AMERICAN MUSEUM Noqvtates PUBLISHED BY THE AMERICAN MUSEUM OF NATURAL HISTORY CENTRAL PARK WEST AT 79TH STREET, NEW YORK, N.Y. 10024 Number 28 1 1, pp. 1-14, figs. 1-2, tables 1-4 February 20, 1985

Variation and Population Structure of the Tourmaline , Heliangelus exortis exortis (Aves, Trochilidae)

ROBERT BLEIWEISS'

ABSTRACT The nominate form of the polytypic Andean tion. Geographic patterns of variability in the the , Helian- polychromatism do not appear to result from in- gelus exortis exortis, exhibits dramatic variation creased variability in putative zones ofhybridiza- in a female polychromatism of throat iridescence tion between differentiated populations. No char- within and among populations that has not here- acter breaks that might indicate noninterbreeding tofore been appreciated. An analysis of the geo- parapatric forms were noted. It is concluded that graphic population structure of samples referable H. e. exortis comprises a single biological entity to H. e. exortis was conducted to determine the and taxon within the polytypic H. exortis complex. biological status and historical process of differ- The evolutionary origins of the complex pattern entiation ofpolychromatically differentiated sam- of geographic variation in H. e. exortis are dis- ples. Geographic variation in additional color and cussed. I suggest that biological attributes, in ad- six mensural characters is discordant within and dition to potential geographic barriers, are nec- between the sexes. The polychromatism shows both essary for a full understanding of the geographic regular and mosaic patterns of geographic varia- differentiation seen in H. e. exortis. INTRODUCTION There are few detailed studies of the geo- categories ofisolates, continua (clines and re- graphic population structure ofAndean lated phenomena), and secondary contact. (Vuilleumier, 1968, 1980a, 1980b, 1984; There are also classes ofgeographic variation Graves, 1982). These studies have been aimed that do not fit these categories but neverthe- primarily at reconstructing the course ofgeo- less manifest striking patterns. This is espe- graphic speciation by describing geographic cially true in topographically complex moun- variation in terms of Mayr's (1959, 1963) tain systems and oceanic archipelagoes.

' Chapman Research Fellow, Department of Ornithology, American Museum of Natural History.

Copyright © American Museum of Natural History 1985 ISSN 0003-0082 / Price $1.90 2 AMERICAN MUSEUM NOVITATES NO. 2811

Unusual forms of geographic variation re- this exceptional form. The present paper treats corded in these regions include changes in geographic variation in the female polychro- sexual dimorphism, sex-limited variation, matism in conjunction with variation in oth- and mosaic, nonclinal variation (Mayr, 1942). er color and mensural characters to examine These complex patterns suggest biological as these questions. well as extrinsic factors as causal agents of geographic variation, and are of general in- ACKNOWLEDGMENTS terest for understanding the full range of fac- I thank the curators of the collections used tors that promote geographic differentiation in this study for providing specimens under and speciation. Because of great interest in their nectar- their care: Dr. F. B. Gill, Academy ofNatural feeding habits, are becoming Sciences, Philadelphia; Dr. J. P. O'Neill, one of the better known avian families from Louisiana State University Museum of Zo- the standpoint ofecology and behaviof. This ology, Baton Rouge; Dr. R. A. Paynter, Jr., knowledge has additional application in stud- Museum of Comparative Zoology; Dr. L. L. ies of the biological basis of geographic dif- Short, American Museum of Natural His- ferentiation, of which hummingbirds in the tory; Dr. D. S. Wood, Carnegie Museum of topographically complex provide Natural History; Dr. R. L. Zusi, National many dramatic examples (Zimmer, 1951a, Museum ofNatural History, Smithsonian In- 1951b, 1952). However, studies of hum- stitution. Drs. W. H. Bossert, L. L. Short, F. mingbird systematics have lagged far behind Vuilleumier, E. E. Williams and Mr. G. C. knowledge of their feeding ecology and be- Mayer made helpful comments on an earlier havior. draft of this paper. Laszlo Meszoly drafted While conducting a survey of geographic figure 1 and Lynn Winter drafted figure 2. variation and speciation patterns in Andean This work was funded in part by a grant from hummingbird genera, I found an unusual form the Frank M. Chapman Fund of the Amer- of color variation in members of the ican Museum of Natural History. Heliangelus. Females of several ex- hibit dramatic variation in the development OF THE of the highly iridescent throat, or "gorget," HELIANGELUS EXORTIS characteristic of males (Bleiweiss, MS). This COMPLEX "polychromatic" variation is particularly Hummingbirds of the genus Heliangelus pronounced within and among populations Gould, 1848, are endemic to mid-elevations ofthe nominate form ofthe polytypic species (1500-3000 m) ofthe Andes from Venezuela Heliangelus exortis, the Tourmaline Sunan- to Bolivia. Current taxonomic treatments gel. Samples from the central section of the recognize seven species in the genus, some of Central and Eastern Cordilleras of Colombia which are highly polytypic (Peters, 1945; show a range of variation from a nonirides- Meyer de Schauensee, 1966). The Tourma- cent to a fully iridescent gorget similar to line Sunangel, Heliangelus exortis, is cur- males, whereas samples from southern Co- rently treated as a polytypic species com- lombia have only discrete noniridescent and prised ofa nominate form with violet-chinned fully iridescent male-like morphs. In con- and rose-throated males occurring in the three trast, the polychromatism is rare or absent Andean ranges of Colombia south through in samples from the eastern slopes ofthe An- the eastern slope of the Ecuadorian Andes, des in Ecuador. and two apparently allopatric forms with or- Intraspecific sex-limited variation and ange-throated males, H. e. micraster and H. variation in sexually dichromatic features are e. cutervensis, in the Andes of southwestern both rare phenomena in birds. The geograph- Ecuador and northern , respectively ic population structure of Heliangelus e. ex- (Meyer de Schauensee, 1966). A number of ortis is therefore of great interest for (1) as- characteristics support the conclusion that H. certaining whether H. e. exortis should be e. exortis and the orange-throated forms are considered a single species and (2) gaining closest relatives in Heliangelus (Zimmer, insight into the process of differentiation in 1951 b; Fitzpatrick, Willard, and Terborgh, 1985 BLEIWEISS: HELIANGELUS EXORTIS EXORTIS 3

