BULLETIN OF THE BRITISH MUSEUM (NATURAL HISTORY) ZOOLOGY VOL. 15 1967-1968 TRUSTEES OF THE BRITISH MUSEUM (NATURAL HISTORY) LONDON: 1970 DATES OF PUBLICATION OF THE PARTS No. i 3 January 1967 No. 2 . . . . . -13 January 1967 No. 3 . . . . . -24 February 1967 No. 4 . . . . . -24 February 1967 No. 5 . . . . . .16 May 1967 No. 6 ..... .28 July 1967 No. 7 ..... 14 November 1967 No 8 ..... 9 January 1968 PRINTED IN GREAT BRITAIN BY ALDEN & MOWBRAY LTD AT THE ALDEN PRESS, OXFORD CONTENTS ZOOLOGY VOLUME 15 PAGE No. i. The evolution, host relationships and classification of the nematode Superfamily Heterakoidea. By WILLIAM G. INGLIS i No. 2. A revision of the Lake Victoria Haplochromis species (Pisces, Cichlidae) . Part VI. By P. H. GREENWOOD 29 No. 3. The Vachell collection of Chinese fishes in Cambridge. By P. J. P. WHITEHEAD & K. A. JOYSEY 121 No. 4. Notes on asteroids in the British Museum (Natural History) V. Nardoa and some other ophidiasterids. By AILSA McGowN CLARK 167 No. 5. New species of Stellicola (Copepoda, Cyclopoida) associated with star- fishes in Madagascar, with a redescription of 5. caeruleus (Stebbing, 1900. By ARTHUR G. HUMES & JU-SHEY Ho 199 No. 6. The cirriped fauna of tropical West Africa. By H. G. STUBBINGS 227 No. 7. Polyzoa (Bryozoa) from West Africa: the Pseudostega, the Cribri- morpha and some Ascophora Imperfecta. By PATRICIA L. COOK 321 No. 8. Caligus longicaudatus Brady, 1899 (Caligidae: Copepoda) By R. R. PARKER 353 Index to Volume 15 369 THE EVOLUTION, HOST RELATIONSHIPS AND CLASSIFICATION OF THE NEMATODE SUPERFAMILY HETERAKOIDEA WILLIAM G. INGLIS BULLETIN OF THE BRITISH MUSEUM (NATURAL HISTORY) ZOOLOGY Vol. 15 No. i LONDON: 1967 THE EVOLUTION, HOST RELATIONSHIPS AND CLASSIFICATION OF THE NEMATODE SUPERFAMILY HETERAKOIDEA BY WILLIAM G. INGLIS, PLD., D.Sc. "VvMv British Museum (Natural Hiory) Pp. 1-28 BULLETIN OF THE BRITISH MUSEUM (NATURAL HISTORY) ZOOLOGY Vol. 15 No. i LONDON: 1967 THE BULLETIN OF THE BRITISH MUSEUM (NATURAL HISTORY), instituted in 1949, is issued in five series corresponding to the Departments of the Museum, and an Historical series. Parts will appear at irregular intervals as they become ready. Volumes will contain about three or four hundred pages, and will not necessarily be completed within one calendar year. In 1965 a separate supplementary series of longer papers was instituted, numbered serially for each Department. This paper is Vol. 15, No. I of the Zoological series. The abbreviated titles of periodicals cited follow those of the World List of Scientific Periodicals. World List abbreviation : Bull. Br. Mus. nat. Hist. (Zool.) Trustees of the British Museum (Natural History) 1967 TRUSTEES OF THE BRITISH MUSEUM (NATURAL HISTORY) Issued 3 January 1967 Price Ten Shillings THE EVOLUTION, HOST RELATIONSHIPS AND CLASSIFICATION OF THE NEMATODE SUPERFAMILY HETERAKOIDEA By WILLIAM G. INGLIS CONTENTS Page APOLOGIA ............ 4 SYNOPSIS ............ 4 INTRODUCTION ........... 4 COMPARATIVE ANATOMY OF THE SUPERFAMILY ..... 5 The male tail ........... 6 The spicules ........... 7 SYSTEMATIC GROUPS WITHIN THE SUPERFAMILY ..... 8 Family Heterakidae . . . . . . . . . .10 Family Aspidoderidae . . . . . . . . .10 Family Ascaridiidae . . . . . . . . . .10 MORPHOLOGICAL AND EVOLUTIONARY TRENDS . . . . . .11 GEOGRAPHICAL AND HOST DISTRIBUTION . . . . . -13 SYSTEMATIC HISTORY .......... 14 ANNOTATED CLASSIFICATION OF THE SUPERFAMILY . . . . .17 Heterakoidea Railliet & Henry, 1912 . . . . . . 17 Heterakidae Railliet & Henry, 1912 . . . . . . 17 Heterakinae Railliet & Henry, 1912 . . . . . 18 Heterakis Dujardin, 1845 . . . . . 19 Odontoterakis Skrjabin & Schikhobalova, 194712 . . 20 Pseudaspidodera Baylis & Daubney, 1922 ... 20 Meteterakinae Inglis, 1958^ . . . . . .21 Meteterakis Karve, 1930 . . . . . .21 Gireterakis Lane, 1917 . . . . . .21 Spinicaudinae Travassos, 1920 . . . . . .21 Spinicauda Travassos, 1920^ ..... 22 Africana Travassos, 1920 . . . . . .22 Moaciria Freitas, 195612 . . . . . .22 Strongyluris Miiller, 1894 ...... 23 Aspidoderidae Skrjabin & Schikhobalova, 1947 . . . . 23 Aspidoderinae Skrjabin & Schikhobalova, 1947 ... 23 Aspidodera Railliet & Henry, 1912 . . . . 23 Ansiruptodera Skrjabin & Schikhobalova, 1947 . . 23 Lauroiinae Skrjabin & Schikhobalova, 1951 . -23 Lauroia Proen9a, 1938 ... .24 Pamspidodera Travassos, 1914 ..... 24 Ascaridiidae Travassos, 1919 ... 2 4 KEY TO THE GENERA OF THE SUPERFAMILY ...... 24 ACKNOWLEDGEMENTS ...... 25 REFERENCES ...... -25 APPENDIX ..... ... 27 ZOOL. 15, i. J 4 W. G. INGLIS APOLOGIA SEVERAL years ago I published a classification of the nematode Superfamily Heter- akoidea (Inglis, 1958) but presented no justification as I always hoped to produce a full revision. It has become increasingly clear that this is never likely to be com- pleted largely because of a shift in my interests in conjunction with a lack of material on which to base such a revision. I therefore present here reasons for the classifica- tion previously proposed and regret being unable to complete the whole work. No illustrations are given as they can be found in Skrjabin, Schikhobalova and Lagodovskaja (1961) and Chabaud (1965). SYNOPSIS The comparative anatomy of the Superfamily Heterakoidea is described and the morpho- logical sequences recognized are used as the basis for the classification proposed. The evolution of the Superfamily is interpreted in terms of (i) intra-host localisation ; (2) major host groups (i.e. amphibia, reptiles, birds and mammals) ; (3) the feeding habits and ecology of the hosts ; and (4) geographical restriction. The Superfamily is classified into three families, five subfamilies and fourteen genera, thus : Heterakidae : Heterakinae (Heterakis, Odontoterakis, Pseud- aspidodera) ; Meteterakinae (Meteterakis, Gireterakis) ; and Spinicaudinae (Spinicauda, Africana, Moaciria, Strongyluris) . Aspidoderidae : Aspidoderinae (Aspidodera, Ansirupto- dera) ; and Lauroiinae (Lauroia, Paraspidodera) . Ascaridiidae (Ascaridia). A key is given to the genera of the Superfamily. The following genera are not accepted largely because the morphology of the specimens does not correspond to the published descriptions, thus : Preterakis Freitas, 1956 ; Heterakoides Freitas, 1956 ; Pareterakis Freitas, 1956 ; Pseudaspidoderoides Freitas, 1956 ; Pseudaspidoderina Freitas, 1956 ; Raillietakis Freitas, 1956 ; Sexansodera Skrjabin & Schikhobalova, 1947 ; Bellaplectana Skrjabin, Schikhobalova & Lagodovskaja, 1961 ; Spinaspidodera Skrjabin & Schikhobalova, 1947 ; Ganguleterakis Lane, 1914 ; and Cheloniheterakis Yamaguti, 1961. INTRODUCTION No full revision of the Heterakoidea or Heterakidae has ever been carried out although several groupings have been proposed largely based on other published work. However, Travassos (1913) did produce a partial revision based largely on the species occurring in Brazil. Other minor changes in the classification were proposed later but no major alterations were suggested for almost thirty-five years when Skrjabin & Schikhobalova (1947, 19470), Freitas (1956), Inglis (1957, 1958) and Chabaud (1957) all advanced new groupings. Most recently the superfamily has been re-grouped once more by Skrjabin, Schikhobalova & Lagodovskaja (1961) and by Yamaguti (1961). All the classifications proposed, except my own, have been based on biblio-taxo- nomic studies while the results presented here are based on a study of virtually all the type material available in the world's collections together with a study of many other specimens. In anticipation of the discussion that follows it may be stressed that the following genera proposed by Freitas (1956) and Skrjabin et al. (1961) cannot stand because the descriptions they classify are contradicted by the morphology of the specimens, thus: Bellaplectana Skrjabin et al., 1961 is a synonym of Moaciria since Aplectana pharyngeodentata Belle, 1957 (of which I have studied the type series) does not THE SUPERFAMILY HETERAKOIDEA 5 possess the so-called teeth in the buccal cavity on which the genus is based. These structures are simply the retracted lips, a common artefact in poorly preserved Spinicauda and Moaciria species. Preterakis Freitas, 1956 falls as a synonym of Africana because the fusion of the caudal alae on the male tail on which it is based does not occur. Such an apparent fusion occurs in many specimens of the super- family but is an artefact due to contraction during fixation and even the artefact does not occur in all the specimens of the type series of Africana astylosterni Sand- ground, 1933. As a consequence the subfamily Preterakinae Freitas, 1956 must also disappear. Finally the following genera proposed by Freitas (1956) are not accepted by Skrjabin et al. (1961) nor by myself, thus: Heterakoides Freitas, 1956 since, although Spinicauda triaculeatus Kreis, 1933 is certainly slightly different from the other species referred to Meteterakis (Inglis, 19580) I still do not feel it warrants generic separation. Similarly the genus Pareterakis Freitas, 1956 is a synonym of Meteterakis since there are no characters which allow it to be distinguished. Pseuda- spidoderoides falls because the anastomozing cordons reported by Chakravarty (1938) appear to be due to contraction of the type specimens, which I have seen. COMPARATIVE ANATOMY OF THE SUPERFAMILY All the species referable to the Superfamily, with the exception of the genera Ascaridia and Lauroia which are discussed later, have the following features in common. The head bears three large, distinct lips, the detailed structure of which varies throughout the family and has been considered elsewhere (Inglis, 1957). In summary, however, the dorsal lip carries two doubled papillae and each ventro- lateral lip carries a single ventro-lateral doubled papilla and a single lateral papilla associated with the amphid. The inner surface of each lip carries an anterior flange which varies in relative size throughout the superfamily. The head in Lauroia, which I did not deal with before, is typical of this pattern with lobes con- necting the lips. It is, however, characterized by the development of three plate- like regions posterior to each lip. This is clearly a development of cordons from one source (see particularly Proenga's (1938) figures of L. travassosi) comparable to the condition in Gireterakis. In Ascaridia the lips are large and stout but otherwise typically Heterakoid. The anterior region of the oesophagus is divided into three separate anterior lobes, previously called " pharyngeal portions ", at the tip of each of which is a cuticular onchium, except in one case. From studies completed since 1957 it now appears that the lips are operated by a system of fibres which are not arranged in a septum as in the Ascaridoidea (Inglis, 1965) but this has not been fully analysed. In addition in the Heterakinae there is generally a pair of lateral papillae lying on the body just posterior to the lip region. The oesophagus, except in Ascaridia, is modified anteriorly into a short region generally called a pharynx and swells fairly evenly posteriorly to form a bulb-like region, containing three valves, which is not clearly offset from the remainder of the oesophagus by a constriction. The distinction between the pharyngeal region, the bulb and the remainder of the oesophagus is the presence of marginal tubes 6 W. G. TNGLIS in the radii of the lumen of the oesophagus in the intervening length between the bulb and pharynx. Into these marginal tubes project a series of paired leaf -like cuticular structures and this arrangement gives the oesophagus its typical longitudinal doubled appearance. The oesophagus is relatively long and narrow with a markedly small posterior bulb in Aspidodera, Ansiruptodera, Lauroia and Paraspidodera; the oesophagus is less markedly narrow in Africana, Gireterakis, Meteterakis, Moaciria, Spinicauda and Strongyluris, and is relatively short and stout in Heterakis, Odontoterakis, and Pseudaspidodera. The oesophagus is club-shaped without a posterior bulb in Ascaridia. The nuclei of the ventral oesophageal glands are doubled. Lateral alae, which may be prominent in the cervical region, are typically present. They are relatively broad in Heterakis, Odontoterakis, Pseudaspidodera and Giretera- kis; are present but relatively narrow in Africana, Spinicauda, Moaciria and Strongyluris in which genera the lateral fields are very prominent and also in Meteterakis and Ascaridia but they are lacking in Aspidodera, Ansiruptodera, Lauroia and Paraspidodera. The nerve ring encircles the oesophagus about one third of its length posterior to the pharynx and the excretory pore opens just posterior to the nerve ring. Small sessile papillae are borne on the general body surface in Africana (few), Meteterakis (few), Spinicauda (very few), Moaciria (many) and Strongyluris (many). Such papillae are, however, frequently difficult to see and they are probably more easily seen in Strongyluris species because such species are relatively larger than the others. The Male Tail A circular pre-cloacal sucker with a cuticular rim is always present (except in Lauroia). Caudal alae are fairly common in the superfamily and, when present, never meet on the ventral surface of the body anterior to the pre-cloacal sucker. In those specimens in which the tail has contracted during fixation such a junction may appear to exist but it is clearly an artefact and is, therefore, of no use as a systematic character. Freitas (1956) lays considerable stress on this " junction " of the caudal alae, going so far as to use it to differentiate genera. I would stress that not only is this an artefact but it is not present in all the specimens of most of the species of which it is considered a diagnostic character. In the genera Heterakis, Odontoterakis and Pseudaspidodera there are, typically, eleven