WHEN BLACK PLUS WHITE EQUALS GRAY: THE NATURE OF VARIATION IN THE VARIABLE SEEDEATER COMPLEX

Transcription

1 Vlume N.2 ORNITOLOGIA NEOTROPICAL 7: , 1996 CiJ' The Netrpical Ornithlgical Sciety WHEN BLACK PLUS WHITE EQUALS GRAY: THE NATURE OF VARIATION IN THE VARIABLE SEEDEATER COMPLEX (EMBERIZINAE: F. Gary Stiles SPOROPHILA) Institut de Ciencias Naturales, Universidad Nacinal de Clmbia, Apartad 7495, Bgtá D.C., Clmbia. Resumen. Las afinidades taxn6micas del Espiguer Variable (Sprphila aurita) yel Espiguer Alifajead (S. americana) han sid discutids pr más de 80 añs. El descubrimient de una zna de hibridizaci6n entre el primer y el Espiguer Gris (S. intermedia) -anterirmente n cnsiderad cm emparentad debid a que el plumaje definitiv del 0" es gris, n blanc y negr -me estimul6 a reexaminar esta cuesti6n. Mi hip6tesis de trabaj era que existiera una relaci6n estrecha entre tdas estas frmas. Esta hip6tesis fue apyada pr la gran similitud mrfl6gica y las distribucines casi perfectamente cmplementarias de tdas ellas, la identificaci6n de tra zna de slapamient y hibridizaci6n limitada, y pr la existencia de variaci6n previamente ignrada dentr de la especie intermedia. Cncluy que S. intermedia es un miembr integral del llamad "cmplej del Espiguer Variable", y que ést cnstituye un grup mnfilétic recncible al nivel de superespecie. Ls patrnes de distribuci6n gegráfica y divergencia mrfl6gica me permiten recncer ls siguientes cuatr alespecies: S. crvina (Espiguer Variable); S. intermedia (Espiguer Gris); S. murallae (Espiguer del Caquetá); y S. americana (Espiguer Alifajead). Cn base en un análises cuantitativ de la variaci6n dentr de S. intermedia, lleg a la cnclusi6n de que la subespecie agustini n es recncibre, cm tampc l es S. c. chcana, de acuerd cn tr estudi. Se prpne una hip6tesis tentativa para la histria zgegráfica del grup. Abstract. The degree f relatinship between the Variable (Sprphila aurita) and Wing-barred (S. americana) Seedeaters has been debated fr ver 80 years. The discvery f a zne f hybridizatin between the frmer and the Gray Seedeater (S. intermedia), nt previusly cnsidered t be clsely related because f its gray rather than black-and-white adult 0" plumage, prmpted me t reexamine this questin. M y wrking hypthesis f a clse relatinship between all f these frms was supprted by their very similar mrphlgy and almst perfectly cmplementar y distributins, the recgnitin f anther zne f limited verlap and hybridizatin, and by heretfre unappreciated variatin within S. intermedia itself. I cnclude that S. intermedia is in effect a member f an enlarged Variable Seedeater cmplex, a mnphyletic unit best recgnized at the superspecies level. Patterns f mrphlgical differentiatin and distributin lead me t prpse the recgnitin f fur allspecies: S. crvina ( = "aurita"), Variable Seedeater; S. intermedia, Gray Seedeater; S. murallae, Caquetá Seedeater; and S. americana, Wing-barred Seedeater. Quantitative analysis f the variatin within S. i. intermedia indicates that the subspecies agustini is nt recgnizable; likewise, synnynúzatin f S. c. chcana, suggested in a previus study, is supprted. A tentative hypthesis fr the histrical zgegraphy f the grup is prpsed. Accepted 5 June INTRODUCTION The genus Sprphila includes sme 30 species f small, bulbus-billed finches that inhabit mstly varius srts f nnfrest habitats like savannas, pen marshes, pastures, and secnd grwth, and cnsume mainly grass seeds (Meyer de Schauensee 1952, Stiles 1983, Stiles & Skutch 1989, Ridgely & Tudr 1989, Ouellet 1992). Althugh the genus ccurs frm suthern Texas t central Argentina, the area f highest diversity is eastern Suth America suth f the Amazn basin (Meyer de Schauensee 1952, Ouellet 1992). Males f Sprphila are mstly bldly patterned in cmbinatins f black, white, gray; buff, and/r 75

2 lles chestnut; females and immatures are usually brwnish, unpatterned, and ften exceedingly difficult t identify when nt accmpanied by adult males. Fr these reasns, the taxnmy f Sprphila has been based almst exclusively n the definitive male plumages, with t4e characters f females and ther aspects f mrphlgy being used haphazardly if at all. Fllwing Meyer de Schauensee (1952), the tendency has been t grup the species accrding t male.plumage patterns ("black-and-white", "gray", etc. ), an arrangement recently further elabrated by Ridgely & Tudr (1989). Hwever, what was riginally devised fr cnvenience has tended t becme accepted as a taxnmic scheme, with the result that where O' plumages differ, tw clsely related species might be placed in different species grups and their relatinship verlked. Such a case is described in this paper. Species limits in the Sprphila aurita -S. americana grup, which might be called the "Variable Seedeater cmplex", have lng been a surce f debate (Chapman 1915, Hellmayr 1938, Meyer de Schauensee 1952, Olsn 1981b, AOU 1983). At the present time, the majr debate centers n whether tw grups f races with black-and-white adult males and widely disjunct distributins, cmprise a single species r tw distinct species (r allspecies). The grups are S. aurita, with 4 r 5 subspecies in Middle and extreme western Suth America, and S. americana, with tw disjunct frms ( ne pssibly including tw races) in Suth America east f the Andes. Treatment f all f these as races f a single species riginated with Meyer de Schauensee (1952), and has been fllwed by many recent authrs including Olsn (1981b) and Ridgely & Tudr (1989). The AOU (1983) disagreed, preferring t treat the tw grups as allspecies f a superspecies in view f differences in the pattern f the O' plumages, and the wide disjunctin in their ranges. In either case, the debate has been limited t these tw grups with black-and-white males. That the slutin t this prblem might lie elsewhere ccurred t me when I examined a series f Sprphila cllected recently at several sites in the upper Rí Sinú drainage S f Tierralta, departament de C6rdba, NW Clmbia (apprx. 7 55'N, 76 20'W), by R. Jiménez and R. Tafur. Included in this cllectin were several putative hybrids between S. aurita and a species~ nt previusly cnsidered t be clsely related, the Gray Seedeater s. intermedia. Once ver my initial amazement, I decided t reevaluate the relatinship between s. americana and aurita in the light f the pssibility that s. intermedia might als be an integral member f the Variable Seedeater cmplex. The bjective f this paper is therefre t present a detailed examinatin f the plumages, external mrphlgy, and distributin f s. intermedia in relatin t thse f s. americana and aurita, in rder t test my wrking hypthesis f a clse relatinship amng all f these frms. TAXONO:MIC HISTORY The first member f this cmplex t be described was s: americana, by Gmelin in 1789 frm a plate by Daubentn in Buffn's "Histire Naturelle d'oiseaux". Originally the species was ascribed simply t " America", but the type lcality was later restricted t Cayenne by Hellmayr (1938). Bnaparte described aurita in 1850, suppsedly frm Brazil, but the type was reexamined by Sclater (1871), wh prnunced it typical f birds frm C Panama, and the type lcality was amended accrdingly. Sclater had described crvina with its black male plumage frm Oaxaca, Mexic in 1859, and tw years later Cabanis described hffmannii (with blackand-white males) frm S Csta Rica. Betwren 1863 and 1865 Lawrence named semicl. laris and frtipes frm Panama (later synnymized with aurita), badiiventris frm E Nicaragua (a synnym f crvina), cllaris frm SW Panama (a synnym f hffmannii), and hicksii, SUppsedly frm C Panama but mre likely frm Buenaventura, Clmbia: see Olsn (1981b). This plethra f names was drastically pruned by Hellmayr (1938), wh nted that crvina ccurred alng the Caribbean slpe f Central America frm Mexic t Panama, and ascribed the great variatin amng males frm C Panama t hybridizatin between crvina and aurita. He accrdingly recgnized nly ne species in Middle America (aurita), with crvina as a subspecles. In Suth America, phthalmica was described frm W Ecuadr by Sclater in 1860, and cnsidered a race f aurita by Chapman (1915, 1917). Chaprnan als (1915) described the subsequently cntrversial frm murallae frm "La

3 IARIATION IN THE VARIABLE SEEDEATER COMPLEX Muralla" ( = Mrelia), Caquetá, SE Clmbia, as a cis-andean race f aurita, cmparing it with phthalmica but suggesting that it might represent a "link" t s. lineata ( = americana). Hellmayr (1938) strngly suggested that murallae might be clser t americana than t aurita. The next step was taken by Meyer de Schauensee ( 1952), wh lumped aurita and its races int americana, cnsidering murallae t represent the "perfect cnnecting link", and thereby setting.the stage fr the current debate. He als described the r,1ce chcana in 1950, fr the birds frm the Pacific slpe f Clmbia S t Valle del Cauca, setting the N limit f phthalmica in extreme SW Clmbia. The recent revisin f this cmplex by Olsn (1981b) clarified the cnfusin surrunding the name aurita. He shwed that the ntriusly variable ppulatin f C Panama is essentially a stable hybrid swarm between the Caribbean crvina and the Pacific hicksii. Since Bnaparte's riginal descriptin f aurita was very vague and the type specimen had disappeared frm the Paris Museum by 1935 (Hellmayr 1938), Olsn argued that it is impssible t assciate the name aurita with either parental ppulatin. This effectively invalidates this name, but t avid cnfusin I will cntinue t use it, in qutes, until the cncluding part f this paper. With better material than that avi!.ilable t previus authrs, Olsn als shwed that hffmannii was a valid race, but chcana was a synnym f hicksii. Hwever, he misinterpreted the statements f Meyer de Schauensee (1950, 1952) regarding the gegraphicallimits f phthalmica and chcana ( = hicksii). In cntrast t the cnfusin and debate surrunding the black-and-white prtin f the Variable Seedeater cmplex, the taxnmic histry f the Gray Seedeater has been relatively tranquil. The earliest name fr the species, grisea Gmelin, was early shwn t be unidentifiable and the name intermedia Cabanis 1851 (type lcality Puert Cabell, Venezuela) has been used nearly exclusively fr ver 75 years. The nly taxnmic cnfusin has been with the Slate-clred Seedeater s. schistacea, which is similar in male plumage but quite distinct in mrphlgy, as recgnized by Meyer de Schauensee (1952). Three races have been split frm nminate intermedia f N and E Clmbia and N Venezuela: insularis Gilliard 1946 fr the birds f Trinidad; bgtensis Gilliard 1946 fr Clm-~ bian ppulatins W f the Eastern Andes and S f the Caribbean castal plain; and agustini Meyer de Schauensee 1947 fr the birds f the upper Magdalena valley f Clmbia. Mst recent debates have cncerned the validity and gegraphic limits f these races (e. g., Olivares 1969). Mre prblematical has been the "blackish race f the Gray Seedeater", anchicayae, described by Miller (1960) frm the Anchicayá Valley f W Valle del Cauca, Clmbia. This frm was lumped int bgtensis by Hilty & Brwn (1986) but ascribed t americana by Ridgely & Tudr (1989). In effect, bth were half right: I cnsider " anchicayae" t represent hybrids between s. i. bgtensis and s. " aurita" hicksii, as discussed belw. In spite f the prblematic nature f " anchica~e", there had been n suggestin prir t the present study that intermedia was clsely related t the Variable Seedeater cmplex, much less a member f it. Since the clue t the prblem was prvided by the Tierralta series mentined abve, the time has cme t describe these specimens in detail and t demnstrate their cnnectin with " anchicayae". THE HYBRIDS: THE TIERRALTA SERIES AND "ANCHlCAYAE" The sample f Sprphila frm the Tierralta area cnsists f 6 adult males, ne O' mlting frm 9-like (first basic?) t definitive plumage, and ne 9. All were cllected between 6 June and 24 July 1991 in secnd-grwth vegetatin alng varius creeks and rivers (Rí Verde, Quebrada Seferin, Quebrada Gaita, Quebrada Chibgad6) at elevatins f m, abve the site fr the Urrá hydrelectric prject, as part f an envirnmental impact study. This general area was heavily frested thrugh the 1970s, but clnizatin and defrestatin have prceeded rapidly during recent years (R. Jiménez, pers. cmm.). The males f the Tierralta series include ne apparently phentypically "pure" intermedia, ne apparently "pure" hicksii, and fur that run the gamut between them (Fig. 1); the yung mlting O' als appears t be attaining an intermediate plumage, clser t intermedia. In mst dimensins the hicksii-like birds tend t be smaller than the intermedia-like birds, but the crrespndence is far frm perfect (Fig. 2). The relatin between plumage phentype and bill clr is als nly apprximate, with intermedia- 77

