Taxonomic revision of Old World members of the feather louse genus Columbicola (Phthiraptera: Ischnocera), including descriptions of eight new species

The feather louse genus Columbicola Ewing is revised and divided into 24 species groups, with descriptions and illustrations provided for the 19 Old World species groups. Eight new species are described; these species and their respective type hosts are: C. browni ex Columba arquatrix Temminck, C. arnoldi ex Macropygia nigrirostris Salvadori, C. galei ex Gymnophaps albertisii Salvadori, C. palmai ex Leucosarcia melanoleuca (Latham), C. mendesi ex Ducula concinna (Wallace), C. reedi ex Ptilinopus magnificus (Temminck), C. davisae ex Treron curvirostra nipalensis (Hodgson), and C. wecksteini ex Ptilinopus rivoli (Prévost). Columbicola juliusriemeri Eichler and Mrosek and C. longiceps sikoraae Eichler are removed from synonymy and recognized as valid species. Columbicola longisetaceus (Piaget) is placed as a junior synonym of C. columbae (L.), C. fradeorum Tendeiro as a junior synonym of C. mjoebergi Eichler, and C. meinertzhageni parvus Tendeiro as a junior synonym of C. meinertzhageni Tendeiro. We provide a key to the 63 valid Old World species, and a table of host associations for the 77 valid Columbicola species of the world.


Introduction
Columbicola Ewing 1929: 116 (Phthiraptera: Ischnocera: Philopteridae) is one of the most speciose genera of chewing lice in the world, with 77 valid species recognized herein. All members of the genus are parasites of pigeons and doves (Columbiformes). Most species of Columbicola are long and slender in shape ( Figure 1). This body form allows the lice to escape from host defense (preening) by inserting themselves between the barbs of the large flight feathers of the host's wings and tail (Nelson andMurray 1971, Clayton 1991). Several unrelated species of Columbicola from Africa and Southeast Asia have wider, rounder bodies and heads. The adaptive significance of this shape, if any, is unknown.
Columbicola is distinguished by 2-3 long metanotal setae on each side (Figures 2 and 3), as well as a bilobed, dorsoanterior head plate (Figure 4), with an associated pair of broad, anterior medial setae (Figure 4a). Abdominal tergites II-IX are medially divided (Figure 1), and most Columbicola show distinct sexual dimorphism. The males of most species are smaller than the females, and most males have an enlarged antennal scape and pronounced spur on the third antennal segment. The spurs enable the male to grasp the female securely around the thorax with his antennae during copulation (Martin 1934).
Like all lice, Columbicola are permanent ectoparasites that pass their entire life cycle on the body of the host. Lice are transmitted to new hosts primarily during periods of direct physical contact, such as that between parent birds and their offspring in the nest (Clayton and Tompkins 1994). Many species of Columbicola are specific to a single genus or even species of host. However, some species of Columbicola are known from several host genera. This lack of specificity may be partly the result of phoresis, i.e. ''hitch-hiking'' on nonspecific hippoboscid flies (Keirans 1975, Clayton et al. 2004). Phoresis may also be responsible for the fact that Columbicola are less host-specific, overall, than several genera of ''body lice'', the other main philopterid lice found on Columbiformes (Johnson et al. 2002). Body lice are seldom observed hitch-hiking on hippoboscid flies (Keirans 1975, unpub data). This simple difference may explain why, in addition to being more host specific than Columbicola, body lice also show less population genetic structure, and less phylogenetic congruence, than Columbicola .
Pigeons and doves and their lice are relatively easy to study both in the field and in captivity. Consequently, the pigeon-Columbicola system has become an important model for research on host-ectoparasite interactions (Clayton 1991, Clayton and Tompkins 1994, 1995. Columbicola lice, like those of other philopterid genera, feed mainly on host feathers. The feather damage they cause has been shown to affect host thermoregulatory ability (Booth et al. 1993), host survival , and host mate choice (Clayton 1990). This impact on host fitness selects for efficient host defensive mechanisms, such as preening. The hosts, in turn, exert reciprocal selection on Columbicola for efficient escape from host defense ), which has recently been shown to reinforce host specificity and cospeciation between Columbicola and their hosts . It goes without saying that studies of specificity and cospeciation depend on a solid alpha taxonomic foundation. The goal of this paper is to expand this foundation for Columbicola.
The framework for Columbicola taxonomy currently in use was first proposed by Hopkins and Clay (1952) in their checklist of chewing lice. Later, Tendeiro (1965) wrote an exhaustive monograph on the genus in which he constructed the first key to Columbicola species of the world. Tendeiro divided the genus into nine species groups based on chaetotaxy, body shape, internal sclerotization, and genitalic structure (a tenth species group was added by Tendeiro (1984)). After 1965, Tendeiro described many additional Columbicola species and is credited with having named over 40% of the known members of the genus (Tendeiro 1967(Tendeiro , 1969(Tendeiro , 1973a(Tendeiro , 1984. Although Tendeiro's work is critical to anyone studying the genus Columbicola, much of it was written in Portuguese and French, making it inaccessible to most workers. Furthermore, many of the features used by Tendeiro to diagnose species groups are difficult to discern and some of these features appear to be convergent adaptations. For these reasons, Tendeiro's keys can be confusing and several of the Columbicola species groups he erected appear to contain unrelated lice. The earliest descriptions by Tendeiro (1965) of species groups for Columbicola were based partly on metanotal setal patterns (Figures 2 and 3) and tenuous features of genitalia (Tendeiro 1984). For example, Tendeiro emphasized the form of the paramere-basal plate connection, which is often difficult to discern. Similar paramere-basal plate arrangements are found in otherwise dissimilar species of Columbicola. By emphasizing such features, Tendeiro's three largest species groups (columbae, passerinae, and gracilicapitis) were amalgamations of unrelated New and Old World species. He later divided these groups into smaller complexes that contained more similar taxa, and which represented more natural species groupings (Tendeiro 1984).
The next review of Columbicola was by Clayton and Price (1999), who examined all of the New World species. They also described five new species and remarked on the apparent species groups. Clayton and Price (1999) included the first English key to a large segment of the genus, i.e. the New World species. The current study is an Old World companion, of sorts, to the Clayton and Price (1999) study. We revise the Old World species of Columbicola using similar criteria to Clayton and Price (1999). In addition to evaluating all of the previously described Old World species, we redescribe the 55 valid ones. We describe and illustrate eight new species and provide a key to the 63 currently valid Old World species of Columbicola. Finally, we provide a table of host associations for the 77 valid species of Columbicola that are currently known.
We divide the Old World Columbicola into 19 species groups, several of which match the species complexes described by Tendeiro (1984). We have attempted to base these groups solely on louse morphology, without consideration of the host associations. Features of particular importance include structure of the dorsoanterior head plate, the metanotal chaetotaxy, and structure and chaetotaxy of both male and female genitalia. Unlike the species groups in Tendeiro (1965), our species groups divide into separate New and Old World assemblages. Nearly all species have been assigned unambiguously to a particular species group, except for four species that were unavailable for examination and whose type descriptions were so vague as to be of little or no use. Two of these species are considered to be nomina dubia. A discussion of each Old World species group precedes the descriptions of its constituent species. When listing the known host genera for each species group, records of questionable validity have been omitted.
For this study, every attempt was made to acquire multiple individuals of each Columbicola species including, whenever possible, the type specimens. When a louse was known from more than one species of host, specimens from all available hosts were sought. More than 1000 lice were examined, representing 60 of the 63 Old World species, including type specimens of 41 of these species. For further details of the material examined see Adams (2002).
Specimens were examined with a Nikon compound microscope equipped with both phase contrast and differential interference contrast capabilities. Setal patterns were recorded and 8-12 measurements (e. g. Figures 5,6) were made using an ocular micrometer. Micro-images of each species were captured using a digital camera, allowing for side by side comparisons of different individuals. Only seven of the 14 strictly New World species were examined directly; however, the detailed descriptions, measurements, and commentary found in Tendeiro (1965) and Clayton and Price (1999) for the other seven species were more than adequate for comparison to the material examined.
In the Description section of the species accounts, all measurements are in millimetres and the abbreviations (Figures 4-6) for measured structures are defined upon first use. For brevity, generic and group features are not repeated in the ensuing species descriptions. When two or more specimens of the same sex were examined, the size range of the structure is given, and, if three or more individuals were measured, the mean is given in parentheses following the range. In a few cases, only a single specimen was available for sphenurus R formosae sphenurus R Ptilinopus occipitalis (*) xavieri Q  (2003). §Host associations from current paper and Price et al. (2003). (*) Indicates type host. A columbae group; B guimaraesi group; C meinertzhageni group; D streptopeliae group; E theresae group; F angustus group; G effeminatus group; H becheti group; I fortis group; J tasmaniensis group; K mjoebergi group; L fradei group; M galei group; N fulmeki group; O veigasimoni group; P palmai group; Q longiceps group; R clayae group; S emersoni group; T baculoides group; U extinctus group; V passerinae group; W gracilicapitis group; X tendeiroi group; Y species group unknown.  5S. roseogrisea (Sundevall), Australia (West Perth Zoo) (1).

