BRITIS! m Bulletin of the British Museum (Natural Histo P^r 9 — • British Museum (Natural History) London 1985 Dates of publication of the parts Nol No 2 No 3 No 4 No 5 No 6 No 7 . 28 June 1984 . 26 July 1984 .30 August 1984 . 27 September 1984 25 October 1984 . 29 November 1984 20 December 1984 ISSN 0007-1498 Printed in Great Britain by Henry Ling Ltd, at the Dorset Press, Dorchester, Dorset Contents Zoology Volume 47 No 1 Miscellanea Page A revision of the genera Trachelostyla and Gonostomum (Ciliophora, Hypotrichida), including the redescriptions of T. pediculiformis (Cohn, 1866) and T. caudataKahl 1932. ByM. Maeda&P. Carey 1 Notes on Atlantic and other Asteroidea. 4. Families Poraniidae and Aster- opseidae. By Ailsa M. Clark 19 The larval and post-larval development of the Thumb-nail Crab, Thia scutellata (Fabricius), (Decapoda: Brachyura). By R. W. Ingle . . 53 Description of a new species of Sylvisorex (Insectivora: Soricidae) from Tanzania. By P. D. Jenkins 65 A new species of the Hipposideros bicolor group (Chiroptera: Hipposideridae) from Thailand. By J. E. Hill & S. Yenbutra 77 No 2 A review of the Mastacembeloidei, a suborder of synbranchiform teleost fishes Part II: Phylogenetic analysis. By Robert A. Travers .... 83 No 3 A review of the anatomy, taxonomy, phylogeny and biogeography of the African neoboline cyprinid fishes. By Gordon J. Howes . . . .151 No 4 The haplochromine species (Teleostei, Cichlidae) of the Cunene and certain other Angolan rivers. By Peter Humphry Greenwood . . . .187 No 5 Miscellanea Notes on testate amoebae (Protozoa: Rhizopoda) from Lake Vlasina, Yugoslavia. By Colin G. Ogden 241 Description of a neotype for the holothurian Oncus brunneus (Forbes MS in Thompson 1840) from Strangford Lough, Northern Ireland (Holothurioidea: Dendrochirotida). By J. Douglas McKenzie . . . 265 Three new species of Varicorhinus (Pisces, Cyprinidae) from Africa. By K. E. Banister .273 Phyletics and biogeography of the aspinine cyprinid fishes. By Gordon Howes 283 New bats (Mammalia: Chiroptera) and new records of bats from Borneo and Malaya. By J.E. Hill &C. M.Francis ... .305 No 6 Anatomy and evolution of the feeding apparatus in the avian orders Coracii- formes and Piciformes. By P. J. K. Burton 331 No 7 A revision of the spider genus Cyrba (Araneae: Salticidae) with the description of a new presumptive pheromone dispersing organ. By F. R. Wanless . 445 Bulletin of the British Museum (Natural History) Miscellanea Zoology series Vol 47 No 1 28 June 1984 The Bulletin of the British Museum (Natural History), instituted in 1949, is issued in four scientific series, Botany, Entomology, Geology (incorporating Mineralogy) and Zoology, and an Historical series. Papers in the Bulletin are primarily the results of research carried out on the unique and ever-growing collections of the Museum, both by the scientific staff of the Museum and by specialists from elsewhere who make use of the Museum's resources. Many of the papers are works of reference that will remain indispensable for years to come. Parts are published at irregular intervals as they become ready, each is complete in itself, available separately, and individually priced. Volumes contain about 300 pages and several volumes may appear within a calendar year. Subscriptions may be placed for one or more of the series on either an Annual or Per Volume basis. Prices vary according to the contents of the individual parts. Orders and enquiries should be sent to: Publications Sales, British Museum (Natural History), Cromwell Road, London SW7 5BD, England. World List abbreviation: Bull. Br. Mus. nat. Hist. (Zool.) Trustees of the British Museum (Natural History), 1984 The Zoology Series is edited in the Museum's Department of Zoology Keeper of Zoology : Dr J. G. Sheals Editor of Bulletin : Dr C. R. Curds Assistant Editor : Mr C. G. Ogden ISBN 0 565 05004 4 ISSN 0007-1 498 Zoology series Vol 47 No. 1 pp 1-82 British Museum (Natural History) Cromwell Road London SW7 5BD Issued 28 June 1984 BRITISH MUSEUM (NAi JN1984 GE:," Miscellanea Contents A revision of the genera Trachelostyla and Gonostomum (Ciliophora, Hypotrichida), including the redescriptions of T. pediculiformis (Cohn, 1866) and T. caudata Kahl, 1932. By M. Maeda & P. Carey 1 Notes on Atlantic and other Asteroidea. 4. Families Poraniidae and Asteropseidae. By Ailsa M. Clark 19 The larval and post-larval development of the Thumb-nail Crab, Thia scutellata (Fabricius), (Decapoda: Brachyura). By. R. W. Ingle 53 Description of a new species of Sylvisorex (Insectivora: Soricidae) from Tanzania. By P. D. Jenkins 65 A new species of the Hipposideros bicolor group (Chiroptera: Hipposideridae) from Thailand. By J. E. Hill & S. Yenbutra 77 A revision of the genera Trachelostyla and Gonostomum (Ciliophora, Hypotrichida), including redescriptions of T. pediculiformis (Cohn, 1866) Kahl, 1932 and T. caudata Kahl, 1932 Masachika Maeda Ocean Research Institute, University of Tokyo, Nakano, Tokyo 164, Japan Philip G. Carey Department of Zoology, British Museum (Natural History), Cromwell Road, London SW7 5BD Introduction Hypotrich ciliates are commonly encountered in many and diverse habitats. As Corliss (1979) noted, the taxonomy of this group is still confused, indeed some species, genera and even families are awaiting assignment to their proper taxonomic positions. When one genus of hypotrich, Trachelostyla, was isolated from marine interstitial sediments during a previous study of the ciliate fauna of the southern coast of England (Carey & Maeda, 1985), some difficulty was encountered in understanding the taxonomic position of this organism and also the closely related genus Gonostomum. Recent revisions of these two genera by Borror (1972) and Buitkamp (1977) have not clarified their position, they have merely confused the taxonomy of the family Holostichidae by either synonymizing conspicuously different organisms in a single genus or by assigning many disimilar species to one taxon. Since Trachelostyla and Gonostomum have been isolated frequently from marine and terrestrial environments (Kahl, 1932; Gellert, 1956), a detailed taxonomic investigation of these two genera should provide useful information for future ecological studies. This work is intended to clarify the confusion between Trachelostyla and Gonostomum, and by virtue of a complete revision, provide a key to species of these two genera. Additionally the morphological features of Trachelostyla pediculiformis (Cohn, 1866) Kahl, 1932 and T. caudata Kahl, 1932 are redescribed. Materials and methods The hypotrich ciliates were collected from interstitial sediments at East Head, West Wittering, National Grid reference SZ7799, a short headland at Chichester Harbour where, in a previous study, six sampling stations had been established (Carey and Maeda, 1985). Stations 1 and 2 faced the English Channel and were exposed to rather intensive wave action, whereas calmer water prevailed at Stations 4, 5 and 6 at the northern- most extremity of the headland. On the eastern shoreline around Station 6 an extensive saltmarsh was present. Earlier results of sand analysis had indicated that the mean grain size of sand in this area varied between 170 um and 250 um. The 'detrital material' or content of particles less than 62 um increased continuously from Station 1 to Station 6. This component was measured as approximately 100 times higher at Stations 5 and 5 than that at Stations 1 and 2. In October 1982 Trachelostyla pediculiformis was found at all the Stations except Station 1 whereas Trachelostyla caudata was isolated from fine sands with low content of detrital material, i.e. Stations 1, 2 and 3. Water temperature of the sediment was recorded as 14°C. Bull. Br. Mm. nat. Hist. (Zool.) 47(1): 1-17 Issued 28 June 1984 2 M. MAEDA & P. G. CAREY In July 1983, the water temperature was 24°C and decaying macroalgae were observed in seawater overlying the sampling area. On this occasion T. pediculiformis was found at all Stations sampled (Stations 1, 2 and 3) while T. caudata was encountered only at Station 1. The ciliates were extracted from sediment cores as soon as possible after sampling, by the seawater-ice method of Uhlig (1968). For the sediment sample collected in July 1983, water at low temperature ( 1 2°C) was used for extraction instead of crushed ice as the tempera- ture drop killed too many organisms (Hartwig et al., 1977). Samples from each Station were retained in seawater under constant aeration for further investigation. Some difficulty in handling the organisms was experienced due to their thigmotactic nature and inate fragility. Rapid transfer using a wide bore micropipette from extraction dish to glass slide overcame this problem. The ciliates were observed using Nomarski interference, phase contrast and brightfield illumination, and recorded on video tape for repeated observation. High resolu- tion optics and video recording enabled observations to be made on living cells without recourse to silver staining techniques. Nomenclature Faure-Fremiet (1961) proposed the new family Holostichidae in the suborder Stichotrichina to include those organisms which possess a row of right and left marginal cirri, an elongated body and macronuclei two to many in number, with differentiated frontal and transverse cirri in most cases. These characteristics are certainly found in the genera Gonostomum and Trachelostyla although Corliss (1979) suggested the transfer of these two genera to the family Oxytrichidae in the suborder Sporadotrichina, because some members of these taxa have distinctive marginal cirri and well developed fronto ventral and transverse cirri. The genus Gonostomum was proposed by Sterki (1878) who transferred Oxytricha affinis Stein, 1859 to Gonostomum affine because of the location and shape of the peristome area and the arrangement of frontoventral cirri. He also proposed that Oxytricha strenua Engelmann, 1862 be changed to Gonostomum strenua. Since then, these two species have been placed into several genera. Kent (1881-1882) pointed out that the name of Gonosto- mum closely resembled those of Gonostoma and Gonostomus, names already employed to designate certain genera of fish and molluscs. Consequently he proposed the new genus Plagiotricha, with P. affinis and P. strenua. The first detailed revision of this group was presented by Gourret and Roeser (1888), which is illustrated in Table 1. These authors did Table 1 Revision of the genus Stichochaeta by Gourret & Roeser, 1888 Genus Stichochaeta Clap. & Lachm. Syn. Gonostomum Sterki Plagiotricha Sav. Kent 1 . Stichochaeta pediculiformis Cohn Syn. Gonostomum pediculiforme Maupas 2. Stichochaeta affinis (Stein) Syn. Oxytricha affinis Stein Gonostomum affine Sterki Plagiotricha (Gonostomum) affinis Sav. Kent 3. Stichochaeta strenua (Englemann) Syn.Oxytricha strenua Englemann Plagiotricha strenua Sav. Kent Gonostomum strenua Maupas 4. Stichochaeta Corsica n. sp. Stichochaeta Corsica n. sp. CILIOPHORA 3 not accept the names Gonostomum or Plagiotricha and placed the two species mentioned above into the genus Stichochaeta as S. affinis and S. strenua. In their revision, which already contained Stichochaeta pediculiformis described by Cohn, 1866, they also described a new species Stichochaeta Corsica. Maupas (1883) decided to transfer S. pediculiformis to the genus Gonostomum noting that a typical member of this genus Stichochaeta cornuta possessed quite different features in the peristome area and also cirral arrangements from that of Gonos- tomum. In 1929 Shibuya found and described a new species which he named Gonostomum andoi. Kahl in 1928 described G. pediculiforme, but in 1932, in contrast to the description of Maupas (1883), mentioned that certain details were sufficient to erect a new genus, Trachelostyla. He pointed out that the three filose cirri in the caudal area which Maupas (1883) described as caudal cirri were in fact dorsal cirri which can be seen from the ventral side. The discontinuity of the two marginal cirral rows in the posterior area is a characteristic feature of Kahl's description of Trachelostyla. Kahl (1932) placed two species in this genus, T. pediculiformis and a new marine species T. caudata. After the genus Trachelostyla was proposed, the nomenclature of this group became simplified in that only the genera Gonostomum and Trachelostyla were recognized by subsequent authors. T. dubia was described from marine interstitial sediments by Dragesco (1954). In publications dealing with the ciliated fauna in soils underlying moss and leaves, Gellert (1942, 1956, 1957) described 5 species which were attributed to the genus Gonostomum, G. algicola, G. spirotrichoides, G. bryonicolum, G. ciliophorum and G. geleii. The second major revision of this group was presented by Borror (1972) which unfortunately was given without detailed explanations (Table 2). He transferred the 4 species of Gonostomum described by Gellert (1956, 1957) to the genus Trachelostyla along with a species of Urosoma and a species of the genus Sticho- tricha. T. pediculiformis was retained as Kahl (1932) proposed. A major divergence from the established taxonomy was proposed in the removal of Oxytricha affinis; Gonostomum affine of Sterki, 1878 and its synonyms, to the genus Gastrostyla. Trachelostyla dubia Dragesco (1954) was also transferred by Borror (1972) to the genus Gastrostyla. Gonostomum Table 2 Revision of the genera Gonostomum and Trachelostvla by Borror, 1972 Genus Gonostomum Sterki, 1878 1. G. strenum (Englemann, 1862) Sterki, 1978 Syn. Oxytricha strenuum Englemann, 1862 Oxytricha tricornis Milne, 1886 Genus Trachelostyla Kahl, 1932 1. T. pediculiformis (Cohn, 1866) Kahl, 1932 Syn. Stichochaeta pediculiformis Cohn, 1866 S. Corsica Gourret & Roeser, 1887 Gonostomum pediculiforme Maupas, 1883 2. T. bryonicolum (Gellert, 1956) n. comb. Syn. Gonostomum bryonicolum Gellert, 1956 3. T. caudata Kahl. 1932 4. T. ciliophorum (Gellert, 1956) n. comb. Syn. Gonostomum ciliophorum Gellert, 1956 5. T. geleii (Gellert, 1957) n. comb. Syn. Gonostomum geleii Gellert, 1957 6. T. macrostoma (Gellert, 1957) n. comb. Syn. Urosoma macrostoma Gellert, 1957 7. T. simplex (Kahl, 1932) n. comb. Syn. Stichotricha simplex Kahl, 1932 8. T. spirotrichoides (Gellert, 1956) n. comb. Syn. Gonostomum spirotrichoides Gellert, 1956 M. MAEDA & P. G. CAREY Table 3 Revision of the genus Trachelostyla by Buitkamp, 1977 Genus Trachelostyla \. Trachelostyla pediculiformis (Cohn, 1866) Kahl, 1932 Synonym: Stichochaeta pediculiformis Cohn, 1866 Gonostomum pediculiforme Maupas, 1883 S. Corsica Gourret & Roeser, 1887 2. Trachelostyla caudata, Kahl, 1932 3. Trachelostyla affine (Stein, 1859) n. comb. Synonym: Oxytricha affine Stein, 1859 Gonostomum andoi Shibuya, 1929 G. affine (Stein, 1859) Kahl, 1932 G. bryonicolum Gellert, 1956 G. ciliophorum Gellert, 1956 G. spirotrichoides Gellert, 1956 G. geleii Gellert, 1957 Gastrostyla affine (Stein, 1859) Borror, 1972 Trachelostyla bryonicolum (Gellert, 1956) Borror, 1972 T. ciliophorum (Gellert, 1956) Borror, 1972 T. spirotrichoides (Gellert, 1956) Borror, 1972 T. geleii (Gellert, 1957) Borror, 1972 T. canadensis Buitkamp & Wilbert, 1974 strenua, and its synonyms including Oxytricha tricornis Milne (1886-1887) was the only species of this genus to be retained. A new species of Trachelostyla, T. canadensis was described by Buitkamp & Wilbert (1974) from prairie soil in Canada and three years later the last major revision of this group was undertaken by Buitkamp (1977), see Table 3. In the latter revision which was devoid of detailed explanatory information, both Trachelostyla pediculiformis and T. caudata were retained, but an unusual step was taken in the establish- ment of Trachelostyla affine n. comb, and the inclusion of seven species in T. affine under various synonyms. It should be noted that this new taxon contained a variety of morpho- logical forms which, when carefully investigated, should not have been synonymized. The last species to have been described in the genus Gonostomum was G. franzi (1982) from Austrian soils. As a result of these earlier revisions there are still species which have a dubious position in these genera. These include Stichochaeta mereschkoviskii Andrusov, 1886 which Kahl (1932) put forward as a possible Trachelostyla, and Trachelostyla rostrata of Lepsi (1964) which is poorly described. A species of Gonostomum, G. parvum was found and described by Lepsi according to Stiller (1977). However detailed information is not yet available for this species. The diagram of Gonostomum franzi given by Foissner (1982) clearly shows the presence of ventral cirral rows extending almost the full length of the body, and should not be included in the genus Gonostomum. Gonostomum geleii described by Gellert (1957) pos- sesses certain characters that suggest its inclusion in the genus Gonostomum, however the shape of the body and transverse cirri indicate that this organism should be placed in another genus, probably Urosoma. Similarly Borror's (1972) decision to include Urosoma macro- stoma and Stichotricha simplex was clearly based on a superficial resemblance of these organisms to Trachelostyla. However the decision to transfer T. dubia of Dragesco (1954) to the genus Gastrostyla in our opinion was correct. A detailed study of these organisms indicates that they should all be excluded from the genus Gonostomum. Jankowski (1979) proposed a new scheme of classification for the order Hypotrichida, including the trans- ference of the genus Trachelostyla to the new subfamily Oxytrichinae. But as detailed infor- mation on these revisions were not given, the Jankowskian scheme of classification was not followed in the present work. CILIOPHORA 5 Although 3 major revisions have been undertaken on this group, little information has been given by authors in support of major taxonomic changes. Certainly with the confusion regarding these genera it is clearly essential that a complete and accurate revision should be undertaken. Diagnosis of the Genus TRACHELOSTYLA Kahl, 1932 Gonostomum Maupas, 1883 pro pane Stichochaeta Gourret & Roeser, 1888 pro pane A free swimming genus of hypotrich with a fragile and elastic, but non-contractile body, 130-1 90 um in length. This organism possesses a narrow neck-like constriction in the anterior region. The peristome area is confined to the left lateral border, its posterior part bending abruptly and extending nearly to the centre of the body. Five to 10 frontoventral cirri are present in the anterior region, but frontoventral cirri in the mid- and posterioventral area are absent. Five or six transverse cirri are distinct in the posterior and form an oblique row. There are two marginal cirral rows which are not confluent posteriorly. No caudal cirri are present. Fine dorsal cirri can be observed from the ventral side. These appear to make two rows at the edge of the body and where these terminate at the posterior end some authors in earlier descriptions have clearly misinterpreted these as caudal cirri. Numerous macro- nuclei are dispersed throughout the body. This genus is found in marine habitats, and may be described as truely interstitial. Key to species of Trachelostyla Posterior region rounded, membranelles of AZM thickened and distinctive . . . . pediculiformis Posterior region narrowed, membranelles of AZM fine and of uniform length caudata Trachelostyla pediculiformis (Cohn, 1866) Kahl, 1932 Stichochaeta pediculiformis Cohn, 1866 Gonostomum pediculiforme Maupas, 1883 Stichochaeta Corsica Gourret & Roeser, 1888 MORPHOLOGICAL DESCRIPTION: This species was found in interstitial sediments at East Head, Chichester Harbour, England, during a taxonomic and ecological survey of marine inter- stitial ciliates, in October, 1982 and July, 1983 (Carey & Maeda, 1985). Water temperatures of the sediment were 14°C and 24°C, respectively. This highly psammophilic species has been commonly encountered in many coastal areas of Europe, Russia, North and South America and the Sea of Japan. The body is characteristically flexible. Its locomotion as well as its body shape is distinct enough to distinguish it from other ciliates, frequently moving forward and backward and repeatedly raising the 'head' region above the substrate, con- stantly searching between the sand grains. The body is elongate, 136-196 um in length (Fig. 1). The anterior region, which comprises one-third of entire body length, is attenuated and produced as a narrower neck-like process. The posterior end is rounded in this species. The peristome area is confined to the left lateral border of the anterior region for most of its length. The posterior quarter of peristome area is bent inwards, and eventually extends to near the centre of the body. The buccal opening is situated at the 'shoulder-like' ridge at the base of the 'neck' region. There are five long membranelles of the AZM at the anterior extremity. A bundle of several long cilia-like mem- branelles are clearly seen in the buccal area which is entirely consistent with Cohn's (1866) published diagram. The very thin, membrane-like dorsal edge of the peristome extends for half the total length of the AZM on the left side. These are transparent and easily overlooked. The anterior area displays three frontal and seven ventral cirri. Maupas (1883), Kahl (1932) and Borror (1972) indicated the number of these cirri as 8, 1 1 and 11, respectively, including M. MAEDA & P. G. CAREY AZM DC TC Fig. 1 Trachelostyla pediculiformis Entire organism, ventral view, (AZM) adoral zone of mem- branelles, (FC) frontal cirri, (VC) ventral cirri, (MC) marginal cirri, (TC) transverse cirri, (DC) dorsal cirri, (BC) buccal cavity, (Ma) macronuclei. Bar represents 10 urn. three distinct frontal cirri. There are no ventral cirri in the middle and posterior of the body as Kahl (1932) described. However in the descriptions of Maupas (1883) and Borror (1972) two ventral cirri just anterior of transverse cirri were present. Five or sometimes six trans- verse cirri make an oblique row from the left hand to the right hand side of the body. The right marginal cirral row starts from the 'shoulder' adjacent to the apical area and left row runs from just below the buccal region. These two marginal rows are not confluent poster- iorly and terminate at the base of the transverse cirri. At both lateral edges rows of very CILIOPHORA 7 fine and long cilia can be observed, they are particularly long in the posterior region. Maupas (1883) designated these long cilia at the posterior end as caudal cirri, however Kahl (1932) found them projecting from the dorsal surface of the body: that is to say these thread-like cilia of the lateral and posterior are bristles originating from the dorsal side. Cohn (1866) found this species and appointed the name Stichochaeta pediculiformis. Maupas (1883) also described this organism and transferred it to the genus Gonostomum as G. pediculiforme because he found it had quite different morphological features in the peristome area, and cirral arrangements from the genus Stichochaeta of Claparede & Lachmann (1858), Kahl redescribed G. pediculiforme in 1928. However, he erected the new genus Trachelostyla in 1932 giving the reason that the organism had a 'neck-forming' narrowed peristome area and its two marginal cirral rows were not confluent at the posterior end of the body. He included two species in this genus, T. pediculiformis and T. caudata. Borror (1972) and Buitkamp (1977) agreed with Kahl (1932) and placed Gonostomum pediculiforme and Stichochaeta pediculiformis as synonyms of T. pediculiformis. In agreement with Kahl (1932), Borror (1972) and Buitkamp (1977), the possession of a narrowed 'head' area, clearly denned transverse cirri, the non-convergent marginal cirral rows posteriorly and the absence of caudal cirri are sufficient reason to retain this species in the genus Trachelostyla. Gourret & Roeser (1 888) described Stichochaeta Corsica as having serried rows of adoral 'cirri', long and fine caudal 'cilia', dorsal 'cilia' and transverse cirri. Careful analysis of their description and diagram reveals the long and fine cilia at the pos- terior extremity of the body are in reality dorsal bristles, not caudal cirri. From a detailed survey of morphological features it is clear that S. Corsica must be considered a synonym of T. pediculiformis. Trachelostyla caudata Kahl, 1932 MORPHOLOGICAL DESCRIPTION: Specimens were found in the fine sediments of East-Head, Chichester Harbour, especially in sands which contained a lower content of detrital material than the chosen biotope of Trachelostyla pediculiformis (Carey & Maeda, 1985). The dates of sampling and the water temperature of the sampling area were the same as that for T. pediculiformis. This psammophilic species was reported to be common in the coastal areas of many European countries and Russia. Locomotion is similar to T. pediculiformis, but the 'head region' is not lifted from the substrate during forward movement. The body, 1 56 urn in length, is fragile and elastic but not contractile (Fig. 2). It is elongated and has a narrowed 'neck-like' anterior and a narrowed 'tail-like' posterior region. The form of peri- stome area is similar to T. pediculiformis which is confined to the left lateral border of the 'head' area, and its posterior bends abruptly, extending backwards to near the centre of the body. Five long membranelles of the AZM are present at the apex of the cell, but the mem- branelles of the lateral portion are rather short and 'brush-like'. The characteristically long membrane which T. pediculiformis possesses in the buccal area is not present in this species. Among the four fronto ventral cirri, 3 make a row, but the fourth cirrus from the apex is slightly separated from the other three. The length of these cirri is approximately the same as that of the marginal cirri. There are no frontoventral cirri in the mid- and posterioventral area. Five thick transverse cirri are present, as Kahl (1932) described, these make a oblique row from the left to the right hand side. The right marginal cirral row starts from the base of the narrowed 'head' and the left marginal row begins just posterior of the end of the peri- stome. The two marginal rows terminate at the base of the transverse cirri. At both the right and left edges of the body, faint dorsal cilia are clearly displayed. These dorsal cilia of the right hand edge run from the top of the apical area, and join the other dorsal row of cilia of the left hand side at the posterior end of the body. Dorsal cilia or cirri are not as long as those of T. pediculiformis and do not extend to any great length in the caudal area. There are 8-10 dorsal cirral rows according to Kahl (1932). A contractile vacuole is situated just posterior of the peristome end. In Kahl's (1932) diagram, 11 macronuclei are shown. M. MAEDA & P. G. CAREY Fig. 2 Trachelostyla caudata Ventral view. Bar represents 10 ^m. This animal was originally found in the sands of Kiel by Kahl (1932). In agreement with both Borror (1972) and Buitkamp (1977) this organism has been retained in the genus Trachelostyla. Diagnosis of the Genus GONOSTOMUM Sterki, 1878 Oxytricha Stein, 1859 pro pane Plagiotricha Kent, 1881-1882 Stichochaeta Gourret & Roeser, 1888 pro pane Gastrostyla Borror, 1972 pro pane Trachelostyla Borror, 1972 pro pane 9 CILIOPHORA 9 A genus of hypotrich with a more or less flexible and elastic, but non-contractile body. The body shape is oval or ellipitical, 60-1 50 um in length. The peristome area is confined to the lateral border and its posterior portion, in most cases, is abruptly bent towards the centre of the body. The AZM in the apical area is composed of long and thick membranelles. There are 3 distinct frontal cirri in the anterior region. Among the 6 to 20 ventral cirri present, one or two cirri are positioned just anterior of the transversals. No ventral cirri are present in the mid-ventral area. Two to four transverse cirri are present, which are not thickened in most species. There are two marginal cirral rows which are confluent posteriorly, or more correctly run posteriorly to meet several elongate caudal cirri. Dorsal cirri are not long enough to be seen from the ventral side of the body. Two oval macronuclei are present and a single micronucleus is situated near each of the macronuclei. A single contractile vacuole is situated just posterior of the end of the peristome area, at the left hand lateral border. The species belonging to this genus are found in salt water, fresh water and terrestrial soils. Key to species of Gonostomum 1 Ventral cirral rows well developed 2 — Ventral cirral rows absent 4 2 At least one ventral cirral row extending further than the end of the peristome. strenua — Ventral cirral rows terminating at the peristome 3 3 Four transverse cirri present affine — Two transverse cirri present algicola 4 Caudal cirri present 5 — Caudal cirri absent ciliophorum 5 AZM bent abruptly inwards at the buccal cavity, membranelles large and distinctive . . bryonicolum — AZM gently curved inwards at the buccal cavity, membranelles not distinctive. . . . spirotrichoides Gonostomum affine (Stein, 1859) Sterki, 1878 Oxytricha affinis Stein, 1859 Plagiotricha affinis Kent, 1881-1882 Stichochaeta affinis Gourret & Roeser, 1888 Gonostomum andoi Shibuya, 1929 Gastrostyla affine Borror, 1972 Trachelostyla affine Buitkamp, 1977 DIAGNOSIS: This free swimming species has been found in a variety of habitats including salt water, soil under leaf litter, moss and heathland. The length recorded by Kahl (1932) was 95-1 15 um, however those of Wenzel (1953) were measured as 63-125 um. The body is elongate and its anterior and posterior ends are narrowed and slightly pointed (Fig. 3). Kent (1881-1882) described the body form as lanceolate. The peristome area is arcuate and is confined chiefly to the left lateral side, its posterior end bending inwards at the tip, terminating near the centre of the body. There are three frontal cirri in the anterior area and one buccal cirrus above the paroral membrane. According to Buitkamp (1977) there is one distinct frontal cirrus in the apical region, but Foissner (1982) in his detailed investi- gation found 3 frontal cirri as Stein (1859), Kent (1881-1882) and Kahl (1932) described. The ventral cirri number 8 to 13 in Stein's (1859) diagram, some of these making a slightly oblique row on the right side of the peristome. There are only 2 ventral cirri in the mid- ventral area, placed just in front of the transverse cirri. Three to five transverse cirri are present, whose length does not extend beyond the posterior end of the body in the original description. However, Kahl (1932), Wenzel (1953) and Foissner (1982) showed this animal having rather longer transverse cirri than those described by Stein (1859). Two marginal cirral rows are confluent posteriorly or terminate as a row of caudal cirri in some descriptions (Wenzel, 1953; Buitkamp, 1977). Two macronuclei are present. A contractile vacuole is situated just below the peristome. 10 M. MAEDA & P. G. CAREY Fig. 3 Gonostomum affine After Wenzel, 1953; ventral view slightly modified. Bar represents 10 ^m. This animal was described by Stein (1859) and given the name Oxytricha affinis. Sterki (1878) noted the characteristic shape of the peristome area and the lack of mid-ventral cirri. In consequence he erected the new genus Gonostomum and transferred O. affinis to G. affine. Kent (1881-1882) pointed out that the name of Gonostomum closely resembled that of Gonostoma and Gonostomus which were already employed to designate certain genera of fish and molluscs. Consequently Kent appointed the new genus Plagiotricha which included P. affinis. Gourret & Roeser (1888) agreed with the action of Kent, but chose to transfer P. affinis to the genus Stichochaeta. Gonostomum andoi Shibuya, 1 929 possesses a slightly different arrangement of cirri on the ventral surface from that of G. affine: that is, it displays a slightly higher number of frontoventral cirri and no cirri just anterior of transversals. But Kahl (1932) and Buitkamp (1977) showed the variation of form in G. affine especially the absence of ventral cirri just anterior of the transverse. Because other features of this organism resemble those of G. affine, we designate G. andoi as the synonym of G. affine. Kahl (1932) retained this organism as Gonostomum affine and gave Gonostomum andoi as its synonym. Borror (1972) transferred this organism to Gastrostyla affine because it possessed one oblique row of ventral cirri. However it is clear that this ventral cirral row in Gonostomum affine is not sufficiently developed to ensure removal to the genus Gastrostyla. Buitkamp (1977) proposed the new combination Trachelostyla affine, however Foissner (1982) retained this species in the genus Gonostomum. CILIOPHORA 1 1 In agreement with Kahl (1932) and Foissner (1982), the confluence of the 2 marginal cirral rows (Stein, 1859), the possession of caudal cirri (Wenzel, 1953), the presence of two ventral cirri just anterior of the transversals and 2 macronuclei are sufficient to retain this organism in the genus Gonostomum. Gonostomum strenua (Engelmann, 1862)Sterki, 1878 Oxtricha strenua Engelmann, 1862 Plagiotricha strenua Kent, 1881-1882 Stichochaeta strenua Gourret & Roeser, 1888 DIAGNOSIS: This organism has a flexible and contractile form, the body being elongate and elliptical (Fig. 4). It has been described from fresh water. The length recorded by Engelmann was 1 50 jim. Its anterior and posterior ends are evenly rounded. The frontal region is slightly thinner and dorsoventrally thicker than the posterior area. The peristome area is very narrow and confined to the left hand lateral edge, its posterior end being inwards, extending nearly to the centre of the body. The adoral zone of membranelles in the lateral area of the peri- stome is longer than that of Gonostomum affine. There are 23-25 frontoventral cirri, most of which form two slightly oblique rows, one row extending two thirds of the total body length, the other row being only half as long. No frontoventral cirri are present in front of the 4 transverse cirri. We assume that among the transverse cirri, possibly two can be counted Fig. 4 Gonostomum strenua After Englemann, 1862; ventral view. Bar represents 10 ^im. 12 M. MAEDA & P. G. CAREY as frontoventrals. The right and left marginal cirri were confluent posteriorly. At the posterior extremity there were 4 long, fine caudal cirri. Two macronuclei are present in the midventral area and one contractile vacuole is situated just below the peristome. Englemann (1862) first described this organism under the name Oxytricha strenua. Sterki (1862) established the new genus Gonostomum and transferred O. strenua to G. strenua on the same basis that O. affinis was transferred to G. affine (see species description, G. affine). Kent (1881-1882) erected the new genus Plagiotricha and proposed P. strenua, and Gourret and Roeser (1888) transferred this organism to Stichochaeta strenua. Kahl (1932) and Stiller (1974) redescribed it as G. strenuum and G. strenua, respectively. Borror (1972) retained only G. strenua in the genus Gonostomum and adopted Oxytricha tricornis Milne, 1886-1887 as its synonym. After careful study of the original description and diagram of O. tricornis it is clear that it should be excluded from Gonostomum and from Oxytricha because of the unusual form taken by peristome. Gonostomum algicola Gellert, 1 942 Trachelostyla canademis Buitkamp & Wilbert, 1974 Trachelostyla affine Buitkamp, 1977. DIAGNOSIS: This species was found in association with green plants on rocks, feeding on flagellates. The length has been recorded as 60-100 um. The body displays an oval or ellipti- cal shape (Fig. 5). The peristome area is confined to the left lateral border and the AZM Fig. 5 Gonostomum algicola After Gellert, 1942; ventral view. Bar represents 10 u CILIOPHORA 13 extends backwards to nearly half the body length. A quarter of the peristome is bent inwards toward the body. In the frontal area 10 thick cirri are usually observed, including 3 frontal cirri and 1 buccal cirrus. Five ventral cirri make a slightly oblique row in the frontal region. Mid- ventral cirri are absent. Transverse cirri number two and there are two caudal cirri pre- sent. There are 1 5 left marginals and 1 8 right marginals. Two macro- and two micronuclei are present. These features are characteristic of the genus Gonostomum. Buitkamp and Wilbert (1974) isolated Trachelostyla canadensis from prairie soil in Canada. Buitkamp (1977) transferred this species to T. affine in his revision of the genus Trachelostyla. The presence of 3 frontal cirri, 1 buccal cirrus, 2 transverse cirri and 2 macro- nuclei, also the lack of ventral cirri in the mid- ventral area indicate this species has a strong similarity to that of Gonostomum algicola although there are 2 caudal cirri instead 3 in G. algicola. These taxonomic features are clearly sufficient to designate this animal as the synonym of G. algicola. Gonostomum spirotrichoides Gellert, 1956 Trachelostyla spirotrichoides Borror, 1972 Trachelostyla affine Buitkamp, 1977 DIAGNOSIS: This species has been found in soil under moss, feeding on bacteria and detritus. The body length is 1 10 um and its form is elongate and cylindrical (Fig. 6). The peristome Fig. 6 Gonostomum spirotrichoides After Gellert, 1956; ventral view. Bar represents 10 um. 14 M. MAEDA & P. G. CAREY region is confined to the left lateral edge and extends backwards to nearly the centre of the body. The AZM comprises 27 membranelles of which 4 membranelles in the apical area are distinctly longer than others. The buccal opening is situated at the posterior extremity of the AZM, which displays a funnel like appearance. There are 3 distinct frontal cirri at the apex of the cell. Among the five ventral cirri present, three cirri are arranged in one oblique row extending from the right to the left of the body. It is not certain whether one cirrus which is situated below the end of this cirral row, is a buccal cirrus. Two isolated cirri are present at the region just below the majority of the ventrals and there is another pair of ventral cirri just in front of the 4 transversals. The number of right marginal cirri is 19, left marginal cirri number 15. Between the two marginal cirral rows there are four caudal cirri situated at the posterior. On the dorsal surface, three cirral rows of long bristles have been described. Two macronuclei with one micronucleus are present at the fronto- and midventral areas, respectively. A contractile vacuole is situated at the left of the posterior end of the AZM. Because of the possession of caudal cirri, the presence of two ventral cirri just anterior of the transversals and two macronuclei, Borror (1972) was clearly in error in attributing this animal to the genus Trachelostyla, as T. spirotrichoides. Buitkamp's (1977) designation of the new combination Trachelostyla affine is also incorrect. Based on the present study this species has been retained in the genus Gonostomum, as G. spirotrichoides. Fig. 7 Gonostomum bryonicolum After Gellert, 1956; ventral view. Bar represents 10 (im. CILIOPHORA 15 Gonostomum bryonicolum Gellert, 1956 Trachelostyla bryonicolum Borror, 1972 Trachelostyla affine Buitkamp, 1977 DIAGNOSIS: The ciliate was found in the humus layer of soil samples and has been desig- nated a detritus feeder, 60 urn, in length. The body is oval, cylindrical and the peristome region is arcuate and confined chiefly to the left lateral border, consequently its posterior end bends abruptly inwards, terminating near the centre of the body where the buccal appar- atus is situated (Fig. 7). In the anterior region there are 3 distinct frontal cirri and 4 ventral cirri, behind which are a separate pair of ventral cirri. One buccal cirrus is situated just above the anterior- most end of the paroral membrane. There is only one ventral cirrus in the midventral area, placed just in front of the four transverse cirri. Two marginal cirri rows terminate as a group of four caudal cirri at the posterior. Four cirral rows of long bristles on the dorsal side have been described. Two macronuclei are present together with one micronucleus, and one contractile vacuole is situated near the termination of the peristome. Borror (1972) and Buitkamp (1977) transferred this species to the genus Trachelostyla as T. bryonicolum and T. affine, respectively. The existence of caudal cirri and a ventral cirrus just anterior of the transversals, indicate that it should be retained in the genus Gonostomum as G. bryonicolum. Fig. 8 Gonostomum ciliophorum After Gellert, 1956: Ventral view. Bar represents 10 (im. 16 M. MAEDA & P. G. CAREY Gonostomum ciliophorum Gellert, 1956 Trachelostyla ciliophorum Borror, 1972 Trachelostyla affine Buitkamp, 1977 DIAGNOSIS: This species has been described as a bacteria and detritus feeder, and was found in the humus layer of soil. The body, 70 um in length, displays a slender and oval form (Fig. 8). The peristome area is confined chiefly to the left lateral border, its posterior portion bending inwards into the body. The buccal apparatus is located at the posterior end of the AZM. Three distinct frontal, one buccal and seven ventral cirri are present on the right of the peristome. There are two ventral cirri in the posterioventral area placed just anterior of the transverse. There are four transverse cirri, two of which were longer than the others. Two marginal cirral rows are confluent posteriorly where no caudal cirri are situated. Two macronuclei with one micronucleus are present, one at the front and the other in the midventral area. Without detailed comments, Borror (1972) transferred this species to the genus Trachelo- styla as T. ciliophorum and Buitkamp (1977) proposed the new combination Trachelostyla affine, again with scant information. Here it has been retained in the genus Gonostomum, as G. ciliophorum despite the revision of Borror (1972) and Buitkamp (1977) who clearly did not take into account the presence of confluent marginals and other diagnostic features. References Andrusov, V. J. 1886. Infusoria of the Bay of Kertch. Trudy Imperatorskago Sankt-Peterburgskago Obschchestva Estestvoispytatelei. S.-Peterburg. (Leningrad) 17: 236-259. Borror, A. C. 1963. Morphology and ecology of the benthic ciliated protozoa of Alligator Harbor, Florida. Achivfur Protistenkunde 106: 465-534. 1972. Revision of the order Hypotrichida (Ciliophora, Protozoa). Journal of Protozoology 19: 1-23. Buitkamp, U. 1977. Uber die Ciliatenfauna zweier mitteleuropaischer Bodenstandorte (Protozoa; Ciliata). Decheniana (Bonn) 130: 1 14-126. & Wilbert, N. 1974. Morphologic und Taxonomie einiger Ciliaten eines kanadischen Prariebo- dens. Acta Protozoologica 13: 201-210. Carey, P. G. & Maeda, M. 1985. Horizontal distribution of psammophilic ciliates in fine sediments of the Chichester Harbour area. Journal of Natural History (in press). Claparede, E. & Lachmann, J. 1857. Etudes sur les infusoires et les rhizopodes. Memoires de I'lnstitut National Genevois 5: 1-260. Cohn, F. 1866. Neue Infusorien im Seeaquarium. Zeitschrift fur wissenschaftliche Zoologie 16: 253-302. Corliss, J. O. 1979. The Ciliated Protozoa: Characterisation, Classification and Guide to the Litera- ture. 2nd ed. Oxford: Pergamon, 455 pp. Dragesco, J. 1954. Diagnoses preliminaires de quelques cilies nouveaux des sables. Bulletin de la Societe Zoologique de France. Paris 79: 62-70. Engelmann, T. W. 1862. Zur Naturgeschichte der Infusionthiere: Beitrage zur Entwicklungsgeschichte der Infusorien. Zeitschrift fur Wissenschaftliche Zoologie. Leipzig 11: 347-393. Faure-Fremiet, E. 1961 . Remarques sur la morphologic comparee et la systematique des Ciliata Hypo- trichida. Comptes Rendus Hebdomadaires des Seances de I 'Academic des Sciences. Paris 252: 3515-3519. Foissner, W. 1982. Okologie und Taxonomie der Hypotrichida (Protozoa: Ciliophora) einiger osterrei- chischer Boden. Archivjur Protistenkunde 126: 19-143. Gellert, J. 1942. Eletegyuttes a fakereg zaldporos bevonataban. Acta Scientiarum Mathematicarum et Naturalium, Universitas Francisco- Josephina Kolozsvdr 8: 1-36. 1956. Ciliaten des sich unter dem Moosrasen auf felsen gebildeten Humus. Acta Biologica, Academiae Scientiarum Hungaricae 6: 337-359. 1957. Ciliatenfauna im Humus einiger ungarischen Laub- und Nadelholzwalder. Annales Instituti Biologici (Tihany) Hungaricae Academiae Scientiarum 24: 1 1-34. CILIOPHORA 1 7 Gourret, P. & Roeser, P. 1888. Contribution a 1'etude des Protozoaires de la corse. Archives de Biolo- gie. Paris 8: 1 39-204. Hartwig, E., Gluth, G. & Weiser, W. 1977. Investigations on the ecophysiology of Geleia nigriceps (Ciliophora, Gymnostomata) inhabiting a sandy beach in Bermuda. Oecologia (Berlin) 31: 1 59-1 75. Jankowski, A. W. 1 979. Systematics and phylogeny of the order of Hypotrichida Stein, 1 859 (Protozoa, Ciliophora). Trudy Zoologicheskogo Instituta, Akademiya Nauk SSSR, Leningrad 86: 48-85. Kahl, A. 1928. Die Infusorien (Ciliata) der Oldesloer Salzwasserstellen. Archiv fur Hydrobiologie 19: 189-246. 1932. Urtiere oder Protozoa I. Wimpertiere oder Ciliata (Infusoria) III. Spirotricha. In Dahl, F. (Ed.) Die Tierwelt Deutschlands und der angrenzenden Meeresteile 25: 399-650. Gustav Fisher, Jena. Kent, W. S. 1881-1882. A Manual of the Infusoria, 2. David Bogue, London, 473-913. pp. Lepsi, I. 1962. Uber einige insbesondere psammobionte Ciliaten vom rumanischen Schwarzmeer-Ufer. Zoologischer Anzeiger. Leipzig 168: 460-465. Maupas, E. 1883. Contribution a 1'etude morphologique et anatomique des infusoires cilies. Archives de Zoologie Experimental et Generate (serie 2) 1: 427-664. Milne, W. 1886-1887. On a new tentaculiferous protozoon and other infusoria, with notes on repro- duction and the function of the contractile vesicle. Proceedings of the Philosophical Society of Glasgow 18: 48-56. Stein, F. 1859. Der Organismus der Infusionsthiere nach eigenen Forschungen In Systematischer Reihenfolge Bearbeitet I. Leipzig, 206 pp. Sterki, V. 1878. Beitrage zur Morphologic der Oxytrichinen. Zeitschrift fur Wissenschaftliche Zoolo- gie. Leipzig (ser. 3) 31: 29-58. Shibuya, M. 1929. Notes on two Hypotrichous ciliates from the soil. Proceedings of the Imperial Academy of Japan, Tokyo 5: 155-156. Stiller, J. 1974. Jarolabacskas csillosok-Hypotrichida. Magyarorszdg Allatvildga Fauna Hungariae 115: 1-189. Uhlig, G. 1964. Eine einfache Methode zur Extraktion der vagilen, mesopsammalen Mikrofauna. Helgoldnder Wissenschafterliche Meeresuntersuchungen. Helgoland 111: 178-185. Wenzel, F. 1953. Die Ciliaten der Moosrasen trockner Standorte. Archiv fur Protistenkunde 99: 70-141. Manuscript accepted for publication 23 September 1983. Notes on Atlantic and other Asteroidea. 4. Families Poraniidae and Asteropseidae Ailsa M. Clark Department of Zoology, British Museum (Natural History), Cromwell Road, London SW7 5BD Introduction The skeletal structure of the poraniid starfishes, upon which the classification relies, is hidden or at least obscured by the more or less thickened body wall and opaque skin. X-radiographs of some North Atlantic specimens have thrown new light on the limits of several taxa. The study has been helped by additional material, from various museums and oceanographic institutes, notably from recent collections of the Discovery, RSS Challenger, Walter Herwig and Alvin, as well as many type specimens. Two inter- family transfers of genera are also made: the Southern Ocean genus Poraniopsis Perrier — hitherto placed in the Echinasteridae despite its name — is now included in the Poraniidae while conversely Poraniella Verrill is transferred from the Poraniidae to the Asteropseidae. In view of previous confusion between these two families, the paper is prefaced by full diagnoses of both. A tabular key for characters of the genera of Poraniidae is appended. Systematic Account Family ASTEROPSEIDAE Hotchkiss & Clark Asteropsidae Perrier, 1884: 154. Gymnasteriidae (pt) Sladen, 1889: 355-356; Perrier, 1894: 327. Asteropidae Fisher, 1908: 90; 1911 (pt): 247-248; Verrill, 1915: 86. Poraniidae (pt) Spencer & Wright, 1966: U69. Asteropseidae Hotchkiss & A. M. Clark, 1976: 266; Blake, 1980: 179; 1981: 381, 391. A family of the order Valvatida with body form stellate, but the young nearly pentagonal, interradial arcs rounded or sometimes with a blunt angle; arms normally five, flat below, convex or carinate above, there being a distinct ventrolateral angle, entirely covered by more or less thickened skin, opaque and tending to obscure the skeleton in larger specimens, R>20mm; abactinal skeleton with primary calycinal plates usually distinguishable, small specimens with compact, flat, initially rounded or hexagonal somewhat imbricating plates (as in the Atlantic genus Poraniella which is only known at R<20 mm) arranged in longi- tudinal series but becoming an open reticulum with linking secondary plates, armed with single carinal spines only (Asteropsis), spaced fine spines (Poraniella), numerous spines (Valvaster) or completely spineless (other genera); abactinal papulae single in small speci- mens but becoming grouped in the skeletal meshes in larger ones; marginal plates well developed, inferomarginals thick wedge shaped, usually projecting to form a ventrolateral angle, variously armed to match the abactinal armament (Poraniella with a horizontal fringe of divergent spines along the edge and a few smaller ones above), superomarginals usually bare and more or less inset but with a few small spines in Poraniella or with multiple spines and often a conspicuous pedicellaria in Valvaster; actinal plates in longitudinal series parallel to the ambulacra, the largest plates and longest series adradial, naked or armed with a few spaced spines; adambulacral plates with two series of spines sheathed in thick skin: pedi- cellariae present in some species, granuliform or (in Petricia and Valvaster) large bivalved Bull. Br. Mm. nat. Hist. (Zool.) 47(1): 19-51 Issued 28 June 1984 19 20 A. M. CLARK (Valvaster also having some elongated tong shaped ones): internally interbrachial septa pre- sent and reinforced by a proximal vertical calcined column in each interradius, joined to the side wall of the disc by a membrane (in Asteropsis at least; septum present but undescribed by Fisher, 1911 in Dermasterias and in Valvaster by Blake, 1980). DISCUSSION: In distinguishing between the Asteropseidae and Poraniidae in 1976, Hotchkiss & Clark (p. 266) failed to realize that Poraniella Verrill, 1914 (then unrepresented in the British Museum collections) shares the arrangement of the series of actinal plates parallel to the furrow characteristic of the Asteropseidae, to which this west indian genus is now referred from the Poraniidae as the only Atlantic representative of an otherwise Indo-Pacific family. All the known specimens of Poraniella are small, whereas the other asteropseids may exceed 70 mm or even 100 mm R. In comparison of a Poraniella echinulata (Perrier) (from the Pillsbury collections in the Lesser Antilles) with a small Asteropsis carinifera (Lamarck) (from the Indian Ocean), both R c. 10 mm, the abactinal plates are similarly arranged in longitudinal rows and hexagonal in shape, though with slightly better developed and more imbricating lobes in Poraniella where there is a well developed armament of fine spaced spinelets or spines on the abactinal, marginal and actinal plates missing in the Asteropsis. Both have the five primary radial plates distinctly enlarged at the head of the midradial series of plates which form a keel in the Poraniella whereas, surprisingly, the rays are quite flat in the small Asteropsis though keeled in larger ones. Also the superomarginals of the small Asteropsis completely overlie the inferomarginals, rather than being inset as in Poraniella, and each bears a relatively stout tubercular spine and some smaller tubercles, though in large Asteropsis these plates are inset and usually naked. The skin investment is thicker throughout in the Asteropsis but the wet preservation (the Poraniella is dry) may account for the difference. Because of the general resemblance of the plating at some stage of the ontogeny, I do not think that the difference in armament and apparently thinner skin in the Poraniella merit more than a generic difference, justifying its inclusion in the Asteropseidae. Indeed, Poraniella is intermediate in armament and skin between Asteropsis and Valvaster, support- ing Blake's inclusion of Valvaster in the family in 1980. Family PORANIIDAE Perrier Poraniidae Perrier, 1893: 849; 1894: 163-164; Verrill, 1914: 17; 1915: 68; Mortensen, 1927: 89-90; Fisher, 1940: 154; Spencer & Wright, 1966 (pt): U69; Hotchkiss & Clark, 1976: 263-266; Blake, 1981:380-381. Gymnasteriidae: Bell, 1893: 21, 78; Farran, 1913: 16. Asteropidae (pt) Fisher, 1911: 247-248. Asteropidae: Koehler, 1921: 40-41; Mortensen, 1933: 249: Fisher, 1940: 136. A family of normally five-rayed Valvatida with body form short-rayed stellate or almost pen- tagonal, interradial arcs rounded, or sometimes angular (in Poraniopsis and some Poranio- morpha); upper side arched, under side flat or, if the body is cushion shaped, slightly convex, a ventrolateral angle more or less distinct; dorsal body wall thickened and skin opaque in all but the smallest specimens; abactinal plates obscured or concealed, when well developed either similar and forming an irregular fairly compact reticulum (Poraniomorpha) or with the ten primary calycinal plates on the disc enlarged and making a pentaradiate pattern from which run irregular carinal (midradial) series of plates linked to the superomarginals by transverse chains of dorsolateral plates sometimes interconnected to form an open reticulum with larger nodal plates (Porania and Poraniopsis), but in some taxa the plates progressively resorbed internally, even completely lost in large specimens, R> 50 mm, and the body wall more or less thickened to give compensatory support, gross armament usually lacking or sparse, only Poraniopsis and occasional Porania with coarse spaced spines on some of the larger plates, sometimes the skin of the upper side with more or less numerous spiniform, papilliform or almost granuliform spinules, forming a continuous coating, rarely even finer ASTEROIDEA 2 1 spicules, this superficial armament not necessarily penetrating the soft tissue to contact the underlying plates; papulae either spaced or clustered, often present intermarginally as well as abactinally; marginal plates either well developed together with the abactinal skeleton (though often becoming hollow) or sharing in general decalcification, when well developed the two series tending to alternate and often flattened in planes at right angles, the infero- marginals then horizontal and alone supporting the venterolateral angle, the superomarginals vertical (e.g. in Poranid) or both series more compact, blocklike, sharing in forming the ambitus, the angle less well marked (Poraniomorpha], or the plates relatively undistinguished except for their longitudinal arrangement, the inferomarginals ventrally aligned and without a distinct angle (Poraniopsis), inferomarginals armed with a horizontal series, sometimes enclosed within the body wall or more or less aborted (Porania and Chondraster), or with multiple spinules usually enlarged into spinelets along the maximum convexity (Poranio- morpha), or with a few coarse spines not forming a horizontal series (Poraniopsis); actinal areas large, the plating obscured by thick skin usually with spaced grooves which sometimes fork or anastomose running from furrow to margin, or pustular, the underlying plates pri- marily arranged in arcs parallel to the margin, the longest series admarginal (abradial), the shortest adoral, spanning the interradius, also forming series corresponding to the grooves, if the skeleton is generally reduced then the actinals are progressively resorbed from within, appearing as rings on the inner face of the body wall of dissected specimens, the last traces of them close to the adambulacrals, actinal surface either naked or armed with a few spaced spines arising from the underlying plates tending to form series parallel to the margin, or with spinelets enlarged from spinules, spinules alone or unarmed; adambulacrals armed with a few sheathed furrow and subambulacral spines, aligned either transversely (e.g. Porania) or longitudinally (Chondraster and Poraniomorpha); pedicellariae unknown; internally interbrachial septa developed, reinforced by vertical plating unless the entire skeleton is reduced. DISCUSSION: Hotchkiss (in Hotchkiss & Clark, 1976: 265-266) was largely responsible for emphasizing the importance of the different arrangement of the actinal plates with the pri- mary series either parallel to the margin as in the Poraniidae, or parallel to the furrow as in the Asteropseidae. The two families are also probably distinguishable by thermal differ- ences since the Asteropseidae are found in shallow water in tropical or warm-temperate seas while the Poraniidae are mainly from cold temperate and boreal seas, only occurring in lower latitudes at greater depths and with a stunted form. This accords with the removal now from the Poraniidae to the Asteropseidae of the west indian Poraniella Verrill, 1914, because of the actinal plating, which is known from a minimum depth of only 20 m compared with c. 175 m for Marginaster pectinatus Perrier, the only poraniid from the same area. Observation of this same character, agreeing conversely with the Poraniidae, in the genus Poraniopsis by Blake (pers. comm.) prompted him to suggest that it be removed from the Echinasteridae to this family. Perrier (1891: K106-107) failed to describe the arrangement of the actinal plates but Madsen (1956: 29) noted that their spines run parallel to the margin. Perrier thought Poraniopsis intermediate between Echinaster and Porania, emphasizing this by using echinaster as a specific name for the type species. Although he cited more characters in which Poraniopsis resembles Porania, more weight must have been given to those such as arm shape shared by Echinaster. Fisher (1911: 261) thought that resemblances to Porania are 'mostly superficial', holding to this view still in 1940 (pp. 154-155) when faced with an unusually well-armed Falkland Is specimen of Porania antarctica (Fig. IB). However, I find there are also important internal characters in which Poraniopsis agrees better with Porania than with Echinaster, comparing specimens from the vicinity of southern South America, as summarized in Table 1 . The abactinal plating is very similar, with the primary calycinal plates forming a pentaradiate pattern on the disc in both and irregular midradial (carinal) series forming part of an open reticulum; some nodal plates bear large spines and there are often superficial spinules in the skin. The latter are better developed in the truly antarctic subspecies P. antarctica glabra though minute ones c. 0-25 mm long are present Fig. 1 A. Poramopsis echinaster Perrier, BM reg. no. 1975.11.12.9, Chinquihue, S Chile, R 36mm; dry, partly denuded, showing spinules between the spines, x 2. B, Porania (Poranid) antarctica magellanica Studer, 1948.3.16.448. Discovery Investigations st. 80, Falkland Is, R 60 mm; wet, showing papulae between the spines, the spinules microscopic, x U. Fig. 2 A, Poraniopsis echinaster (as in 1A), partly denuded. x2. B, Porania (Poranid) pulvillus (O. F. Miiller), 1922.4.10.1, Lousy Bank, SW of Faeroe Is, R c. 36 mm; wet but contracted, the contours of the plates showing their limits, x 1J. C, Poraniella echinulata (Perrier), Pillsbury st. 853, Windward Is, R 10 mm; dry, partly denuded. x3. 24 A. M. CLARK Table 1 Comparison of Poraniopsis with Porania (especially P. antartica) and Echinaster. Agreement with Porania in capitals. Porania Poraniopsis Echinaster 1 r a a 2 C C I 3 s S N 4 d u u 5 a r r 6 M M L 7 D D S 4. Interradial arcs: a angular r rounded Primary calycinal plates: C conspicuous when denuded, linked to form a pentaradiate pattern I inconspicuous Skin: S usually containing a fine superficial secondary armament of small scattered spinules indepen- dent of the underlying plates N nude, armament limited to spines mounted on the plates Superomarginal plates: u relatively unspecialized, similar to the abactinal plates though corresponding in number to the inferomarginals; usually armed with one or more spines d distinct from the abactinals; spineless 5. Profile of ambitus (widest part of body): r rounded, curving into the upper and lower sides; inferomarginals inset somewhat on the ventral side, their spines not forming a continuous fringe a angular, the prominent ventrolateral angle emphasized by a prominent fringe of spines 6. Alignment of actinal plate series and any coarse armament: M in arcs across the interradii parallel to the mar- gin, the admarginal series the longest L in longitudinal lines along each ray parallel to the furrow, the adradial series the longest 7. Adambulacral plate joint faces: D with a deep restricted interadambulacral muscle depression towards the furrow face S with a shallow extensive muscle depression. Characters 6 and 7 carry the greatest weight in my opinion. in the Falklands specimen. There is also frequent alternation of the plates of the two marginal series of both Poraniopsis and Porania and the adambulacral spines are very few, thickly sheathed in skin and aligned at right angles to the furrow, besides the most obvious resem- blances of the thick opaque skin and the actinal plate arrangement. Sectioning of an arm shows that the proximal face of the adambulacral plates in Poraniopsis has a similarly deep interadambulacral muscle depression towards the furrow side to that found in the poraniids examined, quite different from the much wider and more shallow depression in Echinaster (see Blake, 1981, fig. 2, Chondraster and Echinaster, also Figs 7, 8 here). The main differ- ences in Poraniopsis consist of the better-defined rays with an angular rather than curved- interradial arc, the resemblance in shape and armament of the superomarginal plates to the primary (nodal) abactinals and the absence of a distinct ventrolateral angle, the infero- marginals having no horizontal abradial prolongation being instead slightly inset ventrally with the spines of consecutive plates quite discrete, not forming a continuous horizontal fringe. Although the diagnosis of the Poraniidae needs to be somewhat modified to accom- modate it, the balance of evidence, I believe strongly supports the inclusion of Poraniopsis. Madsen (pers. comm.) tells me that Mortensen placed Poraniopsis in the Asteropidae (then including Poraniidae) in the MS catalogue of asteroids in the Zoological Museum, Copenhagen, giving a clue to this disposition by placing Poraniopsis after Chondraster elattosis, well separated from the Echinasteridae in his table of south african echinoderms (1933: 225). Leipoldt (1895) also concluded that Poraniopsis belongs in the Poraniidae but this has generally been overlooked. ASTEROIDEA 25 Ten other nominal genera of poraniids are known from Atlantic waters. With the currently accepted name of the type species, these are: Porania Gray, 1840. P. pulvillus (O. F. Miiller, 1776) (as Asterias) Poraniomorpha Danielssen & Koren, 1881. P. hispida (M. Sars, 1872) (as Asterind) Tylaster Danielssen & Koren, 1881. T. willei Danielssen & Koren, 1881 Marginaster Perrier, 1881. M. pectinatus Perrier, 1881 Chondraster Verrill, 1895. C. grandis (Verrill, 1878) (as Porania) Culcitopsis Verrill, 1914. C borealis (Siissbach & Breckner, 191 1) (as Culcita) Poranisca Verrill, 1914. P. lepidus Verrill, 1914 Pseudoporania Dons, 1936. P. stormi Dons, 1936 Sphaeriaster Dons, 1939. S. berthae (Dons, 1938) (as Sphaeraster) Spoladaster Fisher, 1940. S. brachyactis (H. L. Clark, 1923) (as Cryaster). Two of these, Marginaster and Poranisca, are only known from small specimens, R<20 mm. Verrill (1914: 19) suggests that M. pectinatus is probably 'simply the young of Porania or some similar genus', while his own Poranisca was proposed 'as a matter of con- venience for another group of small young forms belonging to this family, until they can be connected with adults'. However, Downey (1973: 81) found some of the small Margin- asters from the West Indies to be sexually mature. It is not clear from Verrill's account just how he thought Poranisca lepidus differs from Marginaster. His photograph of the largest (syn) type (1914, pi. 1 , fig. 3a) is remarkably similar to the specimen of M pectinatus figured by Downey (1973, pi. 37, fig. A). Both have fairly numerous coarse abactinal spines. Verrill wrote: 'the type is from off the eastern coast of the United States, in 77 fathoms, no, 18,485, Nat. Mus.' A specimen with this catalogue number sent to me as a 'Type' of Marginaster austerus Verrill, 1899, was originally so identified by Verrill but subsequently he wrote 'For. lepidus V. Type' on the back of the earlier label. This label also bears in pencil an illegible Albatross station number beginning with 2 over which has been written 'near sta. 2265' the full station data for which are: 37°07 '40 ' 'N, 74°35 '40 ' 'W (off Chesapeake Bay) 70 fathoms. Since in 1899 (p. 222) Marginaster austerus was cited as from Blake and Albatross stations in the West Indies [my italics], in which area many hauls numbered at around 2350 were made, I think it very likely that the substituted number and hence the more northern type locality given for P. lepidus were misleading. It is significant that Verrill included no less than eight figures of Poranisca lepidus in his 'Starfishes of the West Indies' (1915) without having any mention of the species in his text, while the 'east of the U.S.' locality should have made it completely inappropriate. His plate, 4, fig. 3 shows this specimen, although it was not among the four illustrated in 1914 (pi. 1, fig. 3a-d) some of which have relatively broader inferomarginal plates, leaving little doubt that lepidus is a synonym of Marginaster pectinatus. Poranisca therefore becomes a synonym of Marginaster. A second Albatross specimen in the U.S.N.M. collection labelled as a 'type' of Marginaster austerus (cat. no. 10179 from st. 2333, 'off Havana, Cuba, 169 fathoms') proved to be a Poraniella echinulata (Perrier, 1881), the actinal plates being aligned parallel to the furrow and the five primary radial plates being conspicuously enlarged and convex. Little, if any- thing of Verrill's 1899 description of austerus could have been based on this specimen. The only other extant 'type' of M austerus is in the Peabody Museum, Yale (no. 9858) labelled by Verrill 'sta. unknown. West Indies, coll. A.E.V. Two enlarged photos; also draw- ings'. This has R/r 16-17/10 mm (17/1 1 according to Verrill, 1915: 78) and is almost cer- tainly the specimen with an abbreviated arm shown in his pi. 3, fig. 1 , la, captioned as 'type'. Most of his 1899 description (p. 221) could have been based on it except for descriptions of the primary calycinal plates as distinctly enlarged and the proximal actinal plates as bear- ing rows of spinules — both contradicted in his 1914 and 1915 remarks, the latter specifically referring to the type. Accordingly, this specimen is the most appropriate one for selection as lectotype. Superficially it looks rather different from Marginaster pectinatus (compare Figs Fig. 3 A, B, Marginaster pectinatus Perrier, Pillsbury st. 876, Windward Is, R 1 5 mm; dry, partly denuded. C, Poraniella echinulata (as in 2C), dorsal view. D, E, Marginaster austerus Verrill, P. M. Yale no. 9858, lectotype, 'West Indies', R 16-17 mm; dry. AIIx3. ASTEROIDEA 27 2C, 3C and 3A, B) notably in the more evenly convex upper side after drying — all the dried pectinatus seen being almost flat level with the tops of the superomarginal plates; also the number of marginals is greater, 8-10 rather than 7 or 8 in pectinatus of similar size and the inferomarginals project less, forming only an inconspicuous border in ventral view; lastly, the superomarginals in austerus are armed with spaced spines for their full height, not just at the upper end, whereas pectinatus usually has a bare belt above the inferomarginals, only occasionally a few longer superomarginal spines. The first of these differences may be attributable to an artefact of preservation but the others together could provide a significant difference. However, in general the remaining armament is similar and I suspect that austerus will prove to be a synonym of pectinatus when more material is available. It should be noted that Verrill's pi. 11, fig. 6a (1915) misrepresents the adradial actinal plates as being in line with the furrow; they are parallel to the margin as usual in poraniids. There is a further possibility that Marginaster itself could be a synonym of Porania, to which genus Verrill provisionally referred austerus in 1914 (p. 20). In 1895 (pp. 138-139) he wrote of P. insignis from east of the U.S.A. 'Young specimens- -have more or less numer- ous small, scattered simple spines, both on the dorsal and ventral plates; these plates are distinctly visible, beneath the cuticle, when dried, and the upper marginal plates are rela- tively larger than in the adult. The papulae are few and scattered. In this stage, it agrees in all respects with the genus Marginaster Perrier and Lasiaster Sladen, both of which are probably the young of Porania or Poraniomorpha '. Possibly Lasiaster was added here as an afterthought, since he used the singular 'genus'; it has since been synonymized with Poraniomorpha. A proper comparison of M. pectinatus with small Poranias from off the USA may shed more light on the relationship and the status of the name Marginaster. As for the even smaller (R max. c. 10 mm) geographically 'fringe' species (latitudinally). Marginaster capreensis (Gasco, 1876: 38) from the Mediterranean in 50-c. 600 metres this is very similar in abactinal and marginal armament to M. pectinatus but has more numerous actinal and adambulacral spines. In comparison, the N. european Porania pulvillus loses the few actinal spines found in juveniles more quickly than the american P. insignis. A fourth atlantic Marginaster is M. fimbriatus Sladen, 1889: 365-366, known only from the holotype, R 6 mm, from the Rockall Trough, W of Scotland, in 2487 metres. The name fimbriatus was synonymized with Marginaster capreensis by Ludwig (1897: 190), prompting Mortensen's inclusion of the latter in the british fauna (1927: 92). However, recent collecting in the Rockall Trough has produced several specimens from down to 2070-22 10m which are much more likely to be the same species although the smallest, R 22 mm, is much larger than Sladen's type. I believe that these are referable to the genus Chondraster and conspecific with Chondraster grandis (Verrill, 1878: 371-372), known for off Cape Cod to Cape May, U.S.A. in c. 400-1645 m. One of Farran's three specimens (from Helga st. SR 483) named by him Culcita borealis Sussbach & Breckner (1913: 15-16) also proved to be grandis besides several others from various sources extending to the Bay of Biscay at c. 44°N, 04-5°W. (Biogas VI st. CP 29). Arm sections and X-rays of some of them show a single horizontal row of slender tapering inferomarginal spines, as in Porania, numbering up to 4 on a plate, but these are completely enveloped by the very thick body wall, not individually sheathed; only by excessive shrinkage in preservation does their presence become evident externally. The proximal superomarginals are deeply inset, aligned vertical but somewhat obliquely, tall and flattened, exaggerated in form from those of Porania, while the inferomarginals are flattened horizontally and markedly elongated at right angles to the edge of the body, the spines usually present borne along their abradial ends (see Fig. 7D). The actinal plates are very elongated and overlap to form series linking the inferomarginals with the adambulacrals. A section of a large specimen, R c. 73 mm, shows only a few small, hollow, isolated abactinal plates. X-ray show the density of the other skeletal plates is also low. The adambulacral plates bear usually 2 individually-sheathed furrow spines and 2, sometimes 3, subambulacral spines within a single elongate sheath aligned almost parallel to the furrow, contrasting with the single transverse row in Porania. The smallest specimen, R 22 mm, from the southern Bay of Biscay, is dried, which helps to show a better developed skeleton approximating to that Fig. 4 A, B. Chondrasler grandis (Verrill), 1981.7.20.1, Rockall Trough, R 73mm; partial dorsal and ventral views, x 14. Fig. 6 X-rays of Chondraster grandis (Verrill): A, No details, off Cape Cod, specimen probably dried and shrunken, R c. 15 mm. B, (as in 4). x 1$. ASTEROIDEA 31 Fig. 7 Partial cross section near base of ray viewed from proximal side of: a, Poraniopsis echin- aster Perrier, 79.8.19.6, Magellan Strait, R c. 37mm; b, Porania (Porania) pulvillus (O. F. Miiller), 1950.1 1.3.1, Porcupine Bank, W of Ireland, R 55 mm, the cross-hatched area of the second superomarginal hypothetical, the plate cut in sectioning; c, P. pulvillus, 90.5.7.511, Porcupine st. 8, W of Ireland, R 14 mm; d, Chondraster grandis (Verrill), 1981.7.20.1, Cirolana st. 22, E side of Rockall Trough, R c. 73 mm; e, Porania (Pseudoporanid) stormi (Dons), 1 920. 1 2.28.3 1 , Lousy Bank, SW of Faeroe Is, R 40 mm, the median part of an adjacent complete inferomarginal plate shown by discontinuous lines. i = inferomarginal, s = superomarginal. The scale measures 5 mm. 32 A. M. CLARK Fig. 8 Partial cross sections near base of ray viewed from proximal side of: a, Poraniomorpha (Culcitopsis) borealis (Siissbach & Breckner), IOS st. 50702, Porcupine Seabight, R 19 mm; b, P. (Poraniomorpha) hispida (M. Sars), 98.5.3.223, Trondheim fjord, R 26-5 mm (inferomarginal armament dubbed from another specimen); c, P. hispida rosea Danielssen & Keren, SMBA st. AT 230, Rockall Trough, R 33 mm. Plates well out of the plane of the section are shown by discontinuous lines; i = inferomarginal, s = superomarginal. The scale measures 5 mm. of Porania pulvillus with the primary radial and interradial abactinal plates distinct on the disc and irregular carinal series linked to some of the superomarginals by transverse chains of dorsolateral plates; however, the proximal marginals are of more exaggerated form than in Porania. There are 2-4 inferomarginal spines on each plate and 1, sometimes 2, spines on many plates of the adjacent actinal series, as in some adults of the american Porania insignis and the arctic Tylaster willei. The larger Chondrasters all lack actinal spines and show varying degrees of skeletal reduction. The extent of the papulae is also very variable, from the two narrow bands along each ray shown in Verrill's figure (1885, fig. 44a), to a wide coverage of the upper side except for small midradial and interradial bands; some papulae may even occur between successive superomarginals. Although these specimens are clearly conspecific with Chondraster grandis from the NW Atlantic, they might be subspecifically distinct. An X-ray of an american specimen, R c. 75 mm (Fig. 6A), shows discontinuous series of inferomarginal spines, with 1 or 2 on most interradial and some distal plates, not in between. The inferomarginals appear hollow almost to the abradial extremity in contrast to those of the NE Atlantic specimens but vestiges of actinal and possibly some abactinal plates are similarly distinguishable. Another specimen, from Lydonia Canyon, SE of Cape Cod, R c. 55 mm, shows up to 3 spines on most infero- marginals but no trace of any abactinal or actinal plates; the marginals themselves are ill- defined, whereas a larger similarly dried NE Atlantic individual (from Lousy Bank, SW of the Faeroes), as well as the large wet Rockall Trough specimen, show quite distinct outlines of many actinal plates at least, indicating that the dried and contracted condition is not responsible for the skeletal loss in the american specimen. One of Verrill's two syntypes of C. grandis has been examined, the second is not to be found in the Peabody Museum, Yale. It is considerably shrunken and flattened with all the rays curled upwards and the distal part of one broken off. Although now in alcohol it may have dried up at some time judging from the extreme flattening and shrinkage of the body wall. Mean R is estimated at c. 95 mm; in life it was probably 100+ mm; r is c. 50mm. Superficial spicules all over the upper side give a 'furlike' appearance; on the lower side they are more spaced out, especially proximally. The broken edge of the detached ray shows no sign of any abactinal plates, even in this distal area where resorption is likely to be mini- mized. However, the actinal plates are still fairly well developed, though hollow, and there are 1-3 spines on the abradial ends of four inferomarginal plates from which the tissue has been pared. Indeed, contours corresponding to actinal plates are evident all over the ventral ASTEROIDEA 33 interradii, though it does not follow that the plates remain well developed since similar con- tours may show in poorly preserved specimens of other skeletally deficient poraniids even though X-rays show only vestiges of actinal plates. The papulae are restricted to two narrow bands along each ray, as in Verrill's fig. 44a, 1 885, and the same is true of four other american specimens, three of them from Lydonia Canyon, which is on the south side of George's Bank not far WSW from the type locality. R is c. 55-105 mm, probably at least 60-120 mm in life since they are dried and very shrunken so that the cluster of spinelets around the anus stands out. The number of subambulacral spines ranges from 2 or 3 in the smallest to usually 4 in the largest and the number of oral furrow spines is 6-8 with a single suboral, except in the smallest specimen which has none. Verrill initially (1895: 138) treated Chondraster as a subgenus of Porania but in 1914 (p. 21) evidently accorded it generic rank, being followed in this by H. L. Clark (1923: 274-275) when describing a new species from South Africa. However, in 1959 (p. 160) Madsen thought subgeneric rank to be more appropriate when he described another poraniid, from E Greenland, as Porania (Chondraster) hermanni. Since then (pers. comm.) he has come to believe that hermanni is a Porania sensu stricto and Chondraster generically distinct — mainly on account of the longitudinal arrangement of the adambulacral armament in C. grandis, whereas Porania has these spines in a transverse row. A final atlantic nominal species of Marginaster should be mentioned, namely M. pentago- nus Perrier, 1882: 51) (also 1894: 165-167, pi. 11, fig. 4), the holotype and only recorded specimen of which had R 3 mm, the body form retaining post-larval flattened shape with the inferomarginal plates (numbering 6 on each side in this specimen) alone forming the periphery, the superomarginals being inset on the upper surface and resembling the some- what imbricating polygonal abactinal plates, all bearing a scattering of spinelets. The infero- marginals each bear a comb of 6-8 spinelets along the free edge but apparently inclined downwards. On the under side most actinal plates have one or a few small spinelets and the adambulacrals bear a furrow spine and two or three subambulacral spines in a transverse series. Mortensen (1927: 94) and Tortonese (1965: 167) suggest that pentagonus could be conspecific with the mediterranean M. capreensis but that species has fewer and coarser abactinal spinelets and inferomarginal spines, judging from Ludwig's illustrations (1897, pi. 7, figs 2 1-23). The type locality of M. pentagonus is NW of Finisterre, Spain (c. 44°N, 10-5°W) in 400 metres. The closest geographical record for a poraniid is that of Gallo (1937: 1664) for three specimens from Santander, N Spain, also in 400 m, resembling Culcitopsis borealis (Siissbach & Breckner, 1911: 217-218) although he named them Poraniomorpha hispida (M. Sars, 1872: 26), following Mortensen's synonymy of borealis with hispida in 1912 (p. 258) and 1927 (p. 93). The small and relatively numerous inferomarginal spinelets support inclusion of pentagonus in Poraniomorpha despite the single furrow spines which are probably corre- lated with the small size. However, the status of Culcitopsis needs reassessing since Farran (1913: 15),Koehler(1924: 160-1 61) and Cherbonnier & Sibuet(1973: 1348) have recorded as C. borealis specimens from the Porcupine Seabight (SW of Ireland) and from the NE Bay of Biscay. Mortensen, and also Grieg (1927: 131) had discounted the swollen form and thickened body wall with reduced skeleton of C. borealis as insufficient to warrant more than an infra- specific difference from Poraniomorpha hispida, designating such specimens as forma borealis. Madsen too (pers. comm.) strongly supports such a low rank, believing that borealis is an ecophenotypic form. Certainly most poraniids show some progressive resorption of the skeleton during growth so that even in apparently well-calcified large specimens of Porania and Poraniomorpha the marginals and other plates may be hollow, as evidenced by sectioning or by X-rays — a useful technique for study of this family of thick-skinned asteroids. In addition to the above mentioned authors, others have also commented on the consider- able variation in several directions of Poraniomorpha hispida, notably Djakonov (1946: 163-169, at length, in russian), who compared it with the exclusively arctic P. tumida 34 A. M. CLARK (Stuxberg, 1878: 31), finding intermediate specimens where the ranges of the two overlap (presumably in N Norway and the Barents Sea), mentioning this briefly in his book on russian asteroids (1950: 59, translation 1968: 50). Unfortunately it is not possible to ascertain the adult form of Sars' material since his holotype (from the Lofoten Is, N Norway, 365-550 m) has R only 6 mm. Although it does have a near pentagonal form, R/r 1-2/1 (see Sars, 1877, pi. 8, figs 24-26), this could be true of a more stellate adult when young. However, Grieg (1927: 129-133) makes several references to the 'typical' form, which by implication is a shorter-rayed one since he also refers separately to forma rosea, Danielssen & Keren's holo- type of which had R/r 1 -67/1 and appears relatively stellate in their figures. Djakonov (1950) also described P. hispida as having a massive body, broad disc and broad, very short, rays. In the absence of any evidence to the contrary, this is the form which can be attributed to 'typical' hispida. All nine norwegian specimens in the British Museum collections with well developed skeletons (from Hardanger and Trondheim fjords, from SW of Bergen and off the North Cape) consistently have a pentagonal form, R/r 1-3-1-5/1 (R 7-45 mm). Ostergren (1904: 615) recognized rosea as a distinct variety of hispida, followed by Grieg (1907: 42) who noted that specimens from Bergen and Trondheim fjords as well as from Bohuslan in the vicinity of Oslo fjord (i.e. close to shore) are short-rayed whereas those from the deep area of the Skagerrak (500-600 metres) have relatively long rays. The latter form with R/r c. 2/1 and triangular rays forming angular interradial arcs was illustrated by Mortensen (1927, fig. 53, taken from his earlier 'Danmarks fauna') but does not appear to be found on the continental shelf in british waters, only from the bathyal at 900-1400 metres to the NW and W of the British Isles, from which c. 20 specimens are consistently stellate. These include two small syntypes ofLasiaster villosus Sladen, 1889: 372 (synonymized with Pora- niomorpha hispida by Grieg and others), the specimen from Helga st. SR 506 (Fig. 1 1 B, C) named P. villosa by Farran (1913: 17) and others more recently collected in the Rockall Trough and Porcupine Seabight. The only exception is the holotype of Rhegaster murrayi Sladen, 1889: 368-371 (Fig. 11 D, E) (another synonym of P. hispida) from the Wyville Thomson Ridge in 5 10-790 metres, which has a near pentagonal form, R/r 1 -3/1 but R only c. 14mm. The type locality of Poraniomorpha rosea Danielssen & Koren, 1881: 189-192; also 1884: 67-70, the oldest species-group name for stellate european specimens, is NW of Bergen (61°41'N, 03° 19 'E) in 402 metres, that is in the southern arm of the Norwegian Sea which leads to the Skagerrak. On the basis of this evidence, it is possible that short and long rayed specimens are isolated in different water masses. However, Madsen (pers. comm.) finds considerable overlap in norwegian waters. Nevertheless I believe that rosea can be accorded at least subspecific rank. It should be noted that rosea is antedated by two other names long synonymized with P. hispida, namely Asterina borealis Verrill, 1878: 213-214 and Porania spinulosa Verrill, 1880a: 202-203 (Fig. 11F), based on moderately long-armed type material from american waters N and E of Cape Cod. R/r of the respective types is 12/7 = 1-7/1 (implying a higher value when fully-grown) and 40/23 = 1-75/1. Paucity of american material for comparison prevents a proper assessment of the affinites of specimens from east and west and I can only note that the american ones appear to have the interradial arcs more curved and the tips of the rays blunter than is usual in european ones. In 1895 (p. 139) Verrill noted that spinulosa was taken 'mostly in the warm area' and borealis in the 'cold area' but in 1880 (b: 401) he had recorded both from USFC stations 869 and 879 and in 1914 (p. 18) he remarked that his 1895 notes (p. 139) on a relatively large specimen (R/r 35/23 mm) from the Fishing Banks (c. 45-5°N, 57°W, in 170 fathoms) refer not to borealis but to spinulosa. With such confusion and overlap it is impossible to assess if two infraspecific taxa exist in american waters until more material is available. With regard to the decalcified adult specimens such as have been referred to Culcitopsis borealis (Sussbach & Breckner), the range of these appears to parallel to a great extent that of P. hispida rosea. Twelve samples range from the Porcupine Seabight W of southern Ireland N and E to the Faeroe Channel, Shetlands and N Norway (Lofoten Is) in depths down to c. 1000 metres though with a minimum of only 1 10 metres. These show a near- Fig. 9 A, B, Poraniomorpha (Poraniomorphd) hispida (M. Sars), 9 1 .4. 1 5. 1 , Trondheim fjord, R 32 mm, dorsal and side views, partly denuded. C-F, P. (Culcitopsis) borealis (Siissbach & Breckner): C, (as in 8a); D, 1956.5.25.5, E of Shetlands, R 25 mm, both dorsal views; E, F, 1974.1.4.2, E of Wyville Thompson Ridge, R c. 44mm, side views, E wet, external; F dry, internal. Others all wet, x 1^. Fig. 10 A, Poraniomorpha (Poraniomorphd) hispida (M. Sars) (as in 9A, B), ventral view. B-F, P. (Culcitopsis) borealis (Siissbach & Breckner): B, C, (as in 8a) ventral and side views; D, E, (as in 9D) ventral and side views. All wet, x 1^. Fig. 11 A, Poraniomorpha (Culcitopsis) borealis (Siissbach & Breckner), National Museum of Ireland no. 102.1913, Helga st. SR 223, R c. 40mm, ventral view. B,C, P. (Poraniomorphd) hispida rosea Danielssen & Koren, Nat. Mus. Ireland 403.1913, Helga st. SR 506, R 28 mm, ventral and dorsal views. D-F, P. hispida hispida (?): D, E, holotype of Rhegaster murrayi Sladen, 90.5.7.545, Triton st. 5, Wyville Thomson Ridge, R c. 14 mm, dorsal and ventral views; F, presumed holotype of Porania spinulosa Verrill, Peabody Museum, Yale no. 9867, off Cape Cod Light, R 40 mm. A, wet; others dry; D, Ex 2; others x 1^. Fig. 12 X-rays of Poraniomorpha (Culcitopsis) borealis (Siissbach & Breckner): A (as in 8a); B (as in 9D); C, IOS st. 50601, Porcupine Seabight, R 35 mm; D, 1965.5.24.4, Lofoten Is, R. c 72mm.xli. Fig. 13 X-rays of Poraniomorpha (Culcitopsis) borealis (Siissbach & Breckner): A, IOS st. 9752, Porcupine Seabight, R 40mm; B, 1966.1.13.45, SE of Wyville Thomson Ridge, R. c. 47 mm.x U. 40 A. M. CLARK pentagonal outline, R/r < 1-5/1, are more or less high and cushionlike with the body wall thickened and agree with Siissbach & Breckner's holotype of C. borealis, taken NE of the Shetlands in 1 34-2 1 5 metres. The papulae are in close clusters, the upper surface otherwise appearing fairly smooth superficially but studded with numerous embedded fine spinules visible under magnification, the ventral surface somewhat pustular and the adambulacral spines heavily sheathed. Most of these characters are at variance with 'typical' Poraniomor- pha hispida where the body is flattened though thick and the superficial armament is distinct and almost continuous, covered over with only thin skin. X-rays of six of the borealis-\ike specimens ranging in size from R 19-72 mm are shown in Figs 12 and 13. As would be expected, the maximum calcification of the skeleton appears in the smallest where the interbrachial septa are partly calcified. However, even here many of the marginals and actinals (or abactinals) showing in the interradii have a fairly large cen- tral cavity and the body shape (Fig. IOC) is markedly inflated, much as in Greig's specimen of similar size (1927: 132-133, figs 3-5) from Michael Sars st. 32 (W of Kristiansund, Norway, 400 metres, upon which (rather than his larger ones) I suspect Grieg based his obser- vation that 'the skeleton of the disc is well developed and agrees completely with that of Poraniomorpha hispida ' though he notes that the surface armament is hidden in the thick skin. Although the rate of calcite resorption varies to some extent as shown in the X-rays (compare Figs 12C & 13 A, B), at R>40 mm only vestigial outlines of most plates are evi- dent, at best. This compares with a flattened Trondheim specimen of P. hispida (Figs 9A, B, 10A, 14) with R c. 32 mm in which the abactinal and actinal plates and interbrachial septa appear well calcified and the marginals solid and blocklike. Even in a specimen of hispida with R > 50 mm, from the Skagerrak in 660 metres, X-ray kindly sent by Dr Madsen, the skeletal development still appears much the same except that the interradial plates of one of the marginal series are reduced and hollowed to a similar extent as the corresponding plates of the borealis with R only 19 mm (Fig. 12 A). Clearly there should be considerable similarity in the skeletons of juvenile borealis and hispida, the main skeletal differences Fig. 14 X-ray of Poraniomorpha (Poraniomorpha) hispida (M. Sars) (as in 9A, B). x \{. ASTEROIDEA 41 being in the timing and extent of resorption. However, the resultant morphological difference in well-grown specimens is so marked (compare also the side views shown in Fig. 9B, E) that I find it impossible to believe there is insufficient genetic difference to justify a specific distinction for borealis, as well as a subgeneric one for Culcitopsis within the genus Poranio- morpha. My conclusion that borealis is more than just a form of hispida is supported by a notable enlargement and distal inclination of a single pair of suboral spines on each jaw of most specimens of borealis (see Fig. 11 A), suggesting a modification in feeding habits (perhaps approaching that ofOdontaster which has similar projecting but hyaline- tipped oral spines used for rasping on sponges). A similar modification, but of the apical pair of oral spines, is shared by Poraniomorpha bidens Mortensen, 1932: 9-12 from Greenland (recently taken also in the cold area of the Faeroe Channel NE of the Wyville Thomson Ridge). P. bidens has a furlike coating of very fine superficial spinules of papillae clearly visible through the thin skin, as in P. hispida, but a general body form with tapering pointed rays, as in the other arctic species, P. tumida (Stuxberg) where the superficial armament is much coarser, almost granuliform. Additionally, the colour in life of borealis appears to be generally much paler than that of hispida, the darkest cited being pale orange above; it is more often yellow or pale yellow, white below, whereas hispida is said to be: rose red above, orange below; dark violet-red above with white papulae, reddish-white below, or pale reddish-yellow all over. As mentioned previously (p. 34), intermediates exist in northern Norway between the polymorphic hispida and tumida where the two taxa overlap. Clearly the taxonomy of the entire genus Poraniomorpha needs to be reviewed, not just the Atlantic members within the scope of the present study. It should be noted that in addition to the natural differences correlated with the skeletal development, decalcified specimens can be drastically modified in appearance by changes in preservation. For instance, the large specimen, R 72 mm (Fig. 12D) from the Lofoten Is in the British Museum collection, thought to be P. (C.) borealis is badly flattened with the upper side crinkled and the whole body wall excessively contracted so that the superficial spinules are brought together in an almost continuous coating, as is usual in Poraniomorpha hispida, the identification it had prior to being X-rayed. The extent of shrinkage possible in these poraniids is shown by the holotype of Sphaeriaster berthae (Dons, 1938: 163-164), from N of the Lofoten Is, which had R 90-1 15 mm in life but was only 77-80 mm after preservation in spirit. Change in body shape may also be drastic, as shown by Spoladaster veneris (Perrier), from St. Paul's I, southern Indian Ocean where live specimens may be markedly stellate but preserved ones become pentagonal — see A.M.C., 1976, pi. 6 and pi. 3, fig. 2. Madsen and I have no doubt that S. berthae is synonymous with borealis so that Sphaeriaster itself is a synonym of Poraniomorpha. Dons's second nominal species, S. bjoerlykkei, (1938: 165-168) type locality N of the Shetlands in 300-350 metres, R/r 87/47 = 1-7/1 so fairly stellate, shows a high density of superficial spinules as in our Lofoten Is specimen just mentioned. A new X-ray sent by Madsen shows very faint outlines of plates, much as in large borealis, but he thinks that it is more likely to be decalcified P. tumida', Dons described multiple furrow and subambulacral spines, more than usual in borealis. Another taxon with a very reduced skeleton in larger specimens is Spoladaster Fisher. In 1976 (in Clark & Courtman-Stock: 73) I suggested that Tylaster meridionalis Mortensen, 1933: 249-250, from the same area W of South Africa, based on a specimen with R only 28 mm, is probably a synonym of S. brachyactis (H. L. Clark, 1923: 293-294), of which R is 40-80 mm in the few specimens recorded. Studies now on the growth changes and vari- ation of P. (C.) borealis confirm me in this view. It is noteworthy that no better-calcified Poraniomorphas have been collected in south african waters. S. brachyactis shows some deve- lopment of macroscopic inferomarginal and actinal spines, such as are found in greater numbers in Tylaster willei Danielssen & Koren, 1881: 186, from the northern Norwegian Sea (see Danielssen & Koren, 1884: 64-67). This species too has the underlying skeleton very reduced. These taxa illustrate the ability of poraniids to utilize coarse armament even though this is, at best, articulated only to rudiments of skeletal plates in the thickened body 42 A. M. CLARK wall. Tylaster and the other arctic taxa mentioned come within the range of the series 'Marine Invertebrates of Scandinavia', the asteroid part of which is in preparation by Madsen. Hopefully he will be able to clarify the relationships of these if more material is available. There is yet one more conspicuous example of skeletal reduction in poraniids, exemplified by Pseudoporania Dons. Again I am indebted to Madsen for an X-ray, of the holotype of P. stormi Dons, 1936: 1 7-20, from Trondheim fjord in 300 metres, R 83-96 mm. This shows that the actinal and marginal plates have contracted down into very small, widely separated rods or partially hollow nodules of calcite, the more interradial inferomarginals being appar- ently reduced to a rudiment of their abradial, possibly also adradial ends. This is just the progression I would expect from the condition found in six much smaller specimens, R 19^40 mm, one from the Porcupine Seabight, the others from around the Wyville Thomson Ridge, S of the Faeroes, in depths of 360 (?183)-770 (7927) metres. These too have a smooth surface above and below, apart from well-marked actinal grooving. Sectioning shows the lateral body wall to be extremely thick (Fig. 5B) and X-rays show no signs of inferomarginal spines, even on the distal plates in the smallest (Fig. 1 6A, B), the ambitus being rounded. The body form is flattened and pentagonal whereas Don's specimen has short tapering rays. The papulae are relatively sparse and scattered. The adambulacral plates are armed with single furrow and subambulacral spines (the sheath of the latter continuous with the thick- ened ventral body wall), which appeared to provide a distinction from stormi but Madsen informs me that it too has single spines, Dons' description of 2 + 2 being incorrect. The smallest specimen has the marginals blocklike except for the interradial inferomarginals which project abradially. The sections and X-rays show progressive attenuation of the plates with division of the interradial inferomarginals into two small end pieces, ad- and abradial, by loss of the middle part. This is very different from the resorption shown by Poranio- morpha (C.) borealis, which is almost entirely from the inside, the plates being reduced to hollow, usually rectangular or ovate, shells. The complete absence of any superficial spinules and the small number of adambulacral spines, with only single furrow spines, agrees more closely with Porania than any other genus of the family, though the great thickening of the body wall and the absence of a distinct ventrolateral angle emphasized by a horizontal fringe of individually sheathed inferomarginal spines results in a very different appearance. In 1983 (in Gage et al.: 281) I noted that a specimen of Porania pulvillus (O. F. Miiller, 1776: 234) from the Rockall Bank in 148 metres with R 55 mm has the inferomarginal spines drastically reduced from the usual 3-5 on each plate to only 1 or 2 on some of the more interradial plates and none on the distal plates. Nevertheless, the remaining spines are indi- vidually sheathed and projecting from the ambitus and the usual ventrolateral angle is still distinct, the body wall not being markedly thickened. An X-ray of this specimen (Fig. 18B) shows that a few of the interradial inferomarginal plates are slightly compressed, recalling those of the smallest specimen of stormi (Fig. 16A), though the modification is much less. Madsen (pers. comm.) has found occasional specimens of P. pulvillus from Norway with the inferomarginal spines more or less reduced but the skeleton otherwise well developed. Addi- tionally he has sent X-rays of two other specimens, R probably 25-30 mm, with the inter- radial marginals much reduced, some divided into two parts and the body wall obviously much thickened. Although he finds these akin to Pseudoporania stormi, he considers this to be a synonym of Porania pulvillus. One (from S of Iceland, Thor st. 166) appears to have nearly all the marginals narrowed down and completely lacking spines, much as in the Porcupine Seabight specimen (Fig. 16B) but the other (Troms0 Museum, probably from N Norway) has about 3 inferomarginal plates each side of the very reduced interradial plates in each interbrachial arc with a rhombic abradial part bearing 1, rarely 2, large spatulate spines. In face of such intermediate specimens, there can be little doubt that Pseudoporania should be referred to the synonymy of Porania, in a comparable way to Culcitopsis and Poraniomorpha. However, the general form of adults of the several Porania species, with a distinct ventrolateral angle and the body wall no more than moderately thickened is so Fig. 15 Porania (Pseudoporanid) stormi (Dons): A, B, (as in 5B), dorsal and ventral views; C D. Royal Scottish Museum, Walter Herwig st. 848, S of Faeroe Is, R 19 mm, dorsal and ventral views; E, IOS st. 50601, Porcupine Seabight, R 35 mm. All wet,x 1^. 44 A. M. CLARK Fig. 16 X-rays of Porania (Pseudoporanid) stormi (Dons): A (as in 1 5B, C); B (as in 1 5E). x 1^. consistent that here again I believe a subgeneric distinction is justified, in spite of Madsen's opinion to the contrary. It seems likely that P. stormi is zoogeographically isolated from pulvillus. The present records indicate that pulvillus alone is found on the shelf around the British Isles and in southern Norway N to about Trondheim fjord, to a minimum depth of Fig. 17 Porania (Porania) pulvillus (O. F. Muller): A (as in 5C); B, C, 1950.1 1.3. 1, Porcupine Bank, R 55 mm, dorsal and part ventral views. Both wet,x \\. Fig. 18 X-rays of Porania (Porania) pulvillus (O. F. Miiller): A (as in 5C); B, 1981.3.24.28, Rockall Bank, R 55 mm (the inferomarginal spines only show in one interradius though in fact present in the others; a section of one arm removed), x 1^. ASTEROIDEA 47 5 metres (a new record from Lunna, Shetland Is where it was collected by a diver) and a maximum of c. 250 metres, whereas P. stormi is an upper bathyal species limited to more remote areas, from S of Iceland to the continental slope W of the British Isles and possibly from N Norway. However, there is a much closer morphological resemblance between Porania pulvillus and the american P. insignis Verrill, 1895: 138-139. The former is consistently relatively thin-walled with a well-marked ventrolateral angle. Although the available material suggests that the abactinal skeleton may be more open in the american taxon, the main difference is that some adults retain a few actinal spines whereas these are almost invariably lost at an early stage in P. pulvillus. Madsen (pers. comm.) has independently reached the same conclusion. To provide a summary of the main diagnostic characters used in identification of poraniids, a tabular key to the genera and subgenera now accepted is given here (Table 2). Table 2 Tabular key to the genera of Poraniidae. Alternate columns in lower case. Poraniopsis W s(i) C-O r SD q-j- T Porania (Porania) W(R) n,s(i) O a SC q + / — T Porania (Pseudoporanid) R n o r L q- T Tylaster R i 7 r S*(D) q+ T Spoladaster R i 0(C) r SD q,p( + ) T Chondraster R n o a(r) H q-[+] M Poraniomorpha (Culcitopsis) R i c r L[M] P- M(T) Poraniomorpha (Poraniomorpha) W f C(S) r M s + /- M(T) Marginaster [W] [s] [S] [a] [SC] [q-] U] Note: Square brackets indicate occurrence in small specimens, as throughout in Marginaster; round brackets show the condition in occasional specimens or a modified form. *In Tylaster willei the inferomarginal spines are evidently in triangular groups of three, not a horizontal line. 1 . Abactinal skeleton: H-hidden in live or well-preserved specimens, the W-well developed few large spines wholly within the much thick- R-more or less reduced by resorption, especially in ened body wall large specimens, R > 50 mm L-lacking altogether 2. Superficial dorsal body wall: M-of multiple spinelets clustered along the ventro- f-with very fine continuous or clustered spinules, lateral convexity tubercles or papillae, not necessarily articulated S-of up to 5 large spines in a horizontal row, at to the underlying plates least their tips projecting, forming either a con- i-with fine isolated spinules, not articulated to the tinuous fringe (C) or a discontinuous grouped plates series (D) n-naked and smooth, sometimes with surface spic- 6. Appearance of actinal areas (apart from the ciliated ules so dense as to show a pale colour when dried grooves between the furrows and margins): s-with spaced relatively large spines mounted on p-pustular the larger plates or vestiges of plates q-quite smooth (apart from any macroscopic 3. Papulae: armament) C-clustered s-with fine superficial spinules, usually slightly O-in open groups, short arcs or evenly spaced over spaced wide areas +/— with or without enlarged spinelets or spines S-single in series parallel to the inferomarginals 4. Shape of margin: 7. Adambulacral armament: a-more or less distinctly angular ventrolaterally, T-arranged normally in one series transverse to the corresponding to the horizontally projecting furrow, usually 2 or 3 spines inferomarginals M-with multiple furrow spines on most plates, r-rounded, the inferomarginals hardly, if at all, subambulacral spines variously arranged, paired, projecting or else both series of plates reduced in an oblique but nearly longitudinal series with- 5. Inferomarginal armament: in a common sheath, or transversely 48 A. M. CLARK Nomenclature The classification of the Poraniidae has been complicated not only by the thick skin obscuring the usual diagnostic characters afforded by the skeleton but also by failure to allow for ontogenetic changes and an unwise propensity of certain early workers to give new names to juvenile or small specimens. Consequently, the names of certain species-group taxa are threatened by the possibility that they will be proved to be synonymous or homonymous with prior nominal species, as follows: Poraniomorpha (Culcitopsis) borealis (Siissbach & Breckner, 191 1) is threatened by two possibilities, firstly: Asterina borealis Verrill, 1878 (holotype extant in the Peabody Museum, Yale, R 12mm), long synonymized with P. hispida, may prove to be consubspecific with P. (Poraniomorpha) hispida rosea Danielssen & Koren, 1881, which it antedates. In this eventuality and if the subspecies now proposed is accepted, then borealis Verrill would be a senior species-group homonym within the genus Poraniomorpha. Secondly: Marginaster pentagonus Perrier, 1882 (holotype extant in the Paris Museum, R only 3 mm) may prove to be a senior synonym. The name pentagonus has only been mentioned as a possible synonym since Perrier, 1 894. Porania pulvillus insignis Verrill, 1895 is threatened by: Asterina pygmaea Verrill, 1878 (holotype extant in the Peabody Museum, R only 5 mm), which may prove to be a senior synonym. The name pygmaea has been unused since referred to Poranisca by Verrill, 1914. Porania antarctica Smith, 1876 is threatened by: Astrogonium fonki Philippi, 1858, which Madsen (1956) has little doubt was based on specimens con- specific with P. antarctica magellanica Studer, 1876 but which he assumed are no longer extant in any Chilean collection since they were not mentioned in Meissner's note on Philippi's asteroids of 1898. The name fonki has been unused since 1858 but P. antarctica is widely utilized. Summary of taxonomic confirmations or changes Poraniella Verrill, 1914, referred to the family Asteropseidae from Poraniidae. Poraniopsis Perrier, 1891, referred to the family Poraniidae from Echinasteridae. Poranisca Verrill, 1914, with type species P. lepidus Verrill, 1914, synonyms of Marginaster Perrier, 1881 and M. pectinatus Perrier, 1881. Chondraster Verrill, 1895, confirmed as of generic rank, distinct from Porania Gray, 1840. Poraniomorpha rosea Danielssen & Koren, 188 1 , treated as a subspecies rather than a form or variety of P. hispida (M. Sars, 1872). Culcitopsis Verrill, 1914, type species Culcita borealis Siissbach & Breckner, 191 1, treated as subgenus of Poraniomorpha Danielssen & Koren, 1881. Culcitopsis borealis (Siissbach & Breckner), treated as a separate species rather than a form of Poranio- morpha hispida (M. Sars). Sphaeriaster Dons, 1939, type species Sphaeraster berthae Dons, 1938, synonyms of Poraniomorpha (Culcitopsis) Verrill and P. (C.) borealis (Siissbach & Breckner). Tylaster meridionalis Mortensen, 1933, confirmed as a synonym of Spoladaster brachyactis (H. L. Clark, 1923). Pseudoporania Dons, 1936, type species P. stormi Dons, 1936, a subgenus of Porania Gray. Porania insignis Verrill, 1895, reduced to a subspecies of P. pulvillus (O. F. Miiller). Taxa the affinities of which need further investigation: Asterina borealis Verrill, 1878 and Porania spinulosa Verrill, 1880, as infraspecific taxa within, rather than pure synonyms of, Poraniomorpha hispida (M. Sars). Marginaster austerus Verrill, 1899, in relation to M. pectinatus Perrier. Marginaster fimbriatus Sladen, 1889, in relation to Chondraster grandis (Verrill, 1878). Marginaster pentagonus Perrier, 1882, in relation to Poraniomorpha hispida borealis (Siissbach & Breckner). ASTEROIDEA 49 Sphaeriaster bjoerlykkei (Dons, 1938), in relation to Poraniomorpha hispida borealis (Siissbach & Breckner) and P. tumida (Stuxberg, 1878). Tylaster Danielssen & Keren, 1881, with type species T. willei Danielssen & Keren, 1881, in relation to Chondraster Verrill, 1895 and Porania Gray. Spoladaster Fisher, 1940, with type species Cryaster brachyactis H. L. Clark, 1923, in relation to Poraniomorpha Danielssen & Koren. Acknowledgements A particular debt is owed to Dr F. Jensenius Madsen, Universitetets Zoologisk Museum, Copenhagen, for extended consultation, examination of type material and provision of X-rays for comparison, though our conclusions as to taxonomic weighting of certain characters did not always coincide. Other X-rays were made in the British Museum by Mr G. Howes and Miss B. Brewster, Fish Section. I am also indebted to the following for provision of material for study: Mr D. Billett, Benthic Group, Institute of Oceanographic Sciences, Wormley; Miss S. Chambers, Royal Scottish Museum, Edinburgh; Miss M. E. Downey, United States National Museum, Smithsonian Institution, Washington, D.C.; Dr J. D. Gage, Scottish Marine Biological Association, Oban; Dr. Willard'D. Hartman, Peabody Museum, Yale, New Haven, Conn.; Dr B. Hecker, Lament- Doherty Geological Observatory, Palisades, New York; Mr M. Holmes, National Museum of Ireland, Dublin; Dr M. Sibuet, Centre Oceanologique de Bretagne, Brest; Dr G. Smaldon (formerly of) Royal Scottish Museum and Mr A. C. Wheeler, Fish Section, BM (NH). References Bell, F. J. 1893. Catalogue of the british echinoderms in the British Museum (Natural History). xvii + 202 pp. London: Trustees of the British Museum. Blake, D. B. 1980. On the affinities of three small sea-star families. Journal of Natural History. 14: 163-182. 1981. A re-assessment of the sea-star orders Valvatida and Spinulosida. Journal of Natural History. 15: 375-394. Cherbonnier, G. & Sibuet, M. 1973. Resultats scientifiques de la campagne Noratlante: Asteroides et Ophiurides. Bulletin du Museum National d'Histoire Naturelle, Paris. (Zoologie) No. 76 [1972]: 1333-1394. Clark, A. M. 1976. Asterozoa from Amsterdam and St Paul Islands, southern Indian Ocean. Bulletin of the British Museum (Natural History) (Zoology) 30: 247-261. Clark, A. M. & Courtman- Stock, J. 1976. The echinoderms of southern Africa. 271 pp. London: Publications of the British Museum (Natural History) No. 776. Clark, H. L. 1923. The echinoderm fauna of South Africa. Annals of the South African Museum. 13: 221-435. Danielssen, D. C. & Koren, T. 1881. Fra den norske Nordhavs-expedition: Echinodermer. Nyt Magazin for Naturvidenskaberne. 26: 177-195. 1884. Asteroidea. Den norske Nordhavs-expedition, 1876-1878. 1 19 pp. Christiania. Djakonov, A. M. 1946. [Individual and age variability in some groups of echinoderms.] Trudy Zoologicheskogo Instituta, Leningrad 8: 145-193. 1950. [Sea stars of U.S.S.R. seas.] Opredeliteli po Faune SSSR. no. 34: 1-199. [English translation: Jerusalem. 1968. 183 pp.] Dons, C. 1936. Zoologische Notizen. 26. Pseudoporania stormi n. gen., n. sp. Kongelige Norske Videnskabernes Selskabs Forhandlinger. 8: 17-20. 1938. Zoologische Notizen. 35, 36. Sphaeraster Berthae n. gen., n. sp. Sphaeraster Bjorlykkei n. sp. Kongelige Norske Videnskabernes Selskabs Forhandlinger. 10: 161-168. 1939. Zoologische Notizen. 37. Die Asteriden Gattung Spaeriaster, nomen novum. Kongelige Norske Videnskabernes Selskabs Forhandlinger. 11: 37. Downey, M. E. 1973. Starfishes from the Caribbean and the Gulf of Mexico. Smithsonian Contributions to Zoology No. 126: 1-158. Farran, G. P. 1913. The deep-water Asteroidea, Ophiuroidea and Echinoidea of the west coast of Ireland. Scientific Investigations. Fisheries Branch, Department of Agriculture for Ireland. 6: 1-66. Fisher, W. K. 1908. Necessary changes in the nomenclature of starfishes. Smithsonian Miscellaneous Collections. 52: 87-93. 50 A. M. CLARK 1911. Asteroidea of the North Pacific and adjacent waters. 1 . Bulletin of the United States National Museum. 76(1): vi + 419 pp. 1940. Asteroidea. Discovery Reports. 20: 69-306. Gage, J. D., Pearson, M., Clark, A. M., Paterson, G. L. J. & Tyler, P. A. 1983. Echinoderms of the Rockall Trough and adjacent areas. 1 . Bulletin of the British Museum (Natural History) (Zoology) 45: 263-308. Gallo, V. R. 1937. Sur le genre "Culcitopsis" Verrill (Asteroides). Compte Rendu XIIe Congres International Zoologique. Lisbon, 1935: 1664-1667. Gasco, F. 1876. Descrizione di alcune Echinodermi nuovi o per la prima volta trovato nel Mediterra- neo. Rendiconti dell'Accademia delle Scienze Fisiche e Matematiche. Napoli. 15(2): 9-11. Gray, J. E. 1840. A synopsis of the genera and species of the class Hypostoma (Asterias, Linn.). Annals and Magazine of Natural History 6: 1 75-1 84; 275-290. Grieg, J. A. 1907. Echinodermen von dem norwegischen Fischereidampfer "Michael Sars" in den Jahren 1900-1903 gesammelt. Asteroidea. Bergens Museum Arbog 1906(13): 1-87. 1927. Echinoderms from the west coast of Norway. Nyt Magazin for Naturvidenskaberne. 65: 127-136. Hotchkiss, F. H. C. & Clark, A. M. 1976. Restriction of the family Poraniidae, sensu Spencer & Wright. Bulletin of the British Museum (Natural History). (Zoology) 30: 263-268. Koehler, R. 1921. Echinodermes. Faune de France 1: 1-210. 1924. Les Echinodermes des Mers d'Europe. 1. 360 pp. Paris: Librarie Octave Doin. Leipoldt, F. 1895. Asteroidea der "Vettor-Pisani" Expedition (1882-1885). Zeitschrift Jur Wissen- schaftliche Zoologie. 59: 545-654. Ludwig, H. 1897. Die Seesterne des Mittelmeeres. Fauna und Flora des Golfes von Neapel. 24: 1-491. Madsen, F. J. 1956. Reports of the Lund University Chile Expedition, 1948-49. 24. Asteroidea. Acta Universitatis Lundensis. Andra Afd. N.S. 52(2): 1-53. 1959. A new North Atlantic sea-star, Chondraster hermanni n. sp., with remarks on related forms. Videnskabelige Meddelelser fra Dansk Naturhistorisk Forening. 121: 153-160. Meissner, M. 1898. Ueber chilenische Seesterne. Zoologischer Anzeiger. 21: 394, 395. Mortensen, T. 1912. Report on the echinoderms collected by the Danmark-Expedition at North-East Greenland. Meddelelser om Gronland. 45: 237-302. 1927. Handbook of Echinoderms of the British Isles, ix + 471 pp. London: Oxford University Press. 1932. Echinoderms of the Godthaab Expedition of 1928. Meddelelser om Gronland. 79: 1-62. 1933. Echinoderms of South Africa (Asteroidea and Ophiuroidea). Videnskabelige Meddelelser fra Dansk Naturhistorisk Forening. 93: 2 1 5-400. Muller, O. F. 1776. Zoologiae danicae prodromus. xxxii + 282 pp. Havniae. 6'stergren, H. 1904. Ueber die arktischen Seesterne. Zoologischer Anzeiger. 27: 615-616. Perrier, E. 1881. Description sommaire des especes nouvelles d'Asteries. Bulletin of the Museum of comparative Zoology, Harvard. 9: 1-31. 1882. In: Milne Edwards. Rapports sur les travaux de la Commission chargee par M. le ministre de PInstruction publique d'etudier la faune sous-marine dans le grandes profundeurs de la Mediter- ranee et de FAtlantique. Archives des Missions Scientifiques et Litteraires. Paris. (3)9: 46-49. 1884. Memoire sur les etoiles de mer recueillies dans la Mer des Antilles et la Golfe de Mexique. Nouvelle Archives du Museum d'Histoire Naturelle, Paris (2)6: 127-276. 1 89 1 . Echinodermes. 1 . Stellerides. Mission Scientifique du Cap Horn. 1882-1883. 6: Zoologie. (3) 198 pp. Paris. 1893. Traite de Zoologie. 1 & 2. 864 pp. Paris: Librairie F. Savy. 1 894. Stellerides. Expeditions scientifiques du Travailleur et du Talisman. 43 1 . pp. Paris. Philippi, R. A. 1858. Beschreibung einiger neuer Seesterne aus dem Meere von Chiloe. Archiv Jur Naturgeschichte. 24: 264. Sars, M. 1872. Tillaeg. In: Sars, G. O. Nye Echinodermer fra den Norske Kyst. Forhandlinger i Videnskabsselskabet i Kristiania. 1871: 27-31. 1877. New echinoderms. In: Keren & Danielssen [Eds] Fauna littoralis Norvegiae. 3: 49-75. Bergen. Sladen, W. P. 1889. Asteroidea. Report of the scientific results of the voyage of H. M.S. Challenger, 1873-76. Zoology, 30: 1-935. Spencer, W. K. & Wright, C. W. 1966. Asterozoans. In: Moore, R. C. [Ed.] Treatise on Invertebrate Paleontology. U. Echinodermata 3(1): 4-107. Geological Society of America Inc.: University of Kansas Press. ASTEROIDEA 5 1 Stuxberg, A. 1878. Echinodermer fran Novaja Zemljashaf samlade under Nordenskioldska Expeditio- nerna, 1875 och 1876. Ofversigt af Kongl, Vetenskaps-Akademiens Forhandlingar. Stockholm. 1878(3): 27^0. Sussbach, S. & Breckner, A. 191 1. Die Seeigel, Seesterne und Schlangensterne der Nord- und Ostsee. Wissenschaftliche Meeresuntersuchungen der Kommission zur Wissenschaften Untersuchung der Deutschen Meere. Abteilung Kiel. N.S. 12: 167-300. Tortonese, E. 1965. Echinodermata. Fauna d'ltalia. 6: xv + 422. Bologna: Edizioni Calderini. Verrill, A. E. 1878. Notice of recent additions to the marine fauna of the eastern coast of North America. 1, 2. American Journal of Science. (3) 16: 207-215; 371-378. 1880a. Notice of recent additions to the marine Invertebrata of the northeastern coast of America. Proceedings of the United States National Museum. 2: 165-205. 18806. Notice of the remarkable marine fauna occupying the outer banks off the southern coast of New England. American Journal of Science. (3)20: 390-403. 1885. Results of the explorations made by the steamer "Albatross" off the northern coast of the United States in 1883. Report United States Commissioner of Fisheries. 1883: 503-543. 1895. Distribution of the echinoderms of north-eastern America. American Journal of Science. 49: 127-141; 199-212. 1899. Revision of certain genera and species of starfishes with descriptions of new forms. Transactions of the Connecticut Academy of Arts and Sciences. 10: 145-234. 1914. Revision of some genera and species of starfishes. Annals and Magazine of Natural History. 14: 13-22. 