4 STILES ICN ~ ~---~: rcn 3J~91 ~- UN ,- ICN I'N 31?41 )~uv 1&46 FIG. 1. Plumages f Sprphila males cllected at Tierralta, and a specimen f "anchicayae" (UV 1846). ICN resembles ICN in plumage; ICN is acquiring a plumage like that f ICN Nte resemblance between ICN and uv Fr abbreviatins f museums see text. like birds tending t have paler bilis. The 9 is small in mst dimensins and might thus be clser t hicksii (see belw), but is prbably nt safely assignable t either frm. Miller's (1960) riginal descriptin f anchicayae included a detailed descriptin f his tw males, and a well-executed drawing f the type by G. Christman. This drawing prtrays a bird whse plumage is virtually identical t that f the mst intermediate (ICN 31601) f the Tierralta series (cf. Fig. 1); frm the descriptin, the ther O' is very similar t the type. In additin, a specimen in the cllectin f the Universidad del Valle (UV 1846) has a plumage very like that fmiller's drawing and ICN (Fig. 1). It was taken by J. I. Brrer and G. Catañ at 340 m alng the Ri Anchicayá, very clse t Miller's type lcality, n 22 February I have n dubt that this specimen is anchicayae, which in turn represents hybrids between s. " aurita" hicksii and intermedia, in this case f the race bgtensis (see belw). Mrever, it seems likely that the anchicayae-icn plumage, essentially exactly intermediate between the tw putative parent species, represents the F1 phentype, with the ther Tierralta males resembling mre ne r the ther parental species representing F2 phentypes r backcrsses. It is ntewrthy that nne f the latter phentypes have been taken in the Rí Anchicayá regin, a pint I shall return t belw. The fact that the Tierralta hybrids were taken at several sites within a radius f 78

5 VARIATION IN THE VARIABLE SEEDEATER COMPLEX apprximately 20 km, and represent apparent backcrsses as well as a putative F1, presents much strnger evidence fr a clse relatinship between intermedia and "aurita" than wuld a single islated hybrid individual (cf. Cracraft 1989). This is an imprtant cnsideratin since ccasinal hybridizatin ccurs rather widely in SpQrphila, even with species f related genera (Sick 1963; sre belw). Given the existence f tw znes f limited verlap and hybridizatin between " aurita" and intermedia, I decided that a detailed examinatin f plumage, mrphlgy, and distributin f these tw species in Clmbia and adjacent areas was warranted. In the prcess, I fund that the degree f variatin in plumage f the varius frms, while ften mentined:, had never been precisely quantified, and that such an analysis culd shed light n the taxnmic validity f several f the described subspecies as well as n the relatinship between the Gray and Variable Seedeaters. METHODS Fr this study I examined a ttal f 297 specimens in the fllwing museums (fr each museum, I give in parenthesesthe number f specimens f each frm in the fllwing rder: intermedia/bg tensis Ihicksiilphthal mica/ murallae/ americana/hybrids): Institut de Ciencias Naturales, Universidad Nacinal de Clmbia (64/42/7/33/4/O/6); Muse de la Universidad 79

6 de La Salle, Bgtá (11/12/O/O/16/O/O); Unidad de Investigaci6n "Federic Medem" (UNIFEM) f the Institut de Recurss Naturales (Inderena), Bgtá (5/3/33/O/O/O/O); Universidad del Valle, Cali (O/17/27/O/O/O/1); and the Naturhistrisches Museum in Vienna, Austria (3/0/0/ 6/0/7/0). In additin, Miguel Lentin f the Clecci6n Phelps in Caracas, Venezuela, and Manuel Marín A. f the Museum f Natural Histry f Luisiana State U niversity in Batn Ruge measured 5 specimens f americana and 13 f murallae fr me, giving a ttal sample f 315 specimens. All measurements were taken t the nearest 0.1 mm using dial calipers, f: wing chrd, tail length, depth f bill at nstril, billlength frm nstril, and tarsus length (cf. Baldwin et al. 1931). I chse length frm nstril as my measure f bill length because f the difficulty in btaining accurate measurements f expsed r ttal culmens f such bulbus-billed birds; and bill length frm nstril yielded a significantly lwer cefficient f variatin in mst taxa than did either culmen measurement. Because the sample distributins did nt deviate significantly frm nrmality (X2 tests), and variances were hmgeneus (F-tests), measurements were analyzed by ne-way ANOVA; where significant differences were fund amng samples I used Tukey tests t determine which samples differed significantly (Zar 1984). I determined by Student's t-tests that males and females f all frms differed significantly in wing length but nt in measurements f the bill, tail, r tarsus (P>0.10 in all cases). I therefre cmbined measurements f males and females fr analyzing the latter measurements, but analyzed wing length fr each sex separately. I als cmpared the mean ratis between varius measurements fr all frms; a rati falling utside the 95% cnfidence interval fr the given measurements was cnsidered t differ significantly frm the remaining samples (Zar 1984). I analyzed clrs f females and immature plumages, and the depth f gray in adult males f intermedia, with reference t the clr guides f Smithe (1975, 1981), based n the Munsell system. Fr certain aspects f male patterns (e. g., size f wing speculum), I made cmparisns using t-tests; fr thers (e. g., size and intensity f neck-patches and thrat-bands) I devised subjec- tive scring systems and analyzed these using tw-sample Klmgrv-Smirnv tests (Skal & Rhlf 1981 ). I als made drawings f several representative individuals f all taxa t better illustrate the degree f plumage variatin amng males. Distributins f the different taxa in Clmbia were pltted frm cllecting lcalities f specimens in the afrementined cllectins, frm lcalities given by Meyer de Schauensee ( , 1950, 1952), Nicéfr & Olivares (1975), Serna (1981, 1984), and Negret (1992), and frm unpublished field data f T. McNish, L. Rsselli, P. Salaman, and my wn field ntes. Fr the verall distributins f all frms f the cmplex, I als used infrrnatin frm ffrench 1973 (Trinidad and Tbag), Snyder 1966 (Guyana), Phelps & Phelps 1950 and 1963 (Venezuela), Pint 1944 (Brazil), Meyerde Schauensee (1952), and Ridgely & Tudr (1989), supplemented by Venezuelan and Peruvian lcalities supplied by M. Lentin and M. Marín, respectively. RESULTS Because the descriptins f varius frms in the Variable Seedeater cmplex have been based n ne r a few characters, sme f which have subsequently been shwn t be invalid, I will first characterize in detail the plumages f each recgnizable member f the cmplex, in the prcess attempting t evaluate variatin bth within and between subspecies and species. I will then prceed t a quantitative analysis f external mrphlgy t cmplete the characterizatin f each frm. This in turn will permit me t reexamine the distributins f all frms, crrecting previus errrs due t misidentificatins. Finally, I will attempt a zgegraphic verview f the cmplex and present my taxnmic cncluslns. Clrs, pattems, and plumage sequences. Sprvphila "aurita": Uncertainty regarding the distributinallimits f the races hicksii (including chcana) and phthalmica (Olsn 1981b) reflect insufficient understanding f the range f ppulatin and individual variatin in bth frms. T illustrate this variatin, I have drawn representative examples f the adult male plumages f three Clmbian ppulatins: hicksii frm nrthern Chc6 and nrthern Antiquia, and frm suthwestern Valle del Cauca (Fig. 3); and phthalmica 80

7 VARIATION IN THE VARIABLE SEEDEATER COMPLEX FIG. 3. Plumage variatin in adult males f s. "aurita" hicksii. Left-hand clumn: N Chc -N Antiquia sample; right-hand clumn: W Valle del Cauca sample. Birds are arranged frm palest-thrated t darkest-thrated, tp t bttm. 81

8 ~ STILES frm extreme suthern Cauca and Nariñ (Fig. 4). The latter tw ppulatins represent the clsest knwn apprach f the distributins f these tw races (see belw). These races were previusly separated n the basis f the whiter thrat and narrwer pectral band f phthalmica relative t hicksii (cf. Olsn 1981b), but frm Figs. 3 and 4 this is evidently an versimplificatin. A quantitative analysis f the variatin in these and ther features f the pattern f the adult O' plumage (Table 1) shws verlap in all characters, with the differences between the nrthern and Valle ppulatins f hicksii being cnsiderably less than the differences between the latter and phthalmica. The tw races differ significantly nly in the width f the pectral band, althugh the difference in the degree f dusky t blackish scaling and smudging n the sides and flanks als appraches significance. Tw ther characters are useful in distinguishing these subspecies. Mst individuals f phthalmica (at least 17 f my sample f 22) shw white n the lwer eyelid, varying frm ne r tw feathers t a cbspicuus crescent (cf. Fig. 3); I have seen n hicksii O' with even a vestige f this mark (X2 = 20.25, 1 d. f., P<0.001). AIs, the wing speculum f phthalmica averages significantly lnger than that f hicksii (t = 6.64, P<0.001; cf. Table 2). Using a cmbinatin f characters, I can separate virtually all adult males f the tw races examined t date. I shuld nte here that Meyer de Schauensee (1950) distinguished chcna f Clmbia frm aurita f Panama mainly n the basis f its whiter thrat. Frm the present analysis and the data f Olsn (1981b), I have n dubt that Olsn was crrect in cnsidering chcana a synnym f hicksii. There is a definite difference in clratin f the adult females f the tw frms as well, as nted by Meyer de Schauensee (1952). Females hicksii are duller, mre live-brwn abve (between 123, Raw Umber and 28, Olive / V.M 2038' --- rc.n 30+?5 ;-~ "' rcn 4s+ ~-./ ICN 409'. / ~ ICIt 4099 / /..- ~ IC.N 4091 FIG. 4. Plumage variatin in adult males f s. "aurita" phthalmica; cmpare with hicksii (Fig. 3).