Remarks
Tendeiro (1965) considered C. bacillus a subspecies of C. columbae; however, based on morphometric, genitalic, and setal differences, Clayton and Price (1999) recognized C. bacillus as a valid species. Both sexes have, in general, a greater HL/HW ratio than C. columbae. The anterior mesosomal pore of the males is surrounded by only a narrow band of pigmentation compared to the much wider band found on C. columbae. Females have narrower average dorsoanterior plate widths and head widths than C. columbae, as well as fewer subgenital plate setae. Columbicola bacillus is found on numerous species of Streptopelia throughout Europe, the Middle East, Central Asia, and Africa. It may also have been introduced to the Americas on S. decaocto, which was introduced to Florida from the Bahamas in the 1970's, and now has many established breeding pairs in the southeastern United States (Romagosa 2002). Columbicola bacillus may also be periodically ''introduced'' on escaped S. risoria, the domesticated form of S. roseogrisea (Clayton and Price 1999), which, unlike S. decaocto, does not appear to establish viable long term breeding populations.

Remarks
Although C. stresemanni was originally described as a subspecies of C. columbae, the overall shape and dimensions of this louse are much closer to those of C. claviformis, with which it may, in fact, prove to be conspecific. Even with the poor condition of the available specimen, the separation from C. columbae is straightforward, but there is not enough detail visible to determine its status beyond that. For these reasons, this louse is recognized as a valid species; however, additional well preserved specimens will be needed to examine details of the louse chaetotaxy and genitalia. No males were available for this study, so Figure 13 was redrawn from Eichler (1942b), and the measurements were taken from Tendeiro (1965). Although it is in no way direct evidence of the relatedness of their parasites, it is interesting to note that the respective hosts of C. stresemanni and C. claviformis, Columba bollii and C. palumbus L., are also quite closely related (del Hoyo et al. 1997

Remarks
This species is quite similar to C. columbae and C. bacillus, differing in overall body dimensions, as well as the structure of the male genitalia. It is especially close to C. stresemanni, differing subtly in head shape and antennal length. It should be noted that individuals collected from C. palumbus azorica E. Hartert were slightly shorter than those collected from mainland hosts. They also showed a slightly smaller HL/HW ratio and longer subgenital plate setae; however, these differences are so minor that any formal division between these specimens and those from the mainland would be premature.

Remarks
This louse shares long hair-like PMHS with C. mckeani Tendeiro, but differs in its smaller HL/ HW ratio and distinctive genitalia. This species is clearly a member of the columbae group, as indicated by its mesosomal structure and the numerous subgenital plate setae. It has been recorded from several species in the genus Columba in Central Asia and, although no specimens from the type host are available, the specimens examined for this project were compared to the measurements and drawings given in Tendeiro (1965) and found to be identical. Blagoveshtchensky (1951)

Remarks
In nearly all morphometric dimensions, C. keleri is identical to C. turturis, thus requiring genitalic details to confirm identification. The C. keleri mesosome is more elongate with four pores on each side, compared to C. turturis with a shorter mesosome and only three pores on each side. The broad, ovoid subgenital plate groove of C. keleri contrasts with C. turturis having a laterally compressed groove. Columbicola keleri also has a greater number of subgenital plate setae than C. turturis.

Remarks
This species was originally described by Tendeiro (1965) as a complex of 2 subspecies, C. g. guimaraesi and C. g. grandiusculus. Tendeiro (1967) later added a third subspecies, C. g. vitiensis. After examining numerous individuals of each of the three subspecies, the differences between them were found to be quite noticeable, especially in the variation of the female ventral terminalia. For this reason, each of the three subspecies of C. guimaraesi is recognized as a full species. Although there is some size overlap, C. guimaraesi is the smallest of the guimaraesi group lice. Male identification is based on overall body dimensions and subtle differences in structure of the genitalia. The female is more easily identified by subgenital plate differences in the shape of the groove and the length and number of setae. The mesosome of C. guimaraesi has a reduced thickening on its lateral edges compared to C. vitiensis and C. grandiusculus. Individuals from the type host Chalcophaps indica, and those from C. stephani, were found to be indistinguishable, thus representing a new host record for C. guimaraesi.

Remarks
This species was originally described as a subspecies of C. guimaraesi; however, differences in the body dimensions and female ventral terminalia are distinct enough to warrant species recognition. In addition, the male head setae are distinctly longer in C. grandiusculus than in C. guimaraesi. Columbicola grandiusculus can be separated from C. vitiensis by its slightly smaller HL/HW ratio, broader head, genitalia, and by the shape and chaetotaxy of the female subgenital plate.

