THE CONDOR

JOURNAL OF THE COOPER ORNITHOLOGICAL SOCIETY

Volume 84 Number 1 February 1982

Condor84:1-14 0 The CooperOrnithological Society 1982

SPECIATION IN THE CARBONATED FLOWER-PIERCER (DIGLO,% CARBONARIA) COMPLEX OF THE

GARY R. GRAVES ABSTRACT. -The evolution, differentiation, and present distribution of the Carbonated Flower-Piercer (Diglossa carbonaria) superspecieswere apparently greatly affected by Pleistocene glacial cycles. Populations of the six principal morphotypes are now largely allopatric, although secondary contact occurs in northern Peru (aterrima and brunneiventris)and northwest (brunneiventris and carbonaria). A narrow zone of hybridization between brunneiventris and carbonaria occurs northeast of La Paz, Bolivia. The presence of hybrid-like in- dividuals in allopatric populations from eastern Bolivia (carbonaria), (humeralis), and (gloriosa) may be due to past genetic contact and not to present introgression from nearby forms. The forms aterrima, humeralis, and nocticolor are closely related and conspecific (D. humeralis). The form gloriosa is a distinct allospecies(D. gloriosa). The Colombian population of brunneiventris is subspecificallydistinct from the nominate form. The forms brunneiventrisand carbonaria are closely related, but assuming the existence of a partial isolating mechanism between them, should be recognized as allospecies(D. brunneiventris and D. carbonaria). I present a model of speciation that accounts for the dis- junction of similar phenotypes.I proposethat differentiation in isolation occurred in a number of high Andean forest refuges during Pleistoceneglacials and inter- glacials, and that dispersal and genetic contact occurred with altitudinal changes in timberline forest at the beginningsand terminations of glacial periods.

The Andes Mountains of western South Amer- to northern Argentina. These primarily nec- ica are among the most biotically complex but tarivorous and insectivorous nine-primaried least studied highland tropical regions in the oscines reach their greatest diversity in the world. Several workers have examined the ef- Andes (Colombia south to Bolivia), where up fects of Pleistocene climatic shifts on the dif- to seven speciesoccur on an altitudinal gra- ferentiation of part of the Andean biota (e.g., dient. The high Andean Diglossa carbonaria Haffer 1967, 1969, 1974; F. Vuilleumier superspeciesinhabits “ceja” and timberline 1969a, 1970a, 1981a; B. Vuilleumier 1971; shrubland (1,500 to 4,000 m) from the Sierra Van der Hammen et al. 1973; Van der Ham- Nevada de Santa Marta in northern Colombia men 1974; Simpson 1975; Simpson and Haffer south through Ecuador and Peru to northern 1978; Vuilleumier and Simberloff 1980). Chile and Bolivia (Fig. 1). This polytypic group However, until recently, the paucity of distri- (Table 1) haspreviously attracted the attention butional and life history information pre- of researchers (Zimmer 1929, Vuilleumier vented detailed analysis of avian speciation in 1969b, unpubl. observ.) becauseit provides an the Andes. In the tradition of Chapman’s outstandingexample of speciation in progress. (19 17, 1926) pioneering works on Andean One of the most perplexing problems con- , F. Vuilleumier (1968, 1969b, 1970b, fronted by a systematist is divining the taxo- 197 l), Paynter (1972, 1978), Fitzpatrick nomic status of allopatric taxa (Mayr 1963). (1973), and Carpenter (1976) have now de- The D. carbonaria superspeciesis composed tailed specificpatterns of speciation in several of a confusing variety of largely allopatric groups of subtropical and temperate to pa- groups at and below the specific level that at ramo-puna zone birds. first glance seem to be distributed in a chaotic The flower-piercers (Diglossa) consist of manner (Fig. 1). Since the 1838 description of several groupsof speciesrestricted to the high- the Carbonated Flower-Piercer (D. carbo- land forest and shrubland of the neotropics naria, Lafresnaye and d’orbigny), the system- from Mexico south through Central America atic relationships of named taxa of Diglossa 111 2 GARY R. GRAVES

Zimmer (1929) was the first to recognizethe superspecificrelationship of the six forms of the carbonaria group, which he considered as conspecific on the basis of group characteris- tics. The subsequent classification by Hell- mayr (193 5) followed Zimmer in treating D. carbonaria as a polytypic species. However, in a recent monograph concerning the Diglossa, Vuilleumier (1969b) provided ad- ditional information about the development of isolating mechanisms and secondary con- tact in D. carbonaria. He considered it a su- brunneiventris perspecies with two semispecies, nomencla-

carbonaria turally treated as species: D. humeralis (humeralis, aterrima, nocticolor) and D. car- bonaria (carbonaria, brunneiventris and glo- riosa). In a recent reappraisal of the group, Vuilleumier (unpubl. observ.) considered glo- riosa an allospecies(along with D. humeralis and D. carbonaria). Past systematicstudies have dealt primarily with documenting morphological differentia- tion and location of barriers to gene flow. In this study of the D. carbonaria superspecies, I attempt to determine population structure of CHILE taxa and to correlate speciationwith presumed J Pleistocene climatic events. METHODS FIGURE 1. Present distribution (highly schematic) of the D. carbonariasuperspecies in the Andes Mountains My investigations of phenotypic variation in of western South America (redrawn from Vuilleumier the D. carbonaria superspeciesinvolved the 1969b). The distribution of brunneiventrison the western detailed examination of 1,O 10 museum spec- slope of the Andes between central Peru and northern Chile is based on LSUMZ specimensand personalobser- imens (134 carbonaria, 358 brunneiventris, vations. Colombian brunneiventrisare subspecificallydis- 279 aterrima, 80 humeralis, 127 nocticolor, tinct from the nominate form found in Peru and Bolivia and 32 gloriosa) together with field studies in (Graves 1980). the Peruvian Andes. I conducted field investigations during have been uncertain. On the basisof morpho- 1976-1978 in the Department of Cajamarca, logical similarities and coincidental distribu- northern Peru, as follows (actual field days): tion, Stresemann (in Zimmer 1929) grouped 2-5 September 1976 and 1 September 1977 certain members of the genus into “Formen- at Colmena, 20 km NE Llama, 2,675 m; 2-5 kreise.” The forms gloriosa and humeralis of September 1977 and 8-22 June 1978, near the D. carbonaria superspecieswere grouped, Cutervo, 2,600 m; 6-12 September 1977, NE respectively, with gloriosissimaand lafresnayii of Chota, 2,670 m; and 13-19 September of the D. lafresnayii superspecies (Glossy 1977, along a transect between Lajas and Cu- Flower-Piercer). tervo, 2,600 to 3,000 m.