1915. Report on the starfishes of the West Indies, Florida and Brazil. Bulletin from the Laboratories of Natural History of the State University of Iowa. 7(1): 1-232. The larval and post- larval development of the Thumb-nail Crab, Thia scutellata (Fabricius), (Decapoda: Brachyura) R. W. Ingle Department of Zoology, British Museum (Natural History), Cromwell Road, London SW7 5BD Introduction The Thumb-nail crab, Thia scutellata (Fabricius) has been reported from the west coast of Sweden, off German, Belgium and Netherlands coasts, in the southern North Sea, from western sea areas of the British and Irish coasts, and southwards to the Mediterranean and west coast of Africa (see Christiansen, 1969; Ingle, 1980; Manning & Holthuis, 1981). The larval and early crab stages have been figured by Claus (1876), Cano (1892), Lebour (1928) and Williamson (1915). Except for Claus, these authors also gave very brief descriptions of stages, Williamson's being based on Claus' and Cano's figures; the accounts relate chiefly to plankton caught material and are somewhat superficial. In 1975 Mr T. Farrall, a Deputy Head of Langdon Park School, London, obtained two ovigerous crabs of T. scutellata from the Channel Islands while studying the relationships of this species with the Purple Heart Urchin, Spatangus purpureus O. F. Muller. The live crabs were donated to the British Museum (Natural History) and the eggs of one hatched on 18th September 1975. Sufficient material was reared through to fifth crab stage to enable a complete account to be given of the larval and early post- larval development of this species in the laboratory. Materials and methods The female, from which the larvae and post-larvae were reared, was collected near Vermerette Rock, Herm, Channel Islands in September 1975. The larvae were reared using methods described by Rice & Ingle (1975) and Ingle & Clark (1977). All material was fixed and stored in the preservative formulated by Steedman (1976: 148) and later transferred to 70% alcohol. Drawings and measurements were made with the aid of a camera lucida. Measurements are as follows: T. T. = total lengths of zoeae, measured between tips of dorsal and rostral spines; C. L. = carapace lengths, measured from between eyes to posterio- lateral carapace margin for zoeae, from rostral tip (for megalopa) and from frontal margin (for crab stage), to median posterior margin of carapace. The female and reared material are deposited in the Collections of the Zoology Department, British Museum (Natural History), registration number: 1983:312. Descriptions Thia scutellata (Fabricius, 1793) Thia polita: Claus, 1876: 56, Tav. X, figs 1-7 (1st zoea); Cano, 1892: 7, Tav. II, figs 16-26 (A-E) (lst-3rd zoeae, megal. 1st crab); Thia residuus: Williamson, 1915: 548-50, figs 470-479 (1st and later zoeae, megal. after Claus & Cano); Thia polita: Lebour, 1928: 528-9 (lst-4th zoeae, megal. 1st crab described), fig. 5 (17), PI. I, fig. 1 1 (2nd zoea, coloured), PI. VIII, figs 7-8 (3rd zoeae, megal. 1st, 2nd crab); Bourdillon-Casanova, 1960: 167. Bull. Br. Mus. nat. Hist. (Zool.) 47(1): 53-64 Issued 28 June 1984 53 54 R. W. INGLE FIRST ZOEA Dimensions: T. T. 2-70 mm, C. L. 0-70 mm. Carapace (Fig. la): Dorsal spine long and straight, stout proximally and narrowing distally; rostral spine thinner than dorsal and shorter; lateral spines about \ carapace length; dorso- median elevation present; carapace dorsal margin slightly elevated, a pair of prominent posterio- median setules present. Eyes: Partly fused to carapace. Antennule (Fig. 2b): Unsegmented, with 3 terminal aesthetascs and 2 setae. Antenna (Fig. 2f): Spinous process about 2^ x length of exopod, distal \ spinulate; exopod with 1 long and 1 short seta. Mandible (Fig. 2i): Incisor not differentiated from molar process. Maxillule (Fig. 21): endopod 2-segmented, with 1,6 setae; basal endite with 5 and coxal with 7 setae/spines. Maxilla (Fig. 3c): endopod with broad outer and narrower inner lobe with 5 + 3 setae; basal endite with outer lobe slightly broader than inner, with 4 + 5 setae; coxal endite with outer lobe slightly broader than inner, with 4 + 3 setae; scaphognathite with 4 plumose setae and a very stout posterior plumose projection. First maxilliped (Fig. 4a): Basis with 10 setae arranged 2,2,3,3; endopod 5 -segmented with 2,2,1,2,4+ 1 setae; exopod incipiently segmented with 4 terminal plumose setae. Second maxilliped (Fig. 4c): Basis with 4 setae; endopod 3-segmented with 1,1, 3 + 1 setae; exopod with 4 terminal plumose setae. Third maxilliped: Not developed. Pereiopods: Not developed. Abdomen (Figs le, 2a): 5 segmented + telson; 2nd segment with a pair of dorso- lateral pro- cesses; posterio- lateral margin of 1st truncate, of 2nd-5th rounded to sub-acute and minutely spinulate. A pair of small setae near posterio-dorsal margin of segments 2-5. Telson forks long and thin, each with thin long lateral and dorsal spine; inner medio-lateral margin of telson with 6 setae all similarly plumed; median margin of telson strongly concave. SECOND ZOEA Dimensions: T. T. 3-1 mm, C. L. 0-75 mm. Carapace (Fig. Ib): Posterior margin with 3-4 long setules; two pairs of dorso- median setae now present. Eyes: Now stalked. Antennule (Fig. 2c): Now with 6 acesthetascs and 1 seta. Antenna: Unchanged. Mandible: Unchanged. Maxillule (Fig. 2m): Endopod setation unchanged; outer margin of basal endite with a promi- nent plumose seta, dorso-inner margin with 9 setae/spines; coxal setation unchanged. Maxilla (Fig. 3d): endopod setation unchanged; basal endite with 5 + 5 setae and additional seta in some specimens (some distance from margin); coxal setation unchanged; scaphognathite with 1 1 plumose marginal setae. First maxilliped: Basal and endopod setation unchanged; exopod with 6 terminal plumose setae. Second maxilliped (Fig. 4d): Basal setation unchanged; endopod with 1,1,4+1 setae; exopod with 6 terminal plumose setae. Third maxilliped: Not developed. Pereiopods: Not developed. Abdomen (Fig. If): Sixth segment incipient; 1st segment with dorso- median setule; dorsal- lateral processes on 2nd segment less acute and not curved. Inner medio-lateral margin of telson with additional pair of small setules; lateral spine on telson reduced to a very minute setule. BRACHYURA 55 THIRD ZOEA Dimensions: T. T. 3-8 mm, C.L. 1, 30 mm. Carapace (Fig. Ic): Posterior margin with 5 long setules; three pairs of dorso-median setae now present. Eyes: Unchanged. Antennule. (Fig. 2d): two or 3 of the 6 aesthetascs now sub-distal. Antenna (Fig. 2g): Spinous process slightly less than 3 x length of exopod; endopod now developed as a conspicuous bud. Mandible (Fig. 2j): Incisor and molar processes now differentiated. Maxillule (Fig. 3a): Endopod setation unchanged; basal endite with 11 and coxal with 10 setae/spines. Maxilla (Fig. 3e): Endopod, basal and coxal endite setation unchanged; scaphognathite with 18 setae. First maxilliped (Fig. 4b): Basal setation unchanged; endopod terminal segment now with 5 + 1 setae; exopod with 8 terminal plumose setae. Second maxilliped: Basal and endopod setation unchanged; exopod with 8 terminal plumose setae. Third maxilliped: Represented as a small bud. Pereiopods: now represented as small buds. Abdomen (Fig. Ig): Sixth segment now fully differentiated; first segment with 3 conspicuous dorso-median setules; innermost pair of medio- lateral setules on telson much longer than in previous stage; pleopods represented as conspicuous buds on segments 2-5. FOURTH ZOEA Dimensions: T.T. 4-6 mm, C.L. 1-60 mm. Carapace (Fig. Id): Dorsal and rostral spines stouter than in previous stage and lateral spines smaller; posterior margin of carapace with 11-12 setules. Eyes: Unchanged. Antennule (Fig. 2e): Incipiently 2-3 segmented, ultimate segment with 2 aesthetascs and 1 seta, penultimate with 7 distal and 2 sub-distal aesthetascs; endopod bud developed. Antenna (Fig. 2h): Spinous process about 2^x length of exopod; innermost terminal seta of exopod very long; endopod bud very long. Mandible (Fig. 2k): Now with incipient palp. Maxillule (Fig. 3b): Endopod setation unchanged; basal endite now with 16 setae/spines; coxal setation unchanged. Maxilla (Fig. 3f): Endopod setation unchanged; basal endite with 6 + 6-7 setae; coxal setation unchanged. First maxilliped: Basal and endopod setation unchanged; exopod with 10 distal plumose setae. Second maxilliped: Basal and endopod setation unchanged; exopod with 10 distal plumose setae. Third maxilliped: now bilobed. Pereiopods: longer than previous stage and cheliped dactylus differentiated. Abdomen (Fig. 1 h): Dorsal surface of telson with a pair of small setules; pleopod buds well developed, longer than in previous stage. MEGALOPA Dimensions: C.L. 2-2 mm. Carapace (Fig. 4e,f): Longer than broad; frontal region broad, margins almost parallel; a small median furrow near rostral base; rostrum slightly deflected ventrally; cardiac region with suggestion of a broad tubercle; posterior margin with numerous setae. Eyes: Large. Antennule (Fig. 5a): Peduncle 3-segmented, setose as shown exopod incipiently 3-segmented, with 4,4,4,-5 aesthetascs and 0,1,2 setae respectively; endopod incipiently 2-segmented with 5 terminal setae. 56 R. W. INGLE Antenna Fig. 5b): Peduncle 3 -segmented, with 0,2,1 setae respectively; flagellum 8-segmented with 4,0,2,0,4,0,3,4 setae respectively. Mandible (Fig. 5c,d): Molar and incisor processes not differentiated; mandibular palp (d) 3-segmented, terminal segment long and slightly curved, with 8 setules. Maxillule (Fig. 5e): Endopod long and unsegmented, with 2 sub-terminal and 2 terminal setae all reduced; basal endite with 23-24 setae/spines; coxal endite still with 10 setae/spines. Maxilla (Fig. 50: Endopod now reduced to a sub-acute lobe with 1-2 setae; basal endite with 7 + 7 setae, coxal still with 4 + 3 setae; scaphognathite with 4 1-42 plumose setae, shorter than in last zoeal stage. First maxilliped (Fig. 5g): Coxal and basal segments slightly differentiated coxal with 1 1-12 setae on or near inner margin; basis with 28-29 setae; endopod unsegmented and with 3 distal setae; exopod 2-segmented, distal segment with 3 terminal setae; epipod (not shown) with 3-4 setae. Second maxilliped (Fig. 5h): Coxal and basal segments undifferentiated; endopod with only propodus and dactylus demarcated, propodus with 6 setae, dactylus with 5 spines + 1 seta; exopod 2-segmented, distal segment with 3 setae; epipod (not shown) short, with 2-3 setae. Third maxilliped (Fig. 5i): Inner margin of coxa with 1 and basis with 4 setae; inner margin of ischium with 3-4 small denticles and with 10-11 setae, carpus-dactylus with 7,5,6,4-5 setae respectively; epipod (not shown) long and with 10-11 setae. Pereiopods (Fig. 6a-f): Cheliped stout, setose as shown, propodal palm inflated, inner pro- podal and dactylar margins cut into irregular teeth (b); cheliped without coxal or ischial spines. Pereiopods 2-5 (c-f) short and stout, setose as shown, dactyls terminally very acute and those of 5th with 2 long terminal simple setae in addition to 3 prominent, slightly shorter ones on lower margin. Abdomen (Fig. 4g,h, 6g): With 6 segments + telson and setose as shown, posterio- lateral mar- gins broadly truncate. Telson (Fig. 6g) truncate, about as broad as long, dorsal surface with 2 pairs of median setules. Five pairs of pleopods, exopods with long plumose marginal setae, 1st (Fig. 6h) with 13, 2nd 15, 3rd 14, 4th (Fig. 6i) 11, 5th (uropod, Fig. 6g) with 8 setae respectively; endopods of pleopods 1-4 each with 3 distally placed coupling hooks on internal margins. FIRST CRAB Dimensions: C.L. 2-54 mm. Carapace (Fig. 6j): Longer than broad, frontal region projecting; four pairs of anterio- lateral teeth, lst-3rd acute, 4th small and obtuse. Dorsal surface of carapace smooth and anteriorly with minute setules; margins with well developed plumose setae. Remarks The present laboratory reared material of Thia scutellata agrees in most aspects with pre- vious accounts of the larval and post-larval descriptions of this species (see p. 53), except in the following features. (1) The maxillule of the 1st zoea figured by Claus (1876) shows 1 + 5 setae on the endopod and 4 on the coxa and the setal formula for the maxilla is given as 5 + 2, 3+4, 2 + 3 for endopod, basis and coxa respectively; Claus also figured an incipient 3rd maxilliped in this 1st stage zoea. (2) Lebour (1928) depicts 2 dorsal setules on the 1st abdominal segment of the 3rd zoea whereas all the present specimens have 3 setules. (3) Lebour did not examine 1st stage zoeae and assumed that only one spine was present on each telson fork in this stage although Claus clearly shows two spines in his figure of the 1st zoea. In the present material the lateral spine is minute from the 2nd zoeal stage onward. However, Cano (1892) depicted two prominent spines on the telson forks of the zoea that he attributed to the 3rd stage of T. scutellata. (4) In all previously published figures of the 1st crab stage the carapace anterio-lateral margins are shown as more prominent than observed in the present material and Lebour figured these margins as somewhat irregular in outline. BRACHYURA 57 The zoea of Thia scutellata can be distinguished from those of other brachyrhynchs (except perhaps Atelecydus, see below) described from NE. Atlantic waters on the following com- bined features. (1) Lateral spines on carapace. (2) Dorso- lateral processes confined to 2nd abdominal segment. (3) Two setae on basal segment of 1st maxilliped endopod. (4) Lateral spine on telson forks reduced to a very minute setule in 2nd-4th stages. The megalopa of T. scutellata is less easy to recognize because this stage is inadequately described for many species of NE. Atlantic brachyrhynchs. T. scutellata megalopa has the following combined features. (1) Absence of coxal or ischial spines on the pereiopods. (2) Sternites without spines. (3) Absence of tubercles or spines on carapace. (4) Uropod with 8 setae. (5) The two long terminal setae on the dactylus of the fifth pereiopod have simple apices. Lebour (1928: 528) proposed the family name Thiidae for Thia polita, stating that this species is 'different in many ways, and its larval stages does not fit into any family, although apparently near the Cancridae and the Corystidae'. The family name is now accredited to Dana 1852. Rice (1980: 333), basing his remarks on Lebour's account, suggested that Thia is possibly more closely allied to Corystes than to cancrids or portunids but was unable to comment further because of the absence of adequately described material. The zoeae of T. scutellata have two setae on the basal segment of the first maxilliped endopod, six setae on the distal segment of the maxillule, ten setae on the basis of the first maxilliped, lateral spines on the carapace, 5 + 3 setae on maxilla endopod and the dorso-lateral processes confined to the second segment of the abdomen. These features place them near to the Portunidae and in this respect they key satisfactorily to that part of the key to the brachyuran families as constructed by Rice (1980: 360). Differences separating zoeae of Thia from Atelecydus must await a critical re-examination of zoeae of A. rotundatus since the character listed by Lebour (1928: 487) for separating zoeae of these two genera, i.e. the presence of only one spine on the telson forks in Thia, is no longer valid. Addendum Mr Jose Paula, Faculdade de Ciencias, Lisbon, has recently informed me that a minute third spine is present on the outer telson fork of T. scutellata zoeae collected in Portuguese waters. A re-examination of the present reared material has revealed that this third lateral spinule is just discernible in some specimens. Acknowledgements I wish to express my sincere thanks to Mr T. Farrall for kindly providing the ovigerous specimens of T. scutellata and to Dr A. L. Rice for reading the manuscript. References Bourdillon- Casanova, L. 1960. Le meroplancton du Golfe de Marseille; les larves de crustaces decapodes. Recueil des Travaux de la Station Marine d'Endoume. 30: 1-286. Marseille. Cano, G. 1982. Sviluppo postembrionale dei Dorippidea, Leucosiadi, Corystoidei e Grapsidi. Memoire della Societa Italiana della Scienze delta dei XL. 8(3) 4: 1-14 Roma. Christiansen, M. E. 1969. Marine invertebrates of Scandinavia. No. 2, Crustacea, Decapoda, Brachyura. Universitetsforlaget 143 pp. Oslo. Claus, C. 1876. Untersuchungen zur Erforschung der Genealogischen Grundlage des Crustaceen- Systems. i-viii + 1 14 pp. Wien. Ingle, R. W. 1980. British Crabs. Oxford University Press & British Museum (Natural History) 222 pp. Oxford & London. & Clark, P. F. 1977. A laboratory module for rearing crab larvae. Crustaceana. International Journal of Crustacean Research. 32: 220-222 Leiden. Lebour, M. V. 1928. The larval stages of the Plymouth Brachyura. Proceedings of the Zoological Society of London. 2: 473-560. London. 58 R. W. INGLE Manning, R. B. & Holthuis, L. B. 1981. West African Brachyuran Crabs (Crustacea: Decapodd). Smithsonian Contributions to Zoology i-xii + 379 pp. Washington, D.C. Rice, A. L. 1980. Crab zoeal morphology and its bearing on the classification of the Brachyura. Transactions of the Zoological Society of London 35: 271-424 London. & Ingle, R. W. 1975. The larval development of Carcinus maenas (L.) and C. mediterraneus Czerniavsky (Crustacea, Brachyura, Portunidae) reared in the laboratory. Bulletin of the British Museum (Natural History) (Zoology) 28: 101-119 London. Steedman, H. F. (ed.) 1976. Zooplankton fixation and preservation. In: Monographs on oceanographic methodology. 350 pp. Paris. Williamson, H. C. 1915. VI. Crustacea Decapoda Larven. Nordisches Plankton 18: 315-558. Kiel. Manuscript accepted for publication 18 October 1983 Fig. 1 (p. 59) Thia scutellata (Fabricius): a-d, right lateral aspecies of lst-4th zoeae; e-h, abdomens from dorsal aspects of lst-4th zoeae. Scale (e-h) = 0-1 mm. Fig. 2 (p. 60) Thia scutellata (Fabricius): a, lateral margins of abdominal segments 1-5 of 1st zoea; b-e, antennules of lst-4th zoeae; f-h, antennae of 1st, 3rd & 4th zoeae; i-k, mandibles from dorsal aspects of 1st, 3rd and 4th zoeae; 1, m, maxillules of 1st & 2nd zoeae. Scales = 0-1 mm, a, 1, m to lower; i-k to middle; b-h to upper scale. Fig. 3 (p. 61) Thia scutellata (Fabricius): a, b, maxillules of 3rd & 4th zoeae; c-f, maxillae of lst-4th zoeae. Scale = 0-1 mm. Fig. 4 (p. 62) Thia scutellata (Fabricius): a, 1 st maxilliped of 1 st zoeae; b, 1 st maxilliped, terminal segments of 3rd zoeae; c, d, 2nd maxillipeds of 1st & 2nd zoea; e, dorsal aspect of megalopa; f, carapace of megalopa from right lateral aspect; g,h, abdomen from dorsal and right lateral aspects. Scales =0-1 mm, g,h to upper; a, c, d, to middle; b to lower. Fig. 5 (p. 63) Thia scutellata (Fabricius): megalopa: a, antennule; b, antenna c, ventral aspect of mandible; d, dorsal aspect of mandibular palp; e, maxillule; f, maxilla; g-i, lst-3rd maxil- lipeds. Scales = 0- 1 mm, g-i to uppermost; b to second; a to third; c, e, f to fourth; d to lowermost scale. Fig. 6 (p. 64) Thia scutellata (Fabricius): a-i, megalopa; a-f, lst-5th pereiopods; g, telson and uropods from dorsal aspects; h, i, 1st, 4th pleopods; j, carapace of 1st crab. Scales = 0-1 mm, a, c-f to upper; b, to middle; h, i to lower scale. Fie. 1 Fie. 2 Fi Fig. 4 Fie. 5 Description of a new species of Sylvisorex (Insectivora: Soricidae) from Tanzania Paulina D. Jenkins Department of Zoology, British Museum (Natural History), Cromwell Road, London SW7 5BD Introduction Two species of Sylvisorex are known from Tanzania, S. granti Thomas, 1907 which has been reported from Mount Kilimanjaro and S. megalura (Jentink, 1888) of which specimens from three separate localities have been recorded recently by Howell & Jenkins (in press). In the course of organised collecting in Tanzania, Dr K. M. Howell of the University of Dar-es- Salaam obtained a number of shrews which were submitted to the British Museum (Natural History) for identification. These include a single example of Sylvisorex which on examin- ation proves to differ substantially from all the known species of the genus in size and dental morphology, which is described here as new. The specimen differs externally from other members of the genus in the presence of bristle- hairs on the tail. The absence of such bristle- hairs has been used to distinguish Sylvisorex from Suncus but this distinction must now depend on the cranial differences elaborated by Heim de Balsac & Lamotte (1957) plus the dental characters used by Repenning (1967) and Butler & Greenwood (1979). The most readily applied of these latter characters is the presence of denticulations on the cutting surface of the first lower incisor in Sylvisorex, which are lacking in Suncus; also in Sylvisorex the talon of the upper premolar is more highly developed, the interorbital region broader and the braincase broader and higher relative to skull size. Additionally the hindfeet in Sylvisorex are larger relative to body size with slightly elongated, staggered, separated, metatarsal pads, while Suncus has smaller hindfeet with more oval, more or less adpressed pads. All measurements are in millimetres: the dental nomenclature follows that of Swindler (1976), Butler & Greenwood (1979) and is illustrated in Figure 1. Systematic Section Sylvisorex howelli sp. nov. HOLOTYPE. BM(NH) 82.874 adult of undetermined sex (viscera and external genitalia removed) in alcohol, skull removed; collected 27 April 1982 on Bondwa Peak, Uluguru North Forest Reserve, Uluguru Mountains, Morogoro District, Tanzania, c. 06°54'S 37°40'E, c. 1050m on road through forest by M. K. S. Maige and donated by Dr K. M. Howell. DIAGNOSIS. Small, size intermediate between S. johnstoni (Dobson, 1888) and S. granti; tail with bristle-hairs; braincase shallow and long, relative to skull length; lingual edge of second upper unicuspid projecting beyond that of first, level with lingual edge of third unicuspid; crown area of fourth upper unicuspid smaller than crown area of second upper unicuspid; parastyle of upper premolar low and slender; posterolingual ridge on first lower incisor very prominent, forming a small cusp; talonid of third lower molar reduced. (See Figs 2-8). Bull. Br. Mm. nat. Hist. (Zool.) 47(1): 65-76 Issued 28 June 1984 65 66 P. D. JENKINS (a) buccal cingulum lingual cingulum parastyle protocone paracone distostyle parastyle hypocone metastyle (b) posterior ridge protostylid on posterior ridge metaconid on posterolingual ridge posterolingual ridge posterolingual cingulum anterolingual ridge (c) protoconid paraconid metaconid hypoconid talonid basin entoconid ridge entoconid Fig. 1 Diagrams to show cusp nomenclature: (a) crown view of left upper third and fourth unicuspids, premolar and first molar; (b) lingual view of right lower first and second incisors and premolar; (c) lingual view of right lower third molar. DESCRIPTION. Size small (head and body length 48, tail length 44-5, hindfoot length without claws 11-5, ear length 6-8); dorsally dark brown, the hairs basally grey but brown medially and terminally; ventral pelage paler brown, the hairs with light grey bases and light brown tips; a gradual transition along flanks between colour of dorsum and venter; ears, limbs and dorsal surface of tail dark brown, their ventral surfaces paler, lacking any sharp lateral demarcation; the tail with a cover of short hairs along its entire length, interspersed with longer bristle hairs on the basal two-thirds. INSECTIVORA 67 Fig. 2 Dorsal view of skulls of Sylvisorex. Top row from left to right: S. johnstoni, S. howelli, S. granti and S. megalura; lower row from left to right: S. morio, S. lunaris and S. ollula. Scale in mms. Skull small (see Table 1), mostly lacking any exceptional features, but cranial profile sloping gradually upwards from tip of rostrum to posterior part of inter-orbital region then sloping more steeply to a rounded braincase which is long, not especially broadened and shallower relative to skull length than in other members of the genus (Figs 2-4); mandible with short, broad ascending ramus (Fig. 5). Posterior portion of upper incisor (I1) only slightly wider than remainder of tooth, the dis- tance between posterior part of incisors just greater than the width of one incisor. First upper unicuspid (Un1) sub-oval, with straight-edged lingual cingulum, tapering anteriorly and lacking any posterior cingular ridge; second upper unicuspid (Un2) with broad lingual cingulum, approximately twice as broad as buccal cingulum, its lingual edge projecting beyond that of Un1 and level with that of third upper unicuspid (Un3); Un3 with broad lingual cingulum, the tooth tapering anteriorly and rounded in crown view; fourth upper unicuspid (Un4) with broad lingual cingulum, the tooth almost round in crown aspect and smaller than Un2. Parastyle of upper premolar (P4) low and slender; talon posteriorly and lingually expanded, its posterior edge level with distostyle; least internal distance between premolars (P4-P4) approximately three-quarters of the width of one premolar. Talon of first and second 68 P. D. JENKINS Fig. 3 Ventral view of skulls of Sylvisorex. Top row from left to right: 5. johnstoni, S. howelli, S. granti and S. megalura; lower row from left to right: S. morio, S. lunaris and S. ollula. Scale in mms. upper molars (M1 and M2) only slightly expanded lingually, its lingual edge straight but expanded posteriorly, so that its posterior edge is level with metastyle. Third upper molar (M3) with a long ridge between parastyle and paracone, the ridge between paracone and protocone and that between metacone and mesostyle short, angle between protocone, meta- cone and midline of palate shallow; talon well-developed. Lower incisor (I,) long, slightly curved and approximately the same vertical diameter (in lateral aspect) for most of its length, tapering gradually to its tip; two rounded elevations on posterior ridge, anterior elevation long and low; posterolingual ridge prominent, forming a small cusp, higher than posterior ridge; lingual enamel extension reaching level of protoconid of second lower incisor (I2); lingual groove extending along length of tooth and terminating just anteriorly to notch at base of lateral enamel extension; anterolingual ridge present but poorly developed, not extending onto lateral enamel extension; no posterolingual cingulum. I2 anteroposteriorly slightly lengthened; posterolingual ridge well-developed; no protostylid. Posterior ridge of fourth lower premolar (P4) lacking protostylid; metaconid on posterolingual ridge barely marked. Anterior ridge of entoconid of first and second molars (M, and M2) poorly developed, not divergent from lingual side of tooth, the entoconid conical; postentoconid INSECTIVORA 69 e I a: E 11 o.S "*.« co 0 - •c g o § •^ 3 f-^-t *^** Ji 05 S.2 o s I ^- **- 0 DC - 70 P. D. JENKINS Fig. 5 Lateral view of mandibles of Sylvisorex. Top two rows, above — lingual view of right man- dibular ramus, below — labial view of left mandibular ramus, from left to right: S. johnstoni, S. howelli, S. granti and S. megalura; lower two rows, above — lingual view of right mandibular ramus, below — labial view of left mandibular ramus, from left to right: S. morio, S. lunaris and S. ollula (left mandibular ramus absent). Scale in mms. ledge present, adjacent to base of entoconid; lingual cingulum of M, weakly developed anter- iorly, absent from M2; the buccal cingulum continuing round hypoconid and merging with posterolingual rib on M, but on M2 narrow and merging with posterior part of hypoconid. Talonid of third lower molar (M3) reduced, the talonid basin reduced, the entoconid and the posterolingual rib absent but an entoconid ridge present. ETYMOLOGY. The name of the new species is derived from that of Dr K. M. Howell of the University of Dar-es-Salaam, who kindly donated this specimen. Comparison with other species Key to the species of Sylvisorex 1. Large, condylobasal length (CBL)>23, upper toothrow length (UTL)> 10 S. ollula Smaller, CBL <23, UTL< 10 2. Larger, CBL > 18, UTL>8 Smaller, CBL< 18, UTL< 8 3. Larger, CBL > 20, UTL>9, talonid of third lower molar (M3) more reduced than that of second lower molar (M2), third upper molar (M3) anteroposteriorly compressed, < 7-5% ofUTL S. lunaris Smaller, CBL < 20, UTL<9, talonid of M3 similar to that of M2, M3 not anteroposteriorly compressed, > 9% of UTL S. morio 4. Tail longer than head and body, braincase narrow, braincase breadth (BB) <48% of CBL .V. megalura Tail equal to or shorter than head and body, braincase broader, BB>48% of CBL INSECTIVORA 71 Fig. 6 (a-d) Crown view of left upper unicuspids: (a) S. johnstoni; (b) 5". howelli; (c) S. granti; (d) S. megalura. (e-f) Labial view of left upper second, third and fourth unicuspids and pre- molar: (e) S. granti; (f) S. howelli. Scales 1 mm. 5. Small, CBL< 15, UTL up to 6-5, M3 anteroposteriorly compressed, <8% ofUTL, talonid of M3 reduced to a single cusp S. johnstoni Larger, CBL> 15, UTL>6-5, M3 not anteroposteriorly compressed, >8% of UTL, talonid of M3 less reduced 6. Tail lacking bristle-hairs, braincase deep, braincase height (BH) 4-5-5-0, >27% of CBL, talonid of M3 similar to that of M2 S. granti Tail with bristle-hairs, braincase shallow, BH 4-1, <27% of CBL, talonid of M3 reduced but talonid basin and entoconid ridge present. 5. howelli S. howelli is intermediate in size between the smaller S. johnstoni and the slightly larger S. granti but smaller than all other members of the genus (Table 1 & Figs 2-5). S. morio (Gray, 1862) S. lunaris Thomas, 1906 and S. ollula Thomas, 1913 are readily distinguished from S. howelli by their much greater size and require no further detailed comments; their measurements are included in table 1 for comparative purposes. 72 P. D. JENKINS Fig. 7 (a-d) Lingual view of right lower first and second incisors, (e-h) Lingual view of right lower premolar and first and second molars, (a & e) S. johnstoni; (b & 0 S". howelli; (c & g) S. granti; (d & h) S. megalura. Scales 1 mm. The skull of S. howelli has a moderately broad, shallow, long braincase relative to skull length, in comparison with other members of the genus (Table 2 & Figs 2-4). S. johnstoni has a broad, moderately deep and long braincase; S. granti has a broad, deep, moderately long braincase and S. megalura a narrow, moderately deep and long braincase. INSECTIVORA 73 Fig. 8 Lingual view of right lower third molar: (a) S. johnstoni; (b) S. howelli; (c) S. granti; (d) S. megalura. Scales 1 mm. The ascending ramus of the mandible is rather short and broad in S. howelli (Fig. 5). The ramus in S. johnstoni is high and narrow. Height of the ascending ramus at the coronoid process is less in S. howelli than in S. granti or S. megalura (Table 1). The dentition of S. howelli is distinctive (see description). The main differences between the four species are illustrated in figures 6-8 and discussed below. The degree of development of the lingual cingulum on the upper unicuspids varies from broad in S. johnstoni and S. granti to very broad in S. megalura and S. howelli (Fig. 6). In S. howelli the lingual edge of the second upper unicuspid projects beyond that of the first and is level with that of the third, while in the other three species the lingual edge of the second upper unicuspid does not project as far as that of the first and third unicuspid. The parastyle of the upper premolar is low and slender in S. howelli but medium in height and well developed in the other three species. S. granti is illustrated as an example of the condition in all three species in comparison with S. howelli (Fig. 6). The third upper molar is a large tooth in S. granti and S. megalura, it is somewhat smaller in S. howelli but is anteroposteriorly compressed in S. johntoni (Table 1). The posterolingual ridge of the first lower incisor (Ij) is higher than the posterior ridge and forms a small cusp in S. howelli, unlike the condition in the other three species (Fig. 7). The anterolingual ridge of I, does not extend onto the lateral enamel extension and a posterolingual cingulum is absent in S. johnstoni and S. howelli. In S. granti and S. megalura there is a well developed anterolingual ridge, extending onto the lateral enamel extension to form a posterolingual cingulum. 74 P. D. JENKINS T '_' OX r*l iA s; .3 CO o ooS •* CO CO -s: co o rn o R -s: •^ CO 1 — *o m^ >nrn 2^ lo*^ ONr-~ C IX C IX C |X C |X C IX C IX C a xi _o TJ o U op OJ | D 1 1 1 o - o- tli X INSECTIVORA 75 Table 2 S. johnstoni S. howelli S. granti S. megalura Braincase breadth 7-0-7-5 7-7 8-0-8-6 7-6-8-1 X 7-30 8-28 7-84 n 7 1 11 9 Braincase height 3-8-4-0 4-1 4-5-5-0 4-4-4-7 X 3-87 4-75 4-55 n 7 1 11 9 Braincase length 5-6-5-8 6-8 6-6-7-1 6-8-7-4 X 5-73 6-85 7-11 n 7 1 11 9 Braincase breadth as 49-0-54-0 48-4 49-1-51-9 43-7-47-6 a % of condylobasal X 51-38 50-48 46-01 length n 7 1 11 9 Braincase height as 26-6-28-1 25-8 27-5-30-9 25-9-28-3 a % of condylobasal X 27-24 28-92 26-69 length n 7 1 11 9 Braincase length as 39-6-41-1 42-8 40-7-42-6 40-8-42-0 a % of condylobasal X 40-31 41-72 41-70 length n 7 1 11 9 x^mean; n = sample size. A posterolingual ridge is present on the second lower incisor of S. howelli, it is weakly developed in S. granti and S. megalura but absent from S. johnstoni (Fig. 7). There is no protostylid and a metaconid is barely indicated on the lower premolar of S. howelli; both protostylid and metaconid are lacking in S. johnstoni; both protostylid and metaconid are present in S. granti, while a metaconid only is present in S. megalura (Fig. 7). In all three species except S. howelli, a small cusplet is present in the posterior part of the valley between the posterior ridge and the posterolingual ridge. The talonid of the third lower molar is reduced to a single cusp, the hypoconid, in S. johnstoni (Fig. 8). It is reduced to a small talonid basin and an entoconid ridge in S. howelli. In S. granti and S. megalura less reduction has occurred and the talonid resembles that of the second lower molar. Acknowledgements I should like to thank Dr K. M. Howell of the University of Dar-es-Salaam who has donated a number of shrews to this museum, including the specimen of the new species. I am also grateful to Mr J. E. Hill and Mr I. R. Bishop O.B.E. of the Mammal Section, British Museum (Natural History) for their helpful criticism of the manuscript. References Butler, P. M. & Greenwood, M. 1979. Soricidae (Mammalia) from the early Pleistocene of Olduvai Gorge, Tanzania. Zoological Journal of the Linnaean Society 67: 329-379, 18 figs. Dobson, G. E. 1888. On the genus Myosorex, with descriptions of a new species from the Rio del Rey (Cameroons) District. Proceedings of the Zoological Society of London (1887) (4): 575-578. Gray, J. E. 1862. List of mammalia from the Camaroon (Sic.) Mountains, collected by Capt. Burton, H. M. Consul, Fernando Po. Proceedings of the Zoological Society of London (2): 180-181, pi. 24. Jentink, F. A. 1888. Note 1. Zoological researches in Liberia, a list of mammals, collected by J. Biittik- ofer, C. F. Sala and F. X. Stampfli. Notes from the Leyden Museum 10: 1-58, pis. 1-4. 76 P. D- JENKINS Heim de Balsac, H. & Lamotte, M. 1957. Evolution et phylogenie des soricides Africains 2. La lignee Sylvisorex-Suncus-Crocidura. Mammalia 21(1): 15-49. Howell, K. M. & Jenkins, P. D. (in press). Records of shrews (Insectivora, Soricidae) from Tanzania. African Journal of Ecology. Repenning, C. A. 1967. Subfamilies and genera of the Soricidae. Professional Papers of the United States Geological Survey (565): 1-74. Swindler, D. R. 1976. Dentition of living primates. London, New York, San Francisco. Thomas, M. R. O. 1906. Descriptions of new mammals from Mount Ruwenzori. Annals and Maga- zine of Natural History (7) 18: 136-147. 1907. On further new mammals obtained by the Ruwenzori Expedition. Annals and Magazine of Natural History (7) 19: 1 18-123. 1913. On African bats and shrews. Annals and Magazine of Natural History (8) 11: 314-321. Manuscript accepted for publication 21 October 1983 A new species of the Hipposideros bicolor group (Chiroptera: Hipposideridae) from Thailand J. E. Hill Department of Zoology, British Museum (Natural History), Cromwell Road, London SW7 5BD, United Kingdom Songsakdi Yenbutra Thailand Institute of Scientific and Technological Research, Bangkhen, Bangkok 9, Thailand While Curator of Terrestrial Vertebrates at the Thai National Reference Collection (now a part of the Thailand Institute of Scientific and Technological Research) the late Kitti Thonglongya undertook a survey of the bats of Thailand and their associated parasites, col- lecting extensively throughout the country. For the most part the bats that he collected were reported on the basis of a broad sample by Hill (1975). Those sent to London for examination included an example of Hipposideros from Rat Buri (TNRC 54-2080, now BM(NH) 78.2344) initially referred to H. ater Templeton, 1 848 but which proved to diifer quite clearly either from this species or from H. cineraceus Blyth, 1853 which in some superficial respects it resembled. No further study was undertaken at the time, Hill (1975) listing it as Hipposideros sp. but noting that it could not be referred to ater or to cineraceus from which it differed in certain features of the noseleaf and skull, or in size. Further similar specimens have now been found in the Thai National Reference Collection and more detailed study has established that all represent an undescribed species: additionally, another example in London (BM(NH) 78.2346, formerly TNRC 54-1961) proves also to represent the new species, rather than H. cineraceus as was thought originally. Hipposideros halophyllus sp. nov Hipposideros sp.: Hill, 1975: 27, Rat Buri, Thailand. Hipposideros cineraceus: Hill, 1975: 29 (in part), Phet Buri, Thailand; Lekagul & McNeely, 1977: 165 (in part), fig. 6 1 . HOLOTYPE. rf TNRC 54-3694. Khao Sa Moa Khon, Tha Woong, Lop Buri, Thailand. Collected by Kitti Thonglongya. In alcohol, skull extracted, rear of cranium slightly damaged. OTHER MATERIAL EXAMINED IN LONDON, rfd1, 9 TNRC 54-3696, 54-3697, 54-3705. All from the type locality. In alcohol, skulls extracted. rf BM(NH) 78.2344. Tham Khao Bin, Rat Buri, Thailand. Originally TNRC 54-2080. In alcohol, skull extracted. d1 BM(NH) 78.2346. Tham Khao Yoi, Phet Buri, Thailand. Originally TNRC 54-1961. Skin and skull. OTHER MATERIAL EXAMINED IN BANGKOK. 33 TNRC 54-3695, 54-3706, 99 TNRC 54-3704, 54-3710. All from the type locality. In alcohol, skulls of TNRC 54-3695, 54-3710 extracted. DIAGNOSIS. A member of the bicolor group (Hill, 1963) of Hipposideros, characterized exter- nally by large rounded ears lacking any sharply defined point or any definite fold or thicken- ing at the antitragal lobe; by the absence of lateral supplementary leaflets beneath the an tero- lateral margin of the anterior leaf, which has a shallow median emargination, and by the expansion of the internarial septum to form a small, disc-like structure just anterior Bull. Br. Mus. nat. Hist. (Zool.) 47(1): 77-82 Issued 28 June 1984 77 78 J. E. HILL & S. YENBUTRA to the nostrils. The skull is elongate and narrow, the zygomatic width less than the mastoid width, the braincase moderately inflated anteriorly, with a low, narrow rostrum, broad basis- phenoid depression and wide basioccipital, the tympanic bullae long and narrow and the inflated part of the cochleae elongate rather than subcircular. The new species is similar in size and ear structure to Hipposideros cineraceus or to H. ater but may be distinguished by its internarial structure and elongate tympanic bullae. Among other Asian species it has some resemblance to H. ridleyi in the disc-like expansion of the internarial septum but ridleyi is much larger (length of forearm c. 48 mm) and its internarial disc is relatively much larger, thinner and more saucer-like. DESCRIPTION. Small (length of forearm 35-1-38-2 mm). Ears large, broad and rounded with broad, poorly defined tip, the anterior margin strongly convex, lacking any basal lobe, the posterior margin of the ear slightly convex for much of its length, deflected sharply to a wide, anteriorly rectangular antitragal lobe; a very slight thickening of the integument of the ear at the rear of the antitragal lobe but no definite fold or obvious antitragal modification. Outer (medial) surface of conch haired for about one half or a little more the length of the ear, the hairs proximally rather dense and long, more distally sparser and shorter; a sparse scattering of short hairs along the anterior border of the inner surface of the conch. Muzzle low, not especially broadened, with small, narrow noseleaf lacking lateral sup- plementary leaflets; anterior leaf narrow, rather elongate, its total width about two thirds of the width of the muzzle, widest at a point level with the nostrils, narrowed anteriorly with a small, rounded median emargination. Central part of internarial septum expanded into a small, rounded, lobular and thickened disc-like structure lying in the narial depression slightly anteriorly to the nostrils: laterally this disc is slightly swollen, with a shallow median trough separating its lateral lobes; anteriorly the margins of the expansion curve sharply inwards to join the anterior leaf by a short, constricted and unthickened internarial segment, posteriorly merging similarly but less abruptly with the base of the intermediate leaf. Narial lappets small but evident, the nostrils very slightly pocketed. Intermediate leaf sometimes with four small glands, not sharply demarcated from anterior leaf and not especially elevated or inflated. Posterior leaf high, slightly semicircular, its lower half or a little more divided by three broad, ill-defined septa into four shallow pockets, its upper part smooth; no serrated structure on its posterior face. A prominent frontal sac with horizontal aperture lies immediately behind the posterior leaf in male examples, represented in a female specimen by a small tuft of hair. All but one of the specimens available in London are in alcohol and have been so for some years: the sole dry example also appears to have been preserved in this way, possibly for a considerable period. The dorsal surface is now mid-brown, the hairs pale cream at the base and generously tipped with the brown terminal colour, which doubtless has been bleached to some extent by fluid preservation; the ventral surface is paler, largely lacking any brown, and has a greyish or greyish buff tinge. Skull very small (condylocanine length 12-7-13-1 mm) with inflated, elongate and rather narrow braincase, its length from occiput to narrowest part of postorbital constriction one fifth greater than its greatest width across the mastoids. Postorbital region strongly con- stricted, the rostrum narrow and uninflated, its greatest supraorbital width about two fifths of the mastoid width; anterior and lateral rostral compartments not much inflated, the anterior part of the rostrum in profile a little below rather than level with the junction of the sagittal and supraorbital crests, the upper profile of the rostrum sloping slightly down- ward anteriorly rather than horizontal, and curving smoothly rather than abruptly to the maxillae. Zygomata robust, with a low jugal eminence; anteorbital foramen elongate, closed by a narrow bar of bone. Premaxillae making a broadly V-shaped junction with the maxillae; palate long, narrow, the toothrows strongly convergent anteriorly; palation rounded, level anteriorly with a line joining the anterior faces of m3~3. Mesopterygoid fossa wide, the pterygoids slightly flared; sphenoidal bridge moderate, flanked by elongate apertures; basi- sphenoidal depression wide and shallow, subcircular in outline; basioccipital wide, the CHIROPTERA 79 cochleae widely separated. Tympanic bullae long, narrow, almost platelet-like, anteriorly approaching the rear of the glenoid fossa, in length considerably exceeding the antero- posterior diameter of the exposed part of the associated cochleae, while in width the tympanic bullae are equal to rather less than one half the diameter of the exposed part of the associated cochleae along the other axis; cochleae small, not much inflated, their greatest exposed width about one and three quarter times their distance apart, the exposed part more elliptical rather than subcircular in outline. Dentition with no unusual features; upper incisors weak, their tips convergent, the outer lobe obsolescent; anterior upper premolar (pm2) small, compressed between the canine and the second upper premolar (pm4) or very slightly extruded but nonetheless separating these teeth; posterior ridge of last upper molar one half of the length of its anterior ridge; outer lower incisors rather larger in crown area than inner lower incisors; anterior lower premolar (pm2) about equal in length to second lower premolar (pm4) and one half to three quarters its height. Measurements of the holotype, followed by minima and maxima of the series of six (except where indicated in parentheses) measured in London: length of forearm 37- 1 , 35- 1-38-2; con- dylocanine length 12-7, 12-7-13-1 (5); least interorbital width 2-0, 2-0-2-1; rostral width 3-6, 3-6-3-7; width across anteorbital foramina, 3-6, 3-5-3-7; zygomatic width 7-4, 7-2-7-4; width of braincase 6-9, 6-5-6-9 (5); mastoid width 7-8, 7-6-7-9, d-c1 (alveoli) 3-1,3-0-3-3; m3-m3 4-8, 4-8-4-9; width sphenoidal depression 2-7, 2-6-2-8 (5); width basioccipital 2-68, 2-62-2-77; c-m3 4-8, 4-7^-8; length complete mandible from condyles 8-5, 8-2-8-6 (5); length right ramus from condyle 8-9, 8-9-9-2 (4); c-m3 5-1, 5-0-5-2; length tympanic bulla 2-57, 2-49-2-72, width tympanic bulla 1-08, 0-98-1-13; antero-posterior diameter of cochlea 2-01, 1-80-2-03; transverse diameter of exposed part of cochlea 2-23, 2-21-2-33. ETYMOLOGY. The name of the new species is drawn from ataos, a disc, and 0uXA,ov, a leaf. REMARKS. The noseleaf of this new species is described (p. 165) and illustrated (fig. 61) by Lekagul & McNeely (1977) as Hipposideros cineraceus. The illustration of the skull (p. 166) of H. cineraceus provided by these authors is in fact of that species, having in contrast to the skull of H. halophyllus an inflated, higher rostrum, broad, short tympanic bullae and rounded rather than elongate cochleae. The material of//, halophyllus examined in London has come from the Thai National Reference Collection: all but one of these specimens was identified initially as H. cineraceus in Thailand, the exception being referred formerly to H. ater. Evidently Lekagul & McNeely employed a specimen labelled H. cineraceus from this collection as the basis of their description and illustration of the noseleaf. The new species is similar in size to Hipposideros cineraceus and a little smaller than //. ater, with either of which it can at first inspection be confused. Its ears are similar in size and shape to those of cineraceus but in this species there is a distinct fold or thickening at the antitragal lobe. There is also a slight thickening with a small antitragal fold in the ear of H. ater. The anterior leaf in H. halophyllus is shallowly but distinctly emarginated just above the centre of the upper lip: neither cineraceus nor ater display any such emargination and although in both of these the internarial septum is inflated and sometimes bulbous there is no evidence of the development of any disc-like structure between and slightly in advance of the nostrils, the septum remaining more or less parallel-sided although swollen. Cranially, halophyllus may be distinguished from cineraceus and ater by its less inflated anterior narial compartments and lower anterior rostrum which slopes more gently to the canines, by its narrower, longer tympanic bullae, its more elongate rather than subcircular cochleae, and by its broader basioccipital. DISCUSSION. Modification of the internarial septum into a circular or subcircular disc in Hipposideros occurs in the three African species H. curtus Allen, 1921, H.jonesi Hayman, 1947 and H. marisae Aellen, 1954, and also in one other Asian species, H. ridleyi Robinson & Kloss, 1911 from Malaya and Borneo. All belong to the bicolor group of Hipposideros and except for curtus are allocated by Hill (1963) to the bicolor subgroup, characterized by 80 J. E. HILL & S. YENBUTRA Fig. 1 . Hipposideros halophyllus. 9 TNRC 54-3704. Head x 5 Fig. 2 Hipposideros halophyllus. "' T 'This research was carried out in the Department of Zoology, British Museum (Natural History) and was submitted in partial fulfilment of the requirements for the degree of Doctor of Philosophy in the Faculty of Science, University of London. The author's present address: Department of Anatomy & Cell Biology, St Mary's Hospital Medical School, Norfolk Place, London W2 1PG. Bull. Br. Mus. not. Hist. (Zool.) 47(2): 83-150 Issued 26 July 1984 83 84 R. A. TRAVERS Vertebral column . 125 Dorsal and anal fin spines . . . . . . . . . 127 Caudal fin . .127 Squamation . 129 Myology .... 129 Cephalic muscles . ........ 129 Adductor mandibulae 129 Adductor arcus palatini 1 30 Other features . . . ... 130 Anterior nostril 1 30 II. Intrarelationships in the Mastacembeloidei 131 Proposed changes in classification 1 39 Diagnoses for the Synbranchiformes . 141 Synbranchiformes 141 Synbranchoidei 141 Mastacembeloidei 141 Chaudhuriidae 142 Mastacembelidae 143 Acknowledgements . 146 References 146 Synopsis The Mastacembeloidei or spiny eels (comprising the families Mastacembelidae, Chaudhuriidae and Pillaiidae) is a distinctive group of about 70 freshwater species with a tropical and subtropical Oriental and Ethiopian distribution, currently recognised as a suborder of the perciform fishes. The majority of its 70 species have been placed in a single genus, Mastacembelus, without regard to their genealogical relationships, and the suborder as a whole has never been the subject of a detailed taxonomic or anatomical review. A revision of the genera and families within the suborder, and a reconsideration of its interrelationships within the Percomorpha, are the overall objectives of this study. It is based on numerous anatomical characters drawn from the descriptions in Travers (1984) and involves a comparison of their condition with that of their homologues in other teleostean lineages. This analysis indicates that the mastacembeloids, long associated with the Perciformes, should be reallocated to the Synbranchiformes. A phylogenetic hypothesis of their intrarelationships is also proposed, viz., the Mastacembeloidei can be resolved into two lineages: the Chaudhuriidae (expanded to incorporate two genera) and the Mastacembelidae (divided into two subfamilies representing the Asian and African species). This hypothesis results in the elevation of the sole endemic Chinese species, Mastacembelus sinensis, to a monotypic genus within the Chaudhuriidae, and the generic synonymy ofPillaia (and Garo) with Chaudhuria (but the retention of both the latter taxa as subgenera ofChaud- hurid). The Asian mastacembelid subfamily, Mastacembelinae, retains the genera Mastacembelus and Macrognathus although the former is restricted to six species only, and the latter expanded to include eight species previously included in Mastacembelus. The African species are shown to be phyletically distinct and to warrant the creation of a new subfamily and new genera for two major African sub lineages. Introduction The Mastacembeloidei, or spiny-eels, are a group of freshwater teleost fishes occurring in the Oriental and Ethiopian zoogeographical regions. The Oriental species are widely distrib- uted in SE. Asia and extend from the eastern China seaboard, through the continental islands of Indonesia to the Middle East (Fig. 1). This pattern, together with their recent dis- covery in southern Iran (Coad, 1979), gives these fishes a continuous distribution throughout their Oriental range. The absence of mastacembeloids from the Arabian Peninsula, North Africa and the Horn of Africa, however, has resulted in the geographical isolation of the Oriental fauna from the Ethiopian species. These are restricted to tropical and subtropical MASTACEMBELOIDEI III PHYLOGENETIC 85 waters of Central Africa; embracing the Nilo-Sudan, Upper and Lower Guinea, Zaire, East Coast and Zambezi ichthyofaunal provinces of Roberts (1975). Mastacembeloids occur in a variety of freshwater habitats at high and low altitudes, and are common in riverine and lacustrine environments, streams and ponds (Job, 1941; Sufi, 1956; Matthes, 1962; Poll & Matthes, 1962; Skeleton, 1976). At least four African species are cavernicoles (Poll, 1953, 1958, 1959 & 1973 and Roberts & Stewart, 1976) and show considerable atrophy of the eye tissues, an associated hypertrophy of the superficial parts of the adductor mandibulae musculature and even lack of pigment (Poll, 1973). A group of Asian mastacembeloid taxa exhibit an elaborate burrow- ing mechanism (Job, 1941; Sufi, 1956); however, this habit is not found in all species (Schofield, 1962). Most mastacembeloids appear to be carnivores with a wide range of feeding strategies. Food organisms range from small zooplankton (Job, 1941; Sufi, 1956) through aquatic insect larvae and oligochaetes (Roberts, 1980) to other fishes (pers. obs.) including both eggs and fry (Hamid Khan, 1934) and even mastacembeloid species with a relatively small adult size (Staeck, 1976). To a large extent the prey species are related to the size and developmental stage of the predators. Reproduction in the mastacembeloids is poorly known. There are brief descriptions of the spawning activity in a single species in captivity (Schoenebeck, 1955; Polder, 1963; Franke, 1965), artificially induced spawning (Kochetov, 1982) and the periodic occurrence of vast numbers of juveniles in Lake Tanganyika (Brichard, 1978), but reproductive behaviour in nature has never been observed. Many mastacembeloids are thought to have some form of aerial respiration (Dobson, 1874; Day, 1877; Ghosh, 1933; Hora, 1935), although reports are contradictory and the presence of suprapharyngeal or swimbladder adaptations has never been shown. Whether survival in oxygen deficient waters is by some form of cutaneous respiration (Mittal & Datta Munshi, 1971), or by inanition with respiration at a standstill (Job, 1941), requires further research before an accurate assessment can be made. Fig. 1 Present-day distribution of mastacembeloid Oriental and Ethiopian species. 86 R. A. TRAVERS A pseudobranch, once thought to be absent in mastacembeloids (Day, 1889; Boulenger, 1915: 12) has recently been described in some Asian species (Bhargava, 1953). A spherical sac- like pseudobranch situated above the epithelium roofing the oral cavity was found in all taxa examined. This study is based on the many osteological and myological features shown (Travers, 1984) to be of potential value in reconstructing phylogenetic relationships. These morpho- logical characters will be analysed by comparing their condition with that of their homo- logues (Patterson, 19826) in other groups of percomorph fishes (acanthomorphs) or with the teleosts as a whole (see below). The objectives of these comparative studies, besides providing an analysis of mastacem- beloid interrelationships with other percomorph fishes, was an attempt to revise the intra- relationships of the group. Only by revealing the phylogeny of these fishes can an improved understanding of their biogeographical history be gained (Greenwood, 1983). Although numerous specific features were identified (see Travers, 1984), and are included where possible in the analysis, a formal species level revision was not attempted. Materials and methods Materials The bulk of the comparative outgroup material examined is tabulated in Travers (1981). In addition to these specimens further taxa are listed in systematic order in Table 1; all are given with their register number and codes indicating the type of examination or preparation involved. Unless otherwise indicated, all are BM(NH) registered specimens. For a complete list of the mastacembeloid specimens upon which this study is based, for nomenclatural references and the techniques of morphological analysis, see Travers, 1984. Methods The inter- and intrarelationships of the mastacembeloids were evaluated by the application of a cladistic methodology as originally defined by Hennig, 1950 & 1979 and advocated latterly by Eldredge & Cracraft, 1980, Nelson & Platnick, 1981, and Wiley, 1981. This methodology has, over the last decade, brought increased order to the complexities of teleostean classification regardless of the vociferous debate (e.g. see Nature Correspondence from volume 275 to 292), and the proliferation of conceptual literature (e.g. see Systematic Zoology from 1967 (4): 289-292, to date), that it has engendered. To determine the primitive (plesiomorphic) or derived (apomorphic) nature of mastacem- beloid characters, in the absence of a complete ontogenetic history of all species, a series of outgroup comparisons was made in a manner similar to that discussed by Watrous & Wheeler (1981); see Farris (1982) for a critical review of their study. Information from onto- genetic transformations (de Beer, 1958; Gould, 1977; Nelson, 1978; Rosen, 1982), where available was utilized especially if comparisons could be made with similar developmental stages in outgroup taxa (Fink, 1982; Patterson, 1983). Only those uniquely derived features confined to individuals of a single species were identified as autapomorphic characters. When detected in this restricted sense these char- acters can be used to identify species, regardless of whether or not they are strictly definable (Patterson, 1982a). Where an outgroup has been the subject of a recent study, highly derived species were excluded so as not to be mislead by taxa that are unrepresentative for their group as a whole. In poorly studied assemblages a wide range of representatives was examined in order to avoid overlooking a group which may be masked by a previous incorrect assessment of its characters and hence given an erroneous taxonomic allocation. Outgroup taxa were taken from all major percomorph assemblages. Emphasis was given to perciform lineages, since these represent the largest acanthopterygian groups, and several beryciform -families since the perciforms are generally thought to have been derived from MASTACEMBELOIDEI IK PHYLOGENETIC 87 a beryciform species (Patterson 1964; Greenwood et al, 1966; McAllister, 1968; Zehren, 1979). However, these beryciform taxa are ill-defined and thought by most authors to rep- resent polyphyletic assemblages (Patterson, 1964; Greenwood et al, 1966; Rosen, 1973; Zehren, 1979). Furthermore, the task of analysing such groups is complicated by the large number of taxa involved and also, as Rosen (1973:398) observed, by: '...the lack of co-ordinated comprehensive studies of character complexes throughout the group'. For example, the only synapomorphy so far proposed for the Perciformes is the presence of an interarcual cartilage (Rosen & Greenwood, 1976) and even this has been shown recently to have a far wider occurrence than had been previously thought (Travers, 1981; it may even occur in some ophichthid eels (McCosker, 1977 and M. Leiby pers. comm.). Table 1 Teleostean outgroup taxa (supplementary to those listed in Travers 1981). Species Reg. No. Preparation Arapaima gigas Elops hawaiensis Flops machnata Halosaurus owenii Notacanthus bonapartei Notacanthus sexspinis Anguilla anguilla Etrumeus teres Ostichthys parvidens Holocentrus rufus Zeusfaber Hypoptychus dybowskii Macrotrema caligans Macrotrema caligans Monopterus cuchia Synbranchus bengalensis Synbranchus javanicus Synbranchus marmoratus Mugil cephalus Pholis gunnellus Pholidichthys leucotaenia Pholidichthys leucotaenia Pholidichthys leucotaenia Scytalina cerdale Notograptus guttatus Ptilichthys goodei Stichaeus (Leptoblennius) mackayi Unreg. 1962.4.3: 1-25 1962.8.28: 1-7 1890.6.16: 55 1972.1.26:33-39 1873.12.13:27 1962.6.5: 2-4 1923.2.26: 73-78 1974.5.25: 747-753 1976.7.14: 79-81 1971.7.21: 86-90 1979.11.26: 1-3 1860.3. 19: 943 Type 1908.7.13: 1 Unreg. 1860.3.19: 1477 1862.11.1: 138 1981.1.15: 1117-1119 1913.7.10: 31-34 1981.2.20:446-479 1974.3.14: 1 USNM 206237 USNM Gift USNM 70801 USNM Gift USNM 130266 1896.7.23: 183 DS A A AP MD DS MD A MD MD MD MD AP AP DS MD MD MD MD MD MD A AP AP A/A AP MD Abbreviations Anatomical Aa Anguloarticular Add Mand Adductor mandibulae Asph Autosphenotic Bbl 1-3 Basibranchial 1-3 Bbl KC Basibranchial I keel cartilaginous Bh Basihyal Bh+ Bbl Basihyal fused with basibranchial I BR Branchiostegal rays Bs Basisphenoid C Cleithrum 88 R. A. TRAVERS Com Coronomeckelian DO Dilatator operculi Dsph Dermosphenotic Ect Ectopterygoid End Endopterygoid F Frontal Fa Fimbria FDL Frontal descending lamina FFac Frontal facet Fu Fimbrule Hb 2-3 Hyobranchial 2-3 Hb3 AP Hyobranchial 3 anterior process Hyo Hyomandibula Hyo Add Hyohyoidei adductores HyoMF Hyomandibula metapterygoid flange Hyo SP Hyomandibula symplectic process lop Interoperculum LE Lateral ethmoid LE xsc Lateral extrascapula Lig Ligament LO Levator operculi MC Meckel's cartilage Met Metapterygoid MExsc Medial extrascapula MP Maxillary process Op Operculum P Parasphenoid Pal Palatine Palopt Palatopterygoid PalT Palatine teeth PAW Parasphenoid ascending 'wing' Pop Preoperculum Pr Prootic PrAP Prootic anterior process Pt Pterosphenoid Ptm Posttemporal PtmT Posttemporal tubule Q Quadrate Ra Retroarticular Sc Supracleithrum Sen Pap Sensory papillae Sep Septum (ossified) Sop Suboperculum Sph Sphenotic Sym Symplectic Uh Urohyal UhAP Urohyal ascending process Note on the figures: even stipple-dots indicate the presence of cartilage. The scale on all figures indicates 1 mm. Table of study material A/A A DS MD AP Unreg. Institutional BM(NH) USNM Double stained transparency (alizarin red and alcian blue) Alizarin stained transparency Dry skeleton preparation Muscle dissection (cheek & opercular region) Alcohol preserved specimen not available for dissection or preparation Unregistered specimen held at BM(NH) British Museum (Natural History). United States National Museum, Washington. MASTACEMBELOIDEI IK PHYLOGENETIC 89 Mastacembeloid interrelationships I. Character analyses Osteology NEUROCRANIUM The neurocranium exhibits the general trends (mainly reductional) and particular features that are currently recognised as diagnostic for acanthopterygian fishes (see Zehren, 1979: 163-166 for a recent summary), although they are somewhat masked by the extreme precommissural attentuation of the skull. Of the eleven acanthopterygian neurocranial and circumorbital character states listed by Patterson (1964: 449) and Zehren (op. cit.) eight are readily recognisable in almost all mastacembeloids. A further derived feature found by Patterson (1975: 568) only in the Acanthopterygii is the pons moultoni, a loop of bone on the inner face of the sphenotic which surrounds the anterior semi-circular canal. The medial face of the sphenotic in mastacembeloids houses the anterior semi-circular canal in such a structure and thus conforms to the condition in the Acanthopterygii. Surprisingly, Zehren (1979) did not mention the pons moultoni in his study of the Beryciformes. Elongation of the supraethmoid and vomer in mastacembeloids contributes to the gener- ally pointed snout in these fishes, although the lateral ethmoids are of more usual proportions (Travers, 1984). In other percomorphs with elongate syncrania, including such forms as the congrogadids, pholidichthyids, sphyraenids, acanthurids, luciocephalids and synbranchids, the ethmovomerine region is not always attenuated. In Congrogadus subduceus and Pholidichthys leucotaenia this region is of similar proportions to that found in most other percomorphs, any elongation being restricted to the postorbital neurocranial bones. In Sphyraena obtusata, Acanthurus bahianus and Luciocephalus pulcher the ethmovomerine region is particularly elongated, whilst the braincase proportions are similar to those in most percomorph fishes. This condition is produced in Sphyraena by elongation of the lateral ethmoids, whilst in Luciocephalus the prevomer and nasals are particularly lengthened (Liem, 1967). The anterior region of the parasphenoid is particularly long in Acanthurus and, combined with the long supraethmoid, it contributes to the elongate ethmovomerine region. The vomer in synbranchids was described by Rosen & Greenwood (1976: 49) as a 'long, thin strut' elongated to a point below the pterosphenoids and basisphenoid. This is the only outgroup examined which has a vomerine shaft comparable in length to that in the mastacembeloids (e.g. compare the vomerine length in Rosen & Greenwood, 1976, figs. 57 & 59 with the vomer in other percomorphs). This may well be a synapomorphy of these taxa. Although an elongate ethmovomerine region occurs in a variety of what appear to be phyletically distant percomorph fishes, the method by which elongation is achieved in the mastacembeloids (by extreme lengthening of the vomer and supraethmoid) is not found in any other group (apart from a similar elongation of the vomer in the synbranchids). The sensory canal bearing bones of the ethmovomerine region (i.e. the nasal and 1st infraorbital) are also elongated, and the nasal has a particularly broad dorsal surface in mastacembeloids. The 1st infraorbital is long and tapered in a variety of percomorphs, including the synbranchids, ammodytids and luciocephalids. However, the nasal in these taxa is barely broader than the sensory canal it carries. The concomitant elongation in the mastacembeloids of the supraethmoid, vomer and 1st infraorbital bone, accompanied by a long nasal with a broad dorsal surface, results in a snout unlike that found in any of the other taxa examined. It is thus considered to be a synapomorphy for the group. The mastacembeloid lateral ethmoid is not elongate although the 'ethmoids' (presumably including the lateral ethmoid) were included by Rosen & Greenwood (1976) in their list of 90 R. A. TRAVERS 'greatly attentuated' mastacembeloid neurocranial elements. In fact the lateral ethmoids retain the general percomorph proportions, but are fused in the midline and have an overall tubular shape. In percomorph fishes the lateral ethmoids are generally plate-like bones lying on either side of the supraethmoid and are separated from each other medially by the cartilaginous, posteroventral end (septal cartilage) of the supraethmoid. Medial fusion of the lateral ethmoids in mastacembeloids, dorsal to the cartilaginous pos- terior end of the supraethmoid is an arrangement found also in blennioids, (e.g. Notograptus guttatus) and in gobioids (e.g. Trypauchen wakae). However, the tubular shape of the lateral ethmoids was not found in any other outgroup taxa. The mastacembeloid pterosphenoids have wide lateral faces and are connected ventro- medially (Travers, 1984: 51). Among the outgroups examined (see Table 1) medial pterosphenoid connections were found only in the clupeid Etrumeus, some beryciforms (e.g. Diretmus argenteus and Ostichthys trachypoma; Zehren, 1979) and the anabantoids (e.g. Belontia and Trichogaster leeri; Liem, 1963). In the anabantoids the pterosphenoids are not connected ventrally, but dorsally are linked by long wing- like processes, running transversely across the orbital cavity. In two beryciform families (Diretmidae and Holocentridae) Zehren (1979: 232) found the pterosphenoids to have their ' . . . anterior lower edges meet in the midline closing off the optic fenestra' and, on the basis of outgroup comparisons concluded that this arrangement represented the apomorphic condition (see below). In synbranchids the massive development of the pterosphenoids has been noted by Rosen & Greenwood (1976: 44) who found that, with the basisphenoid, these bones occupy ' . . . somewhat less than half the total length of the neurocranium'. Although the ptero- sphenoid is large in synbranchids it does not contact its partner in the midline. Rosen & Greenwood (1976) also listed the bones involved in the neurocranial lengthening of mastacembeloids, when comparing them with synbranchids, but failed to mention the central role of the pterosphenoids. The wide lateral face of the pterosphenoids and their medial union ventrally (resulting in the ventral half of the optic foramen being bounded by them) is a synapomorphic character of all the mastacembeloids (its absence in Pillaia and Mastacembelus aviceps is discussed on p. 1 10). In the beryciforms in which the pterosphenoids are connected in the midline (discussed above), these bones lack wide lateral faces and their medial connection is thought to be convergent with that in the mastacembeloids. The basisphenoid is a characteristically small, compressed bone in mastacembeloids (Travers 1984: 53). This configuration appears to be directly associated with the median pterosphenoid connection, and represents a derived state of the large Y-shaped bone present in most teleostean lineages. It can be noted that in beryciforms and Etrumeus, taxa also with medially fused pterosphenoids, the basisphenoid retains its plesiomorphic condition. A preorbital (or, more correctly, suborbital) spine that pierces the skin is a characteristic feature of almost all mastacembeloids (Travers, 1984). A preorbital spine is also present in halosauroids (sensu Greenwood, 1977) as described by McDowell (1973) and illustrated by Greenwood (op. cit. figs. 5, 7, 9, 10 & 14), and in some Ostariophysii (e.g. Cobitidae). However, in these fishes it is produced, respectively, from the maxilla or the lateral ethmoid, and is not homologous with the mastacembeloid preorbital spine. Here the spine is the posterior end of the large first infraorbital bone. In no other teleostean taxon is the first infraorbital developed in this way, and the presence of the spine is a further valuable synapomorphy of the mastacembeloids. A long prootic, characterised by a distinct anterior process passing across the anterolateral face of the neurocranium and into the orbital cavity, is typical of most mastacembeloid taxa (Travers, 1984: 58). This is contrary to the opinion expressed by Springer & Freihofer (1976: 37) that, 'In the highly specialised Mastacembelidae and Chaudhuriidae the parasphenoid also lacks an MASTACEMBELOIDEI II: PHYLOGENETIC 91 ascending process but either the prootic is entirely blocked by the pterosphenoid- pleurosphenoid bones from entering the postorbital margin or the pterosphenoid- pleurosphenoid bones are absent (D. E. Rosen, pers. comm.)'. The plesiomorph condition of the teleostean prootic is one in which the anterolateral edge contributes to the posterior rim of the orbit. In a few, and diverse, taxa the anterior margin of the prootic is prevented from bordering the orbit by the ascending parasphenoid process, the pterosphenoid and/or the descending lamina of the frontal. This occurs particularly in taxa with elongate neurocrania, and is seen in some Lates species illustrated by Greenwood (1976: fig. 3). FDL Dsph PrAP Fig. 2 Congrogadus subduceus, otic region of neurocranium; lateral view, left side. A massive pterosphenoid and basisphenoid prevent the anterolateral extension of the prootic from entering the border of the small orbital cavity in the synbranchids (Rosen & Greenwood, 1976). In the Pholidichthyidae (Springer & Freihofer, 1976) the neurocranium is elongate and the prootic, lengthened anteriorly, has a wide lateral face, (Fig. 4a). The anterior region of the prootic in this taxon passess above the lateral face of the basisphenoid, and its anterior edge borders the posterior rim of the orbit. A very large prootic is also present in the congrogadids (e.g. Congrogadus subduceus, Fig. 2), and is distinguished by its long anterior region (between the ventral edge of the pterosphenoid and upper margin of the parasphenoid) which passes from the trigeminofacialis chamber to a point posterior to the rim of the orbit. Thus, the large size of the prootic in mastacembeloids is not in itself exceptional. Rather, the exceptional feature is the anterior region of the prootic overlying the ventrolateral face of the pterosphenoid and being developed into a long anterior process extending into the orbital cavity. This constitutes a major synapomorphic feature of the mastacembeloid neurocranium. The trigeminofacialis chamber lies in the prootic; its development has been discussed by Patterson (1964: 434-438). He concludes that the plesiomorphic condition for teleosts (as seen for example in Flops) was one in which the truncus hyomandibularis, the jugular vein and the orbital artery are separated by individual foramina in the pars jugularis. This condition differs from that in perciforms where the three foramina are confluent. The progressive reduction from four to two external openings in the pars jugularis among beryciform fishes has recently been demonstrated by Zehren (1979: 235) and lends support to Patterson's hypothesis (op. cit.) that ' . . . during the evolution of teleosts there has been a simplification of the pars jugularis'. 92 R. A. TRAVERS A large trigeminal foramen (generally situated anterior to the lateral commissure) and a small facial foramen (generally medial to the lateral commissure) are the only foramina in the pars jugularis of mastacembeloids, which are apomorphic in this respect. The mastacembeloid otic bulla is a small recess in the posteromedial face of the prootic, and completely houses the sacculus in most taxa. Only these fishes among the teleosts, have their saccular recess contained entirely within the prootic. In all others the recess is formed from the prootic, exoccipital and basioccipital. The relative contribution of these three bones does, however, vary from an almost equal contribution to one in which the greater part is derived from the prootic (e.g. congrogadids, tripterygiids of the genus Paraclinus, and Channa obscurd). This housing of the otolith bulla in the prootic only is a synapomorphic character of the mastacembeloids. Precommissural elongation of the neurocranial bones in mastacembeloids involves that region of the autosphenotic lying anterodorsal to the lateral commissure (T ravers, 1984: 69). This part of the sphenotic is produced into a wide anterolateral flange that generally extends across part of the lateral face of the pterosphenoid and descending frontal lamina, but does not enter the posterior border of the orbit. The postorbital process on the sphenotic, as a a FDL Asph FDL PAW Fig. 3 Neurocranium of (a) Ophidian rochei, and (b) Lycodes brevipes. Left side of otic region, lateral aspect. MASTACEMBELOIDEI III PHYLOGENETIC 93 result of its anterolateral flange, has a posterior position relative to that of a similar process in other perciform fishes, for example the labroids (Rognes, 1973) and cichlids (Stiassny, 1981). This posterior position of the postorbital process indicates the extent to which the mastacembeloid sphenotic has expanded anteriorly. In its plesiomorphic state the teleostean autosphenotic forms the posterodorsal corner of the orbit, a condition found in almost all major teleostean lineages, including many beryciform and perciform fishes. Some development of the autosphenotic occurs in several phylogenetically distant acanthomorph lineages (e.g. Ophidian rochei, Fig. 3a; Pholidichthys leucotaenia and Xiphistes mucosus, Fig. 4a & b; Ammodytes tobianus, Fig. 4c), although in none is it developed to an extent comparable with that in the mastacembeloids, and in all the bone, unlike that in mastacembeloids, still forms the posterodorsal margin of the orbit. The ammodytoids, (e.g. Ammodytes tobianus, Fig. 4c) are the only other perciform group found to have a sphenotic with a wide anterolateral face. However, in these taxa the sphenotic forms the posterodorsal margin of the orbit and in this respect it retains the plesiomorph condition. Thus, the condition of the wide anterolateral flange on the sphenotic in mastacembeloids which falls short of the postorbital margin, is a synapomorphy of the group. The mastacembeloids lack a posttemporal fossa (T ravers, 1984: 17). It is a common feature of lower teleosts and apart from having lost its bony roof, is widespread among higher euteleosts. In these fishes the posttemporal fossa generally lies lateral to the supratemporal fossa (Patterson, 1964: 449 & 1975: 392-5). Such fossae are typically found in the perciform neurocranium. The lack of a posttemporal fossa must, by virtue of the widescale presence of this fossa in teleosts, be a secondary loss and as such may be considered an apomorphic feature of these fishes. It is not, however, a feature restricted to these fishes. The fossa appears to have been lost independently in the following acanthomorph lineages: ophidioids (e.g. Campus acus), synbranchoids (e.g. Synbranchus marmoratus\ blennioids (e.g. Notograptus guttatus), trachinoids (e.g. Trachinus viperd) callionymoids (e.g. Callionymus lyrd), acanthuroids (e.g. Acanthurus bahrianus) and channoids (e.g. Channa obscurd). A wide subtemporal recess in the posterolateral wall of the neurocranium, defined in part by the prootic, pterotic and exoccipital, is characteristic of all mastacembeloids (T ravers, 1984) and is the site of origin for the levator musculature of the branchial arches. This recess, which is little more than a shallow lateral concavity, is not thought to be the homologue of the subtemporal fossa in lower teleosts (Forey, 1973; Patterson, 1975). The occipital condyle in all mastacembeloid taxa is in the form of a tripartite, concave socket (Travers, 1984: 71) formed from equal contributions by the exoccipitals and the basioccipital. The hemispherical anterior face of the first centrum articulates in this socket. A tripartite occipital condyle occurs in the majority of teleosts including the more basal assemblages (Forey, 1973: 12-19; Patterson, 1975: 318). Among these fishes the tripartite basi- and exoccipital facets are arranged in a variety of ways. Rosen & Patterson (1969) recognised two general apomorphic conditions of the condyle in acanthomorph lineages. The paracanthopterygians they examined have the exoccipital facets displaced laterally, losing contact with the basioccipital facet, whereas, in the acan- thopterygians they examined the exoccipital facets were displaced dorsally. Thus, the masta- cembeloid arrangement in which the basi- and exoccipital facets are firmly joined into a single tripartite, concave socket is considered to represent an even more derived condition. However, this arrangement is not unique to mastacembeloids as the facets are also developed into a single concave socket in myctophids, polymixids and ammodytoids. In, for example, Myctophum punctatum (illustrated by Rosen & Patterson, 1969: fig. 6 ID) the anterior face of the first centrum is tightly fused with the occipital socket and cranial movement must be considerably restricted. The tripartite occipital socket in Polymixia japonica, a species recently interpreted by Zehren (1979) as a basal percomorph, was also illustrated by Rosen & Patterson (1969: fig. 6 IE). In this taxon the anterior face of the first centrum has expanded into a convex condyle that fits into the occipital socket. This convexity of the first centrum may be the result of 94 R. A. TRAVERS FDL Pt Asph Asph Pt FDL / Bs RAW PrAP Fig. 4 Lateral view of neurocranial otic region in; (a) Pholidichthys leucotaenia, left side; (b) Xiphistes mucosus, right side; (c) Ammodytes tobianus, left side. the osteoid cone (Patterson, 1975) having fused with it and the anterior part of the cone becomes rounded to fit into the occipital facet. This apomorphic development of the anterior face of the first centrum has been taken a stage further in mastacembeloids. Here the anterior face is produced into a hemispherical condyle, that functionally forms a 'ball and socket' joint between the vertebral column and neurocranium. This is a synapomorphy of all mastacembeloids and is important evidence in support of the monophyletic origin of the group (see p. 1 08). In no other acanthopterygians, apart from the ammodytoids (Gosline, 1963), is there a 'ball and socket' joint between the MASTACEMBELOIDEI II: PHYLOGENETIC 95 vertebral column and neurocranium. Although the arrangement of the joint in ammodytoids is of a similarly derived type, in the absence of any further derived characters shared by these fishes it is thought that the resemblance is homoplastic. The absence of a posttemporal bone was first noted by Regan (1912) in Mastacembelus armatus and is confirmed by all mastacembeloid taxa I have examined. In place of the posttemporal, 1-3 ossified tubules occur in most species. Among teleosts, the only other groups lacking a posttemporal bone are the anguilloids, for example Anguilla anguilla, the notacanthids (McDowell, 1973: 137) and some siluroids (G. Howes, pers. comm.). MExsc Ptm LExsc Sc MExsc Ptm LExsc PtT Sc Ptm Sc Fig. 5 Articulation of the left posttemporal bone to the pectoral girdle in: (a) Zoarces viviparus; (b) Lycodes brevipes; (c) Dadyanos insignis. Viewed obliquely from a dorsolateral position. 96 R. A. TRAVERS Sc Ptm Fig. 6 Articulation of the left posttemporal bone in: (a) Monopterus albus; (b) Synbranchm marmoratus. Lateral view of the elements which are depicted in situ. In the true eels the pectoral girdle lies posterior to the cranium, adjacent to the 7th or 8th abdominal vertebrae. The postcranial sensory canal in Anguilla passes through three small dermal tubules lying between the cranium and pectoral girdle. Zoarcoids, for example Zoarces viviparus, Lycodes brevipes and Dadyanos insignis (Fig. 5), are also distinguished by their eel-like shape and posteriorly displaced pectoral girdle. This is connected to the neurocranium (epioccipital) by a narrow, blade- like posttemporal bone, through which the postcranial sensory canal does not run; instead it passess laterally, parallel to the posttemporal and is housed in two ossified dermal tubules. From the condition in zoarcids it seems reasonable to conclude that the ossified post- cranial tubules in mastacembeloids and some anguilloids are not necessarily remnants of the posttemporal. The absence of a posttemporal distinguishes mastacembeloids from all other neoteleosts. Its loss, like that in some anguilloids (discussed above) and notacanthids (McDowell, 1973), is correlated with the posterior position of the pectoral girdle, (see Travers, 1984: 73). The synbranchids also come into the category of eel-like neoteleostean fish with a reduced, posteriorly displaced pectoral girdle. A well-developed, forked, posttemporal connecting the basicranium to the pectoral girdle (generally adjacent to the 3rd & 4th abdominal vertebrae) occurs in Monopterus albus (Fig. 6a) and Ophisternon (Rosen & Greenwood, 1976: 63). On the other hand, in Synbranchus (e.g. S. marmoratus, Fig. 6b) the posttemporal is reduced to a narrow blade-like bone that tapers posteriorly from the basicranium (to which its MASTACEMBELOIDEI II: PHYLOGENETIC 97 anterior end is attached) and is not in contact with the pectoral girdle. The tendency in synbranchids towards reduction in size of the posttemporal is probably related to the position of the pectoral girdle. In its most derived condition (e.g. 5". marmoratus) the posttemporal has lost its connection to the supracleithrum, an arrangment which, although unique to the synbranchids, could represent an early stage in the transformation of this bone to its con- dition in the mastacembeloids (i.e. complete loss). Unfortunately, suitable material of the most plesiomorph synbranchid — Macrotrema elegans — was not available for study. Extrascapular bones are absent in all mastacembeloid taxa, (with the exception of Mastacembelus brachyrhinus, discussed below). The supratemporal arm of the cephalic sensory canal is completely enclosed in a transverse tube that passes across the parietal (T ravers, 1984: 72). This canal surfaces at the medial edge of the parietal, and crosses the supraoccipital as an open channel, joining its partner in the midline. The position of the supratemporal arm of the cephalic sensory canal in mastacembeloids may be the result of the lateral and medial extrascapulars (which generally house the canal) fusing along the posterodorsal surface of the parietal, or because true parietals have been replaced by extrascapulars (McDowell, 1973: 25). Alternatively, the extrascapulars may have been lost and the supratemporal sensory canal enclosed by the parietals during ontogeny. In all mastacembeloids the parietal has a posterolateral flange (Travers, 1984: 72) that lies along the dorsal junction of the pterotic and epioccipital, and encloses the most lateral region of the supratemporal sensory canal. The expansion of the parietal in this region suggests that the extrascapulars may have fused with its posterodorsal surface, presumably in a manner similar to that described by Patterson (1977: 98) for the fusion between the median extrascapula and the supraoccipital in primitive clupeomorphs. Furthermore, in Mastacem- belus paucispinis (Travers, 1984: fig 35a & c) a short, independent tubule, reminiscent of the posterior tubule-like region of the lateral extrascapula in other perciforms (e.g. Perca fluviatilis), is fused to the parietal between the pterotic and epioccipital. In my specimen of Mastacembelus brachyrhinus (uniquely among mastacembeloids) there is an independent lateral extrascapula on the dorsal surface of the parietal (albeit on the left side only, Travers, 1984: fig37a&c). The lateral and medial extrascapulae in higher euteleosts are short tubules which house the supratemporal arm of the cephalic sensory canal system as it traverses the dorsal surface of the neurocranium. Extrascapulae of this type are of widespread occurrence in the gadi- form, batrachiodiform, cyprinodontiform, beryciform, scorpaeniform and perciform assem- blages. In some of these the lateral and medial extrascapulae may be fused to the underlying parietal (dorsal surface), as seen for example, in the zoarcoids (e.g. Zoarces viviparus & Lycodes muraend) and some blennioids (e.g. Malacoctenus delalandei, Haliophus guttatus, Acanthemblemaria maria & Congrogadus subduceus). In all these examples the extra- scapulae, although fused to the parietal, are clearly distinguishable, retaining their overall shape and size. However, in other blennioids including members of the Blenniidae (Springer, 1968) and Stichaeoidea (Makushok, 1958: his Stichaeoidae), extrascapulae are absent and the supratemporal sensory canal lies within the parietal. This arrangement of the supra- temporal sensory canal, apart from the mastacembeloids, is found only in these blennioids among the teleosts examined. In some percomorphs, however, the extrascapulae and the supratemporal arm of the sensory canal are both absent, for example in the gobioids and synbranchoids, as well as in Chaudhuria and Pillaia among the mastacembeloids (see below). Thus, the total absence of discrete extrascapulae in mastacembeloids is a derived feature shared only with some blennioids. The lack of a supratemporal commissural sensory canal in Chaudhuria and Pillaia, taxa in which there appears to be a progressive reduction in the extent of the cephalic sensory canal system (see p. 1 13), may well be a further stage in such a trend (one convergent with that in the gobioids and synbranchoids). JAWS The upper jaw in mastacembeloids is non-protrusile and in this respect is rather exceptional 98 R. A. TRAVERS among percomorphs; indeed, jaw protrusibility is a characteristic feature of most neoteleosts (Liem & Lauder, 1983). Among percomorph fishes, relatively long premaxillary ascending processes and maxillary cranial condyles are absent only in the synbranchids and the mastacembeloids. The simple non-protrusile jaws in these taxa are similar to the plesiomorph teleostean condition seen in the osteoglossomorph, elopiform, clupeomorph and protacanthopterygian assemblages. Although the non-protrusile jaws in mastacembeloids and synbranchids appear to be of this type they have presumably been secondarily redeveloped from protrusile jaws (see Gosline, 1983: 324, discussed below p. 109 & 129). The mastacembeloid dentary has a posteroventral extension (Travers, 1984: 80) which tapers posteriorly. This process extends from the rim of the posterior sensory canal opening and lies along the ventral edge of the anguloarticular. In representatives from nearly all major teleostean lineages (particularly neoteleosts) the posterior opening of the sensory canal in the dentary marks the posteroventral tip of this bone; e.g. in Osteoglossum (Kershaw, 1976); Elops (Forey, 1973); Anguilla; Denticeps (Greenwood, 1968); Cyprinus; Salmo; Maurolicus; Chlorophthalmus; Myctophum; Percopsis; Gadus; Ophidian; Holocentrus and Serranus. In the numerous lower jaws taken from a wide selection of basal teleosts and illustrated by Nelson (1973), only that in Arapaima gigas (Nelson, 1973: fig. 2c & d) shows a process comparable with that in mastacembeloids, and must be considered a homoplasy. The dentary of Arapaima gigas illustrated in lateral view by Kershaw (1976: fig. 20, redrawn after Ridewood 1905) shows no posteroventral extension beyond the posterior opening of the sensory canal (the latter being clearly indicated); however, I have examined a skeletal preparation of Arapaima gigas and the long posteroventral process is clearly present. Of the remaining lower jaws illustrated by Nelson there is a posteroventral extension of the dentary beyond the posterior sensory canal pore (albeit to a lesser extent than in the mastacembeloids) in some engraulids e.g. Coilia mystus and Thrissina baelama (Nelson, 1973: fig. 5C, D&E, F). Both these taxa are characterised by a long, pointed mandible, and the short posterior projection on the dentary may be regarded as homoplastic with that in mastacembeloids and Arapaima. Among euteleosts, the synbranchids are the only group (apart from the mastacembeloids) in which there is a long posteroventral process on the dentary (illustrated by Rosen & Greenwood, 1976: fig. 60 & 61). This process extends beyond the posterior sensory canal opening along almost the entire ventral edge of the anguloarticular. Thus, it seems reasonable to conclude that the posterior extension of the dentary is a synapomorphy uniquely shared by mastacembeloids and synbranchids among euteleostean fishes. The mastacembeloid mandible is also exceptional in the size and position of its coronomeckelian, a large bone (relative to the other mandibular elements) lying across the anterolateral face of the suspensorium. A small coronomeckelian on the medial face of the anguloarticular is of common occurrence throughout the teleostomes (Starks, 1916), and this is taken to represent the plesiomorph condition both with regard to its size and position. The coronomeckelian was shown by Starks (1916) to be the ossified anterior end of the A3 tendon. In support of this view he cited the condition of these elements in Spheroides annulatus ' . . . where the adductor tendon has obviously ossified for a short space leaving an interval of tendon between the ossified portion and the mandible'. This interpretation can also be applied to the arrangement of the coronomeckelian in mastacembeloids, except that in these fishes the tendon has ossified dorsal to the mandible. Tendons may become ossified (forming a sesamoid bone) in regions where tendon movement produces opposing frictional forces, for example where the tendon runs across a bony prominence. The long A3 tendon in mastacembeloids runs across such a prominence — the deep and somewhat bulbous anterolateral face of the ectopterygoid (Travers, 1984: fig. 5). The development of the unusually large, dorsally situated coronomeckelian in mastacem- beloids thus appears to be directly related to the shape of the ectopterygoid. The size and MASTACEMBELOIDEI III PHYLOGENETIC 99 position of the coronomeckelian in Chaudhuria and Pillaia (discussed below), taxa which lack an ectopterygoid with a deep anterolateral face, is evidence in support of this view. Although there is extensive variation in the size and shape of the coronomeckelian among teleosts, (see Fig. 7 a-g) in no other taxon is it comparable with that in mastacembeloids, for which it is taken to be a unique synapomorphy. The dorsal edge of the anguloarticular is straight in most mastacembeloids (Travers, 1984: 82) and lacks a coronoid process in all taxa except Macrognathus, where the coronoid elevation is low and broad based (see Travers, 1984: 80). The anguloarticular lacks a coronoid process in about half the synbranchid species recognised by Rosen & Greenwood (1976: 45), and this characteristic was considered to be of some phylogenetic interest by these authors. They were of the opinion that, 'Such coronoid prominences on the articular of teleosts are common, although in many unrelated fishes with elongate crania and jaws the process is absent'. The absence of a coronoid process on the anguloarticular in synbranchoids was thought by Rosen & Greenwood (1976) to be a plesiomorph feature of the taxon, and the processes present in Monopterus albus and Ophisternon species were thought to be independently gained autapomorphies. A coronoid process is absent on the anguloarticular in a variety of teleosts including the anguilloids (e.g. Anguilla anguilla}, some percoids (e.g. Scarus croicensis), blennioids (e.g. Notograptus guttatus and Pholidichthys leucotaenia [very low projection]), many gobioids (e.g. Gobius niger; Aphia minuta; Amblyopus brousonetti and Crystallogobius linearis), and notothenioids (e.g. Notothenia sema [very low projection]). These observations support the view held by Rosen & Greenwood (op. cit.) that many unrelated fishes lack a coronoid process and that its absence in percomorph lineages is plesiomorphic. PTERYGO-PALATINE ARCH Regan (1912) found that in mastacembeloids the ' . . . pterygoid is movably articulated with the lateral ethmoids external to the palatine', and noted that they are 'very perculiar' in these fishes. The ectopterygoids are large bones characterised by a deep anterolateral face and their direct connection to the lateral ethmoids (Travers, 1984: 83). This direct articulation of the large ectopterygoid (functionally replacing the palatine which lacks articulatory facets; see below) is the sole means by which the anterior end of the suspensorium is joined to the neurocranium in most mastacembeloid taxa. The articulation of the suspensorium to the neurocranium in almost all other euteleosts involves the palatine. Among lower teleosts, the palatine also plays a central role in suspensorial articulation with the neurocranium, although this does not necessarily exclude the ectopterygoid from having an articulatory function in some taxa, as for example in the osteoglossoids. In Osteoglossum there is only a single element in place of the ectopterygoid and palatine, which are, in the opinion of Kershaw (1976: 192), 'indistinguishably connected'. The anterior end of the 'ecto- palatine' bone is ligamentously connected to the lateral ethmoid in Osteoglossum. The ectopterygoid is enlarged anteriorly in some halosauroids (Greenwood, 1977); how- ever, in none does it articulate directly with the ethmoid region. Anguilloids too generally have an elongated suspensorium and in Anguilla anguilla (Fig. 8) its anterior articulation is effected by a long, narrow bone which extends to the ethmoid region and may incorpor- ate the palatine (the palatopterygoid of Matsui & Takai, 1959) because an ossified auto- palatine was not observed in Anguilla. This bone was considered by Norman (1926) to be an endopterygoid. More recently, Leiby (1979 & 1981) has found that in several ophichthid anguilloids the endopterygoid and metapterygoid form a single compound bone which fuses during ontogeny with the anterior edge of the hyomandibula. He identifies the long narrow bone connecting the suspensorium and neurocranium as an ectopterygoid in these fishes (Leiby, 1979: fig. 2F& 1981: fig. 12A). The suspensorium in Anguilla has, superficially, a remarkably similar apearance to that in several of the more highly derived mastacembeloids, particularly Chaudhuria and Pillaia Com MC Aa Com MM\_/ g MC Com MC Com Com, Fig. 7 Medial view of the right coronomeckelian in: (a) Notograptus guttatus; (b) Rhyacichthys aspw, (c) Trichogaster leeri; (d) Sandelia capensis; (e) Luciocephalus pulcher, (f) Hypoptychus dy bows kit; (g) Callionymus lyra. MASTACEMBELOIDEI III PHYLOGENETIC 101 Palopt Hyo Pop HyoSP Fig. 8 Anguilla anguilla, left hyopterygoid arch and preoperculum in lateral view. MP Pal Fig. 9 Congrogadus subduceus, left hyopalatine arch and preoperculum in lateral view. (e.g. compare Travers, 1984: figs. 17 & 25 with Fig. 8). However, its anterior articulation in Anguilla is effected by what is apparently the ectopterygoid (which may have fused with the palatine) articulating with the ethmoid region and maxilla, whereas, in Chaudhuria and Pillaia the ectopterygoid, which is also attenuated (and possibly fused with the palatine), articulates with the lateral face of the vomerine shaft (see Travers, 1984: 83). The endopterygoid is enlarged, functionally replacing the ectopterygoid, in the congro- gadids. In Congrogadus subduceus and Haliophus guttatus (Fig. 9) the anterior end of the bone is connected to the palatine which articulates the suspensorium with the neurocranium. The ectopterygoid in these species is a short, narrow bone connected to the anterior edge of the quadrate, and lacks an anterior point of articulation with the skull. In synbranchids there is a large ectopterygoid which Rosen & Greenwood (1976: 44-48) imply is associated with the absence of the endopterygoid and with the massive development of the basisphenoid. They term this trend one of the 'special' attributes of all synbranchids. In Synbranchus marmoratus (Fig. 10) the ectopterygoid is connected posteriorly (by its anterodorsal, but subdistal process) to a basisphenoid projection, and also has a more usual distal articulation with the palatine. This additional abutment has developed between the ectopterygoid and frontal in Ophisternon aenigmaticum (Gosline, 1983: fig. 3B). The degree 102 R. A. TRAVERS FFac PalT Ect Fig 10 Synbranchus marmoratus, left palatopterygoid articulation to the neurocranium, lateral aspect. of enlargement of the ectopterygoid is particularly great in Monopterus albus (compare Rosen & Greenwood, 1976: figs. 59 & 61). The palatine in Monopterus lies further anteriorly than that in Synbranchus (e.g. com- pare Rosen & Greenwood, 1976: figs. 58 & 59) and its articulatory function (between suspensorium and lateral ethmoid) is taken over by the ectopterygoid in both species. The increase in the size of the ectopterygoid in synbranchids (p. 1 10) must be an apo- morphic feature because it culminates in the anterior displacement of the palatine (associ- ated with a reduction in the size of its maxillary process and its close adherence to the vomerine shaft) and, consequently, the establishment of a direct articulation between the ectopterygoid and lateral ethmoid in some taxa (e.g. Monopterus). Apart from these highly modified synbranchids, a direct articulation between the ectop- terygoid and the lateral ethmoid is an otherwise distinctive synapormorphy found only in the majority of mastacembeloids. A long, thin palatine sutured along the postero lateral face of the vomerine shaft, and lacking articulatory facets, is a characteristic feature of most mastacembeloid taxa. In those euteleosts with a distinct palatine this bone generally has moveable joints with the neurocranium and the upper jaw, viz., with the lateral ethmoid, usually involving a medial palatine facet, and anteriorly with the maxilla (cranial condyle) usually involving a short, often curved, maxillary process developed from the palatine. Although the pala- tine in higher euteleosts is variable in shape and dentition it was found to possess these articulatory functions in most lineages except the mastacembeloids and synbranchids. The mastacembeloid palatine is a straight bone that is sutured to the lateral face of the vomer/parasphenoid junction below the lateral ethmoid to which it may be connected by a weak ascending spur (Travers, 1984: 84). It lacks any moveable articulation and does not have a maxillary process. Posteriorly, the palatine extends below the orbit, is dorsoventrally flattened, and supports the anterior fibres of the adductor arcuspalatini muscle in this region. This highly derived palatine is typical of all mastacembeloids and is presumably associated with the lack of a protrusile upper jaw. The reductional trend in the synbranchid palatine is probably also associated with the loss of protrusibility of the upper jaw. In the more highly derived synbranchids (e.g. Monopterus) both the close adherence of the palatine to the vomerine shaft, and the lack of its connection to the lateral ethmoid are derived features (synapomorphies) shared with the mastacembeloids but with no other higher euteleosteans. Development of the mastacembeloid palatine appears to be at a more derived stage (i.e. no maxillary process or connection to the ectopterygoid) than is the synbranchid MASTACEMBELOIDEI II: PHYLOGENETIC 103 palatine. This arrangement of the palatine in mastacembeloids and synbranchids is closely associated with the development of the ectopterygoid in these taxa and, in combination, these characters provide important evidence in support of an hypothesis of shared common ancestry for the two groups, evidence consistent with the indications provided by the dentary (see p. 98) and posttemporal (p. 96) bones. BRANCHIAL ARCHES In most mastacembeloids a small, round, toothplate is fused to the dorsal surface of hypo- branchial 3 (see also Maheshwari, 1965: fig. 6). Apart from these fishes a fused toothplate features (see p. 103), it could be a reliable synapomorphy for uniting these groups. Its (e.g. Nandus & Badus} and channids (e.g. Channd). The possible means by which this fused toothplate has developed in Nandus, Badus and Channa, and its value as an indicator of phyletic relationships were discussed by Nelson (1969: 496-7). A fused toothplate on hypobranchial 3 is absent in Mastacembelus mastacembelus, Pillaia and in an assemblage of African species (T ravers, 1984: 94). In those species in which the toothplate is present, it is not always opposed by toothplates on epibranchial 1 . Toothplates on that element occur only in the larger predaceous species (e.g. Mastacembelus cunningtoni; Travers, 1984: fig. 64). Although a fused toothplate on hypobranchial 3 is a synapomorphy uniting most mastacembeloids, its mosaic distribution within the group and its occurrence in several phylogenetically distantly related taxa make it a character of limited taxonomic value when considered in isolation. A pair of processes descending from the posterolateral corners of basibranchial 2 are present in most mastacembeloids. The ventral tips of these processes are curved anteriorly and connect by converging ligaments to the posteroventral margin of basibranchial 1 (Travers, 1984: 92). Processes on basibranchial 2 have not been found in any outgroup taxa, their presence is treated as a synapomorphic character of the mastacembeloids, and further evidence in support of their monophyletic origin. DORSAL FIN SPINES The development of a series of spinous rays in the dorsal fin is a feature common to most acanthopterygian fishes. The number of dorsal spines may vary from a short row to a long series along the entire length of the dorsal surface in some fishes, e.g. the 'prickleback', Stichaeus hexagrammus. The spinous dorsal rays, regardless of their number, are generally interconnected by a thin membrane, however, this is absent in mastacembeloids. Isolated dorsal spines appear also in several notacanthid halosauroids. However, the spinous rays in these fishes have been shown by McDowell (1973: 143) to form ' . . . part of a (single dorsal) connected fin ... that is low and so heavily sheathed by scaly skin that only the separate tips of the spines are visible without dissection'. Furthermore, all these fishes lack a series of soft-rays posterior to the spinous part of the dorsal fin. Thus a dorsal fin composed of isolated short, stout spines anterior to a long series of soft rays must be a synapomorphy of the mastacembeloids as it is unique to the group. Myology CEPHALIC MUSCLES A number of myological features common to several mastacembeloid taxa were found to be unique to the group. The mastacembeloid levator operculi originates from the pterotic and inserts on the dorsolateral face of the operculum in all taxa examined (Travers, 1984: 119). The insertion of a levator operculi was described by Winterbottom (1974: 238) as the dorsal or dorsomedial face of the operculum. This is the condition in the majority of teleosts I have examined, and must be considered the plesiomorphic condition in neoteleosteans. A lateral insertion of fibres of the levator operculi, apart from in the mastacembeloids, was only found in species of the phylogenetically distantly related anguilloids (e.g. Anguilla anguilla), synbranchoids 104 R. A. TRAVERS (Liem, 1980a: 83) and blennioids (e.g. Notograptus guttatus). The lateral insertion oUevator fibres in all mastacembeloid and synbranchoid taxa suggests that, together with several other features (see p. 103), it could be a reliable synapomorphy for uniting these groups. Its occurrence in anguilloids and some blennioids in the absence of any further derived char- acters shared by these taxa and the mastacembeloids and synbranchoids, is interpreted as a convergence. A small independent muscle lies between the posterior edge of the preoperculum and anterolateral face of the operculum (ventral to the opercular lateral ridge) in all mastacem- beloids (Travers, 1984: fig. 79a), and is termed the 'musculus intraoperculi' (T ravers, 1984: 120) because of its position within the opercular series. Winterbottom (1974) describes no a AddMand Op Add Mand Sop Fig. 1 1 Ligamentous connection between the preoperculum and operculum in: (a) Cottus gobio; (b) Channa obscura. Lateral view, right side. MASTACEMBELOIDEI IK PHYLOGENETIC 105 muscle in this position in any teleost he examined; its absence is confirmed by all out- group taxa I have examined. Its unique occurrence in the mastacembeloids is a valuable synapomorphy for the group. A ligament between the posterior edge of the preoperculum and the anterolateral face of the operculum occurs in a number of phylogenetically diverse acanthomorph taxa. These include the cottoids (e.g. Agonus cataphractus, with a weak ligament between the preoper- culum and operculum and Coitus gobio with a broad strong ligament; Fig. 1 la), the gobioid Amblyopus broussonetti (in which a broad, weak collagenous band lies between the preoper- culum and operculum) and the channoids (e.g. Channa obscura, with a strong opaque ligament; Fig. lib). An origin of the adductor hyomandibulae from the posteroventral surface of the para- sphenoid is not recorded by Winterbottom (1974: 240), but this condition is typical of all mastacembeloids (Travers, 1984: 125). Also, a musculous insertion of this muscle partly on the medial face of the symplectic seems to be a general feature of all mastacembeloids and one not found in other groups (Winterbottom, 1974 gives the muscle's insertion site as the 'posterodorsomedial face of the hyomandibular'). The arrangement of these features of the adductor hyomandibulae were not always readily discernible in the outgroup taxa I examined they, therefore, remain doubtful additional synapomorphies of the group. The dorsal expansion of fibres of the hyohyoidei adductores above the upper branchio- stegal rays (medial to the suboperculum and operculum) and their insertion on the cleithrum, supracleithrum and posttemporal tubules in mastacembeloids (Travers, 1984: 128) is an unparalleled apomorphic development of this muscle. The general condition of the hyohyoidei adductores in teleosts is as a medial sheet of fibres extending only between the distal portions of the branchiostegal rays. Winterbottom (1974) noted that the fibres may continue dorsally above the posterodorsal ray to attach to the medial faces of some of the opercular bones, but in no teleost did he find the dorsal expansion continuing above the operculum and across the lateral face of the body muscles. I have examined the hyohyoidei in representatives of each major teleostean lineage, particularly those in which the opercular opening is restricted. The anguilloids, e.g. Anguilla anguilla, have a very restricted opercular opening effected in part by the filiform branchiostegal rays which curve up around the posterodorsal corner of the operculum; see McAllister (1968: 79). Fibres of the hyohyoidei adductores pass across the medial face of these long branchiostegal rays, but do not extend further dorsally or insert on the lateral face of the body. McAllister (op. cit.) found a close resemblance between the anguilloid branchiostegal arrangement and that in the myctophoids. Here the hyohyoidei adductores are of general teleostean proportions. The ophidioids (e.g. Carapus acus), zoarcoids (e.g. Zoarces viviparas) some blennioids (e.g. Pholis gunnellus), callionymoids (e.g. Callionymus lyrd) and the tetraodontoids (e.g. Diodori) are all acanthomorph lineages with some restriction of the opercular opening. In none, how- ever, do the hyohyoidei expand across the lateral wall of the body, although in some tetraodontoids they may expand medially and contact their partners in the midline (Winterbottom, 1974). The synbranchids (particularly members of the Synbranchinae; Rosen & Greenwood, 1976) have a very restricted opercular opening. The 4-6 branchiostegal rays may have their distal halves poorly ossified in a number of synbranchids (Rosen & Greenwood, 1976), and although the family is characterised by small opercular bones, the branchiostegal apparatus and its musculature are hypertrophied (Liem, 1980 S ojS'oS 3£3 •s f •§ ^ •o ? 9P 8 e |"5 o > y s ^ -2 VJ £S. >> s: 'C 5" Q i~ -~ c ~ o .2 ^ 5 5U -c 5 ^ G a '— s .« ?i 2P c - o .s 5 >> c ^ js a > H-'- •S ^ X" % c c |§.S "O *O ^> 0 3.> a " 5 S _. vi I •.= • £^ c - . ^ (U §3 a E •5 se -G g.0- g. ^2, c c« -^ '"" v- ^» *• o i S 5 00 '•o C ^ G -C) 5 ^ MASTACEMBELOIDEI III PHYLOGENETIC 135 This character is present in Mastacembelus armatus, M. erythrotaenia, M. mastacem- belus, M. oatesii and M. unicolor. Mastacembelus aiboguttatus is tentatively assigned to this group because of its superficial resemblance to the other members, but lack of material precludes a definite decision on its membership. No other characters were found that make it possible to analyse further the interrelation- ships of these 6 species which, therefore, are treated as an unresolved polychotomy (D). If polychotomies of this type prove to be inherently unresolvable they could reflect an evolutionary reality (Eldredge & Crancraft, 1980). The other sublineage stemming from the 5th node (E to J) is defined by 8 synapomorphies, represented by the 6th node in the cladogram, viz: 6 1 . Posterior region of parasphenoid undivided, apart from its tip, and excavated to form a pit-like depression on ventral surface (p. 111). 62. Expanded ventrolateral face of exoccipital and postero ventral margin of basioc- cipital, associated with deep basicranium (p. 111). 63. Deep basioccipital fossa accommodates anterior end of Baudelot's ligament (p. 1 1 1 ). 64. Dorsal surface of frontal slopes ventrolaterally (p. 1 13). 65. Elongate narrow neural and haemal spines, associated with a deep body (p. 126). 66. Relatively low total vertebral count (see p. 126 & Travers, 1984: table 5). 67. Distinct anterior part of adductor arcus palatini muscleiinserts on the attenuated posterior edge of the 1st infraorbital (p. 130). 68. 6 slender, digitiform fimbriae around rim of each anterior nostril, associated with relatively long rostral appendage (p. 1 30). This sublineage contains the remaining 8 Mastacembelus species recognised by Sufi (1956) and the 3 Macrognathus species recognised by Roberts (1980). It represents a monophyletic assemblage defined by at least 3 unique synapomorphies (6 1 , 67 & 68). Apart from the char- acters visible without dissection (e.g. 64 & 68), and those visible in radiographs (e.g. 65 & 66), the remaining characters could not be checked in every species because suitable material for dissection was not always available. For that reason Mastacembelus guentheri, M. keithi and M. perakensis have been placed with Mastacembelus circumcinctus, M. caudiocellatus and M. maculatus in an unresolved polychotomy (E) pending further research. Detailed studies of Mastacembelus zebrinus, Mastacembelus pancalus, Macrognathus aculeatus, Macrognathus aral and Macrognathus siamensis however, have revealed a nested series of synapomorphic characters which clearly show the phylogenetic affinity of these species (F to J). The 7th node represents the 7 synapomorphies shared by species F to J, viz: 69. Anterolateral expansion of premaxillary tooth-bearing alveolar surface (p. 1 13). 70. Low and broad coronoid process on dentary associated with tendency for its medial face to the toothed (p. 1 1 4). 71. Dorsal edge of anguloarticular with low, broad-based coronoid expansion (p. 1 15). 72. Indented anterolateral edge of quadrate (p. 1 16). 73. Tendency for ventral limb of cleithrum to develop a wide lateral face (p. 125). 74. Loss of the maxillo-mandibular ligament (p. 129). 75. Tendency for A, to become the largest part of the adductor mandibulae muscle, and for tA, to be a long, strap-like tendon extending to rostral appendage (p. 129). This suite of characters could be considered as a single transformation as they appear to be closely correlated (i.e. all can be functionally associated with the jaw mechanism) and may result from a common heterochronic ontogenetic shift. Mastacembelus zebrinus (species F) shows the suite of synapomorphies represented by the 7th node in the cladogram and is defined by two autapomorphies: 76. Extremely wide ventrolateral face of cleithrum producing 'keeled' pectoral girdle (p. 125). 77. Urohyal ascending process long and thin, its dorsal tip contacting basibranchial 2 and anterior margin connected to the posterior edge of the keel on basibranchial 1 (p. 123). 136 R. A. TRAVERS The 8th node represents 3 synapomorphies that unite Mastacembelus pancalus (species G) more closely with the Macrognathus species than to any other mastacembeloid. These characters are: 78. Loss of palatine spur (p. 118). 79. Preopercular sensory canal with 3 central pores at the end of short descending branches (p. 1 19). 80. Basibranchial 1 keel with a cartilaginous ventral edge and direct articulation with the urohyal (p. 122). Although characters 79 and 80 are both neomorphs, character 78 is a reductional one and as such could have arisen independently (see p. 118) as it does in Mastacembelus macula- tus. However, in combination with the two other apomorphic characters it is considered to be a reliable indicator of the close phylogenetic relationship between these species. Mastacembelus pancalus is defined by a single autapomorphy: 8 1 . Faceted connection between anteromedial margin of quadrate and posterolateral face of ectopterygoid (p. 1 18). The 9th node in the cladogram represents 3 synapomorphies that unite the Macrognathus species (H to J). These can be summarised as follows: 82. Fragmentation of premaxillary toothbearing alveolar surface into small pairs of plates along ventral surface of the rostral appendage (p. 1 14). 83. Dorsal edge of anguloarticular notched by facet anterior to posterodorsal corner of the bone (p. 1 15) and flange on lateral commissure of prootic (p. 1 12). 84. Low number of dorsal spines resulting from loss of anterior elements in the series (P. 127). Character 82 is the underlying synapomoprhy of this group and its division into a further 3 separate states (85, 88 & 89) clearly demarcates the Macrognathus species each from the other. Macrognathus aculeatus (species H) can be defined by the state of the group's synapomorphy, and by a 2nd apomorphic character: 85. Rostral toothplates usually in 38-55 pairs (p. 1 14). 86. Extremely large basisphenoid, associated with wide opening to posterior myo- dome (see Travers, 1984: 53 & fig. 30a & b). The 10th node represents a single synapomorphy which indicates the closer relationship of M. aral (species I) and M. siamensis (species J) to one another than either is to Macrognathus aculeatus. 87. Deeply notched posterolateral margin of pterosphenoid forming anterior region of trigeminal foramen (see Travers, 1984: 65 & fig. 29a). Macrognathus aral and Macrognathus siamensis, respectively, may be defined by the state of the group synapomorphy (i.e. 82) in each species: 88. Rostral toothplates: 14-28 pairs (p. 1 14). 89. Rostral toothplates: 7-14 pairs (p. 1 14). The llth node represents 4 synapomorphies uniting members of the 2nd branch of the main dichotomy (the 4th node in the cladogram). The synapomorphies uniting this assemblage, which includes all the African taxa but no others, are: 90. Lack of ascending process on urohyal or direct articulation between this bone and basibranchial 1 (p. 122). 9 1 . Hypural plates, generally 2; tendency for parhypural fusion to ventral edge of lower plate; 8-10 principal fin rays and confluent caudal fin (p. 128). 92. Scapula foramen not completely bone enclosed (p. 125). 93. Tendency to have noticeably more caudal than abdominal vertebrae (p. 126). MASTACEMBELOIDEI II: PHYLOGENETIC 137 Within this assemblage, 2 major subdivisions (K, and L to S in the cladogram) can be recognised. Subdivision K serves, for the time being, as a 'catch-all' assemblage for species united by the presence of synapomorphies 90-93. No other characters could be determined which would allow further subdivision of the group which, therefore, is represented as an unresolved polychotomy. Included in this group are Mastacembelus albomaculatus, M. cunningtoni, M. ellipsifer, M.flavidus, M.frenatus, M. micropectus, M. moorii, M. ophidium, M. plagiostomus, M. platysoma, M. tanganicae, M. zebratus, M. congicus, M. shiranus, M. stappersii and M. vanderwaali. A number of other nominal species (listed in Travers, 1984: table 1), for which material was unavailable should probably also be included. The second major subdivision (L to S) is represented by the 12th node in the cladogram, and is denned by 2 synapomorphies: 94. No toothplate on pharyngobranchial 2 (p. 124). 