The Condor92970-981 Q The Cooperomithological society 1990

PATTERNS OF ELEVATIONAL AND LATITUDINAL DISTRIBUTION, INCLUDING A “NICHE SWITCH,” IN SOME GUANS () OF THE

J. V. REMSEN, JR. AND STEVEN W. CARDIFF Museumof Natural Science,Louisiana State University,Baton Rouge, LA 70803

Abstract. A recentlydiscovered population of Chamaepetesgoudotii (Sickle-winged ) in the easternAndes of southernPeru and northern is anomalousin its high-elevation (above 3,000 m) distribution; populationsfrom central to not-themColombia are found generallybelow 2,100 m. Several hypothesesare evaluated to accountfor the origin of such an unusualdistribution. No evidence for long-distancedispersal or human introduction was found (and previous claims of introduction by ancient civilizations to account for disjunct distributions of two other speciesof New World are disputed). We propose that the current pattern is the result of fragmentation of a once continuousdistribution. Hypotheses concerning the cause of the fragmentation and maintenance of the current pattern are evaluated. The relative consistencyof elevational distributions of Andean speciesover broad latitudinal ranges (and dramatic differences in speciescomposition with changesin elevation) provides circumstantial evidence against an autecologicalhypothesis that would proposethat ecologicalrequirements are met for C. goudotiionly in the two disjunct regions. If potential competitors are restrictedto congeners,no supportcan be found for a hypothesis basedon interspecificcompetition. However, if the field of potential competitorsis expanded to include morphologically similar confamilials, a striking pattern of generally nonoverlap- ping distributions in terms of elevation and latitude is revealed. Most striking is the reversal in elevationsoccupied by C. goudotiiand Penelopemontagnii. The reliance on interpretation for evaluating competing hypothesesmakes such analysesinherently unsatisfying. Key words: Guans;Andes; ; Chamaepetes;patterns of distribution;interspecijc competition.

INTRODUCTION inhabitants and are among the first birds ac- quired for collectors by local hunters or men- Three specimens of the Sickle-winged Guan tioned when inquiring about the local avifauna. (Chamaepetesgoudotii) collected by Cardiff in Thus, the discovery of this population resembles humid Temperate Zone forest of Dpto. La Paz, the rediscovery of the White-winged Guan (Pe- Bolivia, in June 1980 represented an 800&m nelopealbipennis) within sight of the Pan-Amer- southerlyrange extension and establishedthe first ican Highway in Peru (Williams 1980). Both cases records for Bolivia (Cardiff and Remsen 1981). illustrate that the distributional limits of many Such a range extension was not entirely surpris- South American birds are still not well estab- ing becauserecent fieldwork in the forested An- lished. desof northwestern Bolivia has produced 12 new Although this major extension in latitudinal speciesfor Bolivia previously unrecorded south distribution for C. goudotiiwas of zoogeographic of central or southern Peru (Parker et al. 1980, interest, the elevationat which the Bolivian pop- Cardiff and Remsen 198 1, Schulenberg and ulation was discovered was more significant. Remsen 1982, Remsen and Traylor 1983) de- Throughout its range in the Eastern Andes from spite visits by several earlier workers (including to central Peru, C. goudotiioccurs at M. A. Carriker, who spent 12 days at Hichuloma, low elevations, usually below 2,100 m. Yet the at the same elevation as, and only 2 km from, newly discovered southern populations, includ- the guan locality). It was a surprise, however, ing additional specimens collected subsequently that a cracid would pass undetected for so long, in Dpto. La Paz (Remsen 198 5) and Dpto. Puno, only 2 hr on the main road from the city of La southern Peru (specimensat Museum of Natural Paz. Cracids are almost always familiar to native Science, Louisiana State University, LSUMNS) were found 1,000 m higher, at 3,000-3,300 m. ’ ’ Received23 February 1990. Final acceptance15 Such a dramatic latitudinal switch in eleva- July 1990. tional range is unusual in Andean forest birds,