9 VARlATION IN THE VARIABLE SEEDEATER COMPLEX TABLE 1. Cmparisns f different cmpnents f the definitive male plumage between three Clmbian ppulatins f Sprphila.aurita. by Klmgrv-Smirnv tw-sample tests. Ppulatins are: NCA = N Chc6, N Antiquia (hicksi ); WVC = W Valle del Cauca (hicksi ); and SCN = S Cauca and Nariñ (phthalmica). Ppulatins and Plumage Areas Plumage Scres ):; Dmaxl Definitins A. Thrat Pattern n.s..270 n.s. = Entirely white 1 = Chin < 1/4 black = small malar streak. 2 = 1/4-1/2 f thrat black, including 1/2 + f chin. 3 = > 1/2 f entire thrat black B. Breast-Band c. Sides and Flanks = Imcmplete: center f chest gray n.s..460* r white. 1 = Cmplete, v. narrw at center. 2 = Cmplete, mderately brad, especially at sides. 3 = Cmplete, brad }.152 n.s. 17 } = Entirely white I = Gray, < 1/4 with dark smudging. 2 = 1/4-1/2 f area with dark smudging r scaling. D. Rump NCA 11 2 MVC 10 SCN } :}.161 n.s.251 ll.s. = Clear white 1 = Sme gray r dusky smudging. 2 = Gray r black ~ white. 3 = Mstly dusky t blackish with sme white. Abbreviatins fr prbabilities: n.s. = p > 0.10; = 0.10 > 0.05; * = p < 0.05; ** = p = < Brwn); the thrat, chest, and sides are paler (near 123A, Cinnamn; center f chest appraching 26, Clay Clr), shading t buffy-white (near 54, Cream) n the abdmen. Females phthalmica are a richer, mre chraceus brwn abve (brighter than 123, Raw Umber), paler chraceus-tawny (between 123A, Cinnamn and 123B, Clay Clr) n the thrat, breast, and sides, shading t pale chraceus yellw (paler and brighter than 123C, Yellw Ocher) n the belly. Yung females hicksii are brighter abve and mre yellwish belw than adults, whereas yung females phthalmica are duller than their respective adults, such that mst immatures f the tw frrs are indistinguishable. Sprphila intermedia: Gilliard (1946) separated bgtensis frm nminate intermedia n the basis f its suppsedly darker, mre blackish head and upper back, and the presence f white "pst-auricular" patches (absent in intermedia). Gilliard's type was frm Lmitas, Valle del Cauca, and his series included birds frm Vall(', Cauca, Tlima, 83

10 ST~LES TABLE 2. Size f wing speculum (greatest length f expsed white area frm primary cverts t distal end f speculum) in varius samples f the Variable Seedmter cmplex. Taxn Sample x ::I: SD (mm) s. "aurita" hicksii NCA WVC :1: :I: s. " aurita" phthalmica s. americana americana S. a. murallae s. intermedia intermedia s. intermedia bgtensis SCN CAP CPA NCL LLN VMG MCP :1: :1: :1: :1: :1: :I: :1: Samples: NCA= Nrthern Chcó-Nrthem Antiquia; WVC = Westem Valle del Cauca; CAP = Cayenne- Pará, Brasil; CPA = Caquetá, Putumay, Amaznas; NCL = Nrthern Caribbean lwlands; LLN = Llans, Nrte de Santander; VMG = Valle del Magdalena; MCP = Medellin area, Cauca Valley, Pacific slpe. Huila, and Antiquia. Meyer de Schauensee (1947) separated agustini frm bgtensis n the basis f its paler gray upperparts with the crwn nt darker than the back, larger white spts n the sides f the neck, and a larger white area n the abdmen. He gave the range f agustini as the Magdalena Valley, nrth t N Tlima (Hnda). Miller (1947, 1952) did nt cllect any frm f intermedia in the upper Magdalena Valley, and later (1960) stated that specimens f agustini he had examined were nt paler gray than bgtensis, but that agustini was "dubtless recgnizable by the cmbinatin f characters ascribed t it". The prblematic nature f this frm was highlighted by Olivares (1969), wh fund that large series frm the Magdalena Valley and adjacent slpes in N Huila, Tlima, and W Cundinamarca included birds shwing the suppsed diagnstic features f bth bgtensis and agustini, as well as intermediates. He thus cnsidered this entire area t represent a zne f intergradatin, in effect leaving at best a small area in S Huila fr " ".. pure agusurn. I examined quantitatively the variatin in plumage f adult males f s. intermedia thrughut Clmbia in rder t evaluate the validity f agustini and t verify the characteristics f bgtensis relative t thse f intermedia. I fund cnsiderable variatin in the darkness f the gray clratin in males thrughut Clmbia west f the Andes (Table 3). Ppulatins frm the middle Magdalena Valley and frm Valle del Cauca are statistically indistinguishable in the darkness f the gray clratin f the crwn, back, rump, and breast (nly in the latter des the difference even apprach significance). I can als detect n difference between these ppulatins in the size f the white neck spt r in the size f the white area n the abdmen, and the prprtin f bth ppulatins shwing the neck spt is similar (Table 4). Perhaps the prprtin f dark birds with incnspicuus neck spts is slightly higher in Valle, but the range f variatin in bth ppulatins is such that n clear-cut divisin int tw races is pssible. Thus I tentatively cnsider agustini t be a synnym f bgtensis, while acknwledging that the case will nt be entirely clsed until a gd series frm the type lcality f agustini can be included in the analysis. When bgtensis (including the Magdalena Valley birds) is cmpared with nminate inter. media (Fig. 5), a number f interesting facts emerge. First, I culd detect n statistically significant difference between these frms in the depth f the gray clratin, r in the size f the white area n the abdmen (Table 3). The tw races d differ strngly in several ther characters, hwever. The white neck spt is significantly larger and is mre frequently present in bg- 84

11

12 STILES tensis (Table 4). The size f the wing speculum averages significantly larger in bgtensis as well (Table 2j t = 4.53, P<O.OOl). A previusly verlked but highly interesting aspect f the pattern f bgtensis is a band f paler gray acrss the lwer thrat, effectively cnnecting the white spts n the sides f the neck. This mark is lacking altgether in 77 % f my sample f intermedia, the remainder having but a faint trace f paler gray n the thrat. By cntrast, ver 90 % f bgtensis have at least a trace, and ver half shw a definite pale band crssing the thrat, a highly significant difference (Table 4). As in the case f hicksii vs. phthalmica discussed TABLE 4. Cmparisn f aspects f plumage pattern in tw samples f s. i. bgtensis, and in 5: i. intermedia, by Klmgrv-Smirnv tw-sample tests. Fr abbreviatins fr prbabilities see Table 1. SAMPLES, PATTERN PLUMAGE SCORES r Dmax DEFINITIONS WH~TE PATCH, SIDE OF NECK bgtensis Valle del Cauca = absent Magdalena Valley Ttal intetmedia n.$ ** 1 = faint, gray 2 = small, distinct 3 = large, distinct PALE BAND ACROSS THROAT bgten, 'alle del Cauca = absent n.s. 1 = faint, pale area 2 = faint band Magdalena Valley = distinct band Ttal intermedia *' SIZE OF WHITE AREA ON BELL y bgtensis 'alle del Cauca Magdalena Valley Ttal intermedia n.s n.s. = <1/4 white I = 1/ white 2 = 1/ white 3 = 1/2 White 86

13 VARIATION IN THE VARIABLE SEEDEATER COMPLEX ICN 2BISt ICN '62'9 lcn rch 15'831 FIG. 5. Plumage variatin in adult males f s. inte1media. Left: S. i. intermedia; upper bird is "typical. unifrm gray with just a trace f pale n thrat; lwer bird is a mre patterned individual frm Nrte de Santander (see text). Right: S. i. bgtensis. Upper bird is "agustini. phentype, lwer bird is darker "bgtensis. phentype, accrding t Meyer de Schauensee (1947); bth birds were cllected at La Vega, Cundinamarca. abve, use f multiple characters permits the discriminatin f ver 90 % f the males f bgtensis frm thse f intermedia, and I cnsider bgtensis a valid race. Of particular interest is the fact that the pale neck-spts and thrat-band f mst bgtensis ccupy precisely the same area as des the white band acrss the thrat in "aurita»: in fact, in hicksii this mark may rarely be brken int separate neck-spts and thrat-band, much as in bgtensis (UF 3329 in Fig. 3). A ntewrthy aspect f the variatin within intermedia itself is that the prprtin f males with at least a trace f the neck spt and/r thrat band is higher amng specimens frm Nrte de Santander than amng thse frm ther areas (4 f 10 vs. 3 f 20 fr neck spts, 3 f 10 vs. 3 f 20 fr thrat bands). As will be discussed belw, Nrte de Santander is als the nly place where bgtensis has been reprted frm the east slpe f the Andes-Sierra de Perijá; the strnger neck and thrat pattern amng the intermedia frm this area culd be the result f gene flw frm bgtensis. Adult females f the tw races als differ. Thse f intermedia are live-brwn n the head and upperparts (near 29, Brwnish Olive, but paler); the thrat is paler, the malar area brighter and mre chraceus (near 123B, Clay Clr), shading t between 123A (Raw Umber) and 39 (Cinnamn) n the breast and sides, and paler, near 54 (Cream) n the belly. Adult females f bgtensis are paler and grayer; the head and upperparts are a grayish brwn (between 79, Glaucus, and 119B, Dark Drab), the breast and sides a buffier brwn (between 223D, Tawny, and 39, Cinnamn), shading t a duller buffy-white n the abdmen. The amunt f pale, hrn clr t yellw n the bill als may be greater in females f bgtensis: mst have 1/4 r mre f the bill pale, and nearly 1/3 shw extensive pale areas n the maxilla, whereas nearly all intermedia skins examined had less than 1/4 f the bill pale. Hwever, because f pssible pstmrtem changes in bill clr the degree t which this difference may be evident in live birds remains t be determined. Interestingly, as in the case f phthalmica and hicksii abve, immatures f bth sexes are much mre similar than are the adult females. In bth races the first-year birds are live-brwn 87

14 STILES tinged with chraceus abve (nearest 123, Raw Umber); the breast and sides are brighter, mre chraceus (nearest 123B, Clay Clr), shading t a paler buffy yellwish (between 123D, Chamis and 54, Cream) n the abdmen. FeIhales in this plumage average slightly mre yellwish belw than d males, and sex fr sex immatures f intermedia may average slightly brighter verall than thse f rgtensis, but, mst are indistinguishable. Als ntewrthy is tpe great similarity f these immatures t thse f hicksii and phthalmica in the crrespnding plumage: I cannt safely separate mst immatures f any f these frms n the basis f plumage clr. Sprphila americana, including murallae: Of greatest interest here is the extent t which murallae represents the "perfect cnnecting link" between americana and " aurita". U nfrtunately, I was able t examine nly five males and tw females f americana in detail, and hence the degree f individual variatin in this frm is prbably underestimated in the fllwing analyses. Cmpared t males f either race f " aurita", thse f americana have much less heavily marked thrats: 4 f 5 had white thrats with at mst a faint cluding f gray. The black pectral band is usually incmplete in americana, the center f the breast being either white r cluded with gray. The sides and flanks f americana RR