TABLE 1 Distribution and Sample Size for Heliangelus exortis exortis Specimens Examined Cordillera Locality Pooleda West Central Eastb Adults Immatures 1. El Peii6n A West 2. La Aguadita A West 3. El Roble A West 6 4. FusugasugA-Silvania A West 4 5. Hacienda Zuliaba East 7 6. Santa Elena East 2 7. Paramo de Sons6n West 6 2 8. El Zancudo B West 17 1 9. Santa Isabel B West 1 1 10. Laguneta West 43 15 11. Toche; Tolima West 7 6 12. Purace West 1 13. Coconuco West 14. Paletara West 2 15. Almaguer West 6 3 16. Hacienda La Ilusi6n West 8 1 17. La Florida C East 1 18. Coast Range C East 3 19. Cerro Munchique C East 17 1 20. Munchique-El Tambo C East 2 21. Tijeras-Moscopan East 5 22. Guanderal D East 3 23. Cerro Pax D East 8 24. Papallacta E East 7 25. Cuyujua E East 2 26. Baeza E East 3 27. Volcan Sumaco Isolated peak 10 28. San Antonio F East .1 1 29. Ambato F East 1 30. Bafios F East 6 31. Volcan Tungurahua F East 4 32. Planchas East 2 33. Coraz6n West a Letters A-F designate pooled localities used for analysis of geographic variation. b East and west designations under each Cordillera refer to the slope of the range.

1979; Bleiweiss, in prep.). Each can be dis- AND tinguished from most sympatric congeners by MATERIALS METHODS the absence of a white pectoral band, and The present study is based upon 230 spec- uniquely, by white female throat feathers and imens from six museum collections; Amer- under-tail coverts in both sexes. The Purple- ican Museum of Natural History, New York throated Sunangel, H. viola, also lacks a white (AMNH), Academy of Natural Sciences, pectoral band and may be closely related to Philadelphia (ANSP), Carnegie Museum of H. e. exortis (Chapman, 1926). However, H. Natural History, Pittsburgh (CAR), Museum viola is sympatric with the orange-throated ofComparative Zoology, Cambridge (MCZ), members of the H. exortis complex and is Louisiana State University Museum of Nat- thus a distinct species. In my analysis, I define ural History (LSUMZ), and the National Mu- H. e. exortis as all rose-throated populations seum of Natural History, Washington D.C. of the H. exortis complex. (USNM). I used all specimens with adequate 4 AMERICAN MUSEUM NOVITATES NO. 2811

FIG. 1. Geographic distribution ofHeliangelus exortis exortis specimens examined. Locality numbers are identified in table 1. Letters A-F indicate localities pooled for the analysis of geographic variation (see Appendix 1). locality data from throughout the range ofH. on the labels of only two specimens. Imma- e. exortis (table 1, figure 1, see Appendix 1: tures were distinguished by the presence of Specimens Examined). corrugations on the culmen, lacking in adults Study skins were aged and sexed by exter- (Ortiz-Crespo, 1972). Older immatures, rec- nal characters since gonadal data were noted ognized by their few bill corrugations, are 1985 BLEIWEISS: HELIANGEL US EXORTIS EXORTIS 5

TABLE 2 Geographical Variation in Color Characters in Heliangelus exortis exortisa Character' Male Female Throat Under-tail Populationb Throat Classes Nonglittering Coverts A Bogota region R I-IV w w 5 Hacienda Zuliaba R II/IV b Faintly b 6 Santa Elena R - - b 7 Paramo de Sons6n R II/IV b b B El Zancudo, Santa Isabel R I-IV b b 10 Laguneta R I-IV b b 11 Toche R II/IV b b 12 Purace R - - b 13 Coconuco R - - Faintly b 14 Paletarai R IV Faintly b 15 Almaguer R I/IV w w, b 16 Hacienda La Ilusi6n R IV b Faintly b C Cerro Munchique R II/IV b b 21 Tijeras-Moscopan R I/IV w, b w, b D Guanderal, Cerro Pax R I/IV w w 24-31 Eastern Andes of Ecuador R I w w 32 Planchas - I/S(?) w w 33 Coraz6n - IV - w aAdults and immatures. h Populations labeled as in table 1 and figure 1. C Symbols for characteristics of gorget: R = glittering rose with violet chin; S scaled, white with glittering rose edge; w = white; b = bicolored, white with a shining green to dusky mark; w, b = both white and bicolored conditions present. Female throat class symbols described in Materials and Methods. similar to adults in throat color (Bleiweiss, respectively. The female polychromatism 1983), and are included in evaluating pat- consists ofvariation in glittering throat feath- terns of geographic variation in coloration. ers which I quantified by the method devised Immature hummingbirds do differ from by Bleiweiss (1983). Individuals were divid- adults in mensural characters (Stiles, 1983). ed into four classes based upon the number I include measurements for only 2 of special ofglittering throat feathers; class I = 0; class interest from among the 42 examined since II= 1-17; class III = 21-28; class IV = 31- most localities had few immatures (table 1). 50+. Classes II and III have a mixture of Criteria for distinguishing the sexes based glittering and nonglittering feathers while upon measurements and body coloration were classes I and IV generally have only nonglit- developed in Bleiweiss (1983), and the results tering or glittering feathers, respectively. were utilized for this study. Females differ For the analysis ofpopulation structure, six from males in having a lighter green body measurements were taken on study skins plumage and tan bellies, and are consistently judged to be adults: length ofexposed culmen smaller in two dimensions, length of the first (EC), length of culmen from flange ofthe na- primary and length of the outer tail feather sal operculum (CF), length of first primary (rectrix 5); a length of45.5 mm for both mea- (FP), chord length of unflattened wing (WB), surements was found to separate the sexes. and lengths of inner (IT) and outer (OT) tail To describe variation in the brilliance of feather (Baldwin et al., 193 1). iridescence, I use Greenewalt's (1960) terms The mapping of collecting localities was "glittering" and "shining" to refer to irides- accomplished with the aid of recently com- cent reflections of high and low brilliance, piled gazetteers of ornithological collecting 6 AMERICAN MUSEUM NOVITATES NO. 2811 n