I9701 DISTRIBUTION PATTERNS IN GUANS 971 most of which occur within relatively consistent workers, and some expeditions may not record elevational limits throughout most of their lati- the speciesat all but then find it on return visits. tudinal range (Chapman 1926, Graves 1985, Nevertheless, we are reasonably certain that LSUMNS specimens, pers. observ.). In this pa- C. goudotii does not occur at elevations below per, we first describe the pattern of elevational approximately 2,800 m south of 11%. Since 1979, distribution of C. goudotii and other guans (Crac- approximately 190 person-days have been spent idae) of the Andes, and then attempt to deter- by personnel of the Museum of Natural Science, mine why this speciesshows such an odd pattern Louisiana StateUniversity, in humid forestabove of distribution. 3,000 m in the area occupied by the newly dis- covered, high-elevation population of C. gou- METHODS dotii (see Cardiff and Remsen 198 1 and Remsen Specimen localities for montane guans were ob- 1985 for localities in La Paz; the locality in Dpto. tained from museums with major collections of Puno, Peru, is “Valcon, 5 km NNW of Quiaca, Andean birds (see Acknowledgments). We feel 3000 m,” 6-l 8 October 1980). This effort pro- that we located most guan specimens with lo- duced seven specimensand several sight records. cality data sufficiently precise to allow plotting By comparison, approximately 220 person-days of latitude and elevation. Latitudes were taken have been invested at the same latitudes but be- primarily from the ornithological gazetteers of low 2,600 m in La Paz and Puno without re- the Andean countries (Paynter et al. 1975; Payn- cording C. goudotii (2,575 m, Sacramento Alto, ter and Traylor 1977, 1981; Paynter 1982; Ste- 8 km by road north Chuspipata, Dpto. La Paz, phens and Traylor 1983). If the specimen label Bolivia, 20 July-8 August 1979; 2,000 m, Abra did not include elevation, the gazetteers were de Maruncunca, 10 km southwestof San Josede1 used to determine if the elevation could be as- Oro, Dpto. Puno, Peru, 9 November-6 Decem- certained with reasonable precision. A substan- ber 1980; 1,650 and 1,350 m, Serrania Bellavis- tial number of specimens could not be used be- ta, 35-36 and 47 km, respectively, by road north cause their elevations could not be determined. of Caranavi, Dpto. La Paz, Bolivia, 10 June-2 For specimensfor which the elevation was given July 1979 and 10 July-2 August 1980, respec- as a range, e.g., “2000-2400 m,” the midpoint tively). Therefore, we believe that the unusually of the range was used as the elevation, unless the high and compressed elevational range of the specimen was from near an upper or lower limit newly discovered southern population is not a of the ’ overall elevational distribution, in sampling artifact. which case the specimen was not used. We are even more certain that C. goudotii does not occur above 2,300 m elevation north of 11%. RESULTS AND DISCUSSION LSUMNS expeditions in central and northern A plot of specimen localities of C. goudotii on Peru have loggedover 500 person-daysin humid the eastern slope of the Eastern Andes (Fig. 1) forest above 2,300 m without finding any C. gou- reveals that the species’ elevational distribution dotii. switches dramatically from below 2,100 m to What processeswould produce this elevational above 3,000 m at about 11% We are certain that switch? What factors maintain the present pat- the high-elevation population does not represent tern of limits of distribution? These zoogeo- wandering or migratory individuals: guans are graphical questions are difficult to answer be- sedentary birds, with no known cases of long- cause (a) the historical processescan never be distance movement or vagrancy (Delacour and ascertained with complete certainty, leaving in- Amadon 1973). ference as the primary methodology, and (b) ex- Whether the observed pattern is a sampling periments to test hypotheses concerning large- artifact, however, deserves consideration. Al- scaledistribution patterns are usually logistically though C. goudotiican be locally common (John- intractable, especially for flying birds (Diamond son and Hilty 1976), it usually occurs in low 1986a). The remainder of this paper represents densities, and this probably plays a more im- our attempt to answer thesequestions for C. gou- portant role than shynessin determining whether dotii. or not it is recorded at a site. Some members of We consideredthree hypothesesconcerning the a field party may never seethe bird over a period processthat might account for this unusual ele- of weeks, even though seen several times by co- vational “niche switch” pattern of distribution: 972 J. V. REMSEN, JR. AND STEVEN W. CARDIFF