15 VARIATION IN THE VARIABLE SEEDEATER COMPLEX shw significantly less dark markings than hicksii, being mre like phthalmica in this respect; hwever, in americana these markings tend tward a unifrm gray cluding rather than the dusky smudging f phthalmica (and hicksii). The rump f americana is intermediate in clr between that f pthalmica (which has, n average, the whitest rump) and hicksii, in which the rump is darker, averaging mre like the latter.(table 4). The wing speculum f americana averages smaller than that f hicksii, and much smaller than that f phthalmica (Table 2; statistical tests nt perfrmed due t the small sample f americana). The mst striking difference between americana and all frms f " aurita" is the presence f ne t three welldevelped wing-bars in the frmer; n specimen f " aurita" I have seen has even a vestige f such a mark (Fig. 6). In mst f these features, males f murallae resemble thse f americana much mre clsely than thse f "aurita" (Table 4). In bth thrat pattern and the cmpleteness f the pectral band, murallae differs significantly frm bth frms f "aurita" but nt frm americana, althugh in the secnd feature the tw races f the frmer als differ significantly. The sides and f1anks f hicksii are significantly mre heavily marked with black r dusky than are thse f the ther frms, which d nt differ amng themselves; hwever, mst f the sides and f1anks f murallae shw a unifrm gray cluding like that f americana and unlike the dusky smudging f males f "aurita". The rump f murallae averages darker than that f any ther frm, but differs significantly in this respect nly frm that f phthalmica. The wing speculum f murallae averages smaller than that f any ther frm, that f americana being clsest. The nly character by which murallae culd be cnsidered intermediate between americana and " aurita" is its less well-develped wing-bars; but even here, all males murallae I have seen pssess at least ne such bar, cmpared t nne in "aurita" (Table 4). Thus, in this character as well murallae is actually much clser t americana. Females f murallae are duller and brwner verall than are thse f any frm f "aurita". The head and upperparts are dull live-brwn (clsest t 28, Olive Brwn); the thrat, breast, and sides are paler, nearest t 26, Clay Clr but slightly duller, shading t dull buff (near 223D, Tawny) r pinkish-buff (121D). An immature O' is similar but the belly is buffy-white; in nne f the specimens I have seen is there even a trace f yellw r chraceus. Unfrtunately, I was unable t cmpare the females f americana with the clr standards f Smithe (1975,1981). My ntes describe them as veryunifrm brwnish verall: plain du1l live-brwn abve, slightly paler buffy-brwn belw, shading t paler buff n the abdmen. Althugh a mre detailed, standardized descriptin f the 9 plumage f americana is clearly desirable, it appears t be mst similar t that f murallae. Females f bth appear t shw sme resemblance t thse f hicksii, but are very different frm thse f phthalmica. In sum, the status f murallae as the link between americana and "aurita" rests almst entirely upn a single feature: its less we1l-develped wing-bars. In a1l ther aspects f bth O' and 9 plumages, murallae is either much clser t americana r is mre extreme than either americana r "aurita". I therefre cnclude that lumping f these tw grups based n the suppsed intermediacy f murallae is nt justified. A mre general cnclusin frm this analysis is that the uniting f americana and " aurita" at any taxnmic level while excluding intermedia, is likewise untenable. Evidence frm hybridizatin and the similarity f immature plumages argues fr a clser relatinship between intermedia and " aurita" than between either f these and americana. The thrat markings f bgtensis als prvide a link between hicksii and nminate intermedia; in this respect bgtensis is a mre cnvincing "cnnecting link" than is murallae! This analysis clearly indicates that americana and murallae are each ther's clsest relatives, but the relatinship culd be less clse than that between, say, phthalmica and hicksii. The tw ppulatins f the latter tw frms that are clsest gegraphica1lyare als the mst similar in pattern (Table 3), but the same might nt be true f americana and murallae. Clsest gegraphically t murallae are the americana ppulatins f the middle Amawn, althugh the gap is much greater (see belw). The Amaznian ppulatins were split frm americana as the race dispar by Tdd (1922) n the basis f suppsedly larger size, larger wing speculum and whiter rump in 89

16 STILES the males, and duller clratin f the females (brwnish t grayish live vs. rich brwn abve, dull buffy-white vs. strngly buff belw). Pint (1944) did nt accept this split, cnsidering that the differences were bridged by individual variatin; but Meyer de Schauensee (1952) recgnized dispar as being whiter belw and n the rump, and with the black cllar mre prminent t cmplete. A quantitative analysis alng the lines.f thse presented here culd help settle the questin, which in any case is beynd the scpe f this paper. Hwever, t the extent that Tdd and Meyer de Schauensee are crrect regarding at least the characters f the males, the Amaznian birds may differ mre frm murallae in mst features than d thse frm the Atlantic castal districts. Because neither authr attempted t define pssible plumage changes with age, little can be cncluded with respect t the females. Extemal mrphlgy. Fr the analyses f variance f measurements, I was able t amass samples f at least 10 individuals f each sex fr all frms except fr females f americana (n = 2). Samples f hicksii, bgtensis and nminate intermedia were sufficiently large that I was able t divide them gegraphically, t evaluate pssible within-subspecies variatin in measurements as TABLE 5. Cmparisns f features f the definitive male plumages between fur "black-and-white" members f the Variable Seedeater cmplex, by Klmgrv-Smirnv tw-sample tests. Definitins f plumage patterns and abbreviatins fr prbabilities as in Table 1. TAXA AND PLUMAGE AREAS PLUMAGE SCORES }:; Dmax values A. THROAT PATTERN hicksii phthalmica murallae americana O O ~ h vs. =.339 h vs. m =.656* vs. m =.500* m vs. a =.160 ll.s. B. BREAST-BAND hicksii phthalmica murallae americana O O 8 O O O h vs. =.480** h vs. m =.844** vs. m =.501* m vs. a =.054 n.s. c. SIDES AND FLANKS hicksii phthalmica murallae amerzcana 3 O h vs. =.446* h vs. m =.446 vs. m =.136 n.s. m vs. a =.200 n.s. D. RUMP hicksii phthalmica murallae americana 2 O O h vs. =.326 h vs. m =.239 n.s. m vs. a =.545* m vs. a =.364 n.s. 90

17 VARIATION IN THE VARlABLE SEEDEATER COMPLEX TABLE 6. Relative develpment f wingbars in Sprphila americana and murallae. O = nne; 1 = trace (spts, faint scaling); 2 = cntinuus, narrw; 3 = cntinuus, brad. Dmax = frm Klmgrv-Smimv tw-sample tests. Fr abbreviatins see Table 1. TAXA SCORES 2 3 }:; Dmax LESSER COVERTS americana murallae O MIDDLE COVERTS americana murallae, * GREATER COVERTS americana murallae 6, O 2 O * well. The tw samples f intermedia were cis-andean (the L1ans, Nrte de Santander) and trans- Andean (the Caribbean castallwlands); thse f bgtensis were frm the Magdalena Valley (the suppsed range f "agustini") and the Medellin area, Cauca Valley, and Pacific slpe; and thse f hicksii were frm N Chc-N Antiquia, and SW Valle del Cauca. Thus, including the single samples f phthalmica, murallae, and americana, a ttal f nine samples were analyzed (Table 5). Highly significant variatin amng these samples in bill length, bill depth, wing length (the sexes tested separately), and taillength, and significant variatin in tarsus length, were detected by these analyses (Table 6). The results f each analysis will nw be cnsidered in detail. Bill length frm nstril: mst striking was the highly significant separatin f murallae frm all the ther frms n the basis f its lng bill. Next lngest was the bill f americana, which was significantly lnger than that f hicksii but nt f the ther taxa. N significant differences in billlength were fund between any f the frms f "aurita" and intermedia (Table 6). Bill depth: the thickest-billed frm was americana, but its bill was nt significantly deeper than thse f mst samples f intermedia. At the ppsite extreme was murallae, whse bill was nt significantly thinner than mst samples f "aurita"; the latter differed significantly frm intermedia, but nt bgtensis, in bill depth. One f the suppsed characters f agustini, a less massive bill than that f intermedia, is nt supprted by this analysis: the Magdalena Valley ppulatin f bgtensis des nt differ frm intermedia in this respect, whereas the mre western sample des s (Table 6). With the exceptin f murallae, bill depth decreases frm east t west ver all samples (Table 5). Taillength: variatin in this measurement shwed n clear pattern with respect t gegraphy r taxa (Tables 5,6). The shrtest-tailed were the tw samples f hicksii; the lngest-tailed frms were the western sample f bgtensis, murallae, and americana. Within bth intermedia and bgtensis, the tw samples differed cnsiderably (thugh nt significantly) in taillength. Tarsus length: within several taxa (bgtensis, intermedia, murallae vs. americana) tarsi f the mre eastern sample averaged lnger than thse f the mre western birds; in hicksii, birds f the nrthern sample had lnger tarsi than thse f the suthern sample. In general, hwever, variatin in tarsus length was less than in any ther dimensin examined (Tables 5 and 6). Wing length, males: the lngest-winged frm was murallae, which differed significantly frm all

18 TABLE 7. Measurements (mean, standard deviatin, sample siu) f nine samples f the Variable Seedeater cm. plex. Measurements f males and females cmbined except fr wing length. SAMPLE Billlength frm nstril Bill depth Wing length Wing length (a) (9) Tail length Tarsus length intermedia (nrthern) intermedia {cisandean) bgtensis (Magdalena) bgtensis (western) hicksii {nrthem hicksii (Valle del Cauca) phthalmica nurallae 7.25:tO.26 n= :tO.25 n= :tO.27 n= :tO.28 n= :tO.24 n= :tO.25 n= :tO.24 n= :tO.39 n= :tO.31 n=ll 8.27:t.O :t :t :t :t.O.39 n=23 n=20 n= 11 n=29 n= :t.O :t :t :t :t.O.40 n=43 n=23 n=22 n=44 n= :t.O :t :t :t :t.o.37 n=35 n=20 n= 17 n=36 n= :t.O :t :t :t :t.O.50 n=38 n=25 n= 14 n=37 n= :t.O :t :t :t :t.O.58 n=36 n=22 n=17 n=40 n= :t.O :t.O :t :t :t.O.56 n=26 n=18 n=ll n=23 n= :t.O :t.O :t.O :t :t.O.31 n=39 n=26 n=13 n=38 n= :t.O :t :t :t :t.O.72 n=33 n=21 n=12 n=32 n= :t.O :t.O :t :t.O.48 n=10 n=10 n=2 n=ll n=ll thers except the cisandean sample f intermedia. Nearly as lng-winged as the latter was americana; the shrtest-winged were the tw races f " aurita". Amng linear dimensins, wing length is usually cnsidered the best indicatr f verall size amng clsely related frms that d nt differ in migratry behavir. On this basis, the members f the Variable Seedeater cmplex tend t increase in size frm west t east, the main exceptin t this trend being the large size f murallae. Wing length, females: the rder f the samples was almst exactly the same as fr males, with nly tw adjacent samples (the suthern sample f hicksii and phthalmica) changing places; murallae was separated even mre clearly frm the remaining frms by its lng wing (Table 6). The mrphlgical differences detected in the preceding analysis can be placed in clearer prespective by cnsidering the ratis between mean measurements (Table 7). The rati f bill depth t bill length is especially ntable as it separates murallae cmpletely frm all the ther frms, even at the individuallevel (Fig. 7). This is the nly frm with a bill as lng r ln~r than it is deep; all thers have thick, stubby bills, and the deepest bill relative t its length is precisely that f americana. Cnsidering bill depth in relatin t wing length, again murallae (shallwest) and americana (deepest) are at the ppsite extremes f the distributin, with nly the value fr murallae utside the 95 % cnfidence interval f the mean (Table 7). The rati f bill length t wing length again separates murallae frm the remaining frms. Ratis f taillength t wing length and tarsus length t wing length yield patterns less easy t interpret. Relative t wing length, phthalmica had the lngest tail and tarsus; bth samples f intermedia averaged shrt-tailed whereas bgtensis (especially the western sample) was lng-tailed relative t the verall mean rati. The nrthern sample f hicksii averaged shrt-tailed, the suthern sample, lng-tailed (and thus apprached phthalmica in this respect). By cntrast, bth samples f hicksii had relatively lng tarsi, whereas all samples f