TABLE 1. Color patterns of taxa in the D. carbonariasuperspecies.

nterrima noclicolor humeraiis carbonaria gloriosa brunneiventri~

Base of lower mandible Black Bluish gray Bluish gray Bluish gray Bluish gray Bluish gray Rump Black Gray Gray Gray Gray Gray Shoulders Black Black Gray Gray Gray Gray Crissum Black Black Black Rufous Rufous Rufous Lower breast and belly Black Black Black Gray Rufous Rufous Superciliary line Absent Absent Absent Absent Present Present Malar mark Absent Absent Absent Absent Absent Present * Colombian populations(vui/leumier~~ average smaller in wing, tail, and tarsusmeasurements than Peruvian specimens. SPECIATION OF FLOWER-PIERCERS 3

6fI” 67” 66’ SF”

BRUNNEIVENTRIS k +.. -16 ’ %..‘

-MO

FIGURE 2. Collection sites in Bolivia: (1) Caxata, (2) Iquico, (3) Palca, (4) Calacoto, (5) Tanapaca, (6) Chorros, (7) , (8) , (9) Incachaca,(10) , (11) Chapare, (12) , (13) Totora, (14) , (15) Comarapa. For location of the area shown here, see Figure 1. Stippled lines surround the approximate rangesof carbonariaand brunneiventris.Distances between sites where carbonariawas collected and the nearest populations of brunneiventris were measuredalong the lines shown in the figure. The latter are the geographicallines of least resistanceto dispersal above 2,500 m. The long central axis approximates the continental divide. The numbers that appear along the edges of this and other figuresare degreesof south or north latitude and west longitude.

Additional specimens and behavioral ob- In the text, all taxa are referred to by their servations of brunneiventrisand aterrima in subspecific epithet (nocticolor, humeralis, Peru were obtained during 7-18 August 1976 aterrima, brunneiventris,gloriosa, carbo- at 33 km NE Ingenio, 2,225 m, Department naria).The terms “hybrid” and “hybrid-like” of Amazonas; 2-3 July 1976, Upper Rimac are used to denote birds and character states Valley, 3,050 m, Department of Lima; 4-8 thought to resemble the potential results of August 1977, Cordillera Carpish, 2,600 to introgression or hybridization between the 3,200 m, Department of Huanuco; 12-2 1 Oc- above taxa. tober 1977 and 4-15 July 1978, Cerro Chin- guela, 2,500 to 3,000 m, Departments of Piura RESULTS AND DISCUSSION and Cajamarca. PHENOTYPIC VARIATION OF D. CARBONARZA IN My analysis of the populations for possible BOLIVIA introgression is based primarily on selected Published accounts of contact between brun- characteristicsof the plumage: presenceor ab- neiventrisand carbonariain northwest Bolivia senceof a pale superciliary line, humeral patch, indicate that possible isolating mechanisms, and rufous malar mark, and color of the flanks, acquired before secondary contact was estab- wing coverts, rump, breast, crissum and lower lished between the two taxa, were insufficient mandible. I did not use other plumage char- to prevent gene flow. Specimens showing in- acteristics (e.g., size of the black throat patch termediate charactersand describedas hybrids in brunneiventrisand intensity of feather pig- from the Department of La Paz have been mentation) becausethey were too variable due known sincethe 1800’s (Zimmer 1929). Based to fading and wear. Also, I did not usejuveniles on collections of brunneiventrisand carbo- (becauseof their small size) or adults with un- naria phenotypes during 1967-1968, in the molted immature or heavily worn plumage on Department of La Paz, Vuilleumier (1969b, the wings and tail in this study. In addition, unpubl. observ.) concluded that isolating I compared populations with respect to wing mechanisms between the two taxa were in- length (chord), tail length (from point of in- complete and that a narrow zone of hybrid- sertion of central rectrices to tip of the longest ization existed between them. feather), tarsus length, and culmen length Through the courtesy of F. Vuilleumier, I (from the anterior edge of the nostril). I com- examined the specimensthat he collected in pared certain population samples statistically this zone of contact. In addition, I assembled using student’s two-tailed t-test. a representative series of carbonariapheno- 4 GARY R. GRAVES

TABLE 2. Character index to score carbonariaphenotypes.

Character Rufous feathers on M&I score breast and abdomen mark

None None None Feathers with rufous edgingson Trace of rufous on feather Trace only lower abdomen edgesor gray feathers in malar areas Feathers with rufous edgingson Several rufous feathers on 2to5mm abdomen and lower breast both sides Feathers with rufous edgings Extensive rufous malar >6 mm extensively distributed over marks on both sides abdomen and breast types from the Departments of not been recorded east of the Rio de La Paz, and Santa Cruz, the principal range of the I found specimens of carbonariawith “hy- taxon (Fig. 2). Then I used a modified “hybrid brid” characters from near the zone of sec- character index” (Anderson 1949) to test for ondary contact in the Department of La Paz possible introgression of brunneiventrischar- eastward to near the limit of the speciesdis- acters in carbonariaphenotypes. Three char- tribution in the Department of Santa Cruz. If actersoften usedfor determining introgression introgression is responsible for these hybrid were scored numerically with possible values characteristics, one would assume that their ranging from 0 (for typical carbonaria)to 3 frequency and intensity would be greatestnear (typified by a hypothetical intermediate be- the zone of contact, and lower elsewhere.From tween brunneiventrisand carbonaria;Table 2). the specimensat hand, however, the frequency Thus a “typical” carbonariawould score 0, and intensity of these characteristicsare high- whereas a hybrid exhibiting (a) extensive ru- est in samples from eastern populations far fous malar marks (score 3), (b) a long super- removed from possiblesources of genetic con- ciliary line (score 3), and (c) extensive rufous tact with brunneiventris(Fig. 3, Table 3). Sim- feather edgings on the upper breast (score 3) plifying the data by deleting specimens col- would have a total score of 9. It should be lected from dry intermontane valleys (e.g., noted that a pure brunneiventris(see Table 1) Calacoto and Aiquile) does not significantly would also score 9. I scored these character- changethe relationship (Fig. 3). This suggests istics twice, at intervals of several days, to in- that factors other than introgression (e.g., sure precision. pleiotropic and polygenic effects) are respon- Although brunneiventrisphenotypes have sible for maintaining polymorphism in eastern Bolivian populations of carbonaria.Again, consideringonly these specimenscollected on the Amazonian side of the continental divide, TABLE 3. Character index scoresfor carbonariain Bo- the average character index values increase livia. clinally from northwest to southeastalong the axis of the divide. Specimenscollected in wet Estimated AWXige Individuals distance areas have a higher average character index character with to nearest index “hybrid-like” bmnnerventris value than birds from drier locales (1.43 vs. (maximum characters population0 0.86). However, there is no significant differ- Location n score = 9) (% of sample) (km) ence between the two groups in frequency of Caxata 9 0.33 33 “hybrid” character expression (0.49 wet vs. Iquico I 0.71 51 : 0.50 dry). A negative regression relationship Palcaa 4 0.25 25 25 Calacotoa 1 0.00 0 25 would be expected if active introgression of Tanapaca” 1 3.00 100 45 brunneiventrisgenes was occurring in carbo- Chorros 7 0.57 14 135 naria. Parotani” 1 1.00 100 215 The wet, humid conditions in temperate for- Colomi 2.00 100 227 ests of central and eastern Bolivia (e.g., Inca- Incachaca 3: 1.89 50 232 Tiraque 4 2.25 75 240 chaca) may favor certain coadaptedgene com- Chapare 5 0.80 60 250 binations phenotypically expressedas “hybrid Poconaa 11 0.64 55 293 characters.” If “hybrid-like” carbonariaphe- Totora” 4.00 100 314 notypes are the result of coadaptation to wet Aiquile” : 1.00 33 342 Comarapa 9 1.89 61 360 environments, then one would expect hybrid characterindices in populations to changeover * Dry localities. h See Figure 2. a moisture gradient (i.e., from arid interior SPECIATION OF FLOWER-PIERCERS 5