a. C.

FIG. 2. Throat feather types ofmale and female Heliangelus e. exortis. From left to right: (a) glittering rose feather from adult male; (b) white, fringed glittering rose feather from immature "female" from Planchas, southern Ecuador; (c) white, glittering rose feather from adult female; (d) bicolored feather from adult female; (e) fully white feather from adult female.

localities in South America (Paynter and most known limit of rose-throated H. e. Traylor, 1977, 198 1). Museum materials from exortis. The throat feathers ofthis individual specific localities represent small, sometimes are white with narrow glittering rose edges, temporally heterogeneous collections, and are giving the throat a scaled appearance. This referred to here as samples. In the statistical condition differs from all other immature analysis of geographic variation, some geo- male gorgets, which are similar to adults in graphic samples were pooled as noted in table color but reduced in extent. The sex of this 1 and figure 1. Groupings ofgeographic sam- specimen is problematical since the gonads ples were based upon close geographic prox- were greatly reduced in size (1 x 1 mm). Im- imity and absence of an obvious geographic mature hummingbirds have longer wings than barrier between them (Vuilleumier, 1968). adults so the measurements of this specimen The samples were homoscedastic by F-max are certainly more typical of females (FP = (Sokal and Rohlf, 1981) and met other as- 42.8; OT = 41. 1). I judge this specimen to be sumptions for parametric ANOVA. I therefore a female. used one-way ANOVA with a posteriori Scheffe Adult female samples vary dramatically in contrasts to analyze patterns of geographic throat class frequencies, from highly variable variation in the six measurements taken on (I-IV) in the central section of the Central adults. All male samples - five and female and Eastern Cordilleras of Colombia (locali- samples >-four were included in the analysis. ties A, B, and 10), to two pronounced classes (I and IV) at their southern end (localities 14, 15, 21, and D), to predominantly the non- VARIATION IN COLOR iridescent class (I) on the eastern slope of the CHARACTERS Andes in Ecuador (localities 24-31); one of two immature females from the southern- Table 2 summarizes geographic variation most collection in Planchas (locality 32) lacks in color characteristics for each sex. The vari- throat iridescence. ation in throat feather color patterns is shown Only a single immature specimen, with a in figure 2. Males have glittering violet chins fully developed gorget (count 40), has been and rose throats at all localities. There is also collected in western Ecuador, labeled "lower one atypical immature specimen (fig. 2b) la- west side of Corazon," west of Quito. This beled as a male (LSUMZ 82902) from Plan- specimen was collected by Goodfellow and chas, Ecuador (locality 32), at the southern- Hamilton in 1898, and was subsequently de- 1985 BLEIWEISS: HELIANGELUS EXORTIS EXORTIS 7 scribed by Oberholser as a separate taxon He- similar to nonglittering throat feathers, rang- liangelus exortis soderstromi (Oberholser, ing from bicolored to pure white. Geographic 1902). Oberholser believed the specimen to variation in under-tail coverts usually par- be a male. However, Chapman (1917) cor- allels that ofnonglittering throat feathers, with rectly surmised it to be a female, a deter- bicolored under-tail coverts in samples with mination I corroborated by measurements bicolored throat feathers and white under-tail (FP = 45.2; OT = 42.2). Since the majority coverts in samples with white throat feathers of immature and adult females from eastern (table 2). The type ofH. e. soderstromi, which Ecuador lack throat iridescence, this unique has a fully rose throat, has white under-tail specimen from western Ecuador suggests dif- coverts. ferentiation ofH. e. exortis between the East- There is some evidence for intermediacy ern and Western Cordilleras of Ecuador. in the color and correlation ofthese two char- However, the accuracy ofthe collecting lo- acters in the region between Laguneta (lo- cality given for this specimen can be ques- cality 10) and Cerro Pax (locality D): inter- tioned, as Chapman (1917, 1926) noted. Sev- vening samples from both slopes of the eral hummingbird species restricted to the Central Cordillera in its southern section (lo- eastern or western slopes ofthe Andes in Ec- calities 12-15, 21) are highly variable or in- uador were reported from the opposite side termediate. On the west slope, the single male of the Andes in the Quito region in early from Purace (locality 12) had pronounced collections such as those of Goodfellow and dusky centers on the under-tail coverts, clos- Hamilton (see Oberholser, 1902 for exam- est in color to samples to the north. Speci- ples). This situation may have resulted from mens south of Purace from Coconuco, Pal- inaccurate collecting information supplied by etara, and Almaguer (localities 13-15), were native collectors who were extremely active fainter, or pure white like specimens farther in the Quito region (Chapman, 1926) and south (locality D southward). Specimens from who were employed by Goodfellow (Good- the eastern slope in Tijeras (locality 21) show fellow, 1901). Chapman (1917) stated that similar variation and in addition, show both the type of H. e. soderstromi appeared to be white and bicolored class I throats. Further- a native-made skin. Thus, the true geographic more, the correlation between throat and un- origin and potential evolutionary significance der-tail covert coloration appears to break of this specimen remain in doubt. down in this region since Tijeras females have Aside from polychromatic variation, only various combinations of the two throat and two other plumage characters vary: nonglit- under-tail covert conditions (table 2). tering throat feathers of females, and the un- der-tail coverts of both sexes. In Colombia, the nonglittering throat feathers are bico- GEOGRAPHIC VARIATION IN lored, white with a shining green to dusky MEASUREMENTS terminal dot (fig. 2d) in samples from the Table 3 presents descriptive statistics ofthe Western and Central Cordilleras south six mensural characters for adults of both through Laguneta (locality 10), and pure white sexes. Table 4 presents a summary of the (fig. 2e) in the southernmost sample from the results of the one-way ANOVA with Scheffe Central Cordillera in Almaguer (locality 15), contrasts for the six measurements for adults and in all samples in the Eastern Cordillera of both sexes. In males, five measurements southward into eastern Ecuador. While each showed significant heterogeneity among sample has a characteristic pattern, there is means. However, the ScheffTe contrasts among some variation within them. Single individ- means indicate that this is due largely to the uals with large numbers of bicolored throat small size of males from the Cerro Mun- feathers occur at the "white" localities of El chique region of the Western Cordillera (16- Penion (locality 1), Tijeras (locality 21), and calities 17-20 and C). Males from the Cerro Baeza (locality 26), while there is also one Munchique region differed significantly from individual with pale dusky markings at El samples from three prominent massifs: Vol- Zancudo (locality 8), a "bicolored" locality. can Sumaco (locality 27), Hacienda La Ilu- Color variation in the under-tail coverts is sion (locality 16), and the Nevado de Tolima N0 O- - - - 0 - - - ON r- Wo WIrtq ON tt m qt N tn Cl N .O O N qt £o - - £- 0 0 ON 00 In l lC ON N H Co_:- -- - 6 6 -; -; - -6 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 r. C' r-00 N (ON - - NC O0 N -o0 - C ON O0 N rO tu tN - Cs 00 00 cl W) 0 - O m N N £o 00 00 N Cl 0 N - o - l lo tn +l +l N 0-0 N WI N N t- 00 00 00 .C - 0 Wi dB £o - CD(ON CDO7 I" o c o 1. ONo -Oen0 WU) 0 e-Oa)t _ CO