19 VARIATION IN THE VARIABLE SEEDEATER COMPLEX intermedia and bgtensis had tarsi f relatively shrt t mderate length. In bth f these ratis, murallae and americana differed rather strngly: the tail and tarsus f the frmer were shrt relative t the wing, while thse f the latter were lng, relative t the verall mean. Hwever, in neither f these ratis did any frm differ frm the rest as strngly as did murallae in bill shape (Table 7). The verall cnclusin t emerge frm this analysis is that americana, intermedia, and "aurita" are quite similar in verall prprtins, especially bill shape, and differ mainly in size (see abve). Tail length was the nly dimensin in which variatin appeared t be f a mre irregular, perhaps msaic nature. The ne frm that des nt fit the general pattern, especially in bill shape, is murallae: in 4 f the 5 cases this frm is at ne extreme f the distributin f ratis, differing significantly frm the rest in bill and wing length ( and in bth ratis invlving billlength). Of particular interest is the fact that the frm mqst different frm murallae in mst mrphlgical ratis is americana, precisely the mst similar in plumage. This strngly supprts the idea that these tw are much less clsely related than are the ther pairs (hicksiiphthalmica, intermedia-bgtensis) under cnsideratin, as was suggested abve by pssible patterns f plumage variatin within americana itself. Patterns f distributin and areas f verlap. I have pltted all Clmbian cllecting lcalities and reliable sight recrds available t me in Figs. 8 and 9, while Fig. 10 gives the verall ranges f all members f the Variable Seedeater cmplex. In the fllwing paragraphs I will attempt t clear up previus misunderstandings due t misidentificatins. S. «aurita" hicksii: this frm ccurs frm the Glf de Urabá regin (E t extreme W Córdba) and the Panama brder suth in the Pacific lwlands and fthills t S Valle del Cauca; the 0 a.m mu l.n b 6--hi p FIG. 7. Means and plygns enclsing all individuals measured fr billlength vs. bill depth, fr six taxa f the Variable Seedeater cmplex. Taxa are: am = americana; mu = murallae; in = intermedia; b = bgtensis; hi = hicksii; p = phthalmica. 93

20 STILES 94

21 VARIATION IN THE VARIABLE SEEDEATER COMPLE suthernmst r~rded lcality is Rí Raps, ca. 30 km S f Buenaventura. Meyer de Schauensee (1950, 1952) gave the Rí Dagua area (just suth f the latitude f Buenaventura) as the suthern limit f this frm, but Olsn (1981b) misinterpreted this t mean that the Dagua regin frmed "the bundary between chcana ( = hicksii) and phthalmica". Olsn assigned specimens frm Cisners, n the N side f the Rí Dagua, t phthalmica, apparently n the basis f their white thrats (which, by itself, is nt a reliable character -see Table 1) Als, Chapman (1917) had placed birds frm Cisners in phthalmica, but this was befre any ther " aurita" had been taken n this slpe; Meyer de Schauensee himself (1952) included the Cisners birds in chc~na ( = hicksit). I have examined 12 ther specimens taken within a few km f Cisners: all are hicksii. S. «aurita" phthalmica: This frm has been recrded frm the Pacific lwlands and fthills frm extreme sw Cauca (Guapi) S thrugh Nariñ and W Ecuadr t extreme NW Peru. When citing the Rí Dagua regin as the suthern limit f chcana, Meyer de Schauensee (1952) actually placed the nrthern limit f phthalmica in the Rí Patía regin in Nariñ, well S f Guapi. Between the clsest recrded lcalities f the tw frms (Rí Raps and Guapi) there is a gap f ca. 150 km; whetherthis is due t an absence f birds r simply a lack f cllecting remains t be determined. s. i. intermedia: The mst wide-ranging member f the cmplex, this frm ccurs in the nrthern lwlands f Clmbia (frm W C6rdba eastward), Venezuela, and prbably extreme W Guyana; west f the Andes it extends suth t the lwer Magdalena valley; eastwards it ccupies an extensive range in the Maracaib Basin and the Llans f Clmbia and Venezuela. The suthern limits f inte1media are apprximately the Sierra de la Macarena and the Rí Guaviare (Puert Inírida) in Clmbia; this frm extends lcally t ca. 100 km S f the Rí Orinc in 95

22 STILES Venezuela (Phelps & Phelps 1950, 1963). In additin, there are recent recrds f intermedia frm the Territry f Rraima in N Brazil, near the S edge f the Guyana Shield (Silva & Willis 1986, Silva & Oren 1990). This apparently islated, pssibly relict ppulatin (see belw) is the nly ne presently knwn in Brazilian territry. The ppulatin f Nrte de Santander in the W end f the Maracaib Basin shws the greatest resemblance in pattern t bgtensis, as nted abve; the latter has been taken at Las Ventanás near Ocaña n the E slpe f the Sierra de Perijá (Meyer de Schauensee 1952), and cntact betwren the tw races in this area is nt unlikely. It is als highly likely that these races are in cntact ver a brad zne in NW Antiquia and E Santander, given the extensive defrestatin in the lwer and middle Magdalena valley ver the last years. I knw f n recent cllectins frm this regin, hwever. This race is als resident n Trinidad (this ppu" latin is smetimes separated as the race insularis Gilliard, althugh mst recent treatments d nt recgnize this split), where it frmerly ccurred widely but is nwrare due t cage-bird trapping (ffrench 1973), and n the dry islands ffthe N cast f Venezuela (Phelps & Phelps 1950, 1963). S. i. bgtensi5: this race is basically cnfined t the middle and upper Cauca and Magdalena valleys, extending acrss the N end f the Crdillera Central in the deeply incised, warm valleys f the Medellín regin. It ccurs regularly t 1800 m r mre n the adjacent slpes, especially in drier, mre pen areas; this has dubtlessly permitted it t crss t the E side f the Sierra de Perijá via the lw (ca. 1500m) passes near Ocaña, and t reach the Pacific slpe f the Crdillera Occidental in the dry upper valleys f the Dagua and Patía rivers f SW Clmbia. In these valleys, bgtensis extends lcally dwn t belw 400 m, and it is at this lwer extreme that the hybrids with hicksii ("anchicayae") have been taken. Tw specimens frm beynd this range assigned t bgtensis by Meyer de Schauensee ( ), frm Malaguita near the muth f the Rí San Juan N f Buenaventura, and FIG. 10. Overall distributins f members f the Variable Seedeater cmplex in Central and Suth America; abbreviatins fr taxa as in Fig. 7.1 =.aurita" hybrid swarm, C Panama; 2 = area f.anchicayae"; 3 = Tierralta.? = uncertain, pssible zne f cntact. 96

23 VARIATION IN THE VARIABLE SEEDEATER COMP TABLE 8. Results f analyses f variance f measurements f samplesl f members f the variable Seedeater cm. plex, with results f Tukey a psteriri tests2. Measurement F, p Tukey test: mffins rdered frm largest t smallest Billlength frm nstril 49.12, P<O.00l 306 MU AM IN IC BM BW OP HN Hv -//L Bill depth 9.10, P< AM IC IN BM BW HN OP HV Ml Taillength 11.10, P< BW MU AM BM IC OP IN HV HN Taillength 2.83, P<O.OS 302 AM BM IC HN IN MU OP BW H\' Wing length, males 26.44, P<O.OOl 185 MU IC AM IN BW BM HN OP HV II Wing length, females , P<O.OOl 117 MU IC (AM) IN BW BM HN HV OP 1 = Samples are: HN = hicksii, nrthern sample; HV = hicksii, Valle del Cauca sample; OP = phthalmica; BW = bgtensis, ~stern sample; BM = bgtensis, Magdalena Valley sample; IN = intermedia, nrthern Caribbean lwlands sample, IC = intermedia, Cisandean sample; AM = americana, MU = murallae. 2 = Lines jin samples nt significantly different. Number f dashes indicates degree f significance separating samples: /// = P<.OOl; // = P<.Ol; n dash = P<.O5. 3 = americana nt included in analysis; psitin f mean in parenthesis. Quimarí in the muntains n the brder f C6rdba and Antiquia (NW f Tierralta), were later reidentified as s. schistacea (Meyer de Schauensee 1952). In additin, there is an immature 9 (ICN 23891) cllected at Buenaventura by C. J. Marinkelle that is labeled s. i. bgtensis, but frm plumage and measurements it is at least as likely t be hicksii (recall that immatures f these tw are very similar). Unfrtunately, the specimen is missing mst f the tail, and tail length is the mst useful measurement fr distinguishing these frms (cf. Tables 7 and 8). If, asi suspect, this specimen is indeed hicksii, then there is n recrd f bgtensis lwer than ca. 350 m n the Pacific slpe (where I bserved it at El Cauchal, Valle in May 1993). S. a (?) murallae: this frm ccurs in the upper Amazn regin in SE Clmbia, E Ecuadr, NE Per, and adjacent extreme W Brazil. A sight recrd frm "Cañ Duda" ( = Rí Duda?) n the W side f the Sierra de la Macarena ( cited in Hilty & Brwn 1986) is uncnfirmed; hwever, the riverine scrb alng the Rí Duda apparent- 1 y cnstitutes a crridr fr migratins f several species f Sprphila (M. Alvarez, pers. cmm.). S far as is knwn, murallae is strictly a lwland bird f riverine scrb, nt entering the Andean fthills at any pint. A suppsed recrd f this 07