01

01

km from nearest BRUNNEIVENTRIS population

FIGURE 3. Relationships between character index scores of populations of carbonaria and the distance between them and the nearest population of brunneiventris. Sample sizes are in circles. The dotted regression line (y = 0.600 + 0.004x, r2 = 0.05) describes the relationship for carbonaria from all localities. The solid line (y = 0.520 + 0.005x, r2 = 0.07) describes the relationship for specimens from “wet” localities (see Table 3). The coefficient of determination of each line is statistically significantat P < 0.05. valleys to the wet Amazonian slopes).Unfor- Rock Earthcreeper, Upucerthiaandaecola; and tunately, climatological data for the eastern Brown-backed Mockingbird, Mimus dorsalis). slope of the Bolivian Andes are not available. The southeastern distribution of brunnei- Thus, precisecorrelations between rainfall and ventrisis limited by high mountains along the phenotype cannot be examined. continental divide and by the drier valley of Endler (1977) maintained that populations the Rio de La Paz, which is inhabited by car- can differentiate under a strong selection gra- bonaria. It is perhaps not coincidental that dient despite extensive gene flow. The possi- hybrids between brunneiventrisand carbonar- bility that brunneiventrisand carbonariaare ia occur exclusively in or near a zone of eco- exchanging genes, but retaining their pheno- logical transition between relatively wet un- typic identity due to differential selection is disturbed forests on north-facing slopesof the questionable. In other parts of their ranges, Cordillera Hampaturi and drier, disturbed carbonariaand brunneiventrisinhabit a vari- habitat covering the southeasternfacing slopes ety of ecotypes from dry chaparral to exceed- overlooking the Rio de La Paz. The narrow ingly wet pajonal and yet show no such gra- zone between Chojlla (La Emramada) and dient in phenotypic variability. Iquico where F. Vuilleumier (unpubl. observ.) Birds with D. carbonariaphenotypes are collected hybrids of brunneiventrisand car- regularly seen near and below the city of La bonariais no more than 10 km in width. Phe- Paz and in the dry shrubland of the foothills notypically “pure” populations of brunneiven- southwestof the continental divide (T. Parker tris are found 30 km northwest of Chojlla and F. Vuilleumier, pers. comm.; Fig. 4). They at Hichuloma. Of the nine adult specimens also penetrate the high mountain passesof the that resemble brunneiventrisand were col- Cordillera Hampaturi between Nevado lected at Chojlla by Vuilleumier, four are Huayna Potosi and Nevado Illimani, where clearly hybrids between brunneiventrisand they may contact brunneiventris.The north- carbonaria.These hybrids resemble brunnei- western limits of carbonarianear La Paz co- ventriswith much enlarged, diffusely edged incide with the northwestern limits of the Bo- throat patches and reduced rufous malar livian dry intermontane avifauna (including marks. The other five specimensare indistin- Gray-hooded Parakeet, Bolborhynchusay- guishablefrom those in “pure” populations of mara;Red-tailed Comet, Sapphosparganura; brunneiventris.No carbonariaparental types 6 GARY R. GRAVES

In the zone of contact, the differences in mensural characters between carbonaria, brunneiventris,and the Chojlla hybrid popu- lation are not significant (Table 4). This fact and the observed limits of the ranges of the two taxa suggestthat competition prevents geographic overlap. Further study may show that although the breeding ranges of the phe-

LA notypes do not overlap, some non-breeding individuals occasionally stray into the range of their allopatric congener. The zone of sec- ondary contact thus appears to be parapatric with little or no overlap of the parental phe- notypes. Vuilleumier (unpubl. observ.) sug- geststhat former barriers separating the taxa may have been modified by man, permitting 17O , 25 km I secondary contact. As a hypothetical alternative to present day introgression,the increasedfrequency of brun- neiventrischaracters in eastern populations of FIGURE 4. Zone of secondary contact between brun- neiventris(stippled area) and carbonaria (hatched area) carbonaria may be an “echo” of a proto-car- northeast of La Paz, Bolivia. Collecting localities: (1) bonaria genome delineated by a Cochabamba Caxata, (2) Hichuloma, (3) Chojlla, (4) Iquico. Contour Pleistocene refuge. An echo is defined as the lines approximate 4,500 m. For location of this area, see phenotypic expressionof an allelic remnant of Figure 1. Specimenshave not been collected from areas a genetically swamped population. During the designatedby question marks. pleniglacial period, a brunneiventris-likepop- ulation of proto-carbonaria from an assumed have been collected at Chojlla, nor have they Amazonia-facing refuge in Cochabamba may been seen there by F. Vuilleumier (pers. have been separated from a population re- comm.). Another hybrid, perhapsthe most in- stricted to the drier intermontane valley of the termediate specimen available (in the collec- upper Rio Grande. In view of the wide eco- tion of the American Museum of Natural His- logical tolerance of present day carbonaria, it tory; AMNH 793265), was collected 3 km is plausible that proto-carbonaria stock sur- from Iquico on the road to Mina Bolsa Negra. vived in both a “wet” and “dry” refuge. The The seven specimens collected at Iquico are “two-refuge” hypothesisis partially supported clearly carbonaria phenotypes. Vuilleumier by the concordance in ranges of a number of (pers. comm.) describesthe Iquico population differentiated taxa that are restricted to dry as “essentially pure carbonaria . . . although montane steppe and forest edge (i.e., the with brunneiventrisgenes. ” However, some of Brown-capped Tit-Spinetail, Leptasthenura the charactersthat he attributed to hybridiza- fuliginiceps; Streak-fronted Thornbird, Pha- tion or introgression(e.g., a rufous-edgedma- cellodomus striaticeps; and Creamy-breasted lar feather) may be natural variations in the Canastero, Asthenes dorbygni) and the avi- phenotype of carbonaria as in eastern Bolivia. fauna in the wet temperate forest of the Am- Away from the zone of contact, there are few azonian slope, typified by the Rufous-faced indisputable signs of introgression between Antpitta (Grallaria erythrotis) and the Scimi- carbonaria and brunneiventris. tar-winged Piha (Chirocylla uropygialis).These

TABLE 4. Measurements of brunneiventrisand carbonariafrom Bolivia.

Mean (mm) t SE (II) of males LOC&= Wing Tail Tarsus Cldmell brunneiventris A 71.33 * 0.70 61.25 + 0.51 21.17 f 0.21 7.70 f 0.10 (9) (10) (10) (11) brunneiventrisand B 71.04 f 0.60 60.44 + 0.52 21.27 + 0.21 7.90 f 0.11 brunneiventrisx carbonaria (5) (7) (7) (7) carbonaria C 70.62 I!I 0.81 60.56 + 0.90 20.80 + 0.29 8.07 k 0.28 (5) (5) (5) (3) carbonaria D 72.21 + 0.53 62.16 + 0.76 20.51 + 0.32 7.65 f 0.10 (7) (8) (8) (8)

I‘ A = Hichuloma, Pongo, Unduavi; B = Chojlla (la Ennamada) and 3 km NW Iquico; C = Iquico; D = Caxata. SPECIATION OF FLOWER-PIERCERS 7 wet and dry avifaunas probably differentiated in situ during the Pleistoceneglaciations. The “dry” refuge population, derived and isolated from brunneiventris-likeproto-carbonaria, lost part or all of the “ornate” characters of the ancestral stock (e.g., the rufous malar mark), and after the disappearanceof barriers, genet- ically swamped populations of carbonariain the “wet” Cochabamba refuge. Under these conditions, the present-day center of pheno- typic variation in Cochabamba representsthe remnant of former contact between two proto- carbonariagenomes now in equilibrium. Selection between these alternatives is not possible without additional specimens from localities along climatic and geographic gra- dients. AREA ENLARGED

