SYSTEMATICS AND BIOGEOGRAPHY OF THE TYRANNID GENUS TODIROSTRUM AND RELATED GENERA (AVES)

JOHN W. FITZPATRICK 1

Abstract. Of 32 Tody-tyrant species (family positions immediately preceding that of Todi- Tyrannidae), 14 comprise the genus Todirostrum rostrum, reflecting their affinities with Todirostrum and 12 are currently placed within Idioptilon. The capitate; 3) Todirostrum and Idioptilon should remaining species represent six monotypic genera. continue to stand as separate genera; 4) the genera Todirostrum species fall into two major species Microcochlearius and Snethlagea are best merged groups (labelled cinereum and sylvia), and the with Idioptilon, while Oncostoma should continue affinities of one additional, relict species (capi- to be generically recognized. A phylogeny of the tate) lie outside the genus. The distributional Tody-tyrants and a revised classification of the history of each species group can be approximately group are proposed (Figure 4 and Table 1, re- traced from the isolation of two ancestral popu- spectively ) . lations. Alternating contraction and expansion of lowland South American forests, corresponding INTRODUCTION with dry and humid climatic phases, generated new species within each lineage. Their modern Large complexes of closely related, often and distribution were de- compositions patterns morphologically similar species are common termined and by dispersal ability competitive in the diverse avifauna of South America. interactions between related species. Current A number of recent studies have focused distributions and plumage patterns appear to re- flect the history of this radiation within Todi- on ecological relationships between sym- rostrum. patric members of potentially competing Two relict species groups (one includes T. some in species complexes ( examples Cody, capitate, the other is an intermediate assemblage 1974). However, the evolutionary history between Todirostrum and Idioptilon placed in the behind modern distributions and latter genus) illustrate the potential fate of many ecological taxa in the rapidly radiating South American avi- interactions is disregarded in many such fauna. As the number of increases with species studies, at the expense of an adequate cyclic production of new forms within lowland perspective on the problems examined. The forest refuges, each faces increasing competitive aim of the is to reconstruct pressure as it disperses. Certain species and whole present study lineages may be secondarily contracted into local this evolutionary background for a widely distributions by dispersing, competitively superior sympatric group of tropical, insectivorous relatives. The process culminates either in morpho- bird species. logic and ecological divergence by the inferior The Euscarthminae population, or in its eventual extinction. Inter- Tyrannid subfamily mediate stages in this "continental taxon cycle" (Hellmayr, 1926) contains roughly 65 spe- are found in the Tyrannid group examined here. cies, all of which are small and wholly Four specific taxonomic suggestions are offered insectivorous. This forms one of in addition to comments on affinities within subfamily Todirostrum: 1) the genus Ceratotriccus should the four major phylogenetic lineages of be with the Tacnio- merged Idioptilon; 2) genera small Tyrannidae (Smith, 1970), as shown triccus and Poecilotriccus should be moved to in Figure 1. The core of the subfamily

1 consists of an of 32 broad-billed Department of Biology, Princeton University, assemblage Princeton, New Jersey 0S540 species, placed in 10 genera by de Schauen-

Bull. Mus. Comp. Zool, 147(10): 435-463, June, 1976 435 436 Bulletin Museum of Comparative Zoology, Vol. 147, No. 10

"% Ar

Myiarchinae Platyrhynchinae

Serpophag inae

TYRANNIDAE

Cotingidae JU Pipridae suboscine ancestor of Figure 1 Schematic phylogeny of the family Tyrannidae, showing postulated relationships the Euscar- thminae within the "small" lineage. The seven subfamilies shown are those of Hellmayr (1926). Adapted from W. J. Smith (1970).

see (1966). The genera Todirostrum and none at all since Zimmer's brief comments Idioptilon contain 26 of these species (14 on some of them (1940, 1953, 1955). and 12 respectively), and each of the re- As an initial stage in the investigation of maining 6 species is currently placed in a the evolution and modern competitive monotypie genus. Together, the 32 species structure of the group, the distributions of clearly represent a well-circumscribed, all Euscarthminc species were carefully monophyletic group, though their affinities plotted on maps. These distributions were have received little critical attention, and examined in relation to recent theories in Biogeography of Todirostrum • Fitzpatrick 437

South American biogeography. It became the subfamily Euscarthminae. It is now apparent that the patterns of radiation possible to assess the validity of certain of within several subgroups could be traced these genera. Specimens of all but two of with reference to Pleistocene climate and the 32 Euscarthmine species in the Tody- vegetation fluctuations, the occurrence of tyrant lineage were assembled and exam- which has been convincingly demonstrated ined in conjunction with the present study. by Haffer (1969, 1974) and Vanzolini and The following comments and suggestions Williams (1970). are based largely on morphologic patterns Distributional and morpholigic data and biogeographic considerations covered relevant to two particular topics are pre- later. sented in this 1 detailed de- The considered in this paper: ) A species treatment scription of speciation patterns in the genus of the genus Todirostrum Lesson, 1831 are Todirostrum is given. The account includes those listed by de Schauensee (1966) with of three related here- the three 1 T. discussion species following exceptions: ) pictum tofore placed in the monotypic genera is considered to be one of three members of Poecilotriccus, Taeniotriccus, and Cerato- the chrysocrotaphum superspecies, as dis- triccus, and four additional species in the cussed in detail in the speciation section. genus Idioptilon. 2) Certain general impli- It should therefore be listed as an allo- cations of the findings to modern biogeo- species along with nigriceps and chrysocro- graphic theory are discussed as they relate taphum. 2) T. hypospodium Berlepsch, to the evolutionary history of certain mem- 1907 is considered a synonym of T. sylvia bers of the Euscarthmine assemblage. superciliare Lawrence, 1871. Known from a Ultimately, the goals of present and single unsexed specimen from the Bogota forthcoming studies are to determine, 1) trade market, hypospodium was listed as "a the modes of ecological and morphological very doubtful species" bv de Schauensee radiation in this large group of similar, (1966: 365). Both Zimmer (1955) and insectivorous species, 2) the varying com- Hellmayr (1926: 308) considered it a petitive pressures placed on each other by dark individual variant of T. sylvia super- such closely related forms, and 3) the ways ciliare, a fairly common species in the in which various populations selectively Magdalena Valley near Bogota. The un- respond to these pressures. Such analyses likely event of hypospodium s proven clearly depend upon establishing an evo- validity as a species would have little bear- lutionary perspective with which to view ing on the patterns discussed in this paper. modern geographical and ecological con- 3) T. albifacies Blake, 1959 is considered ditions. The present study is a first step, conspecific with T. capitale tricolor ( Berla, therefore, in that it focuses primarily upon 1946). Blake (1959) listed the sex of the the affinities within a subset the collected to date phylogenetic type ( only specimen ) of the Euscarthminae, and secondarily upon as uncertain, possibly male. Its close simi- certain brief dis- ecological implications. A larity to the female of sexually dimorphic cussion of taxonomic considerations ger- T. c. capitale has led Blake and Traylor mane to this the historical analysis precedes (personal communication) to reidentify it reconstruction of in Todirostrum. speciation as a female tricolor. The latter form was— TAXONOMIC COMMENTS heretofore known from a single type unequivocally male—from the Bio Madeira, Division of distinct morphologic groups about 200 km downstream in the same into numerous, often monotypic genera was as the of the rule during the early twentieth century drainage type locality "albifacies" examined the classification of the Tyrannidae (e.g., Hell- (see Fig. 10). Having type mayr, 1926). With several exceptions this of albifacies and the photographs in Bella's has revision within 1946 of the I early taxonomy escaped ( ) description male tricolor, Bulletin Museum of Com para five Zoology, Vol. 147, No. 10

chrysocrotaphum superspecies

e r u m c a I opt

sylvia

turn if ronsl

Nip" N v (russatum)

Figure 2. Head patterns in the three species groups making up the genus Todirostrum. Eleven of four- teen species are illustrated. Dotting about the faces in the sylvia group represents buffy plumage; lighter dots represent yellow plumage in the cinereum group; the crown of capitate female is dark rufous.

consider their similarities to nominate from rich green (in chrysocrotaphum, capitale sufficient to agree with Zimmer maculatum, and nominate calopterum) to 1955 in the two as or ( ) treating populations grey grey-green (cinereum and polio- subspecies. cephalum) to solid black (the isolated

The 14 species in Todirostrum fall into Peruvian race of calopterum) . Throughout three distinct species groups on the basis its range maculatum has numerous thin, of bill characteristics and plumage patterns brown streaks superimposed on its yellow (Fig. 2). The cinereum group consists of breast and belly. Thick, solid black streaks four distinct species—calopterum, cine- form whole or partial necklaces in two reum, poliocephalum, and maculatum— members of the chrysocrotaphum super- and an assemblage of three populations species. All other forms in this species eoi uprising the chrysocrotaphum super- group are clear-breasted. The progenitor species: chrysocrotaphum, nigriceps, and for the species group is therefore assumed pictum. Each of these seven forms has to have been a clear-breasted form—an bright yellow underparts and a dark grey assumption that will be used in reconstruct- or solid black crown. Back color varies ing several speciation sequences to follow. BlOGEOGRAPHY OF ToDIROSTRUM • Fitzpdtlick 439

