HBW 8 - Family text: Dendrocolaptidae (Woodcreepers)

Family text: 

Class AVES
Suborder TYRANNI

  • Small to medium-sized birds with broad rounded wings, long tail with rigid shafts, outer and middle toes united and much longer than inner toe, front claws strongly curved; most with brown plumage, a few grey to blackish, all with rufescent wings and tail.
  • 13-36 cm.
  • Neotropical Region.
  • Forest and woodland, a few in semi-open habitats.
  • 13 genera, 52 species, 270 taxa.
  • 1 species threatened; none extinct since 1600.

The Dendrocolaptidae, the woodcreepers, are part of an endemic Neotropical group of birds known as the tracheophone suboscines. These primitive passerines possess an elaborate tracheal syrinx, of similar form in each one, and have evolved a diverse array of ecological specializations. Most woodcreepers are specialized tree-climbers, extremely well adapted to an arboreal life. Not surprisingly, their centre of diversity lies in the rich, forested lowlands of the Amazon Basin, where up to 19 species occur together. The Dendrocolaptidae were formerly known as "woodhewers", a name derived from early accounts that referred to these birds as "picicules", meaning "little woodpeckers (Picidae)". This name, however, was abandoned in favour of one that more appropriately reflects their habits, which seldom include pecking or hammering in the manner of true woodpeckers.

As with most passerines, the fossil record of woodcreepers is scant, providing no insight into the age of their lineage. Only a few Quaternary fossils are known, all from two sites in Brazil. One of these is the Lapa da Escrivania, near Lagoa Santa, in Minas Gerais, where one fossil from the Pleistocene, of the White-throated Woodcreeper (Xiphocolaptes albicollis), and two others, one Pleistocene and the other Holocene, of the Narrow-billed Woodcreeper (Lepidocolaptes angustirostris) have been discovered. According to H. Alvarenga, specimens of Xiphocolaptes species have been found at another site in Brazil. The recent assertion, by J. Cheneval, that a middle Miocene fossil, Homalopus picoides, first discovered in the mid-nineteenth century at Sansan, in France, represents a dendrocolaptid must be viewed with caution. There is no evidence to suggest that tracheophone suboscines ever existed outside the New World. Any attempt to establish confidently that they did would require more than the extremely fragmentary material available, which includes only the distal end of a left tarsometatarsus and the proximal end of a right humerus. Although the fossil does show certain similarities to tree-creeping birds, S. L. Olson suggested that Homalopus might have been an oscine that was only convergently similar to suboscines. Molecular data clearly indicate that the woodcreepers diverged relatively recently from the ovenbirds (Furnariidae), with the split probably occurring during the Miocene.

Evidence from both morphology and molecular genetics strongly supports a monophyletic assemblage including both woodcreepers and ovenbirds, a much larger group of birds that has radiated into a highly diverse array of niches and forms. This close relationship has resulted in the taxonomic rank of these two groups being unsettled, with many treatments combining them as one family, either under the name Dendrocolaptidae, as in older classifications or, as in more recent ones, under the name Furnariidae; in either case, the woodcreepers have consequently been ranked as a subfamily, the Dendrocolaptinae. Only these two avian families, the Dendrocolaptidae and the Furnariidae, possess two intrinsic muscles of the syrinx, the vocalis dorsalis being absent in other tracheophone suboscines. In addition, analyses of both DNA-DNA hybridization data and sequences of nuclear and mitochondrial DNA support the view that the woodcreepers and the ovenbirds form a closely related group that is distinct from other birds. Nevertheless, dendrocolaptids differ from furnariids in a number of features. Characters traditionally used to diagnose woodcreepers include horns on the processi vocales of the syrinx, these horns being present in no furnariids other than the miners (Geositta); nostrils that are rounded at the rear, a condition known as holorhinal; and outer and middle toes that are united at the base, of similar length, and noticeably longer than the inner toe. All woodcreepers share these features to a degree, but not all ovenbirds can be excluded by them. Notably, several "intermediate" woodcreeper genera, namely Dendrocincla, Deconychura, Sittasomus and Glyphorynchus, exhibit a mixture of morphological characters, having some features of "typical" dendrocolaptids and others that are more typical of furnariids, and particularly of the foliage-gleaners, the six genera of which (Anabacerthia, Syndactyla, Philydor, Anabazenops, Automolus, Hylocryptus) are sometimes considered to constitute the subfamily Philydorinae. Some authors, therefore, cite the aforementioned four dendrocolaptid genera as demonstrating a clear link between the two families.

A diverse array of evidence indicates that all extant woodcreeper genera are more closely related to each other than they are to any furnariid. All dendrocolaptids share a particular configuration of their feather tracts (pterylosis) that distinguishes them from all other passerines. Further, they are unique among the passerines in having extensive ossification of the tendons of the hind limb, an adaptation to their tail-braced, scansorial habits that is reduced or more variable only in the genus Dendrocincla, the members of which less frequently adopt scansorial postures. Platyacarus feather mites are known to parasitize only woodcreepers, with documented hosts in every genus except the monotypic Drymornis; the latter, containing the Scimitar-billed Woodcreeper (Drymornis bridgesii), retains many morphological characters, especially those of the limb muscles, that are primitive among woodcreepers (see Morphology). DNA-DNA hybridization data, coupled with phylogenetic analyses of various morphological and molecular characters, consistently yield results suggesting that all dendrocolaptid genera are monophyletic to the exclusion of furnariids. In contrast, the Furnariidae appear not to be monophyletic with respect to woodcreepers.

The issue of whether to rank woodcreepers as a separate family has been questioned recently by M. Irestedt and colleagues, whose estimated phylogeny based on both nuclear and mitochondrial DNA suggested that the furnariid leaftossers (Sclerurus) are basal both to woodcreepers and to all other ovenbirds. Because of this arrangement, those authors recommended that the Furnariidae include three subfamilies: woodcreepers (Dendrocolaptinae), leaftossers (Sclerurinae), and all other ovenbirds (Furnariinae). An alternative classification, not mentioned by Irestedt and co-workers, would be to accord family rank to each of these groups. The placing of the woodcreepers in a family of their own seems the best course, because woodcreepers form a clearly diagnosable yet diverse group of species that share distinctive morphological and biochemical traits. Moreover, incorporation of them within the ovenbird family would also create too inclusive a group, comprising nearly 300 species, that obscures important evolutionary differences among its members.

Some scientists have speculated that it was in the vast Amazon Basin that the woodcreepers evolved from their supposed nearest living relatives, the foliage-gleaners. In his monograph on the relationships of furnariids and dendrocolaptids, A. Feduccia postulated that, on the basis of similarities in anatomy, behaviour and geographical distribution, ancestral woodcreepers were foliage-gleaners that were capable of tree-climbing, an adaptation that may have greatly reduced competition for feeding space in mixed-species foraging flocks. According to this hypothesis, tree-climbing foliage-gleaners then experienced an extensive radiation in this new foraging space, leading eventually to modern woodcreepers.

Shortly after Feduccia's monograph, E. O. Willis, whose extensive observations of woodcreeper behaviour and ecology are perhaps unparalleled, suggested an alternative view of dendrocolaptid evolution. He noted that woodcreepers of the genus Dendrocincla and leaftossers of the ovenbird genus Sclerurus were strikingly similar in appearance and behaviour. On this basis, he proposed that tree-climbing woodcreepers might have evolved from terrestrial leaftosser ancestors lured upwards to exploit food on bare tree trunks in forests, and perhaps driven there through competition with the evolving thamnophilid antbirds that now dominate ant-following flocks in the understorey. Under this scenario, the genus Dendrocincla represents a primitive woodcreeper that has advanced little. This apparent link with leaftossers was not supported by anatomical analyses of woodcreeper-ovenbird relationships, but evidence of it has emerged recently in analyses of mitochondrial and nuclear DNA sequences. Using these biochemical characters, Irestedt and colleagues have proposed a phylogeny for the tracheophone suboscines. Their study included, among several other taxa, a single leaftosser, the Rufous-breasted Leaftosser (Sclerurus scansor), as well as five species of dendrocolaptid, among them the Plain-brown Woodcreeper (Dendrocincla fuliginosa). Those authors found that the leaftosser emerged as ancestral to all woodcreepers as well as to all other ovenbirds. In spite of sampling limitations, this supports the view that both families may have evolved from a woodcreeper-ovenbird ancestor that bore some similarities to extant species of leaftosser.

On the basis of anatomical and molecular characters, there are two major lineages of woodcreepers. One of these comprises species in the four genera that are "intermediate" between the Dendrocolaptidae and the Furnariidae: Dendrocincla, Deconychura, Sittasomus and Glyphorynchus. The other lineage, known as the "strong-billed" group, comprises all other dendrocolaptid species. The several characters possessed by the "intermediate" taxa that are thought to be ancestral and are shared with the ovenbirds include less stiffened shafts of the tail feathers and reduced ossification of the leg tendons in Dendrocincla, the presence of wingstripes in Sittasomus and Glyphorynchus, and particular features of the skull. The "strong-billed" woodcreeper genera exhibit several derived characters, many associated with their ubiquitous creeping behaviour and trunk-foraging lifestyle, such as the "closed" two-notched sternum, a heavily ossified skull, extensive ossification of the leg tendons, and a well-developed pygostyle.

Whether the "intermediate" lineage is the more basal within the Dendrocolaptidae remains controversial. R. Raikow's estimate of woodcreeper phylogeny, based on characters derived from hind-limb and bill morphology, found that two monotypic genera, Nasica and Drymornis, emerged as basal to all woodcreepers; Drymornis was the more primitive of the two, as determined by its retention of ancestral character states, as opposed to its specialized bill shape and ground-foraging behaviour. By contrast, two recent phylogenetic studies employing both nuclear and mitochondrial DNA sequences supported the placement of Nasica and Drymornis in the "strong-billed" assemblage. Neither study included both genera in the same analysis with other woodcreepers, one having sampled Drymornis and the other only Nasica. A possible explanation for the conclusions of the morphological study is that the wide variability in substrates used by tree-climbing birds has led to a great plasticity in anatomical characters associated with feeding and climbing, rendering characters associated with these traits less useful for phylogenetic analysis.

As noted above, woodcreepers in the genus Dendrocincla may be the most primitive within the family, a hypothesis reflected in most linear classifications. All six species in this apparently monophyletic genus have rectrices with a somewhat less rigid shaft and a reduced spiny tip when compared with other dendrocolaptid genera, which exhibit typically rigid shafts with prominent bare, spiny tips that are often curved ventrally. Members of this genus do not depend so much on trunk-foraging, instead catching most of their food by sallying like a tyrant-flycatcher (Tyrannidae); they use vertical trunks mostly for perching, and perch on horizontal branches more often than do most woodcreepers. Within the genus Dendrocincla, species-level taxonomy is in need of study. Widespread species of Dendrocincla, such as the Plain-brown and the White-chinned Woodcreepers (Dendrocincla merula), have several well-differentiated subspecies, some of which may merit recognition as full species when more comprehensive reviews of geographical variation in genetic and vocal characters become available. Within the White-chinned Woodcreeper, differences in vocal characters, size and iris colour suggest that at least two species may be involved. In view of the vocal and plumage differences documented by Willis, it seems better to treat the Plain-winged Woodcreeper (Dendrocincla turdina) of south-eastern Brazil as being distinct from the Plain-brown Woodcreeper, with which it was formerly considered by most authors to be conspecific. Leaving aside turdina, the Plain-brown Woodcreeper has at least three other distinctive groups of subspecies, any or all of which may ultimately be found to be deserving of separate species status.

Willis also noted that the plumage and behaviour of the two Deconychura species suggest that the genus is intermediate between Dendrocincla and Glyphorynchus, with the Spot-throated Woodcreeper (Deconychura stictolaema) being closer to the latter genus. In partial agreement with this suggestion, Raikow's phylogenetic hypothesis suggested that the two species of Deconychura are not each other's closest relatives, his findings placing them, along with the Olivaceous Woodcreeper (Sittasomus griseicapillus), in clades intermediate between Dendrocincla and Glyphorynchus. Like the Olivaceous Woodcreeper and the Wedge-billed Woodcreeper (Glyphorynchus spirurus), the two Deconychura species were placed each in a separate clade not closely related to any extant dendrocolaptid. Relationships among the ten taxa currently divided between the two species of Deconychura remain uncertain. For example, within the Long-tailed Woodcreeper (Deconychura longicauda), birds from Central America south to north-central Colombia, forming the "typica group" of subspecies, may be more closely related to the Spot-throated Woodcreeper. Furthermore, there is marked vocal variation within the remaining four, rather morphologically uniform, taxa of the Long-tailed Woodcreeper occurring in Amazonia.

Among the remaining "intermediate" genera, the taxonomy of the Olivaceous Woodcreeper is perhaps the one most in need of a comprehensive, systematic revision of all included taxa. The species can be divided into five major groups according to size and gross plumage coloration. This variation, combined with concordant vocal differences among some groups, is sufficiently marked to suggest that full biological species are involved, this being especially so in the case of three of the groups: each of the two groups occupying eastern Brazil, namely reiseri in the north-east and the "sylviellus group" in eastern and south-eastern Brazil, and the disjunct form aequatorialis on the Pacific coast. Of the remaining subspecies of the Olivaceous Woodcreeper, some that appear similar to each other have distinct vocalizations, but it is uncertain how this variation should be interpreted. Lastly, the small yet distinctive Wedge-billed Woodcreeper is known to possess at least two different song types, although gene flow exists between some morphologically distinct subspecies.