3325m THE ZONE OF SECONDARY CONTACT cl* BETWEEN BRUNNEIVENTRIS AND 5. ATERRZMA IN NORTHERN PERU The existence of a zone of secondary contact between aterrimaand brunneiventrisin north- em Peru has been suspectedsince the 1880’s when these two taxa were reported near Cu- 4km tervo, Department of Cajamarca (Zimmer c 1 1929). Evidence of sympatry in the two taxa FIGURE 5. Region of secondarycontact between brun- was obtained at Chira (near Tambillo), when neiventrisand aterrima in the Department of Cajamarca, two specimens of aterrima and one of brun- northern Peru. Collection sites are indicated by closed neiventriswere collected within a day of one (aterrima) and open (brunneiventris)circles. The contour another (see Bond 1955). However, the exact lines enclose areas above 2,600 m. For location of this locations of “Chira” and “Tambillo” are un- region, see Figure 1. known (Vaurie 1972) becausethere are several geographic features and communities within 7 km south-southsoutheast of Cutervo. (Con- 20 km of Cutervo bearing these names. In the tact between breeding populations of brun- past, the distribution of aterrima and brun- neiventrisand aterrimais less likely to occur neiventrisin this zone of contact has been doc- along the ridge south of the Rio Chotano be- umented primarily along roads (aterrima:be- tween Huambos and Lajas becauseof low-el- tween Huambos, Cochabamba, and Cutervo; evation gaps [less than 2,400 m south of brunneiventris:between Chota, Conchan, and Cochabamba, Department of Cajamarca].) Tacabamba). Vuilleumier (unpubl. observ.) Except for a few hedgerows and plantings of recently narrowed the probable zone of contact Eucalyptusand Alnus,the terrain surrounding between the two forms to the 30 km that sep- Cutervo is heavily cultivated. Consequently, arate the Huambos-Cutervo axis from the habitat preferred by both phenotypes of Di- Chugur-Chota area. He (1969b, unpubl. ob- glossais patchy. serv.) concluded from available data and from Near the North Peruvian Low (the lowest field work in this zone of secondary contact pass in the Andes between Venezuela and during 1975 that aterrimaand brunneiventris Chile), Carbonated Flower-Piercers inhabit are reproductively isolated. shrubland and forest edge from about 2,400 During field studiesof D. carbonaria,I and m to timberline at 3,500 m. South of Cutervo other personnel from Louisiana State Univer- this corridor of shrubland narrows to 3 km in sity Museum of Zoology (LSUMZ) collected width, thus bottlenecking the dispersalof ater- 33 specimensof brunneiventrisand 29 of ater- rima and brunneiventris.The relatively dry, rima from the zone of contact between Cu- steep Quebrada Cachulca (upper Rio de tervo and Chota (Fig. 5). Our field work sug- Sucse),intrudes westwardinto this narrow cor- gests that the ranges of brunneiventrisand ridor, further restricting dispersal and gene aterrimaoverlap slightly (parapatry), but that flow. In other words, contact between aterrima significant hybridization does not occur. and brunneiventrisis retarded by a combina- My field work in 1976-1978 indicated that tion of factors including (1) the patchinessof the principal area of contact occurs from 3 to the habitat, (2) a partial geographic barrier, 8 GARY R. GRAVES and (3) the narrownessof the shrub zone link- similarities, this variation cannot be directly ing them together. attributed to introgression.) I made a concerted effort to collect brun- The mensural characters of brunneiventris neiventrisphenotypes encountered within the and aterrimafrom the zone of contact are not rangeof aterrimaand vice versa. Of four brun- significantlydifferent ([A and B] vs. C in Table neiventriscollected (six seen) near Cutervo, 5). Such similarity in their morphological two were adult females (LSUMZ 89 184 and characters,as well as in their foraging behav- 89186) apparently mated to aterrima males. ior, suggeststhat competition prevents geo- In both of the suspected“ interspecific” pair- graphic overlap. However, non-breeding in- ings, I collected the female brunneiventrison dividuals of both phenotypes apparently the territory of a singing male aterrima with wander into the range of their allopatric con- enlargedtestes. One specimen(LSUMZ 89 186) gener. Two specimens of aterrima (LSUMZ had a highly vascularized brood patch and the 85424: adult male, 11.5 km N, 2 km W Lajas, other (LSUMZ 89184) had slightly enlarged 2,800 m; LSUMZ 85416: subadult female, 7 ova (2 mm). (I saw no other brunneiventrisin km N, 3 km E Chota, 2,670 m) were collected the area, and intruding aterrima individuals 6 and 21 km, respectively, southeast of the from adjacent territories were vigorously nearest breeding population of aterrima. chased.) Two days later, I also collected ater- That aterrimaand brunneiventrisinterbreed rima from areas adjacent to these singing occasionally is not surprisinggiven their eco- males and the territorial singingmales as well. logical circumstances. South of Cutervo the None of them shows any signs of hybridiza- conditions for breakdown of isolating mech- tion. Conclusive evidence of successfulmating anisms appear to be ideal-disturbed and is lacking, however, since I did not find nests patchy habitat, and small breeding popula- or young on these territories. I conclude that tions of the two taxa. Important in this dis- brunneiventrisand aterrima may pair if po- cussion is the apparent absenceof hybrids in tential mates of the appropriate phenotype are the zone of phenotypic overlap. Long-term in short supply or not available. alteration of habitat in the zone of contact Such “pairings” may indicate that male (Cutervo was founded in 1560) has had an aterrima tolerate female brunneiventrisbe- unknown effect on the isolating mechanisms cause of interspecific ambivalence. However, of the two taxa. Whether secondary contact my observations of intense inter-phenotypic has strengthened or eroded specific distinct- aggressiondo not support this suggestion.In nesscan only be determined by further study. the zone of contact, aterrima and brunnei- ventrisappear to occupy identical niches. I POPULATIONS IN THE NORTHERN ANDES discernedno differences in the feeding behav-, Of the seven recognized taxa of the D. car- ior of the two taxa at clumps of flowering para- bonariasuperspecies, five are found within a sitic Tristerix(Loranthaceae) or in their inter- 500-km circle centered in northeastern Co- actions with hummingbirds. Furthermore, lombia (Fig. 6). The black populations, ranging territorial male aterrima actively displaced from the Santa Marta Mountains in northern non-territorial brunneiventrisfrom Tristerix Colombia south to northern Peru, occupy a clumps and chasedthem to the borders of the central geographicposition in the array of ru- territory. fous or gray-bellied polytypic forms. A careful examination of the 26 specimens Since its discovery in the early 1900’s, the of aterrima and four brunneiventriscollected Colombian populations of brunneiventrishave immediately south of Cutervo revealed no ob- complicated interpretations of the origin and vious signs of hybridization (comparable to relationships of the superspecies.Occurring brunneiventrisX carbonaria hybrids in Bo- some 1,500 km north of the principal range livia). In contrast to known hybrids of D. car- of brunneiventrisin Peru, the Colombian pop- bonaria,no aterrimaexhibited rufous feather ulations are apparently isolated in the northern edgings on the malar areas, breast, belly, or ends of the Western and Central Cordilleras. crissum. Furthermore, the pale superciliary After a careful examination of extant speci- line, which is conspicuousin most specimens mens, Zimmer (1929:26) stated“ I am not able, of brunneiventris,was absent in all of the spec- therefore, to separate the Colombian and Pe- imens of aterrimathat I examined, including ruvian birds [brunneiventris]even subspecifi- those from the zone of secondary contact. (A tally except on the sole ground of geographic few female aterrimahave gray rumps and un- isolation, which is not adequate for racial dis- dertail coverts and resemble the most mela- tinction.” Vuilleumier (unpubl. observ.) ten- nistic examples of nocticolorfrom the Santa tatively treats the Colombian brunneiventris Marta Mountains of northern Colombia. as a member of a polytypic subspeciesof the However, in the absenceof other phenotypic “carbonaria allospecies” on morphological SPECIATION OF FLOWER-PIERCERS 9 grounds, pending additional information on the evolutionary relationships of the group. I examined 40 specimensof the Colombian brunneiventris from both sides of the Rio Cauca (i.e., from P&ram0 Paramillo, Paramo Frontino, and Hacienda Zulaiba), and found no phenotypic color characteristicsthat were 8QN consistently different from those of Peruvian brunneiventris. Although Zimmer (1929:25) stated that “the size of the Colombian speci- mens falls well within the range of variation of my Peruvian series,” a mensural compari- 4% son of specimens from Colombia and the Chota-Cutervo region of Peru does reveal sig- nificant differences in some characters([D and COLOMBIA E] vs. [A and B] in Table 5). Several individ- uals superficially resemble hybrids of brun- Oo- neiventrisand carbonaria from Bolivia in hav- ing large diffusely-edged throat patches or ‘, ?50 km. disjunct malar stripes. However, many speci- mens of brunneiventrisfrom Colombia cannot be visually discriminated from Peruvian spec- FIGURE 6. Current distribution of north Andean taxa imens. In other words, the small populations of Diglossa:A = nocticolor,B = brunneiventris(D. b. vuil- of brunneiventris from Colombia appear to leumien], C = gloriosa,D = humeralis,E = aterrima. The dashed lines denote elevations above 2,000 m (redrawn exhibit as much phenotypic variation as the from Vuilleumier 1970a). widespread Peruvian populations. Whether this variability is due to introgression from nearby populations of aterrima or intrinsic genotypic variation is not known. In any case, 436793, with “juvenal” primaries) from Pa- the Colombian form warrants recognition at ramo Sonson, Department of Antioquia, Co- the subspecificlevel (Graves 1980). lombia, has gray-tipped lesser wing coverts Populations of the all-black form, aterrima, and a blackish-gray rump contrasting with a occur south of the headwatersof the Rio Porte black lower back. Zimmer (1929) noted that on the Cerro de 10s Parados. The depression several subadult specimensof aterrima from in the Andes (ca. 2,000 m) south of Medellin Laguneta, Department of Caldas, had gray- probably prevents extensive contact between edged coverts. This may represent a trend in brunneiventris and aterrima, but an over- coloration toward the gray-shoulderedhumer- looked population of either form may exist on alis. Unfortunately, a comprehensive seriesof either side of this gap. adult aterrima from the northern end of the A specimen of aterrima (in the collection of Central Cordillera is not available. the United States National Museum; USNM In the Eastern Cordillera, humeralis (a black