95. Less than 5 preopercular sensory canal pores (p. 1 19). Both these are loss features, and each has arisen separately, but never in combination (see p. 1 19 & 124), in several of the species lumped provisionally in species complex K. Since these characters are found in combination only in taxa L to S, they are taken to indicate the phylogenetic unity of the assemblage. Group L to S (12th node in cladogram) can be subdivided into two sub-groups; L and M to S. L is a polychotomy (discussed on p. 123) of at least 15 species viz: Mastacembelus batesii, M. brevicauda, M. flavomarginatus, M. goro, M. greshoffi, M. liberiensis, M. loennbergii, M. longicauda, M. marchii, M. marmoratus, M. niger, M. nigromarginatus, M. reticulatus, M. sclateri, and M. ubangensis, and may possibly include several of the species for which study material was unavailable (see Travers, 1984: table 1). The second group contains species M to S, viz: Mastacembelus paucispinis, M. brachyrhinus, M. brichardi, M. latens, M. crassus, M. aviceps and the undescribed species (see p. 1 1 9). Group L is denned by 2 synapomorphies: 96. Loss of anterior process on hypobranchial 3, and no ligamentous connection to basibranchial 2 (p. 124). 97. Arched ventral processes on basibranchial 2 (p. 123). Although character 96 is a reductional one it is atypical of the condition in most outgroup taxa and is found in no other mastacembeloids. Therefore, it should be treated as a derived feature. Character 97 is a synapomorphic feature for this lineage. No other characters were found which would enable the interspecific relationships of these species to be resolved further, for the present they remain as an unresolved polychotomy. This is not the case for the second group of the 12th dichotomy. Taxa in this branch are united by 3 synapomorphies represented by the 1 3th node, viz: 98. A reduction in the number of preopercular sensory canal pores (p. 1 19). 99. Loss of toothplate on hypobranchial 3 (p. 124). 100. A reduction in the number of caudal vertebrae (p. 126). Although character 99 mimics the plesiomorphic condition, it is thought to be a secondary reduction that has arisen, independently, in some species recognised as part of polychotomy L (see p. 124). In combination, however, these characters are considered to be a unique synapomorphy suite for the species in which they occur (i.e. M to S). Detailed analysis of the species in this group has led to the resolution of their interspecific relationships. The 14th node in the cladogram represents 2 synapomorphies shared by Mastacembelus paucispinis and the undescribed Mastacembelus species. These are taken as indicative of the taxa being more closely related to one another than either is to any other species. These synapomorphies are: 101. Low number of dorsal spinous rays (their loss occurring posteriorly in the series), combined with a long soft rayed dorsal fin that extends across the junction between abdominal and caudal vertebrae (p. 127). 138 R. A. TRAVERS 102. Tendency for the anterior tip of the prootic to form a bridge by contacting the ventral edge of a pedicel on the frontal (p. Ill ). Although character 101 is a synapomorphy for both species, it can be separated into 2 states, each of which is species specific. Thus, Mastacembelus paucispinis (species M) can be defined by its having: 103. 7-10 dorsal spines (p. 127). and the undescribed Mastacembelus species (N) by its having: 104. 15-16 dorsal spines (p. 127). The 1 5th node in the cladogram represents 3 synapomorphies uniting the remaining 5 species, viz: 105. Loss of pleural ribs from the anterior abdominal vertebrae (p. 126). 106. Coronomeckelian very short (p. 1 15). 107. Part A2 of the adductor mandibulae hypertrophied (p. 130). These taxa can be considered a small species flock as they are endemic to the highly specialised rapids environment of the Lower Zaire River and conform to the prerequisite criteria discussed by Greenwood (1984). Of their features, 105 is a reductional one which does occur in some species provisionally assigned to species complex K (see p. 126). Whether this indicates that it has occurred independently in these various taxa or whether it indicates that the species in K may eventually be included in this group (O to S), cannot be determined at present. Mastacembelus brachyrhinus (species O) may be distinguished from other species associ- ated through node 1 5 by a single autapomorphic feature: 108. Dorsal expansion of the pterotic, associated with relatively small parietals, posterior expansion of frontals, and a tendency for the extrascapulae to be independent (p. 97). The 16th node represents the following 4 synapomorphies: 109. Relatively large saccular bulla (accommodated entirely within the prootic, see p. 112). 1 10. Short anterior region of frontal; roofs small orbital cavity (see p. 1 13). 111. Endopterygoid very small and splinter-like (see p. 1 16). 1 12. Ventral limb of cleithrum short and indistinct (p. 125). 113. Tendency to be microphthalmic or cryptophthalmic (p. 1 13 & 130). These synapomorphies unite Mastacembelus brichardi (species P), Mastacembelus crassus (species R) and Mastacembelus aviceps (species S), and probably Mastacembelus latens (species Q) as well, although lack of material precludes a definite allocation of this species. Mastacembelus brichardi may be distinguished from the other species by 2 autapomorphies, viz: 1 14. Loss of pigment (p. 85). 115. Eyes extremely small and lying medial to the levator arcus palatini muscle (see Travers, 1984: 125 & fig. 84b). The remaining taxa in the lineage containing species M to S are Mastacembelus crassus (R), M. aviceps (S) and, probably, M. latens (Q). The 1 7th node in the cladogram represents 6 synapomorphies for these species, viz: 116. Scaleless(p. 129). 1 17. Loss of basisphenoid (p. 1 10). 1 18. Anterior process of prootic very short or absent (p. 111). 1 19. Ectopterygoid with narrow anterolateral face (p. 1 18). MASTACEMBELOIDEI II: PHYLOGENETIC 139 120. Dorsal opening of preopercular sensory canal on the posterior edge of preoperculum (p. 119). 121. Low 'keel' on basibranchial 1 (p. 121). Mastacembelus latens is provisionally included in this group on the grounds that it shares one of these apomorphic characters (116). Until specimens of M. latens are available for dissection, the presence or absence of the other 5 synapomorphies cannot be determined. The 6 synapomorphies shared by Mastacembelus crassus and Mastacembelus aviceps (and possibly M. latens) are all reductional ones. Several of these characters (all, apart from 120 & 12 1 ), or at least the tendency for their manifestation, are seen in Pillaia and in Chaudhuria (to a lesser extent). However, these taxa are clearly more closely related to Mastacembelus sinensis (as discussed on p. 128) and this complex of 6 characters in combination is an indicator of the close phyletic affinity of M. crassus and M. aviceps. Mastacembelus crassus (species R) can be distinguished from Mastacembelus aviceps (S) by a single apomorphic character, viz: 122. Tendency for the trigeminal and facial foramina to be confluent (p. 111). Whereas, M. aviceps can be identified by a combination of 5 apomorphic features, of which three are autapomorphic (i.e. 123, 124 & 125), viz: 123. Extremely small, splinter-like pterosphenoid lying along dorsolateral edge of frontal (p. 110). 124. Loss of neurocranial precommissural lateral wall and trigeminofacialis foramina (Travers, 1984: 68 & fig. 40a & b). 125. Anterior region of maxilla reduced to a weak and thin process tightly connected to premaxilla (p. 1 13). 126. Loss of ventral processes on basibranchial 2 (p. 123). 127. Hypurals fused into single fan-like plate (p. 128). These characters, and that defining M. crassus are all reductional ones; one of them (char- acter 127) also occurs in Mastacembelus ellipsifer (see p. 128), and is a further example of a reductional feature developing independently. Proposed changes in classification Comparison of the present state of mastacembeloid taxonomy (Fig. 18) with the phylogeny of the group (Fig. 20) reveals the necessity for numerous alterations to the existing classifi- cation, if the phylogenetic relationships of the species are to be reflected. These taxonomic and nomenclatural changes can be summarised as follows: Re-allocation of the Mastacembeloidei from the Perciformes to the Synbranchiformes, as a sister taxon of the Synbranchoidei. Expansion of the family Chaudhuriidae to incorporate two genera. Elevation of Mastacembelus sinensis to a monotypic genus of the Chaudhuriidae. The generic synonymy of Chaudhuria with Pillaia (and Garo see p. 109), but retention of both taxa as subgenera of Chaudhuria. Division of the Mastacembelidae into two subfamilies (representing the Asian and African species, respectively). Restriction of the Mastacembelus generic concept to five Asian and one Middle Eastern species only. Expansion of the Macrognathus generic concept to include eight Asian species previously included in Mastacembelus. Creation of a new subfamily for the African taxa. Creation of new genera for the major African sublineages. 140 R. A. TRAVERS .0 I MASTACEMBELOIDEI III PHYLOGENETIC 141 Diagnoses for the Synbranchiformes, its suborders, families, subfamilies and genera Order SYNBRANCHIFORMES Berg 1940 Berg, L. S., 1940. Classification of fishes, both recent and fossil. Trav. Inst. Zoo/. Acad. Sci. U.S.S.R. 5: 1-517, (English translation). DIAGNOSIS. Eel-shaped acanthomorph fishes of small to moderate size (attaining max. length of approx. 1 m). Burrowing and cavernicolous habit commonly displayed. Lack pelvic fins or girdle, with caudal fin reduced or absent. Gill membrane attached to lateral wall of body by expansion of hyohyoidei adductores muscle; restricted opercular opening and insertion of levator operculi on lateral face of operculum. Prominent adductor mandibulae muscula- ture, with part A, lying ventral to A2, the latter tending to encroach across dorsal surface of neurocranium and into orbital cavity. Eyes small and well forward in skull. Anterior and posterior nostrils. Cycloid scales, small and oval, sometimes absent. Neurocranium attenuated, particularly precommissural region involving frontals, pterospenoid, vomer and parasphenoid; dorsal surface lacks crests or any form of sculpturing. Frontals turned down with prominent descending lamina. Infraorbital bones reduced apart from 1st. Palatines joined firmly to vomer in midline; generally tooth bearing. Vomer a long thin strut. Ecto- pterygoid articulates with lateral ethmoid, vomer or both. Non-protrusile upper jaw. Maxilla and premaxilla long and strut-like, with symphyseal and articulatory processes reduced or absent. Dentary with posterior extension along ventral edge of anguloarticular. Pectoral girdle remote from basicranium, posttemporal bone reduced (accompanied by loss of connection to pectoral girdle) or lost. Flexible craniovertebral joint. Dorsal gill arch skeleton positioned posteriorly; lacks first pharyngobranchial bone, with second pharyngobranchial reduced or absent. Numerous vertebrae. Pantropical and subtropical fishes from freshwaters at high and low elevations; some individuals reported from brackish waters; tendency for facultative air breathing and sex re- versal. 83 extant species currently recognised (no fossil record) in two suborders. Suborder SYNBRANCHOIDEI Boulenger, 1904 Boulenger, G. A., 1904. A synopsis of the suborders and families of teleostean fishes. Ann. Mag. nat. Hist. (7), 13: 161-190. DIAGNOSIS. Monotypic suborder with synbranchiform features given above. For annotated account of groups, species diagnoses and key see Rosen & Greenwood (1976: 49-66). Suborder MASTACEMBELOIDEI Greenwood et at, 1966 Greenwood, P. H., Rosen, D. E., Weitzman, S. H. & Myers, G. S. 1966. Phyletic studies of teleostean fishes, with a provisional classification of living forms. Bull. Am. Mm. nat. Hist. 131: 339-456. OPISTHOMI Boulenger, G. A., 1904. Ann. Mag. nat. Hist. (7), 13: 161-190. (subordinal rank). Regan, C. T., 1912. Ann. Mag. nat. Hist. (8), 9: 217-219 (ordinal rank). MASTACEMBELIFORMES Berg, L. S., 1940. Trav. Inst. Zoo/. Acad. Sci. U.S.S.R. 5: 1-517. DIAGNOSIS. Synbranchiform fishes with long dorsal and anal fin composed of isolated spinous rays anterior to long series of soft branched rays. Rostral appendage formed from anterior nostrils (at end of tubular extensions) positioned lateral to a central rostral tentacle. Rim of anterior nostril generally with two wide (fimbrules) and two narrow (fimbriae) flaps of skin. 'Musculus intraopercuW differentiated within opercular series. Elongate ethmovomer- ine region with long nasal and 1st infraorbital bones. Preorbital spine (posterior tip of 1st infraorbital) generally pierces skin. Tubular lateral ethmoids accommodate anterior end of massive nervus olfactorius. Pterosphenoids, joined firmly in midline; frontals completely en- close foramen magnum. Small compressed basisphenoid, occasionally absent. Anterior re- gions of prootic and sphenotic attenuated; former having long process extending into orbital cavity and latter having wide lateral flange. Saccular bulla generally small and house entirely 142 R. A. TRAVERS within the prootic. Extrascapulae (lateral and medial) lost and sensory canal enclosed by parietal. 'Ball and socket' craniovertebral joint. Obliquus superioris with anterior insertion on posterior edge of exoccipital. Posterior end of Baudelot's ligament forked. Very large cor- onomeckelian dorsal to dentary, across lateral face of suspensorium. Pair of ventral processes on basibranchial 2. Hypobranchial 3 with a round toothplate fused to dorsal surface. Pseudo- branch present. Ethiopian and Oriental distribution. Wide range in tropical and subtropical Africa, and in Asia continuously from Middle East to islands of Indonesia, and eastern China seaboard. 68 species currently recognised, in two families. Family CHAUDHURIIDAE Annendale, 1918 Annendale, N., 1918. Fish and fisheries of the Inle Lake. Rec. Ind. Mm. 14: 33-64. TYPE GENUS. Chaudhuria Annendale, 1918. DIAGNOSIS. Derived mastacembeloid fishes with above features but tending to small adult size and reduction or loss of many characters by heterocrony. Rostral appendage reduced or lost. Needle-like parasphenoid posterior processes. Dorsomedial region of exoccipital with perforated dorsal surface and separated from opposite number. 1-2 inner rows of pre- maxillary teeth. Ectopterygoid with narrow lateral face and very long antero-dorsal process. Palatine reduced in size (lacks suborbital flange). Three or fewer epicentral ribs. Interdigi- tating processes of anterior and posterior ceratohyals reduced or lost. No endopterygoid, basibranchial 2 ventral processes or epipleural ribs. Confined to China (including Taiwan and possibly Korea), Thailand, Burma and northern India. Family expanded to incorporate 4 species in two genera. Genus RHYNCHOBDELLA Bloch & Schneider, 1 80 1 Blotch, M. E. & Schneider, J. G., 1801. M. E. Blochii Systema Ichthyologiae iconibus ex illustratum. Post oblitum auctoris opus inchoatum absolvit, correxit, interpolavit . . . J. G. Schneider, Saxo. Berolini. TYPE SPECIES. Rhynchobdella sinensis Bleeker, 1870 DIAGNOSIS. Monotypic chaudhuriid genus. Parietals may contact medially; very large 3rd anal spine equal in size to the 2nd spine and separated from it by four vertebrae. Tip of vertical urohyal ascending process contacts underside of basibranchial 2. Widely distributed in China, Taiwan and possibly N. & S. Korea. Genus CHA UDHURIA Annendale, 1918 Annendale, N., 1918. Fish and fisheries of the Inle Lake. Rec. Ind. Mus. 14: 33-64. Pillaia Yazdani, G. M., 1972. J. Bombay Nat. Hist. Soc. 69(1): 134-135; Talwar, P. M., Yazdani, G. M. & Kundu, D. K., 1977. Proc. Indian Acad. Sci. 85: 53-6. Garo Yazdani, G. M. & Talwar, P. M., 1981. Bull. zool. Surv. India 4(3): 287-288. TYPE SPECIES. Chaudhuria caudata Annendale, 1918 DIAGNOSIS. Highly derived chaudhuriid fishes with particularly small adult size (generally between 40-60 mm.). Extreme reduction of many features appears to mimic plesiomorphic condition e.g. large saccular bulla partly within prootic, exoccipital and basioccipital, coronomeckelian reduced to a small ossicle on medial face of anguloarticular. Loss of numerous features including; pterosphenoid, basisphenoid, frontal ventral lamina, cephalic sensory canal system (reduced or lost), palatine, pharyngobranchial 2 toothplate, dorsal and anal spines, and scales. Lack of ectopterygoid articulation with lateral ethmoid (ectoptery- goid directly contacts lateral face of vomerine shaft); possibly associated with loss of palatine. Posterior end of vomer ventrally depressed; pars jugularis pierced by single relatively large foramen. MASTACEMBELOIDEI III PHYLOGENETIC 143 Three species confined to Thailand, Burma and northern India: Chaudhuria caudata Annendale, 1918 Chaudhuria indica (Yazdani), 1972 Chaudhuria khajuriai (Talwar, Yazdani & Kundu), 1977. Family MASTACEMBELIDAE Gunther, 1861 Giinther, A., 1861. Catalogue of the acanthopterygian fishes. 3. London. TYPE GENUS. Mastacembelus Scopoli, 1777 DIAGNOSIS. Mastacembeloid fishes with features as for the suborder (given above) and, additionally: wide separation of hyomandibula and metapterygoid, associated with large symplectic. 64 species widespread throughout the range of the suborder. Two subfamilies comprise the mastacembeloids Oriental and Ethiopian regions. Subfamily MASTACEMBELINAE subfam. nov. TYPE GENUS. Mastacembelus Scopoli, 1777 DIAGNOSIS. Mastacembelid fishes with distinct caudal fin generally unconnected to posterior ray of dorsal or anal fin. If connected (by membrane) caudal fin rays extend posterior to, and remain distinct from, last posterior dorsal and anal fin ray. Caudal fin skeleton with 4 separate and autogenous hypural plates. Rostral appendage varies from very large to intermediate size. Tendency to be brightly coloured. 1 7 species widely distributed from Middle East to SE. Asia including China and continen- tal islands of Indonesia, arranged in two genera. Genus MASTACEMBELUS Scopoli, 1777 Scopoli, J. A., 1777. Introductio ad Historiam Naturalem, Prague. TYPE SPECIES. Mastacembelus mastacembelus (Banks & Solander, in Russell), 1794; See Wheeler, 1955. DIAGNOSIS: Mastacembeline fishes of moderate to large size (over 50 cm.). Attenuated anterior arm of endopterygoid lies between ectopterygoid and lateral ethmoid connection. Neurocranium broad with rostral appendage of moderate size. 6 species, five widely distributed in SE. Asia including the continental islands of Indonesia: Mastacembelus alboguttatus Boulenger, 1893 Mastacembelus armatus (Lacepede), 1 800 Mastacembelus erythrotaenia Bleeker, 1870 Mastacembelus oatesii Boulenger, 1893 Mastacembelus unicolor (Kuhl & van Hasselt) Cuv. & Val., 1831 and the single Middle Eastern species: Mastacembelus mastacembelus (Banks & Solander, in Russell), 1794. Genus MACROGNATHUS Lacepede, 1800 Lacepede, B., 1 800. Histoire naturelle des poissons. Paris 2. TYPE SPECIES. Ophidium aculeatum Bloch, 1786 DIAGNOSIS. Mastacembeline fishes of moderate to small size (under 50 cm.). Deep bodied with elongate, narrow neural and haemal spines. Rostral appendage large and more elongate than in other mastacembeloids. Six slender and digitiform fimbriae surround rim of each anterior nostril. Distinct anterior part of adductor arcus palatini muscle inserts on attenuate posterior edge of 1st infraorbital. Narrow, but deep, neurocranium with dorsal surface of 144 R. A. TRAVERS frontals sloping ventrally; ventrolateral face of exoccipital and posteroventral margin of basioccipital expanded. Posterior region of parasphenoid undivided (apart from tip) and excavated to form pit-like depression on ventral surface of basicranium (for muscle attach- ment). Deep basioccipital fossa accommodates anterior end of Baudelot's ligament. Vertebral count relatively low. In addition, see diagnosis given by Roberts (1980: 387). Roberts' (1980) Macrognathus generic concept (i.e. Macrognathus aculeatus, M. ami and M. siamensis) now expanded to include eight Oriental species previously assigned to Mastacembelus (i.e. 'Mastacembelus' caudiocellatus, 'M'. circumcinctus, 'M'. guentheri, 'M'. keithi, 'M'. macula- tus, 'M'. pancalus, 'M'. perakensis and 'M'. zebrinus). 1 1 species widely distributed in SE. Asia and continental islands of Indonesia: Macrognathus aculeatus (Bloch), 1786 Macrognathus aral (Bloch & Schneider), 1801; see Roberts, 1980 Macrognathus siamensis (Giinther), 1861; see Roberts, 1980 Macrognathus caudiocellatus (Boulenger), 1 892 Macrognathus circumcinctus (Hora), 1924 Macrognathus guentheri (Day), 1865 Macrognathus keithi (Herre), 1 940 Macrognathus maculatus (Cuv. & Val.), 1 83 1 Macrognathus pancalus Hamilton Buchanan, 1822 Macrognathus perakensis (Herre & Myers), 1937 Macrognathus zebrinus (Blyth), 1859 Subfamily AFROMASTACEMBELINAE subfam. nov. TYPE GENUS. Afromastacembelus gen. nov. DIAGNOSIS. Mastacembelid fishes with confluent caudal fin rays continues with posterior rays of dorsal and anal fin. Caudal fin skeleton generally with two separate and autogenous hypurals, tend to have parhypural fused to lower edge of ventral element and to have only 8-10 principal fin rays. Loss of ascending process on urohyal and direct articulation between urohyal and basibranchial 1 . Scapula foramen not completely bone enclosed. Tendency to have noticeably more caudal than abdominal vertebrae. This subfamily represents the Ethiopian mastacembelid species widely distributed throughout tropical and subtropical regions of the continent. 46 species currently recognised and provisionally arranged in two genera. Genus CAECOMASTACEMBELUS Poll, 1958 Poll, M., 1958. Description d'un poisson aveugle nouveau du Congo Beige appartenant a la famille des Mastacembelidae. Revue Zool. Bot. Afr. 57: 388-392. TYPE SPECIES. Caecomastacembelus brichardi Poll, 1958 DIAGNOSIS. Afromastacembeline fishes of small to moderately large size. With no pharyngo- branchial 2 toothplate and less than five preopercular sensory canal pores. Species with atrophied eye tissues and one (i.e. type for genus) is anoptic. General morphological simplifi- cation (by secondary reduction and loss) occurs in microphthalmic and cryptophthalmic species (parallels condition in Chaudhuria, although not carried to such extremes). Distributed predominently in western half of continent and includes small species flock endemic to lower Zairean rapids. At least 22 species tentatively assigned to this genus, and probably several more, for which study material was unavailable (see Travers, 1984: table 1): Caecomastacembelus aviceps (Roberts & Stewart), 1976 Caecomastacembelus batesii (Boulenger), 1911 Caecomastacembelus brachyrhinus (Boulenger), 1899 Caecomastacembelus brevicauda (Boulenger), 1911 Caecomastacembelus brichardi Poll, 1958 MASTACEMBELOIDEI III PHYLOGENETIC 145 Caecomastacembelus crassus (Roberts & Stewart), 1976 Caecomastacembelus flavomarginatus (Boulenger), 1898 Caecomastacembelus goro (Boulenger), 1902 Caecomastacembelus greshoffi (Boulenger), 1901 Caecomastacembelus latens (Roberts & Stewart), 1976 Caecomastacembelus liberiensis (Steindachner), 1 894 Caecomastacembelus loennbergii (Lonnberg), 1895 Caecomastacembelus longicauda (Boulenger), 1907 Caecomastacembelus marchii (Sauvage), 1 892 Caecomastacembelus marmoratus (Perugia), 1 892 Caecomastacembelus niger (Sauvage), 1878 Caecomastacembelus nigromarginatus (Boulenger), 1898 Caecomostacembelus paucispinis (Boulenger), 1899 Caecomastacembelus reticulatus (Boulenger), 1911 Caecomastacembelus sclateri (Boulenger), 1903 Caecomastacembelus ubangensis (Boulenger), 191 1 Caecomastacembelus sp. and probably: Caecomastacembelus ansorgii (Boulenger), 1905 Caecomastacembelus cryptacanthus (Giinther), 1867 Caecomastacembelus laticauda (Ahl), 1937 Caecomastacembelus sanagali (Thys van den Audenaerde), 1972 Caecomastacembelus seiteri (Thys van den Audenaerde), 1972 Genus AFROMASTACEMBELVS gen. nov. TYPE SPECIES. Mastacembelus tanganicae Giinther, 1893 DIAGNOSIS. Afromastacembeline fishes of moderate to large size; occur predominently from eastern half of continent and include species endemic to Lake Tanganyika. All afromasta- cembeline species, other than those assigned to Caecomastacembelus, provisionally lumped in this 'catch-all' assemblage (which may not be monophyletic) pending closer examination of groups interspecific relationships. At least 16 species tentatively placed in this genus, and probably several more, for which study material was unavailable (see Travers 1984: table 1). Afromastacembelus albomaculatus (Poll), 1953 Afromastacembelus congicus (Boulenger), 1 896 Afromastacembelus cunningtoni (Boulenger), 1906 Afromastacembelus ellipsifer (Boulenger), 1 899 Afromastacembelus flavidus (Matthes), 1962 Afromastacembelus frenatus (Boulenger), 1901 Afromastacembelus micropectus (Matthes), 1962 Afromastacembelus moorii (Boulenger), 1898 Afromastacembelus ophidium (Giinther), 1893 Afromastacembelus plagiostomus (Matthes), 1 962 Afromastacembelus platysoma (Poll & Matthes), 1 962 Afromastacembelus shiranus (Giinther), 1 896 Afromastacembelus stappersii (Boulenger), 1914 Afromastacembelus tanganicae (Giinther), 1893 Afromastacembelus vanderwaali (Skelton), 1976 Afromastacembelus zebratus (Matthes), 1962 and probably: Afromastacembelus moeruensis (Boulenger), 1914 Afromastacembelus signatus (Boulenger), 1905 Afromastacembelus trispinosus (Steindachner), 1911 146 R. A. TRAVERS The revised classification can be summarised as follows: Series: Percomorpha (consists of 1 2 orders) Order: Synbranchiformes Suborder: Synbranchoidei Family: Synbranchidae (see Rosen & Greenwood, 1976) Suborder: Mastacembeloidei Family: Chaudhuriidae Rhynchobdella Chaudhuria Family: Mastacembelidae Subfamily: Mastacembelinae Mastacembelus Macrognathus Subfamily: Afromastacembelinae Caecomastacembelus Afromastacembelus Acknowledgements I am indebted to the Trustees of the British Museum (Natural History), and the Keeper of Zoology for access to the collections and necessary research facilities. Again, it is a pleasure for me to thank the staff in the Fish Section (and all associated with it) for their generous support and tolerance of many ichthyological 'teething' problems. In particular I warmly thank Dr P. H. Greenwood, for painstaking criticism of an earlier draft, and Gordon Howes; both of whom have been a continual source of inspiration. I also take great pleasure in thanking Dr D. R. Kershaw for contributions of a supervisory kind and Prof. A. D. Hoyes for his encouragement and support. Maintenance from Queen Mary College (Drapers' Studentship), the University of London (Univer- sity Studentship and Central Research Fund) and the Godman Exploration Fund (British Museum (Natural History)) is acknowledged with gratitude. Loans and gifts of specimens for this study were generously donated by (in addition to those mentioned in Part I); Dr V. G. Springer (USNM). Finally, my mother (Helen Travers) deserves special mention for typing and retyping well beyond the call of duty. References Alberch, P., Gould, S. J., Oster, G. F. & Wake D. B. 1979. Size and shape in ontogeny and phylogeny. Paleobiology 5: 296-3 1 7. Berg, L. S. 1940. Classification of fishes both recent and fossil. Trav. Inst. Zoo/. Acad. Sci. U.S.S.R. 5: 87-517, (English translation). Bhargava, H. N. 1953. Pseudobranch in Mastacembelus. Curr. Sci. 22: 343-344. Boulenger, G. A. 1915. Catalogue of the freshwater fishes of Africa in the British Museum (Natural History). IV. Taylor & Francis, London. Brichard, P. 1978. Fishes of Lake Tanganyika. T. F. H. Goad, B. W. 1979. A provisional, annotated check-list of the freshwater fishes of Iran. J. Bombay nat. Hist. Soc. 76(1): 86- 105. Gohen, D. M. & Nielsen, J. G. 1978. Guide to the identification of genera of the fish order Ophidiiformes with a tentative classification of the order. NOAA Tech. Rep. NMFS. 417: 1-72. Day, F. 1877. On amphibious and migratory fishes of India. J. Linn. Soc. (Zool). 13: 198-215. 1889. The fauna of British India, including Ceylon and Burma. Fishes (II). London. de Beer, G. R. 1958. Embryos and Ancestors. Oxford Univ. Press. Dobson, G. E. 1874. Notes on the respiration of some species of Indian freshwater fishes. Proc. Zool. Soc.Lond. 1874: 312-321. Eldredge, N. & Cracraft, J. 1980. Phylogenetic Patterns and the Evolutionary Process: Methods and Theory in Comparative Biology. Columbia Univ. Press. MASTACEMBELOIDEI II: PHYLOGENETIC 147 Farris, J. S. 1982. Outgroups and parsimony. Syst. Zool. 31: 328-334. Fink, W. L. 1982. The conceptual relationship between ontogeny and phylogeny. Paleobiology 8: 254-264. Forey, P L. 1973. A revision of the elopiform fishes, fossil and recent. Bull. Br. Mus. nat. Hist. (Geoi). Supplement 10: 1-222. Franke, H. J. 1965. Tropische Mastacembeliden in aquarium. Zusammenfanssung der bisherigen Kenntrisse ueber lebensweise and paarungsethologie. Ichthyologica 4: 3-7. Ghosh, E. 1933. An experimental study of the asphyxiation of some air breathing fishes of Bengal. J. Asiat. Soc. Beng. 29: 327-332. Gosline, W. A. 1963. Notes on the osteology and systematic position of Hypoptychus dybowskii Steindachner and other elongate perciform fishes. Pacif. Sci. 17(1): 90-101. 1968. The suborders of perciform fishes. Proc. U. S. Natn. Mus. 124 (3647): 1-78. 1983. The relationships of the mastacembelid and synbranchid fishes. Japan J. Ichthyol. 29: 323-328. Gould, S. J. 1977. Ontogeny and Phylogeny. Harvard Univ. Press. Greenwood, P. H. 1968. The osteology and relationships of the Denticipitidae, a family of clupeo- morph fishes. Bull. Br. Mus. nat. Hist. (Zool). 16: 213-273. 1976. A review of the family Centropomidae (Pisces, Perciformes). Bull. Br. Mus. nat. Hist. (Zool). 29: 1-81. 1977. Notes on the anatomy and classification of elopomorph fishes. Bull. Br. Mus. nat. Hist. (Zool). 32: 65-102. 1983. The zoogeography of African freshwater fishes: bioaccountancy or biogeography? In: Systematics Association Special Volume No. 23. Evolution, time and space: the emergence of the biosphere: 181-199. R. W. Sims, J. H. Price and P. E. S. Whalley (Eds.). Academic Press, London. 1984. What is a species flock? In: Evolution of Fish Species Flocks. Echelle, A. A. & Kornfield, I. (Eds.). Maine at Orono Univ. Press (In press). Rosen, D. E., Wietzman, S. H. & Myers, G. S. 1966. Phyletic studies of teleostean fishes, with a provisional classification of living forms. Bull. Am. Mus. nat. Hist. 131: 339-456. Giint her, A. 1861. Catalogue of the acanthopterygian fishes. 3. London. Humid Khan, M. 1934. Habits and habitats of food fishes of the Punjab. J. Bombay Nat. Hist. Soc. 37: 655-668. Hennig, W. 1950. Grundziige einer. Theorie der phylogenetischen systematic. Deutscher Zentralver- lag, Berlin. 1979. Phylogenetic Systematics. Univ. of Illinois Press. Hora, S. L. 1935. Physiology, bionomics and evolution of the air-breathing fishes of India. Trans. Nat. Inst. Sci. India. I: 1-16. Humphries, C. J. & Parenti, L. R. In press. Cladistic Biogeography. Oxford University Monographs on Biogeography. Oxf. Univ. Press. Job, T. J. 1941. Life-history and bionomics of spiny eel, Mastacembelus pancalus (Hamilton). Rec. Indian. Mus. 43: 121-135. Johnson, G. D. 1980. The limits and relationships of the Lutjanidae and associated families. Bull Scripps Instn. Oceanogr. tech. ser. 24: 1-1 14. Kershaw, D. R. 1976. A structural and functional interpretation of the cranial anatomy in relation to the feeding of osteoglossoid fishes and a consideration of their phylogeny. Trans. Zool. Soc. Lond. 33: 173-252. Kochetov, S. M. 1982. Spawning spiny eels, Trop Fish. Hobb. 30 (3): 8-13. Leiby, M. M. 1979. Morphological development of the eel Myrophis punctatus (Ophichthidae) from hatching to metamorphosis, with emphasis on the developing head skeleton. Bull Mar. Sci. 29 (4): 509-521. 1981. Larval morphology of the eels Bascanichthys bascanium, (Ophichthidae), with a discussion of larval identification methods. Bull. Mar. Sci. 31 (1): 46-71. Li em, K. F. 1963. The comparative osteology and phylogeny of the Anabantoidei (Teleostei: Pisces). Illinois Monogr. 30: 1-149. 1967. A Morphological Study of Luciocephalus pulcher with notes on gular elements in other recent teleosts. /. Morph. 121: 103-134. 1970. Comparative functional anatomy of the Nandidae (Pisces: Teleostei). Fieldiana (Zool). 56: 1-166. 1980a. Air ventilation in advanced teleosts: biomechanical and evolutionary aspects. In: Environ- mental Physiology of Fishes: 57-91. AH, M. A. (Ed.). Plenum Publishing Corp. New York. 148 R. A. TR AVERS 19806. Acquisition of energy by teleosts: adaptive mechanisms and evolutionary patterns. In: Environmental Physiology of Fishes: 229-334. Ali, M. A. (Ed.). Plenum Publishing Corp. New York. & Lauder, G. V. 1983. The evolution and interrelationships of the actinopterygian fishes. Bull. Mus. Comp. Zooi, 150(3): 95-197. Luling, K. H 1980. Biotop, begleitfauna and amphibische lebensweise von Synbranchus marmoratus (Pisces, Synbrachidae) in seitengewassern des Mittlaren Parana (Argentiniem). Bom. Zooi Beitr. 31: (1/2): 111-143. MacDonald, C. M. 1978. Morphological and biochemical systematics of Australian freshwater and esturine percichthyid fishes. Aust. J. Mar. Freshwater Res. 29: 667-698. Maheshwari, S. C. 1965. The head skeleton of Mastacembelus armatus, (Lacep.). J. zool. Soc. India. 17: 52-63. Makushok, V. M. 1958. The morphology and classification of the northern blennioid fishes (Stichaeoidae, Blennioidei, Pisces). Proc. Zool. Instil. (Trudy Zool. Inst. Akad. Nank. S,S,S,R) 25: 3-129. Matsui, I. & Takai, T. 1959. Osteology of the Japanese eel, Anguilla japonica. J. Shimonoseki Coll. Fish. 8: 56-62. Matthes, H. 1962. Poissons nouveaux ou interessants du lac Tanganika et du Ruanda. Annls. Mus. r. Afr. Cent, (in 8) Zool. No. Ill: 27-88. McAllister, D. E. 1968. The evolution of branchiostegals and associated opercular, gular and hyoid bones and the classification of teleostome fishes living and fossil. Bull. Natn. Mus. Can. (Biol. ser.) 77(221): 1-239. McCosker, J. E. 1977. The osteology, classification, and relationships of the eel family Ophichthidae. Proc. Cai Acad. ScL, 4th ser., 41 (1): 1-123. McDowell, S. B. 1973. Order Heteromi (Notacanthiformes). In: Fishes of the Western North Atlantic. Mem. Sears Fdn. mar. Res. I (6): 1-228. Mittal, A. K. & Datta Munshi, J. S. 1971. A comparative study of the structure of the skin of certain air-breathing freshwater teleosts. J. Zool. Lond. 163: 515-531. Nelson, G. J. 1969. Gill arches and the phylogeny of fishes, with notes on the classification of vertebrates. Bull. Am. Mus. nat. Hist. 141: 475-552. 1973. Relationships of clupeomorphs with remarks on the structure of the lower jaw in fishes. In: Interrelationships of Fishes: 333-349. Greenwood, P. H., Miles, R. S., & Patterson, C. (Eds). Academic Press, London. 1978 Ontogeny, phylogeny, paleontology and the biogenetic law. Syst. Zool. 21: 324-345. & Platnick, N. 1981 Systematics and Biogeography. Cladistics and Vicariance. Columbia Univ. Press, New York. Norman, J. R. 1926. The development of the chondrocranium of the eel (Anguilla vulgaris), with observations on the comparative morphology and development of the chondrocranium in bony fishes. Phil. Trans. R. Soc. B. 214: 369-464. Osse, J. W. M. & Muller, M. 1980. A model of suction feeding in teleostean fishes with some implications for ventilation. In: Environmental Physiology of Fishes: Ali, M. A. (Ed.). Plenum Publishing Corp. New York. Patterson, C. 1964. A review of the Mesozoic acanthopterygian fishes, with special reference to those of the English Chalk. Phil. Trans. R. Soc. B. 247: 213-482. 1975. The braincase of pholidophorid and leptolepid fishes, with a review of the actinopterygian braincase. Phil. Trans. R. Soc. B. 269: 275-579. 1977. Cartilage bones, dermal bones and membrane bones, or the exoskeleton verses the endo- skeleton. In: Problems in Vertebrate Evolution: 77-121. Andrews. S. Mahala, Miles, R. S. & Walker, A. D. (Eds.). Academic Press, London. 1982a. Classes and cladists or individuals and evolution. Syst. Zool. 31: 284-286. 19826. Morphological characters and homology. In: Problems of Phylogenetic Reconstruction: 21-74. Joysey, K. A. & Friday, A. E. (Eds.). Academic Press, London. 1983. How does phylogeny differ from ontogeny? In: British Society for Developmental Biology Symposium 6; Development and Evolution: 1-31. Goodwin, B. C., Holder, N. & Wylie, C. C. (Eds.). Cambridge Univ. Press. Polder, W. N. 1963. Mastacembelus pancalus. Met. Aq. 34(5): 100-104. Poll, M. 1953. Poissons non Cichlidae. Res. Scient. Explor. Hydrobiol. Lac Tanganika (1946-47). 3 (5A): 1-125. 1958. Description d'un poisson avengle nouveau du Congo Beige appartenant a la famille des Mastacembelidae. Revue Zool. Bot. Afr. 57: 388-392. MASTACEMBELOIDEI III PHYLOGENETIC 149 — 1959. Resultats scientifiques des Missions zoologiques an Stanley Pool subsidies par le Cemubac (Univ. libre de Bruxelles), et le Musee royal du Congo (1957-58). III. Recherches sur la faune ichthyologique de la region du Stanley Pool. Annls. Mus. R. Congo Beige. 71 (8): 75-174. — 1973. Les yeux des poisson avengle Africains et de Caecomastacembelus brichardi Poll en particular. Annls. speleol. 28 (2): 221-230. & Matthes, H. 1962. Trois poissons remarquables du Lac Tanganika. Annls. Mus. R. Afr. Cent. (In 8°) Zool. No. Ill: 1-26. Regan, C. T. 1912. The osteology of the teleostean fishes of the order: Opisthomi. Ann. Mag. nat. Hist. (8). 9:217-219. Ridewood, W. G. 1905. On the cranial osteology of the fishes of the families Osteoglossidae, Pantodon- tidae and Phractolaemidae. J. Linn. Soc. (Zool.). 29: 252-282. Roberts, T. R. 1975. Geographical distribution of African freshwater fishes. Zool. J. Linn. Soc. 57: 249-319. 1980. A revision of the Asian mastacembeloid fish genus Macrognathus. Copeia. 1980. 3: 385-391. & Stewart, D. J. 1976. An ecological and systematic survey of fishes in the rapids of the Lower Zaire or Congo River. Bull. Mus. Comp. Zool. 147 (6): 239-317. Rognes, K. 1973. Head skeleton and jaw mechanism in Labrinae (Teleostei : Labridae) from Norwegian waters. Arh. Univ. Bergen. Mat.-Naturv. 1971. Ser. no. 4: 1-149. Rosen, D. E. 1973. Interrelationships of higher euteleostean fishes. In: Interrelationships of Fishes: 397-513. Greenwood, P. H., Miles, R. S. & Patterson, C. (Eds.). Academic Press, London. 1982. Do current theories of evolution satisfy the basic requirements of explanation? Syst. Zool. 31: 76-85. & Patterson, C. 1969. The structure and relationships of the paracanthopterygian fishes. Bull. Am. Mus. nat. Hist. 141: 375-474. & Greenwood, P. H. 1976. A fourth neotropical species of synbranchid eel and the phylogeny and systematics of synbranchiform fishes. Bull. Am. Mus. nat. Hist. 157: 1-69. Schoenebeck, K. J. von. 1955. Neues von den Stachelaalen. Aquar. Terrar. Z. 9: 315-317. Schofield, D. 1962. Spiny eels. Trap. Fish Hobby. 10: 5-10. Skelton, P. H. 1976. A new species of Mastacembelus (Pisces: Mastacembelidae) from the Upper Zambezi River, with a discussion of the taxonomy of the genus from this system. Ann. Cape Prov. Mus. (Nat. Hist.}. 11 (6): 103-1 16. Springer, V. C. 1968. Osteology and classification of the fishes of the family Blenniidae. Smithson. Inst. U.S. Nat. Mus. Bull. 284: 1-85. & Freihofer, W. C. 1976. Study of the monotypic fish family Pholidichthyidae (Perciformes). Smithson. Contr. Zool. 216: 1-43. -, Smith, C. L. & Fraser, T. H. 1977. Anisochromis straussi, new species of protogynous hermaphroditic fish, and synonymy of Anisochromidae, Pseudoplesiopidae and Pseudochromidae. Smithson. Contr. Zool. 252: 1-15. Staeck, W. 1976. Die lebensraume im Tanganjikasee. Aquar. Mag. 1976: 164-168. Starks, E. C. 1916. The sesamoid articular, a bone in the mandible of fishes. Leland Stanford jr. Univ. Publs. Univ. Ser. 1916: 1-40. Stiassny, M. L. J. 198 1 . Phylogenetic versus convergent relationship between piscivorous cichlid fishes from lakes Malawi and Tanganyika. Bull. Br. Mus. nat. Hist. (Zool.) 40: 67-101. Sufi, S. M. K. 1956. Revision of the Oriental fishes of the family Mastacembelidae. Bull. Raffles Mus. 27: 93-146. Travers, R. A. 1981. The interarcual cartilage; a review of its development, distribution and value as an indicator of phyletic relationships in euteleostean fishes. J. Nat. Hist. 15: 853-871. 1984. A review of the Mastacembeloidei, a suborder of synbranchiform teleost fishes. Part I: Anatomical descriptions. Bull. Am. Mus. nat. Hist. (Zool.) 46: 1-133. Vari, R. P. 1978. The terapon perches (Percoidei, Teraponidae). A cladistic analysis and taxonomic revision. Bull. Am. Mus. nat. Hist. 159: 175-340. Watrous, L. E. & Wheeler, Q. D. 1981. The out-group comparison method of character analysis. Syst. Zool. 30: 1-11. Weitzman, S. H. & Fink, W. L. 1983. Relationships of the neon tetras, a group of South American freshwater fishes (Teleostei, Characidae), with comments on the phylogeny of New World characiforms. Bull. Mus. Comp. Zool. 150(6): 339-395. Wiley, E. O. 1981. Phylogenetics. The Theory and Practise of Phylogenetic Systematics. J. Wiley & Sons, New York. 1 50 R. A. TRAVERS \\ interbottom, R. 1974. A descriptive synonymy of the striated muscles of the Teleostei. Proc. Acad. Nat. Sci. Phil. 125 (12): 225-317. Yazdani, G. M. 1976. The upper jaw of Indian hill-stream eel Pillaia indica, Yazdani (Perciformes: Mastacembeloidei). Bull. Zool. Surv. India. 2: 213-214. & Talwar, P. K. 1981. On the generic relationship of the eel-like fish, Pillaia Khajuriai Talwar, Yazdani & Kundu (Perciformes, Mastacembeloidei). Bull. zool. Surv. India. 4(3): 287-288. Zehren, S. J. 1979. The comparative osteology and phylogeny of the Beryciformes (Pisces : Teleostei). Evolutionary Monographs. 1: 1-389. Manuscript accepted for publication 25 May 1983 British Museum (Natural History) Tilapiine fishes of the genera Sarotherodon, Oreochromis and Danakilia Dr Ethelwynn Trewavas The tilapias are cichlid fishes of Africa and the Levant that have become the subjects of fish-farming throughout the warm countries of the world. This book described 4 1 recognized species in which one or both parents carry the eggs and embryos in the mouth for safety. Substrate-spawning species, of the now restricted genus Tilapia, are not treated here. Three genera of the mouth-brooding species are included though in one of them, Danakilia, the single species is too small to warrant farming. The other two, Sarotherodon, with nine species, and Oreochromis, with thirty-one, are distinguished primarily by their breeding habits and their biogeography, supported by structural features. Each species is described, with its diagnostic features emphasised and illustrated, and to this is added a summary of known ecology and behaviour. Conclusions on relationships involve assessment of parallel and convergent evolution. Dr Trewavas writes with the interests of the fish culturists, as well as those of the taxonomists, very much in mind. 580pp, 188 illustrations include halftones, diagrams, maps and graphs. Extensive bibliography. Publication 1983. £50 0 565 00878 1 Publications Sales British Museum (Natural History) Cromwell Road London SW7 5BD England Titles to be published in Volume 47 Miscellanea A review of the Mastacembeloidei, a suborder of synbranchiform teleost fishes Part II: Phylogenetic analysis. By Robert A. Travers A review of the anatomy, taxonomy, phylogeny and biogeography of the African neoboline cyprinid fishes. By Gordon J. Howes The haplochromine species (Teleostei, Cichlidae) of the Cunene and certain other Angolan rivers. By Peter Humphry Greenwood Miscellanea Anatomy and evolution of the feeding apparatus in the avian orders Coraciiformes and Piciformes. By P. J. K. Burton A revision of the spider genus Cyrba ( Araneae: Salticidae) with the descriptions of a new presumptive pheromone dispersing organ. By F. R. Wanless Printed in Great Britain by Henry Ling Ltd., at the Dorset Press, Dorchester, Dorset UM 31 Bulletin of the British Museum (Natural History A review of the anatomy, taxonomy, phylogeny and biogeography of the Africai neoboline cyprinid fishes Gordon J. Howes Zoology series Vol 47 No 3 30 August 1984 The Bulletin of the British Museum (Natural History), instituted in 1949, is issued in four scientific series. Botany, Entomology, Geology (incorporating Mineralogy) and Zoology, and an Historical series. Papers in the Bulletin are primarily the results of research carried out on the unique and ever-growing collections of the Museum, both by the scientific staff of the Museum and by specialists from elsewhere who make use of the Museum's resources. Many of the papers are works of reference that will remain indispensable for years to come. Parts are published at irregular intervals as they become ready, each is complete in itself, available separately, and individually priced. Volumes contain about 300 pages and several volumes may appear within a calendar year. Subscriptions may be placed for one or more of the series on either an Annual or Per Volume basis. Prices vary according to the contents of the individual parts. Orders and enquiries should be sent to: Publications Sales, British Museum (Natural History), Cromwell Road, London SW7 5BD, England. World List abbreviation: Bull. Br. Mus. nat. Hist. (Zool.) Trustees of the British Museum (Natural History), 1984 The Zoology Series is edited in the Museum's Department of Zoology Keeper of Zoology : Dr J. G. Sheals Editor of Bulletin : Dr C. R. Curds Assistant Editor : Mr C. G. Ogden ISBN 0 565 05006 0 ISSN 0007-1498 Zoology series Vol47 No. 3 pp 151-185 British Museum (Natural History) Cromwell Road London SW7 5BD Issued 30 August 1984 >^W^.«-T 3 JA1 ff^^^^^ A review of the anatomy, taxonomy, p|yl6gefl$^$ 34 ] biogeography of the African neoboline cyprinid fishes **^ Gordon J. Howes Department of Zoology, British Museum (Natural History), Cromwell Road, London SW7 5BD Contents Introduction 151 Diagnoses, anatomy and taxonomy of the neoboline genera 152 Neobola .. 152 Chelaethiops 157 Mesobola gen. nov 168 Rastrineobola Phylogenetic relationships of the neoboline group 177 Neoboline distribution and its biogeographic implications 181 Acknowledgements References 1 84 Introduction In a review of the bariliine cyprinids (Howes, 1980), the genus Engraulicypris which previously had contained ten species was recognized as monotypic, its only species, E. sardella confined to Lake Malawi. Those species formerly included in Engraulicypris were re-assigned to the genera Neobola Vinciguerra, Chelaethiops Boulenger, and Rastrineobola Fowler, with the accompanying statement that they formed a monophyletic assemblage whose close relationships were with Asian phoxinines rather than with African bariliines (Howes, 1980: 196). Later, Howes (1983) modified these views and included Neobola, Chelaethiops and Rastrineobola among the bariliines, naming them as the neoboline lineage (see fig. 2 in Howes, 1983). Although a suite of supposed apomorphies characterizing the three genera were given (Howes, 1980: 195), together with lists of their contained species no detailed generic diagnoses were presented. The purposes of this paper are: 1 . To give diagnoses of the genera Neobola, Chelaethiops and Rastrineobola and establish a new genus to contain two species formerly assigned to Neobola. The characters used in these diagnoses are, for the greater part, those involving cranial anatomy. From previous studies (Howes, 1978; 1979; 1980; 1981; 1982; 198 3) and from out-group comparisons made in the course of this work it is clear that in cyprinids cranial characters provide the most pertinent information at all levels of investigation. 2. To review the taxonomy of the included species. Although the neoboline cyprinids are abundant in many lakes and rivers of east, central and west Africa, they are, as compared with other cyprinids poorly represented in collections both in terms of sample sizes and geographic range. The species are small-sized, pelagic zooplanktivores and form an import- ant part of the diet of many piscivores (see Fryer & lies, 1972; Lowe-McConnell, 1975; van Oijen, 1982). Previous taxonomic reviews have been those of Poll (1945) and Whitehead (1 962) but these authors relied for the most part on data compiled from the literature. Almost Bull. Br. Mm. nat. Hist (Zool) 47(3): 151-185 Issued 30 August 1984 151 1 52 GORDON J. HOWES all the material on which this review is based is from the collections in the British Museum (Natural History), and types of nearly all species have been examined. Other specimens used are from the collections of the Central African Museum in Tervuren (MRAC) and the American Museum of Natural History (AMNH). 3. To confirm the supposed monophyly of the neoboline genera and to consider their intra- and interrelationships. 4. To consider, in the light of their phylogenetic relationships, the biogeography of the neoboline taxa and some broader issues of African biogeography. Diagnoses, anatomy and taxonomy of the neoboline genera NEOBOLA Vinciguerra, 1895 Neobola is characterized by the articulation of the lower jaw extending posterior to the centre of the orbit; a dorsally channelled, broad supraethmoid; 10-12 olfactory lamellae on each half of the nasal rosette; 4-7 short gill-rakers on 1st ceratobranchial; small pectoral axial scale with a fleshy ventral border; pharyngeal teeth in two rows, 5-3; scales small with 5-8 parallel radii, 37-41 in the lateral line; the lateral line gently decurved anteriorly and running close to the ventral margin of the trunk with a wave-like irregularity along the caudal peduncle; swimbladder divided by a deep constriction into short anterior and posterior chambers, the posterior extending to above the pelvic fin. CONTAINED SPECIES. N. bottegi, N. Jluviatilis and N. stellae. Cranial anatomy Osteology (Figs 2-4). The ethmo-vomerine bloc is short, deep and triangular, its anterior border has a sloped indentation. The supraethmoid is narrow- waisted, the lateral margins of the bone are concave and raised, thereby forming a shallow dorsal channel; the frontals overlay the posterior border of the bone. The vomer has a concave anterior border which projects only slightly beyond the overlying mesethmoid. Each lateral ethmoid is truncated with the dorsal part of the lateral wall extending anteriorly (Fig. 4). The frontal sensory canal runs along the margin of that bone and is raised above the level of the supraorbital. There are 4 or 5 sensory canal pores; the first is extensive, and all open somewhat laterally. The nasal is long and curved into the concavity of the supraethmoid border; it has no dorsal pores. The infraorbitals are deep (Fig. 6), the sensory canal of the 1st runs along the ventral border of the bone, but in the 2nd, 3rd and 4th the canal runs along the orbital border. In the 5th, the canal passes through the centre of the bone. The supraorbital is broad and long, narrowly separated from the 5th infraorbital. The orbitosphenoids are connected with the parasphenoid via a narrow septum; thepteros- phenoid contacts the ascending process of the parasphenoid across a narrow area of bone. There is a small lateral sphenotic process which provides the greater part of the dilatator fossa. The prootic is large with the anterior trigemino-facialis foramen narrowly separated from its anterior border; there is a long lateral commissure. The posterior myodome is completely floored by the parasphenoid and basioccipital. A small posttemporal fossa is formed by the posterior separation of the dermo and autopterotics. The supraoccipital is small, lacks a crest, and is confined to the posterior slope of the cranium. The jaws are long, the posterior tip of the maxilla extends to, or beyond the centre of the orbit. The maxilla is slender with a low mid-lateral (palatine) ascending process which curved laterally (Fig. 7). The posterior part of the maxilla is spine-like with only a slightly expanded tip which extends to below the centre of the orbit. The premaxilla is slender with a low ascending process and slightly concave ventral border. The dentary is shallow with a long, high, backwardly sloped coronoid process; the dorsal margin of the dentary is gently convex (Fig. 8). The anguloarticular is short with a slightly concave dorsal border. A REVIEW OF THE AFRICAN NEOBOLINE CYPRINID FISHES 153 B Fig. 1 Representatives of neoboline genera: A, Neobola bottegi; B, Mesobola brevianalis; C, Chelaethiops bibie; D, Rastrineobola argentea (drawn from a Lake Kioga specimen). All genera have 7 branched dorsal fin rays (rarely 6 or 8) I 10 pectoral rays; I 7 pelvic rays and 10 + 9 principal caudal rays. Suspensorium (Fig. 9). The hyomandibula is broad and its dorsal facets are deeply separ- ated. There is a weak lateral flange, its recess providing the insertion of the levator arcus palatini muscle. The ento- and metapterygoids have a marked concavity toward the midline; the dorsal border of the entopterygoid is straight, that of the metapterygoid is markedly concave. The palatine is laterally compressed, its anterior head forming a tripartite process, the medial spur of which overlies the concavity between the mid- lateral and anterior ascending maxillary processes. 154 GORDON J. HOWES soc Fig. 2 Crania, in dorsal view: A, Neobola bottegi; B, Chelaethiops elongatus; C, Mesobola brevianalis; D, Rastrineobola argentea. epo = epioccipital; fr = frontal; le = lateral ethmoid; na = nasal; pa = parietal; pte = dermopterotic; soc = supraoccipital; sor = supraorbital. Branchial arches. The ceratobranchials bear 4-8 short gill-rakers. Pharyngeal teeth on the 5th ceratobranchial are caniniform, arranged in two rows (5.3). The operculum (Fig. 9) has a shallow concave posterior border, the ventro-posterior part of the bone being attenuated. A REVIEW OF THE AFRICAN NEOBOLINE CYPRINID FISHES V 155 Fig. 3 Ethmoid region in dorsal view: A, Neobola bottegi; B, Chelaethiops elongatus; C, C. bibie; D, Mesobola brevianalis; E, Rastrineobola argentea; F, Leptocypris niloticus. me = mesethmoid; pe = preethmoid; se = supraethmoid; v = vomer. Scale = 1 mm. Myology (Fig. 13). Section A, of the adductor mandibulae muscle originates from the quadrate and preoperculum. There are two divisions of the section. (1), a triangular anterior part, A,<2«£, which has an aponeurotic constriction below the anterior border of the eye and is bordered by the ligamentum primordium. This anterior segment continues along the maxilla as a narrow, parallel-fibred element and inserts below the outwardly curved palatine process of the maxilla; (2) a posterior part, A^post, extending from the preoperculum to insert partly on the rictal tissue and partly onto the rim of the dentary coronoid process. Muscle A2 is thin and narrowly triangular, inserting into an aponeurosis medial to the anguloarticular; a small Aw section originates from the aponeurosis. The levator arcus palatini is divided by the sphenotic process and inserts onto the lateral hyomandibular flange. The dilatator operculi is small and broadly triangular, originating from the sphenotic spur and inserting onto the small opercular process. Pectoral girdle (Fig. 10) The upright part of the cleithrum is short, the posterior lamellae lacking any central process; the tip of the horizontal limb extends to a point below the parasphenoid ascending process; the supracleithrum is short, articulating halfway along the cleithral limb; the coracoid is shallow with a fretted anterior border.There is no postcleithrum. Vertebral column The 1st vertebra has a flat articulatory face, it bears short lateral processes that are directed somewhat anteriorly. Centra 2 and 3 are fused and bear long, slightly posteriorly curved lateral processes. The caudal fin skeleton (Fig. 17 A), has 6 hypurals, the 6th being minute. The fused preural and ural centra (PU1+U1) bear a small neural spine; the epural is long and slender; paired uroneurals are lacking. The parhypural bears a short but wide hypurapophysis. Dorsal and ventral procurrent rays are well-developed. 156 GORDON J. HOWES epo exo me le os pts ps pro soc Fig. 4 Crania, in lateral view, (above) Neobola bottegi and (below) Chelaethiops elongatus. bop = basioccipital process; exo = exoccipital; lee = lateral ethmoid extension; os = orbitosphenoid; pro = prootic; ps = parasphenoid; pts = pterosphenoid; other abbreviations as in previous figs. Taxonomy Three species are assigned to Neobola, N. bottegi Vinciguerra, 1895; from the Webei Shebeli and Omo rivers; N. Jluviatilis (Whitehead), 1962 from the Athi and Tana rivers, Kenya and N. stellae (Worth ington), 1932 from Lake Turkana, Kenya. Whitehead (1962) separated Neobola Jluviatilis from N. bottegi on its having fewer lateral line scales and a higher number of branched anal fin rays. However, I find there to be no substantial differences in these features. There are 37-40 lateral line scales in N. bottegi cf. 38-40 in N. Jluviatilis. Anal fin rays range in N. bottegi from 14-18 (N14) cf. 19-21 (N18) in N. Jluviatilis. Gill-raker counts similarly have a continuous range; 5-8 in N. bottegi (7 and 8 in specimens from the Webi Shebeli River and 5-6 in other specimens from other localities listed below) cf 4-6 in N. Jluviatilis. In both species there are 12 nasal lamellae on each half rosette. There are, however, differences in the morphology of the pectoral axial scale, it being smaller and more lobate in N. Jluviatilis than in N. bottegi. Pharyngeal teeth in both species are in 2 rows, 4.3 or 5.3 in N. bottegi and 4.2 in N. Jluviatilis. The total vertebral number for both species is 40 or 41 (4 Weberian+14 abdominal + 2 1-22 caudal + the fused preural and ural centrum). It seems likely that N. Jluviatilis is but the southern population of N. bottegi; further collections from northern Kenya and southern Ethiopia should help resolve this problem. A REVIEW OF THE AFRICAN NEOBOLINE CYPRINID FISHES 157 Fig. 5 Crania, in lateral view, (above) Mesobola brevianalis and (below) Rastrineobola argentea. Neobola stellae (Worthington), 1932 is distinguished from N. bottegi and N. fluviatilis by its small size (maximum adult size measured, 25 mm SL), lower number of olfactory lamellae (10 on each half rosette), lower vertebral number (total, 37 or 38 cf. 40-41) and a small, bluntly triangular pectoral axial scale. There are 28-33 lateral line scales. Worthington (1932) gives a count of 34-37, but in fact none of the types have more than 34 lateral line scales even if one includes scales at the base of the caudal fin. Branched anal fin rays number 12-18. Pharyngeal teeth are in 2 rows, 4.2. SPECIMENS EXAMINED. Neobola bottegi; 1902.11.7:22, Imi, Webi Shebeli; 1905.7.25:94-5, Modjo R.; 1905.7.251:1 10-1 12, Wabbi R.; 1905.7.25:106-9, Iraro R.; 1950.1 1.25:5-6, Webi Shebeli; 1982.5.17:9-14, Webi Shebeli. Neobola fluviatilis; 1961.5.3:1 (holotype); 2-6 (paratypes), Athi R. near Kithimani; 1966.7.5:29^2, Athi R. at Yalta: 1966.8.25:6, Tana R. Neobola stellae; 1932.6.13:57-65 (syntypes); 1932.6.13:47-56, all labelled 'Lake Rudolf; 1973.8.6:65-76, Loiengalani; 1981.12.17:2488-2587, Lodge Spit; 1981.12.17:2173-82, Ferguson's Gulf, Lake Turkana. CHELAETH1OPS Boulenger, 1899 Chelaethiops is characterized by its pointed snout and long, shallow jaws, the articulation of the lower jaw extending posterior to the centre of the eye, a dorsally channelled broad to narrow supraethmoid, laterally opening frontal pores, 16-26 olfactory lamellae on each half rosette, few (5) to many (18) gill-rakers on 1st ceratobranchial, elongate pectoral axial scales, pharyngeal teeth long, recurved in three rows, 5.3.2. Scales with 7-9 widely 158 GORDON J. HOWES Fig. 6 Infraorbital bones: A, Neobola bottegi; B, Chelaethiops bibie\ C, Mesobola brevianalis; D, Rastrineobola argentea. spaced radii, 36-42 lateral line scales; the lateral line gently decurved anteriorly and with a pronounced curve above the pelvic fin base, often with a wave-like irregularity along the caudal peduncle; swimbladder has short anterior and posterior chambers, the posterior extending to above the pelvic fin (exceptionally, long in C. minutus, extending to above the anal fin origin). CONTAINED SPECIES. Chelaethiops elongatus, C. bibie, C. congicus, C. minutus and C. rukwaensis. Cranial anatomy Osteology. The ethmoid bloc is short and deep, its anterior border varies interspecifically from shallow to deeply indented (Fig. 3B & C). The lateral edges of the supraethmoid are A REVIEW OF THE AFRICAN NEOBOLINE CYPRINID FISHES 159 B Fig. 7 Maxillae: A, Neobola bottegi; B, Chelaethiops bibie; C, Neobola stellae; D, Mesobola spinifer; E, M. bredoi; F, M. brevianalis; G, Raster -ineobola argentea. Scales in mm. raised and the width of the dorsal channel is also interspecifically variable. The mesethmoid is deep, its arms widely divergent, abutting on large preethmoids. The vomer floors the mesethmoidal indentation. As in Neobola the lateral ethmoid has an anterior extension of its dorso-lateral wall (lee, Fig. 4). Thefrontals are narrower than those of other neobolines, the sensory canal is more highly developed, and there are 4 large, conspicuous pores all of which open laterally. The nasal is large and reduced to the sensory canal tube with one dorsal pore or none. The infraorbitals more closely resemble those of Neobola than any other neoboline genus, both in their width and orbital configuration. The sensory canal, however, runs along the orbital margin of the bones and the 4th infraorbital is longer than in Neobola (Fig. 6B). The neurocranium is similar to that of Neobola except that the sphenotic has a ven- trally directed process, and there is a ventral opening between the basioccipital and the parasphenoid. The jaws are long (Figs 7 & 8); the maxilla has a low, long palatine process and the bone terminates in a spine-like tip. The upper jaw bones are longer and more slender than those of any neoboline genus. The lower jaw closely resembles that of Neobola except that the symphysial process of the dentary is more prominent and the coronoid process is lower and backwardly sloped at a shallower angle. 160 GORDON J. HOWES cp B Fig. 8 Lower jaw bones: A, Neobola bottegi; B, Mesobola brevianalis; C, Chelaethiops elongatus; D, Rastrineobola argentea; E, Leptocypris niloticus. aa = anguloarticular; cp = coronoid process; ra= retroarticular. The suspensorial bones are like those of Neobola except that the hyomandibula has a pronounced concavity to its lower anterior border and only a slight indentation separates the articulatory condyles. The posterior condyle is shaped into a long, triangular process. The palatine, ecto-, ento- and metapterygoids are all longer than those elements in Neobola and the metapterygoid has only a slight anterior dorsal process (cf. Figs 9A & B). The operculum has a rounded dorsal border and a strongly attentuated, rather triangular lower border (Fig. 9B). Myology (Fig. 13). Muscle adductor mandibulae A, is divided into anterior and posterior segments as in Neobola. The anterior part A^nt, extends from the quadrate to insert below the maxillary cleft. The segment contains an aponeurotic constriction below the anterior border of the eye. The aponeurosis is divided so that the orbital part serves as the site of the attachment for the fibres from the lower part of the muscle, and the outer division is continuous with the ligamentum primordium and serves as the site of attachment for the fibres of the anterior part of that muscle. The posterior segment of muscle A,, A^post, originates from the preoperculum and inserts on to the coronoid process of the dentary and on the rictal tissue. There is no Aw section of the adductor mandibulae. The configuration of other jaw and suspensorial muscles are as described for Neobola. Pectoral girdle (Fig. IOC) The pectoral girdle resembles that of Neobola; the coracoid is deeper than in other neobolines and has a markedly fretted antero-ventral margin. There is no postcleithrum. Vertebral column (Fig. 12) The anterior face of the 1st centrum is rounded (cf. flat in Neobola), the centrum bears long lateral processes, the distal tips of which underlie the processes of the fused 2nd and 3rd centra. The lateral processes of the fused centra are curved posteriorly, their distal tips reach- ing to a level with the articulation of the 3rd and 4th centra. The neural complex is upright, the 4th neural spine sloped backward at a low angle (Fig. 12). A large plate-like supraneural, possibly two fused supraneurals, overlies the 4th neural spine and extends backwards to above the 5th; this is followed by 6 or more elements varying in shape from triangular plates to thin rods. The caudal skeleton differs from that of Neobola in elongation of the 6th hypural, the neural spine on the fused preural-ural centrum and the hypurapophysis (Fig. 17B). A REVIEW OF THE AFRICAN NEOBOLINE CYPRINID FISHES hy met pal ent op 161 Fig. 9 Suspensoria in medial view: A, Neobola bottegi; B, Chelaethiops elongatus; C, Mesobola brevianalis; D, Rastrineobola argentea. ect = ectopterygoid; ent = entopterygoid; hy = hyomandi- bula; met = metapterygoid; op = operculum; pal = palatine; po = preoperculum; q = quadrate; sy = symplectic. In Mesobola and Rastrineobola, the symplectic is shown in solid black. Taxonomy Boulenger (1911) included two species in Chelaethiops, viz: elongatus and bibie. Since then only one other species has been described, C. katangae Poll, 1948. Howes (1980) assigned three other species to the genus, Engraulicypris congicus Nichols & Griscom, 1917, Neobola minuta Boulenger, 1906, and Engraulicypris rukwaensis Ricardo, 1939. Chelaethiops elongatus Boulenger, 1899, is characterized by its long and pointed snout and upwardly inclined head (Fig. 19B); a distinct mandibular symphysial notch; 16-17 olfactory lamellae on each half rosette; 5 short gill-rakers on the 1st ceratobranchial; 16-17 branched anal fin rays; 36-38 lateral line scales; axial pectoral scale 30-33% pectoral fin length (Fig. 18C); pectoral fin extending to the origin of the pelvic fin; lower part of posterior opercular border attenuated; pharyngeal teeth in 3 rows, 5.3.2; vertebrae 38 or 39 (4 + 1 1 - 12 + 21 -22 + 1, see p. 156 for method of counting). Distribution of the species is the Zaire drainage. Chelaethiops bibie (Joannis), 1835, is characterized by its narrow supraethmoid with well-developed dorsal channel (Fig. 3C). The snout is pointed but more strongly curved than in C. elongatus and there is a prominent ridge above the eye (Fig. 19 A). There are 12-13 olfactory lamellae on each half rosette; 9 or 10 short gill-rakers on 1st ceratobranchial; 16-17 162 GORDON J. HOWES -scl Fig. 10 Pectoral girdles: A, Neobola bottegi (medial view of disarticulated girdle); B, Mesobola brevianalis; C, Chelaethiops elongatus; D, Rastrineobola argentea (medial views), cl = cleithrum; co = coracoid; mc = mesocoracoid; sca = scaphium; scl = supracleithrum; ptt = posttemporal. bh hb cl Fig. 11 Branchial arches of Mesobola brevianalis. bb = basibranchials; bh-basihyal; C 1 — 5 = ceratobranchials; eb = epibranchials; hb = hypobranchials; if = infrapharyngobranchials. A REVIEW OF THE AFRICAN NEOBOLINE CYPRINID FISHES 163 vl Fig. 12 Anterior vertebrae, in ventral and lateral views, of Chelaethiops bibie. cla = claustrum; ic = intercalarium; oss = ossa suspensorium; pr = pleural rib; sc = scaphium; sn = supraneurals; tr=tripus; v = vertebrae; vp = lateral vertebral process. branched anal fin rays (rarely 14 or 18); 36-38 lateral line scales; the lateral line usually more irregular and sinuous than that of C. elongatus; axial pectoral scale about 20% of pectoral fin length (Fig. 18E); pectoral fin extending to beyond the origin of the pelvic fin; lower part of the posterior opercular border attenuated; pharyngeal teeth in three rows, variants are 5.3.1. or 4.3.1; vertebrae 36-40 (4+ 1 1 - 13 + 19-22 + 1). Daget (1954) described a subspecies from the Upper Niger which he named as C. elongatus brevianalis. Blache & Miton (1960) decided that Daget's taxon represented a species, which comprised two subspecies, C.b. brevianalis and C.b. lerei from the Mayo Kebbi. It would seem, however, that Daget did not consider C. bibie as he makes no mention of the species in his description. There is no difference between C. bibie and Daget's description of C. elongatus brevianalis. It may well be that Blache & Miton's (1960) taxon from the Mayo Kebbi represents a subspecies, or morphologically distinct population, but this fact has yet to be established. There are differences in vertebral counts between Nilotic C. bibie and those from Ghana as follow: Nilotic specimens; 12 or 13 abdominal and 22 caudal, Ghanian specimens; 1 1 or 12 abdominal and 19 or 20 caudal. Chelaethiops bibie occurs in the Nile (including Lake Turkana), Niger (eastern limit uncertain) and Volta. Chelaethiops minutus (Boulenger), 1906, is characterized by its narrow supraethmoid with well-developed dorsal channel; long and downwardly curved snout, its tip extending beyond that of the lower jaw (Fig. 19E); a prominent frontal ridge above the orbit; 25-27 olfactory lamellae on each half rosette; 17 or 18 long gill-rakers on the 1st ceratobranchial; 18-21 branched anal fin rays; 39-42 lateral line scales; axial pectoral scale 22-25% pectoral fin length (Fig. 1 8G); pectoral fin not reaching beyond the origin of the pelvic fin; posterior border of the operculum rounded; pharyngeal teeth in 3 rows, 5.3.2; vertebrae 39 or 40 (4+12-13 + 22-23 + 1). This species differs from other Chelaethiops in gill-raker length and number, an elongate posterior chamber of the swimbladder extending to, or just beyond, the anal fin origin and 164 GORDON J. HOWES A i post Fig. 13 Jaw muscles of (above) Neobola bottegi and (below) Chelaethiops elongatus. A = divisions of the adductor mandibulae muscle; 1 p = ligamentum primordium; lap = levator arcus palatini; the vertical line marks the orbital centre. the articulation of the anterior 3 or 4 supraneural bones. It is endemic to Lake Tanganyika (see Poll, 1953). Chelaethiops congicus (Nichols & Griscom), 1917, is characterized by its short and blunt snout (Fig. 19C); broad supraethmoid with shallow dorsal channel; 12-15 olfactory lamellae in each half rosette; 5-6 short gill-rakers on the 1st ceratobranchial; 16-18 branched anal fin rays; 38-42 lateral line scales; axial pectoral scale 30-33% pectoral fin length (Fig. 18F); pectoral fin extending to the origin of the pelvic fin; upper part of posterior opercular border concave; pharyngeal teeth in 3 rows, 5.4.1 or 5.3.1; vertebrae 41 (4+ 14 + 22 + 1). It appears that although authors have cited C. congicus in their comparative analyses, none has actually examined the type specimens (Poll, 1945; Whitehead, 1962; Ricardo, A REVIEW OF THE AFRICAN NEOBOLINE CYPRINID FISHES 165 Fig. 14 Jaw muscles of (above) Mesobola brevianalis and (below) Rastrineobola argentea. A, lateral and B, medial views. 1939). Poll (1948) described Chelaethiops katangae from the Lufira, Zaire. Poll did not, however, compare his specimens with C. congicus but only with C. elongatus and C. bibie. I find the holotype of C. congicus to be closely similar to the syntypes of C. katangae in all morphological aspects and consider C. katangae to be a synonym of C. congicus. The distribution of C. congicus is the Zaire basin. Chelaethiops rukwaensis (Ricardo), 1939, is characterized by its narrow supraethmoid with well-developed dorsal channel; pointed and curved snout (Fig. 19D); 16-17 olfactory lamellae in each half rosette; 5-6 short gill-rakers on the 1st ceratobranchial; 15-17 branched anal fin rays; 36-38 lateral scales; axial pectoral scale with wavy border (Fig. 18F), 20-25% pectoral fin length; pectoral fin extending to the origin of the pelvic fin; upper posterior border of the operculum shallowly concave, lower border slightly attenuated; pharyngeal teeth in 3 rows, 5.3.1; vertebrae 38^0 (4 + 14-16 + 19 + 1). 166 GORDON J. HOWES A] post Fig. 15 Jaw muscles of (above) Leptocypris niloticus and (below) Raiamas senegalensis. Ricardo (1939) described, from Lake Rukwa, a subspecies of Chelaethiops congicus, commenting: 'It is ... thought best to regard the examples from L. Rukwa as a new subspecies of E. congicus in order to show that they do differ from all forms previously known and that they are more closely related to the Engraulicypris in L. Tanganyika and the Congo than to any of the species found in other lakes or rivers.' From this statement it would seem that Ricardo was regarding the subspecies as a convenience category to demonstrate her opinion of relationship. The comparative material from Lake Tanganyika and the Congo used by Ricardo is in fact composite. The speci- mens from the Congo are those included herein under C. congicus, but those from Lake Tanganyika differ in morphometric and other characters and, in these respects, are closer to the samples from Lake Rukwa. As with the Lake Rukwa specimens, the Lake Tanganyika sample has a pectoral axial scale length of 20-25% the pectoral length, cf. 30-33% in C. A REVIEW OF THE AFRICAN NEOBOLINE CYPRINID FISHES 167 Fig. 16 Jaw muscles of Engraulicypris sardella. congicus, 16-18 olfactory lamellae, cf. 12-15 in C. congicus, and 36-38 lateral line scales, cf. 38-42 in C. congicus. There is an added difficulty concerning the exact provenance of the 'Lake Tanganyika' sample of C. rukwaensis. Neither Poll (1953) nor Brichard (1978) in their respective accounts of Lake Tanganyika fishes record any species of Chelaethiops (their Engraudicypris) other than C. minutus. The record of Lake Tanganyika C. congicus given by Ricardo (1939) is based on the material collected by Christy. No more precise locality is available than 'L. Tanganyika' and it may well be that the specimens were collected in the environs of the lake. Howes (1980) noted that C. rukwaensis was most closely related to an undescribed taxon from Lake Tanganyika. The taxon is the 'L. Tanganyika' sample discussed above. The status of the taxon is difficult to determine on the basis of such a small sample (55 specimens) and with imprecise locality data. It is further complicated by specimens from the Luiche river, Malagarasi system, having characters intermediate between the C. rukwaensis and C. congicus (viz: 38 lateral line scales, 14 branched anal fin rays and 12 olfactory lamellae on each half rosette). The 'L. Tanganyika'- Malagarasi forms may eventually prove to be a morphologically distinct local population (or subspecies) but for the present my study material is identified only as 'Chelaethiops rukwaensis'. SPECIMENS EXAMINED. Chelaethiops elongatus: 1901.12.26:31, Banzyville Ubanghi; 1912.12.6:9-10, Dungu, Uelle; 1919.9.10.241-2, Avakubi, Ituri; 1920.7.12:39^0, Banghi; 1975.6.20:308-40. 34-43, 344-47, Lualaba; 348-49, Ituri Bridge; MRAC 85848-947, Ankoro, Lualaba. Chelaethiops bibie. Nile: 1904.9.26:22-23. Kalioub, north of Cairo; 1907.13.2:1513-9, near Luxor; 1520-23, between Luxor and Asswan; 1524, Asswan; 1525, Kermeh, Nubia; 1526- 1626. Omdurman; 1627, White Nile; 1628-31, Lake No; 1632-51, 52-53, Gondokoro; 1913.11.11:5-7; Khor Barboy; 1924.5.21:1-5, near Cairo; 1981.2.17:1314-1358. Tode- nyang, L. Turkana; 1369-1431, Morago R., Turkana; 2378-2427, west shore, L. Turkana; 2428-2437, Ferguson's Gulf, L. Turkana. West Africa: 1935.5.29:9-18, Kaduna, Nigeria; 1969.11.14:109-23, Volta, Ghana; 1974.1.2:185, Black Volta; 1982.4.13:783-93, Bahindi, Nigeria; 794-797, Sokoto R., 821-830, Sokoto-Rima floodplain; 799-811, 812-820, Rima R., N. Nigeria. Chelaethiops minutus: 1906.9.8:55-60 (syntypes), Mbete (14-24 mm SL); 1955.12.20:1021- 1029, L. Tanganyika (54-66 mm SL); 1982.9.24:61-66, Kigoma (67-5-86-5 mm SL). 168 GORDON J. HOWES -pr ep Fig. 17 Caudal fin skeletons: A, Neobola bottegi; B, Chelaethiops elongatus; C, Mesobola brevianalis; D, Rastrineobola argentea. cpu = fused preural-ural centrum; ep = epural; h6 = 6th hypural; hp = hypurapophysis; ph = parhypural; pr = procurrent rays. Chelaethiops congicus; AMNH 6295 (holotype), Poko, Ubangi; MRAC 78108-9 (syntypes of C. katangae), Kafila, Lufira; BMNH 1919.9.10:238, Avakubi, Ituri; 1919.9.10:239, Busabangi, Lindi; 1902.4.14:47-8, Lindi R.; 1907.4.30:42, Atuwimi R.; 1909.7.9:62, Bumba (Boumba) R., Assobam, S. Cameroon; 1975.6.20:350-402, Lufira R. Chelaethiops rukwaensis; 1942. 12.31.: 19 1-2 10 (syntypes), Lake Rukwa; 1936.6.15:547-67, 'L. Tanganyika'; 1969.1.31:49-105, Lake Rukwa; 1971.6.22:137-138, Luiche R., Malagarasi system. MESOBOLA gen. nov. TYPE SPECIES. Engraulicypris brevianalis Boulenger, 1908, Ann. Natal M us. 1: 231 This genus is uniquely denned by its cranial, jaw bone and jaw muscle morphology. At the same time it can be included with Neobola, Chelaethiops and Rastrineobola on the basis of those synapomorphies denning the neoboline group (p. 177). Mesobola is characterized by a narrow, dorsally channelled supraethmoid; vomer forming a floor to the ethmoid inden- tation; pre-ethmoids directed rostrad; nasal without dorsal pores; frontal canal with 4 or 5 pores; narrow, triangular metapterygoid process; symplectic elongate with expanded tips; A REVIEW OF THE AFRICAN NEOBOLINE CYPRINID FISHES 169 Fig. 18 Pectoral axial scales: A, Neobola bottegi; B, Neobola stellae; C, Chelaethiops elongatus; D, C. rukwaensis; E, C. bibie; F, C. congicus; G, C. minutus. 8-10 olfactory lamellae on each half rosette; pectoral axial scale absent; pharyngeal teeth in 3 rows; swimbladder with long anterior and posterior chambers, the posterior curved downward and terminating above the anus. CONTAINED SPECIES. M. brevianalis, M. spinifer, M. bredoi and M. moeruensis. Cranial anatomy Osteology. The most characteristic feature of the cranium is the morphology of the ethmoid region (Fig. 3D). The ethmoid bloc is hour-glass shaped, the supraethmoid with a deep anterior notch, its lateral walls lamellate and forming a deep channel posterior to the notch. The mesethmoid has a V-shaped anterior indentation, the preethmoids being situated on the tips of the mesethmoid arms and thus considerably extending the depth of the indentation. The vomer floors the notch and where it meets the preethmoids there is a thickening of the bone, thus forming what appears to be medial processes of the preethmoids. In some specimens there is a small vomerine foramen. The lateral ethmoid margin curves antero-ventrally and lacks the dorsal lamellae present in Neobola and Chelaethiops. The frontal canal (Fig. 2C) lies close to the edge of that bone and has 4-5 pores. The size and position of the canal pores is interspecifically variable, but the posterior pore always opens laterally. The nasal is large, lacking dorsal pores. The cranial roof is convex in the area of the fronto-parietal suture; in all other neoboline genera the roof is flat. The infraorbitals (Fig. 6C) are shallow, as in Rastrineobola (cf. deep in Neobola and Chelaethiops). The infraorbital canal runs along the ventral part of the 1st bone, the orbital border of the 2nd and through the centres of the 3rd, 4th and 5th infraorbitals. The orbitosphenoids are connected to the parasphenoid via a deep, narrow septum. The anterior trigemino-facialis foramen indents the border of the prootic and there is a long lateral commissure (Fig. 5). There is a small foramen between the parasphenoid and basioccipital. 1 70 GORDON J. HOWES The jaws are long (Figs 7 & 8); the palatine process of the maxilla varies interspecifically from being low and long, to high and triangular in shape (see p. 180 & Figs 7D-F). The articulation of the dentary with the quadrate is below the centre of the orbit. The dentary is rather deep with a convex dorsal border and a backwardly sloped coronoid process. The anguloarticular has a long, almost horizontal dorsal border. The suspensorium (Fig. 9C). The hyomandibula is narrow with a straight anterior border. The dorsal border is sloped so that the anterior articulatory condyle is at a level lower than that of the posterior condyle. The ento- and metapterygoids are deeper than those of other genera. The entopterygoid is long with a slightly concave dorsal border; the metapterygoid has a narrow, triangular anterior process directed mesad. The most noticeable feature of the suspensorial elements is the length and shape of the symplectic. The bone is excessively elongate with expanded tips where it contacts the quadrate and the hyomandibula. Branchial arches (Fig. 11). The ceratobranchials bear 7-12 long gill-rakers, pharyngeal teeth are arranged in 3 rows with interspecific variation of 4.2.1, 4.3.1 and 5.3.2. The operculum has a concave upper posterior border and an attentuated lower part (Fig. 9C). Myology (Fig. 14). As in Neobola and Chelaethiops, section A, of the adductor mandibulae muscle is divided, but A.{ant lacks the aponeurotic constriction present in those two genera and extends from the quadrate to insert below the anterior cleft of the maxilla. Muscle A, post extends from the lower limb of the preoperculum to insert partly on the coronoid process of the dentary and partly, via a long tendon, into the fascia of \{ant. Other muscles are arranged as in Neobola. Pectoral girdle (Fig. 1 0) The upright part of the cleithrum is narrower than in Neobola and more closely resembles that of Chelaethiops in shape; the coracoid is deep and its posterior border forms a sharp angle with the ventral border. There is no postcleithrum. Vertebral column The 1st vertebra has a flat anterior face, and short, blunt lateral processes. The 2nd and 3rd centra are fused, bearing long, lateral processes with posteriorly curved tips. The caudal fin skeleton closely resembles that of Neobola in having a small 6th hypural and reduced spine on the fused ural centrum, and short hypurapophysis (Fig. 17C). There are no paired uroneurals. Dorsal and ventral procurrent rays are well-developed. Taxonomy Mesobola brevianalis (Boulenger), 1908, is characterized by a maxilla with a high, narrowly triangular mid-lateral (palatine) ascending process (Fig. 7F); 10 olfactory lamellae on each half rosette; absence of pectoral axial scales; 12-15 branched anal fin rays (12 in the type specimen, 14-15 in others); 9-12 gill-rakers on 1st ceratobranchial; 48-50 lateral line scales, the lateral line with a pronounced and abrupt downward curve over the pectoral fin, and another at the caudal fin base; pectoral fin reaching to origin of the pelvic; pharyngeal teeth in 3 rows, interspecific variation, 4.2.1-5.3.2; swimbladder with long anterior and posterior chambers, the posterior curving downward and terminating above the anus; vertebrae 39 or 40 (40+ 13 -14 + 20-22 + 1). Jubb (1967) gives a description, synonymy and distribution for the species. Bell-Cross (1956a) records M. brevianalis from the Kabompo river and an 'isolated' example from Fort Rosebery, off the Luapula river. Jubb (1967) refers to M. brevianalis from the Cunene river, although this species is not listed by Bell-Cross (19656) in his check-list of the fishes of that river. The species is recorded from Zambia, Zimbabwe, Transvaal and Natal. On the western side of South Africa, it occurs below the Aughrabies Falls of the Orange river (Jubb, 1967:127). A REVIEW OF THE AFRICAN NEOBOLINE CYPRINID FISHES 171 Fig. 19 Characteristics of Chelaethiops species: A, C. bibie\ B, C. elongatus; C, C. congicus; D, C. rukwaensis; E, C. minutus. The arrows indicate specific features; length and shape of snout, inclination of head, prominence of frontal ridge, shape of opercular border. Mesobola bredoi (Poll), 1945, is characterized by a maxilla with a high and narrowly triangu- lar palatine process (Fig. 7E); low number of olfactory lamellae, 8 on each half rosette; 1 1-13 anal fin rays; lateral line scales 36-39; 12 long gill-rakers; pharyngeal tooth formula: 4.2.1; vertebral number 37-39 (4+12 — 13 + 19 — 21 + 1). The species is confined to Lake Albert. Mesobola moeruensis (Boulenger), 1915, is characterized by a maxilla with a low, sloped palatine process; 9 long gill-rakers; the lower posterior border of the operculum attenuated; 8 olfactory lamellae on each half rosette; 1 5 branched anal fin rays; lateral line scales 4 1 ; pharyngeal tooth formula, unknown; vertebral number 38 (4+12 + 21 + 1). All characters taken from a syntype. A sample of 10 specimens from Katanga (MRAC 85812-847) named as this species, have gill-rakers less spinose than those of the type specimen and more closely resembles M. spinifer. Mesobola moeruensis is most probably confined to Lake Mweru, and those samples from elsewhere named as this species belong to some other taxon. 172 GORDON J. HOWES 4-6 8-11 Fig. 20 Above, synapomorphy scheme of the neoboline group. Synapomorphies are, I , chan- nelled supraethmoid; 2, frontal canals at edge of the bones with extensive, laterally opening pores; 3, divided A, muscle with complex aponeurotic inclusions; 4, derived ethmovomerine morphology; 5, elongate symplectic; 6, derived jaw morphology; 7, forward shift of jaw articula- tion; 8, anterior extension of lateral ethmoid; 9, aponeurotic constriction of muscle A,an/; 10, enlarged frontal pores; 1 1, hypertrophied supraneurals; 12, divided A, aponeurosis (see text, pp. 179). Below, cladogram showing relationships of the neoboline group. Synapomorphies are, 1, divided A, muscle; 2, pectoral axial scales (secondarily lost in some taxa); 3, A, post muscle insertion divided; 4, derived ethmovomerine morphology (see text, pp. 181). A REVIEW OF THE AFRICAN NEOBOLINE CYPRINID FISHES 173 Mesobola spinifer (Bailey & Matthes), 197 1 , is characterized by a maxilla with a low, sloped palatine process (as in M. moeruensis)', 7-8 olfactory lamellae on each half rosette; 9-1 1 gill-rakers on 1st ceratobranchial; 15-16 branched anal fin rays; lateral line scales, 41-45; pharyngeal tooth formula, 4.3.2.; vertebral number 39-41 (4+13 + 21 -23 + 1). Bailey & Matthes (1971) compare M. spinifer with other species here included in Neobola and Mesobola and conclude that it is most closely related to M. moeruensis. They distinguish spinifer from moeruensis principally on the higher number of gill-rakers. However, in the syntype of M. moeruensis I count 9 rakers on the ceratobranchial, not 7-8 as given by Bailey & Matthes. There is no substantial difference in lateral line scales and branched anal fin ray counts. Bailey & Matthes (1971) also distinguish M. spinifer from M. moeruensis on body depth. The authors give a body depth range of 18-3-2 1-8% of SL for M. spinifer but give no figures for M. moeruensis. The body depth for the only examined syntype is 22-8% of the SL and the range for the Katangan specimens of 'moeruensis ' (see above) is 21 -1-24-7% (mean 23-5%). In numbers of branched fin anal fin rays (16-18) and vertebral count (total 39-40), the Katangan specimens of M 'moeruensis' are closer to M. spinifer than to the type specimen of M. moeruensis. Mesobola spinifer occurs in the Malagarasi and Ruaha drainages (see Bailey & Matthes, 1971). SPECIMENS EXAMINED. Mesobola brevianalis: 1907.4.17:90 (holotype), Mkuzi R., Zululand; 1907.5.15:5-7, Devaard R., Transvaal; 1915.6.29:15-17, Aapies R., at Petronella; 1977.6.27:299-300, Namaini Pan, Pongolo R.; 1977.6.27:307-1256, Pongolo R., below Jazini Dam, Mzinyeni Pan, N. Natal; 1982.4.13:4693-97, Sabe R., Chisumbanje, Zim- babwe; 1978.8.3:158-62, L. Chiuta; MRAC 186489-947, Chambesi R., Rhodesia. Mesobola bredoi: 1969.3.18:1-15, Lake Albert. Mesobola moeruensis, 1920.5.26:84 (syntype), Lake Mweru; Mesobola 'moeruensis ' MRAC 85812-847, Elizabeth ville, Katanga. Mesobola spinifer. 1970.3.10:1 (holotype), Kazima, Malagarasi watershed; 1970.3.10:2-9; 10-11 (paratypes), same locality as holotype. RASTRINEOBOLA Fowler, 1936 Rastrineobola argentea (Pellegrin), 1904 is the type and only species of the genus. It is characterized by an elongate ethmoid region, short, deep jaws, long gill-rakers, 10 olfactory lamellae on each half rosette, absence of pectoral axial scale and a swimbladder with long anterior and posterior chambers, the posterior curving downward and extending to above the anus. Cranial anatomy Osteology (Figs 2, 3 & 5). The ethmoid bloc is long, narrow and deep. The lateral edges of the supraethmoid are raised forming a shallow dorsal channel; posteriorly the bone is overlapped by the frontals. The mesethmoid has an omega-shaped (ft) indentation, the lateral arms abut on long, anteriorly directed preethmoids. The vomer floors the mesethmoid indentation and its anterior border is also strongly indented. The exposed portion of the vomer, between the mesethmoid arms, is sometimes perforated (Fig. 3E). The lateral ethmoid has a marked lateral protrusion (cf. the truncated condition in Neobola and Chelaethiops) from the frontal margin and is sloped anteroventrally. The frontal canal in common with other neobolines lies on the margin of the bone and has 3 dorsally directed sensory pores; the 4th pore opens laterally. The nasal is broad, lacking dorsal pores. The morphology of the neurocranium is essentially like that of Mesobola except in having a flatter cranial surface, smaller sphenotic, straighter edged pterotic, longer basioccipital and parietals. The supraoccipital is also shorter than in Mesobola and bears a slight crest (Fig. 5). There is a ventral opening between the basioccipital and parasphenoid. 174 GORDON J. HOWES Fig. 21 Distribution of Neobola and Chelaethiops. The single dashed line indicates the probable limit of Neobola distribution; the area enclosed by the dotted line is devoid of any records of Neobola species. The Nile marks the eastern boundary of Chelaethiops, that in the west is unknown and is indicated as the Volta, the northern extent is marked by a double-dashed line. The lakes where Chelaethiops species occur are shown in solid black. The infraorbitals (Fig. 6D) are all of approximately the same depth, the 1st is broadly triangular with the sensory canal passing through its centre, the 2nd bears an antero-dorsal process and the canal, as in the 3rd, 4th and 5th, runs through the centre of the bone. The 5th infraorbital is separated from the wide supraorbital. The jaws (Figs 7 & 8) are shorter than in other neobolines. The maxilla bears a high mid- lateral (palatine) ascending process with a convex (cf. concave) posterior border (Fig. 7G). A REVIEW OF THE AFRICAN NEOBOLINE CYPRINID FISHES 175 Fig. 22 Distribution ofMesobola, Rastrineobola, Leptocypris and Engraulicypris. The supposed limits of Mesobola species are indicated by hatched lines, the lakes where they occur by solid black. Exclamation signs mark the western records. The Nile is the eastern boundary of Leptocypris, while the dotted lines indicate the supposed northern and southern boundaries. Rastrineobola occurs only in Lakes Victoria and Kioga, and Engraulicypris in Lake Malawi. Anteriorly the maxilla is not bifurcated as in the other genera where the arms of the bifurca- tion are of equal length. Instead the medial arm is the longer and is directed ventrally. The ventral border of the maxilla is concave. The premaxilla has a short anterior ascending process. The dentary is deep with a convex dorsal border and a high, rounded coronoid process, halfway along the bone. The anguloarticular is long with a horizontal dorsal surface (Fig. 8D). Suspensorium (Fig. 9D). The hyomandibula is broadly triangular, its anterior edge vertical; the articulatory condyles are widely separated, the intervening border of the bone being gently concave. The metapterygoid bears a well-developed, narrow process which is directed medially and overlies the posterior margin of the entopterygoid. The ectopterygoid 1 76 GORDON J. HOWES is a short, deep bone (cf. long and slender in other neobolines). As in Mesobola, the symplectic is elongate with expanded tips and extends the length of the metapterygoid. Branchial arches. The ceratobranchials are short with 12 or 13 long gill-rakers; epi- branchials with 3 or 4; pharyngeal teeth long and recurved, in 3 rows, 4.3.2. Infrapharyngo- branchial 2 is short and almost square, compared with a longer, triangular element in other neobolines. The operculum (Fig. 9D) is long with a sloped and rounded posterior border. Myology (Fig. 14). As in Mesobola, the adductor mandibulae section A, is divided, the anterior part, A, ant, originating from the quadrate and inserting musculosly on to the anterior part of the maxilla; the posterior part, Atpost, originates from the preoperculum to insert in part via a long tendon into the medial fascia of A, ant and in part on to the dorsal rim of the coronoid process of the dentary. Muscle A2 originates from preoperculum and is divided by the levator arcus palatini. The muscle inserts via a long tendon on to the medial face of the anguloarticular; a small segment of muscle runs from the tendon of A2 to the medial rim of the coronoid process, this segment is taken to represent Aw (Fig. 14B). Pectoral girdle (Fig. 10D) The upright part of the cleithrum is short, the tip of the horizontal limb extending to a point below the parasphenoid ascending process. There is a longer upright cleithral lamina than in other neobolines. The coracoid is deeper posteriorly than in any other neoboline and has an irregularly shaped posterior border; there is also a slight fretting of the anteroventral margin in some specimens. Vertebral column The 1st centrum has a flat articulatory face and long lateral processes; the 2nd and 3rd centra are fused with long, recurved lateral processes. There are 38-40 vertebrae (4+16+17 + 19 + 1). The caudal skeleton (Fig. 17D) more closely resembles that ofChelaethiops in having a long 6th hypural; it differs from other neobolines in possessing a bifurcate or trifurcate spine on the fused ural centrum, a bowed epural and anteriorly expanded neural spine on the 1st preural centrum. There is a long hypurapophysis. Taxonomy There is a considerable variation in lateral line scale counts in the several samples examined. Also, there is a higher range of branched anal fin rays in a sample from Lake Kioga than in those from Lake Victoria. The compared samples are, however, small and it is possible that there may be a greater overlap of minimum values than those indicated: LOCALITY N SL RANGE (MM) SCALES IN LL. MEAN ANAL RAYS Mwanza 51 49-0-70-0 42-56 50 12-14 Entebbe 20 38-5-78-7 44-56 49 12-14 Lake Kioga 13 38-0-52-6 40-54 49 14-16 In the sample from Entebbe the largest specimen examined, 78-7 mm SL, has 51 scales compared with 56 in a specimen 52-5 mm SL. SPECIMENS EXAMINED. L. Victoria: 1905.2.28:2-8 (syntypes), Kavirondo Bay; 1906.5.30:146-51, Bunjako; 1908.10.19:8, Sesse Island; 1909.7.27:11, Kavirondo Bay; 1909.11.15:20, Kizumu Bay; 1964.2.20:12-18, Entebbe; 1982.9.24:1-10; 41-50, Entebbe; 1982.9.24:31-40, Katebo; 1982.9.24:11-20, Mwanza; 50 specimens measured at Mwanza but not preserved. Lake Kioga: 1929.4.16:21-22; 1939.3.8:1-10; 1982.9.24:21-30; 1982.9.24:51-60. A REVIEW OF THE AFRICAN NEOBOLINE CYPRINID FISHES 177 Asia V .Q-Za-G-Zi. NN-Za-G-Zi. Za-G. N-Ec-Zi Fig. 23 Reduced area cladogram of African bariliines and neobolines. a = Opsaridium + Asian bariliines; b = Raiamas; c = Leptocypris + Engraulicypris; d = Chelaethiops; e = other neoboline genera. EC = East Coast; G = Guinean; N = Nilo-Sudanian; Q = Quanzan; Za = Zairean; Zi = Zambesian. Phylogenetic relationships of the neoboline group Monophyly The assumed monophyly of Neobola, Chelathiops, Mesobola and Rastrineobola is based on the following synapomorphies: 1. Ethmovomerine morphology. The supraethmoid in all neobolines is narrow and dorsally channelled (see p. 1 52). The plesiomorphic cyprinid supraethmoid is broad and flat (see Howes, 1981). Within the neobolines, the most extreme form of supraethmoid channelling occurs in Mesobola (Fig. 3D). The width and development of the dorsal channel varies interspecifically in Chelaethiops (Figs 3B & C). In Neobola the dorsal channel is least developed. Howes (1979) argues that the overlap of the supraethmoid by the frontals is a synapomor- phy for chelines and possibly Chelaethiops (Engraulicypris of Howes, 1979). However, further study shows a similar condition in several other cyprinid taxa, usually in juveniles, where it is an ontogenetic precursor to the sutured joint of the adult. Where it survives in adults, it must be looked upon as ontogenetic retention and therefore plesiomorphic. 2. The frontal canals situated at the edge of the bone. Frontal canals occupy the border of the bone only in neobolines. The sensory canal openings are few (3-5), of large aperture, and in Neobola and Chelaethiops alone all open laterally (Fig. 2). The common condition in cyprinids is for the frontal sensory canals to be embedded within the bone, distant from its margin and with small dorsal pores. See Howes, 1981 : 17-18 for an account of the plesiomorphic cyprinid frontal. 3. A derived jaw muscle morphology. In all neobolines, the adductor mandibulae A, section is divided into anterior (A,a«0 and posterior (A^post) segments. A^nt originates from the quadrate and inserts on the maxilla; A,/?as/ originates from the lower part of the preoperculum and inserts variously into the lower jaw bone and rictal tissue (see below). 178 GORDON J. HOWES Zairean L. Rukwa LTang'ka Zairean Nilotic congicus rukwaensis minutus elongatus bibie Fig. 24 Cladogram of Chelaethiops species. Synapomorphies are 1, A.{post inserts entirely into the dorsal aponeurosis of A.lant, well-developed frontal canal openings; 2, ethmoid region elongated, dentary with convex margin, articular face of 1st centrum prominently rounded, maxillary valve of increased width; 3, highly developed aponeurotic connection between A{ant and A.tpost, elongate axial scales. The plesiomorphic cyprinid condition of the adductor mandibulae A, is an undivided element with a simple insertion on the maxilla (see Takahasi, 1925; Howes, 1982 : 31 1). The only other cyprinids known to possess a divided A, are certain bariliines (see below). Another, assumed, synapomorphy is the loss of the single pair of uroneurals in the caudal skeleton. All neoboline genera lack these elements that are present in all other cyprinid taxa examined or documented; the only known exception is Engraulicypris sardella. It would appear that several sets of uroneurals is a primitive teleost character (see Patterson, 1968). Amongst other otophysans, characoids have up to 3 pairs, and in siluriformes, they are consolidated with other elements of the caudal skeleton (see Fink & Fink, 1981). In those cyprinid genera considered to be plesiomorphic (Opsariichthys, Opsaridium, Barilius), the paired uroneurals are elongate bones extending to the base of the 5th hypural. In the majority of cyprinid taxa the bones are short, triangular or curved elements. On grounds of commonality amongst cyprinids, the loss of paired uroneurals in neobolines should probably be looked upon as a synapomorphy. Relationships of neoboline genera A series of further derived states of those synapomorphies listed above together with other derived features relate the neoboline genera in the following pattern: Neobola and Chelaethiops share: 1. A derived form of the lateral ethmoid. The dorsal part of the outer wall of the lateral ethmoid is produced anteriorly as a 'shield' covering the posterior portion of the olfactory A REVIEW OF THE AFRICAN NEOBOLINE CYPRINID FISHES 179 Asia N-Ec-Zi-Q Fig. 25 Reduced area cladograms of (above) Vari's (1979) scheme of distichodontid and cithari- nid relationships; a-f indicate a = Citharinidae, b = Distichodontidae, Vari's (fig. 47) genera D-E; c — genera G-I; d = genera J-Q; e = Paradistichodus; f=Xenocharax; (below) Parenti's (1981) scheme for Old-World aplocheiloids, another aplocheiloids; b = Aphyosemion. Area abbrevia- tions are as in Fig. 23. organ (see p. 152 & Fig. 4). Elsewhere amongst cyprinids this condition is known only in Salmostoma, but here it is the entire lateral ethmoid wall, rather than just the dorsal part, that extends forward. This is probably a parallelism with neobolines in view of the opposing synapomorphy scheme of Salmostoma among chelines (see Howes, 1979; 1983). 2. An aponeurotic constriction of muscle segment A{ant. This configuration of the muscle may be a mechanical device allowing A, to circumvent the eye. Otten (1981) presents a diagram hypothesising the functional advantage of A, having an aponeurotic sheet in just such a position as occurs in Neobola and Chelaethiops. In Chelaethiops there is a further derived state whereby the aponeurosis is bifurcated. 3. Enlarged frontal pores opening laterally. Those in Chelaethiops are all laterally directed and the frontal canal is raised above the level of the supraorbital so forming a pronounced ridge in some species (Fig. 19). 4. Hypertrophied supraneurals. The supraneurals are plate-like elements. In Chela- ethiops the bones are larger than in Neobola, their most extreme form is encountered in Chelaethiops minutus where the anterior 3 or 4 are articulated in a similar fashion to those in some chelines (see Howes, 1979). The presence in Chelaethiops of a condylar face on the 1st centrum suggests the facility of cranial elevation, again as in chelines. 180 GORDON J. HOWES Mesobola and Rastrineobola share: 1. A derived ethmovomerine morphology. Mesobola and Rastrineobola both have a deep mesethmoid indentation which in Mesobola is V-shaped and in Rastrineobola omega-shaped (see p. 173). In both genera the arms of the mesethmoid abut on long, anteriorly directed preethmoids which, in effect, extend the depth of the indentation. The mesethmoid inden- tation is floored by the vomer which in both genera is often perforated, the foramen may be small or may coincide with the rim of the overlying mesethmoid. There are also dorsal protruberances of the vomer on either side of the medial notch (Figs 3D & E). 2. An elongate symplectic. The cyprinid symplectic is usually a near-triangular bone (the base of the triangle abutting on the lower tip of the hyomandibula). This form of symplectic occurs in Neobola and Chelaethiops (Figs 9A & B), but in Mesobola and Rastrineobola, the bone is elongate and horizontal rather than forming the usual steep angle (Figs 9C & D). The tips of the symplectic are spatulate. 3. A derived jaw morphology. The maxilla in Mesobola and Rastrineobola has a tall mid-lateral (palatine) ascending process. In this respect the form of the maxilla is near to that postulated for the presumed plesiomorphic cyprinid type (see Howes, 1981: 28). That this is a secondarily derived form in Mesobola and Rastrineobola is suggested by the following observations: (i) In Chelaethiops and Neobola, the sister-taxa to Mesobola and Rastrineobola, the upper jaw bones are long and slender; the maxilla having a long, low palatine process. This is the characteristic morphology for all bariliines and chelines, considered as the close relatives of the neobolines (see below); (ii) There is a marked concavity of the posterior border of the palatine process, and an acute lateral curvature. These features are absent in plesiomorphic cyprinid maxillae, but present in bariliine and cheline taxa; (iii) The presence of a transitional sequence in Mesobola species. In M. spinifer the palatine process is long and low, but even so is still higher than that of Neobola, Chelaeth- iops and bariliines. In Mesobola brevianalis and M. moeruensis, the process is taller and more triangular, and in Rastrineobola, the palatine process is best developed (Figs 9D-G). In Mesobola and Rastrineobola the dentary is deep with a convex dorsal border, a feature particularly marked in Rastrineobola (Fig. 8E). In both genera the anguloarticular is long with a horizontal dorsal surface. The usual cyprinid condition, exemplified by Neobola and Chelaethiops (Figs 8A & C), is for the anguloarticular to be short and deep with a sloped or concave dorsal border. This condition also appears general for teleosts (see Nelson, 1972). In Rastrineobola the jaw articulation is situated so far forward that it lies below the anterior half of the orbit (cf. below, or posterior to the centre in other neobolines). Only in Engraulicypris, amongst bariliines, is the jaw articulation so far forward (see below). A feature also shared by Mesobola and Rastrineobola is the elongation and ventral curvature of the posterior chamber of the swimbladder. Comparative out-group data is too sparse to recognize this feature as a synapomorphy, but a brief survey of cyprinids suggest the condition is, at least, unusual. In summary Neobola and Chelaethiops, and Mesobola and Rastrineobola form paired sister groups related on a synapomorphy scheme involving ethmovomerine, frontal, jaw bone, suspensorial and jaw muscle morphology (Fig. 20). The relationships of the neobolines are with certain bariliines. Relationships of the neoboline group The neobolines share with the bariliine genera Engraulicypris, Leptocypris, Raiamas and Opsaridium the following synapomorphies: 1. A divided adductor mandibulae A, muscle. In Leptocypris and Engraulicypris the fibres of A,/?atf insert partly into the fascia of Auant and partly into the rictal tissue as in the neobolines. In Raiamas, Opsaridium and South-East Asian Barilius, all the fibres of A{post insert into the fascia of A,anf (Fig. 15), a condition taken to represent the predecessor of the more complex arrangement in the neobolines, Leptocypris and Engraulicypris. A REVIEW OF THE AFRICANi NEOBOLINE CYPRINID FISHES 1 8 1 2. Pectoral and pelvic axial scales. Howes (1983) proposes a fundamental dichotomy of the bariliine group based on axial scale type (viz: elongate 'typical' scales versus modified scales in the form of fleshy lobes). Amongst the neobolines, axillary scales are reduced in some species of Mesobola and are lacking in Rastrineobola. In bariliines, axillary scales are absent in Engraulicypris. From the widespread distribution of axial scales amongst neoboline and bariliine genera, it is assumed that their absence in those above cited taxa are indepen- dent losses. Where scales are reduced or absent, this condition is confined to lacustrine species. 3. Derived ethmovomerine morphology. Amongst the bariliines, only Leptocypris and Engraulicypris have an ethmovomerine architecture approaching that of the bariliines. In neither genus is there a dorsal channelling of the supraethmoid as in neoboline genera, but in Leptocypris there is a slight lateral elevation of the bone (Fig. 3F). In some Leptocypris species and in Engraulicypris there is an omega-shaped ethmoid indentation, anterior elongation and perforation of the vomer (see Howes, 1980; 1983). These are features shared with the neobolines Mesobola and Rastrineobola. This pattern of ethmovomerine mor- phology is disjunct throughout Leptocypris, Engraulicypris and neoboline species and may, therefore, be a case of homoplasy. Likewise, a somewhat channelled supraethmoid occurs in a group of South-East Asian and Indian Barilius species. In these taxa, however, the ethmoid bloc is depressed, rather than laterally compressed as in the neobolines, and other characters such as tubercle pattern and palatine morphology suggest that they form a lineage distinct from that of the neobolines, Leptocypris and Engraulicypris. One other feature to be considered is the absence in Engraulicypris of paired uroneurals. It was noted above (p. 178) that these elements are lacking in all neoboline genera, and their loss might be construed as a synapomorphy. If this is the case, then Engraulicypris would appear to be the sister group to the neobolines, thence to Leptocypris. In the synapomorphy scheme presented here (Fig. 20), the caudal fin character has been reserved until such time as more complete comparative data on its distribution are available. In summary, the neobolines form the sister group to that comprising Leptocypris and Engraulicypris, which in turn is the sister group to Raiamas. This combined assemblage comprises the sister group to the South-East Asian bariliines (possibly including some Indian 'Barilius' and Opsaridium. Neoboline distribution and its biogeographic implications Neoboline taxa have a wide distribution in Africa, embracing the 'ichthyofaunal provinces' Roberts (1975) termed Nilo-Sudanian, East Coast, Guinean, Zambesian and Zairean. Neobola species occur in eastern flowing rivers of Ethiopia and Somalia, northern Kenya and in Lake Turkana (Fig. 21). Chelaethiops, the derived sister group of Neobola, is extensively distributed throughout the Zairean, Guinean and Nilo-Sudanian provinces and also includes Lakes Tanganyika and Rukwa (Fig. 21). Mesobola is present in Lake Albert, the Malagarasi, Rufiji and Zambesi systems, and eastern flowing rivers as far south as Natal (Fig. 22). The genus is also known from the Orange River on the western side of Africa, although there is some doubt about the record from the Cunene River; see p. 170. Rastrineobola occurs only in the Lake Victoria basin-including Lake Kioga (Fig. 22). The most notable feature of neoboline distribution is the geographical division of the genera by the eastern Rift system, so that Chelaethiops is the only genus with a continuous westward extension. When depicted as an area cladogram, the branching sequence of neoboline taxa shows a sister-group relationship between Nilotic and Zairean areas and East-Coast and Zambesian (Fig. 23). The sister group of the neobolines, Leptocypris + Engraulicypris, display a congruent area pattern, with Lake Malawi forming the sister-area to the Nilo-Zairean (Fig. 22). Within Leptocypris, those species considered as derived, weynsii, lujae and modestus, are Zairean, again a pattern in agreement with that of the derived neoboline, i.e. Chelaethiops, species; see below. 1 82 GORDON J. HOWES On the eastern side of the Rift, the area relationship signified by the sister group Mesobola + Rastrineobola, is between East Coast-Zambesian + Lake Victoria basin. The distributional pattern of these sister-group pairs is readily explicable on vicariance events involving, in the case of Neobola and Chelaethiops, the formation of the Rift system, and, in that of Mesobola and Rastrineobola, the isolation of the latter in the Victoria basin. Roberts (1975) includes Lake Victoria in the East-Coast province, and this geographical relationship is certainly borne out by the neoboline phylogeny. Chelaethiops also occurs in both the Malagarasi and Lake Rukwa. This distribution can either be interpreted as a subsequent dispersal from Lake Tanganyika or, as representing the result of an interrupted (vicariant) distribution due to the topographical evolution of that lake; see Banister & Clarke, 1980. The synapomorphy scheme of Chelaethiops species (Fig. 24) makes the latter interpretation more economical. Chelaethiops congicus (Zairean) has least derived features and together with C. rukwaensis (Lake Rukwa and Malagarasi) and C. minutus (Lake Tanganyika) forms the sister group to the derived species C. bibie (Nilo- Sudanian) and C. elongatus (Zaire-Guinean); see Fig. 24 for character summary. Mesobola occurs sympatrically with Chelaethiops in the Malagarasi drainage and the Lualaba, it is also disjunct in distribution, being found in the Orange River below the Aughrabies Falls on the western side of the continent. There is a single, unconfirmed record of its presence in the Cunene (see p. 170). Roberts (1975 : 309) supposed this distribution to be the result of dispersal through South-West Africa. It would be more parsimonious to recognise in this pattern the fragmentation of a formerly uninterrupted distribution. In terms of historical biogeography, the neobolines seem uninformative. This is due partly to the genera resolving only into two-area cladograms, and the mostly unresolved interrelationships of their contained species. Indeed, for Mesobola, a species cladogram cannot be constructed as the species are presently recognised only on the basis of mainly, meristic differences. For Neobola, of the three species, one, N. fluviatilis, is possibly a population variant of another, N. bottegi, and the third, N. stellae is a lacustrine endemic. The other reason is that there are few cladograms of African freshwater fishes which can be used in broader comparison with neobolines. Only three phylogenies can be considered as a basis for constructing area cladograms; those of Vari (1979), Parenti (1981) and Howes (1983). Vari (1979) studied the characoid families Citharinidae and Distichodontidae, recognising them as sister groups. Both families are widespread throughout Nilo-Sudanian, Guinean, Zairean, East-Coast and Zambesian provinces. Within the Distichodontidae, most taxa, which according to Vari's scheme of interrelationships are the derived ones, occur in Zairean and lower Guinean regions. As a simplified area cladogram, the distichodontid pattern is one of repeated dichotomy between Nilotic and Zairean-Guinean regions (Fig. 25). Parenti's (1981) geographical analysis of African aplocheiloid cyprinodonts demonstrates a sister-group relationship between Zairean-Guinean (derived forms) and Nilo-Sudanian, Zambesian-East Coast and Quanzan provinces (Fig. 25). The bariliine relationships presented by Howes (1980; 1983) show both Opsaridium and Raiamas with Asiatic relatives and representatives. Opsaridium has as its sister group the South-east Asian Barilius (and possibly some Indian species, see above, p. 180), and Raiamas is represented also in India and Burma. At this high level of universality, the pattern of bariliine distribution is virtually concordant with that presented by Parenti (1981) for Old World aplocheiloids (cf. fig. 23 in Parenti with fig. 47 in Howes, 1980). Interestingly, all these patterns - aplocheiloids, characoids, bariliines and neobolines -exclude the Cape Province (see below). African ichthyogeography and biotic subdivision Too few data are available to form any refined picture of African ichthyogeographical history. Those that are available suggest a widespread, plesiomorphic fauna disrupted by A REVIEW OF THE AFRICAN NEOBOLINE CYPRINID FISHES 183 a Zairean-Nilotic break with a subsequent (or even contemporaneous) Zairean-Guinean fragmentation. To date, studies of African ichthyogeography have been little more than catalogues of endemism and scenarios of dispersals (see Greenwood, 1983 for discussion). The various hypotheses proposed for African freshwater fish distribution are ad hoc assumptions based on the supposedly known histories of past drainage patterns. An example of this approach is given by Livingstone et al (1982) who account for fish distribution patterns by invoking dispersal from refugia, competition and 'powers of dispersal'. These authors find the ' . . . pattern of faunal similarities surprising . . . ' and rather than accepting that this pattern reflects a previously uniformly distributed fauna (Greenwood, 1983) would prefer to see in it a reflection of '. . . more recent faunal exchange'. The same problems have beset discussions of the biogeography of other African faunas and, as for fishes, ecological parameters are seen as the determinate factors in shaping dis- tributional pattern. Two recent examples are works dealing with molluscs (Brown, 1978) and birds (Crowe & Crowe, 1982). According to Brown ' . . . many distributional patterns seem to be dependent mainly on existing ecological conditions'. He does, however, draw attention to the taxonomic relationships between southern east African and Malagasian molluscs. Crowe & Crowe correlate their avian zones with vegetation types, although admitting that the distributional patterns of passerine and non-passerine birds cannot be explained solely on environmental factors; there is no reference to the possible phylogenetic relationships of the respective bird groups. Similarly, the reasoning behind accounts for mammal distribution has been entirely eco- logical. Indeed, Rautenbach (1978) emphasises that faunal interrelationships should be inter- preted from an ecological point of view. No consideration has been given to a vicariographic approach. There is a cladistic analysis of African bufonids by Grandison (1981) which shows that the plesiomorphic lineage, represented by Capensibufo, is restricted to Cape Province. A reduced area cladogram of the other bufonid taxa also reveals repeated east-west dichoto- mies. Interestingly, Grandison remarks of Capensibufo that it may ' . . . have had an austral origin'. It was noted above that some groups of cyprinids, characoids and cyprinodonts are congru- ent in their distribution in being absentees from the Cape Province. The relationships of those freshwater fishes endemic to the Cape are at present largely unknown. According to McDowall (1973) the South African galaxiid, Galaxias zebratus, may have a close phylo- genetic relationship with Brachygalaxias bullocki in Chile, although McDowall would prefer to recognize any affinity between the two as convergence. Reid (pers. comm.) has pointed out that the Labeo umbratus group of the Cape may be more closely related to Asian than to other African Labeo species. The ichthyofaunal peculiarities of the Cape are paralleled by its flora. Miller (1982) in reviewing bryophyte distribution concludes that the Cape was ' ... an island which was left behind and later caught up with the primary continental block It is clear that if the phylogenetic relationships of the endemic Cape fishes were resolved, a general biogeographic pattern for this region would begin to emerge. This example serves to underline Greenwood's (1983) thesis that only by resolving the phylogenetic relation- ships of African freshwater fishes will any understanding be gained of their present-day distributional patterns. Acknowledgements The manuscript benefited greatly from the criticisms, advice and information offered by Drs P. H. Greenwood, K. E. Banister and G. M. Reid My sincere thanks are due to Bernice Brewster, Margaret Clarke, Debbie Green and Adrian Penrose for assisting with morphometric and other data, to Tony and Jane Hopson for supplying additional 184 GORDON J. HOWES data on Lake Turkana specimens, to Dr Lynne Parent! for informative discussions on biogeography, and to Dr Thys van den Audenaerde, Luc de Vos, Dr Donn Rosen and Dr Richard Vari for the loan of specimens. I owe special thanks to Drs Martien van Oijen and Frans and Els Witte for their warm hospitality and assistance in securing specimens at Mwanza. References Bailey, R. G. & Matthes, H. 1971. A new species of Engraulicypris (Cyprinidae) from Tanzania, East Africa. Revue de Zoologie et de Botanique Africaine 83(1-4): 79-83. Banister, K. E. & Clarke, M, A. 1980. A revision of the large Barbus (Pisces, Cyprinidae) of Lake Malawi with a reconstruction of the history of the southern African Rift Valley lakes. Journal of Natural History 14: 483-542. Bell-Cross, G. 1965a. Additions and amendments to the check list of the fishes of Zambia. Occasional Papers, Game and Fisheries Department, Zambia (3): 29-43. \965b. The distribution of fishes in Central Africa. Fisheries Research Bulletin, Zambia 4: 3-20. Blache, & Miton, F. 1960. Sur le statut de quel ques especes de poissons du bassin du Tchad et du bassin adjacent du Mayo Kebbi. Bulletin du Museum National de Histoire Naturelle, Paris 32(5): 395-400. Brichard, P. 1978. Fishes of Lake Tanganyika: 448 pp. TFH Publications, New Jersey. Brown, D. S. 1978. Freshwater Molluscs. In: Biogeography and ecology of Southern Africa (Ed. Werger, M. J. A.). Junk, The Hague: 1 155-1 180. Crowe, T. M. & Crowe, A. A. 1982. Patterns of distribution, diversity and endemism in Afrotropical birds. Journal of the Zoological Society of London 198: 417-442. Daget, J. 1954. Les poissons du Niger superieur. Memoires de I'lnstitut Francais d'Afrique Noire 36: 1-391. Fink, S. V. & Fink, W. L. 1981. Interrelationships of the ostariophysan fishes (Teleostei). Zoological Journal of the Linnean Society of London 72 (4): 297-353. Fryer, G. & lies, T. D. 1972. The cic hi id fishes of the Great Lakes of Africa. Their biology and evolution. 