pairs of caudal papillae, of which seven pairs are long and support broad caudal alae. Two pairs of these papillae, which are always slim, flank the pre- cloacal sucker, four pairs lie lateral to the cloacal opening (later referred to as the para-cloacal papillae), and one pair lies posterior to the cloacal opening, roughly midway between the para-cloacals and the terminal spike of the tail. The remaining four pairs of papillae are typically sessile, two pairs flank the cloacal opening, the peri-cloacal papillae, of which one pair is pre-cloacal and one pair post-cloacal. The remaining two pairs of papillae lie at the base of the terminal spike of the tail with one pair wholly ventral and one pair lateral in position. The phasmids open just anterior to the terminal group of ventral papillae and have generally been THE SUPERFAMILY HETERAKOIDEA 7 described as papillae. In some cases additional papillae occur, in particular a pair of sessile papillae frequently occur anterior to the suctorial pairs and an additional long pair sometimes occurs immediately posterior to the para-cloacals. The tail ends posteriorly in a long, narrow, evenly pointed, non-alate terminal spike. In all four genera the pre-cloacal sucker is relatively large and there is no gubernaculum. Both Spinicauda and Africana have long narrow tails bearing many small sessile papillae and a relatively small pre-cloacal sucker. There are no caudal alae in Spinicauda while those in Africana stop posteriorly at the level of the cloacal opening and are unsupported by papillae. A gubernaculum is present in Spinicauda and a gubernacular mass in Africana (gubernacular mass is a term proposed earlier for the strongly cuticularized lining of the cloaca in the genus Meteterakis (Inglis, 19580)). The tail in Strongyluris is very short with a large pre-cloacal sucker which is directed posteriorly. The caudal alae are very broad and are typically supported by seven pairs of long stout papillae of which the largest are the three most anterior pairs, with the others becoming increasingly smaller posteriorly. There are, in addition, two pairs of ventral sessile papillae peri-cloacal in position and one similar pair just anterior to the beginning of the terminal spike. There is no gubernaculum. The tail in Moaciria is similar but with a long terminal spike and a gubernaculum. The relatively long, narrow tail in Meteterakis is characterized by a small pre- cloacal sucker and narrow caudal alae supported by, typically, three, exceptionally four, pairs of short, stout papillae of which, typically two pairs, exceptionally three pairs, lie about the level of the pre-cloacal sucker and the remaining pair is roughly at the level of the cloacal opening. There are in addition many pairs of small sessile papillae (up to about twenty) on the tail which may represent the caudal complement of the similar papillae which are found scattered generally over the surface of the body. A gubernacular mass is present in all except one (possibly two) species. The tail in Gireterakis is very similar to that of Meteterakis with the same three pairs of swollen papillae supporting the caudal alae but there are much fewer small sessile papillae and there is no gubernacular mass. The tail in the genera Aspidodera, Ansiruptodera, and Paraspidodera is so uniform that one description will cover all three. It is long and thin, tapering evenly to a point. It has no caudal alae, the pre-cloacal sucker is markedly small and there is a large number of small sessile papillae arranged in roughly parallel rows. A guber- naculum is present in all three genera. The tail in Lauroia is highly modified but may be interpreted as being derived from a tail similar to that characteristic of the previous three genera by a great reduction of all its structures. Thus the pre-cloacal sucker is represented by a slight swelling and the number of sessile caudal papillae is very small. The tail in Ascaridia is relatively stout with narrow alae supported by short stout papillae. The pre-cloacal papillae are sessile and there is no gubernaculum. The Spicules The spicules are always equal in length, non-alate, simple and identical in structure in the genera Africana, Ascaridia, Aspidodera, Ansiruptodera, Moaciria, Paraspido- ZOOL. 15, i. J 8 W. G. INGLIS dera, Spinicauda and Strongyluns. They are equal in length and identical in structure, with alae in some cases, in Meteterakis and Africana. They are equal in length and slightly elaborate although identical in structure in Gireterakis while in Heterakis, Odontoterakis and Pseudaspidodera they are frequently unequal in length and may be very dissimilar in Heterakis and Pseudaspidodera but are always simple, needle-like in Odontoterakis. It may be noted here, and will be referred to later, that the general facies of the male tail in Aspidodera, Ansiruptodera, and, to a lesser extent, Paraspidodera is remarkably uniform even to the same general shape of the spicules which are, almost invariably, rather stout with squarish ends. SYSTEMATIC GROUPS WITHIN THE SUPERFAMILY This section should be read in conjunction with the discussion of morphological trends given in the next section, since the classification presented reflects the sequences discussed there. In other words this classification is largely akoluthic (sensu Inglis, 1966(2) in reflecting trends in morphological modification rather than in attempting to assess over-all resemblances or simply utilizing key characters. It does appear, however that an assessment of over-all resemblances would produce a classification the same as the present since virtually the only characters available for analysis are those of the head and the male tail and a classification based on either alone is congruent with one based on the other. That is, if only tails are classified they fall into roughly the same major groupings as do the heads if classified alone. This explanatory discussion of procedure intrudes here because in classifica- tions such as this I find it impossible to separate the discussion of the process of classification from that of the establishment and recognition of morphological trends. In fact the trends were probably recognized before the classification was developed. It is, however, easier to follow the discussion of trends and host re- lationships if a foundation is given upon which they can be discussed. As stressed above most classifications of this superfamily (family) have been mainly based on published descriptions which are in many cases incomplete or inaccurate. A reliable assessment of such descriptions, and the characters upon which to form a classification can only be made after a wide and detailed knowledge of the group under consideration has been obtained. The dangers involved in bibliotaxonomy are exemplified by the gross oversplitting proposed by Freitas (1956) with a multiplicity of new names at all levels of taxa. The classification proposed by Freitas may be considered analytical in that he has given systematic recognition to any differences he found. The classification proposed here is basically synthetic with a stressing of similarities rather than differences although, obviously analysis must precede such a process. The superfamily contains three morphologically distinct groups which differ in the forms of the lips, of the oesophagus and of the male tail. One is characterized (i) by square lips which are connected by lateral lobes, (2) by a cephalic cap (i.e. a thickening of the cuticle at the anterior end of the body) , (3) by a markedly long and narrow oesophagus which expands relatively suddenly into a small posterior tri- valvulate oesophageal bulb and (4) by males with relatively long, narrow, evenly pointed tails without alae, typically bearing many small sessile papillae, (5) a THE SUPERFAMILY HETERAKOIDEA 9 relatively small pre-cloacal sucker, (6) having relatively massive, square ended spicules without alae and (7) possessing a gubernaculum. This group I recognize as the Family Aspidoderidae. The second group is characterized by (i) rounded lips without lateral lobes, (2) no cephalic cap, (3) a relatively short stout oesophagus which merges rather slowly into a relatively large posterior tri-valvulate bulb and (4) by males in which the tail (with the exception of one genus) always carries caudal alae, (5) the spicules are frequently dissimilar and alate, (6) a gubernaculum is generally lacking, and (7) there is a tendency for the pre-cloacal sucker to be prominent and for the caudal papillae to be large. This group I treat as the Family Heterakidae. The third group is characterized by (i) massive lips without anterior cuticular flanges or (2) lateral lobes or (3) onchia at the anterior ends of the pharyngeal portions which are themselves very small, (4) the oseophagus is grossly club-shaped without a posterior oesophageal bulb or valvular apparatus, (5) the male tail carries the usual circular pre-cloacal sucker with a definite cuticular rim but, although the papillae are grouped round the region posterior to this sucker, they are generally stout and rounded without marked lateral caudal alae, (6) there are frequently files of sessile papillae running down the lateral sides of the body. This group I treat as the Family Ascaridiidae. I did not previously accept the Family Ascaridiidae as referable to the Heterakoidea (Inglis, 1958; 19586) and argued that it was Ascaridoid. In this I was clearly wrong. I have now studied the Ascaridoidea (Inglis, 1965, 1965(2) and am convinced by the overwhelming weight of evidence that the Ascaridiidae are heterakoids and that their resemblances to the Ascaridoidea are simply due to the convergent development of a massive body size. The form of the male tail and the presence of paired nuclei in the ventral oesophageal glands are clear features of resemblance with the Heterakoidea and rule against ascaridoid affinities, as does the life history. Equally significant is the structure of the head which in the Ascaridiidae is heter- akoid in plan and shows no point of resemblance with that in the Ascaridoidea, other than gross size. The lips in the Ascaridoidea are operated by a septum of three systems of non-contractile fibres (Inglis, 1965) and the cheilorhabdion (i.e. the cuticle lining the inner surface of the cheilostome: Inglis, 1966, 1967) is never markedly sclerotized. In Ascaridiidae the conditions are quite different. The cheilorhabdion is modified into a definite heavily sclerotized region in each lip while the fibre system, which is present, is not organized as three subsiduary systems but is sparse except in the inter-labial regions from which it fans out into each contiguous lip. Exactly the same occurs in the heterakids and aspidoderids. Further the arcade system and associated cells bears no resemblance to that of the Ascaridoidea (Hartwich, 1957)- I continue to treat the Aspidoderidae as a distinct family in spite of Chabaud (1957; 1965) who considers it a subfamily of the Heterakidae. I did at one time agree with Chabaud over this point but further consideration leads me to conclude that to do this masks the many and manifest differences between the aspidoderids and the heterakids. The forms included within the Aspidoderidae differ in so many features from those referred to the Heterakidae, they form such a specialized, io W. G. INGLIS geographically restricted group and reach such a degree of independent specializa- tion in a form such as Lauroia, that I look on them as being as distinct from the Heterakidae as are the Ascaridiidae. Lauroia is a particularly aberrant genus at first sight and this has led to its being referred to a unique family within a different Superfamily by Skrjabin et al. (1961) but I have studied some of the syntype series of L. travassosi Proenga, 1938 (type species of the genus) and L. intermedia Caballero, 1955. The lips have the inter- connecting lobes typical of the aspidoderids and this may be considered a specialized form of the head found in Paraspidodera. Further the reduction in the structures of the male tail is so obviously secondary, and has involved all structures, that little stress can be laid on this alone. But in L, trinidadensis Cameron, 1939 there is a definite small bump anterior to the pre-cloacal opening which can be considered to be the reduced remnants of a pre-cloacal sucker while the pre-cloacal modification in L. intermedia is even more obviously sucker-like. Family HETERAKIDAE The family Heterakidae, as defined here, contains three distinct morphological groups. The first is characterized by (i) a short, rather stout oesophagus, (2) rather large lips, (3) interlabia (or at least structures which may be interpreted as homo- logous with inter-labia), (4) a rather large pre-cloacal sucker, and (5) very broad caudal alae supported by long, narrow papillae. The second is characterized by (i) rather small lips which are not set-off from the body, (2) by a relatively small pre- cloacal sucker, (3) by a flap over the vulvar opening, (4) an excretory pore which leads into a lobulate excretory vesicle and (5) by rather narrow caudal alae on the male tail which are supported by three rather small fleshy papillae. The third group is much less homogenous than the others but is characterized by (i) relatively small lips (although the anterior cuticular flange may be large) which are clearly off-set from the remainder of the body by distinct shoulders, (2) by prominent lateral lines, (3) no trace of inter-labia, (4) a relatively long, and narrow oesophagus, and (5) spicules which are always identical and relatively simple in structure. The structure of the male tail is more variable and will be discussed more fully below. These three groups are treated as subfamilies, thus: (i) Heterakinae, (2) Meteter- akinae and (3) Spinicaudinae. Family ASPIDODERIDAE The family Aspidoderidae consists of two distinct groups in one of which there are inter-labia modified as cordons while in the other there is no trace of inter-labia or cordons. These two groups are treated as subfamilies: Aspidoderinae and Lauroiinae respectively. Family ASCARIDIIDAE The family Ascaridiidae contains only one genus, Ascaridia. These taxa are diagnosed formally later. THE SUPERFAMILY HETERAKOIDEA n MORPHOLOGICAL AND EVOLUTIONARY TRENDS Species of the superfamily occur in all vertebrate groups from amphibia to birds with a few representatives in mammals, except in South American where the Family Aspidoderidae is restricted to Mammals (marsupials, edentates and histricomorph rodents). In view of the great overall similarities between the component taxa it appears reasonable to consider that the superfamily represents a mono-phyletic group. Within the family Heterakidae there is a sequence along which is a tendency towards a reduction in the number of caudal papillae, with an increase in their size and the associated appearance and increase in the size of the caudal alae. Concomitantly there is a reduction and finally a loss of the gubernaculum, an increase in the relative complexity of the spicules, an increase in the relative size of the pre- cloacal sucker and a loss of papillae on the general body surface with the development of wide lateral alae. There is also a tendency towards the development of cephalic cordons, by the modification of inter-labia associated with the lips, or by the modification of the lips alone. That these trends have developed in the order described is supported by the host distribution of the parasites. The genera Spinicauda, Africana and Meteterakis, all of which are characterized by papillae on the general surface, relatively large numbers of caudal papillae, a gubernaculum or gubernacular mass, poorly developed caudal alae (when such alae are present), a relatively small pre-cloacal sucker, equal and identical spicules and no interlabia, are restricted to reptiles and amphibia. The genera Heterakis, Pseudaspidodera and Odonototerakis, on the other hand, are characterized by no papillae on the body surface, relatively small numbers of elongate caudal papillae, no gubernaculum or gubernacular mass, well developed caudal alae, a relatively large pre-cloacal sucker, spicules which are frequently unequal and dissimilar and almost invariably complex, with inter-labia modified as cordons in Pseudaspidodera and Odontoterakis and are restricted, typically, to birds. In- dependently the genus Gireterakis has developed straight cordons, has no papillae on the general body surface, has very elaborate spicules and is known only from the mammal genus Hystrix. It is now possible to suggest that the form ancestral to the superfamily possessed most of the following characters, although not necessarily all, (i) an anterior cuticular flange which did not project anterior to the mass of the lips; (2) no interlabia; (3) papillae on the general body surface; (4) a relatively long oesophagus with a small posterior oesophageal bulb; (5) a relatively small pre-cloacal sucker on the male tail; (6) no caudal alae on the male tail; (7) equal and identical spicules; (8) many small, sessile caudal papillae on a long, narrow male tail. The genus Spinicauda possesses a combination of characters almost identical with those postulated above for an ancestral form, diverging only in that the anterior cuticular flange of the lips projects slightly anterior to the main mass of the lips, and may be considered as containing the most primitive species of the superfamily (i.e. the most generalized). Non morphological supporting evidence is given by its world-wide occurrance within tropical and subtropical regions and its restriction to reptiles. The characters of the male tail are almost identical with those 12 W. G. INGLIS postulated for an ancestral form and as it is in just those characters that Spinicauda resembles the Aspidoderidae I consider the form of the male tail in that family to be relatively unspecialized. The structure of the head in Spinicauda fits into an almost perfect series of in- creasing elaboration with Africana, in which the anterior cuticular flange is wholly within the limits of the lips, Spinicauda in which the flanges projects slightly anterior to the main mass of the lips and Strongyluris where the flange forms the major part of the lips and dominates the lip-mass completely (Inglis, 1957). It may be pointed out that Africana is a close rival to Spinicauda for the position of the most primitive genus and may be so considered without affecting the argument presented here. The connection between the genera Spinicauda and Strongyluris is accentuated by the forms referred to the genus Moaciria (formerly treated as a subgenus of Spinicauda}. In this latter genus the male tail is intermediate in form between the long narrow tail of Spinicauda and the truncate tail, with broad alae, of Strongyluris. This transitional form of the male tail is associated with a head which is identical in structure with that of Spinicauda. The genus Africana appears to be atypical in being the only genus of the super- family in which the caudal alae are not supported by caudal papillae, although this could be interpreted as an intermediate stage between the forms without alae and those with alae supported by papillae. The similarities between Spinicauda, Africana, Moaciria and Strongyluris are so great and, in addition, can be arranged in such a distinct morphological sequence with fairly clear intermediates culminating in the specialized form of Strongyluris that they are treated as a second evolutionary sequence recognized as the subfamily Spinicaudinae. The remainder of the family Heterakidae falls into two distinct groups, recognized as subfamilies, Heterakinae and Meteterakinae, of which the latter appears to be the result of radiation within a restricted geographical area. The species referred to Meteterakis may be considered more primitive than that in Gireterakis in possessing a gubernacular mass and in having papillae in the general body surface. M. tri- aculeatus without a gubernacular mass being to that extent intermediate between the two genera. The subfamily Heterakinae contains the species which are parasitic in birds and are the most highly evolved parasites, morphologically, within the family. Re- presentatives of the subfamily occur in ground feeding, grain-eating birds throughout the world but have become sufficiently distinct in two geographical regions to warrant the recognition of two genera, Odontoterakis in South American tinamous and Pseudaspidodera in Indian peafowls. The inclusion of Pseudaspidodera within the Family Aspidoderidae by Skrjabin et al. 1961 is completely unacceptable since not only does it differ from the typical aspidoderids in the structure of the head (Inglis, 1957) but also differs in the form of the male tail with its broad, typically heterakid caudal alae, and the form of the dissimilar spicules. The resemblance must in fact be dismissed as the crudest and most superficial convergence of one morphological feature and if Pseudaspidodera is grouped with the aspidoderids so must the genus Odontoterakis, which Skrjabin et al. do not do. THE SUPERFAMILY HETERAKOIDEA 13 It is also worth noting that the species referred to the genus Heterakis fall into two groups characterized by the structure of the spicules. The groups are not those in which the spicules are equal or unequal (as recognized by many authors, Lopez- Neyra (1947), Skrjabin & Schikhobalova (1947), Freitas (1956) and most recently Skrjabin et al. (1961)), but are based on whether the spicules are identical in gross structure although they can be of different lengths (e.i. non-alate without elaborate tips) or are dissimilar (i.e. one, the left, usually bears broad alae and has an elaborate tip while the right is simple and needle-like). This difference is also geographic since the first group is most frequent in the Americas while the other is most characteristic of the European and Asiatic parasites. This difference is also re- flected by the structure of the spicules in the neotropical genus Odontoterakis, in which the spicules are never alate and never have elaborate tips. The oriental genus Pseudaspidodera in contrast is characterized by dissimilar spicules in which the left spicule bears broad alae while the right is needle-like. The family Aspidoderidae represents the radiation of a stock isolated in South American mammals. In this group most of the diversity is confined to the anterior end of the body, and superficially parallels the conditions in Pseudaspidodera and Odontoterakis but there are marked differences in the details. Thus, in the heterakids the cordons, when present, never remain the same width throughout their length even when they anastomose (accepting that they do anastomose in Pseudaspidodera spinosa, see below) . In the family Aspidoderidae the cordons always remain the same width along their full length. The Aspidoderidae appear to have divided into two lines: one in which inter-labia modified as cordons are present and one in which inter-labia have not appeared. Whether the species referred to Lauroia have been derived from the forms grouped in Paraspidodera is an open question but in view of the similarities of the head structures the two genera are classified in the same subfamily, Lauroiinae. The genus Ascaridia is clearly heterakoid in affinity but appears to represent a group which diverged fairly early and probably shows its greatest affinities to the Spinicaudinae. GEOGRAPHICAL AND HOST DISTRIBUTION The superfamily as a whole is cosmopolitan in distribution and occurs in all vertebrate hosts groups from amphibia to mammals and birds, but it is in the latter group that it is most frequent and wide-spread. The distribution of the major subdivisions of the superfamily is interesting since it reflects a combination of host and geographical restrictions. Thus, the subfamily Spinicaudinae is cosmopolitan with forms occurring in all major geographical regions although restricted to reptiles and amphibia. Even the genera within the subfamily tend to be wide-spread. Spinicauda occurs in South America, Africa, Madagascar and Australia, with one species in each area, while Strongyluris is extremely widely spread in tropical and subtropical regions and is represented by a much larger number of species. In contrast Africana is restricted to Africa, and Moaciria to South America, Africa and Madagascar. Strongyluris appears to represent the culmination of the Spinicaudinae i 4 W. G. INGLIS and is as successful as a group of parasites in reptiles as the genus Heterakis is in birds. The subfamily Heterakinae is equally wide-spread but is largely restricted to ground-feeding, grain-eating birds i.e. Galliformes (one species in mammals; H. spalacis and H. macrospiculum are not typical of the genus Heterakis and are best considered incertae sedis although the latter shows similarities to Africana, particularly in the structure of the male tail and spicules.) The genus Heterakis is cosmopolitan and ubiquitous but the other genera of the subfamily are geographically restricted with Odontoterakis in South America and Pseudaspidodera in the Indian region. The family Aspidoderidae is the only group of the superfamily which occurs widely in mammals and it also is geographically restricted to South America while Ascaridia, although cosmopolitan, is most commonly reported from members of the Columbi- formes and Psittaciformes although it is not uncommon in the Galliformes. The patterns suggest that the evolution of the superfamily has been due to a combination of (i) intra-host restriction to the caecum, (2) host restriction to ground-feeding grain-eating birds and (3) geographical restriction. There is no evidence to suggest any close host : parasite parallelism. The impression is rather of a group which has evolved and expanded to occupy all the space available to it within the constrains of the ecology and feeding habits of the hosts (see Inglis, 1965^. Thus the members of the superfamily exist in a specialized, selected, locality within the host, in all birds in which they occur. But they only occur in a restricted range of hosts with similar feeding habits. Then the various smaller taxa are largely determined geographically. Nevertheless the Aspidoderidae represent a special case in that they are not only geographically isolated but also occur in a long isolated group of mammalian hosts, the South American marsupials and the ground-feeding edentates. It should be noted that they do not occur in the ant-eaters or the tree- dwelling sloths. It is interesting to wonder, although impossible to know, if they continue to exist in South America because there is no competition with other nematode parasites in the large intestine where they occur or whether they represent a new group of parasites which has replaced some other group. In other words are they a remnant group or are they the result of a specialized radiation within an isolated geographical and host locality? Be that as it may the Heterakoidea is clearly an old group which is supremely successful within the hosts groups in which it occurs. SYSTEMATIC HISTORY Railliet and Henry (1914) considered that the species of the family Ascaridae for which they had created a new subfamily Heterakinae in 1912 were sufficiently distinct to warrant their further separation into a distinct family, Heterakidae. This family was diagnosed largely on the presence of a pre-cloacal sucker on the male tail and included the genera Heterakis, Ascaridia, Aspidodera, Cissophylus and Subulura, with Strongyluris as a subgenus of Heterakis and Oxynema as