TABLE 5. Measurementsof male brunneiventrisand aterrima.

Mean (mm) i SE (n) Locale” WI%? Tail TaNU Culmen

brunneiventris A 68.14 + 0.55 58.44 f 0.88 21.14 k 0.18 8.08 + 0.15 (8) (7) (8) (8) brunneiventris B 68.19 k 0.66 58.90 f 0.90 21.39 k 0.14 8.29 k 0.10 (8) (7) (8) (8) aterrima C 69.61 Y!I 0.61 60.63 + 0.68 21.16 -t 0.21 8.20 k 0.24 (7) (7) (7) (6) brunneiventris D 66.56 -t 0.62 57.88 k 0.73 20.48 + 0.16 7.88 k 0.11 (11) (12) (12) (12) brunneiventris E 66.60 zk 0.51 57.93 * 0.59 20.12 f 0.20 7.58 f 0.26 (6) (6) (6) (5) Pb 0.02 NS 0.001 0.01 -A = Cutervo-Lajastransect; B = 7 km N, 3 km E Chota; C = Cutervo; D = Paramillo-Frontmo;E = HaciendaZulaiba. bMeasurements of A versusB and D versusE are not significantlydifferent (P > 0.05). The pooledvalues (A + B vs. D + E) werecompared using a two- tailed Student’s“t” test: the P values in the table are for thesecomparisons. Pooled measurementsof brunnerventris(A + B) are not significantlydifferent (P > 0.05) from the measurementsof aterrima (C). 10 GARY R. GRAVES race with gray rump and shoulders) ranges specimensof nocticolorresemble those of hu- from south of Bogota north to the Department meralisin having gray tips on the upper wing of Norte de Santander. The gap between the coverts, thus bridging the difference between humeralispopulations shown in Figure 1 is typical nocticolorand humeralis.I examined probably due to the lack of collecting, because 105 adult nocticolorand found that the per- the Eastern Cordillera from Bogota northward centage of specimenshaving gray feathers in to Venezuela is generally above 3,500 m in the humeral area was lowest in the north- elevation. South of Bogota, several elevational westernmost populations (San Lorenzo), i.e., gaps (ca. 2,000 m) now prevent contact be- those farthest removed from populations of tween humeralisand aterrimaalong the East- humeralis(see Table 6). All available speci- ern Cordillera. To the north, the Tachira mens from Serrania de Perija have at least Depression separates the humeralispopula- some trace of gray edgingson the upper wing tion on Cerro de Tama from gloriosain the coverts, although none have humeral patches Tachira Andes (Vuilleumier, unpubl. observ.). as extensive as those typical of humeralis. In humeralis,the extent of gray on the hu- The black-and-rufous (chestnut) form, glo- meral area and rump is variable, with the dark- riosa,of the Venezuelan Andes is intermediate est individuals resembling nocticolor(e.g., in many charactersbetween the black subspe- USNM 374838, Pamplona, Department of cies, humeralisand brunneiventris.Vuilleu- Norte de Santander). I examined a series of mier (1969b) noted that many individuals of humeralis for signs of introgression from gloriosawith rufous malar spots resembled neighboring “rufous” taxa. Of 65 adult spec- hybrids of brunneiventrisand carbonariafrom imens, 11 exhibit either a pale superciliary line Bolivia. In the present study, 21 of 25 adult or rufous feathers in the &sum (one “Native specimens from scattered localities show Bogota” specimen;one from Norte de Santan- traces of rufous malar spots. Since gloriosais der; four from Santander; one from Boyaca; geographically isolated from other forms of four from Cundinamarca). In USNM 470509 Diglossa,presumably these hybrid-like char- (Chia, Sabana de Bogota), the feathers of the acteristics are not due to ongoing genetic in- crissum have gray centers and broad rufous filtration from nearby black populations. This borders. In a situation analogous to the oc- suggeststhat the “rufous” population (brun- currence of “hybrid-like” carbonariain east- neiventris-like)of the Venezuelan Andes was ern Bolivia, thesespecimens of humeralisfrom contacted by the black proto-humeralisacross scatteredlocales up to 250 km from the nearest the Tachira Depression during the glacial low- “rufous” populations do not seem to be the ering and expansion of the Andean forest. As- productsof introgression.No rufous was found suming that isolating mechanismshad not de- on the belly, breast, or malar areas in nocti- veloped at the time of contact, hybridization color, humeralis,or aterrima. However, in was uninhibited. Gene frequencies subse- common with occasional individuals of noc- quently reached equilibrium during the pres- ticolorand aterrima,several of the humeralis ent interglacial isolation. specimens have faint gray barrings on the flanks, upper tail coverts, crissum, and lower HYPOTHETICAL SCHEME FOR SPECIATION belly. IN THE DIGLOSSA CARBONARIA Paynter (1978) suggestedthat in addition to SUPERSPECIES the north Andean refuges proposed by Haffer Using existing geographic distribution and (1969, 1974) the western slope of the Nevada phenotypic characteristicsof populations and de1 Cocuy (the highest peak of the Eastern geo-historical events, zoogeographerscan re- Cordillera) may have served as a refuge. A constructputative speciation models (Vuilleu- persistent “rufous” refugial population on the mier 198 1b). These models are difficult or im- western slope of the Eastern Cordillera may possibleto test, but I propose the following as have been swampedduring the northward dis- a nucleus for experimentation. In attempting persal of black “aterrima-like” birds from the to reconstruct the evolutionary history of the Napo refuge (eastern Ecuador). The observed D. carbonariasuperspecies, several less likely phenotypic variation in humeralismay again hypothesesmay be rejected in favor of a “best- be an “echo” of former contact between the fit” hypothesis of speciation consistent with two gene pools. An alternative but less satis- the Pleistocene history of the Andes (but see factory hypothesis is that introgression from Dunbar 1980). nearby gloriosais responsible for “hybrid- The ancestral stock of the D. carbonaria like” characters. complex probably evolved in northwestern The gray-rumped form, nocticolor,occurs South America, the present center of distri- in the Serrania de Perija and Sierra Nevada de bution of the genus. During the general uplift Santa Marta. Zimmer (1929) noted that some of the Andes in the late Pliocene and early SPECIATION OF FLOWER-PIERCERS 11