In contrast to the uncertain relationships among the "intermediate" woodcreepers, monophyly of the "strong-billed" clade, exclusive of Nasica and Drymornis (see above), has been corroborated by a number of investigations. A study of mitochondrial DNA (mtDNA) sequences supported the division of the "strong-billed" woodcreepers into two clades, one containing the genera Dendrexetastes, Hylexetastes, Xiphocolaptes and Dendrocolaptes, and the other consisting of Campylorhamphus, Lepidocolaptes and Xiphorhynchus.

Molecular and anatomical characters are strongly indicative of a sister relationship between the genera Hylexetastes and Xiphocolaptes. The first of those contains only two species, both confined to the Amazon Basin, while the four species of Xiphocolaptes are distributed in a variety of forested habitats from Mexico south to central Argentina. Species-level taxonomy within these genera merits further study. In the genus Hylexetastes, two species, the Red-billed Woodcreeper (Hylexetastes perrotii) and the Bar-bellied Woodcreeper (Hylexetastes stresemanni), have traditionally been recognized, although some authors have elevated the subspecies uniformis of the former to a full species, the "Uniform Woodcreeper". In 1995, the taxon brigidai was, on the basis of plumage, described as a new species, "Brigida's Woodcreeper", but its appearance is close enough to that of uniformis that hybrids would be difficult to detect. Vocal similarity among these and other Hylexetastes taxa, coupled with limited morphometric differences, supports treatment of all as subspecies of the Red-billed Woodcreeper. Plumage patterns in Hylexetastes are comparable to those in the sympatric populations of the Amazonian Barred Woodcreeper (Dendrocolaptes certhia), which appear to constitute a single species.

The four Xiphocolaptes have a significantly longer bill than that of the two Hylexetastes, to which they are otherwise nearly identical morphologically. This morphological similarity is reflected in their placement as sister taxa in a recent phylogeny. Unlike species in the genus Hylexetastes, those in the genus Xiphocolaptes differ notably in plumage and structure. Morphology within the genus is constant, all members being large and heavy-bodied, with the tail relatively short, and the bill long, massive and laterally compressed. Xiphocolaptes formed a monophyletic clade in a phylogenetic analysis that sampled three of the four species. Three species are largely restricted to discrete biogeographical regions, within which geographical variation is subtle. The White-throated Woodcreeper has three subspecies in the Atlantic and Planalto forests of eastern and south-eastern Brazil and adjacent portions of Argentina and Paraguay. The Great Rufous Woodcreeper (Xiphocolaptes major), a Chaco species, has four subspecies in Bolivia, south-west Brazil, Paraguay and northern Argentina. The rare Moustached Woodcreeper (Xiphocolaptes falcirostris) is distributed spottily in deciduous forests of the Caatinga region of north-eastern Brazil. Unlike its three congeners, the Strong-billed Woodcreeper (Xiphocolaptes promeropirhynchus) occurs over a vast range and exhibits tremendous geographical variation in plumage and habitat preference. There are currently 25 recognized subspecies, divided into three groups: the "emigrans group" in Central America, the "promeropirhynchus group" living in the Andes of northern South America, and the "orenocensis group" of the Amazonian lowlands and the foothills of the tepuis. C. Cory and C. Hellmayr hypothesized that the "orenocensis group" is a separate species, the "Great-billed Woodcreeper", and others have raised the possibility that all three groups may represent distinct species. Some taxa in this genus have had a tortuous taxonomic history. Recent analysis of plumage characters revealed that the taxon franciscanus, formerly regarded either as a separate species, "Snethlage's Woodcreeper", or as a subspecies of the White-throated Woodcreeper, is instead best treated as the southernmost population of the globally threatened Moustached Woodcreeper. Conversely, a poorly known race of the White-throated Woodcreeper, villanovae from north-eastern Brazil, was formerly considered a subspecies of the Moustached Woodcreeper.

The genus Dendrocolaptes likewise forms a discrete morphological group, but its affinities to other woodcreepers remain uncertain. Some analyses suggest a close relationship between Dendrocolaptes and Hylexetastes, but more recent analyses of anatomical characters favour a sister relationship with Xiphocolaptes and Hylexetastes combined. Analyses of molecular data suggest that Dendrocolaptes occupies a basal position relative to the genera Dendrexetastes and Nasica. Species limits within and between currently recognized species in this genus also remain unsettled. The two barred woodcreepers, with joint populations extending from southern Mexico southwards to the southernmost reaches of Amazonia, were formerly united in a single species, Dendrocolaptes certhia, but differences in morphology and voice between populations occurring north and west of the Andes and those in Amazonia are sufficiently marked that two species are now recognized: the Northern Barred Woodcreeper (Dendrocolaptes sanctithomae) and the Amazonian Barred Woodcreeper, respectively. By contrast, variation in morphology and voice suggests that the form concolor, previously regarded as a separate species under the name "Concolor Woodcreeper", is best treated as a subspecies of the Amazonian Barred Woodcreeper. Species limits within what apparently constitutes a superspecies of streaked birds, the Black-banded (Dendrocolaptes picumnus), Planalto (Dendrocolaptes platyrostris) and Hoffmann's Woodcreepers (Dendrocolaptes hoffmannsi), are uncertain. Morphologically, the little-known Hoffmann's Woodcreeper is unique within the genus, having a long, slim bill and unusual plumage patterns. Although Raikow grouped Hoffmann's with the two barred woodcreepers, its voice, foraging behaviour and biogeography support Willis's assertion that it is more closely related to the Black-banded and Planalto Woodcreepers.

Taxonomic confusion in the Dendrocolaptidae is perhaps best exemplified by the largest genus in the family, Xiphorhynchus. This has a broad distribution, and its 15 members exhibit a diverse range of body sizes and ecological adaptations. A. Aleixo's recent phylogenetic hypothesis, based on variation in mtDNA, helped to clarify many relationships in this vexing genus while at the same time highlighting further challenges. Despite the morphological diversity in the genus, molecular evidence implies that, with the exception of the Straight-billed Woodcreeper (Xiphorhynchus picus) and Zimmer's Woodcreeper (Xiphorhynchus kienerii), all species of Xiphorhynchus share a common ancestor. These results further combine with patterns of gross morphology and aspects of ecology to support the treatment of most species in one of four groups, but the affinities of two additional species remain unclear. The first group comprises an Amazonian radiation of small dendrocolaptids that characteristically follow mixed-species flocks through the understorey. The second group consists of four ecological generalists occurring from Mexico south to southern Amazonia and eastwards to the Atlantic Forest of Brazil; these species are larger than those in the first group, are less tied to the forest understorey, and forage alone or by following either mixed flocks or army ants. The third group contains two largely montane species, the Spotted Woodcreeper (Xiphorhynchus erythropygius) and the Olive-backed Woodcreeper (Xiphorhynchus triangularis), each of medium size and atypical plumage, that may be foraging specialists on mossy trunks.

The Straight-billed and Zimmer's Woodcreepers, representing the fourth group, are sibling species that are sufficiently distinct from other Xiphorhynchus to have been treated for many years in the genus Dendroplex. The recent genetic analysis suggested that these two are more closely related to the genera Campylorhamphus and Lepidocolaptes. Notwithstanding this, the return of the Straight-billed and Zimmer's Woodcreepers to their own genus is not straightforward, because an unfortunate nomenclatural problem complicates the use of the name Dendroplex. Moreover, the bill of Zimmer's Woodcreeper does not fit the diagnosis for Dendroplex, and this species was consequently placed in Xiphorhynchus even by authors who still classified the Straight-billed Woodcreeper in Dendroplex. Further complicating matters, a recent examination revealed that the holotype of Dendrornis [= Xiphorhynchus] kienerii, described in 1856 and subsequently considered a subspecies of the Straight-billed Woodcreeper, matches Dendroplex [= Xiphorhynchus] necopinus, described in 1934 and known as Zimmer's Woodcreeper. Dendroplex necopinus is, therefore, a junior synonym of Dendrornis kienerii, and the current scientific name of Zimmer's Woodcreeper becomes Xiphorhynchus kienerii. Until these various problems are resolved, it is deemed best to retain the Straight-billed and Zimmer's Woodcreepers in Xiphorhynchus.

Affinities of two other species in this genus are uncertain. The Lesser Woodcreeper (Xiphorhynchus fuscus), occurring in the Atlantic and Planalto forests of eastern Brazil and adjacent areas in Paraguay and Argentina, was for many years classified in the genus Lepidocolaptes. In 1983, Willis suggested that, because of its behaviour, it belonged in Xiphorhynchus; more recent analyses of anatomical and molecular characters place its closest relatives as members of the Amazonian radiation that combines the X. ocellatus and X. spixii species complexes. The Striped Woodcreeper (Xiphorhynchus obsoletus) is similar morphologically to species in the radiation of small Amazonian woodcreepers, but genetic analyses have not resolved its affinities.

Interpreted relationships within each of the four groups of Xiphorhynchus woodcreepers are in a state of flux. Although mtDNA data gave credence to the monophyly of the Amazonian radiation in the first group, consisting of the X. ocellatus and X. spixii superspecies groups, the basal taxon is the Lesser Woodcreeper of the Atlantic Forest. Within the radiation, mtDNA data supported a return to the treatment of the first of those superspecies as comprising three species: the Chestnut-rumped (Xiphorhynchus pardalotus), Ocellated (Xiphorhynchus ocellatus) and Tschudi's Woodcreepers (Xiphorhynchus chunchotambo). The molecular data likewise supported the results of an independent analysis of plumage by J. Haffer, which indicated that the X. spixii/elegans species complex should be recognized as two species, the monotypic Spix's Woodcreeper (Xiphorhynchus spixii) of eastern Amazonia and the polytypic Elegant Woodcreeper (Xiphorhynchus elegans) of southern and western Amazonia. The latter now includes not only the nominate race and ornatus, traditionally recognized within X. elegans, but also races juruanus, insignis and buenavistae, all of which have traditionally been placed within X. spixii.

Even more complicated are the taxonomic issues associated with the second group of widespread generalists, and especially the complex now combining the Buff-throated Woodcreeper (Xiphorhynchus guttatus) and the Cocoa Woodcreeper (Xiphorhynchus susurrans). The taxa contained in this complex have been treated variously as constituting either one widespread species or as many as three or four different species. The currently recognized division is unsatisfactory and requires further study. Within the Cocoa Woodcreeper, for example, geographical patterns of song types are more complex than was previously realized, and differences in size do not appear to correspond to current species limits. Taxa within the "susurrans group", occurring in Trinidad, Tobago and the adjacent Venezuelan mainland, are significantly larger than those within the "nanus group" ("Lawrence's Woodcreeper"), which range west from central Venezuela into Central America. Moreover, plumage patterns of the former group are so different that authors who merged taxa in the "nanus group" with the remaining South American populations of the Buff-throated Woodcreeper still recognized the Cocoa Woodcreeper proper as being distinct. Molecular analyses suggest that members of the "nanus group" are closely related to races of the Buff-throated Woodcreeper occurring in eastern Brazil and in the north-eastern part of Amazonia, these being, respectively, the nominate race and the subspecies polystictus and connectens. Because members of the "susurrans group" were not examined in this study, however, relationships between it and the "nanus group" remain uncertain.

Taxonomy is in a state of flux even within the Buff-throated Woodcreeper as presently recognized. Molecular data support the merger of the form eytoni, the "Dusky-billed Woodcreeper", with the western Amazonian populations of the Buff-throated Woodcreeper, a move complemented by variation in plumage coloration but contradicted by marked differences in both bill coloration and vocalizations. These data also indicate a close relationship between the nominate race of the Buff-throated Woodcreeper, from coastal eastern Brazil, and the disjunct forms polystictus and connectens occurring north of the Amazon in Brazil, the Guianas and the Orinoco region of southern Venezuela, all to the exclusion of geographically intermediate Amazonian populations (both eytoni and the "guttatoides subspecies group"). Vocally, the nominate form is relatively similar to populations previously recognized as the Dusky-billed Woodcreeper, with polystictus and connectens similar but of uncertain affinities. Songs of the western Amazonian forms, by contrast, differ strikingly. It is likely, therefore, that the complex of taxa now recognized as the Buff-throated Woodcreeper includes more than one allospecies, but interpretation of variation in voice, morphology and DNA in this widespread dendrocolaptid is in need of more comprehensive sampling.

The other two species in this group of generalists, the Ivory-billed Woodcreeper (Xiphorhynchus flavigaster) and the Black-striped Woodcreeper (Xiphorhynchus lachrymosus), both occur largely or exclusively in Central America. Molecular data suggest that they are each other's closest relatives; together, they are the sister taxon to the Buff-throated Woodcreeper complex.

The genus Lepidocolaptes is a well-defined group of relatively small woodcreepers, all of which have a slender and decurved bill. With the Lesser Woodcreeper now placed in Xiphorhynchus, as mentioned above, anatomical and molecular data indicate that Lepidocolaptes is monophyletic. Although many earlier authors considered South American populations of the Montane Woodcreeper (Lepidocolaptes lacrymiger) to be conspecific with the Spot-crowned Woodcreeper (Lepidocolaptes affinis), these two taxa differ in voice and plumage. In eastern Brazil, the Scaled (Lepidocolaptes squamatus) and Scalloped Woodcreepers (Lepidocolaptes falcinellus) are now regarded as separate species. They were formerly combined as northern and southern populations, respectively, of a single species, but they exhibit diagnosable differences in plumage, notwithstanding a degree of intergradation where the taxa meet abruptly in south-eastern Brazil. Two of the most widespread species, the Lineated Woodcreeper (Lepidocolaptes albolineatus) and the Narrow-billed Woodcreeper, display substantial geographical variation in plumage, and each has many described subspecies. On the basis of plumage differences, Hellmayr hypothesized that populations of the Lineated Woodcreeper from southern, central and western Amazonia, the "fuscicapillus group", represent a distinct species, referred to as the "Dusky-capped Woodcreeper". The songs of the Lineated Woodcreeper also show marked geographical variation. The remaining two of the eight species in this genus are the White-striped Woodcreeper (Lepidocolaptes leucogaster), a Mexican endemic, and the Streak-headed Woodcreeper (Lepidocolaptes souleyetii), which is widespread in Central America and northern South America. In these two, plumage variation is far more subtle, and the latter, despite its broad distribution, exhibits little variation in vocalizations.

Genetics and anatomical features support monophyly of the final genus of the Dendrocolaptidae, Campylorhamphus. On the basis of mtDNA sequences, the five scythebills are most closely related to the genus Lepidocolaptes. Nevertheless, the Greater Scythebill (Campylorhamphus pucherani) not only is the sole member of its genus that has not yet been included in any systematic study, but is also the most divergent both morphologically and vocally. As seems to be the rule among woodcreepers, both of the widespread species, the Red-billed (Campylorhamphus trochilirostris) and the Curve-billed Scythebills (Campylorhamphus procurvoides), exhibit marked geographical variation in plumage, song and habitat preference; each may comprise multiple allospecies. Moreover, habitat preferences and vocalizations imply that some Amazonian populations of the Curve-billed Scythebill may be better treated instead as subspecies of the morphologically similar Red-billed.

Because woodcreeper plumage is cryptically patterned, and because dendrocolaptid anatomy is constrained by foraging behaviour, morphological diagnosis of many species is difficult. There are few avian groups for which the careful analysis of genetic and vocal data is more likely to improve our understanding of systematics and evolution. Future studies must strive to document variation not among a select sample of named taxa but, instead, among populations separated by either vast distances or key barriers to dispersal. Only by using sampling schemes sufficiently thorough to document genetic and vocal variation within populations, and to enable the determination, with a fair degree of certainty, of geographical barriers that correspond to these differences, can one make taxonomic recommendations confidently. Since voice is an indirect measure of reproductive isolation, the analysis of vocal data requires a caution similar to, if not greater than, that demanded in the interpretation of genetic data. Before making taxonomic recommendations based on any type of variation, it is necessary to outline geographical patterns in the characters, to develop a benchmark of variation within populations for later comparison among populations, and to determine, with a reasonable degree of certainty, that the characters being studied are, indeed, important in the process of reproductive isolation. It is essential to bear in mind that, although both genetic and vocal data may be important for understanding species limits, these data are subject to the same pitfalls as are morphological data, our understanding of which has had a century longer to develop.

Morphological Aspects
Although the Dendrocolaptidae are variable in body size, with a total length ranging from 13 cm to 36 cm, most members of the family share a rather uniform body form and plumage coloration. Bill length and shape account for much of the variation. The bill represents more than a quarter of the body length in the Long-billed Woodcreeper (Nasica longirostris), for example, while its extreme curvature in the case of the scythebills results in the distinctive placement of the eyes of those species. The smallest dendrocolaptid is the widespread Wedge-billed Woodcreeper, which, at 13 cm long and weighing 15 g, is not much larger than a Certhia creeper (Certhiidae). The heaviest is the Strong-billed Woodcreeper, up to 169 g in weight and to 35 cm in length, which is a little bigger than a Colaptes flicker of the Picidae family. The majority of woodcreepers are roughly the size of a small thrasher (Mimidae).

Males tend to be larger than females, but broad overlap in both measurements and body mass is typical of most dendrocolaptid species. Pronounced sexual dimorphism in size is exhibited only by the Long-tailed and Spot-throated Woodcreepers, the males of which are considerably larger and heavier than the females. Differences in plumage among woodcreepers are slight, both between young birds and adults and between the sexes. In a few species, especially some Dendrocolaptes, males are recognizable because they raise their crown feathers frequently, producing a somewhat shaggy crest that creates a noticeably different appearance from that of the sleek-headed females. Sexual dimorphism in plumage is otherwise not apparent.

Juveniles differ subtly from adults, but young birds already resemble adults after their first body moult. Immatures tend to be more barred, have a slightly shorter tail, and have a darker or blacker bill that is significantly shorter than the adult's; the last feature is evident for a relatively longer time than are other characteristics of immaturity.

Woodcreeper plumage does not undergo seasonal change. The only perceivable changes in the feathering are the result of wear and fading, some birds appearing paler late in the breeding season, when many feathers, most notably the remiges and rectrices, are extensively worn.

The predominant coloration of most dendrocolaptids is brown, with the wings and tail contrastingly rufescent. Even such species as the Olivaceous, Olive-backed, Spotted and Black-striped Woodcreepers, which are mostly grey, olive or blackish, exhibit the plain, reddish-brown wings and tail typical of the family. As an overlay on their more subdued coloration, woodcreepers may possess a variegated pattern of streaks, spots and bars on the back, nape, head and underparts. Exceptions to this include the Olivaceous Woodcreeper and most members of the genus Dendrocincla, which are largely unmarked. Most species have creamy, whitish or light brown streaks and tear-shaped spots, but the Elegant, Olive-backed and Spotted Woodcreepers are unique in sharing triangular spots on the back and chest. Barring is usually restricted to the belly, and even then is obvious only on species of Dendrocolaptes, Hylexetastes and Xiphocolaptes. The degree of barring reaches its extreme in the Northern and Amazonian Barred Woodcreepers, each being extensively barred above and below. Two features present only on the Olivaceous and Wedge-billed Woodcreepers are buff axillaries and a conspicuous, buff bar running obliquely across all but the outer remiges. This bar, because it is formed by pale bars on the inner webs of the remiges, is mostly concealed when the wings are folded; it is seen most easily from below on the spread wing, a vantage seldom attainable with living woodcreepers in the field.

Some adaptations designed to resist feather wear or degradation in moist environments seem apparent in the plumage. On many woodcreepers, the tips of the remiges, especially those of the outermost primaries that are exposed on the folded wing, are much darker, containing more melanin than is present in the rest of those feathers. There is a slight tendency towards darker coloration in birds occupying regions of higher relative humidity, in accordance with "Gloger's rule", but the pattern is subtle, as has been discussed for some Dendrocolaptes woodcreepers. Species that occupy open habitats may be dramatically countershaded, dark above and light below, in comparison with those that favour the forest interior. The Narrow-billed Woodcreeper, for instance, occurs in open country and has predominantly white underparts. Other examples of dendrocolaptids with relatively pale plumage are the Scimitar-billed Woodcreeper, the Moustached Woodcreeper, and some subspecies of the Red-billed Scythebill that inhabit predominantly open or semi-open landscapes.

Limited data, for a few species, suggest that woodcreepers undergo a single, annual moult that lasts from four to six months. The wings and the tail are typically moulted simultaneously, although wing moult may start first. As with most bird species, the remiges are replaced in sequence from the innermost primary outwards and from the outermost secondary inwards; the rectrices are moulted outwards from the innermost pair. Moult usually follows immediately after the breeding season, but for at least a small number of dendrocolaptids, such as the Northern Barred and Streak-headed Woodcreepers, moult and breeding may overlap. The Amazonian Barred Woodcreeper's cycle of breeding and moult is believed to extend over a period of nine to ten months, while a female Streak-headed Woodcreeper possessing a brood patch was in the early stages of moult. It is thus possible that some species begin the moult process before their young leave the nest. Overlap of moult and breeding may be an adaptation either for a long moult cycle to be completed before mating and singing resume in the next breeding season or, alternatively, to allow the completion of both during a flush of prey abundance.

The most distinctive aspects of woodcreeper morphology are manifest in their adaptations for tree-climbing, or "hitching". To some extent, these adaptations are shared with other tree-climbing birds, such as the woodpeckers, a few genera of ovenbirds (e.g. Margarornis and Pygarrhichas) and creepers (Certhiidae), but with important differences and modifications. All woodcreepers are able to hitch up vertical trunks by flexing their legs and hopping forwards; the long and spiky tail holds them in place when they stop. Rigid shafts of the tail feathers facilitate these tail-braced movements and postures. Among all tree-climbing birds, strengthening of the tail shafts appears to be best developed in woodcreepers, which tend to have proportionately longer rectrices. Unlike other birds that use the tail as a brace, however, the woodcreepers' denuded, spiny rectrix tips are curved downwards like claws, allowing them to contact the substrate at a right angle. The stiffened rectrices with rigid, spiny tips enable the tail to support most of the bird's weight. Woodcreepers that have lost the tail have great difficulty in climbing.

Several characteristics of the foot and leg are adapted for climbing trees. Woodcreepers have an anisodactyl foot, with three toes pointing forwards and one toe backwards, an arrangement typical of all passerines. Their feet differ from those of most passerines, however, in having the outer and middle toes united for much of their length, and of similar lengths, both being much longer than the inner toe. The hallux is reduced. The claws of the three front toes are strongly curved, aiding in clinging, whereas the claw of the hallux is long and often pressed hard against the substrate, providing support. Strong thighs dominated by strengthened flexor muscles, which have developed at the expense of extensor muscles, provide the power for climbing up trunks. Extensive ossification of the leg tendons appears to be an adaptation for resisting stress on the toes and ankle flexors while clinging to trees. Ossification of the tendons in the legs is reduced, or is more individually variable, only in those woodcreepers that do not rely exclusively on tree-climbing for foraging; these include the Tawny-winged Woodcreeper (Dendrocincla anabatina) and, probably, other species in the genus Dendrocincla. Elaborate tree-climbing adaptations allow dendrocolaptids to exploit several ecological niches, but they also greatly restrict a woodcreeper's ability to use other modes of locomotion or the wide variety of perches available to most passerines.

Most woodcreepers are exclusively arboreal, moving between trees with a strong, slightly undulating flight. As may be expected for birds that often fly only short distances and glide frequently, the Dendrocolaptidae have broad wings with rounded tips. Many species, especially those that forage in association with army-ant swarms (see Food and Feeding), often approach or briefly land on the ground. Members of the genus Xiphocolaptes seem to spend more time on the ground than do the majority of woodcreepers, the Great Rufous Woodcreeper being the most terrestrial. Only the Scimitar-billed Woodcreeper, with its relatively short tail and unusual hind-limb musculature, seems suited to a terrestrial lifestyle. This species runs on the ground with little difficulty.

The bill is by far the most adaptively variable feature of dendrocolaptid morphology. Its size and shape are closely correlated with foraging behaviour and have evolved along four major themes: gleaning, flycatching, probing, and chiselling. The Wedge-billed Woodcreeper is the only member of the family that has the bill of a typical gleaner. The Olivaceous and Spot-throated Woodcreepers glean with a short, pointed bill, but, as they also sally after prey flushed from trunks, the bill is somewhat broadened at the base. The Long-tailed Woodcreeper and species in the genera Dendrocincla and Dendrocolaptes have a relatively short and straight flycatcher-like bill, characterized by a weak hook at the tip of the maxilla. That of Dendrocolaptes, in particular, recalls the bill of tyrant-flycatchers (Tyrannidae), being relatively short, broad-based, flattened dorsoventrally, and hooked at the tip. These species frequently sally out for prey on the ground, trunks or foliage, or in the air, catching large arthropods and even small vertebrates (see Food and Feeding). Bills used for probing are the most variable in length and shape, being narrower and more delicate compared with other types; in addition, they tend to be compressed laterally, instead of dorsoventrally. In the case of some species, the bill is extraordinarily long and either straight, as that of the Long-billed Woodcreeper, or extremely decurved, as with the scythebills. Dendrocolaptids possessing a probing bill explore a variety of cavities in tree and fern trunks, dead-leaf clusters, bamboo stems and epiphytes. Willis noted that species with a straight bill tend to probe from a lateral position, whereas those with a decurved bill probe from a vertical position. Many species with a probing bill also glean insects from bark or occasionally sally out after flushed prey.

Bills used for chiselling, seen on members of the genera Hylexetastes and Xiphocolaptes, are a variation of those used for probing. Such bills are laterally compressed and relatively deep and heavy. Species having this type of bill capture large prey on trunks, or tear apart decaying material accumulated on branch forks and in epiphytes. The bill of Hylexetastes is similar to that of Xiphocolaptes, but is shorter. The two Hylexetastes woodcreepers are more generalist foragers, at times sallying out for prey in the manner of many woodcreepers that have a flycatcher-like bill.

Woodcreepers inhabit chiefly forest or woodland, from tropical rainforest in the lowlands to stunted cloudforest near the timber-line, but a few species occur in savanna or other semi-open habitats. Most are restricted to tropical evergreen forest, although some are plentiful in semi-deciduous woodland and gallery forest in drier regions. A small number occur in pine-oak (Pinus-Quercus) woodland and pine forest in the mountains of Central America, and others frequent montane evergreen forest or cloudforest in the Andes and other mountain ranges. The Polylepis woodlands of the high Andes and the temperate forests of southern South America are perhaps the only forested habitats in the Neotropics that are devoid of woodcreepers. As a result of their sedentary nature and their aversion to crossing large unforested gaps, few dendrocolaptid species inhabit offshore islands.

Most woodcreepers are found in lowland forest, but a few occupy higher elevations. The vast majority of this family's members are restricted to tropical lowlands from sea-level to elevations mostly below 1000 m, with some species occurring sparingly to 1500 m. Above this altitude, the Tyrannine (Dendrocincla tyrannina), Olive-backed and Montane Woodcreepers and the Greater Scythebill are among the few that are largely confined to Andean cloudforest, mostly at elevations ranging from 1500 m up to 3000 m but, depending on the species, locally down to 700 m or, in elfin forest, up to 3400 m. In the mountains of Central America, the White-striped and Spot-crowned Woodcreepers likewise frequent cloudforest, pine-oak woodland and pine-fir (Pinus-Abies) forest, ascending locally to 3600 m. Most lowland and montane dendrocolaptids are found exclusively within their preferred elevations, but a few widely distributed species occur from lowland rainforest up to montane cloudforest. The Strong-billed and Black-banded Woodcreepers are the best examples of this pattern, with some populations of each having strictly montane distributions and others being restricted to the Amazonian lowlands.

The rainforests of the Amazonian lowlands are without doubt the centre of woodcreeper distribution, with many localities harbouring in excess of 15 species and some holding more than a third of all woodcreeper species. Maximum diversity is reached at sites where upland terra firme forest and seasonally flooded várzea and igapó forests exist in close proximity. Most woodcreepers occur in the upland forest, but a few, such as the Long-billed, Striped and Zimmer's Woodcreepers, are largely restricted to seasonally flooded and river-edge forests. Within each of these habitats, woodcreepers are often segregated by micro-habitat, often through the use of different strata or unique substrates. The scythebills' specialization on bamboo thickets is a striking example: at some Amazonian sites, both the Red-billed and the Curve-billed Scythebills are to a great extent confined to dense stands of bamboo, mostly Guadua.

Different species of woodcreeper join different types of foraging flocks. Some associate primarily with mixed-species flocks that travel rapidly through the forest, while others join flocks that forage in association with swarming army ants. Species in the genera Xiphorhynchus and Lepidocolaptes often forage with mixed flocks in which most species frequent trunks and branches in the understorey and mid-levels of the forest, but a few, such as the Lineated Woodcreeper, choose instead to seek food in the forest canopy. By contrast, woodcreepers that forage in association with army-ant swarms, examples of which include the White-chinned, Black-banded and Red-billed Woodcreepers, often perch on trunks within a few metres of the ground while awaiting prey, which they capture on or near the ground. Differences among the members of the family in the use of habitat are discussed in greater detail below (see Food and Feeding).

Away from Amazonia, the number of dendrocolaptid species co-existing at a given site drops significantly. In the Atlantic Forest of south-eastern Brazil, for example, no more than eight species are found in sympatry. Likewise, whereas 15-19 woodcreeper species co-exist at many Amazonian sites, only nine have been recorded at the well-worked La Selva Biological Station, in the lowlands of Costa Rica; indeed, only 19 species occur in all of Central America. Regional diversity may be higher where elevational gradients are pronounced because some woodcreeper species show a pattern of altitudinal replacement; at a local level, however, species diversity decreases rapidly at higher elevations.

Most woodcreepers are relatively specialized in their micro-habitat use, with the majority frequenting the understorey and mid-levels of the forest interior. A significant percentage are more flexible in that they regularly frequent forest edge or enter older second growth when it regains the character of mature forest. Far fewer dendrocolaptids inhabit younger second growth, plantations, selectively logged forest or isolated forest fragments, and, in most cases, these are the very species that live in more open habitats.

In addition to the four woodcreepers characteristic of semi-open habitats (see below), there are a few that, despite requiring at least patches of mature forest nearby, seem to be able to exploit a variety of forest types. Near the northern limit of the family's distribution, the Ivory-billed Woodcreeper occurs in a variety of lowland, montane, deciduous, pine-oak and secondary forests, from sea-level to about 2500 m. Willis even speculated that this species' behavioural plasticity may give it an advantage over more specialized woodcreepers in the variable climate typical of northern latitudes. It may be a similar plasticity that has allowed the Olivaceous Woodcreeper to become the most widespread member of the family, with a range extending from near the northern edge of woodcreeper distribution, in western Mexico, to near the southern limit, in northern Argentina, and also from sea-level to nearly 2300 m. The Olivaceous Woodcreeper occurs in lowland rainforest in Amazonia and along the Atlantic coast of Brazil, in deciduous woodland on the Yucatán Peninsula and in the caatinga of north-eastern Brazil, and in montane forest in Central America, the Andes, the coastal ranges of Venezuela and the various ranges of the Brazilian Atlantic Forest. It is also one of few dendrocolaptids that is present on the offshore island of Tobago, although, surprisingly, it is absent from Trinidad.

Only a small number of woodcreepers inhabit predominantly non-forested habitats. The Narrow-billed Woodcreeper is widespread in both savanna (cerrado) and Chaco vegetation in central South America, as well as, more locally, elsewhere on the continent. It also occurs in gallery and dry forests and can even be found in gardens and city parks. Even more plastic in its habitat preferences is the Straight-billed Woodcreeper. This species is largely restricted to coastal mangroves in some parts of its range, but elsewhere it frequents arid scrub, wooded savanna, gallery forest, or seasonally flooded forest. It is equally at home near humans, frequenting plantations, scattered trees associated with agricultural lands, and even trees on the outskirts of some Amazonian cities. The only wooded situation that it avoids is the interior of mature forest. The Streak-headed Woodcreeper fills a similar niche in Central America. It shows a comparable variety of habitat preferences, and appears to be equally comfortable in the proximity of humans. The Scimitar-billed Woodcreeper occurs in arid scrub, savanna, and the Chaco woodlands of southern South America.

Finally, there are four dendrocolaptids that can be found in semi-open woodland. The Great Rufous and Moustached Woodcreepers, in addition to some populations of the Olivaceous Woodcreeper and the Red-billed Scythebill, occur primarily in semi-deciduous and deciduous woodlands, habitats that may be transformed seasonally into semi-open environments.

General Habits
The Dendrocolaptidae are among the most characteristic avian inhabitants of Neotropical lowland forests. Their conspicuous habit of creeping, or "hitching", up trees gives the group its name. This behaviour converges with that of the generally smaller but morphologically similar certhiid creepers of the Holarctic and treecreepers (Climacteridae) of Australia and New Guinea. In each group, the birds climb up tree trunks and branches by grasping the bark with their strong feet and long claws, and using the spine-tipped tail as a prop. Woodcreepers move with short hops, those of some species being so rapid that motion appears smooth. Although they occasionally move in reverse downwards on trunks, especially when descending into nesting or roosting cavities, or when foraging in association with ant swarms, they typically move upwards and outwards, using a direct or spiral route. Trunk-foraging woodcreepers begin near the bottom of their preferred stratum, creep to the top of this level, and then fly to the base of the next tree in order to repeat the process. Most species prefer vertical trunks, while others select horizontal limbs in the canopy, and a few seem to specialize in creeping along the undersides of branches. Only a small number of species, particularly those frequenting open situations, regularly descend to forage on the ground (see Food and Feeding). The flight of most large woodcreepers is strong but somewhat undulating, with rapid flaps interspersed with glides on outstretched wings and spread tail. Smaller species have a more direct and darting flight.

Woodcreepers are generally encountered singly or in pairs, but many species join the mixed-species flocks that are ubiquitous in the lowland forests frequented by those dendrocolaptids. Woodcreepers usually attend flocks of insectivores, rather than those of tanagers (Thraupidae) and other frugivores. Most of them associate with understorey flocks containing Myrmotherula antwrens and, especially, Thamnomanes antshrikes. A few species, most notably the Lineated Woodcreeper, associate with canopy flocks led by greenlets (Hylophilus) or other canopy-dwelling insectivores. Groups of woodcreepers consisting of more than a single pair are in general uncommon, and in most instances these groups represent families, with one or both parents tending dependent young. In the case of Dendrocincla, only one parent raises the young (see Breeding), so that groups of multiple individuals probably represent either a mother and her offspring or aggregations of wandering immatures. By contrast, the larger woodcreepers of the genera Dendrocolaptes, Hylexetastes and, presumably, Xiphocolaptes remain paired throughout the year, and the young of some species may not become independent until the beginning of the year after hatching; thus, when two individuals of a species are seen together, they probably represent pair-members, while small groups are likely to be family parties.

The aggregations of woodcreepers that gather, sometimes in fairly large numbers, over swarms of army ants are anomalous for these typically solitary birds. Ant swarms are often attended by three or fewer Plain-brown Woodcreepers, although it is not unusual for four or five individuals to be present together, and in Trinidad, in the absence of competing thamnophilid antbirds, up to twelve of this species have been found at the same swarm. Similarly, seven or eight White-chinned Woodcreepers have been noted at ant swarms in south-eastern Peru.

Although not particularly secretive, many woodcreepers are nonetheless shy. Most move rapidly and continuously, making them difficult to observe. Only rarely do trunk-foraging woodcreepers remain motionless for more than a few seconds. By contrast, those that forage over ant swarms are more sedate, primarily because they must wait to spot prey flushed by the ants. Accordingly, a disproportionate amount of what is known of the behaviour of the Dendrocolaptidae pertains to birds foraging in association with ants. At the first sign of danger, woodcreepers hide on the rear surface of large trunks while tentatively peering around to investigate the situation. Many are difficult to approach and to study at close quarters, but some become accustomed to the presence of an observer, especially during extended periods of observation while they forage over army ants.

The roosting behaviour of this family has been studied for only a few species in Costa Rica, by A. F. Skutch, and in eastern Amazonia, by Y. Oniki. Woodcreepers roost solitarily, in cavities similar to the ones used for nesting (see Breeding). Natural cavities are utilized more often than are old woodpecker holes, possibly because they are less conspicuous. The young do not return to the natal nest after fledging, but instead they seek new sites for roosting. Although documented roost-sites have been in cavities relatively close to the ground, this may reflect ease of observation as much as it does availability of suitable cavities or particular preferences. A captive Straight-billed Woodcreeper roosted in an upright position, with its head tucked into its back feathers. This posture is probably the typical one adopted by all roosting woodcreepers.

Maintenance behaviour among the members of this family is poorly known, with most information being only anecdotal. Various woodcreepers scratch the head by the indirect method, reaching over a wing with one foot while gripping the substrate with the other. In addition, many dendrocolaptids have been observed to perform "anting", in which they rub small, presumably noxious items on the wing or tail feathers; sometimes, the items are then eaten. Anting behaviour is often thought to be a means of removing ectoparasites of the plumage or skin, but Willis suggested that it may serve to wipe distasteful secretions from prey, while H. Sick proposed the possibility that formic acid released during anting may be a physiological stimulant.

Regardless of the true function of anting, and despite a dearth of knowledge of the natural history of the Dendrocolaptidae, woodcreeper parasites are fairly well studied. G. Bennett and colleagues have studied blood parasites in many woodcreeper species. Likewise, R. Price and D. Clayton reviewed an array of feather lice, chiefly of the genus Rallicola, that are known from woodcreeper hosts. Moreover, L. Kudon described a new genus of feather mites, Platyacarus, that is specific to woodcreepers, and detailed host-parasite relationships for many new species in this genus. Within the family, the absence of Platyacarus mites on the Scimitar-billed Woodcreeper provides support for the ancestral position of the genus Drymornis, for these mites infest at least some members of all other dendrocolaptid genera. Finally, M. Marini and D. Couto correlated ecological parameters and ectoparasite infestation for several woodcreepers from central Brazil.

Some woodcreepers are remarkably aggressive towards conspecifics and even towards other woodcreepers. During his studies of ant-followers, Willis documented numerous instances of both intraspecific and interspecific aggression among dendrocolaptids, while N. Pierpont studied interspecific interactions in a guild of woodcreepers occurring in Peru. Not surprisingly, both found that larger species are generally dominant, but some highly aggressive Xiphorhynchus species periodically attack the larger, less agile and possibly somewhat more docile species in the genus Dendrocolaptes. The Buff-throated, Black-striped and Tschudi's Woodcreepers seem to be especially aggressive. Pierpont speculated that the aggressive nature of Tschudi's Woodcreeper may be necessary for it to defend territories against the larger Elegant Woodcreeper, which generally dominates it, even forcing it into marginal habitats.

Most woodcreepers are highly territorial, but "territorial" defence is not necessarily of the same kind as that demonstrated by passerines of temperate regions. For example, among species of understorey-dwelling Xiphorhynchus studied by Pierpont in south-eastern Peru, territorial defence manifests itself in the eviction of intruders from understorey flocks. In the case of the Elegant Woodcreeper, the dominant species in this guild, pairs excluded intruders from two understorey flocks, but they allowed intrusion into the area, not only by other Xiphorhynchus, but also by conspecifics, provided that these intruders did not attempt to join their flocks. Thus, aggression and "territory" defence are largely limited to the immediate proximity of these flocks. One can only speculate why many woodcreepers sing only briefly, at dawn and dusk, from favoured perches near their nests or roost-sites. Song does appear to be seasonal for at least some species, their dawn bouts of singing being longer during the breeding season but quite brief at other times. Perhaps the availability of cavities used for roosting and, especially, for nesting is a limiting factor, much as are foraging opportunities in mixed-species flocks. If so, woodcreepers may concentrate their energy on defence of niche parameters that are most limiting. Future investigation of this atypical singing behaviour should provide an insight into the importance of territoriality among the Dendrocolaptidae.

Most woodcreeper songs are neither complex nor musical, but they are among the most characteristic components of dawn and dusk choruses in Neotropical forests. Woodcreeper songs consist primarily of simple rattles or trills, those of a few species instead comprising clear whistles. Some species emit rapid and repetitive renditions of a single note. Others have songs that take on the character of a whinny or a laugh by initially ascending and then descending, before slowing at the end. The song of the Long-billed Woodcreeper is atypical, being a simple series of three to five drawn-out whistles each described as "whoooóóóooo", "twoooooóoo" or "wheeeeéeer". Only a few dendrocolaptids regularly utter more complex vocalizations, with the song of the Brown-billed Scythebill (Campylorhamphus pusillus) perhaps representing the pinnacle of complexity in the vocal capacities of the family: a soft, twittering trill in the background is combined with a series of loud, quavering whistles in the foreground, yielding a song described either as "wheéwhipwhipwhipaweé, at-t-t-t-t-weeaweéaweé" or as "twe-weo-wéó-weo weo-we-we-we-we-we". The Scimitar-billed Woodcreeper is likewise atypical for a woodcreeper, as the loud, descending jumbles of notes that represent its song are often given in a duet.

The frequency range of woodcreeper songs is neither particularly high nor particularly low. M. Palacios and P. Tubaro documented the range as being 840 Hz to 8830 Hz for the songs of all but a few species, and they found an inverse correlation between song frequency and both body mass and bill length, the latter even after removing the effects of body size. Therefore, songs with lower frequencies are given not only by the larger species, but also by those with the longest bill. Those authors suggested that, in addition to the well-known trend for larger birds to have deeper-voiced songs, the increased length of the vocal tract corresponding with a longer bill may result in resonating properties that affect song frequency. The correlation was complicated, however, by a tendency for species living in open habitats, such as wooded savanna, to have higher-frequency songs than those occupying forested habitats. The correlation of bill length and song frequency was significant, therefore, only when two open-country species, the Narrow-billed Woodcreeper and the Scimitar-billed Woodcreeper, were excluded from analyses.

Woodcreepers stand out among Neotropical birds for their tendency to sing during twilight hours. Apart from a few crepuscular bird species noted for singing in the pre-dawn darkness, such as forest-falcons (Micrastur) and potoos (Nyctibius), woodcreepers are among the first birds to sing at dawn and the last to sing at dusk. Many dendrocolaptids begin singing at the first hint of light in the sky, and their dawn chorus is over even before sunlight hits the forest canopy. Several members of the family, particularly those in the genera Dendrocincla, Dendrexetastes, Hylexetastes, Xiphocolaptes and Dendrocolaptes, as well as many Xiphorhynchus species, sing for no more than a few minutes in the near-darkness of pre-dawn, after which they begin their daily foraging. Many woodcreepers seldom sing during the day, so that forging an association between voice and visual characters is extremely difficult, making the field identification of many dendrocolaptids a real challenge. Most species that sing through the day do so intermittently, and, in many cases, those species that sing during the daytime do not sing extensively at dawn. Regardless of time of day, the Olivaceous Woodcreeper typically sings intermittently, and, despite a lack of quantitative data, it seems hardly more likely to sing at dawn than to sing after sunrise. Unlike many others in the family, the Olivaceous Woodcreeper rarely delivers more than one or two songs at a time or from the same perch. Instead, this woodcreeper forages actively and sings rarely, so that singing individuals are often difficult to follow. Some Lepidocolaptes and several of the smaller Xiphorhynchus species likewise sing intermittently during the day, but generally not for extended periods at dawn or dusk.

Willis found that some woodcreepers sing at dawn and dusk from perches near the centre of their territory, often near a roost-site. In these instances, song may function as a means of marking a bird's nesting or roosting territory, and, in the absence of other birds invading this territory, little singing is necessary. Territorial interactions are often accompanied by song, but Pierpont found that various species of Xiphorhynchus evicted intruders from mixed-species flocks, rather than from a territory itself (see General Habits). In such cases, constant singing of the kind used by birds defending a patch of suitable habitat may be unnecessary.

Despite the fact that their songs are given largely at dawn and dusk, woodcreepers are not necessarily quiet during the day, because most utter a variety of calls. The better-known species, such as the Buff-throated Woodcreeper and the closely related Cocoa Woodcreeper, regularly utter three or four different calls, and occasionally several others. Pair-members of the Buff-throated Woodcreeper are regularly encountered while foraging together, often among mixed-species flocks. Although these birds rarely sing during the daytime, mates appear to communicate with each other by using characteristic "long calls" that are given intermittently throughout the day, in all months of the year. It is possible that reports of singing by both sexes of some woodcreepers, including the Buff-throated, in fact involve individuals emitting long calls rather than true songs, which appear to be given only by the males, and primarily during the breeding season. The most common call of the Buff-throated Woodcreeper is a descending series of roughly five to seven loud, sometimes paired, whistles and has been described as "wheer, wheer, whip-whip-wip-wip" or "fee-a-wip, fee-a-wip, wip-wip". This species also utters, in a variety of contexts, especially when going to roost, a piercing whistle or "short call" described as sounding like "pyeeeu" or "stool".

Homologues of both the long call and the short call are given not only by other Xiphorhynchus species, but also by the Plain-brown Woodcreeper of the genus Dendrocincla. The diverse repertoire of the Plain-brown Woodcreeper includes both a short, descending trill, which many authors consider to be the song, and a long, incessant sputtering that may last for several minutes at a time, which many regard as a call. The true function of these vocalizations remains unclear, but the conditions under which the sputtering "trill" is given, generally at dawn, suggest that it represents the song of the species. By contrast, the short trill is given throughout the day, and apparently by immatures and adults of both sexes.

Considering its infrequent nature, the role of song in maintaining pair-bonds is uncertain. Most woodcreepers appear to be monogamous, and many remain paired throughout the year, but this is not true of all species. All members of the genus Dendrocincla for which data are available appear to form only brief pair-bonds, and this appears to be the case also with the Olivaceous Woodcreeper and possibly other species (see Breeding). The Tyrannine Woodcreeper has a remarkably loud song that it delivers for several hours daily through the breeding season, usually from the same perches on exposed ridges. Although apparently territorial when singing, males do not react to tape playback outside their normal hours of singing. On the basis of these findings, from a study in the Colombian Andes, Willis and Oniki suggested that these birds may mate on "exploded", or dispersed, leks that have come about as a result of the difficulty of finding mates in the sparsely populated and densely vegetated environment frequented by the species (see Breeding). Continuous singing over a protracted period is also typical of the Cocoa Woodcreeper, another dendrocolaptid that is believed not to form extended pair-bonds. This scant evidence suggests that the frequency with which songs are given and the importance of the song in mate attraction are inversely proportional to the duration of the pair-bond.

Woodcreeper songs contradict the long-standing belief that suboscine songs are highly stereotyped and geographically invariable. Recent work has documented a higher degree of individual variation in woodcreeper song than had ever been suspected. Moreover, in addition to the variations on a theme resulting from individual and micro-geographical differences, the songs of geographically well-separated populations of what are now recognized as the same species may show remarkable differences, or variations on different themes.

Indeed, marked geographical variation appears to be the rule, rather than the exception, not only in the songs of many woodcreepers but also in those of many other Neotropical suboscines. The significance of this variation, however, remains elusive. Many populations that differ vocally also differ, albeit often subtly, in morphology. In a few instances, research has revealed that some vocal differences among populations correspond to the existence of cryptic species, but the extent of geographical variation in suboscine vocalizations is still unknown. Moreover, although a few North American tyrant-flycatchers have innate songs, subsequent studies on bellbirds (Procnias) imply that some suboscine songs are learnt. Furthermore, it seems reasonable to believe that even innate vocalizations should vary geographically, since there is no reason why they should differ fundamentally in this respect from other genetically controlled characters such as aspects of plumage or morphology. Interestingly, this does seem to be the case with the very few suboscines in which it has been investigated. Recent studies have revealed that the songs of certain tyrant-flycatchers of the genus Empidonax, as well as those of the furnariid Pale-breasted Spinetail (Synallaxis albescens), do exhibit a limited amount of geographical variation.

Geographical vocal variation is marked and easily perceptible in the songs of the Plain-brown, Long-tailed, Olivaceous, Wedge-billed, Northern Barred, Planalto, Ocellated, Buff-throated, Cocoa, Straight-billed and Lineated Woodcreepers, and in those of both the Curve-billed and the Red-billed Scythebills. The songs of many other dendrocolaptid species vary more subtly. Exceptions seem to be the Cinnamon-throated Woodcreeper (Dendrexetastes rufigula) and members of the genera Hylexetastes and Xiphocolaptes, the songs of which vary minimally over large regions, despite significant variation in plumage. Even interspecific differences are subtle in the last two genera.

Significant geographical variation in the songs of what was until recently recognized as a single species, the "Barred Woodcreeper", provided the first clue that Dendrocolaptes certhia, as then constituted, did in fact represent more than one allospecies. Three remarkably different song types are known from these birds, the differences representing not variations on a theme but, instead, completely different themes. First, Amazonian populations utter a rapid whinny or laugh, generally consisting of 8-15 simple notes that rise and then fall in pitch. Secondly, barred woodcreepers from Central America and north-western South America deliver a series of usually three to five protracted whistles with a sharply inflected ending. Finally, on the basis of a few recordings made by the late P. Schwartz, birds from the Maracaibo Basin of north-western Venezuela sing a series of protracted whistles on one pitch, the notes degrading into a harsh chatter. C. Marantz and Willis have recently identified a suite of morphological and behavioural characters that correspond with these song differences, and avian taxonomists now recognize two species, the Northern Barred Woodcreeper and the Amazonian Barred Woodcreeper.

Marantz is currently undertaking work aimed at characterizing geographical variation in the songs of the Buff-throated Woodcreeper and its allies, some of which may lead to the recognition of similar cryptic species. In this case, there appear to be three to five distinct song types. Some populations of the Buff-throated Woodcreeper deliver a rapid three-part whinny consisting of an introduction in which the notes increase in frequency and speed, a middle part during which frequency and speed are relatively constant, and a conclusion that descends and slows. Each part is highly variable in length, even within individual birds, and sometimes one or more parts are lacking altogether, so that analysis of geographical variation is a challenging task.

The Olivaceous Woodcreeper offers an even greater challenge. Preliminary studies reveal that it has at least six different song types. These may be broadly classed in two groups: the trills sung by populations in Central America and adjacent South America, and series of clear whistles given by birds in the rest of the South American range. Although all of these populations have been recognized for more than 75 years as belonging to a single species, geographical variation in plumage is striking, and it is likely that this species may comprise perhaps as many as five allospecies.

It seems premature to draw conclusions solely on the basis of geographical patterns in woodcreeper vocalizations. The high degree of individual variation complicates assessment of geographical differences, and the behavioural context of vocalizations is little understood. A tendency towards interspecific territoriality adds further confusion, and in no case is there any sound indication of which vocalizations, if any, serve as reproductive isolating mechanisms in the Dendrocolaptidae. Some calls of the Buff-throated Woodcreeper vary less from one region to another than do its songs, a pattern displayed by many oscine passerines, the Song Sparrow (Melospiza melodia) being just one example. If these calls are used for species recognition and mate choice, then significant differences in songs may not be a good indication of evolutionary relationships; preliminary genetic data for the Buff-throated Woodcreeper support this scenario. The function of woodcreeper song, therefore, remains a mystery. The limited understanding of the evolutionary relationships of the Dendrocolaptidae, and of the behavioural context and genetic basis of their vocalizations, lends weight to the view that taxonomic changes should await analyses of multiple, independent data sets. These must encompass, in addition to vocalizations, both morphological characters, including plumage, bare parts and skeletal features, and genetic characters.

Food and Feeding
The name "woodcreeper" is appropriate for all but a few members of the family. The Dendrocolaptidae spend much of their time in creeping along trunks and branches of trees and large shrubs. Species in the genera Xiphorhynchus and Lepidocolaptes seek their food in a manner typical of woodcreepers. With few exceptions, these species forage while creeping upwards along trunks or outwards along major branches. They either glean prey from the bark's surface or probe into crevices, knotholes, vine tangles, moss clumps, or clusters of bromeliads or other epiphytes. There is, however, some degree of specialization with regard to substrate use. The Buff-throated Woodcreeper, for example, probes or rummages through clusters of dead leaves, whereas the Spotted Woodcreeper tends to forage on trunks heavily laden with moss. Other Xiphorhynchus species, and some in the genus Xiphocolaptes, may specialize on foraging among bromeliads and other epiphytes.

Gleaning and probing of bark, although characteristic of woodcreepers, are not the sole means of obtaining food. Pierpont distinguished two guilds of woodcreepers on the basis of their primary foraging technique, either picking or sallying. Although the species that primarily pick will occasionally sally after flushed prey, and those that sally also regularly pick prey, the two techniques provide a means of partitioning resources. These behavioural differences are associated with specific genera. Species that forage principally while climbing trunks, limbs and vines, either by picking items from the surface or by probing into the wood or other substrate, include the members of the genera Xiphocolaptes, Xiphorhynchus and Lepidocolaptes. Woodcreepers in the genera Glyphorynchus, Nasica, Drymornis, Dendrexetastes and Campylorhamphus are probably best treated as pickers with specialized types of foraging behaviour or substrates. Sallying is best suited to "sit-and-wait" predators. Among the woodcreepers, this behaviour is most often associated with attendance at swarms of army ants, which are used as "beaters" to flush prey that is easily captured. Sallying is especially well developed in the genera Dendrocincla, Hylexetastes and Dendrocolaptes, which include those species that depend on following ants. Sittasomus and Deconychura woodcreepers also sally, but these birds follow mixed-species flocks, not ants. Whereas ant-followers spend much of their time perched motionless, waiting for prey to be flushed, the Olivaceous and Long-tailed Woodcreepers hitch along trunks in the typical manner, suggesting that their sallies are directed at prey flushed during the bird's ascent.

Species in both foraging guilds partition food resources by exploiting different micro-habitats and strata. Among the pickers, for example, scythebills frequent bamboo thickets, where the exceptionally long and strongly decurved bill of the genus is well suited to probing stalks in search of arthropods hidden in the hollow interior. The equally long but straight bill of the Long-billed Woodcreeper may likewise facilitate its probing of clusters of bromeliads along horizontal limbs of canopy trees. The massive but somewhat less unusual bill of species in the genera Xiphocolaptes and Hylexetastes may allow these woodcreepers to exploit resources that are inaccessible to the smaller and weaker-billed species of Xiphorhynchus, such as grubs inside decaying wood. Whereas most Xiphorhynchus species forage along trunks and large limbs in the understorey and at mid-levels of the forest interior, the Black-striped Woodcreeper forages primarily in the upper strata of the forest, sometimes making forays into the canopy. Several species of the closely related genus Lepidocolaptes likewise forage predominantly in the canopy, a habit especially prevalent when Xiphorhynchus woodcreepers also occupy the site. Pierpont suggested that the absence of interspecific aggression between the Lineated Woodcreeper, the sole Lepidocolaptes that is widespread in the Amazon Basin, and the various species of Xiphorhynchus that she studied was the result of habitat-partitioning by strata. Despite overlap with Xiphorhynchus in both substrate use and size of prey taken, the Lineated Woodcreeper avoids contact with these larger birds by foraging in the canopy above them. It is possible that, where they co-exist with Xiphorhynchus, the smaller Lepidocolaptes may be pushed up into the canopy.

Similar partitioning by strata has been suggested for the two smallest dendrocolaptids. The Wedge-billed Woodcreeper forages primarily by gleaning tiny prey from the surface of large trunks in the understorey. By contrast, the Olivaceous Woodcreeper seeks food largely in the upper levels of the forest, especially at sites where it co-exists with the former. The degree to which the Olivaceous Woodcreeper is forced into the canopy by the Wedge-billed Woodcreeper remains untested, but the Olivaceous Woodcreeper forages more by sallying than by picking, and the substrate preferences of these two are not nearly so similar as they are for many larger woodcreepers.

In an early study of substrate use in a community of woodcreepers in south-eastern Brazil, M. de L. Brooke demonstrated an even finer partitioning within the trunk-foraging species of the understorey. He found that species differed in the diameter of trunks that they used, in the distance for which they ascended these trunks, and in their tendencies to forage on trunks or on branches.

Despite the morphological uniformity of the woodcreepers (see Morphological Aspects), many localities harbour a variety of species. Five different pickers in the genus Xiphorhynchus occupied Pierpont's study site in south-eastern Peru. Some of these are segregated by habitat, but three understorey species occur syntopically, partitioning the site by means of interspecific aggression, rather than by habitat preference, diet or foraging technique. Indeed, Pierpont demonstrated that the Elegant and Tschudi's Woodcreepers both prefer terra firme forest and that both used the same techniques to take prey of the same size from the same substrates. In a few instances, individuals of these two species experienced reciprocal turnover of their territories, demonstrating conclusively that the same sites are suitable to both species. Although they two occurred at the same sites, Tschudi's Woodcreepers preferred mixed-species flocks that travelled through the forest at a significantly faster pace than that of flocks attended by Elegant Woodcreepers. The Striped Woodcreeper was the third species found at the site, where it was largely restricted to palm swamps; radio-tracking, however, revealed that this species repeatedly attempted to invade territories of Elegant Woodcreepers in neighbouring terra firme forest. Although the Striped Woodcreeper sallied to a greater extent, it likewise took prey of comparable size to that captured by the Elegant and Tschudi's Woodcreepers and from similar substrates. Pierpont further demonstrated that overlap in diet and substrate use was positively correlated with interspecific territoriality and aggression, meaning that those species most likely to compete over resources were also the most likely to exclude one another by using aggression. Pierpont's work on understorey Xiphorhynchus woodcreepers provided an important insight into the way in which closely similar species can co-exist, not only by partitioning resources and micro-habitats, but also through interspecific aggression.

A few dendrocolaptids, principally those frequenting open habitats, regularly descend to the ground. The Scimitar-billed Woodcreeper is the family's archetypical ground-forager. Unlike other woodcreepers, this species eschews foraging among mixed-species flocks of arboreal insectivores, choosing instead to look for food on or near the ground in association with terrestrial ovenbirds. Indeed, it is adept at running on flat surfaces through the scattered vegetation of its preferred habitat, but, as other woodcreepers, the Scimitar-billed does investigate fallen logs, branches, bromeliads, and cavities in trees. Most of its prey appears to be taken from the earth's surface, typically by probing in soft soils. The Great Rufous Woodcreeper, which frequents the same open savannas and Chaco woodland as those inhabited by the previous species, also forages terrestrially on occasion. Although it does not spend so much time on the ground as does the Scimitar-billed Woodcreeper, it does descend regularly to rummage through leaf litter, especially in open habitats. When on the ground, the Great Rufous Woodcreeper usually hops rather clumsily, unlike the Scimitar-billed Woodcreeper, which runs about swiftly and seems better adapted for a terrestrial lifestyle.

It is possible that most woodcreepers descend to the ground only rarely because they are awkward on horizontal surfaces, and thus highly vulnerable to predators. A most unexpected observation was that of an Ivory-billed Woodcreeper foraging on mudflats exposed when mangroves were cleared.

The majority of accumulated knowledge of the foraging behaviour of woodcreepers is based on detailed observations made by Willis and Oniki of birds foraging over ant swarms. Although those authors' findings may be biased slightly with regard to those woodcreepers that do not normally follow ants, the basic foraging dichotomy appears to be stereotyped. Even over ant swarms, sallying species dart after flushed prey, whereas picking species, which forage over ants opportunistically, take prey primarily by gleaning or probing as they hitch along trunks low over the swarm. Woodcreepers that normally forage at mid-levels may descend to levels near the ground to consume flushed prey. By all accounts, the White-chinned Woodcreeper is an obligate ant-follower. The Red-billed, Black-banded, Hoffmann's, Amazonian Barred, Plain-brown, Plain-winged, Tawny-winged and Ruddy Woodcreepers (Dendrocincla homochroa) also follow army ants, but none of them exclusively so. The true ant-followers share a similar mode of foraging: they descend to vertical perches near the ground and await flushed prey. When a prey item is spotted, the woodcreeper sallies out to capture it, whether it is on the ground, on a nearby trunk, amid foliage, or in mid-air. Much like picking woodcreepers, ant-followers use interspecific aggression as a means of partitioning foraging opportunities over swarming ants. Larger species, especially the Red-billed and Black-banded Woodcreepers, take the premier positions, low down over the front edge of the swarm. Willis noted that, in the absence of large antbirds (Rhegmatorhina) and bare-eyes (Phlegopsis) near Manaus, in Brazil, the smaller White-chinned Woodcreeper forages on slim trunks near the ground, but in in southern Amazonia, where large antbirds are present at swarms, this woodcreeper forages higher up and from broader trunks. Willis reported similar exclusion and release for the Plain-brown Woodcreeper. In Trinidad, with its dearth of ant-followers, the Plain-brown Woodcreeper forages near the ground and in large aggregations, habits that are rare where this species shares swarms with aggressive antbirds and larger woodcreepers. Unlike the White-chinned Woodcreeper, the Plain-brown Woodcreeper restricts its ant-following to swarms that move through its territory, and it often picks prey items near the periphery of the swarm instead of sallying for prey near the centre.

Willis and others have speculated that some dendrocolaptids characteristic of mixed-species flocks may be attracted to the activity of birds following the ants, rather than to the ants themselves and the prey flushed by them. Some of these woodcreepers appear to be unsure how to take advantage of ant swarms, with many species typical of flocks apparently lacking the patience required to forage over ants successfully. These birds spend little time with the ants before departing to seek a mixed-species flock.

Woodcreepers are chiefly insectivorous. In a comprehensive study of the diets of Brazilian birds, O. Schubart and colleagues listed many arthropods as being consumed by woodcreepers. F. Haverschmidt likewise provided orders and families of arthropods found in the stomachs of woodcreepers occurring in Surinam, and, more recently, A. Chapman and K. Rosenberg, O. Rocha O. and E. Peñaranda, and F. Puebla-Olivares produced detailed analyses of stomach contents for select species. In addition, B. Poulin and colleagues detailed emetic samples vomited from the stomachs of woodcreepers captured in northern Venezuela, and Willis has observed a variety of prey taken by birds of several species foraging over ant swarms. These studies reveal that woodcreepers typically take arthropods and that they occasionally take small vertebrates, and yet, with the exception of prey size, the diet differs little across species.

The most common food items are grasshoppers and crickets (Orthoptera), beetles (Coleoptera), cockroaches (Blattodea), ants (Formicidae), spiders (Araneae) and, in the case of some species, lizards. Woodcreepers take smaller quantities of leafhoppers and cicadas (Homoptera), bugs (Hemiptera), termites (Isoptera), earwigs (Dermaptera), ant-lions (Neuroptera), bees and wasps (Hymenoptera), flies (Diptera), moths (Lepidoptera), caterpillars and other insect larvae, dragonflies (Odonata), centipedes (Chilopoda), millipedes (Diplopoda), scorpions (Scorpiones), pseudoscorpions (Pseudoscorpiones), and even small snails (Mollusca) and crabs (Decapoda). Fruit pulp and small seeds have been found in the stomachs of a few dendrocolaptids, but it is unlikely that any woodcreeper consumes vegetable matter routinely. In a study of the stomach contents of five woodcreeper species from southern Mexico, Puebla-Olivares documented a suite of prey types for each. The Tawny-winged Woodcreeper took 26 types of prey, the Ivory-billed 20, the Olivaceous 18, and the Wedge-billed 12. The most important items in the diet of each species were spiders, beetles and wasps for the Tawny-winged Woodcreeper; spiders, caterpillars and various orthopterans for the Ruddy Woodcreeper; ants and beetles for the Wedge-billed Woodcreeper; beetles, spiders and ants for the Olivaceous Woodcreeper; and snails, beetles and orthopterans for the Ivory-billed Woodcreeper.

It is likely that all but the smallest woodcreepers take small vertebrates, in addition to arthropods. Small lizards, especially anoles (Anolis), constitute the most common vertebrate prey, especially for the ant-followers. In one study, vertebrates represented roughly 10% of the diet of the Plain-brown Woodcreeper. Skutch found that small lizards formed the bulk of items brought to nestling Tawny-winged Woodcreepers, although parents may preferentially feed young with such large prey. Some authors have speculated that lizards foraging at ant swarms are the most likely victims. Besides eating lizards, woodcreepers have been observed to take a variety of small frogs, salamanders, and even a few small snakes. Without question, the most unusual instance of a woodcreeper taking a vertebrate involved a Great Rufous Woodcreeper, observed as it pecked at and then ate a small bat.

Probably the biggest interspecific differences in dendrocolaptid diets are those between species that forage near the ground, over ant swarms, and species that pick or probe on trunks or other arboreal substrates. These differences are subtle, but, not surprisingly, ant-followers take a larger proportion of terrestrial prey, such as grasshoppers and crickets, cockroaches, spiders and vertebrates, whereas arboreal woodcreepers consume proportionately more beetles, bugs and larvae. The White-chinned Woodcreeper is an opportunistic feeder, with the relative proportions of most prey taken approximating those of prey flushed by swarming ants, but it shows a strong predilection for grasshoppers and crickets, a weak liking for spiders, and a strong disdain for cockroaches. In the case of the Plain-brown Woodcreeper, however, deviation of the diet from samples of prey flushed by ants may reflect either a higher degree of selectivity or, more likely, a weaker dependence on ant swarms. Like the White-chinned Woodcreeper, the Plain-brown shows a preference for grasshoppers and crickets and a strong disdain for cockroaches, but, unlike that species, it takes many more beetles but many fewer spiders and ants than would be expected on the basis of samples of flushed prey.

The diets of sympatric dendrocolaptids that feed by picking are extremely similar. The Buff-throated and Elegant Woodcreepers show an almost complete overlap in food preferences, despite searching different substrates.

Chapman and Rosenberg concluded that the specialized foraging techniques used by woodcreepers do not result in specialized diets, at least not at the level of their study. A more detailed investigation, by Puebla-Olivares, revealed that dietary overlap was high but not complete, and that prey size may differ more than prey type. That researcher, as well as Rocha O. and Peñaranda, found that the diminutive Olivaceous and Wedge-billed Woodcreepers took primarily small prey, that the Long-tailed Woodcreeper fed on small and intermediate-sized items, and that the White-chinned, Tawny-winged, Ruddy, Ivory-billed and Buff-throated Woodcreepers captured prey of intermediate and large sizes. Pierpont, too, found variation in prey size across a suite of woodcreepers, with prey size positively correlated with body mass in sallying species, but only weakly so in woodcreepers that forage by picking.