641 pp. Oliver & Boyd, Edinburgh. Grandison, A. G. C. 1981. Morphology and phylogenetic position of the West African Didynamipus sjoestedi Andersson, 1903 (Anura, Bufonidae). Monitore Zoologico Italiano NS Suppl. 15 (11): 167-215. Greenwood, P. H. 1983. The zoogeography of African freshwater fishes: Bioaccountancy or biogeogra- phy?: 79-199. In: Evolution in time and space: the emergence of the biosphere (Eds Sims, R. W., Price, J. H. & Whalley, P. E. S.). Systematics Association special volume No. 23. Academic Press, London and New York. Howes, G. J. 1978. The anatomy and relationships of the cyprinid fish Luciobrama macrocephalus (Lacepede). Bulletin of the British Museum of Natural History (Zoology) 34 (1): 1-64. 1979. Notes on the anatomy, phylogeny and classification of Macrochirichthys macrochirus (Valenciennes) 1844 with comments on the Cultrinae (Pisces, Cyprinidae). Bulletin of the British Museum of Natural History (Zoology) 36 (3): 147-200. 1980. The anatomy, phylogeny and classification of the bariliine cyprinid fishes. Bulletin of the British Museum of Natural History (Zoology) 37 (3): 129-198. 1981. Anatomy and phylogeny of the Chinese Major Carps Ctenopharyngodon Steind., 1866 and Hypopthalmichthys Blkr., 1860. Bulletin of the British Museum of Natural History (Zoology) 41 (1): 1-52. 1982. Anatomy and evolution of the jaws in the semiplotine carps with a review of the genus Cyprinion Heckel, 1843 (Teleosti: Cyprinidae). Bulletin of the British Museum of Natural History (Zoology) 42 (4): 299-335. 1983. Additional notes on bariliine cyprinid fishes. Bulletin of the British Museum of Natural History (Zoology), 45 (2): 95-101 . Jubb, R. A. 1967. Freshwater fishes of southern Africa. 248 pp. Balkema, Cape Town and London. Livingstone, D. A., Rowland, M. & Bailey, P. E. 1982. On the size of African riverine fish faunas. American Zoologist 22: 361-369. Lekander, B. 1949. The sensory line system and the canal bones in the head of some ostariophysi. Ada Zoologie a 30: 1-131. Lowe-McConnell, R. H. 1975. Fish communities in tropical freshwaters. 337 pp. Longmans, London. A REVIEW OF THE AFRICAN NEOBOLINE CYPRINID FISHES 185 McDowall, R. M. 1973. The status of the South American galaxiid (Pisces, Galaxiidae). Annals of the Cape Provincial Museums, Natural History 9 (5): 91-101. Miller, H. A. 1982. Bryophyte evolution and geography. Biological Journal of the Linnean Society of London 18 (2): 145-196. Nelson, G. J. 1972. Relationships of clupeomorphs, with remarks on the structure of the lower jaw in fishes. In: Interrelationships of fishes (Eds Greenwood, P. H., Miles, R. S. & Patterson, C). Academic Press, London and New York: 333-349. Okedi, J. Y. O. 1979. The Engraulicypris 'Dagaa' fishery of Lake Victoria: A pilot study of the Ukerewe Island-complex 'Dagaa' fishing industry. Appendix A. In: Annual Report 1978/79 Fresh- water Fisheries Research Institute, Ministry of Natural Resources and Tourism, Tanzania: 25-75. Often, E. 1981. Vision during growth of a generalized Haplochromis species: H. elegans Trewavas 1933 (Pisces, Cichlidae). Netherlands Journal of 'Zoology 31 (4): 650-700. Parenti, L. 1981. A phylogenetic and biogeographic analysis of cyprinodontiform fishes (Teleostei, Atherinomorpha). Bulletin of the American Museum of Natural History 168, Art. 4: 341-557. Patterson, C. 1968. The caudal skeleton in Lower Liassic pholidophorid fishes. Bulletin of the British Museum of Natural History (Geology) 16 (5): 201-239. 1977. Cartilage bones, dermal bones and membrane bones, or the exoskeleton versus the endo- skeleton. In: Problems in Vertebrate Evolution (Eds Andrews, S. M., Miles, R. S., Walker, A. D.) Linnean Society Symposia, series (4): 77-120. Poll, M. 1945. Descriptions de cinq especes nouvelles de Cyprinidae du Congo beige apportenant aux genres Barbus et Engraulicypris. Revue de Zoologie et de Botaniques Africaine 38 (3-4): 298-3 1 1 . 1948. Poissons recuellis au Katanga par H. J. Bredo. Bulletin du Musee Royal d'Histoire Naturelle de Belgique 24 (21): 1-24. 1953. Poissons non Cichlidae. Explorations Hydrobiologique du lac Tanganyika 3 (5 A): 1-251. Rautenbach, I. L. 1978. A numerical re-appraisal of the southern African biotic zones. In: Ecology and taxonomy of African small mammals (Ed. Schlitter, D. A.). Bulletin of Carnegie Museum of Natural History. No. 6: 175-187. Ricardo, C. K. 1939. The fishes of Lake Rukwa. Journal of the Linnean Society of London, Zoology, 40 (257): 625-657. Roberts, T. R. 1975. Geographical distribution of African freshwater fishes. Zoological Journal of the Linnean Society of London 57: 249-319. Takahasi, N. 1925. On the homology of the cranial muscles of the cypriniform fishes. Journal of Morphology 40: 1-109. van Oijen, M. J. P. 1982. Ecological differentiation among the piscivorous haplochromine cichlids of Lake Victoria (East Africa). Netherlands Journal of Zoology 32 (3): 336-363. Vari, R. P. 1979. Anatomy, relationships and classification of the families Citharinidae and Disticho- dontidae (Pisces, Characoidea). Bulletin of the British Museum of Natural History (Zoology) 36 (5): 261-344. Whitehead, P. J. P. 1962. Two new river fishes from eastern Kenya. Annali del Museo civico Storia Naturale Giacomo Doria 73: 98-108. Manuscript accepted for publication 6 December 1983 British Museum (Natural History) Tilapiine fishes of the genera Sarotherodon, Oreochromis and Danakilia Dr Ethelwynn Trewavas The tilapias are cichlid fishes of Africa and the Levant that have become the subjects of fish-farming throughout the warm countries of the world. This book described 41 recognized species in which one or both parents carry the eggs and embryos in the mouth for safety. Substrate-spawning species, of the now restricted genus Tilapia, are not treated here. Three genera of the mouth-brooding species are included though in one of them, Danakilia, the single species is too small to warrant farming. The other two, Sarotherodon, with nine species, and Oreochromis, with thirty-one, are distinguished primarily by their breeding habits and their biogeography, supported by structural features. Each species is described, with its diagnostic features emphasised and illustrated, and to this is added a summary of known ecology and behaviour. Conclusions on relationships involve assessment of parallel and convergent evolution. Dr Trewavas writes with the interests of the fish culturists, as well as those of the taxonomists, very much in mind. 580pp, 188 illustrations include halftones, diagrams, maps and graphs. Extensive bibliography. Publication 1983. £50 0565008781 Publications Sales British Museum (Natural History) Cromwell Road London SW7 5 BD England Titles to be published in Volume 47 Miscellanea A review of the Mastacembeloidei, a suborder of synbranchiform teleost fishes Part II: Phylogenetic analysis. By Robert A. Travers A review of the anatomy, taxonomy, phylogeny and biogeography of the African neoboline cyprinid fishes. By Gordon J. Howes The haplochromine species (Teleostei, Cichlidae) of the Cunene and certain other Angolan rivers. By Peter Humphry Greenwood Miscellanea Anatomy and evolution of the feeding apparatus in the avian orders Coraciiformes and Piciformes. By P. J. K. Burton A revision of the spider genus Cyrba (Araneae: Salticidae) with the descriptions of a new presumptive pheromone dispersing organ. By F. R. Wanless Printed in Great Britain by Henry Ling Ltd., at the Dorset Press, Dorchester, Dorset Bulletin of the British Museum (Natural History The haplochromine species (Teleostei, Cichlidae) of the Cunene and certain other Angolan rivers Peter Humphry Greenwood Zoology series Vol47 No 4 27 September 1984 The Bulletin of the British Museum (Natural History), instituted in 1949, is issued in four scientific series, Botany, Entomology, Geology (incorporating Mineralogy) and Zoology, and an Historical series. Papers in the Bulletin are primarily the results of research carried out on the unique and ever-growing collections of the Museum, both by the scientific staff of the Museum and by specialists from elsewhere who make use of the Museum's resources. Many of the papers are works of reference that will remain indispensable for years to come. Parts are published at irregular intervals as they become ready, each is complete in itself, available separately, and individually priced. Volumes contain about 300 pages and several volumes may appear within a calendar year. Subscriptions may be placed for one or more of the series on either an Annual or Per Volume basis. Prices vary according to the contents of the individual parts. Orders and enquiries should be sent to: Publications Sales, British Museum (Natural History), Cromwell Road, London SW7 5BD, England. World List abbreviation: Bull. Br. Mus. nat. Hist. (Zool.) Trustees of the British Museum (Natural History), 1984 The Zoology Series is edited in the Museum's Department of Zoology Keeper of Zoology : Dr J. G. Sheals Editor of Bulletin : Dr C. R. Curds Assistant Editor : Mr C. G. Ogden ISBN 0565 05007 9 ISSNO°07-1498 ^Nopp ,87-239 British Museum (Natural History) Cromwell Road London SW7 5BD Issued 27 September 1984 The haplochromine species (Teleostei, Cichlidae) of the Cunene and certain other Angolan rivers *o\ f "^w /? •» Peter Humphry Greenwood Department of Zoology, British Museum (Natural History), Cromwell Road, London SW7 5BD . BRIT Contents Introduction 187 Methods and materials 188 The haplochromine species of the Cunene river 189 Thoracochromis Greenwood, 1979 189 Thoracochromis buysi (Penrith), 1970 190 Thoracochromis albolabris (Trewavas & Thys van den Audenaerde), 1969 . 197 Orthochromis Greenwood, 1954 206 Orthochromis machadoi (Poll), 1967 206 Pseudocrenilabrus Fowler, 1934 213 Pseudocrenilabrus philander (Weber), 1897 214 Serranochromis Regan, 1920 216 Serranochromis (Sargochromis) Regan, 1920 216 Serranochromis (Sargochromis) gracilis sp. nov 225 Serranochromis (Serranochromis) Regan, 1920 228 Serranochromis (Serranochromis) thumbergi (Castel.) 228 Serranochromis (Serranochromis} macrocephalus (Blgr) 229 Serranochromis (Serranochromis) angusticeps (Blgr) and S. (S.) robustus jallae(E\gr) 229 Zoogeographical considerations 229 Appendix I 233 The generic status of various Angolan haplochromine species previously referred to Haplochromis by Regan (1922), Trewavas (1973) and Bell-Cross (1975). Appendix II The generic status of Pelmatochromis welwitschi Blgr, 1898 234 Acknowledgements 238 References 238 Introduction L 'Angola est un plateau d' ou descendant de nombreux Jleuves et rivieres qui cachet encore bien des secrets Poll (1967:16) Despite Poll's (1967) extensive monograph on the fishes of Angola, and the work of Trewavas (1964, 1973) and Bell-Cross (1975) much has still to be learnt about the biology, taxonomy and zoogeography of the haplochromine cichlid species in this region of Africa (see Greenwood, 1979). A basic inventory of the species has been worked out, but many taxa are known only from the type specimen, by a limited number of type specimens or by some specimens whose locality is recorded no more precisely than 'Angola'. Above all, almost nothing is known about the phyletic relationships of the species, and hence zoogeographical conclusions based on them are correspondingly uncertain. Judging from the present-day hydrography of Angola, especially the isolated rivers which discharge directly into the Atlantic, and the numerous tributaries emptying into the Zaire system, one might expect a high degree of localized endemicity in the various rivers. In other Bull. Br. Mus. nat. Hist. (Zool.) 47(4): 1 87-239 Issued 27 September 1984 187 188 P. H. GREENWOOD words, the physical background would seem ideal for promoting vicariant speciation, a situation that is, indeed, suggested by some of the taxonomic data already available. Thus it was with considerable pleasure and interest that I accepted an invitation from Dr M. Penrith (then of the Windhoek Museum) to study a large collection of cichlid fishes from, principally, the Cunene river. The Cunene is one of the least studied Angolan rivers, and physiographically is one of the most isolated in the country. The collection has provided an opportunity to redescribe a number of species on the basis of many more specimens than were previously available, and to confirm the presence in the Cunene of species either not previously recorded from there or recorded with some uncertainty. Also, it has established that several species have an extensive distribution within the river itself, their ranges stretch- ing from or near the river mouth, almost to its northernmost tributaries. Less has been learnt about the relationships of the endemic Cunene species, and the collec- tion has underlined the still unsatisfactory taxonomic situation surrounding species of the Serranochromis subgenus Sargochromis. However, a species currently of indeterminable status 'Haplochromis' welwitschii (Blgr), whose type specimen is probably from the Cunene, can now be referred to the genus Chetia, a taxon otherwise known from the Limpopo system in South Africa (see Appendix II), and a tributary of the Zaire system (see Balon & Stewart, 1983). In terms of species numbers and morphological diversity, the Cunene seems to have a haplochromine fauna more complex than that in any other Angolan river and, indeed, more diverse than that of the Zambezi-Kafue systems. The new collection also apparently corroborates existing ideas that the Cunene fauna, on a broad zoogeographical scale, has phyletic affinities with both the Zaire and the Zambezi drainage systems (Trewavas, 1964, 1973; Bell-Cross, 1975; Roberts, 1975). However, sister-group relationships for the endemic haplochromines both within and outside the different sytems still cannot be established. Zoo- geographical problems are compounded by the fact that precise specific identification is impossible for most Cunene representatives of the Serranochromis (Sargochromis) species complex, and is unlikely to be obtained until more specimens, coupled with data on male breeding colours, are available from the different river systems within and outside Angola. Methods and materials Methods. Measurements and counts generally are those used in my other papers on haplo- chromine fishes (see Greenwood, 1981). Measurements relating to the neurocranium are those used in Greenwood (1980: 4-6); an additional measurement used here, ethmovomerine length, is taken directly from the anterior tip of the vomer to the most ventrolateral point on the lateral ethmoid bone. When the length of the ascending premaxillary process is given for whole specimens it is measured directly from the dentigerous surface of the bone, between the teeth on either side of the midline, to the distal tip of the processes (determined through the skin by moving the premaxillae gently forwards and downwards). When this length is taken from a skeletal preparation, the depth of the dentigerous arm is excluded, and the reference points are those illustrated in Greenwood (1980: 5, fig. 2), viz from the distal tip of the processes to a hori- zontal line drawn level with the upper margin of the dentigerous arm immediately posterior to the basal region of the ascending processes. Measurements of the lower pharyngeal bone are those employed by Bell-Cross (1975: 410), viz: length is taken along the median axis of the bone, from the anterior tip of its shaft to a line drawn transversely between the tips of the posterior horns. Lower pharyngeal breadth is measured directly between the outer edges of the two horns. A feature not previously mentioned in the description of haplochromine species is the anal sheath scales. My attention was drawn to these scales when examining part of the type series of Tilapia steindachneri Blgr (see p. 190). In these specimens, a distinct but shallow sheath of small, almond-shaped scales, aligned in a single row, lies between the anal fin base and the ventral row of body scales. The long axes of the scales are arranged horizontally, and the scales are either imbricate or spaced, sometimes widely spaced. CUNENE RIVER HAPLOCHROMINE SPECIES 189 It seems that anal sheath scales occur in a number of haplochromine lineages. A sheath, or at least some characteristically almond-shaped scales, has been found in species from Lakes Victoria, Tanganyika and Malawi, as well as in some fluviatile taxa. Sheath length varies intraspecifically to a considerable extent. It can be present along almost the entire base of the fin, or it may be confined to the base of the spinous part (apparently the common- est condition). Often it is represented merely by a few isolated scales. Since the scales are easily dislodged, the latter condition may be artefactual. The taxonomic value of the anal sheath, in whatever form it is present, cannot yet be assessed. Materials. All the type specimens of Angolan haplochromine species in the BM(NH) collec- tions were examined, as were other specimens identified as conspecific with these types or with type specimens of Angolan species held in other institutions. Likewise, all the BM(NH) material of Serranochromis and Pseudocrenilabrus species was studied, together with the type and other specimens of 'Haplochromis'' darlingi, a species first recorded from Angola by Poll (1967). In addition, the following material was borrowed from the Zoological Museum of Hamburg (ZMH) and the Musee Royal de 1'Afrique Centrale, Tervuren (MRAC). ZMH 4599 Haplochromis species (7 specimens) Cunene R. 4599a Haplochromis species (1 specimen), Cunene R. at Capelongo. 1300 Serranochromis angusticeps (1 specimen), Cunene R. at Capelongo. 1307 Serranochromis angusticeps (1 specimen), Cunene R. at Capelongo. 1718 Serranochromis robustus jallae Cunene R. at Miilongo Fiirt. 1719 Serranochromis thumbergi (2 specimens) Cunene R. at Capelongo. The identity of all these Serranochromis specimens has been confirmed. 1 722 Haplochromis frederici (4 specimens), Cunene R. at Capelongo. Two specimens are members of the Serranochromis (Sargochromis) giardi-codringtoni complex (see pp. 2 1 7-224 below), and two probably can be referred to S. (Sargo.) coulteri (Bell-Cross). Musee Royal de 1'Afrique Centrale (MRAC), Tervuren. MRAC 154779-780 Haplochromis welwitschii, Sanguenque Uembe Cuanaa, Angola. MRAC 66470 Haplochromis schwetzi, holotype, Cuango R., Angola. MRAC 163992; 164013-016; 164023-026; 164027-032; 164033-39 Haplochromis schwetzi, paratypes (26 specimens), Cuango R., Angola. MRAC 163981-986 Haplochromis darlingi, Lac Calundo, Angola. Other material examined is listed in the text. Regrettably, as a result of extensive reorgani- zation now been carried out in the fish collections of the Vienna Museum, it was impossible to examine the types of two Steindachner (1866) species: Chromis humilis and Chromis acuticeps. Both taxa are described, simply, as coming from Angola. Fortunately the types of both species were carefully examined by my colleagues Dr Ethelwynn Trewavas, and later by Mr G. Bell-Cross (now of the Port Elizabeth Museum, South Africa). I have been able to use the notes and recollections of both these people, to whom I am most indebted. The haplochromine species of the Cunene river THORACOCHROMIS Greenwood, 1979 Several Angolan species, currently referred to the genus Haplochromis, show the diagnostic features of Thoracochromis, viz an abrupt size change between the small scales on the chest and the larger scales on the ventrolateral aspects of the flanks, a marked anteroventral embayment of the cheek squamation (with, in some species, a narrow, horizontal naked area lying between the cheek scales and the preoperculum), and the absence of true ocelli, but not discrete spots, on the anal fin or adult males (see Greenwood, 1979: 290-292). 190 P. H. GREENWOOD The Angolan species now placed in Thoracochromis are: Haplochromis lucullae (Blgr), 1913; H. albolabris Trewavas & Thys van den Audenaerde, 1969; H. schwetzi Poll, 1967, and H. buysi M.-L. Penrith, 1970. Haplochromis lucullae was treated as a junior synonym of H. acuticeps (Steindachner, 1866) by Regan (1922: 255), but the species has been informally 'resurrected' by several recent authors, notably Trewavas (1964: 8-9, 1973: 31), Penrith (1970: 170-171) and Bell- Cross (1975: 427). Unfortunately, I have not been able to examine the holotype of H. acuticeps (see p. 1 89) but from Dr Trewavas' comments, based on detailed examination of that specimen, its separation from lucullae, at least at the species level, is justified (see also Trewavas, 1973: 31). Regrettably, neither Steindachner's (1866) original description, nor Trewavas' later reexamination of the acuticeps type specimen provide any information on the nature of the size-change at the chest-abdominal scale transition line, nor are there data on the nature of the cheek squamation. Thus it is impossible to comment on the generic assignment of ''acuticeps'. Steindachner's figure, however, suggests that the scale transition is of the Thoracochromis type. With one exception (Th. schwetzi), and unlike species of Thoracochromis from the Nile, Lake Turkana and the Zaire river system, none of the Angolan species has more than 4 or 5 upper lateral line scales each separated from the dorsal fin base by one large and one small scale. This low number is thought to represent the primitive condition, the higher number (8 or 9 scales) occurring in the other species being the derived one (Greenwood, 1979: 291). As compared with the Nilo-Zairean taxa most Angolan species have more scales in the lateral-line series and higher modal counts for this feature. Again, an exception is Th. schwetzi, whose lateral line counts are like those in the Nilo-Zairean species; interest- ingly, Th. schwetzi occurs only in the Cuango river, an Angolan affluent of the Zaire system. In all other meristic and morphometric features the Angolan Thoracochromis do not lie outside the range of variability found in other members of the genus. The significance, if any, of the differences in squamation cannot be assessed until more data are available from those Angolan species which are currently represented only by one or a few type specimens. Species previously referred to Haplochromis and which are not members of Thoraco- chromis, are discussed in Appendix I. Thoracochromis buysi (Penrith), 1970 SYNONYMY Haplochromis buysi Penrith, M.-L., 1970. Cimbebasia, ser A, 1 (7): 168-171, plate 2; fig. 1. Holotype: SM5099, a specimen 75mm standard length, from the Cunene river mouth. Paratype: SAM 25243, a specimen 6 1 mm SL from the same locality. This specimen is now damaged extensively, and was not used in the redescription of the species. It is, however, conspecific with the holotype. Tilapia steindachneri (part) Boulenger, 1913. Ann. Mag. nat. Hist. (8) 12: 483. Five of the syntypical specimens only (BMNH 1907.6.29:141-5, from the Que river). The largest specimen, 104.5 mm SL, alone is in reasonable condition. Although Boulenger (1913) did not select a holotype, he did later (1915) designate one specimen as 'Type' in the caption to a figure of that specimen. The fish in question is one of the syntypes from Mossamedes which Regan (1922) included in the species Sargochromis mellandi. Thus the inclusion of the five Que fishes in the synonymy of Thoracochromis buysi (Penrith), 1970 raises no question of nomenclatural priority for Boulenger's earlier name ' steindachneri'. Haplochromis acuticeps (part): Regan, 1922. Ann. Mag. nat. Hist. (9) 10: 255 (the syntypical specimens of T. steindachneri noted above; BMNH 1907.6.29:141-5). DESCRIPTION. Based on 46 specimens, including the holotype, 44-0-1 18-0 mm standard length. Depth of body 29-4-34-7 (M = 32-0)% of standard length, length of head 30-4-36-4 (M = 31-5)%. Dorsal head profile gently curved (almost straight in a few specimens), sloping at an angle of 35°-40° to the horizontal, the angle increasing with the fish's srze. The upper margin of CUNENE RIVER HAPLOCHROMINE SPECIES 191 Fig. 1 Thoracochromis buysi. Adult male (1984.2.6: 24). Drawn by G. J. Howes. Scale = 20 mm. the eye is coincident with, or lies immediately below the dorsal profile of the head, but never extends above it. The extent to which the curve of the profile is interrupted by the intrusion of the premaxillary ascending process varies, but is never marked and may be influenced by preservation methods. Preorbital depth 18-5-26-0 (M = 22-9)°/o of head length, showing slight positive allometry with standard length; least interorbital width 16-6-23-6 (M=19-4)% of head. Preorbital depth is generally greater than least interorbital width, but in some individuals the measure- ments are equal; in no specimen examined is the interorbital width greater than the preorbital depth (cf. Th. schwetzi where the interorbital is wider than the preorbital is deep). Snout length shows clear cut allometry with standard length. The range for the whole sample is 31-0-39-0% of head; in specimens less than 70mm SL (n=17) it is 31-0-35-3 (M = 33-l)% and in larger individuals (71-0-1 18-0 mm SL, n = 29) it is 34-5-39-0 (M = 36-7%). The snout is from 1-0-1-3 times longer than broad (modal range 1-0-1-1). Eye diameter is negatively allometric with standard length; for the whole sample it is 25-4-36-2% of head length, in fishes <70 mm SL it is 28-6-36-2 (M = 33-6)%, and in larger individuals 25-4-33-3 (M = 28-3)%. Cheek depth is 18-2-27-8 (M = 22-7)% head, and shows no obvious allometry with standard length. Caudal peduncle length is 16-2-22-0 (M= 19-0)% of standard length, and 1-3-1-9 (modal range 1-5-1-7) times its depth. Mouth horizontal or almost so, the lips slightly but noticeably thickened, the jaws equal anteriorly. The posterior tip of the maxilla reaches a vertical closer to the anterior orbital margin than to the nostril, rarely extending to a vertical through the margin of the orbit. Lower jaw 1-5-2-0 (mode 1-8) times longer than broad, its length 33-3-39-0 (M = 36-0)% of head length. Ascending processes of the premaxilla 25-7-34-3 (M = 30-6)% of head. Gill-rakers and pharynx. There are 8-10 (mode 10), relatively short and moderately stout gill-rakers in the outer row on the lower part of the first arch; the lowermost one or two rakers are smaller than the others. The rakers are transversely elongate, with the upper surface produced into two or three cusp-like projections. Microbranchiospines are present. In his original description of Tilapia steindachneri (see synonymy above), Boulenger (1913) gave the gill-raker count as 13-14, a count repeated in his redescription of 1915. These figures, however, apply only to those syntypes which Regan (1922) ultimately referred to Sargochromis mellandi. The remaining syntypes, which I refer to Th. buysi, have only 9 or 10 rakers. 192 P. H. GREENWOOD The dorsal pharyngeal epithelium is thickened and thrown into well-defined, approxi- mately longitudinal furrows, the crests of the ridges often further developed into low papillae. Immediately anterior to the toothed upper pharyngeal bones of each side, the buccal roof is produced into a prominent pad which, however, has neither the size nor the shape of the visor-like hanging pad found in certain cichlid genera (see Trewavas, 1974: 389-392, and Greenwood, 1983: 265-267). Scales are ctenoid below the level of the lower lateral-line, cycloid above it and on the chest. The chest scales are small, except for a midventral row of slightly larger scales, and are noticeably smaller than those on the ventrolateral aspects of the flanks and on the belly. The size transition is abrupt and takes place along a line connecting the pectoral and pelvic fin insertions, or a little behind that line. There are 32 (rare) to 36 (rare) scales, modally 34, in the lateral-line series, 4^-6^ (modally 5 or 5£) between the dorsal fin origin and the upper lateral-line, and 7-9 (mode 8) between the pectoral and pelvic fin bases. Cheek with 3-5 (mode 4) scale rows, the scaled area with a clearly demarcated, naked embayment anteroventrally. Each of the last 3 or 4 pored scales in the upper lateral-line is separated from the dorsal fin base by one large and one small Fins. Dorsal with 14 (fl), 15 (f!7), 16 (f26) or 17 (f2) spinous and 10 (f2), 11 (f24), 12 (f!8) or 13 (£2) branched rays. Anal with 3 spinous and 7 (f8), 8 (f37) or 9 (fl) branched rays. In all but three of the 46 specimens examined, small, almond-shaped sheath scales are present at the base of the anal fin (see p. 188 above). The horizontal extent of these varies from a row extending along the entire spinous part of the fin and reaching the 3rd-5th branched ray, to a few isolated and often non-imbricating scales at the base of either or both the spinous and the anterior part of the soft fin; sometimes only one or two scales are present and then usually at the base of the first one or two spines. The pectoral fin length is 19-6-26-4 (M = 2H)% of standard length, 61-3-80-0 (M = 70-2)% of head length. The pelvic fins have the first branched ray longer than the second, most noticeably so in adult males, but never produced into a filamentous extension, and never reaching to the origin of the anal fin. The caudal fin generally is subtruncate, but is almost truncate in a few specimens; it is scaled on its proximal third to half. Teeth. The outer row in both jaws of fishes up to ca 90 mm SL is composed, mostly, of relatively slender, unequally bicuspid and gently recurved teeth. The small minor cusp is angled away from the vertical axis of the major cusp (Fig. 2 A). In teeth from the upper jaw, the tip of the major cusp usually lies within, or but slightly beyond, the vertical formed by the outer margin of the tooth; in lower jaw teeth, however, the tip often lies well outside that line, as it may occasionally do in upper jaw teeth as well. Posteriorly in the premaxillary outer row of most specimens there are from 2 to 8 unicuspid teeth. These teeth, unlike those in Astatotilapia (see Greenwood, 1979), are not noticeably enlarged nor are they caniniform. Although some unicuspids are present laterally and anteriorly in the jaws of fishes less than 90 mm SL (especially those in the 75-90 mm range), their frequency only increases in specimens over 90 mm SL, becoming the predominant form in fishes more than 1 10 mm SL; even in these specimens, however, a few weakly bicuspid teeth are present in both jaws. The unicuspids are slender and slightly recurved, and do not have the near-cylindrical neck and crown of typical caniniform teeth. Both uni- and bicuspid teeth often show pronounced wear at the tip of the crown. There are 42-66 (modal range 50-62) teeth in the outer premaxillary series, the number not showing any clear-cut allometry with the fish's standard length. Inner series. It is difficult to generalize about tooth form in these rows because there is both a change with growth and, apparently, some inter-population differences as well. Most fishes less than 85 mm SL have a predominance of slender tricuspid teeth in the inner rows; the median cusp of these teeth is longer and broader-based than are the cusps flanking CUNENE RIVER HAPLOCHROMINE SPECIES 193 0-5 Fig. 2 Outer row jaw teeth of: A, Thoracochromis buysi; B, Th. albolabris; C, Orthochromis machadoi. A & B are anterior premaxillary teeth, C, a tooth from the anterior part of the dentary. it. A few slender bicuspid and weakly bicuspid, nearly unicuspid teeth are interspersed amongst the tricuspids, especially in fishes over 70 mm SL. Such teeth become more frequent in specimens between 75 and 85 mm SL. In specimens from certain localities, however, this admixture of tricuspids, weakly tricuspids and bicuspids is found in much smaller fishes, even among individuals as small as 47 mm SL. Fishes above ca 80 mm SL from all localities show a further increase in the number of bi- and unicuspid inner teeth, coupled with a decline in the number of tricuspids. These latter also tend to be less distinctly tricuspid, the median cusp gaining in dominance over the lateral ones. Specimens more than 90 mm SL have an essentially unicuspid inner dentition, although a few bicuspid and weakly bicuspid teeth persist; only the largest fish examined, 118-0 mm SL, has the inner rows composed solely of unicuspids. Anteriorly and anterolaterally the inner teeth are arranged in 2 (mode) or 3 rows, rarely in a single irregular row. Posteriorly in both jaws, however, only a single row of teeth is present. All but a few of the specimens examined have the dental mucosa greatly thickened with the result that just the tips of the teeth are visible. That this situation is a preservation artefact, cannot be overruled. Lower pharyngeal bone and dentition. The lower pharyngeal bone has an approximately triangular and equilateral dentigerous surface; the anterior shaft is short (Fig. 7A). Except for about the posterior four or five teeth in the median tooth rows, the pharyngeal teeth are slender, compressed and cuspidate, and are closely spaced. The exceptional teeth are dis- tinctly coarser and larger than their lateral congeners, but still retain a cuspidate crown. Sometimes a few posterior teeth in the rows immediately lateral to the median row are slightly coarser than the other lateral teeth. The pharyngeal bone itself is not enlarged, and has slender posterior horns. Osteology. Neurocranium. Overall skull morphology in this species (Fig. 10A) departs slightly from the generalized haplochromine type (Greenwood, 1979: 274) in being more slender, with a shallower and narrower otico-occipital region, narrower interorbital and ethmoid regions, and in having a lower and less expansive supraoccipital crest. Also the dorsal skull profile, from the anterior tip of the supraoccipital bone to the tip of the vomer, slopes less steeply (ca 30° compared with ca 45° in the case of Astatotilapia nubila or A. bloyeti; c/Fig. 10A with fig. 6 in Greenwood, 1979). Expressed as percentages of neurocranial length, the orbital depth is 34-8-36-3%, pre- orbital depth 17-4-20-8%, preotic skull length 63-6-66-6%, ethmoverine length 27-0-27-4%, 194 P. H. GREENWOOD depth of otic region 37-5-40-0%, width of otic region 50-0%, and greatest height of supra- occipital crest 16-5-18-2% (Data from three skulls, 22-0, 23-0 and 24-0 mm neurocranial length; for definition of measurements see Greenwood, 1980: 4-5). The apophysis for the upper pharyngeal bones is of the Haplochromis type (Greenwood, 1978); in two of the three skulls examined the basioccipital contribution to the facet is large, but in the third it is greatly reduced. Suspertsorium (Fig. 3 A). There is a distinct gap between the palatine and entopterygoid bones, an unusual features so far recorded only in members of the Ophthalmotilapia assemblage of Lake Tanganyika, and in at least some Lethrinops species (Lake Malawi); for a discussion of this feature see Greenwood (1983: 254-6, and 279). The hyomandibula in Th. buysi (Fig. 3C) has a fairly well-developed anterior flange, but one which is less expansive than that in Orthochromis machadoi (see Fig. 3E). Jaws. The dentary (Fig. 11 A) is a slender bone, with its alveolar surface flared out- wards so as to form a shelf-like overhang above the bone's lateral face. There is no mental projection in the symphysial region, which is, however, a little swollen. The premaxilla (Fig. 8A) has no outstanding features. Its ascending processes are long (almost one fifth longer than the dentigerous arm) and have a slight but obvious posterior Fig. 3 A, Suspensorium of Thoracochromis buysi; B, that of Th. albolabris, both in left lateral view. C, D & E, hyomandibula, in left lateral view, of C, Th. buysi; D, Th. albolabris; E, Orthochromis machadoi. Scale in mm. CUNENE RIVER HAPLOCHROMINE SPECIES 195 inclination at an angle of about 10° from the vertical. The dentigerous arms are laterally compressed and are not expanded anteriorly and anteroventrally to form a beak-like process. Caudal fin skeleton. All the hypurals are free in 22 of the specimens radiographed but in some hypurals 3 and 4 are very closely apposed to one another, and in two others hypurals 1 and 2 are fused. In another fish, hypurals 1 and 2 seemingly are fused distally but are free proximally, as are hypurals 3 and 4. All these observations were made from radiographs thus rendering it difficult to distinguish with certainty between actual fusion and close apposition. Vertebrae. Excluding the fused PU, + U, centra, there are 30 (f8), 31 (flO), 32 (f4) or 33 (fl) vertebrae, comprising 13 (f2) or 14 (f21) abdominal and 16 (f7), 17 (flO), 18 (f5) or 19 (f 1) caudal elements. The syntypical specimens of Tilapia steindachneri, from the Que river, are excluded from these counts; here the range is 28 (fl), 29 (fl) and 31 (f3), comprising 12 (fl), 13 (fl) or 14 (f3) abdominal, and 1 5 (f 1 ) or 1 7 (f4) caudal centra. In her original description of Th. buysi, Penrith (1970: 169) gives the vertebral count (including PL^ + Uj) for the holotype as 16+18; I have checked this figure on a radiograph made in the BM(NH), and find that my count, including the PU, + U, elements, is 14+ 18. Coloration. No information is available on live colours. For material fixed in formol and preserved in alcohol, the coloration is: Females and immature males, with a light brown (beige) ground colour which often becomes silvery on the belly and the flanks below the midlateral line. The intensity and presence of the silvery pigment may depend on factors of preservation since in some specimens it is absent, the beige colour merely lightening on the lower half of the belly. In those specimens which are silvery, faint traces of silver are present on the cheek and, more intensely, on the operculum. Traces of from 8-12 vertical bars are visible on the flanks and caudal peduncle; some of these bars extend almost to the ventral body profile, but most fade and disappear slightly below the level of the midlateral line. The intensity and clarity of the bars is very variable in the sample as a whole, but are reasonably constant within any one sample. Dr Michael Penrith (in litt) has observed, for the small cichlids of the Cunene, that coloration is generally darkest in fishes from the upper reaches of the river. All fins are greyish-hyaline, the dorsal with darker pigmentation between the spines, and dark maculae between the branched rays; the lappets are darker than the areas between the spines. The caudal fin is faintly maculate, with a dark posterior margin; this marginal band is most obvious when the fin is closed. The anal has dark lappets, and some indication of a dark margin to the anterior region of the soft part as well. In some males there are 5 or 6 dark spots, arranged, somewhat irregularly, in two rows on the soft part of the fin; the distal row lies a little above the fin's margin, the proximal row (usually with fewer spots) lies along the middle of the fin. There is no indication of a clear surround encompassing each of the spots, which thus cannot be considered true ocelli. The pelvic fins sometimes have a peppering of dark chromatophores which are most obvious in males. Sexually active males. In the few sexually active males examined, the overall coloration is much darker than that in females and inactive males. Scales above the midlateral line are outlined in dark brown, the vertical barring is moderately intense, the dorsum of the head and the entire snout is dark, almost dusky, as are the rami of the lower jaw and the anterior two-thirds of the lower lip. The branchiostegal membrane and the chest are dusky, but are lighter than the dentaries. The cheek is brown and only a little darker than the ground colour of the body. The membrane between the dorsal fin spines is almost black, but the lappets are hyaline; the soft part of the fin is densely maculate, the spots having a clear centre and a narrow, very dark brown surround. The proximal two-thirds of the caudal fin is covered in similar maculae, but its distal third is a somewhat dusky hyaline; the posterior margin is dark. The anal fin has a dusky hyaline ground colour showing between the large number, 8-10, of pale spots, each with a narrow, dark surround. The spots are arranged in three irregular rows (with from one to three spots in each) on the soft part of the fin. The anal spots are about 196 P. H. GREENWOOD four times larger than the biggest maculae occurring posteriorly on the dorsal fin. The pelvic fins are very dusky, almost black over the proximal half of each fin. The pectorals are hyaline. DISTRIBUTION. Thoracochromis buysi is known only from the Cunene river, including one of its tributaries, the Que. DIAGNOSIS AND AFFINITIES. Within the genus Thoracochromis, this species would seem, at least anatomically, to be a relatively primitive member of the group. This is particularly so when comparisons are made with species from the Nile and Lake Turkana (see Greenwood, 1979: 293-4). Unfortunately, insufficient information is available for many species occurring in Lake Mweru and the Zaire system to indicate what level of relationship might exist between them and Th. buysi. What information we have, however, does not suggest that a sister-species relationship is likely. In their overall appearance and most anatomical characters, two of the other Angolan Thoracochromis species, Th. schwetzi and Th. lucullae, are very similar to Th. buysi. Neither species occurs in the Cunene river, and only Th. schwetzi is well-represented by numerous specimens. Thoracochromis buysi differs from Th. lucullae mainly in having the depth of the pre- orbital bone greater, and not less than, the interorbital width (or, rarely, equal to it), and in details of its dentition. In Th. lucullae, the outer row jaw teeth have a relatively larger and broader minor cusp, with the result that its teeth are less unequally bicuspid than are those in Th. buysi. There are indications from the few available specimens of Th. lucullae that the scales are larger than in Th. buysi; there are 32 lateral-line scales in Th. lucullae as compared with a modal count of 34 or 35, rarely 32 or 36, in Th. buysi, and the cheek scale rows are generally more numerous in Th. buysi (3-5, mode 4, c/3 in Th. lucullae). Neurocranial form, and the osteological features of the jaws, are similar in the two species, although the narrower interorbital region in the skull of Th. buysi is very obvious. The suspensorium is damaged in the only skeleton of Th. lucullae, so it is impossible to check whether or not that species has a palatopterygoid gap (see p. 194 above). Thoracochromis buysi is also very similar to Th. schwetzi, which differs, however, in having unicuspid jaw teeth in specimens of a much smaller size. It also differs in generally having only a single series of inner teeth anterolaterally in both jaws, compared with the modal condition of 2 or 3 rows in Th. buysi. Like Th. lucullae, Th. schwetzi has a shallower preorbital bone than Th. buysi (1 5-8-20-0% head length, cf 18-5-26-0, M = 22-9% in Th. buysi) the depth of which is always less than the least interorbital width. In all three species the basic shape of the bicuspid outer row jaw teeth is similar, as is that of the tricuspid inner teeth. The unicuspid outer teeth in Th. schwetzi, however, are more slender and cylindrical in cross-section than are the unicuspids occurring in the other two species. The overall similarity of these three species, each from a different river system, might suggest that each is the vicariant (i.e. replacement) sister-species of the others. However, it must be stressed that there are as yet no synapomorphic characters known to be uniquely shared by the three taxa, and so their possible sister-species relationship cannot be established unequivocally. Thoracochromis buysi differs from Th. albolabris, the fourth Angolan member of the genus, in several characters, all of which are autapomorphic for Th. albolabris. The most obvious of these are the greatly thickened, often lobate lips of Th. albolabris, the very small chest scales in that species, and its narrower, near V-shaped dental arcades (see p. 201). Regrettably, I have not been able to examine the holotype of Steindachner's species 'acuti- ceps\ whose generic status thus remains unknown. It is clear from Penrith's (1970: 171) comments, and from Dr Trewavas' personal examination of the acuticeps type specimen, that Th. buysi differs from it in several features. In particular, Th. buysi has a less massive lower pharyngeal bone with smaller median pharyngeal teeth, and its lower jaw is shorter than in 'acuticeps\ CUNENE RIVER HAPLOCHROMINE SPECIES 197 The precise locality from which the type of 'acuticeps' was collected is unknown, and no descriptions of additional material have ever been published, apart, that is, from Regan's (1922) account which is clearly based on a polyspecific sample (see Trewavas, 1973: 31, and personal observations). Study material and distribution records Registered material Museum register number Locality Staatsmuseum Windhoek 5099 (Holotype) SAM 25243 (paratype) BMNH; P = collection number: Cunene river mouth Cunene river mouth 1984.2.6:1-7 P1682 Cunene R., Ondurusu falls (17° 24 'S, 1 3° 56 'E). 1984.2.6:8 P1347 Cunene R., Ondurusu falls (17° 24 'S, 13°56'E). 1984.2.6:9-14 PI 403 Cunene R., Ondurusu falls (17° 24 'S, 13° 56'E). 1984.2.6:15-18 P1422 Cunene R., Ondurusu falls (17° 24 'S, 13° 56 'E). 1984.2.6:19-23 P589 Cunene R., below Ruacana falls (17° 24'S, 14° 13 '£). 1984.2.6:24 P89 1984.2.6:25-27 P121 Cunene R., 3 miles west of Swartbooisdrif. (17° 19 'S, 16° 58'E). 1984.2.6:28-38 PI 780 Foz do Cunene (17° 15'S, 1 1° 43 '£). 1984.2.6:39-52 P1781 Foz do Cunene (17° 15'S, 11°43'E). 1984.2.6:53 P1841 Cunene R., Otjinungwa (17° 12 'S, 12° 20 'E). 1984.2.6:54 PI 126 Cunene R., Matala Dam, Luceque (14° 36 'S, 15° 18'E). 1984.2.6:55-56 PI 128 Cunene R., Matala Dam, Luceque (14° 36 'S, 15° 18'E). 1984.2.6:57-62 P1157 Cunene R., Chitapua (14° 23 'S, 15° 18'E). 1984.2.6:63 PI 186 Cunene R., Jamba-ia-Homa (13° 46'S, 15°30'E). Unregistered material Locality Number of specimens Collection no. Cunene R., 10 miles west of Ruacana (17° 26' S, 14°05'E). 2 P611 Cunene R., Ondurusu falls (17° 24'S, 13° 56'E). 1 P1416 Cunene R., Ondurusu falls (17° 24'S, 13° 56'E). 4 P1347 Foz do Cunene (17° 15'S, 11°43'E). 10 — Cunene R., Otjinungwa (17° 12 'S, 12° 20' E). 1 P1324 Cunene R., Otjinungwa (17° 12 'S, 12°20'E). 1 P1325 Cunene R., above Epupa falls (17° GO'S, 13° 15'E) 1 P696 Cunene R. , near Cafu ( 1 6° 30 ' S, 1 5° 1 0 ' E). 1 P808 Cunene R., 82 km west of Ondurusu falls ( 1 5° 59 ' S, 1 3° 22 ' E). 5 P1289 Cunene R., Chitapu (14° 23'S, 15° 18'E). 3 P1116 Cunene R., Chitapu (14° 23 'S, 15° 18'E). 1 PI 176 Thoracochromis albolabris (Trewavas & Thys van den Audenaerde), 1969 SYNONYMY. Haplochromis albolabris Trewavas & Thys van den Audenaerde, 1969. Mitt. zool. St Inst. Hamb. 66: 237-239, figs 1 & 2, plate 13. DESCRIPTION. Based on 18 specimens, ca 30-1 2 1-0 mm SL. Since 11 of these specimens 198 P. H. GREENWOOD were badly distorted before or during preservation, morphometric features are taken from 7 individuals only, 37-0-121-0 mm SL; these include the holo- and paratype of the species. Information on dentition, lip form and various meristic characters are, however, sup- plemented by data taken from the distorted specimens. Because the sample size from which the morphometric characters are derived is so small, ranges but not means or modes are given for those features. Depth of body 29-7-34-5% of standard length, length of head 32-0-36-7%. Dorsal head profile straight or very gently curved, sloping at an angle of 30°-35° to the horizontal, its outline sometimes interrupted by the slight prominence of the premaxillary ascending processes. The upper margin of the orbit lies distinctly below the outline of the head. Preorbital depth 16-0-22-4% of head length, least interorbital width 16-0-21-4%. Pre- orbital depth and interorbital width are equal in 2 specimens, the preorbital depth is greater in 4 others, and the interorbital width is greater in one specimen (see p. 191 above). Snout length 36-0-44-0% of head length and 1-3-1-6 times its breadth; in one exceptional specimen the length-breadth ratio is 1-9. Eye diameter 25-0-32-0% of head length, the largest eye being that in the smallest specimen measured (37-0 mm SL). Cheek depth is 19-6-24-2% of head length. The caudal peduncle is 1-3-1-7 times longer than deep, its length 17-0-19-3% of standard length. With such a small sample it is impossible to detect any features which might vary allometri- cally with standard length. The highest percentage ratios for preorbital depth, interorbital width, and pectoral fin length (see below) are, however, those for the smallest individual measured. The lips exhibit a wide but continuous variation in form, from those (as in the type specimens) which are clearly much thickened, but uniformly so (Fig. 4), those which are not only thickened but are produced medially into prominent lobes (Fig. 5). This latter con- dition is identical with that occurring in extreme individuals of Lobochilotes labiatus (Lake Tanganyika) and Paralabidochromis chilotes (Lake Victoria) and in the one Melanochromis labrosus (Lake Malawi) available for study. The intermediate stages of lip development seen in Th. albolabris are also encountered in the first two species listed above; M. labrosus is known from too few specimens to allow comment on that point. The degree of lip development in Th. albolabris, at least with respect to lobe formation, is not obviously size correlated; uniformly and distinctly thickened lips are present in Fig. 4 Thoracochromis albolabris', after Trewavas & Thys van den Audenaerde (1969). Drawn by G. J. Howes. Scale = 20 mm. CUNENE RIVER HAPLOCHROMINE SPECIES 199 Fig. 5 Thoracochromis albolabris. Variations in lip and lobe development. Scale in mm. specimens throughout the size range available. Some large specimens, for example the holo- type 96 mm SL and the paratype 121 mm SL, have unlobed lips; incipient lobes are present in some specimens 33-76 mm SL, and moderate lobes are developed in a fish 77 mm long. Other individuals in the size range 30-40 mm show no trace of lobes, but the lips are developed to an extent comparable with those in the much larger type specimens. Fully developed lobes are found in fishes of 73, 92, 93 and 121 mm SL. Because some large specimens lack lobes it is impossible to determine whether or not small specimens without lobes are at an early stage of future lobe development. But, judging from the sample studied it seems probable that lobes are not well-developed in fishes of less than 45-50 mm standard length. Since considerable variation in the extent of lobe development is recorded for other cichlid species (see Greenwood, 198 1 for the Lake Victoria species), I have no doubt that this sample of Cunene fishes is conspecific, particularly since all share other, and uniquely derived features as well. The mouth is horizontal or nearly so, the posterior tip of the maxilla either reaching a vertical passing close to the anterior orbital margin, or one lying about midway between it and the nostril. The ascending processes of the premaxilla are 34-5-41-0% of head length, the length of the lower jaw 35-2-40-0% and its length-breadth ratio is 1-3-1-7. Gill-rakers and pharynx. There are 1 1 (fl), 12 (f3), 13 (f3), 14 (f3), 15 (f4) or 17 (f 1) rakers in the outer row on the lower part of the first gill-arch; one specimen has 9 rakers on one side and 1 1 on the other. Except for the lower 4 or 5 elements, the rakers are short and stout to moderately stout, transversely elongate, and with the upper surface generally thrown 200 P. H. GREENWOOD into 2 or 3 cusps. The lower 4 or 5 rakers are reduced to little more than low knobs; the size-range of the lower rakers is positively correlated with the length of the fish, being barely visible in specimens less than 35 mm SL. The microbranchiospines are sometimes difficult to detect, but although small are always present. As in Th. buysi, the dorsal epithelium of the pharynx is thickened, noticeably corrugated and papillose. The prepharyngeal pads are well-developed and are like those of Th. buysi in size and shape. Scales are weakly ctenoid on the flanks below the upper lateral-line, and are very weakly ctenoid to cycloid on the caudal peduncle and above the upper lateral-line on the flanks. In larger specimens the degree of ctenoidy is weaker than in smaller individuals. The size- transition between the scales on the chest and those on the ventrolateral aspects of the flanks is abrupt; the chest scales are very small and deeply embedded. Two specimens appear to have small naked areas ventrolaterally on the chest, but closer examination shows that the scales in these areas are more deeply embedded than elsewhere in the region. There are 32 (f 1), 33 (f7), 34 (f5) or 35 (f3) scales in the lateral-line series, 4£ to 6£ (generally 5 or 5£) between the dorsal fin origin and the upper lateral-line, and 8 to 10 between the pectoral and pelvic fin insertions. The cheek has 3 to 5 scale rows (usually 3), generally embedded deeply in the thickened skin. Anteroventrally there is a distinct but sometimes small naked embayment, and there is always a narrow naked strip between the cheek scales and the preoperculum. With one exception, all specimens have one large and one small scale between each of the last 4 or 5 scales in the upper lateral-line and the dorsal fin base. In the exceptional fish there are, on one side, two large scales between the lateral-line and the fin, and on the other side two large and one small scale. Fins. Dorsal with 14 (fl), 15 (f!4 ) or 16 (fl) spinous and 10 (f2), 11 (f9) or 12 (f5) branched rays, the anal with 3 spinous and 7 (f4), 8 (f 1 0) or 9 (f 1 ) branched elements. All but three fishes (30-35 mm SL) have either a well-developed scale sheath at the base of the anal, or isolated sheath scales present in that region, even in specimens as small as 33-34 mm SL. The pectoral fin is 1 8-9-25-0% standard length, 56-0-66-8% of head length. The pelvic fins have the first branched ray only a little longer than the second, or rarely, the two are of equal length. In large adult males, however, the first ray is clearly longer than the second, but is not drawn out into a distally filamentous projection. The caudal fin is subtruncate and is scaled over its proximal third to half, or exceptionally, over its proximal two-thirds. Teeth. The dental arcade in Th. albolabris (Fig. 6) is more nearly 'V- than 'U' shaped, especially in specimens with fully lobate lips. A similar correlation of arcade shape with lip development is seen in other species, mentioned above, with lobed lips (Greenwood, 1959: 209, and other observations). The outer teeth (Fig. 2B), in virtually all the specimens have worn cusps, thus making it difficult to describe accurately the form of the crown. Unworn teeth in fishes over 75 mm SL are either unicuspid or weakly bicuspid; in bicuspids the minor cusp, although reduced as compared with that cusp in Th. buysi, does show a similar angling away from the major cusp. Fishes in this size range generally have unicuspid teeth posteriorly in the premaxillary series. As in Th. buysi, these and other unicuspids are slender and relatively compressed, with attenuated rather than pointed crowns; that is, they could not be described as caniniform. From the sample available it seems that in fishes less than 75 mm SL the majority of unworn teeth are distinctly but unequally bicuspid, with a cusp shape and arrangement like that in the weakly bicuspid teeth of larger individuals. Teeth situated posteriorly in the outer premaxillary row of these smaller fishes are weakly bicuspid. There are 22-64 teeth in the outer premaxillary series, the number probably having a positive correlation with standard length, although in one of the largest specimens (121 mm SL) there are only 50 teeth, and a specimen of 37 mm has 60. CUNENE RIVER HAPLOCHROMINE SPECIES 201 B Fig. 6 Premaxilla, in occlusal view, of: A, Thoracochromis buysi; B, Th. albolabris to show differ- ence in outline of dental arcade; only the basal region of the ascending processes is shown. Scales in mm. The inner teeth are arranged, anteriorly and laterally, in from 1 to 4, usually 2, rows in the upper jaw, and in 1 to 3, usually 2, in the lower jaw; posterolaterally in both jaws only a single row is present. Fishes less than 70 mm SL have mostly weakly tricuspid inner teeth in which the larger median cusp is flanked by slight, shoulder-like projections. Fishes over 70mm SL may have an admixture of tri- and weakly bicuspid teeth or of bi- and, predominantly, weakly bicuspids, or most of the teeth may be unicuspids. Lower pharyngeal bone and dentition (Fig. 7B). The anterior shaft is short, the dentigerous area triangular and varying from slightly longer than broad, through equilateral, to slightly broader than long. Some variation exists in the stoutness of the bone. In the holotype it is moderately stout, the stoutest seen amongst the specimens available. In other, and larger, fishes the bone is but slightly thickened, while in the majority of specimens it shows no marked departure from the generalized condition seen in Th. buysi. There is variation, too, in the degree to which teeth in the median pair of tooth rows are enlarged or coarsened. Certain of the, larger specimens (including the holotype) have some of these teeth distinctly enlarged with near-molariform crowns, but in the majority of speci- mens the teeth are slender and cuspidate, those situated in the posterolateral corners of the dental field being more closely spaced than elsewhere. Osteology. The neurocranium is similar to that of Th. buysi (Fig. 10B) in its general form, but has a lower supraoccipital crest, a flatter (that is, less concave) interorbital skull roof, and a convex not concave surface to the posterior part of the roof contributed by the frontals. The apophysis for the upper pharyngeal bones is of the Haplochromis type, but in some specimens the basioccipital contribution to the facet is rather small (although never as small as in the Tropheus type of apophysis; see Greenwood, 1978). Expressed as percentages of neurocranial length (18-5 mm in the one skull available for measurement) the orbital depth is 37-9%, the preorbital depth 23-3%, the ethmovomerine 202 P. H. GREENWOOD Fig. 7 Lower pharyngeal bones, in occlusal and ventral views, of: A, Thoracochromis buysi; B, Th. albolabris; C, Orthochromis machadoi. Scale in mm. length 28:7%, the depth of the otic region 39-9%, the width of the otic region 51-3%, and the greatest height of the supraoccipital crest 13-5%. Suspensorium. Unlike that in Th. buysi, the suspensorium in Th. albolabris (Fig. 3B) has no gap between the palatine and entopterygoid bones, and the posterior margin of the pala- tine is gently curved rather than rectangular. The anterior flange on the hyomandibula in Th. albolabris is slightly more expansive than in Th. buysi. Jaws. Relative to that in Th. buysi, the maxilla in Th. albolabris is foreshortened (Fig. 8D & E). The premaxilla (Fig. 8B) has elongate ascending processes which are about 1-25 CUNENE RIVER HAPLOCHROMINE SPECIES 203 Fig. 8 Premaxilla (upper row) of: A, Thoracochromis buysi; B, Th. albolabris; C, Orthochromis machadoi. Maxilla, in left lateral and dorsal view (middle and lower rows respectively) of: D, Th. buysi; E, Th. albolabris. Scale in mm. times the length of the dentigerous arms, and which slope strongly backwards at an angle of about 25° to the vertical. In occlusal view the two dentigerous arms are arranged so as to form a 'V'-rather than a 'U'-shaped outline (Fig. 6). These arms are not inflated, but anteriorly and somewhat anteroventrally each is produced forward beyond the base of the ascending process to form a prominent, shelf-like beak (Fig. 8B). The dentary (Fig. 1 IB) is like that in Th. buysi, but the rami of each side meet to form a more nearly 'V-shaped occlusal surface in Th. albolabris, and the anterior 'shelf is more prominent. Caudal fin skeleton. All five hypurals are separate from one another in six of the nine specimens examined (1 alizarin specimen and 8 radiographs), but in the three others, hypurals 3 and 4 are so closely apposed as to appear almost fused. Vertebrae. Excluding the fused PU, and U, centra, there are 30 (f2) or 3 1 (f5) vertebrae, comprising 13 (fl) or 14 (f6) abdominal and 16 (f2), 17 (4) or 18 (f 1) caudal elements. Coloration. Live colours are unknown, and the coloration can only be described for formol fixed and alcohol preserved specimens. 204 P. H. GREENWOOD There are no sexually active fishes in the entire sample available for study. Immature and sexually quiescent individuals of both sexes have a similar coloration. The ground colour is light brown (beige), darkening to greyish-beige on the dorsum of the head in some individ- uals. The body is crossed by 7-10 bars, often of irregular outline and shape, and either vertically or somewhat obliquely aligned. Some may be simple bars, others are narrowly triangular with, in any one specimen, the apex directed either dorsally or ventrally. Several specimens have a faint and narrow midlateral stripe extending the whole length of the body; a second narrow band runs along the upper lateral-line scale row, interconnecting the bars over the posterior half of the body. In some specimens there is a rather broad but faint lachrymal stripe running from the anteroventral margin of the orbit to the angle of the jaw; others have, in addition, a narrow bar across the snout between the anterodorsal margins of the orbits. All the fins are hyaline, the spinous dorsal with dark streaks between the rays, and dark lappets; the soft dorsal is fairly densely maculate, the maculae either solid, or light-centred with a ring of dark marginal spots circumferentially. The caudal is densely and darkly macu- late over most of its length, with, in at least some individuals, a dark posterior margin. In one of the males examined there are numerous light-centered but dark margined spots arranged rather irregularly in two or three rows; none of the spots has a clear surround. Other males, and most females, have the anal fin hyaline, but in some almost the anterior half of the anal fin is a light greyish-sooty colour. The pectoral and pelvic fins are hyaline, but in one male there are faint dusky areas over about the anterior third of the pelvic fins. The figure of the holotype, a male, published by Trewavas and Thys van den Audenaerde (1969: plate 1) gives a good impression of a darkly pigmented fish; the specimen, however, is now somewhat faded. DISTRIBUTION. Known only from the Cunene river, see also p. 205. DIAGNOSIS AND AFFINITIES. Specimens of Th. albolabris with lobed lips are immediately distinguishable from all other cichlids occurring in the Cunene river. Those specimens with- out obvious lobes are recognisable by the degree to which the lips are thickened and by the near 'V'-shaped dental arcade, both features which, of course, also serve to diagnose lobe-lipped individuals as well. Thoracochromis albolabris is further distinguished from all Cunene haplochromines, except Orthochromis machadoi (see p. 206), by the very small size of its chest scales, and from all, except the species of Serranochromis, by its high gill-raker counts (1 1-17). From the Serranochromis species it is distinguished by various dental characters and several morphometric features as well. From Orthochromis machadoi, Th. albolabris is distinguished by its head shape, dental pattern, absence of naked areas on the chest, and by having the first branched pelvic ray at least equal in length to the second ray, and generally a little longer than it. When first describing Th. albolabris, Trewavas and Thys van den Audenaerde (1969) compared the taxon with a then undescribed haplochromine which they considered to be ' . . . very close to our species'. The undescribed species was Th. buysi, from which Th. albolabris can be separated readily on the several features noted above, particularly the high gill-raker count. The two taxa do share certain features, for example, a similar tooth morpho- logy and various meristic and morphometric similarities, but none can be considered as synapomorphies indicative of a sister-species relationship between these species. Indeed, the resembtences are all in plesiomorph characters. When compared with the Thoracochromis species occurring in other parts of Africa (Greenwood, 1979), Th. albolabris does exhibit a large number of derived features, for example, the hypertrophy of its lips, the high gill-raker count, the beaked premaxilla with its strongly angled ascending process, and the near 'V-shaped dental arcade. Again, none of these features is an intrageneric synapomorphy; as autapomorphies, none can be used to CUNENE RIVER HAPLOCHROMINE SPECIES Study material and distribution records 205 Museum register number Locality ZMH 1784(Holotype) Stockholm Museum, NRM NNN/95 1989, 7151 BMNH; P = collection no. 1972.9.27:89 1984.2.6:101 1984.2.6:87-89 1984.2.6:90 1984.2.6:91 1984.2.6:92-98 P710 P1483 P609 P1276 PI 389 (one an alizarin prep.) 1984.2.26:99-1 00 P643 (one used for skeletal prep.) CuneneR., Matala (14° 43'S, 15°04'E) Southern Angola; locality unknown. Collected by M. Frohlich. Cunene R., above Epupa falls (17° OO'S, 13° 15'E). CuneneR., 1 mile east of Epupa falls (17° OO'S, 13° 15'E). Cunene R., Ondurusu falls (17° 24' S, 13° 56'E). Cunene R., Ondurusu falls (17° 24' S, 13° 56'E). Cunene R., 32 km west of Odurusu falls (17° 15'S, 13° 42'E). Cunene R., 54 km west of Ondurusu falls. CuneneR., 27 miles west of Ondurusu falls (17° 10'S, 13°33'E). establish the intrageneric relationships of Th. albolabris. That Th. albolabris shows such a high number of derived features would seem to negate the views of Trewavas and Thys van den Audenaerde (1969: 238), who considered that it should be ranked amongst the general- ised haplochromines. For the moment, Th. albolabris must remain a phylogenetic isolate amongst its congeners. In the hyperdevelopment of its lips, and in other, possibly correlated characters, Thorac- ochromis albolabris resembles Paralabidochromis chilotes of Lake Victoria, Cyrtocara lobochilus and C. euchilus (Lake Malawi), Melanochromis labrosus (Lake Malawi) and Lobochilotes labiatus (Lake Tanganyika). In each of these lakes there are other species, too, some still undescribed, which resemble Th. albolabris in having hypertrophied lips as well as sharing certain dental and osteological features with that taxon. I have made detailed comparisons of Th. albolabris with all these species, and find that on the basis of various characters and character combinations, arguments can be deduced which would militate against postulating Th. albolabris as having a close phylogenetic relationship with any but one of them, namely Melanochromis labrosus. For example, the dentition and the morphology of the jaws, together with details of the neurocranial osteology in Lobochilotes labiatus, Cyrtocara euchilus and Paralabidochromis chilotes differ quite trenchantly from the condition seen in Th. albolabris, and in the two latter species there are differences in squamation patterns as well. Few anatomical details could be studied in Cyrtocara euchilus, a species represented in the BMNH collections only by the type speci- men. It differs from Th. albolabris in having an Astatotilapia-type of chest squamation (see Greenwood, 1979: 270; fig. 1), and there also appear to be differences in the dental morphology of the two species. On the basis of their sharing the greatest number of derived features, the Malawian Melanochromis labrosus would thus seem to be the species most closely related to Th. albolabris. Unfortunately, as M. labrosus is known only from one specimen, its osteology could not be studied in any detail. For that and other reasons, especially our great ignorance of cichlid interrelationships in general, I would not develop any further the suggestion that Thoracochromis and Melanochromis labrosus might be closely related. 206 P. H. GREENWOOD ORTHOCHROMIS Greenwood, 1954 This genus, originally defined on the basis of one species, Orthochromis malagaraziensis (David), was later expanded to include three other taxa (see Greenwood, 1979: 295-7). At that time the Cunene river species, O. machadoi (Poll), was known only from the two types and one other specimen. The greatly increased number of 0. machadoi specimens now available for study requires that two of the diagnostic features for the genus be modified. In my 1979 paper (page 296) it was indicated that all the upper lateral-line pore-bearing scales in Orthochromis are each separated from the dorsal fin base by not more than one large and one much smaller scale. (The few anterior scales in the upwardly curving portion of the lateral-line are excluded from that generalization.) In O. machadoi, however, only the last 9-12 pored scales of the upper lateral-line are separated from the fin in this way, the more anterior scales having two large scales between them and the fin base; in one specimen only the last 3-5 scales have less than two equal-sized scales in that position. Apart from this exceptional specimen the number of lateral-line scales separated from the dorsal fin by less than two scales in Ortho- chromis machadoi is still high as compared with the usual condition in haplochromine genera, and represents a situation otherwise only found in Ctenochromis (see Greenwood, 1979: 287). Even here the modal number of lateral- line scales involved (8 or 9) is lower than in O. machadoi. Although O. machadoi can be described as having a relatively elongate and slender body (see Greenwood, 1979: 296) its body depth is now known to range as high as 34-5% of standard length (c/the maximum of 30% cited in Greenwood, 1979), and the mean depth for the specimens sampled is 31-1% SL. Apart from these modifications, no other data derived from the enhanced O. machadoi collection necessitates changes in the generic characters enumerated in Greenwood (1979). The additional material does, however, reinforce earlier conclusions that this species is the least derived member of the taxon (Greenwood, 1979: 298). As far as I am aware, the new material also provides the first indication, in nature, that any species of Orthocromis is a female mouth-brooder (see Staeck, 1983: 178 for comments on the behaviour, in an aquarium, of an unidentified species resembling O. polyacanthus). A female O. machadoi, 40-5 mm SL, from Folgares (14° 55' S, 15° 06' E) is carrying fry in the buccal cavity. The characteristically distorted buccal cavity in some females from other localities also suggests that these fishes were carrying young at the time of their capture. Orthochromis machadoi (Poll), 1967 SYNONYMY. Haplochromis machadoi Poll, 1967. Publicacbes cult. Co. Diam Angola no. 75: 313-315, fig. 152. Orthochromis machadoi (Poll): Greenwood, 1979. Bull. Br. Mus. nat. Hist. (Zool.) 35: 295-299. DESCRIPTION. Based on 35 specimens, 38-0-65-0 mm SL, excluding the type and paratype which were examined previously (Greenwood, 1979). Various characters, such as squama- tion patterns, fin shape, body form, and dentition were checked on many other specimens, most of which were too distorted to use for morphometric purposes. Depth of body 27-2-34-5 (M = 3M)% of standard length, length of head 28-5-35-0 (M = 32-3%). Dorsal head profile sloping upwards at an angle of 40°-50° to the horizontal, its outline not broken by the ascending premaxillary processes, and generally straight until a point above the middle of the orbits, after which it is gently curved; in some specimens the lower part of the profile is also slightly curved. The upper margin of the orbit lies distinctly below the level of the dorsal head profile. Preorbital depth 13-7-29-9 (M=18-2)% of head length; in one extreme individual the preorbital is only 12-5% of head length. Least interorbital width 12-5-20-0 (M=16-0)% of CUNENE RIVER HAPLOCHROMINE SPECIES 207 Fig. 9 Orthochromis machadoi. Drawn by G. J. Howes. Scale = 20 mm. head, generally narrower than the preorbital is deep, but occasionally the measurements are equal. Snout 0-8-1-1 times as long as it is broad (modally 1-0 times), its length 29-2-35-7 (M = 3 1-8)% of head. Eye. diameter 24-2-33-3 (M = 28-2)% of head, showing no obvious allometry with standard length. Cheek depth 20-8-33-3 (M = 25-4)%. Caudal peduncle 1-2-1-5 times longer than deep (modally 1-4-1-5 times), its length 15-4-20-5% of standard length. Mouth horizontal or almost so, the lower jaw generally a little shorter than the upper when the mouth is closed. Lips thickened (but not so noticeably as in Th. albolabris) with, in a few specimens, the dorsal margin of upper lip overlapping the ventral margin of the pre- orbital bone. The posterior tip of the maxilla reaches a vertical through the anterior margin of the orbit or, in a few specimens, extending either a little beyond, or not quite as far as that level. In the latter situation, however, the maxillary tip always reaches to a level well posterior to the nostril. Lower jaw length 3 1 -0-41-3 (M = 37-0)% of head, and 1 -4-1 -8 (mode 1 -5) times its breadth. The length of the premaxillary ascending processes is 23-8-31-1 (M = 26-8)% of the head. Gill-rakers and pharynx. There are 7 (f2), 8 (f!7), 9 (f!3) or 10 (f3) gill-rakers in the outer row of the lower part of the first gill-arch; the lower 1-3 rakers are reduced, the others short and stout, and not expanded transversely across the arch; the crown of each raker is simple or, less commonly, crenulate. The prepharyngeal pads in O. machadoi are moderately developed and are comparable with those in the Cunene Thoracochromis species, but the pharyngeal epithelium is not noticeably thickened, neither is it distinctly papillate or thrown into longitudinal furrows. Well-developed microbranchiospines are present. Scales. Above the upper lateral-line the scales are weakly ctenoid or cycloid, those below that level are ctenoid except for the cycloid scales on the chest and belly. The chest scales are very small and sharply demarcated in size from those on the ventrolateral aspects of the anterior flanks and belly. The ventromedial scales on the belly are markedly smaller than those on the ventral aspects of the flanks, and show an almost imperceptible size gradient with the scales on the chest (see Greenwood, 1979, fig. 3 for an illustration of the Ortho- chromis chest-belly squamation pattern; the belly scales in O. machadoi are relatively larger 208 P. H. GREENWOOD than those in the species depicted). Most specimens examined have, bilaterally, a naked area in the chest squamation. The size of this area shows some intraspecific variability, but in no specimen is the entire lateral region of the chest naked. There are 29 (fl), 30 (f!5), 31 (f!2), 32 (f4) or 33 (fl) scales in the lateral-line series, 6H£ between the upper lateral-line and the dorsal fin origin, 8-1 1 between the pectoral and pelvic fin bases, and 2-4 (modal range 2-3) rows on the cheek. Cheek squamation pattern is often irregular, but a horizontal naked strip is always present between the scale rows and the preoperculim, as is a small, anteroventrally situated, naked embayment of the scaled area. There are never more than one large and one small scale between the dorsal fin base and each of the last 6-12 (modally 9-1 1) pored scales of the upper lateral-line; in one exceptional specimen, however, only the last 3 and 5 scales on the two sides respectively are separated from the fin in this manner, the other scales in the series being separated by 2 large scales of equal size. Fins. Dorsal with 15 (f3), 16 (f20) or 17 (flO) spinous, and 9 (f8), 10 (f20) or 11 (f5) branched rays; anal with 3 (in one specimen 4) spinous and 7 (fl), 8 (f27) or 9 (f5) branched elements. In only 2 of the many specimens examined were traces of anal sheath scales observed (see p. 1 88), and then only as a single scale in each fish. Caudal fin rounded or strongly subtruncate, scaled on its proximal half or, rarely, two-thirds. Pelvic fins with the second branched ray longer than the first, and occasionally, the third also as long as the second; the pelvic spine and the first two rays of the fin are covered by greatly thickened skin. Pectoral fin 18-3-23-9 (M = 21-5)°/o of standard length, 57-0-81-3 (M = 67-2)% of head length. Teeth. In both jaws the outer row is composed of relatively slender, close-set and unequally bicuspid teeth (Fig. 2C). Generally from 1-3 slender unicuspids are situated posteriorly in the upper jaw. The number of teeth in the premaxillary outer row varies from 32-50 (modal range 46-50), the number showing weak positive correlation with the fish's standard length. The inner series in both jaws are densely arranged in 2-4 (usually 3 or 4) rows anteriorly and anterolaterally, reducing to a single row posteriorly. The teeth are all tricuspid, with the median cusp slightly larger than the lateral ones. Lower pharyngeal bone and dentition. The dentigerous surface is broader than long (ca 1-3 times), and the anterior shaft of the bone is short (Fig. 7C). Its teeth are cuspidate, fine and compressed, with only those in the posterior transverse row noticeably larger and coarser than the others. The teeth in the two median rows are barely coarser than those situated laterally. Osteology. I have been able to compare, intragenerically, the skeleton of Orthochromis machadoi only with that of O. malagaraziensis (from the Malagarazi river, Tanzania). In all features which could be checked, the two species are virtually identical. The neurocranium (Fig. IOC) is notable for its low supraoccipital crest, fairly steeply sloping (ca 45°) dorsal profile, gently rounded transverse profile of the skull roof anterior to the supraoccipital crest, and the slightly convex rather than concave camber to that part of the roof situated between the parietal crests and the base of the supraoccipital crest. The apophysis for the upper pharyngeal bones is of the Haplochromis-type, with small but definite contributions from the basioccipitals. Expressed as percentages of neurocranial length in the two 1 3 mm long skulls examined, the orbital depth is 38-0%, the preorbital depth 23-0%, the ethmovomerine length 30-8%, the depth of the otic region 46-0 and 48-5%, the width of the otic region 53-8%, and the height of the supraoccipital crest 15-4% (measured in only one skull). Suspensorium (Fig. 3E). The palatopterygoid region is relatively foreshortened, and the hyomandibula has an expanded anterior flange which extends well beyond the level of the bone's anterior articulatory facet. This expansion of the'hyomandibula is probably correlated with the enlarged levator arcus palatini muscle. The entopterygoid and palatine bones are in contact. CUNENE RIVER HAPLOCHROMINE SPECIES 209 Fig. 10 Neurocranium, in left lateral view, of: A, Thoracochromis buysi; B, Th. albolabris; C, Orthochromis machadoi. Scale in mm. Jaws. The dentigerous arms of the p re maxilla are slightly inflated, the ascending processes are shorter than the dentigerous arms, and have but a slight posterior inclination (Fig. 8C). The lower jaw (Fig. 11C), as compared with the generalized haplochromine condition, appears foreshortened in lateral view, with the coronoid arm of the dentary originating rela- tively far forward, and the region surrounding the dentary's division into horizontal and coronoid arms somewhat inflated. The outer tooth row extends posteriorly onto the anterior half, or slightly less, of the coronoid process. 210 P. H. GREENWOOD The dorsal gill-arch skeleton was examined in 3 cleared specimens, double stained with alizarin-red and alcian-blue. Its most outstanding feature is the very greatly reduced cartilagi- nous extension from the anterior border of epibranchial II. In all African cichlids examined so far there is an expansive cartilaginous flange developed from this epibranchial, the flange extending forward and ventrally beyond the head of epibranchial I (see Stiassny, 1981: 294- 296, and 1 982: 430-432 for an account and figures of this feature in cichlids and related taxa). Fig. 11 Lower jaw, in left lateral view, of: A, Thoracochromis buysi; B, Th. albolabris; C, Orthochromis machadoi. Scale in mm. CUNENE RIVER HAPLOCHROMINE SPECIES 2 1 1 In O. machadoi virtually no such extension is present, the margin of epibranchial II having merely a narrow strip of hyaline tissue (which is partly stained by the alcian blue). The strip is very slightly wider on the right than on the left side of the branchial skeleton, but on neither side is it broader or more extensive than the cartilaginous strip shown by Stiassny (1982: fig. 3) as occurring in the South American species Cichla ocellaris. Regrettably, no double stained preparations are available for other Orthochromis species, and the feature cannot be checked on the dry skeletons available. Its significance, phylogenetically speaking, remains to be investigated. Caudal fin skeleton. All eight specimens radiographed have hypurals 1 and 2, and 3 and 4 fused, as do the three alizarin transparencies examined. Such extensive and consistent fusion of hypural elements would seem to be a characteristic feature of the genus Ortho- chromis (Greenwood, 1979: 297). Vertebrae. Excluding the fused PU, and U, centra, there are 28 (f2), 29 (f!7) or 30 (f!3) vertebrae, comprising 12 (f21) or 13 (fl 1) abdominal and 16 (f6), 17 (f20) or 18 (f6) caudal elements. Breeding. As noted earlier, O. machadoi is apparently a female mouth-brooder. The smallest sexually active female recorded (apart from the brooding individual) is 46 mm SL, the smallest male, 52 mm. However, the preservation state for the majority of specimens examined precluded accurate determinations of sexual maturity or activity. Hence these figures can give only rough indications of the size at which sexual activity begins. Coloration. Information is only available for formol fixed and alcohol preserved speci- mens. There are apparently no sexually correlated differences in preserved coloration; sexually active adults of both sexes are represented in the samples studied. The ground coloration is a light brown (beige), usually shading to yellowish-brown ventrally. There are about twelve broad and dark 'bars' crossing the upper two-thirds to three-quarters of the body's lateral surface, the breadth of the 'bars' being about twice the width of the paler spaces between them. The 'bars' are not solid, but are formed from the dark outlines to the scales underlying the area which they occupy. The centres of these scales are lighter, so that the overall effect is to produce a diamond-mesh pattern of dark reticula- tions which, in places, are condensed to give the appearance of vertical bars. The centre of each 'bar' is darkest and is somewhat expanded anteroposteriorly, thus creating the appear- ance of an interrupted mid-lateral stripe; a similar but narrower and fainter stripe runs along the course of the upper lateral-line. At the base of the caudal fin there is a somewhat vertically elongate blotch. The snout is crossed by two bands, the upper of which is the broader and becomes almost continuous with the darkly bordered scales on the nape and the posterior interorbital regions of the dorsum. A broad and intense lachrymal stripe runs from the orbit to the angle of the jaws, and is continued dorsally as a short blotch above the posterodorsal margin of the orbit. Post-orbitally there is a narrow horizontal bar continuous with the dark dorso-posterior margin of the operculum. Dorsal, caudal and anal fins are a dusky hyaline, with the dark pigment most concentrated, almost into streaks, between the spinous rays of the dorsal fin. The lappets of that fin are clear. The anal is, apparently, without discrete maculae, and there are no indications of any ocellus-like spots. In males the pelvic fins are dark over their anterior third, and in females are hyaline with a faint dusting of small melanophores over the greater part of their surface. Within the material examined there is considerable variation in the intensity of dark pigmentation, especially that contributing to the vertical 'bars'. However, within any one sample the intensity is generally constant. Whether this variation is attributable to differing preservation techniques, or is a reflection of populational differences or different environ- ments, remains an open question. DISTRIBUTION. The species is known only from the Cunene river; see also Poll (1967: 23 & 314), and p. 212 for detailed distribution records within the Cunene. 212 P. H. GREENWOOD Study material and distribution records of (>. machadoi Registered material BMNH; P = collection no. Locality 1972 ,9 .27:90-91 Cunene R., Ondurusu falls ( 1 7° 24 ' S, 1 3° 56 ' E). 1984 .2 .6: 102-103 P612 Cunene R., Ondurusu falls (17° 24 'S, 13° 56 'E). 1984 .2 .6: 104-108 PI 192 Cunene R., Calueque(17° 16'S, 14° 30'E) 1984 .2 .6: 109 P672 Cunene R., 45 miles west of Ondurusu falls (17° 03 'S, 13°30'E). 1984 .2 .6: 110-112 P696 Cunene R., above Epupa falls (1 7° 00 'S, 1 3° 1 5 'E). 1984 .2 .6: 113 P713 Cunene R., above Epupa falls (17° GO'S, 13° 15 'E). 1984 ,2 .6: 114-115 P716 Cunene R., above Epupa falls (17° GO'S, 13° 15°E). 1984 .2 .6: 116-131 P1341 Cunene R., 82 km west of Ondurusu falls (16° 59' S, 13 °22'E). 1984 ,2 .6: 132-141 PI 176 Cunene R., in the Chitapua falls (14° 23 'S, 15° 18 'E). 1984 ,2 .6: 142-145 P918 Cunene R., Folgares(14°55'S, 15°06'E). 1984 .2 .6: 146 P882 Cunene R., Folgares(14°55'S, 15°06'E). Unregistered material Locality Number of specimens Collection no. Cunene R., 10 miles west of Ruacana falls (17° 26'S, 14° 05 'E). 2 Cunene R., 5 km west of Ondurusu falls (17° 24' S, 13° 56 'E). 10 Cunene R., Ondurusu falls (17° 24 'S, 13° 56 'E). 12 Cunene R., Calueque (17° 16 'S, 14° 30 'E). 2 Cunene R., 32 km west of Ondurusu falls (17° 15'S, 13° 23 'E). 7 Cunene R., 32 km west of Ondurusu falls (17° 15 'S, 13° 23 'E). 9 Cunene R., 32 km west of Ondurusu falls (17° 15 'S, 13° 23 'E). 9 Cunene R., 32 km west of Ondurusu falls ( 1 7° 1 5 ' S, 1 3° 23 ' E). 17 Cunene R., Otjinungura (17° 12 'S, 12° 20 'E). 12 Cunene R., Otjinungura (17° 12 'S, 12° 20 'E). 1 Cunene R., 45 miles west of Ondurusu falls (17° 03' S, 13°30'E). 12 Cunene R., above Epupa falls (17° GO'S, 13° 15 'E). 14 Cunene R., above Epupa falls (17° GO'S, 13° 15 'E). 8 Cunene R., above Epupa falls (17° GO'S, 13° 15 'E). 2 Cunene R., above Epupa falls (17° GO'S, 13° 15 'E). 1 Cunene R., above Epupa falls (17° GO'S, 13° 15 'E). Cunene R., above Epupa falls, and 1 mile east. 4 Cunene R., above Epupa falls, and 1 mile east. 1 Cunene R., 82 km west of Ondurusu falls ( 1 6° 59 ' S, 1 3° 22 ' E). 2 Cunene R., 82 km west of Ondurusu falls (16° 59 'S, 13° 22 'E). 18 Cunene R,, 82 km west of Ondurusu falls (16° 59'S, 13° 22 'E). 10 Cunene R., 82 km west of Ondurusu falls (16° 59'S, 13° 22 'E). 2 Cunene R., Folgares(14° 55'S, 15° 06'E). 10 Cunene R. , Folgares ( 1 4° 5 5 ' S, 1 5° 06 ' E). 13 Cunene R., Folgares (14° 55'S, 15° 06' E). 4 Cunene R., Chitapua ( 1 4° 1 3 ' S, 1 5° 1 8 ' E). 2 P604 P1392 P1481 PI 192 P1276 P1278 P1296 P1299 P1843 PI 324 P670 P705 P715 P697 P710 P1181 P708 P706 P1290 P1341 P1361 P1290 P918 P1345 P882 P1116 CUNENE RIVER HAPLOCHROMINE SPECIES 2 1 3 DIAGNOSIS AND AFFINITIES. Orthochromis machadoi is easily distinguished from other Cunene haplochromines by several features. Amongst these may be noted its very small chest scales which are continuous with the small scales ventrally on the belly; the bilateral naked patches on the chest; the pelvic fins with the second and often the third branched ray longer than the first; the thickened skin covering the pelvic spine and first two or three branched rays; by the absence of discrete spots or ocelli on the anal fin of male fishes, and osteo- logically, by the low supraoccipital crest, the generally convex dorsum of the skull, and the greatly reduced cartilaginous projection from the anterior face of the second epibranchial bone. In his original description of the species, Poll (1967: 314) commented on the similarity between the preserved colour pattern of O. machadoi and that of Pseudocrenilabrus philander (for which see p. 2 1 5). The new material of both species from the Cunene river, and elsewhere as well for Ps. philander, certainly confirms that similarity (but not the supposed phylogenetic affinity of the two species, as was suggested by Poll). However, the species are readily distinguished by their squamation patterns, especially that on the chest and belly, the sharp size differentiation between chest and ventrolateral flank scales in O. machadoi, the nature of the pelvic fins in that species, and in the different modal lateral-line scale counts for the two species (see p. 215 below). Furthermore, Pseudocrenilabrus philander is dis- tinguished from O. machadoi and all other Cunene haplochromines in having 4 and not 5 lateral-line canal openings in the lachrymal (1st infraorbital) bone. Despite the large amount of O. machadoi material now available, little more can be told about its possible intrageneric relationships (see Greenwood, 1979: 298). The species appears to be the least derived member of the genus, but even that supposition cannot be tested until more is known about its congeners. PSEUDOCREN1LABR US Fowler, 1934 The diagnosis of this genus is based essentially on ethological features associated with the spawning habits of its constituent species. For details of these see Wickler (1963), in which paper he also erected the genus Hemihaplochromis, now treated as a synonym of Pseudo- crenilabrus (see Trewavas, 1973: 33-36 for a full account of the taxon's nomenclatural history). The sole morphological reflection of these ethological peculiarities is found in the absence of discrete spots, or of ocelli, on the anal fin of adult males. Instead, these markings are replaced functionally and morphologically by an orange or scarlet tip to that fin, a feature not readily discernible in preserved specimens. There are, however, two other features which, in the context of the Cunene haplochromines, serve to identify the genus, viz, the presence of only four openings in the first bone of the preorbital series (i.e. the lachrymal), and a tendency for there to be some, often several, pore-less scales in the lateral-line series. In those rare specimens with the entire lateral-line pored, only the four-pored lachrymal serves for instant diagnosis. The close superficial resemblance of the preserved coloration in Cunene Orthochromis and Pseudocrenilabrus species was commented on above; see also p. 215 below. Currently, three species of Pseudocrenilabrus are recognised, Ps. multicolor (Schoeller, 1903), Ps. ventralis (Nichols, 1928) and Ps. philander (Weber, 1897). Their combined distri- butions extend, latitudinally, from the Nile to Natal, South Africa; no species occurs in north-west Africa. The species-level taxonomy of Pseudocrenilabrus has not been revised for many years, and nothing is known about the phylogenetic relationships of the genus. Three subspecies of Ps. philander have been recognised (see Trewavas, 1936: 73, & 1973: 33-36), of which two, Ps. philander dispersus (Trewavas) 1936 and Ps. p. luebberti (Hilgendorf, 1902) occur within the Angolan region. The former subspecies is found in several rivers, including the 214 P. H. GREENWOOD Cunene (Poll, 1967), but the latter is apparently restricted to sink-holes in the neighbourhood of Otavifontein, Namibia. On the basis of one subspecifically diagnostic feature, the length of the premaxillary ascending process, the Cunene fishes I have examined would be referable to the subspecies dispersus, as was the Angolan material examined by Poll (1967). However, the Cunene material has a modal dorsal fin ray spine count of 15, and includes some specimens with 16 spines. On that character it should be referred to the subspecies luebberti. Since these features appear to be the only trenchant ones on which the subspecies can be recognised (Trewavas, 1936: 7, and personal observations), I would argue that there is little to be gained from their formal recognition. A complete taxonomic overhaul of the genus is required, a revision that must take into account coloration and ethological features as well as anatomical ones, and must be based on numerous specimens from many localities. Until that revision is completed, I would also defer any decision on the validity of the third sub- species, Ps. philander philander (Weber) 1897, a taxon apparently restricted to Natal and Mozambique. For these various reasons, and for others noted by Trewavas (1973: 33), the Angolan populations described here are recognised simply as Ps. philander. Pseadocrenilabrus philander (Weber), 1 897 SYNONYMY. See Trewavas (1936: 73) DESCRIPTION. Based on 23 specimens from three localities (see p. 216). Since only ten fishes are undistorted the morphometric analysis is derived from those specimens alone, but meris- tic data were taken from the whole sample. Also, because of the small sample size, only ranges are given for morphometric characters. Where there are indications of populational differences in certain features, or where these fishes differ from those described by Poll (1967), comments are given between square brackets. Poll's material did include some specimens from the Cunene, but most came from the Cuango, Cuilo and Cassai river systems. Depth of body 32-9-39-0% of standard length, length of head 34-2-38-8%. Dorsal head profile sloping at an angle of 40°-45° to the horizontal, straight or gently curved from the nape to a point above the anterior orbital margin, then more strongly curved below that point. Preorbital depth 12-9-18-2% of head length, the least interorbital width 19-2-25-8%. [Fishes from Jamba bridge, Cutato river, Cubango drainage, have a wider interorbital, 23-1-25-8, mean 21-6% of head, than those from the Cunene river at Calueque and from near Cafu; the range given by Poll for his entire sample is 18-1-24-6%.] Snout 0-8-1-0 times as long as broad, its length 23-1-33-3% of head length. Eye diameter 30-8-36-4% of head, cheek depth 16-7-25-0%. Caudal peduncle 1-1-1-4 times longer than deep, its length 14-5-16-7% of standard length. Mouth horizontal or almost so, the jaws equal anteriorly when the mouth is closed, the lips slightly thickened. The posterior tip of the maxilla reaches a vertical through the anterior orbital margin, or very nearly so. Lower jaw 33-1-39-9% of head, 1-2-1-6 times longer than broad. The ascending processes of the premaxilla are 23-1-31-6% of head length. Gill-rakers and pharynx. There are usually 8, rarely 9, short, stout and relatively com- pressed gill-rakers in the outer series on the lower part of the first gill-arch [Poll's count is 8-10, mode 8]. Microbranchiospines are present and obvious. The pharyngeal epithelium is not greatly thickened, nor is it deeply folded and papillose; the prepharyngeal pads are moderately developed. Scales. There is a very gradual size transition between the scales on the chest and those on the ventrolateral aspects of the flanks and the ventral surfaces of the belly. The chest scales, although smaller than the belly scales, are not markedly smaller. CUNENE RIVER HAPLOCHROMINE SPECIES 2 1 5 The lateral line-series has 27-30 (mode 28) scales. [In specimens from the Cutato river, 6-1 1 scales in the upper lateral-line, and 9-1 1 in the lower line, are without pores. Specimens from the two Cunene localities (see p. 216) have all the upper lateral-line scales pored, but from 2-7 scales in the lower line may lack openings. Poll makes no comments on this feature.] There are 4-5^ scales between the upper lateral-line and the dorsal fin origin, and 5 or, more frequently 6, between the pectoral and pelvic fin bases. The cheek has 3 rows of large scales which either cover the entire area or have a small naked embayment anteroventrally [Poll gives the range of cheek scale rows as 3-5]. There is never more than one large and one small scale between each of the last 8-14 upper lateral-line scales and the base of the dorsal fin [Poll makes no comments on this feature]. Fins. Dorsal with 14, 1 5 or 16 (mode 1 5) spinous and 8, 9 or 10 (modally 9 or 10) branched rays, anal with 3 spinous and 7-9 (mode 8) branched elements [The range given by Poll is: dorsal 13-15 spinous and 9-1 1 branched rays, modes 14 and 10 respectively; anal 3 spinous and 7-9, mode 8, branched rays.] No anal sheath scales were observed. Caudal fin rounded, scaled on its basal quarter to third. Pectoral fin 22-5-25-4% of stan- dard length, 61 -5-69-2% of head length. Pelvics with the first branched ray slightly, but obviously longer than the second ray. Teeth. In the outer row of both jaws the teeth are relatively stout and unequally bicuspid. The minor cusp lies at a slight angle to the broad-based major cusp; posteriorly in the upper jaw the last few teeth usually are unicuspid. There are 28-36 teeth in the outer row of the premaxilla [Poll gives a range of 35 to 63 teeth in this row, and indicates that the number is positively correlated with the fish's length; his sample included specimens longer than any recorded above]. Inner row teeth are tricuspid, with the middle cusp noticeably larger than the others, and are arranged in a single row in each jaw (except for one specimen which has a double row anteriorly and laterally in the lower jaw). Lower pharyngeal bone and dentition. The shaft of the bone is relatively longer than that in any of the Cunene river or other Thoracochromis species; its length is contained about 1^ times in the length of the median tooth row (c/l|-2 times in Thoracochromis species). This feature contributes to the less attenuated appearance of the bone when it is compared with that of a Thoracochromis specimen. The dentigerous surface is triangular and equilateral; the teeth are slender, compressed and cuspidate, those of the two median series being only a little coarser than the teeth in the lateral rows [Poll describes the teeth in his material as being 'conique', but I suspect this is an error]. Osteology. No skeletal material has been prepared. Vertebral counts (made from radio- graphs of the Cunene specimens) are: 25 (f 1), 26 (f2), 27 (f3) and 28 (f2), comprising 12 (f5) or 1 3 (f 3) abdominal and 1 3 (f 1 ), 1 4 (f 3) or 1 5 (f4) caudal elements. It has proved impossible to produce radiographs suitable for observing the extent, if any, of hypural fusion and apposition. Coloration. Only preserved colours are known for this material. Superficially, the color- ation and colour pattern in Ps. philander closely resemble those of Orthochromis machadoi (see p. 211). They differ, however, in a number of details. The vertical bars on the flanks and caudal peduncle are solid; that is, the entire exposed surface of the scales underlying a bar is pigmented and not, as in O. machadoi, only the scale margins. Consequently, the diamond-mesh pattern so characteristic of O. machadoi is absent in Ps. philander. The bars in Ps. philander are less intense than those in O. machadoi, except in the region where each is expanded antero-posteriorly to form a faint and horizontal mid-lateral stripe; this res- tricted area of intensity results in Ps. philander having a more intense and discrete mid-lateral stripe. Similarly the upper longitudinal stripe is more distinct in Ps. philander than in O. machadoi, at least over the anterior half of the stripe's course. The species also differ in having the spot at the caudal fin base in Ps. philander less elongate and more distinct than that in O. machadoi. 216 P. H. GREENWOOD The dorsal fin coloration in Ps. philander differs in having light spots on a dark background, the spots covering the membrane between the spines, and also in having light, slightly curved and narrow bands sloping obliquely across the dark membrane of the branched dorsal fin; the bands extend, with regular spacing, to the fin's posterior tip. The caudal fin also is banded, light on dark, but with the bands arranged vertically. The anal, pelvic and pectoral fins are hyaline, but the anal may be faintly banded by darker stripes running obliquely ventro-dorsad across most of its area. This anal banding is probably confined to males, or at least is more obvious in that sex; it seems to be the only sexually dimorphic feature apparent in the preserved coloration, apart from the faint, light spot at the posterior ventral tip of the anal which is sometimes visible in males. Like Orthochromis machadoi, Pseudocrenilabrus philander has a prominent, but narrow, dark lachrymal, stripe; it lacks, however, the well-defined supraorbital continuation seen in O. machadoi. Unlike O. machadoi there are no transverse bands on the snout, which is dark grey in Ps. philander. DISTRIBUTION. Widely distributed in southern Africa; for Angolan localities, see Poll (1967), and below. DIAGNOSIS AND AFFINITIES. See discussion on pp. 213-214 above. Study material and distribution records Museum register number Locality BMNH; P = collection no. 1984.2.6:64-74 Cunene R., pump station at Calueque (17° 16 'S, 14° 30 'E). 1984.2.6:75-76 Cunene R., near Cafu, Angola (16° 30'S, 15° 10'E). 1984.2.6:77-84 Cutato R., Jamba bridge, Angola (Cubango drainage) 1984.2.6:85-86 Okovango R., at Callindo, Angola SERRANOCHROM1S Regan, 1920 The Serranochromis generic concept was redefined and expanded by Greenwood (1979: 299 et seq) to incorporate a number of species previously included in Haplochromis. These latter species, some of which had, at an earlier date (Regan, 1920) been placed in the genus Sargochromis, are now considered to form a subgenus of Serranochromis. Reasons for uniting these various taxa phylogenetically are given by Greenwood (op. cit.), but an additional synapomorphy has been found as a result of comparative studies related to the Cunene representatives of both subgenera. This synapomorphy is the presence, in all species and most individuals, of two to four enlarged median teeth in the first row of the inner premaxillary series; these enlarged teeth are displaced a little anteriorly relative to the others in the row, and so come to lie between the first inner series and the outer tooth row (see figs 4, 9 & 13 in Trewavas, 1964). The majority of specimens I have examined show this displacement quite distinctly, although in a few it may appear only as an obvious irregularity in an otherwise uniformly curved tooth row. SERRANOCHROMIS (SARGOCHROMIS) Regan, 1920 For a description and diagnosis of the subgenus, and a list of its constituent species, see Greenwood (1979: 303-305). Considerable difficulty was experienced when attempting to identify the nine CUNENE RIVER HAPLOCHROMINE SPECIES 2 1 7 Serranochromis (Sargochromis) specimens represented in the Penrith collection. Seven of these specimens were collected in the Cunene river, and two are from an affluent of the Cubango river. The latter apparently represent an undescribed species. The most recent species-level revision of the taxa involved is that by Bell-Cross (1975), who recognised seven species. Only two of these, S. (Sarg.) giardi (Pellegrin) and S. (Sarg.) coulteri (Bell-Cross) are recorded from the Cunene system, with the latter taxon endemic to it. Judging from Bell-Cross' descriptions, all seven Serranochromis (Sargochromis) species are identifiable, when alive, by their distinctive adult male coloration. For preserved material Bell-Cross provides a key employing what, apart from colour differences, seem to be the principal diagnostic features of the taxa. Using this key all the species, either individually or as small groups, should be determinable when, in various combinations, the form of the neurocranial pharyngeal apophysis, the morphology of the lower pharyngeal bone, the nature of its dentition and the shape of the head, are taken into account. Members of species groups, however, are usually separable only on the basis of their geographical distribution. For all species, the ranges of most meristic and morphometric characters show considerable overlap, although there is sometimes a distinction to be found in the mean or modal values for certain features. Such fine distinctions cannot, of course, be utilized when, as in the case of the present collection, only a few specimens are available for study. From Bell-Cross' key there would seem to be little difficulty in distinguishing between the two species he recognised as occurring in the Cunene, namely S. (Sarg.) giardi and S. (Sarg.) coulteri. The former is characterized by its massive lower pharyngeal bone with an exten- sively molarized dentition, and its characteristically 'butterfly'-shaped pharyngeal apophysis (Bell-Cross, 1975: fig. 1 & pp 450-454; also p. 456). In sharp contrast, S. (Sarg.) coulteri has a weakly developed lower pharyngeal bone, the teeth of which are '. . . sharp and pointed' (Bell-Cross, 1975: 455; elsewhere (p. 429] the teeth in the median rows are described as '. . . slightly enlarged and not molariform'). The pharyngeal apophysis in this species does not depart from the modal Haplochromis-type (see Bell-Cross, 1975: fig. 1; also Greenwood, 1978: 303). I would certainly confirm Bell-Cross' description of the S. (Sarg.) giardi pharyngeal bone and dentition, but must disagree with his account of those features in S. (Sarg.) coulteri. Using the material (both whole specimens and skeletal preparations) on which Bell-Cross (1975: 426-431) based his description of S. (Sarg.) coulteri I find that, apart from the smallest fish in the size range it covers (105-214 mm SL), at least the two median tooth rows, and often most teeth in the next two lateral rows, are composed of enlarged teeth with flat, molar-like crowns and cylindrical necks (Figs 12-15). Morphologically, these teeth are quite distinct from the compressed, shouldered and thus virtually bicuspid teeth occurring in the lateral and posterior dentigerous fields of the bone. Only in the smallest specimen (105 mm SL, BMNH 1975.6.19: 1-13; DA68) do the median teeth retain slight indications of a minor cusp, and a discrete posterior cusp (Fig. 13). In other words, apart from their manifestly coarser appearance, the median teeth in this specimen are like those situated posteriorly and medio-laterally on the bone. This specimen alone, would accord with the description Bell-Cross gives for the pharyngeal dentition in S. (Sarg.) coulteri. Thus, I cannot agree with Bell-Cross' description, in his key, of the pharyngeal teeth in S. (Sarg.) coulteri as being 'sharp and pointed' (even the lateral teeth are bicuspid), and would consider his description in the text ('teeth slightly enlarged') to be an understatement, at least with respect to fishes more than 105 mm SL. Likewise, I would argue against his statement (loc. cit.) that the crowns are 'not molariform'; they are molariform in large specimens, and could be described as 'submolariform' in all others except the smallest fish. With that correction made, the situation regarding the specific identification of preserved Serranochromis (Sargochromis) material is rendered more difficult. The lower pharyngeal dentition in S. (Sarg.) coulteri is, in fact, like that in many specimens of supposed S. (Sarg.) codringtoni and S. (Sarg.) mellandi, and there are no trenchant meristic or morphometric features which can be used to separate elements of the trio. 218 P. H. GREENWOOD Fig. 12 Serranochromis (Sargochromis) coulteri. Lower pharyngeal bone of the holotype, a specimen 216 mm SL from the Cunene river system. Scale in mm. An examination of several specimens identified either as codringtoni or mellandi shows that there is a wide range in the degree to which the lower pharyngeal dentition is molarized, and a correlated variation in the degree to which the bone is enlarged. In some specimens both bone thickening and tooth molarization exceed that found in S. (Sarg.) coulteri, but in many others there is complete overlap with the conditions found in that species. In one respect, overall tooth shape, the enlarged teeth in S. (Sarg.) coulteri do seem to differ from those in the other two species. Whereas the teeth in 5. (Sarg.) mellandi and S. (Sarg.) codringtoni are short and broad, those in S. (Sarg.) coulteri are relatively taller and more slender, features particularly obvious in smaller specimens. Also, in S. (Sarg.) coulteri the posterior horns of the pharyngeal bone appear to be rather more slender than those in the other two species, even when the dentition and bone itself are at comparable levels of hypertrophy. On the basis of overall pharyngeal tooth morphology, I would therefore refer, at least tentatively, four Cunene river specimens (76-5, 98-0, 130-0 and 132-0 mm SL respectively to S. (Sarg.) coulteri; for detailed distribution records see p. 230. These presumed S. (Sarg.) coulteri differ in some respects from the specimens described by Bell-Cross (1975), but it should be recalled that two are smaller than any of the specimens he examined. In two of the new specimens (76-5 and 98-0 mm SL) eye diameter is larger (30-3% of head length), and the snout is shorter (32-1 and 33-0% head in the specimens CUNENE RIVER HAPLOCHROMINE SPECIES 219 Fig. 13 Serranochromis (Sargochromis) coulteri. Lower pharyngeal bone from a specimen 105 mm SL (1975.6.19: 1-3; DA68, ex Cunene river system). Scale in mm. respectively); the four specimens have slightly shorter lower jaws (33-9-37-0% head cf 31 -4-42-6% in the Bell-Cross material). All other morphometric features lie within the ranges given by Bell-Cross. In some meristic characters the new material also lies outside the ranges given by Bell- Cross (1975: table in appendix (iii)). Three of the four specimens (including the two smallest) have 12 or 13 gill-rakers on the first arch (9-11, mode 10, according to Bell-Cross), and all have higher lateral-line scale counts (33 or 34, c/30-31, mode 31). Clearly some of these discrepancies might be attributable to personal differences in the way counts and measurements were made, and others could be attributed to the smaller size of two specimens I examined. Parenthetically it should be mentioned that Bell-Cross' (1975: 429) statement that the first branched pelvic ray in S. (Sarg.) coulteri reaches the origin of the anal fin, does not hold for specimens other than the type; probably the length of this ray is correlated with an individual's sex and, in the case of adult males, with the level of sexual activity. Also, it should be noted that the photograph of the type specimen reproduced in Bell-Cross (1975: plate 4) was taken before the fish was set and preserved. The distended mouth shown in the picture distorts the dorsal head profile which, in the preserved specimen, is like that in most other Serranochromis (Sargochromis) species; see, for example plates 5, 6 and 8 in Bell-Cross (1975). Of the remaining new Serranochromis (Sargochromis) specimens, one large fish (270 mm 220 P. H. GREENWOOD Fig. 14 Serranochromis (Sargochromis) coulteri. Lower pharyngeal bone from a specimen 139 mm SL (1975.6.19: 1-3; DA59, ex Cunene river system). Scale in mm. SL, BMNH 1984.2.8: 1) should, on the basis of its extremely hypertrophied lower pharyngeal bone and the extreme molarization of its teeth (Fig. 16), be identified as S. (Sarg.) giardi (see Bell-Cross, 1975: 451-454). In its general appearance too, particularly the almost rounded head profile, this specimen resembles S. (Sarg.) giardi from localities outside the Cunene system. However, it differs from those fishes in certain morphometric features, as it does from the few Cunene specimens currently identified as S. (Sarg.) giardi. These other Cunene specimens comprise three small fishes (the largest 86-0 mm SL) from Ponang Kuma, Mossamedes (probably Donguena, 17° 03 'S, 14° 40 'E, according to Dr Michael Penrith, in litt). They were previously identified by Boulenger (1915: 408) and by Regan (1922: 263) as Sargochromis angolensis (Steindachner), but were reidentified as Haplochromis giardi by Bell-Cross (1975). The new 270 mm long specimen comes from below the Ruacana falls (17° 24 'S 14° 13 'E) is larger than any of the giardi material examined by Bell-Cross, all of which, save that from Ponang Kuma, came from the Okovango, Zambezi or Kafue river systems. It also differs from those specimens in having a greater preorbital depth (26-1% of head c/19-4-22-3, mean 21-2%), and a longer lower pharyngeal bone (43-2% of head, c/37-6-41-8%). The pharyngeal apophysis, however, has the specifically characteristic 'butterfly' shape described by Bell- Cross (1975: 453 and 41 1; fig. 1), and is of the extreme 'butterfly' type (as might be expected from the great hypertrophy of the pharyngeal mill) which otherwise occurs in fishes from the Zambezi (see Bell-Cross, 1975; fig. 1). If it be assumed that the deeper preorbital bone (lachrymal), antf the longer pharyngeal CUNENE RIVER HAPLOCHROMINE SPECIES 221 Fig. 15 Serranochromis (Sargochromis) coulteri. Lower pharyngeal bone from a specimen 183 mm SL (1975.6.19: 1-13; DA54, ex Cunene river system). Scale in mm. bone in the new Cunene fish are both correlates of that specimen's large size (or are simply examples of individual variability), this fish could be identified as S. (Sarg.) giardi. However, certain other Cunene specimens throw some doubt on that conclusion. One of these specimens is another large fish, 190mm SL (BMNH 1984.2.6: 149), and comes from a locality 45 miles west of Ondurusu Falls (17° 03 'S, 13° 30'E). It too has an hypertrophied lower pharyngeal bone with an extensively molarized dentition, and its pharyngeal apophysis approaches the extreme 'butterfly' type, being intermediate between the Kafue and Upper Zambezi forms illustrated by Bell-Cross (1975, fig. 1). The lower pharyngeal bone, however, proves to be less massive, and its dentition less molarized than in S. (Sarg.) giardi of a comparable size. It is more massive and further molarized than in a 260 mm specimen of S. (Sarg.) codringtoni, but only slightly more massive and molarized than in a 194mm specimen of the same species. Similar results are obtained when the Cunene specimen is compared with examples of S. (Sarg.) mellandi. The 190 mm Cunene specimen, like the 270 mm fish discussed earlier, differs from indi- viduals in other populations of S. (Sarg.) giardi in having a deeper preorbital (27-1% of head, cf 18-0-22-0, mean 19-9% in S. (Sarg.) giardi) but unlike the larger Cunene fish it also differs in having a somewhat shorter lower jaw (34-2% of head, c/36-0-42-6, mean 39-6%). Although the lower pharyngeal bone and dentition in the 1 90 mm Cunene specimen are comparable, except for the bone's greater length (which does lie within the giardi range), with those in some specimens of codringtoni and mellandi, the short lower jaw would seem to exclude the specimen from either of these taxa, as well as from giardi. Its preorbital depth, 222 P. H. GREENWOOD Fig. 16 Lower pharyngeal bone from a giardi-like member of the Serranochromis (Sargochromis) giardi-codringtoni species complex, 270 mm SL, ex Cunene river (1984.2.8: 1). Scale in mm. excessive for giardi is, however, only a little greater than the maximum recorded for either mellandi or codringtoni. The larger (270 mm) specimen, it will be recalled, has a more massive pharyngeal mill than is found in specimens of either mellandi or codringtoni, but that 'gap' is bridged by the bone and its dentition in the 190 mm fish. Lower jaw length in the larger fish, however, lies within the ranges for giardi, codringtoni and mellandi, and its preorbital depth lies within the ranges for codringtoni and mellandi, but outside that for giardi. In other words, on morphometric characters and in the nature of the pharyngeal mill, the two giardi-\\ke fishes from the Cunene seem to show, either in the themselves or by providing bridging features, characters of three Serranochromis (Sargochromis) species, only one of which (giardi) is thought to occur in that river. This situation is by no means clarified when two further specimens, 122 and 190 mm SL, from the Hamburg Museum collections, are taken into account. These fishes (ZMH 1 722, collected in the Cunene at Capelongo) have greatly hypertrophied lower pharyngeal bones with extremely molarized dentitions. On those criteria the specimens fall within the range of variation encountered within S. (Sarg.) giardi, mellandi and codringtoni (Fig. 17) but are perhaps nearest giardi. The length of the bone (33-5% of head length) for the larger of the two Hamburg specimens, however, is below that for giardi of a comparable length, is near but slightly shorter than that for codringtoni, and is well within the range for mellandi. The CUNENE RIVER HAPLOCHROMINE SPECIES 223 Fig. 17 Serranochromis (Sargochromis) codringtoni. Lower pharyngeal bone from a specimen 183 mm SL (1975.6.19: 1-3; ex Kafue river). Scale in mm. length of the bone in the smaller fish, 33-3% of the head, falls within the recorded ranges for comparable sized mellandi and codringtoni, but is below that for giardi of the same length. Preorbital depth in the larger Hamburg specimen (25-0% of head) is somewhat greater than that recorded for giardi, but closely approaches that for codringtoni and mellandi; in the smaller fish, the preorbital depth (21-0% of head) lies within the range for comparable sized giardi, codringtoni and mellandi. Lower jaw length in the larger Hamburg fish (35-8% of head) is slightly below that of giardi, and also below that of codringtoni, but is much shorter than in mellandi; the smaller specimen, however, has a jaw length (38-0% of head) within the ranges for all three species. Taken in concert, the features in the two Hamburg specimens, and those of the two Cunene fishes, certainly seem to break down the principal morpho-anatomical differences between Serranochromis (Sargochromis) giardi and S. (Sarg.) codringtoni, and, indeed, those between these species and S. (Sarg.) mellandi. This conclusion casts doubts on any possibility of identifying the new Cunene specimens with enlarged pharyngeal mills. The small specimens from Ponang Kuma (see above, p. 220) can be identified as giardi on the basis of various 'key' characters. But, in the absence of specimens at sizes intermediate between them and the larger fishes discussed above, even that identification is uncertain. 