Pleistocene, proto-carbonaria stock probably TABLE 6. Gray coloration in the humeral area of noc- adapted to the newly formed shrublandsin the ticolor. temperate zone. Diglossinestock from the pre- Ind$i$ds Individuals existing South Venezuelan Highlands may showine gray-tipped gay inl have invaded the emerging Andes. Mayr and lesser wing humeral area Phelps (1967) proposed that D. duidue of the Location n COVeItS (% of sample) Venezuelan Tepuis is the product of an inva- San Lorenzo 13 5 38 sion of the flower-piercers from the Andes and Paramo de Churuqua 10 6 60 is related to D. carbonaria.Vuilleumier (1969b) Paramo de Macotama 29 21 72 considered D. duidue to be related to either San Miguel 6 3 50 Rio Guatapuri 9 ; 78 the lafresnayii or the carbonaria superspecies. Mamancanaca 4 75 Since the Venezuelan Highlands predate the Siminchucua 8 I 88 Andes, it is possible that the Andean carbo- Chirua 3 3 100 naria and lafresnayii superspecieswere prod- Sierra PerijA (South Teta)* 6 6 100 ucts of a double invasion from the Venezuelan LagunadeJunco* 11 11 100 Highlands. The geographic distribution and ’ Serrania de P&j&. partially overlapping phenotypic complexes of the carbonaria and lafresnayii superspecies may indicate that they are undergoing a pro- bamba refuges apparently underwent a series cess similar to the “taxon cycles” that occur of progressiveor sudden melanic phenotypic on islands (Wilson 196 1). However, this in- changes, either through natural selection or terestingpossibility cannot be verified without genetic drift. a thorough study of speciation in the lufres- There is ample evidence of widespread nayii superspecies. Pleistocene vulcanism in the northern Andes Following their adaptation to the emerging (see Folster and von Christen 1977, Folster et temperate zone shrubland, groups of flower- al. 1977, and Folster and Hetsch 1978). The piercers may have spread southward along the deposition of successivelayers of volcanic ash Andean chain to about 205 latitude (i.e., to between 32,000 and 10,000 years B.P. in the the current southern distributional limit of the Western, Central, and Eastern Cordilleras of D. carbonaria superspeciesin the Department Colombia and adjacent Ecuador, darkened the of Chuquisaca, Bolivia; Fig. 1). Examination shrubland and . One can imagine of present distributions of the superspecies microphyllous timberline cloud forest hun- suggeststhat the ancestraltaxon resembledthe dreds of kilometers from the centers of present-day form brunneiventris,which now volcanic activity covered with precipitating has two isolated relictual populations at the ash. Chapman (1926) cautiously suggestedthat northern end of the Western and Central Cor- central Andean populations of brunneiventris dilleras of Colombia some 1,500 km north of were devastated by volcanic activity, thus al- its extensive Peruvian range (Fig. 1). This dis- lowing a new form (aterrima) to fill the dis- tributional pattern suggeststhat the formerly tributional void. In the Napo refuge, the mu- continuous range of the brunneiventris-like tation of the “ornate” proto-carbonaria to a (rufous-bellied) proto-carbonaria was frag- melanistic form, which matched the environ- mented so that groups are now separated by ment more closely, could have been stabilized the black form aterrima in Colombia, Ecua- rapidly by natural selection. The preponder- dor, and Peru. ance of high Andean taxa exhibiting dull gray, It is not known if proto-carbonaria was iso- brown, and black plumage probably is not co- lated in North Andean glacial refuges before incidental (e.g., Cinclodes,Scytalopus, Turdus, its Andes-wide dispersal.However, during the Diglossa, Phrygilus, Catamenia). Of the 38 first major glacial period following its dis- north Andean passerineslisted by Vuilleumier persal,proto-carbonaria was probably isolated and Simberloff (1980: Appendix 7) only the in a number of highland refugia along the east- Eastern Meadowlark (Sturnella magna) and ern slope of the Andes (Fig. 7). A myriad of the Andean Siskin ( [Carduelis] spines- other lesspersistent, smaller, and isolated ref- tens) (both with yellow underparts)are brightly uges also probably existed in moist forest colored. Johnson and Brush (1972) suggested pockets at the heads of river valleys on the a similar selective advantage in dark color easternslope and to a lesserextent on the west- morphs of Chlorospingusfrom the volcanic ern slope. Some differentiation probably oc- highlands of Costa Rica. In contrast, the in- curred in these transient refugia, which were complete melanization of carbonaria in the obliterated during the climatic extremes of gla- Cochabamba region of Bolivia may be unre- cial cycles. The rufous-breasted “ornate” lated to volcanism. proto-carbonaria of the Napo and Cocha- As periods of glaciation began and cooler 12 GARY R. GRAVES