Knowledge of the breeding biology of woodcreepers is limited. Much of what is known is derived from many years of observation by Skutch in Costa Rica, supplemented by studies by Oniki and Willis in South America. Two mating systems have been identified in the family. Most woodcreepers, especially the larger species, appear to be socially monogamous, with pair-bonds maintained throughout the year and both sexes taking part in all aspects of nesting and the rearing of young. By contrast, members of the genus Dendrocincla, and perhaps a few other woodcreepers, apparently do not form long-term bonds, instead being polygamous and, in the case of one species, possibly using leks.

In the cases of the Streak-headed and Spot-crowned Woodcreepers, the two species for which the data are most complete, the mates share the tasks of nest-building, incubation, feeding of the nestlings, and care of the fledglings. It appears that these details apply also to other Lepidocolaptes species, as well as to members of the genera Glyphorynchus, Dendrocolaptes and Hylexetastes. A Streak-headed Woodcreeper with abnormal plumage that nested on Skutch's ranch in Costa Rica allowed him to elucidate the relative effort expended by the male and the female. On the assumption that female woodcreepers incubate during the night, Skutch found that, in one season, the male spent much less time in incubating during the day than did its mate; in a subsequent season, however, the sexes shared incubation duties more equally. Both parents brooded the newly hatched young, but the female not only brooded them for longer periods but also continued to brood them after the male ceased doing so. Provisioning rates were comparable for both parents. Although data for other dendrocolaptids are not so complete, several workers have noted that juveniles of a variety of woodcreeper species associate with both parents, and, in the case of the large species in the genera Dendrocolaptes and Hylexetastes, the young may remain with their parents for an extended period.

As mentioned in the preceding paragraphs, the breeding system of the genus Dendrocincla appears to differ from that of most of the Dendrocolaptidae. In the cases of the Plain-brown and the White-chinned Woodcreepers, the partners associate for only a brief period before one parent, presumably the female, takes on the duties of incubating the eggs and rearing the young. Willis noted that interactions begin aggressively, but the female soon accepts the male's advances. Some pair-members investigate possible nest-sites together, but the pair-bond seems to represent little more than this. Skutch's observations of another Dendrocincla species, the Tawny-winged Woodcreeper, confirm that one parent incubates and raises the young without assistance from its mate.

Although single-parent families are well known in the genus Dendrocincla, the reasons for the emancipation of males from the task of parental care remain unclear. Unlike most lekking species, which feed their young with fruit or other abundant but often dispersed resources, woodcreepers focus instead on insects and small lizards, which are unlikely to be abundant in most circumstances. How females of species having the latter kind of diet succeed in provisioning the nestlings remains, therefore, a puzzle. Willis speculated that the high degree of aggression displayed by these species, combined with dispersed but abundant food available at army-ant swarms, promotes this presumably polygamous system; even among those woodcreepers that forage primarily in association with ants, however, single-parent families are typical for some species but not for others. The Tyrannine Woodcreeper has taken this mating system a step further: during the breeding season, the males broadcast unusually loud songs from prominent ridgetops. Although unexpected for a plainly marked, insectivorous species, this behaviour recalls a mating system in which males gather on exploded, or dispersed, leks, a system which Willis and Oniki proposed as being the optimal one for a scarce species dependent on dispersed resources and occurring in a thickly vegetated environment. One can only speculate on the importance of phylogenetic inertia in forcing this species into a difficult situation, one in which the young are raised by a single parent in an environment less suited to this lifestyle than are the lowlands frequented by other Dendrocincla species.

The breeding cycle is complex. The Dendrocolaptidae, taken as a whole, breed in every month of the year, with year-round breeding likely to be the rule when considering the entire ranges of widespread species such as the Plain-brown, Wedge-billed and Olivaceous Woodcreepers. It is important, therefore, to examine patterns not only by species, but also by region. Northern woodcreepers breed primarily during the boreal spring and summer, and southern species during the austral spring and summer. Seasonality of breeding by populations living in equatorial regions is more difficult to assess, but it appears to be associated with seasonal rains. The majority of species breed either late in the dry season or early in the wet season, but some species in the Guianas and nearby regions breed during both the long and the short dry seasons. In addition, despite concordant seasonality of the rains in central and eastern Amazonian Brazil, the breeding seasons of woodcreepers may be, on average, somewhat earlier in the year in southern Amazonia than they are near Manaus. Indeed, Oniki and Willis found little seasonality of nesting near Belém, in extreme eastern Amazonia.

Other aspects of breeding biology are remarkably constant throughout the family. All woodcreepers nest in cavities, in which they lay small clutches of white, unmarked eggs in a shallow cup, or on a bed of bark flakes or wood shavings. Woodcreepers place their nests in a variety of hollows. They are partial to natural cavities in rotting stumps or those created when branches break off trees, but various dendrocolaptids have nested in abandoned termite nests, between buttress roots, amid vine tangles, within hollow bamboo poles, and at the bases of palm fronds. There is no evidence that woodcreepers excavate their own cavities, even though some of them may enlarge the entrance to a pre-existing cavity. Several species use abandoned woodpecker holes, while the Scimitar-billed Woodcreeper has nested in an abandoned nest of a Rufous Hornero (Furnarius rufus) and in a chimney. The Narrow-billed Woodcreeper, characteristic of open habitats, occasionally nests on bridge supports, cement columns, and a variety of other man-made structures.

Most woodcreeper nests are near the ground, a fact which may reflect a preference for low sites, or the availability of suitable cavities, or the ease of locating such sites. Of dendrocolaptid nests described in the literature, the majority have had an entrance within 5 m of the ground, and few have been above 10 m. The entrances to some nests are only just above the ground, and, in a small number of instances, nest-cavities are subterranean.

The nest itself is either a shallow cup, lined sparsely with dark rootlets, dried leaves, stems or other plant fibres, or merely a bed of chips or flakes of wood or bark. Deep cavities may be filled with bark chips and wood flakes, upon which the nest is placed. The filling was nearly 50 cm deep in a cavity contested by both a pair of Streak-headed Woodcreepers and a female Tawny-winged Woodcreeper. The nest entrance is often a long, narrow slit, which can cause adults to contort the body when entering. Skutch documented that some nests have an entrance at the top, rather than at the side, so that the occupants are exposed to the elements. He speculated that roosting sites of this type indicate that protection from predators may be more important than protection from the elements, but Oniki found that flooding from heavy rains was a key cause of nestling mortality among Wedge-billed Woodcreepers near Belém.

As mentioned above, woodcreepers lay plain white eggs. Some authors have commented that the eggs can have a pinkish or greenish cast, but most have noted little more than a slight gloss. Looking at all members of the family, the eggs vary in shape and in size, the latter weakly correlated with body size of the bird. Clutches typically comprise two eggs, less frequently three. Three-egg clutches are regularly laid only by the Scimitar-billed Woodcreeper and members of the genus Lepidocolaptes, and a four-egg clutch may be unique to the Narrow-billed Woodcreeper. Clutches containing a single egg are occasionally laid by a variety of dendrocolaptids, but only some of the largest species, such as the Red-billed Woodcreeper, may routinely lay a single egg. There is no apparent correlation between clutch size and the system of parental care; since most woodcreepers lay either two or three eggs, however, such a correlation would be difficult to confirm, given the paucity of data. The Scimitar-billed Woodcreeper lays on consecutive days, but the Wedge-billed Woodcreeper lays on alternate days.

With most woodcreepers, both sexes incubate the eggs. In the case of Dendrocincla species and possibly a few other genera, however, only one parent, presumably the female, does so. Detailed observations by Skutch at a nest of Streak-headed Woodcreepers revealed that the eggs were tended for 60-66% of the observation period. In one season, Skutch observed 13 incubation sessions by the apparent female, these averaging 38 minutes, with a range of 7-72 minutes, but only three sessions by the male, which averaged 28 minutes and ranged from 15 to 42 minutes. In a subsequent season, the female of this same pair completed eleven sessions, with a mean duration of 27 minutes and a range of 5-57 minutes; the male's eight sessions lasted for 16 minutes on average, with extremes of 6 and 37 minutes. The nest was left untended 14 times in the first season, for periods of 2-41 minutes, with a mean of 21 minutes, and 17 times during the second, when periods of absence were similar, at 4-41 minutes, with an average of 17 minutes. The behaviour of a pair of Spot-crowned Woodcreepers was similar, although these birds tended the nest for 82% of the observation period, with the incubating bird more often waiting for its mate to arrive before it departed from the nest.

Incubation patterns shown by the Tawny-winged Woodcreeper are different. Skutch reported that the lone parent, presumably the female, spent longer periods in incubating, but also longer periods away from the nest. During one morning of observation, the sitting bird remained on the eggs for three periods of 47-87 minutes' duration, these separated by recesses of 26 and 39 minutes. In all, the incubating bird tended the eggs for just over 60% of the time, a figure comparable to the total time that both members of the Streak-headed Woodcreeper pair spent on their eggs.

Not uncommon among woodcreepers is a tendency for incubating adults to return to the nest with a small piece of wood or bark. In fact, this behaviour was used by Skutch as a common method of determining the time of hatching, because the pieces of nest material were replaced by food items as soon as the young hatched and needed to be fed. The continued addition of inedible items to the nest throughout the incubation period is most interesting, but its function remains unknown.

The incubation and nestling periods of woodcreepers are poorly known, but those species in which the pair-members co-operate in these duties have shorter periods than do those in which only one parent cares for the young. From the limited information available, the relatively large Scimitar-billed Woodcreeper has the shortest incubation period, lasting 14 days, but its nestling period is of an average duration of 21 days. The shortest nestling period in the family is that of the tiny Wedge-billed Woodcreeper, the young of which leave the nest after 17 days. Xiphorhynchus and Lepidocolaptes species of comparable size have incubation and nestling periods of comparable length. For example, the Elegant Woodcreeper incubates for at least 16 days and tends its chicks for 18-19 days, while the Cocoa Woodcreeper tends its nestlings for 20 days, during which a single parent may raise the young alone; in Lepidocolaptes, the Streak-headed Woodcreeper incubates for 15 days and tends its nestlings for 18-19 days, the respective periods for the closely related Spot-crowned Woodcreeper being 17 days and 19 days. Species in the genus Dendrocincla require longer, probably because the single parent has less time to devote to incubation and to the feeding of its brood. For example, the Tawny-winged Woodcreeper incubates the eggs for 20 or 21 days and broods and feeds the chicks for 24 days, and the Plain-brown Woodcreeper tends its nestlings for 23-25 days.

Dendrocolaptid nestlings are fed with a variety of prey, usually one item at a time, and in most cases by both parents. In species of Dendrocincla and possibly other genera, however, only one parent feeds the chicks. There is a tendency for the adults to deliver small prey to younger nestlings, but larger prey to older young. Some authors have further suggested that the parents provision young with relatively large items, perhaps compensating for their intermittent visits to the nest. The large size of some prey sometimes makes identification possible. Skutch noted that a Tawny-winged Woodcreeper provided its young with insects when they were small, but mostly with small lizards, and a few spiders, after the chicks were a week old. The Northern Barred and Cocoa Woodcreepers also bring small lizards to the nest, but the Streak-headed and Spot-crowned Woodcreepers deliver chiefly brown insects, and the Wedge-billed Woodcreeper delivers tiny items that are almost impossible to identify in the field. The stomach contents of two nestling Lesser Woodcreepers comprised spiders, orthopterans and a beetle larva.

Provisioning rates vary, depending in part on the number of parents involved. Skutch found that, during a five-hour period, a female Streak-headed Woodcreeper brooded its two-day-old chicks for 131 minutes, after which the male brooded them for 43 minutes, and the parents brought food to the nest 16 times in total. During a subsequent four-hour period, the female alone brooded the eight-day-old young only briefly, and the parents visited the nest 21 times. Finally, when the young were 15 days of age, they were not brooded at all, even during the night, and the parents delivered food to the nest 32 times during a three-hour period. Provisioning rates and prey types were similar for the Spot-crowned Woodcreeper. By comparison, a female Tawny-winged Woodcreeper likewise brooded her young frequently when they were newly hatched, but her provisioning rates were substantially lower than were those of the two Lepidocolaptes species. During multiple six-hour periods, the female visited the nest six times when it contained a single day-old nestling, nine times when there were two nestlings aged 3 days, twelve times when the young were aged 9 days, and 16 times when they were 17 days old. Thus, provisioning rates by a lone parent were less than half those at Lepidocolaptes nests attended by two parents. The Tawny-winged Woodcreeper may, however, compensate by bringing larger prey to the nest. Skutch also reported that nestlings of species with biparental care are much more vocal than are those tended by a single parent. Perhaps young with single parents are quieter because of the increased risk of predation.

Fledgling woodcreepers remain with their parents for an extended period, those of large species apparently being dependent on their parents for longer than are fledglings of small species. At the extreme, young Red-billed Woodcreepers may remain with their parents until early in the year after fledging, but for most species the period of dependence probably lasts for no more than a few months. There is no evidence that fledglings return to their natal cavities to roost; instead, they either locate or are shown alternate roosting sites.

When dual parental care is the case, the parents either will travel together with their offspring, as a group, or will split up, with each parent then accompanying a single fledgling. For those species in which the young are raised by a single parent, groups of three birds probably represent mothers with their offspring. Willis noted that family groups of ant-followers forage away from ants when the young are only recently fledged, but attend swarms together when fledglings are older. He also noted a moderate degree of aggression between parents and young, manifest in some species as the young being dominant over one parent, presumably their father, but subordinate to the other, presumably their mother. Over time, aggression by parents probably increases until, finally, the young leave the natal territory.