Remarks
Columbicola vitiensis is morphologically closest to C. grandiusculus. The males can be separated by subtle differences in overall body size, and especially by differences in head dimensions. The females can be separated by overall body dimensions and ventral terminalia shape and chaetotaxy.

meinertzhageni species group
This group consists of four species from the host genera Streptopelia, Treron, Turtur, and Columba. Like the previous two groups, they have anterior marginal head carina complete and rounded ( Figure 35); body elongate ( Figure 1); each side of metanotum with two long, two short setae ( Figure 2). Male antenna with enlarged scape, spur on third segment ( Figure 35); genitalia similar to streptopeliae group, with mesosome laterally divided, each side with three pores and one small spine ( Figure 37). Female subgenital plate groove broadly rounded, each side with 2-5 short to long setae ( Figure 38).

Description
Male head as in Figure

Remarks
This species was originally described by Tendeiro (1959) as a complex of four subspecies: C. m. meinertzhageni, C. m. parvus, C. m. meridionalis, and C. m. longantennatus. The nominate form, C. m. meinertzhageni, was described from three host species: Streptopelia semitorquata, Columba arquatrix Temminck, and Treron calvus delalandii (Bonaparte). Comparison of specimens from Columba arquatrix and the type host, S. semitorquata, indicates that the former are longer overall, with narrower heads, greater HL/HW ratio, greater APW, greater GW, and much longer subgenital plate setae. We therefore elevate these specimens to the status of a new species, Columbicola browni ex Columba arquatrix. We have not been able to study specimens of C. m. meinertzhageni from Treron calvus delalandii. Therefore, we continue to recognize the record of C. m. meinertzhageni from T. c. delalandii.
Comparison of type specimens of the four subspecies of C. meinertzhageni reveals sufficient differences between three of these (C. m. meinertzhageni, C. m. meridionalis, and C. m. longantennatus) to recognize them as full species. In contrast, C. m. parvus ex Turtur chalcospilos is represented by three poorly preserved specimens that Tendeiro separated from C. m. meinertzhageni on the basis of smaller size. Re-examination of these specimens shows considerable overlap with the nominate subspecies; we therefore consider C. m. parvus to be a junior synonym of C. m. meinertzhageni.

Remarks
The individuals used to describe C. browni were originally part of the paratype series for C. meinertzhageni. They are differentiated by larger overall body proportions and, in the case of females, much longer subgenital plate setae. The dimensions of a much smaller male specimen with a constricted dorsoanterior head plate were not included in the above description, despite the fact that the specimen is mounted on the same microslide as the others. The morphology of this specimen places it well within the range of C. meinertzhageni. The fact that two species of lice are mounted on the same slide is evidence of co-occurrence on a single host individual or, perhaps more likely, contamination of lice from different host individuals.

Etymology
This species is named for Paul Brown, The Natural History Museum, London, England, in appreciation of his assistance lending us the large number of specimens required for this study.

Remarks
This species combines the small dimensions of C. meinertzhageni with the longer subgenital plate setae of C. browni.

Remarks
The description of this louse was originally based on a single male specimen that we have also examined. Its genitalia are largely obscured by debris in the body, making detailed study impossible. However, sufficient detail is visible to show that the genitalia are similar to those of C. meinertzhageni, while dimensions of the genitalia and other features of this louse are much larger than those of C. meinertzhageni.

streptopeliae species group
This group consists of four species found on the host genera Streptopelia and Oena.

Description
Male head as in Figure 48; clypeus narrow, lateral edges rounded ( Figure

Remarks
This species, along with C. capicolae, C. oenae, and C. senegalensis, is readily recognized by its unique head shape and ovoid body, in conjunction with a two long, two short metanotal setal pattern. These differences led Clay and Meinertzhagen (1937) to erect the genus Soricella for the subspecies Soricella s. streptopeliae and S. s. capicolae, while recognizing that these taxa were closely related to Columbicola. Hopkins and Clay (1952) moved Soricella into Columbicola because they found that Soricella was ''linked with it by intermediates and cannot be kept separate''. Because of consistent differences in clypeal and head shapes, overall sizes, and some genitalic structures, we recognize each of the four named subspecies as a valid species. These species can be separated by examination of the clypeal region, in conjunction with overall body dimensions. The lone female ex S. decipiens represents a new host record both for the species and species group.

Remarks
This species was originally described as a subspecies of Soricella streptopeliae; however, due to the distinctly wider and rounder clypeal region, C. capicolae is now considered a valid species.

Remarks
Both sexes of C. oenae can be distinguished by their small size, shortened anterior head region, and rounded clypeus ( Figure 57). Males have shorter, stockier parameres than other streptopeliae group lice, whereas females have fewer subgenital plate setae. To date, this is the only streptopeliae group louse not found on doves of the genus Streptopelia.   Tendeiro (1965) based his description of this louse on two males and three females collected from S. senegalensis on the Islands of São Tomé and Príncipe in the Gulf of Guinea. Unfortunately, the original type specimens, which were deposited at the Centro de Zoologia da Junta de Investigaçõ es do Ultramar, Lisbon, Portugal, cannot be located. Further collecting will be necessary in order to compare this louse directly to other streptopeliae group lice. Data in the above description are from Tendeiro (1965), and Figure 61 was redrawn from his figure in order to make identification of C. senegalensis easier. We feel confident of the validity of this species, given the unique shape of its clypeus and the square clypeal indentation. Neither Dickinson (2003)  Material 2 males, 2 females, ex S. senegalensis, India (1). 5 males, 2 females, ex S. capicola, Botswana, Republic of South Africa (2). 7 males, 17 females, ex S. chinensis suratensis (J. F. Gmelin), India (1). 1 male, ex S. vinacea, Ghana (1). 5 males, 4 females, ex S. orientalis meena (Sykes), Nepal (1).

Remarks
Columbicola theresae is one of the most widespread members of the genus, recorded from Streptopelia throughout Africa, the Middle East, and Central Asia. Tendeiro (1965) records C. theresae from Oena capensis, S. decipiens, and S. tranquebarica. The specimens we examined from S. orientalis and S. vinacea represent new host records for this species. The shape of the male mesosome and the pattern of its pores are distinctive. The female subgenital plate groove often, but not always, shows small irregular indentations along its lateral edges.

Remarks
The description of C. deboomi was based on a single male specimen. The elongate mesosome, in conjunction with the arrangement of its pores, is unique. Ventral terminalia as in Figure 69; subgenital plate groove narrow, oval anteriorly, widening posteriorly with smooth lateral edges, and without lateral setae. TL, 2.69-3.01 (2.83).

Remarks
The broad, triangular mesosome of the male genitalia is instantly recognizable. Unfortunately, the female is much more difficult to identify using the structure of the subgenital plate and overall body dimensions. The presence of small subgenital plate setae on some C. hoogstraali Tendeiro, in conjunction with non-overlapping host ranges, sometimes allows the females of these two louse species to be separated, but this is not always possible. Tendeiro (1965) noted subtle differences in the shapes of the clypeal region and in the overall length of male C. orientalis from the two (allopatric) host species. However, our study of additional specimens shows sufficient overlap in these traits to convince us that lice from the two hosts are, in fact, conspecific. Material 4 males, 4 females (including holotype male, allotype female of C. hoogstraali), ex S. picturata, Madagascar, Reunion Island (3).

Remarks
The mesosome of the male C. hoogstraali is distinctive, whereas the subgenital plate groove of the female is nearly identical to those of C. theresae and C. orientalis. The female C. hoogstraali can be tenuously recognized using dimensions of the head and by the existence of several minute lateral subgenital plate setae. Tendeiro (1965) originally placed this species within the columbae subgroup. However, the shape of the mesosome and its serrated, sclerotized edges, and the lack of obvious subgenital plate setae, make this species more like other members of the theresae group. Ventral terminalia as in Figure 73; subgenital plate groove bell-shaped, narrow anteriorly, broad posteriorly; one individual with 3 short lateral subgenital plate setae (0.010), others without such. TL, 2.55-2. 62 (2.58).

Remarks
The shape and pattern of the male mesosome are unique. The bell-shaped subgenital plate groove of the female is shared with C. carrikeri Tendeiro, but the anterior portion is narrower in C. cicchinoi. We are suspicious of the C. cicchinoi record from Thailand, which is outside the range of the type host, S. bitorquata: western islands of Indonesia and the Philippines (del Hoyo et al. 1997). Although Tendeiro recorded a male as the holotype, one of the females available for examination was accidentally mislabeled as such, and, after checking with Tendeiro's original description, this individual is now listed as part of the paratype series. Material 7 males, 3 females (including holotype male, allotype female, 4 male, 1 female paratypes of C. carrikeri), ex T. chalcospilos, Kenya, Zimbabwe, Mozambique (3). 3 male, 5 female paratypes of C. carrikeri, ex T. abyssinicus (Sharpe), Uganda (1). 1 male, ex T. tympanistria (Temminck), Ghana (1).

Remarks
Males of C. carrikeri can be separated from C. theresae by the shape of the mesosome and parameres and the arrangement of the mesosomal pores. Although similarly shaped, the anterior portion of the female subgenital plate groove is wider in C. carrikeri than C. cicchinoi.

angustus species group
This group consists of six species from the host genera Phaps, Macropygia, Gallicolumba, Reinwardtoena, and Ocyphaps. They have the anterior marginal head carina complete, either rounded ( Figure 81) or indented ( Figure 78); body elongate ( Figure 1); each side of metanotum with two long, two short setae (Figure 2). Male antenna with enlarged scape, spur on third segment; mesosome triangular, tightly wedged between long, thin parameres, with two or four pores (Figures 76, 82

Remarks
The genital structure of both sexes is distinctive. It is worth noting that the type host, P. chalcoptera, is known to harbor two different species of Columbicola: C. angustus from western Australia and C. tasmaniensis Tendeiro from southeastern Australia. The degree of range overlap between these two species, if any, is unknown. ventral row of small setae anterior to mandibles. Thorax with PW, 0.23; MW, 0.29. Genitalia as in Figure 80; parameres long and straight; mesosome triangular, with 2 pores; GW, 0. 088 47.

Remarks
Tendeiro (1984) based his description on two males and two females. Although the genitalia are clearly visible in the male (holotype), the ventral terminalia, including the subgenital plate, are obscured in the female specimen we examined. We are therefore unable to confirm Tendeiro's (1984) description of the subgenital plate as having a narrow posterior indentation, widening slightly posteriorly. However, the broad head shape, in conjunction with the indented dorsoanterior head plate, and row of setae above the mandibles in the male, should make identification of this species straightforward.  Material 1 male, ex R. reinwardtii griseotincta E. Hartert, New Guinea (1).

Remarks
The male of C. taschenbergi is one of the largest in the genus. Its size, low HL/HW ratio, long setae, and mesosomal structure make identification of C. taschenbergi straightforward. Thorax with PW, 0.22-0.23 (0.223); MW, 0. 28-0.31 (0.300

Remarks
The holotype of C. exilicornis is a female, recorded from an unidentified species of tern (Sterna sp.), which presumably represents an erroneous host record. Other specimens have been collected from several species of Macropygia, and additional collecting has yielded this species on additional host genera, particularly in the Philippines. For example, Tendeiro (1984) recorded a pair of C. exilicornis on Gallicolumba jobiensis (A. B. Meyer). Specimens we examined from G. striata and P. amethystinus both proved to be new host records. Columbicola exilicornis shares features with the new species C. arnoldi, such as the medially expanded anterior head carina and thickened sagittal band. The males of these species can be readily separated by details of the mesosome; however, the females are currently inseparable. Tendeiro (1965) considered C. juliusriemeri Eichler and Mrosek to be a synonym of C. exilicornis, despite never having studied the single male (holotype) specimen from which the former species was described (the holotype has apparently been lost). We feel Tendeiro's action was premature, and that C. juliusriemeri must retain species status pending the acquisition and study of additional specimens. Drawings of the holotype by Eichler and Mrosek, although minimal in detail, show some possibly informative characters (see C. juliusriemeri account later in this revision).

Remarks
Although quite similar to C. exilicornis, the distinct tongue-like extension of the mesosome is unique among lice in the angustus group. Its occurrence on two separate hosts collected years apart convinces us of the validity of this species. The females are currently indistinguishable.

Etymology
This species is named for Don C. Arnold, Survey Entomologist, Oklahoma State University, in appreciation for his assistance with the loan of large numbers of lice for this study.

Remarks
The males of this species are the only Columbicola among those with two long and two short metathoracic setae that do not have an enlarged scape or spur on the third antennal segment. The genitalia of both sexes are unique. The lack of an enlarged male scape may be associated with the reasonably close overall size of the two sexes. The females in many other species of Columbicola are distinctly larger than the males and, when they mate, the male slides under the female, grasping her around the thorax with his antennae. A male C. effeminatus would probably not be able to grasp a female with its filiform antennae, implying a different mating strategy for these lice.

becheti species group
This group contains a single species from the host genus Ducula. It has both sexes with thickened sagittal band posterior to dorsoanterior head plate (Figures 98, 100)  Material 4 males, 4 females (including holotype male, allotype female, 1 male, 1 female paratypes of C. becheti), ex D. goliath, New Caledonia (2).

Remarks
The very large size of this louse, in conjunction with the structure of the genitalia and the male antenna, is distinctive. Despite having only two long metathoracic setae, the existence of small lateral sclerites around the mesosome hints at a distant relationship between this louse and those in the longiceps group.

fortis species group
This group contains a single species from the host genus Otidiphaps. It has a very broad head, with both anterior and posterior thickenings (Figures 102, 103); body oblong; each side of metanotum with two long, two short setae (Figure 2

Remarks
Tendeiro (1965) was unable to study this species directly because the type specimens were deposited in the collection of the University of Halle and were almost certainly destroyed in World War II. The species was later redescribed and illustrated by Emerson and Price (1979); it is easily recognized by the structure of the head and genitalia.

Remarks
The distinctive genitalia of both sexes make identification of C. tasmaniensis straightforward. No other species of Columbicola shows such asymmetry of the male genitalia; the band-like subgenital plate of the female is also unique. Some aspects of the mesosomal structure, such as the shape of the anterior portion and arrangement of the pores, may indicate a distant relationship between this louse and those in the mjoebergi group.
All of the specimens we examined were from southeastern Australia or Tasmania, although both host species are more widely distributed. Interestingly, P. chalcoptera is parasitized by C. angustus in Western Australia, suggesting geographic specificity by different species of Columbicola on a single widespread host species.

mjoebergi species group
The two species in this group are both found on the host genus Geopelia. They have the anterior head carina rounded, complete; body elongate ( Figure 1); each side of metanotum with two long, two short setae (Figure 2). Male antenna with enlarged scape, spur on third segment ( Figure 108); mesosome rectangular, lateral edges thickened, each side with three pores (Figure 110). Female subgenital plate groove broad, elongate; without lateral setae (Figures 111, 113).

Remarks
Males of C. mjoebergi are distinguished by their unique mesosomal structure; females are distinguished by the pentagonal shape of the subgenital plate groove, and by the lack of lateral subgenital plate setae. Tendeiro (1973b) described the species C. fradeorum from a single male taken from S. chinensis tigrina, along with several specimens of C. fulmeki Eichler, the normal species found on this host. Although Eichler provided an adequate description of C. fradeorum, he apparently did not compare it directly to other species of Columbicola. Our comparison of the holotype to several C. mjoebergi specimens indicates no difference in gross morphology or genitalic structure, leading us to synonymize C. fradeorum with C. mjoebergi. Collection of the single specimen from S. chinensis is not altogether surprising, given that S. chinensis and G. striata are both often kept in captivity in Thailand. Straggling of lice between species of captive birds is known to be somewhat common.
Columbicola mjoebergi may also prove to be conspecific with C. timorensis Tendeiro. However, the limited number and poor quality of C. timorensis specimens prevents us from drawing a firm conclusion, as discussed below.

Remarks
Columbicola timorensis was originally described by Tendeiro (1979) as a subspecies of C. turturis, but the rationale for this classification was never given. Despite the poor condition of the two available specimens, C. timorensis is smaller, has a shorter subgenital plate groove, and lacks subgenital plate setae. Hence, C. timorensis is distinct from C. turturis and merits consideration as a different species.
In contrast, the relationship of C. timorensis to C. mjoebergi is unclear. The specimens of C. timorensis are poorly prepared and contain a good deal of debris that obscures the terminalia. Overall, C. timorensis and C. mjoebergi are quite similar morphologically and, although the C. timorensis subgenital plate groove appears low and broad, this may be an artifact of preparation. Additional collecting and future examination of specimens may well show these forms to be conspecific. However, until such specimens are available, the most appropriate course of action is to continue recognizing the two forms as separate species.

fradei species group
The single species of this group is from the host genus Columba. It has the anterior head carina rounded, complete; body elongate (

Remarks
Columbicola fradei has been recorded from both of the disjunct East and West African populations of the type host, C. larvata. The ovoid mesosome of the male genitalia is distinctive. The wide, slightly peaked subgenital plate groove of the female, together with the lack of subgenital plate setae, and overall body dimensions, should allow for reliable identification.

galei species group
The single species of this group is from the host genus Gymnophaps. It has the anterior marginal head carina rounded, complete (Figures 116, 117); body elongate; each side of metanotum with two long, two short setae (Figure 2). Male antenna with enlarged scape, spur on third segment ( Figure 116); mesosome broadly triangular with four pointed anterior projections; no pores ( Figure 118). Female subgenital plate groove elongate, smoothly rounded, each side with 5-7 medium to long (0.017-0.024) setae ( Figure 119).

Description
Male head as in Figure 116;

Remarks
The unique mesosomal structure of the male, in conjunction with the elongate subgenital plate groove and long lateral setae of the female, should make C. galei readily identifiable. The relationship of this species to other species groups is unclear.

Etymology
This species is named for Jon H. Gale, Animal Facility Supervisor, University of Utah, in appreciation of his invaluable assistance with many host-parasite related projects.
14. fulmeki species group The single species in this group is from the host genera Streptopelia and Geopelia. It has the anterior marginal head carina rounded, complete; body elongate; each side of metanotum with two long, two short setae (Figure 2

Remarks
The male genitalia of C. fulmeki are unique, consisting of several apparent components between the long, thin parameres. The female subgenital plate groove is also distinctive, being broadly rounded with a small anterior extension, and bordered by a number of minute setae (often difficult to discern). Tendeiro (1979) listed multiple records of this louse from G. maugei, and additional specimens were recently recorded from S. orientalis. The latter represents a new host record for C. fulmeki.

veigasimoni species group
The single species of this group is from the host genus Phapitreron. It has the anterior marginal head carina rounded, complete (Figures 122, 124); thickened region immediately posterior to dorsoanterior head plate; each side of metanotum with two long, two short setae (Figure 2). Male antenna with enlarged scape, spur on third segment ( Figure 122); mesosome laterally spread, each side with two pores; parameres long, narrow ( Figure 123). Female subgenital plate groove broadly rounded, with slight posterior constriction; each side with 0-2 minute (0.005-0.008) setae ( Figure 125).

Remarks
The description of C. veigasimoni was based on a single male from P. amethystinus; an additional male from P. leucotis was later described by Tendeiro (1969). The shape of the mesosome is unique among Columbicola having the two long, two short metanotal setal pattern. Until now, however, there has been no description of the female. The female is similar to the male, showing many of the same structural features of the head, such as the broad width and thickened band just posterior to the dorsoanterior plate. The female is also distinctly larger than the male. Its subgenital plate groove is wide, with a slightly angled anterior edge. Of the two female specimens, one had two pairs of lateral subgenital plate setae, while the other had none.

palmai species group
The single species of this group is from the host genus Leucosarcia. It has the anterior marginal carina rounded, complete; body elongate; each side of metanotum with three long, one short setae (Figure 3). Male antenna with enlarged scape, spur on third segment (Fig 126); mesosome narrow, with lateral portions curving back towards the midline; each side with two pores (Figure 127). Female subgenital plate groove broad, anteriorly narrowed and evenly rounded, much as in Figure 125; each side with 3-5 long setae (0.020-0.027).

Description
Male head as in Figure 126;

Remarks
Columbicola palmai can be identified by its overall size, uniquely placed medioposterior head setae and distinctive male genitalia. The long female subgenital plate setae and shape of the groove, in conjunction with three long metanotal setae, distinguish female specimens of C. palmai from other Columbicola. Figure 128 was drawn from the allotype female, which unfortunately was overcleared when prepared as a specimen. Recently collected specimens revealed further details of female morphology and chaetotaxy.