-. +1 - Cl - +l +1 +1 +1 +1 +1 +1 +l +l +l +l +l +l +l +l +1 +1 o+1 +1o~ 0 +1 +1+1 OC - C 0 N- Cl 00 N 00 Rt r- - - C 00 - C oE_7s -- £ ^b - - O C' "t N r Cl -Cl - N Cl t t tn en (. 0

q4.0 Wi 0 tn < N 't _N tn _ _t Un_ N t o r0F b o ooo - o xo ~~~o C7- N n "o a, 0 + +1 ON 00 N - - - N ocliN -4 -4 -4 -4 N N-l C - ON +l +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 00 ON o C* I'l 0 'I tn - M tn "o M "t 0 m 0 r- - Cl - vE ur un N Cl ON 00 0 ~O N Wi ~: +1 0 H.- en N N0 't-0N t- t- l- 00 .j) N 00 Co N- 66 -oNC 0 +l +1 +1 +1 +1 +1 +1 +1 +1 +1 +l +1 +1 +1 +1 +1 +1 Wi "T-00en}en_ ~r-"o00 O0 NO0en0 en cq00e-C.N ent eq tnenna NeIt000N N00 cq00 0 ovr -0000N 00 ovo'<>0 al,ala,to o - o~oo 000o0r tn a l 0 ClCl - - - ON Ie en en Cl-reC- Q +1 I-I C0 CD l +1 +1 N +1+1 +1 - N N e}n N000 (- N0 - N qTl 0- w 0 l "T Co-t" o m ,T t "o m N 'T 4n kn me IR It I- tn OTt N "o - Cl Cl - Wi 0 L +1 ON 0 0 +l +l +1 +1 0 +l +l +l +l +l +l +l +l +l +l +l +l +l +1 +1 a"s tn 00 N - t- CD c en 00 'T0 " 00 . tn 00 ON o oo r_ o6 o6 o6o6 6 o6 6O6 o6O6r0 o6 o6 o6 Cl 0n C1 ~0~O ~N - ClC00O- ~--4N ~N Cl - 40 00 00 'O0 en 00 00 00 00 00 00 00,0000 0 i en; en v N C I* 1- 0 N - o cj. o o C)-O' ON 00 00 enN 'Ir00 e - r4 -0 en tn 11 ooW; W; W; W; W te tt IrIC;)tl u +1 +1 +1 +1 -4-. . I. -. -. .-.-. . . . "DO 00 00 Cl Cl Wi 0