224 P. H. GREENWOOD Clearly the situation is confused, and is unlikely to be clarified without studying a lot more material, supported by data on breeding coloration, from all areas in which the species giardi, mellandi and codringtoni have been recorded. Until that revision is effected, I would prefer to recognize the Cunene specimens with hypertrophied pharyngeal mills only as members of a Serranochromis (Sargochromis) giardi- codringtoni species-complex, that complex to include 5". (Sarg.) mellandi. Certainly it would be unrealistic to refer the Cunene specimens to any one species in that complex; to describe them as a new species would confuse the issues involved. I suspect that specimens from Lake Calundo, Angola, described by Poll (1967) and identified by him as Haplochromis mellandi, are also members of the ' giardi- codringtoni1 complex. Poll's figure (1967: fig. 50, p. 310), and his remarks about the deep preorbital in these fishes, reinforce my suspicions. Bell-Cross (1975: 436) thought that Poll's specimens should be referred to S. (Sarg.) codringtoni, a taxon which I would include in the complex under discussion. At this point it is appropriate to mention certain type specimens of Boulenger's (1913) species Tilapia steindachneri (see p. 189). The specimens in question (BMNH 1907.6.29: 176-9; from the Donguena swamps) were later referred to Sargochromis mellandi (now Serranochromis (Sargochromis) mellandi] by Regan (1922: 263), a decision with which I would concur, allowances being made for the species-level problems discussed above. All are small fishes (52-0-64-0 mm SL). One, the largest, was illustrated by Boulenger (1915: 210; fig. 1 34) and designated 'Type' in the caption to the figure accompanying this redescrip- tion of the species. That action I treat as the subsequent designation of a type specimen since none was chosen when the species was first described (Boulenger, 1913). The figured specimen and three others from Donguena swamp are easily distinguished from the remaining syntypes of Tilapia steindachneri, collected in the Que river, a tributary of the Cunene. The Que fishes are now referred to Thoracochromis buysi (Penrith); see p. 1 90. The type and three syntypes of Tilapia steindachneri from Donguena swamp have enlarged lower pharyngeal bones, with most teeth in the two median rows enlarged and molariform or submolariform, and thus resemble those in Serranochromis (Sargochromis} codringtoni and S. (Sarg.) mellandi. Considering the small size of these fishes, and the degree to which their lower pharyngeal bones are enlarged, it seems likely that, as adults, they would have relatively massive to massive pharyngeal bones, and a highly molarized dentition. Unfortunately there is no way in which the four specimens can be given a positive specific identification, especially in the light of what is now known about the possible complex of Serranochromis (Sargochromis) species in Angola (see p. 217). Under the circumstances it would seem inadvisable to consider the synonymy of Tilapia steindachneri (in part) with Serranochromis (Sargochromis) mellandi as well established. The question should be left open until the Angolan l giardi- codringtoni' complex is resolved. From that complex could well emerge a taxon which would take the name ' steindachneri" . Finally, attention must be given to another Serranochromis (Sargochromis} specimen amongst the new material from the Cunene river. This fish, 126 mm SL (BMNH 1984.2.6: 154) is apparently referable to the species S. (Sarg.) greenwoodi (Bell-Cross), a taxon not previously recorded from Angola; all other records are from the Upper Zambezi, Kafue and Okavango river systems. Apart from the new species to be described below, S. (Sarg.} greenwoodi is unique amongst the species of this subgenus in having, at least in most popu- lations, a fine lower pharyngeal bone with no noticeably enlarged or molarized teeth (see Bell-Cross, 1975:425). The Cunene specimen, from Matala Dam, Luceque (14° 36'S, 15° 18'E) is an adult male. It has a relatively slender lower pharyngeal bone (length 27-6% of head length) whose dentition is composed mainly of fine bicuspid teeth; a few teeth in the posterior part of the two median rows are somewhat enlarged, but, like the others, are distinctly bicuspid. In all morphometric features, including a deep preorbital bone (27-6% of head), long snout (42-5% of head), long lower jaw (42-5% of head), and long ascending premaxillary processes (34-5% of head), the specimen falls within the range for S. (Sarg.) greenwoodi from other localities. CUNENE RIVER HAPLOCHROMINE SPECIES 225 Its dental and meristic characters are also within the range of that species, and include the high gill-raker count of 1 4. Despite our apparently identical methods for counting the lateral- line scale series (Bell-Cross, 1975: 410), I make the number of scales in S. (Sarg.) greenwoodi 33-36, and not, pace Bell-Cross, 29-31. The Cunene fish has a count of 33. Although this fish resembles S. (Sarg.) greenwoodi in all features ascertainable from a single, preserved specimen, its precise status will be uncertain until more material is available, and data are obtained on the live colours of breeding males. The two remaining Serranochromis (Sargochromis) specimens in the Penrith collection were obtained from the Cubango river drainage basin. They are conspecific, but do not seem referable to any of the species or populations so far described. I am loath to create a new taxon on only two specimens, especially since the taxonomy of the subgenus Sargochromis is in such an uncertain state. However, the specimens are very distinctive and thus would seem to justify their recognition as members of a new species. Serranochromis (Sargochromis) gracilis sp. nov. (Fig. 18) HOLOTYPE. An adult female, 1 16-5 mm standard length, from the Cutato river at Jamba bridge (Cubango drainage), Angola; BMNH 1984.2.6: 147. Paratype: an adult female 118-0 mm SL, from the same locality; BMNH 1984.2.6: 148. The trivial name, from the Latin, meaning slender or simple, refers to the body proportions of the type specimens, and to the relatively unspecialized nature of the pharyngeal dentition. DESCRIPTION. Based on the two types specimens only. Depth of body 33-1 and 34-3% of standard length, length of head 36-9 and 37-3%. Dorsal head profile gently curved and sloping at an angle of ca 30°-35° to the horizontal. Preorbital depth 18-6 and 19-3 of head length, least interorbital width 18-2 and 18-6%. Snout 1-1 and 1-2 times longer than broad, its length 31-8 and 33-7% of head. Eye diameter 25-0 and 25-6% of head, depth of cheek 22-7 and 23-3%. Fig. 18 Serranochromis (Sargochromis) gracilis. Holotype. Drawn by G. J. Howes. Scale = 18 mm. 226 P. H. GREENWOOD Caudal peduncle 1-4 and 1-5 times longer than deep, its length 15-4 and 16-1% of standard length. Mouth very slightly oblique, sloping at an angle of ca \ 5° to the horizontal; lips slightly thickened, jaws equal anteriorly when the mouth is closed. Posterior tip of the maxilla reach- ing a vertical through the anterior orbital margin; premaxilla not beaked anteriorly, its ascending processes breaking, slightly, the dorsal outline of the head, their length 31-8 and 32-5% of the head. Gill-rakers. Ten in the outer row on the lower part of the first gill arch, the lowest one or two rakers greatly reduced, the following 5 or 6 either stout and short (holotype) or rela- tively slender, the uppermost 3 rakers either flattened, with the crown produced into 2 or 3 cusps, or flattened and anvil-shaped (holotype). Microbranchiospines are present. Scales on the anterior part of the body below the lateral-line are weakly ctenoid, but are cycloid above that level. Scales on the posterior part of the body are cycloid. When ctenoid, the scales have the cteni distributed over most of the exposed parts. The chest scales are not noticeably small, and have a gradual size gradient with those on the belly and ventrolateral aspects of the flanks. Lateral-line with 34 scales, cheek with 4 rows of large scales which cover the area except for a small naked embayment anteroventrally. There are 5 scales between the dorsal fin origin and the lateral-line, 7 between the pelvic and pectoral fin bases. Only the last (holo- type), or the last two, pored scales of the upper lateral-line are separated from the dorsal fin base by one large and one small scale, the others being separated from the fin by at least 2 large scales of almost equal size. Fins. Dorsal with 15 spinous and 13 branched rays, anal with 3 spines and 10 branched elements. No anal sheath scales are present in either specimen. Pectoral fin 22-3 and 22-9% of standard length. Caudal subtruncate, scaled over its basal two-thirds or three-quarters. First branched pelvic ray not reaching the vent. Teeth. The outer row in both jaws is composed of relatively slender and compressed teeth. In most the crown has a sharp, fine point and a low, laterally placed shoulder; other teeth in the row are more distinctly bicuspid, the shoulder being replaced by a discrete minor cusp. Posteriorly, there is a short edentulous region on the premaxilla. About 50 and 54 teeth are present in the premaxillary outer row. The inner teeth are mostly tricuspid or weakly bicuspid, but a few unicuspids occur anter- iorly; the teeth are arranged in a single row in both jaws, that of the upper jaw extending posteriorly beyond the anterolateral section of the bone's dentigerous surface. Medially in the upper jaw there are two enlarged teeth situated between the inner and outer tooth rows. Lower pharyngeal bone and dentition (Fig. 19): The bone is not enlarged, its teeth are distinctly cuspidate, with the minor cusp present as either a well-demarcated shoulder or a discrete cusp. Teeth in the two median rows are manifestly coarser and larger than those situated laterally and postero- laterally. The dentigerous area is a little longer than broad, giving it a more nearly isoscelene than equilateral outline. The length of the bone, measured in one specimen, is 28-0% of the head length. Osteology. No skeleton is available but both specimens were radiographed. Excluding the fused PU, and tl, centra, there are 31 vertebrae, comprising 15 abdominal and 16 caudal elements. In the one specimen dissected the neurocranial pharyngeal apophysis is of the typical Haplochromis-lype, although the basioccipital contributions to the facet are not extensive. Judging from the radiographs, the neurocranium has a relatively protracted ethmovomer- ine region. Comparison with figures in plates 1 & 2 of Bell-Cross (1975) suggests that of the skull of S. (Sarg.) gracilis is nearest that of S. (Sarg.) greenwoodi, a resemblance confirmed by comparisons with actual skulls. Coloration. Live colours are unknown. The two formol preserved and alcohol fixed speci- mens have a pale beige ground coloration which darkens dorsally and lightens ventrally; the dorsal surface of the head and snout is greyish. Traces of up to eight vertical bars cross the flank and caudal peduncle, and are most obvious over the midlateral surfaces of the flanks; CUNENE RIVER HAPLOCHROMINE SPECIES 227 Fig. 19 Serranochromis (Sargochromis) gracilis. Lower pharyngeal bone. Scale in mm. dorsally the bars merge with the darker ground colour, ventrally they terminate along a hori- zontal line drawn posteriorly from the base of the pectoral fin. A pronounced opercular spot is present, but no other cephalic markings are apparent. The entire dorsal fin is densely flecked with brownish to reddish streaks and elongate spots, the markings becoming more discrete on the soft parts of the fin; the lappets to the spinous dorsal have a similar dark pigmentation. The caudal fin, to about its distal quarter, is darkly spotted with elongate ovoid maculae; the distal quarter of the fin is hyaline, with traces of a dark but narrow marginal band, more obvious in one specimen than the other. The pectoral, pelvic and anal fins are hyaline; the soft part of the anal is lightly maculate, the spots fairly regularly arranged in two rows. Breeding. Both specimens are adult females at an advanced stage of oogenesis; the left and right ovaries are equally developed. DISTRIBUTION. Known only from the Cutato river, Angola. DIAGNOSIS. The relatively slender, unthickened lower pharyngeal bone of S. (Sarg.) gracilis, with its distinctly bicuspid teeth, sets the species apart from all members of the subgenus except S. (Sarg.) greenwoodi and some individuals of -S. (Sarg.) coulteri (see p. 217). The compressed outer row teeth in the jaws, the persistence in that row of bi- and weakly bicuspid teeth in specimens over 100mm SL, and the persistence in such individuals of numerous tricuspid teeth in the inner tooth rows, distinguish S. (Sarg.) gracilis from both 5". (Sarg.) greenwoodi and S. (Sarg.) coulteri, and from the other species as well. The gently sloping dorsal head profile, and the shallow body of S. (Sarg.) gracilis are further diagnostic features. More specifically, S. (Sarg.) gracilis is distinguished from the two species with fine pharyngeal bones and dentition as follows: From S. (Sarg.) greenwoodi by its shorter pectoral fins (22-3-22-9% SL c/25-6-29-2, mean 27-7%), shorter snout (3 1 -8-33-7% head, c/37-4-40-7, M = 39-3%), shallower preorbital bone (18-6-19-3% head, c/25-3-27-3, M = 26-3%), narrower interorbital width (18-2-18-6% head, cf2 14-1 3-9, M = 22-8%), and by its fewer gill-rakers (10 cf 12-1 5, mode 15). From S. (Sarg.) coulteri it is distinguished, apart from differences in the pharyngeal bones, by its shorter pectoral fins (22-3-22-9% SL, c/25-4-29-3, M = 27-3%), much shorter snout (31-8-33-7% head, c/36-9-40-4, M = 38-7%), slightly shallower preorbital depth (18-6-19-3% head, cf 20-6-22-1, M = 21-4%), and somewhat longer lower jaw (44-2-46-6% head, cf 37-4-42-6, M = 39-6%). 228 P. H. GREENWOOD AFFINITIES. It is difficult to suggest any phylogenetic relationships for S. (Sarg.) gracilis within the Sargochromis assemblage. In part this is because virtually no anatomical infor- mation is available for the new species, and in part because it seems to share no uniquely derived features with any other member of the group. The characters which it does share with 5. (Sarg.) greenwoodi and S. (Sarg.) coulteri, in particular those shared with the former species, appear to be plesiomorphic ones associated with the generalized type of pharyngeal jaws present in both taxa. For the moment S. (Sarg.) gracilis can only be considered a possible candidate for sister-species relationship with S. (Sarg.) greenwoodi. If that were established, the two species would then constitute a sister group to all other Serranochromis species. The relationships of S. (Sarg.) coulteri, because of its relatively derived pharyngeal dentition, would be with other members of the subgenus. SERRANOCHROMIS (SERRANOCHROMIS) Regan, 1920 For a description and diagnosis of the subgenus see Greenwood (1979: 299 et seq.). Two species, Serranochromis (Serranochromis) thumbergi (Castelnau) and S. (S.) macro- cephalus (Blgr) are represented in the Penrith collection. Several of the specimens were badly distorted in preservation, but it has been possible to identify them by using a combination of various and mutually exclusive characters. Because of their poor preservation no detailed descriptions of the species can be given. The material differs little from that described by Trewavas (1964) from other localities, but where differences were observed, or could be observed, these will be noted. In the distribution maps published in her monograph, Trewavas (1964) records S. (S.) thumbergi and S. (S.) robustus jallae from the Cunene (op. cit.: fig. 26), and S. (S.) macro- cephalus, with S. (S.) angusticeps, from a short, isolated and westward flowing river north of the Cunene (op cit. figs 27 & 28 for the species respectively). This river opens to the sea near the town of Mossamedes. Since the Angolan material which Trewavas lists for the two latter species (op. cit.: 34 & 40) bears only the locality 'Mossamedes', her indication of their occurrence near that town seems reasonable enough (as would, for the same reason, her record of S. (S.) robustus jallae in the same river). However, according to Dr Penrith (in lift). . . 'A problem with some early collections, especially Ansorge's, has been the use of Mossamedes. This could refer to the town of that name; it could and usually did, refer to the district, a district that in the nineteenth century comprised most of southern Angola. Maps show two rivers flowing into the sea near Mogamedes, the Bero and the Giraul, and a third, the Curoca, entering the sea slightly further south. None of these rivers is perennial, and with the exception of some isolated stretches are dry for much of the year. Most older references to "Mossamedes" therefore probably refer to the Kunene river'. The same doubt must affect the presumed localities for certain Angolan S. (S.) robustus specimens. These were also collected by Ansorge from 'Mossamedes', and on Penrith's argu- ments could have eome from the Cunene river, thus casting doubt on Trewavas' record of the species in the small river north of the Cunene. Other S. (S.) robustus material collected by Ansorge, however, is from Donguena, and therefore is definitely attributable to the Cunene river system, as is the Hamburg Museum specimen (ZMH 1718) from Miilongo- Fiirt. Possibly we should accept with reservation the presence of S. (S.) robustus, S. (S.) macrocephalus and S. (S.) angusticeps in the small northern river indicated on Trewavas' (1964) maps until their presence is confirmed by further collections. No specimens identifiable as 5". (S.) angusticeps are included in the new collection, but the identification of two Cunene river specimens, collected by Ladiges in 1959 (ZMH 1300 & 1 307), as S. (S.) angusticeps can be confirmed. Serranochromis (Serranochromis) thumbergi (Castel.) For the species as a whole, Trewavas (1964: 24 & 26) described the coloration of specimens CUNENE RIVER HAPLOCHROMINE SPECIES 229 preserved in alcohol. She noted that vertical markings on the body may be absent or, if present, are much fainter than the longitudinal ones, and are confined to the upper part of the body. Two of the four Cunene specimens (168-183 mm SL; all from the Matala Dam, Luceque) have the vertical bars predominating, while the two other fishes have the longi- tudinal midlateral stripe as the dominant component. In one of the latter specimens, both the vertical and horizontal markings are faint, and equally so. The total vertebral counts (excluding the PU, and U, centra) in all four specimens is 36, comprising 18 abdominal and 18 caudal elements in three specimens, and 19 and 17 centra respectively in the fourth fish. These figures are in agreement with those given by Trewavas. The identity of two S. (5). thumbergi specimens in the Hamburg Museum (ZMH 1719) was confirmed. Both were collected by Ladiges from the Cunene river at Capelongo, Angola (14° 55 'S, 15° 06 'E). In one, the horizontal markings of the colour pattern predominate, but in the other, the horizontal and longitudinal components are equally intense. Serranochromis (Serranochromis) macrocephalus (Blgr) Only one large specimen (160mm SL; from Luceque) is represented in the collection, the remaining 21 individuals are much smaller (31-1 10 mm SL) and come from a number of different localities (see below, p. 230). It is regrettable that the majority of small specimens are so badly distorted as to render them unsuitable for morphometric analysis. Little information is available on allometric and other growth changes in any Serranochromis species. Total vertebral counts (excluding PU, and Uj) in the specimens radiographed, which in- cluded the largest and the smallest fish, are 32 (fl 1) and 33 (10), comprising 15 (fl 1) or 16 (flO) abdominal and 16 (f4), 17 (f!3) or 18 (f4) caudal elements. Trewavas (1964: 29) gives the total counts in this species as 31-33, and the range for abdominal centra as 15 or 16; the only caudal count she records is 1 7 (all these figures are adjusted from Trewavas' original counts so as to exclude PU, and U, centra). The low number of abdominal vertebrae in S. (S.) macrocephalus, and hence the low total count, serves as a further feature distinguishing this species from the superficially similar S. (S). robustus jallae. One of the specimens examined (from a locality 45 miles west of Ondurusu Falls) is unique amongst all the Serranochromis (Serranochromis) specimens in the collection in having anal sheath scales present (see p. 188 above). Serranochromis (Serranochromis) angusticeps (Blgr) and S. (S.) robustus jallae tBlgr) As noted earlier, neither of these species is represented in the new collection. However, the presence of both species in the Cunene river has been confirmed on the basis of specimens in the Hamburg Museum collections (see p. 189). Zoogeographical considerations Before considering what light the new material might throw on the zoogeography of the Cunene river fish fauna, attention must be given to four species which Poll (1967: 23) lists as occurring in that river. The species involved are Haplochromis darlingi (Blgr), H. angp- lensis (Steindachner), H.frederici (Castelnau) and H. mellandi (Blgr). The .last named species has been discussed already in connection with the Cunene Serranochromis (Sargochromis] (p. 224). The presence of Haplochromis darlingi (now Pharyngochromis\ see Greenwood, 1979: 310), a species otherwise known only from the Zambezi, is based on Poll's redetermination of specimens first identified by Pellegrin (1936: 60) as Pelmatochromis welwitschi. Poll (op. cit.) also identified several more specimens from various Angolan rivers as H. darlingi. His 230 P. H. GREENWOOD Study material and distribution records for Serranochromis (Sargochromis) species Museum register number Locality BMNH; P = collection no. giardi-codringtoni complex 1984.2.8:1 1984.2.6:149 P 984 Cunene R., below Ruacana falls (17° 24' S, 14° 13 'E). Cunene R., 45 miles west of Ondurusu falls (17° 03 ', 13° 30 '£). coulteri: 1984.2.6:150 P809 Cunene R., nr Cafu ( 1 6° 30 ' S, 1 5° 1 0 ' E). 1984.2.6:151 PI 094 Cunene R., Matala dam at Luceque (14° 36 'S, 15° 18 'E). 1984.2.6:152 P1095 Cunene R., Matala dam at Luceque (14° 36' S, 15° 18'E). 1984.2.6:153 PI 120 Locality unknown. greenwoodi: 1984.2.6:154 Cunene R., Matala dam at Luceque (14° 36 'S, 15° 18'E). Study material and distribution records for Serranochromis (Serranochromis) species Museum register number Locality BMNH; P = collection no. thumbergi: 1984.2.8:2 P1551 1984.2.8:3. P1091 1984.2.8:4 P1093 1984.2.8:5 P1096 macrocephalus: 1984.2.8:6-12 P665 1984.2.8:13-17 P669 1984.2.8:18 P670 1984.2.8:27 P899 1984.2.8:19-20 P808 1984.2.8:21 P1092 1984.2.8:22-23 PI 116 1984.2.8:24-26 PI 179 Cunene R., at Luceque (14° 40 'S, 1 5° 07 'E). Cunene R., Matala dam at Luceque (14° 36 'S, 15° 18'E). Cunene R., Matala dam at Luceque (14° 36' S, 15° 18'E). Cunene R., Matala dam at Luceque (14° 36' S, 15° 18'E). Cunene R., 45 miles west of Ondurusu falls (17° 03 ', 13C Cunene R.,45 miles west of Ondurusu falls ( 17° 03 ', 13C Cunene R., 45 miles west of Ondurusu falls (17° 03^', 13C Cunene R., Calueque (17° 16 'S, 14° 30 'E). Cunene R., Calueque ( 17° 16'S, 14°30'E). Cunene R., Matala dam at Luceque (14° 36 'S, 15° 18'E). Cunene R., Chiatapu (14° 23'S, 15° 18'E). Cunene R., Jamba-ia-Homa (1 3° 46 'S, 1 5° 30 'E). 30 'E). 30 'E). 30 'E). CUNENE RIVER HAPLOCHROMINE SPECIES 231 description of these fishes, in particular the presence of large anal spots in males, led me to express some doubts about their true identity (Greenwood, 1979: 311). Now, having examined some of Poll's material (from Lake Calundo; MCA: 163987-986), these doubts are reinforced. Certainly the specimens do resemble Ph. darlingi in some respects, but the anal fin markings are unlike those of Zambezi Ph. darlingi, while the lower pharyngeal bone and dentition in the Angolan fishes are, respectively, less well-developed and less molarized. Thus, at least until more is known about intraspecific variation in Zambezi darlingi, and until the live coloration of specimens from Angola and elsewhere is recorded, I would defer any inclusion of Ph. darlingi amongst the haplochromine species of Angola. The Angolan 'darling?, I suspect, probably represents an undescribed species distinct from that in the Zambezi, and one whose phylogenetic and hence generic relationships are at present indeterminable. Poll's inclusion of Haplochromis angolensis in the Cunene fauna stems from Boulenger's (1915: 409) identification of three specimens from Mossamedes as that species. It is these specimens which Bell-Cross (1975: 451) reidentified as Haplochromis giardi, and which are discussed on page 220 above. The single type specimen of Steindachner's angolensis is now lost (see Bell-Cross, 1975: 426) and the true identity of the taxon is unlikely to be determined because the original description is totally inadequate for that purpose. The inclusion of ango- lensis in the Cunene faunal list would, therefore, seem to be rendered null and void. As a result of Bell-Cross' reidentification of the specimens involved, the record for H. angolensis should be replaced by one for Serranochromis (Sargochromis) giardi. But, as discussed on p. 224 above, there are certain doubts about the specific identity of giardi-\\\x fishes in the Cunene. The record for Haplochromis frederici (now Serranochromis [Sargochromis] greenwoodi, see Bell-Cross, 1975, and Greenwood, 1979) was presumably based on Ladiges' (1964: 268) identification of four fishes from Capelongo as H. frederici. I have examined these specimens (ZMH 1722), and find that two (190 and 122 mm SL) should be referred to the Serrano- chromis (Sargochromis} giardi-codringtoni complex (see p. 222) and that the other two (132 and 76 mm SL) can be referred, provisionally, to S. (Sarg.) coulteri (Bell-Cross). Interestingly, S. (Sarg.) greenwoodi is present amongst the new Cunene material (see p. 224). The new cichlid material from the Cunene river, and the taxonomic changes which it has necessitated, throw very little fresh light on the zoogeographical relationships of the river's haplochromine fauna (but see, Appendix II). The species of Serranochromis (Serranochromis) occurring in the Cunene also occur in the Upper Zambezi system (including the Okovango river), the Kafue, the Upper Zaire drain- age, and in the other Angolan rivers. At the species, but not the subspecies-level, one taxon S. (S.) robustus extends to Lake Malawi and the Shire river (see Trewavas, 1964 and 1973; Poll, 1967). Because of the confused species-level taxonomy of Serranochromis (Sargochromis), little zoogeographical information can be derived from the species of that subgenus in the Cunene. It seems likely, nevertheless, that their relationships are with taxa from the Zambezi and Kafue systems, the Cunene fishes being either conspecifics or, if specifically distinct, their vicariant counterparts. The genus Pseudocrenilabrus has an extraordinarily wide distribution in Africa (Nile, Lakes Victoria, Edward, George and Malawi, the Zambezi, Limpopo and Zaire basins, vari- ous Angolan rivers, the Okavango and Orange rivers, and the rivers of Natal and Kwazulu). Once again, inadequate species-level taxonomy precludes any fine zoogeographical analysis (see p. 214). The Cunene and other populations of Pseudocrenilabrus appear to be referable to Ps. philander, a species with a wider, but more southerly distribution than its congener Ps. multicolor which is confined to the Nile, Lakes Victoria, Edward, George and to streams and small lakes in Uganda. The Cunene Pseudocrenilabrus species is certainly quite distinct from Ps. ventralis, a species endemic to the Zaire river (Nichols, 1928). That several of the Cunene and other Angolan haplochromines previously classified in the genus Haplochromis are now referred to Thoracochromis (p. 189) is possibly of some 232 P. H. GREENWOOD biogeographical significance. Thoracochromis has a wide distribution encompassing the Nile (including Lake Albert), Lake Turkana, Lakes Edward and George, Lake Mweru, and the lower Zaire drainage system (see Greenwood, 1979: 293). It has not so far been found in the Zambezi. The two Cunene species, Th. buysi and Th. albolabris, are both endemic to that system; unfortunately their phyletic relationships cannot yet be determined. Morphologically, Th. buysi is a generalized species, and could well be the local, that is vicariant member of a group including at least two other Angolan taxa, Th. lucullae and Th. schwetzi. There are indications from the material I have examined that other species will eventually be added to the group. Thoracochromis albolabris, in sharp contrast, is a highly derived taxon (see p. 204) whose specialized features, being autapomorphic ones, do not help in establishing its relationships within the genus. Certainly there are no characters indicative of affinity with any Zairean taxa, nor with its Angolan congeners. Indeed, the only relationship suggested is with Melano- chromis labrosus of Lake Malawi (see p. 205). That possibility cannot be elaborated further until more is known about the anatomy and osteology of M. labrosus. Finally, there is Orthochromis machadoi, another endemic species, and a member of another genus with Zairean connections (see Greenwood, 1979: 297). Unlike Thoraco- chromis, Orthochromis is otherwise confined to the Upper Zaire system (extending that drainage, in an historical context, to include the Malagarazi river which now empties into Lake Tanganyika). Orthochromis machadoi is the least derived member of the genus, and thus cannot be linked, as a sister-species, with any of its congeners. Its colour pattern suggests a possible affinity with O. malagaraziensis of the Malagarazi river, Tanzania (see Greenwood, 1979: 298), but the value of that character for establishing a true phyletic relationship is still untested. In brief, and on a broad scale, it seems that the new collection corroborates earlier sugges- tions of the Cunene cichlid fauna's affinities with those of the Zambezi and Zaire systems (Trewavas, 1964 & 1973; Bell-Cross, 1975; Roberts, 1975). Any finer resolution of those affinities will depend on the acquisition of many more data leading to greater precision in establishing interspecific relationships. For the moment it is much easier to recognise differences, that is endemicity, than it is to assess phylogenetic affinities. Any remarks made about levels of endemicity for particular rivers in Angola must perforce be cautious ones since the area is still poorly collected. The situation is also complicated by the absence of precise locality data for some apparently 'good' species currently repre- sented by one or a few specimens. From the information now available, the Cunene has at least four endemic haplochromine species, namely, Orthochromis machadoi, Serranochro- mis (Sargochromis) coulteri, Thoracochromis buysi and Th. albolabris, and there are indi- cations of two or possibly three other endemic species as well. Even without the inclusion of these undescribed taxa, the Cunene has the highest number of endemic haplochromines for any Angolan river (see Poll, 1967; Trewavas, 1973). Several difficulties are encountered when attempting to compare the Cunene haplochro- mine fauna with that of other rivers in Angola, in particular those which flow westward and empty directly into the Atlantic. In part these problems stem from the inadequacy of existing collections, and in part from the poor documentation of earlier collections. For example, as species occurring in westward flowing rivers, other than the Cunene, Poll (1967: 23) lists, under the generic name Haplochromis, the taxa acuticeps, fasciatus, humilis, lucullae, multiocellatus and welwitschi. The type locality for acuticeps is recorded only as Angola, and the species has not since been identified in material from any westward flowing river. Likewise, humilis, known from the holotype only, has not been recorded since its original description; again, the type locality was given merely as Angola. Haplochromic fasciatus (now Thoracochromis) does not appear to have been collected in any westward flowing Angolan river (pace Poll's reference to Regan [1 922], who cites its distribution only as 'Lower Congo'). There are doubts about the identity of certain specimens referred to the species by Regan (see Greenwood, 1979: 293), but the types are from the lower Zaire drainage. CUNENE RIVER HAPLOCHROMINE SPECIES 233 Poll's (op. cit.) listing of Haplochromis welwitschi in western rivers other than the Cunene is probably attributable to uncertainty about the exact provenance of the holotype, for long the only known example of the species. It seems likely that, apart from Poll's specimens taken in tributaries of the Cubango river (Zaire drainage) and other Angolan rivers flowing into the Zaire system, the only other recorded locality for the species is the Cunene river (see Appendix II). Finally, there is the problem of Haplochromis angolensis. This is discussed fully on p. 220. In brief, the specimens to which Poll refers were misidentified by Boulenger (1915), and anyhow came from the Cunene drainage and not one of the other westward flowing rivers in which Poll records the species' presence. Thus, in effect, Poll's list of five haplochromine species in western rivers other than the Cunene is reduced to two, lucullae and multiocellatus; these taxa may be referred to as 'small haplochromines' in contradistinction to those whose individuals reach a much larger adult size (that is, species of Serranochromis). The 'small haplochromines' are represented in the Cunene by three endemics, Thoracochromis buysi, Th. albolabris and Orthochromis machadoi, and the non-endemic Pseudocrenilabrus philander. Using adult size as a criterion, the non-endemic Chetia welwitschi (see Appendix II) should also be included as a Cunene 'small haplochromine'. If what appear to be two undescribed species (so far represented by inadequate samples) are also included, the total number of Cunene 'small haplochromine' species is seven, five more than in any of the other westward flowing rivers of Angola. For the 'large haplochromines' in the other rivers, Poll (1967: 23) lists a total of five taxa, viz four species of Serranochromis (which would now be referred to the nominate subgenus of that taxon), and Haplochromis mellandi (now referred to the subgenus Sargochromis of Serranochromis). The four Serranochromis (Serranochromis) species also occur in the Cunene (see p. 228 above). Because of the confused situation surrounding the taxonomy of Angolan Serranochromis (Sargochromis) species it is not certain whether S. (Sarg.) mellandi occurs in the Cunene, or, indeed, whether it is present in Angola at all. A mellandi-\ike taxon is present in some Angolan rivers, and may be in the Cunene as well. It is possible that five S. (Sargochromis) taxa occur in that river, namely a giardi-\\kt and a mellandi-\ike species, together with S. (Sarg.) coulteri, a species close to, if not conspecific with S. (Sarg.) greenwoodi and S. (Sarg.) gracilis. Thus, from the data currently available, the Cunene haplochromine fauna, in terms of species numbers, is probably richer than that of all the other westward flowing rivers com- bined, and also richer than that of those rivers which ultimately flow into the Upper Zaire river basin. Indeed, at the species level, haplochromine diversity in the Cunene is higher than in the Zambezi-Kafue system, a situation attributable mainly to the greater number of 'small haplochromine' species present in the Cunene, three of which are members of Zairean genera (Thoracochromis and Orthochromis) not represented in the Zambezi and Kafue rivers. In conclusion it can be noted that the genus Pseudocrenilabrus is present in the Cunene, but not in other westward flowing rivers of Angola, whilst Hemichromis which does occur in those rivers (Poll, 1967) is apparently absent from the Cunene. The reverse pattern to that of Hemichromis holds for Orthrochromis, a genus represented in the Cunene by the endemic O. machadoi, but one which seemingly is absent from other Angolan rivers empty- ing directly into the Atlantic. Possibly these paradoxes could be resolved if more was known about the ecological requirements of the species involved. Appendix I The generic status of various Angolan species referred to Haplochromis by Regan (1922), Poll (1967), Trewavas (1973) and Bell-Cross (1975) Four species in this category have now been placed in the genus Thoracochromis, see pp. 1 89-206 above. The generic status of certain other species was reviewed in Greenwood 234 P. H. GREENWOOD (1979). There, reasons were given for transferring H. thysi Poll (1967) and all the 'large Haplochromis' species revised by Bell-Cross (1975) to the subgenus Sargochromis ofSerrano- chromis. Also in that paper, H. machadoi Poll (1967) was recognised as an Orthochromis species (see also p. 206 above). The generic status of the other Angolan ' Haplochromis' species remains uncertain. Haplochromis humilis (Steindachner), 1866, recorded only as being from Angola, is not currently available for reexamination (see p. 189). Bell-Cross' (1975: 426) comments, coupled with Steindachner's original description and accompanying figure, suggest that the specimen should probably be referred to a species of Serranochromis (Sargochromis). The type of//, angolensis (Steindachner), 1865 is lost (see Bell-Cross, 1975: 426); the original description is so inadequate that the type specimen's identity cannot be determined at either generic or specific levels. The type and only specimen of Haplochromis multiocellatus Blgr, 1913 has a gradual size- change of scales in the transition area between chest and belly squamation. Such a pattern would exclude the species from Thoracochromis (see Greenwood, 1979: 290), a genus with which it shares no other diagnostic features either. That the type appears to have true anal ocelli of the kind found in Astatotilapia and in the majority of Lake Victoria haplochromines (see Greenwood, 1979: 274-5), would also seem to argue against its inclusion in Thoraco- chromis. However, details of anal fin markings are difficult to ascertain in preserved speci- mens, and the absence of true ocelli in all Thoracochromis species has yet to be established on the basis of fresh or live material. Thus, the generic status of ' Haplochromis' multi- ocellatus must remain undetermined, except in so far as it cannot be referred to the genera Haplochromis, Thoracochromis or Serranochromis as defined by Greenwood (1979). At a lower taxonomic level, it should be recalled that Trewavas (1973: 31) believes the taxon to be a junior synonym of '//' acuticeps (whose generic status is, at least for the moment, also uncertain (see above, p. 1 90)). Haplochromis welwitschi (Blgr) 1898, is discussed in Appendix II. Appendix II The generic status of Pehnatochromis welwitschi Blgr, 1898 Surprisingly, this species has never been formally transferred to any other haplochromine genus, despite the fact that it clearly is not a member of the genus Pelmatochromis as cur- rently defined, and despite its obvious membership of the genus Haplochromis as defined by Regan (1920 & 1922). It has, however, been referred to as Haplochromis welwitschi by several workers, notably Poll (1967), Trewavas (1964) and Bell-Cross (1975). When reviewing the generic status of fluviatile haplochromine species (Greenwood, 1979: 310 & 313), I commented on the possible generic affinities of P. welwitschi, particularly in the light of Trewavas' (1974: 9; fig. 1) suggestion that it might be related to Serranochromis (Greenwood, 1979: 306). At that time I considered whether the species might be included in the genus Chetia Trewavas (1961) but expressed reservations about its formal transfer to that taxon until further material could be studied. Having now had the opportunity to examine specimens from Angola, kindly lent to me by the MRAC, I would withdraw those reservations. My reasons for doing so are based on Chetia flaviventris, type species of the genus, sharing with P. welwitschi a number of character combinations which neither genus shares with other haplochromine taxa. These include dental characters, features of the squamation pattern, vertebral numbers, and the type of spotting on the anal fin of male fishes. The recently described Chetia mola (see Balon & Stewart, 1983) also shares these features, differing only in its hypertrophied pharyngeal jaws and dentition, and in correlated modifications to the neurocranial apophysis for the upper pharyngeal bones. The entirely unicuspid oral dentition of available welwitschi specimens is like that in larger CUNENE RIVER HAPLOCHROMINE SPECIES 235 Chetia flaviventris specimens. In C. flaviventris, the unicuspid outer row teeth appear in specimens of a very small size (see Trewavas, 1961); since none of the welwitschi material is less than 100 mm SL, it is not possible to check whether this species, too, shows a similar precocity in dental ontogeny. Neither welwitschi nor C. flaviventris has enlarged and serially displaced median teeth in the inner premaxillary tooth row (see p. 216). Apart from the last 1 to 4 scales in the upper lateral- line series, all the scales in that row of both species are each separated from the dorsal fin base by at least one small and two large scales of equal size, comparable in that respect with the pored scales below them. The lower pharyngeal bone in C. flaviventris and in welwitschi is fine and relatively narrow, with long and delicate (welwitschi) or relatively delicate posterior horns (c/Fig. 20 with fig. 19 in Greenwood, 1979). None of the median row teeth in welwitschi is much coarser than the teeth situated laterally in the dental field, but in C. flaviventris some median teeth are slightly coarser than those elsewhere on the bone. The morphology of the pharyngeal teeth is similar in both species; only those teeth in the posterior half of the dental field have a well-marked shoulder or a distinct minor cusp, the others being essentially unicuspid with a sharp and oblique crown. Like C. flaviventris, P. welwitschi has numerous spots on the anal fin of male fishes. In the holotype of P. welwitschi, now faded, there are about 8-10 spots arranged in a short upper and a longer lower row; the MRAC specimens examined have an estimated 13-15 spots, in several irregular rows, scattered over the greater part of the fin, but these fishes are larger than the holotype (see p. 237 below). Finally, both C. flaviventris and welwitschi, when compared with all the other fluviatile haplochromine species except Serranochromis, show a tendency towards an increase in the number of abdominal vertebrae (see Greenwood, 1979). The three welwitschi specimens all have 14 or 15 abdominal elements, most of the C. flaviventris examined have 15, although 14 were counted in one. It must be stressed that none of the distinctive characters shared by Chetia flaviventris and P. welwitschi is uniquely synapomorphic for the two species (see Greenwood, 1979: 308-3 10 Fig. 20 Chetia welwitschi. Lower pharyngeal bone. Numerous teeth are missing, but their sites of attachment are indicated. Scale in mm. 236 P. H. GREENWOOD for further discussion). Taken together, however, the various characters appear to be shared only by these two species and C. mola. Since none has any detectable feature or features shared uniquely by it and any other genus, it would seem reasonable to consider them con- generic until such times as their implied monophyly can be refuted. There would certainly seem to be no grounds for recognising welwitschi as representing a distinct lineage, and thus a distinct genus. Chetia welwitschi (Blgr) 1898 SYNONYMY. Pelmatochromis welwitschi Boulenger, 1898. Proc. zool. Soc. London: 149, pi. xix; idem, 1915. Cat. Afr. Fw. Fishes, 3: 397, fig. 268. Haplochromis welwitschi'. Trewavas, 1964. Annls Mus. r. Congo Belg, Ser. 8vo, Zool. no. 25: 1-58; Poll, 1967. Publicacbes cult. Co. Diam. Angola no. 75: 1-381; Bell-Cross. 1975. Occ. Pap. natn. Mus. Rhod. ser. B. 5 (7): 405^64; Greenwood, 1979. Bull Br. Mus. nat. Hist. (Zool.) 35 (4): 265-322. DESCRIPTION. Poll (1967: 307-309; fig. 149) gives a detailed account of the Angolan material he examined. The table and comments below refer only to the holotype (BMNH 1864.7.13: 62) and the two Angolan specimens from Sanguenque Uembe Cuanaa, Angola, loaned by the MRAC (154779-780), which were not included in Poll's (1967) redescription of the species. In three morphometric features (smaller eye, deeper preorbital and longer lower jaw), the MRAC specimens differ slightly from the holotype, but these discrepancies could be due to the larger size of the former. There is, however, a marked discordance in caudal peduncle length when Poll's (1967) figures for other material are compared with those obtained from Fig. 21 Chetia welwitschi (Blgr). Holotype; after Boulenger (1915). Drawn by G. J. Howes. Scale = 20 mm. the specimens described above. Poll gives the caudal peduncle length as 30-0 and 32-6% of the standard length; one can only conclude that these figures are typographical errors. There are 34 scales in the lateral-line series of the holotype, and 32 in the MRAC speci- mens. In the original description, Boulenger (1898) gives a count of 32 for the type, but this was changed to 33 in his 1915 description of the species. No anal sheath scales are present in the MRAC fishes, but a few scattered scales, on both sides of the fin, are preserved in the holotype. None was found in the type and two paratypes of Chetia flaviventris \ examined. CUNENE RIVER HAPLOCHROMINE SPECIES 237 Holotype MRAC Specimens BMNH:1864.7.13:62 154779-780 Standard length 102-0 126-0 146-0 Depth of body* 33-3 34-1 34-2 Length of head* 32-3 34-1 34-2 Preorbital bone deptht 24-3 20-9 21-0 Least interorbital widthf 21-2 23-3 22-0 Snout lengthf 36-3 34-9 34-0 Snout length/breadth 1-1 1-0 0-9 Eye diameterf 22-8 18-6 20-0 Cheek depthf 33-3 32-6 32-0 Lower jaw lengthf 42-5 47-1 48-0 Lower jaw length/breadth 1-7 1-8 2-0 Caudal peduncle length/depth 18-6 17-0 17-5 Caudal peduncle length/breadth 1-4 1-5 1-5 Length of premaxillary ascend, processes! 33-0 30-2 30-0 Pectoral fin length* 18-1 20-6 20-5 t 56-0 60-5 60-0 * = percentage of standard length; t = percentage of head length The holotype of C. welwitschi has 9 short and relatively stout gill-rakers on the lower part of the first gill-arch; there are 9 rakers in one of the MRAC specimens, and 1 1 in the other. Microbranchiospines are present in all three fishes, but are smaller and less obvious in the holotype. Many outer row teeth are missing in the holotype, but I would estimate that about 40 were once present in the premaxilla. There are about 56 premaxillary teeth in the MRAC specimens (Poll [1967] gives a count of 68 in the two fishes he examined). The holotype has 30 vertebrae (excluding the fused PU, and U, centra), comprising 15 abdominal and 15 caudal elements; the two other specimens both have 29, 14 of which are abdominal, and 15 are caudal elements. In Chetiaflaviventris the counts are 30-32, mode 31, comprising 14 or 15 (mode 15) abdominal, and 15-17 (modes 16 and 17) caudal centra. Hypurals 1 and 2, and 3 and 4 are apparently fused in the MRAC specimens radiographed; in the holotype all are free, but 3 and 4 are closely apposed. All hypurals are free in the Chetiaflaviventris material radiographed. The pharyngeal apophysis of C. welwitschi holotype is of the Haplochromis-type, and is broad, with a laterally expansive contribution from the parasphenoid; its structure was not examined in the MRAC material. No information is available about the vertebral apophysis for the retractor arcuum branchialis muscle. All three specimens are males and, apparently, are adult. The 146 mm SL fish is sexually active. In none does the first pelvic ray extend posteriorly beyond the anus. This ray in the 146 mm specimen is distinctly longer than the 2nd ray, but it is not filamentous; in the two smaller fishes, the first ray is but slightly longer than the second. The occurrence and pattern of the numerous anal spots is described above (p. 235). Poll's drawing (1967: fig. 149) of another Angolan specimen shows, in contrast, only three or four spots confined to the posterior part of the soft anal fin. This may be the result of an artist's error stemming from the extreme difficulty one experiences in arranging the lighting to reveal such faintly pigmented areas. There is some uncertainty about the type locality otChetia welwitschi. Originally recorded as being collected by Welwitsch from Fluilla, Angola (Boulenger, 1898), it was later thought that Fluilla had been a misspelling of Huilla, a town and district in the- south-western part 238 P. H. GREENWOOD of the country (Bell-Cross, 1975: 427). If that is so, then the holotype would be from the Cunene drainage. With the addition of Poll's material, the range of C. welwitschi is now extended to include the Zaire drainage system as well (Poll, 1967: 307). If the generic placement of P. welwitschi is correct, then it is the only Angolan haplo- chromine, except Pseudocrenilabrus philander, belonging to a taxon also occurring in the Limpopo drainage (the type material ofChetiaflaviventris is from a dam on the Sterkstroom river, a tributary of the Crocodile). That remark is made, of course, on the assumption that the specimens Poll identified as Haplochromis darlingi are not referable to the genus Pharyngochromis, the genus in which H. darlingi is now placed (Greenwood, 1979: 310). Neither the inclusion of Pelmatochromis welwitschi in Chetia, nor the discovery of C. mola, provide any information on the phyletic relationships of the genus. As Trewavas (1964) suggested, its relationships would seem to be with the Serranochromis generic com- plex, but the characters on which that suggestion was made are still of unproven value in a truly phylogenetic scheme of classification (Greenwood, 1979: 309). Acknowledgements Once more it is a pleasure to thank my colleague Gordon Howes for the time and skill he has given to producing the figures, and carrying out so many of the tedious activities involved in producing this paper. I am also greatly indebted to Bernice Brewster, who assisted Gordon Howes with the radiographic work, and has aided me in many other tasks as well. Outside the Museum, my gratitude goes to Dr M. Penrith at whose invitation this study was under- taken, and who has answered so readily my many requests for information about the collection, even producing a detailed gazetteer of localities. My thanks must also go to Dr H. Wilkens of the Hamburg Museum, Dr Paul Skelton (Albany Museum, Grahamstown), Drs Thys van den Audenaerde and Guy Teugels (MRAC, Tervuren) and Dr S. Kullander (Stockholm Museum) all of whom kindly lent material in their care. References Balon, E. K. & Stewart, D. J. 1983. Fish assemblages in a river with unusual gradient (Luongo, Africa — Zaire system), reflections on river zonation, and description of another new species. Envir. Biol. fishes 9 (3^): 225-252. Bell-Cross, G. 1975. A revision of certain Haplochromis species (Pisces: Cichlidae) of Central Africa. Occ. Pap. natn. Mus. Rhod. ser. B. 5(7): 405-464. Boulenger, G. A. 1898. A revision of the African and Syrian fishes of the family Cichlidae. Proc. zool. Soc.Lond. 1898: 132-152. 1913. Descriptions of five new cichlid fishes from Africa. Ann. Mag. nat. Hist. (7) 12: 482^85. 1915. Catalogue of the fresh-water fishes of Africa 3. London. Greenwood, P. H. 1954. On two species of cichlid fishes from the Malagarazi river (Tanganyika), with notes on the pharyngeal apophysis in species of the Haplochromis group. Ann. Mag. nat. Hist. (12) 7:401-414. 1959. A revision of the Lake Victoria Haplochromis species (Pisces, Cichlidae), Part III. Bull. Br. Mus. nat. Hist. (Zool.) 5: 179-218. 1978. A review of the pharyngeal apophysis and its significance in the classification of African cichlid fishes. Bull. Br. Mus. nat. Hist. (Zool.) 33: 297-323. 1979. Towards a phyletic classification of the 'genus' Haplochromis (Pisces, Cichlidae) and related taxa. Part I. Bull. Br. Mus. nat. Hist. (Zool.) 35: 265-323. 1980. Towards a phyletic classification of the 'genus' Haplochromis (Pisces, Cichlidae). Part II. Bull. Br. Mus. nat. Hist. (Zool.) 39: 1-101. 1981. The haplochromine fishes of the east African lakes. Munich & London. 1983. The Ophthalmotilapia assemblage of cichlid fishes reconsidered. Bull. Br. Mus. nat. Hist. (Zool.) 44: 249-290. Ladiges, W. 1964. Beitrage zur Zoogeographie und Oekologie der Siisserwasserfische Angolas. Mitt, zool. St Inst. Hamb. 61: 221-272. CUNENE RIVER HAPLOCHROMINE SPECIES 239 Nichols, J. T. 1928. A few fishes from the northeast corner of the Congo basin. Am. Mus. Novit. 309: 1^. Pellegrin, J. 1936. Contribution a 1'ichthyologie de PAngola. Archos Mus. Bocage 7: 45-62. Penrith, M.-L. 1970. Report on a small collection of fishes from the Kunene river mouth. Cimbebasia, ser. A. 1(7): 165-176. Poll. M. 1967. Contribution a la faune ichthyologique de 1'Angola. Publicacdes cult. Co. Diam. Angola no. 75: 1-381. Regan, C. T. 1920. The classification of the fishes of the family Cichlidae -I. The Tanganyika genera. Ann. Mag. nat. Hist. (9) 5: 33-53. 1922. The classification of the fishes of the family Cichlidae -II. On the African and Syrian genera not restricted to the Great Lakes. Ann. Mag. nat. Hist. (9) 10: 249-264. Roberts, T. R. 1975. Geographical distribution of African freshwater fishes. Zool. J. Linn. Soc. 57 (4): 249-3 19. Staeck, W. 1983. Cichliden III: Entdeckungen und Neuimporte. Wuppertal. Steindachner, F. 1866. Ueber einege neue Susswasserfische von Angola. Verh. zool-bot. Ges. Wien 16:761-771. Stiassny, M. L. J. 1981. The phyletic status of the family Cichlidae (Pisces, Perciformes): A compara- tive anatomical investigation. Neth. J. Zool. 31: 275-314. 1982. The relationship of the neotropical genus Cichla (Perciformes, Cichlidae): a phyletic analysis including some functional considerations. J. Zool. Lond. 197: 427-453. Trewavas, E. 1936. Dr Karl Jordan's expedition to South- West Africa and Angola: the fresh-water fishes. Novit. Zool. Tring 60: 63-74. 1961. A new cichlid fish in the Limpopo basin. Ann. S. Mus. 46 (5): 53-56. 1964. A revision of the genus Serranochromis Regan (Pisces, Cichlidae). Annls. Mus. r. Congo Beige. Ser. 8vo, Zool. no. 125: 1-58. 1973. II. A new species of cichlid fish of rivers Quanza and Bengo, Angola, with a list of the known fishes of these rivers and a note on Pseudocrenilabrus natalensis Fowler. Bull. Br. Mus. nat. Hist. (Zool.) 25: 27-37. 1974. The freshwater fishes of rivers Mungo and Meme and Lakes Kotto, Mboandong and Soden, West Cameroon. Bull. Br. Mus. nat. Hist. (Zool.) 26: 329^19. & Thys van den Audenaerde, D. F. E. 1969. A new Angolan species of Haplochromis (Pisces, Cichlidae). Mitt. zool. St Inst. Hamb. 66: 237-239. Wickler, W. 1963. [Classification der Cichlidae, am Beispiel der Gattungen Tropheus, Petrochromis, Haplochromis und Hemihaplochromis n. gen. (Pisces, Perciformes). Senckenberg. biol. 44: 83-96. Manuscript accepted for publication 1 May 1984 British Museum (Natural History) Tilapiine fishes of the genera Sarotherodon, Oreochromis and Danakilia Dr Ethelwynn Trewavas The tilapias are cichlid fishes of Africa and the Levant that have become the subjects of fish-farming throughout the warm countries of the world. This b9ok described 41 recognized species in which one or both parents carry the eggs and embryos in the mouth for safety. Substrate-spawning species, of the now restricted genus Tilapia, are not treated here. Three genera of the mouth-brooding species are included though in one of them, Danakilia, the single species is too small to warrant farming. The other two, Sarotherodon, with i nine species and Oreochromis, with thirty-one, are distinguished primarily by their breeding habits and their biogeography, supported by structural features. Each species is described, with its diagnostic features emphasised and illustrated, and to this is added a summary or known ecology and behaviour. Conclusions on relationships involve assessment ol parallel and convergent evolution. Dr Trewavas writes with the interests of the fish cultunsts, as well as those of the taxonomists, very much in mind. 580pp 1 88 illustrations include halftones, diagrams, maps and graphs. Extensive bibliography. Publication 1983. £50 Publications Sales British Museum (Natural History) Cromwell Road London SW7 5 BD England Titles to be published in Volume 47 Miscellanea A review of the Mastacembeloidei, a suborder of synbranchiform teleost fishes Part II: Phylogenetic analysis. By Robert A. Travers A review of the anatomy, taxonomy, phylogeny and biogeography of the African neoboline cyprinid fishes. By Gordon J. Howes The haplochromine species (Teleostei, Cichlidae) of the Cunene and certain other Angolan rivers. By Peter Humphry Greenwood Miscellanea Anatomy and evolution of the feeding apparatus in the avian orders Coraciiformes and Piciformes. By P. J. K. Burton A revision of the spider genus Cyrba (Araneae: Salticidae) with the descriptions of a new presumptive pheromone dispersing organ. By F. R. Wanless Printed in Great Britain by Henry Ling Ltd., at the Doi^ hester. Dorset Bulletin of the British Museum (Natural History) Miscellanea Zoology series Vol47 No 5 25 October 1984 The Bulletin of the British Museum (Natural History}, instituted in 1949, is issued in four scientific series. Botany, Entomology, Geology (incorporating Mineralogy) and Zoology, and an Historical series. Papers in the Bulletin are primarily the results of research carried out on the unique and ever-growing collections of the Museum, both by the scientific staff of the Museum and by specialists from elsewhere who make use of the Museum's resources. Many of the papers are works of reference that will remain indispensable for years to come. Parts are published at irregular intervals as they become ready, each is complete in itself, available separately, and individually priced. Volumes contain about 300 pages and several volumes may appear within a calendar year. Subscriptions may be placed for one or more of the series on either an Annual or Per Volume basis. Prices vary according to the contents of the individual parts. Orders and enquiries should be sent to: Publications Sales, British Museum (Natural History), Cromwell Road, London SW7 5BD, England. World List abbreviation: Bull. Br. Mus. nat. Hist. (Zool.) © Trustees of the British Museum (Natural History), 1984 The Zoology Series is edited in the Museum's Department of Zoology Keeper of Zoology : DrJ. G. Sheals Editor of Bulletin : Dr C. R. Curds Assistant Editor : Mr C. G. Ogden ISBN 0 565 05008 7 ISSN 0007-1498 British Museum (Natural History) Cromwell Road London SW7 5BD Zoology series Vol47 No. 5 pp 241-329 Issued 25 October 1984 p Miscellanea Contents Page Notes on testate amoebae (Protozoa: Rhizopoda) from Lake Vlasina, Yugoslavia. By Colin G.Ogden 241 Description of a neotype for the holothurian Oncus brunneus (Forbes MS in Thompson 1840) from Strangford Lough, Northern Ireland (Holothurioidea: Dendrochirotida). By J. Douglas McKenzie 265 Three new species of Varicorhinus (Pisces, Cyprinidae) from Africa. By K. E. Banister 273 Phyletics and biogeography of the aspinine cyprinid fishes. By Gordon Howes . . . 283 New bats (Mammalia: Chiroptera) and new records of bats from Borneo and Malaya. By J. E. Hill & C. M. Francis 305 '/s|;^ ^•'•'V ^C:/ •• ¥1*1 250CTI984 \if\ -..- — •— -j V i H i ^ / .^O^SfiiES <>J X ^r/^AnTi Notes on testate amoebae (Protozoa: Rhizopoda) from Lake Vlasina, Yugoslavia Colin G. Ogden Department of Zoology, British Museum (Natural History), Cromwell Road, London SW7 5BD Introduction The morphology of some testate amoebae from Lake Vlasina, Yugoslavia has been reported in an earlier work (Ogden & Zivkovic, 1983). This present study extends the survey to include recently collected specimens, some of which have already been used (Ogden, 1983#) in a reappraisal of the genus Pontigulasia. The author was fortunate to be guided on his visit in September, 1982, by Dr A. Zivkovic who had collected the samples for the earlier work. Two contrasting sites were sampled, one at the southern extremity of this artificially created lake and the other at the northern. The southern site was an area of damp to wet pasture, with small ponds or pools scattered across it at intervals. It was one of several similar level expanses of land consisting of mixed grasses and patches of Sphagnum bordering the lake margin, used for grazing but possibly flooded or waterlogged in winter. The northern site was a small triangular inlet, which appeared to be in the process of being naturally reclaimed. It was covered with a floating mass of vegetation, including tall coarse grasses and small shrubs, but being mainly a platform of Sphagnum. It was possible to walk over the surface with care, but at each step water oozed up from below, the whole being buoyant and unsteady. The level of the lake was quite low during the visit, which was clearly seen from the water level marks around the shoreline, reflecting the end of a long dry summer. The large floating masses of peat which appeared on the lake surface after the initial flooding, described by Milovanovic & Zivkovic (1956), are now stranded well above the normal level of the lake. One such mass was seen straddled across the water course of a small stream as a solid block about 7 m high and 15 m long. The main purpose of this study was to look particularly for specimens belonging to the family Difflugiidae, and to compare the organic cement patterns found in species ofDifflugia with those already described (Ogden, 19836) from sites in the British Isles. Such a comparison was considered to be a reasonable test of the validity of these patterns as a good taxonomic character. Nevertheless, several examples of other species of testate amoebae were examined and compared with those described earlier by the author. The samples were rich in speci- mens, and the numbers examined here and elsewhere (Ogden, 1983#) represent only a small proportion of those present. Materials and methods Samples of Sphagnum moss and water plants were collected from several points at each site. There was no apparent difference between samples taken over a large area at one site, but there were a few differences between the two sites. Individual specimens were extracted from the samples as being representative of the population, and animals described are a selection of those present. They were examined optically either by Nomarski interference or bright- field illumination, and by scanning electron microscopy. Preparative techniques for scanning electron microscopy have been described elsewhere (Ogden, 19836), treated specimens were examined in either a Cambridge Stereoscan SI 80 or an Hitachi S800 microscope. Bull. Br. Mus. nat. Hist. (Zool.) 47(5): 241-263 Issued 25 October 1984 242 C. G. OGDEN Systematic section An alphabetical arrangement has been adopted for both genera and the species in each genus. The numbers given for specimens measured refer to those which were examined in detail, and descriptions are included for those specimens which have not been the subject of earlier studies. When animals are found to be similar to a previous description that reference is quoted. Arcella arenaria Greeff, 1866 One specimen examined: diameter of shell 99 urn, depth 30 urn, diameter of aperture 16 urn, similar to Plate 1, Ogden & Hedley, 1980. Arcella discoides Ehrenberg, 1 843 One specimen examined: diameter of shell 114um, depth 34 um, diameter of aperture 46 um, similar to Plate 7, Ogden & Hedley, 1980. Arcella rotundata var. stenostoma Deflandre, 1928 Four specimens examined: diameter of shell 52-61 um, depth 33-36 um, diameter of aper- ture 12-14 urn. The specimens agree with the original description given by Deflandre (1928), except that the domed region was covered with small regularly spaced angular facets. Obser- vations on clonal cultures of Arcella vulgaris led Ogden (1984) to suggest that this feature was unreliable as a specific character and it is treated as such here. Assulina muscorum Greeff, 1888 One specimen examined: shell length 53 um, breadth 31 um, diameter of aperture 11 um. This specimen agrees well with the description of Ogden & Hedley (1980) and Ogden (198 1). It would appear from the detail shown in Figs 1 & 2 that small shell plates may be incorporated in the organic cement which characteristically surrounds the aperture. Figs 1 & 2 Assulina muscorum: Two views of the aperture to illustrate the organic cement margin, with possibly small shell plates (arrowed) incorporated x6000 and x7000. TESTATE AMOEBAE 243 Centropyxis aerophila Deflandre, 1929 Five specimens examined: length of shell 54-70 um, breadth 37-60 um, depth 27-43 um, diameter of aperture 21-35 um, similar to Plate 13, Ogden & Hedley, 1980. Centropyxis cassis (Wallich, 1864) Two specimens examined: length of shell 108 & 122 jam, breadth 84 & 98 um, depth 61 um, diameter of aperture 53 & 49 um, similar to Plate 14, Ogden & Hedley, 1980. Centropyxis hirsuta Deflandre, 1929 Two specimens examined: length of shell 86 & 88 um, breadth 51 & 73 um, depth 40 & 42 um, diameter of aperture 27 & 31 um, similar to Plate 18, Ogden & Hedley, 1980. Centropyxis platystoma Penard, 1890 Two specimens examined: length of shell 61 & 75 um, breadth 33 & 36 um, depth 28 & 30 um, diameter of aperture 22 um, similar to Plate 19, Ogden & Hedley, 1980. Cyclopyxis kahli (Deflandre, 1929) One specimen examined: diameter of shell 9 1 um, depth 62 um, diameter of aperture 24 um, similar to Plate 24, Ogden & Hedley, 1980. Cyphoderia ampulla (Ehrenberg, 1 840) Three specimens examined: length of shell 87-109 um, breadth 39-55 um, diameter of aperture 15-16 um, similar to Plate 94, Ogden & Hedley, 1980. Cyphoderia laevis Penard, 1902 DESCRIPTION. The shell is colourless, retort-shaped and circular in cross section, being widest at about the mid-point and tapering to both extremities (Figs 3 & 5). A distinct neck is formed by the sharp curvature of the shell towards the aperture (Fig. 4). It is composed of oval shell plates which are irregular in outline and size, with the occasional elongate plate scattered amongst them (Fig. 6). An apparently structureless organic cement binds these plates together without any overlapping. The aperture is circular (Fig. 5) and has an even margin composed of small plates (Fig. 3). MEASUREMENTS (in um). Based on one specimen: length of shell 64, breadth 27, diameter of aperture 9. REMARKS. This specimen is similar in size to those described by Penard (1902). It differs from his description because he stated that the curvature of the neck was hardly distinguish- able from the general outline of the shell, whereas in the present specimen the shell has a distinct neck. Nevertheless, the shell plates are small and irregular in both size and distri- bution, which is in agreement with the original description. In the absence of other animals, the present specimen is considered to be conspecific with C. laevis. Difflugia acuminata Ehrenberg, 1838 Four specimens examined: body length 228-274 um, breadth 85-106 um, diameter of aperture 37-46 um, similar to Figs 3, 8 & 9, Ogden, \919a. The distinctive organic cement pattern that distinguishes this species is shown in Figs 7 & 8. C. G. OGDEN Figs 3-6 Cyphoderia laevis: Fig. 3, lateral view of aperture x3000; Fig. 4, view of shell to show curvature of outline x 1 700; Fig. 5, apertural view x 1300; Fig. 6, detail of shell surface x 10000. Difflugia amophoralis Cash & Hopkinson, 1909 Two species examined: body length 108 & 1 10 urn, breadth 67 & 70 urn, diameter of aperture 32 & 37 um, similar to Fig. 19, Ogden, 19836. These specimens agree well with the earlier description (Ogden, 19836) being composed mainly of quartz, but they also incorporate some spherical siliceous flagellate cysts. TESTATE AMOEBAE 245 Figs 7 & 8 Difflugia acuminata: Detail of organic cement pattern x 34000 and x 39000. Figs 9-11 Difflugia bryophila: Fig. 9, lateral view to illustrate basic outline x960; Figs 10 & 1 1 , detail of organic cement network x 27000 and x 49000. 246 C. G. OGDEN Difflugia bicornis Penard, 1 890 Fifteen specimens examined: body length 63-9 1 um, breadth 36-5 1 jim, diameter of aperture 18-25 um, similar to Fig. 2, Ogden & Zivkovic, 1983, but all of these specimens have only one aboral spine. Difflugia brevicolla Cash & Hopkinson, 1909 Eight specimens examined: body length 84-11 9 um, breadth 79-1 00 urn, diameter of aperture 36-46 um, similar to Figs 13-17, Ogden, 1980. Difflugia bryophila (Penard, 1 902) Five specimens examined: body length 99-1 36 um, diameter of aperture 17-22um. The specimens are typical of those already described (compare Fig. 9 with Fig. 1, Ogden, 19836), but the organic cement is now seen at higher magnification to be a network with a distinct pattc-ra (Figs 10 & 11). Difflugia capreolata Penard, 1902 One specimen examined: body length 237 um, breadth 164 um, diameter of aperture 65 urn, this specimen is similar to Fig. 3, Ogden & Zivkovic, 1983. Difflugia distenda Ogden, 1983 Three specimens examined; body length 199-222 um, breadth 9 1-108 um, diameter of aperture 47-54 um, similar to Fig. 21, Ogden, 19836. Difflugia elegans Penard, 1 890 Six specimens examined: body length 95-129 um, breadth 55-66 um, diameter of aperture 30-36 um, similar to Plate 55, Ogden & Hedley, 1980 and Figs 5 & 1 1, Ogden, 19790. Difflugia elegans var. angustata Gauthier-Lievre & Thomas, 1958 DESCRIPTION. The shell is usually transparent, elongate with an aboral horn, composed of angular quartz and an occasional diatom frustule (Fig. 12). An organic cement network (Fig. 14) similar to that seen in D. elegans (Fig. 15) is present (see Ogden, 1979a). The only feature that differentiates these specimens from D. elegans is the elongate shape of the body and the absence of a constriction near the aperture (Figs 12 & 13). MEASUREMENTS (in um). Based on four specimens: body length 95-118, breadth 44-51, diameter of aperture, 24-27. REMARKS. Although the slimness of the body and the smaller aperture give reduced ratios for both, breadth/body length (0-44) and diameter of aperture/body length (0-24) when compared with the type species (0-59 + 0-07; 0-33 ±0-04, see Ogden, 1979a) there are too few specimens on which to base an accurate assessment of the specificity of this variety, and it is therefore described as such. Difflugia lanceolata Penard, 1 890 Three specimens examined: body length 162-1 73 um, breadth 60-67 um, diameter of aperture 24-32 um, similar to Fig. 6, Ogden, 19836. Difflugia lismorensis Playfair, 1918 Two species examined: body length 144 & 1 5 1 um, breadth 96 & 89 um, diameter of aperture 37 & 39 um, similar to Fig. 8, Ogden & Zivkovic, 1983. TESTATE AMOEBAE 247 14 Figs 12-14 Difflugia elegans var. angustata: Fig. 12, lateral view, note the almost parallel sides x840; Fig. 13, apertural view x900; Fig. 14, detail of organic cement x 34000; Fig. 15, Difflugia elegans, detail of organic cement x 34000. 248 C. G. OGDEN Figs 16-18 Difflugia oranensis: Fig. 16, lateral view showing the aboral projection x 1500; Fig. 17, portion of surface to illustrate the distribution of organic cement x 12000; Fig. 18, detail of organic cement x 45000. TESTATE AMOEBAE Difflugia lucida Penard, 1 890 249 Two specimens examined: body length 61 & 69 um, breadth 38 & 44 jam, depth 29 um, diameter of aperture 22 & 25 um, similar to Fig. 43, Ogden, 19836. Difflugia manicata Penard, 1 902 Two species examined: body length 55 urn, breadth 46 & 47 um, diameter of aperture 19 & 20 um, similar to Fig. 50, Ogden, 19836 and Fig. 9, Ogden & Zivkovic, 1983. Difflugia oranensis Gauthier-Lievre & Thomas, 1958 Difflugia mamillaris var. oranensis Gauthier-Lievre & Thomas, 1958 DESCRIPTION. The shell is transparent, slim, elongate, slightly widened at the aperture taper- ing to the mid-body point and then swelling slightly until it tapers gently to a small aboral projection (Fig. 16). It is composed mainly of flattened pieces of quartz, with an occasionl piece of flattened diatom frustule, arranged to give, in general, a smooth regular outline. The particles are bound together by organic cement, which has a typical network pattern with walls 0-12um thick and a mesh with a diameter of about 0-10um (Figs 17 & 18). The aperture is circular and usually surrounded by small particles. MEASUREMENTS (in um). Based on six specimens: body length 79-93, breadth 20-32, diameter of aperture 13-15. REMARKS. In describing the variety oranensis Gauthier-Lievre & Thomas (1958) stated that it was compressed and had an elongate aperture, those described here although agreeing well in all other respects, disagree on these two points. However, the present specimens are con- sidered to be fragile and liable to collapse or distort if handled roughly, sufficient for them to agree with the original description. For this reason it is considered that they be treated as being identical with D. oranensis. Three smaller specimens, measuring 61-73 um in length, 3 1-35 um in breadth and having apertures 13-1 5 urn in diameter, of a similar shape are also tentatively assigned to D. oranensis. These specimens differ not only in length but have a more pronounced swelling in the aboral region (Figs 19 & 20). They share a similar organic cement pattern with the specimens of D. oranensis described above. It may well be that they represent a distinct species but there are insufficient specimens to make adequate comparisons. Figs 19-20 Difflugia oranensis (?): Fig. 19, view to show the more swollen mid-body region (compare with Fig. 16) x 1 100; Fig. 20, apertural view x 1600. 250 C. G. OGDEN Figs 21 & 22 Difflugia parva: Detail of organic cement units x 37000. Difflugia parva (Thomas, 1954) Five specimens examined: body length 168-224 um, breadth 77-103 um, diameter of aper- ture 24-40 jim, these specimens agree with the description already given by Ogden (19836). The detail of the small mesh, covering the inner portion of each network enclosure, is now clearly seen in Figs 21 & 22. Difflugia paulii Ogden, 1983 Five specimens examined: body length 85-130 um, breadth 28-46 um, diameter of aperture 14-22 um, these specimens have the typical cigar-shape (Fig. 23) and smooth shell, the cement is seen here as a network (Fig. 24). They agree well with the original description (Ogden, 19836). Difflugia pecac sp. nov. DESCRIPTION. The shell is transparent, elongate almost cylindrical except that the widest diameter is in the latter third of the body length (Fig. 25), the aboral extremity may be either smoothly rounded or occasionally pointed (Fig. 27). It is often slightly compressed in the first third of the body length, but some specimens show no compression at all, and the aper- ture is therefore either elongate or oval in outline (Fig. 26). The structure is composed mainly of thin flattish pieces of quartz, flattish diatom particles are frequently included, arranged to make a smooth surface with small particles surrounding the apertural opening. Organic cement binds these particles and is usually restricted to small areas between these neatly arranged pieces, but may sometimes be seen as a network. In small areas it appears as an arrangement of rings (Fig. 28), but in a larger area it is seen as a typical network of walls and depressions (Figs 29 & 30). TESTATE AMOEBAE 251 Figs 23 & 24 Difflugia paulii: Fig. 23, lateral view to illustrate typical cigar-shape x700; Fig. 24, detail of organic cement network x 27000. MEASUREMENTS (in um). Based on twenty-two specimens: body length 62-84, breadth 25-36, depth 19-30, diameter of aperture 12-22. REMARKS. The fragility of this species may account for the variable degree of compression seen in the anterior third of the body. The other alternative is that these specimens may be comprised of two closely similar forms, one slightly compressed and the other having a more rounded aboral extremity. Nevertheless, as a whole they appear to represent a pre- viously undescribed species. The compressed nature of D. pecac makes it most similar to D. lucida. It differs in body shape, having a regular slim outline with almost parallel sides and a smoothly rounded aboral extremity, whereas D. lucida is characteristically oval in outline and has a curved extremity (see Ogden, 19836). ETYMOLOGY. This species is named after the unknown fisherman, pecac = fisherman (Serbo-croatian), who kindly ferried us across Lake Vlasina in his boat and assisted us in our landing on the floating mass of vegetation. Difflugia petricola Cash, 1909 Sixteen specimens examined: body length 80-105 um, breadth 58-74 urn, diameter of aperture 21-29 um, similar to Figs 4-6, Ogden & Fairman, 1979. Difflugia pristis Penard, 1902 Eighteen specimens examined: body length 41-50um, breadth 22-39 urn, diameter of aperture 9-16 um, similar to Fig. 12, Ogden, 19836. Difflugia pulex Penard, 1902 Five specimens examined: body length 33-39 um, breadth 22-31 um, diameter of aperture 10-14 um, similar to Fig. 14, Ogden, 19836. 252 C. G. OGDEN Figs 25-30 Difflugia pecac sp. nov.: Fig. 25, lateral view to show the almost parallel sides x 1 500; Fig. 26, apertural view x 1 800; Fig. 27, view to illustrate the pointed aboral extremity and lateral compression x 1300; Figs 28-30, detail of organic cement network, each unit (arrowed) appears to be composed of a central ring surrounded by five or six similar rings x 35000, x 35000 & x 30000. TESTATE AMOEBAE Difflugia rubescens Penard, 1 89 1 253 Three specimens examined: body length 57-80 um, breadth 41-44 um, diameter of aperture 15-20 urn, similar to Plate 66, Ogden & Hedley, 1980. Detail of the organic tooth-like structures surrounding the aperture (Figs 31 & 32) are included here to show that this character is a distinct rim of smooth cement. The 'teeth' are irregular in distribution but can often be precise in construction (Fig. 33), and the cement is limited to a narrow band (Figs 32 & 34). Figs 31-34 Difflugia rubescens: Fig. 31, latero-apertural view showing the ring of 'tooth-like' structures x2800; Fig. 32, apertural view to show uneven distribution of 'teeth' x3800; Figs 33 & 34, detail of featureless organic cement rim and 'teeth' x 14000 & x 17000. 254 C. G. OGDEN Figs 35 & 36 (?) Difflugia: Fig. 35, lateral view x700; Fig. 36, apertural view, note the invaginated aperture x 800. Difflugia ventricosa Deflandre, 1926 Two specimens examined: body length 221 & 237 urn, breadth 191 & 219 um, diameter of aperture 85 & 1 14 um, similar to Fig. 39, Ogden, 1983&. (?) Difflugia sp. A single specimen 78 um long, 66 um breadth and diameter of aperture 23 um, was found which appears to represent a previously unknown species. The shell construction (Fig. 35) is typical of a Difflugia-type shell, but the invaginated aperture (Fig. 36) is more typical of the Centropyxis-type shell. It is here considered as a possible species of Difflugia. Euglypha acanthophora (Ehrenberg, 1841) Four specimens examined: shell length 75-85 um, breadth 38-41 um, diameter of aperture 19-22 um, similar to Figs 1-5, Ogden, 1981. Euglypha cristata Leidy, 1874 One specimen examined: shell length 54 um, breadth 1 5 um, diameter of aperture 8 um, similar to Plate 79, Ogden & Hedley, 1980. There are six apertural plates on this specimen, each being thickened at the denticular margin (Fig. 37) and having a small protrusion at the opposite margin. As noted earlier (Ogden & Hedley, 1980) the spines that protrude from the aboral extremity appear to be flexible, in this instance the flattened structures are curved back over the body surface (Fig. 38). Euglypha rotunda Wailes, 1911 Three specimens examined: shell length 37^0 um, breadth 21-32um, depth 15-16um, diameter of aperture 8-9 um, similar to Plate 82, Ogden & Hedley, 1980. Euglypha tuberculata Dujardin, 1841 Six specimens examined: shell length 64-82 um, breadth 28^47 um, diameter of aperture 13-20um, similar to Plate 84, Ogden & Hedley, 1980. A clonal culture of this species was TESTATE AMOEBAE 255 Figs 37 & 38 Euglypha cristata: Fig. 37, latero-apertural view to illustrate detail of apertural plates x 8000; Fig. 38, aboral region showing the spines folded onto the shell surface x 5000. maintained in the Museum for several months. Specimens from this culture were identical to those found from the wild, although in culture several specimens having a double comple- ment of shell and apertural plates were observed. This phenomenon of larger individuals has already been reported from other cultures of siliceous testate amoebae (Hedley & Ogden, 1973). Heleopera petricola Leidy, 1879 Seven specimens were examined: shell length 95-102 um, breadth 52-78 urn, depth 35-44 um, diameter of aperture 29^4 um, similar to Plate 27, Ogden & Hedley, 1980. Hyalosphenia papilio Leidy, 1875 Two specimens examined: shell length 1 12 & 1 18 um, breadth 67 & 70 um, depth 26 um, diameter of aperture 30 & 32 um, similar to Plate 25, Ogden & Hedley, 1980. Lesquereusia epistomum Penard, 1902 Two specimens examined: shell length 1 16 & 127 um, breadth 67 & 75 um, diameter of aper- ture 20 & 33 um. Although only two specimens were examined this was one of the most abundant species present in the samples, being easily distinguished by the distinct neck and ovoid shape (Figs 39 & 4 1 ). The typical cement network (Fig. 40) is the same as that described for L. spiralis by Ogden (19796). Lesquereusia spiralis (Ehrenberg, 1 840) Four specimens examined: shell length 94-108 um, breadth 75-97 um, diameter of aperture 20-32 um, similar to Plate 32, Ogden & Hedley, 1980. 256 C. G. OGDEN Figs 39-41 Lesquereusia epistomium: Fig. 39, lateral view, note the distinct neck x 700; Fig. 40, detail of organic cement x 25000; Fig 41, apertural view x700. Fig. 42, Quadrulella symmetrica: lateral view of elongate specimen x500. Nebela dentistoma Penard, 1 890 Two specimens examined: shell length 104 & 108um, breadth 67 & 77 urn, depth 51 & 56 urn, diameter of aperture 21 & 23 um, similar to Plate 37, Ogden & Hedley, 1980. Nebela lageniformis Penard, 1 890 DESCRIPTION. The shell is transparent, colourless or slightly yellow, ovoid with a distinct neck (Fig. 43) and laterally compressed (Fig. 44). It is composed of a mixture of circular, oval or elongate shell plates, the illustrated specimen is atypical being composed of mainly elongate plates. These plates are usually distributed at random and may overlap each other, a featureless cement binds them together (Fig. 47). The aperture has a small lip (Fig. 45) which is surrounded by a small rim of organic cement (Fig. 46). It is oval or like an elongated slit (Fig. 46). MEASUREMENTS (in um). Based on five specimens: shell length 98-120, breadth 50-69, depth 39-43, diameter of aperture 9-1 2 x 19-25. TESTATE AMOEBAE 257 43 Figs 43-47 Nebela lageniformis: Fig. 43, view to show the general outline with distinct neck x 1200; Fig. 44, apertural view x900; Fig. 45, detail of neck and apertural collar x2500; Fig. 46, view showing apertural opening and surrounding organic collar x2500; Fig. 47, detail of shell surface x 800. 258 C. G. OGDEN REMARKS. The specimens described here are in good agreement with both of Penard's earlier descriptions (1890 & 1902). Nebela penardiana Deflandre, 1936 Six specimens examined: shell length 99-1 69 um, breadth 50-86 urn, depth 27-62 um, diameter of aperture 17-32 um, similar to Plate 42, Ogden & Hedley, 1980. Nebela vitrae Penard, 1899 One specimen examined: shell length 126um, breadth 80 um, depth 45 um, diameter of aperture 26 um, similar to Plate 46, Ogden & Hedley, 1980. Netzelia tuberculata (Wallich, 1864) Netzel, 1983 Five specimens examined: shell length 105-139 um, breadth 86-1 20um, diameter of aperture 28-40 um, similar to Plate 67, Ogden & Hedley, 1980. This species originally described as Difjlugia tuberculata, was mentioned by Ogden (19796) as being a possible candidate for the genus Netzelia Ogden, 1979 because of the earlier ultrastructural studies of Eckert & McGee-Russell (1974) and Owen & Jones (1976). More recent studies by Netzel (1983) and Meisterfeld (as reported at the Illrd International Workshop on Testate Amoebae in Aachen, 1983) have shown that this species produces its own siliceous idiosomes plus the organic building material that binds these idiosomes together, and they therefore propose its transfer to the genus Netzelia. Paraeuglypha reticulata, Penard, 1902 Difflugia delicatula Gauthier-Lievre & Thomas, 1958 DESCRIPTION. A few live animals were observed, which had a large central nucleus in the posterior half of the cytoplasm, and a band of dense particles positioned in front of this across the mid-body region (Fig. 48). The latter feature is identical to the pigment zone seen in species of the family Euglyphidae. No pseudopodia were observed as the aperture was surrounded by an assemblage of particulate material (Fig. 49). The shell is transparent and pale yellow in colour, elongate ovoid in shape tapering at the aperture to a small collar and aborally to a thin slightly tapered central spine or projection (Fig. 50). The projection is built as an extension of the shell and varies in both length and shape, specimens with single and bifurcate points have been seen (Fig. 54). A mixture of small and medium-sized, oval, circu- lar and elongate shell plates, cover the shell surface (Fig. 52). An occasional diatom frustule has also been seen in this assemblage. Organic cement binds these particles together and is seen as a smooth sheet between some junctions. The aperture is round and usually surrounded by a small raised collar (Figs 51 & 53). MEASUREMENTS (in um). Based on 14 specimens: body length 79-98, breadth 34-44, diameter of aperture 12-19. REMARKS. Paraeuglypha was erected by Penard (1902) to accommodate a new species P. reticulata, an unusual animal which had shell plates similar to those described for Euglypha, but a thin aboral spine. The shell was yellow, elongate, pyriform or fusiform and not com- pressed, the shell plates were round or oval but often with irregular outlines. There was a zone of 'grains brillants', now referred to as the pigment zone, just anterior to the nucleus. Filiform pseudopodia were observed, but most frequently the aperture was surrounded by an assemblage of extraneous material. Penard considered it to be a rare species as he only found it in three samples. The present description agrees well within the animals described by Penard, there was a pigment zone in live animals and the shells are similar in colour, size and shape. TESTATE AMOEBAE 259 Figs 48-52 Paraeuglypha reticulata: Fig. 48, optical micrograph of living animal to show the dark pigment zone (arrowed) x500; Fig. 49, optical micrograph illustrating the nucleus (arrowed), extraneous material surrounds the aperture x500; Fig. 50, lateral view, note the small apertural collar and distinct aboral protuberance x 1 300; Fig. 5 1 , detail of apertural collar X4500; Fig. 52, detail of shell surface x6000. C. G. OGDEN Figs 53-56 Paraeuglypha reticulata: Fig. 53 apertural view x 1 300; Fig. 54, detail of aboral protu- berance, note the mixture of different shell plates x 3700; Fig. 55, specimen with shell composed almost entirely of circular shell plates x 1600; Fig. 56, detail of shell surface x7500. TESTATE AMOEBAE 261 A new species of Difflugia, D. delicatula was described by Gauthier-Lievre and Thomas (1958) as having an ovoid or ellipsoid shell with a long anterior spine. It was covered with small polygonal particles and one specimen had used small circular plates from the diatom Echinornic crassipes. Again it was considered to be a rare species. It would appear that this description was based on observations of shells alone, as observations of living animals were not recorded. It is also in good agreement with the present specimens, being composed of a mixture of different sized small particles, and is here considered to be a synonym of P. reticulata. The most interesting feature of the present specimens is the presence of a pigment zone, which imply that the animal has a facility for producing its own shell plates. Nevertheless, it does not appear to produce a shell constructed only of shell plates arranged in a regular, evenly spaced shell as seen in specimens belonging to the family Euglyphidae, but instead has an apparent mixture of shell plates and extraneous material. This assemblage of particles in some instances resembles a mixture more frequently seen in species of Nebela, but here the resemblance ends because of their irregular arrangement. The arrangement is most simi- lar to that seen in small fragile specimens belonging to the family Difflugiidae, but members of these two groups have lobose pseudopodia. In summary, it would appear that this species represents a link between those animals that have lobose pseudopodia, produce their own shell components and can also incorporate extraneous particles, for example Netzelia and Lesquereusia (see Ogden, 1979/?), with those which have filiform pseudopodia and a shell composed of arranged shell plates. Plagiopyxis penardi Thomas, 1955 One specimen examined: shell length 88 urn, breadth 43 um, diameter of aperture 34 urn, similar to Figs 7-11, Ogden, 1984. Pseadodifflugia gracilis Schlumberger, 1845 Two specimens examined: shell length 63 & 72 um, breadth 42 & 48 um, diameter of aperture 41 & 36 um, similar to Plate 76, Ogden & Hedley, 1980. Quadrulella symmetrica (Wallich, 1863) Four specimens examined: shell length 88-1 35 um, breadth 49-62 um, depth 33-37 um, diameter of aperture 19-24 um, similar to Plate 47, Ogden & Hedley, 1980. Three of these specimens were more elongate (Fig. 42) than the typical specimen shown in Plate 47 (Ogden & Hedley, 1980), and such specimens have been described as Q. symmetrica var. longicollis by Chardez (1964). However, these shells could represent the extra large specimens often seen in species of Euglypha, for example E. tuberculata as described on p. 254 above, and for that reason are not differentiated from the typical Q. symmetrica. Sphenoderia lenta Schlumberger, 1845 Two species examined: shell length 45 & 50 um, breadth 38 & 34 um, diameter of aperture 16 & 14 um, similar to Plate 89, Ogden & Hedley, 1980, and Figs 50-54, Ogden, 1984. Sphenoderia sp. Two specimens exmined: shell length 40 & 45um, breadth 28 & 27 um, diameter of aperture 14 um, similar to those specimens described by Ogden (1984, Figs 45^9) having fewer shell plates than the typical S. lenta. Trinema enchefys (Ehrenberg, 1838) Six specimens examined: shell length 39-60 um, breadth 22-29 um, depth 18-27um, diameter of aperture 10-13 um, similar to Plate 91, Ogden & Hedley, 1980. 262 C. G. OGDEN Discussion The agreement found between the organic cement patterns of species of Difflugia collected in Britain and Yugoslavia establishes this feature as a good taxonomic character. These organic elements are produced by the animal as individual units, they are used as such or combined assemblages to bind the shell components together. The difference in the structure of these elements is probably related to the type of construction built by a particular species, and there appears to be some correlation between the cement used by species which construct similar shells. This similarity does not necessarily relate to shape, although some are similar, but more especially to the type of material used and the way it is blended together in con- struction. For example, Difflugia parva and D. cylindrus have similar organic cement pat- terns and construct elongate, cylindrical shells having a rough surface using a mixture of angular quartz particles, whilst D. mamillaris and D. ampululla have another type of organic cement but construct different shaped smooth shells using mainly flattish quartz particles. Recent ultrastructural studies (Netzel, 1983) on shell formation in Netzelia oviformis have shown that these organic elements are prefabricated in the Golgi apparatus and transported through the cytoplasm to the plasmalemma. They are extruded and placed by finger-like processes between adjacent inorganic idiosomes, to which they become attached and form the shell casing. This behaviour is similar to the positioning of siliceous plates in shell formation by species ofEugylpha as described by Hedley & Ogden (1974) and Ogden (1979c). The finding of Paraeuglypha in the present samples appears to suggest another type of shell formation in the Class Filosea. To date all members of the family Euglyphidae are con- sidered to produce shell plates of uniform outline. These may be circular, oval or elongate, within a range of different sizes, a mixture of which may be incorporated in one shell, and some species additionally produce spines. Such siliceous shell plates are usually organized into distinct patterns, or regular arrangements, and no extraneous material is incorporated into the shell. Paraeuglypha comes close to the definition but is differentiated by producing irregular shaped shell plates and it also appears to incorporate some extraneous material. Acknowledgements I am most grateful to Dr A. Zivkovic for her kindness and generosity in organising our expedition to Lake Vlasina and to Miss D. Zivkovic for acting as interpreter and guide during my visit. References Cash, J. & Hopkinson, J. 1909. The British Freshwater Rhizopoda and Heliozoa. Vol. ii Rhizopoda part 2. 166pp. The Ray Society, London. Chardez, D. 1964. Thecamoebiens (Rhizopoda, Testacea). Expi hydrobiol. L. Bangweolo-Luapula 10 (2): 1-77. Deflandre, D. 1928. La genre Arcella Ehrenberg. Morphologic-Biologic. Essai phylogenetique et systematique. Arch. Protistenk. 64: 152-287. Eckert, B. S. & McGee- Russell, S. M. 1974. Shell structure in Difflugia lobostoma observed by scanning and transmission electron microscopy. Tissue Cell 6: 2 1 5-22 1 . Gauthier-Lievre, L. & Thomas, R. 1958. Les genres Difflugia, Pentagonia, Maghrebia et Hoogenraadia (Rhizopodes, testaces) en Afrique. Arch. Protistenk. 103: 241-370. Hedley, R. H. & Ogden, C. G. 1973. Biology and fine structure of Euglypha rotunda (Testacea: Protozoa). Bull. Br. Mus. nat. Hist. (Zool.) 25: 1 19-137. Milovanovic, D. & Zivkovic, A. 1956. Limnolska ispitivanja baraznog jezera na Vlasini. Zborn. Rad. Inst. Ekol. Biogeogr. 1 (5): 1-47. Netzel, H. 1983. Gehausewandbildung durch mehrphasige Sekretion bei der Thekamobe Netzelia oviformis (Rhizopoda, Testacea). Arch. Protistenk 127: 351-381. Ogden, C. G. 1979a. Comparative morphology of some pyriform species of Difflugia (Rhizopoda). Arch. Protistenk 122: 143-153. TESTATE AMOEBAE 263 — 19796. Siliceous structures secreted by members of the subclass Lobosia (Rhizopodea, Protozoa). Bull. Br. Mus. nat. Hist. (Zool.) 36: 203-207. — 1979c. An ultrastructural study of division in Euglypha (Protozoa: Rhizopoda). Protistologica 15: 541-556. 1980. Shell structure in some pyriform species of Difflugia (Rhizopoda). Arch. Protistenk. 123: 455^70. — 1981. Observation on clonal cultures of Euglyphidae (Rhizopoda, Protozoa). Bull. Br. Mus. nat. Hist. (Zool.) 41: 137-151. — 1983#. The significance of the inner dividing wall in Pontigulasia Rhumbler and Zivkovicia gen. nov. (Protozoa: Rhizopoda). Protistologica 19: 215-229. 1984. Shell structure of some testate amoebae from Britain (Protozoa: Rhizopoda). Journal of Natural History 18: 341-361. Ogden, C. G. & Fairman, S. 1979. Further observations on pyriform species of Difflugia (Rhizopoda). Arch. Protistenk. 122: 372-381. Ogden, C. G. & Hedley, R. H. 1980. An Atlas of Freshwater Testate Amoebae. British Museum (Natural History)^ London & Oxford University Press, Oxford. 222pp. Ogden, C. G. & Zivkovic, 1983. Morphological studies on some Difflugiidae from Yugoslavia (Rhizopoda, Protozoa). Bull. Br. Mus. nat. Hist. (Zool.) 44: 341-375. Owen, G. & Jones, E. E. 1976. Nebela tuberculata comb. nov. (Arcellinida), its history and ultrastructure. /. Protzool. 23: 485-487. Penard, E. 1 890. Etudes sur les Rhizopodes d'eau douce. Mem. Soc. Phys. Hist, nat Geneve. 31: 1-230. 1902. Faune Rhizopodique du Bassin du Leman. Geneve 700pp. Manuscript accepted for publication 20 March 1984 Description of a neotype for the holothurian Ocnus bmnneus (Forbes MS in Thompson, 1840) from Strangford Lough, Northern Ireland (Holothurioidea;Dendrochirotida) J. Douglas McKenzie Department of Zoology, Queen's University of Belfast, Belfast BT7 INN, Northern Ireland Introduction Ocnus bmnneus was first described in a brief footnote on p. 100 of a paper by Thompson on molluscs and other invertebrates of Ireland as follows: Holothuria brunnea Forbes MS Holothuria brown, angulated, suckers 6 to 8 in each row, tentacula long whitish, pinnated towards their extremities. Forbes. This minute Holothuria, generally under an inch in length, is the most common species taken by dredging in the loughs of Strangford and Belfast. However, Forbes himself in 1841 (p. 230) recorded the species from the Firth of Clyde and the Isle of Man when he gave a fuller description, referring it to the new genus Ocnus (cited as 'Forbes & Goodsir') together with Holothuria lactea Forbes & Goodsir, 1839. The name Ocnus was subsequently either synonymized with Cucumaria or inadmissably used for species other than the two named above until Panning (1949) revived it, designating brun- neus Forbes as type species. However, no type material is known to exist and considerable confusion has arisen over the identity of this little holothurian. Herouard (1889) considered brunneus to be a valid species of Cucumaria but Mortensen (1927) could find no differences, except in colour, between it and lacteus and regarded them as synonymous. Despite this, albeit with reservations, he accepted Cucumaria brunnea: sensu Koehler, 1921, as a valid species though not conspecific with O. brunneus (Forbes). Mortensen's opinion has led to specimens agreeing with the description of brunneus being described and figured under the name of Cucumaria lactea in several popular shore guides (Eales, 1952; Barrett & Yonge, 1958; Barnes, 1979). However, Cherbonnier (1951) examining (presumably) small brown specimens from Roscoff, Brittany, including three named brunnea by Herouard, concluded that these are conspecific with young C. planci (Brandt, 1835) from Mediterranean and Belgian localities. He therefore synonymized brunneus with planci, which he retained in Cucumaria. Panning (1971: 33-34) maintained Ocnus as a genus distinct from Cucumaria and discussed the prob- lem at some length, suggesting initially that three species together extend between Senegal and mid-Norway: O. planci ranging from Senegal to the west of England, O. lacteus from the east of England to Scandinavia, and O. brunneus in southern England and northern France (without mention of Northern Ireland). However, he finally agreed with Cherbonnier that brunneus represents young planci and the latter name, being older, should be accepted for the type species of Ocnus. Rowe (1970) adopted Mortensen's synonymy of brunneus with lacteus but accepted the genus Ocnus. In view of the controversy about the validity and identity of O. brunneus and its status as the nominal type species of Ocnus Forbes and Goodsir in Forbes, 1841 , the establishment of a neotype for brunneus in the absence of original Forbes or Thompson material is highly desirable. Specimens matching the description and figure of brunneus (Forbes, 1841) have Bull. Br. Mus. nat. Hist (Zool) 47(5): 265-272 Issued 25 October 1 984 266 J. DOUGLAS MCKENZIE frequently been found in Strangford Lough, usually on Modiolus modiolus (L.) shells. This sea lough is part of the type locality for brunneus as specified by Thompson (1840). From this material, a neotype of O. brunneus is now described and the status of this taxon dis- cussed. Brief descriptions of O. lacteus, O. planci and Aslia lefevrei (Barrois) are also given. Apart from material taken in Strangford Lough by the author, specimens of O. lacteus, planci and brunneus studied were kindly lent by: Miss A. M. Clark of the British Museum (Natural History), London; Mr B. Picton of the Ulster Museum, Belfast, and Dr B. O'Conner of University College, Galway. Systematic descriptions Ocnus Forbes & Goodsir in Forbes, 1841 Ocnus brunneus (Forbes in Thompson, 1 840) Holothuria brunnea Forbes in Thompson, 1840 : 100 (footnote). Ocnus brunneus: Forbes, 1841; 229-230, fig. ICucumaria brunnea: Herouard, 1889 : 682. ICucumaria brunnea: Koehler, 1921. Cucumaria lactea (pt.): Mortensen, 1927 : 402-403. ICucumaria planci: Cherbonnier, 1951 : 39-40. DIAGNOSIS. Tube feet in single rows; spicules predominantly knobbed buttons with more than four holes; body colour russet brown. DESCRIPTION. Neotype: BM(NH) reg. no. 1982.7.29.1 Strangford Lough; 15-25 metres. (Figs la, 2a-d, 3a). Body length in life, excluding introvert, 7mm; diameter 2mm for most of the length, the body truncated. Shrinkage in 70% alcohol was minimal. Fully extended intro- vert with tentacles almost equal to body length. Tentacles ten, eight long and two ventrally situated ones short. Ambulacra slightly raised, each bearing around ten large tube feet ending in conspicuous suckers, in an almost straight row. Body smooth, rather stiff and angular, pentagonal rather than circular in cross section, posterior end rounded. Spicules very numer- ous in the body wall and also present in the tube feet and tentacles. In the body wall predomi- nantly knobbed buttons with more than four holes (fig. 2a), tightly packed together, small, almost flat baskets (fig. 2b) superficial to the buttons. Less numerous, but not uncommon, large, smooth fenestrated plates (fig. 2c) also found and in the tentacles smooth fenestrated rods (fig. 2d). Overall body colour in life a russet brown, fading to greyish brown in spirit; tentacles, posterior end and tube feet creamy white, becoming darker on preservation; the ventral inter- ambulacra a darker brown than the rest of the body. In life small brown bodies noticeable within the tentacle trunks. The neotype being so small, the description of the internal anatomy which follows has been obtained from other, dissected specimens. When contracted the introvert occupies up to one third of the internal body volume. The retractor muscles are large. One large Polian vesicle is always found. Posterior to the introvert, the gut swells to form a smooth, thick walled stomach — Forbes' 'gizzard'. At the anterior end of the stomach there are a large number of small bodies that are seen as papillae in larger specimens (fig. 3a) but are only brown, warty structures in small specimens. In larger individuals these structures lose their distinctive coloration and are the same creamy-yellow as the rest of the internal anatomy. The gut loops once behind the stomach with a mesentery connecting the ascending and des- cending sections. In no specimens from Strangford Lough were gonads found and in only one was there evidence of a respiratory tree. Even in the smallest individuals the knobbed spicules almost invariably have more than four holes, eight being the most common number. Occasionally individuals of a much lighter brown than the russet commonly encountered were found. HOLOTHURIOIDEA 267 Fig. 1 Ocnus brunneus (a) drawn from life, colour russet brown; O. lacteus (b) preserved specimen from Strangford Lough, colour white; scale for (a) & (b) 1 mm; O. planci (c) preserved specimen from Naples, colour russet brown with faint, darker spots; Aslia lefevrei (d) preserved specimen from Garvellachs, Scotland, colour dirty-white to fawn with dark tentacles; scale for (c) & (d) 10 mm. All the animals examined were dredged from areas of between 15-25 metres in depth, characterized by the presence of the horse mussel Modiolus modiolus to whose valves (either alive or dead) specimens were often found adhering. Other holothurian species, including O. lacteus, were often taken in the same dredge samples. Captured specimens of O. brunneus crawl around aquaria and extend their tentacles regardless of season, suggesting that the species does not hibernate. In the summer months brunneus appeared to be more common and individuals larger than in the winter. Similar specimens from Galway and the north-west of Scotland have now been identified as O. brunneus. 268 J. DOUGLAS MCKENZ1E I- •I Fig. 2 Ocnus brunneus (a) knobbed buttons from the body; (b) flat basket from the body; (c) flat plate from the body; (d) perforated rod from the tentacles. O. planci (e) & (f) knobbed buttons from the body; (g) flat baskets from the body. Scale bar 80 urn. a be Fig. 3 Stomachs from Ocnus brunneus (a); O. lacteus (b); O. planci (c). Ocnus planci (Brandt) DIAGNOSIS. Tube feet in distinct double rows, length up to 15 cm; spicules predominantly knobbed buttons with more than four holes; overall colour russet-brown, sometimes with darker spots; skin relatively smooth. (Figs Ic, 2e-g, 3c) Ocnus lacteas (Forbes & Goodsir) DIAGNOSIS. As found in Strangford Lough. Tube feet in single zig-zag rows; length up to 4 cm; spicules predominantly knobbed buttons with four holes, basket spicules almost flat; overall colour brilliant white. (Figs Ib, 3b, 4a-d) Aslia Rowe Aslia lefevrei (Barrois) DIAGNOSIS. Tube feet in double rows but individual tube feet often also found in interradii; length up to 15 cm; spicules predominantly knobbed buttons with four holes, basket spicules very deep in shape; colour from dirty white to black. (Figs Id, 4f-h) HOLOTHURIOIDEA 269 Fig. 4 Ocnus lacteus (a) knobbed buttons from the body wall of Strangford specimens; (b) flat basket from body; (c) large rod from body; (d) flat plate from body; (e) knobbed buttons from the body wall of Icelandic (large) specimens. Aslia lefevrei (f) knobbed buttons from body; (g) flat plate from body; (h) deep baskets from body. Scale bar 80 urn. N.B. The knobbed buttons of Aslia and all the Ocnus species may show variation within individuals in the number of perforations in each button but it is the most common button form within an individual that is important in species identification. Discussion The illustration of O. brunneus and its accompanying text in Forbes (1981) are sufficiently detailed to confirm that the species described above is identical to that described by Forbes. Two possibilities exist as to the status of brunneus: either it is a valid species or it is synony- mous with another. While O. brunneus is found with lacteus in some localities (Herouard, 1889; pers. obsv.) their geographical ranges differ: brunneus extending only to the outer Hebrides of Scotland but lacteus being known from as far north as Iceland (Einarsson, 1948). In Strangford Lough the two species are similar in size and, to a lesser extent, in shape but are usually very distinct in colour. The suggestion that Forbes was wrong to recognize two species and that brunneus and lacteus are synonymous stems entirely from Mortensen (1927). Unfortunately the char- acters on which this conclusion was based are not discussed; nor does he mention the locality from which his specimens were obtained. His illustrations of the knobbed spicules of lacteus show more than four holes, contradicting his description (Mortensen, 1927: Fig. 23 1 (3)) and suggesting a possible reason for his opinion that the two are synonymous. Strangford speci- mens ofO. lacteus of similar size to those of brunneus mainly have only four holes in their knobbed plates, these closely resembling the plates found in Aslia lefevrei. Large specimens from Iceland which I have named as lacteus (Ulster Museum, ZB 234, ace no. Mu 45 1) have knobbed spicules similar to those illustrated by Mortensen (1927) and are distinguishable from the knobbed plates of O. brunneus only in their larger size (Fig. 4e). Interestingly, in the largest of these specimens (>2-5 cm) the tube feet show the suggestion of the formation 270 J. DOUGLAS MCKENZIE of double rows though most are still in obviously single, zig-zag rows. Possibly Mortensen compared brunneus with large (or Scandinavian?) lacteus and assumed that the variation in the plates of lacteus would be paralleled in brunneus. The warts and papillae on the stomach of O. brunneus were absent from all the specimens of lacteus examined (Fig. 3). The single polian vesicle found in brunneus may only reflect early development but in lacteus of similar size two vesicles were always present. These differences, added to those of colour, indicate strongly that lacteus and brunneus cannot be considered synonymous. The only other species of Ocnus known from western European waters is O. planci (Brandt, 1835) mainly found in the Mediterranean (Koehler, 1921, type locality unknown), though Cherbonnier (195 1) recorded it from Belgian waters. In British seas A. lefevrei is often misidentified as O. planci and I have yet to see an undoubted specimen from the British Isles. If large adults are present in this area then they must be very rare to have escaped the notice of diving naturalists and the dredge. Furthermore, O. planci is known to have direct development, without a planktonic stage (Mortensen, 1927). As planci has never been recorded from Strangford Lough, and is very unlikely to occur there, it seems improbable that the brunneus found there are juvenile planci. However, three specimens in a sample collected from Liverpool Bay in 1889 (BM(NH) 89.12.3.40-42) and one from Galway Bay (O'Conner, pers. collection) are intermediate between brunneus and planci, having spicules indistinguishable from specimens of brunneus from Strangford Lough and planci from Naples (BM(NH) 98.5.3.252-5) but have, or are forming, double rows of tube-feet while they resemble brunneus in shape and colour. The largest of these four specimens is 31 mm long. Gonads are also forming. These intermediates suggest that brunneus may be a small form of planci, possibly a non-breeding, juvenile form, but, unlike more southern populations, those from the British Isles probably seldom achieve much more than one centimetre in length and only sexually reproduce under ideal conditions. This reproduction could be the source of all the new individuals in the populations but it seems more likely that asexual reproduction makes the major contribution in maintaining local populations. Asexual repro- duction through fission is well known in O. planci though its contribution to the population biology of this species is unknown (Hyman, 1955). O. brunneus is also capable of fission (pers. observ.) and when found on Modiolus shells several individuals may be found on the same valve. A specimen ofO. lacteus in the Ulster Museum is preserved in the act of dividing and is especially interesting in that a smaller individual is dividing from the side of a larger one rather than constricting in the middle and pulling into two halves as has been observed in brunneus. O. lacteus is also found on Modiolus shells but more usually in dead barnacle shells adhering to the valves. Again two or more individuals are often found in the same barnacle shell. While some other mechanism rather than asexual reproduction could explain such clumping, the importance of fission in the population dynamics of Ocnus species deserves further study. If the brunneus populations are self maintaining without the adult planci form then there is the very interesting possibility of paedomorphic separation of brunneus and planci. O. lacteus may represent such a paedomorphically evolved species deriving from a larger form with double rows of tube-feet, similar to A. lefevrei. The above does not, of course, preclude the possibility that such separation has already occurred and brunneus is already a distinct taxon. At present, however, the evidence is insufficient to allow resolution of this point. Further investigations of brunneus are clearly warranted; electrophoresis would be an inter- esting technique to employ in this and attempts to promote growth of brunneus into recog- nizable planci using heated aquaria would be interesting. The relationship between Ocnus and the monotypic genus Aslia also merits examination. Acknowledgements I would like to thank Dr Boaden for the use of facilities at the University Marine Station, Portaferry; Dr O'Conner and Mr Picton for the loan of material and for much useful discussion. I am grateful HOLOTHURIOIDEA 271 to Dr Gotto for reading the manuscript and his comments on it. My special thanks go to Miss Clark for her lengthy correspondence on this matter and her most valuable criticisms and discussion. References Barnes, R. 1979. The Natural History of Britain and Northern Europe: Coasts and Estuaries. 224 pp. London. Barrett, J. & Yonge, C. M. 1958. Collins Pocket Guide to the Seashore. 272 pp. London. Cherbonnier, G. 1951. Inventaire de la faune marine de Roscoff. Bryozoaries-Echinodermes. Travaux de la Station Biologique de RoscofflS Suppl. 4: 1-15. Eales, N. B. 1950. The Littoral Fauna of Great Britain. 305 pp. Cambridge. Einarrison, H. 1948. Echinoderma. The Zoology of Iceland IV: 70 pp. Copenhagen and Reykjavic. Forbes, E. 1841. A History of British Starfishes and other Animals of the Class Echinodermata. 270 pp. London. Herouard, E. 1889. Recherches sur les Holothuries des cotes de France. Archives de Zoologie Experi- mental et Generate (2) 7: 535-704. Hyman, L. H. 1955. Echinodermata: The Coelomate Bilateria: The Invertebrates. Vol. IV. 763 pp. New York. Koehler, R. 1921. Echinodermes. Fauna de France. 1: 210pp. Paris. Mortensen, T. 1927. Handbook of the Echinoderms of the British Isles. 471 pp. Oxford. Panning, A. 1971. Bemerkungen iiber die Holothurien — Familie Cucumariidae (Ordung Dendro- chirota). 6. Mitteilungen aus dem Hamburgischen Zoologischen Museum und Institut 67: 29-51. Rowe, F. W. E. 1970. A note on the British species of cucumarians involving the erection of two new nominal genera. Journal of the Marine Biological Association of the United Kingdom 50: 683-687. Thompson, W. 1840. Contributions towards a knowledge of the Mollusca Nudibranchiata and Mollusca Tunicata of Ireland with descriptions of some apparently new species of Invertebrata. Annals of Natural History S: 84-102. Manuscript accepted for publication 12 December 1983 Three new species of Varicorhinus (Pisces, Cyprinidae) from Africa K. E. Banister Department of Zoology, British Museum (Natural History), Cromwell Road, London SW7 5BD Introduction The genus Varicorhinus is contentious and a re-evaluation of it is in preparation. During the course of this work the existence of three new taxa became apparent and to avoid side issues and to make the larger work more coherent, they are described below. Varicorhinus jubae sp. nov. A recent sample of 13 fishes from the Juba river, near Sidam, Ethiopia consisted of 10 speci- mens of Barbus gananensis Vinciguerra and three specimens referable to Varicorhinus. No species of the latter genus have been recorded from the Juba system. TYPICAL SERIES. Holotype BMNH 1976.7.1:13; two paratypes BMNH 1976.7.1:14-15, respectively of 135, 101 and 77 mm SL. The fish were collected by Drs Yalden, Largen and Morris from Juba river, close to the Sidam-Bale bridge 05°45 ' N, 39°37 ' E, altitude 1200 m. The collectors describe the locality as 'permanently flowing river through Acacia woodland. The fish were caught in a shallow stretch where the water flows rapidly over shingle'. ETYMOLOGY. The specific name alludes to the Juba river and is treated as a feminine Latin noun. Fig. 1 The holotype of Varicorhinus jubae. Bull. Br. Mm. nat. Hist. (Zool.)47(5): 273-282 Issued 25 October 1984 274 K. E. BANISTER Description The description is based on the three fishes noted above. Measurements and abbreviations conform to those detailed in Banister, 1973. Apart from the standard length itself, all figures are a percentage of the standard length. 