freely and populations of carbonaria from sev- eral of the less widely separated refuges (e.g., the three Peruvian refugesin Fig. 7) came into contact with each other. However, during peaks of glaciation, the climate was extremely dry (Van der Hammen 1974) and the size of habitable areas again shrank, restricting dis- persion and gene flow between these groupsof birds (Graves, unpubl.), even though some highland refugia probably remained confluent with nearby lowland refugia (Haffer 1974). It is probable that the Bolivian refuge remained Y isolated, perhapsblocked by the extensive gla- cial formation east of present day Lake Titi- caca. B. Vuilleumier (197 1) hypothesized that glaciers and tongues of ice created enough of a gap in the puna vegetation east of Titicaca to have caused divergence of populations on either side of it. The melanistic Napo group in Ecuador (D in Fig. 7) expanded northward into Colombia and invaded the southern portion of the West- ern Andes, as well as the Central Andes. The “rufous” carbonaria of the Venezuelan (MC- BOLIVIA rida) refuge (4 in Fig. 7) was relatively isolated from the Colombian Andes by the Rio Torbes Valley which acts as a filter barrier. However, it is probable that during the humid, but cool climatic conditions at the beginning and end of glacial periods, when temperate vegetation retreated and high Andean forest was more CHILE extensive, gene flow across this barrier was c’ considerable. The Serrania de Perija and the Sierra Nevada de Santa Marta in Colombia FIGURE 7. Areas proposed as major refugial centersof did not uplift until the late Pleistocene (B. differentiation for high Andean birds inhabiting wet tem- Vuilleumier 197 1) and probably underwent perate areas during arid parts of the middle-to-late Pleis- tocene Period. A = Venezuelan Andes, B = Nechi, only one glaciation. In the Western Cordillera, C = Magdalena?, D = Napo, E = Huallaga, F = Junin, habitat islandsfluctuated greatly in size during G = Cuzco, H = Cochabamba. Refuges B, D, and the glacial cycles, apparently leading to the ex- “Eastern Peruvian Refugia” (E, F, G) were probably con- tinction of either black or rufous forms of car- fluent with the lowland refuses of Haffer (1974). Other bonaria in the Departments of Cauca and An- smaller and less effective potential refuges(including the Sierra Nevada de Santa Marta, Serrania de Perija, and tioquia in Colombia. Melanistic stock from the Central Cordillera of Colombia; the upper and lower Eastern Cordillera (C in Fig. 7) probably dis- Marafi6n, upperApurimac, and Tarija ofnorthwestern Ar- persed northward during the last glacial cycle, gentina) are not depicted. Arrows depict repetitiousroutes contactingthe rufous population of the MCrida of dispersalfrom refugial areasduring the onset of glacia- tion. Suture zones should be detected between major ref- refuge. The rufous carbonaria on the Nechi uges,provided that no smaller refugesoccupied the inter- refuge (at the northern end of the Western and vening territory. The upper MaraA6n corridor (X) is the Central Cordilleras; B in Fig. 7) remained iso- major dispersal route between the eastern and western lated during this period. slopesof the Central Andes in Peru. Polylepis wood land The last glacial extreme again isolated var- at the head of river valleys cy) probably formed important dispersalbridges (“islands”) acrossthe northern Altiplano ious taxa of carbonaria in refugial areas that in central and southern Peru during interglacial periods. probably lie within their current ranges.These groups subsequently became phenotypically distinct and developed isolating mechanisms within these refuges,probably during the Late climatic conditions prevailed, temperate vege- Pleistocene and Holocene. More recently, in tation retreated from the highlandsand alpine the last 8,000 to 10,000 years, amelioration of vegetation became more extensive, i.e., ad- climatic conditions has allowed secondary vanced to lower elevations. This probably per- contact between several of them (brunneiven- mitted high Andean birds to disperse rather tris and aterrima in northern Peru, and brun- SPECIATION OF FLOWER-PIERCERS 13 neiventrisand carbonariain northwestern Bo- Moore Laboratory of Ornithology (Occidental College), livia; see Fig. 1). Museum of Comparative Zoology at Harvard University, Museum of Vertebrate Zoology at Berkeley, Peabody The validity of this model of speciation may Museum at Yale University, and the United States Na- be tested by comparing the distribution of tional Museum. Diglossaand other birds that inhabit forest Funds for my field work were provided by Louisiana and shrubland in the high Andes and have State University, J. S. McIlhenny, B. M. Odom, H. I. and L. R. Schweppe,E. W. Mudge, and AeroPeru. My work similar dispersalcapabilities. was also underwritten by the Direction General Forestal y de Fauna of the Ministerio de Agricultura of Peru. TAXONOMIC CONCLUSIONS On the basisof existing information, I believe LITERATURE CITED that the D. carbonariasuperspecies consists of three or possibly four allopatric or parapatric ANDERSON,E. 1949. Introgressivehybridization. Wiley, New York. taxa at or near the species level. The black BOND, J. 1955. Notes on Peruvian Coerebidae and forms, nocticolorand hurneralis,are very Thraupidae. Proc. Acad. Nat. Sci. Phila. 107:35-55. closely related and only subspecifically dis- CARPENTER,F. L. 1976. Ecology and evolution of an tinct. The entirely black aterrima is a well- Andean hummingbird (Oreotrochilusestella). Univ. marked subspeciesof D. humeralis,which thus Calif. Publ. Zool. 106:1-74. CHAPMAN,F. M. 1917. The distribution of -life in includes nocticolor,humeralis, and aterrima. Colombia: a contribution to a biological survey of The forms brunneiventrisand carbonariahy- South America. Bull. Am. Mus. Nat. Hist. 36: l-729. bridize onlv, in a narrow zone of overlan in CHAPMAN,F. M. 1926. The distribution of bird-life in northwestern Bolivia. Assuming the existence Ecuador: a contribution to a study of the origin of Andean bird-life. Bull. Am. Mus. Nat. Hist. 55: l-784. of “partial” isolating mechanismsbetween the DUNBAR,M. J. 1980. The blunting of Occam’s Razor, two taxa, they should be treated nomencla- or to hell with parsimony. Can. J. Zool. 58: 123-128. turally as semispecies(taxonomic species in ENDLER,J. A. 1977. Geographic variation, speciation, sensuShort 1969). The Colombian population and clines. Princeton Univ. Press, Princeton, NJ. of brunneiventrisis subspecifically distinct FITZPATRICK,J. W. 1973. Speciation in the genus Och- thoeca(Aves: Tyrannidae). Breviora 402: l-l 3. from Peruvian brunneiventris.The Venezue- FOLSTER,H., AND H. VON CHRISTEN.1977. The influence lan form gloriosais a distinct allospeciesre- of Quatemary uplift on the altitude zonation of moun- lated to both black and rufous forms. tain soils on diabaseand volcanic ash in humid parts of the Colombian Andes. Catena(Giessen) 3:233-263. DIGLOSSA CARBONARIA SUPERSPECIES FOLSTER,H., AND W. HETSCH. 1978. Paleosol sequences in the Eastern Cordillera of Colombia. Quat. Res. I. humeralisallospecies (D. humeralis) (N.Y.) 9:238-248. subspecies-humeralis(D. h. humer- FOLSTER,H., W. HETSCH,AND E. SCHRIMPFF.1977. Later alis) Quaternarypaleosols in the Western and Central Cor- subspecies- nocticolor(D. h. noctico- dillera of Colombia. Palaeogeogr.Palaeoclimatol. Pa- laeoecol. 21:245-264. - - lor) GRAVES.G. R. 1980. A new subsneciesof Didossa (car- subspecies-aterrima(D. h. aterrima) bonmia) brunneiventris.Buli. Br. OmGhol. Club II. brunneiventrisallospecies (D. brunnei- 100:230-232. ventris) HAFFER,J. 1967. Speciation in Colombian forest birds subspecies- Peruvian form (D. b. west of the Andes. Am. Mus. Novit. 2294:1-55. HAFFER,J. 1969. Speciation in Amazonian forest birds. brunneiventris) Science 165:131-137. subspecies-Colombian form (D. b. HAFFER,J. 1974. Avian speciation in tropical South vuilleumieri) America. Publ. Nuttall Gmithol. Club, No. 14. III. carbonariaallospecies (D. carbonaria) HELLMAYR.C. 1935. Cataloaueof birds of the Americas IV. allospecies and adjacent islands. Field Mus. Nat. Hist. Publ., gloriosa (D. gloriosa) Zool. Ser. 13:l-258. JOHNSON,N. K., AND A. H. BRUSH. 1972. Analysis of ACKNOWLEDGMENTS polymorphism in the Sooty-capped Bush . I thank J. W. Eley, M. Sanchez, T. S. Schulenberg,and Syst. Zool. 21:245-262. M. D. Williams for field assistancenear Cutervo in 1977 MAYR, E. 1963. speciesand evolution. Harvard and 1978, and S. Hilty and fellow graduate students for Univ. Press, Cambridge, MA. information. K. C. Corkum, J. Haffer, W. J. Harman, T. MAYR, E., AND W. PHELPS.1967. The origin of the bird A. Parker III, J. V. Remsen, T. S. Schulenbergand an fauna of the south Venezuelan highlands. Bull. Am. anonymous referee provided helpful comments on the Mus. Nat. Hist. 136:269-328. manuscript. PAYNTER,R. A., JR. 1972. Biology and evolution of the I esoeciallv thank S. Y. Graves. F. Vuilleumier. J. P. Atlapetesschistaceus species-group (Aves: Emberizi- O’Nei& and-the late G. H. Lowery, Jr., for support and nae). Bull. Mus. Comp. Zool. 143:297-320. advice during this study. PAYNTER, R. A., JR. 1978. Biology and evolution of the For the loan of specimens,I gratefully acknowledgethe avian genus Atlapetes (Emberizinae). Bull. Mus. Academy of Natural Sciencesof Philadelphia, American Comp. Zool. 148:323-369. Museum of Natural History, Carnegie Museum in Pitts- SHORT, L. L. 1969. Taxonomic aspectsof avian hybrid- burgh, Delaware Museum of Natural History, Field Mu- ization. Auk 86:84-105. seum of Natural History in Chicago, Los Angeles County SIMPSON, B. 1975. Pleistocenechanges in the flora of the Museum, LouisianaState University Museum ofZoology, high tropical Andes. Paleobiology 1:273-294. 14 GARY R. GRAVES