Remarks
Tendeiro (1965,1979,1984) recorded C. longiceps from 12 different host species, nearly all in the genus Ducula. Due to their large size and unique male genitalic structure, we recognize specimens from D. concinna (Wallace) and D. rosacea (Temminck) as the separate new species C. mendesi. We examined specimens from four additional host species and, even though some differences were found among their lice, the limited number of specimens and subtlety of the differences does not warrant separation into different species. An additional difficulty is that we were unable to obtain specimens from the type host D. perspicillata. Although some measurements from such specimens are available in Tendeiro (1965), there is not enough information to allow us to do a thorough evaluation. Further collecting may reveal additional species level differences among the C. longiceps populations on different Ducula species. Eichler (1952b) proposed a new genus and species, Parasoricella wolffhuegeli ex Ducula luctuosa, as a form intermediate between Columbicola and the then recognized genus Soricella. Unfortunately, the description of this louse was inadequate and no illustrations were provided. Hopkins and Clay (1953) placed P. wolffhuegeli in Columbicola after they synonymized Soricella with Columbicola. Based on its known host and the few details of head shape available, Tendeiro (1965) synonymized C. wolffhuegeli with C. longiceps. While synonymizing a species without detailed examination is impertinent, Tendeiro is probably correct in his placement of C. wolffhuegeli with C. longiceps. In this situation, we believe it appropriate to follow Tendeiro's lead. Columbicola longiceps is recognized by its size, elongate shape, indented anterior head margin, and genitalic structure. In addition to the normal Ducula hosts, Tendeiro (1979) recorded single individuals of C. longiceps from Treron phoenicopterus (Latham) and Streptopelia chinensis tigrina.

Remarks
Columbicola mendesi is quite similar to C. longiceps, but is distinctly larger, with the male genitalia having a narrower mesosome and thicker transverse sclerites. The Columbicola from D. concinna and D. rosacea were originally described by Tendeiro (1965) as part of the C. longiceps complex; however, the differences between these Columbicola and C. longiceps from other hosts are sufficient to warrant species level recognition. Although no individuals from D. rosacea were available for study, measurements of 1 female and 2 males, together with a photo of the male genitalia, are provided in Tendeiro (1965). These individuals were from hosts on the Island of Flores, Indonesia; they are identical in size and male genitalic structure to the C. mendesi found on D. concinna from the Kei Isles.

Etymology
This species is named for Luis F. Mendes, Instituto de Investigação Científica Tropical, Lisbon, Portugal, in appreciation for his invaluable assistance with the acquisition of rare and long missing specimens.

Remarks
Like C. longiceps, C. cavifrons is known from several forms of Southeast Asian Ducula and it, too, may represent a species complex. Taschenberg (1882) listed both D. aenea and D. badia as type hosts in his description of C. cavifrons, which was based on specimens from both of these host species. Years later, Eichler (1942b) described C. longiceps sikoraae from D. b. badia, which was later synonymized with C. cavifrons by Tendeiro (1965). Our examination of specimens from the two type hosts reveals consistent male and female genitalic differences. These differences warrant recognition of the lice from these hosts as different species: C. cavifrons and C. sikoraae Eichler (see below). However, this creates a problem concerning identity of the lice originally described by Taschenberg as C. cavifrons, since his original material is unavailable for study. The two host species have broadly overlapping ranges, and one cannot assume isolation between C. sikoraae on D. badia and C. cavifrons on D. aenea. For this reason, we recognize both D. aenea and D. badia as possible hosts of C. cavifrons. Tendeiro (1965Tendeiro ( , 1984 recorded an additional six species of Ducula as hosts of C. cavifrons. Unfortunately, we were unable to acquire specimens of C. cavifrons from these other hosts. Since Tendeiro's descriptions and measurements for the specimens he examined are quite similar to C. cavifrons, we continue to recognize them as members of this species. C. cavifrons can be identified by its size, by the structure of the anterior head margin, and by the shape of the male mesosome and female subgenital plate groove. Tendeiro (1979) also recorded a single C. cavifrons female from S. chinensis tigrina. Material 3 males, 6 females, ex D. b. badia, Malaysia (2). 11 males, 13 females, ex D. b. griseicapilla Walden, Thailand (8).

Remarks
Eichler (1942b) originally described C. sikoraae as a subspecies of C. longiceps. He based this solely on host association, ignoring the fact that Taschenberg (1882) had already described C. cavifrons from this host. Hopkins and Clay (1952) later recognized both C. cavifrons and C. sikoraae as full species; however, Tendeiro (1965) synonymized C. sikoraae with C. cavifrons. After examining many specimens from multiple hosts, we are elevating C. sikoraae to full species status. Columbicola sikoraae consistently shows a much narrower male mesosome, as well as a different shaped female subgenital plate groove, compared to C. cavifrons. Material 1 male, 1 female (including holotype male of C. xavieri), ex P. occipitalis, Philippines (2).

Remarks
Columbicola xavieri was described from a single male. Although similar to other members of the longiceps group, C. xavieri is distinctly smaller with a unique genitalic structure. The above description is the first for a female C. xavieri. Unfortunately, debris within the body prevents study of the details of this specimen. The female was identified based on a shared host and similarities in the internal sclerotization of the head between this louse and the holotype male. The shape of the subgenital plate groove is most similar to C. sikoraae, but the groove is narrower in C. xavieri. (Figures 144-147) Columbicola cavifrons harrisoni Tendeiro 1965: 354. Type host: Ducula pinon jobiensis (Schlegel).

Remarks
Tendeiro (1965) described C. harrisoni from two host species, Ptilinopus magnificus (Temminck) and D. pinon. Based on consistent differences in both the male and female genitalia, the specimens from the two hosts are considered to represent distinct species: C. harrisoni ex D. pinon and C. reedi ex P. magnificus. Columbicola harrisoni was originally described as a subspecies of C. cavifrons; however, Tendeiro (1984) elevated C. harrisoni to full species status.

Description
Male head as in Figure

Remarks
Females of C. paradoxus are unknown and, of the three male specimens in existence, one is missing its head. This species is similar to C. gourae, yet differs in having a more elongate preantennal region, higher HL/HW ratio, and different genitalia. Tendeiro (1965) originally placed C. paradoxus in its own species group, based on medial placement of the spiracles and the novel host association. However, the similarity of head, setal, and genitalic structures between C. paradoxus and other members of the longiceps species group is striking. Although spiracles on the available C. paradoxus specimens are indeed closer to the midline of the body than in some other Columbicola, they are close to the position of the spiracles in members of the longiceps group. Indeed, these similarities were remarked upon by Tendeiro (1984), who believed that C. paradoxus arose from within the longiceps group and should be considered part of it. We agree, and have placed C. paradoxus in the longiceps group.
This record of C. paradoxus, from what we consider to be an erroneous noncolumbiform type host, is based on three specimens of lice from a single Figbird (Sphecotheres vieilloti) collected at an unknown locality. Figbirds are passeriform songbirds (Oriolidae) that are unrelated to Columbiformes. In recent years the third author (DHC) and colleagues have collected over a dozen Figbirds, as well as specimens of other members of the Oriolidae, at localities in Northern Australia and Queensland. All of these birds were thoroughly checked for lice and, although some were infested, none had Columbicola (unpub data).
Figbirds are common birds that are often present in mixed species foraging flocks with Ducula fruit pigeons and Ptilinopus fruit doves in Northern Australia (DHC pers obs). The erroneous host record could be the result of contamination during an early collecting trip. Alternatively, it may be a case of lice ''straggling'' onto Figbirds by phoretic dispersal on hippoboscid flies (see introduction). Regardless of the source of the error, the true host of C. paradoxus remains a mystery, pending the collection of additional material.

Description
Male head as in Figure 158

Remarks
Columbicola clayae can be identified by its distinctive genitalic structure. To date, it is the only member of this complex known to occur on either the African mainland or the Arabian Peninsula. Although Tendeiro (1960Tendeiro ( , 1965 recorded C. clayae from several subspecies of Treron ''australis '', both Dickinson (2003) anddel Hoyo et al. (1997) split this host species into T. calvus, with many subspecies on the African mainland and Arabian Peninsula, and T. australis (L.), which is restricted to Madagascar and its neighboring islands. All records of C. clayae from T. ''australis'' are, in fact, from subspecies of T. calvus, and there are no known records of Columbicola from T. australis (sensu stricto). Small numbers of C. clayae were also recorded from Oena c. capensis by Tendeiro (1960) and Streptopelia s. semitorquata by Tendeiro (1965).

Remarks
All of the paratypes of C. davisae were originally part of the paratype series for C. elbeli Tendeiro. Both sexes of C. davisae can be distinguished from C. elbeli by details of the genitalia. The abrupt curvature of the posterior portion of the parameres, broadly rounded triangular mesosome, and lack of a ventral mesosomal protuberance are distinctive. The ventral terminalia of the female is most similar to that of C. clayae. However, the posterior edges of the subgenital plate groove are nearly parallel, in contrast to the angled edges of C. clayae.

Etymology
This species is named for Monika Davis, Monterey, California, in great appreciation for her continuous support and assistance with this project.

Remarks
Columbicola insularis is known only from a pair of females and a single nymph collected in 1954. Although Tendeiro recorded the institutions in which these specimens were to be deposited, their current location is unknown. For this reason, all measurements and re-drawings were taken from Tendeiro (1965). Normally, splitting a species based on such a small number of specimens, especially specimens unseen, would be unwise. However, the differences between C. insularis and C. clayae are clear enough that the former can be recognized as distinct from the latter. Nonetheless, without additional specimens to examine, the position of C. insularis in relation to other species of Columbicola is uncertain.

Remarks
Males of C. elbeli can be identified by the smooth curve of the parameres, in conjunction with the structure of the mesosome. Females are indistinguishable from C. sphenurus. Tendeiro (1965) originally described C. elbeli from eight host species. Since then, three species have been split off: C. phoenicopterae Tendeiro, C. sphenurus Tendeiro, and C. davisae. Columbicola elbeli specimens from the remaining host species vary morphologically, although the variation is subtle and overlap is common. However, future collecting of additional material could reveal that C. elbeli is still a complex of closely related species and subspecies.
Columbicola elbeli was originally known only from species of Southeast Asian Treron. Recently, however, a series of C. elbeli collected in Malaysia from Ptilinopus jambu has come to light. This series is part of the K. C. Emerson Collection (OSU) and was the source of an erroneous record of C. emersoni Tendeiro from P. jambu (Tendeiro 1965). We compared the genitalia, chaetotaxy, and overall dimensions of these lice to other C. elbeli specimens and found them to be indistinguishable. The host data label for these specimens is incomplete and poorly written, bringing the validity of the record into some question. Nevertheless, we have chosen to recognize the record as legitimate.

Remarks
Columbicola phoenicopterae was initially described as a subspecies of C. elbeli and, because an individual male from T. pompadora phayrei was mounted along with three male and three female specimens of C. elbeli, it was listed as part of the C. e. elbeli paratype series. Tendeiro (1984) later elevated C. e. phoenicopterae to species status and, while re-examining the specimens, he identified this individual. Columbicola phoenicopterae can be distinguished from C. clayae by the structure of the mesosome and by the broader, rounder subgenital plate groove. Description Similar to C. elbeli, differing in structure of male genitalia. Male with HW,HL,; HL/HW, 1.96. Thorax with PW,MW,. Genitalia as in Figure 171; parameres elongate, thickened anteriorly, with small lateral indentations; mesosome thick, ''V' indented anteriorly;GW,0.113. TL,HL,; HL/HW, 1.87-2.00 (1.96). Thorax with PW,MW,. Subgenital plate groove ovoid, edges rough. TL,.

Remarks
Several of the specimens used to describe C. sphenurus were originally part of the type series for Columbicola elbeli. C. sphenurus was later recognized as a separate species because of differences in the male genitalia of lice from T. pompadora, the type host for C. elbeli, versus lice from T. sphenurus, the type host of C. sphenurus. Tendeiro (1984) also recorded C. sphenurus from T. formosae Swinhoe. Measurements for the description above, except genitalic width, were taken from three males and seven females in Tendeiro (1984

Remarks
While superficially resembling C. elbeli, the male mesosomal and female subgenital plate structures distinguish C. wardi. It is interesting to note that, upon re-examining the type series of C. e. elbeli, Tendeiro (1984) discovered that the male specimen from Treron oxyurus (Temminck) was actually identical to C. wardi, which he had described nearly 20 years earlier.

Remarks
Columbicola emersoni is widespread on members of the genus Ptilinopus. The structure of the male genitalic mesosome is distinctive. The shape of the female subgenital plate groove varies, with the roughly triangular pattern being consistent, but the exact shape and texture of the lateral edges of the plate groove being more variable. Much like C. elbeli and C. longiceps, these differences in overall morphology are most apparent when comparing C. emersoni specimens from different host species. This is particularly evident when dealing with the single male from P. tannensis, which is a new host record for C. emersoni. Although most other C. emersoni males examined for this paper, as well as those recorded in Tendeiro (1965), are well under 2.15 in length, this individual was substantially larger at 2. 35. In every other respect this specimen is identical to C. emersoni males from other hosts. Additional collecting may reveal that these size differences are consistent between populations on different host species, in which case the lice from P. tannensis may eventually be recognized as a different species.

Remarks
Columbicola wecksteini is similar to C. emersoni; however, males are easily distinguished by their reduced scape and unique mesosomal structure. The anterior portion of the female subgenital plate groove is rounder and broader in C. wecksteini than in C. emersoni. Due to the limited number of female C. emersoni specimens available for study, we used the measurements given in Tendeiro (1965) for comparing the two species. According to these measurements, C. wecksteini is distinctly longer than C. emersoni.

Etymology
This species is named for Jason D. Weckstein, The Field Museum, Chicago, in recognition of his work with avian lice and in appreciation for his assistance with the collection of ectoparasites for numerous projects.