a subgenus of Subulura. No groupings higher than genera were proposed. THE SUPERFAMILY HETERAKOIDEA 15 Travassos (1920) introduced two new genera, Spinicauda and Africana, and later in the same year (Travassos, 19200) he divided the family Heterakidae into three subfamilies : Heterakinae, with Heterakis, Ganguleterakis , Aspidodera, Para- spidodera and Gireterakis ; Spinicaudinae nov. for Spinicauda, Africana and Strongyluris ; Subulurinae for Subulura, Heteroxynema, Oxynema and Numidica. The last subfamily differed from both the others in having in the males an elongate pre-cloacal sucker without a definite rim while the other two had a pre-cloacal sucker with a definite cuticular rim. The diagnostic differences between the other two subfamilies were vague and appear to have been mainly the restriction of the species of the first to warm-blooded hosts and those of the second to cold-blooded hosts. Yorke & Maplestone (1926) considered the subfamily Subulurinae a family and did not recognize the subfamily Spinicaudinae, thus leaving the family Heterakidae with no groups higher than genera. Baylis & Daubney, slightly later in the same year (1926), still treated the subulurids as a subfamily but this was not generally accepted. No further radical changes were proposed in the classification of the Heterakidae until Skrjabin and Schikhobalova (1947) recognized two subfamilies: Heterakinae and Aspidoderinae nov. This grouping was largely based on published descriptions and the subfamilies were considered distinct almost wholly on the presence or absence of cephalic cordons. In 1957 I described the comparative anatomy of the head in the family and argued that the classification of Skrjabin and Schikhobalova was unnatural because cephalic cordons appeared to have been developed in- dependently three times. I then proposed that the genera Aspidodera, Ansirupto- dera, Sexansodera and Paraspidodera be placed in a separate family, Aspidoderidae with two subfamilies : Aspidoderinae for the first three of the above genera and Paraspidoderinae nov. for the fourth. Independently Freitas (1956) reviewed the family and proposed four new families, four new subfamilies and five new genera (four with one species each and one with two species). This classification was, and still is, unacceptable. There are five families of which two have only one species, there are eight subfamilies of which three have only one genus each, there are twenty- two genera of which eleven have only one species each and four have only two species each. This classification was based largely on published descriptions and every error of observation and every faulty description appears to have been recognized by a distinct genus, at least, and in some cases a subfamily or even family. Then Chabaud (1957), in considering the classification of the suborder Ascaridina introduced a new superfamily Heterakoidea which he considered to contain two families and four subfamilies thus: Heterakidae, with four subfamilies Heterakinae (with sixteen genera or subgenera), Aspidoderinae (with four genera or subgenera), Lauroiinae (with one genus) and Schneidernematinae (with one genus); and the family Ascaridiidae with one genus, Ascaridia. This classification is largely that of Inglis (1957) and Freitas (1956) re-arranged. In 1958 I proposed the classification which I am justifying here and will not discuss it now accept to draw attention to the fact that Schneidernema and Morgascaridia (a genus I proposed for Paraspidodera sellsi, Morgan, 1928 : Inglis, 19586) were 16 W. G. INGLIS removed from the Heterakidae first to the Ascaridiidae (Inglis, 19586) and later to the Superfamily Seuratoidea (Inglis & Chabaud, 1958) as members of a distinct sub- family Schneidernematinae. Most recently Skrjabin, Schikhobalova & Lagodovskaja (1961) have presented a re-arrangement of the Heterakoidea, with four families, thus : Heterakidae : Heter- akinae (Heterakis, Ganguletemkis and Odontoterakis); Meteterakinae (Meteterakis; and Gireterakis}. Aspidoderidae : Aspidoderinae (Aspidodera, Ansiruptodera and Sexansodera) Spinaspidoderinae (Spinaspidodera and Pseudaspidodera}. Spinicaudidae : Spinicaudinae (Spinicauda, Africana and Paraspidodera} ; Preter- akinae (Preterakis). Strongyluridae : Strongylurinae (Strongyluris and Moaciria}. In addition they refer Lauroia to a distinct family within the Cosmocercoidea and introduce a new genus, Bellaplectana for Aplectana pharyngeodentata Belle, 1957. It should, perhaps, be pointed out that Yamaguti (1961) has grouped the Heter- akoidea in yet another way but this does not warrant serious consideration except to dismiss the new genus Cheloniheterakis. This genus was introduced for two old and inadequately described species from Testudo. No other heterakids are known from this host genus in which members of the family Kathlaniidae are fairly common parasites. As the two species referred to Cheloniheterakis possess circular precloacal suckers, which are common in Kathlaniids it is more probable they are referable to that family. Be that as it may the two species are species dubia and the genus must be ignored as unrecognizable. The Skrjabin et al. classification does warrant consideration and I will give my arguments against accepting it. Part of it can be dismissed quickly and easily as due to classifying errors of description. The arguments against the genus Bella- plectana have been adumbrated at the beginning of this paper but to recapitulate briefly the so-called teeth in the head, upon which the genus is based, do not exist they are simply the retracted lips in poorly fixed specimens. The genus Preterakis and the associated subfamily Preterakinae must disappear since the fusion of the caudal alae on the ventral surface of the male body is a fixation artefact and does not occur in all the specimens of the type series. The remaining arguments are more expressions of opinion and of assessing the weight of the morphological and other evidence, thus : The reference of the aspidoderids and Pseudaspidodera to the same family is a continuation of the position taken by Skrjabin & Schikhobalova (1947) but is still un-acceptable and morphologically indefensable. If this grouping is advocated there is no real reason for not referring Odontoterakis to the same family, but this genus is referred by Skrjabin et al. (1961) to the Heterakidae. I stress, and repeat what I wrote before, the only resemblance between the Aspidoderidae and Pseud- aspidodera is in the common possession of structures called by the same name, i.e. cordons. The family Aspidoderidae (sensu Skrjabin et al.} is otherwise characterized (in addition to cordons) by the presence of a cephalic cap, inter-connecting lobes to the lips, a long slim oesophagus and, in the male, is further characterized by equal and identical square-ended spicules, no caudal alae and the presence of many sessile THE SUPERFAMILY HETERAKOIDEA 17 papillae. In all these respects Pseudaspidodera differs from the Aspidoderids and in all the corresponding features is identical with the genera referred by Skrjabin et at. to the subfamily Heterakinae. Equally the reference of Paraspidodera to the Spinicaudinae is unacceptable since its only resemblance to the other genera of that subfamily is in lacking caudal alae and cephalic cordons. In all other respects, the structure of the head, the oesopha- gus, and the male's tail and spicules it is indistinguishable from the forms referred to the Aspidoderidae. The reference of Lauroia to the superfamily Cosmocercoidea is equally contradicated by the morphology. The structure of the anterior end of the body is very similar to that of the Aspidoderidae and the remnants of a pre- cloacal sucker are clearly present in at least two species. The separation of Strongyluris and Moaciria (which I now accept as a distinct genus) into a separate family is equally unacceptable since Strongyluris so clearly represents the end of a morphological sequence running from Spinicauda to Moaciria to Strongyluris. The structure of the head in Moaciria is identical with that in Spinicauda while the male tail forms an almost perfect intermediate between that genus and Strongyluris. I therefore propose the following classification for the Superfamily Heterakoidea. ANNOTATED CLASSIFICATION OF THE SUPERFAMILY On the basis of the arguments presented above the following classification is proposed. Notes are appendaged to each group to draw attention to points of interest and to explain some of the names used. In these notes some of the argument given above is briefly repeated so that this section is more or less complete in itself. HETERAKOIDEA (Ralliet & Henry, 1912) Chabaud, 1957 Ascaridida : three large distinct lips, one dorsal with two double papillae, two ventral with a double ventral papilla and a single lateral papilla each : lining of mouth cavity (i.e. cheilorhabdion) sclerotized and forming a flange along the anterior edge of each lip in all forms except Ascaridia ; anterior end of oesophagus divided into three projecting lobes, one to each lip, except in Ascaridia ; nuclei of ventral oesophageal glands double ; no intestinal or oesophageal appendices. MALE : circular pre-cloacal sucker with a definite cuticular rim ; two spicules. FEMALE : eggs not embryonate in utero, thick-shelled ; vulva opening about middle of body. LIFE HISTORY : Direct, with or without a migration within the final host. HOSTS : Restricted to the intestine (more specifically to the caecum in birds and the large intestine in mammals) of ground feeding birds and mammals. In amphibia (anurans) and reptiles (saurians) restricted to terrestrial feeders. GEOGRAPHICAL DISTRIBUTION : Cosmopolitan as a superfamily. i8 W. G. INGLIS HETERAKIDAE Railliet & Henry, 1912 Heterakoidea : lips rounded, not connected by lateral lobes ; cordons when present poorly developed and not remaining the same width throughout their length ; oesophagus relatively short and stout. MALES : caudal alae typically present ; number of caudal papillae relatively low ; spicules may be unequal and dissimilar ; gubernaculum generally reduced or absent ; tail generally relatively short. HOSTS : amphibia, reptiles and (mainly) birds. One species in mammals. DISTRIBUTION : Cosmopolitan as a family. TYPE GENUS : Heterakis Dujardin, 1845. HETERAKINAE Railliet & Henry, 1912 Heterakidae : lips not off-set from body ; anterior cuticular flange of lips generally not projecting beyond main lip mass ; inter-labia, or their homologues, present. MALE : precloacal sucker relatively large ; caudal alae broad, supported by long, narrow papillae ; gubernaculum lacking ; spicules frequently dissimilar. HOSTS : mainly birds, one exceptional species in mammals (rodents). DISTRIBUTION : Cosmopolitan. The subfamily is a homogeneous group in which three subgroups can be recog- nized. Odontoterakis is South America, Pseudaspidodera in India and Heterakis which is cosmopolitan. The latter genus is characterized by lacking definite cephalic cordons such as characterize the other two genera. But there is a tendency for there to be marked " bumps " in the inter-labial spaces which may be considered as precursors of the inter-labia which are modified as cordons in the other two genera. It is worth pointing out that what are here called cordons are identical in form and origin with the so-called labial grooves of some members of the Ascari- doidea, e.g. Porrocaecum and Multicaecum. These labial grooves are also formed from two sources, one the inter-labia and the other the lips, and also consist of open grooves. The morphological differences between Pseudaspidodera and Odontoterakis tend to be slight but in the former genus, in addition to the cordons being more strongly developed, the left spicule is always shorter than the right, is always broadly alate while the right is slim and needle-like, and generally has an elaborate posterior tip. In contrast the spicules in Odontoterakis are always identical in structure, never bear alae, and always end posteriorly in simple points. I therefore interpret the two groups as having developed in specialized host groups within isolated geographical areas since Pseudaspidodera appears to have arisen in peafowls in India and Odontoterakis to have arisen in tinamous in South America. Both host groups being typically ground dwelling forms with poor powers of locomotion. THE SUPERFAMILY HETERAKOIDEA 19 Heterakis Dujardin, 1845 Heterakinae : lips without definite inter-labia or cordons. MALES : spicules frequently dissimilar but may be equal and identical. TYPE SPECIES : Ascaris gallinarum Schrank, 1788. HOST AND GEOGRAPHICAL DISTRIBUTION : ground feeding birds, mainly Galli- formes, throughout the world. Several attempts have been made to divide this genus into two on the basis of the relative lengths of the spicules, species with equal spicules to one genus and with unequal to another. Both Lopez-Neyra (1947) and Skrjabin & Schikhobalova (1947^) independently attempted to do this, using the name Ganguleterakis for species with equal spicules. Madsen (1950), in a generally excellent and outstanding publication, commented upon this unfavourably pointing out the great difficulty in many cases in deciding whether the spicules should be called equal or unequal. Freitas (1956) attempted to over-come this difficulty by defining his two subgenera, Heterakis and Raillietakis, so that unequal spicules were considered to be those in which the difference in length was at least one third the length of the shorter spicules. Even with this qualification it is difficult to imagine that equality or its lack is a sufficiently fundamental character upon which to recognize genera or subgenera. Skrjabin et al. (1961) continue to recognize two groups on just this difference in the lengths of the