ts 0 - (q T 0 .Qt, - = -q -4 14 - 1985 BLEIWEISS: HELIANGELUS EXORTIS EXORTIS 9

region (localities B and 10). There was no significant differentiation among other local- ities within the main portion of the range. Sample sizes for females were smaller and ^ £O r N t _ the pattern less consistent than for males. Four N ON T v* 00 o o o cz t of six characters showed significant hetero- 0 +1 +l +1 +1 +1 +1 geneity among means. In contrast to males, CU co ! t 00 females from El Zancudo (locality B) and La- m 00 00 90 Os _ N guneta (locality 10) are most divergent, dif- fering significantly from localities in the East- o~ o N_ - - ern Cordillera of Colombia and the eastern slopes of the Ecuadorian Andes. -_ N N _ m 0000o . . . H 0 C5 . POPULATION STRUCTURE +l +l +1 +1 +1 +1 toC N N- _ The complex population structure that rt _ 00 emerges for H. e. exortis is indicated by at least four different patterns among the var- o N 0 eN N ,,, ious characters:

un _ en t (1) Differences between the sexes in geo- graphic variation ofcolor and mensural char- acters. Males do not vary in any throat color +1 +l +1 +1 +1 +1 CZ -NoN N - feature, whereas females are highly variable stlJ tn 00 en00 .tr Z in throat coloration within and between lo- 2! *4 calities. In mensural characters, males from z the Cerro Munchique region (locality C) in

en^ C : CN - - the Western Cordillera ofColombia are most en tn w vO N 00o N 10ri0 distinctive (smallest) in five characters. Fe- 0-. N 0 males from Cerro Munchique are not signif- m +1 +l +1 +1 +1 +1 ~0 0 icantly different from females from any other E- cU 'O Nlb 00 00 00 N locality. Rather, females from the Central CU Cordillera (localities B and 10) are the most distinct, differing from four (B) and three (10) 'IT tn t t other localities, respectively. ON W tN n .CU (2) Lack of concordance in the geographic C0 e N O Nt variation among color and between color and U0 +1 +1+1 +1+1 +1 d Cd mensural characters within females. Among el 04 0 -O N - in CU 00 color characters, changes throat class fre- c7 00 0jN7- - m 00N quencies are not paralleled by changes in non- u glittering throat feathers (white or bicolored). For example, the white feather condition is C.) U3C observed throughout the Eastern Cordillera Ecc 'O N tn 00 of Colombia through the eastern slope of the +1 cq+l oo+1 oooo+1 +1 +1 n Andes in Ecuador, whereas the throat class CU N o e00 N frequencies change dramatically from highly r;or;o t;o variable (I-IV), to two discrete classes (I and IV), to predominantly class I. No samples in the Eastern Cordillera from the Bogot'a region (locality A) through the east Andean slope in Ecuador differed signifi- cantly in mensural characters even though there is a dramatic change in throat color 10 AMERICAN MUSEUM NOVITATES NO. 2811