24 STILES frm frm Satá, far t the nrth in the muntains f Byacá (Nicéfr & Olivares 1975), is errneus: I have examined this specimen, which is s. luctusa. S. americana: this frm inhabits the narrw strip f castallwlands and the adjacent lwer river valleys f NE Suth America, frm the Orinc delta f extreme E Venezuela t the muth f the Amazn in Pará, Brazil; and alng bth.banks f the Amazn, including the lwer parts f several majr tributaries, E as far as the area f Mana:us. The Amaznian ppulatins might be separable as the race dispar, althugh this pint remains cntrversial (Pint 1944, Meyer de Schauensee 1952) in the absence f a quantitative analysis. At least in Venezuela, americana is restricted t frest edge and secnd grwth adjacent t mangrve swamps (Meyer de Schauensee & Phelps 1979). The species is als resident n Tbag, ccurring in a greater variety f pen habitats (ffrench 1973), as is als the case in Guyana (Snyder 1966). There are tw ld recrds, apparently f strays, frm the N cast f Trinidad (Sclater 1871). A gap f ca. 500 km separates the westernmst ppulatins f americana near Manaus frm the nearest ppulatins f murallae f extreme W Brazil and NE Peru. The mst revealing cntact znes are thse where a race f intermedia ccurs with ne f the frms with black-and-white males. s. i. bgtensis meets hicksii alng the R Anchicayá ( and perhaps elsewhere) in SW Valle del Cauca, Clmbia, where at least three hybrids have been taken between 340 and 450 m. That ten years elapsed between Miller's cllecting f "anchicayae" and the Brrer-Catañ specimen suggests either that spradic hybridizatin ccurred ver a perid f years, r that a stable but very lcalized hybrid ppulatin exists. The unifrmity f these hybrids in plumage, particularly cmpared t the Tierralta series, suggests that all are F1 hybrids: there is n suggestin f backcrssing. Als, during tw weeks in this general regin in May 1993, I saw nly apparently "pure" hicksii and bgtensis. I therefre believe that hybridizatin between these frms is at best rare at present. Hwever, the hicksii-like thrat pattern shwn by mst bgtensis (but nt intermedia) might indicate a perid f frequent hybridizatin and intrgressin in the past, fllwing which islating mechanisrns have been nearly perfected. The very narrw zne f verlap may indicate different eclgical preferences (wetter areas fr hicksii, drier and mre pen areas fr bgtensis). These differences might be further reinfrced by eclgical incmpatibility as reflected in their extremely similar bilis (and presumably, seed preferences). T judge frm the range f phentypes present (Fig. 1), hybridizatin -including backcrssing -is much mre frequent between hicksii and nminate intermedia in the Tierralta regin. The tw frms have prbably cme int cntact in this area nly very recently (see abve), and in effect we might be seeing a recreatin f the initial cntact between hicksii and bgtensis in Valle. That apparently "pure" parental phentypes ccur in the Tierralta cntact zne suggests that hybridizatin is nt ttally free, hwever. Mre infrmatin frm this area, especially a quantitative study f O' phentypes and sng, wuld be highly interesting; unfrtunately, the area is at present very dangerus due t guerrilla activity. By cntrast, there is n evidence fr hybridizatin between intermedia and either murallae r americana. Althugh bth murallae and intermedia have bren reprted frm the Rí Duda W f the Sierra de la Macarena, it is unlikely that either breeds there, at least regularly (see abve), and n ther pssible cntact has been fund (r seems likely, given their apparently different eclgical preferences). On present knwledge, a gap f ca. 200 km separates the islated (?) ppulatin f intermedia in N-C Brazil frm the americana ppulatins alng the Amazn. The intervening area f Brazil has been prly cllected, hwever. Althugh americana and intermedia appear t verlap in extreme E Venezuela and adjacent Guyana, the tw have never been taken at precisely the same lcalities (M. Lentin, in litt.) and in fact may nt ccur sympatrically, at least fr breeding. Althugh intermedia has been fund in a variety f pen and secnd-grwth habitats in Venezuela, it seems nt t ccur in the specific habitat f americana (Phelps & Phelps 1963, Meyer de Schauensee & Phelps 1979). Of particular interest is the cmplete replacement between the tw n Trinidad and Tbag, suggestive f cmpetitive exclusin: their extremely similar verall mrphlgy, 98

25 VARIATI( INTHE especially their bills, might preclude cexistence -especially n islands (cf. Grant 1968). Frm the scanty data available, it appears that america. na and intermedia might be acting as parapatric species; clearly eclgical studies cmparing the situatin n Trinidad and Tbag with that in eastern Venezuela wuld be very valuable. Whether these species have ever hybridized is als mt: it is perhaps suggestive that the.suppsed distinguishing characteristic f insula. ris, the regular pssessin f a white neck-patch, recalls the situatin in bgtensis. Again, a quantitative analysis f plumage characteristics might prve helpful. The verall cnclusin frm this analysis f distributin is that the Variable Seedeater cmplex des indeed appear t represent a grup f very clsely related frms. There is almst perfect cmplementarity f distributins; the very narrw znes f verlap are characterized by hybridizatin (perhaps mre frequent where cntact has been mre recent) r, pssibly, parapatry in the case f americana and intermedia. The mst islated member f the cmplex, in distributin as well as mrphlgy, is murallae. BIOGEOGRAPHY OF THE VARIABLE SEEDEATER COMPLEX Viewing the verall distributins f all frms f the Variable Sredeater cmplex as a unit (Fig. 10), an interesting pattern emerges. In the central part f this distributin ccur the tw frms with gray male plumages (intermedia and bgtensis), while all f the frms with black-and-white O' plumage ("aurita", murallae, and americana) ccupy peripheral psitins. Mrever, all f the areas ccupied by the latter frrns are wetter (annual rainfall mm r mre) than thse inhabited by the frms with gray males, as might be expected frm Glger's Rule. Anther pint f interest is that n member f the cmplex ccupies the Guyana Shield, gelgically the ldest regin f nrtheastern Suth America. The area t the nrth f the Shield, n the castal slpe, is ccupied by americana, which cntinues arund its suthern periphery alqng the Amazn. Even the islated ppulatin f intermedia in Rraima, Brazil is at best n the very edge f the Shield (Silva & Oren 1990). Als, murallae has been recrded t the W and S f the Sierra de Chiribiquete f SE Clmbia, which gelgically pertains t the same frmatin as the Venezuelan tepuis (Estrada & Fuertes 1993), but nt in the Chiribiquete regin itself (Stiles et al. 1995). The areas ccupied byamericana alng the nrtheastern cast, and by" aurita" alng the Pacific cast, are als amng the gelgically yungest regins f the cntinent. Severallines f evidence, including turnver f fssil avifaunas (Brdkrb 1971, Vuilleumier 1984), mlecular data (Sibley & Ahlquist 1~90), and present distributin patterns (Haffer 1967, 1974, 1985) indicate that mst extant species f birds, especially scine passerines, riginated during the Pleistcene. Therefre, an understanding f the Quaternary histrical gegraphy f nrthern Suth America is essential fr develping a hypthesis fr the evlutin f the Variable Seedeater cmplex. By the end f the Plicene, the majr patterns f tpgraphic relief f nrthern Suth America were essentially mdern: the Andes and related ranges had nearly r quite attained their present elevatins, and a land cnnectin with Central America had been established. Fr lwland taxa, faunal exchange between the Amazn- Orinc basins and the Pacific slpe r with Central America was nw pssible nly via the lwlands nrth f the Andes r, t a lesser degree, via the Prculla Pass (presently ver 2000 m) in N Peru (Haffer 1967,1979). During mst f the Plicene, much f the Amazn basin was cvered by a large lake r epicntinental sea but by the Pleistcene, depsitin f huge amunts f sediments derived frm the nascent Andes had transfrmed this area int an alluvial plain dissected by large rivers, and pssibly with extensive wetlands (cf. Simpsn 1979, Haffer 1979, Fjeldsa 1985). The alternating glacial and interglacial perids f the Pleistcene undubtedly had a decisive influence upn dispersal and speciatin.in the Variable Seedeater cmplex. During glacial perids, cler and drier cnditins prbably prevailed ver mst f Suth America; lwered sea levels increased the lwland areas alng the casts. During the warmer, wetter interglacials, higher sea levels resulted in the submergence f mst f the Amazn valley, frming a lng gulf that extended west nearly t the Andes (Simpsn 99

26 STILES 1979, Fjeldsa 1985). Paleclimatic data frm the last glacial maximum (LGM) f ca. 18,000 years ag shw that arid cnditins prevailed in sme parts f the Arnazn basin, but that the currently arid lwlands f N Clmbia and N Venezuela had a mre mesic climate (Schubert 1988, Krnberg et al. 1991). The cler climates f the LGM lwered vegetatin znes by m and temperatures by 6-7 C (van der Hammen 1974). Whether r nt ne accepts in detail the scheme f frest refugia f Haffer (1974, 1985), it seems inevitable that during glacial perids savannas and ther pen vegetatin types expanded at the expense f clsed evergreen frest, with the reverse ccurring during the interglacials. Thus, fr nnfrest lwland birds like Sprphila, glacial perids prbably facilitated dispersal thrugh the lwlands but restricted crssing f muntain barriers. The interglacials prbably tended t fragment distributins f savanna taxa, and t the extent that muntain slpes became frested, may nt have facilitated dispersal f such taxa acrss muntain passes. Thus, it seems likely that mst r all dispersal ccurred thrugh the lwlands during glacial perids, thrughut the Pleistcene amng members f the Variable Seedeater cmplex. Based upn present-day patterns f species diversity (see Meyer de Schauensee 1952, Ouellet 1992), the rigin and early radiatin f the genus Sprphila prbably ccurred n and arund the Brazilian Shield. The present-day species f this regin mst similar in plumage and mrphlgy t the members f the Variable Seedeater cmplex is S. cllaris, a widespread, cmmn, and variable inhabitant f grassy marshes and wet savannas (Ridgely & Tudr 1989). This species was cnsidered t be "allied t" s. americana by Hellmayr (1938). Differentiatin f the Variable Seedeater cmplex prbably began when a thickbilled sprphiline related r ancestral t cllaris mved nrthward frm the Brazilian Shield and clnized the marshes r riverine habitats f the Amazn basin during an early glacial perid f the Pleistcene. During the subsequent interglacial, when the Amazn became a deep bay r gulf, this ppulatin became adapted t shreline habitats, pssibly assciated with mangrve as is americana tday; the expansin f frests t the suth islated this ppulatin frm the prt-cllaris f the shield. During this perid, this prt-americana ppulatin prbably ccupied apprpriate habitat all arund the gulf, thugh its expansin nrthward alng the Atlantic cast might nt have ccurred untillater, as the castal savannas f NE Suth America ccur n Pleistcene sediments (Sarmient 1983). Part f this ppulatin may have cntinued t ccupy wet savannas, particularly n the nrth side f the Amazn gulf. With the next (last?) glacial cycle and the expansin f these savannas this ppulatin spread nrthward TABLE 9. Ratis f mean measurements f members f the Variable Seedeater cmplex. Abreviatins fr samples as in Table 8. RATIOS IN IC BM BW HN HV OP MU AM x 95% Cnfidence Interval Bill depth/ bill length Bill length/ wing length Bill depth/ wing length Taillength/ wing length Tarsus length/ wing length