135 mm 101 mm 77 mm 30-6 27-3 26-3 23-3 23-3 24-7 5-9 6-4 7-3 8-5 7-9 7-6 8-9 7-9 8-6 23-7 21-8 21-4 18-5 19-3 17-8 12-2 10-9 11-7 8-1 7-4 7-2 2-9 1-9 2-6 3-7 2-9 4-5 31-9 32-6 30-3 The head is short and the snout bluntly rounded. The body is relatively deep and, in the largest specimen, gives the fish the appearance of having been a powerful swimmer. The two smaller specimens are more fusiform. Two pairs of short barbels are present. Small, off- white tubercles are distributed over the sides of the snout in all three fishes. The mouth is wide and ventral. The lower jaw is short, its anterior edge gently curved and covered with a sharp-edged horny sheath (Fig. 2). The 101 mm SL specimen has a gut length of 200 mm (the same length as in a sympatric Barbus gananensis of 103 mm SL). All the specimens are either sexually immature or quiescent. There are 39(fl) or 40(f2) vertebrae including those of the Weberian mechanism. Dorsal fin. There are four unbranched and 10 branched rays (f3). The last unbranched ray is ossified into a stout, straight, smooth spine. When erect, the dorsal margin of the fin is concave. A low sheath of scales envelops the base of the fin. The first ray is in advance of the vertical from the pelvic fin origin. Fig. 2 Ventral view of the head of the holotype of Varicorhinus jubae. NEW SPECIES OF VARICORHINUS 275 4mm Fig. 3 The striations of the fifth scale from the row above the lateral line of the holotype. The Anal fin has three unbranched and five branched rays (f3). The last branched ray of the holotype is split to the base giving the appearance of a sixth ray. Squamation. In the lateral line series there are 26, 27 and 3 1 scales. From the dorsal mid- line to the lateral line there are 4^ (O) scales and from there to the ventral mid-line also 4^ (f3) scales. There are 12 (D) scales around the least circumference of the caudal peduncle. Between the lateral line and the pelvic fin base there are 1^ (f2) or 1\ scales. The pattern of the scale striations is shown in Fig. 3. Pharyngeal bones and teeth. The pharyngeal teeth number 2.3.5-5.3.2 (f3). The pharyngeal bone is depicted in Fig. 4 where it can be seen that it is smaller than that of an equivalently sized Barbus gananensis. Gill rakers. On the lower limb of the first gill arch there are 16 (fl) or 17 (f2) long, hooked gill rakers. Coloration. Alcohol preserved specimens are dark grey-brown dorsally, silver-grey laterally and pale grey ventrally. The fins are pale grey. In life, the fishes were intensely silver, darken- ing dorsally. DISTRIBUTION. This species is known only from the type locality. Comparison with Barbus gananensis The last re-description of Barbus gananensis (Banister, 1973) was based on only three long- preserved specimens so the extra ten specimens mentioned above (BMNH 1976.7.1:3-12) were very useful. These specimens 71 mm to 125 mm SL have shallower caudal peduncles (1 1-0-12-9 cf 13-1-15-0% SL) than the previous sample as well as fewer lateral line scales (26 (f5), 27 (f3), 28 (fl) and 29 (fl) c/29 (f2) or 3 1 (fl) suggesting that the previously reported range may be atypical. All other counts and measurements lie within the published limits but the head length, snout length and the length of both pairs of barbels lie at the upper ends of the ranges. In overall appearance, Varicorhinus jubae and Barbus gananensis resemble each other more closely than either does its respective congeners. Although such resemblances are unquantifiable, the reality can be seen by comparing Figs 1 & 6. The head is excluded from these comments. 276 K. E. BANISTER Fig. 4 Comparison of the pharyngeal bones of Varicorhinus jubae 101 mm SL (left), and Barbus gananensis 103 mm SL. 2mm Fig. 5 The gill rakers on the first ceratobranchial of Barbus gananensis (A) 103 mm SL, and Varicorhinus jubae (B) 101 mm SL. NEW SPECIES OF VARICORHINUS 277 The most conspicuous differences between V.jubae and B. gananensis occur in characters related to feeding. The wide mouth, sharp edge to the lower jaw, smaller barbels and higher number of gill rakers in the new species are all characters that are associated with scraping, epilithic dietary habits. The disparity in the pharyngeal bone size has been noted above and further discussed in Banister, 1972. The presence of a Varicorhinus species bearing a close superficial resemblance to a sympa- tric Barbus species has been observed in other instances. Varicorhinus ruwenzorii from the Ruimi river, Uganda, has the same striking colour pattern as Barbus somereni (Banister, 1972). From the Upemba region of Zaire a similar pairing phenomenon is shown by Varico- rhinus upembensis and Barbus gestetneri (Banister & Bailey, 1979). Although I have no immediate explanation for this, the phenomenon will be treated in detail in the reassessment of the genus referred to above. Varicorhinus clarkeae sp. nov. HOLOTYPE. A fish of 151 mm SL, No. 164456 in the Musee Royale de 1'Afrique Centrale, Tervuren, Belgium. Specimen No. 164457 is designated a paratype. Both specimens (pre- viously registered as Varicorhinus ensifer) were collected in the Rio Cunje, an affluent of the Cuanza, Ceilunga, Angola. ETYMOLOGY. Named in honour of Mrs Margaret Clarke who gave so much assistance during the course of these researches. Description Based on the two specimens of 1 5 1 and 1 6 1 mm SL. The morphometric data (holotype first) is as follows: D = 22-5, 23-6; H = 2O5, 20-5; 1 = 4-0, 6-3; IO = 8-0, 7-5; MW = 7-3, 6-8; Pct= 19-3; 18-6; Cpl= 14-6, 18-0; Cpd = 9-3, 10-6; Snt = 7-3, 6-8; Ab= 1-7, 1-0; Pb = 3-6, 2-5; Dfin = 21-8, 18-0. The body is shallow and nearly circular in cross section. The lower jaw has a slightly curved anterior edge is covered with a thin horny layer. Papillae are present on both lips, those on the upper lip decrease in size anteriorly and are larger than those on the lower jaw which are confined to a strip behind the horny sheath. The snout is fleshy and a thin Fig. 6 Barbus gananensis 103 mm SL. 278 K. E. BANISTER ventral flap covers the anterior edge of the upper jaw. There are no tubercles nor tubercle scars on the snout but what could be scars of small tubercles are present on the skin covering the lachrymal bone in the holotype. Two pairs of small barbels are present. There are 41(f2) vertebrae, including those of Weberian mechanism. Dorsal fin. There are 4 .unbranched and 9 branched rays (f2). The last unbranched ray is ossified into a thin, short, straight spine. There is no raised sheath of scales around the base of the dorsal fin. The Anal fin has three unbranched and five branched rays (f2). Squamation. The lateral line has 33 or 35 scales and is relatively high on the body (see below). There are 5£ (f2) scales between the dorsal mid-line and the lateral line and 5^ (f2) from there to the ventral mid-line. Three scales lie between the lateral line and the pelvic fin (f2). Twelve scales encircle the least circumference of the caudal peduncle. The scale striations are parallel. Gill rakers. There are 16 gill rakers on the lower limb of the first gill arch of the holotype. Pharyngeal bones and teeth. The pharyngeal teeth number 2.3.5-5.3.2. The posterior faces of the crown of the second and third teeth of the inner row (12 & 13 in the system of Banister & Clarke, 1980) have a horseshoe shaped ridge which becomes exaggerated on 14, 15, 112 and 113. Coloration. In alcohol preserved specimens the back is dark brown, the belly and flanks lighter brown. The demarcation between the two browns is very marked on the posterior part of the fish. DISTRIBUTION. Known only from Cunaza, Ceilunga, Angola, approximately 12°00'S, 17°40'E. Diagnosis and affinities The presence of papillae on the lips and a horny edge to the lower jaw suggests an affinity with Varicorhinus ensifer Boulenger and the species described below as this combination of features appears to be unique to these three species. Varicorhinus darkeae can be dis- tinguished from V. ensifer by a shorter, thinner dorsal fin spine and a more cylindrical body. 30mm Fig. 7 Varicorhinus darkeae the holotype. NEW SPECIES OF VARICORHINUS Varicorhinus robertsi sp. nov. 279 TYPICAL SERIES. Holotype, a fish of 65 mm SL from the Sanga waterfalls at the tailwaters of the hydroelectric dam at Sanga on the Inkisi river, Zaire, BMNH 1983.3.30:20; paratypes 18 specimens 38-7-64-0 mm SL from the same locality, BMNH 1983.3. .30.21-38. All the specimens were collected by Drs Tyson Roberts and Don Stewart on June 26 1973. ETYMOLOGY. Named after the ichthyologist and collector Dr Tyson Roberts. Description The description is based on the 19 specimens listed above. L D H I 10 MW Pet Cpl Cpd Snt Ab Pb Dfin 26-0 25-8 7-0 6-2 5-8 22-1 17-7 10-5 8-4 + 2-3 29-1 s.d 1-3 1-1 0-9 0-6 1-0 1-0 1-3 0-5 0-5 0-3 2-6 s.e. 0-3 0-3 0-2 0-1 0-2 0-2 0-3 0-1 0-1 0-1 0-6 range mm 38-7-65-0 24-1-28-8 23-4-27-8 5-9- 8-4 5-2- 7-4 5-9- 9-6 20-2-24-2 15-5-20-8 9-4-11-7 7-6- 9-4 1-6- 2-7 25-2-34-4 20mm Fig. 8 Varicorhinus robertsi the holotype. 280 K. E. BANISTER A further 9 specimens were collected (18-4-38-7 mm SL) but are not included in this description because their poor condition precluded accurate measurement. The body is compressed, its greatest depth is immediately in front of the dorsal fin. The dorsal profile is more convex than the ventral profile. The snout is fleshy, rounded and extends in front of the mouth. The ventral face of the head is concave just behind the mouth. The upper jaw bears rows of papillae (Fig. 9), one row appears on the lip of small fishes (circa 30 mm SL) and the number of rows increases with size. The lower jaw has papillae behind the sharp-edged horny sheath. None of the specimens has tubercules. There are minute anterior barbels. In the ten specimens radiographed there are 38 (fl), 39 (f5) or 40 (f4) vertebrae including those of the Weberian mechanism. Dorsal fin. There are 4 (f!9) unbranched and 8 (f2) or 9 (fl 7) branched rays. The last simple ray is formed into a thin, smooth spine. The dorsal fin origin is anterior to that of the pelvic fin. The anal fin has 3 unbranched and 5 (f!9) branched rays. Squamation. In the lateral line there are 39 (f9), 40 (f5), 41 (f4) or 43 (fl) scales. From the dorsal mid-line to the lateral line there are 6\ (f!2) or 1\ (f5) scales and from there to the ventral mid-line 6^ (f4) or 1\ (f!4) scales. Between the lateral line and the pelvic fin base there are 4 (f4), 4^ (f!3) or 4 (f2) scales. Eighteen (fl) or 16 (fl6) scales encircle the least circumference of the caudal peduncle. Scale counts were not obtainable on all specimens. There are few striations on the scales and their disposition is shown in Fig. 10. Pharyngeal bones and teeth. The teeth number 2.3.5-5.3.2. The last tooth on the inner row (15) has a conspicuous sulcus on the crown in the plane of the tooth. The pharyngeal bone is shown in Fig. 1 1 . Gill rakers. There are 10 (fl), 11 (f3) or 12 (G) on the lower limb of the first gill arch in the 7 specimens examined. Coloration. The body colour in alcohol preserved specimens is pale brown, only slightly darker on the back than on the belly. A narrow, dark strip covers the ridge of the back from Fig. 9 Ventral view of the head of the holotype of Varicorhinus robertsi to show the papillae. NEW SPECIES OF VARICORHINUS 281 1mm Fig. 10 The fifth scale from the row above the lateral line of the holotype of Varicorhinus robertsi. 2mm Fig. 11 Varicorhinus robertsi pharyngeal bone from the largest paratype. the occipital region to the dorsal fin origin. The smaller specimens have a dark spot at the base of the tail fin. The centre of the operculum is transparent and the gills are visible through it. The paired fins are hyaline. The caudal fin lobes have a streak of dark pigment and the dorsal fin has a dark, dorsal margin. The live colour is unknown. DISTRIBUTION. This species is known only from the type locality, the Sanga waterfalls, 4°50'S, 14°47' E.Zaire. DIAGNOSIS. From the other Varicorhinus species with papillae on the lips this species is easily distinguished by the presence of smaller scales. 282 K. E. BANISTER Acknowledgements Firstly, I wish to thank the collectors for their care in preserving and bringing back the speci- mens. Secondly, I would like to thank my colleagues, Dr P. H. Greenwood, Mr G. J. Howes and Mrs Margaret Clarke for their help and advice. The drawings are the work of Jack Shroeder (7), Mandy Holloway (8) and Gordon Howes: to them and the photographer, Mr Paul Richens, I extend my thanks. I am very grateful to Ms Julie Hacker for typing the manuscript. References Banister, K. E. 1972. On the cyprinid fish Barbus alluaudi Pelegrin: a possible intergeneric hybrid from Africa. Bulletin of the British Museum (Natural History) Zoology 24 (5): 261-290. A revision of the large Barbus (Pisces, Cyprinidae) of east and central Africa. Bulletin of the British Museum (Natural Historv) Zoologv 26 (1): 1-148. Banister, K. E. & Bailey, R. G. 1979. Fishes collected by the Zaire River Expedition, 1974-1975. Zoological Journal of the Linnean Society 66: 205-249. Banister, K. E. & Clarke, M. A. 1980. A revision of the large Barbus (Pisces, Cyprinidae) of Lake Malawi with a reconstruction of the history of the southern African Rift Valley lakes. Journal of Natural History 14: 483-542. Manuscript accepted for publication 20 December 1983 Phyletics and biogeography of the aspinine cyprinid fishes Gordon Howes Department of Zoology, British Museum (Natural History), Cromwell Road, London SW7 5BD Introduction This study was initiated by the examination of specimens of the Hwang-Ho dace, Leuciscus mongolicus (Kessler, 1876) in the collections of the American Museum of Natural History. Even a cursory examination cast doubt on their assignment to the genus Leuciscus, and a more detailed anatomical study, reported herein, indicates that the species should be allocated to a new genus. Study of this material has also led to a reappraisal of the genus Leuciscus and to a discussion of the phyletics and zoogeography of the aspinine cyprinids. Methods and materials As in previous studies (Howes, 1978, 1980, 1981) a wide range of both barrelled and non- barbelled Old World cyprinids have been examined. The materials listed in those papers have been re-examined together with more recently prepared material. In addition, skeletal preparations of many Nearctic taxa have also been studied. The following species were dissected, or skeletal preparations examined. Abramis brama\ Algansea tincella; Aspiolucius esocinus; Aspiopsis merzbacheri; Aspius aspius; Aspius vorax; Chondrostoma nasus; Elopichthys bambusa; Gila bicolor; G. copei; G. crassicauda; G. cypha; G. elegans; G. nigrescens; G. pandora; G. robusta (these include Michigan Museum specimens); Lavi- nia exilicauda; Leuciscus borysthenicus; L. cephalus; L. idus; L. fellowesii; L. lehmanni; L. leuciscus; L. orientalise L. schmidtii; L. smyrnaeus; L. souffia; L. svallizae; L. waleckii (including the syntypes of L. waleckii sinensis in the Swedish Museum of Natural History); Luciobrama macrocephalus; Mylo- cheilus caurinus; Ochetobius elongatus; Oreoleuciscus humilis; O. pewzowi; O. potanini\ Orthodon microlepidotus; Pelecus cultratus; Pogonichthys macrolepidotus; Ptychocheilus lucius; P. grandis; P. oregonensis\ Rhynchocypris variegatus; Tinea tinea; Tribolodon brandti; T. jouyi; Xenocypris argenteus. The concept of Leaciscus and the status of Leuciscus mongolicus (Kessler) The daces and chubs of the genus Leuciscus Cuvier, 1817 form the most speciose group of Palearctic cyprinid fishes, there being at least 36 nominal species (the number listed in the BMNH catalogues). The majority of Leuciscus species are alike in having moderately deep and stout bodies, broad cranial bones (including the ethmoid and supraorbital), a ventrally directed basioccipital process with a well-formed masticatory plate, biserially arranged uncinate pharyngeal teeth, and a short-based anal fin. Comparative studies (Howes, 1978, 1980, 1981) suggest that these 'diagnostic' characters are plesiomorphic for non-barbelled cyprinids. Leuciscus, as presently recognized, cannot be defined by a set of unique characters and is therefore a non-monophyletic assemblage. One species of 'Leuciscus', L. mongolicus (Kessler, 1876), has, however, a suite of derived characters that sets it apart from the corpus of Leuciscus species (detailed in Table I). Some of these specializations are shared with genera of the aspinine group sensu Howes, 1978 (see pp. 29 1 below) and are as follows: Bull. Br. Mus. nat. Hist. (Zool.) 47(5): 283-303 Issued 25 October 1984 284 G. J. HOWES Table 1 Character Leuciscus spp. 'Leuciscus ' mongolicus Cranial width (% of its length from tip of ethmoid indentation to the posterior border of the superaoccipital) Supraethmoid Mesethmoid Maxillary mid-lateral ascending process Dentary pores Pterosphenoid Orbitosphenoid keel Basioccipital process Dilatator fossa Epioccipital process 4th infraorbital canal Operculum Adductor arcus palatini muscle originates from: Gill-rakers 40% mode of 20 specs representing 6 species (110-2 30 mmSL) with slightly concave lateral border deep with wide anterior notch low 5-6 short, lacking shelf and remote from frontal border deep ventrally sloped short, sphenotic with straight posterior border short, rounded follows contour of orbit dorsal border short, posterior border rounded parasphenoid and prootic few, stout and simple Gap between branchial arch and restricted (Fig. 1 1 ) bucco-pharyngeal roof Genital papilla not pronounced 61% mode of 9 specs (72-147 mm SL) narrow-waisted shallow with V-shaped notch high 8 elongate, with lateral shelf and extended to the frontal border shallow horizontal long, sphenotic with concave posterior border extended, triangular divergent from orbital border dorsal border long, posterior border concave in outline pterosphenoid, parasphenoid and prootic many, slender with a crenate medial membrane extensive (Fig. 1 1 ) elongate, with deeply folded border The cranium of 'Leuciscus ' mongolicus more closely resembles that of Aspius than that of any other Leuciscus species (Fig. 1 ). This resemblance is due to its narrowness, particularly that of the suprathemoid and the frontals which are markedly concave above the orbit; exten- sion of the posterior border of the epioccipital; extent of the dilatator fossa; width of the pterosphenoid; shallowness of the orbitosphenoid keel and the horizontal plane of the basioccipital process. ASPININE CYPRINID FISHES me 285 le so Fig. 1 Crania of A, Leuciscus leuciscus; B and C, Genghis mongolicus. A and B in dorsal, C in ventral views. Scale = 5 mm. epo = epioccipital, exo = exoccipital, f = frontal, le = lateral eth- moid, me = mesethmoid, pro = prootic, pte = pterotic, pts = pterosphenoid, se = supraethmoid, so = supraorbital, soc = supraoccipital, sp = sphenotic. Cranial width varies in cyprinids and cannot, by itself, be treated as a synapomorphy. However, among the aspinine genera (Aspius, Elopichthys, Pseudaspius, Aspiolucius and Luciobrama; Howes, 1978) there is, associated with a narrow cranium, marked concavity of the frontal border above the orbit and an elongate ethmoid region. The lateral margin of the supraethmoid is deeply concave and the mesethmoid arms prominently extended forming a V-shaped anterior notch. All these features are characteristic of 'Leuciscus' mongolicus (Figs 1 & 2). An aspinine synapomorphy (Howes, 1978) is the posterior extension of the epioccipital which, in combination with the lengthened parietal and flattened lateral portion of the supraethmoid, forms an occipital platform (postcranial platform of Howes, 1978). In 'Leucis- cus' mongolicus the occipital platform is not developed to the extent that it is in aspinine genera, but nonetheless this feature is absent in Leuciscus species. The dilatator fossa in 'Leuciscus' mongolicus makes a deep excavation into the frontal (Fig. 1 B); the sphenotic is expanded posteriorly with a deeply concave posterior margin. This fossa morphology is unlike that in other Leuciscus species where the frontal is only shallowly indented and the sphenotic is short with, at best, a slightly concave posterior border. A long, deep indentation of the frontal and a deeply concave sphenotic are characteristics of the aspinine dilatator fossa (see Howes, 1978, figs 25 & 26). 286 G. J. HOWES me Fig. 2 Anterior cranial region of A, Leuciscus leuciscus and B, Genghis mongolicus in lateral views. Scale = 5 mm. oss = orbitosphenoid 'septum', pe = preethmoid, v = vomer; other abbre- viations as in Fig. 1 . The lateral extension of the pterosphenoid is an aspinine synapomorphy, a feature most marked in Aspius and Elopichthys. The most extreme pterosphenoid expansion occurs in Elopichthys where the bone is also exposed dorsally and forms the site of origin for part of the adductor mandibulae musculature (see Howes, 1978: 32 & 53). In 'Leuciscus' mongoli- cus the pterosphenoid is narrowly separated from the margin of the overlying frontal and resembles the condition in Aspius (cf Fig. IB with fig. 27B in Howes, 1978). As in the aspi- nine genera, the pterosphenoid is elongate, being longer than the orbitosphenoid and having a prominent, downwardly curved lateral shelf (Fig. 2B). The usual cyprinid condition is for the orbitosphenoid and pterosphenoid to be of equal length or for the pterosphenoid to be shorter and without a lateral shelf. The orbitosphenoids in 'Leuciscus' mongolicus are joined to the parasphenoid by a shallow keel or 'septum', thus contrasting with the condition in other Leuciscus species where a deep keel is present (Fig. 2A). Again, this condition resembles that of the aspinines where contact between the orbito- and parasphenoid is direct or via a shallow keel (see Howes, 1978:3 1). The anterior myodome in Leuciscus' mongolicus has suffered reduction as a consequence of the depressed anterior part of the cranium. In other Leuciscus species the myodome is a spacious cavity. In the arrangement of its infraorbital bones, 'Leuciscus' mongolicus resembles the aspinines more closely than it does other Leuciscus species (Fig. 3 A). Synapomorphic for aspinine genera is the elongate 3rd infraorbital, the wide separation of the 4th infraorbital from the orbital border, and its divergent angle (Howes, 1978, fig. 20). In 'Leuciscus' mongolicus there is a similar elongation of the 3rd infraorbital and a divergent ASPININE CYPRINID FISHES 287 Fig. 3 Infraorbital series of A, Genghis mongolicus; B, Leuciscus leuciscus; C, Aspiopsis merzba- cheri; D, Tribolodon brandti. 4th element. Leuciscus species, in common with most cyprinids, have the 4th infraorbital canal in a near vertical alignment and following the contour of the orbital border (see Howes, 1978, fig. 21). The posterior extent of the bone is variable. What may represent the plesio- morphic condition is illustrated in bariliines (Howes, 1980, figs 29 A & B). In these taxa the 4th infraorbital canal also takes a near vertical course and the lamellar part of the bone covers the adductor musculature. In aspinines and 'Leuciscus' mongolicus the 4th infraorbital is also expanded, its posterior margin being attenuated, but the canal runs through the centre of the bone at a divergent angle to the orbit (Fig. 3 A). This particular configuration of the infraorbital in the aspinines and 'Leuciscus' mongolicus is a correlate of the elongated posterior cranial bones, particularly the pterosphenoid. The result has been a backward shift of the pterotic-infraorbital canal commissure and a re-alignment of the 4th and 5th infraorbital canals. The arrangement of the adductor arcus palatini (AAP) musculature in 'Leuciscus ' mongo- licus differs from that of other Leuciscus species in having its origin not only from the more usual sites viz the prootic and parasphenoid, but also from the pterosphenoid. The anterior part of the muscle originates tendinously from the prominent lateral pterosphenoid shelf and the lateral surface of the parasphenoid ascending process. Insertion of the muscle is on to the lateral face of the posterior margin of the entopterygoid and the entire dorso-lateral face of the metapterygoid (Fig. 4A). The muscle is thick and convex; posteriorly it is confluent with the adductor hyomandibularis which runs from the prootic to the medial face of the hyomandibula. In Leuciscus species and the majority of cyprinids, the AAP extends from the base of the neurocranial part of the parasphenoid to its orbital portion. Often, the muscle is continuous with the adductor hyomandibularis which originates from the posterior part of the prootic. 288 G. J. HOWES .ptss aap Fig. 4 Origins and insertions of the adductor arcus palatini muscle in A, Genghis mongolicus; B, Oreoleuciscus pewzowi and C, Pogonichthys macrolepidotus, lateral views. In A, the dotted line indicates area of origin of the 'adductor hyomandibularis' portion of the muscle from the cranium; the dash-dot line, its insertion on the hyomandibula. In B, the tendinous origin of the muscle is indicated by dashed lines, aap — adductor arcus palatini muscle, ent = entopterygoid, hy = hyomandibula, met = metapterygoid, ptss = pterosphenoid shelf. In cyprinids investigated, only in genera of the aspinine group (except Luciobrama which lacks an AAP; see Howes, 1978:24), and in Oreoleuciscus, Tribolodon and Pogonichthys does the AAP originate from the pterosphenoid. For these taxa the condition is considered to be synapomorphic. There are 3-4 gill-rakers in 'Leuciscus' mongolicus on the 1st epibranchial, with 8-9 on the outer and 12 on the inner side of the 1st ceratobranchial. Those on the inner side closely intermesh with the 13 or so rakers on the outer edge of the 2nd ceratobranchial. The gill- rakers are well-developed with a strong, triangular bony spine supporting a thin medial mucosal membrane, the border of which is convex and crenate. The membrane is most developed on the posterior rakers of the ceratobranchial (Fig. 5A). In Leuciscus leuciscus and the majority of its congeners, the bony core of the gill-raker is a short, flat, almost equilateral triangle invested by mucosal tissue. This is in contrast to the 'L.' mongolicus morphotype, where the bony part of the raker is exposed laterally. Berg's (1949) diagnosis of Leuciscus gives gill-rakers as 'short, few (6-30)'; in his key there are three species with more than 13 rakers, viz. bergi, lindbergi and schmidtii. No specimens of the two former species are available to me, but in L. schmidtii the posterior gill-rakers are of the same morphotype as in 'L. ' mongolicus. Gill-rakers with a crenate border to the medial membrane are present also in genera of the aspinine group and in Aspiopsis, Tribo- lodon, Oreoleuciscus and Pogonichthys. Comprison of gill-raker types in several cyprinid taxa has shown that this form of raker is comparatively rare. Normally the mucosal mem- brane has a plain concave border, but in some taxa, e.g. Cyprinus, the rakers have a thick and highly folded mucosal membrane. Dendritic and pulvinate gill-raker membranes are also common, the tissue often papillose as in abramines. However, in these taxa the pulvinate membrane meets a thick longitudinal septum (see Zander, 1903 for description ofAbramis) and the rakers lie close together so that the fimbriate medial margins form a sieve. ASPININE CYPRINID FISHES 289 Fig. 5 Gill-raker morphology in A, Genghis mongolicus; B, Oreoleuciscus humilis; C, Tribolodon brandti; D. Aspius aspius; E, Aspiopsis merzbacheri; F, Pogonichthys macrolepidotus. Allx 12, except B= x25. A and E, left, others right 1st ceratobranchial. Those features analysed above which serve to distinguish 'Leuciscus' mongolicus from other Leuciscus species are specializations shared with genera of the aspinine group (see below, p. 291. 'Leuciscus' mongolicus is, however, excluded from that group since it lacks the three synapomorphies denning it, namely, a high vertebral number, and numerous frontal, nasal sensory pores, and elongated pterosphenoid. As such it is necessary to assign 'Leuciscus' mongolicus to a new genus: GENGHIS gen. nov. TYPE SPECIES. Squalius mongolicus Kessler, 1876. In: Prejevalsky, N. Mongolia i strana Tangutov 2 (4): 21, pi. II. Type locality, Dalai Nor. ETYMOLOGY. After Genghis (Khan), below whose rampart lies the type locality, Lake Dalai. DIAGNOSIS. Medium-sized cyprinid fish (the largest specimens measured, 225-5 mm SL), slender-bodied, distinguished from other non-barbelled cyprinids by a combination of the following features: somewhat humped nuchal profile; dorsal cranial profile gently sloped; mouth set obliquely at 45°; border of the 4th infraorbital attenuated and widely separated from the orbit; elongate pterosphenoid, laterally expanded and possessing a lateral shelf from which originates the adductor arcus palatini muscle; operculum with attenuated lower posterior border; gill-rakers spinous with crenate medial border; long gape between the bran- chial arch and pharyngo-buccal roof (Fig. 11); lateral line scales large; caudal fin deeply emarginate. 290 G. J. HOWES Fig. 6 Outline drawings of A, Genghis mongolicus; B, Aspiopsis merzbacheri; C, Oreoleuciscus humilis. Genghis mongolicus (Kessler, 1876) (Fig. 6A) Squalius chinensis Prejevalski, 1876 Mongolia istrana Tangutovl: 135 nomen nudum Squalius mongolicus Kessler, 1876. In: Prejevalsky, N. Mongolia i strana Tangutov 2 (4): 21, pi. II, fig. 2. Squalius chaunchicus Kessler, 1876 Ibid: 23 Leuciscus farnumi Fowler, 1889 Proc. Acad. nat. Sci. Philad.: 179 Leuciscus mongolicus, Berg, 1912. Fauna de la Russie 3: 92 ^Leuciscus mongolicus Oshima, 1926 Zoological Magazine 38: 100 Leuciscus (Idus) sp. Miyadi, 1940 Fishes of Manchuria In: Hydrobiological Investigations of Kwantung Province and Manchurian Empire: 41, fig. 30. Leuciscus waleckii suiyuani Mori, 1941 Zoological Magazine 53 (3): 183, fig. 2. NOTES ON SYNONMY Squalius Heckel, 1843 is a synonym of Leuciscus Cuvier, 1817. Berg (1912: 92 & 110) referred Squalius mongolicus Kessler to Leuciscus and considered S. chaunchicus Kessler a synonym, an opinion endorsed by Banarescu (1970). Fowler (1899) ASPININE CYPRINID FISHES 291 described Leuciscus farnumi from three specimens, the holotype being from Lake Dalai. Fowler makes no mention of Kessler's species from that lake and his description is undoubtedly that of mongolicus. Oshima (1926) described a new species named as Leuciscus mongolicus making no refer- ence to the Kessler species which he had obviously overlooked. Oshima's species was later synonymized by Mori (1934) with Leuciscus waleckii (Dybowski, 1869). From Oshima's description it is difficult to tell if his species is Kessler's mongolicus or is Dybowski's waleckii as was assumed by Mori. Mori's concept of Leuciscus waleckii certainly appears to corre- spond with Dybowski's description of that species, and his illustration shows a fish that is clearly different from Kessler's mongolicus. Attempts to trace Mori's and Oshima's speci- mens have so far failed. Whether or not Oshima's Leuciscus mongolicus is synonymous with Kessler's Squalius ( = Leuciscus) mongolicus, the fact remains that the name proposed by Oshima is a junior homonym for which I propose, as a nomen novum, jeholi as indicative of that species' locality. Banarescu( 1970) recognized Mori's (1941) subspecies L. waleckii suiyuani as a synonym of L. mongolicus, an opinion with which I would concur. I can also confirm Banarescu's (1970) opinion that Leuciscus waleckii sinensis Rendahl (1925) which also occurs in the Hwang-Ho drainage does indeed belong to that species complex. The possibility that Rendahl's subspecies might also be synonymous with Kessler's L. mongolicus has been dis- pelled by examination of the three syntypes which conform in every respect with Dybowski's Leuciscus waleckii. DESCRIPTION. The following description amplifies that given under the generic diagnosis and is based on the following 18 specimens; AMNH 10907, 4 specs 138-146 mm SL; BMNH 1983.3.1:3-4, 2 specs 130-5; 147 mm SL (ex AMNH 10907); AMNH 10913, 7 specs 72-1 21 SL: AMNH 10908, 3 specs 93, 106, 145 mm SL; AMNH 10906, 2 specs 154, 222 5 mm SL. All from Paotow (Pao-t'ou), Suiyan Province, Inner Mongolia; collected by C. H. Pope. As % of SL: body depth 20-1-27-7 (M25-5); head length, 26-2-29-2 (M26-2); caudal peduncle length, 15-9-21-9 (Ml 8-3), caudal peduncle depth as % of its length, 49-2-63-9 (M58-0); as % of head length, interorbital width 24-5-32-0 (M28-8); snout length 21-3-29-6 (M25-7); eye diameter 17-8-23-6 (M22-1); opercular length 29-7-35-5 (M32-9). Gill-rakers spinous, 3-4 on 1st epibranchial 18-10 on 1st ceratobranchial; extensive gap between branchial arch and bucco-pharyngeal roof (Fig. 12); pharyngeal bones slender, teeth biserial, slender, hooked numbering 5-3 (B), 4-2 (fl), 4-3 (f2). Scales 9-10/50-52/7-8; Kessler gives a lateral line count of 54 and it is apparent from his figure that he was counting those pore-bearing scales extending onto the base of the caudal fin. Banarescu (1970) gives a count of 52-54 for the type specimen of G. mongolicus. My counts are those of the standard length. Dorsal fin with III, 7 (f!5), III, 8 (f3) rays; anal fin with III, 8 (f2) or III, 9 (f!6) rays. Pectoral rays I, 16 (f!3), I, 17 (O), I, 18 (f2), I, 19 (fl); pelvic rays I, 8 (f9) I, 9 (f7). Swimbladder is two-chambered, the posterior chamber reaching to above the genital open- ing. Genital papilla prominent with plicate margin. Small pectoral flap and an elongate pelvic axial scale present. Caudal fin emarginate, lobes pointed. Distribution. The type locality is the lake Dalai (now Hu-lun Ch'ih) in the Nei Monggol, China, 48° N, 117°E; it lies in the plain between the Mongolian Plateau and the Ta Hsing-an-ling Shan-mo (Greater Khinghan) range. According to Berg (1949) there is no out- flow of the lake. However, two main rivers flow into Dalai, that from the Mongolian Plateau is the Kerulen (Herlen or Ko lu-lun-Ho) and that from Lake Buyr in the Khingan moun- tains, is the Orxon (Wu-erh-Shun-Ho). The specimens examined are all from Paotow (Baotaou) on the Hwang-Ho river, some 88° south of the type locality. Relationships of Genghis and the aspinines The character analysis given above suggests that Genghis is closely related to the assemblage of five genera recognised by Howes (1978) as the aspinine group, viz.; Luciobrama, Pseud- 292 G. J. HOWES aspius, Aspiolucius, Aspius and Elopichthys. The derived characters uniting these genera are (1) posteriorly and laterally extended pterosphenoid, (2) elongation of posterior cranial bones with development of an occipital platform, (3) unique configuration of the infraorbitals, (4) high vertebral number, and (5) many nasal and frontal sensory pores. Characters (1H3) are possessed by Genghis and have already been discussed; (4) and (5) are shared only amongst the aspinine group genera, and are discussed below. Character (4). All aspinines have a total vertebral number in excess of 50 (51-55). Apart from Pelecus, with 52 and Ochetobius with 6 1 , no other cyprinid has such a high vertebral count (see Howes, 1978). In the aspinines the increase is in the abdominal vertebrae. In three genera, however, namely Aspius, Elopichthys and Aspiolucius, there is a high number of caudal vertebrae, 22-24 cf. 2 1 in Luciobrama and Pseudaspius, a figure that compares with the modal number of other cyprinids viz. 24 (calculated in part from figures published in Howes, 1978, table 1, and from unpublished data). Pelecus and Ochetobius have both high abdominal and caudal counts. Genghis has a total of 45 vertebrae. Character (5). Aspinine genera share a high number of nasal and frontal sensory pores, respectively 8-10 and 10-22. These counts are unusually high amongst cyprinids; in general the nasal is a short bone with 2-3 pores (exceptionally 6 in some cultrines) and the commo- nest number of frontal pores is 5-6. Some abramine taxa have 9-10 frontal pores (e.g. Hypo- phthalmichthys, see Howes, 1981 : 17), but the frontal morphology of the abramines and their recognition as monophyletic on the basis of other synapomorphies, suggests an independent derivation of increased sensory pore numbers. High numbers also occur in Oreoleuciscus (see below) and in species of the Nearctic genera Ptychocheilus, Gila, Lavinia and Pogo- nichthys. These genera, as is the case with aspinines, tend to have elongate crania and it may be that increased pore number is a straightforward correlation with cranial length. This does not always follow, however, since many cheline, bariliine, cultrine and schizothoracine taxa also have lengthened crania but show no sign of an increase in frontal pore number. By itself it would be dubious to treat a high frontal pore number as synapomorphic but in combination with increased numbers of nasal and mandibular pores it seems a valid synapo- morphy for aspinine taxa. Whether this is also the case for the high frontal pore number in the Nearctic taxa demands further investigation. A character overlooked by Howes (1978) when considering aspinine group synapo- morphies is the extreme development of the coronomeckelian bone. The usual cyprinid con- dition is for the bone to be small and irregularly shaped, with a medial shelf on to which inserts the tendon of muscle A2. Among the aspinines, the coronomeckelian bone of Lucio- brama is the most derived, being a long, almost boomerang-shaped element with a wide medial shelf (Fig. 7C). In Elopichthys the coronomeckelian has an irregular shape but with a long anterior process. The shape of the bone in Aspius is most like that of Genghis, being broadly triangular with a wide medial shelf (Figs 7A & B). Departure from the general cypri- nid condition also occurs in Tribolodon, Oreoleuciscus, Pogonichthys and Ptychocheilus where it is long and triangular (Figs 7D-F). A similarly shaped bone is present in some Phoxinus species. The value of this character is difficult to judge, as its development may be related to the insertion of the adductor muscle. From the various teleost jaws figured by Nelson (1973) it seems that there is much variability in the size of the coronomeckelian; in some plesiomorphic groups (e.g. gonorynchids, esocoids, amiids) the bone appears insig- nificant, whilst in others (hiodontids, albulids, argentinoids) it is extensive. All that can be said is that in aspinines and the other taxa considered above, the coronomeckelian is of a particularly unusual (and possibly derived) shape which may represent a synapomorphy. The relationships of Genghis plus the aspinines must now be considered. Howes (1978) thought Oreoleuciscus (Fig. 6C) the most likely candidate as the sister group of the aspinines. It is now clear that this is not the case since Oreoleuciscus possesses none of those characters uniquely shared by Genghis and the aspinine group. Nonetheless, Oreoleuciscus has a close affinity with these taxa as it shares with them and with Tribolodon both an elongate ptero- sphenoid bearing a lateral shelf from which originates part of the adductor arcus palatini musculature, and crenate gill-rakers. The configuration of the infraorbitals in Oreoleuciscus ASPININE CYPRINID FISHES 293 Fig. 7 Lower jaws, medial views to show coronomeckelian bone (cm) of, A, Genghis mongolicus; B, Aspius vorax; C, Luciobrama macrocephalus; D, Oreoleuciscus humilis; E, Pogonichthys macroleptidotus; F, Tribolodon brandti. Scales = 5 mm. resembles that of the aspinines and Genghis in that the 3rd infraorbital extends, almost hori- zontally, well past the posterior border of the orbit with a consequent separation of the 4th infraorbital from the orbital margin. However, the 4th infraorbital is reduced to its canal tube and is not diagonally aligned, as in the aspinines and Genghis (cf. Figs 3A, C & D with fig. 22 in Howes, 1978). A simple sister-group relationship between Oreoleuciscus and Genghis plus the aspinines is disrupted by the monotypic genus Aspiopsis (Fig. 6B). Only a single syntype is available for examination and only those characters visible without dissection and from radiographs can be ascertained. Howes (1978 : 60) followed Berg (1949) in considering Aspiopsis to be synonymous with Leuciscus, but a reappraisal of Aspiopsis makes it clear that it is a distinct genus and must be included amongst the assemblage of genera discussed here. Aspiopsis is characterized by a rather elongate body, an operculum with attenuated posterior border, small, imbricate scales (70 in the lateral line), numerous gill-rakers (27 on the 1st ceratobranchial) with a crenate medial membrane, and a papillate lateral buccal membrane, particularly over the preopercular area adjacent to the gill-arch. Radiographs reveal an elongate cranium. In the shape and configuration of its infraorbitals, Aspiopsis closely resembles Genghis and the aspinine genera (Figs 3C). The 1st infraorbital, however, bears a V-shaped depression on its dorsal margin, a feature encountered elsewhere only in Tribolodon. In these characters, apart from the latter, Aspiopsis most clearly resembles Oreoleuciscus. Lack of dissectable material precludes investigating the site of origin of the adductor arcus palatini muscle in Aspiopsis. Assuming that this muscle does originate from the pterosphenoid, then Aspiopsis would be considered as the sister-lineage to Oreoleuciscus. On the basis of their cranial elongation Aspiopsis and Oreoleuciscus appear most closely related to Genghis and the aspinines; however, the derived 1 st infraorbital morphology which 294 G. J. HOWES Fig. 8 Subtemporal fossae of A, Tribolodon brandti; B, Ptychocheilus oregonensis; C, Pogo- nichthys macrolepidotus; D, Aspius vorax, dashed lines indicate extent of the anterior chamber. Aspiopsis shares with Tribolodon places the Aspiopsis-Oreoleuciscus lineage in an unre- solved trichotomy, with Genghis and the aspinines on the one hand, and Tribolodon on the other. The trichotomy may be resolved when the anatomy of Apsiopsis is better known. Tribolodon, as well as sharing a derived infraorbital feature with Aspiopsis also shares with the aspinines, Genghis, Oreoleuciscus and the Nearctic genus Pogonichthys, a pterosphenoid ASPININE CYPRINID FISHES 295 sop Fig. 9 Suboperculum, medial views showing anterior process (sop) of A, Tribolodon brandti; B, Pogonichthys macrolepidotus (210mm SL); C, Ptychocheilus grandis (UMZM 181 929-5, 315mmSL). origin of the adductor arcus palatini muscle, crenate margined gill-rakers, and a triangular coronomeckelian bone. In addition, Pogonichthys, shares with Tribolodon a derived form of subtemporal fossa, suboperculum and caudal skeleton as follows: The subtemporal fossa in Tribolodon and Pogonichthys has an anterior extension into the autopterotic and sphenotic (Fig. 8). The extension is in the form of a narrow, finger-shaped chamber filled with a plug of fat. A sphenotic contribution to the subtemporal fossa was recognized by Howes (1982) as a synapomorphic condition for an assemblage of barbelled carps, named the squaliobarbine group. In these taxa however, the subtemporal fossa has a different shape in that the fossa is more extensive, with the prootic and exoccipital con- tributing substantially to its roof. Furthermore, in the squaliobarbines, part of the levator posterior muscle originates from the sphenotic chamber. In Tribolodon and Pogonichthys the levator posterior takes its origin dorsally from the pterotic and epioccipital only, and posteriorly from the exoccipital — as in the case ofCyprinus shown by Eastman, 1971 — , the sphenotic is not involved. This particular type of fat-filled sphenotic chamber in Tribolodon and Pogonichthys has not been discovered in any other cyprinid examined. In Ptychocheilus, there is a lateral cavity of the subtemporal fossa in the pterotic and this too contains a fatty plug (Fig. 8). Amongst the aspinine genera the subtemporal fossa is small and trianguloid in outline (Fig. 8). The suboperculum in Tribolodon and Pogonichthys has a club-shaped antero-dorsal process (Figs 9A & B). Normally this part of the bone is rounded, or, if produced, it is in the form of a slender triangle. A similarly shaped subopercular process also occurs in Ptychocheilus (Fig. 9C). The caudal skeleton in Pogonichthys and Tribolodon exhibits hypertrophy of the preural neural spines. In Pogonichthys the neural spines of the 2nd-4th preural centra are thickened and antero-posteriorly lengthened and articulate distally with hypertrophied procurrent rays (Fig. 10A). In Tribolodon the 2nd and 3rd preural neural spines bear prominent anterior lamellae (Fig. 10B). Cyprinids are generally conservative in the morphology of the caudal skeleton and hypertrophy of the preural neural spines is rare. Often, however, there is a 296 G. J. HOWES npu2 pr ep Fig. 10 Caudal skeletons of A, Pogonichthys macrolepidotus; B, Tribolodon brandti; ep = epural, pr = procurrent ray, npu2 = neural spine of 2nd preural vertebra, un = uroneural. Scales = 5 mm. double neural spine on the 3rd preural centrum, and this feature regularly occurs among aspinine species. It is possible that the hypertrophied condition of the spines in Tribolodon and Pogonichthys is due to coalescence of the spines into a single unit. The Nearctic genera Ptychocheilus and Lavinia also display thickening of the preural neural spines and Lavinia has hypertrophied dorsal procurrent rays, a feature shared with Pogonichthys. There is further evidence to suggest Pogonichthys and Ptychocheilus have close phylogenetic links. ASPININE CYPRINID FISHES 297 Ptychocheilus shares with Tribolodon and Pogonichthys the club-shaped subopercular anterior process (see above & Fig 9). Ptychocheilus also has crenate and papillate gill-rakers. In Ptychocheilus oregonensis the anterior fibres of the adductor arcus palatini muscle origi- nate from the lower part of the pterosphenoid, but there is no prominent pterosphenoid shelf like that in Pogonichthys. The cladistic relationships of the Nearctic 'aspinines' Pogonichthys and Ptychocheilus have yet to be ascertained. Apparent synapomorphies linking Ptychocheilus with the Nearctic genera Mylopharodon and Gila (part) have been reported. Hopkirk (1973) pointed out the similarity of jaw and gill-raker structure with Mylopharodon and Illick (1956) drew attention to the looped canal on the 1st infraorbital in both Ptychocheilus and Gila robusta. According to Illick (1956) this canal configuration does not exist in other Nearctic taxa and I have not found such an erratic course of the canal in any Old- World cyprinid taxon. To summarize; Genghis represents the sister lineage of the aspinine genera, Aspius, Elopichthys, Pseudaspius, Aspiolucius and Luciobrama, which in turn from one part of a triad whose two other lineages are Aspiopsis + Oreoleuciscus and Tribolodon + Pogonichthys. It cannot as yet be determined which of these two lineages is the closest relative of Genghis and the aspinine group. This impasse is expressed as an unresolved trichotomy in the cladogram (Fig. 12). Now that the aspinine group sensu Howes (1978) are seen to form one part of a more extended monophyletic assemblage, it is necessary to broaden the concept of the aspinine group so as to embrace Genghis, Aspiopsis, Oreoleuciscus, Tribolodon and Pogonichthys. The wider relationships of the aspinines are presently unclear. However, in discussing the characters which distinguish ''Leuciscus'' mongolicus from other Leuciscus species it was noted that two species, L. lehmanni and L. schmidtii possess a gill-raker morphology similar to that of Genghis mongolicus. Furthermore, these two species have an extensive gap between Fig. 11 Ventral view of 1st gill-arch (operculum raised) to show extensive opening between it and the bucco-pharyngeal roof in A, Genghis mongolicus and B, Aspius aspius, and the res- tricted opening in C, Leuciscus leuciscus. 298 G. J. HOWES f 10 Fig. 12 Cladogram depicting the hypothesized relationships of the aspinine group. Synapo- morphies: (1) Crenate margined gill-rakers; AAP muscle originates from pterosphenoid shelf; elongate, triangular coronomeckelian bone (this character of dubious polarity). (2) Derived infraorbital configuration; elongate posterior cranial bones. (3) Further derived state of infraorbital morphology; well-formed occipital platform; elongate and laterally expanded ptero- sphenoid. (4) Vertebrae 51-55; extensive contact between pterosphenoid and parasphenoid; nasals elongate with 6-10 pores. (5) Elongation of occipital region and lower jaw; tunnel-like post-temporal fossa. (6) Extreme divergence of 4th and 5th infraorbital canals; elongation of ento- and metapterygoid; complex development of LAP muscle. (7) Extensive aortic foramen in basioccipital process; 13-16 supraneurals. (8) Papillate lateral buccal membrane; attenuated operculum. (9) Concavity of 1st infraorbital (snared only with Tribolodori). (10) Club-shaped subopercular process; hypertrophy of preural neural spines. the branchial arch and pharyngo-buccal root', in contrast to the restricted space of other Leuciscus species (Table 1 & Fig. 11). It seems likely that 'Leuciscus' lehmanni and 'L. ' schmidtii (both from Central Asia) may represent the sister group to the aspinines, and that the whole assemblage is the sister group to an, as yet, unidentified monophyletic unit within the all-embracing 'Leuciscus'. These ideas can only be tested by a revision of 'Leuciscus' (see comments below in Conclusion section). ASPININE CYPRINID FISHES 299 .—*--:> — -• Fig. 13 Distribution of the aspinine group in Asia (hatched, below) and western North America (solid black, above). The numbered zones on the Asian map refer to the disposition of accreted continental plates as proposed by McElhinny et at. ( 1 98 1 ) and Leith ( 1 982). Double-dashed lines indicate the boundary of the Siberian craton. Plates; 1 = Iranian, 2= Afghanistan, 3 = Kazakh- stanian, 4 = Tarim, 5 = Qinhai-Tibet, 6 = Si no- Korean, 7 = Yangtze, 8 = SE Asian, 9 = Sikhote Alin, 10 = Kolyma, 1 1 = Kamchatka. Map of Asia drawn on Zenithal equal-area projection. Biogeography of the aspinines Distribution within Asia The most significant feature of aspinine distribution within Asia is its east-west dichotomy (Fig. 13). Aspius lies to the west, covering much of Europe and extending south to the Tigris. Aspiolucius occurs sympatrically with Aspius in the Amu Darya (see Coad, 1981). The majority of aspinine genera are distributed east of the Mongolian plateau; Elopichthys, the 300 G. J. HOWES Fig. 14 Area cladogram of the aspinine group. sister genus of Aspius; Pseudaspius and Luciobrama, the relatives of Aspiolucius, all lie in the Sino-Korean region (including the Amur; the Siberian and China subregions of Mori, 1936). Luciobrama extends south to Hanoi. The plesiomorphic aspinines, Genghis and Tri- bolodon also occur in eastern Asia. The former in northern China and the latter along the coastal margin of the Yellow Sea and the Sea of Japan and the Pacific coasts of Japan and Sakhalin. Only two genera, Aspiopsis and Oreoleuciscus occur in Central Asia, being con- fined, respectively to a small area bordering Sinkiang and the landlocked basins of the Upper Ob and Bya in Mongolia. An area cladogram of the Asian aspinines shows a repeated dichotomy in the two lineages between western and eastern Asia (Fig. 14). The lineage of central Asian genera forms part of an unresolved trichotomy with the East Asian and Japanese-American branches and so is uninformative as to its area relationships. Trans- Pacific links The phyletic relationships established here between Tribolodon and Pogonichthys supports the hypotheses of Miller (1959; 1965), Hopkirk (1973), Gosline (1974) and Howes (1980) that a close relationship exists between some western North American and Japanese and Chinese cyprinid taxa. This Pacific link is the only one so far known for members of the Cyprinidae, although a well-known relationship exists elsewhere within cyprinoids, namely that between Chinese and American catostomids (see Patterson, 1981). The area cladograms presented by Patterson (198 1) for various Nearctic and Palearctic freshwater teleosts, suggest closer links between western-North America and eastern America and Europe than with Asia. ASPININE CYPRINID FISHES 301 Links between eastern Asia and western-North America are, however, forthcoming from cladistic relationships established amongst various insect groups. Ross (1974) demonstrates a Pacific link for caddisflies and Edmunds (1981) in discussing the distribution of mayflies points to the relationships between Eurasian and Nearctic genera as displaying strong Pacific vicariant patterns. Tuxen's (1977) analysis of proturans points out an 'unexplained' geographic relationship between Japanese and a Nearctic species of Baculentulus. Explanations for an eastern Asia-western-North American faunal association may be attri- buted to dispersal or vicariance. At present too few phylogenetic data are available to dis- criminate between these alternative explanations. It has generally been accepted that the Bering land connection has been the principal route for faunal dispersal from late Cretaceous (see Cox, 1974 : 86 concerning 'Asiamerica'). More recent notions have proposed that several continental plates (or terranes) have occupied what is now the Pacific Ocean and that these elements are now accreted to the margins of the Asiatic and American cratons (see dis- cussions in Nur & Ben-Avraham, 1981 and Jones et al, 1982). Thus, Asia and North America are hybrid continents and from a vicariant point of view areas of related biotic endemism within those continents should mark former plates and their associations. The numbers of plates and their former dispositions, and whether or not there were supercon- tinents Gondwanaland and Pacifica are hotly disputed subjects amongst geologists (see McElhinny et al. 1981; Batten & Schweichert, 1981; Leith, 1982; Audley-Charles, 1983; Kerr, 1983). The distributional pattern of the aspinines within Asia and between Asia and western- North America provides general support for a vicariant explanation involving continental plate displacements. However, only more congruent cladograms of east Asiatic and western- North American biota will favour such an explanation. Conclusion Further progress with understanding the relationships of leuciscine and aspinine cyprinid fishes depends on: (1) A revision of Leuciscus. Such will not be an easy task since apart from the strictly practi- cal problem concerned with lack of adequate samples of Russian and eastern Asian species in Western European museums, the would-be reviser faces the taxonomic problem of dealing with what is seemingly a plesiomorphic assemblage of species. (2) Conduct a more wide-ranging cladistic analysis of Palearctic and Nearctic non-barbelled cyprinids. (3) Consolidation of the trans-Pacific link hypothesis through a more wide-ranging vicariance analysis of other biotas. Acknowledgements My sincere thanks are due to Drs Keith Banister, Humphry Greenwood, Gordon McGregor Reid and Colin Patterson for their discussions and criticisms of this paper. I am most grateful to Dr T. Abe for supplying literature references and information on Korean collec- tions and to Dr S. Kullander for the loan of type material from the Swedish Natural History Museum. My warm thanks go to Drs Reeve Bailey and Gareth Nelson for allowing me to work on and borrow material from the collections in their care at, respectively, the University of Michigan Museum of Zoology and the American Museum of Natural History. These thanks are extended to the several staff and students of those institutions for their help and generous hospitality. References Audley-Charles, M. G. 1983. Reconstruction of eastern Gondwanaland Nature 306: 48-50. Banarescu, P. 1970. On the systematics and synonymy of the Hwang-Ho dace, Leuciscus mongolicus (Kessler) (Pisces, Cyprinidae). Annali Museo Civico di Storia Naturale di Genova 78: 47-52. 302 G. J. HOWES Batten, R. L. & Schwichert, R. A. 1981. Discussion of A. Nur and Z. Ben-Avraham's paper. In: Vicar- iance Biogeography: a critique (G. Nelson & D. E. Rosen, Eds), 359-366. Berg, L. S. 1912. Faune de la Russie. Pisces 3. St. Petersburg. 846 pp. 1949, Freshwater fishes of the US.S.R. and adjacent countries. 2 (English translation, 1964), 496 pp. Coad, B. W. 1981. Fishes of Afghanistan, an annotated check-list. National Museum of Canada Publictions in Zoology 14: i-v, 1-26. Cox, C. B. 1973. Vertebrate palaeodistributional patterns and continental drift. Journal of Biogeogra- phy, 1 (2): 75-94. Eastman, J. T. 1971. Pharyngeal bone muscles of the carp, Cyprinus carpio. Journal of Morphology, 134(2): 131-140. Edmunds, G. F. 1981. Discussion of R. Melville's paper. In: Vicariance Biogeography: a critique (G. Nelson & D. E. Rosen, Eds), 285-297. Fowler, H. W. 1899. Notes on a small collection of Chinese fishes. Proceedings of the Academy of Natural Sciences, Philadelphia, 179-182. Gosline, W. A. 1974. Certain lateral-line canals of the head in cyprinid fishes, with particular reference to the derivation of North American forms. Japanese Journal oflchthylogy 21 (1), 9-15. Hopkirk, J. D. 1973. Endemism in fishes of the Clear Lake region of Central California. University of California Publications in Zoology 96, 1-137. Howes, G. J. 1978. The anatomy and relationships of the cyprinid fish Luciobrama macrocephalus (Lacepede). Bulletin of the British Museum (Natural History) Zoology, 34 (1), 1-64. 1980. The anatomy, phytogeny and classification of the bariliine cyprinid fishes. Bulletin of the British Museum (Natural History) Zoology 37 (3), 129-198. 1981. Anatomy and phylogeny of the Chinese Major Carps Ctenopharyngodon Steind., 1866 and Hypophythalmichthys Blkr, 1860. Bulletin of the British Museum (Natural History), Zoology 41 (1), 1-52. 1982. Anatomy and evolution of the jaws in semiplotine carps with a review of the genus Cypri- nion Heckel, 1843 (Teleostei: Cyprinidae). Bulletin of the British Museum (Natural History) Zoology 42 (4): 299-335. Illick, H. J. 1956. A comparative study of the cephalic lateral-line system of North American Cyprini- dae. American Midland Naturalist 56 (1), 204-223. Jones, D. L., Cox, A., Coney, P. & Beck, M. 1982. The growth of Western North America. Scientific American 247 (5), 50-64. Kerr, R. A. 1983. Suspect terranes and continental growth. Science 222, 36-38. Kessler, K. T. 1876. Descriptions of the fishes collected by Col. Prejevalsky in Mongolia. In: Preje- valsky, N. Mongolia i Strana Tangutov 2 (4), 1-36. St. Petersburg. Leith, W. 1982. Rock assemblages in Central Asia and the evolution of the southern Asian margin. Tectonics I (3), 303-318. McElhinny, M. W., Embleston, B. J. J., Ma, X. H. & Zhang, Z. K. 1981. Fragmentation of Asia in the Permian. Nature, 293: 212-216. Miller, R. R. 1959. Origin and affinities of the freshwater fauna of Western North America. In: Zoo- geography (C. L Hubbs, Ed.), American Association for the Advancement of Science Publication No. 51, 187-222. 1965. Quaternary freshwater fishes of North America. In: The Quaternary of the United States (H. E. Wright and D. G. Frey, Eds). Princeton, 569-581. Mori, T. 1934, The fresh water fishes of Jehol. Report of the first scientific expedition to Manchoukou. Section V. Part 1 , 6 1 pp. 1936. Studies on the geographical distribution of freshwater fishes in eastern Asia, 86 pp. 1 94 1 . A new species and a new subspecies of Cyprinidae from North China. Zoological Magazine, Tokyo 53(3), 183. Nelson, G. 1973. Relationships of clupeomorphs, with remarks on the structure of the lower jaw in fishes. In: Interrelationships of fishes (P. H. Greenwood, R. S. Miles & C. Patterson, Eds), London, 333-349. Nur, A. and Ben-Avraham, Z. 1981. Lost Pacifica continent: A mobilistic speculation. In: Vicariance Biogeography; a critique (G. Nelson & D. E. Rosen, Eds), 341-358. Oshima, M. 1926. Notes on a small collection of freshwater fishes from east Mongolia. Zoological Magazine, Tokyo 38 (450-451), 99-104. Patterson, C. 1981. The development of the North American fish fauna — a problem of historical bio- geography. In: Chance, Change and Challenge: The evolving biosphere (P. H. Greenwood & P. L. Forey, Eds), 265-281. ASPININE CYPRINID FISHES 303 Prejevalsky, N. 1876. Mongolia i Strana Tangutov 1 (in English translation as, Mongolia, the Tangut country, and the solitudes of northern Tibet. Translated by E. D. Morgan, London 1876, 287 pp; p. 197, fishes). Rendahl, H. 1925. En nyid (Leuciscus [Idus] waleckii sinensis n. subsp.) fran Kina. Fauna och Flora (5): 193-197. Ross, H. H. 1974. Biological Systematics. 345 pp. Tuxen, S. L. 1977. The genus Berberentulus (Insecta, Protura) with a key and phylogenetical consider- ations. Revue d'Ecologie et de Biologic du Sol, 14 (4), 597-61 1. Zander, E. 1903. Studien iiberdas Kiemenfilterbei Susswasserfischen. Zeitschrift fur Wissenschaftliche Zoologie. Leipzig, 75, 233-257. Manuscript accepted for publication 7 February 1984 New bats (Mammalia: Chiroptera) and new records of bats from Borneo and Malaya J. E. Hill Department of Zoology, British Museum (Natural History), Cromwell Road, London SW7 5BD, United Kingdom C. M. Francis Wildlife Branch, Forest Department, P.O. Box 311, Sandakan, Sabah, East Malaysia Introduction Although numerous bat species are known from Sabah, many records (Medway, 1977) origi- nate in collections made from Mount Kinabalu or from the montane region of which it forms a part. Bats from the lowland areas of Sabah have received less attention and moreover many records are of cavernicolous species. One of us (CMF) has since 1981 been working as a Canadian volunteer (CUSO) for the Wildlife Branch of the Sabah Forest Department. While the principal activity in Sabah has been the management of edible-nest swiftlets (Collocallid) and the study of the rain forest avifauna, it has also been possible to capture and study bats, of which a proportion has been retained as museum specimens for taxonomic and record purposes. Some 65 species have been obtained, including three hitherto undescribed, some, including a new subspecies, representing new distributional records for Borneo, and others of taxonomic or more local distributional interest. This account is limited to the novelties that have been obtained, to those specimens that constitute new records for Borneo, and those that represent taxa so far poorly known. A single new record for Malaya has also been included. The majority of the bats were caught using harp-style traps (Tuttle, 1974; Tidemann & Woodside, 1978). Usually these were placed across narrow trails or streams in the rain forest, or near cave openings. Some bats were also caught in standing mist nets, others by hand or with a butterfly net at their roosts in caves. All trapping has been done at or near ground level, with no work being attempted in the canopy. Most of the collecting has been carried out either at Gomantong or Sepilok. The first of these, several miles north of the Kinabatangan River, about 20 miles south of Sandakan, has a network of limestone caves surrounded by both primary and secondary forest. Large colonies of several species of bats roost in the caves, while many other species are found in the vicinity. Bats from the Gomantong Caves were collected in 1929 by F. N. Chasen, and in 1930 by Senior Forest Ranger P. Orolfo (Chasen, 1931). More recently, Gomantong has been visited by the members of a Japanese group, in 1976 and 1979, the bats collected being reported by Kobayashi et al (1980). These workers also obtained bats at the Madai Caves, whence some of those reported here originate. Sepilok is a virgin jungle reserve located near Sandakan and maintained by the Sabah Forest Department as a research area and wildlife sanctuary. Limited collecting has also been carried out at other sites which are summarized in Table 1 . Many of the more interesting specimens have been donated to the British Museum (Natural History), as have the holotypes of the new taxa described in this paper. All are denoted by their accession numbers, prefixed BM(NH). The final disposition of the remain- der of the specimens examined in London has yet to be decided: all are indicated in this account by their original collector's numbers, prefixed CMF. Duplicates of some of these, together with those specimens that were not sent or brought to London since they represent Bull. Br. Mus. nat. Hist. (Zool.) 47(5): 305-329 Issued 25 October 1984 306 J. E. HILL & C. M. FRANCIS Table 1 Major localities whence bats have been obtained Gomantong 5 °31 'N, 118 °04 'E. 30m Limestone caves, in pri- mary dipterocarp forest Baturong 4 °42 'N, 118 °00 'E. 80m Limestone caves, in pri- mary dipterocarp forest Madai 4 °43 'N, 118 °09 'E. 80m Limestone caves, in logged dipterocarp forest Panggi 5 °31 'N, 118 °18 'E. 150m Limestone caves, in logged dipterocarp forest Segarong 4 °34 'N, 118 °25 'E. 30m Limestone caves near the sea, in secondary forest and mangrove Sepilok 5 °52 'N, 117 °56 'E. 30m Primary dipterocarp forest Lumerau 5 °12 'N, 118 °52 'E. 40m Primary dipterocarp forest Sungei Labau, Witti Range 5 °16 'N, 116 °30 'E. 300m Primary dipterocarp forest Silabukan 5 °11 'N, 118 °47 'E. 300m Primary dipterocarp forest Rinangisan, Crocker Range c. 5 °40 'N, 116 °20 'E. 110m Montane forest Kinabalu National Park 6 °02 'N, 116 °32 'E. 1500m Montane forest Menggadong 4 °58 'N, 115 °28 'E. 30m Kerangas Forest species already well known in Borneo or of common occurrence will be maintained as a reference collection at Sepilok by the Sabah Forest Department. All measurements are in millimetres: those of individual teeth have been made with a stereoscopic microscope and traversing micrometer stage, others with a dial micrometer. Systematic section Rousettus spinalatus Bergmans & Hill, 1980 Rousettus spinalatus Bergmans & Hill, 1980 : 95. In or near Medan, or in or near Prapat, N Sumatra. SPECIMEN EXAMINED, c? CMF 0100 Panggi, Sabah (in alcohol, skull extracted). REMARKS. This is the second example of Rousettus spinalatus to be recorded from Borneo, the first being a subadult female obtained at Niah Great Cave in Sarawak and reported by Bergmans & Hill in the original account. As in the original specimens, the wings in this example from Sabah are inserted on the spinal line and there is no longitudinal dorsal band of fur down the centre of the back. Dentally this specimen agrees closely with the example from Niah, although in the Sabah specimen, as in that from Niah, m1 is slightly wider than the corresponding tooth in R. amplexicaudatus, not narrower as it is said to be in the original description. As in the Niah example, nV and m3 are longer, wider and more massive than in R. ample xicaudatus, and m2 is approximated more closely to the rear edge of the palate at the zygomatic insertion than in that species. A specimen (9 CMF 0109) of R. ample xicaudatus was collected at Panggi on the day following that on which this example of/?, spinalatus was obtained. Measurements: length of forearm 86-9; thumb (c.u.) 25-3; IIm 33-4; II1 7-1; II2 5-4; IIIm 50-6; III1 31-7; IIP 41-4; IVm 49-2; IV 24-0; IV2 28-4; vm 47-7; V1 22-4; V2 26-4; length of ear 17-6; length of tibia 33-3; length of foot (c.u.) 17-9; greatest length of skull 34-9; condylobasal length 32-8; condylo- canine length 31-7; rostral length (front of orbit to prosthion) 12-0; rostral length (front of orbit to tip of nasals) 11-4; palatal length 17-5; palatilar length 16-8; length palation — incisive foramina 16-0; length palation — basion 13-0; lachrymal width 10-0; least interorbital width 8-1; least postorbital width 7-8; zygomatic width 22-2; width of braincase 14-8; mastoid width 13-8; orbital diameter 8-6; c'-c1 (crowns) 6-7, (alveoli) 6-4; c'-c1 (internally, cingula) 3-8; pm4-pm4 (internally) 5-1; m2-!!!2 (crowns) CHIROPTERA 307 10-5; (alveoli) 10-4; width of mesopterygoid fossa 4-5; c-m2 (crowns) 12-9, (alveoli) 12-7; length of com- plete mandible from condyles 25-6; length right ramus from condyle 27-1; mandibular height (angular process to tip of coronoid process) 1 1-5; c-m3 (crowns) 14-3, (alveoli) 14-2. Length/width of cheekteeth: pm3 2-19/1-31; pm4 2-33/1-89; m1 2-82/2-07; m2 2-04/1-54; pm, 1-19/1-28; pm3 2-16/1-25; pm4 2-38/1-62; m, 2-51/1-56; m2 2-21/1-61; m3 1-63/1-20. Chironax melanocephalus (Temminck, 1825) Pteropus melanocephalus Temminck, 1825 : 190, pis 12, 16, figs 3, 4. Bantam, Java. SPECIMEN EXAMINED. 9 CMF 830105.1 Sepilok, Sabah (in alcohol, skull extracted). REMARKS. Chironax melanocephalus has not before been reported from Borneo although known from southern Thailand, Malaya, Sumatra, Nias Island, Java and Sulawesi. The species was reviewed in some detail by Hill (1983) who reported further specimens from Sumatra, Java and Sulawesi and provided comparative notes. This specimen agrees with those from the mainland, from Sumatra and from Java in the presence of a small antero-external supplementary cusp on the second upper premolar (pm3) rather than with Sulawesian examples from which this cusp is absent. In most points of wing structure it is similar to specimens from the Malay Peninsula, Sumatra and Sulawesi, conforming to these in the relative lengths of the third, fourth and fifth metacarpals rather than with Javan examples in which these digital components are relatively slightly shorter, the third metacarpal especially so. The second phalanges of the fourth and fifth digits, how- ever, are relatively a little shorter than those in most specimens from the mainland and from Sumatra, in this respect approaching or agreeing with specimens from Java and Sulawesi. This female specimen has diffuse pale orange neck tufts (from specimen in alcohol): other- wise dorsally the pale based pelage is tipped with dark brown except on the head where the pelage is blackish, while ventrally it is pale greyish brown, becoming buffy on the throat and flanks. Measurements (wing indices in parentheses): length of forearm 45-6 (1000); IIIm 31-6 (693); III1 22-4 (491); III2 28-4 (622); IVm 30-5 (669); IV1 16-9 (371); IV2 16-7 (366); Vm 30-2 (662); V1 14-8 (325); V2 15-5 (340); greatest length of skull 22-7; condylobasal length 22-0; condylocanine length 2 1 -0; length front of orbit-tip of nasals 5-4; length orbit-nares 5-0; length orbit-gnathion 6-9; palatal length 1 1-6; length palation-incisive foramina 9-8; length palation-basion 8-6; lachrymal width 6- 1 ; least interorbital width 4-4: least postorbital width 5-1; zygomatic width ; width of braincase 9-8; mastoid width 10-6; orbital diameter 5-8; c'-c1 (crowns) 4-5, (alveoli) 4-2; (cingula, internally) 2-1; pm4-pm4 (crowns) 6-7, (alveoli) 6-2, (internally) 4-0; m'-m1 (crowns) 6-4. (alveoli) 6-0; width mesopterygoid fossa 3-1; c-m1 (crowns) 7-3; length complete mandible from condyles 15-7; length right ramus from condyle 16-6; coronoid height 8-5; c-m2 (crowns) 8-2. Hipposideros bicolor bicolor (Temminck, 1834) Rhinolophus bicolor Temminck, 1834: 19, pi. 1, fig. 3; 1835: 18 (further description). Anjer coast, northwestern Java (Tate, 1941). SPECIMENS EXAMINED. IY-> "3- oo — • oo ,1, r^- in TJ- -<3- '^ op — ON op op — ^ ^ §* J3«C"Sl«-S^JrJ22233I s§ ^^ & •3 i Cy s ^ "3 o J « '** | "sPilsS ti: c « 8 ^ | i oo r-4 7~T"T"TTv'?irlr?riirT1? I I oo — ^r o •o c |.S 3) £ S3 2 1 O 73 ci: c ^ 8 •C "O 2'* OJ o o JS -C "oo "oo c c o c ^ _0> *J — C C <£ C O Cy o3 ">>"> c is •« o ^ p ^ 2 ."2 '•= o • — o 2 j= •- E j= o o c jS O 03 U G- ^ E — CS D. ^ O ""> O 'u, ?f -i i 326 J. E. HILL & C. M. FRANCIS (in alcohol, skull extracted); d BM(NH) 83.358 (in alcohol, skull extracted), 9 CMF 830206.2 (in alcohol) Lumerau; all Sabah. REMARKS. This species has been known hitherto from southern Thailand, Malaya and Sumatra. Specimens from Borneo agree in most respects with those from Thailand and Malaya but are generally a little smaller. Additionally, the braincase in Bornean examples is slightly more elevated frontally than in those from the mainland, with the frontal profile a little more deeply concave; the palate is slightly narrower, especially posteriorly; the outer upper incisor (i3) is considerably reduced to a narrow spicule; and the maxillary teeth are rather smaller, particularly the molars. Measurements of CMF 821106.06, BM(NH) 83.77, 83.358 (including skulls) and CMF 830206.2 (forearm only) (in that order), with those (in parentheses) of three (except where indicated) specimens from the Malayan peninsular: length of forearm 31-6, 32-4, 32-3, 32-5, (32-7-34-2); greatest length of skull 14-3, 14-2, 14-4, (14-5-15-2); condylobasal length 13-2, 12-9, 13-1, (13-1-13-8); condylocanine length 13-0, 12-8, 13-0, (13-0-13-7); palatal length 7-6, 7-7, 7-9, (7-9-8-5); least interorbital width 3-6, 3-6, 3-7, (3-8^-2); zygomatic width 8-2, — , 8-3, (8-5, 8-7, two only); width of braincase 6-8, 6-6, 6-9, (7-0-74); mastoid width 7-2, 7-0, 7-2, (7-3-7-5); c'-c1 (alveoli) 3-3, 34, 34, (3-5-3-7); m3-m3 5-0, 5-1, 5-0 (5-3-5-5); c-m3 5-7, 5-6, 5-7, ((5-9-6-1); length complete mandible from condyles 10-1, — , 10-0 (10-5, 10-6, two only); length right ramus from condyle 10-4, — , 104, (10-5-11-0); c-m3 6-1, 6- 1,6-1, (6-3-6-6). DISCUSSION. Phoniscus atrox appears to have been represented in the literature until now by no more than six specimens: the holotype and one other from the type locality in Sumatra, two from Klong Bang Lai, Patiyu, southern Thailand reported in detail by Kloss (1916) and subsequently examined again by Hill (1965), by a subadult from the Ulu Gombak Forest Reserve in Selangor, Malaya, reported by Medway (1969) and finally by one from the Tekam Forest Reserve, Pahang, Malaya, recorded by Hill (\914a) and by Medway (1978). Those reported by Kloss (1916) from southern Thailand and by Hill (1974a) and Medway (1978) from Pahang are now in the British Museum (Natural History) (BM(NH) 20.7.3.7-8, 71.1135). A direct comparison of specimens from the Malayan peninsular and examples from Sumatra has yet to be made but following Kloss (1916) peninsular specimens have been referred to P. atrox with which apparently they agree in all essential respects. Unfortunately, Miller gave no details of the skull or its dimensions in the original account. It seems likely that adequate material might demonstrate that the Bornean and mainland populations are subspecifically distinct but both need to be compared with specimens from Sumatra before any proper decision can be reached. The only other record of Phoniscus from Borneo is of a specimen of P. jagorii javanus (Thomas, 1880) from Riam, Kotawaringin district, central Kalimantan, listed by Tate ( 1 94 1 : 597) who remarks elsewhere (p. 589) in the same paper that there were two specimens from Borneo in the Archbold Collections. However, H. M. van Deusen (in Medway, 1977) reported that only one of these is from Borneo, as listed by Tate: the other is from Bali, as Tate's list shows. Acknowledgements We especially wish to thank the Canadian volunteer organization CUSO and the Game Branch of the Sabah Forest Department for supporting one of us (CMF) during his field work in Sabah, the Sabah Forest Department particularly for permitting the collecting of bats. Mr Patrick Andau, the Assistant Chief Game Warden, and his rangers provided continuous encouragement and valuable support in the field, as well as supplying two bat traps and mist nets. The Sabah National Parks Department kindly allowed the collection of bats in the Kinabalu National Park. We are indebted also to Mr Frank Rozendaal and Dr C. J. Smeenk of the Rijksmuseum van Natuurlijke Historic, Leiden, who have generously placed at our disposal specimens ofMurina obtained by Mr Rozendaal while visiting Sabah. CHIROPTERA 327 References Allen, G. M. 1913. A new bat from Tonkin. Proceedings of the Biological Society of Washington 26: 213-214. & Coolidge, H. 1940. Mammal and bird collections of the Asiatic Primate Expedition. Mammals. Bulletin of the Museum of Comparative Zoology, Harvard College, Cambridge, Massachusetts 87: 131-166. Andersen, K. 1918. Diagnoses of new bats of the families Rhinolophidae and Megadermatidae. Annals and Magazine of Natural History; including Zoology, Botany and Geology, London (9) 2: 374-384. Bergmans, W. & Hill, J. E. 1980. On a new species of Rousettus Gray, 1821, from Sumatra and Borneo. Bulletin of the British Museum (Natural History), Zoology, London 38: 95-104, 5 figs, 1 tab. Blyth, E. 1853. Report of Curator, Zoological Department. Journal of the Asiatic Society of Bengal, Calcutta 22: 408-4 17. Chasen, F. N. 193 1 . On bats from the limestone caves of North Borneo. Bulletin of the Raffles Museum, Singapore No. 5: 107-114. 1940. A Handlist of Malaysian mammals. Bulletin of the Raffles Museum, Singapore No. 15: i-xx, 1-209, map. Corbet, G. B. & Hill, J. E. 1980. A world list of mammalian species. British Museum (Natural History)/ Comstock Publishing Associates (Cornell University Press), London/Ithaca. Davis, D. D. 1962. Mammals of the lowland rain-forest of North Borneo. Bulletin of the Raffles Museum, Singapore No. 31: 1-129, 20 figs, 23 pis. Dobson, G. E. 1874. Description of new species of Chiroptera from India and Yunan. Journal of the Aisatic Society of Bengal, Calcutta 43: 237-238. 1877. Notes on a collection of Chiroptera from India and Burma, with descriptions of new species. Journal of the Asiatic Society of Bengal, Calcutta 46: 310-3 13. I Herman, J. R. & Morrison- Scott, T. C. S. 1951. Checklist of Palaearctic and Indian mammals 1758-1946. British Museum (Natural History), London. Gray, J. E. 1846. Catalogue of the specimens and drawings of Mammalia and birds of Nepal and Tibet, presented by B. H. Hodgson . . . to the . . . Museum. British Museum, London. Heller, K-G. & Volleth, M. (1984). Taxonomic position of "Pipistrellus societatis' Hill, 1972 and the karyological characteristics of the genus Eptesicus (Chiroptera, Vespertilionidae). Zeitschrift fur Zoologische Systematik und Evolutionsforschung, Frankfurt am Main 22, (1): 65-17, 5 figs, 1 tab. Hill, J. E. 1962. Notes on some insectivores and bats from Upper Burma. Proceedings of the Zoological Society of London 139: 1 19-137, 5 tabs. 1963. A revision of the genus Hipposideros. Bulletin of the British Museum (Natural History), Zoology, London 11: 1-129, 41 figs, 2 tabs. 1964. Notes on some tube-nosed bats, genus Murina, southeastern Asia, with descriptions of a new species and a new subspecies. Federation Museums Journal, Perak (1963), 8: 48-59, 4 pis. 1965. Asiatic bats of the genera Kerivoula and Phoniscus (Vespertilionidae), with a note on Keri- voula aerosa Tomes. Mammalia, Paris 29: 524-556. 1969. The generic status of Glischropus rosseti Oey, 1951 (Chiroptera: Vespertilionidae). Mam- malia, Paris 33: 133-139. 1972. The Gunong Benom Expedition 1967. 4. New records of Malayan bats, with taxonomic notes and the description of a new Pipistrellus. Bulletin of the British Museum (Natural History), Zoology, London 23: 21-42, 3 tabs. 1974a. New records of bats from southeastern Asia, with taxonomic notes. Bulletin of the British Museum (Natural History), Zoology, London 27: 127-138, 2 tabs. 19746. A new family, genus and species of bat (Mammalia: Chiroptera) from Thailand. Bulletin of the British Museum (Natural History), Zoology, London 27: 301-336, 8 figs, 4 tabs. 1976. Bats referred to Hesperoptenus Peters, 1869 (Chiroptera: Vespertilionidae) with the descrip- tion of a new subgenus. Bulletin of the British Museum (Natural History), Zoology, London 30: 1-28, 5 figs, 4 pis. 1983. Bats (Mammalia: Chiroptera) from Indo- Australia. Bulletin of the British Museum (Natural History), Zoology, London 45: 103-208, 14 tabs. & Topal, G. 1973. The affinities of Pipistrellus ridleyi Thomas, 1898 and Glischropus rosseti Oey, 1951 (Chiroptera: Vespertilionidae). Bulletin of the British Museum (Natural History), Zoology, London 24: 447-456. Hodgson, B. H. 1847. On a new species of Plecotus. Journal of the Asiatic Society of Bengal, Calcutta 16: 894-896. 328 J. E. HILL & C. M. FRANCIS Honacki, J. H., Kinman, K. E. & Koeppl, J. W. 1982. Mammal species of the world. A taxonomic and geographical reference. Allen Press Inc. /Association of Systematics Collections, Lawrence, Kansas. Husson, A. M. 1955. Note on the cotypes of Vespertilio harpia Temminck, 1840 (Mammalia, Chiro- ptera, genus Harpiocephalus). Zoologische Mededelingen, Leiden 33: 121-125, 1 tab. Jentink, F. A. 1890. On a collection of mammals from Billiton. Notes from the Leyden Museum, Leiden 12: 149-154, 1 pi. 1897. Zoological results of the Dutch Scientific Expedition to Central Borneo. II. Mammals. Notes from the Leyden Museum, Leiden 19: 26-66, 2 pis. Kloss, C. B. 1916. On a collection of mammals from Siam. Journal of the Natural History Society of Siam, Bangkok 2: 1-36. Kobayashi, T., Maeda, K. & Harada, M. 1980. Studies on the small mammal fauna of Sabah, East Malaysia. I. Order Chiroptera and genus Tupaia (Primates). Contributions from the Biological Laboratory, Kyoto University, Kyoto 26: 67-82, 2 figs, 5 tabs, 4 pis. Koopman, K. F. 1983. A significant range extension of Philetor (Chiroptera, Vespertilionidae) with remarks on geographical variation. Journal of Mammalogy, Baltimore 64: 525-526. kuhl, H. 1819. Die deutschen Fledermause. Annalen der Wetterauischen Gesellschaji jiir die Gesammte Naturkunde, Frankfurt am Main 4: 1 1-49, 185-215, 2 pis. Lawrence, B. 1939. Collections from the Philippine Islands. Mammals. Bulletin of the Museum of Comparative Zoology, Harvard College, Cambridge, Massachusetts 86: 28-73. Lyon, M. W. 1916. Mammals collected by Dr W. L. Abbott on the chain of islands lying off the western coast of Sumatra, with descriptions of twenty-eight new species and subspecies. Proceedings of the United States National Museum, Washington 52: 437-462. Maeda, K, 1980. Review on the classification of little tube-nosed bats, Murina aurata, group. Mamma- lia, Paris 44: 53 1-551, 16 figs. Medway, Lord. 1965. Mammals of Borneo. Field keys and an annotated checklist. Journal of the Malaysian Branch of the Royal Asiatic Society, Singapore (1963) 36 (3): i-xiv, 1-193, 9 figs., 24 pis, 5 tabs, map. 1969. The wild mammals of Malaya and offshore islands including Singapore. Oxford University Press, Kuala Lumpur, Singapore, London. 1977. Mammals of Borneo. Field keys and an annotated checklist. Monographs of the Malaysian Branch of the Royal Asiatic Society, Singapore No. 7: i-xii, 1-172, 10 figs, 24 pis, frontis. 1978. The wild mammals of Malaya (Peninsular Malaysia) and Singapore. Oxford University Press, Kuala Lumpur, Oxford, New York, Melbourne. Miller, G. S. 1898. List of bats collected by Dr W. L. Abbott in Siam. Proceedings of the Academy of Natural Sciences of Philadelphia 316-325. 1905. A new genus of bats from Sumatra. Proceedings of the Biological Society of Washington 18: 229-230. 1911. Description of six new mammals from the Malay Archipelago. Proceedings of the Biological Society of Washington 24: 25-28. Peters, W. 1 86 1 . Berichtet iiber die von Hrn. F. Jagor bisher auf Malacca, Borneo, Java und den Philip- pinen gesammelten Saugethiere aus den Ordnungen der Halbaffen, Pelzflatterer und Flederthiere. Monatsberichte der Koniglichen Preussischen Academic der Wissenschaften zu Berlin 706-712. 1866. Uber einige neue oder weniger bekannte Flederthiere. Monatsberichte der Koniglichen Preussischen Akademie der Wissenschaften zu Berlin 16-25. 1 869. Mittheilung iiber die von dem Hon. Marquis Giacomo Doria in Sarawak auf Borneo gesam- melten Flederthiere. Monatsberichte der Koniglichen Preussischen Akademie der Wissenschaften zu Berlin ( 1868), 626-627. 1872. Mittheilung iiber neue Flederthiere (Phyllorhind micropus, Harpyiocephalus huttonii, Mur- ina griseus, Vesperugo (Marsipoloemus) albigularis, Vesperus propinquus, tenuipinnis). Monats- berichte der Koniglichen Preussischen Akademie der Wissenschaften zu Berlin 256-264. Phillips, C. J. 1967. Occurrence of the least horseshoe bat, Hipposideros cinceraceus, in Sabah (North Borneo). Journal ofMammology, Baltimore 48: 667-668, 1 fig. Robinson, H. C. & Kloss, C. B. 191 1. On new mammals from the Malay Peninsula and adjacent islands. Journal of the Federated Malay States Museums, Kuala Lumpur 4: 241-246. Sly, G. R. 1975. Second record of the bronze tube-nosed bat (Murina aenea) in peninsular Malaysia. Malayan Nature Journal, Kuala Lumpur 28 (3-4): 217. Tan, K. B. 1966. Stomach contents of some Borneo mammals. Sarawak Museum Journal 12: 373-385. CHIROPTERA 329 Tate, G. H. H. 194 la. Results of the Archbold Expeditions. No. 39. Review of Myotis of Eurasia. Bulletin of the American Museum of Natural History, New York 78: 537-565, 2 figs. 194 \b. Results of the Archbold Expeditions. No. 40. Notes on vespertilionid bats. Bulletin of the American Museum of Natural History, New York 78: 565-597, 4 figs. Taylor, E. L. 1934. Philippine land mammals. Monographs of the Bureau of Science, Manila No. 30: 1-548, 17 figs, 24 pis. Temminck, C. J. 1824-1841. Monographies de Mammalogie (Vol. I 1824-1827; Vol. II 1835-1841). Dufour & D'Ocagne, Paris. 1834. Over een geslacht der Vleugelhandige Zoogdieren, Bladneus genaamd. (Rhinolophus Geoff., Cuv., Illig., Desm.; Vespertilio Linn., Erxl.; Noctilio Khul). Tijdschrift voor Naturrlijke Geschiedenis en Physiologic, Amsterdam 1(1): 1-30, 1 pi. Templeton, W. E. 1848. In Blyth, E., Report of Curator, Zoological Department. Journal of the Asiatic Society of Bengal, Calcutta 17 (1): 247-255. Thomas, O. 1880. Description of a new bat from Java, of the genus Kerivoula. Annals and Magazine of Natural History; including Zoology, Botany and Geology, London (5) 5 472^73. 1898. Description of a new bat from Selangore. Annals and Magazine of Natural History; includ- ing Zoology, Botany and Geology, London (7) 1: 360-362. 1 905. A new genus and two new species of bats. Annals and Magazine of Natural History; includ- ing Zoology, Botany and Geology, London (7), 16: 572-576. 1916. List of Microchiroptera, other than leaf-nose bats, in the collection of the Federated Malay States Museum. Journal of the Federated Malay States Museums, Kuala Lumpur 7: 1-6. 1923. Scientific results from the Mammal Survey. No. XLI. — On the forms contained in the genus Harpiocephalus. Journal of the Bombay Natural History Society, Bombay 29: 88-89. Turtle, M. D. 1974. An improved trap for bats. Journal of Mammalogv, Baltimore 55: 475-477, 1 fig. Tidemann, C. R. & Woodside, D. P. 1978. A collapsible bat trap and a comparison of results obtained with the trap and with mist-nets. Australian Wildlife Research, Melbourne 5: 355-362, 5 figs, 1 tab. Van Deusen, H. V. 1961. New Guinea record of the tube-nosed insectivorous bat, Murina. Journal of Mammalogy, Baltimore 42: 531-533. Waterhouse, G. R. 1845. Descriptions of species of bats collected in the Philippine Islands, and pre- sented to the society by H. Cuming, Esq. Proceedings of the Zoological Society of London 3-10. Wroughton, R. C. & Ryley, K. V. 1913. Scientific results from the mammal Survey. III. A. — A new species of Myotis from Kanara. Journal of the Bombay Natural History Society, Bombay 22: 13-14. Yoshiyuki, M. 1970. A new species of insectivorous bat of the genus Murina from Japan. Bulletin of the National Science Museum, Tokyo 13: 195-198, 1 fig. 1983. A new species of Murina from Japan (Chiroptera: Vespertilionidae). Bulletin of the National Science Museum, Tokyo, Ser. A. (Zoology) 9: 41-150, 1 fig, 1 pi, 2 tabs. Manuscript accepted for publication 26 March 1984 British Museum (Natural History) Tilapine fishes of the genera Sarotherodon, Oreochromis and Danakilia Dr Ethelwynn Trewavas The tilapias are cichlid fishes of Africa and the Levant that have become the subjects of fish-farming throughout the warm countries of the world. This book described 41 recognized species in which one or both parents carry the eggs and embryos in the mouth for safety. Substrate-spawning species, of the now restricted genus Tilapia, are not treated here. Three genera of the mouth-brooding species are included though in one of them, Danakilia, the single species is too small to warrant farming. The other two, Sarotherodon, with nine species, and Oreochromis, with thirty-one, are distinguished primarily by their breeding habits and their biogeography, supported by structural features. Each species is described, with its diagnostic features emphasised and illustrated, and to this is added a summary of known ecology and behaviour. Conclusions on relationships involve assessment of parallel and convergent evolution. Dr Trewavas writes with the interests of the fish culturists, as well as those of the taxonomists, very much in mind. 580pp, 1 88 illustrations include halftones, diagrams, maps and graphs. Extensive bibliography. Publication 1983. £50 0565008781 Publications Sales British Museum (Natural History) Cromwell Road London SW7 5BD England Titles to be published in Volume 47 Miscellanea A review of the Mastacembeloidei, a suborder of synbranchiform teleost fishes Part II: Phylogenetic analysis. By Robert A. Travers A review of the anatomy, taxonomy, phylogeny and biogeography of the African neoboline cyprinid fishes. By Gordon J. Howes The haplochromine species (Teleostei, Cichlidae) of the Cunene and certain other Angolan rivers. By Peter Humphry Greenwood Miscellanea Anatomy and evolution of the feeding apparatus in the avian orders Coraciiformes and Piciformes. By P. J. K. Burton A revision of the spider genus Cyrba (Araneae: Salticidae) with the descriptions of a new presumptive pheromone dispersing organ. By F. R. Wanless Printed in Great Britain by Henry Ling Ltd., at the Dorset Press, Dorchester, Dorset Bulletin of the British Museum (Natural History) Anatomy and evolution of the feeding apparatus in the avian orders Coraciiforme; and Piciformes P. J. K. Burton Zoology series Vol47 No 6 29 November 1984 The Bulletin of the British Museum (Natural History), instituted in 1949, is issued in four scientific series, Botany, Entomology, Geology (incorporating Mineralogy) and Zoology, and an Historical series. Papers in the Bulletin are primarily the results of research carried out on the unique and ever-growing collections of the Museum, both by the scientific staff of the Museum and by specialists from elsewhere who make use of the Museum's resources. Many of the papers are works of reference that will remain indispensable for years to come. Parts are published at irregular intervals as they become ready, each is complete in itself, available separately, and individually priced. Volumes contain about 300 pages and several volumes may appear within a calendar year. Subscriptions may be placed for one or more of the series on either an Annual or Per Volume basis. Prices vary according to the contents of the individual parts. Orders and enquiries should be sent to: Publications Sales, British Museum (Natural History), Cromwell Road, London SW7 5BD, England. World List abbreviation: Bull. Br. Mus. nat. Hist. (Zool.) © Trustees of the British Museum (Natural History), 1984 The Zoology Series is edited in the Museum's Department of Zoology Keeper of Zoology : DrJ.G. Sheals Editor of Bulletin Assistant Editor Dr C. R. Curds Mr C. G. Ogden ISBN 0 565 05009 5 ISSN 0007-1498 British Museum (Natural History) Cromwell Road London SW7 5BD Zoology series Vol47 No. 6 pp 33 1-443 Issued 29 November 1984 , , .. f.i r j- , Anatomy and evolution of the feeding apparatus in P. J. K. Burton British Museum (Natural History), Tring, Hertfordshire HP23 6AP Contents Synopsis Introduction Material examined Abbreviations PART 1 : Feeding behaviour Alcedinidae Todidae Momotidae Meropidae Leptosomatidae Coraciidae Upupidae Phoeniculidae Bucerotidae Galbulidae Bucconidae Capitonidae Indicatoridae Ramphastidae Picidae PART 2: Anatomy Osteology and arthrology of the skull Jaw muscles M. adductor mandibulae externus Coraciiformes Piciformes M. pseudotemporalis superficial Coraciiformes Piciformes M. pseudotemporalis profundus and M. adductor posterior M. pterygoideus Coraciiformes Piciformes M. protractor pterygoidei et quadrati M. depressor mandibulae Tongue and hyoid apparatus Tongue Hyoid skeleton Hyoid muscles M. mylohyoideus M. ceratohyoideus M. stylohyoideus M. serpihyoideus M. branchiomandibularis M. genioglossus (NA 30 N 333 335 336 336 338 338 339 340 340 341 341 342 342 342 343 344 345 346 346 347 349 349 359 359 361 365 367 367 369 371 373 374 379 381 383 383 383 385 387 387 388 388 389 390 391 Bull. Br. Mus. not. Hist. (Zool.)47(6): 331-443 Issued 29 November 1984 331 332 P. J. K. BURTON M. ceratoglossus 392 M. ceratoglossus anterior and M. hypoglossus medialis 392 M. hypoglossus obliquus 392 M. tracheohyoideus and M. tracheolateralis 393 M. thyrohyoideus 394 The neck and its musculature 394 Cervical vertebrae 394 Cervical muscles 396 M. biventer cervicis 396 M. spinalis cervicis and M. splenius colli 396 M. splenius capitis 396 Mm. pygmaei 396 Mm. ascendentes cervicis 397 M. longus colli ventralis 397 M. flexor colli brevis 397 M. complexus 398 M. rectus capitis superior 398 M. rectus capitis lateralis 398 M. rectus capitis ventralis 398 Mm. intertransversarii and Mm. inclusi 398 PART 3: Functions and evolution 399 Bill and skull 399 Overall form 399 Kinesis 40 1 Safety devices 402 Protraction stops 403 Retraction stops 403 Support of the jugal bar and palatines 404 Desmognathy 405 Pterygo-palatine articulation 405 Lower jaw 406 Jaw muscles 407 M. adductor mandibulae externus 407 M. pseudotemporalis superficialis 410 M. pseudotemporalis profundus and M. adductor posterior 41 1 M. pterygoideus 41 1 M. protractor quadrati et pterygoidei 413 M. depressor mandibulae 4 1 3 Tongue, hyoid and hyoid musculature 414 Tongue and hyoid skeleton 414 Hyoid musculature 415 M. mylohyoideus 415 M. ceratohyoideus 415 M. stylohyoideus 415 M. branchiomandibularis 415 M. genioglossus 416 M. ceratoglossus 4 1 7 M. ceratoglossus anterior and M. hypoglossus medialis 417 M. hypoglossus obliquus 417 M. tracheohyoideus and M. tracheolateralis 418 M. thyreohyoideus 418 The neck 418 'Cervical vertebrae 418 Cervical musculature 419 M. biventer cervicis 419 M. spinalis cervicis and M. splenius colli 420 M. splenius capitis 420 Mm. pygmaei 420 Mm. ascendentes cervicis 421 FEEDING APPARATUS IN CORACIIFORMES & PICIFORMES 333 M. longus colli ventralis 421 M. flexor colli brevis 421 M. flexor colli profundus 421 M. complexus 421 M. rectus capitis superior 422 M. rectus capitis lateralis 422 M. rectus capitis ventralis 422 Mm. intertransversarii and Mm. inclusi 422 PART 4: Systematic review 423 Introduction 423 Typical Coraciiformes 423 Alcedinidae 423 Todidae 424 Momotidae 425 Meropidae 425 Leptosomatidae 426 Coraciidae 426 Hoopoes, Wood-hoopoes and Hornbills 426 Upupidae 428 Phoeniculidae 428 Bucerotidae 428 Jacamars and Puffbirds 429 Galbulidae 430 Bucconidae 43 1 Typical Piciformes 43 1 Capitonidae 432 Indicatoridae 432 Ramphastidae 433 Picidae 434 Phylogeny and affinities with other orders 435 Phylogeny 435 Morphology of the stapes 437 Classification 438 Acknowledgements 438 References 439 Synopsis The fifteen families recognized by Peters (1945, 1948) as comprising the orders Coraciiformes and Piciformes form the subject of this study, which is presented in four parts. The first summarizes avail- able information on feeding behaviour from fieldwork and the literature. In the second part, which constitutes the main results of the investigation, a detailed account of the anatomy of jaws, tongue and neck is given. Part three examines the anatomical findings from functional and evolutionary standpoints. Overall skull form is discussed in relation to diet, posture and requirements for vision. Kinetic coupling is achieved either via the postorbital ligament, or by modifications of the quadrate-mandible articulation (Bucorvus and Megalaimd); factors affecting the degree and type of coupling are examined. Bony stops and other safety devices present in the kinetic apparatus are reviewed, and their effectiveness assessed. Functional properties of the desmognathous palate are examined; it is noted that this condition is present in large-billed forms, and in families which regularly beat prey against a perch. A distinctive form of pterygo-palatine articulation is present in the Indicatoridae and Picidae; the need for studies of its ontogeny is pointed out. The lower jaw lacks structural features permitting gape enlargement by bowing the mandibular rami, despite the large food items swallowed by many species; possible reasons for this are discussed. Maintenance of the quadrate-mandible articulation is analysed; a medial brace or a deepened medial quadrate condyle are compared as possible alternative forms of protection in groups consuming large or active animal prey. M. adductor mandibulae externus is the most complex and variable jaw muscle. A detailed compari- 334 P. J. K. BURTON son of its components is made between the families studied here, and published descriptions for birds of other orders. On this basis, certain features (especially the postorbital lobe, narrow M.a.m.e. ventralis and laterally expanded M.a.m.e. caudalis) are considered primitive; the Phoeniculidae and Picidae show these features particularly well. Simpler architecture of this muscle is considered derived, but there are indications that its condition in most Coraciiformes and the Galbuloidea was derived indepen- dently from that seen in other Piciformes. The functional significance of complex vs. simple architec- ture of the muscle is discussed. M. pseudotemporalis superficialis has a mainly lateral origin in typical Coraciiformes and the Galbuloidea, although the whole muscle is much reduced in some groups, and occasionally absent in the Galbulidae. By contrast, a medial extension of origin is developed in typical Piciformes. M. pseudotemporalis profundus is generally parallel fibred, varying mainly in size. It is absent in many Alcedinidae as a consequence of changes in skull form. M. pterygoideus dorsalis is clearly divided into lateral and medial parts, except in the Upupidae, Phoeniculidae and Ramphastos. The functional significance of division, and the relative proportions of the two components are dis- cussed. In several families, M.pter.dors.lat. has attachment to the maxillopalatine, permitting the muscle to exert force on the upper jaw directly, rather than through the palatal complex — a feature previously undescribed in birds. Bipinnate structure of M.pter.dors.lat. in several families may be related to small amplitude jaw movements or to isometric contraction. A slip attached directly to the skull base (retractor palatini) is present in the Upupidae, Phoeniculidae and Bucerotidae, derived from M.pter.dors.med. and vent.med. M.pter.dors.med. is divided into anterior and posterior portions in the Coraciidae and Leptosomatidae. M. protractor is much enlarged in the Upupidae, Phoeniculidae and Picidae. In the first two families, this is related to foraging by 'gaping', and is associated with a bulky M. depressor, but in the Picidae, enlargement is a consequence of the muscle's shock absorbing function during hammering, and M. depressor is of normal size. The tongue is much reduced in the Alcedinidae, Upupidae, Phoeniculidae and Bucerotidae. This may be related to diet in the Alcedinidae, and to bill reinforcement in the Upupidae and Phoeniculidae. Brush tongues occur in several families, showing no obvious correlation with diet or feeding methods. The Picidae show reduction of the tongue itself, but great elongation of the basihyal and hyoid horns to form a probing organ capable of great extension, with a capacity for fine manipulation at the tip. Comparison with other families of typical Piciformes gives some insight into the evolution of this fea- ture. Most Capitonidae, the Ramphastidae and the Indicatoridae possess an entoglossum of distinctive and unusual form. M. ceratohyoideus is lacking in the Alcedinidae, Indicatoridae and Picidae. M. stylohyoideus, the main tongue retractor, is lost in the Bucerotidae, which rely largely on head jerking to propel food items backward to the oesophagus. It is also absent in the Coraciidae, Galbulidae and Bucconidae, where it is functionally replaced by a slip from M. serpihyoideus; in Leptosomus, both this slip and an M. stylohyoideus are present. M. branchiomandibularis ranges from very highly developed in the Picidae to virtually absent in the Alcedinidae. Its sites of origin vary extensively among the Coraciiformes, and may be related to the history of the sporadically distributed M. genioglossus. M. ceratoglossus anterior and M. hypoglossus medialis are best developed in the Meropidae and Galbuli- dae. M. hypoglossus obliquus is considerably elongated in the Upupidae, Phoeniculidae and typical Piciformes; contracting synergically with M. ceratoglossus, this imparts rigidity to the tongue/basihyal unit during forceful probing actions. Extreme development of this condition is seen in the Picidae, with some approach in the Indicatoridae. M. tracheohyoideus originates either on the clavicle (Coracii- formes and Galbuloidea) or on the sternum (typical Piciformes). In the Bucerotidae it originates on both: this may be the primitive condition from which the two alternatives are derived. The cervical vertebrae show reduction or loss of fused ribs in the Bucerotidae and Picidae, and in the former, the first and second vertebrae are fused. A sub-vertebral canal is well developed in the Picidae; though enclosing the carotids, its primary function is probably to provide increased surface for attachment of the highly developed M. longus colli ventralis. The evolution of neck muscle attachments is discussed; reduction in attachment sites is probably a derived state in most cases, but the possibility that increased numbers of attachments might evolve should perhaps not be discounted. M. biventer cervicis is strikingly enlarged in the Alcedinidae, presumably to maintain posture of the heavy head and bill; the Cerylinae and some Daceloninae possess a transverse aponeurosis between right and left muscles. The muscle is also enlarged in Jynx and Indicator, but in specialized excavators (Upupidae, Phoeniculidae, other Picidae) it is reduced. M. splenius capitis has additional origin on 3 in Phoeniculus and some Bucerotidae. M. longus colli is highly specialized in the Picidae, with short slips eliminated and long ones enlarged, enabling the neck to be held rigidly flexed during hammering. M. complexus is highly developed in the Alcedinidae, with origin as far back as 8 or 9 in some; the mechanics and adaptive significance of this are discussed. FEEDING APPARATUS IN CORACIIFORMES & PICIFORMES 335 M. rectus capitis lateralis is unusually small in the Galbuloidea, where its role may be partly taken over by M. splenius capitis. In the fourth and final part of the paper, the principal features of feeding apparatus structure are summarized family by family, and their adaptive and taxonomic significance are discussed. The con- cluding section examines the picture of phylogeny which emerges for the assemblage as a whole. Three main phylectic lines are envisaged; the first consists of the first six Coraciiform families of Peters (1945) together with the Galbuloidea. This line is itself split between the Alcedinidae, Todidae, Momotidae and Meropidae on the one hand, and the Coraciidae, Leptosomatidae, Galbulidae and Bucconidae on the other. The Upupidae, Phoeniculidae and Bucerotidae are considered to represent a second major line, in which the first two families are more closely allied, while the third line consists of the four remaining Piciform families. Among these, the Indicatoridae and Picidae are closely allied, with Jynx occupying a somewhat uncertain position between the two. The Ramphastidae are considered directly derived from the Capitonidae. In the classification proposed on the basis of this phylogeny, the Brachypteraciinae and Jynginae are given family rank. The Ramphastidae are, for the present, retained as a family. Introduction The avian orders Coraciiformes and Piciformes, as defined by Peters (1945, 1948) comprise 15 families of mainly small to medium sized birds, remarkable for their morphological diver- sity. This diversity is nowhere more clearly shown than in the structure of the feeding appar- atus, so that they form an ideal subject for studies of feeding adaptations and functional anatomy. Curiously, though, relatively scant attention has been paid to these matters, except in the case of the Picidae. Probably, this is because questions of pure taxonomy have always been uppermost, for although some alliance between the two orders has long been generally accepted, the interrelationships of their component groups have never been satisfactorily clarified. The study presented here has concentrated on the adaptive aspects hitherto neglected, but at the same time, has furnished extensive new evidence relating to the evolution and affinities of the various families. A detailed historical review of the classification has been given by Sibley & Ahlquist (1972), and need not be repeated here. However, some resume of the present state of knowl- edge is necessary, and can be conveniently grouped under the headings of three possible types of relationship. 1. Relationships between families within an order. More problems are presented by the Coraciiformes, which fall roughly into three groups. Most heterogeneous of these is that made up by the Alcedinidae, Todidae, Momotidae and Meropidae, which in most respects show only a loose affinity, but share a distinctive stapes form (Feduccia, 1977, 1980). The Coraciidae (including the Brachypteraciinae, treated as a family by some authors) seem fairly definitely allied to the Leptosomatidae (see Cracraft, 1971), but both appear rather distantly related to the preceding families. The Upupidae, Phoeniculidae and Bucerotidae are usually considered to be allied, but the connection between these three and the rest of the order seems less obvious. Within the Piciformes, relationship between four of the families (Capitonidae, Indica- toridae, Ramphastidae and Picidae) seems fairly well established, but the position of the Galbuloidea (Galbulidae and Bucconidae) is more open to question (but see Steinbacher, 1937). 2. Relationships between Coraciiform and Piciform families. Sibley & Ahlquist (1972) have produced biochemical evidence for affinity between the Galbuloidea and Alcedinidae, and in view of the uncertain position of the former relative to the rest of the Piciformes, the possibility of a connection with the Coraciiformes must be taken seriously. 