(1) long-distance dispersal, (2) human introduc- of C. goudotii was deliberately or accidentally tion, and (3) vicariance. We then consideredtwo introduced by primitive peoples. Becauseguans hypothesesthat might explain how such an un- are potential sources of food for humans, this usual pattern might be maintained: (1) autecol- hypothesis deservesspecial consideration. ogy and (2) interspecific competition. A thorough evaluation of the human intro- duction hypothesis would require extensive, original research on the anthropology and ar- HISTORICAL HYPOTHESES chaeology of the region. Such an analysis is be- LONG-DISTANCE DISPERSAL yond our expertise. In our opinion, however, in- One hypothesis that might account for the cur- troduction by humans seems unlikely for two rent disjunct pattern of distribution is that it was reasons. First, the human population density of the result of long-distance dispersal. In other the humid, forested Eastern Andes is currently, words, perhapssome individuals dispersinglong and probably always has been, relatively low. distances from one area successfullycolonized The steepnessof the slopes, the high degree of the other without colonizing intermediate areas. cloud cover, and, at upper elevations, the cold We find no support for long-distance dispersal temperatures all make this region unfavorable for two reasons.First, the gapsin the distribution for agriculture. So, the region in general does not of C. goudotii do not correspond to gaps in cur- seem to have been a likely area for intensive rent habitat availability. The areas from which human activity. Second, efforts at domesticating the speciesis absent are all filled with seemingly cracids have remained largely untried owing to appropriate habitat (humid montane cloud-for- their low clutch size and aggressivebehavior to- est) that is occupied by bird assemblagestypical ward conspecifics, thereby requiring relatively of such habitat. Only a few, narrow, arid canyons large enclosures(J. P. O’Neill, pers.comm.). That (e.g., the Marafion and Urubamba valleys) in- cracidsare not ideal for domestication, however, terrupt an otherwise continuous blanket of mon- does not preclude their transport as cagebirds tane forest from to central Bolivia. from place to place by primitive societies. Therefore, a dispersalhypothesis requires special We also are not convinced that Haemig’s (1978, pleading to explain why theseintermediate areas 1979) casesof proposedhuman introduction are have not been colonized by gradual expansion valid. The disjunct distribution patterns of C. from either or both directions. Second,guans are dickeyi and Q. mexicanusare not nearly the zoo- perhaps the least likely long-distance dispersers geographic anomalies that Haemig believed. among the generally sedentary Andean forest Cyanocorax dickeyi belongs to a superspeciesof birds. All cracids fly poorly and are probably blue/black-and-white jays, the members of which incapable of lengthy flight; C. goudotii is typical are distributed in a disjunct, relictual pattern in ofguans in this respect(pers. observ.). We cannot dry habitats in South America (Goodwin 1976); rule out the possibility that a group of guans C. dickeyi is simply the most isolated of these dispersedby walking or by a seriesof short flights taxa. At least four other taxa (Killdeer (Charudri- acrosshundreds of kilometers of habitable forest us vociferus),White-winged Dove (Zenaida asia- to colonize forest nearly 1,000 m above or below tica), Cave Swallow (Petrochelidonj&a), and their normal elevational distribution without the Dives superspeciesof blackbirds show dis- colonizing intervening elevations or latitudes. tribution patterns similar in some ways to that This prospect, however, must be viewed as ex- of C. dickeyi and its apparent closestrelative, C. tremely unlikely for birds for which there are no mystacalis. In all four cases,a taxon in the dry instances of extralimital occurrences. lowlands of the Pacific coast of South America has a sister taxon in Middle America. Other ex- INTRODUCTION BY HUMANS amples of taxa found in Middle America with Haemig (1978, 1979) proposedthat introduction relatives isolated in arid northwestern Peru- by primitive peoplesaccounts for the anomalous, southwestern are: (1) Aimophila spp. disjunct distributions of two speciesof Mexican and A. stolzmanni, (2) Tachycineta albilinea al- birds, the Great-tailed Grackle (Quiscalus mex- bilinea and T. a. stolzmanni, and (3) Forpus cy- icanus)and the Tufted Jay (Cyanocorax dickeyi). anopygiusand F. coelestis.The comparability of The analogous hypothesis here would propose these distribution patterns depends on knowing that one of the elevationally disjunct populations the phylogenies of the taxa involved, including DISTRIBUTION PATTERNS IN GUANS 973 whether the two jays are sister taxa; these rela- American indigenous peoples as well as with tionships are currently unknown. In our opinion, “educated” peoples is sprinkled with cases of the only anomaly in the Cyanocorax case is not amusing, but obviously erroneous, explanations the disjunction itself but that the Middle Amer- of past and present natural phenomena, in spite ican representative shows such a restricted dis- of detailed and accurate knowledge of natural tribution. history in other cases. As for the isolated, high elevation population of Q. mexicanus, similar disjunct, montane val- RELICTUAL DISTRIBUTION ley populations of primarily lowland speciesare The relictual distribution hypothesis would ex- found in several places in the Western Hemi- plain the disjunct distribution of C. goudotii by sphere,particularly in the savanna region of Bo- proposing that the once continuous distribution gota, Colombia, and the intermontane valleys of became fragmented by extirpation in portions of central Bolivia. The Mexican plateau region in- the range. A direct test of this hypothesis would habited by the grackle is also inhabited by dis- require documentation of former presenceor ab- junct, highland populations of normally lowland sence of C. goudotii in the current elevational species(e.g., Little Blue Heron Egretta caerulea, and latitudinal gaps in its range. In the absence American Bittern Botaurus lentiginosus, Red- of an extensive fossil record from the eastern head Aythya americana, Ruddy Duck Oxyura slopesof the Andes, evaluation of this hypothesis jamaicensis, Yellow Rail Coturnicopsnovebora- is not possible. We can find no evidence of his- censis,Virginia Rail Rallus limicola, Clapper Rail torical change in the distribution of C. goudotii Rallus longirostris,Northern Jacana Jacana spi- within the brief period of scientific sampling of nosa, Sedge Wren Cistothorusplatensis, Marsh Andean birds, roughly the last 100 years, al- Wren C. palustris, Song Sparrow Melospiza me- though the failure of previous workers to find the lodia, and Red-winged Blackbird Agelaius southern population might be interpreted this {phoeniceus}gubernator [Moore 1945; Dicker- way. man 1966; Ripley 1977; Banks and Dickerman Nevertheless, we favor the relictual distribu- 1978; AOU 1983; Williams 1989; S. 0. Wil- tion hypothesis by default. Although we cannot liams, in litt.]) for which a hypothesis of human provide direct evidence for once-continuous dis- introduction is untenable. Like the grackle, al- tribution followed by subsequent extinction in most all these speciesare associatedwith aquatic some areas, we believe that this best accounts or marshy habitats, as was the extinct Quiscalus for the current disjunct pattern becausewe have palustris. a plausible mechanism for the extinction of pop- Furthermore, blackbirds (Icteridae) are noto- ulations in the gaps (see below). riously adept at colonizing areas modified by hu- man activity (Orians 1985). In particular, the ECOLOGICAL HYPOTHESES Great-tailed Grackle has expanded its range dra- matically within the last 50 years (Pratt et al. AUTECOLOGY 1977, Garrett and Dunn 198 1, Holmes et al. The two disjunct regionsinhabited by C. goudotii 1985). Therefore, we feel that both of Haemig’s might be the only two in which its ecological examples are readily explained within the con- requirements are met. This hypothesis would text of broader distribution patterns, and that propose, therefore, that this speciescan survive there is no need to invoke human intervention. only at 3,000-3,300 m in southern Peru and Bo- Concerning Haemig’s data in support of hu- livia, and at 500-2,200 m from central Peru to man introduction, we note that it is entirely cir- Colombia. cumstantial and speculative in the case of the Although our fragmentary knowledge of the jay; in fact, Phillips (1986) dismissed Haemig’s biology of C. goudotiidoesnot permit strongtests hypothesis outright. In the case of the grackle, of such an hypothesis,available data do not sup- the direct evidence comes from a friar’s 1577 port it. We find it difficult to believe that the collection of stories told to him by Aztec cold, stunted forest near timberline in the south- informants (and not the friar’s direct observa- em Andes is in some critical aspectmore similar tions). Although the plausibility and detail of the to the warm, tall, lower montane forest in central latter are intriguing, we regard such folklore with Peru than either area is to immediately adjacent distrust. Our personal experience with South areas with similar habitat and climate. No other 974 J. V. REMSEN, JR. AND STEVEN W. CARDIFF speciesof Andean bird shows a pattern similar overall environmental dissimilarity between the to that of C. goudotii.Indirect evidence for eco- two regions makes such a possibility unlikely. logical dissimilarity between the high- and low- If the elevational distribution of C. goudotii elevation regions occupied by the disjunct C. showed some gradual latitudinal change in the goudotii populations comes from analysis of direction of increasing elevation with increasing sharedspecies. Of the resident bird speciesat the distance from the equator, such a pattern could C. goudotiilocality at 3,050 m in La Pax, Bolivia be used as circumstantial evidence for the aut- (Remsen 198 5), 70 species occur 1,100 km to ecology hypothesis, which would predict that the north in Dpto. Huanuco, where C. goudotii changing elevational distribution is a response is a low-elevation species. Of these 70 shared to some clinally varying climatic or ecological species,only 24 (34%) also occur at the upper- parameter. The locality data, however, do not most locality for low-elevation populations of C. show such a pattern. South of the equator, ele- goudotii,at 2,100 m in Dpto. Hu6nuco (LSUMNS vational limits appear to be more or lessconstant specimens). All 24 of the speciesfound at 2,100 (Fig. l), and the switch to high elevations takes m are basically high-elevation speciesnear the place rather abruptly. Unfortunately, the poor lower limit of their elevational range; all also resolution of the locality data prevents deter- occur at 3,050 m in Dpto. Huanuco. Only one mination of whether the elevational shift occurs, of the 70 sharedspecies occurs as low as the lower as we suspect,on opposite sides of a major An- elevational limit of C. goudotii,at about 1,000 dean river canyon. Given such a shift, the aut- m. To illustrate further the latitudinal stability ecology hypothesis would require major clima- of community composition at a given elevation tological or ecological differences on opposite in the Andes, our comparison of the resident canyon sides, an unlikely possibility that none- avifauna at 3,050 m in Dpto. La Paz with the thelesscannot be excluded at present. speciescomposition at the same elevation at a Ifthe latitudinal boundaries between guan taxa locality (Cordillera Golan; LSUMNS specimen are formed by dry canyons that might be barriers data) much further north than Huarmco (1,700 to dispersal, then these barriers could maintain km northwest of La Paz in extreme northern the current, parapatric pattern of distribution Dpto. Amazonas, Peru) revealed that 54 species without invoking other factors. However, there and seven additional allospeciesreplacements are are no barriers to dispersal across elevational shared.This reflectsa general pattern among An- boundaries-these boundaries must be main- dean forest birds (and montane forest avifaunas tained by some ongoing, ecological process. in general): similar elevations support a similar The elevational niche switch posesthe greatest species composition over a broad latitudinal difficulty for proposing that autecology is this range, but localities a few hundred meters apart ongoing process.It seemsunlikely that a species in elevation tend to support different avifaunas that occurs at lower elevations throughout most (Terborgh and Weske 1975, Parker et al. 1985). of its distribution would suddenly find theselow- Becauseno other speciesof Andean forest bird er elevations unsuitable in one section of its dis- shows a similar niche switch, and because very tribution while finding upper elevations suitable. few speciesare sharedby the two disparateregions That the reverse would occur in another guan, occupied by C. goudotii,we are unable to con- Penelopemontagnii, in a geographically coinci- struct a plausible reason why the two C. goudotii dent manner compoundsthe unlikelihood of such populations are found only where they are. Of a scheme (see next section). course, until the biology of the speciesis better known, we will never be able to eliminate the INTERSPECIFIC COMPETITION possibility that some critical food plant, nest site, Perhaps C. goudotiiis restricted to two disjunct etc., is present only in the two disjunct regions, areasbecause it is prevented from occupying the or that some species-specific,unconquerable par- intervening geographic areas and elevations by asite or predator is found everywhere but the two competition with other species. Terborgh and regions. As noted by Connor and Bowers (1987), Weske (1975) used a natural experiment, a a mosaic distribution suchas that shown by these mountain range isolated from the main Andes, guans could also be produced or maintained by to examine factors responsible for limiting the a comparable mosaic distribution of ecological elevational distributions of Andean birds. They requirements. Nevertheless,we maintain that the concluded that as many as two-thirds of the spe- DISTRIBUTION PATTERNS IN GUANS 915

4000 EASTERN ANDES

0

3000

- 20 - 15 -10 -5 0 5 10

LATITUDE

FIGURE 1. Specimenlocalities of five speciesof montane guansin the EasternAndes from northern Venezuela and Colombia to northern . Each point representsa single specimen; when these are from precisely the same locality, the points are offset slightly but contiguous(also applies to Figs. 24). Latitudes range from central Bolivia to northern Venezuela; latitudes south of the equator are indicated by negative numbers. ties’ rangeswere influenced by interspecific com- 16”3O’S, apparently from the Rio Apurimac to petition and that many of the presumed com- the Rio La Paz, their elevational relationship is petitors were not congeners. reversed, with the recently discovered popula- If one were to examine this hypothesis in the tion of C. goudotiinot known from below 3,000 manner frequently used by community ecolo- m. This switch in elevational distribution ap- gists,namely to examine the distribution of con- parently correspondswith subspeciesboundaries geners, the interspecific competition hypothesis in P. montagnii,but does not in C. goudotii.The is not supported for C. goudotii.The only other southernmostsubspecies of C. goudotii,C. g. rufi- speciesin the genus Chamaepetesis the Black ventris,ranges from Dpto. Huanuco to Dpto. La Guan (C. &color), found in the mountains of Paz, Bolivia, thus bridging the elevational switch Costa Rica and Panama (Delacour and Amadon (Cardiff and Remsen, unpubl. data). 1973). If, however, one examines other species Chamaepetesgoudotii has not been recorded of guans in the Andes, a pattern emergesthat is south of 16”3O’S, in spite of intensive collecting consistent with the interspecific competition hy- in Dpto. Cochabamba, where P. montagniioc- pothesis. cursas high as 3,200 m. Farther south, the Red- A plot of elevation and latitude for guan spec- faced Guan (P. dabbenei)replaces P. montagnii imens from the Eastern Andes (Fig. 1) showsthe (contra Delacour and Amadon 1973) from Dpto. following features. First, although there is some Tarija, Bolivia, south to Prov. Salta, Argentina; overlap among various species,the pattern that in fact, dabbeneiis treated as a subspeciesof is most evident is that the two most widespread montagniiby some authors (Olrog 1960, Vuil- species,the Andean Guan (Penelopemontagnii) leumier 1965; cf. Vaurie 1967). The records of and C. goudotii,replace each other at approxi- P. montagniifrom Salta (Olrog 1960; Vaurie mately 2,100 m from central Colombia to south- 1967, 1968) pertain to misidentified specimens em Peru. However, from about 11% to about of P. dabbenei(M. Nores, in litt.). Elsewhere in 976 J. V. REMSEN, JR. AND STEVEN W. CARDIPP

3000 the Eastern Andes, the Bearded Guan (P. bar- CENTRAL ANDES (E. SLOPE, bata) is known from two localities, in extreme northern Peru in an area where P. montagnii does not occur (Parker et al. 1985). At the extreme north end of the Andes in northern Colombia and Venezuela, the Band-tailed Guan (P. argy- rotis) occurs at elevations inhabited by C. gou- dotii farther south and below elevations typically inhabited by P, montagnii.

1 2 3 On the eastern slope of the Central Andes of N. LATITUDE Colombia (Fig. 2), C. goudotii has been collected from about 1,600 to about 2,600 m. Penelope

4040-I montagnii has been collectedonly at 2,500-2,600 : CENTRAL ANDES (w. SLOPE) m at three northern localities. With so few lo- R-i- calities, little can be said except that the two specieshave not been collected at the same lo- cality and that C. goudotii is more widespread and may occupy elevations generally below those occupied by P. montagnii. On the western slope of the Central Andes (Fig. 2), the same tentative conclusions apply except that in the south, the few specimen records available suggestthat C. 1 2 3 4 5 6 7 0 goudotii occursabove P. montagnii. Because vir- N. LATITUDE tually all locality data from the Central Andes FIGURE 2. Specimenrecords of two speciesof guans come from specimens taken 50-60 years ago, on the eastern slope and western slope of the Central before the advent of accurate mapping of ele- Andes of Colombia. Boxes enclose localities for Pe- vational distributions, we believe that all locality nelopemontagnii. data from this region should be interpreted with

4000

WESTERN ANDES 0

2000

/

0.’ ...... ’ “. .‘ *. .‘ 1 -5 0 - 10 5 IO LATITUDE FIGURE 3. Specimen records of three speciesof guans in the Western Andes of Colombia, Ecuador, and northwestern Peru. Latitudes south of the equator are indicated by negative numbers. DISTRIBUTION PATTERNS IN GUANS 911

3000- SANTA MARTA MTNS. x xX xX x :: 0

10.0 11.0

N. LATITUDE FIGURE 4. Specimenrecords of two speciesof guansin the SantaMarta Mountainsof northeasternColombia.

caution. Furthermore, throughout the Andes calities, but whether or not they are truly syn- many professional collectors apparently ranged topic is unknown. In the Coastal Range of Ven- hundreds of meters above or below a given col- ezuela (not mapped), where C. goudotii does not lecting locality, yet all specimens bore the same occur, P. argyrotis occupies all elevations from locality name. 300 m to the summits at ca. 2,000 m (Schafer In the Western Andes (Fig, 3), C. goudotii gen- and Phelps 1954). erally occurs below elevations occupied by P. Therefore, the general pattern throughout the montagnii in Ecuador and Colombia, with some Andes is that only one species of Penelope or elevational overlap; however, at only one locality ChamaepetesOCCLUS at most, but not all, local- have both speciesbeen collected. Neither species ities, and that C. goudotii occurs below one of occurs south of about 4”3O’S, where P. barbata the three speciesof Penelope,except in extreme occurs exclusively and occupies both the high southern Peru and northern Bolivia, where the elevations typical of P. montagnii and the low situation is reversed.Although sucha “repulsed” elevations typical of C. goudotii. pattern of species’distributions is consistentwith In the Santa Marta Mountains, an isolated, the predictions of the interspecific competition major mountain range in northern Colombia, hypothesis, it is also consistent with other hy- two species of guans occur, C. goudotii and P. pothesesthat propose a parapatric distribution argyrotis (Fig. 4). Although Todd and Carriker of autecological factors, such as habitat suitabil- (1922) reported that the two speciesoverlapped ity, predation, and diseases(Connor and Bowers broadly in elevational distribution, a plot of 1987). specimen localities (Fig. 4) shows that on the With only a cursory knowledge of the ecologies southern side of the Santa Martas, only P. ar- of guans of the Andes, it is difficult to know gyrotis occurs,and C. goudotii is found only on whether these speciesare likely to be competi- the northern side. There, however, it does over- tors. Examination of available data, however, lap broadly in elevational distribution with P. favors consideration of all guan speciesas po- argyrotis but generally occurs above it. The two tential competitors. All are similar in overall size specieshave been collected together at five lo- and proportions, especially in bill length (Table 978 J. V. REMSEN, JR. AND STEVEN W. CARDIFF

Ortalis motmot P. a. agyrotis P. a. colombiana Penelope barbata C. g. sanctaemariae C. g. tschudii C. g. rufiventris C. g. goudotii P. m. montagnii

P. m. sclateri

P. m. atrogutaris

P. m. brooki P. m. plumosa C. g. tagani , Penelope dabbenei I Oreophasis derbianus ““‘.““..““““...,.,,...,,.,,,,,,,,,,,,,...,,.....,...... ~.”...... ,..,..~...... ~....“.~...... ~~..~~...,~...... ,.....,...... “.”...,...“.....“Crax.,...., mitu 3.0 2.0 1.0 0.0 morphometricdistance FIGURE 5. UPGMA phenogramof five speciesof guansof the Andes. Speciesof three other generaof Cracidae are also included for comparison. Charactersused were wing length,bill length,tarsus length, and tail lengthof adult males;measurements were taken from Vaurie (1968). Copheneticcorrelation = 0.981.

1); C. goudotii has a disproportionately shorter Penelope (P. argyrotis colombiana in this case). tarsus, but is otherwise similar to Penelope spp. On the contrary, Johnson and Hilty (1976) found in size and shape.In fact, a phenogram based on that in the Santa Marta Mountains C. goudotii wing, tail, tarsus, and bill measurements (Fig. 5) forages only in arboreal situations, although it showsthat all subspeciesof C. goudotii are phe- regularly descendsto the ground to cross open notypically more similar to some forms of Pe- spaces. nelope than are some Penelope speciesto each Stomach contents are available only for C. other. Also, we can find no conclusive evidence goudotii and P. montagnii. Examination of these to support the inferences of Todd and Carriker showsthat both species’diets are roughly similar, (1922) that C. goudotii is more terrestrial than with fruits in the l-10 mm size range and other vegetable matter predominating in both (Fig. 6). Foraging observations for both thesespecies and TABLE 1. Measurements of some guans of the An- P. barbata (T. A. Parker, pers. observ.) indicate des. All measurementsare lengths rounded to nearest that they feed on small fruits taken primarily in 1 mm for adult malesand are taken from Vaurie (1968). trees.Thus, our ecologicaldata, admittedly weak, The largest and smallest subspeciesof Penelopemon- are consistent with potential interspecific com- tagnii and Chamaepetesgoudotii are included to en- compassthe range in variation in these two species. petition between C. goudotii and P. montagnii. Not included in our analysis was the Wattled TX- Guan (Aburria aburri), which overlaps broadly Species(n) Wing Tail sus Bill in latitudinal and elevational (600-2,500 m) dis- Penelopeargyrotis (2 1) 276 251 53 27 tribution with various Penelope speciesand C. Penelopebarbata (6) 261 257 51 28 goudotii. If and how it differs ecologically from Penelopemontagnii other montane guans is not known. plumosa (16) 240 233 54 21 Although information on the distribution, P. m. sclateri(19) 263 240 57 26 Penelopedabbenei (7) 303 301 68 27 ecology, and morphology of guans of the Andes Chamaepetesgoudotii is most consistent with the interspecific com- fagani (16) 241 205 58 28 petition hypothesis, such an interpretation cre- C. g. mfiventris(2) 269 255 66 28 ates problems in terms of “common sense” nat- DISTRIBUTION PATTERNS IN GUANS 979

C. goudofii (N= 7) pJ P. monfagnii W = 12)

60-

nc ” seeds (l-10 mm) seeds (1 l-20 mm) seeds (21-30 mm) leaf parts other veg. matter Diet Items

FIGURE 6. Representationof diet items in cropsand stomachsof specimensof Chamaepetesgoudotii and Penelopemontagnii. The seedswere presumablyenclosed by fruit parts,which appearedto form the bulk of the “other vegetablematter.”