27 VARlATION IN THE VARIABLE SEEDEATER COMPLEX thrugh the present-day Llans and nt the nrthern castal plain, and thence int the Magdalena and Cauca Valleys and Central America, and dwn the Pacific slpe f Suth America. With the lwered sea levels, bth the land cnnectin t Central America and the castal plains were much brader than at present, facilitating this expansin. That this expansin culd have bren quite rapid, in gelgical terms, is suggested by the rapid range expansins f varius pen-cuntry birds in recent years as man has remved frest barriers (cf. Kiff 1975, Stiles & Skutch 1989 fr examples). It is als prbable that the range f intermedia n the N side f the Amazn was nce mre extensive than at present; the ppulatin f N-C Brazil, apparently islated frm ther ppulatins f the species by 300 km r mre, may represent a relict f a nce cntinuus distributin. The increasing humidity and expanding frests f the next (present?) interglacial fragmented this ppulatin, first islating the Central American and Pacific cast birds frm thse f the rest f Suth America. That segment ccupying the Llans and the Caribbean cast f present-day Clmbia and Venezuela, expsed t a drier and/r mre seasnal climate, evlved grey male plumage t becme the current species intermedia. In Central America, the ppulatins f the Pacific slpe became islated frm thse f the mre humid Caribbean slpe. Increasing humidity selected fr mre black pigmentatin in the latter ppulatin, prducing crvina; n drier Pacific slpe, the black-and-white male plumage was retained. Stilllater in this interglacial, frest barriers further fragmented the Pacific ppulatins, giving rise t the currently recgnized subspecies hffmannii, hicksii, and phthalmica. Gene flw between the Magdalena- Cauca and Caribbean slpe ppulatins f intermedia was prbably interrupted at least intermittently as well; the frmer develped mre patterned gray plumage, resulting in bgtensis. In any case, it seems likely that islatin between intermedia and bgt~sis was never as cmplete as between either f them and the Central American-Pacific slpe ppulatins. Gene flw between the intermedia and bgtensis might have ccurred fr a cmparatively lng time in the brad expanse f the lwer Magdalena Valley r via the lw passes near Ocaña in Nrte de Santander. The final stage f this prcess wuld then have been the establishment f secndary cntact between c1"vina and hicksii in central Panama, perhaps reflecting defrestatin in preclmbian r clnial times; the hybrid swarm in this area apparently has been stable fr well ver a century (Olsn 1981b), and there are sme histrical indicatins that when the Spaniards arrived, the Indiaqs maintained extensive savannas in this area by burning (Bennett 1968). Secndary cntact between hicksii and bgtensis is pssibly mre recent, the spread f the latter nt the Pacific slpe having been facilitated by lw passes in the Crdillera Occidental and dry, rain-shadw valleys. It is pssible that a perid f mre extensive hybridizatin and intrgressin ccurred in the past (which might help t accunt fr the mre patterned plumage f bgtensis relative t intermedia), fllwing which islating mechanisms between these tw frms have been nearly perfected. The mst recent cntact zne is that between intermedia and hicksii arund Tierralta, prbably dating frm n mre than 20 years ag; cntinued mnitring f the interactins f the tw frms in this zne will be critical t dcumenting future trends, and will help t clarify their status. The rigin f murallae was almst certainly quite independent f that f the intermedia- "aurita" clade. The mst likely sequence f events was fr a part f the riginal prtamericana ppulatin ccurring arund the head f the Amazn gulf during an early interglacial perid, t have becme islated as the sea level drpped during the fllwing glacial perid. This ppulatin may well have ccupied a humid regin just east f the Andes during this glacial perid, when it became adapted t wet riverine scrub. Certainly the present-day distributin f murallae crrespnds clsely t that f a species presumed t have differentiated in the Nap refugium f Haffer (1974, 1985). Hwever, given the degree f mrphlgical differentiatin f murallae, I suspect that it has remained islated in apprximately its present distributin fr cnsiderably lnger than the interval since the last glacial maximum. The nly way in which murallae culd frm a cnnecting link between amencana and "aurita" wuld be via 101

28 ;TILES dispersal f a prt-murallae stck t the Pacific slpe via the Nrth Peruvian Lw (Prculla Pass) and subsequent expansin nrthward n the Pacific slpe. Hwever, neither a detailed examinatin f plumage and mrphlgy nr the present-day eclgy f murallae supprt this alternative. In fact, had such a dispersal event ccurred, ne culd make at least as strng a case fr the end result being the Thick-billed Seedeater, s. peruviana, rather than "aurita". This hypthesis fr the evlutin f the Variable Seedeater cmplex is summarized graphically in Fig. 11. It seems plausible in that it accunts fr the results f analyses f plumage, mrphlgy, and distributin. Hwever, there are numerus pints f uncertainty and dubtless alternative scenaris culd be devised. Althugh the general sequence f events may be crrect, the timing is bviusly tentative; a lnger time span, invlving mre glacialinter-glacial cycles and perhaps extending back int the Plicene, might apply. Hwever, studies that suggest such a time frame (e. g., Capparella 1988, Bates & Zink 1994) deal mainly with subscine taxa rather than with the pssibly mre recent radiatin f the nine-primaried scmes. In cnclusin, evidence frm hybridizatin, mrphlgy, and distributin are cmpatible with the hypthesis that the enlarged Variable Seedeater cmplex represents a mnphyletic unit; this cnclusin is in turn cmpatible with a plausible histrical bigegraphic hypthesis. Genetic infrmatin will prbably be required 102

29 VARIATIC IN THE VARIABLE SEEDEATER COMPLE t test these hyptheses and perhaps t prvide mre satisfactry alternatives. Further analyses will 'als be required t determine the relatinship f this cmplex t such taxa as peruviana f Pacific Suth America and several Brazilian frms, especially cllaris, as well as trquela f N Central America. Perhaps the mst likely candidate fr inclusin in the Variable Seedeater (s.l.) clade is the similarly heavy-billed White-cllared Seedeater, s. trquela. Given that the currently arid Caribbeari littral f Clmbia and Venezuela apparently hada mre mesic climate during at least the last glacial maximum (Schubert 1988), it is pssible that americana nce ccurred mre widely here, and might even have reached Central America prir t the arrival f the "auri. ta" grup. Because trquela has wing-bars, its direct ancestry might lie with americana rather than with " aurita". U ntil very recently trquela was cnfined t Mexic and N Central America; it ccurs in sympatry with crvina in E Mexic and with hffinannii in W Csta Rica. Within the last 20 years trquela has expanded its range t bth slpes f W Panama, greatly increasing the area f sympatry with "aurita", but t my knwledge n hybrids between these frms have ever been recrded. Subsequently, intermedia might have replaced (r displaced) americana n the N Caribbean littral as this area became increasingly arid. The nly change required in the abve scheme t accmmdate trquela wuld be t place the arrival f americana n the N cast f Suth America smewhat earlier, with subsequent fragmentatin f its range by frest expansin in S Central America and arid cnditins n the Caribbean littral. Mlecular genetic data, as well as a thrugh study f its plumage and mrphlgy, will be required t clarify the affinities f trquela. TAXONOMIC CONCLUSIONS The majr cnclusins frm the abve analyses may be summarized as fllws: a) the Variable Seedeater cmplex may be regarded as a mnphyletic grup cmprising tw clades: americana (including dispar) and murallae; and intermedia, bgtensis, crvina, hicksii, phthalmica, and hff mannii; b) agustini and chcana are synnyms f bgtensis and hicksii, respectively; c) the species intermedia (including bgtensis) and "aurita" are mre clsely related t each ther than either is t americana r murallae; the latter des nt cnstitute the "cnnecting link" between americana and "aurita"; and d) the relatinship between americana and murallae is less clse than either that between intermedia and bgtensis, r that between the fur races f " aurita". It is pssible that trquela is als a member f the americana clade, but further study is required n this pint; fr lack f data, I d nt include it here. I feel that the mst apprpriate level at which t recgnize the Variable Seedeater cmplex as a taxnmic unit is the superspecies. Because f the limited, spradic nature f the hybridizatin between bgtensis and hick.sii, and the very lcalized zne f hybridizatin between the latter and intermedia, in which parental phentypes apparently ccur in appreciable frequencies ( cf. Shrt 1969), intermedia and "aurita" are apprpriately cnsidered allspecies. The apparent parapatry between americana and intermedia als suggests that they culd be cnsidered allspecies (Haffer 1992). I als cnclude that the mrphlgical distinctness and islated distributin f murallae indicate that this frm merits cnsideratin as an allspecies rather than as simply a subspecies f americana. I thus cnsider the Variable Seedeater cmplex t represent a single (americana) superspecies cmprising the allspecies americana (Wing-barred Seedeater), intermedia (Gray Seedeater), "aurita" (Variable Seedeater), and murallae. Tw questins f nmenclature remain: the crrect species name fr "aurita", and an apprpriate English vernacular name fr murallae. I accept the arguments f Olsn (1981b) regarding the invalidity f the name aurita; unfrtunately, he effectively sidestepped the questin f what t call " aurita" shuld this taxn be accrded ( all )species rank. The next ldest name fr a member f this grup, and the ldest that applies unequivcally t a recgnizable taxn, is crvina Sclater The Variable Seedeater in the strict sense shuld therefre be called Sprphila crvina, which wuld include the subspecies crvina, hffinannii, hick.sii, and phthalmica. I supprt Olsn's (1981b) recmmendatin that the name aurita be retained slely as an infrmal designatin fr the hybrid swarm f central Panama.~ 103

30 STILE In the past, murallae has been called "La Muralla Seedeater" (Chapman 1915) r "La Murelia Seedeater" (Hellmayr 1938), bth f which are garbled versins f the name f the type lcality f Mrelia in the Department f Caquetá, SE Clmbia. Because murallae has been cnsidered a subspecies f americana in all recent treatments, n mre apprpriate specieslevel name has been prpsed. Rather than.simply crrecting these ld names, I prpse a mre inclusive tpnym, Caquetá Seedeater. Mrelia is a rather bscure lcality in any case, and histrically the name "Caquetá" referred t mst f SE Clmbia, the wide area drained by the Rí Caquetá. The taxnmy and nmenclature prpsed here may be summarized as fllws: Sprphila americana Superspecies Sprphila murallae Chapman Caquetá Seedeater. Sprphila americana (Gmelin 1789). Wingbarred Seedeater. Included subspecies: americana, dispar Tdd 1922 (pssibly a synnym f americana) Sprphila intermedia Cabanis Gray Seedeater. Included subspecies: intermedia (synnym: insularis Gilliard 1946), bgtensis Gilliard 1946 (synnym: agustini Meyer de Schauensee 1947). Sprphila crvina (Sclater 1859). Variable Seedeater. Included subspecies: crvina (synnym: badiiventris Lawrence 1865), phthalmica Sclater 1860, h.ffmannii Cabanis 1861 (synnym: cllaris Lawrence 1865), hicksii Lawrence 1865 (synnym: chcana Meyer de Schauensee 1950). Sprphila "aurita" (Bnaparte 1850) t be used fr individuals with phentypes intermediate between S. c. crvina and S. c. hicksii frm C Panama. Synnyms: semicllaris Lawrence 1863 and frtipes (Lawrence 1865). I emphasize that, as with the bigegraphical scenari abve, this classificatin is presented as a hypthesis, which it wuld be mst desirable t test using mlecular genetic infrmatin. As nted abve, several ther taxa als shuld be investigated as pssible relatives r even members f the Variable Seedeater cmplex, especially tr. quela, which culd easily be accmmdated in the abve classificatin shuld the evidence S warrant. DISCUSSION The prcess f hybridizatin in the Variable Seedeater cmplex. This study was stimulated by the recgnitin f hybridizatin between the species crvina and intermedia. It is wrth explring hw hybridizatin culd ccur between tw species whse adult male plumages are s different (at least at first sight) that their clse relatinship had previusly gne unsuspected. Ptential islating mechanisms include size (intermedia being slightly larger), adult O' plumages and sng (given that the females prbably play the decisive rle in mate chice, the lesser differences in adult 9 plumages are prbably less significant in this regard). The sng f intermedia is a rather de1iberate, meldius carling f slurred whistles, trills, and chatters; that f hicksii is a mre hurried, jumbled medley f whistles, trills, and sputters (pers. bs.; see als ffrench 1973, Ridgely 1981, and Hilty & Brwn 1986). On the ther hand, such sng differences might in themselves be ineffective islating mechanisms in species like emberizine finches in which sng learning undubtedly ccurs. Mrever, the tw species are very similar in verall mrphlgy, especially bill shape, and their immature plumages are virtually indistinguishable. This might be especially significant as males (and presumably females) Sprphila are knwn t breed in this plumage (cf. Stiles & Skutch 1989). It des nt seem implausible that a yung male culd becme imprinted n the "wrng" sng in a cntact zne, then sing and attract a heterspecific mate befre attaining the definitive plumage. Members f the Variable Seedeater cmplex ften cexist with up t 5 ther species f Sprphila plus species f Tiaris, Vlatinia, and/r Oryzbrus, at least at sme times f year (Meyer de Schauensee & Phelps 1979, Ridgely 1981, HiltY & Brwn 1986, Stiles 1983, Stiles & Skutch 1989, Ouellet 1992). They nearly always have the thickest and/r mst bulbus bills f these assemblages. Bill shape culd therefre be imprtant in species recgnitin and mate chice, much as apparently ccurs in the related Galápags finches Gespiza (Ratcliffe & Grant 1983). Thus, the very similar bills f crvina and inter- 104