3. Relationships with other orders. Many such have been proposed in the past, but only three seem at present to merit further investigation. These are, the possible connections between the Indicatoridae and the Cuculiformes (Sibley & Ahlquist, 1972), between the Trogoniformes and the Coraciiformes (Sibley & Ahlquist, 1972; Feduccia, 1977) and 336 P. J. K. BURTON between the Piciformes and the Passed formes. To keep this study within practical limits, only members of the Coraciiformes and Piciformes have been included. However, the find- ings presented here have some relevance to two of these suggested relationships, arguing strongly against any honeyguide — cuckoo connection (p. 433), but providing tentative support for associating the trogons with the Coraciiformes (p. 437). The first two parts of this paper are factual, and the following two interpretative. A survey of available information on feeding behaviour is followed by the detailed descriptive anatomy which constitutes the main results of the investigation. In the third part, the anatomical systems studied are re-examined from a functional and evolutionary perspective, and in the final part, the families are reviewed in turn, considering both adaptive and taxonomic aspects. The concluding discussion deals with the principal taxonomic problems, and presents a phylogeny and classification based on the results of the study. The taxonomic nomenclature and classification used is that of Peters, but a problem exists in regard to the naming of structures. This manuscript was already largely complete when the Nomina Anatomica Avium (Baumel, et al., 1979) appeared, and not all the anatomical names used were in agreement with it. The revision needed to achieve total agreement with the N.A.A. would have been a daunting proposition, but it is hoped that the lesser adjustment which has been adopted will suffice. This is to give the N.A.A. equivalent in brackets following all headings in Part 3 which use a superseded term. Material examined The anatomical specimens that were available for this investigation are mostly those listed by Blandamer & Burton (1979); the main addition to the orders studied since this list was published is a specimen of Bombylonax breweri (Meropidae) presented by Dr C. H. Fry. At least one specimen of every genus in the alcoholic collection has been dissected, and several species in the case of large genera such as Halcyon; skeletons of all available genera have also been examined. Four species absent from the collections of the British Museum (Natural History) were loaned by the American Museum of Natural History; these were Galbula tombacea, Brachygalba lugubris (Galbulidae); Nystalus maculatus and Hypnelus bicinctus (Bucconidae). Birds of several other orders have been dissected to check specific points, and these are noted in the text. Abbreviations Jaw muscles, figs 7-24 amec M. adductor mandibulae externus caudalis amer M. adductor mandibulae externus rostralis amerl M. adductor mandibulae externus rostralis lateralis amerm M. adductor mandibulae externus rostralis medialis amerp M. adductor mandibulae externus rostralis, postorbital slip amert M. adductor mandibulae externus rostralis temporalis amev M. adductor mandibulae externus ventralis amevc shared fibres of M. adductor mandibulae externus ventralis and caudalis ap M. adductor posterior (N.A.A.: M. adductor mandibulae caudalis) dm M. depressor mandibulae pr M. protractor pterygoidei et quadrati prl M. protractor, part 1 pr2 M. protractor, part 2 psp M. pseudotemporalis profundus pss M. pseudotemporalis superficialis pssl M. pseudotemporalis superficialis, lateral part pssm M. pseudotemporalis superficialis, medial part FEEDING APPARATUS IN CORACIIFORMES & PICIFORMES 337 ptd M. pterygoideus dorsalis ptdl M. pterygoideus dorsalis lateralis ptdm M. pterygoideus dorsalis medialis ptdmp M. pterygoideus dorsalis medialis posterior ptmxp M. pterygoideus dorsalis, maxillopalatine slip ptv M. pterygoideus ventralis ptvl M. pterygoideus ventralis lateralis ptvle venter externus of M. pterygoideus ventralis lateralis ptvm M. pterygoideus ventralis medialis ptvms medial slip of M. pterygoideus ventralis medialis qml Lig. quadrate- mandibulare rp retractor palatini portion of M. pterygoideus dorsalis medialis Hyoid skeleton figs 25-26 has basihyal (N.A.A.: basibranchiale rostrale) cebr ceratobranchiale ent entoglossum epbr epibranchiale uro urohyal (N.A.A.: basibranchiale caudale) Tongue muscles figs 27-30 bmd M. branchiomandibularis bmdl M. branchiomandibularis, lateral origin bmdm M. branchiomandibularis, medial origin cgl M. ceratoglossus cglap aponeurosis of M. ceratoglossus chy M. ceratohyoideus (N.A.A.: M. interceratobranchialis) hyp M. hypoglossus obliquus mhy M. mylohyoideus (N.A.A.: M. intermandibularis) sehy M. serpihyoideus sehya M. serpihyoideus, anterior slip sthy M. stylohyoideus thy M. thyrohyoideus (N.A.A.: M. cricohyoideus) thyd M. thyrohyoideus, dorsal part thyv M. thyrohyoideus, ventral part trhy M. tracheohyoideus (N.A.A. nomenclature: see text) Cervical muscles fig. 3 1 asc Mm. ascendentes cervicis (N.A.A. : M. cervicalis ascendens) co 6-8 M. complexus, slips of origin from vertebrae 6 to 8 itr M. intertransversarius res M. rectus capitis superior (N.A.A.: M. rectus capitis dorsalis) spin M. spinalis cervicis (N.A.A.: M. longus colli dorsalis) spla Mm. splenii accesorii splc M. splenius capitis 338 P. J. K. BURTON PART 1 Feeding behaviour This section presents a digest of available information on feeding behaviour in each family, derived largely from a survey of the literature, but also to some extent from personal obser- vations carried out in the course of this study. Some of the latter have already been published at greater length elsewhere (e.g. Burton, 1977, 1979), while others are in preparation. The account given here concentrates mainly on the actions performed by jaws, tongue and neck during feeding. Details of diet and some general ecological background are given, but to pursue these aspects further, it will be necessary to consult the references given. ALCEDINIDAE Despite their name, most kingfishers do not fish. Of the three sub-families, only the Cerylinae is exclusively piscivorous. The Alcedininae also hunt mainly by plunge diving, though in several species (e.g. Alcedo leucogaster, A. cristata, A. meninting), it appears that the prey is aquatic insects and other organisms rather than exclusively fish. Some (e.g. Ceyx spp.) apparently eat mainly terrestrial or aerial insects. The Daceloninae are predominantly terrestrial feeders, and only a few, notably Pelargopsis spp. hunt regularly by plunge diving. Unfortunately, precise information is lacking on the feeding behaviour of most members of this large and diverse subfamily, even such specialized forms as the nocturnal Melidora, or Clytoceyx which apparently excavates mud for worms with its short but massive bill. Further details on feeding habits throughout the Alcedinidae are provided in the review by Fry (1980). Whether the prey taken are aquatic or terrestrial, the feeding strategy employed is essen- tially the same throughout the family; a likely area is scanned from a perch, and periodically a sally is made to capture prey. Searching and sallying are nearly always directed downward to prey on the ground or in the water. A few kingfishers show more versatility; Halcyon pileata, for instance, seeks prey with frequent head movements directed to foliage above as well as below (Burton, 1979). Several species capture airborne insects, e.g. Ceyx spp., even H. chloris (Burton, 1979). Aquatic prey may be sought in hovering flight, particularly among the Cerylinae, notably C. rudis (Douthwaite, 1976). Prey captured are usually large relative to the size of the bird, including reptiles, nestling birds and small rodents in the case of the larger Daceloninae. Capture rates are fairly low, particularly in large species. Prey are commonly beaten against the perch before consump- tion, with a sharp sideways action. Dacelo novaeguineae drops snakes to the ground repeatedly to kill them, and there are records of two individuals cooperating to kill a snake (Keast, 1969). There is some evidence that scorpions may have the sting removed before consumption (Burton 1979, H. concreta), but details on the treatment of venomous prey are generally lacking. Fish are swallowed head first, after being oriented along the bill axis. For A. atthis, many photographs exist showing fish grasped in the bill (those with the head forward are probably destined for nestlings). In at least some, e.g. Massny (1977), the jaws appear almost parallel, with the lower projecting slightly beyond the upper, evidence of strong retraction of the upper jaw. Plunge diving, ideally, is accomplished by an almost vertical drop from the perch. How- ever, in species which fish from perches, many dives are strongly oblique. The author has seen A. atthis make captures after a slanting flight of nearly 10 metres from a low perch; such instances imply considerable skill in compensating for refraction. There is presumably some relation between the height of dives, and the depth of penetration into the water, though this relationship is unlikely to be a simple one due to habits of prey and availability of perches. However, larger species tend to dive from greater heights than smaller ones, and presumably thus have access to deeper swimming fish. In A. atthis, at least, the nictitating FEEDING APPARATUS IN CORACIIFORMES & PICIFORMES 339 membranes cover the eyes on entry to the water, and if the fish evades capture, inanimate objects may be briefly grasped, suggesting that the prey cannot be seen during the final moments of the attack. The bill is held slightly open as the water is entered (see Junge & Lutken, 1974). An important use of the bill in all the Alcedinidae is nest excavation. Usually the nest burrow is excavated in a bank of sand or earth. The process has been well described by Eastman (1969). The tunnel is started by flying at the bank and striking it with the bill. Once enough material has been removed to gain a purchase with the feet, excavation continues from a perched position, until the bird is far inside the tunnel, working with the bill, and using the feet to shovel out material. Some kingfishers at least occasionally excavate wood. Excavation of a tree nest site by H. Moris has been described in detail by Harrisson (In Smythies, 1960) and England (pers. comm.) kept a pair of captive Kookaburras, Dacelo gigas, which cut through an inch thick board of elm wood. The writer has found a nest of Ceyx pusillus in the trunk of a rotten tree in lowland rainforest in New Guinea. TODIDAE The feeding behaviour and ecology of the todies have been described in great detail by Kepler (1977) from whose paper this account is condensed. Todies forage in a variety of brush or forest habitats, favouring situations with an abundance of twigs, branches and leaves. Insects are captured from all available substrates; in descending order of preference these are leaves, twigs and branches, the air, inflorescences and fruits, and the ground. Perches chosen are those not normally covered by close-touching leaves, and affording a view both above and below. Foraging heights for most species are generally substantially below canopy level; T. subulatus forages higher than T. angustirostris, and this difference is most marked where both occur together. Kepler describes two principal feeding methods and three minor ones. Most frequent of all is leaf feeding; the bird perches with bill inclined upwards at angles up to 45° from the horizontal, and scans the undersides of leaves and twigs above it. Seeing an insect, it flies up to the leaf or twig, snaps its bill, and continues in an unbroken arc to another perch. Occasionally insects are taken from the tops of leaves in a downward swoop. Air feeding, second in importance, consists of flycatcher like sallies to capture a passing insect, usually returning to a different perch. Minor feeding methods are hovering in front of feeding sub- strates, snapping prey from tree trunks, and sliding along perches to glean prey in warbler fashion. After a capture, the insect is usually swallowed in flight, but they kill larger ones either by flicking their heads to and fro or by vigorous beating against twigs for up to a minute. Feeding rates are greatest in xeric scrub, averaging 1-7 per minute, and slowest in rain- forest, averaging 1-0 per minute. Capture rate averages about 41% in dry scrub and 33% in rainforest. Flight distances range from about 1 -9 m to 2-6 m., varying somewhat according to species and habitat. Where T. subulatus and T. angustirostris occur together, the former flies longer distances, illustrating another feature of their behavioural divergence. Prey recorded for T. mexicanus includes 1 1 insect orders, spiders, and small amounts of miscellaneous items, including plant material. Diptera account for about 31% of the adult diet, and Coleoptera for 23%. Food given to nestlings includes much fewer Diptera, but considerably more Momoptera and Lepidoptera. The bills are also used for excavating nest tunnels in banks, and occasionally in rotten wood. Sometimes several centimetres of mosses and liverworts have to be torn away to expose the soil. The burrows are usually curved, and Kepler records an average length of 30.5 cm in wet forests, slightly shorter (26-9 cm) in dry scrub. Excavation usually lasts several weeks, the tunnel growing at about 0-5 cm per day. When starting a tunnel, the birds hover in front of the bank, jabbing it with their bills in bouts of 2-5 thrusts. Loose debris is scraped out with the feet, though occasionally large particles may be carried out in the bill. 340 P. J. K. BURTON MOMOTIDAE Baryphthengus spp. are generally birds of the middle to lower stories of the forest, feeding on large animal prey with some fruits. Johnson (1954) noted B. martii in attendance near ant swarms, perching quietly watching for prey animals moving out to escape the ants; these were then taken from tree trunks or palm leaves. One bird which took a large scorpion spent five minutes killing the creature before swallowing it whole; Johnson's account does not mention treatment of the sting. Small lizards, orthoptera, caterpillars, wasps and various fruits are other food items recorded by Wetmore (1968) for B. martii. Closely similar to Baryphthengus in diet and habits, Momotus spp. range into relatively open areas of second growth or into isolated thickets, as well as forest. Wetmore (1968) and Skutch (1964) record that small birds are exceptionally taken as prey; other items recorded are similar to those listed for B. martii. Prey are banged on the perch vigorously and repeatedly, either to quieten them, or to break off wings. Skutch (1964) notes for M. momota that seeds of the nutmeg family are swallowed whole, to digest the thin aril surrounding the seed, which is later cast up intact. Though most hunting seems to be done by watching from a perch, M. momota may be seen to forage for extended periods on the ground, hopping about with tail cocked up in a manner reminiscent of a thrasher, Toxostoma. Skutch (1964) noted an individual of M. momota on the ground, apparently searching for a dropped prey item, pushing aside fallen leaves with alternate sweeps of the bill to left and right. Hylomanes is the least known genus of the family. Wetmore (1968) says it is found alone or in pairs, resting quietly in open undergrowth in humid forest. Prey recorded by him include a spider, a snail, Coleopter, Orthoptera, Homoptera, caterpillars and ants. Motmots also use their bills to excavate long burrows in earth banks for nesting. An account of excavation in M. momota is given by Skutch (1964). MEROPIDAE The most thorough survey of Bee-eater feeding behaviour is contained in a comparative study of the various African species by Fry (1972). Except for Asian and Australasian Meropidae, the details which follow are condensed from his account, with extensive verbatim passages. Small bee-eaters, such as M. pusillus, M. variegatus, M. orientalis, M. boehmi and M. reviolii feed by 'fly-catching'. They keep watch for flying insects from low perches, usually within 2 metres of the ground, and make brief sorties of seldom more than a few metres, returning to the perch with their victim and immobilizing it there. Like all bee-eaters, they are extremely adept at catching the prey, which seldom escapes. Middle size Merops — M. hirundineus, M. oreobates, M. bulocki, M. bullockoides, M. albi- collis and M. gularis-a\so behave like flycatchers, returning to the perch to beat their prey. But they select more elevated vantage points, fly farther after insects, and range more widely, foraging during the course of a day over hundreds of acres or along as much as a mile of wooded watercourses. Upon alighting these species swallow small insects, for example the 5 mm long, stingless 'sweat-bees' (Trigona spp.) without treatment, but they immobilize larger insects before eating them. Among M. bulocki, the mean of 66 counts of feeding flights, made in and out of the breeding season in all weathers, was 5-2 sorties every 10 minutes. Since these birds average 1 1 waking hours a day in active feeding and make very few sorties without a capture, an individual M. bulocki must eat about 340 insects a day or 124,000 insects a year. Large birds such as M. superciliosus, M. malimbicus, M. apiaster and M. nubicus tend to use tree top perches and to range even more widely, over a few square miles daily. They are also more prone to hunt in continuous flight. M. apiaster may be seen twisting and wheel- ing in flight high over the savannas, catching probably soft-bodied insects such as flying ants and termites. With a large and hard bodied insect, they must return to a perch to immobilize it by beating and if necessary, devenoming it before eating it. (Fry, 1969). The two species of Nyctiornis are less aerial than the smaller bee-eaters. The habits of FEEDING APPARATUS IN CORACIIFORMES & PICIFORMES 341 N. athertoni are summarized by Smythies (1953); much of its food is obtained while clamber- ing about in trees, taking insects directly from leaves or flowers. Prey include various large insects, such as beetles, in addition to bees. N. amicta is more strictly a forest bird, preferring areas where trees are more thinly spaced, especially near the edge of streams or swamps. (Smythies, 1953, 1960; Robinson, 1928). Found in pairs or small groups, it perches at a height of 3 to 6 metres, darting out occasionally to seize prey; it will also clamber about in trees at times like N. athertoni. Its flight is noticeably less swift and more laboured than the smaller bee-eaters. Bee-eaters nest in burrows excavated in cliffs or flat ground. Soil is loosened with the bill and kicked out with the feet; Fry (1972) observes that they will often support the body on a tripod of bill and carpal joints, enabling both feet to be used together. He also notes that a frequent cause of death in all species is getting the head and bill jammed sideways in the tunnel. LEPTOSOMATIDAE The little we know of the feeding habits of Leptosomus is largely contained in the accounts by Rand (1964), Forbes- Watson (1967) and Appert (19680). The Cuckoo-roller inhabits forest and brushland, where it stays largely in the tree tops, and is often seen in small parties. According to Rand (1964), the food consists of chameleons, locusts, caterpillars and other large insects. The stomach is often lined with hairs from caterpillars. Large caterpillars, and probably other prey, are held in the bill and beaten against a perch to subdue them before being swallowed. Forbes- Watson (1967), who watched a nest of Leptosomus discolor found that the two young were each being fed six chameleons per day during the period of his observations. He comments that there must either be considerably more chameleons than one would sus- pect, or the birds must have a large hunting territory and be extremely efficient at finding chameleons. Neither Rand (1964, 1936), Benson (1960) or Forbes- Watson (1967) record other lizards as prey, although Forbes- Watson notes that these were abundant and con- spicuous on the ground in the vicinity of the nest he watched. Probably most, if not all, the food, is taken above ground-level. Appert's (1968a) observations give a similar picture except that at a nest he watched, caterpillars seemed to be the main food brought by the male to the incubating female, and by both sexes to the young. CORACIIDAE The rollers fall into three distinct groups as regards way of life and feeding behaviour. Typical rollers, Coracias, watch for prey on the ground from an elevated perch, fly to seize it, and then consume it there or return with it to the perch; more rarely, they may catch aerial prey, or forage for some time on the ground. Broad-billed rollers, Eurystomus, also watch for their prey from a perch, but feed almost entirely on insects captured in flight. When dense swarms pass by, they may stay on the wing for several minutes, making repeated captures while sail- ing or dashing about with a buoyant and agile flight action. The ground-rollers, confined to Madagascar, are a distinctive group of genera sometimes separated as a family (Brachypteraciidae). They are entirely ground living, and at least some are believed to be partly nocturnal. Detailed studies of roller feeding behaviour are lacking, though some quantitative infor- mation on foraging in E. orientalis will be presented in a forthcoming paper on the Artamidae (Wood Swallows) with which it often associates (Burton, in prep.). A good survey of food and ecology in some West African species is given by Thiollay (1971). Coracias spp. had taken Coleoptera, Orthoptera, ants and termites in roughly similar quantities; E. glaucurus had consumed winged ants and termites in great numbers, but otherwise, prey was predominantly coleopteran. Termites, coleoptera and cicadas are recorded for E. orien- talis. Other accounts bear out this general picture, though larger prey (scorpions, lizards, 342 P. J. K. BURTON etc.) are often recorded in the case of Coracias. Exceptionally, Eurystomus has been recorded diving into water to take fish. Apart from the fact that ground rollers do not normally hunt from perches, prey and feeding methods seem unlikely to differ in essentials from those of Coracias. The evidence for their nocturnal habits is partly based on native sources which Rand (1964) considered unreliable, but crepuscular and nocturnal behaviour has been confirmed in Uratelornis by Appert( 19686). Coracias, Eurystomus and probably Brachypteracias nest in natural cavities, and hence do not use the bill for excavation. However Uratelornis certainly excavates a breeding tunnel, and probably also Atelornis pittoides (Rand, 1964). The breeding habits of A. crossleyi are unknown. UPUPIDAE The hoopoe forages principally on the ground, and the following account of its feeding tech- niques is condensed from the account by Skead (1950). Hoopoes probe assiduously through- out the day, with the bill tip slightly open but not usually penetrating the soil to any great depth in and around grass tussocks. In softer soil the bill may be inserted full length. Prey items are held at the bill tip for a second before being tossed back into the mouth with a short, sharp jerk of the head. Skead once watched a male digging in one spot for five minutes, excavating a hole so deep that his forehead disappeared from view. The prey extracted was pounded heavily on the ground to break it up into pieces small enough to swallow. Hoopoes nearly always probe in short grass rather than long. Probing in wood is rare, but they fre- quently and successfully probe around the bases offence posts. Dry cow- pats are flicked over with a deft sideways sweep of the bill, or tipped over backwards. Skead has occasionally seen hoopoes fly up from the ground to hawk flying termites. Prey listed in this and other accounts include a large proportion of insect larvae, especially coleopterous, mole crickets, scorpions and lizards. PHOENICULIDAE Unlike the hoopoe, wood-hoopoes virtually never forage on the ground. Phoeniculus spp. seek food principally on tree trunks and branches, clambering about over them and probing into crevices and behind loose bark. Prey taken is chiefly insects, including beetles, termites and ants, but small fruits are occasionally eaten. The scimitar bills, Rhinopomastus spp., are even more specialized, with a particular fondness for acacias. The behaviour of R. cyanomelas is well described by Brown (1969) who writes as follows: Normally the bird is seen climbing about in the tops of acacia trees, or clinging to the bark as it works up or down, searching minutely for insects. It then behaves more like a creeper or tit than a wood-hoopoe, sometimes hanging upside-down on a branch, or again probing crannies in the bark with head held downwards. When feeding it constantly probes small holes and cracks delicately and gently, or pushes the bill round under stems and the globose flowers of the acacias. Evidently it survives mostly on very small insects. Brown goes on to describe how the bird works its way along branches of the ant-gall acacis, A. drepanolobium, probing with its fine bill into the tiny openings by which ants enter and leave the galls. Presumably small grubs or pupae are obtained in this way. BUCEROTIDAE Tockus spp. are generally fairly small and have less enlarged and elaborate bills than most other hornbills. Their diets are wide, and their feeding methods highly versatile. Kemp (1976) has described these in some detail, and distinguishes the following techniques, which are quoted verbatim: 1 . PICKING — The food item is picked up where it is found in the vegetation or on the ground, while the bird is standing. FEEDING APPARATUS IN CORACIIFORMES & PICIFORMES 343 2. DIGGING — Standing on the ground, the bird pushes the closed bill into the substrate and then flicks the dirt to one side, partly opening the bill at the same time. Food items are thus exposed. 3. LEVERING — On the ground, the head is lowered to one side and the closed bill slid under an object. Raising the head causes the bill to work as a lever to turn over the object. This is used when the object is too large to be moved by the digging action, and exposes the food items underneath. 4. CHASING — Pursuing an active food item on the ground. 5. SWOOPING — Flying down from a perch to obtain a food item that has been noticed on the ground below. 6. PLUCKING — Picking up a food item from the ground or vegetation without landing. 7. HAWKING — Catching a flying food item whole on the wing. Less detail is available for other members of the family, but all authors who have watched large hornbills feeding stress the dexterity and delicacy with which the huge bill can be used. For example, Fogden (1969) writing about Anthracoceros malayanus describes a technique used for dealing with potentially dangerous prey such as snakes, centipedes or scorpions. The creature is held by the very tip of the bill and repeatedly squeezed throughout its length, working from end to end several times. The extremities are given a particularly vigorous squeeze, ensuring that the head or tail sting are completely crushed. Objects such as twigs and leaves are often manipulated in 'play' in the same way. Similar dexterity is also shown in dealing with fruits, which the birds are capable of peeling using the bill alone. Objects to be swallowed, whether fruit or animal, are swallowed by tossing them into the air with the bill tip, then shifting the head and open bill to catch them at the rear of the buccal cavity. This picture generally holds good for most hornbills, whether omnivorous or primarily fruit eaters, but a few show additional specializations. Ground hornbills (Bucorvus) are adept at finding and excavating for underground wasps' or bees' nests; honey comb is removed and carried away impaled on the bill tip. Rhinoplax vigil, with its relatively short bill and heavy solid casque appears to be adapted for wood excavation, though substantiating field observations are scanty at present. An activity common to most hornbills though not Bucorvus is that of plastering mud or excreta around the entrance of the nest cavity to wall in the incubating female, using the sides of the bill to smear or pat material into place. GALBULIDAE Jacamars typically forage in flycatcher style, by making aerial sallies from a perch to catch passing insects. The most detailed published account of feeding behaviour is that for G. dea by Burton (1977). This species feeds mainly from relatively high perches at the edges of clear- ings, rather than from the low perches in scrub or forest favoured by most other jacamars. Feeding flights usually commence with a dive from the perch, followed by a swift pursuit, often involving rapid and agile changes of direction. Occasional flights are made directly upwards. After a capture, the bird returns to a perch with an undulating flight; perches were changed after about 50% of the flights. Flight distances ranged from about 1 to 1 2 metres, mostly from 4^ to 9 metres, and were of short duration — 1-3 to 5-6 seconds, averaging 3-2. Only a single capture occurred in all flights observed. Frequency of flights averaged 21-8 per hour over 5 hours, considerably greater than in Chelidoptera tenebrosa (Bucconidae) which feeds by aerial sallies in similar habitat to G. dea, and is of similar body size. Prey captured were frequently fairly large, including butterflies, dragonflies and sizable wasps. Large prey were sometimes banged on the perch several times, but back and forth rubbing to devenom Hymenoptera has only been recorded once. Other species of Galbula, and probably also Brachygalba and Jacamaralcyon, show simibehaviour, though as perches are generally lower, upward flights are more frequent, and perched birds generally sit with the bill inclined upwards at about 45° (about 20° in G. dea). Activity rates are probably higher in smaller species; the writer's limited observations on G. galbula suggest a flight frequency twice that of G. dea, and much more rapid head move- ments while sitting perched, though the length and duration of flights is generally less. As 344 P. J. K. BURTON in G. dea, prey generally seem to average fairly large for the size of the bird, with Hymenop- tera predominating. Also, as Wetmore (1968) points out, jacamars are one of the few groups of birds which regularly catch and eat butterflies. Jacamerops aurea differs from other jacamars in its more sluggish behaviour. Detailed quantitative observations are difficult to compile for this species, which is usually encoun- tered perched moderately high (6 to 10 metres) at the edge of a small clearing in forest. It may sit motionless for several minutes before making a short sally and returning to a new — and frequently invisible — perch. Usually prey taken in these sallies are seized from foliage rather than in flight. Galbalcyrhynchus, one of the most distinctive genera of jacamars, is also one of the least well known, though it may be surmised that its exceptionally heavy bill denotes a preference for large or hard bodied prey. Jacamars excavate nest burrows using the bill usually in earth banks. Skutch (1937, 1963), writing about G. ruficauda, notes that the bill is used for loosening particles of earth, which are periodically kicked out with the feet, though occasional small lumps are carried out in the bill. Nests are sometimes found in termitaria, though it has not been proved that these were actually excavated by the jacamars themselves in such a hard material. BUCCONIDAE A detailed account of feeding in Chelidoptera has been given by Burton (1977). This is the most aerial, and one of the most active of puffbirds, resembling more the Wood Swallows (Artamidae) of the Oriental and Australasian regions in its general behaviour and way of life. Most of its prey are small insects, including many Hymenoptera, captured in flight. Though less frequently using agile manoeuvres to catch prey than most jacamars, its sorties are frequently more prolonged, and may sometimes exceed a minute in length, with several captures interspersed by bouts of soaring. Perches commonly used are dead branches near the tops of trees in clearings or at forest edges, on which small groups of the birds often sit close together. Prey are mostly swallowed in flight, and rubbing or beating against the perch was not observed. Amongst other puffbirds, Monasa shows the closest approach to Chelidoptera in feeding behaviour (Skutch 1972; Burton, unpubl.), making frequent aerial sallies, which, however, are usually to seize prey from foliage, tree trunks and branches, or sometimes the ground, and less frequently to capture flying insects. They associate in small parties, usually up to a maximum of six and generally return to a new perch after a feeding sortie, with the effect that the whole group gradually moves in straggling fashion through the forest. Sallies usually involve simply a glide on set wings, occasionally with a brief hover as prey is taken, though they are capable of rapid manoeuvring when necessary. Prey include many cicadas, orthop- tera, beetles, dragonflies and occasionally butterflies, and also some spiders and millipedes. In contrast to Chelidoptera, larger prey are often beaten or rubbed against the perch, for periods of up to a minute. This is by no means confined to venomous species, but if intended to remove wings and legs is not strikingly successful; I have seen even butterflies swallowed whole after a bout of beating. A curious item of behaviour noted by Skutch is the feeding of adult birds by other members of the foraging group. Nonnula spp. are small members of the family which glean insects off foliage from undergrowth to lower tree crowns, often moving actively through cover as they forage, but sometimes waiting for a time on a perch like other puffbirds. Wetmore ( 1 968) records orthop- tera, caterpillars, earwigs, small beetles, membracids and spiders from stomachs. Equally diminutive, Micromonacha lanceolata may be similar in habits, though no information is available. Most other members of the family are larger, more heavily built birds which somewhat resemble forest kingfishers of the genus Halcyon in their habits. Normal foraging strategy is to sit on a tree perch at moderate height, motionless but for regular head movements, often waiting for many minutes until prey is seen. After a short flight to seize this, the bird FEEDING APPARATUS IN CORACIIFORMES & PICIFORMES 345 returns to the same or another perch. Prey taken are usually large insects, obtained mostly from branches or leaves, though Bucco spp. perhaps take more from the ground or low herbage. Puffbirds nest in burrows excavated in earth, or occasionally termitaria. The process of excavation has been observed only in Notharcus pectoralis (Skutch, 1948), working on a termitarium. Although this pair were invisible when inside the chamber, Skutch judged from sounds that material was removed by biting or tearing, as well as by hammering. He also notes that in three other species, Malacoptila panamensis (Skutch, 1958), Monasa mor- phoeus (Skutch, 1972) and Chelidoptera tenebrosa (Skutch, 1948), no excavated earth is found around the burrow entrance, suggesting that this is carried away by the birds. Sticks and dead leaves are placed around the entrance by these three species, a habit especially strongly developed in Monasa. CAPITONIDAE Barbets occupy a variety of tropical habitats ranging from rainforest to acacia savannahs and dry scrubland. The diet of most species appears to be primarily fruit with some insects, but because of the difficulties of observation in many of their habitats, information on feeding techniques is derived largely from captive birds (See especially England, \913a & b, 1975 1976, 1977). In the wild, the numerous species of figs are probably a major food source for many forest barbets. An interesting photograph in Thomson (1964) shows an individual of Megalaima zeylanica carrying four figs simultaneously in its bill. In more open or arid habitats, various berries are recorded, as well as leaves and shoots. Tinker-barbets, Pogoniulus spp., often use the berries of'mistletoes as a staple item of diet. Insect items recorded include many Orthop- tera, as well as beetles, ants, etc. Termites are probably important to Trachyphonus spp., at least one species of which nests in termite mounds. Captive barbets subsist satisfactorily on a variety of prepared fruits, but England notes for several species that the diet switches largely to insects just before and during breeding. It appears that the young are fed mainly on insects at least until the later stages of fledging, as Skutch (1944) observed for Semnornis frantzii in the wild. Under aviary conditions, meal- worms and locusts are the principal insect food supplied, and their treatment by the barbets is of some interest. Mealworms are often 'mandibulated', that is, crushed and softened by running them from one end to the other several times through the bill tip, which makes many rapid small amplitude bites on the insect. This process is used especially, though not exclusively, when the mealworm is to be fed to fledglings. Treatment of locusts varies from one species to another. Thus, England (1975) notes that Eubucco bourcieri holds the victim on the perch by one foot, or sometimes both, and very occasionally raises a foot to the beak, usually to remove and clasp food which requires more tearing up before swallowing; Trachyphonus erythrocephalus holds the locust in its bill by a leg or a wing and shakes it violently until the body comes off, repeating the process until it is devoid of appendages; Megalaima chrysopogon seizes and mandibulates the locust, occasionally throwing it up to catch it another way round, then finally swallows it whole. The writer watched M. zeylanica and Psilopogon pyrrholophus both confronted with a locust for the first time; M. zeylanica treated the insect exactly as did England's M. chrysopogon, while P. pyrrholophus appeared nonplussed, and after a few desultory attacks finally ignored it. More interestingly, a captive specimen of Semnornis rhamphastinus treated the locust in a similar way to England's Eubucco, using a foot to hold it on the perch. The extraordinary pronged bill tip was then used with great dexterity to systematically pluck all appendages off the locust's body, attacking them near the base, where they were attached to the thorax. The insect was then consumed without further treatment. Use of the feet has been noted also in S. frantzii (Skutch, 1944), which apparently is mainly vegetarian in the wild, consuming fruits, berries and the petals of flowers. Skutch notes that by contrast, E, bourcieri in the Costa Rican forests is insectivorous, spending its time well 346 P. J. K. BURTON up among forest trees, where it investigates curled or rolled dead leaves, either probing the interior, or biting them on the outside to drive out prey. The Asian Calorhamphus fuligino- sus appears to have similar habits, foraging in small parties, and investigating crevices and crannies of all kinds, often behaving in an acrobatic manner more reminiscent of a tit than a barbet (Hume & Davison, 1878). Megalaima haemacephala is recorded as making ungainly aerial sallies after insects (Hyatt, in Gooders, 1969-71). Most barbels use their bills for nest excavation, as well as for feeding. In the great majority, nest holes are bored in rotten timber, even colonially in at least one genus (Gymnobucco). Trachyphonus spp. excavate in banks, termite mounds, or the ground. The form of the nest chamber and its entrance varies, but it is generally more irregular than those of woodpeckers (Picidae). An account of excavation techniques given for Tricholaema diadematum by England (19730) is probably fairly typical. He writes '...each would attack the wood, hammering away until a piece was loosened at one end; this was then seized and torn off and, when a large beakful was free, it was carried to the far end of the flight and dropped onto an ever-increasing heap. Amazingly, small pieces of wood were swallowed and regurgi- tated on to the heap; practically nothing remained beneath the hole to give the site away to enemies.' Similar techniques have been observed under natural conditions for Semnornis frantzii by Skutch (1944). Use of the bill in grasping or consuming wood, and the habit of carrying debris away are important differences from the excavation methods of woodpeckers. Excavation continues during the fledging period as noted by England and by Skutch (1944). The debris thus produced serves the function of soaking up the nestlings' droppings, and soiled wood chips are regularly carried away. In general, the nesting habits of barbets appear more varied and flexible than those of woodpeckers, even if their architecture is less perfect. Fuller details of their interesting breeding biology are given by Skutch, and in the series of papers by England. INDICATORIDAE The honeyguides are remarkable for their ability to use wax as a food source (see Friedmann, 1955 for a full account). In the case of the genera Indicator, Melichneutes and Melignomon this is derived from the combs of wild bees; the bees themselves, and their larvae and pupae are also eaten. At least two species of Indicator, both African, have evolved a relationship with ratels and men in which these predators are guided to the bees' nests by the birds, which can then feed with ease on the remains of the nests after they have been plundered. This is by no means the only food source of honeyguides; a variety of insects are eaten, often captured on the wing. The smaller sharp-billed Honeyguides (genus Prodotiscus) do not attack bees' nests, but nevertheless obtain a considerable amount of wax by eating scale insects (Hemiptera, Homoptera, Coccoidea). Honeyguides are parasitic in their nesting habits, and the nestlings of at least two species of Indicator possess well developed hooks at the tips of both jaws which are used to bite the young of the host, eventually causing death; a detailed account of this behaviour in /. minor is given by Friedmann (1955). The hooks are lost early in the fledging period. It is not known whether Prodotiscus spp. behave similarly, or possess such hooks, although Maclean (1971) found none in a week old P. regulus. RAMPHASTIDAE Toucans are primarily fruit eaters, but also take a variety of small animals, including insects, reptiles, and the eggs and nestlings of other birds. Fruits are plucked with the mandible tips, then tossed back with an upward jerk of the head. Despite its size, the bill can be used with some precision, as when birds pass food from one to another. Skutch (1972) suggests for R. swainsonii that the size and colours of the bill have an intimidatory value, useful when nest robbing. He also notes that toucans are unable to turn the head back in flight for self defence, a fact which small birds take advantage of when mobbing them. Skutch mentions two FEEDING APPARATUS IN CORACIIFORMES & PICIFORMES 347 Swainson's Toucans with severely broken bills which nevertheless survived in the wild for at least 1\ years and 7 weeks respectively. Feeding behaviour is essentially similar in the smaller toucans such as Aulacorhynchus spp., and Skutch (1967) has provided a description of nest excavation by A. caeruleogularis. Rotten wood was the chosen substrate, and was excavated by pecking and hammering, blows being often (perhaps always) delivered with the jaw tips slightly parted. Only the foreparts were moved when delivering blows, not the whole body. The birds also seemed to bite away pieces of rotten wood. Large billfuls of excavated material were removed at intervals and carried some distance away as in many barbets. No unusual function in feeding has been ascribed to the exceptionally well developed tongue, although several observers, the writer included, have made a point of looking out for some special form of tongue action. PICIDAE The Jynginae (wrynecks) do not excavate, either for food or to make nest holes, but rely entirely on the long protrusible tongue to take food, either from the surface, or from tunnels and crevices. Ants make up a large proportion of the diet of small insects. Though neck movements while feeding appear unremarkable, both threat and courtship displays involve extraordinary contortions of the neck. Piculets (Picumninae) principally exploit fine twigs, vines, leaf petioles, etc., hammering them in typical woodpecker fashion to gain access to ants and grubs. The tail is not used to prop the body while hammering as in woodpeckers. A good general account of foraging by Picumnus olivaceus is given by Skutch (1969). He also quotes Miller (1947) who notes that in Columbia this species shows great versatility, including nuthatch like behaviour in which the bird moves down tree trunks head first. Sasia spp. show a fondness for fallen tree trunks and logs, as well as for fine twigs and bamboo. Verreauxia evidently forages in a gener- ally similar way to other piculets, but is said to prefer grubs to ants (Bannerman, 1951). All piculets excavate their own nest holes either in trees or bamboo. Woodpeckers (Picinae) make up the largest of the three subfamilies. They subsist typically on a diet of wood and bark-boring insects, but some, especially Melanerpes lewis and M. formicivorus, take aerial insects, and other foods include acorns and other nuts, and even cambium and sap, obtained by boring holes through bark (Sphyrapicus spp.). The bill is used not only for feeding, but also for nest excavation and to produce 'drumming' — an instru- mental sound with territorial significance. An excellent account of both these activities is given by Sielmann (1958). The feeding process also involves sophisticated techniques using the highly modified and protrusible tongue. A detailed account of woodpecker feeding methods has been given by Spring (1965); an important earlier study was made by Burt (1930). Spring concentrated on three species — Sphyrapicus varius, Dendrocopos villosus and Picoides arcticus. He described the action of hammering carefully, and used cine photography to assess the contribution of body and neck to the blows. The contribution of the neck was greatest in Sphyrapicus, which holds its body closer to the trunk than the other two species; it consequently strikes less powerful blows, but since most of its food is obtained near the surface, deep excavation is not needed. Spring notes that leg and foot adaptations enabling the body to be held well off the trunk impose a penalty of decreased climbing efficiency in the species he studied, but he could not demon- strate any difference in blow delivery between D. villosus and D. pubescens, or between P. arcticus and P. tridactylus, despite some anatomical differences. Personal observations of a wide range of woodpeckers seem to indicate that a substantial part of the force of hammering is provided by body movement in a majority of species. An interesting variant of blow deliv- ery is noted for Campephilus principalis by Allen & Kellogg (1937) and Tanner (1942). This species forages largely by knocking off flakes of bark with sideways blows to reveal insects underneath. 348 P. J. K. BURTON Several species, notably the Acorn woodpecker Melanerpesformicivorus and Dendrocopos spp. open nuts, acorns and fir cones by wedging them into cracks or holes in bark, then at- tacking them with the bill. An interesting account of the development of this behaviour in a captive D. major is given by Sielmann (1958). Nuts were carried to the chosen 'anvil' in the bill; if the anvil was inadequate, the nut was held under the breast feathers while the bill was used to chisel out more wood. Pine cones were set upright, and the scales broken off one by one, while seeds were picked out with bill or tongue. Sielmann (1958) describes the use made of the tongue by several European species as revealed by sophisticated cinematography. A feature of particular interest emerging from these observations is the brief duration of most tongue movements. Although these can be forceful enough to impale beetle larvae, or to create a 'drumming' noise on the side of a box housing a captive Dryocopus martins, tongue protrusion is not long sustained, but con- sists instead of a series of very rapidly repeated darting extensions. Sielmann suggests that this is due not to any inability to maintain the tongue in a state of protrusion, but to the necessity for keeping it adequately coated with the sticky secretion from the highly developed salivary glands. FEEDING APPARATUS IN CORACIIFORMES & PICIFORMES 349 PART 2 Anatomy Osteology and arthology of the skull Literature concerning functional aspects of the skull structure among birds in general is dis- cussed more fully in Part 3 (pp. 399-402), but some important papers may be briefly noted here. Barnikol (1952) dealt with overall form of the skull, Bock (1964) with cranial kinesis and Yudin (1961) with movements of the mandibular rami. Kinetic stops were reviewed by Fisher (1955) and secondary articulation of the mandible by Bock (1960a). The lacrimal- ectethmoid complex was thoroughly surveyed by Cracraft (1968), who should be referred to for details of the complex and its modifications in the various families of Coraciiformes and Piciformes. Studies on probing adaptations in the Charadrii (Burton 1974a) and gaping adaptations in the Icteridae (Beecher 1953a) and Callaeidae (Burton \914b) are of special relevance to the Upupidae and Phoeniculidae. Descriptive and comparative cranial osteology of birds belonging to the Coraciiform- Piciform assemblage has been well covered in the literature, although functional studies are mainly confined to the Picidae. A good summary of nineteenth century literature and views is given by Beddard (1898). Many of the papers from this era dwelt largely on palatal struc- ture, pursuing the ideas first introduced by Huxley (1867). Particular attention was given to the woodpecker palate, with discussions by Garrod (1872), Parker (1875) and Shufeldt (1891). Other important studies include those of Murie (1872a & b, 1873) on various Coraciiform families, and of Shufeldt (1884) on Ceryle alcyon. Publications on various aspects of osteology in the two orders have continued to appear at intervals in the present century, and most of these are mentioned in the reviews of literature included by Sibley and Ahlquist (1972). Important general studies include those by Lowe (1946, 1948) and Verheyen (19550 & b), while J. Steinbacher (1937) investigated the Galbulidae and Bucconidae, and interest in the Picidae has continued with papers by Burt (1930), Beecher (19536), Spring (1965) and Zusi & Marshall (1970). The study of rollers, ground-rollers and cuckoo- rollers by Cracraft (1971) included a thorough examination of cranial osteology. A detailed and important work on the skull of a single species of hornbill is that by Manger Cats-Kuenen (1961) dealing with Rhinoplax vigil. Ligaments associated with jaw articulations and other regions of the skull have been surveyed by Lebedinsky (1921) and Bock (1964). Those occurring in the Coraciiformes and Piciformes are listed below; they are present throughout, unless otherwise stated, in which case their distribution is given in the family by family summary of skull morphology which follows. 1. The postorbital ligament runs from the tip of the postorbital process to the external process of the mandible, slightly anterior to its articulation with the quadrate. It is absent or reduced in a number of the families studied here. 2. The occipitomandibular ligament is situated at the dorso-medial edge of M. depressor mandibulae, running from the exoccipital process to the posterior face of the internal process of the mandible. 3. The external jugomandibular ligament is a short band connecting the posterolateral face of the jugal bar with the external process of the mandible. It passes superficial to the postorbital and internal jugomandibular ligaments. It is absent in the Picinae, and very weak in the Jynginae and Picumninae. 4. The internal jugomandibular ligament takes the form of a strong band running around the posterior edge of the interface between quadrate and mandible at their articulation. At its lateral end it is attached to the jugal bar, anterior to the external jugomandibular, and medially to the posterodorsal rim of the mandible. Sesamoid bones are commonly found within the ligament (see e.g. Burton 1973). 5. The quadratomandibular ligament is apparently unique to the Bucerotidae (q.v.). It 350 P. J. K. BURTON •§s " 2 u" S Jo ^» ^ s: JD £ O o o> "§ s ^£ > -2 el 3 ^ on DC - FEEDING APPARATUS IN CORACIIFORMES & PICIFORMES 353 medial edge. Medial condyle a narrow oval in section, and deep, extending well ventral to the pterygoid articulation, and somewhat produced anteriorly to give a hook shaped profile when viewed laterally. No retroarticular process. Medial brace present, postorbital ligament rather weak. LEPTOSOMATIDAE. The Cuckoo-roller's bill is of medium length and fairly deep, resembling those of typical Coraciidae except in the anterior placement of the nostril. Strongly desmog- nathous, with wide palatines, and the quadrate with a short, fairly broad postorbital process. Medial condyle oval, rather narrow and deep. Orbits relatively very large. The postorbital process is long, reaching the jugal bar, but without an anterior spur for the suborbital process. No retro-articular process. Postorbital ligament stout; medial brace absent. For a comparison of skull morphology in this and the preceding family see Cracraft (1971). CORACIIDAE. In Coracias and the Brachypteraciinae the bill is typically of moderate length, fairly deep, and with gently decurved tomia. In Eurystomus it is snorter, but greatly widened. Completely desmognathous. The quadrate has a short, wide orbital process, and a deep, oval medial condyle. The cranium is generally robust and heavily ossified, with a long, stout post- orbital process, usually reaching the jugal bar, but with a small protuberance anteriorly for attachment of the suborbital ligament. The whole skull is markedly broadened in Eury- stomus, and its palatines are conspicuously expanded. No retroarticular process. Medial brace well developed, postorbital ligament broad. UPUPIDAE. Bill long, slender, pointed and decurved, with a heavily developed rhamphotheca totally occluding the lumen of the bill for most of its length. Weakly desmognathous, with a narrow zone of maxillopalatine fusion. The quadrate has a short but rather narrow orbital process somewhat expanded at its medial tip, and the thin medial condyle barely projects below the pterygoid articulation. Lateral and posterior condyles are fused into a single cres- centic structure. The postorbital process is weak, and lies very close to the zygomatic process. Retroarticular process long, extending well behind the quadrate. Vestigial basipterygoid processes sometimes present. The postorbital ligament is of moderate breadth and the occipito-mandibular ligament is conspicuously well developed. The medial brace is poorly developed, and oriented more nearly parallel to the long axis of the skull than in other groups. PHOENICULIDAE. Bill varying from medium length, conical and slightly decurved (P. aterri- mus) through long and moderately decurved (P. purpureus, P. bollei) to long, slender and strongly sickle shaped (Rhinopomastus spp.); in all cases, the ramphotheca is strongly developed, occluding the lumen as in Upupa. Strongly desmognathous in Phoeniculus, less so in Rhinopomastus. Quadrate orbital process and condyles as in Upupidae, though lateral and posterior condyles are somewhat more distinct. Postorbital and zygomatic processes very close together, the former much reduced. Retroarticular process very long. Postorbital ligament moderately developed, very strong occipito-mandibular ligament, but medial brace absent. BUCEROTIDAE. Despite the great variability of the casque in hornbills, the bill itself is rather similar throughout the family moderately long, deep and laterally compressed, with a more or less decurved profile, and tapering to a point. In Rhinoplax, the bill is relatively much shorter and more conical than in other hornbills. The skull is robust and heavily ossified, with a doubly desmognathous palate — that is, with maxillopalatines and anterior palatines solidly welded to form a complete bony roof to the proximal part of the upper jaw. The quadrate has a fairly short and rather pointed orbital process. The medial condyle is rounded, but not especially prominent, the lateral crescentic, but clearly distinct from the deeper posterior condyle. Postorbital process well developed. Vestigial basipterygoid processes in some (Starck, 1 940). Retroarticular process varying from moderate to barely indicated. Postorbital ligament extremely broad, occipito-mandibular ligament strongly developed, medial brace absent. In Bucorvus, there appears to be some kinetic coupling via the structure 354 P. J. K. BURTON 5 55 2 3 "^ V s: a • <*-".£ it U ^ S ^ S Cx O 2 1- % § S J s ^ OJD O JZ FEEDING APPARATUS IN CORACIIFORMES & PICIFORMES 355 of the quadrate/mandible articulation; in a dried skull, the lower jaw stays firmly in place once fitted, and in life this would seem to make it impossible for one jaw to move indepen- dently of the other. The quadratomandibular ligament appears to be unique to this family, although possibly analogous to one of the snap closing jaw ligaments described by Bock & Morony (1972) in the Tyrannidae. GALBULIDAE. Bill typically long, slender, sharply pointed and straight. Much deeper and heavier in Galbalcyrhynchus, with a strong dorsal ridge; shorter, deep and somewhat decurved in Jacamerops. Fully desmognathous, tending to doubly desmognathous in Galbalcyrhynchus. Posterior cranium extends well posterior to the foramen magnum, and dorsal to the orbit, producing a sloping profile as in the Meropidae, though less pronounced. This feature is less obvious in the deep billed Galbalcyrhynchus, though the rearward projec- tion of the occipital region is still marked. Quadrate with a short and very broad orbital process which has a long medial edge. Medial condyle narrow and deep, extending well ven- tral to the articulation with the pterygoid. Postorbital process long and stout, nearly reaching the jugal bar. No retroarticular process. Postorbital ligament fairly strong. Medial brace absent. BUCCONIDAE. Bill very variable in form; heavy, shrike like and slightly hooked (e.g. Nothar- cus); moderately long and slightly decurved (Monasa); long and rather flattened (Nonnula); short, broad and slightly decurved (Chelidopterd). In hook billed types, the hook ends bluntly, with a U-shaped cross section, which may be more or less developed into a notched tip; this is especially well shown in Notharcus. The lower jaws of hooked billed puffbirds are often emarginated at the tip. Strongly desmognathous. Quadrate with a very short but broad orbital process, its wide medial edge nearly in line with the posterior edge, giving a rather pointed appearance. The medial condyle is extremely deep, ending in a rounded tubercle, the lateral and posterior condyles merged. The postorbital process is long, curved and broad, often meeting the jugal bar. No retroarticular process. Postorbital ligament strong. Medial brace absent. CAPITONIDAE. Bills usually massive, of medium length, but wide and deep, with a strongly curved culmen. Pogoniulus spp. have bills of more moderate proportions. Notches or serrations on the tomium of the upper jaw are seen in Lybius spp., while in Semnornis there is a remarkable arrangement in which the hooked tip of the upper jaw fits into the notched tip of the lower when the bill is closed. Gymnobucco and Stactolaema spp. have a sharp raised ridge on the base of the culmen in the males. Most barbets have divided palates, though the maxillopalatines have a narrow region of fusion in some Pogoniulus spp., Capita, Eubucco and Semnornis. The vomer has a broad, bifurcated tip resembling that of passerines; hence, the palate may be termed aegithognathous (Beddard, 1 898). The quadrate has a long, slender orbital process. The medial condyle does not project far below the pterygoid articulation, but in some species of Megalaima it has a sharply defined lateral edge which engages with the lower jaw to give some measure of kinetic coupling. A postorbital ligament is, however, absent in all barbets examined. The postorbital process is long and well defined, and lateral and posterior quadrate condyles are more or less merged into a single S-shaped ridge. The medial brace is absent. INDICATORIDAE. The bill is of medium length and fairly robust except in Prodotiscus, where it is weak and somewhat decurved. The skull is generally similar to that of the Capitonidae, though more lightly constructed and less heavily ossified than in most barbets. The maxillo- palatines remain separate; the quadrate has a slender orbital process, a shallow medial condyle, and its lateral and posterior condyles are merged as in barbets. The junction of pterygoid and palatine has a distinctive form (shared with the Picidae), in which the pterygoid foot is elongated, overlapping the posterior end of the palatine dorsally and medially for a considerable distance. The postorbital ligament is weak, but distinct, and the postorbital process well developed. There is no medial brace. 356 P. J. K. BURTON K B ^ >; If a a is ? * ft> C c o J= FEEDING APPARATUS IN CORACIIFORMES & PICIFORMES 357 .§•1 i s: O s^ &c o o s: 5 « ^ .s _ S- ^ F3 fe S m <3 <-> "2 73 i/^ &^ ^ ^ cd •V ^3 _J 358 P. J. K. BURTON a.s 3s s: c y ^ * rf* % s « '> S3 c ill C ** i- WD ^T3 Q ^ FEEDING APPARATUS IN CORACIIFORMES & PICIFORMES 359 RAMPHASTIDAE. The use of R. toco as an advertising symbol has made the bills of toucans familiar even to non-ornithologists. All are basically similar — long and deep, with a strongly decurved tip. There is less lateral compression than in the superficially somewhat similar bills of the Bucerotidae. The upper tomium is strongly serrated, the lower more weakly; the serrations slope more steeply anteriorly than posteriorly. In Andigena laminirostris the tip of the lower mandible ends with a U-shaped cross section which slightly overlaps the curved tip of the upper on each side. The palate is thoroughly desmognathous, with the maxillopalatines fused along a consider- able length of the midline, and the anterior part of the palatines and vomer fused onto the continuous bony sheet lining the upper jaw. As in the Capitonidae the orbital process of the quadrate is long and slender; the medial condyle fairly shallow, and the posterior and lateral condyles merged into a single curved ridge. As in the barbets too, the postorbital ligament is absent, although there is a strong postorbital process. The medial brace is also absent. PICIDAE. Extensive differences exist between Jynx and the remainder of the family. The bills of typical woodpeckers and piculets are short to moderately long, straight, and tapering evenly to a blunt point; the rhamphotheca is hard and well developed. The bills of wrynecks resemble those of woodpeckers in their straight, pointed form, but are proportionately much smaller and more lightly constructed, with large bony nares, and a minimum of rhampho- thecal reinforcement. Other features of skull structure also show considerable disparity between Jynx on the one hand and the woodpeckers and piculets on the other. Their principal common features are in palatal structure; all have reduced maxillopalatines and reduced or absent vomer. (For discussion of the latter see Beddard, 1898: 186-187). As in the Indicatoridae the pterygoid foot overlaps the palatine extensively at its posterior end. However, the following specialized features are limited to the Picinae and Picumninae: a pronounced, or even overlapping, fore- head at the fronto-nasal hinge; a spur on the pterygoid from which arises part of M. protrac- tor; a quadrato-jugal articulation which is extended posteriorly behind the lateral condyle, with consequent enlargement of the quadrate body, especially in the Picinae; and, in the Picinae only, a strong forwardly directed process extending from the posterior lateral rim of the auditory capsule, with ligamentous attachment to the external process of the mandible. In some species this process apparently articulates with the posterior surface of the lateral process of the quadrate, which is markedly broadened in woodpeckers. The orbital process of the quadrate is long and slender in all three sub-families, but is strongly curved in the Jynginae. The postorbital process and ligament are weak, though distinct, in Jynx, but both are quite strongly developed in woodpeckers and piculets. The medial brace is absent in all. Jaw muscles The basis for most detailed studies of avian jaw muscles was laid down by Lakjer (1926). His system of nomenclature is adhered to as far as possible, but with modifications stemming from more recent work, notably that of Starck & Barnikol (1954) and Richards & Bock (1973). As noted in the introduction, N.A.A. terms (Baumel et al, 1979) are given in head- ings where they differ from terms used here. The descriptions given here are effectively the first for most families of Coraciiformes and Piciformes with the exception of those for the Bucerotidae (Starck, 1940) and Picidae (Beecher, 19536). M. adductor mandibulae externus (abb: M.add.mand.ext.) This muscle is of complex structure, and several subdivisions can be recognized. Various systems of terminology and criteria have been employed by different authors for naming and distinguishing these subdivisions. The system followed here is that used by Richards & Bock 360 P. J. K. BURTON (1973). Starck & Barnikol (1954) demonstrated the presence of three major aponeuroses in the muscle in birds of several orders. These aponeuroses are recognizable in the families considered here, and the numbering used corresponds to that of Stark & Barnikol. (See also Table 1). M. adductor mandibulae externus rostralis (M.a.m.e.rost.) is the most dorsal portion of the muscle. It can be further divided into three parts: (a) M. adductor mandibulae externus rostralis temporalis (M.a.m.e.rost. temp.) has a fleshy origin from the temporal fossa, from the base of the postorbital process, and from the dorsal surface of Aponeurosis 2. It inserts on the coronoid process of the mandible via a flattened tendon (Aponeurosis 1) which begins within the temporal region as the raphe of a bipinnate fibre arrangement. (b) M. adductor mandibulae externus rostralis lateralis (M.a.m.e.rost.lat.). Origin is from the lateral edge of the zygomatic process and the dorso-lateral surface of Aponeurosis 2. Insertion is made either through the lateral part of Aponeurosis 1 , or through an aponeurosis covering the antero- lateral surface of the muscle, and continuous medially with Aponeurosis 1. Table 1. Divisions of M. adductor mandibulae externus used in the present paper, and their equivalents in some previous studies. Further tables of synonymy are given by Starck and Barnikol (1954). See also Baumel et al., 1979. Burton, 1974 (Charadrii) Starck & Barnikol 1954 (several orders) Lakjer ( 1 926); Goodman & Fisher, 1962 (Anatidae) M.a.m.e. rostralis Part M + Part A dorsal Aponeurosis 1 portion Part of superficialis M.a.m.e. rostralis temporalis Part M, external temporal Aponeurosis 1 portion, external temporal superficialis, la portion (Levatoranguli oris). M.a.m.e. rostralis lateralis Part A, dorsal Aponeurosis 1 portion not present M.a.m.e. rostralis medialis Part M, rostral Aponeurosis 1 portion medialis M.a.m.e. postorbital lobe, dorsal part Not named; absent in most species Aponeurosis 1 portion superficialis, Ic portion (retractor anguli oris) M.a.m.e postorbital lobe, ventral part. Not named; absent in most species. Aponeurosis 1 portion superficialis, Ib portion M.a.m.e. ventralis Part A, ventral Aponeurosis 2 portion Included with M.a.m.e. caudalis PartB M.a.m.e. caudilis, Part B, lateral lateral expansion expansion Aponeurosis 3 portion Aponeurosis 3 portion superficialis la portion; rost. temp. + vent, treated as a bipinnate muscle. M. adductor mandibulae posterior profundus "In discussions throughout this paper I shall use these terms in their German form, to indicate their origin clearly, to avoid possible ambiguities inherent in their English translations, and because they are less clumsy. FEEDING APPARATUS IN CORACIIFORMES & PICIFORMES 361 amerp amert pr ptdm pss amev ptv|e 10mm 5mm Fig. 7 Chloroceryle americana, jaw musculature: A, lateral view; B, dorsal view. (c) M. adductor mandibulae externus rostralis medialis (M.a.m.e.rost.med.) originates from the lateral edge of the posterior will of the orbit and the medial edge of the temporal fossa. The origin is fleshy, and may also involve an aponeurosis on the medial surface. The insertion is on the anterior coronoid process, posterior and medial to that of M. adductor mandibulae externus rostralis temporalis, usually via a lateral aponeurosis which is continuous with Aponeurosis 1 . M. adductor mandibulae externus ventralis. (M.a.m.e.vent.) Origin is from the medial and ventral surfaces of Aponeurosis 2, which begins as a strong flat tendon arising from the zygo- matic process, and fans out anteriorly over the lateral surface of the muscle. The main region of insertion is a wide area on the lateral side of the mandible; fibres arising more posteriorly run forwards and downwards to insert on the dorsal surface of Aponeurosis 3 and could equally be considered with M.a.m.e. caudalis. M. adductor mandibulae externus caudalis. (M.a.m.e.caud.) Origin is fleshy from the otic process of the quadrate, and insertion is usually made principally via Aponeurosis 3 which lies on the dorso-lateral side of the muscle. The insertion is on the dorsal edge of the mandible, behind and below that of Aponeurosis 1. In some species, there is an additional aponeurosis of insertion (Aponeurosis 3a) on the medio- ventral surface of the muscle; it may be separate from Aponeurosis 3 at the insertion, or may fuse with it. There is frequently an additional aponeurosis (Aponeurosis 4) in the muscle, which arises on the otic process, and functions as the raphe of a bipinnate fibre arrangement. In the Picidae, this aponeurosis, and the part of the muscle medial and ventral to it, are expanded anteriorly to insert widely on the lateral surface of the mandible. This will be further explained under the family heading. CORACIIFORMES The main general points of difference from the Piciformes concern the lateral region of the muscle. M.a.m.e.rost.lat. is never extensively developed, and in many cases, scarcely to be distinguished. M.a.m.e.vent. fans out further anterior, so that it does not usually conceal M.a.m.e.caud., as in many Piciformes. M.a.m.e.caud. itself is narrow, but usually bipinnate. Aponeurosis 3 may join Aponeurosis 1 medially, and the dorsal edge of Aponeurosis 4 commonly joins Aponeurosis 2 at its medial edge. ALCEDINIDAE. Alcedo atthis. M.a.m.e. rost.temp. is highly developed, the right and left muscles meeting in the midline, at the back of the skull; Aponeurosis 1 extends far back in the temporal region as the raphe of a strongly marked bipinnate system. There is a well developed distinct slip (referred to here as the postorbital lobe) also bipinnate, arising at the 362 amepl P. J. K. BURTON amert ptdm ptvle ptmxp B pss Fig. 8 Dacelo gaudichaud, jaw musculature: A, lateral view; B, dorsal view. lateral edge of the posterior orbital wall, its raphe arising from a blunt process corresponding in position with the postorbital process, the postorbital ligament being absent. The dorso- medial surface of the postorbital lobe is covered by an aponeurosis which joins Aponeurosis 1 anteriorly. M.a.m.e.rost.lat. is absent. M.a.m.e.caud. is small and has only one aponeurosis (Aponeurosis 3), which is vertically oriented. It inserts through this and a small fleshy attachment. Other members of the family show an essentially similar structure, including a slip arising from the postorbital process, although a postorbital ligament is present in the Cerylinae and Daceloninae. In these families also, a distinct, though small M.a.m.e.rost.lat. can be dis- tinguished, fused posteriorly with M.a.m.e.rost.temp. It is best developed in Dacelo, in which it is aponeurotic on its lateral surface, and extends a little way over the dorsal part of M.a.m.e.vent. In the Daceloninae, M.a.m.e.caud. is better developed; Aponeurosis 4 is present and an Aponeurosis 3a was discernible in Halcyon chloris. TODIDAE. Todus viridis. The origin of M.a.m.e.rost.temp. is much reduced in extent, but retains clear bipinnate structure. M.a.m.e.rost.lat. is more substantial than in most families of the order, and its aponeurosis of insertion is, in effect, the lateral part of Aponeurosis 1 which has no lateral expansion. Aponeurosis 2 is weakly developed. M.a.m.e.caud. is thin and of simple structure, with only one aponeurosis (Aponeurosis 3). Ventrally and medially there is some fusion with M. adductor posterior. MOMOTIDAE. Momotus momota. M.a.m.e.rost.temp. extends back over about half the cranium posterior to the orbit, but is unusually bulky in the orbital region, bipinnate struc- ture continuing well forward into this area; there is also a small but distinct postorbital lobe. M.a.m.e.rost.lat. is also substantial, and has a strong lateral aponeurosis of insertion. M.a.m.e.caud. is much reduced, and only Aponeurosis 3 can be discerned; this is broad and vertically oriented, its medial edge joining Aponeurosis 1 dorsally. There are no fibres medial to Aponeurosis 3. Electron platyrhynchum and Baryphthengus ruficapillus are similar, but lack a distinct postorbital lobe. MEROPIDAE. Nyctiornis amicta. M.a.m.e.rost.temp. has a long but narrow area of attachment, extending almost to the midline of the cranium posteriorly. It is strongly bipinnate in struc- ture. Aponeurosis 1 is sharply downturned where it reaches the orbit. M.a.m.e.rost.lat. is vestigial, and M.a.m.e.vent. is thus almost completely exposed laterally. M.a.m.e.caud. is bulky and bipinnate, with three aponeuroses; 3 and 3a which insert separately, though close together, and Aponeurosis 4 arises from the origin, its dorsal edge joining the medial edge of Aponeurosis 2. FEEDING APPARATUS IN CORACIIFORMES & PICIFORMES ptdm pr amert amerl amev 5mm Fig. 9 Todus todus, jaw musculature: A, lateral view: B, dorsal view. Other bee-eaters appear essentially similar. M.a.m.e.rost.temp. is even narrower in other genera, and the downward bend in Aponeurosis 1 at the border of the orbit is absent or un- noticeable. The insertion of M.a.m.e.ext.vent. is generally rather small in extent. LEPTOSOMATIDAE. M.a.m.e.rost.temp. has a very limited temporal origin, in which no obvious bipinnate structure can be distinguished. Aponeurosis 1 is partially overlapped in the orbit by fibres of M.a.m.e.rost.med. M.a.m.e.rost.lat. is absent. M.a.m.e.vent. possesses the unusual feature of a small aponeurosis on the lateral side arising from the postorbital ligament (which is extremely broad). This joins Aponeurosis 2 anteriorly; it is largely con- cealed by a lobe overlapping it laterally from below. This lobe consists of fibres inserting on the dorsal surface of Aponeurosis 3, some arising from Aponeurosis 2, and some on the otic process, and hence part of M.a.m.e.caud. The more medial fibres of M.a.m.e.caud. insert along the medial edge of Aponeurosis 3, which is its only aponeurosis in this family. CORACIIDAE. Coracias benghalensis. M.a.m.e.rost.temp. covers about three-fifths of the dis- tance between the orbit and the posterior midline of the skull. Its bipinnate structure is not well marked. M.a.m.e.rost.lat. is absent. M.a.m.e.vent. has a rather small area of insertion, covering only about half the depth of the mandible. M.a.m.e.caud. has a double aponeurosis of insertion (Aponeuroses 3 laterally and 3a medially), the two fusing at the point of attachment. Its fibres are arranged bipinnately about Aponeurosis 4. In Eurystomus M.a.m.e.rost.temp. is more extensive, nearly reaching the skull midline; nevertheless, bipinnate structure is not well marked. The insertion of M.a.m.e.vent. is rather larger, and there is a distinct M.a.m.e.rost.lat.; its fibres are short, arising along the dorsal edge of Aponeurosis 2, and running forwards and up at a steep angle to Aponeurosis 1. Brae hyp teracias is similar to Coracias. UPUPIDAE. Upupa epops. M.a.m.e.rost.temp. consists of two portions. The main portion, occupying the temporal fossa, is much reduced by comparison with the majority of families included in this study, the temporal fossa itself being extremely small. An additional portion, corresponding to the postorbital lobe described in the Alcedinidae, arises from the base of the postorbital process, and a concave region on the rim of the orbit just dorsal to this. Both the main portion and the dorsal slip are bipinnate. The strong raphe of the dorsal slip arises from the tip of the postorbital process, anterior to the postorbital ligament, while the raphe of the main portion is, of course, Aponeurosis 1. M.a.m.e.rost.med. is narrow and M.a.m.e.rost.lat. is only moderately developed. M.a.m.e.vent. has the normal wide insertion on the mandible. M.a.m.e.caud. is bipinnate, and has lateral and medial aponeuroses of insertion (Aponeuroses 3 and 3a). 364 P. J. K. BURTON amert amer B 5mm 5mm Fig. 10 Momotus momota, jaw musculature: A, lateral view; B, dorsal view. PHOENICULIDAE. Phoeniculus purpureus. As in Upupa, the postorbital process is relatively close to the zygomatic process, so that the short temporal portion of M.a.m.e.rost.temp. is very narrow. However, the postorbital lobe is bulkier and more complex in structure, consist- ing of two parts. The more dorsal of these consists of fibres originating dorsal and anterior to the postorbital process, and these insert via an aponeurosis (la) situated on the dorsal and medial surfaces, on the dorsal edge of the mandible. The more ventral part consists of fibres originating from the ventral surface of an aponeurosis (Ib) attached near the base of the postorbital process. This part fans out across the dorsal lateral surface of the mandible, and aponeurosis Ib covers part of this broad fleshy insertion laterally. Aponeurosis 1 itself is broad, and is extended laterally to provide insertion also for the poorly developed M.a.m.e.rost.lat. M.a.m.e.vent. is much reduced by comparison with most birds. Apo- neurosis 2 is weak, and the fleshy insertion on the mandible is a narrow one, between the broad fleshy insertion of the postorbital lobe (ventral portion), and the anterior part of M.a.m.e.caud. This latter division of the muscle inserts by two strong aponeuroses (1 and 3a), the dorsal and lateral one (3) attaching just onto the lateral surface of the mandible. Between them lies a raphe (Aponeurosis 4) which spreads out onto the lateral ventral surface of the mandible, the fibres medial to it making a broad fleshy insertion. P. aterrimus is generally similar to P. purpureus, but the following differences may be noted: The dorsal part of the postorbital lobe does not insert only by Aponeurosis la, but has in addition a narrow band of fibres extending and inserting along the dorsal lateral edge of the mandible. Aponeuroses 1 and 3 fuse at the insertion; M.a.m.e.vent. is still more reduced. The lateral lobe covered by Aponeurosis 4 is very large, but Aponeurosis 3a is rather weak. Rhinopomastus resembles P. purpureus, but M.a.m.e.vent. is even more reduced, and the temporal origin is almost absent. BUCEROTIDAE. Tockus erythrorhynchus. M.a.m.e.rost.temp. extends back about two thirds of the distance from the orbit to the midline of the cranium. The insertion of M.a.m.ext. on the mandible is narrow, and limited to its dorsal edge. Aponeuroses 1 and 3 are very strong and prominent, and the insertions of both are clearly visible in lateral view. There is no distinct M.a.m.e.rost.lat. A band of fibres arising at the tip of the postorbital process travels a- short way to insert at a large angle on the lateral edge of Aponeurosis 1. M.a.m.e.vent. is much reduced, amounting to only a narrow band of fibres inserting between Aponeuroses 1 and 3; Aponeurosis 2 is moderately developed. M.a.m.e.caud. is strongly bipinnate, and the dorsal edge of its raphe (Aponeurosis 4) joins Aponeurosis 2 medially. There is a strong Aponeurosis 3a inserting medial to Aponeurosis 3, and an additional aponeurosis on the medial surface of M.a.m.e.caud. FEEDING APPARATUS IN CORACIIFORMES & PICIFORMES 365 amert amec / / ^¥!%/ \ IKP pss ptdl amev B 5mm 5mm Fig. 11 Merops superciliosus, jaw musculature: A, lateral view; B, dorsal view. PICIFORMES A feature of the muscle in many Piciformes is the extensive M.a.m.e.lat, which forms a flattened sheet covering much of M.a.m.e.vent., though usually easily separable from it. The anterior part is usually strongly aponeurotic on its lateral surface. M.a.m.e.vent. fans out relatively close to the origin; it frequently conceals the lateral surface of M.a.m.e.caud. Aponeurosis 3 is usually broad, and M.a.m.e.caud. well developed. A raphe (Aponeurosis 4) is often present, but a medial aponeurosis of insertion (Aponeurosis 3a) is lacking in most groups. GALBULIDAE. Galbula dea. M.a.m.e.rost.temp. has a fairly long area of origin covering about three fifths of the space between the orbit and the posterior midline of the skull. Its structure is clearly bipinnate. M.a.m.e.rost.lat. is absent. Aponeurosis 2 is broad, and the fleshy inser- tion of M.a.m.e.vent. is extensive, but fans out some distance from the origin. M.a.m.e.caud. is bipinnate; its medial fibres insert on a weak Aponeurosis 3a, which dorsally joins the medial edge of Aponeurosis 1 near the insertion. Aponeurosis 3 itself is narrow. Aponeurosis 4 is fairly well developed, joining Aponeurosis 2 dorsally. These features — which differ in several respects from the general characters of the Pici- formes noted above — are shown with little variation by the various other jacamars examined. In Jacamerops and Galbalcyrhynchus M.a.m.e.temp. is more extensive, reaching close to the posterior midline of the skull, and Aponeurosis 2 is extensive; In Brachygalba, M.a.m.e.temp. is somewhat reduced, covering only about half the posterior cranium. BUCCONIDAE. Chelidoptera tenebrosa. Generally similar to the Galbulidae in most respects. M.a.m.e.rost.temp. extends back about two-thirds of the distance from the orbit to the skull midline. M.a.m.e.vent. is noticeably bulky, but Aponeurosis 2 is not extensive. Other puffbirds also resemble the Galbulidae (and differ from other Piciformes) in general features. M.a.m.e.rost.temp. reaches the skull midline in Nystalus spp. and Notharcus macrorhynchos, and nearly as far in Hypnelos and Notharcus tectus. It is rather short (com- parable with Chelidoptera} in Nonnula. M.a.m.e.vent. is notably bulky also in Nystalus and Notharcus, and distinctly elongated in Monasa spp. M.a.m.e.rost.med. is unusually broad in Nystalus spp. CAPITONIDAE. Megalaima haemacephala. M.a.m.e.rost.temp. has an origin extending nearly 366 P. J. K. BURTON ptdmp amert amev ptdl ptvle B pss ptdl amerl Fig. 12 Leptosomus discolor, jaw musculature: A, lateral view; B, dorsal view. back to the midline, and is clearly bipinnate. M.a.m.e.rost.lat. is wide and flattened, covering more than half the area of M.a.m.e.vent. as seen laterally. Its surface is largely covered anteriorly by a strong aponeurosis continuous dorsally with Aponeurosis 1. M.a.m.e.vent. fans out close to its origin, concealing M.a.m.e.caud. from view laterally. Aponeurosis 3 is broad and vertically oriented, M.a.m.e.caud. forming a rather thin sheet. There is a raphe (Aponeurosis 4), its dorsal edge joining the medial edge of Aponeurosis 2. The muscle is similar in other barbets examined including Pogoniulus. In Semnornis, a fleshy slip from M.a.m.e.caud., arising lateral to Aponeurosis 4 extends a short way onto the lateral surface of the mandible. INDICATORIDAE. Indicator indicator. M.a.m.e.rost.temp. extends only about halfway to the skull midline. M.a.m.e.rost.med. is bulky, and M.a.m.e.rost.lat. is wide and flattened with an extensive lateral aponeurosis as in barbets. M.a.m.e.vent. has an elongated area of inser- tion. M.a.m.e.caud. is visible laterally; Aponeurosis 3 is short, and there is a weak bipinnate arrangement, although a clear Aponeurosis 4 cannot be detected. /. maculatus, I. minor and Melichneutes are similar. RAMPHASTIDAE. Selenidera maculirostris. M.a.m.e.rost.temp. spans about three-quarters of the space between the posterior rim of the orbit and the posterior midline of the skull. Aponeurosis 1 is sharply angled down at the edge of the orbit. M.a.m.e.rost.lat. forms a broad flattened sheet as in barbets, but with a more extensive lateral aponeurosis; this covers much of the lateral surface of M.a.m.e.vent. M.a.m.e.caud is similarly flattened, with a broad, vertically oriented Aponeurosis 3; it is bipinnate, with a distinct Aponeurosis 4, but no Aponeurosis 3a. Other toucans examined are closely similar. PICIDAE. Jynx torquilla. M.a.m.e.rost.temp. has only a very small area of origin. In the orbital region, Aponeurosis 1 is almost buried by dorsal fibres. M.a.m.e.rost.lat. and M.a.m.e.vent. are similar in their extent and form to those of the Capitonidae and Indicatoridae. M.a.m.e.- caud. is fairly bulky, visible laterally, and weakly bipinnate, though no Aponeurosis 4 can be detected. Aponeurosis 3 is short and superficial, and there is no Aponeurosis 3a. Dendrocopos major. The origin of M.a.m.e.rost.temp. is small, and there is a vestigial postorbital slip. Aponeurosis 1 is broad dorsally, but partly covered by dorsal fibres; their superficial aponeurosis fuses with Ap.l anteriorly. M.a.m.e.rost.lat. is fairly narrow. M.a.m.e.vent., also narrow, extends well forward, concealing Aponeurosis 3 which inserts on a prominent ridge on the dorsal lateral surface of the mandible. FEEDING APPARATUS IN CORACIIFORMES & PICIFORMES 367 amert ptdm \ ptdmp ptdl ptvle B 5mm 5mm Fig. 13 Coracias abyssinica, jaw musculature: A, lateral view; B, dorsal view. M.a.m.e.caud. is highly developed, and Aponeurosis 4 extends a long way lateral and anterior to the rest of the 'muscle, with fibres fanning out from its medial surface to insert on the mandible over a wide area ventral and posterior to M.a.m.e.vent. Aponeurosis 3a is broad and strong, inserting posterior and medial to Aponeurosis 3, on the dorsal edge of the mandible. Other woodpeckers examined — including Picumnus — are very similar in all respects. M.a.m.e.caud. is somewhat reduced in Campephilus malherbi. M. pseudotemporalis superficialis The origin is fleshy, and often also aponeurotic from the posterior wall of the orbit. Insertion is made via a narrow tendon on the dorsal medial surface of the mandible. This typically lies lateral to the bulk of the muscle, whose fibres attach on it in unipinnate fashion. The relationship of the muscle to the course followed by the Vth (trigeminal) cranial nerve is of interest. Typically the nerve lies more or less lateral to the muscle, separating it from M. adductor mandibulae externus. In all the families considered here (with a slight modifi- cation in some jacamars and puffbirds) the tendon eventually passes medial to the nerve and inserts below it, though this is not so in all bird groups (e.g. some Charadrii; Burton, 1974a). CORACIIFORMES In most members of the order, the origin is mainly well to the lateral side of the orbit, and generally rather high. On the medial side there is usually a group of fibres originating con- siderably lower, but this only constitutes a small proportion of the bulk of the muscle. The Vth cranial nerve usually runs under the muscle towards its medial side (not lateral as in most birds), and is unusually near the dorsal surface by comparison with many other groups. The ramus pterygoidei and ramus mandibularis of the nerve typically run close alongside each other for a considerable distance in most members of this order, and the tendon of the muscle in many cases runs between them. ALCEDINIDAE. The muscle is long and narrow, and its fleshy region extends much further anteriorly than in the other families of the order. In Alcedo atthis and Halcyon chloris there is a high lateral origin with an aponeurosis along its medial edge from which fibres diverge forwards on each side, those on the medial side coming from a distinct medial lobe with a relatively low origin on the orbit. The Vth cranial nerve runs under this medial lobe. In Clytoceyx rex the medial and lateral lobes remain distinct for much of the length of the muscle, and each is bipinnate. 368 P. J. K. BURTON pr2 ptvle A B 5mm 5mm Fig. 14 Upupa epops, jaw musculature: A, lateral view; B, dorsal view. TODIDAE. The muscle is small in size and elongated in shape. Most of its bulk originates fairly high in the orbit; a small region on the medial side originates lower than the rest. The Vth cranial nerve runs along the medial side of the muscle, separating it at the origin from M. protractor quadrati et pterygoidei. The muscle is unipinnate in structure, fibres inserting on a lateral aponeurosis. About level with the posterior end of M. pseudotemporalis profundus this aponeurosis merges with the long slender tendon of insertion. MOMOTIDAE. The form of the muscle resembles that of the Todidae, but it is relatively con- siderably larger, the origin extending high up in the orbit. There is a distinct, though small, medial region with a much lower origin; this is bipinnate in Momotus and Baryphthengus, but of simple structure in Electron. The fibres of the muscle insert on an aponeurosis which cover the whole anterior surface of the muscle, merging into the tendon of insertion about level with the posterior limit of M. pseudotemporalis profundus. The Vth cranial nerve runs under the medial edge of the muscle. MEROPIDAE. Similar to the preceding two families, but ranking between them in the relative size of the muscle and the dorsal extent of its origin. The medial region is very much reduced, and the muscle lies almost entirely lateral to the Vth cranial nerve. LEPTOSOMATIDAE. The muscle resembles that of the Alcedinidae in form. It is long and narrow, and its fleshy region extends far forward. The lateral origin contributes less to the total bulk of the muscle than in other Coraciiformes. The posterior part of the muscle shows some bipinnate structure about a weak raphe attached to the orbit in the medial third of the muscle. The muscle is joined near its insertion by a slip from the lateral side of M. pseudotemporalis profundus. The Vth cranial nerve runs under the muscle towards its medial side. CORACIIDAE. Similar to the Meropidae in form and situation, but the fleshy region extends further forward — to about the midpoint of M. pseudotemporalis profundus — in Coracias benghalensis and Eurystomus glaucurus. It is smaller, and the fleshy region more limited, in Brachypteracias squamigera. UPUDIDAE. The muscle is bulky and less flattened than in the preceding families, and the Vth cranial nerve consequently lies much deeper. The lateral origin is well developed, and the fibres are attached in unipinnate fashion to a strong lateral aponeurosis which merges into the tendon of insertion about level with the posterior end of the short M. pseudo- temporalis profundus. FEEDING APPARATUS IN CORACIIFORMES & PICIFORMES 369 amev ptvle amec pss ptd amer B psp 5mm 10mm ptvl ptvm ptvms 5mm Fig. 15 Phoeniculus purpureus, jaw musculature: A, lateral view; B, dorsal view; C, ventral view. PHOENICULIDAE. Very similar to Upupa in both species examined. BUCEROTIDAE. Similar in form to that of the Upupidae and Phoeniculidae, but the medial edge is strongly emarginated by a crista on the wall of the orbit at about the junction of this muscle and M. protractor quadrati et pterygoidei. PICIFORMES The origin of M. pseudotemporalis superficial is in general more medially situated than in the Coraciiformes, and never extends very high up the orbit wall. In many species the medial side of the muscle is produced as a narrow extension across part of the origin of M. protractor quadrati et pterygoidei. The fifth cranial nerve lies deeper than in most Coracii- formes (except in the Galbulidae and Bucconidae), and generally runs well towards the lateral side of the muscle. The ramus pterygoidei of the nerve usually turns mediad away from the ramus mandibularis after only a short distance, well posterior to the starting point of the muscle's tendon of insertion (except in the Galbulidae, Bucconidae and Picidae). 370 P. J. K. BURTON amert ame pr rp psp qm ptdl amer Psp ptvle B ptdm 10mm 5mm Fig. 16 Tockus erythrorhynchus, jaw musculature: A, lateral view; B, dorsal view. GALBULIDAE. The muscle is vestigial in most jacamars, and entirely absent in one specimen of Jacamaralcyon. In this vestigial condition, it consists of a very small and narrow fleshy slip attached low on the orbit wall, immediately lateral to the fifth cranial nerve which, as in the Coraciiformes, lies superficial, with the ramus pterygoidei and ramus mandibularis running close together for some distance. A long and extremely slender tendon beginning level with the posterior end of M. pseudotemporalis profundus runs approximately between the ramus pterygoidei and ramus mandibularis of the nerve to insert just below the latter. In Jacamerops the tendon is forked at the insertion, one fork passing above the nerve to its attachment. The muscle is best developed — but still very small — in Galbalcyrhynchus. BUCCONIDAE. The muscle is also vestigial in the puffbirds, and entirely absent in all three specimens of Nonnula. In some species (Notharcus macrorhynchus, N. tectus, Hypnelus, Malacoptila), it is somewhat better developed, and approximately the shape of an equilateral triangle, though still very small. The form of the fifth cranial nerve, and its topological relations with the muscle are the same as in the Galbulidae. In Monasa morphoeus there is a forked insertion like that in Jacamerops. CAPITONIDAE. M. pseudotemporalis superficial is of moderate size, with its origin low on the orbit wall, extending medially to partially cover the origin of M. protractor quadrati et pterygoidei. The fifth cranial nerve lies deep on the lateral side of the muscle. The tendon of insertion begins about level with the midpoint of M. pseudotemporalis profundus. The medial extension is particularly marked in Megalaima haemacephala. The muscle is bipinnate in Lybius bidentatus, the raphe arising from the origin; there is some fleshy inser- tion as well as tendinous, but all medial and ventral to the ramus mandibularis of the fifth cranial nerve. There is a strong bipinnate medial slip arising from a prominent crista on the orbital wall in Semnornis, with a fairly broad fleshy insertion, also medial and ventral to the nerve. A similar condition is seen in Trachyphonus, as noted by Starck and Barnikol (1954). INDICATORIDAE. The muscle is similar in form to that of Megalaima haemacephala, with a long medial extension at the origin. It is slightly bipinnate in /. indicator, and strongly so in /. maculatus and Melichneutes in which the raphe arises from a prominent crista on the wall of the orbit. In /. minor, however, the muscle is of simple structure, with fibres FEEDING APPARATUS IN CORACIIFORMES & PICIFORMES 371 psp ptdm amert ptdl amerl Fig. 17 Galbula ruficauda, jaw musculature, dorso- lateral view. attached in a unipinnate arrangement on the lateral aponeurosis. The fifth cranial nerve, lies lateral to the muscle, and deep. The tendon of insertion begins well forward. RAMPHASTIDAE. Bulky and bipinnate in the three species examined, but similar in general form to barbets. The raphe is in the medial third of the muscle, arising from a strong crista on the orbit wall. The fleshy part of the muscle extends far forward, and there is a small amount of fleshy insertion additional to the tendon. The fifth cranial nerve runs deep, roughly under the midline of the muscle. PICIDAE. In Jynx, the muscle is of simple structure, with a long medial extension at the origin, much as in Megalaima haemacephala. In all Picinae and Picumninae examined, it is tri- angular in shape, and the origin is more lateral, with no medial extension. It is, however, of simple structure. The fifth cranial nerve runs under its midline. M. pseudotemporalis profundus and M. adductor posterior (N.A.A.: M. adductor poster ior=M. adductor mandibulae caudalis) These two muscles are closely associated, both functionally and morphologically, and it is convenient to treat them together. Like M. pterygoideus, they both act to lower the upper jaw at the same time that they raise the mandible. Since both attachments of each muscle move, the terms origin and insertion are arbitrary, but following usual practice they are considered as originating on the quadrate and inserting on the mandible. The origin of M. pseudotemporalis profundus is a fleshy one from the dorso- lateral surface of the orbital pro- cess of the quadrate, and often also from a dorsal aponeurosis attached to the medial edge of the process (generally rather a weak in the birds described here). It is otherwise largely a parallel-fibred muscle although a narrow band of fibres along the lateral or medial edges often shows a unipinnate arrangement. The insertion is a wide fleshy one on the medial surface of the mandible; the muscle often overlaps the dorsal edge of the mandible to some extent. M. adductor posterior takes its ori- gin from the proximal region of the orbital process and from the quadrate body. Lakjer (1926) regarded N. pterygoideus as separating this muscle from M. pseudotemporalis profundus, and his criterion is followed here, although Stark & Barnikol (1954) have suggested a different basis for discriminating the two. The insertion of M. adductor posterior is fleshy, and lies typically on the dorsal surface of the mandible, just posterior to the insertion of M.a.m.e.- caudalis, and to the medial surface of the mandible immediately dorsal to that of M. pterygoideus dorsalis medialis. However, in many species, there is more or less overlap onto the lateral surface of the mandible, ventral and posterior to the wide insertion of M.a.m.e.ventralis. 372 P. J. K. BURTON amert ptvle ptdl pss amer amev amev pwie B A 5mm 5mm Fig. 18 Monasa morphoeus, jaw musculature: A, lateral view; B, dorsal view. ALCEDINIDAE. M. pseudotemporalis profundus is absent in all members of the Alcedininae and several species of Cerylinae. It is, however, present and reasonably well developed in all members of the Daceloninae, and appears particularly broad in Dacelo gigas, D. leachii and Clytoceyx rex. In the Cerylinae, the muscle is absent in all species of Chloroceryle, and in Ceryle rudis. It is present in Ceryle torquata, C. alcyon and C. maxima; no specimen of C. lugubris was available. In these species it is very small, with a narrow and largely aponeurotic attachment to the tip of the orbital process; it is very clearly separated from M. adductor posterior by N. pterygoideus and loose connective tissue. M. adductor posterior is of normal development in all the Alcedinidae, without extensions onto the lateral surface of the mandible. In members of the Cerylinae lacking M. pseudotemporalis profundus there is a vestigial, thin orbital process, and the origin of M. adductor posterior extends right to its tip; the identity of the more medial fibres is in no doubt, because they are all clearly traversed by N. pterygoideus. In the Alcedininae, the orbital process is virtually absent, and the origin of M. adductor posterior is consequently limited to the quadrate body. TODIDAE. M. pseudotemporalis profundus appears fairly broad in dorsal view, but there is little indication of a dorsal aponeurosis. M. adductor posterior is fairly bulky and just visible laterally under M. adductor mandibulae externus caudalis. MOMOTIDAE. M. pseudotemporalis profundus and M. adductor posterior are of similar rela- tive development to those of the Todidae in Momotus and Baryphthengus. In Electron, M. adductor posterior shows a slight extension onto the lateral surface of the mandible, with a faint aponeurosis at its posterior edge. MEROPIDAE. M. pseudotemporalis profundus is fairly wide, and of even width as seen dor- sally. Its dorsal aponeurosis is best developed in Nyctiornis, where there is a slight indication of bipinnate fibre arrangement anteriorly. M. adductor posterior is invisible in lateral view. LEPTOSOMATIDAE. M. pseudotemporalis profundus is relatively small and narrow. From the lateral edge of the medial tip of the orbital process of the quadrate arises a narrow slip which runs outward to fuse with M. pseudotemporalis superficialis near its insertion. M. adductor posterior is barely visible in lateral view. CORACIIDAE. M. pseudotemporalis profundus is wide at the origin, tapering anteriorly to a rather narrow insertion. Its dorsal aponeurosis is well developed. M. adductor posterior is almost totally concealed in lateral view. UPUPIDAE. M. pseudotemporalis is small and narrow with a weak dorsal aponeurosis, and M. adductor posterior invisible laterally. FEEDING APPARATUS IN CORACIIFORMES & PICIFORMES 373 Phoeniculidae. Similar to Upupa, but M. pseudotemporalis profundus rather better developed. BUCEROTIDAE. Tockus erythrorhynchus. M. pseudotemporalis profundus is relatively small, and its dorsal aponeurosis is weak. However, it shows a marked bipinnate fibre arrangement about a raphe arising on the orbital process. M. adductor posterior is concealed in lateral view. GALBULIDAE. M. pseudotemporalis profundus is fairly small, and M. adductor posterior is barely visible laterally. BUCCONIDAE. In most puffbirds, the medial edge of the post orbital process is long, and M. pseudotemporalis profundus is very wide, as viewed dorsally. It is probably best developed in Nystalus spp. An exception is Hypnelus bicinctus, in which it is relatively much narrower, and the medial edge of the orbital process is shorter than in other members of the family. M. adductor posterior is invisible laterally in all species. CAPITONIDAE. M. pseudotemporalis profundus is fairly broad in most species although the dorsal aponeurosis is generally weakly developed, apparently least well developed in Trachy- phonus vaillantii. M. adductor posterior has a small extension onto the lateral surface of the mandible in Psilopogon pyrrholophus, Semnornis spp., Gymnobucco pelli and Pogoniulus leucolaimus; the extension .is a moderately large one in Capita niger, lying ventral and posterior to M. adductor mandibulae externus caudalis. INDICATORIDAE. Indicator indicator. M. pseudotemporalis profundus is narrower than in most barbets. M. adductor posterior can be seen laterally under M.a.m.e. caudalis. RAMPHASTIDAE. M. pseudotemporalis profundus is similar in form and development to most barbets in Selenidera maculirostris. In Ramphastos toco it appears narrow dorsally, especially at the insertion, but is deep nevertheless. M. adductor posterior is just visible laterally in both. PICIDAE. In all species examined, including Jynx, M. pseudotemporalis profundus is of moderate size, and M. adductor posterior rather small, and invisible laterally. M. pterygoideus General structure. Lakjer (1926) recognized four principal divisions of this complex muscle, and his system is followed here. In most of the species examined, the aponeuroses of the muscle correspond closely to those described by Zusi (1962) and Burton (1974) for various Charadriiformes, and these are numbered according to Zusi's system. M. pterygoideus nor- mally has its origin on the palatine and pterygoid, and inserts on the posterior part of the mandible. In some birds, a slip from this muscle inserts on the basitemporal plate of the skull. This slip, at one time referred to as 'M. retractor palatini', is considered under (5) below: (1) M pterygoideus dorsalis lateralis. (M.pter.dors.lat.) This division of the muscle is typically intimately connected with M. pterygoideus ventralis lateralis, and the two were thus treated as a single unit by Zusi (1962) and Burton (\914a). Fibres assigned to M.pter.dors.lat. are those originating on the dorsal surface of the palatine. They insert on the medial surface of the mandible immediately anterior to those of M. pterygoideus dorsalis medialis. The insertion is a broad and fleshy one; an additional surface for insertion is usually provided by a dorsal aponeurosis (Ap. M). (2) M. pterygoideus ventralis lateralis (M. pter. vent. lat.). Origin is from the lateral edge and caudo-lateral tip of the palatine, usually via a strong aponeurosis (Ap. A1). This aponeur- osis generally extends some way across the ventral surface of the muscle, and is often fused with Ap. A2 (see (4)). Its medial edge is commonly infolded, marking the boundary with M. pterygoideus ventralis medialis, and acting as a raphe, from which fibres of M. pterygoideus ventralis diverge caudad. 374 P. J. K. BURTON amert pss amev 5mm Fig. 19 Trachyphonus darnaudii, jaw musculature: A, lateral view; B, dorsal view. The insertion of M.pter.vent.lat. is fleshy, around the ventral edge of the mandible, and commonly on its lateral surface as well. The laterally inserting fibres form a distinct lobe referred to as the 'venter externus' of the muscle (Hofer, 1950). (3) M. pterygoideus dorsalis medialis. (M. pter. dors, med.) Origin is principally from the pterygoid, but commonly includes the caudal tip of the palatines. The origin is fleshy, and often aponeurotic as well, and the insertion, at the base of the internal process of the mandible, is fleshy, and also via a dorsal aponeurosis (Ap. O). Fibres originating on the antero- lateral face of the pterygoid and on the palatine diverge from those originating on the postero-medial face of the pterygoid. These two areas of the muscle are termed, respect- ively, M.pter.dors.med.anterior and M.pter.dors.med.posterior by Richards and Bock (1973). The anterior section is usually much the bulkier of the two. M.pter.dors.med. is separated from M.pter.dors.lat. by a groove, or sometimes a clear gap, into which, in most birds, the ramus pterygoideus of the Vth cranial nerve posses. In some species, the groove is absent, and the lateral and medial parts of M. pterygoideus dorsalis are continuous, as are Aponeuroses M and O. (4) M. pterygoideus ventralis medialis. (M. pter. vent. med.). The ventral surface of the palatine provides the main surface for the origin of this division of the muscle; there is usually also a strong aponeurosis (Ap. A2) attached to the palatine at the anterior boundary of the muscle and continued back some way across its ventral surface. The insertion, on the anterior face of the internal process of the mandible, is partly fleshy, but also includes a strong apo- neurosis (Ap. N) attached to the medial tip of the process, and running into the muscle as a raphe. The fibres of M. pter. vent. med. are oriented more nearly parallel to the long axis of the skull than those of the rest of the muscle. (5) Retractor palatini slip. Attachment of part of M. pterygoideus to the basitemporal plate has been reported in the Psittaci formes (Hofer, 1950; Burton, 1974c), Columbiformes (Didunculus; Burton, 1974c), Trogoniformes (Burton, 1974c), Bucerotidae (Starck, 1940) and many passerines (Moller, 1930, 1931; Engels, 1940; Fiedler, 1951; Bock, 19606, etc.). This is clearly a feature which has evolved independently in several lines, and it is therefore not surprising that the structure and origins of the slip vary considerably. It generally derives from part or all of M.pter.vent.med., but may also include fibres of M.pter.dors.med. (Bock, 19606). CORACIIFORMES A number of radical modifications in the structure of M. pterygoideus occur in this order, of which the most noteworthy are the shift of part of the origin of M. pterygoideus FEEDING APPARATUS IN CORACIIFORMES & PICIFORMES 375 ptdm amert amerl ptvl amev psp B 10mm 5mm Fig. 20 Indicator maculatus, jaw musculature: A, lateral view; B, dorsal view. lateralis to the maxillopalatine in the Alcedinidae, Phoeniculidae and Bucerotidae; and the development of a large retractor palatini portion in the Upupidae, Phoeniculidae and Bucerotidae. ALCEDINIDAE. The structure of M. pterygoideus in the Kingfishers departs extensively from the general structure given above. In the descriptions which follow, M.pter.dors.lat and M.pter.vent.lat. are considered as a single unit, M. pterygoideus lateralis (M.pter.lat.) for purposes of clarity. Alcedo atthis: The groove separating M.pter.dors.med. from M.pter.lat. is long, and is oriented at only a small angle (approx. 28°) to the midline of the skull. M.pter.dors.med. is continued far forward, to the anterior margin of the palatine, which is raised into a promi- nent crest. M.pter.lat. is attached to the palatine fleshily and by a medial aponeurosis to the antero-dorsal crest and the dorsal surface anterior to this. The rest of its attachment is made via a lateral extension of the ventral aponeurosis (ap. Al) to the dorsal surface of the palatal mucosa near the rictus; and to the transverse flange on the ventral surface of the maxilla which marks the posterior limit of the internal ramphotheca of the upper jaw. There is also a small amount of fleshy attachment at the medial tip of the maxillo-palatine. M.pter.- lat. is of bipinnate structure. The raphe (which runs approximately along the mixline of the muscle) arises from a very strong aponeurosis attached to the lateral surface of the mandible along the dorsal margin of the well developed venter externus; this aponeurosis also gives rise to the dorsal aponeurosis, Ap. M. The raphe is continued, weakening, to the anterior end of the muscle which is forked. The medial fork consists of fibres arising medial to the raphe, and it is this fork which provides the attachments on the palatine and maxillo- palatine. The lateral fork, consisting of fibres arising lateral to the raphe, provides the attachments on the palatal mucosa and maxilla via Ap. 1 . The orientation of M.pter.dor.med. is nearly the same as that of M.pter.vent.med. — a con- trast to the majority of birds in which the fibres of the two run at a sharp angle to one another. Consequently the two are largely fused in Alcedo. The most distinct part of M.pter.vent.med. is the medial region arising on the posterior edge of the pterygoid, and attached on the medial tip of the internal process of the mandible. The rest of the muscle arises by a short aponeur- osis from a limited area at the posterior end of the ventral surface of the palatine, and is continuous with M.pter.dor.med. above it, its origin being an extensive fleshy one on the dorsal surface of the palatine. The combined M.pter.med. is bipinnate about the horizontally oriented raphe Ap. N, fanning out from the medial tip of the internal process of the mandible. Ap. N is situated in the dorsal quarter of the muscle. 376 P. J. K. BURTON ptdl ptdm amert ptmxp psp pss amer amev amec 10mm 10mm Fig. 21 Selenidera maculirostris, jaw musculature: A, lateral view; B, dorsal view. The structure of M. pterygoideus in Alcedo may be thus summarized: M. pterygoideus lateralis: Origin: Fleshy and aponeurotic on the antero-dorsal crest of the palatine, the medial tip of the maxillo-palatine, the palatal mucosa and the ventral surface of the base of the maxilla. Insertion: fleshy and aponeurotic on the lateral surface of the mandible (venter externus) and a limited region of the ventro-medial surface of the mandible near the base of the internal process. Structure: bipinnate about a raphe arising on the lateral surface of the mandible. M. pterygoideus dorsalis medialis: Origin: fleshy on the dorsal surface of the palatine. Insertion: on the internal process of the mandible, fleshily and via. Ap. N. M. pterygoideus ventralis medialis: Origin: mainly aponeurotic from the ventral surface of the palatine, and fleshy from the posterior edge of the pterygoid. Insertion: fleshy and via Ap. N on the internal process of the mandible. Structure: M.pter.dors.med. and M.pter.vent.med. form a unit which is bipinnate about Ap. N. Other Alcedinidae: M.pter.lat. is of bipinnate structure in all members of the family examined except Tanysiptera galatea. The raphe runs roughly along the midline of the muscle in the Cerylinae and Alcedininae, but is shifted more or less to the lateral side in the Daceloninae. The attachment to the maxillo-palatine is small in the Alcedininae, and limited to its medial tip, while in the Cerylinae it is absent altogether. In the Daceloninae, however, this attachment is much more extensive, reaching an extreme in Dacelo (especially D. novaeguinae and D. leachii), Melidora and Clytoceyx rex. In these species, the attachment is broad, extending over much of the length of the maxillo-pallatine, and is made by a dorsal aponeurosis on the posterior edge of the maxillo-palatine, and fleshily on its ventral surface. The attachment to the palatine in these species is extremely limited, and takes the form of a narrow slip, its fibres diverging rostro-mediad from those attached to the maxillo-palatine. The main raphe of M.pter.lat. in Dacelo, Clytoceyx and Melidora lies lateral and ventral to its position in other kingfishers, so that the remainder of the muscle, attaching on the mucosa and maxilla via Ap. Al is largely hidden by the more bulky section originating on the maxillo-palatine. In Tanysiptera galatea the attachment to the maxillo-palatine is pre- sent, and about as well-developed as in most Halcyon spp., but the attachment to the mucosa and maxilla is entirely absent. M.pter.dors.med. extends to the antero-dorsal crest of the palatine in all kingfishers examined. The groove between this muscle and M.pter.lat. makes a relatively small angle FEEDING APPARATUS IN CORACIIFORMES & PICIFORMES amert pss amec B 5mm Fig. 22 Jynx torquilla, jaw musculature: A, lateral view; B, dorsal view. to the midline of the skull in all Alcedininae and also Cerylinae (e.g. Chloroceryle amazona, 26°). However, among the Daceloninae, the groove is more steeply angled to the midline (i.e. more 'normal') in Halcyon and Tanysiptera (e.g. 35°, Halcyon pileata; 33°, Tanysiptera galated) — in part a consequence of the relatively shorter pterygoids and more posteriorly situated palatines in the Daceloninae. In these genera, the distinction between M.pter.dors.- med. and M.pter.vent.med. is clearer, as a result of the different orientation of their fibres. Nevertheless, despite similar skull morphology, Dacelo and Clytoceyx show the smallest angle of groove to midline of all (approx. 22° — 23° in both genera), due to the shift of much of the origin of M.pter.lat. from the palatine to the maxillo-palatine. In all other features, M. pterygoideus conforms closely to the description for Alcedo atthis throughout the Alcedinidae. TODIDAE. Todus viridis. M.pter.dors.lat. is of simple structure, with fibres converging evenly from the origin to the insertion near the base of the internal process of the mandible. M.pter.- dors.med. has origin only from the posterior end of the palatine and the groove separating it from M.pter.dors.lat. is oriented at about 40° to the midline of the skull. The ventral surface of M.pter.vent.med. shows a slight separation of fibres originating on the medial crest of the palatine, and those arising from its ventral surface. There is scarcely any venter externus of M.pter.vent.lat. MOMOTIDAE. Momotus momota. M.pter.dors.lat. is bipinnate, with the raphe well over to the lateral side; part of its origin is on the dorsal surface of the palatal mucosa, lateral to the palatine. M.pter.dors.med. has little attachment on the palatine, and the groove separat- ing it from M.pter.dors. lat. is steeply angled to the midline (approx. 50°). From the ventral aspect. M.pter.vent.med. shows marked separation into a lobe originating on the medial crest of the palatine, and that originating on the rest of the palatine. Most attachment of M.pter.- vent.lat. is at the base of the internal process, including the very strong Ap. M. The venter externus is well developed. Baryphthengus ruficapillus is very similar. In Electron platyrhynchum, however, M.pter.- dors.lat. is of simple structure, and the groove separating it from M.pter.dors.med. runs further anterior, at an angle of about 33° to the midline. MEROPIDAE. Nyctiornis amicta. M.pter.dors.med. extends forward to the antero-dorsal crest of the palatine: the groove separating it from M.pter.dors.lat. is oriented at approx. 30° to the midline. M.pter.dors.lat. itself is of simple structure, and its area of insertion is mainly at the base of the internal process of the mandible. There is hardly any development of the venter externus. Ventrally, M.pter.vent.med. bulges out laterally, ventral to Ap. Al. 378 P. J. K. BURTON amert ptdm pss dm psp amerl B 3mm Fig. 23 Sasia abnormis, jaw musculature: A, lateral view; B, dorsal view. Aerops albicollis is similar, but M.pter.dors.med. has even more extensive origin on the palatine, and is clearly broader and more bulky than M.pter.dors.lat. The groove between the two runs at approx. 26° to the midline. LEPTOSOMATIDAE. Dorsally, no clear groove can be distinguished separating M.pter.dors.lat. and M.pter.dors.med. Both are of simple structure, with fibres converging evenly towards the insertion low on the medial surface of the mandible, and at the base of the internal process. M.pter.vent.lat. has a moderately developed venter externus. LEPTOSOMATIDAE. M.pter.dors.med. is much reduced. There is a distinct posterior section as in the Coraciidae, and this is the only part of the muscle reaching the basiphenoid rostrum; the anterior section has origin only on the posterior half of the pterygoid, and is concealed by M. pseud.prof. in dorsal view. M.pter.vent.lat. has a small venter externus. CORACIIDAE. Coracias benghalensis. M.pter.dors.med. extends well forward on the palatine, not quite reaching its antero-dorsal crest. The groove separating it from M.pter.dors.lat. is oriented at the relatively shallow angle of 35° to the midline. Fibres of M.pter.dors.med. originating on the postero-medial surface of the pterygoid are angled nearly parallel to the midline, forming a distinct posterior section. There is little development of a venter externus, and the insertion of M.pter.vent.lat. includes the medial surface of the mandible, as well as the base of the internal process. Ap. M. is inserted quite high on the medial surface of the mandible. In Eurystomus glaucurus and Brachypteracias squamigera, M.pter.dors.med. also shows a distinct posterior section, but has little attachment on the palatine. The groove between this and M.pter.dors.lat. is steeply angled to the midline (49° in Eurystomus, 44° in Brachypteracias}. There is no venter externus in Eurystomus, but a well developed one in Brachypteracias. UPUPIDAE. Upupa epops. There is no groove or other indication of demarcation between M.pter.dors.lat. and M.pter.dors.med. The pterygoid ramus of the 5th cranial nerve enters the muscle in a very posterior position immediately level with the anterior edge of the orbital process of the quadrate. The lateral fibres of M.pterygoideus dorsalis have a substantial area of insertion on the medial surface of the mandible, anterior to the base of the internal process, and the insertion of M.pter.vent.lat. includes a well developed venter externus. The posterior region of M.pter.dors.med. and the medial region of M.pter.vent.med. are modified to form a large retractor palatini portion, with substantial fleshy attachment on the basitemporal plate. The retractor palatini fibres arise partly on the medial half of the ventral surface and FEEDING APPARATUS IN CORACIIFORMES & PICIFORMES 379 on the medial crest of the palatine; and also on the postero-medial edge (including part of the dorsal surface) of the pterygoid. The part arising on the pterygoid is bulky, and is promi- nent in dorsal view. The lateral half of M.pter.vent.med. is of normal structure, arising mainly on the lateral half of the ventral surface of the palatine, and inserted on the internal process of the mandible, with a raphe (Ap. N) fanning out from the medial tip of the internal process. PHOENICULIDAE. Phoeniculus purpureus. As in Upupa, there is no separation of lateral and medial parts of M.pter.dors. This muscle is, however, strongly bipinnate, the raphe running in its outer quarter, roughly parallel to its lateral edge. The raphe is continuous medially with Ap. N, and dorsally with Ap. M. Ap. N. has an extended region of attachment to the mandible in the form of a crista along the dorsal edge of the anterior face of the internal process. The origin of M. pterygoideus dorsalis extends a long way forward, and its anterior edge is attached fleshily and by a dorsal aponeurosis to the whole ventral edge of the maxillo- palatine. M.pter.vent.med. also has a long area of origin on the ventral surface of the palatines, and there is a thin superficial band of fibres running from the groove between M.pter.vent.med. and M. retractor palatini to the mucosa in the midline. (Fig. 9c, ptvms). Other features, including the structure of M. retractor palatini, resemble Upupa epops. P. aterrimus resembles P. purpureus, but M.pter.dors. is not bipinnate, though attached to the maxillo-palatine. The venter externus is bulky, with a strong lateral aponeurosis which has a group of fibres attached in unipinnate fashion along its dorsal edge. Rhinopomastus cyano- melas shows attachment of M.pter.dors. to the maxillo-palatine, but as in P. aterrimus, the muscle lacks bipinnate structure. The venter externus is rather weaker than in the two previous species, with no obvious lateral aponeurosis or pinnate structure. BUCEROTIDAE. Tockus erythrorhynchus. M. pterygoideus dorsalis is divided into lateral and medial parts by a clear groove running at approx. 40° to the midline of the skull. The pterygoid ramus of the Vth cranial nerve enters M.pter.dors.med. some way posterior to the groove, rather than passing into the groove itself as in all other birds considered in this study (except those showing no separation of medial and lateral parts). M.pter.dors.med. has an origin confined to the anterior edge of the pterygoid, but the retractor palatini portion of the muscle has an extensive dorsal origin including not only the posterior edge of the ptery- goid, but also the medial edge of the posterior end of the pterygoid. The retractor palatini is pinnate in structure, with the raphe arising on the posterior tip of the palatine. The raphe is visible on the lateral face of the dorsal attachment of the muscle, fibres dorsal to it running more or less parallel with it, while those ventral to it diverge forwards to attach on the pterygoid. M.pter.dors. lat. is weakly bipinnate about a broad dorsal aponeurosis (Ap. M.). Anteriorly there is an extensive fleshy origin along the whole length of the maxillo-palatine. M.pter.- vent.lat. has an extremely bulky venter externus, bulging out prominently on the lateral surface of the mandible. M.pter.vent.med., apart from the retractor palatini portion, is of normal structure, with a strong raphe (Ap.N). PICIFORMES The main departures from the general description are seen in the Galbulidae, in which M. pterygoideus lateralis shows extreme reduction; and in the Capitonidae and Ramphastidae, in which attachment of M.pter.dors.lat. to the maxillo-palatine is found. GALBULIDAE. Galbula dea. M.pter.dors.lat. and M.pter.vent.lat. form a single unit which is greatly reduced in size by comparison with other birds. It consists of a narrow slip originating via Ap. A 1 along the lateral edge of the palatine, and inserted on the ventro-medial and ventral edge of the mandible at the base of the internal process. In dorsal view, the groove separating M.pter.lat. from M.pter.dors.med. is oriented at the very narrow angle of 10° to the midline of the skull. M.pter.dors.med. extends appreciably further forward on the pala- tine than M.pter.lat. The pterygoid ramus of the Vth cranial nerve enters the muscle through 380 P. J. K. BURTON amert \ ' \ \ x^*"^^& \ I m amer amev Fig. 24 Dendrocopos major, jaw musculature: A, lateral view; B, dorsal view. the groove near the orbital process of the quadrate in the normal way. The remainder of the muscle is of normal structure. In Jacamerops aurea there is a partial groove extending across the antero-medial half of the M.pter.dors.med. well back, and oriented at approx. 30° to the skull midline. M.pter.lat. is least reduced in Galbalcyrhynchus leucotis, in which the groove separating it from M.pter.dors.med. is angled at 15° to the midline. BUCCONIDAE. Chelidoptera tenebrosa. M. pterygoideus dorsalis is of normal structure, with the groove separating lateral and medial parts situated well back, and angled at 45° to the skull midline. M.pter.dors.med. extends only onto the posterior end of the palatine. Ven- trally, M.pter.vent.med. shows marked separation between fibres originating on the main ventral surface of the palatine, and those originating on the medial crest, which pass super- ficial to the main part of the muscle. M.pter.dor.lat. inserts low, on the ventral edge of the mandible. The venter externus is very small. In all other Bucconidae examined, M. pterygoideus dorsalis shows an approach to the Galbulidae, with M.pter.dors.med. extending further onto the palatine, and the groove separating it from M.pter.dors.lat. oriented much more parallel to the skull midline than in Chelidoptera. In most species, the groove is noticeably curved, its posterior quarter or so being more steeply angled to the midline than the rest. Extremes are reached in Notharcus macrorhynchus and N. tectus, and in Nonnula ruficapilla, in which the angle of the groove to the midline is approx 20° over most of its length, and M.pter.dors.med. reaches the antero- dorsal crest of the palatine, and is noticeably larger than M.pter.dors.lat. Less extreme angles, and smaller extensions of M.pter.dors.med. onto the palatine are seen in, e.g. Hypnelus rufi- collis (approx. 30°), Malacoptila panamensis (27°), Monasa morphoeus (26°), Nystalus chacuru (25°). In Nystalus chacuru, N. radiatus and N. maculatus, M.pter.dors.lat is weakly bipinnate about a broad dorsal aponeurosis derived from Ap. M. The venter externus is weakly developed throughout, and entirely absent in Hypnelus ruficollis and Nonnula ruficapilla. CAPITONIDAE. Megalaima haemacephala. M.pter.dors.med. is very narrow, and situated far back, its boundary with M.pter.dors.lat. angled at 55° to the skull midline. Its origin never- theless includes substantial attachment to the posterior end of the palatine. M.pter.dors.lat. is of simple structure, but its anterior end has attachment, fleshily and by a short, weak dorsal aponeurosis, to the medial 2/3 of the ventral edge of the maxillo-palatine. There is no venter externus. M.pter.vent.med. is of normal structure. In most other barbels examined, M.pter.dors.med. is relatively larger, and extends further FEEDING APPARATUS IN CORACIIFORMES & PICIFORMES 381 forward, the angle to the midline of its boundary with M.pter.dors.lat. ranging from 45° (Trachyphonus vaillantii) to 35° (Pogoniulus leucolaima, Gymnobucco pelli). The extent and mode of attachment to the maxillo-palatine shows interesting variation. No attachment was found in Psilopogon pyrrholophus, Gymnobucco pelli, Trachyphonus vaillantii or Capito niger. In Lybius bidentatus there is a small amount of attachment on the ventral surface of the maxillo-palatine. In Megalaima virens there is a distinct attach- ment, but occupying rather less of the maxillo-palatine than in M. haemacephala. In Pogoniulus leucolaima the attachment is made by a distinct, if narrow, dorsal slip. The attachment is best developed in Semnornis frantzii and S. ramphastinus. In both species there is a distinct and well developed dorsal slip attaching on the medial half of the ventral edge of the maxillopalatine, fleshily and by a strong dorsal aponeurosis. Lateral to the slip, the origin of the muscle extends onto the dorsal surface of the palatal mucosa, near the rictus. In S. rhamphastinus there is a narrow band of fibres lateral to Ap. M and diverging forward from it in unipinnate fashion. InoiCATORiDAE. Indicator indicator. M. pterygoideus shows no unusual modifications, although extending well forward on the dorsal surface of the palatine, there is no attachment to the maxillo-palatine. M.pter.dors. is clearly divided into lateral and medial portions by a groove running at approx. 