Six species are included here in the sylvia plumage of capitale differs from that of all species group, though the affinities of T. its congeners (partially illustrated in Fig. senex remain in doubt. All have grey 2). The male is entirely black above with crowns and light grey underparts, lighten- a white breast and bellv, while the female ing into white or pale yellow on the belly is rich green above with a rufous crown, and undertail coverts. Backs of all species pearly-grey face and flanks, and white are dull green except for that of the south- underparts washed with pale yellow under ern Venezuelan endemic russatum, which the tail. This pronounced sexual dimor- has become dark olive-brown. The species phism is not only unique in the genus, but group consists of one rather distinct, poly- is quite rare in the entire family Tyranni- T. and five dae. T. stands as typic species, sylvia, ( including capitale an aberrant mem- senex) morphologically similar, allopatric ber of Todirostrum, and actually appears populations here referred to collectively as to have closer affinities with two species— the "latirostre complex." Important plum- Poecilotriccus ruficeps and Taeniotriccus age differences occur in the amount of andrei—heretofore placed rather far from buffy or cinnamon coloration in the face it within the Euscarthminae. and throat, almost entirely lacking in sylvia The monotypic genus Poecilotriccus was (Fig. 2). The bill of sylvia is basally wider originally split from Todirostrum on the than in the latirostre complex, and its basis of the bushy rufous crown, relatively breast and flanks show more uniform grey shorter bill, and somewhat shortened outer than in any of the other species. The ex- primaries of ruficeps (Ridgway, 1907). treme similarity between pluml)eiceps and However, in basic plumage pattern ruficeps russatum is unequivocal (Zimmer, 1940). is remarkably similar to the female of T. I agree with Mayr and Phelps (1967) in capitale (see Fig. 3), and capitale shows considering the latter to be a darkened, the same short outer remiges. Besides its Roraiman counterpart of plumbeiceps, darkened male plumage, capitale differs which is recognized here by placing both from ruficeps only in its uniquely wide bill within the plumbeiceps superspecies. For and much paler underparts. In general the present I place senex in this group as shape the bill of ruficeps (Fig. 3) closely well, on the basis of its buffy facial and approximates those of most other Todi- throat plumage (Hellmayr, 1910). rostrum species. Taeniotriccus andrei, a rare in The similarity between patterns of T. species collections, is divergent from all other fumifrons and T. latirostre, especially the Euscarthmine species, especially in its southeastern populations of the latter, is deep, heavy bill. A number of striking equivalent to that in the previous super- similarities it shares with both capitale and species. They are clearly closer to each ruficeps, however, provide virtually con- other than is either to the species plumbei- clusive evidence for the close relationships ceps superspecies. Particularly considering among these three species: 1) T. andrei is the between latirostre and parapatry fumi- sexually dimorphic; like capitale, the males frons along the Amazon (Fig. 9) it seems are almost wholly black while females show appropriate to treat these two as members more olive above (Fig. 3). 2) The bushy, of a second latirostre within in both sexes of superspecies ( ) deep rufous crown andrei, the latirostre complex. though interrupted in the center by a black Todirostrum capitale stands apart from crest, matches exactly the crown color of the remaining members of the genus. Its both ruficeps and capitale females. 3) extremely wide bill is shared only by senex Plumage patterns on the underparts of all (Hellmayr, 1926: 309), a species that does three species are extremely similar, and not otherwise appear closely related. The resemble no other Euscarthmine species. Bulletin Museum of Comparative Zoology, Vol. 147, No. 10

20 mm

Figure 3. Plumage patterns and bill shapes of species comprising the Todirostrum capitate lineage. Both capitate and Taeniotriccus andrei are sexually dimorphic. Crowns of all but male capitate are rich rufous (heavy dots). Yellow on the breast (light dots) is most pronounced in Poecilotriccus ruficeps, but present in all forms. The hidden white stripe in the wing of capitate males (left) further reflects the probable af- finities of this species with andrei (right). The bill of ruficeps (bottom) approaches the typical Todirostrum shape most closely, while the much heavier bills of capitate (middle) and andrei (top) have diverged con- siderably from those of all other Euscarthmine species. This may be associated with their long isolation as relict species within their rapidly diversifying tyrannid lineage.

The throat of andrei is light tawny, fading relict form to the south even retains a through light grey into a wash of yellow complete one (Figure in Berla, 1946). 4) on the belly, but broken by a broad, black The striking yellow wing stripe in andrei pectoral band. This pattern is precisely appears at the outset to be unique, but even repeated in ruficeps, though the breast this character is present, if reduced, in band is rather faint and the yellow below capitate; the outer webbings of capitale quite rich. Furthermore, nominate capitate specimens' innermost remiges are broadly shows a broken black breast band, and the edged with yellow. When the full wing Biogeography of Todirostrum ' Fitzpatrick 441

is opened, however, yellow bases on addi- long tails. Ceratotriccus furcatus is a rare, tional remiges are revealed, appearing even isolated species, restricted to the coastal in the second and third secondaries in some highlands of southeastern Brazil. Morpho- specimens. This hidden yellow band cor- logically similar in most respects to the responds exactly with the conspicuous wing other "brown" species, furcatus was origi- pattern of andrei (see Fig. 3). 5) All three nally placed in its own genus on the basis of species have black thighs, which are absent the lengthened outer rectrices, which result in all remaining Euscarthmine species. All in a somewhat forked tail. While slightly also have pale tarsi, though this character wider distally than in most Idioptilon spe- is not unique to these three forms. 6) In cies, the relatively short bill of furcatus addition to these morphologic comparisons, shows a close affinity to that genus, despite biogeographic peculiarities indicate that all its original description as "Todirostrum three species are secondarily restricted in furcatum." The clear morphologic similari- their distributions (Fig. 10). In sum, the ties it shares with the other four members evidence strongly indicates that these spe- of the "brown" species group are supple- cies represent relicts from a single, ancestral mented by biogeographic considerations; divergence from a Todirostrum precursor. all species, including furcatus, are second- Affinities among the three species are arily restricted in their distributions (Fig. closer than those between the lineage as a 11, discussion in Idioptilon section). whole and Todirostrum or Idioptilon. For Recognizing furcatus as one of the repre- this reason the current taxonomic positions sentatives in a single lineage, the other of the three species are open to question. members of which are all in the genus Possibly, they should be placed together in Idioptilon, it seems appropriate at this time a separate genus from Todirostrum. While to consider it a member of that genus. I it seems the likely course at present, addi- propose, therefore, that Ceratotriccus Caba- tional behavioral and anatomical study is nis, 1874 be merged with Idioptilon required before effecting this rather major Berlepsch, 1907, and that the species taxonomic shift. For the present, therefore, henceforth be recognized as Idioptilon fur- it seems sufficient to place the relict genera catum (Lafresnaye, 1846). Toec Hot rictus Berlepsch, 1884 and Taenio- Idioptilon kaemferi is included as a race triccus Berlepsch and Hartert, 1902 im- of I. mirandae by de Schauensee (1966), in a linear both Zimmer 1953 and mediately before Todirostrum even though ( ) arrangement of the Tyrannidae, and to be- Eisenmann (in de Schauensee, 1966: 366) gin Todirostrum with the primitive species consider it specifically distinct. The forms capitale. are undoubtedly close relatives (specimens Species in the genus Idioptilon Berlepsch, of both species examined), but they occur 1907 separate into two distinct plumage in very local distributions in eastern Brazil types, referred to here as the "brown" and separated from each other by more than "green" assemblages. The "brown" species 2,000 km. Until further information can be group will be dealt with briefly in the obtained, they should be recognized as speciation analyses to follow, as it appears distinct, and are treated here as members to be rather closely related to Todirostrum. of the mirandae superspecies. The five members of this group are "Cera- Todirostrum senex (known only from the totriccus furcatus," Idioptilon rufigulare, I. type specimen, stored in the Vienna mirandae, I. kaemferi, and I. granadense. Museum) may actually be another in this a They share conspicuously buffy eye-rings, group of relict species, rather than mem- tawny brown facial and throat plumage ber of the plumbeiceps superspecies as (darkened into a black throat in grana- previously speculated. This is suggested by face throat as well as its dense), light olive backs, and relatively its very rusty and 442 Bulletin Museum of Comparative Zoology, Vol 147, No. 10

exceeding rarity. Its unique bill, however, is necessary before the validity of this is wider than that of any modern Idioptilon division can be more accurately reassessed. species, and the hypothesis cannot be More typical Idioptilon species comprise further examined until the type can be the "green" assemblage, to which subgroup compared critically with the other repre- Microcochlearius josephinae, SnetJdagea sentatives of the relict species group. minor, and Oncostoma cinerigulare are The hypothesis that the "brown" Idio- closely related. Oncostoma—the Bentbill— ptilon assemblage comprises a primitive represents a rather divergent Central lineage is supported by the species' distinct American offshoot from Idioptilon, and its similarities to certain members of the genus generic separation appears to be warranted. Todirostrum. In overall plumage pattern, Smith (in a letter) concurs on the basis of and particularly in bill shape, the "brown" the unique vocalizations of this form in group contains species that are morpho- Panama. The first two monotypic genera, logically intermediate between typical on the other hand, are clear continental ("green") Idioptilon species and those in representatives of the "green" assemblage. the Todirostrum latirostre complex. T. Both were separated early in this century plumbeiceps and T. russatum were, in fact, on the basis of minor variations from the originally placed in the genus "Euscarth- basic Idioptilon bill structure. At. jose- momis," which was later synonymized with phinae has an excessively wide bill, and S. Idioptilon. Before the transfer of these spe- minor shows peculiarly circular, exposed cies into Todirostrum (Zimmer, 1940), nostrils in many specimens. In all other Hellmayr (1926: 317) had written that characters, however, they precisely match (Idioptilon) rufigulare was nearest to those of the "green" Idioptilon species Euscarthmomis plumbeiceps. Furthermore, group. Recognizing their phyletic positions Idioptilon mirandae was originally placed within one subgroup of a larger, generically in the genus Todirostrum (Snethlage, 1925). recognized taxon, I prefer to merge Micro- Hellmayr (1926: 305) compared it most cochlearius Chubb, 1919 and Snethlagea closely with Todirostrum fumifrons, but Berlepsch, 1909 with Idioptilon Berlepsch, noted that it "provides the passage to 1907. Affinities of josephinae remain in Euscarthmomis." some doubt, but it appears certain that The latirostre complex and the "brown" minor is closest to /. spodiops and 7. zos- Idioptilon species group may actually re- terops within this species group, based on flect a morphologic pattern carried by an similarities between the bill and nostril ancestral form which was to give rise to structures in these three species. Little the entire Todirostrum and Idioptilon lin- attention is given here to the biogeography eages. Their current similarity would thus of the "green" assemblage, pending field have resulted from each conserving certain studies of most species. characters in their common ancestry. In order to summarize the phylogenetic the existence of inter- Despite clearly affinities within it, a diagram of the entire mediate forms, the generic separation of Tody-tyrant lineage as commented upon Todirostrum and Idioptilon, questioned by in this paper is shown here as Figure 4. de Schauensee (1970), Short (1975), W. While additional behavioral and anatomical J. Smith (personal communication) and data may modify it slightly, the gross re- others, should be retained for the present. lationships indicated are defendable at this The separation prevents informative super- time. Phyletic relationships are considered species and species group divisions from in this paper primarily as a step toward being buried in nomenclature. Additional elucidating some biogeographic implica- information, particularly regarding display tions of Tyrannid distributions. Still, when behavior and vocalizations of most species, such a synthesis is possible, it should be BlOGEOGRAPHY OF ToDlROSTRUM ' Fitzpatlick 443

The TODY-TYRANTS

Figure 4. Generic relationships in the Tody-tyrants as commented upon in the text. Dashed lines enclose each lineage whose generic separation warrants continued recognition. Thus "Ceratotriccus," "Micrococh- learius" and "Snethlagea" are here grouped into the genus Idioptilon. The "brown" species group contains species morphologically intermediate between Todirostrum and Idioptilon, and is placed within the latter genus on the basis of overall plumage patterns and wing formulae. incorporated into the basic taxonomy of the SPECIATION IN TODIROSTRUM under With this in mind, the group study. Members of the genus Todirostrum occur in 4 includes dotted lines diagram Figure from southern Mexico south throughout enclosing the generic groups by which, in tropical and lower subtropical South the should now be my view, Tody- tyrants America to southern Brazil, northern Ar- recognized. gentina, and southwestern Bolivia. They The above are suggestions incorporated are small, secretive birds ( 16-22 cm, 5-9 into a revision of the broad-billed Euscarth- gm) generally inhabiting the middle and mine tyrannids presented here, along with lower strata of tropical forests and tangled summaries of the distribution type of each underbrush along forest openings and species, as Table 1. edges. They forage mostly in thick vege- 444 Bulletin Museum of Comparative Zoology, Vol. 147, No. 10

Table 1. Revised classification of the Tody- tation with periodic, rapid flits, picking tyrants (Tybaxntdae: subfamily Euscarth- from the undersides of mixae), with the distbibution type of each arthropod prey species indicated. Species known only from overhanging leaves. This characteristic type their locality abe so marked. columns foraging method makes use of their pecu- \\l> symbols abe adapted fbom mayb and liarly long, spatulate bills, the most striking Short (1970) as follows: 1 = locally dis- feature of the tributed, w = widespread; column a = mono- morphologic genus. Habitat typic species, Column B = uncomplicated preferences of each species are summarized POLYTYPIC SPECIES ( NUMBEB OF DESCRIBED RACES in Table 2. INDICATED IN PARENTHESES), COLUMN C = Reconstruction of the isolation and dif- STBONGLY DIFFEBENTIATED POLYTYPIC SPECIES ferentiation of and their ( NUMBEB OF POPULATIONS IN PARENTHESES), populations, Column D marked with X if species is a mem- patterns of dispersal leading to today's ber OF A SUPERSPECIES. BRACKETS LINK MEMBERS geographic distributions, are based on four OF SUPERSPECIES. assumptions which must be carefully noted Species B D at the outset. The first two are closely and critical: Poecilotriccus ruficeps w(4) related, Taeniotriccus andrei w 1) The primitive Todirostrum ancestor, Todirostrum and all descendent forest inhabiting spe- capitate 1(2) cies, were restricted during continental dry dnereum species to small forest group periods "refugia" within culoptcrum 1(2) their former ranges. This can be inferred nigriceps 1 from the observation that, with two ex- chrysocrotaphum w(5) ceptions discussed presently, modern dis- pictum w tributions in this occur cinereum* w(7) genus exclusively within the forested poliocephalum 1 regions of South maculatum w(5) America. While many populations are sylvia species group currently widespread, most would have sylvia w,l(5) been significantly restricted during any lot i rost re w(6) in which the areas of continental fumifrons 1(2) period plumbeiceps w(4) humid vegetation were reduced. Substantial russatum 1 evidence now indicates that vast shrink- senex (type) 1 age of the continental forests during Idioptilon Plio-Pleistocene times did indeed occur "brown" species group (geological, zoological, botanical, and furcatum 1 paleobotanical data summarized by Haffer, rufigulare 1 1974 . It ) appears safe to suppose, therefore, mirandae 1 X that the distributions of kaemferi (type) 1 X forest-inhabiting Todirostrum granadense w(7) precursors were affected by "green" species the fragmentation of their favored habitat. group 2) Modern populations with local dis- nidipendulum w(2) tributions within the continental forests striaticolle w(5) reflect their of and spodiops 1 directly places origin zosterops w(4) differentiation. A number of these local minor w(2) Todirostrum populations currently exist, aenigma (type) 1 and all of them coincide with refugium inomatum (type) 1 X~] sites Haffer margaritaceiventer w(7) XJ proposed by (1974: 145). orbitutum w Actually, a species originating in one area josephinae 1 could have been secondarily restricted to Oncostoma cinerigulare w(2) another, wholly different region, thus giv- Includes T. viridanum, recently found sympatric with the cinereum in mv. Venezuela by Phelps. ing impression of having differentiated BlOGEOGRAPHY OF ToDIROSTRUM ' Fltzpatfick 445

Table 2. Known or described habitat preferences of species in the cenus Touirostrvm. Ref- erence LIST FOR MOST SPECIES IS NOT EXHAUSTIVE, BUT SELECTED TO INCLUDE THE MOST IMPORTANT OR THOROUGH DESCRIPTIONS AVAILABLE. "LOCALITY DATA" REFERS TO DESCRIPTIONS OF COLLECTING LO- CALITIES OR NOTES ON SPECIMEN LABELS.

Species Habitat References

chrysocrotaphum tropical; forest canopy, Skutch (1972), Terborgh superspecies forest edges, plantations and Weske (1969), Haver- schmidt (1968), pers. obs.

calopterum upper tropical; forest locality data, pers. obs. undergrowth near mountains

cinereum tropical, lower subtropical; Skutch (1930), Slud (1960), scrub, open forest undergrowth: Miller (1963), Terborgh obs. coastal mangroves ( Cent. Amer. ) and Weske (1969), pers. poliocephalum same as cinereum Sick (1968)

maculatum forested 1955 tropical; creek Haverschmidt ( ) , edges, marshy river bottoms; locality data, pers. obs.

coastal mangroves ( Guianas ) sylvia tropical, lower subtropical; Miller (1947), Gilliard dense thickets, scrub, forest (1959), Slud (1960, 1964), understory pers. obs. lativostre tropical; dense forest under- Olivares (1955), Terborgh

growth, thick tangles, cane- and Weske ( 1969, pers. brakes along rivers coram.), locality data fumifrons tropical; tangled underbrush, Haverschmidt (1968), open thickets, savana edge Novaes (1970) plumheiceps subtropical; thick brush, Wetmore (1926), locality forest edge data

in russatum subtropical; atop cerros Mayr and Phelps ( 1967 ) Venezuelan tableland cap it ale tropical; forest locality data

there. Since we have no means of prov- tinction rates in each refuge increased ing otherwise, it is assumed here that any (MacArthur and Wilson, 1967). Hence, if secondary restriction of a species' distri- a species that was initially split into four bution coincided with the area in which or five separate populations survived at all, it was initially isolated. Potential historical it may have done so in only one or two of inaccuracies resulting from this assumption the refugia. are of little consequence to the present 4) Foraging modes and habitat prefer- study, since only the most recent geo- ences of certain forest populations shifted graphic relationships among species are or generalized to include more open habi- of concern here. tats during isolation (because of increased 3) Splitting populations during periods competition within the shrinking forest of forest contraction resulted in extinctions islands), permitting colonization of open of daughter populations in some refuges. scrub regions during the dry periods. This As forest "islands" decreased in size the applies particularly to populations of T. smaller areas became decreasingly capable cinereum and T. sylvia, both of which cur- of supporting the regional fauna, and ex- rently inhabit vegetation types that may 446 Bulletin Museum of Comparative Zoology, Vol. 147, No. 10

chrysocr

colopteru

CINEREUM SPECIES GROUP

DRY

Figure 5a.

Figure 5. Hypothesized patterns of speciation within the cinereum (5a) and sylvia (5b) species groups in Todirostrum. Isolation of ancestral populations (shaded areas) during two dry periods and their directions of dispersal from initial refuge sites are diagrammed. During dry periods populations of cinereum and sylvia dispersed through regions of open vegetation (solid arrows). Dashed lines outline hypothesized distribu- tions of these species during a recent dry period. The distribution of cinereum was restricted during a pre- vious humid period, when the central Amazonian forest had returned. This range contraction (see dashed arrows away from Amazonia in 5a) resulted in the isolation of a population in southeastern Brazil, which be- came T. poliocephalum. During humid periods ancestral populations of the modern forest-inhabiting species dispersed (dashed arrows), but contracted again during a recent dry period (upper right diagrams in 5a and 5b). Todirostrum maculatum might not have been fully restricted to a refugium during the recent dry phase given its current preference for mangroves and streamside vegetation, and the ancestral distribution shown here (heavily dotted region in 5a, with question mark) is speculative. have been widespread on the continent to two or three cycles would be impossible during the dry phases. to determine from modern distribution With each climatic cycle, a new magni- data. For this reason, the following re- tude of complexity is naturally added to the construction begins subsequent to the iso- distribution patterns in any phyletic group. lation of three species group precursors 1 fence precise ranges of forms existing prior from the single Todirostrum ancestor. BlOGEOGRAPHY OF TODIROSTRUM ' Fttzpatlick 447

latirostre

SYLVIA SPECIES GROUP

plumbeiceps

Figure 5b.

Figure 5 provides a diagrammatic summary divergent races, the nominate (eastern of the speciatiou patterns derived in detail Ecuador and northern Peru) slightly more in the following sections. widespread than T. c. pulchelhim (south- eastern Peru; see Fig. 6). The northern Cinereum Species Group race resembles the other members of this group, with a green back, black Four populations originally isolated from species crown, and clear The the precursor of this group have apparently yellow underparts. Peruvian in contrast, has a survived, currently represented by five spe- race, striking facial with dark malar and cies. T. poliocephalum probably originated pattern stripes chestnut loral and its from a southerly advance of the T. deep spots, upper- are black. the cinereum progenitor during a subsequent parts entirely Assuming that dry period. species originated in one of the two regions T. calopterum currently consists of two to which it is currently restricted (as- 448 Bulletin Museum of Comparative Zoology, Vol. 147, No. 10

Figure 6. Distribution of three Todirostrum species in the cinereum species group. The absence of cinereum in central Amazonia may be due to the presence there of T. maculatum (see text). Diagonal hatching = cinereum; dots = poliocephalum; squares = two populations of calopterum (individual localities).

sumption 2 above), calopterum appears to and its uniquely divergent plumage attests have originated in an eastern Ecuadorian to a long separation from the northern race. forest refuge (Haffer's large "Napo" re- Distribution and current geographic vari- fugium). This hypothesis would account ation in the chrysocrotaphum superspecies for the similarity of the northern race to are illustrated in Figure 7. The clear yellow the other species in the cinereum group. underparts of the species group precursor The species probably then colonized south- have been retained in nigriceps and three ward along the post-glacially expanding races south of the Amazon River. Particu- forests east of the Andes. Its continuous larly below its confluence with the Rio north-south distribution was .severed by a Negro, the Amazon is a complete geo- return of dry conditions that split the east- graphic barrier separating streaked from \jndean forest again into refuges. The unstreaked populations. Geographic iso- southerly population has remained isolated. lation even between the two forms on either BlOGEOGRAPHY OF ToDIROSTRVM ' Fttzpattick 449

Figure 7. Distribution of the Todirostrum chrysocrotaphum superspecies. Representations of the main mor- phologic types are shown in their approximate geographic positions. From left to right, the five described races of chrysocrotaphum are: guttatum, chrysocrotaphum, neglectum, similis, and illigeri. See text for de- tailed analysis of this variation.

side of the Rio Negro has been borne out rently clear-breasted birds. A Peruvian by collections along the lower part of the precursor (of chrysocrotaphum) would pre- river. The superspecies thus presents a sumably have had sufficient time before reliable record of the different forest colonizing across the Andes to evolve cer- regions within which its initial populations tain of the patterns it carries today. T. differentiated. nigriceps, however, lacks any trace of the As previously surmised, the breast streaks distinct eye stripe of neglectum. Further- of pictum and guttatum evolved second- more, nigriceps retains the white throat arily on the previously clear-breasted form. characteristic of most other Todirostrum The original chrysocrotaphum stock was species but lost in neglectum. Thus it ap- therefore isolated either in the southeastern pears that the superspecies precursor dif- Peruvian forests or in one of the north- ferentiated in one of the refuges of northern western Colombian forest refuges since Colombia (see Fig. 5a), probably during these two areas retain clear-breasted popu- the same period in which the initial calo- lations. During a period of forest expansion pterum population was isolated in the Napo the population then spread through the refuge. western Amazonian region (where streaks Following initial isolation of the species, later evolved) to the other region of cur- the clear-breasted population crossed the 450 Bulletin Museum of Comparative Zoology, Vol. 147, No. 10

Andes and probably spread through much breast streaks and loral spots to the west, of its present Amazonian range. The sub- before open vegetation in middle Amazonia sequent pattern of differentiation, caused (Haffer, 1967) again separated the eastern by the return of dry conditions, can be and western populations. Upon subsequent surmised from Figure 7. A distinctive popu- dispersal of both populations back toward lation called pictum now ranges from the the Rio Negro, guttatum and pictum then Guianas west to the Rio Negro. T. c. excluded each other on their respective neglectum, with its unstreaked breast, yel- banks. While a complete series of speci- low eye stripe, and solid black forecrown, mens from northern Brazil may still prove occurs from eastern Bolivia north to the that the forms interbreed there, the avail- Amazon. The remaining races variously able evidence indicates that isolation is combine characters from both forms. The complete. T. pictum thus represents as clear necklace and loral spot of pictum have a geographic representative as nigriceps spread westward, while the eye stripe of and chrysocrotaphum within this interest- neglectum has spread north and east. This ing superspecies. distribution of characters indicates that Todirostrum cinereum, the most wide- forest refuges in the Guianas and in south- spread species in the genus, apparently de- ern Peru were the major centers for Ama- veloped its more open habitat preferences zonian differentiation during the last major during its initial isolation from the species dry period. group precursor. The closely related As the forest returned, the southern southeast Brazilian endemic, T. poliocepha- population spread eastward to the Atlantic lum, occupies similar vegetation types, coast, where it is represented today by the commonly including garden bushes and race illigeri. The characters of pictum have shrubbery (Sick, 196S). This latter form crossed the upper Amazon in western Brazil differs from cinereum only in its more and Peru, hence certain specimens still conspicuous yellow lores and distinctly assignable to neglectum show traces of the greener back and tail. Their distributions necklace and loral spots. Loral spots have overlap only marginally in southeastern now spread eastward again along the south- Brazil (see Figure 6) and the southeasterly ern bank of the Amazon, currently termi- occurrence of cinereum appears to be a nating in the race called similis west of Rio reinvasion from the northwest after the Tapajoz. T. c. guttatum, in eastern Ecua- population that became polioceplialum dor and northwestern Brazil, shows the full had been isolated. array of characters from both original The hypothesis that cinereum originated populations (see Fig. 7). All racial char- in the north (Fig. 5) is supported by two acters in chrysocrotaphum shown in the additional lines of evidence. First, the two figure are clinal. branches of its range, the races coloreum The current distribution of characters in the south and cearae to the north, differ near the mouth of the Rio Negro and north rather sharply in color and measurements, into southern Venezuela remains somewhat while populations to the west of each dif- puzzling. The breast and loral markings of ferentiate along clines. The indication is pictum have moved westward, while the that the north and south Brazilian forms conspicuous eye stripe of neglectum re- colonized independently from a common mains restricted to the west bank of the population northwest of both, and have not Rio Negro and Rio Branco. This is probably shared recent contact with each other. due to the sequence by which pictum ex- Second, cinereum is extremely common in panded westward, and may indicate a a variety of habitats in northern South double invasion of the region from the America, and occurs in tropical regions Guianas. The first invasion contributed throughout Central America as far north as Biogeography of Todirostrum • Fitzpatrick 451

Figure 8. Distribution of Todirostrum maculatum. Compare its extensive Amazonian distribution with that of 7. cinereum (Fig. 6), a possible ecological competitor which appears to be excluded from central Amazonia. Diagonal hatching — overall geographic range of maculatum. Dots indicate known collecting sites for the species, including one questionable locality.

Vera Cruz, Mexico (Griscom et ah, 1957). Peru and comments by Haverschmidt in certain of it. In southeastern While fairly common regions ( 1968 ) already support Brazil, it is not nearly as widespread there Peru (dept. Madre de Dios) maculatum as might be expected had it originated in inhabits thickly vegetated, early succes- the south. The species appears to be better sional margins along rivers, especially adapted to the more northerly scrub bordering beaches, where it forages one to regions, having secondarily advanced south ten meters from the ground (Fitzpatrick, and east into Brazil. unpublished data). In many respects these The most recent reappearance of the habits strongly resemble those of cinereum Amazonian forest probably broke a nearly in the Apurimac Valley in central Peru continuous distribution of cinereum in (dept. Ayacucho), where maculatum does western Amazonia, where it now occurs not occur (Terborgh, personal communi- only along the base of the Andes. cation). Furthermore, while cinereum is a It currently appears that the occurrence common mangrove inhabitant in Central of Todirostrum maculatum along river and northwestern South America, it is ap- margins in Amazonia contributes to the parently restricted to the drier habitats in absence of cinereum throughout much of Surinam, where 1 maculatum has replaced it that region (compare Figures 6 and 8). as the common mangrove Todv-tyrant Comparative ecologies of these species are (Haverschmidt, 1968). presently being studied to test this hy- While collecting in the cerrado region of pothesis, but preliminary field data from central Brazil has been spotty, that area 10 452 Bulletin Museum of Comparative Zoology, Vol. 147, No.

does contain a certain fraction of endemic, Sylvia Species Group another locally adapted species, including The distribution of sylvia coincides with small Euscarthmine flycatcher (Euscarth- the scrublands of southern Central America It mus rufomarginatus; Sick, 1965, 1966). and northern South America (Fig. 9). This be that the absence of cinereum in may species apparently shifted its ecological central Brazil is also due to the presence preferences in a fashion similar to T. cine- there of a better of com- adapted complex reum, during or immediately following its peting species. original isolation. In a subsequent diy The and differentiation of T. origin period it then spread north into Central maculatum can be inferred by examining America and south along the Atlantic variation the Ama- its present clinal along coast. This expanded distribution was in- zon shown in This (distribution Fig. 8). terrupted by the return of the modern consists of a crown gradually lightening forests, which restricted nonforest inhabi- from the black-crowned amacurense, race, er 1967 . Two tants as elaborated by Haff ( ) in the Orinoco delta and (Eisenmann uncommon, relict populations mark the to the or Phelps, 1971) light grey green once larger range of sylvia: one inhabits crowns in of eastern Peru (Zim- signatum the dry, inland savanna region of the 1940: There also to be mer, 5-7). appears Guianas (T. s. sylvia) and the other, the a concomitant of the breast broadening caatinga scrub of northeastern Brazil (T. streaks toward the west. Both clines leave sylvia schulzi). the birds much closer to the Guyana pre- The ecological position of sylvia relative sumed of the plumage species group to other small Tyrannid species is of con- in the taxo- percursor (as hypothesized siderable interest, though its significance nomic that macu- comments), indicating to the species' patterns of distribution and latum in a northeastern coastal originated abundance has yet to be worked out. Slud refuge. Lightening of the crown would indicates that its ( 1964 ) sylvia generalizes have occurred as the subsequently, species habitat requirements to include more open Amazonia. spread westward throughout shrubbery and second growth forests in is now abundant in The species mangroves regions where T. cinereum or Oncostoma the Atlantic coast of the Guianas and along cinerigulare (a Euscarthmine species closely the Venezuela, and occurs along margins related to Todirostrum) or both are locally of streams and rivers in the Amazon basin. absent. This observation, which warrants Preference for the river margin habitat further field investigation, suggests that the perhaps explains why, in contrast to sylvia is competitively inferior in habitats canopy-dwelling chrysocrotaphum repre- shared with these relatives. This may ex- or absence of in sentatives, maculatum does not occur in plain the rarity sylvia otherwise suitable habitats throughout distinctly isolated geographic forms. Dur- northern South America, where cinereum ing periods of complete separation between is common and widespread. The disappear- true forest populations, such as those in ance of sylvia in these areas, initiated by the chnjsocrotaphum superspecies, the dis- the reduction of its habitat into local tribution of maculatum was probably more islands, may thus have been catalyzed by continuous along locally moist stream the presence of a highly successful, con- thickets, and canebrakes. Thus edges, generic competitor. have under- maculatum populations might T. latirostre is the only wide ranging spe- gone only limited periods of isolation dur- cies in the "latirostre complex" and its pat- ing Pleistocene dry phases, if in fact the tern of variation gives strong evidence for species was fragmented at all during that the sequence by which the complex di- time. verged. In eastern Ecuador and north- BlOGEOGRAPHY OF ToDIROSTRUM • Fitzpdtlick 453

Figure 9. Distribution of the six species comprising the Todirostrum sylvia species group. All five members of the "latirostre complex" (which excludes sylvia) remain allopatric, though T.—senex is known only from its type locality on the Rio Madeira (Borba). Diagonal hatching— downward to left plumbeiceps superspecies— (triangles = collecting localities for russatum); stippling latirostre superspecies (open circles collecting localities for fumifrons); diagonal hatching downward to right = sylvia (X =z collecting localities for the two eastern races, sylvia and schulzi). eastern Peru, its dark grey crown, cinna- forest-edge habitats east of the range of mon-buff auricular and ocular regions, light latirostre. The most informative feature of buffy throat, and tawny wingbars all the plumage pattern of latirostre is its re- strongly resemble the plumage of T. plum- markable resemblance, in the region of the beiceps. Fading of each of these characters former Napo refuge in eastern Ecuador, to south and east of Ecuador results in south- that of the plumbeiceps superspecies. This easternmost specimens that closely ap- is evidence that the Ecuadorian population proach the plumage of T. fumifrons. The of latirostre most closely represents the latter is a rare species inhabiting scrub and primitive one, and that the species origi- 454 Bulletin Museum of Comparative Zoology, Vol. 147, No. 10

nated there. This hypothesis is strengthened itself (see Fig. 9). The other cis- Andean by the generally northeastern Amazonian species, again reduced to its refuge in distribution of the species today. eastern Ecuador, left an isolate in north- Two alternative patterns of speciation in eastern Brazil which differentiated into the this species group can be hypothesized. two races of T. fumifrons. Today, latirostre First, a recent dry period could have iso- has followed the forest and river valleys lated each separate population from a back eastward into central and southern single, wide-ranging species. T. latirostre Amazonia following its last isolation. Its would then have spread eastward into its plumage in the drier forested valleys of the current range following the return of Brazilian cerrado appears to be responding forests there. This hypothesis, however, to pressures similar to those which affected fails to account for the particular similarity fumifrons. In these areas it has become between plumbeiceps and russatum, which dull and faded and thus strongly resembles appear to be immediately descendant from the pattern of its sister species. a form already distinct from the latirostre- The present distribution of latirostre fumifrons ancestor. Furthermore, plumbei- reaches, but does not overlap, those of to the south to ceps is currently represented by a Bolivian pluml)eiceps and fumifrons population which itself has differentiated the east. It appears to represent a situation into recognizable races (Zimmer, 1940). in which close congeners, adapted to their This Bolivian isolation must have occurred respective place of origin, exclude each during a major dry period which had al- other across an intermediate area. Con- ready isolated russatum in Venezuela, since firmation of this hypothesis requires precise Bolivian specimens of plumbeiceps are ecological information from different areas much closer to those from southeastern within the ranges of each respective spe- Brazil than to the isolated russatum. cies, and at their junctions. These measure- It appears, instead, that a parent stock ments, for most or all species in the group, will to of the for this species group was initially split into be crucial the completion a population north of the Andes, which be- biogeographie analysis initiated here. came sylvia, and one to the southeast which gave rise to the latirostre complex. As open Capitate Species Group vegetation continued to spread, the cis- Todirostrum capitole clearly originated Andean form was reduced to isolated in the forests of the Napo refuge, to which populations in the eastern Ecuadorian area it remains almost exclusively re- forests and in southeastern Brazil (see Fig. stricted (Fig. 10), and appears to represent 5b). This hypothesized sequence is con- a deep forest relict of a third population sistent with the observed similarity between isolated from an original Todirostrum an- plumbeiceps and the population of lati- cestor. From the Napo region, its distri- rostre occupying the region of the Napo bution has probably undergone expansions refuge. As forests returned, the Napo spe- and contractions, but only two specimens cies spread eastward and the Brazilian form have been collected that attest to the spe- spread north and west. As a subsequent cies' occurrence elsewhere. The southerly dry period began (probably the last major population, T. capitale tricolor, originated continental arid the southern phase), spe- in a south Peruvian refuge following a cies had left an isolate on the Venezuelan southward colonization by the Napo popu- and highlands (now russatum) possibly lation. It has now extended along the low- one in central Brazil (senex) before being land forests of the Rio Madeira drainage reduced to its Bolivian-Brazilian range. into western Brazil, where it apparently Further habitat reduction then caused the remains rare. current gap in the range of plumbeiceps Several divergent characteristics of the Biogeography of Todirostrum ' Fitzpatrick 455

andrei

Figure 10. Distributions of Todirostrum capitate, Taeniotriccus andrei, and Poecilotriccus ruficeps. The latter two genera are monotypic. Morphologic and biogeographic evidence (see text and Figure 3) indicate that these three species comprise—a single, relict Tody-tyrant lineage. Vertical hatching = distribution of Poecilo- triccus ruficeps (small dots collecting localities); horizontal hatching — distribution of two populations of Todirostrum capitate (triangles — collecting localities); stippling = distribution of Taeniotriccus andrei (open circles = collecting localities).

species, discussed in the taxonomic com- cally common within its range, its distri- ments above, suggest that the capitale bution is restricted to a narrow forest zone lineage is at least as old as the two major between 1900 and 2700 m in the Andes Todirostrum species groups. Most im- from northern Peru to extreme western portant, the pronounced sexual dimorphism Venezuela (Fig. 10). Without adequate in capitale implies a long period of be- field data it is impossible to speculate on havioral as well as morphologic divergence, the precise ecological pressures constrain- and attests to the species' long isolation ing the distribution of ruficeps. However, from the other Todirostrum species. In fact, nearly every Euscarthmine species occupies the capitale lineage is probably much older lowland habitats, and ruficeps undoubtedly than the two larger species groups, and differentiated originally as a tropical forest appears to contain two additional, equally species. Narrow altitudinal distributions divergent relicts, as shown in the taxo- like that of ruficeps may result from in- nomic section. In addition to their peculiar vasions into a forest species' original range plumage patterns and bill structures, each by one or more species with superior com- of the three species exhibits a secondarily petitive potential. Individuals of the origi- restricted distribution, further indicating nal species that are able to survive and the age and relict position of this lineage reproduce at higher altitudes eventually relative to the other two species groups. provide refuge for the species, gradually Although Poecilotriccus ruficeps is lo- forced from its original lowland habitat 456 Bulletin Museum of Comparative Zoology, Vol. 147, No. 10

through competition. All such shifts are interesting and revealing distribution pat- limited by analogous competitive pressures terns in the relict "brown" species group from species already occurring higher on (Fig. 11) will be discussed briefly here. the slopes. Data provided by Terborgh Idioptilon rufigulare is a perfect example (1971), Terborgh and Weske (1975), and of an excessively rare montane relict spe- Diamond (1972) give compelling evidence cies. It is known from six specimens, all of for the action of competition along alti- which were collected between 1200 and tudinal gradients on the distributions of 1500 m on isolated Andean ridges in Peru montane birds. and extreme northern Bolivia. This re- Presumably following range expansions stricted altitudinal distribution was recently and contractions by capitate and andrei, confirmed by J. Terborgh (personal com- both of these species appear to have be- munication) in central Peru, where the come rather rare and local. This, too, species was uncommon but present at 1500 probably resulted from radiation within m on a small ridge separated from the main closely related Tyrannid species placing Andes by 100 km of lowland forest. Per- increasing competitive pressures on the haps once a wide-ranging species, rufi- earlier evolutionary line. As additional gulare is now restricted to a narrow, middle successful species were periodically added altitude zone on certain slopes which, by to the forest fauna, certain species neces- virtue of their reduced avifauna, are rel- sarily decreased in abundance. Their overall atively free of lower and higher altitude distributions may thus have been reduced competitors (Terborgh and Weske, 1975). to regions of particularly favorable habitat Few ecological data are available for the (possibly the case with andrei) or re- 7. mirandae superspecies. Clearly, however, stricted to ever-diminishing zones in which the limited ranges of the two members, some minimum population level could be separated by more than 2000 km, reflect maintained (as in capitale). These two a tiny fraction of their common ancestral species thus appear to be engaged in—and distribution in eastern Brazil. Similarly, possibly losing—an "evolutionary race" as furcatum (formerly "Ceratotriccus" fur- their related competitors continue to radi- catus) currently occupies an extremely re- ate. As their populations decline, they stricted range within the wooded coastal either diverge sufficiently in a direction hills of Rio de Janeiro, Brazil. that reduces this competition (causing the Of the "brown" Idioptilon species, only peculiarly divergent structures, plumages, granadense does not appear to have be- and presumably behavior of both capitale come significantly restricted in its distri- and andrei), or they are ultimately driven bution. Morphologically, however, this to extinction by their competitive environ- species is the most divergent in its genus, ment. A slightly more detailed description and, more significant, it is one of only three of this process is presented in the discussion in the genus that occur at higher elevations. section. (The other two, rufigulare and spodiops, are rare mountain relicts.) In isolated north IDIOPTILON AND RELATED SPECIES Andean ranges, granadense occurs in an alti- tudinal zone about 1000 m wide 1500 to As previously discussed, species in ( 2400 m in the Sierra Nevada de Santa Idioptilon fall into two plumage types, the 1800 to 2900 in the "brown" and "green" assemblages, which Marta, Colombia; m the main correspond phyletically with the species Sierra de Perija, Venezuela). On groups in Todirostrum. The "brown" line- Andean slopes, in a more crowded competi- age appears to be phylogenetically inter- tive community, its altitudinal limits are mediate between Todirostrum and the more restricted (22S0 to 2600 m, west of typical ("green") Idioptilon lineage. The the Apurimac River, central Peru; Terborgh BlOGEOGRAPHY OF ToDIROSTRUM • FttZpaMck 457

Figure 11. Distributions of the five populations in the "brown" Idioptilon species group, showing known collecting localities for each species. All but granadense are extremely rare, relict species. The upper sub- tropical altitudinal range of granadense (2200-2600 m in central Peru) differs from all other Idioptilon species, and appears to be a response to the radiation of assemblages of lowland competitors which has driven the remaining members of this species group close to extinction. Small dots — granadense- large dots = rufigulare; squares = mirandae superspecies (closed = mirandae, open = kaemferi); triangles = "Ceratotriccus" (Idioptilon) furcatus.

and Weske, 1975). It remains an upper the ancestral lineage. Four members have cloud forest inhabitant throughout its been fragmented or compressed into re- range, however, and this habitat is shared stricted distributions within the forested by no other close relative. tropics and subtropics. The fifth was Distribution and morphologic divergence secondarily forced up into an altitudinal in the brown Idioptilon species group ap- zone that remains unique to the genus. pear analogous to those characterizing the As discussed in the next section, the pres- capitale lineage in Todirostrum. Members sures forcing these early species into their of both groups show relict distributions or relict conditions apparently resulted from divergent characters, or both, compared the continued production of new, competi- with the other members of their respective tively superior relatives within the tropical genera. As in the capitale group, the five forests. brown Idioptilon species apparently com- The remaining members of the genus prise an early phylogenetic divergence from Idioptilon, together with the monotypic 458 Bulletin Museum of Comparative /.oology, Vol 147, No. 10

genera Microcochlearius, Snethlagea, and b. Todirostrum latirostre — T. fumi- Oncostoma, constitute the "green" assem- frons blage. These eleven species appear to form c. Todirostrum plumbeiceps — T. rus- another monophyletic subgroup within the satum broad-billed Euscarthmine Tyrannids. Ad- 3) Closely related populations currently ditional data on behavior and ecological in contact but not overlapping (para- relationships are required before affinities patric). and distributional patterns in this difficult a. Todirostrum chrysocrotaphum — T. group can be satisfactorily commented pictum upon. Research is currently underway b. Todirostrum latirostre — T. plum- which may provide some of this informa- beiceps tion. 4) Well-differentiated species, partly sympatric but apparently excluding DISCUSSION each other geographically over a broad contact zone; simi- The Todirostrum capitale lineage and the presumably lar "brown" Idioptilon species group are ex- ecologically. a. Todirostrum maculatum — T. cine- treme cases illustrating the final of a stages reum speciation-dispersal-extinction process that b. Todirostrum cinereum — T. appears to be a continental analogue of the polio- "taxon cycle" characterizing island faunas cephalum c. Todirostrum cinereum — T. (Wilson,' 1961; Ricklefs and Cox, 1972). In sylvia but this section, some illustrative examples of (entirely sympatric, sylvia's much contracted in the main stages in the process precede a range presence of see discussion in brief discussion of the dynamics of this im- cinereum; speci- ation portant phenomenon. section) 5) Well-differentiated or In the following outline, nine types of species, mostly distributional relationships are delimited entirely sympatric; many examples, between which summarize the biogeographic "life especially species groups. a. Todirostrum maculatum — T. history" of certain lowland South American chry- bird species. By no means are these cate- socrotaphum superspecies b. Todirostrum maculatum — T. lati- gories meant to include all probable inter- rostre mediate stages in the process to be dis- c. — cussed; rather, only those stages are chosen Todirostrum plumbeiceps T. which characterize members of the Tody- poliocephalum tyrant group dealt with in this paper, and 6) Species presumably widespread at the examples are included under each cate- one time, now reduced to one or more gory. localized ranges; populations prob- ably declining; often morphologically 1) Newly isolated populations, presum- divergent. ably still representing a single species. a. Todirostrum capitale a. races of Todirostrum sylvia b. Todirostrum calopterum b. races of Iclioptilon margaritacei- c. Cera- venter ("green" assemblage) Idioptilon furcatum (formerly totriccus 2) Populations apparently isolated prior ) to d. mirandae the last major dry period; signifi- Idioptilon superspecies Lowland still rather wide- cantly diverged but not yet in second- 7) species ary contact. spread, but very rare; either reduced a. Todirostrum nigriceps — T. chry- to low absolute density or restricted socrotaphum to rare habitats. Bioceography of Todirostrvm • Fitzpatrick 459

a. Taeniotriccus andrei populations during a subsequent climatic b. Microcochlearius josephinae cycle. 8) Species with wide altitudinal toler- A geometric increase in species numbers ance, currently forced up to higher theoretically associated with periodic frag- altitudes by lowland competitors. mentations of dispersing populations is Example: probably Todirostrum naturally damped by ecological factors. calopterum pulchellum, though con- Members of each phylogenetic lineage face clusive data are lacking; northern increasing competitive pressure as the num- race still occurs in lowland forests, ber of closely related species on the con- but pulchellum is rare and restricted tinent increases. The probability that to upper tropical zone in southeast- regions potentially available for dispersal ern Peru. might already be occupied by related forms 9) Species related to lowland forms but increases with the rising number of species now restricted to narrow, higher alti- in each taxon. In each such lineage, there- tudinal zone in the Andes. fore, dispersal of new forms grows increas- a. Idioptilon rufigulare (1200-1500 m) ingly restricted. b. Poecilotriccus ruficeps (1900-2700 Survival of each population hinges upon m) one of two requisites: 1) it must occupy, c. Idioptilon granadense (1800-2900 either initially or secondarily, an ecological m, Venezuela; 2200-2600 m, Peru) position sufficiently different from those occupied by sympatric species; or, 2) it Plio-Pleistocene radiation of the sub- must maintain a competitive superiority oscines occured with apparently relatively sufficient to cause the decline and eventual little competition from other avian groups extinction of ecologically similar species in South America. Five major suboscine (e.g., MacArthur, 1972). Radiation within families comprise well over half of the a taxon, and over regional habitats in continent's passerine fauna, and the 302 is thus mediated by continual ex- one general, Tyrannid flycatchers, representing tinctions of new as well as established tenth of South America's 2900 land bird populations. Extinctions in this context species (Smith and Vuilleumier, 1971), result from competitive interactions in two radiated into an almost unparallelled array situations: 1) the regional dispersal of of morphologic groups (summarized by species into areas occupied by competi- Keast, 1972). effects tively inferior species, and 2) island Haffer's extensive work on the zoogeog- within the shrunken areas of the forest raphy of lowland South America has given refuges (MacArthur and Wilson, 1967). us a clearer of how such radiation picture The dynamic interplay between speci- came that about, emphasizing speciation ation and extinction has, in short, caused has been a for millions of repetitive process net species compositions within most taxa continental years. During dry periods, pre- and over entire geographic areas to move associated with world-wide sumably glacial toward equilibrium levels. This process is forest were into iso- episodes, species split diagrammatically illustrated in Figure 12. lated populations each with the potential A number of phenomena variously associ- for differentiating into a new species. Dur- ated with the process described here are ing humid climatic phases, species inhabit- reflected in the current distributions of ing open, nonforest vegetation underwent bird species in lowland South America. similar fragmentation. Faunal dispersal Four points will be briefly discussed that within expanding vegetation zones resulted appear to be critical to our interpretation of in the splitting of newer species into addi- the nine categories of tyrannid distributions tional arrays of potentially differentiating presented at the beginning of this section. Bulletin Museum of Comparative Zoology, Vol. 147, No. 10

RELATIVE NUMBER OF SPECIES

TIME

Figure 12. Schematic changes in the relative number of species in the Amazonian forest over time. Shaded areas represent periods of forest fragmentation, associated with dry climatic periods, which split forest species into numerous potentially differentiating populations. New species disperse as the forest returns. These periodic increases in species number are gradually offset by extinctions as competition increases, causing the overall Amazonian diversity to level off at some potential equilibrium value (A). One cycle is shown in detail in the upper left: as the forest contracts, island effects reduce slightly the total number of species (B); differentiation of remaining populations causes an increase in the number of reproductively iso- ex- lated forms (C); as these disperse with the returning forest, competitive interaction forces species toward tinction, gradually reducing the total number of species (D), which would theoretically smooth out to some stable stable species composition (E) if the forest remained intact for a long enough period. Note that this value increases by smaller increments with each cycle. Heavy arrow indicates hypothesized present position along the species abundance curve.

1 As the of increases as each from its of ) number species spreads respective place within a large region like the Amazonian origin. In certain groups of species, each forest the number of forms with small, maintains sufficient competitive advantage localized ranges correspondingly increases. in its original geographic location to ex- Examples of these were used by Haffer clude the others from its range. This results 1969 in his forest in the mosaics of illus- ( ) reconstruction of refuge parapatric species locations, but few discussions of the eco- trated abundantly by Haffer (1974), and logical implications of such endemics have by the Tody-tyrants in categories 3 and 4 appeared to date. Endemic species within in the examples presented above. the Amazonian forest result from the mutu- In other cases, the competitive superiority pressing effects on dispersal potential of a dispersing species over certain others sxerted on each other by related species may gradually restrict the distribution of Biogeography of Todirostrum • Fitzpatrick 461

the latter as the superior species grows in- Given their inability to stably coexist within creasingly widespread. Ultimately the in- a habitat, the pair would thus exclude each ferior species will vanish completely, but other altitudinally (categories 8 and 9). This not before a period of gradually diminish- process might happen repeatedly along ing success within its limited range. Be- mountain ranges, resulting in the "stacking'' cause of the variable and potentially great of species along altitudinal gradients docu- of this to find at mented 1971 and length period, we expect by Terborgh ( ) Diamond any point in time species representing (1972). various stages of this contraction (category 4) All changes by which a species shifts 6 above). Alternatively, the same gradual away from competitors occur most readily decrease in a species' abundance could oc- in species with small ranges, where gene cur without a corresponding decrease in its flow is sufficiently localized to allow rapid geographic distribution. This type of de- spread of more competitively successful cline might characterize species occupying morphological or behavioral adaptations. peculiar, locally distributed habitats, each It therefore becomes increasingly easy for a patch of which would require separate declining species to diverge into a new colonization by the complex of superior ecological role as its range is reduced, re- competitors. A species' overall population sulting in the "evolutionary race" intro- could thus decline toward extinction while duced earlier with reference to the relict still occupying a wide range (category 7). and divergent T. capitcde lineage. As a 2) An invaded, declining species can species grows decreasingly capable of alter its mode of habitat utilization and thus competing successfully within the com- escape the competitive pressure causing its munity its range or abundance or both decline. In its extreme, this evolutionary diminish. In species with very small ranges strategy might result in a total habitat shift. (category 6), genetic pressures normally These types of ecological shifts are dis- inhibiting rapid changes might be suffi- cussed by Hamilton (1962) and MacArthur ciently low to permit the spread of adapta- et al. (1966) among others. As discussed tions enabling the declining species to in the varied literature regarding "character coexist with its former competitors. Species displacement" (reviewed by Grant, 1972), that are unable to change fast enough or changes in morphology or behavior or both far enough eventually "lose the race" and are normally associated with such ecologi- decline to extinction. cal shifts as in above . of the of the last ( category 6, ) Because recency major 3) A shift in altitudinal tolerance can climatic shift, numbers and distributions of afford refuge for certain populations other- South American bird species are far from wise doomed to extinction. This process is stable at this time (Fig. 12). Certain local discussed with respect to Diamond's data regions of open vegetation in Venezuela, on New Guinea bird distributions by Mac- for example, presently much reduced from Arthur (1972: 73). As an invading lowland their former area, apparently remain richer species becomes increasingly common, suc- in species diversity than might be ex- cessful populations of the diminishing spe- pected on the basis of their small area. cies situated up a mountain slope might Competition and extinction rates presum- begin adapting to microclimatic differences ably remain high in these areas. Forest associated with the slightly higher altitude. regions that remained intact during Pleisto- If the species grows sufficiently more suc- cene dry periods appear to be zones in cessful at the higher altitudes than its low- which endemic species, which have sur- land competitor, each will have assumed a vived but are declining, now co-occur with geographic refuge within which it cannot a dispersing, pan-Amazonian fauna. The re- be forced toward extinction by the other. sult is an unstably large number of species 462 Bulletin Museum of Comparative Zoology, Vol. 147, No. 10