31 IARIATION IN THE VARIABLE SEEDEAT media culd lend themselves t errrs f identificatin, especially amng yung males in which the distinct clr differences f adults have nt yet becme manifest. Similarity in bill shape might als indicate similar seed preferences; depending upn the abundance f the mst preferred seeds, this culd facilitate cccurrence r cmpetitive exclusin. In new pastures and secnd grwth n recently defrested grund,.high residual sil fertility ften results in very high grass seed abundance fr a time; this culd prmte cexistence and hybridizatin fllwing frest remval, as at Tierralta. Were bill shape demnstrated t be imprtant in mate chice, this wuld cnstitute anther argument in favr f accrding (all)species status t murallae. It wuld be mst interesting t cnduct experiments n mate preferences amng members f the Variable Seedeater cmplex, similar t thse perfrmed with Galápags Finches by Ratcliffe & Grant (1983). Beynd the ~riable Seedeater cmplex. Numerus taxnmic prblems remain in the genus Sprphila, as reviewed in varying detail by Meyer de Schauensee (1952), Ridgely & Tudr (1989), and Ouellet (1992). Many f these include questins f the status f allpatric, highly lcalized, and/r prly knwn frms r pssible hybrids, much as in the Variable Seedeater cmplex. Many f these prblems stem frm the fact that the taxnmy f the genus is still based almst exclusively upn the adult O' plumages (and all t ften upn rather superficial examinatins theref) Detailed studies f mrphlgy, and f the plumages f bth sexes, have nt yet been undertaken fr mst f the prblematical frms. Hpefully, this paper will help t stimulate and rient such studies, which can perhaps mst apprpriately be carried ut by wrkers resident in the Netrpics and able t engage in judicius cllecting. In additin, mlecular and genetic studies will surely shed further light upn difficult prblems, r cases where mrphlgical and plumage data appear t be smewhat at variance (as in the case f murallae vs. americana). Mst r all species f Sprphila are assciated with pen habitats and feed largely upn grass seeds, their breeding seasns being largely timed t seeding episdes (Skutch 1950, Stiles & Skutch 1989). Hence, their distributins have undubtedly been affected by Pleistcene climatic fluctuatins, prbably inversely t thse f humid frest birds. T date, mst zgegraphic analyses f Netrpical birds have fcused upn the latter, leading t cncepts such as the frest refugia f Haffer (1974, 1985) and thers, and the imprtance f riverine barriers (Capparella 1988). N cmparably brad and detailed analysis f the nnfrest avifauna has yet appeared, and until data frm mrphlgy, genetics, palynlgy and gelgy can be applied t such a synthesis, ur understanding f the histrical zgegraphy f grups like Sprphila will remain speculative. Cmplicating factrs in such an analysis fr Sprphila in particular include the migratry behavir f many frms (briefly reviewed by Ridgely & Tudr 1989; see als Silva 1995), nmadic mvements f thers in respnse t spradic seeding events f grasses r bambs (e.g., S. schistacea in Csta Rica: Stiles & Skutch 1989), and recent changes as a result f frest clearance. The latter may bring int cntact frms islated since (presumably) the last glacial perid (e. g., hicksii and intermedia ), and the resulting interactins might clarify their status. In any case, the genus Sprphila wuld seem t represent a singularly apprpriate grup fr further taxnmic and bisystematic analysis. A final pint wrth cmment here is the prpsal f Olsn (1981a) t unite the genera Oryzbrus and Sprphila. He stated that the ranges in bill prprtins f the tw genera are cmparable, and that hybrids between the tw genera had been reprted frm Brazil. His Fig. 1 purprtedly demnstrated a cntinuum in bill shape between the tw genera. Hwever, I believe he verlked a significant difference in patterns f bill allmetry between the tw grups. The thickness f the bill is directly prprtinal t the curvature f the culmen in Sprphila but nt in Oryzbrus, in which culmens are at mst slightly curved. Thus, the species f Sprphila mst like Oryzbrus in bill thickness are the mst different in bill shape. I therefre favr cntinued recgnitin f tw genera, but agree with Olsn (1981a) and Sick (1963) that they are clsely related, as evidenced by their nests, eggs, plumage sequences, and sngs (Stiles & Skutch 1989); they shuld be placed cntiguusly in any linear sequence. 105

32 STILES ACKNOWLEDGMENTS Fr access t specimens in their care, I thank the fllwing: Hn. Rque Casallas, (Muse de la Universidad de La Salle), Hernand Chiriví and Jrge Mrales (UNIFEM, INDERENA); Dr. Humbert Alvarez-López (Universidad del Valle), and Dr. Herbert Schifter (Naturhistrisches Museum, Vienna). Manuel Marín A. and Miguel Lentin prvided lcality data and measurements f specimens in the cllctins f the Luisi~a State University Museum f Natural Histry (Batn Ruge) and the Clecci6n Ornit16gica Phelps (Caracas), respectively. Artur Rdríguez prvided curatrial assistance at the Institut de Ciencias Naturales; Ricard Jiménez supplied eclgical infrmatin n the Tierralta area. Fr unpublished lcality data, I thank Thmas McNish, Lreta Rsselli, and Paul Salaman. My visit t Vienna was made pssible by grants frm COLCIENCIAS (Bgtá) and BirdLife Internatinal fr attendance at the XXI Internatinal Ornithlgical Cngress, and thrugh the hspitality f Santiag and Elisabeth Gaviria. I thank J. M. Cards da Silva fr cnstructive cmments n the manuscript. REFERENCES American Ornithlgists' Unin Check-list f Nrth American Birds. 6th editin. American Ornithlgists' Unin, Washingtn, D. C. Baldwin, s. L., Wrley, L. G., & H. C. Oberhlser Measurements f birds. Sci. Publ. Cleveland Mus. Natl. Hist. 2: Bates, J. M., & R. M. Zink Evlutin int the Andes: mlecular evidence fr species relatinships in the genus Leptpgn. Auk 111: Bennett, C. F Human influences n the wgegraphy f Panama. Univ. Calif. Press, Iber-Americana 51: Brdkrb, P Origin and evlutin f birds. Pp in Farner, D. S., King, J. R., & K. C. Parkes (eds.). Avian bilgy. Vl. 1. New Yrk. Capparella, A. P; Genetic variatin in Netrpical birds: implicatins fr the spciatin prcess. Pp in. Ouellet, H. (ai.). Prc. XIX Intl. Ornithl. Cngr. Chapman, F. M Descriptins f apparently new Clmbian birds IV: Bull. Amer. Mus. Natl. Hist. 34: Chapman, F. M The distributin f bird-life in Clmbia: cntributin t an rnithlgical survey f Suth America. Bull. Amer. Mus. Natl. Hist, vl. 36. Cracraft, J Speciatin and ntlgy: the empirical cnsequences f alternative species cncepts fr understanding patterns and prcesses f speciatin. Pp in Otte, D., & J. A. Endler ( eds. ). Speciatin and its cnsequences. Sunderland. Estrada, J., & J. Fuertes Estudis btánics en la Guyana clmbiana v. Ntas sbre la vegetación y flra de la Sierra de Chiribiquete. Revta. Acad. Clmb. Cienc. 18: ffrench, R A guide t the birds f Trinidad and Tbag. Wynnewd. Fjeldsa, J Origin, evlutin, and status f the avifauna f Andean wetlands. Pp in Buckley, P. A., Fster, M S., Ridgely, R. S., Mrtn, E. S., & F. G. Buckley (eds.). Netrpical rnithlgy. Ornithl. Mngr., n.36. Gilliard, E. T Tw new gray seedeaters frm Suth America. Auk 63: Grant, P. R Bill size, bdy size, and the eclgical adaptatins f bird species t cmpetitive situatins n islands. Syst. Zl. 17: Haffer, J Zgegraphical ntes n the "nnfrest. lwland bird faunas f nrthwestern Suth America. Hrner 10: Haffer, J Avian speciatin in trpical Suth America. Publ. Nuttall Ornithl. Club, n.14. Haffer, J The quaternary bigegraphy f trpicallwland Suth Arnerica. Pp in Duellman, W. E. (ed.). The Suth American herpetfauna: its rigin, evlutin, and dispersal. Mus. Natl. Hist. Univ. Kansas, Mngr. 7. Haffer, J Avian zgegraphy f the Netrpical lwlands. Pp in Buckley, P. A., Fster, M. S., Ridgely, R. S., Mrtn, E. S., & F. G. Buckley (eds.). Nectrpical rnithlgy. Ornithl. Mngr., n.36. Haffer, J Parapatric species f birds. Bull. Brit. Ornithl. Cl. 112: Hellmayr, C. E Catalgue f birds f the Americas. Field Mus. Natl. Hist., Zl. Ser. 13, Part 11. Hilty, S. L., & W. L. Brwn A guide t the birds f Clmbia. Princetn. Kiff, L. F Ntes n suthwestern Csta Rican birds. Cndr 77: Krnberg, B. I., Benchiml, R. E., & M. I. Bird Gechemistry f the Acre subbasin sediments: windw n ice-age Amawnia. Interciencias 13: Meyer de Schauensee, R New r little-knwn Clmbian birds. Prc. Acad. Natl. Sci. Phila. 99: Meyer de Schauensee, R The birds f the Republic f Clmbia. Caldasia 22-26: Meyer de Schauensee, R Clmbian wlgical survey, part 7. A cllectin f birds frm Blivar, Clmbia. Prc. Acad. Natl. Sci. Phila. 102: