcuador: The role of edaphicheteroge- mm or less;tail lessthan 52 mm (usuallyless neity of terrafirme forest.Auk 115:559-576. than 51 ram); centralrectrices less than 5 mm WHITNEY,B. M., AND J. E PACHECO.1994. Behavior wide at midpoint and entirelyblack with or 890 WHITNEYET AL. [Auk, Vol. 117

without pale apical fringes; female crown neck and back, this hue continuing posteriorly lacking discrete,whitish marks ..... H. sellowi through flanks. Center of belly whitish (palest lb Bill width greaterthan 3.0 mm; centralrectri- area of underparts),returning to grayishthrough cesmore than 5 mm wide at midpoint (often undertail coverts. No indication of yellow or buff exceeding6 mm) and havingnarrow, whitish in flanks or undertail. Central rectrices (only one lateral fringes and (almost always) small, present) about 4 mm short of fully grown; rest of white tips; femalecrown with discrete,whit- tail fully grown.Tail weakly graduated,with outer ish marks ...... 2 rectrices about 14 mm shorter than central pair 2a Tail less than 47 mm; graduationmaximum (estimated length of central rectrices), all others lessthan 17 mm (lessthan 15 mm may be typ- within about 5 mm of length of central pair. Outer ical); femalelacking distinctochraceous wash rectrices white with black bases (ca. •/• of feather), on breast(weak tinge with grayishwash may black most extensive on proximal vanes. White be typical) ...... H. pileatus tips on successivepairs decreasing from outer- 2b Tail greaterthan 52 mm (perhapsvery rarely most to innermost.Central rectrix with only a mi- as short as 51 mm and usually 54 to 60 ram); nute white spotat tip, but with distinctwhite edg- graduation maximum greater than 17 mm es, most conspicuouson proximal margin (seeFig. (usually 20 to 25 mm); femalewith distinct 2 for extent and pattern of black and white on rec- ochraceouswash on breast,usually pervading trices). Wing coverts same deep-black as crown, mostof underpartsand sidesof head (rarely eachfeather marked with a conspicuouswhite tip, lacking a distinct wash, appearing mostly more extensiveon distal vane. Light tips on tiny whitish) ...... H. atricapillus feathers at bend of wing and on lesser wing co- verts appear as scattered spots, but arrangement of light tips on median and greater covertsmore APPENDIX2. Descriptionand distribution of Herp- regular, imparting two well-defined wing bars. silochmuspileatus (Lichtenstein, 1823). We havelo- Scapulars sharply tricolored, gray on proximal catedonly four specimens(three males and onefe- vanes, and black with a contrastingwhite margin male) of H. pileatus,including the lectotypein the on distal vanes, forming a conspicuouswhite MNHB. Several points of occurrenceare docu- stripe overlying proximal ends of wing bars. Bend mented with tape recordings.We suggestTran- of wing white, greater primary covertsblackish. coso,Bahia (16ø35'S,39ø06'W), as the precisetype Alula black with contrastingwhite margin on dis- locality. tal vane reachingtip. Remigesblackish with nar- Adult male row but conspicuouswhitish fringes on distal vanes. Inner secondaries black, marked with con- See cover. MPEG 54042 is described. Color des- spicuouswhite margins on distal vanes,cotermi- ignations from direct comparison with Munsell nous with white-margined scapulars. Iris dark soil color charts (1994). Plumage fresh and un- brown; upper mandible blackishwith grayish to- abraded. Crown (exceptfor a few minute whitish mia, lower mandible grayish;legs and feet bluish- feathers near base of bill) deep, semiglossyblack. gray, color of soles not recorded. Selected mea- Loral region (including nasal tufts) and cotermi- surements included in Table I. nous superciliary stripe bordering and sharply One other male, AMNH 5381, is available for di- contrasting with crown, white, widest (to about rect comparison.Although this specimenwas col- 3.5 mm) and becoming slightly grayish behind lected more than 165 years before MPEG 54042, its eye. Preocular spot (at anterior edge of eyering) wing and tail were more abradedat the time of col- and 2.3-mm wide postocular stripe to a point lection, and it was obviously prepared for display about I1 mm posterior to orbit same deep black as in a case(legs wired and bent, glasseyes), there is crown. Facialregion samewhitish as superciliary only a slight indication of fading of the plumage stripe, merging posteriorly with grayer sides of (black areas somewhat"flat," lacking semiglossy neck. Entire upperparts from nape to rump me- effect of fresh specimen, but not foxed toward dium-gray (Gley chart 2: 5/5PB) with hint of brownish). It matchesthe above descriptionquite black-and-white feathers of interscapular patch closely.Its complete,fully grown tail (tip of right (most are white with black tips most extensiveon outer rectrix missing) has a graduationmaximum distal vane) barely showing through (pattern of 14.2 mm. Color photographsof the lectotypeof varies with handling of specimen).Throat whitish H. pileatusare closely similar in proportions and (feathers with lightly grayish tips), essentially plumage features to the above two males. Prepa- concolor with facial and superciliary regions. ration for mounting of the lectotype and AMNH Breast slightly darker gray (whitish feather cen- 5381 is so remarkably similar as to suggestthat ters contribute to a weakly mottled effect), espe- these two specimensfrom the 1830s were pre- cially at sides,which are nearly as dark as sidesof pared by the samehand. October2000] Herpsilochmuspileatus Complex 891

Adult female extentof thishue makeschart matching especially difficult); sides, flanks, and undertail coverts con- See cover and Figure 2 (left side). MPEG 54043, color but slightly paler. Center of belly whitish. the only specimenknown to us, is described. Tail as in male described above, but black more ex- Plumagefresh and unabraded.Tiny feathersof fo- tensiveon outer rectrices(covering more than the recrown black with light buff (2.5Y 7/6) tips, im- basal half of the feather); graduation maximum parting a weakly streaked effect. Crown deep 13.5 mm (centralrectrices within ca. 2 mm of fully black with irregularly scattered,short but con- grown). Wings and soft parts as describedfor male spicuouswhitish streaksor spotsformed by white above. Selected measurements included in Table 1. distal marginsof somefeathers. Loral region (in- cludingnasal tufts) samelight buff as forecrown, blending posteriorly into whiter superciliary Distribution stripe and facial region. Preocular spot and pos- tocular stripe same deep black as crown. Sidesof Herpsilochmuspileatus is known only from the neck and entire upperparts same medium-gray coastal,sandy-soil forests of Bahiasouth of the city (Gley chart2: 5/5PB) as male describedabove only of Salvadorand BaiaTodos os Santos(northernmost slightly duller. Throat and malar region creamy- point ca. 25 km by road north of Valen•a: 13ø01'S, white, appearingweakly mottled owing to black 38ø59'W)south at least to Barra do Cai (ca. 17ø00'S, featherbases and white tips. Breastgrayish with a 39ø10'W;Fig. 4). weak overlay of buff (nearest 5Y 7/3, but limited