SIMPSON,B., AND J. HAFFER. 1978. Speciation patterns VUILLEUMIER,F. 1970b. Generic relationsand speciation in the Amazonian forest biota. Annu. Rev. Ecol. Syst. patternsin the caracaras(Aves: Falconidae).Breviora 91497-518. 355:1-29. VAN DERHAMMEN, T. 1974. The Pleistocenechanges of VUILLEUMIER,F. 1971. Generic relationships and spe- vegetation and climate in tropical South America. J. ciation patterns in Ochthoeca,Myiotheretes, Xolmis, Biogeogr. 1:3-26. Neoxolmis,Agriornis, and Musciiaxicola. Bull. Mus. VAN DERHAMMEN, T., J. H. WERNER,AND H. VANDOM- Comn. Zool. 141:181-232. MELEN. 1973. Palynological record of the upheaval VUILLEUMIER,F. 1981a. Speciation in birds of the high of the Northern Andes: a study of the Pliocene and Andes. Proc. XVII Int. Ornithol. Congr. Lower Quatemary of the Colombian Eastern Cor- (1978):1256-1261. dilleraand the earlyevolutionofits high-Andeanbiota. VUILLEUMIER,F. 1981b. Reconstructing the course of Palaeogeogr.Palaeoclimatol. Palaeoecol. 16:l-24. speciation. Proc. XVII Int. Ornithol. Congr. VAURIE, C. 1972. An ornithological gazetteer of Peru (1978):1296-1301. (based on information compiled by J. T. Zimmer). VUILLEUMIER,F., AND D. SIMBERLOFF.1980. Ecologyvs. Am. Mus. Novit. 2491:1-36. history as determinants of patchy and insular &stri- VUILLEUMIER,B. S. 1971. Pleistocene changes in the butions in high Andean birds. D. 235-379. In M. K. fauna and flora of South America: Science Hecht, W. CT Steere, and B. ‘Wallace [eds.], Evolu- 173:77l-780. tionary biology. Vol. 12. Plenum, New York. VUILLEUMIER,F. 1968. Population structure of the As- WILSON,E. 0. 1961. The nature of the taxon cycle in the thenesflammulata superspecies(Aves: Fumariidae). Melanesian ant fauna. Am. Nat. 95:169-193. Breviora 297: l-2 1. ZIMMER,J. T. 1929. Variation and distribution in two VUILLEUMIER,F. 1969a. Pleistocene speciation in birds speciesof Diglossa.Auk 46:21-37. living in the high Andes. Nature 223: 1179-l 180. VUILLEUMIER,F. 1969b. Systematicsand evolution in Museum of Zoology Louisiana State University,Baton Diglossa (Aves, Coerebidae). Am. Mus. Novit. Rouge, Louisiana 70893. Presentaddress: Department of 238 1:l-44. BiologicalScience, Florida State University,Tallahassee, VUILLEUMIER,F. 1970a. Insular biogeographyin conti- Florida 32306. Accepted for publication 5 November nental regions. I. The Northern Andes of South 1980. America. Am. Nat. 104:373-388.

Condor8414 0 The Cooper Ornithological Society 1982

RECENT PUBLICATIONS

Annotated Checklist of the Birds of Arizona. SecondEdi- The Birds of Northeastern Utah.-William H. Behle. tion/Revised and Expanded.-Gale Monson and Allan R. 1981. Occasional Publication No. 2, Utah Museum of Phillips. 1981. University of Arizona Press, Tucson. 240 Natural Historv. Universitv of Utah. Salt Lake Citv. Utah. p. Paper cover. $5.95. This authoritative checklist is the 136 p. Paper cover. $10.95: This work suppleme& Behle revised edition of a work first published in 1964. Docu- and Perry’s comprehensive Utah Birds (noted in Condor mented species are distinguished from hypothetical 77:371), being a detailed report on the hitherto little- species.The entry for each speciesgives its seasonaland known avifauna of the northeastern part of the State. A distributional status, the subspeciesrepresented, and a long introduction presentsseveral aspectsof the area, its summary of important records, especially in the past 20 ecology, and its birdlife. Several tables summarize infor- years. An appendix gives a specialdiscussion of the races mation on habitat usageand density of hawks and owls. of the Red Crossbill (Loxia curvirostra)in Arizona, in- The speciesaccounts cover seasonalstatus, ecological oc- cluding descriptionsof two new subspecies.This pocket- currence,a summary of specimensobtained, other perti- size book will be an indispensablefield reference for se- nent localities where the specieswas seen, and a resumt rious birders in Arizona. Index. of previously published records. For some speciesthere are remarks on natural history and/or systematics.A list of referencesis given, but regrettably no map. Although this book will be useful to seriousbirders, it is primarily intended for curatorsand researcherswho are working on the distribution and of birds in Utah and ad- jacent states.