In common with most birds frequenting the interior of Neotropical forests, dendrocolaptids are largely or exclusively sedentary. There is no evidence to suggest that any woodcreeper species is truly migratory, although limited, anecdotal evidence indicates that some do wander short distances. A small degree of altitudinal movement probably takes place near mountains in Central America, especially in Mexico, near the northern limit of the family's distribution. The White-striped Woodcreeper, for example, has occurred at unusually low elevations in southern Sonora and, similarly, the Spot-crowned Woodcreeper has wandered to the Caribbean lowlands from Mexico south to Costa Rica. Records of the Ruddy and Long-tailed Woodcreepers from La Selva Biological Station, in Costa Rica, suggest some downslope movement, as these species normally frequent higher elevations in nearby foothills. The Black-banded and Strong-billed Woodcreepers have shown similar downslope movement in Honduras. Conversely, records from Monteverde Reserve, also in Costa Rica, imply that some lowland species, such as the Wedge-billed Woodcreeper, move upslope after breeding. Nonetheless, long-term studies at various sites have documented the year-round presence of most species at most sites, demonstrating that movement, when it does occur, probably represents wandering by a few individuals, rather than migration by the population as a whole.

Few woodcreepers occur on islands, even continental ones only a few kilometres offshore, further attesting to their sedentary nature. Only Trinidad harbours more than two dendrocolaptids, and even there the complement of five woodcreeper species is greatly reduced in relation to that found in the forests of coastal Venezuela, within sight of the island. Apart from Trinidad, the only islands supporting woodcreepers are Tobago, which has one species, and Isla Margarita, off north Venezuela, where two species are found; in all three cases, the taxa concerned represent endemic subspecies. The loss of at least one, and possibly two, species from Barro Colorado Island since its isolation from the rest of Panama testifies to the inability of woodcreepers to colonize suitable habitat across even a narrow water barrier.

A further demonstration of the sedentary nature of woodcreepers is provided by ringing studies. Ringing returns from a variety of sites have involved individuals recaptured on the same territories several years after they were initially ringed. Moreover, all five Tawny-winged Woodcreepers and three of four Olivaceous Woodcreepers captured at a site in Mexico returned to their territories shortly after having been being deliberately displaced.

Relationship with Man
Considering their inconspicuous songs, drab plumage, small size, retiring habits, and the forested habitats that the Dendrocolaptidae frequent, the relationship between woodcreepers and humans is predictably limited. Only the most astute human denizens of tropical forests distinguish between woodcreepers and woodpeckers. For example, Amazonian campesinos and caboclos often refer to woodcreepers as "carpinteros" or "pica-paus", the same terms as those that they use for woodpeckers. Some observers are aware of the generally reddish coloration of woodcreepers, and the better woodsmen, or mateiros, notice that these birds rarely peck at wood in the manner of woodpeckers, but in general dendrocolaptids evade detection.

Woodcreepers are not hunted in any numbers, probably because even the largest species are not worth the effort. Even the moderately large Moustached Woodcreeper, the only dendrocolaptid considered to be globally threatened (see Status and Conservation), is believed to be adversely affected by habitat loss, but not by hunting. The dull plumage patterns and unremarkable songs of woodcreepers afford them little popularity among aviculturists; likewise, their earth-toned feathers are unlikely to be prized by indigenous peoples. Thus, the only interaction between man and woodcreepers is an indirect one: the adverse impact of man that results from the loss and fragmentation of the mature forests that most dendrocolaptids need for their survival (see Status and Conservation).

Status and Conservation
Because woodcreepers are mostly inconspicuous birds of the forest interior, direct human exploitation does not represent a threat to them. Among their natural enemies, snakes seem to be regular predators at woodcreeper nests, feeding on both eggs and nestlings, and forest raptors probably take a few juveniles and adults. Without doubt, the major threat to the long-term survival of the members of this and many other avian families is loss and fragmentation of the forests in which they live. Forest clearance and fragmentation, primarily through logging and for agriculture, have increased tremendously in the last 50 years throughout the Neotropics, and many projections indicate a continuing escalation in habitat destruction for several decades to come.

The Moustached Woodcreeper is the only dendrocolaptid currently listed as globally threatened, being designated as Vulnerable. Not only is this species uncommon and local throughout its range, but it requires intact or, at most, only slightly disturbed tracts of semi-deciduous or dry forest that grows on the richest soils in the interior of north-eastern Brazil. Worryingly, forests in this region are being converted at an alarming rate to charcoal, used to fuel the steel and pig-iron industries. In addition, large-scale irrigation projects promote clearance of dry forest for agriculture. As a consequence, it is estimated that less than 5% of the area originally covered by tropical dry forest in north-eastern Brazil will remain intact in the near future. Illustrating the rapidity of the destruction of dry forest, and the concomitant threat to the Moustached Woodcreeper, a large tract of undisturbed forest near Coribe, in the state of Bahia, harboured a substantial population of these birds in 1988, but had been completely cleared by 1993. Under such pressures, local populations of the Moustached Woodcreeper disappear or, at best, are reduced and fragmented, a situation that seriously threatens the species' long-term survival. Although the Moustached Woodcreeper is protected under Brazilian law, the most important step towards its preservation will be the establishment of a network of conservation areas across the interior of north-eastern Brazil aimed at protecting undisturbed patches of dry forest. Despite the recent creation of several reserves in the region, most notably Serra do Baturité State Environmental Protection Area, in Ceará, and Cavernas do Peruaçu National Park, in Minas Gerais, a network of reserves still did not exist in 2001.

The only woodcreeper currently regarded as Near-threatened is the Greater Scythebill. This poorly known species is uncommon to rare and locally distributed along a narrow and discontinuous belt of montane forest in the middle and upper elevations of the central and northern Andes. It is likely that some of its populations were extirpated as a result of recent clearance of large tracts of montane forest along river valleys in northern Peru, Ecuador and Colombia. Additional data are sorely needed in order to enable an accurate assessment of this species' conservation status.

The Tyrannine Woodcreeper is another uncommon and local dendrocolaptid restricted to montane forest in the Andes. It was included in a preliminary "blue list" for Colombia, and probably merits consideration as being globally threatened. The Tyrannine Woodcreeper is somewhat better known than the Greater Scythebill, unlike which it has been reported, if only rarely, in older second growth, at forest edges and even in clearings, perhaps indicating some degree of tolerance of habitat disturbance. Regardless of that possibility, any species that occupies a narrow elevational belt of montane forest in the Andes is at moderate risk, not only from habitat loss, but also from episodic natural catastrophes such as earthquakes and volcanic eruptions.

Because woodcreepers are sedentary and, in most cases, favour the forest interior, even locally abundant species are vulnerable to forest destruction and fragmentation. For example, a protected forest fragment in south-eastern Brazil was inhabited by the Plain-winged, Olivaceous, Planalto and Lesser Woodcreepers and the Black-billed Scythebill (Campylorhamphus falcularius), but all five species disappeared from the site between 1975 and 1992. Possible explanations were the loss and degradation of most of the undisturbed forest at the site, combined with the fragment's extreme isolation, which prevented recruitment and establishment of immigrants from neighbouring source areas. Only those species capable of crossing large unforested gaps, a category which excludes these woodcreepers, were able to recolonize the remaining 250-ha fragment. Local extinction of the Northern Barred Woodcreeper on Barro Colorado Island, following its isolation when Gatún Lake was created during the construction of the Panama Canal, further illustrates the inability of woodcreepers to disperse across a water barrier. These and other studies illustrate that the detrimental effects of habitat fragmentation can result in the complete disappearance of woodcreepers even from protected sites.

Habitat fragmentation is a serious threat to woodcreepers even in vast areas of undisturbed forest in the Amazon Basin. In central Amazonia, small fragments of forest may not contain enough active army-ant colonies to support species of bird that rely on them. Hence, dendrocolaptids such as the White-chinned, Red-billed and Black-banded Woodcreepers disappear rapidly from fragments smaller than 100 ha, and their numbers are much reduced even in larger fragments. Species that forage mostly with mixed flocks, such as the Spot-throated Woodcreeper, are also adversely affected by forest fragmentation, largely because most flocks wander over an area of at least 10 ha. Mixed-species flocks often disband following fragmentation, with the result that species that are obliged to forage with flocks disappear quickly, leaving only those that are efficient solitary foragers. Only dendrocolaptids possessing sufficient flexibility in their diet and foraging behaviour, such as the Plain-brown, Wedge-billed and Chestnut-rumped Woodcreepers, persist in small fragments in similar or slightly increased numbers.

Moreover, fragments are not nearly so isolated from continuous forest at Amazonian sites as they are in the Atlantic Forest of south-eastern Brazil, where rapid and extensive development around some of the largest cities in the world, most notably São Paulo and Rio de Janeiro, has reduced once continuous forest to small, isolated patches. The loss and fragmentation of forest is even more evident farther north, in coastal eastern and north-eastern Brazil, where little intact forest remains and many woodcreeper populations are threatened. Most at risk in the coastal lowlands of eastern Brazil are several endemic subspecies: race taunayi of the Plain-brown Woodcreeper, cuneatus of the Wedge-billed Woodcreeper, and the nominate races of both the Buff-throated Woodcreeper and the Red-billed Scythebill. Farther inland, the northernmost population of the Scaled Woodcreeper, the subspecies wagleri, occurs together with the Moustached Woodcreeper in woodland that is severely threatened, although wagleri appears to be less sensitive to the degradation and fragmentation of woodland than is the larger species.

Less invasive methods of forest exploitation, such as selective logging, can also have an adverse effect on woodcreepers. The direct removal of trees, and incidental loss occurring when roads are built for transporting timber out of logging concessions, combine to simplify the forest's structure, principally through drastic changes in understorey and mid-storey vegetation. Even when not removed completely, the understorey and mid-storey of selectively logged forest receive an increased amount of solar radiation, leading to micro-climatic changes that disrupt the life-cycle of the arthropods upon which understorey woodcreepers rely. Because the forest canopy is in effect an edge environment, such species as the Lineated Woodcreeper, which forage primarily in the canopy, may be better able to tolerate the degradation and fragmentation that result from selective logging. Besides food scarcity, a loss of suitable nesting sites may explain why fewer species of woodcreeper inhabit second growth and selectively logged forest when compared with primary forest. The preferred nesting sites of woodcreepers, both natural cavities and holes excavated by woodpeckers, are scarcer in the absence of large trees and decaying stumps. Some researchers have suggested that nestboxes will help to maintain woodcreepers in degraded habitats, although this idea has not been tested. It is probably no coincidence that those few dendrocolaptid species that frequent disturbed habitats have been found nesting in a variety of atypical sites, including man-made structures (see Breeding).

Although loss and fragmentation of forest threaten most woodcreepers, a few species are adapted to more open situations. Most notably, the Narrow-billed Woodcreeper, a typical inhabitant of cerrado, is gradually extending its range in south-eastern Brazil to occupy sites that were, until recently, blanketed by dense Atlantic Forest. This species quickly exploits resources not available naturally in the wild. It is also one of few woodcreepers that regularly nest in man-made structures, such as bridge supports and cement columns (see Breeding), and it has been observed while foraging on insects attracted to streetlights and refuse barrels. The Streak-headed and Straight-billed Woodcreepers also frequent open habitats, and both seem to have little difficulty in co-existing with humans. One author speculated that the Tawny-winged Woodcreeper was likely to expand its range southwards, because it is less sensitive to habitat disturbance than are the aggressive antbirds that now exclude it at some sites.

Numerous studies have shown that species diversity, not only of woodcreepers but also of most other avian groups, is maximized when there is a diverse array of continuous and undisturbed habitats in close proximity. Protection of large tracts of undisturbed forest is, therefore, the best way in which to ensure the long-term survival of these birds. Given the inability of most woodcreepers to cross even the narrowest of gaps in forest, a connectivity of forest fragments is also important. Recent work has shown, for example, that narrow forest roads represent enough of a gap to delimit the territories of understorey flocks that some woodcreepers join. Since many of the Dendrocolaptidae are sensitive to even minimal changes in habitat quality, their persistence is a natural indicator that can be useful for monitoring ecosystem health throughout Neotropical forests and woodlands.

General Bibliography
Aleixo (2002), Ames (1971), Bennett & Lopes (1980), Bledsoe, Raikow & Crowell (1997), Bledsoe, Raikow & Glasgow (1993), Bock (1994), Brodkorb (1978), Cheneval (2000), Clayton et al. (1992), Clench (1995), Feduccia (1969, 1970, 1973, 1985), Harper (1989), Hayes & Argaña (1990), Heimerdinger & Ames (1967), Ihering (1915), Irestedt (2002), Irestedt et al. (2001), Kudon (1982a, 1982b, 1982c, 1982d, 1982e), Marantz (1992), Marini & Couto (1997), McFarlane (1963), Nores (2000), Olrog (1972), Palacios & Tubaro (2000), Peters (1951), Pierpont (1983, 1986), Poulin et al. (1994a, 1994b), Price & Clayton (1993), Raikow (1991, 1992, 1993, 1994a, 1994b), Sibley & Ahlquist (1990), Sibley & Monroe (1990, 1993), Tellkamp (1998), Todd (1948), Tubaro et al. (2002), Willis & Oniki (1978).