.'ROBERT'__K. SELANDER AND DONALD R. GILLER

BULLETIN OF-THE AMERICAN MUSEUM OF-NATURAL HISTORYi VOLUME 124,:ARTICLE 6NEWV YORK: 1963

SPECIES LIMITS IN THE WOODPECKER GENUS CENTURUS (AVES)

SPECIES LIMITS IN THE WOODPECKER GENUS CENTURUS (AVES)

ROBERT K. SELANDER Research Fellow, 1961, Department of Ornithology The American Museum of Natural History Associate Professor, The University of Texas Austin, Texas

DONALD R. GILLER Graduate Student, Department of Zoology The University of Texas Austin, Texas

BULLETIN OF THE AMERICAN MUSEUM OF NATURAL HISTORY VOLUME 124 : ARTICLE 6 NEW YORK 1963 BULLETIN OF THE AMERICAN MUSEUM OF NATURAL HISTORY Volume 124, article 6, pages 213-274, text figures 1-17, plates 53-56, tables 1-8

Issued March 4, 1963 Price: $1.50 a copy INTRODUCTION

THE PRINCIPAL OBJECTIVE of the present FIELD WORK AND MATERIALS study was to ascertain the systematic rela- tionships among populations of woodpeckers Field work was undertaken in M6xico in of the genus Centurus occurring in the United June and July, 1958, for the primary purpose States, Mexico, and northern Central Amer- of studying the relationship of Centurus auri- ica. Because the distributional pattern of frons and C. uropygialis in Aguascalientes and species of this complex is strongly allopatric Jalisco; but some time was devoted to C. and the morphologic criterion of relation- chrysogenys in Colima and Nayarit, C. hypo- ship is notably unreliable, considerable con- polius in Puebla, and C. aurifrons in Coa- troversy as to species limits has developed huila, San Luis Potosi, and Tamaulipas. Our among taxonomists working with museum investigations are based on 231 specimens specimens alone. But recent field investiga- collected in Mexico in 1958, all of which have tions in zones of contact between allopatric or been deposited in the American Museum of narrowly sympatric forms, together with Natural History, and on material obtained on analyses of morphologic variation in speci- loan from the following institutions: mens, have yielded important information The American Museum of Natural History that contributes to our understanding of (A.M.N.H.), through the courtesy of Dr. systematic relationships within the genus. Dean Amadon and Mr. C. E. O'Brien; British The present report is primarily concerned Museum (Natural History) (B.M.N.H.), with species of Centurus occurring in the through Dr. R. W. Sims; Chicago Natural United States and in northern and central History Museum (C.N.H.M.), through Dr. Mexico, namely, C. carolinus, C. aurifrons, C. M. E. Traylor; Donald Dickey Collection uropygialis, C. hypopolius, and C. chryso- (University of California, Los Angeles) genys; but, in addition, we have treated C. (D. C.), through Dr. T. R. Howell; Museum hoffmanni of Costa Rica and Nicaragua and of Zoology, Louisiana State University have included systematic comments on sev- (M.Z.L.S.U.), through Dr. R. J. Newman; eral other forms referred to the genus. The Museum of Zoology, University of Michigan geographic distribution of species considered (M.Z.U.M.), through Dr. R. W. Storer; in detail in this report is indicated in figure 1, Museum of Vertebrate Zoology, University of and specimens of several species are shown in California, Berkeley (M.V.Z.), through Dr. plates 53 to 55. A. S. Leopold and Dr. F. A. Pitelka; and Research on Centurus was supported by the United States National Museum (U.S.N.M.), National Science Foundation (Grant 3448) through Dr. H. Friedmann. and the Research Institute of the University In addition, specimens in the Museum of of Texas (Grant URI-871). The study was Comparative Zoology, Harvard University, completed in 1960, while the senior author and the Peabody Museum, Yale University, was a Research Fellow in the Department of were examined in October, 1960, through the Ornithology of the American Museum of courtesy of Mr. J. C. Greenway, Jr., and Dr. Natural History. S. D. Ripley, respectively.

217 i ) ill I.."I I,

L.%

0a}

c)Io 04C)

0 a)U) en 04

*wn 0 0U) *I- 04

,04.4 GENERIC LIMITS

THERE IS LITTLEAGREEMENTamOng systemat- TABLE 1 ists as to the number of genera in which WING/TAIL RATIOS IN MALES of Centurus species of the Centurus-Tripsurus-Melanerpes complex are to be included. In many early Population Wing/Tail' classifications, as well as in some of recent date (Todd, 1946), three genera are main- C. hoffmanni 2.03 tained, but Peters (1948) has lumped all three C. aurifrons santacruzi under the name Melanerpes. Other workers SE. Honduras-Nicaragua have recognized two genera, combining Trip- (33)b 1.68 surus with Centurus (Mexican Check-List El Salvador (32) 1.75 Committee, 1957, p. 36) but maintaining NW. Honduras (31) 1.76 S. Guatemala (30) 1.82 Melanerpes, and still others (e.g., Wetmore, Central Guatemala (29) 1.76 1957, p. 51) unite Tripsurus and Melanerpes SE. Chiapas (22) 1.77 but regard Centurus as distinct. Mean, samples 22, 29-33 1.76 In the generic diagnoses of Tripsurus and C. aurifrons dubius Centurus prepared by Ridgway (1914), there Mean, samples 23, 25 1.65(1.62-1.69)0 is no character or combination of characters C. aurifrons grateloupensis providing clear distinction between the two Mean, samples 12, 15, 16 1.69(1.68-1.70) groups of species. The tail is supposedly two- C. aurifrons aurifrons thirds as long as the wing in Centurus and less Mean, samples 3, 5, 10, 14 1.71(1.69-1.75) than one-half as long in Tripsurus, but, in C. aurifrons polygrammus fact, the tail is slightly less than one-half as Mean, samples 18-21 1.74(1.67-1.85) C. carolinus (Texas) 1.64 long as the wing in C. hoffmanni and C. rubri- C. uropygialis capillus; and it is slightly more than one-half Sonora 1.58 as long in T. pucherani (see table 1). The Nayarit-Sinaloa 1.60 extent of feathering of the orbital region is Jalisco-Aguascalientes 1.63 another feature purportedly distinguishing Zacatecas 1.70 the two groups: in Tripsurus this region is C. hypopolius (Puebla) 1.56 bare, and in species of Centurus it is suppos- C. chrysogenys (Nayarit) 1.71 edly well feathered. However, as acknowl- C. pygmaeus edged by Ridgway himself, this region is fully Yucatfin 1.70 as bare in C. chrysogenys as in T. pucherani. Cozumel Island 1.62 in Bonaca Island 1.72 Moreover, it is almost as extensively bare C. rubricapillus C. rubricapillus and C. superciliaris, and an Panami 2.11 intermediate condition is seen in C. hypopo- Isla El Rey 2.12 lius. Similarly, the structure of feathers of the C. pucherani (Nicaragua) 1.92 malar apex is variable in both genera and C. chrysauchen (Costa Rica) 2.04 cannot serve as a diagnostic character. Nor do the shape and size of the nostril (sup- a Calculated from mean wing and tail lengths of posedly broadly oval in Centurus and oval or samples. ovate and smaller in Tripsurus) provide a b Numbers in parentheses refer to sample areas in- dicated in figure 2. basis for separation. a Wetmore (1957) argues for the mainte- Range. nance of Centurus as a genus distinct from pes. He fails, however, to point out that these Melanerpes (in which he apparently includes characters will not permit separation of Trip- Tripsurus) on the basis that the regularly surus and Centurus. barred back and the soft and "blended" In urging that "such apparently well- breast feathers of Centurus contrast with the characterized groups" as Centurus and Trip- uniformly colored back and the hair-like surus should not be combined with Melaner- feathers of the breast in "typical" Melaner- pes, Todd (1946, p. 297) suggests that "more 219 220 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 124 weight should be given to color pattern." But that (1) both the adult male and female usu- when we compare the species assigned to ally sleep in the nest with their eggs or young Centurus and to Tripsurus, it is soon apparent (rarely roosting alone in separate cavities); that similarities in color and pattern between (2) the nest is kept clean until the young take Centurus chrysogenys and Tripsurus pucherani wing and later; and (3) fledglings return to actually provide the strongest argument for roost in the nest cavity with their parents and combining the two nominal genera. Speci- may continue this habit until the approach of mens of these two woodpeckers are shown in the next breeding season. plate 54. Similar features are: (1) dorsal bar- Although Skutch's observations are of con- ring of black and white; (2) extensive black siderable interest and should be extended to on side of head (in males of C. chrysogenys other woodpecker species, there is reason to this color is limited to the superciliary region, doubt that the reported differences in life whereas it also extends over the postorbital history of Central and South American spe- region in T. pucherani; in females of both cies of Tripsurus and Centurus are in any species, the black passes over the occipital sense "fundamental" or of unusual signifi- region as a band, which is wider in Trip- cance from a systematic standpoint. Also, we surus); (3) yellow frontal region; (4) naked, call attention to the fact that Skutch's con- black orbital region; (5) dark breast and clusions in regard to generic classification throat; and (6) barred flanks, the bars being were based on observations of only a few darker and more extensively distributed individuals representing less than half of the anteriorly in Tripsurus. total number of species assigned to these The chief difference in color between these genera. Hence, the degree of constancy of species involves the presence of yellow on the these differences in the two groups of species malar region and chin in C. chrysogenys and remains to be determined, and we note that its absence in T. pucherani, but this is a trivial at least one of the traits is subject to individ- difference, for the character is subject to ual variation within a single species. Thus, for geographic variation in C. chrysogenys and is, T. chrysauchen, Skutch reports several in- in fact, almost absent in birds from Morelos stances in which the female of a pair roosted (R. T. Moore, 1950, p. 109). The amount of in a cavity apart from that in which the male yellow on the malar region and chin also incubated and brooded at night, and he also varies individually within wide limits in found a female of T. pucherani roosting alone another species, C. pygmaeus (see p. 255). in a separate cavity while the male of the pair Skutch (1943, 1948) has argued against incubated. On this account, we are reluctant combining Tripsurus and Centurus on the to attach much taxonomic significance to basis of "fundamental" differences in certain roosting locations. aspects of life history. In his studies of C. Differences in nest sanitation are, as rubricapillus, C. hoffmanni, and C. aurifrons Skutch points out, related to the fact that the in Central America, Skutch notes (1948, pp. nesting cavity of the species of Tripsurus 257-258) the following features of life history studied is used as a family "dormitory" contrasting with those of Tripsurus: (1) the (Skutch, 1961a) by the parent birds and male alone attends the nest at night, the fledged young of the species of Tripsurus female roosting elsewhere; (2) the nest is kept studied but not by those of Centurus. There- clean during incubation and the early part of fore, the basic question concerns the degree of the nestling period, but fecal matter is often taxonomic significance to be attached to dif- allowed to accumulate as the nestlings grow ferences in the length of time that fledglings older and the parents no longer enter the nest remain with their parents. Since the time of cavity to feed them; and (3) in C. rubricapil- dispersal of the young varies among species of lus, the young are not led back to roost in the some other genera of birds, it seems reason- nest after they fledge. (Information on this able to suspect that it may have no particular aspect of behavior is not available for C. hoff- value as a taxonomic character at the generic manni and C. aurifrons.) level in woodpeckers. For example, in wrens By way of contrast, in T. pucherani, T. of the genus Campylorhynchus (Selander, cruentatus, and T. chrysauchen, Skutch finds MS), the young of some species (e.g., C. brun- 1963 SELANDER AND GILLER: CENTURUS 221 neicapillus) disperse from the parental terri- ably placed in Centurus by taxonomists who tory soon after fledging, while in other species maintain Tripsurus is obviously more closely (e.g., C. zonatus) they remain for more than a allied to T. pucherani than to any species year, being associated in family groups with included in Centurus. If the genus Tripsurus their parents and assisting, as helpers at the is to be maintained, the species chrysogenys nest, in rearing broods in the next breeding must be transferred to it from Centurus. Even season (Skutch, 1935, 1961b). An intermedi- if such transfer were made, however, further ate condition is seen in Campylorhynchus difficulties arise in the classifying of the spe- rufinucha, in which, as in the species of Trip- cies hypopolius, which, like chrysogenys, surus studied by Skutch, the young of the bridges the gap between "typical" Tripsurus year disperse at the beginning of the next and "typical" Centurus. It has invariably breeding season. Interspecific variation in the been assigned to Centurus, but, as noted be- time of dispersal of the young is also found in low, it shows a resemblance to chrysogenys in jays of the genus Aphelocoma (Pitelka, plumage pattern, in relative reduction in size 195 la). In birds, the time of disbanding of of the hyoid apparatus, and in some other family groups and dispersal of young is features. At least pending comprehensive presumably adaptively related to structure studies of internal anatomy and further in- and other aspects of the dynamics of the vestigations of life history and behavior, we populations, but little is known of the nature see no course but to combine the two genera; of this relationship or of the selective forces Tripsurus Swainson (1837) is regarded as a that determine the time of dispersal. synonym of Centurus Swainson (1837). Skutch (1948, p. 256) reports seeing indi- We have nothing to offer regarding the viduals of T. cruentatus in Ecuador give a problem of the relationship between Centurus "magnificent exhibition of aerial flycatching, and Melanerpes, except to note that, in the in their strong flight and intricate maneuvers series C. pucherani-pulcher-chrysauchen-flavi- rivalling the Neotropic Kingbirds (Tyrannus frons, there is progressive simplification of mekancholicus) which were engaged in the plumage pattern that involves the reduction same occupation at the same time." This of dorsal barring and the extension of red on observation is of special interest, since Mel- the belly. This trend leads finally to C. anerpes erythrocephalus frequently feeds in cruentatus (including "rubrifrons," which is this fashion, while species of Centurus in probably a color phase of cruentatus rather Mexico and the United States do little fly- than a distinct species; Haverschmidt, 1957), catching. In an earlier paper (Selander and a species resembling in several respects Giller, 1959b) we suggested that the rela- Melanerpes portoricensis of Puerto Rico. In tively long wing of M. erythrocephalus is an C. cruentatus, the only vestiges of bars are adaptation for stronger flight in connection light smudges on the outer webs of the outer with aerial feeding, and it now appears that rectrices and a few light bands at the tips of the same may be true for species of Tripsurus. the inner secondaries. This trend in variation, Skutch (1948, p. 226) mentions that T. chry- together with the well-known similarity of sauchen captures insects on the wing, rising pattern of the juvenal plumage of Melanerpes above treetops and twisting and looping in erythrocephalus to that of the juvenal and the air, and Slud (1960, p. 84) reports fly- adult plumages of Centurus species (Amadon catching by T. pucherani in Costa Rica. At and Eckelberry, 1955, p. 74) suggests a close present we do not know whether these species phyletic relationship between the groups of indulge in this habit more frequently than species presently classified in these separate does C. rubricapillus or other species of Cen- nominal genera, but we are reluctant to lump turus, although this seems likely. the two until a more thorough study of the In sum, one species (chrysogenys) invari- situation has been made. GEOGRAPHIC VARIATION IN CENTURUS A URIFRONS (WAGLER)

MORPHOLOGIC VARIATION and systematic nized and specimens showing intermediacy of relationships among populations currently characters were regarded as interspecific assigned to the species C. aurifrons have been hybrids. considered in brief reviews by Ridgway (1881, Ridgway's second classification was ac- 1914), Griscom (1932), and Wetmore (1943). cepted by Cory (1919) and many other au- The principal problem concerns the relation- thors, but, in 1932, Griscom, examining large ship of populations of red-naped and nar- series of specimens, reached conclusions rowly barred birds (dubius) occurring on the almost identical to those originally stated by Yucatin Peninsula and yellow-naped and Ridgway in 1881, and a study by Wetmore broadly barred birds of the Mexican Plateau (1943) later supported these findings. None- and Texas (aurifrons) and the arid regions of theless, some authors (Stone, 1932; Bond, the Isthmus of Tehuantepec and interior 1936; Sutton, 1940; and Brodkorb, 1943) Chiapas (polygrammus). In addition, there is continued to hold the more conservative view the related problem of the systematic status which maintains dubius and santacruzi as of the closely allied Costa Rican and Nicara- specifically distinct from aurifrons, and, as guan form hoffmanni, which is considered recently as 1951, Sutton again expressed the below. opinion that dubius and aurifrons are separate Working with a total of 62 specimens, species. Meanwhile, several additional races Ridgway (1881) correctly determined that were described, and in 1948 Peters recognized two of the extreme types, aurifrons and du- a total of 12 subspecies of C. aurifrons (in bius, are connected by intergradation of char- addition to hoffmanni, which was also re- acters in specimens from Veracruz and Gua- garded as a race of C. aurifrons). temala (santacruzi of the older literature), Nothing would be gained at the present and on this basis he considered the whole time by a review of the nomenclatural history complex to be a single species, relegating the of the aurifrons complex or by a discussion of nominal species dubius and santacruzi to races the characteristics and distributions of the of aurifrons. However, this essentially mod- named races. Therefore, we propose for the ern treatment subsequently was abandoned moment to pass over the matter of subspecies by Ridgway (1885a, 1885b, 1887, 1888, and and, instead, to examine individual and geo- 1914) in favor of a more conservative scheme graphic variation in size and several charac- of classification similar to that adopted by ters of color and pattern in order to determine Hargitt (1890) and Salvin and Godman the extent of intergradation among these (1888), in which four species, aurifrons, poly- markedly differentiated, allopatric popula- grammus, dubius, and santacruzi, were recog- tions. METHOD OF ANALYSIS OF VARIATION Within the total geographic range of the sent one category of belly color and one cate- aurifrons complex, 34 sample areas have been gory of frontal region color, but all other designated, as shown in figure 2. Our samples reference specimens were C. aurifrons. The include specimens from all parts of the range number of categories represented by reference except Roatin, Barbareta, and Utila Islands. specimens varied with the character, as fol- To permit quantitative analysis of variation lows: in color and pattern, we first determined the NAPE COLOR: Fourteen categories, from limits of variation in these characters, then yellow (category 0) to red (category 13). selected series of reference specimens (listed FRONTAL REGION COLOR: Eight categories, in the Appendix) representing equally spaced from yellow (0) to red (7). steps or segments in total spans of variation. BELLY COLOR: Twelve categories, from Specimens of C. hoffmanni were used to repre- yellow (0) to red (11). In analyzing variation 222 0 cn to C: CZ Q k- -th st: , Z3ek ..5 :34 %. %z -i?

ceo .4 0 *

o0

Cd ce

~ce

.U)

223 224 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 124 in this character, we considered only the hue based almost entirely on adult males. The of the color, not the value, intensity, or ex- variation in females follows closely that tent of distribution in the abdominal region. shown by males, there being no significant DORSAL TAIL PATTERN: Seven categories, sexual dimorphism in color or pattern except representing a range from maximum white in the markings of the head. The female lacks barring and streaking (0) to black with no the red coronal patch of the male, and, in all white marks (6). populations in which the nape of the male is Widths of white and black bars were meas- not pure red, the nape color of the female is ured in the center of the dorsal region. A purer yellow than that of the male. There is a number of standard linear measurements moderate degree of sexual dimorphism in size, were made from the skins, including the arc of with females averaging smaller than males in the unflattened wing and bill length from the all linear dimensions. anterior edge of the nostril to the tip of the There are no appreciable age differences in bill. In addition, notes were taken on the plumage or size following the postjuvenal color of the breast and flanks and on several molt, which involves complete replacement of other geographically variable characters of the juvenal plumage. Because none of the color and pattern for which quantitative species of Centurus is migratory, we have been treatment was not attempted. able to use in our studies specimens taken at Our analysis of geographic variation is all seasons of the year.

VARIATION IN CHARACTERS Individual and geographic variation in changes seen in nape color. In Texas and over characters of color and pattern is shown in the Mexican Plateau, the red coronal patch is figures 3 through 7, and data on size are pre- relatively narrow and is separated from the sented graphically in figures 8 through 11. In yellow or orange nape by a broad band of all figures, numbers following locality names gray. In southern Tamaulipas and in central refer to sample areas designated in figure 2. and eastern San Luis Potosi, the patch aver- Available weights are presented in table 2. ages wider and usually has at least a narrow NAPE COLOR: On the Mexican Plateau and central connection with the nape. There is a in Texas, the nape of males is yellow or much greater degree of confluence between orange, with mean sample scores ranging colored areas in material from Veracruz, but from 2.5 in northwestern San Luis Potosi many specimens have small gray patches in (sample 10, fig. 3) to 5.9 in northern Coahuila the occipital area between the crown and the (sample 6). Individual variation is relatively nape. In birds from Tabasco, the Yucatan great, as it is in other populations having a Peninsula, and Central America, the two yellowish nape. Extending from northwestern colored areas are broadly joined and the coro- San Luis Potosi to Campeche (sample 24), nal patch is also relatively wide. there is a cline in which yellow (2.5) passes On the Pacific side of the Isthmus of Tehu- through orange to an almost pure red (12.4), antepec (samples 17 to 19) and in central and with the steepest gradient occurring between eastern Chiapas (samples 20 and 21), nape central San Luis Potosi (sample 11) and east- color is on the yellow side, being similar to ern San Luis Potosi (sample 12). The nape is that of birds from the Mexican Plateau. This red and shows little variability throughout development reaches an extreme in eastern the Yucatan Peninsula, but the color tends Chiapas (sample 21), where the color is al- toward orange, with some individuals having most pure yellow (0.4). As in other yellow- a distinctly orange nape, in Guatemala (sam- naped populations, the nape and coronal ples 28 to 30, and 34), southeastern Chiapas areas of the male are separated by a gray (sample 22), El Salvador (sample 32), and band. Honduras (samples 31 and 33). FRONTAL REGION COLOR: Variation in this Geographic variation in extent of connec- character parallels that of nape color, except tion between the highly colored coronal and in Central America (samples 22, 30, 31, 32, nape regions in males parallels the clinal and 33), where it shows a more marked shift 1963 SELANDER AND GILLER: CENTURUS 225

NAPE 10 DURANGO (7) | -1,,, 3.8 N.TEXAS (I) 44 8 12

BIG BEND (4) 4.5 _ 4.8 CENT. TEXAS (2)

FORT CLARK (5) 4.5 5.5 S. TEXAS (3)

N.COAHUILA (6) 5.9 6.8 & TAMPS. (9)

E. COAH..N.LEON (8) 4.4 L3 JALISCO (14)

NW S.L.R 10) 2.55- /2 AGUASCALIENTES (13)

CENT. S.L.P. (11) U1_55.7

E. S.LP. (12) 9.3 - 5.0 N. OAXACA (17)

CENT. VERACRUZ (15) _ 10.5 _ _r.4 S. OAXACA (18) - S, VERACRUZ (16) I.o0 5.6 SW. CHIAPAS (19)

TABASCO-NKCHIS. (23) 1 1.3 2.4 CENT. CHIAPAS (20)

CAMPECHE (24) 12.4 0.4 E. CHIAPAS (21)

YUC.-Q. ROO (25) _102.1 - - 9.2 W. GUATEMALA (34)

COZUMEL IS. (26) .- .._ 12.4 10.8 SE.CHIAPAS (22)

BRIT. HONDURAS (27) 12.4 I11.0 S. GUATEMALA (30)

N.GUATEMALA (28) 10.9 EL SALVADOR (32)

CENT GUAT. (29) _ 11.5k 1- 1.2 SE. HOND.-NIC. (33)

NW. HONDURAS (31) ^. 3.0 hoffmcnni 10 4 8-a 12 4 8 12 FIG. 3. Individual and geographic variation in nape color in males of Centurus aurifrons and Centurus hoffmanni. Categories of nape color are indicated in text; mean sample scores are shown at right of base lines of histograms. Numbers in parentheses refer to sample areas indi- cated in figure 2; arrows indicate geographic contact of populations sampled. 226 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 124 TABLE 2 WEIGHTS (IN GRAMS) OF Centurus aurifrons Males Females Samplea No. Mean Range No. Mean Range Central Texas (2) - - 1 90 S. Texas (3) 4 82.4 (79-85) 1 79 E. Coahuila-Nuevo Le6n (8) 1 86 Aguascalientes (13) 9 90.5 (81-98) 10 77.6 (67-84) Jalisco (14) 9 86.5 (77-96) 11 77.9 (66-83) NW. San Luis Potosi (10) 21 86.5 (73-99) 10 77.9 (70-90) Central San Luis Potosi (11) 8 82.6 (73-99) 4 72.8 (66-77) E. San Luis Potosi (12) 10 85.2 (69-102) 7 73.8 (65-79) Central Veracruz (15) 1 93 - - Tabasco-N. Chiapas (23) 2 81.5 (81-82) 1 72 Campeche (24) 1 85 - - - Yucatin-Quintana Roo (25) 3 81.1 (79-83) 3 70.2 (66-74) 9b 87.7 (81-95) 8b 76.4 (71-84) Cozumel Island (26) 2c 86.5 (84-89) 1I 79 British Honduras (27) 2 88.8 (85-93) N. Guatemala (28) 3 83.4 (82-85) 2 77.5 (77-78) Central Guatemala (29) - 1 80 - SE. Guatemala (30) ld 92 2 86.0 (85-87)

a Numbers in parentheses refer to sample areas indicated in figure 2. b Data from Paynter (1955, p. 165); specimens from Quintana Roo and Campeche. * Data from Paynter (1955, p. 166). d Data from Tashian (1953, p. 204). to orange and yellow scores (fig. 5). Again again, the purest yellow is found in birds from there is a cline from yellow (1.4) to red (6.3) eastern Chiapas (1.3). between northwestern San Luis Potosi and DORSAL TAIL PATTERN: Variation in dorsal Campeche, and the population of eastern tail pattern (fig. 5) is not correlated with that Chiapas represents an extreme in average of characters previously considered. In most expression of yellow (1.0). samples from the Mexican Plateau and BELLY COLOR: Variation in belly color (fig. Texas, well over half of the specimens fall in 4) follows that of the two characters previ- category 6, with no white markings present; ously discussed, with two notable exceptions. mean sample scores range from 5.5 to 6.0. First, in Texas and over the Mexican Plateau, This character is slightly more variable in San belly color varies from yellow to orange, with Luis Potosi, and most specimens from central an extreme on the yellow side being reached and southern Veracruz (samples 15 and 16) in western Texas (samples 4 and 5; scores, 1.0 show some white marks. From Tabasco and and 0.9) rather than in northwestern San Luis northern Chiapas (sample 23) southeast to Potosi. Second, the cline connecting the northern Guatemala, there is a return to an yellow-orange color of northwestern San Luis essentially unmarked condition (6.0). Then in Potosi (2.9) and the red (10.5) of Campeche Central America a shift to a moderate has its steepest gradient not in San Luis amount of white marking occurs, with mean Potos! but between central and southern sample scores ranging from 3.0 in southeast- Veracruz (samples 15 and 16), where the shift ern Chiapas (sample 22) to 5.5 in central is from an average score of 6.0 to one of 9.3. Guatemala (sample 29). The only populations The transition from red to orange in Central in which conspicuous white markings repre- America is roughly equivalent to that shown sent the usual condition are those of the in the color of the frontal region; and, once Pacific side of the Isthmus of Tehuantepec 1963 SELANDER AND GILLER: CENTURUS 227

BELLY

10 DURANGO t7,) 5 , ,,,,,,,, 2.4 1.6 N.TEXAS (I) 0 2 6 8S 'IO 01, 4 6 8 10

BIG BEND (4) *_ 1.0 1.4 CENT. TEXAS (2)

FORT CLARK (5) 0.9 2.0 S.TEXAS (3)

N. COAHUILA (6) 1.7 3.8 S. TAMPS. (9)

E.COAH.-N. LEON (8) 2.3 2.2 JALISCO (14)

NW. S.L.P (10) 2.9 2.4 AGUASCALIENTES (13)

CENT S.L.R (11) a3.0

E. S.LP (12) _ 4.2 4.0 K OAXACA (17)

CENT VERACRUZ (15) _ 6.0 3.3 S.OAXACA (18)

S. VERACRUZ (16) 9.3 3.6 SW. CHIAPAS (19)

n z TABASCO-N. CHIS. (23) 9.9 2.6 CENT CHIAPAS (20)

CAMPECHE (24) 10.5 1.3 E.CHIAPAS (21)

YUC-9. ROO (25) ] 10.5 3.2 W GUATEMALA (34)

COZUMEL IS. (26) 10.8 4.7 SE CHIAPAS (22)

BRIT HONDURAS (27) 10.8 5.3 S.GUATEMALA (30)

N. GUATEMALA (28) L 10.3 5.8 EL SALVADOR (32)

CENT GUAT. (29) 8.0 4o 4.0 SE. HOND.-NIC. (33)

NW. HONDURAS (31) , ,, , 5.7 4.9 hoffmann 0 2 4 6 8S '10 2 4 6 8 10

FIG. 4. Individual and geographic variation in belly color in males of Centurus auri- frons and Centurus hogmanni. Categories of belly color are indicated in text; mean sam- ple scores are shown at right of base lines of histograms. Numbers in parentheses refer to sample areas indicated in figure 2; arrows indicate geographic contact of populations sampled. 228 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 124

FRONTAL REGION TAIL

(7) 2.4 2.20 i) e7) , 5.6 5.6 (I) 024602.1.446 024d6 0 24 6 (4) i 1. 1.6 (2) (4) 5.5 5.8 (2)

(5) i 1.8 | 1.9 (3) (5) 5.5.959(3) _ I I (6) 2.3 1. 5 C9) (6) 5.7 6.0 (9)

(8 p 1.9 1.5 C14) (8) E .7 I5,9 14

(10) 1.4 1. 1.5 (13) (10) 5 5.7(13)

(11) 2.3 (11) _ 5.5

_ _2.7 (17) C12) 5.1 2.0 (17)

(15) ]... 4.8;.4A 2.4 (18) (IS) . 3.9 1.1 (18) (16) (12)_~~~~3.7t 5.8 L 1.6 (19) (16) . m 4.2 - 1.8 (19) (23) 6.0 1.8 (20) (23) 51 4 0.2(20)

(24) .L 6.3 1 1.0 (21) (24) l 5.7 1 0.0(21)

C25) 6 1 1.5 (34) (25) 6.0 MI 4.5 (34)

(26) 6.4 3.0 (22) (26) 1 6.0 3. 3.0 (22)

(27) 6.3 . 2.9 (30) (27) 6.0 M 3.9(30)

(28) 2 6.5 1 2.5 (32) (28) 1 6.0 h 3.5(32)

(29) X 5.0 L 3.3 (33) (29) - 5.5 3.3 (33)

(31) | 1.3 - 0.1 hoff. (31) , 4.7.1,. 0.9 hoff. 2 4 6 'P2 4 6 0 ,2 4 6 0 2 4 6 FIG. 5. Individual and geographic variation in frontal region color and dorsal tail pat- tern in males of Centurus aurifrons and Centurus hoffmanni. Categories of frontal region color and of dorsal tail pattern are indicated in text; mean sample scores are shown at right of base lines of histograms. Numbers in parentheses refer to sample areas indicated in figure 2; arrows indicate geographic contact of populations sampled. 1963 SELANDER AND GILLER: CENTURUS 229 and interior Chiapas (samples 17 to 21), and Yucatan Peninsula. it will be noted that all specimens in the With respect to variation in size in the sample from eastern Chiapas fall in category aurifrons complex, the following points seem 0. worthy of comment. DORSAL BARRING OF BLACK AND WHITE: WING LENGTH (FIG. 8): Among northern The widest black bars are found in specimens samples, wing length is greatest in northern from Texas and the Mexican Plateau, with a Texas (sample 1) but is almost equaled by maximum mean width of 3.94 mm. being samples from Jalisco, Aguascalientes, and reached in the sample from Aguascalientes northwestern San Luis Potosi. Wing length (fig. 6). Between central San Luis Potosi and decreases clinally from eastern San Luis southern Veracruz, there is clinal decrease in Potosi south along the Caribbean coast of width from 3.73 to 2.65 mm. A moderate in- Mexico to Tabasco and northern Chiapas crease in width is apparent in southeastern (sample 23) but returns to moderate size on Chiapas, El Salvador, and Honduras, with the Yucatan Peninsula and in Guatemala. It sample averages of 2.78 to 3.10 mm. Black is large again in El Salvador but even smaller bars are also moderately wide in samples from in southeastern Chiapas and in Honduras. In the Pacific side of the Isthmus of Tehuan- southern Oaxaca and interior Chiapas, wing tepec and interior Chiapas, where average length appears to vary clinally from 138.3 width ranges from 2.68 to 3.13 mm. mm. in southern Oaxaca (sample 18) to 133.0 The width of the white bars (fig. 7) exhibits mm. in eastern Chiapas (sample 21). a pattern of geographic variation roughly TAIL LENGTH (FIG. 9): This character is comparable to that of the black bars. Again less variable geographically than wing length. the most prominent bars are found in samples Again the maximum size is seen in northern from Texas and the Mexican Plateau, but the Texas, but the tail length is actually rather clinal shift to narrower white bars begins in uniform, averaging between 75 and 80 mm., northwestern San Luis Postosl rather than in in most samples from central and western eastern San Luis Potosi, as in the case of the Texas south to central Veracruz. Nearly black bars. Width increases slightly in south- maximum size is achieved on Cozumel Island, eastern Chiapas and El Salvador but returns with clines connecting this population with to a very narrow condition (1.13 mm.) in shorter-tailed populations to the north and northwestern Honduras. south. COLOR OF THE BREAST AND SIDES: The BILL LENGTH (FIG. 10): This character color of the under parts varies from light gray shows a pattern of geographic variation not in Texas and on the Mexican Plateau to a unlike that of wing length. Note that, whereas much darker olive or brownish olive-gray in birds from southern Oaxaca and interior Central America and on Cozumel Island, and Chiapas have wing lengths almost equaling samples from the Yucatan Peninsula and those of birds from Texas and the Mexican southern Veracruz are more or less inter- Plateau, in bill length they are distinctly mediate in this character. We note that the smaller. color is subject to considerable seasonal varia- BILL DEPTH: This was the least variable tion as a result of wear and soiling. linear dimension studied. The most notable The gray color of the forehead and anterior feature is the comparatively large size in superciliary region is no darker, and may in specimens from Cozumel Island and El Sal- fact be lighter, in the ventrally dark birds of vador. the Yucatan Peninsula and Central America TARSUS (FIG. 11): Variation in this charac- than in specimens from populations having ter exhibits a pattern similar to that of bill pale gray under parts, and these areas are length. almost white in some specimens from the 230 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 124

BLACK BARS

DURAINGO (7) 3.80 3.74 N.TEXAS (I) 2 a5 1 11 . I BIG BIEND (4) - - 3.70 3.86 CENT TEXAS (2)

3. -MNL FORT CLARK (5) 3.85 3.61 S.TEXAS (3) 'I_

N. COAHU)ILA (6) 3.75 - 3.68 S.TAMPS. (9)

E.COAH.-N. LEON (8) 3.87 _ 3.71 JALISCO (14)

NW. SIL.P. (10) 3.71 - 3.94 AGUASCALIENTES (13)

CENT S. L.P. (I1) 3.73

b E. S. L.P. (12) 3.22 _ t 2.70 N. OAXACA (17) CENT VERACFRUZ (15) _ 2.77 2.68 S. OAXACA (18)

S. VERACFRUZ (16) M 2.65_ tE 2.89 SW. CHIAPAS (19) TABASCO- CHiIS. (23) 2.65 3.13 CENT. CHIAPAS (20) /1 CAMPECIHE (24) _ 2.90 3.13 E. CHIAPAS (21)

YUC.-Q. R 00 (25) 2.76 U- 2.48 W. GUATEMALA (34)

COZUMEL IS. (26) _ 2.38 2.88 SE. CHIAPAS (22)

BRIT. HONDUR,AS (27) 2.62 /3 2.80 S. GUATEMALA (30)

N. GUATEMAILA (28)S 2.68 2.78 EL SALVADOR (32)

CENT. GUiAT. (29) * 2.45 - 2.68 SE. HOND.- NI C. (33)

mm v NW. HONDURIAS (31) . .. 3.10 3.36 hoffmanni 2.0 3.0 4.0 i.11I FIG. 6. Individual and geographic variation in width of dorsal black bars in males of Centurus aurifrons and Centurus hoffmanni. Mean measurements of samples are shown at right of base lines of histograms. Numbers in parentheses refer to sample areas indicated in figure 2; arrows indicate geographic contact of populations sampled. 1963 SELANDER AND GILLER: CENTURUS 231

WHITE BARS 5

DURANGO (7) I I I I 1 1 2.66 a s M 2.72 N. TEXAS I ii 1aa n| 1 I i i i i i n S (I) LO i-. 1.0 2 3.0 BIG BEND (4) 3.05 5__ 2.84 CENT. TEXAS (2)

FORT CLARK (5) 2.69 2.60 S TEXAS (3)

N. COAHUILA (6) 2.73 o 2.50 S. TAMPS. (9)

E.COAH.-N.LE6N (8) 2.61 2.05 JALISCO (14) - 0 k NW. S.L.P. (10) 1.98 1_ 2.28 AGUASCALIENTES (13)

CENT S.LP. (11) 2.17

E S.L.R (12) 1.95 Ma 1.70 N. OAXACA (17) t CENT VERACRUZ (15) 1.54 1.90 S. OAXACA (18)

S. VERACRUZ (16) 1.35 1.96 SW. CHtAPAS (19)

TABASCO-CHIS. (23) 1.37 2.39 CENT CHIAPAS (20) t CAMPECHE (24) ---i 1.20 2.30 E.CHIAPAS (21)

YUC.-Q ROO (25) 1.22 _/_ 188 W, GUATEMALA (34) COZUMEL IS. (26) J . 1.34 1.43 SE. CHIAPAS (22) BRIT HONDURAS (27) 1.35 _ 1.45 S. GUATEMALA (30)

N GUATEMALA (28) 1.35 1.68 EL SALVADOR (32)

CENT. GUATEMALA (29) 1.25 _ 1.28 SE.HOND.-NIC.(33)

NW. HONDURAS (31) 1.13 2.19 hoffmanni ' I l_ I I 1.0 .0 3.0 1. 6. 3.0 FIG. 7. Individual and geographic variation in width of dorsal white bars in males of Centurus aurifrons and Centurus hoffmanni. Mean measurements of samples are shown at right of base lines of histograms. Numbers in parentheses refer to sample areas indicated in figure 2; arrows indicate geographic contact of populations sampled. WING 115 120 125 130 135 140 145 1.5 130 135 140 145 I.. . Big Bond, W Texas(4) 1(2) Fort Clark Region($ _-I . (8) N. Texas (I) N. Coahuilo (6) 7 7 (5) Centrol Texas(2) S.Texas (3) E.Coahuila-NtLeon (8) Durango (7) S.Tamoullpas (9) ±i(4) NW SL Potos' (10) Jalisco (14) Central SL. Potos((II) 7-. Aguascalentes (13) E.&L Potosi (12) I Central Veracruz (15) S Veracruz (16) Tabasco-NChiapas (23) Campeche (24) uca4tl-Q Roo (25) (2), 1 (2) N. Oaxaca (17) Cozumel Is. (26) It1(5) S. Oaxaca (18) British Honduras (27) *;* I 2~~~~) (8) SW. Chiapas (19) NGuatemolo (28) I (6) (8) Centrol Chiopas(20) Central Guatemala (29) \ (P.)I (9) E, Chiopas (21) SEChiapas (22) El Salvador (32) I Honduras (31) -- Costa Rica-Nicaroguo (hoffmanni)

115 120 125 130 135 140 145 125 130 135 140 145 FIG. 8. Individual and geographic variation in wing length in males of Centurus aurifrons and Centurus hoffmanni. Diagrams show range, mean, standard deviation, and twice standard error of mean (solid rectangle). Geographic contact of populations is indicated by thin lines connecting means. TAIL 55 65 ,S5 9 0 5 8 9- Ng Bnd,WTesos(4) (2)

Fort Clark Region tS) 8I N. Texas (1) N. Coohuila (6) (5) Central Teuas(2) S Taos(3) E-Coohuilo- N.Le 18) - sI(5) Durango (7) SwTomoulipos(9) 1/(4) NWSL Potos( (10) Jollsco 04) Centraol SL Ftosi lln) Aguaolientes ()13 E.L Potosi (12) Central Veracruz(IS)/1 aVeracruz (16) Tobono-NChlkpos(23) Campeche (24) Wcot4n-Q.Roo(25) N. Oaxaca (17) Cozumel IL (26) | 5) .Oaxaca (18) (L.(5) British Hlaiduras (27) SW Chiapas (19) N.Guotenwlo (28) (8) Cetral Chiapas (20) Central Guotemal (2) (29) _. 1/ E Chiopas (21) SE.CNopos (22) El Solvodor (32) Honduros (31) Costa Rica-Nicoraguo (hoffmanni)

55 65 8 S 0 70 75 80 85 95 FIG. 9. Individual and geographic variation in tail length in males of Centurus aurifrons and Centurus hoffmanni. Diagrams show range, mean, standard deviation, and twice standard error of mean (solid rectangle). Geographic contact of populations is indicated by thin lines connect- ing means. BILL LENGTH 18.5 205 2,524.5 265 5 35 3255 25 W 32S llig Bend, W.Texaos (4) , I (2) Fort Clark Region (5) KTNpS (I) N. Coohuila (6) , Centrol Texas t) S,Texas (3) w-IX E,Coahuila- NAL.d (B) ~~~~~~~~~~~~~~~~~(5) Durango(7) STamoulipas (9) . K\ _ _ _ _ _L_ _ _ _ (5) NW SL Potosi (tO) Jalisco (14) Centrol SL. Potost(lI) Aguascolientes (13) E.SL Potosi (12) Control Verocruz (15) SVerocruz (16) Tobasco-N.Chiopos (23) Compeche (24) Yucaton-Q.Roo(25) ,4\___ (2) N. Oaxoco (17) Cozumel Is (26) >I (5) , S. Oaxoco (18) British Honduros (27) SW Chiopos (19) klGuatenolo (28) Central Chiopos (20) Control Guotenolo (29) . E Chlopos (21) SE Chiopos (22) El Solvodor (32) Honduros (31) Costo Rico-Nicoroguo 111 1 (hoffmanni) .-~~~~~~~--~~~---t/ 18.5 20.5 225 245 26.5 265 36.5 32S h. 2.5 {~~~~~~9)245 265 2=5 305 32S FIG. 10. Individual and geographic variation in bill length in males of Centurus aurifrons and Centurus hoffmanni. Diagrams show range, mean, standard deviation, and twice standard error of mean (solid rectangle). Geographic contact of populations is indicated by thin lines connecting means. TARSUS B 19 20 21 22 23 24 25 26 21 22 23 24 25 26 i I i i i i I Big Bend, WTexas(4) j7L (2) Fort Clark Region(5) - z1~ (8) N. Texas() N Coahuila (6) Central Texos(2) S Texos(3) '/ (5)5 E.Coohuio- NKLe(8) Durongo(7) S.Tamoullpas(9) NWS. Potos((l0) Jollsco (14) Central S.LPotost(ll) Aguascollentes(3) E.S.L. Potos((12) Centrol VeroruzllS) S.Veracruz (16) -\, ~ ~ ~ 9 (8) Tabosco-N.Chiopos(23) Campeche(24) l ~ ~ ~ I 9 Yucat&Q. Roo (25) N,Oaxaco (17) Cozumel IL (26) (3) S.Oaxaca (18) British Honduras(27) SW Chiapas(19) N.Guaemlo (28) Central Chiapos(20 Central Guatemala(29) EChiapos (21) SE.Chiopos (22) El Salvador (32) Honduras (31) Coaso Rica-Nicragwu (hoffmanni)

18 19 20 21 22 23 24 25 26 21 22 23 24 25 26 FIG. 11. Individual and geographic variation in tarsus length in males of Centurus aurifrons and Centurus hoffmanni. Diagrams show range, mean, standard deviation, and twice standard error of mean (solid rectangle). Geographic contact of populations is indicated by thin lines connecting means. 234 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 124 SUBSPECIES AND INTERGRADATION In recent years the validity of the sub- classification essentially similar to that origi- species concept has been seriously questioned nally proposed by Ridgway in his first review by a number of systematists and evolutionists of the species in 1881! (see discussions by Simpson, 1961, pp. 171- 176; and Inger, 1961). If the category of sub- species is to have a continuing function in Centurus aurifrons aurifrons (WAGLER) systematics, it is apparent that systematists To populations distributed from Texas must attempt to reach agreement on the south over the Mexican Plateau to Jalisco, definition of subspecies and the purpose the Michoacan, Mexico, Hidalgo, and north- category is to serve. As a result of our work western San Luis Potosi, we propose to apply with Centurus and certain other genera of the name C. a. aurifrons (Wagler, 1829, col. birds, we doubt the justification of the use of 512), with type locality at Ismiquilpam, subspecific names as a "tool" in "describing" Hidalgo. These populations are characterized geographic variation in single characters or by yellow to orange nape, frontal region, and minor morphologic differences among popu- belly; pale under parts; an absence (or at best lations. In our view, one proper function of slight development) of white markings on the the trinomial is to indicate a known or pre- tail; broad to moderately broad black and sumed conspecific relationship between allo- white barring dorsally; and relatively long patric populations that are sufficiently well wing and bill. differentiated in a number of morphologic, The name C. a. incanescens was proposed by ecologic, physiologic, or behavioral char- Todd (1946, p. 298) for birds of the Big Bend acters as reasonably to suggest that they are area of Trans-Pecos Texas, with type locality to some meaningful degree semi-independent 12 miles south of Marathon, Brewster evolutionary units not distantly removed County, and the range of this race was subse- from the species level of evolution. Ideally, quently extended by Wetmore (1948) to we should like to see the subspecific category include central Texas. Characters by which used for the organization of geographic popu- Todd sought to distinguish this race from C. lations of a species into phylogenetic units, as a. aurifrons are: "underparts paler gray ... ; suggested by Lidicker (1960, p. 161), but crown also paler gray; forehead paler yellow; whether this can be accomplished for any and nape orange rufous .. . to aniline yellow large number of species remains to be deter- [versus ochraceous orange to Mars yellow in mined. In any event, used in the manner birds from the Mexican Plateau]; and yellow suggested, the subspecific category at least of underparts paler." serves the practical function of preventing Comparing moderate numbers of speci- morphologically or otherwise well-marked mens from western and central Texas with a populations of a species from being elevated series from the Mexican Plateau, Wetmore nomenclaturally to the species level, with the (1948) found that the color of the frontal resultant obscuring of their true evolutionary region, nape, and belly was not significantly relationships with populations from which different, a conclusion fully supported by our they are not fully isolated by intrinsic bar- analysis of these characters (figs. 3-5). How- riers to genetic interchange (Mayr, 1954). In ever, it was Wetmore's opinion that C. a. in- the final analysis, such use may prove to be canescens could be maintained as a race on the the only justification for trinomials. basis of slightly paler gray under parts, nar- A dozen subspecies of Centurus aurifrons rower black and wider white dorsal barring, are currently recognized (Peters, 1948), but, and an average difference in number of black considering the major features of geographic markings on the upper tail coverts. We have variation in this species, we believe that, at not analyzed variation in the last of these least as far as the continental populations are characters, but it was evident from examina- concerned, only four names should be applied. tion of our material that this is merely an Thus, we find ourselves in the unusual posi- average difference, as previously noted by tion of advocating a return to a scheme of Wetmore (1948). As shown in figure 6, the 1963 SELANDER AND GILLER: CENTURUS 235 width of the black bars actually shows little Centurus aurifrons grateloupensis (LESSON) geographic variation from northern Texas As indicated above in our discussion of south to Aguascalientes and Jalisco. It is true, patterns of variation, the characters of the however, that birds from northern and cen- strikingly marked races C. a. aurifrons and C. tral Texas have wider white bars, on the a. dubius (as herein defined) intergrade com- average, than do specimens from the southern pletely in clinal fashion over an extensive part of the Mexican Plateau (see fig. 7), and area from central San Luis Potosi south to they tend to be paler gray ventrally. But in southern Veracruz. Largely because it is con- our opinion these differences do not constitute venient to have a name by which to refer to grounds for formal racial separation. these intergradient populations, we follow general practice in recognizing C. a. grate- Centurus aurifrons dubius (CABOT) loupensis (Lesson, 1839, p. 41), with type locality "Mexico," but, contrary to general To Centurus a. dubius (Cabot, 1844, p. practice, we would apply this name to all 164), with type locality at Uxmal, Yucatan, populations from central San Luis Potosi we assign populations from Tabasco, north- south to southern Veracruz. Wetmore (1943) ern Chiapas, and northern Guatemala north and others recognize a second race, C. a. verae- through the Yucatin Peninsula to Cozumel crucis (Nelson, 1900, p. 259), with type local- Island. Thus defined, the race C. a. dubius is ity at Coatzacoalcos, Veracruz, but we prefer characterized by red nape, with the coronal to follow Griscom (1932) in considering it a and nape regions confluent in males; red synonym of C. a. grateloupensis. frontal region and belly; dorsal tail pattern with few if any white markings; moderately dark under parts; and narrow black and Centurus aurifrons santacruzi BONAPARTE white dorsal barring. In size there is consider- Our analysis of characters has demon- able variation among populations assigned to strated considerable geographic variation in this race. populations distributed from southeastern Birds from Cozumel Island, Yucatan, are, Chiapas east through southern and central with some justification, generally distin- Guatemala to El Salvador and Honduras. guished racially as C. a. leei (Ridgway, 1885a, This variation is especially evident in linear p. 22). In the characters of color and pattern dimensions, width of dorsal bars, and frontal we have studied in detail, birds from Cozumel region color, which ranges from yellow (mean Island are scarcely different in any degree sample score, 1.3) in northwestern Honduras from those of the adjacent mainland popula- to orange (5.0) in central Guatemala. How- tions, but most specmens from the island are ever, dorsal tail pattern and color of the nape darker ventrally, and, as noted by Ridgway and belly are rather stable throughout Cen- (1914), their upper tail coverts tend to be tral America. It is noteworthy that Central more heavily marked with black. In addition, American birds are closely matched in color some males show a tendency for fusion of the and pattern by those of samples of C. a. grate- red crown and frontal regions, thus parallel- loupensis from Veracruz. For example, mean ing a condition more fully developed and scores for belly color range from 4.7 to 5.8 in consistent in the related species C. carolinus Central America, and the mean sample score of the eastern United States. The sample from for birds from central Veracruz is 6.0. It Cozumel Island averages slightly larger in all seems consistent with our general plan of dimensions except wing length than those treatment to use only a single racial name, C. from mainland Yucatan, Quintana Roo, and a. santacruzi (Bonaparte, 1837, p. 116), with British Honduras. But in the over-all picture type locality "Guatemala," for these popula- of variation in Centurus aurifrons, the average tions. morphologic differences between popula- Birds from the Caribbean slope of northern tions on the Yucatan mainland and Cozumel Honduras have been distinguished racially by Island seem relatively minor, and we prefer Ridgway (1888, p. 582) as C. a. pauper, with not to apply a separate name to the latter type locality at Trujillo, Honduras, and this population. name has also been applied to birds from the 236 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 124 Motagua Valley of Guatemala (Skutch, type locality at San Bartolo[me], Tehuan- 1945). Characters supposedly distinguishing tepec, Oaxaca, to the yellow-naped and C. a. pauper from C. a. santacruzi are smaller boldly barred birds of the arid southern part size and narrower white dorsal bars, but when of the Isthmus of Tehuantepec and the arid birds from northern Honduras are compared interior valley of Chiapas (samples 17 to 21, with those from adjacent regions (Guatemala, fig. 2), although it has been customary to El Salvador, and southeastern Chiapas), the recognize an additional race, C. a. frontalis following picture emerges: in length of wing, Nelson, 1900, p. 257), with type locality at tail, bill, and tarsus, birds of the sample from San Vicente, Chiapas. The relatively minor Honduras average smaller than those of any morphologic differences distinguishing these other sample of the species C. aurifrons, al- populations, some of which are shown in our though they are almost matched by the sam- analyses of characters, have been adequately ple from southeastern Chiapas, and in bill discussed by Wetmore (1943), and "inter- depth the Honduran sample averages slightly gradation" is reported in a small sample of smaller than samples from adjacent regions. birds from the Monserrate region of Chiapas The white bars on the back are narrower (Edwards and Lea, 1955, p. 44). (1.13 mm.) than in any other sample, but the As herein defined, C. a. polygrammus is in white bars of birds from southeastern Nic- most characters similar to C. a. aurifrons, aragua are only slightly wider (1.28 mm.). from which it differs most conspicuously in Nape color (11.3) is probably not significantly having a greater amount of white in the tail. different from that of birds from El Salvador (10.9), southeastern Chiapas (10.8), central POPULATIONS NOT REPRESENTED Guatemala (11.5), and southern Guatemala BY SPECIMENS EXAMINED (11.0); frontal region color (1.3) is slightly Birds of the species Centurus aurifrons from yellower, on the average, than in El Salvador Roatan and Barbareta Islands off the Carib- (2.5), southeastern Chiapas (3.0), and south- bean coast of Honduras, are generally recog- ern Guatemala (2.9); belly color (5.7) is not nized as a distinct race, C. a. canescens (Sal- significantly different from that of samples vin, 1889, p. 370), which reportedly is similar from El Salvador (5.8) and southern Guate- to C. a. dubius of the Yucatan Peninsula but mala (5.3); and the dorsal tail pattern (4.7), differs in having wider white dorsal bars and while darker than that of birds from El Salva- the outer webs of the inner primaries spotted dor and southeastern Chiapas, is almost with white. matched (5.5) in the sample from central Gua- The population of C. aurifrons on Utila temala. All facts considered, there would Island, Honduras, resembles that of the adja- seem to be little justification for nomen- cent mainland (C. a. santacruzi) but has been clatural separation of the northern Honduran distinguished racially by Bond (1936, p. 360), population. as C. a. insulanus, on the basis of the follow- The name fumosus Nelson (1900, p. 258), ing characters: larger size; male with forehead with type locality at Huehuetan, Chiapas, is and superciliary region decidely white (versus also regarded as a synonym of santacruzi, gray in C. a. santacruzi); female with forehead following Ridgway (1914) and other authors. and fore parts of crown whiter; and upper Three badly worn females from Progreso, parts on the average "more extensively" Guatemala, have unusually yellowish napes barred with white. Bond also notes that there and bellies (nape color near categories 5 and "is very little white on the rectrices, this 6), which suggests some geographic differen- sometimes entirely lacking," but the same is tiation in the arid district of the Motagua true for our small sample from the mainland. Valley, as previously noted by Griscom (1932, Measurements given by Bond indicate that p. 229). Among Guatemalan specimens, only the Utila Island birds are indeed larger than a female from Antigua (A.M.N.H. No. those from mainland Honduras, with wing 394517) matches the Progreso birds. lengths ranging from 130 to 138 mm. in three Centurus aurifrons males and 127 to 128 mm. in three females; polygrammus CABANIS tail lengths are given as 75 to 85 mm. in males We prefer to apply a single racial name, C. and 76 to 80 mm. in females (compare with a. polygrammus (Cabanis, 1862, p. 326), with sample 31 in figs. 8 and 9). 1963 SELANDER AND GILLER: CENTURUS 237 ZONES OF RACIAL INTERGRADATION Our studies fully support the view that all polygrammus from eastern Chiapas are only populations of the aurifrons complex belong slightly larger than those of C. a. santacruzi to a single species, since even the most highly from central and northern Guatemala, being differentiated forms have been shown to in this respect intermediate between samples intergrade in zones of contact. We have al- of the latter race and those of C. a. polygram- ready discussed in detail the complete and mus from central Chiapas and Oaxaca. gradual transition from C. a. aurifrons of the Beginning in eastern Chiapas and extend- Mexican Plateau to C. a. dubius of the Yuca- ing west through central Chiapas and south- tan Peninsula through populations assigned ern and northern Oaxaca to southern Vera- to C. a. grateloupensis. There is also obvious cruz, evidence exists of clinal merging of clinal intergradation from C. a. dubius to the characters of the races C. a. polygrammus and extreme C. a. santacruzi types of the Pacific C. a. grateloupensis, with a marked shift oc- slope of Central America through Guatema- curring between the latter two regions. One lan populations. In nape color (fig. 3), which specimen, a male (A.M.N.H. No. 388879) is a representative character, there is a grad- from Tequisistlan, Oaxaca, not included in ual cline from a mean sample score of 12.4 our numbered samples (fig. 2), is a good inter- (almost pure red) in British Honduras grade between the races, as are several in our through 12.0 in northern Guatemala, 11.5 in sample from northern Oaxaca, and the rather central Guatemala, and 11.0 in southern marked variability of the large sample of C. a. Guatemala to 10.8 in southeastern Chiapas polygrammus from southern Oaxaca is pre- and 10.9 in El Salvador. sumed to reflect flow of genes south across the Centurus a. polygrammus does not touch C. Isthmus of Tehuanetepec from C. a. grate- a. aurifrons geographically, but it is reason- loupensis. Probably the steepest part of the able to assume that these two morphologi- gradient will be found near Matias Romero, cally similar forms are conspecific, especially where the arid vegetation types of the Pacific since the former appears not to be reproduc- side of the Isthmus give way to the mesic and tively isolated from C. a. santacruzi, with humid types of the Caribbean side (Selander, which it is in contact at several points. 1959, p. 403). Between C. a. polygrammus of eastern Intergradation between C. a. santacruzi of Chiapas and C. a. santacruzi of the coastal southeastern Chiapas and C. a. polygrammus region of southeastern Chiapas, the Sierra of southwestern Chiapas probably occurs not Madre de Chiapas, culminating in Volcan far south of the Laguna de la Joya near To- Tacana on the Chiapas-Guatemala border, nala, a locality that marks the division be- probably constitutes an almost absolute bar- tween the arid and mesic sections of the rier to gene flow. Therefore it is not surprising Pacific coastal plain of Chiapas. Specimens to find that the two populations are, as our from this region are too few in number fully analysis has indicated, strikingly different in to document this transition, but two speci- their characters. Similarly, specimens of C. a. mens from Pijijiapan show intergradation, polygrammus of eastern Chiapas differ con- although on the whole they are referable to spicuously from birds in our sample of C. a. C. a. santacruzi. santacruzi from central Guatemala, but a As yet we have no data on intergradation series of four males and seven females from between C. a. polygrammus and C. a. dubius, Sacapulas and Quiche, in central-western which presumably occurs in northern Chiapas Guatemala (sample 34; fig. 2), although ref- just north of the hills forming the northern erable to C. a. santacruzi, definitely ap- border of the central valley of Chiapas. proaches C. a. polygrammus and tends to In sum, intergradation between morpho- bridge the gap between the two. Intergrada- logically dissimilar populations of the C. auri- tion is particularly evident in nape, belly, frons complex clearly supports the hypothesis and frontal region colors (figs. 3-5), and that aurifrons, polygrammus, santacruzi, du- one male (A.M.N.H. No. 406693) from Saca- bius, and grateloupensis are races of a single, pulas has the red crown patch distinctly highly polytypic species. Presumably the separate from the colored nape area, as in C. insular forms canescens and insulanus are also a. polygrammus. In size, specimens of C. a. races of this species. 238 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 124 ECOLOGIC DISTRIBUTION OF CENTURUS A URIFRONS Over its extensive geographic range, this elevations in Veracruz, and coyol woodland species occupies many of the major plant on the arid coastal plain of central Veracruz. formations occurring from sea level to about Regarding the ecologic distribution of C. a. 7500 feet elevation, but nowhere does it in- dubius on the Yucatan Peninsula, Paynter habit dense forest. The highest recorded ele- (1955) notes that it is chiefly an inhabitant of vation appears to be 7800 feet, which it tall and moderately tall deciduous forest but reaches in Durango. The nominate race oc- also occurs in second growth in the rain forest. curs for the most part in xeric Sonoran Zone It is absent from the interior of heavy rain associations, including mesquite and other forest and from low scrub types. Van Tyne arid woodlands in Texas (Hamilton, 1962b) (1938) reports its occurrence in clearings in and Tamaulipas (Sutton and Pettingill, the forests of Peten, Guatemala, and we have 1942), mixed oak-juniper-mesquite wood- found it common along humid forest edges land (Texas), arid pine-oak woodland (Sierra and in humid woodland at Teapa, Tabasco, de Tamaulipas; Martin, Robins, and Heed, and Palenque, Chiapas. 1954), and the yucca-cactus-juniper-creosote Like C. a. grateloupensis, the Central Amer- bush association of Coahuila and San Luis ican race C. a. santacruzi is confined chiefly to Potosi. In addition, it is especially numerous the tropical zone, within which it occurs in wherever willows, cottonwoods, cypresses, or both humid and arid divisions and inhabits a other riparian trees are found, as in Palo Duro variety of vegetation types. In the Guate- Canyon, northern Texas, and in Zacatecas malan highlands, it occurs in coffee planta- (Webster and Orr, 1954) and Nuevo Leon tions at 3200 feet (Wetmore, 1941), and (Sutton and Pettingill, 1943). Where tele- Tashian (1953) found it common at all locali- phone or other poles are available for nesting, ties that he visited in southeastern Guate- it may occur in areas where there are no large mala. Deignan (1936) reports that it is abun- trees (Quillin and Holleman, 1918). In towns dant on the humid northern coast of Hon- it nests and forages in elms, oaks, pecans, duras, except in heavy forest. Skutch (1945) junipers, chinaberries, and other trees. found it in the humid lower Motagua Valley; Centurus a. polygrammus, like C. a. auri- there are also a few records from the arid frons, which it closely resembles morphologi- upper part of this valley. In El Salvador cally, also inhabits xeric and mesic vegetation (Dickey and van Rossem, 1938) it is a com- types and is often associated with mesquite. mon and generally distributed resident of all Edwards and Lea (1955) found it in a variety types of woodland in the Arid Lower Tropical of habitats on the central plateau of Chiapas. Zone and occurs locally in the Arid Upper Centurus a. grateloupensis inhabits both Tropical Zone up to 4500 feet; it is especially arid and humid tropical regions, occurring in abundant in the lower coffee district, along tropical forest at Xilitla, San Luis Potosi hedgerows dividing cultivated fields, and (Davis, 1952), coffee plantations at medium about small farms of the lower hill country.

DISCUSSION The barred pattern of the back and wings vegetation, and the background of trunks and in Centurus is probably cryptic, and its geo- branches to which the woodpeckers are ex- graphic variation shows a reasonably close posed is relatively light in color. But in more correlation with regional variation in climate humid regions where denser broad-leaved and vegetation type. Boldly patterned types vegetation creates greater areas of shadow (C. a. aurifrons and C. a. polygrammus), and where the color of the vegetation tends to which are in gross aspect much lighter than be darker, the dark appearance of the nar- the narrowly barred types (C. a. dubius and rowly barred forms is probably at a selective C. a. santacruzi), are found in arid regions advantage. In similar fashion, we assume that where light penetrates deep into the middle the value and hue of the color of the breast and lower strata of relatively open woodland and sides are adaptive, providing effective 1963 SELANDER AND GILLER: CENTURUS 239 counter-shading by being darker in more ments are biochemically similar carotenoids humid areas and lighter in those of greater (Test, 1942; Fox, 1953; Rawles, 1960), the aridity. Variation in the highly colored yellow production of which may be modified by or red belly may also be related to the prob- minor genetic changes, as evidenced by the lem of counter-shading, but, since the color of wide range of individual variation found in a the belly is not correlated with that of other single population. Moreover, in flickers (Co- areas of the under parts, we suspect that it laptes), as in certain other species of birds, may have some social signal function as well. variation in red and yellow pigments is influ- In the genus Centurus as a whole, the ex- enced by diet (Lester Short, in conversation). tent of white barring on the tail tends to be There seems to be little justification in at- directly related to the degree of boldness of taching much significance to a character so the dorsal pattern, but even in C. aurifrons subject to variation and so easily modified there are some notable exceptions. For ex- genetically and environmentally. ample, conspicuous white tail bars occur in C. To account for the present disjunct distri- a. polygrammus but are lacking in both the bution of the yellow-naped and broadly boldly patterned C. a. aurifrons and the nar- barred, arid-adapted populations of C. auri- rowly barred C. a. dubius. Since the tail is frons and the closely related species C. hoff- largely concealed by the wings when a wood- manni (see below), Wetmore (1943) has pecker is perched, and the pattern is fully re- proposed that contact between yellow-naped vealed only in flight or in display, it is possible and red-naped populations is secondary, and that the pattern is relatively unimportant he would explain the gap in distribution be- from the standpoint of concealment and has, tween C. a. polygrammus and C. hoffmanni in instead, some significance as a "flash" signal. terms of a crowding out and replacement of The brightly colored, sexually dimorphic yellow-naped populations by an invasion of regions of the head presumably have major C. a. santacruzi. Most of the resemblances significance in sex and species "recognition" among C. a. aurifrons, C. a. polygrammus, and and in agonistic and courtship display. Even C. hoffmanni involve characters that are so, it may be that the dark red or orange presumably adaptive in arid situations and frontal and nape regions of C. a. dubius and could therefore have developed independ- C. a. santacruzi have been selected because ently, but, considering also the close similari- they are less conspicuous to predators against ties of these forms in head pattern, we are dark backgrounds than are the lighter yellow inclined to accept Wetmore's hypothesis that shades of other populations inhabiting more these disjunct populations are monophyletic arid regions. and were formerly connected by yellow-naped Wetmore (1943) believes that the red ab- populations. In the case of the C. a. auri- domen found in C. a. dubius is the primitive frons-C. a. polygrammus disjunction, it is condition in the species, with the yellow and curious that there is not now a population of orange-yellow color in C. a. santacruzi and the C. aurifrons complex in the arid Pacific other races representing a secondary modifi- coastal region of southwestern Mexico, where, cation. The fact that the belly is red in species instead, we find C. chrysogenys, a species of " Tripsurus" and in several other species of resembling C. a. aurifrons in color and pat- Centurus may support this speculation, but it tern but actually not at all closely related to should be noted that the yellow-bellied and it. Perhaps the hypothetical connecting popu- yellow-naped condition also occurs in several lation of the C. aurifrons complex was re- species of Centurus, including C. hoffmanni, placed by C. chrysogenys, the latter invading C. uropygialis, and C. chrysogenys. Extensive western Mexico from the east, where a close speculation on this point seems unwarranted relative, C. pucherani, is found today. in view of the fact that yellow and red pig- OTHER SPECIES OF CENTURUS SYSTEMATIC STATUS OF CENTURUS HOFFMANNI CABANIS IN THE ABSENCE of information on the be- contact in Nicaragua, but of course we cannot havior and ecology of Centurus hoffmanni, neglect the possibility of hybridization in a judgment of its systematic status must rest narrow zone, with no readily detectable intro- entirely on the unsatisfactory criterion of gression. morphologic evidence. This form was de- In color and pattern, C. hoffmanni shows a scribed by Cabanis (1862, p. 322) as a distinct greater over-all resemblance to C. a. poly- species, but it was soon lumped with C. auri- grammus than to any other population of C. frons, to which it is closely allied, by Baird, aurifrons (pl. 53), but at the same time it does Brewer, and Ridgway (1874) and Ridgway not differ greatly from certain populations of (1881). In more recent years, some authors C. a. aurifrons. For example, nape color (fig. (for example, Peters, 1948, and Eisenmann, 3), which varies from yellow to orange (mean 1955) have considered it a subspecies of C. score, 3.0) in C. hoffmanni, is almost matched aurifrons, while others, notably Griscom by samples of C. a. aurifrons from Jalisco (1932) and Wetmore (1943), have suggested (3.9) and northwestern San Luis Potosi (2.5). that it is specifically distinct. On the basis of As noted by Wetmore (1943, p. 277), the red our studies of Centurus, we would concur with coronal patch of males of C. hoffmanni is the latter treatment, but solution of this usually more or less confluent with the yellow problem must await further research in the or orange color of the nape, whereas there is a field. bar of gray separating these two areas in C. a. Centurus hoffmanni is probably in contact, aurifrons and C. a. polygrammus. The belly if not locally sympatric, with C. aurifrons color (fig. 4) of C. hoffmanni (4.9) is on the santacruzi in northwestern Nicaragua, where average a less pure yellow than that of C. a. the two forms have been collected at localities polygrammus (eastern Chiapas, 1.3) but is only 55 miles apart (Jalapa and San Rafael closely matched by that of C. a. santacruzi del Norte) in a region lacking obvious barriers from southeastern Chiapas (4.7). The under to distribution (fig. 2). Considering the na- parts are darker than those of C. a. polygram- ture of patterns of distribution and contact mus and those of C. a. aurifrons, yet lighter between C. aurifrons and C. carolinus in than those of C. a. santacruzi. The frontal re- Texas (Selander and Giller, 1959b) and be- gion color (0.1) is a purer yellow than that of tween C. aurifrons and C. uropygialis in west- C. a. polygrammus (eastern Chiapas, 1.0) or central Mexico (see below), we predict that that of any other population of C. aurifrons C. hoffmanni and C. aurifrons will be found (fig. 5). The dorsal tail pattern shows a con- coming together in a very sharply defined siderable amount of white (0.9), and, among zone, with no extensive area of sympatry. samples of C. aurifons, it is equaled only by It has been claimed (Salvin and Godman, C. a. polygrammus (mean scores of 0.0 to 1.8 1888, p. 421) that specimens of C. aurifrons in different samples). The black and white santacruzi from El Salvador show a "diver- barring (figs. 6 and 7) is relatively heavy but gence" toward C. hoffmanni as regards averages somewhat narrower than that of C. amount of white in the tail, but in fact neither a. aurifrons and C. a. polygrammus. our large series from that region nor our small Individuals of C. hoffmanni are smaller samples from southeastern Honduras and than those of any population of C. aurifrons extreme northwestern Nicaragua (Jalapa) (figs. 8-11). The nearest approach is seen in show evidence of intergradation with C. hoff- C. a. santacruzi from northwestern Honduras, manni in this or any other character (see figs. and it is significant that specimens of this race 3-11). Nor do specimens of C. hoffmanni from from El Salvador and southwestern Honduras Nicaragua differ in any significant degree are considerably larger than C. hoffmanni. from those of Costa Rica. Clearly this evi- Perhaps the most striking character of C. dence strongly suggests that the two forms do hoffmanni is its tail length, which is not only not interbreed, if, as seems likely, they are in shorter than that of C. aurifrons but also 240 1963 SELANDER AND GILLER: CENTURUS 241 much shorter relative to wing length or to distribution of C. hoffmanni in Costa Rica, tarsal length. The wing to tail ratio for C. Carriker (1910) notes: ". . . this species takes hoffmanni is 2.03, whereas comparable ratios the place of ... [Centurus] pucherani on the for samples of the other species invariably are Pacific slope and the central plateau region, smaller, as shown in table 1. As noted previ- although the ranges of the two birds slightly ously by Griscom (1932), specimens of C. a. overlap, the present bird occurring occasion- santacruzi from southern Guatemala ap- ally on the Caribbean slope (Juan Vinas and proach C. hoffmanni in wing/tail proportion, Guapiles). It is the common woodpecker of but we find it significant that birds from El the highlands and the Pacific coast, although Salvador and Honduras do not, although in not so abundant in individuals as ... pu- general the tail is slightly shorter relative to cherani on the eastern side. It is also found the wing in C. a. santacruzi and C. a. poly- in more open country, rarely being seen in grammus than in more northern races. heavy forest." Cherrie reports that it occurs up to an altitude of 6500 feet in the interior. ECOLOGIC DISTRIBUTION On the Pacific side of southern Costa Rica, C. Concerning the geographic and ecologic hoffmanni is replaced by C. chrysauchen.

RELATIONSHIP OF CENTURUS A URIFRONS (WAGLER) AND CENTURUS UROPYGIALIS BAIRD Modern classifications carry Centurus auri- California, it inhabits timbered bottomlands frons and C. uropygialis as separate species, of the Colorado River, desert washes where but this treatment has been questioned by ironwood and paloverde reach large size, and van Rossem (1934, p. 410), who, on the basis desert mesas where there are large cacti of two "intermediate" specimens in the Brit- (Grinnell and Miller, 1944). With the plant- ish Museum (Natural History), a female from ing of cottonwoods and other trees in the Calvillo, Aguascalientes, and a male from Imperial Valley, it has extended its range "Santana," near Guadalajara, Jalisco, has northward in this century (van Rossem, suggested a conspecific relationship. Van 1932). In Arizona, it is the most common Rossem has also called attention to the occur- woodpecker in desert lowlands (Gilman, rence of traces of yellow in the nape patch 1915), nesting commonly in saguaros and in and nasal region of specimens of C. uropygi- other cacti in the Tucson area and also in alis, a feature that could indicate interbreed- trees within the city. It extends up into the ing and introgression with C. aurifrons. lower part of the Transition Zone at 5000 feet Before presenting our findings on the rela- in the Whetstone Mountains of Arizona, but tionship of these species in zones of contact, it is more numerous in lowland valleys and on we provide a brief review of the ecologic mesas (Smith, 1908). In Sonora, as through- distribution of C. uropygialis for comparative out its range, it is found in cottonwoods, purposes. willows, and other riparian elements, and it also occurs in cactus associations (van Ros- EcOLOGIc DISTRIBUTION OF sem, 1945). Centurus uropygialis Seven races of C. uropygialis are currently This species inhabits arid regions and is in recognized (Mexican Check-List Committee, all respects a close ecologic counterpart of C. 1957), although, compared to C. aurifrons, a. aurifrons. It has an altitudinal range from this species is not by any means highly vari- sea level to about 5000 feet but occurs pri- able geographically. marily in tropical and Lower Sonoran zones below 3000 feet. In Baja California, it is com- CONTACT ZONES mon in many desert formations in which There appears to be no contact between the cactus and mesquite are dominant, especially two species in northern M6xico, where Cen- where the large cardons are found, and it turus aurifrons occupies much of the plateau also nests in suburban gardens and in palms and is separated from C. uropygialis by the and yuccas (Bancroft, 1929). In southern Sierra Madre Occidental (fig. 2). But the race 242 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 124 C. uropygialis sulfuriventer penetrates inland interspecific hybrid (see below). in a series of low river basins and barrancas at Adults were in breeding condition, and, the highly dissected southern tip of the Sierra with the exception of the hybrid, both mem- Madre, establishing contact with C. aurifrons bers of pairs that were collected proved to be- aurifrons in western Aguascalientes and cen- long to the same species. Pairs were spaced at tral Jalisco (fig. 12). intervals of approximately 50 yards along AGUASCALIENTES CONTACT: Field work in the river, and the habitat situations in which this area was begun on June 9 and continued the two species occurred did not differ appre- through June 14. In the vicinity of the city of ciably. Moreover, the spacing of pairs sug- Aguascalientes, we found only C. aurifrons, gested that territories of the two species were which was moderately common in riparian mutually exclusive. We detected no differ- cottonwoods in agricultural areas and prob- ences in foraging behavior. ably also occurred in cactus and yucca forma- The vocalizations of the two species were tions in gullies east of the city. Driving west similar, but identification was usually possi- from Aguascalientes on the road to Jalpa, ble, since the calls of C. uropygialis were Zacatecas, we entered the Calvillo Valley higher pitched and softer. The over-all degree through a narrow pass about 33 kilometers of difference in the calls was roughly equiva- west of Aguascalientes. Dry oak woodland lent to that distinguishing the vocalizations was the dominant vegetation type on hillsides of C. aurifrons and C. carolinus in their zone at the pass, but in arroyos this was replaced of contact in Texas (Selander and Giller, by a yucca-acacia-tuna cactus association in 1959b). which a pair of C. aurifrons was taken and On June 11, camp was moved from La another individual was seen 34 kilometers Panadera southwest to the settlement of La west of Aguascalientes. Cuesta, 5 miles southwest of Jalpa, Zacatecas, Between the pass and Calvillo, the valley a locality on the Rio Juchipila, approximately widens and gradually decreases in elevation 30 miles from Calvillo. This region is more from 6000 to 5400 feet. At El Sauz, 38 kilome- arid than that at Calvillo and lies at a lower ters west of Aguascalientes, both species were elevation, approximately 5000 feet. At this found together in small stands of cottonwoods locality, as in the Calvillo Valley, Centurus and willows bordering two small reservoirs. woodpeckers were confined to riparian habi- Four pairs of C. aurifrons were seen, and tats and trees near houses, none being ob- seven adult specimens in breeding condition served in cactus and scrub vegetation on the were collected. We also obtained an adult hillsides, where they were replaced by the female and a juvenile of C. uropygialis. ladder-backed woodpecker (Dendrocopos sca- From El Sauz, we continued west to La laris). Thirty-four specimens of Centurus were Panadera, a settlement on the Rio de Mal- collected, all of which are C. uropygialis; the paso (a tributary of the Rio de Calvillo, which other species was not recorded. Specimens of in turn flows into the Rio Juchipila), 60 C. uropygialis were previously reported from kilometers west of Aguascalientes, or about 2 the Jalpa area (Jalpa, Moyahua, and Apozol) kilometers east of Calvillo, where we camped by Webster (1958). on June 9 and 10. Both species were present JALISCO CONTACT: An intensive search for in narrow stands, 40 to 50 feet tall, of riparian another zone of contact between Centurus cottonwoods, willows, and guamuchiles along species was made in the Guadalajara region the Rio de Malpaso, which at this locality is (fig. 12) from June 15 to 20. This heavily only a few yards wide, and a few birds were populated region has been so drastically seen in orchards near houses. The results of altered by human activity that only a few our investigations at this locality can be areas of habitat suitable for Centurus remain; summarized as follows: these consist of timber along streams and Centurus uropygialis outnumbered C. auri- clumps of trees at settlements. We were able frons four to one: along a 2-mile stretch of the to obtain a series of C. aurifrons in willows river, we found 20 individuals of C. uropygi- and cypresses along the Rio Grande de San- alis and five of C. aurifrons, and we also col- tiago near the town of Puente Grande, 20 lected an adult female that appears to be an miles southeast of Guadalajara, and we found ¾' ~~~~~~030

~C/) ' .~) N a cl) % ~~~~~~~~~~~~~~~) . )U -~~~~Ga-.0 0---~~~~~~~~~~~~~~~~~~4 to; zeci~ ~~.. 0 aC z

U) < 0-

0

0 a~ ~ ~ ~ Ga~ ~ ~ ~ U) b~~~~~~~~~~~~~~~~~~~~~~~*~c I..*~~~~~~~~~.q 1,.) 0 f*~~~~~~~~~~~~~~~ 0 ~~~~c AlI'\l-~t* )I~ 4-a ';~-Z *C.l* \ d.-01

'~~.alw4..JI)~~~~~~~~****,** *...o *~~~~~~~~-~~ ~ ~ ***-~~~~j\ ~~~ o f a) (J~~Cs A '~~~~~~~~~~~~~~~~~~~~~~~~N i" e r.

cn y&0~~~~~~~d r~ ao4*u~~~.~~2( W bO.~~C t%~~~~~~~~~~~~~~~~~~~~ eY 4 U2:

vzfz.. ~Cd~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~x.

CP~ ~~~~~~~~l ~

~~~ H~~

I

243 244 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 124 C. uropygialis common in willows along irri- C. uropygialis and nearly equals that of C. gation canals in cane fields near La Vega, 21 aurifrons. Extremes of variation in body miles west of Guadalajara, but we were un- weight also show overlap. able to locate either species at Santa Ana 2. COLOR OF HEAD AND UNDER PARTS: Tepititlan, a locality that we presume to be These areas average somewhat browner and the "Santana" (Sta. Ana) from which the less neutral gray in C. uropygialis. In males of "intermediate" specimen in the British Mu- C. aurifrons, the frontal and nape regions are seum was obtained in 1889. yellow or orange-yellow, and the red crown After a considerable period of searching, patch is relatively large; in males of C. uro- we discovered a previously unknown area of pygialis, the red crown patch is only three- contact 5 miles northeast of Zapopan, on quarters of the area of that of C. aurifrons, the rim of the Barranca de Oblatos of the Rio and the nape and frontal regions either lack Grande de Santiago, about 10 miles north of yellow or are tinged with very pale brownish Guadalajara. The physiography and vegeta- yellow. Similarly, in females of C. uropygialis, tion types of this barranca have already been the nape and frontal regions normally lack described in connection with an avifaunal yellow but may be tinged with pale brownish report (Selander and Giller, 1959a). In the yellow. barranca, the floor of which is at 3000 feet in 3. UPPER TAIL CONVERTS: These feathers elevation, only C. uropygialis was found, and are regularly barred with black in C. uropygi- here we collected a series of 12 specimens. On alis, whereas they are white or show black the rim of the barranca, at an elevation of blotches or imperfect bars in C. aurifrons. about 5000 feet, both species occurred in a 4. TAIL PATTERN: The dorsal tail pattern small grove of pecans, willows, mangos, and of C. uropygialis shows regular white bars on figs along a small stream. The surrounding the inner webs of the central rectrices, and on area is covered with cactus and other typical the outer web there is either a series of white arid plateau vegetation in which C. aurifrons bars, a wide streak, or some intermediate occurs sparingly in the complete absence of condition. In C. a. aurifrons, the tail is black the other species. dorsally or has a few white smudges or small, In the grove of mesic vegetation near the imperfect bars on the inner webs of the cen- stream, C. uropygialis outnumbered C. auri- tral rectrices, and the outer webs of these frons five to one, notwithstanding the fact feathers lack white. Ventrally, C. uropygialis that the locality is on the plateau at the ex- has four or more white bars crossing both treme upper altitudinal limit of C. uropygi- webs of the outer rectrix (number 5), whereas, alis, and the habitat is not greatly, if any, in C. a. aurifrons, there are one or two white different from the habitats occupied by C. bars suggested by blotches of white, two more aurifrons elsewhere in the Guadalajara re- or less complete bars crossing both webs, and, gion. In this stand, we collected four adult rarely, a third incomplete bar on the inner males, five adult females, and four juveniles web. of C. uropygialis. Two adult females of C. The hue, intensity, and distribution of aurifrons were taken, but males of this species yellow in the belly region are similar in both were not seen. species, as are the pattern of dorsal barring and the dark barring on the thighs, posterior MORPHOLOGIC DIFFERENCES flanks, and under tail coverts. AND HYBRIDIZATION An examination of large series of specimens In comparing Centurus a. aurifrons from from the zones of contact in Aguascalientes the Mexican Plateau with C. uropygialis from and Jalisco fails to demonstrate an extensive Jalisco, Nayarit, and southern Sinaloa, we intergradation of characters, a fact indicating noted the following differences: that the two woodpeckers have evolved in- 1. SIZE (FIGS. 13, 14): Centurus uropygialis trinsic reproductive isolating mechanisms is smaller, especially in wing and bill length, and are to be classified as full species. The but there is some overlap, sex for sex, in all variation in size of specimens from the con- dimensions, and particularly in the length of tact zones and adjacent regions is shown in the tail, which is disproportionately large in figures 13 and 14; note that neither species BULLETIN AMER. Mus. NAT. HIST. VOL. 124, PLATE 53

-IMMIZI..

illillim2ilillowl.,Klw.;.

.", ,71 -I i

Adult male specimens of Centurus aurifrons and Centurus hoffmanni in dorsal view. From top to bottom: Centurus a. santacruzi (El Salvador); C. a. dubius (YucatAn); C. a. grateloupensis (San Luis Potosi); C. a. aurifrons (Aguas- calientes); C. a. polygrammus (Chiapas); and C. hoffmanni (Costa Rica) BULLETIN AMER. Mus. NAT, Hisr. .^O-L.E--14.VOL. 124, PLTPLATE 54

Adult specimens of Centurus in dorsolateral view. From top to bottom: Male C. chrysauchen (Costa Rica); female and male C. pucherani (Costa Rica); and female and male C. chrysogenys (Nayarit) BULLETIN AMER. Mus. NAT. HIST. VOL. 124, PLATF, 55

Top to bottom, dorsal view: Centurus uropygialis, adult female, La Panadera, 2 kilometers east of Calvillo, Aguascalientes; presumed hybrid of C. aurifrons and C. uropygialis, adult female, same locality; C. aurifrons, adult female, same locality; Chryserpes striatus, male; Centurus hypopolius, male; Centurus radiolatus, male; Centurus carolinus (A.M.N.H. No. 39026), female variant from Florida, showing black occipital band; Centurus s. superciliaris, female BULLETIN AMER. Mus. NAT. HIST. VOL. 124, PLATE 56

A A &

A' A

Interspecific variation in degree of sexual dimorphism in size of bill. Adult specimens in dorsal view. Top row: Left, male and female Chryserpes striatus (Hispaniola); right, male and female Melanerpes herminieri (Guadelupe Island, Lesser Antilles). Middle row: Left, male and female Melanerpes portoricensis (Puerto Rico); right, male and female Xiphidiopicus percussus (Cuba). Bottom row: Left, male and female Centurus chrysogenys (Nayarit); right, male and female Piculus rubiginosus (Guatemala) 1963 SELANDER AND GILLER: CENTURUS 245 MALES WING TARSUS 122 126 130 134 138 142 21 23 25 A B (7) L.I .' (9) C (6) - ...L (6) D E n r n . rfAnlm urn us n in n n I _ F

n n roo in . G (7) .i ,frn 3. .m U H I .3 1 I

BILL LENGTH BILL DEPTH 21 23 25 2r 29 31 7.0 8.0 9.0 i I A B (9) (8) C (6) - (6) D E n n n n n inn. I n nnfnln n mis F n rnm ni0* G H Ir _. .. I *

FIG. 13. Measurements of males of Centurus uropygialis and Centurus aurifrons from zones of sympatry and adjacent regions, showing mean, range, twice standard error of mean, and standard deviation. Letters designate localities, as follows: A, southern Sonora (C. uropygialis); B, Sinaloa and Nayarit (C. uropygi- alis); C, La Vega and Bolafios, Jalisco (C. uropygialis); D, Jalpa region, Zacatecas (C. uropygialis); E, Calvillo Valley, Aguascalientes (open histograms, C. uropygialis; solid histograms, C. aurifrons); F, Guadalajara region (both species, as designated in E; circle, presumed interspecific hybrid, B.M.N.H. No. 88-10-10-521); G, northeastern Jalisco and central Aguascalientes (C. aurifrons); H, northwestern San Luis Potosi (C. aurifrons). tends to approach the other in size where the variation in the nape character, we segregated two are sympatric. 607 specimens of C. uropygialis into three Van Rossem (1934) implies that the pres- categories, from 0, with no yellow evident in ence of traces of yellow in the nape and fron- the nape area, to 2, with a light wash of tal regions of specimens of C. uropygialis from brownish yellow visible. A fourth category (3) southern Sinaloa, Nayarit, and Jalisco indi- was represented by a presumed female hybrid cates introgression from C. aurifrons, but our specimen from Aguascalientes (see below), analysis of variation in these characters does but no specimen of C. uropygialis was as- not support this interpretation. To measure signable to that category. 246 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 124 FEMALES WING TARSUS 118 122 126 130 134 20 22 24 A B C D E n n n n n O n rl1¶?mn *; I F n n n n 11III1 .

G -(3) (6) H

BILL LENGTH BILL DEPTH Is 20 22 24 26 6.0 7.0 8.0 A . , ~~~I , ,, B C i (4) - (3) D

E n onn nn3 I ni IN A nnnn n 14s

F nnnmu*nn a mmmm _ nnf an1n. G - I (6) L.I (5) H

FIG. 14. Measurements of females of Centurus uropygialis and Centurus aurifrons from zones of sympa- try and adjacent regions. Letters designate localities, as follows: A, southern Sonora (C. uropygialis); B, Sinaloa and Nayarit (C. uropygialis); C, La Vega and Bolafios, Jalisco (C. uropygialis); D, Jalpa region, Zacatecas (C. uropygialis); E, Calvillo Valley, Aguascalientes (open histograms, C. uropygialis; solid histograms, C. aurifrons; triangle, presumed interspecific hybrid, R.K.S. No. 3127; circle, presumed in- terspecific hybrid, B.M.N.H. No. 98-3-14-515); F, Guadalajara region (both species, as designated in E); G, northeastern Jalisco and central Aguascalientes (C. aurifrons); H, northwestern San Luis Potosi (C. aurifrons). The distribution of specimens in the three With reference to the question of introgres- classes is shown in table 3, in which it will be sive hybridization, we wish to emphasize the noted that yellow pigment is evident more fact that, although the frequency of occur- frequently in worn than in unworn skins. In rence of yellow on the nape is perhaps slightly males, for example, 81 per cent of unworn lower in birds from Baja California than in skins but only 56 per cent of worn specimens other samples, there is actually little geo- fall into category 0, showing no trace of yel- graphic variation in this character in the low on the nape. We also find yellow more species as a whole. If the occurrence of yellow often evident in males than in females. in the nape region is a result of current or 1963 SELANDER AND GILLER: CENTURUS 247 relatively recent hybridization, it is difficult Of the three questionable specimens, the to explain the fact that yellow occurs as fre- one most likely to be a hybrid rather than quently in specimens from Arizona and Cali- merely an unusual variant of one of the spe- fornia as in those from Aguascalientes, Zaca- cies (see comments on this problem by Brown tecas, and Jalisco, states near the zones of and Wilson, 1956) is an adult female (R.K.S. contact with C. aurifrons. A more plausible No. 3127; see pl. 55) taken at La Panadera, 2 hypothesis is that the yellow represents a kilometers east of Calvillo, on June 9, 1958. vestige of a character in the process of being Its measurments are as follows: wing, 127; lost in C. uropygialis or, in other words, a tail, 71; bill length, 25.5; bill depth, 7.4; remnant of a character that was more fully tarsus, 22.9 mm.; weight, 82 grams. In weight developed in a population ancestral to pres- and most linear dimensions (fig. 14), this ent-day C. uropygialis. specimen falls near the average for C. auri- PRESUMED HYBRIDS: The intermediate frons, but the wing and tail lengths are near characters of three specimens from the zones the observed minima for that species. The of contact, the two in the British Museum color of the head and under parts is matched mentioned by van Rossem (1934, p. 410) and by specimens of C. aurifrons, but in place of a one collected by us in 1958, suggest that conspicuous yellow nape crescent there is a intrinsic mechanisms isolating the two species narrow, dull grayish yellow patch quite un- reproductively are not fully effective and that like that seen in any other specimen of C. occasional hybridization does occur. aurifrons; and in the nasal region just poste-

TABLE 3 VARIATION IN AMOUNT OF YELLOW ON NAPE IN Centurus uropygialis Worn Specimens Unworn Specimens Locality Sex No. Category No. Category 0 1 2 0 1 2

Aguascalientes (Calvillo Valley) ei 10 5 4 1 - -

9 7 5 1 1 - - Zacatecas (Jalpa) 12 6 5 1 9 12 9 3 - - - - - Jalisco (Guadalajara region) c 8 6 1 1 9 6 4 2 - - Sinaloa-Nayarit-W. Jalisco 7 3 2 2 5 3 1 1 9 4 3 1 - S. Sonora 11 8 2 1 17 13 4 9 12 8 4 9 9 - N. Sonora ci 25 12 11 2 18 10 8 9 14 9 5 20 18 2 Arizona ci 49 26 17 6 51 42 7 2 9 41 27 11 3 36 33 3 Nevada ci 7 2 5 8 7 1 9 0 - - 5 4 1 California ci 17 8 7 2 17 14 3 9 16 11 5 18 15 3 N. and central Baja California ci 11 7 3 1 9 8 1 9 6 5 1 6 6 S. Baja California ci 41 28 10 3 29 27 2 9 32 26 6 11 11 - Totals ci 198 111 67 20 154 124 27 3 9 150 107 38 5 105 96 9 0 Percentages c - 56 34 10 81 17 2 9 - 72 25 3 - 91 9 0 248 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 124 TABLE 4 VARIATION IN DORSAL TAIL PATTERN IN Centurus aurifrons and Centurus uropygialis No. of Tail Pattern Categories Mean Form, Sex, Age, Locality Speci- 0 1 2 3 4 S 6 7 8 9 10 11 Score mens Centurus aurifrons e?Adult, Aguas.-Jalisco 17 - - - 2 - 15 10.8 eJuvenal, Aguas.-Jalisco 3 ------3 11.0 9Adult, Aguas.-Jalisco 18 - - - - 1 2 5 2 8 9.8 9Juvenal, Aguas.-Jalisco 3 - - 1 2 9.7 Presumed interspecific hybrid 9Adult (R.K.S. No. 3127) 1 - - 1 - 6 Centurus uropygialis ei'Adult, Aguas.-Jal.-Zac. 32 1 3 3 11 11 3 - - - - 3.2 dcJuvenal, Aguas.-Jal.-Zac. 18 5 5 6 1 1 .1.3 9Adult, Aguas.-Jal.-Zac. 29 4 5 9 5 5 1 - - - - 2.2 gJuvenal, Aguas.-Jal.-Zac. 11 4 4 2 1- - - 1.2 e?Adult, Nayarit-S. Sinaloa 13 - 2 3 5 3 - - 2.7 gAdult, Nayarit-S. Sinaloa 6 - 2 2 1 1 - - 2.2 0 Full white bars on both webs of central rectrices 1 White bars confluent near outer margin of outer web on proximal half of feather, "trapping" incom- plete bars of black (see specimen of C. uropygialis, pl. 55) 2 White bars confluent near outer margin along entire outer web 3 Broad streak of white on outer web, with small patches of black (representing greatly reduced bars) adjacent to rhachis 4 Like category 3 but without patches of black 5 Streak of white on outer web shortened to three-quarters of length of web 6 Streak of white on outer web narrower than in category 5 and shortened to one-half of length of web (see hybrid specimen, pl. 55) 7 Four 3-4 mm.-wide white patches (representing incomplete bars) on inner web proximally 8 Four 2 mm.-wide patches of white near outer margin of inner web 9 Single 2-3 mm.-wide patch of white on inner web 10 Single 1 mm.-wide patch of white on inner web 11 No white markings on central rectrices (see specimen of C. aurifrons, pl. 55) rior to the nostrils there are small patches of few are more or less comparable.) Note that this same color. In the character of the nape the female R.K.S. No. 3127 alone falls in and nasal region, the specimen falls nearer C. category 6 between C. aurifrons (categories 7 uropygialis than C. aurifrons, although it is to 11) and C. uropygialis (categories 0 to 5). not matched by any of several hundred speci- The upper tail coverts of the presumed hybrid mens of the former species. female are also intermediate in pattern (pl. The tail pattern of this female is also inter- 55), having black bars that are more perfectly mediate and is not duplicated in any available formed than those of C. aurifrons yet nar- specimen of either species. This is shown in rower and less uniform in width than those of table 4, in which specimens are assigned to 12 C. uropygialis. categories that represent progressive steps in In view of the obvious intermediacy of this the reduction of white on the central rectrices, specimen in a number of characters, the pos- a feature closely correlated with the amount sibility of its being a variant of C. aurifrons or of white on the ventral surface of the outer C. uropygialis seems remote, and we have rectrices. (These are not the same categories accordingly identified it as an interspecific of tail pattern used in the analysis of geo- hybrid. It is worth recording that the bird graphic variation in C. aurifrons, although a was collected in the same tree with an adult 1963 SELANDER AND GILLER: CENTURUS 249 male of C. uropygialis, to which possibly it like, and the tail pattern does not differ from was mated; the male had a brood patch and that seen in many specimens of C. uropygialis. testes measuring 5 mm. in length. In prepar- However, the under parts and forehead are ing the hybrid specimen, we found a fully grayer (less brownish) than in any specimen formed brood patch and an ovary that was of C. uropygialis in comparable plumage, and moderately enlarged and normal in appear- the nape appears to have much more yellow ance. These data strongly suggest that eggs than do specimens of that species. Probably it had been laid. If in fact the bird is a hybrid, it would fall in category 3 of nape color, but a seems probable that it is an F1, and the ap- more critical appraisal of this feature is pre- parent absence of introgression in the Calvillo cluded by the fact that the feathers of the Valley populations sampled in 1958 would nape region are not fully grown. further suggest that the occasional hybrids The male specimen (B.M.N.H. No. 88-10- formed are incapable of producing viable 10-521) was collected at "Sta. Ana," Jalisco, young. on February 23, 1889, by W. Lloyd. It is The two presumed hybrids in the British marked "jv." (juvenile) by the collector but Museum (Natural History) to which van is actually an adult bird. It measured as fol- Rossem (1934) referred were examined by us lows: wing, 132; tail, 83; bill length, 23.9; bill in 1958 but unfortunately were not available depth, 8. 1; and tarsus, 22.0 mm.; no weight is when we made our detailed analysis of varia- given. In wing length, it is intermediate be- tion in nape color and tail pattern in C. uro- tween mean measurements of the two species pygialis and geographic variation in C. auri- (fig. 13). Our notes indicate that this speci- frons. It is our opinion, however, that van men is also more or less intermediate in color Rossem (1934) was probably correct in inter- and pattern but rather on the C. uropygialis pretating these specimens as hybrids, al- side. though neither is so clearly intermediate in In conclusion, it may be of interest to re- its characters as is our female from the Cal- cord the incidence of hybrids in the two zones villo Valley. of contact. From the Calvillo Valley between The British Museum specimen from Cal- Jalpa, Zacatecas, and the city of Aguascali- villo (B.M.N.H. No. 98-3-14-515) is a female entes, we have examined 15 adults of C. auri- collected by W. B. Richardson in August, frons, 18 adults and four juveniles of C. uro- 1888. It was marked "imm." (immature) by pygialis, and the two presumed female hy- the collector and is in a late stage of fall molt, brids. Hence, 5 per cent of the 39 specimens either postjuvenal or postnuptial. It meas- from the Calvillo Valley are hybrids. From ured as follows: wing, 125; tail, 72; bill length, "Sta. Ana" and our locality near Zapopan, 20.6; and tarsus, 21.3 mm.; weight was not Jalisco, we have examined 20 specimens (six recorded by the collector, and bill depth can- adults of C. aurifrons, nine adults and four not be measured. In most dimensions this juveniles of C. uropygialis, and the one pre- specimen falls in a position intermediate sumed male hybrid), 5 per cent of which between mean values for the two species (fig. show hybrid origin. 14). In color and pattern, it is C. uropygialis-

RELATIONSHIP OF CENTURUS A URIFRONS (WAGLER) AND CENTURUS CAROLINUS (LINNAEUS) It has been suggested (Brodkorb, 1957) ties: Port Lavaca area, Calhoun County; that these forms are conspecific but they are Victoria area, Victoria County; Giddings, Lee in fact full species, as demonstrated by sym- County; Austin region, Travis County; San patry in central Texas (Selander and Giller, Angelo, Tom Green County; Colorado City, 1959b). Mitchell County; Carbon area, Eastland All available breeding records of Centurus County; and Vernon, Wilbarger County. in Texas are shown in figure 15. Both C. auri- In winter C. carolinus tends to wander frons and C. carolinus have been found nest- south and west of its breeding range, being ing in the same area at the following locali- reported at several localities in Nueces, Bee, 250 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 124

* Centurus carolinus o Centurus aurifrons

O 10 20 40 6.0 so

FIG. 15. Distribution of Centurus carolinus and Centurus aurifrons in Texas, based on specimens examined and numerous sight records. Goliad, Aransas, San Antonio, Comal, and tion in C. carolinus has been studied by Kendall counties, and as far south as Browns- Burleigh and Lowery (1944) and Koelz ville and Starr counties along the Rio Grande. (1954); also, see remarks by Todd (1946). In the zone of distributional overlap, which At the time of our earlier report (Selander is, at least in the Austin region, only a few and Giller, 1959b), we had recorded no indi- miles wide, both species are closely similar in vidual that suggested mixed ancestry, but in size and in nesting and foraging behavior, and April, 1959, such a bird, a female, was seen on they hold mutually exclusive territories. the campus of the University of Texas in Details concerning the habitat distribution Austin in company with, and apparently and behavior of these species can be found in mated to, a male of C. carolinus. The dorsal our earlier paper (1959b). Geographic varia- tail pattern of the bird in question was typical 1963 SELANDER AND GILLER: CENTURUS 251 of that of C. carolinus in Texas, with con- with C. uropygialis, it is highly unlikely that spicuous white bars on the central rectrices, these two woodpeckers would interbreed if but the nape was yellowish orange rather they were sympatric. than red. This peculiar nape color suggested Centurus hypopolius resembles C. uropygi- mixed ancestry, but the probability of the alis in lacking highly colored nape and frontal bird's being a hybrid is reduced by the fact regions, but it differs in lacking yellow on the that there are several similar specimens of C. belly. It is smaller than C. uropygialis, al- carolinus in the collections of the American though there is a general similarity between Museum of Natural History from localities the two species in body proportions, the wing far removed from the range of C. aurifrons. and tail being comparatively long in both. These are as follows: The presence of a black superciliary patch, a Female (A.M.N.H. No. 362499), Hillsboro, Iowa, relatively naked orbital ring, and a splotch of February 2, 1899; nape pale yellowish orange color (red) on the malar region suggests a Female (A.M.N.H. No. 362453), Eau Gallie, relationship with C. chrysogenys rather than Florida, February 2, 1914; nape pale yellowish with C. uropygialis. The eye color of adults is orange also closer to that of C. chrysogenys: in C. Two males (A.M.N.H. Nos. 362441, 362442), hypopolius the adult iris is dull brown; in C. same locality, January 29, February 2, 1914; chrysogenys, burnt sienna or orange-brown; in orange nape, frontal, and coronal regions Female (A.M.N.H. No. 362510), Danfuskie Is- C. uropygialis, reddish brown. In C. aurifrons, land, South Carolina, November 24, 1904; nape the iris is dark red in Mexican populations orange but may vary to reddish brown in Central Female (A.M.N.H. No. 362525), Warsaw, Illinois; American birds, according to Dickey and van nape orange Rossem (1938) and data on specimen tags. Female (A.M.N.H. No. 362479), Delight, Ar- The iris is also dark red in C. carolinus. In kansas, March 29, 1913; nape orange juveniles of all species, the iris is dull brown. Female (A.M.N.H. No. 362435), Amelia Island, There is considerable interspecific variation Florida, December 13, 1905; nape reddish in the length and relative massiveness of the orange hyoid apparatus in Centurus. In skinning In view of these facts, and when it is con- specimens we have noted a general similarity sidered that our questionable Texan bird had in C. aurifrons, C. carolinus, and C. uropygi- a tail pattern typical for C. carolinus, it is alis in this regard, but, in C. chrysogenys and probable that it was an individual variant of even more markedly in C. hypopolius, this that species rather than a hybrid. Without structure is relatively less massive and the further explanation, Peterson (1960) has horns are shorter. recently stated that hybrids between C. caro- The vocalizations of C. hypopolius were not linus and C. aurifrons "are said to occur." studied in detail, but we recorded one distinc- Possibly this statement is based on sight tive call consisting of an indefinite number of records of variant individuals of the type just harsh "che" notes delivered with rising inflec- described, but, in any event, we have no tion. This call was quite unlike any given by knowledge of published reports of hybridiza- other species of Centurus with which we are tion or of the existence of supposed hybrid familiar in the field. In their harshness the specimens. vocalizations of C. hypopolius suggest those Centurus of C. chrysogenys, but the calls are given more hypopolius (WAGLER) softly, being in this regard similar to those of Peters (1948) considered this form to be C. uropygialis. conspecific with Centurus uropygialis, but ECOLOGIC DISTRIBUTION: Throughout its with few exceptions (Blake, 1953; Davis and relatively restricted geographic range in Russell, 1953) recent authors (Sutton, 1951; southwestern and central M6xico (fig. 1), C. Amadon and Eckelberry, 1955; Eisenmann, hypopolius occurs in arid regions and is found 1955; and Mexican Check-List Committee, in habitats structurally similar to some of 1957) have not followed this action. It is our those occupied by C. aurifrons and C. uro- opinion that, while C. hypopolius may have pygialis in northern M6xico. It was common some not very distant phyletic connection in yucca and cactus formations near Tepanco, 252 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 124

Puebla, in July, 1958, and Amadon and above it actually seems to be not at all closely Eckelberry (1955) found it nesting in cactus allied to the aurifrons-uropygialis-carolinus along the highway near Izuicar de Mata- superspecies and apparently has closer phy- moros, Puebla. Apparently it is in contact letic ties with C. pucherani and other species with C. aurifrons in the Valley of Mexico, and of the " Tripsurus" group. it is also locally sympatric with C. chrysogenys Geographic variation in this species has in Guerrero, Morelos, and possibly southern been studied by R. T. Moore (1950), who Puebla (fig. 1). In Morelos, C. hypopolius is a advocates the recognition of three races. The common resident of the desert "chaparral," measurements of specimens that we have where it is frequently found on cacti, at me- examined are shown in table 5. dium elevations (up to 7000 feet), while C. In preparing specimens of C. chrysogenys, chrysogenys is restricted to lower elevations we noted the following peculiarities of struc- (generally below 4700 feet) and has its center ture and color as compared to C. aurifrons, of distribution in riparian communities along C. uropygialis, and C. carolinus: iris of adults the Rio Balsas and its tributaries. However, orange-brown (rather than red or reddish locally the two species occur together, as at brown); orbital region naked; legs and feet Alpuyeca, Morelos (Davis and Russell, 1953). with a more greenish (less bluish) cast; sub- cutaneous and body fat distinctly more Centurus chrysogenys (VIGORS) orange (less yellow); head much more difficult This handsome species (pl. 54) bears a to pass through neck in skinning, and skin superficial resemblance to the yellow-naped bound to body by stronger fascia, especially races of Centurus aurifrons, but as noted dorsally on legs; hyoid apparatus relatively

TABLE 5 MEASUREMENTS (IN MILLIMETERS) AND WEIGHTS (IN GRAMS) OF Centurus hypopolius AND Centurus chrysogenys No. Wing Tail Bill Length Bill Depth Tarsus Weight Centurus hypopolius Pueblaa e Adult 6 124.0 79.7 20.98 6.84 20.27 51.6b (122-125) (77-86) (19.7-22.6) (6.4-7.8) (19.5-21.1) (50-54) 9 Adult 6 124.3 81.8 19.76 6.18 19.25 50.0 (122-128) (76-88) (18.4-21.3) (6.0-6.5) (18.4-20.5) (46-54) Centurus chrysogenys San Blas, Nayaritc c Adult 13 119.4 70.0 21.58 7.33 20.91 70.5 (117-123) (68-71) (19.5-24.5) (6.9-7.5) (20.4-22.4) (65-73) 9 Adult 14 115.1 68.6 18.80 6.75 19.77 62.6 (113-118) (65-71) (17.9-20.6) (6.5-7.1) (18.5-20.5) (58-70) Colima-S. Jalisco e Adult 4 120.3 67.0 21.48 7.40 21.38 76.8 (118-122) (65-68) (20.9-22.1) (7.3-7.5) (20.5-22.2) (73-80) 9 Adult 2 116.0 65.5 18.45 6.50 18.90 59.0 (115-117) (63-68) (18.4-18.5) (6.5-6.5) (18.6-19.2) (54-64) Guerreod ci Adult 8 125.3 71.4 21.4 9 Adult 4 121.6 70.2 20.8 a July 13, 1958. bAdditional data from Davis and Russell (1953, p. 97): three adult males from Morelos, taken in July and August, weighed 48, 48, and 53 grams. An adult male from Mitla, Oaxaca, weighed 47 grams. Fifteen juvenal males in post- juvenal molt, collected on July 13, 1958, in Puebla, weighed 45.5 (41-51) grams. 0 June, 1958. d Data from Ridgway (1914, p. 90). Additional data: three adult males from Morelos weighed 80, 80, and 85 grams, and two adult females from Morelos weighed 68 and 73 grams (Davis and Russell, 1953, p. 97). 1963 SELANDER AND GILLER: CENTURUS 253 weaker in structure; and mucus-secreting state, C. uropygialis ranges from near sea glands (GI. picorum; see Bock, 1961, p. 357) level to 4000 feet, and records of C. chryso- below rami of upper mandibles relatively genys indicate an altitudinal distribution from smaller. The last two characters are associ- sea level to 3500 feet, as at Los Pieles and ated with a relatively small bill, which sug- Cosala, although there is one doubtful report gests that this species does less excavating for of its occurrence as high as 5000 feet (W. D. food in trees than do the other species men- Miller, 1905). Both species have been taken tioned. at Mazatlan, Rosario, Labrados (McLellan, In the field we noted that C. chrysogenys 1927), Escuinapa, and the "Rio Juanna was more prone to perch crosswise on twigs Gomez" (W. D. Miller, 1905). In Nayarit, than were other species studied, but its gen- however, we found distinct altitudinal segre- eral demeanor was not otherwise different. gation of the species, with C. chrysogenys be- Vocalizations were loud and often rather ing confined to the coastal plain proper below harsher than those of the other species, and 1000 feet, and C. uropygialis occupying me- the "location call" was distinctive, consisting dium-elevation slopes of the Sierra Madre be- invariably of three or four notes (two or three tween 1000 and 5000 feet (see figs. 12 and short ones followed by a long one) rather than 16A). At the latter elevation near Gua- a rapidly pulsed series of seven or eight notes dalajara, C. uropygialis is in contact with given by the other species (see Selander and C. aurifons. Giller, 1959b, fig. 1, p. 111). South of the range of C. uropygialis (the ECOLOGIC DISTRIBUTION: On the Pacific southernmost point of which is at Cocula, coastal plain of Mexico, C. chrysogenys is an Jalisco; see fig. 12), C. chrysogenys ranges abundant resident of the Arid Tropical Zone, upward to 4500 feet, as shown in figure 16B where it occurs in scrub land among giant and C, which is based on two road transects cacti, as at Autlin, Jalisco (Zimmerman and made in 1958, one (B) extending from Salinas, Harry, 1951), along the edges of deciduous on the coast, northeast to Guadalajara forest, and in association with a variety of through Autlan and Cocula, and another (C) palms. In the San Blas area, Nayarit, we extending from the Manzanillo area, Colima, found it common in mixed groves of palms, north-northeast to Guadalajara through figs, and other trees along the coast. Nesting Colima, Ciudad Guzman, and the Sayula in coconut and other palms is also reported by basin. Upper tropical zone habitats in which Bailey (1906) and Lamb (1909). Commonly C. chrysogenys was found at Tecotlan and at this species also nests in cacti (McLellan, Pihuamo and Tuxpan were not visibly differ- 1927). As it ranges inland in the drainage ent from those occupied by C. uropygialis in basin of the Rio Balsas and elsewhere, it Nayarit and in Jalisco west of Guadalajara. appears to become restricted to riparian In the Manzanillo-Guadalajara transect, habitats. For example, 10 miles southwest of actual contact between C. chrysogenys and C. Juchitlan, Jalisco, we found it nesting in aurifrons was not recorded, but it undoubt- willows along a small stream, but it was ab- edly occurs in the vicinity of San Sebastian, sent from stands of tuna cactus on the sur- for we found C. aurifrons in a pecan orchard a rounding hillsides. In Morelos, Davis and mile north of that city, and C. chrysogenys Russell (1953) noted that it is most common was common in tropical riparian forest at in riparian associations. Tuxpan and Pihuamo. The former locality is only 20 DISTRIBUTIONAL AND ECOLOGIC RELATION- miles south of San Sebastian and in SHIPS AMONG Centurus chrysogenys, the same valley. Between San Sebastian and Centurus Tuxpan, the typical plateau flora is replaced uropygialis, AND by tropical vegetation in which banana and Centurus aurifrons fig trees are common. Centurus chrysogenys breeds north in south- ern Sinaloa for an undertermined distance Centurus rubricapillus CABANIS AND and has been taken as far north as Cosala, 75 Centurus pygmaeus RIDGWAY miles north of Mazatlan. It is sympatric with Centurus rubricapillus resembles C. hoff- C. uropygialis over a considerable area in manni and certain races of C. aurifrons in Nayarit and southern Sinoloa. In the latter color and pattern but is much smaller than 254 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 124 aE= chrysogenys 1iiia uropygialis ~t-JAIVK n i i ri f rn n a I 600IO I lwGuadalajara 400

200 .1Oincon

~ an Bias A San Blas- Guadalojora Tronsect

6004

4004

20(

Salinas- Guadalajar.a Transect

Manzonillo Guadalajara Transect

FIG. 16. Altitudinal distribution of three species of Centurus along three road transects in west-central Mexico. Arrows indicate points of contact between species. these forms. Geographic variation in C. rubri- many respects a miniature of C. aurifrons capillus has recently been reviewed by Wet- polygrammus. more (1957), who recognizes four subspecies, Whether C. pygmaeus of the northern justifiably synonymizing several racial names Yucatin Peninsula and Cozumel Island and based on individual rather than geographic its dark insular representative, tysoni, on variation. As in C. aurifrons, the color of the Bonaca Island, Honduras, are conspecific nape and belly varies from yellow to red, and with C. rubricapillus is problematical, since there is in South American populations and in these forms are allopatric (fig. 1). Peters C. r. seductus of Isla El Rey, Pearl Islands, (1948) and several others combine the two, PanamA, a strong tendency for separation of but in Wetmore's opinion (1957) the nar- the red coronal and the orange or yellow nape rower dorsal barring and the disproportion- regions by an occipital band of gray. This ately longer tail (table 1) of C. pygmaeus tendency is most fully expressed in the race argue for specific status. Perhaps the wisest C. r. paraguanae of Venezuela, which is in course is to follow Wetmore until close simi- 1963 SELANDER AND GILLER: CENTURUS 255 larity in courtship, vocalizations, and other Adults of C. pygmaeus are highly variable aspects of the behavior of these forms is dem- with respect to the distribution and intensity onstrated. of yellow pigment on the frontal region, malar In color and pattern, C. pygmaeus is re- region, and chin. In a series of 47 specimens markably similar to C. a. dubius, with which examined, 32 had the entire frontal region it is sympatric (see below). It is in fact a mini- intense yellow or orange-yellow, and the color ature of that form and differs in color and extended posteriorly to the malar region and pattern from C. rubricapillus in much the chin. In many of these 32 specimens there was same way that C. a. dubius differs from C. a. yellow pigment around the eye, and in a few a santacruzi. spot of yellow was also present on the auricu- Parenthetically, we wish to state that, for lars. In 10 specimens only the nasal tufts were reasons given in part elsewhere in this report, yellow and the color was much less intense, we cannot accept Sutton's suggestion (1951) and five specimens exhibited an intermediate that pygmaeus "represents" C. aurifrons and condition, with a moderate amount of yellow that the other Yucatin form, dubius, is spe- in the frontal region but little on the malar cifically distinct from C. aurifrons. region or chin. ANTILLEAN SPECIES OF CENTURUS Centurus superciliaris (TEMMINCK) forms, superciliaris and caymanensis, would When Peters' (1948) readiness to lump interbreed and successfully exchange genes if allopatric forms in the genus (for example, they were brought into sympatry. uropygialis and hypopolius; rubricapillus and The striking over-all resemblance of C. s. pygmaeus; and aurifrons and hoffmanni) is nyeanus and C. carolinus of the eastern considered, it is surprising that he maintained United States is not properly emphasized by the Grand Cayman form, caymanensis, as a Ridgway (1914), who compared the former species distinct from C. superciliaris of Cuba only with other Antillean woodpeckers. They and its racial representatives on the Isle of are similar in size, although the tail in C. s. Pines (C. s. murceus), Abaco Island (C. s. nyeanus is on the average disproportionately blakei), Watling Island (C. s. nyeanus), and short. Indeed, were it not for the aforemen- Great Bahama (C. s. bahamensis). These tioned black superciliary spot, the more ex- forms possess a superciliary (or superauricu- tensively bare orbital region, and the lighter lar) patch of black, which in the female is forehead of C. s. nyeanus, it would be difficult joined on each side by a black occipital band. to distinguish specimens of the two forms. In characterizing caymanensis, Ridgway In examining large series of specimens of C. (1914) correctly noted that there is "no trace carolinus, one finds further evidence of close of black superciliary spot," but he neglected relationship between C. carolinus and the C. to mention that an occipital band of black is superciliaris complex in the form of individual suggested in females, the feathers of this variants showing slight to moderate develop- region being black medially but grayish at the ment of the black occipital band characteris- tips. In C. s. nyeanus, the superciliary spot is tic of females of the latter species. Frequently almost completely masked by grayish feather in C. carolinus the gray of the occipital region tips, and, except for this feature and the more is "intermixed" with black, as previously conspicuous wash of yellow dorsally in cay- noted by Ridgway (1914), but occasionally manensis, the two forms are closely similar. there is a continuous black band, as in a fe- Bridging the gap between C. s. nyeanus and male (A.M.N.H. No. 39026) from Sebastian C. s. superciliaris of Cuba, both as regards River, Florida (see pl. 55), and in females prominence of the black superciliary spot and from Silspee, Texas (A.M.N.H. No. 362524), body size, are the forms C. s. blakei and C. s. Sapelo Island, Georgia (A.M.N.H. No. bahamensis. We therefore see no course but to 362495), and Hawkinsville, Florida (A.M.- regard the whole complex as a single species, N.H. No. 367476). although one might justifiably entertain A similar black occipital band is also doubts as to whether the most dissimilar weakly indicated by a few partly black feath- 256 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 124 ers in approximately 10 per cent of female black. Also, the iris color is unusual, being specimens of C. aurifrons from Cozumel Is- bright yellow or gold in adults. Other struc- land and the Yucatan Peninsula. tural peculiarities of Chryserpes striatus have In C. superciliaris, as in C. carolinus, the been described by W. D. Miller (1915). adult iris is red or reddish brown, and the One characteristic of Chryserpes that sets it tarsus and toes are olive. apart from Centurus is an unusually marked degree of sexual dimorphism in bill size, a Centurus radiolatus (WAGLER) feature that, surprisingly enough, was over- The Jamaican form of Centurus resembles looked by W. D. Miller (1915) and other C. aurifrons dubius, showing in its white fore- authors. The possible adaptive significance of head and sides of the head and in its dark this feature is discussed below. under parts "exaggeration" of characters Information on the ecologic distribution present in that race of C. aurifrons (compare and habits of this woodpecker is summarized pls. 53 and 55). The belly varies from yellow by Wetmore and Swales (1931) and Wetmore to red, and the iris is red. and Lincoln (1933). It is reportedly one of the most common and widely distributed native birds of Hispaniola, inhabiting mangrove THE STATUS OF Chryserpes swamps and swampy woods on the coast, striatus (MtLLER) desert and thorn scrub lowland vegetation The Haitian Woodpecker (pl. 55) is so types, coffee plantations, wooded hillsides, unusual in color and in some aspects of struc- and pinelands of the interior hills and moun- ture that it is difficult to judge its affinities. tains, "seemingly with no preference" (Wet- Recent classifications, including that of Bond more and Swales, 1931, p. 291). It nests in (1956, 1961), carry it as a Centurus, but with- cacti, palms, and a variety of trees. out question it is not closely allied to any Of special interest is a report by Wetmore species of that genus or of Melanerpes, and (Wetmore and Lincoln, 1933) of colonial there would seem to be much justification for nesting in Chryserpes, a feature in which it placing it in its own monotypic genus, Cry- differs from Centurus species: "At Terrier serpes, as W. D. Miller (1915) has advocated, Rouge .., these birds were surprisingly at least until morphologic and behavioral abundant ... and we were astonished to find studies provide further clues to its relation- a dozen pairs going in and out of nesting holes ships with other woodpeckers. There is some in a single dead tree trunk standing in an possibility that its affinities are with " Chioro- open space, the holes being 3 to 10 meters nerpes" (=Piculus, part, of Peters, 1948). In from the ground and in some cases less than a Chryserpes, features not found in Centurus meter apart. There was no question that the include the absence of barring on the flanks woodpeckers were colonizing, as the trunk and under tail coverts; the absence of a highly was a veritable apartment house with the colored frontal region, despite the fact that birds clambering actively over its surface and the nape (as well as the coronal and occipital flying back and forth to near-by woodland." regions in the male) is red; the short, red Vocalizations of Chryserpes have been de- upper tail coverts; and the unusual olive and scribed (Wetmore and Swales, 1931) as loud, yellow colors of the body plumage and wings. rollicking, rolling calls, often interspersed The Centurus-like barred pattern of the with tree-toad-like guttural sounds, sugges- dorsum and wings was probably acquired in- tive of the calls of flickers (Colaptes), ant- dependently in Chryserpes, the resemblance eating woodpeckers (Balanosphyra), and the to species of Centurus being a matter of con- red-headed woodpecker (Melanerpes erythro- vergence. As compared with species of Cen- cephalus). Elsewhere Wetmore (Wetmore turus, the bill of striatus is straighter and the and Lincoln, 1933, p. 44) mentions the culmen is more sharply ridged, and the color "laughing" calls of the Haitian woodpecker of the bill is peculiar, the lower mandible be- and notes that this species does not "drum" ing horn color or gray basally rather than as do species of Centurus. DISCUSSION SIGNIFICANCE OF ECOLOGIC RANGE IN DISTRIBUTION AND SPECIATION

MUCH MODERN SYSTEMATIC RESEARCH deals developed in geographic isolation will lead the with congeneric species in zones of secondary two newly evolved species to select different contact where there is active testing of repro- sorts of places in which to live once secondary ductive isolating mechanisms (A. H. Miller, contact is established. (This argument is 1955a) and where overlapping ecologic spheres analogous to that of systematists who main- may lead to interspecific interactions, includ- tain that reproductive isolating mechanisms ing not only competition in the strict sense of are fully perfected in geographic isolation and the endeavor of individuals to obtain a com- need not, therefore, be reenforced by selection mon resource in short supply (Milne, 1961) when secondary contact is established.) How- but also aggressive and other interference ever, an alternative and perhaps more con- interactions (Hutchinson and MacArthur, vincing view, first developed in detail for 1959; Ripley, 1961) which may not be related birds by Lack (1944, 1949), holds that com- to resources in short supply but which in any petitive interactions in zones of secondary event result in decreased population density contact play a fundamental role in the diver- of one or both species. If populations evolving gence of ecologic ranges. More importantly, in geographic isolation are to coexist as spe- this latter view maintains that the limits of cies after establishing secondary contact, the ecologic sphere of a species may ulti- they must be reproductively isolated to main- mately be set by competitive pressures of tain their adaptive genetic systems, and at other, sympatric species, although they may the same time there is some reason for believ- be determined proximally by instinctive or ing that they must be ecologically isolated conditioned preferences (A. H. Miller, 1942). (compatible) through segregation to different Admittedly, most of the evidence for inter- habitats or by exploitation of different niches specific competition in wild populations is within the same general habitat type (Lack, indirect, but in the opinion of many workers 1947, p. 146). In short, they must differ in it is compelling. Among various lines of evi- their ecologic spheres (A. H. Miller, 1955b; dence are numerous examples of expansions Dixon, 1961) or ranges, and in using these of ecologic ranges and unusual numerical terms we refer to aspects of the Grinnelian relationships of species belonging to unbal- and Eltonian "niche" (Udvardy, 1959) as anced faunas of oceanic islands or isolated well as to the habitat distribution of a species. mountains (Amadon, 1950; Lack and South- There is presently considerable disagree- ern, 1949; A. H. Miller, 1955b; Crowell, ment among systematists regarding the im- 1961); the demonstration (Pitelka, 195 1b) of portance of population interactions in the competition for space in the form of inter- origin and development of reproductive iso- specific territorialism leading to mutual de- lating mechanisms (see, for example, Blair, pression of population densities in Allen and 1958; J. A. Moore, 1957; Mayr, 1959; Bogert, Anna hummingbirds (Calypte anna and 1960; Sibley, 1961). There is no greater har- Selasphorus sasin); and the demonstration of mony among views concerning the degree of striking niche specificity among groups of significance to be attached to competitive and sympatric congeners (Hartley, 1953; Gibb, other interactions in the evolution of ecologic 1954; Betts, 1955; and Dixon, 1961, on isolation between species. Proponents of Parus; MacArthur, 1958, on Dendroica; and density-independent and "probabilistic" the- Cade, 1960, on Falco). In addition, we must ories of population control (Andrewartha and at least cite recent contributions of Mac- Birch, 1954) and certain other ecologists Arthur (1957), Hutchinson (1958), and Elton (Udvardy, 1951; Bowman, 1961) have argued (1946), among others, to the problem of the that it is not necessary to invoke competition role of interspecific competition in commun- as a cause of ecologic segregation of sympatric ity organization and evolution. Of particular species, since different habits and preferences interest is MacArthur's demonstration that a 257 258 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 124 model based on the assumption of contiguous ing behavior, habitat selection, and morphol- but non-overlapping niches fits the data on ogy (bill size). (2) When two species with sim- abundance of bird species in tropical forests ilar ecologic spheres meet over an extensive better than do models based on other assump- area that provides sufficient habitat diversity tions. This finding, in conjunction with to permit displacement, selection may pro- Lack's conclusion (1954) that food supply is a mote unilateral or mutual "exaggeration" of major factor in the limiting of population size morphologic and behavioral differences func- in birds, argues strongly for competition as an tioning to lessen detrimental interactions. (An important factor in community organization. example is provided by Vaurie's study, 1951, In regard to the problem of reproductive of nuthatches of the genus Sitta.) (3) When isolation, it is clear that the efficiency of an species having similar ecologic spheres be- isolating mechanism developed prior to a come secondarily sympatric in a region of nar- secondary contact of populations may be row habitat diversity, an adjustment involv- critical in determining the subsequent course ing displacement of characters may not be pos- of evolutionary development of the popula- sible, and one species will ultimately replace tions, especially when two populations be- the other as a result of interspecific competi- come sympatric over a very limited area. tion. (4) A fourth possible outcome of the When the zone of sympatry is narrow and the interaction of closely related species involves isolating mechanisms are less than fully effec- the meeting of species of such similar behav- tive, a stable situation may arise in which ior and structure that they are ecologically selection prevents extensive introgression and incompatible within normal ranges of habitat the two populations remain connected by a diversity. Even in areas of gently sloping narrow zone of hybridization. Even if the environmental gradients, a sharp line of con- hybrids are selectively inferior to the parental tact between adjacent ranges is established, types, selection cannot enhance the isolation. with little if any actual sympatry. In the Dispersal of pure types into the contact zone absence of sympatry, reenforcement of incipi- constantly disrupts the effects of reenforcing ent ecologic isolation cannot be effected, and selection, and at the same time there is little the situation may remain stable for long chance that the reenforced genotypes will periods, with neither species being able to spread back into pure allopatric segments of invade the range of the other. the parental populations, since they are there Our studies of Centurus have revealed no selectively neutral if, indeed, not disadvan- unequivocal evidence of reenforcement of tageous (J. A. Moore, 1957). either reproductive or ecologic isolating In similar fashion, the outcome of second- mechanisms. However, patterns of geo- ary contacts between species faced with the graphic distribution among Centurus species problem of making ecologic adjustments to provide circumstantial evidence that closely reduce competition and other forms of inter- related, morphologically similar species may ference may depend on the degree of ecologic be ecologically incompatible. Indeed, the isolation attained at the time contact is estab- allopatric distribution of closely related spe- lished. It may also depend on the degree of cies is sufficiently marked to suggest that habitat diversity in the zone of sympatry these woodpeckers may encounter special (Amadon, 1950; Dixon, 1961; Hamilton, problems in achieving the ecologic adjust- 1962a) and on the extent of the area over ments necessary for sympatry. which sympatry is initially established, which Within the group of species usually placed in turn may be related to the character of in the genus Centurus (sensu stricto, as dis- gradients of environmental factors in the re- tinct from the "Tripsurus" group), there is gion of contact. Considering the general prob- only one case of sympatry over an extensive lem of ecologic isolation in birds, we may area. This occurs on the Yucatan Peninsula lump the results of secondary contacts of and Cozumel Island (see fig. 1), where C. species into four categories: (1) If two species aurifrons dubius is sympatric with C. pyg- have evolved sufficiently distinctive ecologic maeus. It is significant, we believe, that the spheres, extensive sympatry may be estab- two woodpeckers are markedly dissimilar in lished with minor if any adjustments in forag- size, being in fact extremes in this regard 1963 SELANDER AND GILLER: CENTURUS 259 within the genus as a whole. The larger form, grams). The degree of size difference is less C. a. dubius, has an average body weight of 87 than that between C. aurifrons and C. grams (table 2), while the smaller species pygmaeus on the Yucatin Peninsula but con- weighs only 47 grams, and in bill length and siderably greater than that between C. auri- other linear dimensions the two sympatric frons and C. urpygialis, species that are species show no overlap whatever. allopatric. Ecologic distribution and foraging behavior All other cases of sympatry over a consider- of these species have not been studied in able area involve one species belonging to the detail, but the conspicuous size differences Centurus group and one belonging to the suggest that the sizes if not the types of food "Tripsurus" section of the genus. For ex- taken are different, and the isolating effect of ample, C. pucherani (a " Tripsurus" type), the presumed feeding difference appears to be ranging from Colombia north through the supported by partial habitat isolation. Ac- Caribbean side of Central America, Oaxaca, cording to Paynter (1955, p. 165), C. auri- and Puebla, is sympatric with C. aurifrons in frons, which is the most abundant wood- southeastern Mexico. Little is known of C. pecker on the Yucatin Peninsula, is most pucherani in M6xico, but to judge from pub- common in tall and moderately tall deciduous lished reports it is decidedly rare and is con- forest and in second growth within the zone of fined to rain-forest habitats, whereas C. auri- rain forest, whereas C. pygmaeus is most frons is abundant and exhibits a much greater common in short and moderately tall decidu- latitude in habitat selection. South of the ous forest and in coastal scrub. The latter range of C. aurifrons, C. pucherani is report- species occurs only sparingly in the tall forest edly abundant in "semi-open" rain forest in habitat "preferred" by C. aurifrons. Thus the Costa Rica (Slud, 1960, p. 98) and has also species reach greatest population densities in been reported from rain forest in Nicaragua different habitats, but both are found in the (Howell, 1957, p. 87). It is probably not with- same stands of vegetation, in which, however, out significance that C. pucherani and the they exhibit stratal separation in foraging, race of C. aurifrons with which it is sympatric the smaller species foraging usually within a in southeastern M6xico differ markedly in few feet of the ground (Blake, 1953, p. 298), size. Wing length in C. pucherani from Mexico while the larger species forages at higher averages 112 mm. in males and 110 mm. in levels. females; bill lengths are 24.0 mm. for males, In sum, ecologic isolation between these 23.3 mm. for females. Corresponding meas- sympatric species is apparently achieved by urements for C. aurifrons dubius from the habitat segregation, stratal separation for same region are: wing length, 133 mm. in foraging, and, as suggested by differences in males, 127 mm. in females; bill length, 32.2 size, specialization on food of different sizes mm. in males, and 28.3 mm. in females (meas- and perhaps types. urements from Ridgway, 1914). The two species are similar in color and Too few data are available to warrant a pattern and show parallel variation geograph- discussion of ecologic relationships among the ically, each being ventrally dark on Cozumel widely distributed species Centurus rubri- Island. Griscom (1926, p. 9) reports that the capillus and three "Tripsurus" types with notes of the smaller species are distinctly which it is sympatric: C. chrysauchen in Costa weaker than those of the larger species. Noth- Rica and Panam6; C. pulcher (C. chrysauchen ing is known of their territorial relationships. puicher, according to Peters, 1948, p. 166) in The distributional relationships of C. chrys- Colombia; and C. cruentatus in Venezuela. ogenys and C. hypopolius are discussed above Available data on the contact between C. (p. 252). There is some overlap in ranges, but rubricapillus and C. chrysauchen in Central effective ecologic isolation is achieved by America indicate isolation by marked habitat marked habitat differences. In the region of differences (Carriker, 1910; Skutch, 1948; sympatry, the two species exhibit a consider- Aldrich and Bole, 1937, p. 81). able size difference (e.g., the weights of males Within the Centurus carolinis-aurifrons- of C. hypopolius range from 48 to 53 grams; uropygialis complex (superspecies), sharply those of males of C. chrysogenys, from 80 to 85 defined zones of contact suggest that distribu- 260 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 124 tional limitations are imposed by inter- all parts of C. carolinus territory in the east- specific interactions. In Texas, C. aurifrons ern United States is easily within the evolu- ranges north to the Panhandle and east to the tionary potential of C. aurifrons. Ecologic central part of the state, where it is in contact incompatability of species is also suggested with C. carolinus along a broad front in a by the distributional relationships of C. auri- region environmentally intermediate between frons and C. uropygialis, as discussed above. semiarid areas to the southwest and mesic (A comparable situation that involves appar- areas to the east (fig. 15). Nowhere is there ent ecologic incompatability of two species of extensive overlap in ranges, but locally, as at pocket gophers of the genus Geomys has been Austin, the species are sympatric in a zone studied by Kennerly, 1959; and Stresemann, from 2 to 5 miles wide (Selander and Giller, 1939, cites several cases in which the presence 1959b). The two species populations un- of one species of bird seems to exclude other doubtedly are adapted to different ranges of species from the area; see also Kendeigh and temperature, humidity, and other environ- Baldwin, 1937.) mental factors, but the fact that both occur in In the Centurus carolinus-aurifrons-uro- normal population densities up to the line of pygialis complex, as in the genus as a whole, contact suggests that neither is at its limits of the evolution of reproductive isolating mech- tolerance for these factors. anisms apparently tends to proceed more The two species in Texas are nearly identi- rapidly than does the evolution of ecologic cal in all linear dimensions and in weight but isolation, with the result that the distribution show consistent differences in tail pattern and and, hence, indirectly the evolution, of these in coloration of the head, the nape being red species may be influenced to a significant in C. carolinus and yellow or orange-yellow in degree by interspecific interactions with C. aurifrons. The vocalizations are distinc- congeners. tive, although not markedly dissimilar, and In concluding this section, we wish to take are thought to serve, along with the morpho- the opportunity to amend and clarify our logic differences, as a basis for species dis- views on the ecologic significance of inter- crimination. As noted above, hybridization or specific territorialism in birds (Simmons, heterospecific pairing is unknown. 1951). Like several other authors, we have Where the two species meet, the frequency been guilty of assuming that species mani- of contact of individuals is limited by the festing interspecific territorialism are neces- occurrence of C. aurifrons in more xeric sarily demonstrating ecologic incompatibility habitats, such as mesquite woodland, and C. (Selander and Giller, 1959b), but it now carolinus in more mesic situations, such as seems probable that in certain cases this type oak woodland. But locally they occupy the of interspecific response may merely be the same stands of vegetation, which demon- fortuitous result of behavioral or morphologic strates a significant overlap in habitat occur- similarities releasing territorial behavior. If rence and in other aspects of their ecologic the appearance, displays, or vocalizations ranges. Moreover, within these stands, the (social releasers) of two species establishing two species behave territorially as a single sympatry are sufficiently similar, individuals species population, manifesting intense inter- of one species will respond to those of the specific territorialism. other species as to their own species, and When all lines of evidence are considered, it mutually exclusive territories will be estab- is difficult to escape the conclusion that C. lished. Some competition for space may re- aurifrons and C. carolinus are ecologically sult. When the resulting spacial segregation incompatible. This is not to suggest that C. of the two species is advantageous to one or aurifrons would immediately occupy all of the both species in reducing competition for food range of C. carolinus if the latter were re- or some other resource provided by the terri- moved. But in view of the close similarities of tory, selection will act to maintain the com- the two species and the high degree of ecologic mon social releasers and response mechanisms adaptability exhibited by C. aurifrons within which insure interspecific territorialism its present large geographic range, there is (Dixon, 1961, p. 200). But if the mutually reason to believe that occupancy of some or exclusive spacing is unnecessary in terms of 1963 SELANDER AND GILLER: CENTURUS 261 reducing interspecific competition for food or narrow, the disruptive effect of gene flow other resource, it seems probable that selec- from large populations adjacent to the zone tion would act to eliminate the interspecific may prevent selection from reducing the response by reducing the degree of similarity similarities, and interspecific territorialism in appearance or behavior of the two species. may persist although it serves no useful eco- However, where the zone of sympatry is very logic function.

SEXUAL DIMORPHISM IN SIZE Species of Centurus and other genera of tail exceeds that of bill length (Selander, woodpeckers show an impressive degree of 1958). (5) When we select those forms that variation in sexual dimorphism in size, par- show unusually large degrees (16% to 21%) ticularly in bill dimensions (pl. 56). Among of sexual dimorphism in bill length, the in- species considered in the present report, teresting fact emerges that four of the six extremes are represented by Centurus hypo- forms are insular endemics and the two re- polius, in which there is no significant sexual maining forms occur in an insular-like situa- difference in bill size, and Chryserpes striatus, tion on the southern end of the peninsula of in which the bill of the female is markedly Baja California. Maximum dimorphism is shorter than that of the male. In table 6, shown by Chryserpes striatus of Hispaniola, in sexual dimorphism in wing, tail, tarsus, and which bill length of the female is 21.3 per cent bill length is shown for Centurus species and less than that of the male, notwithstanding for several other genera of New World wood- the fact that sexual difference is only 9.6 per peckers. To express dimorphism, we have cent in tarsal length and is even less in other calculated the percentage difference between body dimensions. Relatively great sexual mean measurements of males and females. differences are also seen in bill depth (12.1%) From these data the following conclusions and bill width (11.7%). When data on bill emerge: (1) The degree of sexual dimorphism length in Chryserpes are graphed (fig. 17), it is varies within wide limits among various gen- evident that there is no observed overlap era, species, and subspecies. Unfortunately, between the sexes. weights are not available for all forms that we In Melanerpes portoricensis of Puerto Rico wish to compare, but tarsal length may serve and Melanerpes herminieri of Guadelupe as a moderately accurate index to general Island, Lesser Antilles, dimorphism in body body size. (2) In most species of woodpeckers, size, as reflected by tarsal length or wing males average slightly larger than females, length, is slightly greater than in other wood- the difference in tarsal length amounting to peckers studied (with the possible exception about 5 per cent on the average. In a few of Chryserpes), and dimorphism in bill length species, the female is as large as or slightly greatly exceeds that of other dimensions, larger than the male. (3) There is no apparent being about 19 per cent. In Xiphidiopicus relationship between general body size and percussus of Cuba, the tarsal length difference degree of sexual dimorphism in bill length or amounts to only 1.7 per cent, yet in bill length other dimensions; that is, larger species are the female is 15.2 per cent smaller than the not necessarily more dimorphic than smaller male. species. (4) In the great majority of species, Sexual dimorphism in Centurus uropygialis sexual dimorphism in size is greater in bill appears to be slightly greater than in any length than in other dimensions, amounting other species of Centurus, and dimorphism in on the average to about 9 per cent. Moreover, bill length reaches an extreme in C. u. brews- there is a greater range of intraspecific and teri of southern Baja California, in which it is interspecific variation in degree of dimor- 16.7 per cent; the nearest approach is in C. u. phism in bill length than in other linear uropygialis of western Mexico, with 12.9 per dimensions. Similar findings are reported for cent difference in bill length. Dendrocopos the Hawaiian honeycreepers (Drepaniidae) scalaris also shows marked dimorphism in bill by Amadon (1950, p. 187), but in the Icteri- length in southern Baja California. dae sexual dimorphism in length of wing and It is also noteworthy that the piculet 262 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 124

Nesoctites micromegas, a genus and species and the insular species of Melanerpes, and in endemic to Hispaniola and IGonave Island, is these forms (as in the woodpeckers men- dimorphic in size (with t-he female being tioned) dimorphism in bill length greatly larger than the male), whereas species of the exceeds that in wing length and in dimensions continental piculet genus .1 2icummus are es- of other body parts. As Amadon has noted sentially non-dimorphic in s3ize. (1950, p. 188), the significance of increased In discussing factors affecting sexual differ- dimorphism in these species is obscure; but, ences in size in birds, Amatdon (1950, 1953, since two of the three are larger than related 1959) has called attention to unusual degrees forms having less marked sexual differences in of sexual dimorphism in ccertain Hawaiian size, he does consider the possibility that we honeycreepers and in the VVest African sun- are dealing with cases of non-selective allome- bird Cyanomitra thomensi.s of Sao Tom6 try (Rensch, 1950) in which ontogenetic Island: growth rates responsible for sexual dimor- BILL phism in size in adults have been continued VING LENGTH phylogenetically to some extent in rapidly Pseudonestor xanthophrys 9 23 evolving insular forms perhaps subject to a Hemignathus procerus 5 22 minimum of competition and predation. But, Cyanomitra thomensis 12 27 as did Amadon (1959, p. 533), we believe that a purely allometric interpretation of this It is of some interest that the degree of phenomenon is unsatisfactory, for, although dimorphism in these insular species is roughly allometric factors may well be involved as a equivalent to that found in ( 'hryserpes striatus mechanism by which evolutionary changes in

TABLE 6 SEXUAL DIMORPHISM IN SIZE IN WOODPECKERS (Percentage difference between mean measurements of males and females.a) Form Locality Sample Bill Size5 Wingig TiTail Lengthc Tarsusaru Melanerpes portoricensis Puerto Rico 20-18d 4.9 5.9 18.9 10.0 Melanerpes herminieri Guadelupe Island, Lesser Antilles 20-13d 6.2 1.2 19.0 11.3 Melanerpes erythrocephalus E. United States 32-21 2.2 1.6 2.6 4.8 Balanosphyra f. formicivorus Mexico 41-44 4.4 1.3 6.7 6.2 Asyndesmus lewis W. United States 16-14 2.5 5.4 3.7 3.5 Chryserpes striatus Hispaniola 53-54d 4.4 +3.8 21.3 9.6 Xiphidiopicus p. percussus Cuba 10-10 1.8 +4.0 15.2 1.7 Xiphidiopicus p. insulae-pinorum Isle of Pines 4-4 1.1 +4.5 13.8 5.9 Piculus rubiginosus yucatanicus SE. Mexico-Central America 18-24 1.3 2.4 5.1 1.4 Veniliornis fumigatus sanguinolentus SE. Mexico-Panami 31-17 0.1 1.6 4.7 2.9 Centurus s. superciliaris Cuba 10-10 1.9 0.8 10.8 3.1 Centurus s. blakei Abaco Island 10- 9 2.8 0.9 11.3 4.8 Centurus s. caymanensis Grand Cayman 10- 9 3.1 1.4 9.5 5.1 Centurus carolinus E. United States 41-40 2.1 0.8 7.4 4.1 Centurus radiolatus Jamaica 10-10 2.0 2.7 6.2 4.0 Centurus aurifrons dubius Yucatin Peninsula 14-15 4.6 3.6 12.1 7.2 Centurus a. aurifrons N. M6xico-Texas 40-32 3.2 3.6 7.9 3.0 Centurus a. polygrammus Oaxaca-Chiapas 12-10 4.2 5.0 12.3 6.3 Centurus a. santacruzi Central America 12-10 1.2 0.3 7.6 2.2 Centurus hoffmanni Costa Rica-Nicaragua 13-18 1.8 3.7 9.9 4.4 Centurus rubricapillus seductus Isla El Rey 10- 8 2.0 1.6 10.4 6.3 Centurus rubricapillus rubricapillus Panam&-Costa Rica 20-20 4.1 5.7 11.4 3.7 1963 SELANDER AND GILLER: CENTURUS 263 TABLE 6-(Continued)

Form Locality SampleSize Wing Tail LengthBill Tarsus Centurus pygmaeus rubricomus Yucatdn 10-10 1.4 1.1 11.6 3.7 Centurus uropygialis brewsteri Baja California 10-10 3.8 6.0 16.7 7.1 Centurus u. uropygialis W. Mexico-Arizona 33-21 2.4 9.9 12.9 5.3 Centurus hypopolius Central Mexico 7-4 1.1 +4.2 0.8 5.4 Centurus c. chrysogenys Sinaloa-Nayarit 10-10 1.6 1.8 7.1 4.2 Centurus p. pucherani Honduras-Panam& 30-15 1.1 0.7 6.6 3.0 Centurus chrysauchen Panam&-Costa Rica 15-11 2.0 2.7 9.3 3.6 Centurus flavifrons South America 12-11 - 5.1 Centurus "rubrifrons"f South America 15-10 - 9.6 Neoceleus fernandinae Cuba 5-5 2.0 4.1 6.6 1.0 Colaptes auratus chrysocaulosus Cuba 10-10 0.8 1.1 2.1 1.1 Colaptes a. auratus SE. United States 22-22 0.4 1.4 4.1 2.2 Colaptes c. chrysoides S. Baja California 10-8 1.6 4.9 2.3 +1.1 Colaptes c. mearnsi Calif.-New Mexico 10-10 0.9 1.3 3.2 0.4 Colaptes cafer rufipileus Guadalupe Island 6-7 +1.7 +0.4 +0.2 +2.6 Colaptes cafer cafer California 10-10 1.8 3.0 4.0 7.7 Colaptes cafer saturator NW. United States 10-10 1.7 1.4 5.3 4.3 Colaptes mexicanoides Chiapas-Guatemala 16- 8 2.0 1.1 6.8 3.8 Dendrocopos scalaris lucasanus S. Baja California 10-10 2.7 +1.3 15.6 7.3 Dendrocopos s. eremicus N. Baja California 10- 5 4.9 1.6 11.2 7.4 Dendrocopos s. cactophilus SW. United States- N. Sonora 35-36 2.4 +2.6 12.8 5.5 Nesoctites micromegas Hispaniola 10-10 +4.3 +6.4 + 9.6 +2.9 Picummus olivaceus flavotinctus Costa Rica 10-10 0.8 0.0 0.0 0.0

a Measurements from Ridgway (1914), except as noted. b Indicates numbers of males and females measured; for example, "20-18" designates a sample of 20 males and 18 females. Exposed culmen, not culmen measured from anterior rim of nostril. d Specimens in the American Museum of Natural History, Museum of Comparative Zoology, and the United States National Museum. 0 Specimens in the American Museum of Natural History. f See page 221 for nomenclature. dimorphism are effected, it is our view that of wing length and tarsal length does not bill size in the two sexes would remain under seem consistent with such a hypothesis. the influence and control of selective forces at It would be interesting to know if increased all times. In any event, in the case of the dimorphism in bill size in the insular wood- Caribbean woodpeckers, it seems unlikely peckers has been achieved by unilateral or by that the marked sexual dimorphism in bill mutual displacement in the sexes. Since body size can be attributed to non-selective posi- weights are not available, bill size relative to tive allometry accompanying increase in body body mass cannot be compared directly, but size, for the insular birds are not larger than comparison of relative bill size is possible if less dimorphic continental types with which we assume that tarsal length is indicative of they can be compared. body mass. In Melanerpes bill length/tarsus In the absence of information indicating ratios (table 7) indicate that in insular types that the mating system is anything but monog- the bill of the male is relatively longer and the amous in the insular woodpeckers, it does bill of the female is relatively shorter than in not seem possible to invoke increased sexual the continental congener M. erythrocephalus selection as a causative agent in the increased and the related continental form Balanos- sexual dimorphism in size. Moreover, the fact phyra formicivorus. Chryserpes has no known that dimorphism in size of bill outstrips that close relative for comparison, but when it is 264 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 124 BILL LENGTH Chryserpes striraus n U1 n 1= 1, n-rJI, r-- 0 22 24 26 i2 30 32 34 36

Cenlurus curifrons

22 24 26 28 30 32 34 36 22 I 24 2;6 28 30 32 314 36 a

BILL DEPTH C. striatus C. curilrons

______n ri6.070 8.0 6.0 7.i .0 9.0 FIG. 17. Comparison of bill dimensions in Chryserpes striatus and Centurus a. aurifrons (sample area 3: southern Texas, Tamaulipas, and Nuevo Le6n). Dark squares, males; light squares, females. Scale is in millimeters. compared with the continental species Cen- the insular species is related to major differ- turus aurifrons, there is, as in the case of ences in the roles of the sexes in construction Melanerpes, a suggestion of a two-way dis- of roost or nest cavities, but it is at least placement in bill size. Apparently evolution equally probable that the dimorphism is of the insular types from less dimorphic an- somehow related to sexual differences in for- cestral forms has involved a decrease in the aging behavior. relative size of the bill in the female and an Some examples of sexual dimorphism in increase in the relative size of the bill in the size that is apparently related to differences male. We may tentatively conclude that the in foraging and feeding behavior in birds have increased sexual dimorphism in bill size in been given by Rand (1952) and Storer (1952), these woodpeckers has been achieved by a who regard such differences as a means of mutual, two-way displacement, but this reducing intraspecific competition for food hypothesis requires testing by a comparison (but see critical comments by Cade, 1960, pp. of the ratios of bill length to body mass. 241-246). The classical case is the now- extinct Heterolocha acutirostris, in which the POSSIBLE ADAPTIVE SIGNIFICANCE long, thrasher-like bill of the female was used Although the bill of birds is primarily a for probing into crevices, while the shorter, feeding device, it has a number of secondary icterid-like bill of the male was used for dig- functions, including its use as an aggressive ging into wood for insects. If selection can weapon or object of display and as a tool in produce extreme sexual differentiation of the nest building. Therefore, evolution of bill size degree shown by Heterolocha, it seems entirely may be determined by a variety of selective reasonable to expect lesser degrees of differ- forces, and it is likely that the functional ence arising by selection in other species of significance and adaptiveness of sexual dimor- birds. phism in bill size and in other dimensions in In populations showing increased sexual woodpeckers will remain unknown until com- dimorphism in bill size, variability within parative studies of ecology and behavior of each sex group will determine whether or not the insular and continental types are avail- the total span of bill length (or other dimen- able. Possibly the increased dimorphism in sion) for the combined sexes is also increased. 1963 SELANDER AND GILLER: CENTURUS 265 The comparison (table 8) of coefficients of TABLE 8 variability in bill dimensions in Chryserpes COEFFICIENTS OF VARIABILITY striatus and Centurus a. aurifrons (sample area 3) suggests that increased dimorphism in Species Sex Bill length Bill depth Chryserpes has not been accompanied by a change in variability within either sex group. Chryserpes striatus c' 4.29 5.25 Since variability is approximately equal in 9 5.39 5.23 the two species, increased dimorphism in bill Centurus aurifrons e 2.87-5.86 2.47-4.92 length in Chryserpes to a point of non-overlap between the sexes results in an increase of approximately one-third in the total span of populations. Thus reduced competition for bill lengths in the population (see fig. 17). food between the sexes could have selective If the total span in bill dimensions is re- value on both the individual and population lated to the total span of sizes or types of food level. items taken, the increased sexual dimorphism In attempting to develop a working hy- in Chryserpes may have the following ecologic pothesis to account for interspecific variation significance: (1) the total span of food sizes or in degree of sexual dimorphism in size in types taken by the population is increased woodpeckers, we have been impressed with over that taken by Centurus aurifrons and the fact that the most striking dimorphism is other moderately dimorphic woodpeckers; shown by three insular species, each of which and (2) the degree of overlap in food taken by is the only resident species of typical wood- the two sexes is reduced as each sex specializes pecker on its respective island. Thus Chrys- to a greater degree on a separate range of food erpes striatus is the only woodpecker of His- sizes or types. If food supply is limiting to paniola, apart from the endemic piculet woodpecker populations, non-overlap in sizes Nesoctites, which is nuthatch-like in foraging or types of food items taken by the sexes may habits, and the yellow-bellied sapsucker, facilitate maintenance of optimally large which is a winter visitant; on Puerto Rico, where Melanerpes portoricensis is the only TABLE 7 resident form, the sapsucker occurs as a rare BILL LENGTH/TARSUS RATIOS winter visitant; and apparently no other IN SEVERAL WOODPECKERS species but the resident Melanerpes herminieri is known from Guadelupe Island (Bond, Bill 1961). Since these islands are ecologically Species Sex TarsusLength/ diversified and therefore seemingly provide habitats for several woodpecker species, it is Insular species tempting to speculate that freedom from Melanerpes portoricensis ci 1.27 competition with species of similar adaptive 9 1.15 type has permitted the insular species to Melanerpes herminieri ci 1.28 achieve a wider and more thorough exploita- 9 1.17 tion of available food resources through evo- lutionary divergence of the sexes with regard Chryserpes striatus 1.42 to foraging behavior, which in turn has been 9 1.23 accompanied by increased sexual dimorphism in size, and particularly in bill dimensions. Continental species However, it may be gratuitous to assume that Melanerpes erythrocephalus e 1.19 unoccupied "woodpecker niches" are avail- 9 1.21 able on these islands. If one is willing to as- sume for the sake of Balanosphyraformicivorus oi 1.20 argument that the food 9 supply in habitats occupied by the insular 1.19 forms is relatively poor, it seems possible that Centurus aurifrons 1.36 the increased sexual dimorphism and pre- 9 1.28 sumed reduction in intraspecific competition for food represents an adaptive response 266 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 124 essential to the maintenance of the resident phenomenon of general occurrence among populations and is quite unrelated to the insular species. A preliminary survey of the absence of competition with other wood- passerine avifauna of the Antilles, based on pecker species. In other words, it is possible size data in the literature, suggests that it is that the insular forms have managed to not, for none of the species appears to be survive in otherwise suboptimal habitats clearly more dimorphic than related continen- through adaptations that minimize the ex- tal forms. This evidence strongly suggests tent of overlap between the sexes in exploita- that the marked dimorphism in certain in- tion of limited food resources. sular woodpeckers does not exemplify a In future investigations into the adaptive principle of broad application to insular significance of sexual dimorphism in bill di- faunal types but instead is somehow related mensions in birds, it will be desirable to deter- to factors that are peculiar to woodpecker mine if increased sexual dimorphism is a behavior or ecology. SUMMARY THIS STUDY is concerned with the systematic distinct from C. aurifrons. relationships among certain populations of Field work in restricted zones of sympatry woodpeckers of the genus Centurus occurring between Centurus aurifrons and C. uropygialis in the United States, Mexico, and Central in Aguascalientes and Jalisco indicates that America. Significant conclusions arising from the two forms are specifically distinct. Ap- this investigation, which has involved field proximately 5 per cent of the specimens taken work in Mexico in 1958 and examination of in the zones of sympatry show evidence of large numbers of museum specimens, are as hybrid origin, but gene exchange between the follows: two forms is not extensive, if in fact it occurs The generic name Tripsurus is regarded as at all. The presence of traces of yellow pig- a synonym of Centurus, since there is no ment on the nape of specimens of C. uropygi- known character or combination of characters alis throughout its range is interpreted not as that distinguishes species currently assigned evidence of present-day introgression of genes to these two nominal genera. Moreover, one from C. aurifrons but as a vestige of a char- species, chrysogenys, that is invariably as- acter more fully developed in a population signed to Centurus is actually more closely ancestral to C. uropygialis and currently related to species currently assigned to Trip- being lost in that species. surus. Centurus (sensu lato) is closely allied A specific relationship of Centurus aurifrons with, and possibly inseparable from, Melan- and C. carolinus is demonstrated by their erpes. sympatry without interbreeding in a narrow Quantitative analysis of morphologic char- zone in Texas. acters of specimens that represent popula- Contrary to the suggestion of Peters tions of the aurifrons complex fully supports (1948), Centurus hypopolius is probably not the view of Ridgway (1881) and Griscom conspecific with C. uropygialis, and it may be (1932) that these populations belong to a more closely allied to C. chrysogenys, with single, highly polytypic species, since even which it is sympatric in Puebla and Morelos. the most strongly differentiated forms (C. a. The Haitian woodpecker, Chryserpes stria- aurifrons and C. a. dubius) intergrade in zones tus, which has been assigned to Centurus by of contact. It is advocated that the number of Bond and others, is not closely allied to any subspecies for continental populations be species of Centurus or Melanerpes and should limited to five; these are C. a. aurifrons (in- be placed in the monotypic genus Chryserpes, cluding C. a. incanescens of recent classifica- as advocated by W. D. Miller, pending fur- tions), C. a. dubius (including C. a. leei of ther study of its relationships. Cozumel Island, Yucatin), C. a. grateloupen- The significance of ecologic range or ampli- sis (including C. a. veraecrucis), C. a. santa- tude in distribution and speciation in birds is cruzi (including C. a. pauper and C. a. fumo- discussed. For Centurus it is suggested that sus), and C. a. polygrammus (including C. a. the evolution of reproductive isolating mech- frontalis). anisms tends to proceed more rapidly than The color and pattern of the plumage in does the evolution of ecologic isolation, with Centurus species are presumably cryptic, and the result that competition and other inter- geographic variation in these characters specific interactions prevent extensive sympa- shows a reasonably close correlation with try of closely related species of the genus. regional variation in climate and vegetation The possible ecologic significance of in- type. The brightly colored, sexually dimor- creased degrees of sexual dimorphism in size, phic areas of the head (and perhaps also the particularly in bill length, in certain insular brightly colored belly region) presumably woodpeckers is discussed, but, in the absence have epigamic significance. of comparative ecologic and behavioral stud- It is probable that Centurus hoffmanni of ies, no satisfactory explanation of the phe- Costa Rica and Nicaragua is specifically nomenon can be given. 267 APPENDIX Reference specimens used in the analysis of southwest of Veracruz, Veracruz, March 13, 1941. color and pattern in C. aurifrons and C. hoff- (5) M.Z.L.S.U. No. 11276; Axtla River [=Rio manni (p. 222) are listed below, with category Huichihuayan], San Luis Potosi, April 29, 1948. numbers in parentheses. All specimens are (6) U.S.N.M. No. 359746; Tres Zapotes, Vera- cruz, March 7, 1940. (7) M.Z.U.M. No. 137879: adult males of C. aurifrons, except as noted. Belize town, British Honduras, January. NAPE COLOR: Fourteen categories: (0) BELLY COLOR: Twelve categories: (0) U.S.- M.Z.U.M. No. 109172; 10 kilometers west of N.M. No. 193924; San Vicente, Chiapas, April. Tuxtla Guti6rrez, 800 meters, Chiapas, March (1) M.V.Z. No. 124893; 11 miles northwest of 24, 1941. (1) M.Z.U.M. No. 109173; 10 kilometers Jourdanton, 700 feet, Atascosa County, Texas, west of Tuxtla Guti6rrez, 800 meters, Chiapas, January 30, 1952. (2) M.V.Z. No. 124891; 11 March 24, 1941. (2) U.S.N.M. No. 193916; Can- miles northwest of Jourdanton, Atascosa County, job, Chiapas, March 26, 1904. (3) R.K.S. No. Texas, January 30, 1952. (3) M.V.Z. No. 41531; 3188; 5 kilometers west of Aguascalientes, Brownsville, Texas, January 27, 1911. (4) Aguascalientes, June 14, 1958. (4) R.K.S. No. M.Z.L.S.U. No. 15217; Tepeyac, San Luis 3203; Puente Grande, Rio Santiago, Jalisco, June Potosi, June 14, 1950. (5) U.S.N.M. No. 34898; 16, 1958. (5) D.C. No. 23495; Sabinas, Coahuila, San Jose?, Costa Rica, no date (C. hoffmanni). (6) February 15, 1910. (6) U.S.N.M. No. 183337; D.C. No. 16563; Mt. Cacaguatique, Dept. San Nuevo Laredo, Tamaulipas, December 29, 1901. Miguel, El Salvador, December 11, 1925. (7) (7) R.K.S. No. 3060; 11 miles southeast of Pear- M.Z.U.M. No. 107699; Esperanza, Escuintla, sall, Frio County, Texas, March 22, 1958. (8) Chiapas, January 2, 1940. (8) Female, U.S.N.M. M.Z.L.S.U. No. 11274; 1 mile west by north of no. 154831; Guichicovi, Oaxaca, June 24, 1895. Xilitla, 2500 feet, San Luis Potosi, January 27, (9) M.Z.U.M. No. 108032; Panzos, 30 meters, 1947. (9) M.Z.L.S.U. No. 18067; Rancho Sabinal, Guatemala, March 5, 1940. (10) M.Z.U.M. No. 200 feet, San Luis Potosi, November 16, 1951. 100194; Minatiltlfin, Veracruz, April 26, 1939. (10) D.C. No. 16593; Mt. Cacaguatique, Dept. (11) U.S.N.M. No. 167436; Cozumel Island, San Miguel, El Salvador, December 14, 1925. (11) Yucatfin, April. M.Z.U.M. No. 109179; 8 kilometers southwest of DORSAL TAIL PATTERN: Seven categories: (0) Veracruz, Veracruz, March 21, 1941. (12) D.C. M.Z.U.M. No. 109171; 4 kilometers west of No. 16546; Mt. Cacaguatique, Dept. San Miguel, Tuxtla Gutierrez, 800 meters, Chiapas, April 21, El Salvador, December 10, 1925. (13) M.Z.U.M. 1941. (1) U.S.N.M. No. 57836; Chihintan [?], No. 70584; Cayo, British Honduras, March 16, Tehuantepec, Oaxaca, November 20, 1868. (2) 1931. M.Z.U.M. No. 137894; Tehuantepec, Oaxaca, FRONTAL REGION COLOR: Eight categories: (0) December 21, 1916. (3) M.Z.U.M. No. 102385; U.S.N.M. No. 132513; Turrialba, Costa Rica, Uni6n Juirez, 1300 meters, Volcfin Tacanfi, Chia- May (C. hoffmanni). (1) U.S.N.M. No. 154936; pas, March 18, 1939. (4) U.S.N.M. No. 359745; Lagos, Jalisco, June 27, 1896. (2) U.S.N.M. No. Tres Zapotes camp, Veracruz, January 20, 1940. (5) 158579; Forlon, Tamaulipas, May 28, 1898. (3) M.Z.L.S.U. No. 16905; Puente de Dios, San Luis M.Z.L.S.U. No. 11261; 1 mile west (by highway) Potosi, March 25, 1957. (6) M.Z.U.M. No. of Huizache, 4300 feet, San Luis Potosi, February 137886; Belize town, British Honduras, April 16, 1948. (4) M.Z.U.M. No. 109180; 8 kilometers 10, 1931.

LITERATURE CITED ALDRICH, J. W., AND B. P. BOLE, JR. lection. Ibid., vol. 100, art. 3, pp. 393- 1937. The birds and mammals of the western 452. slope of the Azuero Peninsula. Cleveland 1959. Significance of sexual differences in size Mus. Nat. Hist. Sci. Publ., vol. 7, 196 among birds. Proc. Amer. Phil. Soc., pp- vol. 103, pp. 531-536. AMADON, D. 1950. The Hawaiian honeycreepers (Aves, AMADON, D., AND D. R. ECKELBERRY Drepaniidae). Bull. Amer. Mus. Nat. 1955. Observations on Mexican birds. Condor, Hist., vol. 95, art. 4, pp. 151-262. vol. 57, pp. 65-80. 1953. Avian systematics and evolution in the ANDREWARTHA, H. G., AND L. C. BIRCH Gulf of Guinea: the J. G. Correia col- 1954. The distribution and abundance of ani- 268 1963 SELANDER AND GILLER: CENTURUS 269

mals. Chicago, University of Chicago BROWN, W. L., JR., AND E. 0. WILSON Press, 782 pp. 1956. Character displacement. Syst. Zool., BAILEY, H. H. vol. 5, pp. 49-64. 1906. Ornithological notes from western Mex- BURLEIGH, T. D., AND G. H. LOWERY, JR. ico and the Tres Marias and Isabella 1944. Geographical variation in the red- Islands. Auk, vol. 23, pp. 369-391. bellied woodpecker in the southeastern BAIRD, S. F., T. M. BREWER, AND R. RIDGWAY United States. Occas. Paper Louisiana 1874. A history of North American birds. State Univ. Mus. Zool., no. 17, pp. Boston, Little, Brown and Co., vol. 2, 293-301. 590 pp. CABANIS, J. BANCROFT, G. 1862. Uebersicht der im Berliner Museum 1929. The breeding birds of central Lower befindlichen Vogel von Costa Rica. California. Condor, vol. 32, pp. 20-49. Jour. fur Ornith., vol. 10, pp. 321-336. BETTS, M. M. CABOT, S., JR. 1955. The food of titmice in oak woodland. 1844. [Description of three new species of Jour. Animal Ecol., vol. 24, pp. 282-323. woodpeckers.] Proc. Boston Soc. Nat. BLAIR, W. F. Hist., vol. 1, pp. 164-165. 1958. Mating call in the speciation of anuran CADE, T. J. amphibians. Amer. Nat., vol. 42, pp. 1960. Ecology of the peregrine and gyrfalcon 27-51. populations in Alaska. Univ. California BLAKE, E. R. Publ. Zool., vol. 63, pp. 151-290. 1953. Birds of Mexico: a guide for field identi- CARRIKER, M. A., JR. fication. Chicago, University of Chicago 1910. An annotated list of the birds of Costa Press, 644 pp. Rica, including Cocos Island. Ann. Car- BOCK, W. J. negie Mus., vol. 6, pp. 314-915. 1961. Salivary glands in the gray jays (Peri- soreus). vol. CORY, C. B. Auk, 78, pp. 355-365. 1919. Catalogue of birds of the Americas and BOGERT, C. M. the adjacent islands. Pt. 2, no. 2. Publ. 1960. The influence of sound on the behavior Field Mus. Nat. Hist., zool. ser., vol. 13, of amphibians and reptiles. In Lanyon, pp. 317-607. W. E., and W. N. Tavolga (eds.), Ani- mal sounds and communication. Amer. CROWELL, K. Inst. Biol. Sci. Publ., no. 7, pp. 137-320. 1961. The effects of reduced competition in BONAPARTE, C. L. birds. Proc. Natl. Acad. Sci., vol. 47, 1837. Description of new or interesting birds pp. 240-243. from South America and Mexico. Proc. DAVIS, L. I. Zool. Soc. London, pp. 108-122. 1952. Winter bird census at Xilitla, San Luis BOND, J. Potosi, Mexico. Condor, vol. 54, pp. 1936. Resident birds of the Bay Islands of 345-355. Spanish Honduras. Proc. Acad. Nat. DAVIS, W. B., AND R. J. RUSSELL Sci. Philadelphia, vol. 88, pp. 353-364. 1953. Aves y mamiferos del Estado de More- 1956. Check-list of birds of the West Indies. los. Rev. Soc. Mexicana Hist. Nat., Philadelphia, Academy of Natural Sci- vol. 14, pp. 77-147. ences of Philadelphia, 214 pp. DEIGNAN, H. G. 1961. Birds of the West Indies. Boston, 1936. Notes on a small collection of birds from Houghton Mifflin Co., 256 pp. the Republic of Honduras. Auk, vol. 13, BOWMAN, R. I. pp. 55-62. 1961. Morphological differentiation and adap- DICKEY, D. R., AND A. J. VAN RoSSEM tation in the Gal'apagos finches. Univ. 1938. The birds of El Salvador. Publ. Field California Publ. Zool., vol. 58, pp. 1-302. Mus. Nat. Hist., zool. ser., vol. 23, pp. BRODKORB, P. 1-609. 1943. Birds from the Gulf lowlands of south- DIXON, K. L. ern Mexico. Misc. Publ. Univ. Michigan 1961. Habitat distribution and niche relation- Mus. Zool., no. 55, pp. 5-88. ships in North American species of 1957. Birds. In Blair, W. F., et al., Vertebrates Parus. In Blair, W. F. (ed.), Vertebrate of the United States. New York, Mc- speciation. Austin, University of Texas Graw-Hill Book Co., Inc., pp. 361-613. Press, pp. 179-216. 270 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 124 EDWARDS, E. P., AND R. B. LEA area of Nicaragua. Condor, vol. 59, 1955. Birds of the Monserrate area, Chiapas, pp. 73-111. Mexico. Condor, vol. 57, pp. 31-54. HUTCHINSON, G. E. EISENMANN, E. 1958. Concluding remarks. Cold Spring Har- 1955. The species of Middle American birds. bor Symp. Quant. Biol., vol. 22, pp. 415- Trans. Linnaean Soc. New York, vol. 427. 2, pp. 1-128. HUTCHINSON, G. E., AND R. MACARTHUR ELTON, C. 1959. Appendix: on the theoretical signifi- 1946. Competition and the structure of eco- cance of aggressive neglect in inter- logical communities. Jour. Animal Ecol., specific competition. Amer. Nat., vol. vol. 15, pp. 54-68. 93, pp. 133-134. FOX, D. L. INGER, R. F. 1953. Animal biochromes and structural col- 1961. Problems in the application of the sub- ors. London and New York, Cambridge species concept in vertebrate taxonomy. University Press, 190 pp. In Blair, W. F. (ed.), Vertebrate specia- GIBB, J. tion. Austin, University of Texas Press, 1954. Feeding ecology of tits, with notes on pp. 262-285. treecreeper and goldcrest. Ibis, vol. KENDEIGH, S. C., AND S. P. BALDWIN 1937. Factors affecting yearly abundance of 96, pp. 513-543. GILMAN, F. M. passerine birds. Ecol. Monogr., vol. 7, 1915. of pp. 91-124. Woodpeckers the Arizona lowlands. KENNERLY, T. Condor, vol. 17, pp. 151-163. E., JR. 1959. Contact between the ranges of two GRINNELL, J., AND A. H. MILLER allopatric species of pocket gophers. 1944. The distribution of the birds of Cali- Evolution, vol. 13, pp. 247-263. fornia. Pacific Coast Avifauna, no. 27, KOELZ, W. N. 608 pp. 1954. Ornithological studies II. A new sub- GRIScOM, L. species of red-bellied woodpecker from 1926. The ornithological results of the Mason- Texas. Contrib. Inst. Regional Explora- Spinden Expedition to Yucatan. Part tion, vol. 1, p. 32. II. Chinchorro Bank and Cozumel Is- LACK, D. land. Amer. Mus. Novitates, no. 236, 1944. Ecological aspects of species-formation pp. 1-13. in passerine birds. Ibis, vol. 86, pp. 1932. The distribution of bird-life in Guate- 260-286. mala. Bull. Amer. Mus. Nat. Hist., vol. 1947. Darwin's finches. London, Cambridge 64, pp. 1-439. University Press, 208 pp. HAMILTON, T. H. 1949. The significance of ecological isolation. 1962a. Species relationships and adaptations In Jepsen, G. L., E. Mayr, and G. G. for sympatry in the avian genus Vireo. Simpson (eds.), Genetics, paleontology, Condor, vol. 64, pp. 40-68. and evolution. Princeton, Princeton 1962b. The habitats of the avifauna of the University Press, pp. 299-308. mesquite plains of Texas. Amer. Mid- 1954. The natural regulation of animal num- land Nat., vol. 67, pp. 85-105. bers. London, Clarendon Press, 343 pp. HARGITT, E. LACK, D., AND H. N. SOUTHERN 1890. Catalogue of the birds in the British 1949. Birds on Tenerife. Ibis, vol. 91, pp. Museum. Scancores, containing the 607-626. family Picidae. London, Taylor and LAMB, C. Francis, 597 pp. 1909. A glimpse of bird life on the west coast HARTLEY, P. H. T. of Mexico. Condor, vol. 12, pp. 74-79. 1953. An ecological study of the feeding habits LESSON, R. P. of the English titmice. Jour. Animal 1839. Oiseaux rares ou nouveaux de la collec- Ecol., vol. 22, pp. 261-288. tion du Docteur Abeill6 A Bordeaux. HAVERSCHMIDT, F. Rev. Zool., vol. 2, pp. 40-43. 1957. List of the birds of Surinam. Publ. LIDICKER, W. Z., JR. Found. Sci. Res. Surinam and Nether- 1960. An analysis of intraspecific variation lands Antilles, Utrecht, no. 13, 153 pp. in the kangaroo rat Dipodomys merri- HOWELL, T. R. ami. Univ. California Publ. Zool., vol. 1957. Birds of a second-growth rain forest 67, pp. 125-218. 1963 SELANDER AND GILLER: CENTURUS 271 MAcARTHUR, R. from central Mexico. Proc. Biol. Soc. 1957. On the relative abundance of bird spe- Washington, vol. 63, pp. 109-110. cies. Proc. Natl. Acad. Sci., vol. 47, NELSON, E. W. pp. 293-295. 1900. Descriptions of thirty new North Ameri- 1958. Population ecology of some warblers of can birds, in the Biological Survey col- northeastern coniferous forests. Ecol- lection. Auk, vol. 17, pp. 253-270. ogy, vol. 39, pp. 599-619. PAYNTER, R. A., JR. McLELLAN, M. E. 1955. The ornithogeography of the Yucatan 1927. Notes on birds of Sinaloa and Nayarit Peninsula. Bull. Peabody Mus., vol. in the fall of 1925. Proc. California 9, pp. 1-347. Acad. Sci., vol. 16, pp. 1-15. PETERS, J. MARTIN, P. S., C. R. ROBINS, AND W. B. HEED 1948. Check-list of the birds of the world. 1954. Birds and biogeography of the Sierra de Cambridge, Massachusetts, Harvard Tamaulipas, an isolated pine-oak habi- University Press, vol. 6, 259 pp. tat. Wilson Bull., vol. 66, pp. 38-57. PETERSON, R. T. MAYR, E. 1960. A field guide to the birds of Texas. Bos- 1954. Notes on nomenclature and classifica- ton, Houghton Mifflin Co., 304 pp. tion. Syst. Zool., vol. 3, pp. 86-89. PITELKA, F. A. 1959. Isolation as an evolutionary factor. 1951a. Speciation and ecologic distribution in Proc. Amer. Phil. Soc., vol. 103, pp. American jays of the genus Aphelocoma. 221-230. Univ. California Publ. Zool., vol. 50, pp. 195-464. MEXICAN CHECK-LIST COMMITTEE 1951b. Ecologic overlap and interspecific strife 1957. Distributional check-list of the birds of in breeding populations of Anna and Mexico. Pt. 2. Pacific Coast Avifauna, Allen hummingbirds. Ecology, vol. 32, no. 33, pp. 1-436. pp. 641-661. MILLER, A. H. QUILLIN, R. W., AND R. HOLLEMAN 1942. Habitat selection among higher verte- 1918. The breeding birds of Bexar County, brates and its relation to intraspecific Texas. Condor, vol. 20, pp. 37-44. variation. Amer. Nat., vol. 76, pp. 25- RAND, A. L. 35. 1952. Secondary sexual characters and eco- 1955a. Concepts and problems of avian sys- logical competition. Fieldiana (Zool.), tematics in relation to evolutionary pro- vol. 34, no. 6, pp. 65-70. cesses. In Wolfson, A. (ed.), Recent RAWLES, M. E. studies in avian biology. Urbana, Uni- 1960. The integumentary system. In Marshall, versity of Illinois Press, pp. 1-22. A. J. (ed.), Biology and comparative 1955b. The avifauna of the Sierra del Carmen physiology of birds. New York and of Coahuila, Mexico. Condor, vol. 57, London, Academic Press, vol. 1, pp. pp. 154-178. 189-240. MILLER, W. D. RENScH, B. 1905. List of birds collected in southern Sin- 1950. Die Abhbngigkeit der relativen Sexual- aloa, Mexico, by F. H. Batty, during differenz von der Kdrpergr6sse. Bonn 1903-1904. Bull. Amer. Mus. Nat. Hist. Zool. Beitr., vol. 1, pp. 58-69. vol. 21, art. 22, pp. 339-369. RIDGWAY, R. 1915. Three new genera of birds. Ibid., vol. 1881. A review of the genus Centurus, Swain- 34, art. 17, pp. 515-520. son. Proc. U. S. Natl. Mus., vol. 4, pp. MILNE, A. 93-119. 1961. Definition of competition among ani- 1885a. Description of some new species of birds mals. In Mechanisms in biological com- from Cozumel Island, Yucatan. Proc. petition. Symp. Soc. Exp. Biol., no. 15, Biol. Soc. Washington, vol. 3, pp. 1-4 pp. 40-61. (reprint). MOORE, J. A. 1885b. Catalogue of a collection of birds made 1957. An embryologist's view of the species on the Island of Cozumel, Yucatan, by concept. In Mayr, E. (ed.), The species the naturalists of the U. S. Fish Com- problem. Publ. Amer. Assoc. Adv. Sci., mission steamer Albatross, Capt. Z. L. no. 50, pp. 325-338. Tanner, Commander. Proc. U. S. Natl. MOORE, R. T. Mus., vol. 8, pp. 560-583. 1950. A new race of Melanerpes chrysogenys 1887. A manual of North American birds. 272 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 124 Philadelphia, J. B. Lippincott Co., 631 tral American birds. Auk, vol. 62, pp. PP. 8-37. 1888. Catalogue of a collection of birds made 1948. Life history of the golden-naped wood- by Mr. Charles H. Townsend on islands pecker. Ibid., vol. 65, pp. 225-260. in the Caribbean Sea and in Honduras. 1961a. The nest as a dormitory. Ibis, vol. 103a, Proc. U. S. Natl. Mus., vol. 10, pp. 572- pp. 50-70. 597. 1961b. Helpers among birds. Condor, vol. 63, 1914. The birds of North and Middle America. pp. 198-226. Pt. 6. Bull. U. S. Natl. Mus., vol. 50, SLUD, P. 882 pp. 1960. The birds of Finca "La Selva," Costa RIPLEY, S. D. Rica: a tropical forest locality. Bull. 1961. Aggressive neglect as a factor in inter- Amer. Mus. Nat. Hist., vol. 121, art. 2, specific competition in birds. Auk, vol. pp. 53-148. 78, pp. 366-371. SMITH, A. P. SALVIN, 0. 1908. Some data and records from the Whet- 1889. A list of the birds of the islands of the stone Mountains, Arizona. Condor, vol. coast of Yucatan and of the Bay of 10, pp. 75-78. Honduras. Ibis, ser. 6, vol. 1, pp. 359- STONE, W. 379. 1932. The birds of Honduras with special ref- SALVIN, O., AND F. D. GODMAN erence to a collection made in 1930 by 1888. Biologia Centrali-Americana: Aves. Lon- John T. Emlen, Jr., and C. Brooke don, vol. 2, 598 pp. Worth. Proc. Acad. Nat. Sci. Phil- SELANDER, R. K. adelphia, vol. 84, 291-342. 1958. Age determination and molt in the boat- pp. tailed grackle. Condor, vol. 60, pp. 355- STORER, R. W. 376. 1952. Variation in the resident sharp-shinned 1959. Polymorphism in Mexican brown jays. hawks of Mexico. Condor, vol. 54, pp. Auk, vol. 76, pp. 385-417. 283-289. [MS.] Speciation in wrens of the genus STRESEMANN, E. Campylorhynchus. Unpublished doctoral 1939. Die Vogel von Celebes. Teil. I. Jour. fur dissertation, 1956, University of Cali- Ornith., vol. 88, pp. 1-135, 389-487. fornia, Berkeley. SUTTON, G. M. SELANDER, R. K., AND D. R. GILLER 1940. Birds of Tamazunchale, San Luis 1959a. Avifauna of the Barranca de Oblatos, Potosi. Wilson Bull., vol. 52, pp. 221- Jalisco, Mexico. Condor, vol. 61, pp. 223. 210-222. 1951. Mexican birds: first impressions. Nor- 1959b. Interspecific relations of woodpeckers man, University of Oklahoma Press, in Texas. Wilson Bull., vol. 71, pp. 107- 282 pp. 124. SUTTON, G. M., AND 0. S. PETTINGILL, JR. SIBLEY, C. G. 1942. Birds of the Gomez Farias region, south- 1961. Hybridization and isolating mechan- western Tamaulipas. Auk, vol. 59, pp. isms. In Blair, W. F. (ed.), Vertebrate 1-34. speciation. Austin, University of Texas 1943. Birds of Linares and Galeana, Nuevo Press, pp. 69-88. Leon, Mexico. Misc. Publ. Louisiana SIMMONS, K. E. L. State Univ. Mus. Zool., no. 16, pp. 273- 1951. Interspecific territorialism. Ibis, vol. 93, 290. pp. 407-413. TASHIAN, R. E. SIMPSON, G. G. 1953. The birds of southeastern Guatemala. 1961. Principles of animal taxonomy. New Condor, vol. 55, pp. 198-210. York, Columbia University Press, 247 TEST, F. H. PP. 1942. The nature of the red, yellow, and SKUTCH, A. F. orange pigments in woodpeckers of the 1935. Helpers at the nest. Auk, vol. 60, pp. genus Colaptes. Univ. California Publ. 257-273. Zool., vol. 46, pp. 371-389. 1943. The family life of Central American TODD, W. E. C. woodpeckers. Sci. Monthly, vol. 56, pp. 1946. Critical notes on the woodpeckers. Ann. 358-364. Carnegie Mus., vol. 30, art. 17, pp. 1945. Incubation and nestling periods of Cen- 297-317. 1963 SELANDER AND GILLER: CENTURUS 273 UDVARDY, M. D. F. WEBSTER, J. D. 1951. The significance of interspecific com- 1958. Further ornithological notes from Zaca- petition in bird ecology. Oikos, vol. 3, tecas, Mexico. Wilson Bull., vol. 70, pp. 98-123. pp. 243-256. 1959. Notes on the ecological concepts of WEBSTER, J. D., AND R. T. ORR habitat, biotope and niche. Ecology, 1954. Summering birds of Zacatecas, Mexico, vol. 40, pp. 725-728. with a description of a new race of VAN ROSSEM, A. J. Worthen sparrow. Condor, vol. 56, pp. 1932. The Gila woodpecker in the Imperial 155-160. Valley of California. Condor, vol. 35, WETMORE, A. p. 74. 1941. Notes on the birds of the Guatemalan 1934. Critical notes on Middle American highlands. Proc. U. S. Natl. Mus., vol. birds. Bull. Mus. Comp. Zool., vol. 77, 89, pp. 523-581. pp. 387-490. 1943. The birds of southern Veracruz, Mexico. 1945. A distributional survey of the birds of Ibid., vol. 93, pp. 215-340. Sonora, Mexico. Occas. Paper Louisiana 1948. The golden-fronted woodpeckers of State Univ. Mus. Zool., no. 21, 379 pp. Texas and northern Mexico. Wilson Bull., vol. 60, pp. 185-186. VAN TYNE, J. 1957. The birds of Isla Coiba, Panamfa. 1938. The birds of northern Peten, Guate- Smithsonian Misc. Coll., vol. 134, no. 9, mala. Misc. Publ. Univ. Michigan Mus. pp. 1-105. Zool., no. 27, pp. 1-47. WETMORE, A., AND F. C. LINCOLN VAURIE, C. 1933. Additional notes on the birds of Haiti 1951. Adaptive differences between two sym- and the Dominican Republic. Proc. patric species of nuthatches. Proc. Xth U. S. Natl. Mus., vol. 82, pp. 1-68. Internatl. Ornith. Congr., 1950, pp. WETMORE, A., AND B. H. SWALES 163-166. 1931. The birds of Haiti and the Dominican WAGLER, J. Republic. Bull. U. S. Natl. Mus. vol. 1829. Beitrage und Bemerkungen zu dem 155, pp. 1-483. ersten Bande seines Systema Avium. ZIMMERMAN, D. A., AND G. B. HARRY Isis von Oken, vol. 22, no. 5, columns 1951. Summer birds of Autlan, Jalisco. Wilson 505-519. Bull., vol. 63, pp. 302-314.