spicules but it is still unacceptable. It is so very clearly a biblio- graphical key character resorted to in an attempt to reduce the number of species within each genus and can only have been used because it is one of the few characters which can be determined from literature rather than specimens. The structure and relative lengths of the spicules, nevertheless afford good characters in the delimitation of species within the genus used in conjunction with the number of caudal papillae, the relative size and position of the pre-cloacal sucker and the relative length of the male tail. Many authors have laid stress on the position of various caudal papillae, e.g. the number of pre-cloacal papillae, but as Maplestone (1932) and Madsen (1950), among others, have pointed out, the value of these characters is very limited. The typical arrangement of the caudal papillae is described above but it should be noted that it is relatively common for an additional pair to occur between the para-cloaca and suctorial groups or for one of the pairs of para-cloacals to be missing. There is also a tendency, in the species occurring in South America, for a pair of long papillae to be present just anterior to the peri-cloacal group, a pair which I consider homologous with the anterior pair of peri-cloacals since in specimens on which the former pair of papillae are present the latter pair is consistently absent. Both Maplestone (1932) and Madsen (1950) have pointed out that it is, in most cases, impossible to find characters by which the females of Heterakis may be distinguished. This applies throughout the entire superfamily. I have been unable to find any character, or combination of characters by which females alone can be identified. In spite of my criticism of the attempts to divide the genus Heterakis on the basis 20 W. G. INGLIS of the relative lengths of the spicules two subgroups can be recognized : (i) one in which the spicules are identical in structure and never bear alae and (2) one in which the spicules are dissimilar with the right slim and needle-like and the left broadly alate. In this latter group the posterior end of the left spicule is frequently modified into a hook or barb. In both groups the spicules may be unequal in length but in the first group when there is any modification on the posterior end it is always identical in both spicules. Thus on this basis H. gallinarum and H. isolonche, which only differ in the relative lengths of the spicules, fall together while on the basis of un- equality they are completely separated. Lane (1914) introduced the generic name Ganguleterakis for the species generally known as H. spumosa Schneider, 1866. There is a case for recognizing this genus as it contains the only species of Heterakis from mammals (H. spalacis and H. longi- spiculum incertae sedis] but, except for a reduction in the number of caudal papillae, in which it is paralleled by H. psophiae Travassos, 1913, it is a typical Heterakis. I do not, therefore, recognize this genus. Odontoterakis Skrjabin & Schikhobalova, 19470: Heterakinae : interlabia modified as non-recurrent cordons. MALES : spicules simple non-alate and identical in structure. TYPE SPECIES : Heterakis crypturi Baylis, 1944. HOST AND GEOGRAPHICAL DISTRIBUTION : Birds, mainly tinamous, in South America. Note : A restudy of the types of Heterakis interlabiata Ortlepp, 1923 has convinced me that Mendonga (1953) is correct in treating it as a synonym of H. isolonche. Pseudaspidodera Baylis & Daubney, 1922 Heterakinae : inter-labia modified as recurrent cordons. MALES : spicules dissimilar and unequal, right slim and needle-like, left with broad alae. TYPE SPECIES : Pseudaspidodera pavonis Baylis & Daubney, 1922. HOST and GEOGRAPHICAL DISTRIBUTION : Phasianid birds, India. P. jnanendrae Chakravarty, 1938 was described as having anastomosing cordons but in the type, a female, the head is very contracted and it is probable that the apparent fusion is an artefact. P. spinosa Maplestone, 1932 was described as having spines in the cordons which anastomose. As a result a genus Spinaspidodera was proposed by Skrjabin & Schikhobalova (1947) for it. There is some justification for this but in view of the small number of species involved and as Maplestone's species has never been rediscovered and the types are lost I prefer not to recognize this genus. It is worth noting that the cordons in this case, if they do definitely anastomose, vary in width along their length and do not remain the same width as in the members of the Aspidoderidae. THE SUPERFAMILY HETERAKOIDEA 21 METETERAKINAE Inglis, 19580 Heterakidae : head with three rounded lips, without interlabia ; oesophagus relatively long with a small postoesophageal bulb ; excretory pore opening into a large lobulate vesicle ; lateral alae running full length of body. MALE : spicules equal in length ; caudal alae present, supported by three pairs, exceptionally four, of large fleshy papillae one pair lateral to the cloacal opening and typically two pairs, exceptionally three, lateral to the small pre-cloacal sucker ; a large number generally about twenty pairs, of small sessile papillae also present. FEMALE : vulvar opening covered by a flap developed from the anterior lip ; tail relatively very long. TYPE GENUS : Meteterakis Karve, 1930. HOST AND GEOGRAPHICAL DISTRIBUTION : Amphibia, reptiles and mammals (Hystrix} in the oriental region (s.l. : India, Ceylon, China, Japan, East Indies, Burma) . Meteterakis Karve, 1930 Meteterakinae : head without cordons ; body generally with many small sessile papillae scattered over surface. MALES : spicules identical ; indefinite gubernacular mass developed from the walls of the cloaca except in M. triaculeata ; posterior lip of cloacal opening covered by small granulations. TYPE SPECIES : Meteterakis govindi Karve, 1930. HOST AND GEOGRAPHICAL DISTRIBUTION : amphibia and reptiles, as for sub- family. Gireterakis Lane, 1917 Meteterakinae : anterior cuticular flange of lips not projecting anterior to main lip mass ; three straight, simple cordons, one arising from each inter-labial space. MALE : number of sessile caudal papillae relatively small ; spicules elaborate ; no gubernacular mass present. TYPE SPECIES : Gireterakis girardi Lane, 1917. HOST AND GEOGRAPHICAL DISTRIBUTION i Hystrix species in India. SPINICAUDINAE Travassos, 1920 Heterakidae : lips off-set from body ; no inter-labia or cordons ; lateral fields prominent ; papillae on body. MALE : spicules equal, identical and simple ; papillae do not support caudal alae when alae present except when the tail is short and the pre-cloacal sucker is directed posteriorly. TYPE GENUS : Spinicauda Travassos, 1920. HOST AND GEOGRAPHICAL DISTRIBUTION : Amphibia and reptiles ; cosmopolitan in the tropics and sub-tropics. 22 W. G. INGLIS This subfamily is morphologically the least uniform, and therefore the most difficult to diagnoze, in the superfamily Heterakoidea. This is mainly because Spinicauda is morphologically generalized while Strongyluris is highly modified. A trend in the modification of the head is apparent from Africana, with a simple anterior cuticular flange, to Spinicauda and Moaciria in which the flange projects slightly beyond the lip mass to Strongyluris in which the flange forms the major part of the lip. Moaciria forms an almost perfect intermediate stage between Spinicauda and Strongyluris since it resembles Spinicauda in the presence of a gubernaculum and in having the same head form and resembles Strongyluris in the posteriorly directed pre-cloacal sucker and the presence of caudal alae (narrow) supported by longish papillae. I treated Moaciria as a subgenus of Spinicauda in 1958 but now prefer to treat it as a distinct genus. Spinicauda Travassos, 19200 Spinicaudinae : anterior cuticular flange projects slightly beyond the main mass of the lips. MALE : tail relatively long, tapering evenly to a point ; pre-cloacal sucker relatively small ; gubernaculum present ; caudal papillae small and sessile. TYPE SPECIES : Ascaris Spinicauda Rudolphi, 1819. HOST AND GEOGRAPHICAL DISTRIBUTION : Reptiles in South America, Africa, Madagascar and Australia (Queensland), i.e. tropical and subtropical. Africana Travassos, 1920 Spinicaudinae : anterior cuticular flange not projecting anterior to lips ; lateral alae present. MALES : pre-cloacal sucker small ; spicules equal and identical, alate ; guberna- cular mass present ; small sessile papillae on tail ; caudal alae not supported by papillae ; caudal alae stop about level of cloacal opening. TYPE SPECIES : Heterakis africana Gendre, 1909. HOST AND GEOGRAPHICAL DISTRIBUTION : Amphibia and reptiles in tropical and subtropical Africa. Moaciria Freitas, 19560 Spinicaudinae : anterior cuticular flange projecting slightly anterior to mass of lips. MALE : tail relatively long with relatively few stoutish papillae supporting narrow caudal alae ; posteriorly directed pre-cloacal sucker ; gubernaculum present. TYPE SPECIES : Moaciria alvarengai Freitas, 19560. HOST AND GEOGRAPHICAL DISTRIBUTION : Reptiles in South America, Africa and Madagascar. THE SUPERFAMILY HETERAKOIDEA 23 Strongyluris Mueller, 1894 Spinicaudinae : anterior cuticular flange projecting beyond main lip mass so that it forms major portion of lip ; sessile papillae scattered over body surface. MALE : posterior end obliquely truncate ; pre-cloacal sucker relatively large and directed posteriorly ; caudal alae well developed, supported by rather long, stout caudal papillae ; spicules equal, without alae ; no gubernaculum or gubernacular mass present. TYPE SPECIES : Strongyluris brevicaudata Mueller, 1894. HOST AND GEOGRAPHICAL DISTRIBUTION : Mainly reptiles, rarely amphibians, cosmopolitan within the tropics and subtropics. ASPIDODERIDAE Skrjabin & Schikhobalova, 1947 Heterakoidea : lips square connected by lateral lobes ; cuticle at anterior end of body thickened to form a cephalic cap ; oesophagus long and narrow with a small posterior oesophageal bulb ; cordons, when present, the same width along their lengths. MALES : no caudal alae ; gubernaculum generally present ; spicules equal in length and identical in structure, generally massive ; generally many small sessile papillae on long, narrow tail ; caudal papillae all about the same size. TYPE GENUS : Aspidodera Railliet & Henry, 1912. HOSTS AND GEOGRAPHICAL DISTRIBUTION : Marsupials, edentates and occasionally hystricomorphs in South America and related areas. ASPIDODERINAE Skrjabin & Schikhobalova, 1947 Aspidoderidae: cephalic cordons present, lateral lobes of lips complex. TYPE GENUS : Aspidodera Railliet & Henry, 1912. DISTRIBUTION : As family. Aspidodera Railliet & Henry, 1912 Aspidoderinae : cephalic cordons recurrent and anastomosing. TYPE SPECIES: Aspidodera scoleciformes Diesing, 1851. Ansiruptodera Skrjabin & Schikhobalova, 1947 Aspidoderinae : cephalic cordons not anastomosing. TYPE SPECIES : Aspidodera ansiruptodera Proenga, 1937. LAUROIINAE Skrjabin & Schikhobalova, 1951 Aspidoderidae : cephalic cordons lacking ; lateral lobes of lips simple. TYPE GENUS : Lauroia Proensa, 1938. 24 W. G. INGLIS HOST AND GEOGRAPHICAL DISTRIBUTION : Edentates and hystricomorphs, as the family. Lauroia Proenca, 1938 Lauroiinae : cephalic cap modified into three " plates " which are slightly undercut posteriorly. This modification can be interpreted as due to the develop- ment of cordons which do not involve inter-labia as in Girder akis. MALE : posterior end simple without a gubernaculum, relatively few papillae, and a reduced pre-cloacal sucker. TYPE SPECIES : Lauroia travassosi Proenga, 1938. Paraspidodera Travassos, 1914 Lauroiinae : cephalic cap not modified as " plates ". MALE: posterior end relatively complex, with a gubernaculum; relatively large number of caudal papillae and a well developed pre-cloacal sucker. TYPE SPECIES : Paraspidodera uncinata Rudolphi, 1819. ASCARIDIIDAE Travassos, 1919 Heterakoidea : lips large and stout ; no distinct anterior cuticular flange ; oesophagus club-shaped without a posterior bulb. MALE : caudal alae narrow supported by short stout papillae. TYPE GENUS : Ascaridia Dujardin, 1845. HOST AND GEOGRAPHICAL DISTRIBUTION : Birds, cosmopolitan with rare species in mammals. KEY TO THE GENERA OF THE SUPERFAMILY HETERAKOIDEA (1) (2) Without a tri-valvulate posterior oesophageal bulb . . . ASCARIDIA (p. 24) (2) (i) With a tri-valvulate posterior oesophageal bulb . (3) (3) (18) Head without a cephalic cap, without interconnecting lobes to lips . . (4) (4) (9) Anterior cuticular flange of lip not projecting anterior to main lip mass, no papillae on body, lateral alae present, broad caudal alae supported by long, thin papillae ........... (5) (5) (6) Head without cordons or labial grooves .... HETERAKIS (p. 19) (6) (5) Head with cordons or labial grooves . . . . . . (?) (7) (8) Spicules dissimilar, left with broad alae, right needle-like, restricted to India and associated regions ...... PSEUDASPIDODERA (p. 20) (8) (7) Spicules similar, never alate, restricted to South America ODONTOTERAKIS (p. 20) (9) (15) Narrow lateral alae, papillae on body surface, caudal alae when present either not supported by papillae or supported by many short, stoutish papillae . (10) (10) (n) No caudal alae on male tail . . . . . SPINICAUDA (p. 22) (n) (12) Caudal alae not supported by papillae .... AFRICAN A (p. 22) (12) (15) Caudal alae supported by stout papillae, male tail obliquely truncate, and sucker directed posteriorly ......... (13) THE SUPERFAMILY HETERAKOIDEA 25 (13) (14) Male tail without long terminal spike, anterior cuticular flange major part of lip ... STRONGYLURIS (p. 23) (14) (13) Male tail with long terminal spike, anterior flange not major part of lip MOACIRIA (p. 22) (15) (9) Caudal alae supported by 3-4 stout papillae ...... (16) (16) (17) Straight cordons running posteriorly from interlabial spaces, spicules very Complex ........ . GlRETERAKIS (p. 2l) (17) (16) No cordons, spicules not very complex .... METETERAKIS (p. 21) (18) (3) Head with a cephalic cap, interconnecting lobes to lips .... (19) (19) (22) Head without cordons .......... (20) (20) (21) Head with three cuticular " plates ", pre-cloacal sucker reduced, or lacking LAUROIA (p. 24) (21) (20) Head without cuticular " plates ", sucker normal . . PARASPIDODERA (p. 24) (22) (19) Head with cordons .......... (23) (23) (24) Cordons not anastomosing ..... ANSIRUPTODERA (p. 23) (24) (23) Cordons anastomosing ....... ASPIDODERA (p. 23) ACKNOWLEDGEMENTS I wish to express my thanks to the following who kindly allowed me to study specimens under their care or helped in locating specimens : Dr. E. Belle, Macdonald College, Quebec ; Dr. P. L. G. Benoit, Musee Royal du Congo Beige, Tervuren ; Professor J. J. C. Buckley, London School of Hygiene and Tropical Medicine ; Professor E. E. Bird, University of Georgia, U.S.A. ; M. le Professeur A. G. Chabaud, Museum National d'Histoire Naturelle, Paris ; Dr. R. N. Chaudhuri, Director of the Calcutta School of Tropical Medicine ; Dr. B. S. Chauhan, Zoological Survey of India, Calcutta ; Dr. E. Diechmann, Museum of Comparative Zoology, Harvard ; Dr. L. Forcart, Naturhistorisches Museum, Basel ; Dr. E. Ford, Director of the School of Public Health and Tropical Medicine, Sydney, Australia ; Dr. H. M. Hale, Late Director of the South Australian Museum, Adelaide ; Dr. G. Hartwich, Zoological Museum, Berlin ; The late Dr. S. L. Hora, Zoological Survey of India, Calcutta ; Dr. R. Kellogg, Director of the United States National Museum, Washing- ton ; Professor W. Kershaw, Liverpool School of Tropical Medicine ; Dr. E. Kritscher, Naturhistorisches Museum, Vienna ; Dr. J. T. Lucker, U.S. Department of Agriculture, Beltsville, Maryland ; Dr. J. F. Teixeira de Freitas, Institute Oswaldo Cruz, Brazil ; Mrs. P. M. Thomas, University of Adelaide, Australia ; Professore L. Travassos, Institute Oswaldo Cruz, Brazil ; Professore Umberto d'Ancona, Institute di Zoologia, Padua ; M. le Professeur M. Vachon, Museum National d'Histoire Naturelle, Paris ; Professor H. Vogel, Tropeninstitut, Hamburg. In particular my thanks are due to the following for their kindness and assistance : M. le Professeur A. G. Chabaud and his wife during repeated visits to Paris ; Herr Dr. G. Hartwich and his wife while working in Berlin ; Mrs. P. M. Thomas while working in Adelaide, South Australia ; and Dr. B. S. Chauhan during a short stay in Calcutta. REFERENCES BAYLIS, H. A. & DAUBNEY, R. 1922. Report on the parasitic nematodes in the collections of the Zoological Survey of India. Mem. Indian Mus. 7 : 263-347. - 1926. A synopsis of the families and genera of nematoda. London : British Museum (Natural History), i-xxxvi, 1-277. 26 W. G. INGLIS BELLE, EDWARD A. 1957. Helminth parasites of reptiles, birds and mammals in Egypt. IV. Four new species of oxyurid parasites from reptiles. Canad. J. zool. 35 : 163-169. CABALLERO Y C., E. 1955. Estudios helmintologicos de la region oncocercosa de Mexico y de la Repiiblica de Guatemala. Nematoda. 9* parte. Ada zool. Mex. 1 : 1-5. CAMERON, T. W. M. 1939. Studies on the endoparasitic fauna of Trinidad mammals. VI. Parasites of edentates. Canad. J. Res. 17 D : 249-264. CHABAUD, A. G. 1957. Sur la systematique des nematodes du sous ordre des Ascaridina parasites des vertebres. Bull. Soc. zool. Fr. 82 : 243-253. - 1965. Superfamily des Ascaridoidea, in Traite de zoologie. Anatomic, systematique, biologic. Embranchement des Nemathelminthes (Nematelmia Carl Vogt 1851 Nemathel- minthia Gegenbaur, 1859) ou Aschelminthes (Aschelmintha Grobben, 1910). 4(3), 988-1025. CHAKRAVARTY, G. K. 1938. On a nematode, Pseudaspidodera jnanendrae n.sp. from the peafowl (Pavo cristatus). Parasitology 30 : 167-170. DIESING, C. M. 1851. Systema helminthum. II. Vildebonae. i-vi, 1-588. DUJARDIN, F. 1845. Histoire naturelle des helminthes ou vers intestineaux . Paris. 1-654. FREITAS, J. F. TEIXEIRA DE. 1956. Notas sobre " Heterakidae "; Railliet & Henry. 1914. (Nematoda, Subuluroidea). Rev. bras. Biol. 16 : 461-482. - 1956(1. Novo parasitos de reptil de Ihla Fernando de Noronha : Moaciria alvarengai g.n., sp.n. (Nematoda, Subuluroidea). Rev. bras. Biol. 16 : 335-339. GENDRE, E. 1909. Notes d'helminthologie africaine (Deuxieme note). P.V. Soc. linn. Bordeaux 63 : 33-41. HARTWICH, G. 1954. Die Vorderdarmstrukturen, des Excretions-system sowie der Kopfbau der Ascariden und ihre taxonomische Bedeutung. Wiss. Ztsch. Martin-Luther Univ. Halle-Wittenb. 3 : 1171-1211. INGLIS, WILLIAM G. 1957. The comparative anatomy and systematic significance of the head in the nematode family Heterakidae. Proc. zool. Soc. Land. 128 : 133-143. - 1958. A revision of the nematode superfamily Heterakoidea. Ann. Mag. nat. Hist. (io)12 : 905-912. - 19580. A revision of the nematode genus Meteterakis Karve, 1930. Parasitology 48 : 9-31- 19586. A redescription of the nematode Paraspidodera sellsi Morgan, 1927, and its re- moval to a new genus Morgascaridia. J. Helminth. 32 : 65-72. - 1965. The function and developmental significance of the cephalic septum in the Ascaridoidea (Nematoda). Proc. linn. Soc. Lond. 176 : 23-26. - 19650. The comparative anatomy of the ascaridoid cuticle. Bull. Soc. zool. Fr. 89 : 317- 338. - 19656. Patterns of evolution in parasitic nematodes. in Evolution of Parasites, $rd Symposium, Brit. Soc. Parasit. pp. 79-124. - 1966. The origin and function of the cheilostomal complex in the nematode Falcaustra stewarti. Proc. linn. Soc. Lond. 177 : 57-64. - 19660. The observational basis of homology. Syst. Zool. 15 : 218-228. - 1967. The relationships of the Seuratoidea (Nematoda). /. Helminth, in press. - & CHABAUD, ALAIN G. 1960. Sur la position systematique des Schneidernematinae (Nematoda). Ann. Parasit. hum. comp. 35 : 428-429. KARVE, J. N. 1930. Some parasitic nematodes of frogs and toads. Ann. trop. Med. Parasit. 24 : 481-491. KREIS, H. A. 1933. Em neuer parasitischer Nematode aus Corucia zebrata (Scincidae; Reptilia). Ganguleterakis triaculeatus n.sp. Z. Parasitenk. 6 : 332-338. LANE, CLAYTON. 1914. Suckered round-worms from India and Ceylon. Indian J. med. Res. 2 : 655-669. - 1917. Gireterakis girardi (n.g., n.sp.) and other suckered nematodes. Indian J. med. Res. 4 : 754-765. Lopez-Neyra, C. R. 1947. Helmintos de los Vertebrados Ibericos. II. Granada, pp. 413- 802. THE SUPERFAMILY HETERAKOIDEA 27 MADSEN, HOLGAR. 1950. Studies on species of Heterakis (Nematodes) in birds. Dan. Rev. Game Biol. 1 : 1-42. MAPLESTONE, P. A. 1932. The genera Heterakis and Pseudaspidodera in Indian Hosts. Indian J. med. Res. 20 : 403-420. MENDONCA, J. MACHADO DE. 1953. Heterakis isolonche Linstow, 1906 e Heterakis gallinae (Gmelin, 1790), agentes causais de Tiflite verrucosa em faisoes no Jardin Zoologico de Distrito Federal. Mem. Inst. Osw. Cruz 51 : 675-704. MUELLER, A. 1894. Helminthologische Beobachtungen an bekannten und unbekannten Entozoen. Arch. Naturgesch. 60, 5, 1 : 113-128. ORTLEPP, R. J. 1923. Two new nematodes collected in the Zoological Gardens of London. /. Helminth. 1 : 61-64. PROEN9A, M. C. 1937. Revisao do genero Aspidodera Railliet et Henry, 1912 (Nematoda : Subuluroidea) . Mem. Inst. Osw. Cruz. 32 : 427-438. 1938. S6bre um novo typo de Heterakinae Railliet et Henry, 1912 (Nematoda : Subulu- roidea). Livr. Jubil. Prof. L. Travassos, Rio de Janeiro, pp. 419-420. RAILLIET, A. & HENRY, A. 1912. Quelques nematodes parasites des reptiles. Bull. Soc. Path. exot. 5 : 251-259. - & HENRY, A. 1914. Essai de classification des Heterakidae. Extr. IX e Congr. Int. Zool., Monaco, pp. 674-682. RUDOLPHI, C. A. 1819. Entozoorum synopsis cui Accedunt Mantissa Duplex et Indices Locupletissimi. Berlin, i-iv, i-8n. SKRJABIN, K. I. & SCHIKHOBALOVA, N. P. 1947. (A reconstruction of the systematics of the nematode family Heterakidae). Dokl. Akad. Nauk. SSSR 58 : 718-721. (in Russian). - i947a. (A division of the nematodes belonging to Heterakis into generic components). Dokl. Akad. Nauk SSSR. 58 : 1865-1867. (In Russian). - 1951. (A reconstruction of the classification of the suborder Oxyurata Skrjabin, 1923). Trud. gel'mint. Lab. 20 : 5-8. (In Russian). - & LAGODOVSKAJA, E. A. 1961. [Principles of Nematodology. X. Oxyurata of Animals and Man.] Moscow Acad. Sci. U.S.S.R. (in Russian). TRAVASSOS, LAURO. 1913. Sobre as especies brasileiras da subfamilia Heterakinae Railliet et Henry. Mem. Inst. Osw. Cruz. 5 : 271318. - 1914. Contribui9ao para o conhecimento da fauna helmintologica brasileira. III. Novo genero da familia Heterakidae Railliet et Henry. Mem. Inst. Osw. Cruz 6 : 137-142. 1920. Contribui9ao para o conhecimento da fauna helmintologica brasileira. IX. Sobre as especies do genero Spinicauda n.g. Mem. Inst. Osw. Cruz. 12 : 44-50. 1920^. Esb69o de uma chave geral dos nematddeos parasites. Rev. Vet., Rio de J. 10 : 59-70- YAMAGUTI, S. 1961. Systema helminthum. III. The nematodes of vertebrates. New York, London. I. pp. 1-680, II. 681-1261. YORKE, W. A. & Maplestone, P. A. 1926. The nematode parasites of vertebrates. London, i-x, 1-53- APPENDIX While this paper was in press Chabaud (Alain G.) and Dollfus (Robert Ph.) have published the description of a new genus and species of heterakid (1966, Hatterianema hollandei N.G., N.SP., nematode heterakide parasite de rhynocephale. Bull. Mus. nat. hist. Nat. Ser. 2, 37 : 1041-1045) which they refer to the subfamily Meteterakinae. However, from the description the species is referable to the Spinicaudinae rather than Meteterakinae since i) the lips appear to be offset from the body, 2) there are no caudal alae on the male tail, 3) a gubernaculam is present, 4) there is no 28 W. G. INGLIS flap over the vulva in the female. In the key given above H. hollandei comes out at Spinicauda from which Hatterianema appears to differ largely in not possessing a cuticular flange projecting beyond the main mass of the lips and the large number of caudal papillae. In fact Hatterianema corresponds almost perfectly to the hypothetical ancestral form deduced above. PRINTED IN GREAT BRITAIN BY ADLARD & SON LIMITED BARTHOLOMEW PRESS, DORKING JAN 1967 A REVISION OF THE LAKE VICTORIA HAPLOCHROMIS SPECIES (PISCES, CICHLIDAE) PART VI P. H. GREENWOOD BULLETIN OF THE BRITISH MUSEUM (NATURAL HISTORY) ZOOLOGY Vol. 15 No. 2 LONDON: 1967 A REVISION OF THE LAKE VICTORIA HAPLOCHROMIS SPECIES (PISCES, CICHLIDAE) N <^ L H^ BY P. H. GREENWOOD British Museum (Natural History) ' k Pp. 29-119; 24 Text-figures BULLETIN OF THE BRITISH MUSEUM (NATURAL HISTORY) ZOOLOGY Vol. 15 No. 2 LONDON: 1967 THE BULLETIN OF THE BRITISH MUSEUM (NATURAL HISTORY), instituted in 1949, is issued in five series corresponding to the Departments of the Museum, and an Historical series. Parts will appear at irregular intervals as they become ready. Volumes will contain about three or four hundred pages, and will not necessarily be completed within one calendar year. In 1965 a separate supplementary series of longer papers was instituted, numbered serially for each Department. This paper is Vol. 15, No. 2 of the Zoological series. The abbreviated titles of periodicals cited follow those of the World List of Scientific Periodicals. World List abbreviation : Bull. Br. Mus. nat. Hist. (Zool.) Trustees of the British Museum (Natural History) 1967 TRUSTEES OF THE BRITISH MUSEUM (NATURAL HISTORY) Issued 13 January, 1967 Price i 15$. A REVISION OF THE LAKE VICTORIA HAPLOCHROMIS SPECIES (PISCES, CICHLIDAE) PART VI By P. H. GREENWOOD CONTENTS Page INTRODUCTION .......... 31 Haplochromis spekii (Blgr) ........ 32 Haplochromis pachycephalus sp. nov. ...... 39 Haplochromis maculipinna (Pellegrin) ...... 43 Haplochromis boops sp. nov. . . . . . . . 47 Haplochromis thuragnathus sp. nov. ...... 49 Haplochromis xenostoma Regan ....... 51 Haplochromis pseudopellegrini sp. nov. ...... 56 Haplochromis altigenis Regan ....... 60 Haplochromis dichrourus Regan ....... 65 Haplochromis paraguiarti sp. nov. . . . . . . 69 Haplochromis acidens sp. nov. . . . . . . . 73 Haplochromis prognathus (Pellegrin) ...... 78 Haplochromis argenteus Regan ....... 84 Haplochromis squamulatus Regan. . . . . . . 87 Haplochromis barbarae sp. nov. ....... 93 Haplochromis tridens Regan and Trcwavas ..... 97 Haplochromis orthostoma Regan . . . . . . . 100 Haplochromis parorthostoma sp. nov. . . . . . . 103 Haplochromis apogonoides sp. nov. ...... 105 DISCUSSION ........... 108 SUMMARY . . . . . . . . . . . 117 ACKNOWLEDGEMENTS . . . . . . . . . 117 APPENDIX: A statotilapia nigrescens Pellegrin, 1909 . . . . 118 REFERENCES ........... 119 INTRODUCTION THIS is the second of two papers dealing with the piscivorous species of Haplochromis in Lake Victoria. In the first part (Greenwood, 1962) representatives of the principal groups of piscivores were considered, and the main morphological trends within the trophic grade were discussed. The present paper covers the remaining species which have been studied to date ; undoubtedly more piscivorous Haplochromis species will be discovered, particularly amongst the as yet poorly sampled species of the deeper waters. Not every species considered here is a piscivore ; those of other trophic groups are included simply because the species have the morphology of a piscivore, and pre- sumably evolved from the same stem as their fish-eating relatives. ZOOL. 15, 2. 