in these zones compared with the avifauna Parkes (Carnegie Museum), Melvin Tray- in the surrounding forest. Terborgh (1973) lor (Field Museum), Richard Zusi (United discusses the kinetics of these historical States National Museum), and Frank Gill components to local and continental species (Academy of Natural Sciences, Philadel- diversity' with respect to plant communi- phia). I thank John Terborgh and John ties. Weske for information on the behavior of Evidence from the Tody-tyrant distribu- certain Peruvian tyrannids, and for their tions presented here supports the assertion generous loan of a rare specimen in Weske's that avian distributions within Amazonia collection. are not currently in a steady state. Certain The Biology Department at Harvard populations are probably still dispersing, kindly financed my travel to the museums other local endemics appear doomed to ex- in Philadelphia and Pittsburgh. tinction. However, huge time periods are LITERATURE CITED undoubtedly required for dispersing spe- H. E. 1946. Uma nova do cies to force others entirely out of existence. Berla, especie genero Todirostrum Lesson, 1831. Summa Indeed, the continent is not likely to re- Brasil. Biol. 1: 125. main in its condition long enough present Blake, E. 1959. A new species of Todirostrum to settle into a stable for its fauna ever from Peru. Nat. Hist. Misc. (Chicago) No. composition. Rather, a new cycle of iso- 171: 1. M. L. 1974. and the struc- lations and dispersal will again scramble Cody, Competition ture of bird communities. Princeton: Prince- the distributions of its species, and the ton University Press, viii -f- 318 pp. outlined above will be- dynamic processes de Schauensee, R. M. 1966. Species of birds gin on a new array of populations. These of South America and their distribution. Nar- geographic processes appear to have been berth, Pennsylvania: Livingston Publ. Co., xvii 577 the prime generators of the diverse avian + pp. . 1970. A guide to the birds of South composition inhabiting the South American America. Wynnewood, Pennsylvania: Living- lowlands. ston Publ. Co., xiv + 470 pp. Diamond, J. 1972. Avifauna of the eastern ACKNOWLEDGMENTS highlands of New Guinea. Publ. Nutt. Ornith. Club No. 12., 438 pp. This is drawn from senior paper my Eisenmann, E. and W. H. Phelps. 1971. Una honors submitted to the thesis, Biology nueva subespecie de Todirostrum maculatum Department of Harvard University. Prior del delta del Orinoco. Bol. Soc. Venez. Cienc. to and during its preparation I was gener- Natur. 29: 186-194. Gilliard, E. T. 1959. Notes on some birds of ously advised and assisted by Raymond A. northern Venezuela. Amer. Mus. Novit. No. to whom 1 remain in- Paynter, Jr., deeply 1927: 1-33. debted. I also thank Ernst W. Mayr, John Grant, P. R. 1972. Convergent and divergent A. Smith, John Terborgh, Melvin Traylor, character displacement. Biol. J. Linn. Soc. 4: and David Woodruff for their helpful dis- 39-68. H. and R. Moore. cussions and comments on earlier drafts of Griscom, L., Friedmann, 1957. Distributional checklist of the birds of this I particularly appreciate the paper. Mexico. Pac. Coast Avif., No. 29: 1-435. constructive offered the suggestions by Haffer, J. 1967. Zoogeographical notes on the anonymous reviewer of an earlier manu- 'nonforest' lowland bird faunas of northwest- ern South America. Hornero, 10: 315-333. script. . 1969. Speciation in Amazonian fore;:t I am to each of the grateful following birds. Science 165: 131-138. who made available all Euscarth- curators, ~. 1974. Avian speciation in tropical mine specimens under their care and kindly South America. Publ. Nutt. Ornith. Club, No. 12. viii 390 loaned large fractions of them for my use + pp. Hamilton, T. H. 1962. Species relationships in Cambridge: Wesley Lanyon (American and adaptations for sympatry in the avian Museum of Natural History), Kenneth genus Vireo. Condor 64: 40-68. Biogeography of Todirostrum • Fitzpatrick 463

Haverschmidt, F. 1955. Notes on the life Skutch, A. 1930. The habits and nesting activi- history of Todirostrum maculatum in Surinam. ties of the Northern Tody-flycatcher in Auk 72: 325-331. Panama (Todirostrum cinereum). Auk 47:

. 1968. The birds of Surinam. Edin- 313-322.

burgh: Oliver and Boyd. . 1972. Studies of tropical American Hellmayr, C. 1910. The birds of the Rio birds. Publ. Nutt. Ornith. Club No. 10, 228 Madeira. Novit. Zool. 17: 257-428. pp. P. . 1926. Catalogue of birds of the Ameri- Slud, 1960. The birds of Finca "La Selva" cas, vol. 5. Field Mus. Nat. Hist. Zool. Ser. Costa Rica, a tropical wet forest locality. Bull. No. 242: 1-517. Amer. Mus. Nat. Hist. 121: 49-148.

Keast, A. 1972. Ecological opportunities and . 1964. Birds of Costa Rica, distribution dominant families as illustrated by the neo- and ecology. Bull. Amer. Mus. Nat. Hist. tropical Tvrannidae (Aves). Evol. Biol. 5: 128: 1-430. - 229-277. Smith, W. J. 1970. Tyrannidae suggestion for MacArthur, R. H. 1972. Geographical Ecol- linear ordering. Unpublished manuscript. 15 ogy: patterns in the distribution of species. pp. New York: Harper and Row, xviii + 269 pp. axd F. Vuilleumier. 1971. Evolution- and E. O. Wilsox. 1967. The theory ary relationships in some South American of island biogeography. Princeton: Princeton ground tyrants. Bull. Mus. Comp. Zool. 141 :

University Press, xi -+- 203 pp. 181-232." L. 1966. , H. Recher and M. Cody. Sxethlage, E. 1925. Neue Vogelarten aus On the relation between habitat selection and Nord-Brasilen. Journ. f. Ornith. 73: 264-274. species diversity. Amer. Nat. 100: 319-332. Sxyder, D. E. 1966. The birds of Guyana. Mayr, E. axd W. H. Phelps. 1967. The origin Salem: Peabody Museum. of the bird fauna of the south Venezuelan Terborgh, J. 1971. Distribution on environ- highlands. Bull. Amer. Mus. Nat. Hist. 136: mental gradients: theory and a preliminary 273-327. interpretation of distributional patterns in the and L. L. Short. 1970. Species taxa of avifauna of the Cordillera Vilcabamba, Peru. North American birds. Publ. Nutt. Ornith. Ecology 53: 23-40.

Club, No. 9, 127 pp. . 1973. On the notion of favorableness in Miller, A. H. 1947. The tropical avifauna of plant ecology. Amer. Nat. 107: 481-501. Colombia. Auk the upper Magdalena Valley, axd J. S. Weske. 1969. Colonization of 64: 351-381. habitats Peruvian birds. — secondary by Ecology . 1963. Seasonal activity and ecology of 50: 765-782. of the avifauna an American equatorial cloud axd . 1975. The role of competi- forest. Univ. Cal. Publ. Zool. 66: 1-78. tion in the distribution of Andean birds. Novaes, F. C. 1970. Analise ecologica de uma Ecology 56: 562-576. avifauna da regiao do Rio Acara, estado do Vaxzolixi, P. axd E. E. Williams. 1970. The Para. Bol. Mus. Par. Emilio Goeldi 69: 1-52. geographic differentiation and evolution of Olivares, A. 1955. Algunas aves de la Comi- the Anolis chrysolepis species group. Arq.

. : 19: 1-298. saria del Vaupes ( Colombia ) Caldasia 7 Zool. Sao Paulo 259-275. Wetmore, A. 1926. Observations on the birds Ricklefs, R. axd G. W. Cox. 1972. The taxon of Argentina, Paraguay, Uruguay, and Chile. cvcle in the land bird fauna of the West Bull. U. S. Natl. Mus. 133: 1-448. Indies. Amer. Nat. 106: 195-219. Wilsox, E. O. 1961. The nature of the taxon Ridcway, R. 1907. Birds of North and Middle cycle in the Melanesian ant fauna. Amer. Nat. America, vol. 4. Bull. U. S. Natl. Mus. 50: 95: 169-193. 1-973. Zimmer, J. T. 1940. Studies of Peruvian birds, Short, L. L. 1975. A zoogeographic analysis of No. 34: the genera Todirostrum, Euscartli- the South American Chaco avifauna. Bull. mornis, Snetldagea, Poecilotriccus, Lopho- Amer. Mus. Nat. Hist. 154: 16.3-352. triccus, Myiornis, Pseudotriccus, and Hemi- Sick, H. 1965. A fauna do cerrado. Arq. Zool. triccus. Amer. Mus. Novit. No. 1066: 1-23.

Sao Paulo 12: 71-93. . 1953. Notes on Tyrant Flycatchers.

. 1966. As aves do cerrado como fauna Amer. Mus. Novit. No. 1605: 1-16. Brasil. 38: arboricola. Anais Acad. Cienc . 1955. Preliminary draft: Checklist of 355-363. species and subspecies in the Tyrannidae, -list of Birds . 1968. As aves do Rio de Janeiro. Sep. for J. Bond (ed. ). Peters' Check Arch. Mus. Nac. 53: 99-160. of the World vol. 8.