TABLE 4 Results of One-Way ANOVA for Geographic Variation in Mensural Characters in Heliangelus exortis exortis Significantly Different Character Males Females Length of exposed culmen 16 > C B, 10 > A Length of culmen from flange B, 10 > C B, 10 > F Length of first primary 27 > C B < A, 10, E Length of wing cord 27 > C B, 10 < E Length of outer tail feather E > C, B ns ns = not significantly different. Characters showing significant heterogeneity among means at 0.05 level of significance are listed on the left-hand side of the table. Inner tail feather (IT) for both sexes, and the outer tail feather (OT) of females show no significant variation. Locality numbers and pooled locality letters refer to table 1. Symbols indicate size relationships. classes present. Conversely, females from El populations in secondary contact. The vari- Zancudo differ from females in the Eastern able sample from Bogot'a represents the prob- Cordillera (locality A) in several measure- able northern limit of H. e. exortis in the ments but both show a highly variable poly- Eastern Cordillera and cannot be a contact chromatism (classes I-IV). region between two differentiated' popula- (3) Regular patterns of geographic varia- tions. tion in single characters in both sexes. Both nonglittering female throat feathers and un- DISCUSSION der-tail coverts in both sexes vary together Regular patterns of geographic variation in a clear geographic pattern. Bicolored non- permit evaluation of the degree of genetic glittering throat feathers in females and bi- continuity and history of differentiation of colored under-tail coverts in both sexes are geographically variable forms. Clinal varia- found in the Western and Central Cordilleras tion provides evidence of the interplay of south to Almaguer (locality 15), whereas these local differentiation and gene flow among features are exclusively white in the Eastern intraspecific populations. Conversely, con- Cordillera south through the east Andean cordant breaks or step clines in the variation slope in Ecuador. ofone or more characters suggest a reduction (4) Mosaic patterns ofgeographic variation or cessation of gene flow between popula- in the polychromatism. There is progressive tions, from which species status can be in- loss ofpolychromatic color classes from north ferred in the absence of behavioral data on to south in the best sampled region, from the reproductive isolation. Concordant patterns central section of the Central and Eastern ofgeographic variation among different char- Cordilleras of Colombia south into Ecuador. acters and taxa can be correlated with phys- This pattern constitutes a regular loss ofvari- ical environmental features that could serve ability in terms ofthe number ofcolor classes, as extrinsic barriers to gene flow. but shows no inherent directionality in the The complex population structure of He- color classes present. Thus it does not con- liangelus e. exortis is a combination of reg- form to the classical definition of a cline ular, but also nonconcordant and even non- (Mayr, 1963). Moreover, while increased classical patterns of character variation that phenotypic variability has often been docu- do not easily fit into the above categories. mented in zones of secondary contact be- This complexity makes an evaluation of the tween two differentiated populations (Mayr, genetic continuity and history of differentia- 1963; Short, 1969; Gill, Stokes, and Stokes, tion along traditional lines difficult. More- 1973; Endler, 1977), high variability (classes over, the discordant patterns of geographic I-IV) in polychromatic samples does not ap- variation within and between the sexes make pear to result from hybridization between an analysis of differentiation based upon ex- 1985 BLEIWEISS: HELIANGELUS EXORTIS EXORTIS I1I trinsic barriers to gene flow difficult, since grees of genetic discontinuity between pop- different barriers have to be invoked to ex- ulations resulting from the general plain different characters and patterns within topographic complexity of the Andean re- and between the sexes. gion. This could be caused or at least en- Clearly, the range of population structures hanced by historical changes in the continuity with dramatic differentiation extends beyond of Andean habitats, which are believed to the traditional classifications of clines, step have undergone repeated fractionation and clines, and distinct isolates. Although H. e. rejoining during the Pleistocene (Vuilleumier exortis presents an especially complex pat- and Simberloff, 1980). This complex history tern, the paucity of detailed studies of pop- could be difficult to decipher from patterns ulation structure makes it difficult to assess of geographic variation alone. how prevalent these cases are among birds. The deep arid valley ofthe Rio Magdalena Such patterns are part of the broader range is one barrier promoting differentiation be- of phenomena that must be considered in tween Andean adapted forms in the Eastern studies of the forces promoting geographic and Central Cordilleras of Colombia (B. S. differentiation. In lieu of these consider- Vuilleumier, 1971) and could be an extrinsic ations, what can be said of the genetic con- barrier promoting differentiation in H. e. ex- tinuity and history of differentiation in H. e. ortis. Isolation across this barrier is impli- exortis? cated in the differentiation ofpopulations into There is no evidence to contradict the sta- those with white under-tail coverts and throat tus ofH. e. exortis as a single biological entity feathers in the Eastern Cordillera of Colom- within the H. exortis complex. Heliangelus bia and Ecuador, and those with bicolored e. exortis does not show concordant regional ones in the Western and Central Cordilleras differentiation which would indicate a sig- of Colombia. Females from the central sec- nificant reduction or cessation of gene flow tion of the Central Cordillera are most di- and thus either incipient or completed spe- vergent in measurements from females in the ciation. Rather, there is little concordant dif- Eastern Cordillera of Colombia, thus sup- ferentiation between males and females or porting the role of the Rio Magdalena valley even among different characters within each as an effective barrier to gene flow between sex. Such major features of differentiation as these two populations. the presence ofpolychromatic populations in The major dichotomy among females of Colombia and nonpolychromatic popula- polychromatic populations in Colombia and tions in Ecuador are not paralleled by similar largely nonpolychromatic ones in Ecuador dichotomies in other female characters, or by cannot be similarly explained. However, the males, whose plumage is remarkably uniform similar high variability in polychromatic throughout the entire range of the taxon. samples (classes I-IV) from the same latitude Positive evidence of gene flow as indicated in the central section ofthe Central and East- by intergradation between differentiated pop- ern Cordilleras ofColombia is noteworthy in ulations is less clear from the available data. light ofthe differentiation between these pop- Intergradation in characteristics ofthe under- ulations in other characteristics. These two tail coverts and nonglittering throat feathers populations do not appear to be in contact in the southern portion of the Central Cor- since the variable Eastern Cordillera samples dillera of Colombia is suggested, but is based (locality A) form the northern range terminus upon few specimens. for H. e. exortis in the Eastern Cordillera with As for the origin of the pattern of differ- no obvious northern connection with the entiation, it is tempting to interpret the gen- Central Cordillera, while to the south pop- eral lack of concordance in variation among ulations are characterized by morphic vari- characters as evidence that differentiation has ation (classes I and IV). Given the apparent occurred in situ. However, caution is needed effectiveness of the Rio Magdalena valley as in interpreting whether geographic differen- a physical barrier, the similar high levels of tiation is primary or secondary (Endler, 1977). polychromatic variation may have been de- Alternatively, intraspecific patterns of vari- rived independently or have been selectively ation may be due to minor or fluctuating de- maintained despite secondary isolation. 12 AMERICAN MUSEUM NOVITATES NO. 2811