2 32 P. H. GREENWOOD Also included in this paper is a species apparently endemic to the Lake Kyoga system. This step was necessary because of its close relationship with a previously undescribed species from Lake Victoria. Some of the individual species described below, and in the previous paper, would seem to be so far removed from the generality of Victoria Haplochromis species as to justify their elevation to generic rank. Indeed, it could be argued that even some of the species-complexes have attained this level of differentiation. However, I do not think that the question can be dealt with until the whole Lake Victoria Haplo- chromis species-flock has been described. Even then, I doubt whether it will be possible to make any such divisions, at least generically. Perhaps a number of subgeneric groups could be justified on phyletic grounds, but these will be difficult to define. The situation closely resembles that encountered by Trewavas (1964) in the genus Serranochromis. However, I do not believe that her solution to the Serranochromis problem, the recognition of a gradal genus, is applicable to the situation amongst the piscivorous Haplochromis of Lake Victoria, particularly because the boundary between these species and any ancestral grade (or grades) would be even more obscure and indefinite than that separating Serranochromis from the Haplochromis of central Africa. Further complications are introduced when one considers the generic status of " Haplochromis " species outside Lake Victoria (and this includes the Lake Nyasa species in all their complexity) relative to the possibly polygeneric Haplochromis species of Lake Victoria. Haplochromis spekii (Boulenger), 1906 (Text-fig, i) Pelmatochromis spekii (part) Boulenger, 1906, Ann. Mag. nat. Hist., (7), 17, 440 ; Idem, 1915, Cat. Afr. Fish, 3, 416, fig. 285. (Lectotype.B.M. (N.H.), reg. no. 1906.^5.30.296, and pro- bably one paralectotype, 1906.5.30.307). Haplochromis spekii (part) : Regan, 1922, Proc. zool. Soc. Londn., 179 (same specimens as above) . Haplochromis serranoides Regan, 1922, op. cit. (Lectotype, B.M. (N.H.) 1911.3.27.17, and probably the two paralectotypes 1904.5.19.52-3). ? Paratilapia serranus (part) : Boulenger, 1915, Cat. Afr. Fish, 3, 334 (two specimens, 1904.5.19.52-3, see above). ? Haplochromis serranus (part): Regan, 1922, op. cit. 174 (paralectotype of P. spekii 1904.5.30.307, see above). LECTOTYPE : a male, 191-0 mm. S.L., from Bunjako, collected by Degen ; B.M. (N.H.) reg. no. 1906.5.30.296. NOTE ON THE SYNONYMY \ Certain small specimens (those indicated above with an interrogation mark) are included in the synonymy with some uncertainty. Using the diagnostic characters currently available, small preserved specimens of H. spekii cannot readily be separated from similar sized specimens of H. serranus. Regan (1922) distinguished H. spekii from H. serranoides on two characters : the maxillary extending to below the anterior quarter of the eye (barely reaches anterior orbital margin in H. serranoides}, and, the caudal peduncle longer than deep (as long as deep in H. serranoides}. Additional material shows that the A REVISION OF THE LAKE VICTORIA HAPLOCHROMIS SPECIES 33 difference in caudal peduncle proportions is easily masked by intraspecific variability ; furthermore, I am unable to confirm the marked differences in caudal peduncle proportions which Regan found in the lectotypes of the two species. The difference in the posterior extension of the maxilla is valid for the lectotypes, but it must be noted that the jaws in H. spekii type are somewhat distorted because of a deformed right preorbital bone. Again, more material has shown that the maxilla has a variable posterior extension which links the extremes shown by the lectotypes of the two species. In all other characters, including the dentition and the preserved colour patterns, the two type specimens show no trenchant differences, and I consider them to be conspecific. Because of their small size (74 and 114 mm. S.L.) the paralectotypes of H. serran- oides have not been included in the redescription. I think it probable that these specimens are referable to the species. A similar problem is posed by three para- lectotypes of H. spekii (B.M. [N.H.] reg. nos. 1906.5.30.301, and 1906.5.30.297-8, of standard lengths 101-0, 91-0 and 79-0 mms. respectively). The two latter may perhaps be specimens of H. serranus, and the former is probably referable to H. spekii. However, until more is known about the characteristics of smaller specimens of H. spekii, I consider it inadvisable to give a definite identity to these three fishes. DESCRIPTION : based on 44 specimens (including the lectotype of the species, and the lectotype of H. serranoides) , 128-220 mm. standard length. Depth of body 32-8-39-8 (mean, M = 35-6) per cent of standard length, length of head 36-1-39-3 (M = 37'4) per cent. Dorsal head profile straight, sloping at an angle of 30-35, the premaxillary pedicels from barely to moderately prominent and interrupting the profile. Preorbital depth 18-0-24-2 (M = 20-7) per cent of head, least interorbital width 22-0-26-0 ( M = 23-3) per cent. Snout 1-2-1-3 times as long as broad, its length in fishes < 190 mm. S.L., (N = 25), 34-0-40-6 (M = 36-8) per cent of head, and in larger fishes (N = 19) 36-0-42-5 (M = 39-1) per cent. Eye diameter in fishes < 200 mm. S.L. (N = 34) 17-3-22-6 (M = 20-0), and in larger individuals (N = 10) 15-7-19-4 (M = 18-0) per cent of head ; ratio of eye/preorbital 0-8-1-3 (M == i-o). Depth of cheek 25-7-32-9 (M = 29-5) per cent of head. Caudal peduncle 16-7-19-8 (M = 17-9) per cent of standard length, 1-1-1-5 (modal range 1-2-1-3) times as long as deep. Mouth horizontal or slightly oblique, jaws equal anteriorly or the lower projecting slightly, its length 49-2-61-3 (M = 53-8) per cent of head, 1-7-2-6 (modal range 1-9-2-1) times as long as broad. Mental symphysis smooth or with a slight pro- tuberance. Premaxilla sometimes a little expanded medially but never beaked. Posterior tip of the maxilla reaching a point near the vertical through the anterior orbital margin or occasionally reaching this level (see also note on synonymy, p. 32). Gill rakers: stout or moderately stout, the lower I to 3 reduced ; 8 or 9 (rarely) on the lower part of the first gill arch. Scales ; ctenoid ; lateral line with 30 (f.i), 31 (f.8), 32 (23), 33 (f.n) or 34 (f.i) ; 34 P. H. GREENWOOD r 1 3 -t-> rt 3 35 cheek with 3 (rare)-5 (mode 4) rows. Six to 8 (mode 7) between the upper lateral line and the dorsal fin origin, 5-9 (mode 8) between the pectoral and pelvic fin bases. Fins : Dorsal with 24 (f.i2), 25 (f-3o) or 26 (f.2) rays, comprising 14 (f.i), 15 (f.2Q) or 16 (f.i4) spinous and 8 (f.i), 9 (f.2i) or 10 (f.22) branched rays. Anal with ii (f.2), 12 (f-3o) or 13 (f.g) rays, comprising 3 spinous and 8 (f.2), 9 (f.3o) or 10 (f.g) branched elements. Pectoral 27-0-33-3 (M = 29-3) per cent of standard length. Pel vies with the first and second branched rays produced, slightly so in females but the first ray protracted and thread-like in males. Caudal subtruncate, scaled on its basal half. Teeth : In all specimens examined, both the inner and outer teeth are unicuspid, those of the outer row stout and strongly curved. The smallest fish (128 mm. S.L.) shows faint indications of lateral cusps on some teeth in the inner rows. There are 44-70 (M = 55) teeth in the outer row of the upper jaw ; inner teeth in this jaw are arranged in 3-5 (usually 3 or 4) rows, and in the lower jaw in 2 or 3 (rarely i or 5) rows. Osteology. The neurocranium of H. spekii is identical with that of H. serranus, that is, of the generalized predator type showing affinity with the skull of H. guiarti (see Greenwood, 1962). The lower pharyngeal bone is triangular, its dentigerous surface broader than long. The lower pharyngeal teeth are relatively fine, cylindrical in cross-section and weakly bicuspid ; some teeth are almost uniscuspid, with the larger cusp elongate and conical. The teeth are arranged in 22-24 rows. Vertebral counts (precaudal and caudal) for six specimens are 13 + 16 (f-3), 13 + 17 (f.2) and 12 + 17 (f.i). Coloration. Live coloration is unknown. Preserved specimens : Males (adult and sexually active) : Ground colour overall dusky, including the entire head, both jaws and the branchiostegal membrane ; very faint indications of a broad midlateral stripe visible behind the operculum to the beginning of the caudal peduncle where it merges with the dark general body colour. Dorsal fin dark except for the distal third to half of the soft part which is hyaline with dark spots and dashes. Caudal dark on its basal two-thirds, yellowish distally. Anal light dusky except for the distal quarter to third of the soft part which is hyaline ; 4 or 5 moderately large ocelli (dead white), usually arranged in two rows or one irregular row. Pelvic fins dusky. Adult (but sexually quiescent males) have a variable ground coloration which, how- ever, is always lighter than that of sexually active fishes. The snout and jaws are darker than the flanks which vary from dusky to light golden-brown ; branchiostegal membrane dark, but sometimes only in the region below the operculum. Dorsal fin dark, the lappets black, and the soft part often with close-set dark spots or dashes. Anal variable, from dusky to yellowish ; ocelli whiteish-grey, 2-5 in number and arranged as in active fishes. Pelvics usually dusky but of a variable intensity ; when light, the pigment concentrated over the spine and the first two branched rays. Females (adult and juvenile) : brownish above (and on the head and snout), shading to silvery-brown or greyish-silver on the lower flanks, belly, chest and operculum ; branchiostegal membrane greyish. A faint midlateral band (of 36 P. H. GREENWOOD variable depth and of irregular outline) runs from behind the operculum to the caudal fin origin ; there is also a very faint upper longitudinal band running slightly above the upper lateral line visible in some specimens. All fins are brownish-yellow the soft dorsal darkly maculate. Caudal dark brown on its proximal two-thirds (because of the dense maculation in that region). Immature males are coloured like females except that the longitudinal stripes are more distinct, and some specimens have very faint traces of 4 or 5 vertical bars crossing the longitudinal stripes on the flanks ; these bars extend from the back to a level about half way towards the ventral outline. The pelvic fins are faintly sooty. Ecology : Habitat. Haplochromis spekii occurs over both hard and soft substrates, but seems to show a slight preference for the former. Few specimens were collected from nets operated over exposed beaches, most coming from gill-nets set in sheltered areas where the water was 10-30 ft. deep. Some specimens were taken from more exposed areas, but not from deeper water. Food. Of the 42 fishes examined (from 24 localities), 22 contained food. Twenty- one of these had fed exclusively on small fishes (identified in 8 guts as Haplochromis species, in a further 8 as Cichlidae, and in one as a cyprinid). The exceptional fish contained unidentifiable fish remains and fragments of an ephemeropteran larva (probably Povilla adusta) . Breeding. All specimens < 150 mm. S.L. are immature, as is one specimen of 182 mm., but others > 150 mm. are mature. Both sexes reach the same maximum adult size. Affinities. The close relationship between H. spekii and H. serranus has been noted already (see above p. 33). There is complete overlap in most characters but the differential growth trends shown by two characters are such that this overlap is considerably reduced in fishes more than 120 mm. S.L. The two characters are depth of preorbital, and eye diameter as proportions of head length. In H. spekii both are, generally, larger than in H. serranus when specimens of the same size are compared. However, even in these characters there is still some overlap, and, from the sample studied, it seems likely that neither is a reliable diagnostic character when fishes < 120 mm. S.L. are compared. The difference between H. spekii and H. serranus (in the size range 120-205 mm.) is perhaps best shown by the ratio of eye diameter to preorbital depth, viz., 0-8-1-3 (mean i-o) for H. spekii, and 1-1-1-5 (mean 1-3) for H. serranus. Two other characters seem to show interspecific differences in their modal values, (i) In H. serranus the posterior tip of the maxilla usually lies below the eye or reaches to the vertical through the anterior orbital margin ; in H. spekii it rarely reaches as far posteriorly as the orbital margin (ii) Haplochromis serranus has a very prominent mental protuberance, but this bump is much weaker, if it is developed at all, in H. spekii. In many specimens of H. serranus the mental bump is so prominent that, in lateral view, the anterior margin of the dentary has a marked backward slope thus emphasizing the acuteness of the head profile ; in H. spekii the anterior margin of the dentary is, generally, almost perpendicular and so the tip of the head seems blunter than in H. serranus. A REVISION OF THE LAKE VICTORIA HAPLOCHROMIS SPECIES 37 Unfortunately it is impossible to compare the live colours of adult males from the two species ; preserved coloration is similar. This information, together with more field data on niche preferences, and small specimens of H. spekii, will be necessary before the precise relationships (or perhaps conspecifity) of the two species can be determined. If H. spekii and H. serranus were allopatric it would be tempting, on the information available, to consider them conspecific. However, experience with other sympatric species in Lake Victoria suggests that such slight morphological differences as are known between H. spekii and H. serranus can be the only ones manifest by biologically distinct species. Haplochromis spekii is more easily distinguished from other members of the H. serranus species complex. From H. victorianus it is recognizable by its larger adult size (some H . spekii are juvenile at a size near the upper adult limits for H. victorianus}, larger head (36-i-39'3> M =37'4% S.L.,c/. 