47° to the skull midline. The origin of M.pter.dors.med. is confined to the pterygoid (see p. 355 regarding structure of the pterygo-palatine junction in this family and the Picidae). M.pter.dors.lat. is bipinnate, with the raphe in the lateral fifth or quarter of the muscle. M.pter.vent.lat. and med. are of normal structure, and there is no venter externus. M.pter.dors.lat. is scarcely bipinnate in /. maculatus, and not all in /. minor. Melichneutes resembles /. indicator. RAMPHASTIDAE. Selenidera maculirotris. M.pter.dors.med. is narrow, and situated well back, its boundary with M.pter.dors.lat oriented at 45° to the skull midline. M.pter.dors.lat. itself has an extensive attachment to the whole ventral edge of the maxillo-palatine in addition to its origin on the palatine. This attachment is made by a bulky raised dorsal slip, fleshily, and by a strong dorsal aponeurosis. The slip overlaps the lateral part of the palatine origin on its medial side. Beneath it lies a separate thin sheet of fibres with origin on the dorsal surface of the mucosa round the rictus, fleshily, and by a ventral aponeurosis (an extension of Ap. Al). There is scarcely any development of a venter externus. The rest of the muscle is of normal structure. Ramphastos toco is similar, but the slip originating on the maxillo- palatine is relatively even broader. The groove separating lateral and medial parts of M.pter.dors. is absent. PICIDAE. Jynx torquilla. M.pter.dors.med. is narrow, and separated by a gap of about 10° from M.pter.dors.lat. The midline of this gap is oriented at about 65° to the skull midline. The origin of M.pter.dors.med. is confined to the pterygoid. Otherwise, the muscle shows no unusual modifications, and M.pter.dors.lat. is of simple structure. M.pter.vent.lat. and med. are of normal structure; there is no venter externus. Dendrocopos major has a strikingly elongated M.pter.dors.lat., extending forward almost to the midpoint of the nostril. This has a narrow band of fibres lateral to Ap.M., diverging forwards in a unipinnate arrangement. It is separated by a groove oriented at 40° to the skull midline from M.pter.dors.med., which has its origin confined to the pterygoid. There is no maxillo-palatine attachment or other unusual features. M.pter.vent.lat. and med. are of normal structure, and there is a moderately developed venter externus. Other woodpeckers examined are similar. In Picumnus olivaceus M.pter.dors.lat. is somewhat shorter, and of simple structure; otherwise, M. pterygoideus resembles that of the Picinae. M. protractor pterygoidei et quadrati. This muscle originates from the interorbital septum and the posterior wall of the orbit, medial to M. pseudotemporalis superficialis. Its insertion is on the dorsal surface of the posterior end of the pterygoid, and on at least the region of the quadrate body immediately 382 P. J. K. BURTON adjacent to the pterygoid; there is frequently also insertion on the postero-medial surface of the orbital process. In some birds, the regions of the muscle originating on the interorbital septum and the posterior orbit respectively show more or less differentiation into two distinct parts. In such cases, the part originating on the septum is referred to as 'M. protractor 1', and that arising on the posterior orbit as 'M. protractor 2'. The insertion of M. protractor 1 is a narrow one on the dorso-medial surface of the posterior end of the pterygoid, and is partly made via an aponeurosis; that of M. protractor 2 is wider and fleshy, on the quadrate body and usually also the postero-medial surface of the orbital process. The lateral fibres of M. protractor 2 may pass dorsal to the insertion of M. protractor 1. Medial fibres of M. protractor 2 may attach on the aponeurosis of M. protractor 1 in a pinnate arrangement. ALCEDINIDAE. The muscle is narrow in form, but relatively fairly large. It shows no differen- tiation into parts 1 and 2, and has no attachment to the orbital process, which is in any case vestigial or absent in the Alcedininae and several Cerylinae. TODIDAE. M. protractor is well developed, and its origin on the orbittum and interorbital septum is high. It shows slight differentiation into two parts, but there is only limited attachment to the orbital process, low down, near its base. MOMOTIDAE. M. protractor is narrow, its origin extending high on the orbit and interorbital septum. It has no attachment to the orbital process. In Baryphthengus it shows slight bipinnate structure, but there is no real differentiation into two parts. MEROPIDAE. M. protractor is large, and has insertion over the entire medial surface of the orbital process. However, it shows no differentiation into two parts. LEPTOSOMATIDAE. The muscle is relatively rather small, but has extensive attachment to the orbital process. There is no clear division into two parts. CORACIIDAE. M. protractor is relatively small. It has attachment to the postero-lateral half of the orbital process, but shows no differentiation. UPUPIDAE. M. protractor is extremely large, its origin extending high, and nearly to the anterior end of the orbit. It is differentiated into two parts; part 1 is surrounded near its inser- tion by a strong aponeurosis, which is strongest along the posterior edge and partly ossified in some specimens. Fibres of part 1 run ventrally and posteriorly to their attachment on this aponeurosis, and fibres of part 2 run ventrally and anteriorly to attach to its other side in a strongly bipinnate arrangement. There is, however, no attachment to the orbital process. PHOENICULIDAE. Similar to Upupa, but part 1 extends rather less far forward, especially in Rhinopomastus. Parts 1 and 2 are even further differentiated in Rhinopomastus, lateral fibres of Part 2 passing across the dorsal side of Part 1 at its insertion, but as in Phoeniculus, there is no attachment to the orbital process. BUCEROTIDAE. M. protractor is quite small, and shows no differentiation or attachment to the orbital process. GALBULIDAE. The muscle is fairly large, but generally of simple structure; there is an indication of differentiation into two parts in Galbalcyrhynchus and Jacamerops. There is attachment to the full length of the orbital process. BUCCONIDAE. The development of M. protractor varies considerably. It appears to be best developed in Hypnelus bicinctus, in which the origin extends high on the interorbital septum, and there is a distinct division into parts 1 and 2. Part 2 is attached over most of the orbital process except the medial half of its medial edge. It is attached to the entire medial surface of the orbital process in Nonnula ruficapilla, although there is no differentiation into two parts. It is well developed also in Chelidoptera, and somewhat less so in Nystalus. In Notharcus, Malacoptila and Monasa, M. protractor is relatively small, and has little attachment on the orbital process. FEEDING APPARATUS IN CORACIIFORMES & PICIFORMES 383 CAPITONIDAE. M. protractor is generally narrow, and of simple structure, with no attachment to the orbital process. In Psilopogon, its aponeurosis appears particularly well developed along its anterior edge, and fibres insert on it in unipinnate fashion. Indicatoridae. Very similar to most Capitonidae. RAMPHASTIDAE. Similar to the Capitonidae. PICIDAE. In Jynx. M. protractor resembles that of the Capitonidae. It is narrow and fairly small, with no subdivision or attachment to the orbital process. In the remaining members of the family, including Picumnus, the muscle is greatly enlarged, and clearly divided into two parts, and attached to the entire medial surface of the orbital process. Part 1 attaches high and far forward on the interorbital septum. Its insertion is by a very strong aponeurosis arising from a long spur on the pterygoid. Part 2 arises from the posterior orbital wall; some of its medial fibres insert on the aponeurosis of Part 1 , but most on the orbital process. In some, the area for insertion is slightly increased by a small aponeurosis extending medially from the tip of the orbital process, and merging with the aponeurosis of Part 1. M. depressor mandibulae This muscle is fairly small, and of uniform structure throughout the species examined, with the exception of the Upupidae and Phoeniculidae. In general, it originates from the exoccipital and insects over a wide area on the posterior end of the articular, including the posterior surface of the internal process of the mandible. Both origin and insertion are mainly fleshy. In the birds examined here, surface aponeuroses are in most cases only moderately or weakly developed. They occur on the dorsal surface of the muscle (arising from the ventral edge of the exoccipital) and near the insertion, on the lateral and central surfaces (arising from the lateral and ventral edges of the posterior face of the articular). In the Upupidae and Phoeniculidae, the muscle is much larger than in the other families studied, extending at the origin well up onto the temporal region of the squamosal. The area for insertion is increased by the development of a long retroarticular process — a backward extension of the articular, continued some distance posterior to the internal process. Only the medial face of the retroarticular process is occupied by M. depressor, whose fibres run rostro- lateral to meet it. Lateral and dorsal aponeuroses are well developed, but show no infolding or internal branches as in, e.g. the Huia (Callaeidae, Heteralocha acutirostris; see Burton 1 974/?) and some Charadrii (Burton 1974a). However, in the Phoeniculidae, some ventral fibres pass under the edge of the lateral aponeurosis and insert in a narrow band along the ventral border of its lateral surface. The only other family showing any approach to the Upupidae and Phoeniculidae is the Bucerotidae. In this family, although the muscle shows no marked enlargement, there is a definite indication of a retroarticular process, the postero- ventral corner of the articular projecting backwards, rather than being emarginated as in all the remaining families. Tongue and hyoid apparatus Although the morphology of the horny tongue has been described for a number of Coracii- form and Piciform birds (e.g. in Beddard, 1898), anatomical details of the hyoid skeleton and musculature are largely lacking, except in the case of the Picidae (Leiber, 1907a & b). Nomenclature used in descriptions largely follows that of Bock (1972) and Richards & Bock (1973) N.A.A. equivalents (Baumel et al. 1979) are given where they differ from terms used here. Tongue The horny tongue varies greatly among the families studied. Extreme tongue reduction is seen in the Alcedinidae, Upupidae, Phoeniculidae and Bucerotidae. Very long tongues are 384 P. J. K. BURTON ent Fig. 25 Hyoid skeletons of Coraciiformes and Galbuloidea. Cartilage revealed by bulk staining with New Methylene Blue is indicated by stippling, (a) Alcedo atthis; (b) Todus todus; (c) Momotus momota; (d) Aerops albicollis; (e) Leptosomus discolor, (f) Coracias benghalensis; (g) Upupa epops; (h) Phoeniculus purpureus; (j) Buceros bicornis; (k) Galbula dea\ (1) Monasa morphoeus. seen in several families, such as the Coraciidae (particularly the ground rollers), Momotidae, a few barbets and the Ramphastidae. In absolute values, the tongues of the larger toucans must rank among the longest of all birds. The extraordinarily long and extensible tongues of the Picidae cannot be directly com- pared with those of other families, since much of their length is taken up by the greatly elongated basihyal, and part of the hyoid horns; the horny tongue itself is much reduced, though still a vital part of the feeding apparatus. Surveys of the varied morphology of wood- pecker tongues and hyoid apparatus have been provided by Steinbacher (1934, 1935, 1941, 1955, 1957). Brush tongues occur in several families. Brush structure is highly developed, involving both tip and more or less of the sides in most Momotidae and the Ramphastidae. The latter family show this feature in extreme form; the tongue of Ramphastos toco, for example is a remarkable and beautiful structure, delicately tipped and fringed with fine laciniae for 80% of its considerable length. The Coraciidae, especially the ground rollers, also have quite well FEEDING APPARATUS IN CORACIIFORMES & PICIFORMES 385 ent Fig. 26 Hyoid skeletons of Piciformes other than Galbuloidea. (a) Psilopogon pyrolophus; (b) Pogoniulus bilineatus; (c) Lybius bidentatus; (d) Trachyphonus darnaudii; (e) Indicator indicator; (f) Selenidera maculirostris; (g) Jynx torquilla; (h) Sasia abnormis; (j) Colaptes auratus. developed brush tongues, and a moderate brush tip is also seen in Capita and Semnornis among the Capitonidae. Slight indications of brush structure at the tongue tip are seen in the Meropidae and Jacamerops. An African barbet (Buccanodon melanolophus) has a bifid tongue tip. Barbed tongues are characteristic of Picidae other than Jynx. Apart from these, tongues are of simple structure for most of their length in all other species, that is to say, in the Alcedinidae, Todidae, Upupidae, Phoeniculidae, Bucerotidae, Bucconidae, Indicatoridae and most Galbulidae and Capitonidae. Basal barbs are normal as in most birds, but are particularly numerous in the Phoeniculidae and Bucerotidae, occur- ring on the dorsal surface as well as the margins in Phoeniculus purpureus and hornbills. Basal barbs are lacking in the Coraciidae, as noted by Beddard (1898). Many species show a slight indication of a median groove dorsally, corresponding with a ridge on the horny lining of the upper jaw, but a particularly strongly marked groove is characteristic of the Bucconidae, and quite noticeable also in Jacamerops and Galbula (Galbulidae). Hyoid skeleton The hyoid skeletons of representative species of each family are depicted in Figs. 19 & 20. 386 P. J. K. BURTON bmdm bmdl mhy amev ptvle sthy sehy bmd Fig. 27 Todus todus. Ventral view of superficial tongue muscles. These illustrations eliminate much of the need for description; however, a few points deserve comment. The entoglossum generally tends to be more heavily ossified in the Piciformes than in the Coraciiformes, where any ossification is mainly confined to its posterior half. In general, the postero-lateral processes are longer in the Coraciiformes, producing more of an 'arrowhead' shape than in the Piciformes. The size of the entoglossum generally reflects the development of the tongue, but its form shows various features of interest in some Piciformes. The peculiar broadened tip of the entoglossum in Monasa is perhaps of minor significance; its sharply demarcated cartilaginous anterior and bony posterior much resemble the entoglossum ofGalbula. Among the typical Piciformes, a more intriguing modification is seen; this takes the form of a strongly broadened anterior section, in some cases with backwardly directed lateral processes, as though an additional ento- glossum had been grafted onto the front of the original. Of the Capitonidae checked for this feature, Psilopogon and Trachyphonus darnaudii showed it in a strongly developed form, and Megalaima virens moderately, while it was absent in Lybius bidentatus. Among other families, it is strongly developed in Selenidera, and moderately in Indicator. Within the Picidae, the entoglossum is very short and narrow, though with a slightly narrowed 'waist' posteriorly in Jynx and Sasia. The basihyal is rather short and often broad in most Coraciiformes, and in the Galbulidae and Bucconidae. Broadening reaches an extreme in the Alcedinidae, where the basihyal is broader than long, forming a flat plate, with just a tiny narrow projection anteriorly to articulate with the entoglossom. The basihyal has well marked lateral flanges in the Todidae. In the Upupidae and Phoeniculidae, the basihyal has a characteristic 'hour-glass' form, also seen, though less strongly, in the Bucerotidae. In the typical Piciformes, the basihyal is FEEDING APPARATUS IN CORACIIFORMES & PICIFORMES 387 bmd sehy trhy 5mm Fig. 28 Coracias abyssinica. Ventral view of tongue muscles to show anterior slip of M. serpihyoideus. generally simple in shape, and rather long and narrow, reaching an extreme in the Picidae. Pogoniulus and Indicator are exceptions in which the basihyal is rather short. Other parts of the hyoid skeleton require little comment. It may be noted that the great length of the hyoid horns seen in the Picidae has been attained mainly by elongation of the epibranchiale. Hyoid muscles M. mylohyoideus (N.A.A.: M. intermandibularis) This very thin muscle is a sheet of fibres extending across the floor of the buccal cavity between the medial surfaces of the two mandibular rami, to each of which it is attached fleshily along a narrow line dorsal to M. branchio-mandibularis (and M. genioglossus where present). Its posterior end may merge with M. serpihyoideus. There appears to be no median raphe of insertion as in Loxops (Richards & Bock, 1973). The muscle is weakly developed in most of the families examined here, and was barely detectable in Coracias benghalensis, Galbula dea and Megalaima haemacephala. It appeared best developed in the Meropidae, Indicator and the larger Capitonidae. However, comparisons are rendered difficult by the varying states of contraction of the muscle in different specimens, resulting from the common practice of putting cotton wool plugs into the buccal cavity of specimens at the time of collection. 388 P. J. K. BURTON bmd bmd sthy sehy sehy bmd bmd B 10mm 10mm Fig. 29 Megalaima zeylanica, tongue musculature: A, ventral view; B, dorsal view. Left M. ceratoglossus free from basihyal and pinned aside. M. ceratohyoideus (N.A.A.: M. interceratobranchialis) The muscle originates fleshily on the ceratobranchiale close to its articulation with the epibranchiale. It passes along the ceratobranchiale, surrounded by M. branchiomandibularis, then diverges mediad to insert on a median raphe with M. ceratohyoideus from the opposite side. The insertion lies deep to M. serpihyoideus and M. mylohyoideus, and immediately ventral to the urohyal. M. ceratohyoideus is absent in the Indicatoridae and Picidae, and feebly developed in the Alcedininae, in which insertion is on the posterior edge of the basihyal and anterior tip of the urohyal, rather than on the midline. Otherwise it shows little noteworthy variation among the species studied. M. stylohyoideus This muscle has a common fleshy origin with M. serpihyoideus on the postero-ventral edge of the mandible. It appears as a narrow slip which diverges from the anterior edge of M. serpihyoideus and runs to a fleshy insertion on the dorso-lateral surface of the basihyal, typically at its anterior end, and lateral to that of M. thyrohyoideus. This is its condition in the families studied here, with the following exceptions: FEEDING APPARATUS IN CORACIIFORMES & PICIFORMES 389 bmd 5mm Fig. 30 Indicator maculatus. Dorsal view of tongue musculature to show M. hypoglossus obliquus. ALCEDINIDAE. Insertion is on the dorsal surface of the anterior tip of the cerato- branchiale. TODIDAE. M. stylohyoideus is extremely slender and absent in one specimen. MOMOTIDAE. Insertion on the dorso-lateral surface of the posterior end of the basihyal. CORACIIDAE, Galbulidae, Bucconidae: M. stylohyoideus is apparently replaced by a slip from M. serpihyoideus (q.v.). UPUPIDAE. Extremely slender. BUCEROTIDAE. Absent. PICIDAE. Generally much reduced, with no attachment on the hyoid skeleton, and probably absent in some (see Leiber, 1907# & b). M. serpihyoideus Origin is on the postero-ventral edge of the mandible, medial to that of M. stylohyoideus. The insertion is on a medial raphe, the right and left muscles together forming a shallow, forwardly directed 'V. The anterior tip is in close proximity to M. mylohyoideus posterior, and may merge with it. The muscle is similar in all the groups examined except the Cora- ciidae, Leptosomatidae, Galbulidae and Bucconidae. In these families, a narrow slip diverges from the anterior edge of the muscle (which is very narrow in Leptosomus) to insert around the ceratobranchiale-basihyal articulation. The insertion is in fact on the ventral surface of the anterior tip of the ceratobranchiale in all except Eurystomus and Leptosomus, in which attachment is on either side of the posterior ventral surface of the basihyal. Where M. stylo- hyoideus is absent (Coraciidae, Galbulidae, Bucconidae) this slip corresponds with it in origin, but is quite distinct in its completely ventral insertion, rather than a lateral or dorsal 390 P. J. K. BURTON one. However, the presence of both this slip and a normal M. stylohyoideus in Leptosomus makes it clear that the slip is not just a M. stylohyoideus with altered insertion. M. branchiomandibularis In all birds previously studied, M. branchiomandibularis ( = M. geniohyoideus) originates fleshily on the medial surface of the mandible just anterior to the venter externus of M. ptery- goideus ventralis medialis (if present), and division at the origin into an anterior and a posterior portion is common. This is also its origin in some of the families investigated here, but a modified form of origin (either far anterior on the mandible, or from the ventral surface of the buccal mucosa) is seen in others. In this modified form, the muscle is flattened in the horizontal plane, whereas in its typical condition, flattening is less pronounced and oriented nearer to the vertical plane. In all cases, insertion is fleshy on the distal part of the epibranchiale. The muscle is twisted around both ceratobranchiale and epibranchiale. M. branchiomandibularis is generally the largest tongue muscle, and in the Picidae reaches an enormous size due to the great elon- gation of the hyoid horns. Variations in the size and position of these horns are described by Steinbacher (1934 et seq.). ALCEDINIDAE. Alcedo. Origin entirely on the medial surface of the mandible and extending far forward nearly to the mandibular symphysis. There is no clear division into two parts. The hyoid horns, and therefore the area of insertion, are short. Other Alcedininae resemble Alcedo. In the Cerylinae, Megaceryle spp., Chloroceryle amazona and C. inda resemble Alcedo. Chloroceryle americana and C. aenea show great reduction of M. branchiomandibularis. The origin is limited to a small area on the ven- tral medial edge of the mandible, just anterior to the venter externus of M. pterygoideus ventralis medialis. In C. aenea the muscle is reduced to a tiny slip, and could almost be termed vestigial. In the majority of Daceloninae, origin on the mandible is limited to the anterior region near the symphysis, but there is additional origin from the ventral surface of the buccal mucosa. In these species, the origin of the muscle is flattened in the horizontal plane, rather than the more vertical one shown where the mandible is the main site of origin. In Pelargopsis, Tanysiptera and Lacedo, attachment appears to be mainly from the mandible, as in Alcedo; in Halcyon saurophaga it is almost entirely from the mucosa. TODIDAE. In three specimens of T. todus the origin is divided into two clear parts. The bulkier part is attached partly on the medial surface of the mandible, but mainly on the mucosa, extending well anterior. The smaller part is a slip running close to the midline, left and right sides being in contact in some specimens. This medial slip resembles an M. genioglossus, but is distinguished by its insertion on the epibranchiale, to which it runs medially and dorsally relative to the main part. In a fourth specimen, the origin is undivided and confined to the mucosa. MOMOTIDAE. Momotus. The origin of M. branchiomandibularis is not subdivided, and is mainly on the medial surface of the mandible, with a small amount of attachment to the mucosa anteriorly. The insertion is normal. In Baryphthengus, the origin is roughly half from the mandible and half from the mucosa. In Electron, the mucosa is the main site of origin, with only a small amount of attachment to the mandible. MEROPIDAE. Origin is largely on the mucosa, extending far anterior, with only a small area of attachment on the mandible. There is no obvious subdivision of the muscle. LEPTOSOMATIDAE. The origin of M. branchiomandibularis is of typical form, confined to the medial surface of the mandible, and showing clear division into anterior and posterior portion. CORACIIDAE. Origin mainly on the mucosa (entirely in Coracias), extending anteriorly nearly to the symphysis. No subdivision. FEEDING APPARATUS IN CORACIIFORMES & PICIFORMES 391 UPUPIDAE. Origin entirely on the mandible, but extending far anteriorly. No sub- division, and rather narrow. PHOENICULIDAE. Similar to the Upupidae. BUCEROTIDAE. Origin is from a bony ledge at the mandibular symphysis, and the muscle is narrow and flattened in the horizontal plane. In Tockus erythrorhynchus, the left and right muscles meet in the midline at the symphysis, lying superficial to the origin of M. genioglossus, which is also on this ledge. In Buceros bicornis and Bucorvus lead- beaten, the left and right muscles remain separate at the origin, and do not overlap M. genioglossus. GALBULIDAE. Origin resembles that in the Bucerotidae, being mainly from a medially situated bony ledge at the mandibular symphysis, though with some attachment also to the mucosa. The left and right muscles do not quite meet at the midline, although there is no. M. genioglossus. The gap is widest in Galbalcyrhynchus. The muscle is narrow and horizontally flattened. BUCCONIDAE. As in the Galbulidae, except that the narrow gap between right and left muscles is occupied by a well-developed M. genioglossus. CAPITONIDAE. M. branchiomandibularis originates entirely from the medial surface of the mandible, and consists of a single portion. INDICATORIDAE. As in the Capitonidae, but there is some subdivision into two portions. RAMPHASTIDAE. As in the Capitonidae. PICIDAE. Despite the enormous size of this muscle in the Picidae, its origin is basically a conventional one on the medial surface of the mandible, though extending from the anterior edge of M. pterygoideus ventralis medialis forward to the mandibular sym- physis. M. geniothyreoideus (Leiber \9Qla & b) is apparently a derivative of M. branchiomandibularis, originating anteriorly on the medial surface of the mandible, dorsal to M. branchiomandibularis, and inserting on the dorsal surface of the trachea, just posterior to the larynx. M. genioglossus This muscle is absent from many groups of birds, and where present, is often difficult to detect. Its origin is fleshy, or by a short aponeurosis, from the medial surface of the mandible, at or near the symphysis, and it runs along the ventral side of the buccal mucosa to insert in the region of the tongue base. It runs adjacent to a vein, and where poorly developed could well be confused with it. This vein is apparently a branch of the mandibular vein (Hughes 1934), joining it near the posterior end of the lower jaw. The vein appears to be present on the right side only in most of the families studied (both sides in Alcedo atthis; left side only in Halcyon chelicuti), but could not be traced in many specimens. M. genioglossus was found in eight of the fifteen families investigated here, as follows: MOMOTIDAE. Vestigial, left and right sides originating close together on the mandibular symphysis. It fades out short of the tongue base. MEROPIDAE. Similar to the Momotidae; best developed in M. pusillus. CORACIIDAE. Present in Brachypteracias spp., Attelornis and LJratelornis, and quite well developed. The origin is on the mandibular symphysis, where the right and left muscles unite. Insertion is on the ventral side of the mucosa lateral to the tongue base. UPUPIDAE. A single, slender median muscle arising at the symphysis, and fading out on the mucosa just short of the tongue base. The site of origin is the ventral surface of a well developed bony ledge, level with the dorsal edge of the mandible. PHOENICULIDAE. Origin is by a short aponeurosis from the symphysis, on a less distinctly marked bony ledge than in the Upupidae. The muscle is better developed and the right and left muscles unite only near the origin. Insertion is on the mucosa near the base of the tongue. BUCEROTIDAE. A single median muscle, but very well developed. Origin is at the sym- physis, from a more or less well marked bony ledge, shared also by M. geniohyoideus, 392 P. J. K. BURTON which lies ventral (superficial) to M. genioglossus in Tockus, and lateral to it in Bucorvus and Buceros. Insertion is on mucosa near the tongue base in Tockus, but around the posterior end of the basihyale in Bucorvus and Buceros. The muscle is particularly bulky in Bucorvus. BUCCONIDAE. Present throughout, left and right muscles having a combined origin on the symphysis, and insertion on the mucosa lateral to the tongue base. RAMPHASTIDAE. A vestigial median genioglossus was found in Selenidera, with origin at the symphysis, but fading out on the mucosa short of the tongue base. PICIDAE. M. geniothyreoideus, included here under M. branchiomandibularis could alternatively be a modified form of M. genioglossus — see discussion, p. 435. M. ceratoglossus In all birds so far studied, this muscle originates fleshily entirely on the dorso-lateral surface of the ceratobranchiale. It is a pinnate muscle whose short fibres insert on a lateral aponeur- osis. This merges anteriorly into a strong tendon inserting on a tubercle on the ventral side of the posterior end of the paraglossale. The dorso-medial surface of the muscle is concave, to fit the ceratobranchiale. In the families investigated here the extent of origin is the principal feature showing variation. The origin is from the entire length of the ceratobranchiale in the Todidae, Lepto- somatidae, Coraciidae, Bucerotidae, Galbulidae, Bucconidae and Picidae. The muscle stops more or less short of the anterior tip of the ceratobranchiale in the Alcedinidae, Momotidae, Upupidae and Phoeniculidae. In the remaining families there is more or less attachment also to the basihyal. In the Indicatoridae there is a slight amount of attachment to the basihyal just at its articulation with the ceratobranchiale. In the Capitonidae, there is attachment along the entire ventro-lateral surface of the basihyal except in Trachyphonus (posterior half of basihyal and ceratobranchiale only) and in Psilopogon (origin only on ceratobranchiale). In the Ramphastidae, origin includes the full length of the basihyal. M. ceratoglossus anterior and M. hypoglossus medialis The small muscles at the base of the tongue have given rise to considerable confusion in the literature. The definitions followed here are those employed by Burton (1974#). M. ceratoglossus anterior is the name given to a small group of fibres arising from the tendon of M. ceratoglossus just before its attachment to the ventral surface of the paraglossale. These fibres insert partly on the entoglossum, but mainly on a strong median aponeurosis which runs the length of the ventral surface of the tongue. M. hypoglossus medialis is an unpaired median muscle arising on the anterior tip of the basihyal and inserting on the median aponeurosis. Both muscles, together with their median aponeurosis, are absent in the majority of families studied here. M. ceratoglossus anterior is present in the Meropidae, in which it is very small, and the Leptosomatidae, in which it is well developed. In the Ramphastidae, fibres from the main part of the muscle extend onto the paraglossale, but there is no distinct ceratoglossus anterior, and no median aponeurosis. M. hypoglossus medialis is found in the Meropidae, Galbulidae and Bucconidae; it is well developed in the first, smaller in the second, and extremely small in the third. M. hypoglossus obliquus In all birds previously studied this muscle originates on or around the basihyale and inserts fleshily on the postero- lateral tip of the paraglossale. Two main forms occur; in Type 1 (Burton, 1974#), the right and left muscles merge in the midline, forming a continuous band of fibres which passes under the basihyale. In Type 2, the right and left muscles meet ven- trally, but remain separate. In Type 1, the muscle is generally shorter than in Type 2, and confined to the anterior part of the basihyale. Most of the Coraciiformes examined conform reasonably closely to one or other type. However, the Upupidae, Phoeniculidae and Piciformes (with the exception of the Galbulidae and Bucconidae) show a highly distinctive modification of Type 2, in which the right and FEEDING APPARATUS IN CORACIIFORMES & PICIFORMES 393 left sides are entirely separate not even meeting ventrally and originate far back on the basihyale or in some also on more or less of the ceratobranchiale. In this modified form, the muscle is long and narrow, reaching an extreme in the Indicatoridae and Picidae, with its fibres oriented at only a small angle to the basihyale. This contrasts with its normal form, which is short and stout, the fibres being oriented at a large angle to the basihyale — 90° in the case of the more anterior fibres. ALCEDINIDAE. The muscle is of Type 2, but rather poorly developed, so that right and left sides usually fail to meet at the ventral midline, and leave the anterior half of the short basihyale exposed. TODIDAE. Reasonably well developed, and of Type 1 , completely covering the anterior half of the basihyale. MOMOTIDAE. Type 2, completely covering the short basihyale. MEROPIDAE. Type 2, but short, covering slightly less than half of the rather long basihyale. LEPTOSOMATIDAE. Type 2, but short and rather small, covering slightly over half the short basihyale. CORACIIDAE. Type 2, the left and right sides just failing to meet, but otherwise covering all of the short basihyale. UPUPIDAE. The left and right muscles are completely separate, originating on the posterior basihyale, and the head of the ceratobranchiale; however the muscle is narrow and its total bulk is small, the basihyale being short. PHOENICULIDAE. The two sides are quite separate, and origin is on the posterior basi- hyale, and the ceratobranchiale anterior to the origin of M. ceratoglossus. The muscle is somewhat shorter in Rhinopomastus than in Phoeniculus. BUCEROTIDAE. The muscle is vestigial. The right and left muscles are separated by a wide gap, and their origin is confined to the anterior half of the basihyale. M. hypoglossus obliquus is completely concealed by the anterior fibres of the bulky M. ceratoglossus. GALBULIDAE. Type 2, covering the entire basihyale. BUCCONIDAE. Type 2, covering the entire basihyale. CAPITONIDAE. The muscle is of Type 2, but long and narrow, with origin well back on the posterior half of the basihyale. The origin extends onto the anterior tip of the cerato- branchiale in Pogoniulus. INDICATORIDAE. The muscle is very extensively developed, its origin including the posterior basihyale and the entire dorsal surface of the ceratobranchiale as far as its articulation with the epibranchiale. The origin is medial and dorsal to that of M. cerato- glossus. Insertion is by a strong, narrow dorsal aponeurosis, which extends back some distance into the origin, acting as the raphe of a bipinnate fibre arrangement. RAMPHASTIDAE. The muscle is long and narrow, with an extensive origin on the posterior basihyale and anterior ceratobranchiale. PICIDAE. From examination of Jynx torquilla, Dendrocopos major and Picumnus oliva- ceus, it seems quite clear that M. ceratoglossus superior, described by Leiber (1970# & b) in various Picidae, is in fact simply a very long M. hypoglossus obliquus. This point is considered in more detail in the discussion (pp. 4 1 7-4 1 8). M. tracheohyoideus and M. tracheolateralis (N.A.A. terminology: see text) These two muscles insert close together on the cricoid cartilage, and both run the length of the neck. M. tracheohyoideus arises on the sternum or clavicle, and runs lateral to M. tracheolateralis, adhering closely to the skin of the neck. M. tracheolateralis arises on the syrinx and runs along the ventro-lateral surface of the trachea. The site of origin of M. tracheohyoideus is the clavicle in the Alcedinidae, Todidae, Momotidae, Meropidae, Leptosomatidae, Coraciidae, Upupidae, Phoeniculidae, Galbulidae and Bucconidae. The origin is usually a fairly wide fleshy one along a narrow line in the medial half or third of the anterior edge of the clavicle and right and left muscles remain 394 P. J. K. BURTON separate. The origin is fleshy on the antero-ventral tip of the keel of the sternum or the clavicular symphysis in the Capitonidae, Indicatoridae, Ramphastidae and Picidae. This is generally a narrower origin than that on the clavicle and right and left muscles meet at the origin. In the Bucerotidae, both sites of origin are utilized, and the muscle forms an extensive sheet over most of the ventral side of the neck, closely adhering to the skin. In most of the families considered here, M. tracheohyoideus inserts fleshily on the medial surface of the ceratohyal, near its anterior end, and on the dorso- lateral surface of the cricoid between the two heads of origin of M. thyrohyoideus. In the Indicatoridae, attachment to the ceratohyal is slight, and in the Capitonidae (except Psilopogon) and Ramphastidae it is absent altogether. The complex insertions of M. tracheohyoideus in the Picidae are thoroughly described by Leiber (1907a & b). (According to Baumel et al, 1979 (N.A.A., Systema respiratorium, Note 45), the muscle originating on the clavicle should be called M. cleidohyoideus and that arising on the sternum should be M. sternohyoideus. However, the evidence of this study (particularly the condition in the Bucerotidae) seems to indicate that both are derived conditions of the same muscle, and the common term of M. tracheohyoideus for both has been given preference. See also Part 3, p. 4 18). M. tracheolateralis inserts on the ventral surface of the cricoid, medial to the ventral origin of M. thyrohyoideus. It appears to be absent in Coracias. In the Upupidae, M. tracheolater- alis merges with M. tracheohyoideus, and they insert together on the ventral cricoid and the anterior tip of the ceratohyal. M. thyrohyoideus (N.A.A.: M. cricohyoideus) This muscle originates from the cricoid cartilage by ventral and dorsal heads. The ventral head arises from the ventral surface of the cartilage, and the dorsal head from its lateral sur- face. Insertion is on the dorso-lateral surface of the anterior end of the basihyale — typically medial and slightly posterior to the insertion of M. stylohyoideus if present. Little variation was encountered among the forms studied here. The neck and its musculature The most important general account of the avian neck is that by Boas (1929), providing the foundation for all subsequent studies. The account of passerine cervical anatomy by Palmgren (1949) is also of great value. Detailed studies of small groups of species have been made by Zusi (1962), Zusi & Storer (1969) and Burton (19746). While this paper was nearing completion, a detailed account for several Picidae was published by Jenni (1981). Cervical vertebrae Despite the wide range of feeding apparatus modifications shown by the head in the two orders, the cervical vertebrae are relatively uniform in their major features. Denned as ver- tebrae which bear either no movable ribs, or movable ribs not articulating with the sternum, there are either fourteen or fifteen (Beddard, 1898). The cervical vertebrae of birds fall into three fairly well defined sections, with different functional properties (Boas, 1929). Section I (the most anterior) can only be flexed downward, Section II only upward, and Section II. mainly downward, though anteriorly upwards as well. In the Coraciiformes and Piciformes, as in most birds studied, Section I usually consists of the first five vertebrae, 5 often appearing transitional, as in the Picidae; Boas (1929) gives a first section count of four for Picus viridis. Section II comprises vertebrae 6 to 9 or 10, and Section II the remainder. Descriptions of the morphological features which control the range of bending are given by Zusi & Storer (1969); although referring to a grebe (Podilymbus) possessing several more vertebrae than Coraciiform or Piciform birds, the structural modifications concerned are essentially similar. Some noteworthy variations among the Coraciiformes and Piciformes may now be sum- marized. In Segment I, the first vertebra (the 'atlas') is, as in all birds, a simple ring of bone without a neural spine; in the Bucerotidae, it is fused with 2. Vertebrae 2 and 3 invariably FEEDING APPARATUS IN CORACIIFORMES & PICIFORMES 395 splc spla res spin asc co 6-8 5mm Fig. 31 Alcedo atthis. Superficial anterior cervical musculature. M. rectus capitis lateralis and ventralis and M. biventer cervics not shown. M. complexus freed from cranium and turned aside. bear tall neural spines, and so often, does 4. Neural spines are continued further back, in some cases, and extend to 6 or 7 in the Alcedinidae, Phoeniculidae, some Capitonidae and Picidae. The spines normally decrease evenly in height from front to back, though in some woodpeckers there is actually an increase from 2 to 4, and then a steady reduction in height. A series of fused ribs is generally present from 3 or 4 onwards (5 in Leptosomus and Upupa), but these are reduced in the Bucerotidae and absent in the Picidae. Some anterior vertebrae may possess bony struts connecting the transverse process with the dorso- lateral crest, or the costal process with the lateral crest, or both. Exactly which vertebrae varies widely; the dorsal strut occurs most commonly on 3 and 4, the ventral on 5, 6 and 7. Only two species examined (Phoeniculus purpureus, Colaptes auratus) had a dorsal strut on 2, and only two had ventral struts posterior to 7 (Phoeniculus purpureus, on 8; Upupa epops on 8, 9 and 10). Dorsal and ventral struts occur together, if at all, usually only on 4, but Nystalus chacuru is remarkable for having both together on 5, 6 and 7. Hypapophyses occur anteriorly and posteriorly, but are absent from the middle region of the neck. The anterior ones are usually found on vertebrae 2 to 4, though often on 2 to 3 only. That on 2 is often distinctly bifurcate, the two projections pointing backwards. Those on 3 and 4 may have a similar, but less pronounced shape. In the Picinae, very deep hypapo- physes are found on 2 to 5. The ventral side of the atlas (1) also bears a hypapophysis like process, which is frequently bifurcate posteriorly. This provides insertion for M. flexor colli brevis, and is particularly strongly developed in the Picinae. Posteriorly, hypapophyses extend back into the thoracic region, starting usually on 9 or 10, but on 8 in Jynx, 1 1 in Nyctiornis, Upupa and Selenidera, and on 12 in Tockus. In the Picidae (other than Jynx), the subvertebral bridge which is present from 5 or 6 rearwards shows some indication of a hypapophysis on all vertebrae, but this is only well denned on 10 (the last vertebra with such a bridge) and vertebrae posterior to it. The subvertebral bridge in the Picidae has long been known (see, e.g. Beddard, 1898), but has apparently not been noted in other Piciform families. However, there is a complete subvertebral bridge on 8 in Indicator, with an incomplete bridge also on 6, 7 and 9. In Jynx, a subvertebral bridge is present only on 8 and 9. 396 P. J. K. BURTON Cervical muscles Dissection of neck muscles is arduous and time consuming, and it was found impractical to examine the full range of species used in other parts of this study. The species actually investigated are referred to in appropriate places in the text, and group names are used only where an adequate range of species has in fact been dissected. For similar reasons, three muscles (Mm. splenius accessorius, Mm. intercristales and Mm. interspinales) have been omitted, while for some others (M. longus colli ventralis, Mm. intertransversarii and Mm. inclusi) the level of detail studied has been deliberately limited. M. biventer cervicis Origin is from the dorsal aponeurosis of M. spinalis in the region of 13, and the muscle inserts on the dorsomedial edge of the occipital deep to M. complexus. As its name implies, this muscle has two fleshy bellies, at origin and insertion, while the middle region of the muscle consists of a flat strap like tendon. The length of this tendon relative to the fleshy bellies varies considerably. The muscle is best developed in the Alcedinidae, in which the muscle shows a further peculiarity first noticed by Cunningham (1870), in Ceryle. This is a short, strong aponeurosis linking the right and left muscles, near the posterior end of the anterior bellies. This feature was noted in various other kingfishers by Beddard (1896), and has been further checked in a wide range of other species for the present study. The transverse apo- neurosis has been found throughout the Cerylinae, but in none of the Alcedininae examined. Within the Daceloninae, it was found in Tanysiptera, Cittura and Halcyon sancta ( = Sauro- patis vagans) by Beddard, and in the present study also in Melidora; it is evidently absent in most, perhaps all, other members of the subfamily. M. biventer is also well developed in the Indicatoridae and Jynx. A weak biventer with short fleshy bellies and a long tendon is seen particularly in the Upupidae, Phoeniculidae, Meropidae, Galbulidae and some Picidae. The anterior belly usually meets or lies close to its contralateral partner in the midline at the insertion, but right and left are separated by a wide gap in the Galbulidae, Meropidae (except Nyctiornis) and Picumninae. M. spinalis cervicis and M. splenius colli (N.A.A.: M. longus colli dorsalis) M. spinalis arises from the neural spines of vertebrae 14 to 18, and inserts by a series of slips on the anapophyses of 2, and of 6 to 13 (5 to 13 in Malacoptila panamensis). M. sple- nius colli has no independent insertion, but consists of a series of slips joining the slip of M. spinalis inserting on 2. There are from three to five of these slips, arising from the lateral surface of the neural spines of successive vertebrae, the most anterior in all cases being 4. M. splenius capitis In most species examined, this muscle arises from the neural spine of 2, and inserts on the posterior surface of the skull deep to M. biventer and M. complexus. In Phoeniculus pur- pureus and Tockus erythrorhynchus there was additional origin from the neural spine of 3. Right and left muscles generally lie close together in the midline, but a wide gap exists in Melittophagus pusillus. The muscle is relatively very large in the Galbulidae. Cruciform structure of M. splenius capitis, as described by Burton (1971) is clearly evident in Nonnula ruficapilla, but barely discernible in most other species examined. Mm. pygmaei (N.A.A.: M. longus colli dorsalis, pars profundd) Mm. pygmaei are small, weak muscles, originating from the neural arches of vertebrae between 1 1 and 8, and inserting on the lateral edges of the transverse-oblique crests of vertebrae which are most commonly the second ones anterior to the vertebrae of origin. Asymmetrical distribution of Mm. pygmaei is frequent, and they are completely absent in most of the species examined here. They were found only in some species of Meropidae, Coraciidae, Leptosomatidae, Galbulidae and Bucconidae. Their occurrence in single specimens of various species in these families is as follows: MEROPIDAE: Melittophagus pusillus. Left side, from 8 to 6 and 9 to 7; right side, from 8 to 6, 9 to 7, 10 to 8 and 1 1 to 9. FEEDING APPARATUS IN CORACIIFORMES & PICIFORMES 397 Nyctiornis amicta. Absent. LEPTOSOMATIDAE: Leptosomus discolor. Left side, from 9 to 7 and from 10 to 8; right side, from 8 to 6. CORACIIDAE: Coracias benghalensis. On both sides, from 8 to 6, 9 to 7 and 10 to 7. Eurystomus orientalis. From 9 to 7 on the right; absent on the left. GALBULIDAE: Galbula ruficauda. Left side, from 9 to 6; right side from 9 to 6 and from 8 to 6. Galbalcyrhynchus leucotis. Present on the left side only, from 9 to 6 and from 8 to 6. BUCCONIDAE: Malacoptila panamensis. From 8 to 6 and from 9 to 7 on both sides. Absent in Notharcus macrorhynchos, Chelidoptera tenebrosa and Nonnula ruficapilla. Mm. ascendentes cervicis (N.A.A.: M. cervicalis ascendens) Originating on the diapophyses of all cervical vertebrae except 1 to 5 these muscles each consist of two slips inserting on the anapophyses of the second and third vertebrae anterior to the origin. Thus, the most anterior muscle, originating on 6, inserts on 3 and 4. The series continues posterior to the neck as Mm. ascendentes thoracicis. No noteworthy variation was discovered among the species dissected. M. longus colli ventralis This large and complex muscle extends along most of the ventral side of the neck. The main part arises by a series of fleshy slips, starting in the thoracic region, from the sublateral pro- cesses, hypapophyses and the anterior region of the centra of successive vertebrae. Insertion is made by long tendons attaching on the cervical ribs; each vertebra sends a slip to join each of the tendons traversing it. In the anterior region of the neck, shorter slips arise from costal processes or sublateral processes, inserting tendinously on ribs or hypapophyses. Some of these anterior slips should perhaps be treated as M. flexor colli profundus; however, their siting and arrangement is so variable among the species studied that it seems impracticable to distinguish this as a separate muscle from M. longus colli. Complete dissection of M. longus colli to enumerate the siting and arrangement of all slips throughout the study species, would be extremely time consuming. For the purposes of the present study, a more general examination was considered sufficient. The thoracic region was excluded, but particular attention was paid to the short anterior slips. The main features revealed are* as follows: In all species, the most anterior insertion is on the hypapophysis of 2. Insertion on 3 is also on the hypapophysis, but on 4, insertion on the hypapophysis was found only in Todus, Leptosomus, and the representative species of Meropidae, Buccon- idae. Galbulidae and Picidae; in the remainder, attachment on 4 is to the rib, or costal process. The slips attaching on 2, 3 and 4 are, in most species, short ones, arising typically from the sublateral process of 6, though the slip to 2 arises from the costal process of 5 in Upupa, and that to 3 from the sublateral process of 7 in Phoeniculus. The main part of the muscle, (consisting of long tendons and their associated fleshy slips) has its most anterior attachment typically on the rib of 5. In the Bucconidae and Galbulidae, long slips attach only as far forward as 6, but extend to 3 in Upupa and Todus, and to 2 in Alcedo, Indicator, Jynx and Trachyphonus. In Momotus, long slips attach on 2 and 4, but not on 3, while in the Picidae (other than Jynx), the entire muscle consists of long and very strongly tendinous slips attaching forward to 2. M. flexor colli brevis (N.A.A.: M. flexor colli lateralis) The bulk of this muscle originates from the lateral strut of 3 and the lateral processes of 4 and 5; origin on 5 is absent in Alcedo atthis and Malacoptila panamensis, while in Campe- philus there is additional origin from the lateral strut of 6. The muscle also includes a smaller, medial portion, originating ventrally (typically on sublateral processes) and separated from the main body of the muscle by anterior slips of M. longus colli. (This portion may well represent a M. flexor colli profundus, merged with M.f.c. brevis; otherwise this muscle is 398 P. J. K. BURTON absent in the species studied). This medial portion appears to be absent in the pecies of Alcedinidae, Upupidae, Phoeniculidae, Bucerotidae, Bucconidae and Galbulide dissected, and in some Picidae (Sphyrapicus, Sasid). It is present in the remainder, but varies consider- ably in development and in its relation to adjacent muscles (especially M. longus colli). In the species examined, it is best developed in Trachyphonus darnaudii, in which it actually originates posterior to the lateral portion, on the sublateral process of 6. In other species possessing it, the vertebrae of origin are the same as the lateral part except in Momotus, in which the very short medial portion originates from the postlateral process of 3. Insertion is on the hypapophysis like process on the ventral side of the atlas (processus latus). M. complexus The sites of origin of this muscle vary considerably among the families investigated. No single species utilizes all the possible sites of origin, which are the anapophysis of 3; the transverse process and lateral strut of 4, and the diapophyses of 5, 6, 7, 8 and 9. Insertion of M. complexus is on the dorsal edge of the occipitals, superficial to M. biventer. The muscle is best developed in the Alcedinidae, in which its origins have been checked in a wide range of species. Origin is from 5, 6 and 7 in the Cerylinae and Daceloninae, and from 8 as well in the Alcedininae; attachment to 9 was found in a single specimen ofAlcedo atthis. Origin of M. complexus posterior to 7 has not previously been reported in any bird. M. complexus is poorly developed in the Picinae (notably Campephilus) and in Indicator. M. rectus capitis superior (N.A.A.: M. rectus capitis dorsalis) This muscle lies superficial and anterior to M. flexor colli brevis, which it much resembles; the two could well be considered as divisions of a single muscle. Origin is from the lateral surface of neural arch 1, from the anterior surface of anapophyses 2 and 3, and in some species from the lateral strut of 4. Insertion is on the posterior edge of the basitemporal plate. M. rectus capitis lateralis This muscle inserts on the lateral edge of the exoccipital usually immediately lateral to M. complexus. Origin is commonly from the hypapophyses of 2, 3 and 4, and in Leptosomus and Upupa from 5 as well. However, the number of sites of origin is reduced in many species; 3 and 4 only are occupied in the Cerylinae and Coracias, and 2 and 3 only in the Todidae, Meropidae, Eurystomus and Jacamerops. Except for Jacamerops, reduction is even more marked in the Galbulidae and Bucconidae, with origin only from 3 in the species available for dissection. The muscle shows a corresponding reduction in bulk in these two families, and in Notharcus macrorhynchos and Galbalcyrhynchus leucotis it can fairly be described as vestigial. M. rectus capitis ventralis Origin is from the ventral surface of 1, the hypapophyses of 2, 3, 4 and 5, and commonly also the sublateral process of 6. The latter point of origin is lacking in the Todidae, Momot- idae, Meropidae, Phoeniculidae, Bucerotidae, Capitonidae and Ramphastidae. Insertion is on the basitemporal plate, anterior to M. rectus capitis superior. Both this muscle and M.r.c. lateralis are notably bulky in the Picinae, the very deep hypapophyses providing an increased area for their origin. Mm. intertransversarii and Mm. inclusi The Mm. intertransversarii connect successive vertebrae, running in most cases between their lateral processes. In the middle of their range, they are of complex, multipinnate struc- ture (see .especially Zusi & Storer, 1969). Mm. inclusi are similarly situated, but lie deep to Mm. intertransversarii, and are mostly divided into dorsal and ventral parts (Mm. inclusi superiores and inferiores). They end anteriorly one or two vertebrae before the Mm. inter- transversarii. The most anterior M. intertransversarius in Coraciiform and Piciform birds is usually that between 5 and 4; in many species, lateral fibres from 6 also reach vertebra 4. The anterior limit is vertebra 3 in the Capitonidae, Picidae and Coraciidae. FEEDING APPARATUS IN CORACIIFORMES & PICIFORMES 399 PARTS Functions and evolution Bill and skull The functional significance of bill form in birds is sometimes obvious, but more often rather hard to understand, even when it is distinctive or unusual. Bill modifications in the Coracii- formes and Piciformes are discussed at family level in the Systematic review; problems examined in this section concern more fundamental aspects of jaw action and cranial morphology. Ideally it should be possible to integrate information on osteology and arthrology with that on muscle action to produce a complete picture of jaw actions and the ways in which they are controlled. In practice, however, this would require a very full understanding of the forces developed by muscles and their sequence of action, and of the physical properties of joints and ligaments. So far, the only analyses of avian jaw action which approach this ideal standard are those made for the Mallard (Anas platyrhychos) by Zweers (1974, 1977). Work of this type, involving in vivo experimentation and electro- myography was beyond the scope of the present investigation, but the discussion which follows will at least serve to point out problems which would repay more intensive study. Overall form Barnikol (1952), following earlier work by von Kripp (1935) distinguished two general types of avian skulls which he called 'streckschadel' (stretched skull) and 'knickschadel' (hooked skull)*. He defined them in terms of two main criteria. In the 'streckschadel' type, the fora- men magnum is situated at the back of the skull, and the brain axis makes a relatively small angle with the bill axis; in the 'knickschadel', the foramen magnum is situated more ven- trally, and the brain axis makes a larger angle with the bill axes. Barnikol's examples of the former include Cygnus, Ixobrychus and Phalacrocorax; to them might well have been added various other members of their orders, as well as most of the Gaviiformes, Podicipediformes and some Procellariiformes and Charadriiformes. They are predominantly relatively primi- tive types specialized for aquatic feeding. The 'knickschadel' is much more widespread; Barnikol selects Strix, Opisthocomus and Scolopax as examples. He proposed, essentially, that the 'streckschadel' was a relatively archaic skull form which had become exagerrated for functional reasons in certain types, particularly fish eaters. He suggested, furthermore, that positioning of M. adductor mandibulae was an important underlying factor; evolution of the extreme 'streckschadel' allows the external temporal region of the muscle to be sited vertically above its points of attachment on the lower jaw. In the examples he chose, increas- ingly vertical attachments of M. add.mand.ext. are correlated with an increasingly vertical orientation of the postorbital ligament, which is finally sited slanting backwards to the mandible in Phalacrocorax. The Coraciiformes and Piciformes are more advanced birds than any of Barnikol's 'strecks- chadel' types and the majority have skulls conforming fairly well to 'knickschadel' mor- phology. Nevertheless, there are significant variations in the orientation of major skull components, and these deserve to be considered in more detail. Among this assemblage, the Alcedinidae stand out as the clearest example of modification towards 'streckschadel' form, with skulls stretched out along the bill axis, and the palatal apparatus lying almost in the same plane as the tomia of the closed bill. However, difficulties are encountered when Barnikol's criteria are applied. By comparison with, for example, a Megalaima barbet, Ceryle has M. add.mand.ext. and the postorbital ligament less vertically oriented, while the situation of the foramen magnum scarcely differs. The elongated form of the Ceryle skull *In discussions throughout this paper I shall use these terms in their German form, to indicate their origin clearly, to avoid possible ambiguities inherent in their English translations, and because they are less clumsy. 400 P. J. K. BURTON is, instead more clearly manifested in the low orbit and cranium, and in the arrangement of the kinetic apparatus, with the quadrates sited far back, behind the orbit. In ventral view, the pterygoids of Ceryle enclose a considerably smaller angle posteriorly than either Megalaima or any other bird included in this study. It seems significant that the only members of the Coraciiform/Piciform assemblage to clearly show 'streckschadel' features of skull morphology are the one family whose members include fish eaters. Moreover, these features are much more strongly marked in the more thoroughly piscivorous Cerylinae and Alcedininae than in the Daceloninae. The functions and biological roles of 'streckschadel' modifications deserve much more study, but in the present instance, the following suggestions can be made: (a) The combination of low orbits with a palate and bill oriented in a straight line brings the line of vision close to the primary axis of orientation of the bill (see Bock 1964: 28). This is probably of value in minimizing parallax error in a situation which is already compli- cated by refraction between air and water, and which may require great speed and accuracy in countering evasive action by prey. (b) The skull increases in width from front to back; hence, by siting the quadrates, and thus the mandible articulation as far back as possible, the gap between the posterior ends of the mandibular rami is maximized. This may be an adaptation to swallowing large whole fish, since the rami themselves seem to have only a limited capacity for outward bowing. The long flat palate may also be helpful in this regard, since fish are long food items with less capacity for bending than much of the large prey, e.g. insects, frogs, taken by other Coraciiformes. (c) The gradual, even taper of jaws and skull which is achieved through these modifica- tions may be of some value in streamlining, to permit efficient movement through water. The posteriorly situated quadrates of Megalaima mentioned above, are a rather unusual feature among the Coraciiform/Piciform assemblage, though characteristic of many Capito- nidae and the Ramphastidae as well as Kingfishers. As in the Alcedinidae, this may well be an adaptation for swallowing large food items, in this case fruits, by increasing the gap between the mandibular rami. The accompanying feature of palatines, pterygoids and jugals lying more or less in the same plane as the tomia of the closed bill is more widespread, though only in the Alcedinidae does this involve the 'streckschadel' feature of a reduced angle between the pterygoids. It is seen particularly among the Coraciiformes, other than the Upu- pidae, Phoeniculidae and Bucerotidae, and in the Galbuloidea. In the Todidae, Galbulidae and still more the Meropidae, this is combined with a high cranium and very ventrally situ- ated foramen magnum to produce skulls with an extraordinary profile, the front of the skull sloping down sharply to a low and rather flattened upper jaw base. This shape, and the position of the foramen magnum, is perhaps related to perching posture, in which the bill is normally tilted up at a considerable angle as the bird scans the sky for prey. Viewed dorsally, considerable variation is seen in the extent to which the skull narrows from posterior to anterior. Narrowing in front of the orbits must be connected largely with the extent of forward vision. The most pronounced narrowing by far is seen in the Phoenicu- lidae, and is probably connected with feeding by gaping. Lorenz (1949) pointed out that pass- erine 'gapers' in the families Sturnidae and Icteridae look forward between the mandibles as they part them when feeding, and Beecher (1953a) showed that their ectethmoids are notched to increase their capacity for forward vision. It is puzzling that the Upupidae do not also have the skull narrowed in front of the orbits, since they feed in a similar manner, but on the ground. Perhaps the substrates in which they gape impose forces necessitating a more broadly based upper jaw. The skull is also quite markedly narrowed in front of the orbits in the Indicatoridae, Jynx and many woodpeckers, again probably enabling them to look forward at prey or a substrate at close range. Moderate narrowing from posterior to anterior is seen in the Meropidae and Galbulidae, but in this case probably reflects more the distinct broadening of the posterior cranium in these families, a feature believed to be related to neck muscle action (q.v.). Most FEEDING APPARATUS IN CORACIIFORMES & PICIFORMES 401 other families examined have skulls which are relatively very broad anterior to the orbits, and this is in most cases undoubtedly related to the need for a large gape. In the case of the Alcedinidae this creates a problem, since as explained earlier, one function of their 'streckschadel' skulls seems to be to bring the line of vision as close as possible to the primary axis of orientation of the bill; the advantage of this adaptation would seem somewhat offset by the resulting reduction of forward vision. A partial compensation may be provided by the emarginated shape of the lacrimal. Kinesis The mechanics and functional significance of cranial kinesis in birds have been discussed at length by Bock (1964), who distinguishes two principal mechanisms. In uncoupled kinesis, the movements of the jaws are entirely independent; in coupled kinesis, depression of the lower jaw cannot take place without simultaneous raising of the upper. The principal struc- ture on which coupled kinesis depends is the postorbital ligament, which is present in the majority of birds. An alternative way of achieving coupling is by means of an interlocking device at the quadrate/mandible hinge. This arrangement is uncommon, occurring sporadi- cally amongst several unrelated groups of birds; these are listed by Bock (1964), and to this list must now be added Bucorvus and Megalaima (see p. 29-30). Even where a postorbital ligament is present, uncoupled kinesis is possible when the liga- ment is unloaded (slack), a condition which occurs when M. protractor is contracting ahead of M. depressor. The mechanisms involved, and the variety of possibilities between strict coupling and total independence, are discussed by Zusi (1967), while Zweers (1974) provides detailed analysis of a closely coupled system in the Mallard. Some independent jaw action is even possible in forms with articulation lock coupling, such as Bucorvus and Megalaima, through spreading of the lower jaw rami. A critical point concerns the extent of retraction possible for the upper jaw in birds with coupled kinesis. Bock suggested that retraction past the normal closed position was impossible for birds possessing a functional postorbital liga- ment, but Zusi (1967) musters observational evidence to the contrary. Certainly at least some Coraciiformes and Piciformes are capable of this feat (see below). Presumably there is some relation between the extent of development of a postorbital ligament or articulation lock, and the extent and frequency of independent jaw action, but the relation is not a simple one, as indicated by the distribution of these features in families considered here. However, the possible biological roles of these mechanisms among Coraciiform and Piciform birds will be briefly reviewed as far as possible. (a) COUPLED KINESIS. A closely coupled system appears particularly suitable where rapid jaw action is needed (Bock, 1 964), and would seem particularly important for those species, mainly Coraciiform, which capture agile or aerial prey. The required acceleration must be achieved with minimum sacrifice of power, since the prey may need to be gripped strongly once captured — unless the initial snap of the jaws has sufficed to kill it, as must often be the case with small insects. Kinetic coupling may also be of special importance where force- ful protraction of the upper jaw is required, in species which forage by 'gaping', i.e., inserting the jaws into the substrate and then opening them. Coupling in this situation enables the force of M. depressor mandibulae to be added to that of M. protractor (see discussions by Zusi, 1959, 1967; Manger Cats-Kuenen, 1961; Bock, 1964); this is of greatest importance in the Upupidae, Phoeniculidae and some Bucerotidae. Bucorvus, which possesses a well developed articulation-lock device obtains most of its food on the surface, but probably requires closely coupled jaw action for swift prey seizure. The significance of the weak articu- lation lock in Megalaima spp. is more puzzling, especially as they may have a significant need for independent jaw action in some circumstances (see below). An aspect of coupled kinesis which has not received enough attention is the role of the musculature in guidance. This emerges very clearly from the work of Zweers (1974) on the Mallard, Anas platyrhynchos. The very strongly coupled kinetic system of this bird depends for its effectiveness on the action of the internal adductors in guiding the movements of quad- 402 P. J. K. BURTON rate, pterygoids and palatines. Exactly how the guiding and power roles of the jaw muscula- ture would be fulfilled among the varied members of the assemblage studied here can scarcely be surmized at present. An understanding of this point would require much more precise information of their jaw movements than is at present available — a level of study possible only under controlled conditions in the laboratory. (b) UNCOUPLED KINESIS. The independent jaw action made possible by uncoupled kinesis is important where skilful manipulation of food is at a premium, since upper and lower jaws are thus enabled to vary the angle and position in which they meet one another. A particu- larly valuable faculty in some birds is that of depressing the upper jaw past the normal closed position, with the lower jaw itself still partly depressed. The jaws can then be brought paral- lel, or even opposed with their tips closer together than the basal region. In bird photographs, evidence of retraction past the normal closed position is occasionally seen, when the tip of the lower jaw appears to extend beyond that of the upper, despite clearly undamaged bill tips. Uncoupled kinesis is likely to be most needed when dealing with slippery or awkwardly shaped prey, or when several objects need to be gripped along the length of the bill. Thus, the weak development of the postorbital ligament in the Alcedininae and Cerylinae may be related to the need for skilful manipulation when dealing with fish. Retraction past the closed position is of limited extent in kingfishers, due to an effective retraction stop (see below), but apparently does occur; in a photograph of Alcedo atthis by Massny (1977) the lower jaw tip extends appreciably further forward than the upper. Among the Piciformes, similar problems may arise in dealing with large numbers of prey. Photographs of Jynx with a mass of ants held in the bill (Koffan, 1960) demonstrate independent jaw action excellently. The photograph shows the upper jaw tip retracted far behind the lower in a most convincing way. Note that the diagram of Sphyrapicus with the lower jaw projecting while closed in Spring (1965) must be of the bird with a damaged bill which the author refers to on p. 486; the position shown could not be attained by any normal jaw action. Barbets holding several fruits at once in the bill seem to face a similar problem (photograph in Thomson, 1964), but although most lack a postorbital ligament, they are not able to retract past the closed position, as further retraction is halted by a bony stop (see below). It is puzzling, though, that some Capitonidae have evolved a degree of articulation coupling, which has the same effect as a postorbital ligament, and for that matter, that the Ramphastidae have not. Safety devices During the processes of feeding and nest excavation, the skull is subjected to a variety of forces which have a profound influence on its morphology. Under the general heading of safety devices are included various osteological or arthrological features which serve to confine jaw and palate movements within safe limits, or otherwise to withstand potentially hazardous forces. Constraints, on movement are basically of two kinds; 'stops' are devices which limit the range of normal movement, especially that of kinesis, while guides and braces control direction and prevent abnormal movement caused by interaction with prey or objects in the environment. Stops are to be regarded as the final, ultimate limit on movement; in practice, the muscles and ligaments involved will normally arrest a movement before these extremes are reached. A review of kinetic stops, principally in waterfowl, is given by Fisher (1955). As will be seen below, special modifications are present in woodpeckers to sustain forces incurred during excavation for food. It is interesting to note in passing that many other mem- bers of the Coraciiformes and Piciformes excavate wood or other hard substrates to create nest sites, yet lack special adaptations for doing so; even the frail skulls of the Todidae can cope with burrowing (Kepler, 1977). Presumably this is because nest excavation can be spread over a long enough time to allow even rather feeble excavating actions to suffice. Thus, the requirements of feeding remain the overriding force determining the architecture of skull and jaws. FEEDING APPARATUS IN CORACIIFORMES & PICIFORMES 403 Protraction stops A potential stop on protraction in nearly all species is provided by the orbital process of the quadrate; when this touches the wall of the orbit, all forward movement of the palate must inevitably cease. How significant this is in practice is hard to judge. Some birds, such as typical Larids (Zusi, 1962), Corvus (Fisher, 1955) and some hornbills (present study) have a swelling on the orbital wall which would make early contact with the orbital process. In other birds, (notably the Alcedinidae in the present study), the position of the orbital process, far from the orbital wall, would render it quite useless as a stop. In between are a large number of birds in which the orbital process could possibly act in this way, but which lack special modifications. It is quite certain, however, that the orbital process cannot act as a stop in species which have M. protractor quadrati (q.v.) attached along its full length. In this connection it is interesting to compare the Upupidae and Phoeniculidae with the Picidae (except Jynx). All three families have M. protractor greatly enlarged, and would seem to run an appreciable risk of dangerous over- protraction in the course of foraging. In the Upupidae and Phoeniculidae, the orbital process is free; it could well serve as a protraction stop, particularly in Rhinopomastus. These two families show no other obvious limitation on protraction, and it is possible that during the relatively slow action of 'gaping' there is little danger of sudden uncontrolled increases in this force. By contrast, in the Picidae, M. protractor quadrati has attachment along the full length of the orbital process, and although it lies close to the orbital wall, it hardly seems possible for the two to touch, and thus stop protraction. Nevertheless, woodpeckers are at consider- able risk. Hammering is assumed (Spring, 1965) normally to generate retraction forces on the bill, opposed by M. protractor, but a slight error, or unexpected irregularity in the wood might well add suddenly to protraction, instead of counteracting it. To safeguard against such eventualities, woodpeckers have an alternative stop mechanism in the form of a fronto-nasal hinge in which the frontal bulges out to overlap the upper jaw. This is not equally developed in all woodpeckers; Burt (1930) has demonstrated in a range of species that increasing devel- opment of this feature is correlated with increased size of M. protractor, decreased kinesis and greater dependence on hammering as a foraging method. Zusi (1962) has proposed a rather similar stop mechanism in Rhynchops, and the form of the fronto-nasal hinge suggests the same principle in various Strigidae and a few Falconidae. [The greatly developed casque of many Bucerotidae provides an alternative stop mechanism in the region of the fronto-nasal hinge, though this may not be its primary function; see Manger Cats-Kuenen, 1961]. Retraction stops The fundamental stop on retraction occurs when upper and lower jaws meet; in addition, for most birds studied here, contact between the nasal bar and the dorsal anterior rim of the orbit (formed by lacrimal, ectethmoid or frontal) provides a further stop. This latter stop is discussed at length by Cracraft (1968), who points out that no locking mechanism exists which would eliminate muscular effort in retraction — a function of stops which had been proposed by Fisher (1955). In any case it is doubtful if such a device would provide any significant advantage, for reasons discussed elsewhere (Burton, 1978). The lacrimal- ectethmoid complex provides the firmest stop when a large ectethmoid forms or supports the stop. This condition is best realized here in the Upupidae, Phoeniculidae, Bucerotidae and Piciformes (other than Galbulidae and Bucconidae). The lacrimal forms the stop in the remainder except for the Momotidae which lack it, and rely on the frontal; even where it is much enlarged, as in the Alcedinidae and Coraciidae, it probably always produces a more resilient stop than the ectethmoid. In the Picidae, an additional stop is provided for many species by the very long anterior process of the opisthotic, which meets and articulates with the quadrate. Cracraft questions why retraction stops are necessary at all, since the lower jaw would appear sufficient. He suggests that external forces encountered during feeding may produce unusual movements which require this safeguard. Among the families considered here, large 404 P. J. K. BURTON ectethmoids and firm retraction stops occur mainly in birds which excavate for their food (gapers and hammerers) and in fruit eaters; the latter may perhaps incur hazardous external forces during the wrenching movements sometimes needed to detach fruits from trees. Another situation in which retraction stops may be needed, not considered by Cracraft, would occur in uncoupled kinesis if the upper jaw is retracted past the normal closed position while the mandible remains partly depressed. In the Alcedinidae, this apparently occurs to a moderate extent, and the stop, formed by a large lacrimal, is indeed effective in bringing further retraction to a halt. It is interesting to note, though, that in a specimen of Ceryle alcyon with missing lacrimals, retraction much more extensive than could conceivably occur in life was possible without damage. Still more surprising, in Jynx, which regularly uses extensive retraction past the closed position, there is apparently no functional stop other than the limit imposed by the structure of the articulation of the quadrate and cranium. The fronto-nasal hinge here is simply flexible enough to sustain extensive retraction. Experiments with a freshly dead bird indicate a figure of approx. 20°-25° for the maximum retraction obtainable by muscle action, but by manipulation this could be extended to 55° without damage. Barbets and toucans mostly lack a postorbital ligament, and presumably have uncoupled kinesis, at least where there is no articulation coupling. Despite this, they appar- ently lack the capacity for retraction past the closed position, with retraction halting totally when the posterior medial surface of the ventral bar contacts the frontal, supported by a substantial ectethmoid. Support ofthejugal bar and palatines Cracraft (1968) has described various modifications of the lacrimal-ectethmoid complex which appear to have the function of bracing the jugal bar against dorsally, and in some cases medially directed forces. He quotes Bock's (1960a) suggestion that the brace may pre- vent disruption of the jugal when food is being crushed, but notes 'of interest, however, is the fact that many species, which are close relatives of birds with a brace, lack the brace, but yet probably would derive advantage from possessing one'. This problem still remains unresolved, and this review only serves to re-emphasize it. Most Coraciiformes and Pici- formes possess some sort of jugal brace, with the exception of the Momotidae, Capitonidae and, perhaps, Ramphastidae. In the latter family, the ventro-lateral corner of the ectethmoid is produced into a process which makes contact with the jugal bar where it broadens to join the maxillo-palatine — a position which scarcely seems to require bracing. Nor does this process seem to act as a retraction stop, for when skulls are manipulated, retraction is halted by the frontals before the maxillo-palatine makes effective contact with it. The slip of M. pterygoideus attached on the maxillo-palatine passes under the ectethmoid just medial to this process, but would be well flattened during strong retraction. The Capitonidae and Ramphastidae eat a large proportion of fruit, and apparently swallow much of it whole; possibly for these birds, the danger of disruption when crushing food is fairly small. The case of the Momotidae is much more puzzling, as much of their animal food surely requires fairly powerful crushing. Perhaps there is some positive selective advantage for losing the brace in this case, but it is difficult to see what this might be. Bracing of the palatines by a wide ectethmoid is seen in a number of bird families feeding on aerial insects, e.g. Caprimulgidae, Apodidae, Hirundinidae. Cracraft suggests that it may serve to protect the palatines against impacts with moving prey. The Coraciiformes and Piciformes, although including various aerial insect feeding groups, show few clear examples of such a brace. The two principal families which feed in this way, the Meropidae and Galbulidae, catch their prey with the tip of a long bill, and forces on their palatines are unlikely to be abnormal. The Meropidae totally lack any indication of such a brace, while in the Galbulidae, Galbula has a wide ectethmoid lying near the palatine (though part of M. pterygoideus dorsalis intervenes between the two) but this is much reduced in Galbalcyrhynchus. A possible ectethmoid brace exists, however, in Indicator and Jynx, though its function is unclear; perhaps it is connected in the latter with the large mouthfuls of insect prey which are collected to feed nestlings. In the Coraciidae, the anterior ventral FEEDING APPARATUS IN CORACIIFORMES & PICIFORMES 405 edge of the inflated lacrimal lies close to the palatine. In Coracias and the Brachypteraciinae, it is so far forward that it probably adds little to the support already available from the fused maxillo-palatines (see below). In Eurystomus, it lies somewhat further back, and may be of some value to this specialized aerial feeding genus, with its strongly widened palatines. Chelidoptera feeds in a similar way, but its ectethmoids are situated too far above the palatines to have any value as a brace. Desmognathy Much attention was given during the nineteenth century to the variations of palatal structure first described in detail by Huxley (1867). Attempts to interpret these at the time followed mainly phylogenetic reasoning, and surprisingly little attention has been paid to them since, despite the greater concern given to functional anatomy by more recent workers. Among the Coraciiformes and Piciformes the presence or absence of the desmognathous condition (in which the maxillo-palatines are fused in the midline) is a feature of considerable interest. It would appear that this condition is derived from a schizognathous or aegithothognathous one rather than the reverse; various intermediate stages can be seen in the Capitonidae. This condition is found throughout the Coraciiformes, but within the Piciformes it is seen only in the Galbuloidea, the Ramphastidae, and, weakly developed, in some Capitonidae. The Ramphastidae, like the Bucerotidae, have the anterior part of the palatines fused as well, producing the condition which Beddard (1898) termed 'doubly desmognathous'. This is pre- sumably in some way a consequence of the evolution of massive bills in these two families, though whether through mechanical necessity or for ontogenetic reasons is uncertain. The desmognathous forms include a large proportion of birds which feed on active, and often large animal prey, and it is possible that desmognathy provides some support against stresses incurred through killing and consuming such prey. Significant forces would be those acting across the bill axis; those acting along it are more a feature of birds which excavate for food, and resistance to them would scarcely be improved by midline fusion of the maxillo-palatines. Those acting across the bill include upward forces, produced by grip- ping or crushing prey, and lateral ones, either directly across or twisting (i.e. with a dorsal or longitudinal component). Simple upward forces would meet some resistance from well developed maxillo-palatines, but these need not necessarily be joined; the crucial feature of fusion seems most relevant where lateral forces are involved. An important factor here may be the technique of beating prey against the perch, used by many species to kill or immobilize their victims. This is commonly performed sideways, and though the prey is the primary target, the bill often makes contact with the perch as well. The forces involved are necessarily large, since feeble blows would fail in their purpose. It is perhaps significant that passerines lack midline fusion of the maxillo-palatines, their expanded vomers supporting them only against moderate dorsally directed forces. Although some passerines, such as crows and shrikes, regularly take large and vigorous prey, their methods of subduing it do not include the highly developed beating behaviour of the Coraciiformes and Galbuloidea. Desmognathy has also some relevance to the evolution of an M. pterygoideus dorsalis with attachment on the maxillo-palatine, as seen in several Coraciiform and Piciform groups. Such an attachment requires at least a strong maxillo-palatine; since unilateral action could impose severe stress on the maxillo-palatine, desmognathy would seem to be almost essential as well. In fact, the majority of species in which M. pterygoideus is attached on the maxillo- palatine are desmognathous. Exceptions occur only among the Capitonidae, in which slight muscle attachment on the maxillo-palatine is present in some species which are scarcely or not all desmognathous; any increase in such attachment, however, might be expected to be closely accompanied by increased desmognathism. This may have been a key factor in the evolution of the Ramphastidae from barbet ancestors (see Systematic Review, pp. 433-434). Pterygo- palatine articulation The Indicatoridae and the Picidae (including Jynx) differ from all other families studied here in the form of the pterygo-palatine articulation, with a long pterygoid foot overlapping the 406 P. J. K. BURTON palatine dorsally and medially. This arrangement would seem less prone to disarticulation than the short articulation of other families, but it is far from clear why it has only arisen in these two groups. The change may have involved transfer of the epipterygoid from the palatine (with which it normally fuses) to the pterygoid, but no nestlings were available young enough to compare its ontogeny between these two families and others. Lower jaw The movements of the lower jaw are generally simpler than those of the upper, except insofar as the two are correlated by kinetic coupling. An exception to this is seen in birds which possess the capacity to widen the gap between the mandibular rami by the action of M. ptery- goideus ventralis medialis; the pull of this muscle on the internal process of the mandible rotates it on the quadrate articulation, bowing the rami outwards. Though this mechanism has been described from various unrelated taxa, such as the Procellariiformes and Charadrii- formes (Yudin, 1961), Columbiformes (Burton, 19746), Caprimulgiformes (Btihler, 1970, 1980) and probably Pelecanus (Burton, 19776), it appears to be absent, or scarcely developed in most Coraciiformes and Piciformes. Flexible zones within the mandibular rami are obviously a prerequisite, but are virtually lacking in the majority of species examined. A small degree of flexibility which would permit moderate bowing seemed to be present in skulls of Alcedo, Leptosomus and Jynx, and with fresh material might be demonstrated in a few more. Nevertheless, this faculty seems to be of little significance in the feeding methods of the Coraciiform-Piciform assemblage. This is somewhat surprising in view of the many members of this assemblage which swallow large prey whole. Presumably flexible rami would be incompatible with other mechanical demands, such as those of prey beating. It should also be noted that Coraciiform and Piciform birds do not feed their young by regurgitation, a process which involves mandible bowing in some groups of birds. For the majority of birds, there is relatively little danger of excessive depression of the lower jaw — an event only likely to be brought about by unusual environmental forces, such as those encountered by Rynchops nigra (Zusi, 1962). Consequently, there is no system of bony stops, the main limitation on movement being the post orbital ligament, acting in con- junction with the upper jaw apparatus. However, the lower jaw is exposed to other hazardous forces, for which various support mechanisms exist, as follows: (a) The internal and external jugomandibular ligaments. The internal ligament resists backward disarticulation; the external resists downward and perhaps forwardly directed forces. Both are present throughout the families studied except the Picidae. Their lack of the external jugomandibular ligament is presumably directly related to the development of the opisthotic ligament which partly takes over its function. (b) The occipito- mandibular ligament. Universally present (though very weak in Jynx), and very strongly developed in the Upupidae, Phoeniculidae and Bucerotidae. In at least the first two of this group of families, the large size of the ligament is likely to be related to feeding by gaping and probing. The occipito-mandibular ligament resists forward disarticulation of the mandible, and forces tending to cause this are most likely to reach significant levels when the bill is being extracted from some substrate (see Burton, 1974a for discussion of this situation as it affects probing shorebirds). (c) The quadrate condyles. The primary role of the condyles is, of course, in articulation of the mandible in normal movement. However, the form of the medial condyle in some families studied here suggests a possible additional function in resisting disarticulation. This is most clearly seen in the Bucconidae, in which the medial condyle projects ventrally well beyond both the other condyles and the pterygoid articulation. It is flattened in the same plane as the pterygoid, and oriented to project slightly forwards and outwards; the tip is swollen and rounded. Less extreme versions are seen in the Galbulidae, Coraciidae and Leptosomatidae and to some extent even in the Momotidae and Meropidae. Experimentation with prepared skulls quickly shows that this deep medial condyle would be ineffective in resisting anteriorly or posteriorly directed forces, but firmly resists a force FEEDING APPARATUS IN COR AC 1 1 FORMES & PICIFORMES 407 acting medially. Forces acting across the jaw in this way may well arise during prey beating, a technique regularly employed by these families, or when swallowing large prey. Kingfishers lack a deepened medial condyle however, despite the fact that they beat fish vigorously, and consume quite large ones whole. It should be emphasized that a medial condyle of this form must inevitably have other functions as well. It must play an important part in coupled kinesis, although it is unable to bring about coupling in itself, like the specialized medial condyle of Bucorvus and some barbets. Its shape is highly efficient for spreading the mandibular rami during jaw closure, though admittedly this also happens perfectly well in birds with an unspecialized medial condyle. Its depth, in the Bucconidae especially, brings the centre of rotation of the lower jaw unusually far below the cranium; the significance of this is unknown. It certainly merits much further study, for it may be an important feature in the evolution of the Galbuloidea and Coraciiformes (see Systematic Review, p. 430). (d) The medial brace. This feature was first described in Rynchops nigra by Shufeldt (1890) and subsequently in a wide range of birds by Bock (1960a). Its functions in Rynchops were further studied by Zusi (1962). As Bock points out, all structural features of the brace suggest that it serves to prevent disarticulation of the mandible when the bill is opened. He further proposes that it compen- sates for inadequate protection by the quadrate condyles when the jaw is widely opened. The distribution of the brace among the families examined here supports these proposals. The brace occurs only in families which regularly feed on large or active prey, and often need to open the jaws widely; it is absent in groups which are primarily fruit eaters and in those which probe or excavate for insect prey. Three families appear to form an exception; these are the Leptosomatidae, Galbulidae and Bucconidae, all of which lack a medial brace. However, these are also families possessing a very deep medial quadrate condyle. This pro- vides a high degree of protection for the articulation, even with jaws open (see above), and also has the effect of holding the lower jaw so far from the basitemporal region of the cranium that secondary articulation is impossible. On the basis of this reasoning, a deep medial condyle or a medial brace might seem to be alternative derived mechanisms fulfilling the same function. They are not necessarily mutually exclusive, however. The Coraciidae, for example, seem clearly to have a common origin with the Leptosomatidae, and still possess a fairly deep medial condyle (rather reduced in Eurystomus), but have a medial brace as well. Perhaps during the evolution of the Coraciidae, the medial condyle has been reduced due to some other factor(s) to the point at which a basitemporal articulation — and thence, a medial brace — could develop. The medial brace thus needs to be treated with some caution if used as a taxonomic character. Jaw muscles M. adductor mandibulae externus M. add. mand. ext. raises the mandible; it also retracts the mandible against the quadrate, rocking it backwards about its articulation with the cranium, and thus aiding retraction of the palate and depression of the upper jaw. (M.a.m.e. caud, arising on the quadrate, is unable to affect the upper jaw in this way). It is generally the most complex of the jaw muscles, and its structure varies greatly among the Coraciiformes and Piciformes. Fortunately, suf- ficient information is now available from other orders to give at least an outline picture of its evolutionary history. This wider perspective is essential, for the structural diversity of the muscle cannot be adequately interpreted in purely functional terms. Among species studied here, the most complex M. add. mand. ext. is seen among the Phoeniculidae. In this family, the muscle includes several well defined components which are absent from the great majority of birds of other orders so far studied. An important exception is provided by the Anatidae; many members of this family possess a complex M. 408 P. J. K. BURTON add. mand. ext. which is remarkably similar in its components and arrangement to that of Phoeniculus and Rhinopomastus. An approach to this condition is also shown by some Charadriiformes, e.g. Chionis (Burton, 1974a). In the comparative discussion which follows, reference should be made to Table I for clarification of the terminology used for this muscle by previous workers; a fuller account of synonymy for M. add. mand. ext. is given by Starck &Barnikol(1954). The complex condition of M. add. mand. ext. seen in the Phoeniculidae and Anatidae is characterized by two main features: 1. The presence of a postorbital lobe, i.e. a section originating directly from the postorbital process, dorsal to M.a.m.e. rost. temp. This is represented in some families (e.g. Alcedinidae, Upupidae) by a single slip, but in its fully developed form (e.g. Phoeniculidae, Anatidae), there are two parts. The dorsal part arises fleshily from the postorbital process, and is attached to the mandible via an aponeurosis, while the more ventral and medial one arises by an aponeurosis from the postorbital process, and its fibres fan out across the lateral surface of the mandible. 2. A relatively narrow insertion of M.a.m.e. vent. The fan shaped sheet of fibres arising from Ap. 2 which constitutes this division of the muscle in most birds is replaced by similar sheets also inserting on the lateral surface of the mandible, but arising from either the postorbital lobe (ventral part), or from an expanded M.a.m.e. caudalis, or even from M. adductor pos- terior. This arrangement, but without a postorbital lobe, is also seen in Picidae (other than Jynx). Regarding the first of these features, although the evolutionary history of the postorbital lobe has been little discussed by previous workers, its presence in the Anatidae at least is well documented. It was noted by Lakjer (1926) in Oedemia ( = Melanittd) nigra, and included in his terminology for M. adductor mandibulae externus superficialis, which is approximately equivalent to M.a.m.e. rostralis as understood here. The same terminology was followed by Goodman and Fisher (1962) in their study of waterfowl. A complication found in the Anatidae (but not the Phoeniculidae) arises from the coalescence of the post- orbital and zygomatic processes. This has merged M.a.m.e vent, with M.a.m.e rost. Starck and Barnikol (1954, Fig. 1 1) regard M.a.m.e. vent. ( = Ap. 2 portion) in Anas platyrhynchos as the lower half of a bipinnate muscle whose dorsal part is M.a.m.e. rost. temp. ( = Ap. 1 portion, ex. temp.). Lakjer (1926) and Goodman and Fisher (1962) treat the whole unit as M.a.m.e. superf., la portion ('levator anguli oris'). If Starck and Barnikol are correct, which I believe to be the case, then M. add. mand. ext. medialis (Mil) of Lakjer and of Starck and Barnikol is not equivalent to M.a.m.e. vent. — as, on terminological grounds it should be — but, roughly, to M.a.m.e. rost. med. Turning to the second feature noted above, particular interest centres on the contributions of M.a.m.e. caudalis and M. adductor posterior to the more ventral and posterior sheet of muscle on the lateral surface of the mandible. In the Phoeniculidae, (as in the Picidae, dis- cussed later) this is provided entirely by fibres from M.a.m.e. caudalis. Interestingly, this is also the case in Chionis (Charadrii), leading Yudin (.1965, fig. 71) to label it incorrectly as M. add. mand. medialis ( = M.a.m.e. vent.). (See Burton, 1974: 122). Among the superficially similar Anatidae, however, the situation is more complicated. Here, M.a.m.e. caudalis itself is not widely expanded, but is closely associated, or even fused with an extensive lateral insertion of M. add. post. This has led to some confusion. Lakjer (1926) figured Oedemia ( — Melanittd) nigra, which happens to be a species in which M.a.m.e. caudalis is somewhat reduced, and completely overlaid anteriorly by M. add. post.; his figures 121 and 122 depict this correctly, a point I have checked by dissection. Goodman and Fisher (1962) who follow Lakjer's terminology, also label the fibres on the ventral lateral surface of the mandible as M. add. post, in other waterfowl, though in the two of their figured species which I have dissected (Spatula clypeata, Mergus merganser) the posterior edge of FEEDING APPARATUS IN CORACIIFORMES & PICIFORMES 409 this group of fibres is contributed by M.a.m.e. caud. (But note that Goodman and Fisher's 'a' and 'b' parts of M. add. post, do represent a real division of this muscle in M. merganser). Starck and Barnikol (1954) go to the opposite extreme by labelling the equivalent region in Anas platyrhynchos as 'Ap. 3 post.' ( = M.a.m.e. caud.). In this species (and S. clypeata) M. add. post, and M.a.m.e. caud. are thoroughly fused, but most of the lateral fibres come from M. add. post., as in other ducks. In a goose, however, (Anser anser, fig. 29), Starck and Barnikol label the lateral ventral fibre sheet as M. add. post. Clearly there is scope for much further investigation of the relative development of these two muscles, and not only in waterfowl. Although the similarity between M. add. mand. ext. in the Phoeniculidae and Anatidae does not extend to all points of detailed structure, there are still enough features in common to require explanation, the more so since clear echoes of the same condition are seen in some Charadriiformes. These three groups can scarcely be considered as close relatives, and it is difficult to envisage convergence as a factor in such totally different birds. More probably, the complex state of M. add. mand. ext. in the three groups should be regarded as a primitive feature, and similarly any traces of it which remain in other families, e.g. the postorbital slip in kingfishers, or the narrow M.a.m.e. vent, and extensive M.a.m.e caud. in the Picidae. Probably future studies will reveal similar conditions in other groups, and perhaps shed further light on the evolution of components of M. add. mand. ext. As a corollary of this view, the simpler structure seen in most birds is a derived one; this is characterized by a narrowly inserting M.a.m.e. rost. lacking any postorbital lobe, and M.a.m.e vent, which alone forms the sheet of fibres inserting on the lateral surface of the mandible, and a short M.a.m.e. caud., situated deep and partly concealed by M.a.m.e vent. This general derived condition is widespread, and has clearly arisen more than once, and probably many times. It can therefore scarcely be regarded as a useful taxonomic character in itself, any more than can the presumed primitive state. Nevertheless, detailed comparison reveals differences between groups which are consistent enough to be significant, such as those between the Coraciiformes and Piciformes. Such consistent differences may indicate that the simplified condition was derived independently in the two groups. It may be noted that the Phoeniculidae still retain the primitive condition, and thus the Upupidae and Bucerotidae which seem closely related to them (see p. 426-429) probably represent two more independent lines in which a simplified condition has been attained among the Coraciiform/Piciform assemblage. It is interesting to observe that here again, as in many other respects, the Galbulidae and Bucconidae resemble the Coraciiformes rather than the Piciformes. The position within the Picidae is also somewhat unclear; the derived condition is only fully developed in Jynx, which resembles barbets, toucans and honeyguides in its wide, flattened M.a.m.e. rost. lat., and in the form of M.a.m.e. vent. The implications of this are discussed further in the systematic review (p. 434). So far, there has been little consideration of functional aspects. In the foregoing discussion, an underlying assumption has been that the components of M. add. mand. ext. function simply to contribute to the forces of adduction and retraction, rather than possessing distinc- tive additional functions of their own. This is, no doubt, an oversimplification, but one that probably comes fairly close to the truth, leaving unaffected the general evolutionary picture which has thus far been depicted. Any distinct local functions of components of M. add. mand. ext. would probably concern stresses on the mandibular ramus; thus, for example, the balance between the forces of M. add. mand. ext. on the lateral surface of the mandible and M. pseud, prof, on the medial surface must in some degree influence the structure of both muscles. However, adduction and retraction remain overriding factors, the contribution of components varying with their situation. The more dorsal parts of the muscle have a greater moment arm, and consequently provide more force in adduction; they also act over the greatest distance, and generally show fairly simple fibre arrangements. In passing, it may be noted that the most dorsal component of all (in the absence of a postorbital lobe), M.a.m.e. rost. temp., normally has strongly bipinnate structure. This is the inevitable result of the 410 P. J. K. BURTON way in which its fibres are attached, around the edges of the temporal fossa. A parallel fibred structure would waste much of the potential surface for origin. Ventral parts, particularly M.a.m.e. caud. have a shorter moment arm, and also act over a shorter distance; in conse- quence, strongly multipinnate fibre arrangements are characteristic of this section of the muscle (see Gans & Bock, 1965). Although components within a complex M. add. mand. ext. like that of Phoeniculus may have only limited significance individually, the overall result is a bulky and presumably strong muscle, with extensive aponeurotic surfaces for fibre attachment. Birds which have attained a simpler structure in this muscle have lost some potential sites for origin or inser- tion of fibres, and may, when required, create new ones by branching of existing aponeuroses. This is well illustrated within the passerines (e.g. Bock, 1960&) or among waders (Burton, 1974a). In such cases of secondary structural complexity, a region of the muscle which is often elaborated is M.a.m.e. rost. med. — as, for example, the 'Ap.D. slip' of the Scolopacinae described by Burton, \914a. Within the Coraciiformes and Piciformes, M.a.m.e. rost. med. is relatively poorly developed; perhaps because M.a.m.e. rost. temp, has remained well developed in many members of the two orders. The exact extent of M.a.m.e. rost.temp. does, nevertheless, vary a good deal among the birds studied here, and, as in many groups, there appears to be a distinct correlation with body size. (And also with reduction of the lateral portion of M. pseudotemporalis super- ficialis). A very limited temporal origin is characteristic of small species, and is often associ- ated with overall simplicity of muscle structure, and reduction in pinnate fibre arrangements. This is, no doubt, simply one of the many consequences of the relation between body size (weight varying as the cube of linear dimensions) and strength of components (proportional to cross sectional area, and hence their square). It is tempting to speculate, however, that fluctuations in size within the line ancestral to a species may have influenced its present structure. To take M. add. mand. ext. as an example, structural simplification may in many cases have arisen within small species; components which disappeared in such species could have thus been lost forever, even if their descendants subsequently evolved larger body size. M. pseudotemporalis superficialis Among the families considered here, this muscle ranges from a bulky structure of complex architecture to a slip which is vestigial or even absent. A crucial point in understanding this variation appears to be the relative development of lateral and medial regions of the muscle. A highly developed lateral lobe can with confidence be regarded as a primitive feature of the muscle, being exhibited by several unrelated non-passerine groups. In some of these — e.g. among the Laridae — it extends substantially onto the temporal surface of the skull, deep and somewhat dorsal to M.a.m.e. rost.temp. Reduction of this lateral portion appears to be an evolutionary trend common to many avian groups, and generally strongly correlated with reduction of M.a.m.e. rost.temp. In the groups surveyed, a lateral portion is best developed in some families of Coraciiformes, and within this order there is little development of a medial region, even where the lateral origin is much reduced. Where marked reduction has occurred, the whole muscle has become dorsal and superficial in character, lying mainly lateral to the Vth cranial nerve. The Galbulidae and Bucconidae resemble the Coraciiformes in this respect, but the condition is even more strongly marked, the muscle being vestigial in many species, and absent in some. In typical Piciformes, by contrast, there is generally a well marked medial extension of the origin, as though in compensation for loss of the lateral; the whole muscle is generally more medial and deep in position, lying more or less medial to the Vth cranial nerve. The factors underlying these differing trends are hard to discern in our present state of knowledge. It is by no means clear why reduction of one part of M. pseudotemporalis super- ficialis should be compensated by enlargement of another, if, indeed, this is what has actually happened in the Piciformes. A trend towards reduction and loss of the muscle as seen in the Coraciiformes appears more reasonable; it would seem relatively simple for M. add. mand. ext. and M. pseudotemporalis profundus to take over the task of M. pseudotemporalis FEEDING APPARATUS IN CORACIIFORMES & PICIFORMES 4 1 1 superficialis in providing adduction force with a component medial to the mandibular ramus. Some answer may eventually emerge from a deeper understanding of cranial architecture. M. pseudotemporalis profundus and M. adductor posterior Like M. pterygoideus, M. pseudotemporalis profundus acts both to raise the lower jaw and (through its action on the orbital process of the quadrate) to depress the upper jaw. Relative development of the muscle is hard to assess; its apparent size in dorsal view is certainly a most unreliable guide. In the majority of birds, its excursion is a long one, a fact reflected in its largely parallel-fibred structure; the main exception here is shown by Tockus erythror- hynchus in which the muscle is small, with its insertion relatively close to the centre of rotation of the mandible. Otherwise, the most significant variation is the total absence of the muscle in many Alcedinidae. This is certainly related to the form of the skull in this family, in which palatines and pterygoids lie almost in the same plane, and the quadrate has swung backwards (by comparison with other birds), so that the orbital process lies very close to the mandible. Only a small M. pseudotemporalis profundus is therefore possible in any case, and loss of the muscle is not surprising. Even more extreme versions of this skull form are seen in a number of other fish eating groups (Barnikol, 1952), and loss of M. pseudotemporalis profundus has also occurred in some of these, e.g. Sula, Phalacrocorax (Hofer, 1950). A similar quadrate shift, accompanied by loss of M. pseudotemporalis profun- dus has also taken place (though for different morphological reasons) in the Psittaci formes (Hofer, 1950; Burton, 1974c). It is interesting to note that the muscle is present in some Cerylinae, but absent in others; detailed observation and comparison of feeding actions in members of this subfamily might produce interesting results. M. adductor posterior, by reason of its situation so close to the quadrate- mandibular joint and consequent short moment arm, can play little part in adduction, but has been suggested to act in providing support for the lower jaw on the quadrate (Zusi, 1962). It shows relatively little variation among species studied here. The lateral expansion seen in some barbets would somewhat increase its capacity for adduction; this may be a primitive feature (see discussion under M. add. mand. ext.). M. pterygoideus Because this muscle is attached at one end to the mandible, and at the other to the palatal complex, it acts both to raise the lower jaw and to retract the palate and thus depress the upper jaw. The siting and structure of components of the muscle may favour one or the other of these two main actions, and functional accounts of M. pterygoideus inevitably con- centrate on these two aspects. Although an analysis in these terms goes part way to explaining the structural complexity of the muscle, it leaves a number of questions unanswered in the case of the Coraciiformes and Piciformes. Satisfactory answers will probably require a much more detailed understanding of jaw mechanics than we at present possess. Particular interest centres on the division of M. pterygoideus dorsalis into lateral and medial portions, and on the modifications of each. Two families examined (Upupidae and Phoeniculidae) showed no division into two portions; nor does Ramphastos toco (Ramphas- tidae). All but one of the remaining families have the muscle divided by a groove into which N. pterygoideus passes, shortly after diverging from the mandibular ramus of the Vth cranial nerve. The exception is the Bucerotidae, in which a clear groove divides the muscle, but the nerve penetrates it an appreciable distance posterior to the groove. (If the raphe of the bipinnate M. pter. dorsalis in Phoeniculus purpureus is considered homologous with the dividing groove, then this genus shows a similar condition). A similar situation was found among the Charadrii studied by Burton (1974a). A reasonable explanation for such a dichot- omy would be to regard the undivided condition as primitive, and the two positions of nerve relative to groove as alternative derived conditions; however, the possibility cannot be excluded that a divided condition may return to an undivided one. Even less clear is the functional relevance of the division. Physical separation by a groove implies that these two parts of the muscle perform quite 412 P. J. K. BURTON distinct actions. M. pter. dors. lat. is oriented more nearly parallel to the long axis of the skull, and hence would act more effectively as a depressor of the upper jaw; at the same time, its more anterior placement on the mandible would improve its capacity for adduction. M. pter. dors. med. seems badly situated from both points of view; in the case of the passerine genus Loxops, Richards & Bock (1973) suggest that it functions mainly to rotate the ptery- goid (to which it is largely attached) during kinesis. However, in many of the birds examined in this study, M. pter. dors. med. extends far onto the palatine, and may be so oriented as largely to overcome its disadvantageous position for palate retraction. In such cases, the muscle is functionally almost equivalent to a completely undivided M. pter. dorsalis. Groups showing this condition of M. pter. dors. med. may have M. pter. dors. lat. attached to the maxillo-palatine (see below) or much reduced, as in the Galbulidae. In some Galbulidae, M. pter. dors. lat. is approaching a vestigial condition, and interestingly, one of them (Jaca- merops aured) shows a partial division of M. pter. dors. med. in a similar situation to the groove which divides M. pter. dors. lat. from M. pter. dors. med. in the majority of birds. (Total loss of M. pter. dors. lat. would leave N. pterygoideus in the curious situation of entering the muscle via its anterior edge.) The evolutionary reasons for this development remain obscure, and do not seem to be correlated in any obvious way with changes in skull architecture. The Bucconidae generally resemble the Galbulidae in the narrowing and reduc- tion of M. pter. dors, lat., yet one of them (Chelidoptera tenebrosd) surprisingly retains a M. pter. dorsalis divided by a groove in the normal position. A feature of much interest revealed by this study is the attachment of M. pter. dors. lat. to the maxillo-palatine in several families. This feature has not been previously described in any bird, even in Starck's (1940) study of the Bucerotidae. Attachment to the maxillo- palatine was found in the Alcedinidae (excluding the Cerylinae), the Phoeniculidae, Bucer- otidae, many Capitonidae and Ramphastidae. The significance of this modification is not immediately clear. The maxillo-palatine may simply act as a useful additional surface for muscle attachment, permitting the development of a bulkier M. pter. dors. lat. However, it should be noted that by attaching on the maxillo-palatine, M. pter. dors. lat. depresses the upper jaw directly, and not via the medium of the palatal complex. It is possible that such a fundamental innovation may have other advantages than simply to increase the force for adduction of the lower jaw and depression of the upper. Possibly a careful comparative study of jaw movements in related birds differing in this feature (similar sized members of the Daceloninae and Cerylinae would be highly suitable) might shed further light on this point. It may be noted that the presence of a bipinnate M. pter. dors. lat. is not correlated with attachment on the maxillo-palatine, although both features occur together in most Alcedini- dae and the Bucerotidae — and in Phoeniculus purpureus, in which the bipinnate structure of the entire M. pterygoideus dorsalis is functionally equivalent to that of M. pter. dors. lat. alone. Other groups in which bipinnate structure occurs are some Momotidae and Indicatori- dae, and more moderately in the Picidae (except Picumnus and Jynx). The feature might be expected to relate to a need for powerful contraction over short distances, as when the extent of jaw opening is small, or in isometric contraction with the jaws held open. Small amplitude jaw movements are probably important in many kingfishers and woodpeckers, and maintenance of a position with jaws slightly parted may be essential during drumming or chiselling by woodpeckers (Spring, 1965; Bock, 1964, 1 966). Prolonged isometric contrac- tion would also seem to play a part in manipulation of large fruits by some hornbills, though toucans face similar problems and do not have a pinnate M. pter. dors. lat. Finally, the retractor palatini slip of M. pter. vent. med. requires comment. Among the groups considered here, this feature occurs only in the Upupidae, Phoeniculidae and Bucer- otidae. However, a similar modification is found also in many passerines (e.g. Fiedler, 1951; Bock, 1960) and some other orders (Burton, 1974c). The families showing the feature here differ from others in the sheer size of the retractor palatini slip, and in its extensive attach- ment on both palatine and pterygoid; it appears to comprise all of M. pter. dors. med. post, plus part of M. pter. vent, med., whereas in passerines, M. pter. dors. med. post, remains FEEDING APPARATUS IN CORACIIFORMES & PICIFORMES 413 (at least partly), in unmodified form (see Richards & Bock, 1973). (The presence in Tockus erythrorhynchus of a raphe similar to Ap. N is a further point of difference.) The retractor palatini differs from all other parts of M. pterygoideus in that its action is purely to retract the palate and depress the upper jaw; it cannot adduct the mandible at all, nor can it bring about mandible howing as described by Yudin (1961), Zusi (1962) and Biihler (1970). Thus, it would be of great importance in any feeding action for which independent operating of the upper jaw was required. Small manipulative actions in long-billed birds are probably often of this type (Burton 19740), and its presence in the Upupidae and Phoeniculidae may be related to such needs in these probing feeders. (But note that there is no retractor palatini in the many Scolopacidae which employ a combination of deep probing and skilful manipu- lation in feeding; their rhynchokinetic upper jaws are more than adequate to meet these needs.) The Bucerotidae obtain their food in a variety of other ways, but their retractor palatini may be a heritage from ancestral forms possibly rather similar to the Upupidae and Phoeniculidae. M. protractor quadrati et pterygoidei The action of this muscle is to pull the pterygoid forward and medially, and to rock the quadrate forward and upward about its articulation with the cranium. These movements are communicated to the upper jaw, causing it to be raised. Noteworthy development of the muscle occurs in the Upupidae and Phoeniculidae; and in the Picidae. In the former two families, the condition of M. protractor is almost certainly related to their probing habits. These two families also have a large M. depressor mandibulae attached to a long retroarticular process. Both features are adaptations for 'gaping', i.e. open- ing the bill against the resistance of a substrate in order to excavate a wider hole in which to seek prey. Similar modifications are seen in some Charadrii, e.g. Scolopax (Marinelli, 1928; Burton, \974a) and a variety of Passeriformes, e.g. the Icteridae (Beecher 1953