Ural history. It challenges our imaginations to the high-elevation populations of C. gouaotii’ in concoct a situation with a firm basis in natural northern Bolivia and southern Peru as the vestige history that proposesthat speciesA outcompetes of a time before the appearanceof P. montagnii, B at high elevations in one portion of the range, when C. goudotii was the only guan through most but B outcompetes A at these same elevations of its range; this high-elevation, relictual popu- in the other portion, with the reversesituation at lation of C. goudotii may be in the process of low elevations. Why would C. goudotii be able being eliminated by the expansion of P. mon- to outcompete P. montagnii at high elevations tagnii. Such a hypothesis can be supported only in the southern Andes but the situation be com- by gathering evidence on historical changes in pletely reversedin essentially contiguous habitat the distribution of the two putative competitors north ofabout 11‘S? If P. montagnii excludes C. or by genetic data that show which taxon is the goudotii from upper elevations throughout most oldest. of the Andes, why is it not able currently to ex- Although we favor the interspecific competi- clude it from such elevations in southern Peru tion hypothesis,especially with a post-hoc boost and Bolivia? from the nonequilibrium hypothesis articulated One solution to the dilemma posedby the guan above, we emphasize that we are uneasy with niche shifts is to challenge the unstated assump- suchan interpretation becausethe testsmustered tion that the system is in equilibrium. As artic- in its support are weak. We realize, however, that ulated by Wiens (1984), recent thinking con- such uneasinessis probably inevitable given the cerning community patterns and processesis nature of such “natural experiments” that are based too often on assumptions of equilibrium. not amenable to direct experimentation (but see Certainly, if we change our outlook in the An- Diamond [ 1986b] for the advantages of natural dean guan situation to one of disequilibrium, experiments). We also realize that our conclu- then previous objections to all three hypotheses sionsare influenced greatly by interpretation. For can be countered in part. Perhaps we can view example, the evidence for competition is weak 980 J. V. REMSEN, JR. AND STEVEN W. CARDIFF or open to interpretation: (1) the dietary data, ACKNOWLEDGMENTS which addressonly the size of food, are not suf- Thanks are extended to the late Babette M. Odom, ficiently precise to exclude the possibility that John S. McIlhenny, H. Irving Schweppe,and the late major taxonomic differencesexist in food items; Laura R. Schweppe for their financial support of (2) data on feeding-site selection are nonexistent; LSUMNS field research.We are grateful to the Direc- ci6n de Ciencia y Tecnologia and the Academia Na- and (3) the distributional data show that C. gou- cional de Ciencias, La Paz, Bolivia, for permission to dotii is sympatric with Penelopein some areas, work in Bolivia and to the Direction General Forestal particularly outside the Eastern Andes. The im- y de Fauna of the Ministerio de Agricultura, Lima, portance that one attaches to these points is in- Peru, for authorizations to work in Peru. Special rec- evitably based to a degree on one’s subjective oanition goesto Gaston Beiarano for his aid and con- c&n for our work in Bolivia-andto Tom and Jo Heindel biases. Data are also nonexistent on nest-site se- for their hosnitalitv in La Paz durina 1979 and 1980. lection, predators, parasites,or diseases.Only by Dolores M. he Qmntela and Carlos-E. Quintela pre- obtaining more refined data that permit stronger pared the initial figures, and Robert M. Zink guided tests of the competing hypotheseswill the level the UPGMA analysis.We are grateful to the following for specimen loans and providing specimen locality of analysis be elevated from interpretation to data from their museums: M. B. Robbins and F. B. science. Gill (Academy of Natural Sciences,Philadelphia), D. Regardlessof the rather unsatisfying results of Willard and J. W. Fitzpatrick (Field Museum of Nat- our attempt to test various hypothesesconcem- ural History), J. M. Loughlin and K. C. Parkes (Car- ing the underlying processes,we believe that it negie Museum), G. R. Graves (U.S. National Muse- um), A. V. Andors, J. Bull, and L. L. Short (American is important to report this intriguing pattern of Museum of Natural History), the late R. W. Schreiber distribution of guans of the Andes. Although sev- and K. Garrett (Los Angeles County Museum of Nat- eral examples of elevational niche shifts have ural History), R. A. Paynter, Jr. (Museum of Com- been reported in birds of New Guinea and the parative Zoology), J. Fjeldsa (Zoologisk Museum, Co- penhagen),D. A. Bell and N. K. Johnson (Museum of southwesternPacific (Diamond 1970, 1972; Di- Vertebrate Zoology), Charles G. Sibley (Peabody Mu- amond and Marshall 1977), we have found only seum), and E. Alabarce (Instituto Miguel Lillo, Tu- two other casesof elevational niche reversalin cuman). We thank M. Nores and S. 0. Williams for birds from anywhere else in the world. In the unpublished data, and J. M. Bates, J. W. Fitzpatrick, Eastern Andes of northern Colombia and Ven- S. J. Hackett, C. A. Marantz, T. A. Parker, J. P. O’Neill, R. M. Zink, and an anonymous reviewer for critical ezuela and in the Coastal Range of Venezuela, comments on the manuscript. two speciesof brush-finches (Atlapetesbrunnei- nuchaand A. torquatus)reverse their usual rel- ative position found in the rest of the Andes, LITERATURE CITED is where brunneinucha the low-elevation species AMERICAN ORNITHOLOGISTS ’ UNION. 1983. Check- and torquatusthe high-elevation species(Schafer list of North American birds. 6th ed. American and Phelps 1954; Remsen and S. Graves, unpubl. Ornithologists’ Union, Washington, DC. data). A reviewer pointed out to us that Dia- BANKS,R. C., ANDR. W. DICKERMAN.1978. Mexican mond’s (1973) map of Melidecteshoneyeater dis- nesting records for the American Bittern. West. Birds 9: 130. tribution in New Guinea indicated that two spe- CALLAHAN,J. R. 1977, Diagnosis of Eutamiasob- cies, M. ochromelasand M. rufocrissalis,show scurus(Rodentia: Sciuridae).J. Mammal. 58: 188- an elevational niche switch. Also, Andrew Krat- 201. \ ter (pers. comm.) has pointed out to us one ex- CARDIFF, S. W., AND J. V. REMSEN, JR. 1981. Three bird soecies new to Bolivia. Bull. Br. Omithol. ample of an elevational niche switch in mam- Club io1:304-305. mals: two speciesof Eutamiaschipmunks reverse CHAPMAN, F. M. 1926. The distribution of bird-life their relative elevational distributions in the in Ecuador. Bull. Am. Mus. Nat. Hist. 55: l-784. mountains of southern California (Callahan CONNOR, E. F., ANDM. A. BOWERS.1987. The spatial consequencesof interspecific competition. Ann. 1977). Zool. Fenn. 24:213-226. Although such anomalies might be regarded DELACXXJR,J., AND D. AMAD~N. 1973. as minor becauseof their rarity, they potentially and related birds. American Museum of Natural provide important insights into factors that gov- History, New York. em distributional limits. They represent“natural DL~MOND,J. M. 1970. Ecological consequencesof island colonization by Southwest Pacific birds. I. experiments” that may reveal causal factors, or Types of niche shifts. Proc. Natl. Acad. Sci. 67: at least provide direction for subsequent re- 529-536. search. DIAMOND,J. M. 1972. Avifauna of the easternhigh- DISTRIBUTION PATTERNS IN GUANS 981

lands ofNew Guinea. Publ. Nuttall Omithol. Club PAYNTER,R. A., JR., ANDM. A. TRAYLOR,JR. 1981. 12:1-43. Ornithological gazetteerof Colombia. Museum of DIA~~OND,J. M. 1973. Distributional ecologyof New Comparative Zoology, Cambridge, MA. Guinea birds. Science 179~759-769. PAYNTER,R. A., JR., M. A. TRAYU)R, JR., AND B. D~OND, J. M. 1986a. Evolution of ecologicalseg- WINTER. 1975. Ornithological aazetteer of Bo- regation in the New Guinea montane avifauna, p. livia. Museum of Comparative -Zoology, Cam- 98-l 25. In J. Diamond and T. J. Caseleds.1. Com- bridge, MA. munity ecology. Harper and Row, New York. PHILLIPS,A. R. 1986. The known birds of North and DL~MOND,J. M. 1986b. Overview: laboratory ex- Middle America. Published by the author, Den- periments, field experiments, and natural experi- ver, CO. ments, p. 3-22. In J. Diamond and T. J. Case PRA’IT, H. D., B. ORTEGO,AND H. D. GUILLORY. 1977. [eds.],Community ecology.Harper and Row, New Spreadof the Great-tailed Grackle in southwestern York. Louisiana. Wilson Bull. 89:483485. DIAMOND,J. M., ANDA. G. MARSHALL. 1977. Niche REMSEN,J. V., JR. 1985. Community organization shifts in New Hebridean birds. Emu 77:61-72. and ecologyof birds of high elevation humid cloud DICKERMAN.R. W. 1966. A new subsneciesof the forest of the Bolivian Andes, p. 733-756. In P. A. Virginia Rail from Mexico. Condor 68:2 15-2 16. Buckley,M. S. Foster, E. S. Morton, R. S. Ridgely, GARRETT,K., AND J. DUNN. 1981. Birds of southern and F. S. Buckley [eds.], Neotropical ornithology. California. Los AngelesAudubon Society,Los An- Omithol. Monogr. No. 36. American Omitholo- geles. gists’ Union, Washington, DC. GOODWIN,D. 1976. Crowsof the world. Cornell Univ. REMSEN,J. V., JR., AND M. A. TRAYLOR,JR. 1983. Press, Ithaca, NY. Additions to the avifauna of Bolivia. Part 2. Con- GRAVES,G. R. 1985. Elevational correlates of spe- dor 85:95-98. ciation and intraspecific geographic variation in RIPLEY.S. D. 1977. Rails of the world. David R. in Andean forest birds. Auk 102:556- Gddine, Boston, MA. 579. - SCHAFER,E., ANDW. H. PHELPS.1954. Aves de Ran- HAEMIG,P. D. 1978. Aztec emperor Auitzotl and the cho Grande. Bol. Sot. Venez. Cienc. Nat. 16:3- Great-tailed Grackle. Biotrooica 10:1 l-l 7. 167. HAEMIG,P. D. 1979. Secret of ihe Painted Jay. Bio- SCHULENBERG,T. S., AND J. V. REMSEN,JR. 1982. tropica 11:81-87. Eleven bird speciesnew to Bolivia. Bull. Br. Or- HOLMES,J. A., D. S. I)oBKrN,AND B. A. WILCOX. 1985. nithol. Club 102:52-57. Second nesting record and northward advance of STEPHENS,L., AND M. A. TRAYLOR,JR. 1983. Omi- the Great-tailed Grackle (Quiscdus mexicanus)in thologicalgazetteer of Peru. Museum of Compar- Nevada. Great Basin Nat. 45:483484. ative Zoology, Cambridge, MA. JOHNSON,T. B., AND S. HILT. 1976. Notes on the TERBORGH,J., AND J. S. WESKE. 1975. The role of Sickle-winged Guan in Colombia. Auk 93:194- competition in the distribution of Andean birds. 195. Ecology 56:562-576. MOORE,R. T. 1945. The TransverseVolcanic Biotic TODD, W.E.C., ANDM. A. CAIUUKER,JR. 1922. The Province of central Mexico and its relationships birds of the Santa Marta Region of Colombia: a to adiacent nrovinces. Trans. San Dieno Sot. Nat. study in altitudinal distribution. Ann. Carnegie Hist.-10:217-236. Mus. 14:1-611. OLROG,C. C. 1960. Penelope montagnii en la Ar- VAURIE,C. 1967. Systematicnotes on the bird family gentina (Aves, ). Neotronica 6:58-59. Cracidae. No. 6. Reviews of nine soeciesof Pe- O&S, G. 1985.. Blackbirds of the Americas. Univ. nelope.Am. Mus. Novit. 2299. of Washington Press, Seattle. VAURIE. C. 1968. Taxonomv of the Cracidae. Bull. PARKER,T. A., III, J. V. REMSEN,JR., AND J. A. Am. Mus. Nat. Hist. 138:213-214. HEINDEL. 1980. Seven bird speciesnew to Bo- VUILLEUMIER,F. 1965. Relationships and evolution livia. Bull. Br. Omithol. Club 100:160-162. within the Cracidae(Aves, Galliformes). Bull. Mus. PARKER,T. A., III, T. S. SCHULENBERG,G. R. GRAVES, Comp. Zool. 134:l-27. AND M. J. BRAUN. 1985. The avifauna of the WIENS,J. A. 1984. On understandinga non-equilib- Huancabambaregion, northern Peru, p. 169-l 97. rium world: myth and reality in community pat- In P. A. Buckley, M. S. Foster, E. S. Morton, R. terns and processes,p. 439457. In D. A. Strong, S. Ridgely, and F. S. Buckley [eds.], Neotropical [eds.], Ecologicalcommunities. Conceptual issues ornithology. Omithol. Monogr. No. 36. American and the evidence. Princeton Univ. Press, Prince- Ornithologists’ Union, Washington, DC. ton, NJ. PAYNTER,R. A., JR. 1982. Ornithological gazetteer WILLIAMS,M. D. 1980. First description of the eggs of Venezuela. Museum of Comparative Zoology, of the White-winged Guan, Penelopealbipennis, Cambridge, MA. with notes on its nest. Auk 97%X89-892. PAYNT~R,R. A., JR., ANIJM. A. TRAYLOR,JR. 1977. WILLIAMS,S. O., III. 1989. Notes on the rail Rallus Ornithological gazetteer of Ecuador. Museum of longirostristenuirostris in the highlandsof central Comparative Zoology, Cambridge, MA. Mexico. Wilson Bull. 101:117-120.