In addition to extrinsic environmental fac- havior in Heliangelus. Similarity in these tors, biological attributes of the species also conditions across major river barriers may seem important for explaining the broader explain the lack ofcorresponding differentia- aspects of differentiation, including the non- tion in the polychromatism. Other patterns, concordant and nonclassical patterns of geo- such as the differentiation of populations in graphic variation. Adaptive intrapopulation several mensural and color characters be- color variation may result from the coarse- tween the eastern Andean chain and the Cen- grained (sensu MacArthur, 1972) utilization tral and Western Cordilleras, could be proxi- of the environment by these small-bodied mally caused by isolation across intervening homeotherms; hummingbirds are known to river valleys. carefully select microhabitats for the purpose These speculations still require confirma- of thermoregulation and foraging (Gass and tion or rejection from field studies on the Montgomerie, 1981). Moreover, when the function ofplumage coloration in humming- sexes have very different ecological and de- birds. However, the population structure of mographic characteristics, one would expect Heliangelus exortis exortis demonstrates the this to produce sex-limited patterns of vari- need to expand explanations of geographic ation at several levels. The prevalence ofpro- differentiation beyond the role of extrinsic nounced sexual dimorphism in size and color environmental features, to include hypothe- in hummingbirds indicates that each sex is ses based upon the biological properties of under some different selection pressures. the organism and the biological function of These differences seem likely to produce dif- specific characters. ferent patterns ofgeographic variation in the two sexes as well. The effectiveness of barriers also depends TAXONOMIC CONCLUSIONS upon intrinsic biological properties (Mayr, I consider Heliangelus e. exortis to com- 1963). Birds are generally considered to be prise a single taxon. I recognize no new taxa more mobile than other vertebrates but there based upon changes in the frequency of dif- is much variation in sedentariness among dif- ferent female color classes, even when only ferent groups. Some of the most dramatic one class is present, since color classes are examples of microgeographic variation, in- morphologically similar over broad areas. I cluding changes in the frequency of intra- agree with Chapman (1917) and regard H. e. population color variants, occur in several soderstromi as a junior synonym of H. e. ex- families of birds in which nectar is an im- ortis since the specimen on which soder- portant component of the diet (the Coerebi- stromi is based is obviously an iridescent fe- dae, Wunderle, 1981; the Zosteropidae, Gill, male color form. The white under-tail coverts 1973). Thus, nectar-feeding habits may be ofthis suggest its affinity with specimens particularly conducive to geographic differ- from eastern Ecuador. Though it also has a entiation over short distances, such as is also rather short bill [table 3(B)], this is probably seen in H. e. exortis. due to its immaturity. One of the two im- Females of hummingbird species with matures from Planchas differs from males and characteristically bright plumage in both sexes females of other localities in its scaly throat ("bright monomorphic") differ from dull- feathering. This difference may be taxonom- colored females of dichromatic species in ically significant. However, more material is being territorial exploiters of nectar during needed to characterize this population and the nonbreeding season (Wolf, 1975). I have verify the proper sex of this specimen (fe- argued elsewhere that feeding territoriality in male?). females of bright monomorphic humming- Chapman (1917) noted the variation in birds suggests that females with plumage like nonglittering throat feathers but thought that males in polychromatic species are also ter- the bicolored condition indicated immatu- ritorial (Bleiweiss, 1983). Thus, geographic rity. My data indicate that color differences variation in the polychromatism may reflect in the nonglittering throat feathers and un- corresponding variation in floral resources or der-tail coverts exist among adult birds, and ecological competitors that affect foraging be- vary in a regular geographic pattern, with bi- 1985 BLEIWEISS: HELIANGEL US EXORTIS EXORTIS 13 colored feathers found in populations of the Bleiweiss, Robert Western and Central Cordilleras and white 1983. Variation and evolution in some An- feathers in the Eastern Cordillera and the east dean hummingbird genera. Ph.D. the- Andean slope of Ecuador. The similar color sis, Harvard Univ., Cambridge, Mas- variation from white to bicolored in both sachusetts. MS. Iridescent polychromatism in a female nonglittering throat feathers and under-tail hummingbird: Is it related to feeding coverts suggests that these characters are in- strategies? Submitted MS. terdependent, so their concordant pattern Chapman, Frank M. does not seem sufficient to warrant taxonom- 1917. The distribution of bird-life in Colom- ic recognition in the absence of other highly bia: a contribution to a biological survey correlated variation. More sampling is re- of South America. Bull. Amer. Mus. quired from the critical southern section of Nat. Hist., vol. 36, x + 729 pp. the Central Cordillera of Colombia to deter- 1926. The distribution ofbird-life in Ecuador. mine the geographic pattern where bicolored A contribution to the study ofthe origin are in apparent contact. of Andean bird-life. Bull. Amer. Mus. and white forms Nat. Hist., vol. 55, xiii + 784 pp. Endler, John A. 1977. Geographic variation, speciation and APPENDIX 1: SPECIMENS clines. Princeton, New Jersey, Prince- EXAMINED ton Univ. Press, ix + 246 pp. Colombia: Locality A: AMNH 121632-34, Fitzpatrick, John W., David E. Willard, and John 121636-39; ANSP 168270; LSUMZ 33621; W. Terborgh USNM 255902-03. Hacienda Zuliaba: USNM 1979. A new species of hummingbird from 425771-78. Santa Elena: AMNH 133159-60. Par- Peru. Wilson Bull., vol. 91, pp. 177- amo de Sonson: USNM 436231-33, 436235-39. 186. Locality B: AMNH 111686, 111688; ANSP Gass, Clifton Lee, and Robert D. Montgomerie 73749-50; CAR 70320-23, 70367-68, 70372-73, 1981. Hummingbird foraging behavior: deci- 70381-84, 70396, 70437. Laguneta: AMNH sion-making and energy regulation. Pp. 111658, 111660, 111664, 111668-72, 111675, 159-194. In Kamil, A. C., and T. D. 111678-84; ANSP 153270-99, 153301-05, Sargent (eds.), Foraging behavior: eco- 153360, 153556, 153874; USNM 255905-06, logical, ethological, and psychological 255908-09. Toche: ANSP 153306-15, 153317- approaches. Garland STPM Press, 543 19. Purace: USNM 389240. Coconuco: ANSP pp- 142594. Paletara: USNM 389242-43. Tijeras- Gill, Frank B. Moscopan: USNM 389241,446299-301,446230. 1973. Intra-island variation in the Mascarene Almaguer: AMNH 115940, 115942-47, 185040, White-eye, Zosterops borbonica. A.O.U. 255904. Hacienda La Ilusi6n: USNM 436240-46, Monograph, no. 12, vi + 66 pp. 436247-48. Locality C: AMNH 109492-98; ANSP Gill, Frank B., F. J. Stokes, and C. Stokes 129814, 142592-93, 144754-58; CAR 137296- 1973. Contact zone and hybridization in the 98, 137327; LSUMZ 38574, 44665-66; USNM Jamaican hummingbird, Trochilus po- 446225-28. Locality D: ANSP 161991-93, lytmus (L.). Condor, vol. 75, pp. 170- 164796, 164798-802, 164804-10. 176. Ecuador: Locality E: AMNH 163096, 17597 1- Goodfellow, W. 73, 247236-37; USNM 174013-17, 180105-06. 1901. Results of an ornithological journey Volcan Sumaco: AMNH 175970, 185031-39, through Colombia and Ecuador. Ibis, 185041. Locality F: AMNH 35540, 437688, vol. 8, pp. 300-319. 483691; ANSP 145523-32; LSUMZ 33621. Plan- Graves, Gary R. chas: LSUMZ 82902-03. Corazon: USNM 1982. Speciation in the Carbonated Flower- 174008. piercer (Diglossa carbonaria) complex in the Andes. Condor, vol. 84, pp. 1- LITERATURE CITED 14. Greenewalt, Crawford H. Baldwin, S. Prentiss, Harry C. Oberholser, and 1960. Hummingbirds. Garden City, New Leonard G. Worley York, Doubleday and Co., Inc., x + 250 1931. Measurements of birds. Cleveland, PP. Ohio, Sci. Publ. Cleveland Mus. Nat. MacArthur, Robert H. Hist., vol. 2, ix + 165 pp. 1972. Geographical ecology. Patterns in the 14 AMERICAN MUSEUM NOVITATES NO. 2811