Remarks
Columbicola curtus was originally designated as a subspecies of C. emersoni; however, the differences in head shape are distinctive enough to warrant recognition of C. curtus as a different species. The uniquely broad, anteriorly indented head is clearly visible, not just on the adults but also on the two nymphs mounted with them. The quality of these mounts is poor, with many fine details obscured. The female antennae are broken off at the scape, and the head is slightly twisted, preventing a reliable measurement of head width. Despite these problems, we have attempted to illustrate the female head as accurately as possible ( Figure 183).  Figure 186; subgenital plate groove long, rounded anteriorly, each side with 1-3 medium setae (0.010-0.012). TL, 2. 52-2.65 (2.58).

Remarks
Although no specimens were available from the type host, the specimens we examined from A. sganzini were identical to the measurements, photos, and figures of C. brygooi provided in Tendeiro (1967). Hence, this account represents the first description of the female C. brygooi, as well as a new host record for this species.

Unknown species group
Specimens of the following four species are unavailable or useless. Furthermore, the descriptions for these four species are insufficient to allow determination, and the type host is certainly incorrect in the case of the final two species. (Figures 187, 188) Columbicola juliusriemeri Eichler and Mrosek 1958: 140. Type host: Turacoena manadensis (Quoy and Gaimard).

Description
Male head as in Figure 187, apparently lacking posterior medial setae. Genitalia as in Figure 188; mesosome triangular, parameres outwardly expanded before curving posteriorly toward mesosome. Female unknown.

Remarks
This louse was originally described from a single male collected from T. manadensis on the Island of Peleng. The location of the holotype is unknown. Based solely on Eichler's drawings, Tendeiro (1965) placed C. juliusriemeri as a junior synonym of C. exilicornis. This was done despite differences in the shape of the parameres and the uncertainty of various other features, such as the arrangement of the metanotal setae and the questionable existence of a medially swollen marginal carina. Until additional specimens are collected and studied, the most prudent course is to recognize C. juliusriemeri as a valid species. Figures 187 and 188 were redrawn from Eichler and Mrosek (1958).

Description
Male head as in Figure 189 Tendeiro (1980) described this louse based on a male and female pair of poorly mounted specimens from Muana, the Democratic Republic of the Congo. They were recorded as being deposited in the Royal Museum of Central Africa, Tevuren, Belgium, but we were unable to obtain them for study. All measurements were taken from Tendeiro (1980). and ''obliteration'' of the transverse anterior suture by the dorsoanterior head plate. The genitalia are poorly defined, with the only description of the male genitalia commenting on its elongate parameres, mesosome, and sclerotized lateral structures. The female subgenital plate groove was described as elongate, narrow anteriorly, and widening posteriorly.

Remarks
Columbicola fradei has also been recorded from C. larvata, but it differs in head chaetotaxy, as well as the structure of the genitalia. Tendeiro (1980) originally placed C. obliteratus in the columbae group, but he subsequently left it out of the columbae complex (Tendeiro 1984

Remarks
According to Hopkins and Clay (1952), the type specimen of this species is an unidentifiable Columbicola nymph. We therefore consider this name to be a nomen dubium.

Remarks
The series from which this louse was described has been lost and, unfortunately, neither the figure nor the written description is adequate to allow conclusive identification. For this reason, we support Tendeiro's (1965) declaration of this species as a nomen dubium.

Discussion
The 77 members of the genus Columbicola are divided into 24 species groups based solely on louse morphology. Host associations are shown in Table I. The structure of the male mesosome is the most consistently informative character in the designation of the species groups. Within each group, the basic morphology of the mesosome is fairly uniform and rarely shows overlap between groups. Variation in the shape and chaetotaxy of the female subgenital plate groove also provides support for species group assignments, as do the arrangement of the metanotal setae and body shape and size.
Columbicola species groups are found in more or less restricted geographic regions, with no group containing both New and Old World species. The only exceptions involve Old World species such as C. columbae, C. fulmeki, and C. bacillus, which have spread to the New World on introduced Old World hosts. The region with the greatest diversity of Columbicola species is Australasia and, more specifically, locations in and around New Guinea. This pattern makes sense because New Guinea has the greatest diversity of columbiform genera and subfamilies on earth (Gibbs et al. 2001). Adams (2002) provides a table that summarizes the geographic distributions of all Columbicola hosts, which can also be determined from del Hoyo et al. (1997) or Gibbs et al. (2001).
When Columbicola species groups are examined relative to their geographic range and the host genera they contain, several patterns emerge. First, geographically widespread species groups, such as the columbae, theresae, and emersoni groups, are usually restricted to a few closely related host genera. This pattern might be expected for species groups with long histories of association with particular groups of birds .
At the opposite end of the spectrum are species groups, such as the angustus and extinctus groups, that are more geographically restricted, but associated with a wider variety of host genera. For example, members of the angustus group are found on a variety of genera of Southeast Asian and Australasian Columbiformes. Interestingly, there is also a fairly extensive record of angustus group lice being collected from non-columbiform birds, although they likely are not true parasites of these birds. These records may be cases of straggling via phoretic dispersal on hippoboscid flies. Although there is no direct evidence that phoretic dispersal can lead to establishment on a different host species, it is worth noting that Keirans (1975) recorded over 400 instances of lice attached to hippoboscids. According to Maa (1963), over half of all known species of avian hippoboscids (55 of 97) are found in the Oriental and Australasian regions.
Another interesting pattern involves host species with more than one species of Columbicola. For example, many species in the genus Streptopelia have records of two, if not three, species of Columbicola. In such cases, the lice are almost always from different species groups. Whether these lice can coexist on a single host individual or population for an extended period of time is unknown. A similar situation has been shown for the New World dove genus Leptotila, on which multiple species of Columbicola have been recorded from each of several dove species.
In some cases, an individual species of louse is regularly recorded from more than one genus of host (e.g., C. macrourae). A study of the structure of these populations may show that both host association and geography are important components in the development of genetic differentiation, a process that may eventually lead to full reproductive isolation. Future work may show similar patterns of divergence in some of the more widespread Old World lice, such as C. longiceps and C. emersoni.
Ten of the Columbicola species groups are monotypic, and these show several distinct patterns as well. Five of these groups are on monotypic host genera, whereas two others are on genera that are quite small and are endemic to islands. The remaining three monotypic species groups are on host species that have either a very limited distribution (C. becheti) or have been recorded on only a small portion of the known host range (C. fulmeki and C. tasmaniensis).
Currently, Columbicola records are lacking for about half the pigeon and dove species in the world. There are also nine genera of Columbidae from which we have no records of Columbicola. These genera are concentrated around Australia, New Guinea, and the South Pacific Islands, although one genus, Uropelia, is found in central South America. Future research may reveal that many of the more widespread and less host specific lice are actually complexes of genetically divergent but morphologically similar species. Additional collecting and the combined use of morphometric and genetic techniques will provide further clarification of both the evolutionary history and taxonomy of these insects.