distribution ofspecies. New York, Har- Vuilleumier, Fran9ois per and Row, Publishers Inc., xviii + 1968. Population structure of the Asthenes 269 pp. flammulata superspecies (Aves: Fur- Mayr, Ernst nariidae). Breviora, no. 297, pp. 1-21. 1942. Systematics and the origin of species. 1980a. Reconstructing the course ofspeciation. New York, Columbia Univ. Press, xiv + Act., XVII Cong. Inter. Ornith. Deutch. 334 pp. Ornith.-Gesell. pp. 1296-1301. 1959. Trends in avian systematics. Ibis, vol. 1980b. Speciation in birds of the high Andes. 101, pp. 293-302. Separatum Ex Act., XVII Cong. Inter. 1963. species and evolution. Cam- Ornith. Deutch. Ornith.-Gesell., pp. bridge, Massachusetts, Harvard Univ. 1256-1261. Press, xiv + 797 pp. 1984. Patchy distribution and systematics of Meyer de Schauensee, Rudolphe Oreomanes fraseri (Aves, ?Coeribidae) 1966. The species of birds of South America of Andean Polylepis woodlands. Amer. and their distribution. Narbeth, Penn- Mus. Novitates, no. 2777, pp. 1-17. sylvania, Livingston Publishing Co., Vuilleumier, Frangois, and Daniel Simberloff xvii + 577 pp. 1980. Ecology versus history as determinants Oberholser, Harry C. of patchy and insular distributions in 1902. Catalogue of a collection of humming- high Andean birds. Pp. 235-388. In M. birds from Ecuador and Colombia. Proc. K. Hecht, W. C. Steer, and B. Wallace U.S. Natl. Mus., vol. 24, pp. 309-342. (eds.), Evol. Biol., New York, Plenum Ortiz-Crespo, F. I. Press, vol. 12. 1972. A new method to separate immature and Wolf, Larry L. adult hummingbirds. Auk, vol. 89, pp. 1975. Female territoriality in the Purple- 851-857. throated Carib. Auk, vol. 92, pp. 511- Paynter, Raymond A., Jr., and Melvin A. Traylor, 522. Jr. Wunderle, Joseph M., Jr. 1977. Ornithological gazetteer of Ecuador. 1981. An analysis ofa morph ratio cline in the Cambridge, Massachusetts, Mus. Comp. Bananaquit (Coereba flaveola) on Gra- Zool., Harvard Univ., viii + 152 pp. nada, West Indies. Evolution, vol. 35, 1981. Ornithological gazetteer of Colombia. pp. 333-344. Cambridge, Massachusetts, Mus. Comp. Zimmer, John T. Zool., Harvard Univ., v + 311 pp. 195 la. Studies of Peruvian birds. No. 60. The Peters, James Lee genera Heliodoxa, Phlogophilus, Uro- 1945. Check-list ofbirds ofthe world. Vol. V, sticte, Polyplancta, Adelomyia, Coeli- Cambridge, Massachusetts, Harvard gena, Ensifera, Oreotrochilus, and To- Univ. Press, xi + 306 pp. paza. Amer. Mus. Novitates, no. 1513, Short, Lester L. pp. 1-45. 1969. Taxonomic aspects of avian hybridiza- 195 lb. Studies of Peruvian birds. No. 61. The tion. Auk, vol. 86, pp. 84-105. genera Aglaeactis, Lafresnaya, Pteroph- Sokal, R. R., and F. J. Rohlf anes, Boissonneaua, Heliangelus, Er- 1981. Biometry. San Francisco, California, W. iocnemis, Haplophaedia, Ocreatus, and H. Freeman. Lesbia. Ibid., no. 1540, pp. 1-55. Stiles, F. Gary 1952. Studies of Peruvian birds. No. 62. The 1983. Systematics ofthe southern forms ofSe- genera Patagona, Sappho, Polyony- lasphorus (Trochilidae). Auk, vol. 100, mus, Ramphomicron, Metallura, Chal- pp. 311-325. costigma, Taphrolesbia, and Aglaiocer- Vuilleumier, Beryl Simpson cus. Ibid., no. 1595, pp. 1-29. 1971. Pleistocene changes in the fauna and flora of South America. Science, vol. 173, pp. 771-780.