33'5-36-o, M =34-8%), deeper cheek (257-32-9, M = 29-5% head, cf. 22-5-26-2, M 24-6%), longer lower jaw, (49-2-61-3, M = 53-8% head, cf. 44-0-51-8, M = 47-1%), smaller eye in fishes < 200 mm. S.L. (17-3-22-6, M = 20-0% head, cf. 21-7-26-2, M 24-6%), shorter pectoral fin (21-4-28-9, M = 25-1% S.L., cf. 26-2-32-7, M 30-4%), and by having fewer and more curved outer teeth in the upper jaw (44-70, M = 55 teeth, cf. 64-86, M == 74). From H. maculipinna, H. spekii differs in its larger adult size, longer head (36-1-39-3, M =37-4% S.L., cf. 32-6-37-0, M =35'5%)> longer snout (34-0-40-6, M = 36-8% head, cf. 30-3-37-. M = 337%). deeper cheek (25-7-32-9, M = 29-5% head, cf. 23-2-29-8, M 25-3%), longer lower jaw (49-2-61-3, M =53-8% head, cf. 43-3-52-8, M 48-3%), and lower eye/preorbital ratio (0-8-1-3, M == I'O, cf. 1-3-1-6, M == 1-5). Although H. spekii resembles H. bartoni a little more closely in morphometric characters than it does H. victorianus, the species show a greater difference in neuro- cranial form. The neurocrania of H. victorianus and H. spekii are virtually identical, but that of H. bartoni is nearest the typical " prognathus "-group type (see p. 109). Morphometrically, H. spekii differs from H. bartoni in having a broader interorbital region (22-0-26-0, M = 23-3% of head, cf. 17-0-21-0, M == 18-6%), and a somewhat smaller eye (17-3-22-6, M = 20-0% head, cf. 20-3-24-1, M = 22-5%). Also H. spekii has a lower modal number of spinous dorsal fin rays (15 cf. 16). From the third member of the H. serranus species group, H. nyanzae, H. spekii differs in its larger adult size, larger head (36-1-39-3, M = 37-4% S.L., cf. 33'6-36'7> M = 35-4%), deeper cheek (25-7-32-9, M = 29-5% head, cf. 24-4-27-6, M = 25-9%), longer lower jaw (49-2-61-3, M =53-8% head, cf. 45-0-51-6, M =48-0%), and a lower modal number of spinous dorsal fin rays (15 cf. 16). Although typical specimens of H. spekii and H. gowersi are not readily confused (compare text-fig. I with text fig. 13 in Greenwood, 1962) there is one specimen whose appearance is such that I am unable to place it in one species or the other ; it is even intermediate in the two quantifiable morphological characters (body depth and inter- orbital width) showing the greatest interspecific differences. Haplochromis gowersi and H. spekii differ markedly in neurocranial form, but without dissection this character cannot be checked with sufficient precision in the unique intermediate P. H. GREENWOOD specimen. For the present, the possibility cannot be overruled that this fish is an interspecific hybrid. Phyletically, Haplochromis spekii appears to be a derivative from an H. serranus- like ancestor, the principal difference between the species being the larger adult size attained by H. spekii. Note : Gilchrist and Thompson (1917) record six specimens of Pelmatochromis spekii Blgr. from the Magalies river, Transvaal. I have not examined these speci- mens, but clearly they cannot be referred to Haplochromis spekii (Blgr.). Judging from their locality, it seems probable that they are specimens of Chetia flaviventris Trewavas. Dr. Trewavas is of a like opinion (personal communication). STUDY MATERIAL AND DISTRIBUTION RECORDS Museum and Reg. No. B.M. (N.H.) 1906.5.30.296 (Lectotype) B.M. (N.H.) 1911.3.27.17 (Lectotype H. serranoides) B.M. (N.H.) 1966.3.9.1-4, 20-21, 30-35, 39-49 B.M. (N.H.) 1966.3.9.5 B.M. (N.H.) 1966.3.9.11-14 B.M. (N.H.) 1966.3.9.28-29 B.M. (N.H.) 1966.3.9.8-10 B.M. (N.H.) 1966.3.9.22 B.M. (N.H.) 1966.3.9.27 Locality UGANDA Bunjako .... Victoria Nile Napoleon Gulf, near Jinja . Beach near Nasu Point (Buvuma Channel) . Off S. tip of Ramafuta Island (Buvuma Channel) . Karenia, near Jinja (Napoleon Gulf) Pilkington Bay . Thruston Bay Buka Bay (Buvuma Channel) Collector Degen Mel land E.A.F.R.O. E.A.F.R.O. E.A.F.R.O. E.A.F.R.O. E.A.F.R.O. E.A.F.R.O. E.A.F.R.O. B.M. (N.H.) 1966.3.9.17-19 B.M. (N.H.) 1966.3.9.7 B.M. (N.H.) 1966.3.9.15-16 B.M. (N.H.) 1928.5.24.413-5 KENYA Kisumu (Kavirondo Gulf) . Naia Bay (Kavirondo Gulf) Sagorony (Kavirondo Gulf) . Ulambwi Bay (Kavirondo Gulf) E.A.F.R.O. E.A.F.R.O. E.A.F.R.O. Graham B.M. (N.H.) 1966.3.9.345 B.M. (N.H.) 1966.3.9.36-38 B.M. (N.H.) 1966.3.9.6 TANZANIA Beach near Majita . . . E.A.F.R.O. Between Ghogororo and Isanga River E.A.F.R.O. Mwanza (Capri Bay) . . . E.A.F.R.O. B.M. (N.H.) 1966.3.9.23-26 LAKE VICTORIA Locality unknown E.A.F.R.O. A REVISION OF THE LAKE VICTORIA HAPLOCHROMIS SPECIES 39 Haplochromis pachycephalus sp. nov. (Text-fig. 2) HOLOTYPE : an adult male, 199 mm. standard length, from 40 ft. of water off Kazima island, Uganda. B.M. (N.H.) reg. no. 1966.2.21.9. DESCRIPTION : based on the holotype and fourteen other specimens 150-232 mm. standard length. Depth of body 36-5-42-5 (M = 39-1) per cent of standard length, length of head 35-6-39-5 (M = 36-3) per cent. Dorsal head profile straight to moderately concave, the concavity exaggerated by the prominent premaxillary pedicels ; nuchal region prominent and gently convex, prenuchal region sloping at 30-35. Cephalic lateral line pores large, especially on the preorbital and preopercular bones, less so on the dentary. Preorbital depth 18-9-22-5 (M = 20-8) per cent of head, least interorbital width 24-6-31-3 (M = 27-8) per cent. Snout 1-1-1-4 (mode 1-2) times broader than long, its length 32-4-38-2 (M = 35*9) per cent of head ; eye diameter 18-8-22-2 (M = 20-6) per cent, depth of cheek 26-4-36-1 (M == 30-8) per cent. Caudal peduncle 13-2-16-0 (M == 14-8) per cent of standard length, 1-0-1-3 (mode 1-2) times as long as deep. Mouth oblique, sloping at an angle of 35-45 (mode 40). Jaws equal anteriorly or lower projecting slightly, its length 51-5-58-4 (M = 55-0) per cent of head, 1-5-1-9 (one specimen 2-2) times as long as broad. Posterior tip of maxilla reaching the vertical through the anterior orbital margin or nearly so. Gill rakers : stout, the lower I or 2 sometimes reduced, the upper 3 or 4 sometimes expanded ; 8 or 9 (7 in one specimen) on the lower part of the first gill arch. Scales ctenoid ; lateral line with 32 (f.3), 33 (f.4), 34 (f.4) or 35 (f.3), cheek with 5 or 6 (rarely 4) rows. Nine or 10 (less frequently 7, 8 or io|) between the upper lateral line and the dorsal fin origin, 7 or 8 (less frequently 6 or 9) between the pectoral and pelvic fin bases. Fins. Dorsal with 23 (f.i), 24 (f.6), 25 (f.6) or 26 (f.i) rays comprising 15 (i.g) or 16 (f.5) spinous and 8 (f.i), 9 (f.io) or 10 (f.3) branched rays. Anal with n (f.2) or 12 (f.i2) rays, comprising 3 spines and 8 (f.2) or 9 (f.12) branched elements. Pectoral 21-6-30-9 (M = 24-6) per cent of standard length. Pelvics with the first branched ray produced in sexually active males, slightly so in females and quiescent males. Teeth. The outer row in both jaws is composed of unicuspid, slender and slightly curved teeth. There are 60-80 (M = 70) teeth in the outer row of the upper jaw. Teeth in the inner rows are small, unicuspid, curved (strongly so in the upper jaw) and implanted obliquely. There are 4 or 5 (less frequently 2 or 3) rows in the upper jaw and 2 or 3 (rarely 4) in the lower. Osteology. No complete skeleton is available. The lower pharyngeal bone is triangular, with its dentigerous surface slightly broader than long (most markedly so in the smallest fish), or rarely, as long as broad. Lower pharyngeal teeth fairly coarse, their crowns weakly cuspidate and barely compressed ; some teeth in the two median rows are almost conical. The teeth are arranged in 18-22 rows. 4 o P. H. GREENWOOD CS 1 1 N i A REVISION OF THE LAKE VICTORIA HAPLOCHROMIS SPECIES 41 Vertebral counts in 5 specimens are : 13+16 (.3), 14 + 16 (f.i) and 14 + 17 (f.i). Coloration : Live colours are unknown. Preserved material : Adult males. Ground colour variable and probably dependent on sexual state (also affected by preservation). Body greyish with black belly, chest and branchiostegal membrane, or the black replaced by a sooty-grey (in such specimens the branchiostegal membrane may be dark grey and flecked with sooty blotches) ; very faint traces of a broad midlateral stripe on the flank, originating behind the operculum and extending to the caudal fin origin. Dorsal fin greyish, dark lappets and maculae on the soft part. Caudal greyish, darkly maculate between the rays. Anal yellowish with a faintly sooty base, especially on the anterior part and around the ocelli ; the latter are dead-white, 5-9 (usually 5 or 6) in number and arranged in from i to 3 irregular rows. Pelvics blotched sooty to entirely black (the latter condition associated with the darkest body coloration). Two fishes (both from the same net haul) are more sexually active than the others. Both are a very dark brown, almost uniformly so except for a black belly, chest and branchiostegal membrane. The spinous dorsal is a very dark brown (nearly black), the soft part is lighter and has a yellowish margin. Caudal light brown with lighter maculae on the basal three-quarters. Anal very faint pink, with a narrow black basal streak which expands in the region of the proximal row of ocelli and extends in amongst them ; the ocelli are whiteish. Females (adult but quiescent ; based on two specimens only). Ground colour silver-grey becoming creamy on the chest and belly. Entire head (including the lower jaw) brownish with darker and irregular mottling. Body also mottled with sooty blotches, the effect being generally irregular except that on the flanks the blotches have some faint organisation into near vertical, broad bands extending from the dorsal outline to almost the ventral outline. There is some resemblance between this coloration and that of H. cavifrons (see Greenwood, 1962), although in H . pachycephalus the effect is less definitely that of freckling. Dorsal fin yellowish- grey with sooty freckling and blotching. Caudal densely and darkly blotched on its proximal third to half, greyish and darkly maculate distally. Anal greyish-yellow, with a narrow, sooty band along its base, and a duskiness over the spinous part ; both fishes have two large and distinct, dead white ocelli (an unusual feature in females). Pelvics hyaline with irregular sooty blotches. Ecology. Habitat. The species is known from four localities ; all are some distance off-shore but close to islands. The specimens all came from nets set on a soft bottom at depths of 100-120 ft., except in one locality where the collection was made after the use of explosives. In this instance the charge was set off in about 40 ft. of water over a rock shelf with deeper water on its off-shore side. Food. Of the 13 specimens examined (from 5 localities) six contained food in the stomach or intestines. All yielded macerated fish remains. Fragments of Haplo- chromis species were identified from three guts, a cyprinid fish in a fourth, and cichlid remains in two others. Breeding. Little information is available about the breeding habits of H. pachy- cephalus. All specimens except the smallest (a male, 150 mm. S.L.) are mature. The two largest fishes (232 mm. and 228 mm. S.L.) are males. 42 P. H. GREENWOOD Affinities. Haplochromis pacliycephalus is, at least on superficial characters and those detectable on a radiograph, related to the H. serranus species group (see p. 109). From H. serranus, H. pachycephalus is distinguished by its broader snout, broader interorbital (24-6-31-3, M = 27-8% head, cf. 20-4-26-8, M = 23-3%) and lower jaw (length/breadth ratio 1-5-1-9 cf. 1-8-2-5), and its smaller nuchal scales. From H. victorianus it differs in its broader interorbital (24-6-31-3, M = 27-8% head, cf. 21-5-24-5, M=22-6%), deeper cheek (26-4-36-1, M 30-8% head cf. 22-5-26-2, M = 24-6%) and longer lower jaw (51-5-58-4, M = 55-0% head, cf. 44-0-51-8, M =3 47-1%) ; the lower jaw is also broader in H. pachycephalus. From H. spekii and H. maculipinna, the oblique mouth and broad snout of H. pachycephalus serve as immediately diagnostic characters, although the snout in H. maculipinna is broader than in other members of the " serranus "-group (being as much as i -i times broader than long, but generally as long as broad). As with other members of the group, H. maculipinna and H. spekii have a narrower interorbital region than H. pachycephalus ; H. maculipinna also has a larger eye (24-0-31-7, M = 26-3% head cf. 18-8-22-2, M s= 20-6% in H. pachycephalus) but the larger adult size reached by H. pachycephalus may influence this character. The same superficial characters (including the oblique mouth) serve to distinguish H. pachycephalus from H. bartoni and H. nyanzae, the former a member of the " prognathus " group, the latter a " serranus " group member. Haplochromis boops and H. thuragnathus (both " serranus "-group species) closely resemble one another (see pp. 50) and H. pachycephalus. Both differ from H. pachy- cephalus in the following characters : a narrower interorbital, shorter snout, larger eye, and larger nuchal scales. All three species have the snout broader than it is long. From the evidence available, H. pachycephalus would seem to be derived from an H. serranus-like ancestor, the principal morphological changes being an increase in mouth size coupled with greater obliquity of the mouth angle. The larger cephalic lateral line pores of H. pachycephalus are probably correlated with the deep water habitat of the species (as compared with H. serranus and its immediate allies). STUDY MATERIAL AND DISTRIBUTION RECORDS Museum and Reg. No. Locality Collector UGANDA B.M. (N.H.) 1966.3.9.175-177 Off S. tip of Ramafuta Island (Buvuma Channel) . . . E.A.F.R.O. B.M. (N.H.) 1966.3.9. 174 Deep water off Dagusi Island . E.A.F.R.O. B.M. (N.H.) 1966.2.21 .9 Off Kazima Island . . . Uganda (Holotype) Fisheries Dept. B.M. (N.H.) 1966.3.9.166-169, Off Kazima Island . . . Uganda 171-173 Fisheries Dept. TANZANIA B.M. (N.H.) 1966.3.9.170 Off Godziba Island . . ' , E.A.F.R.O. LAKE VICTORIA B.M. (N.H.) 1966.3.9.178 Locality unknown ,. . . . E.A.F.R.O. A REVISION OF THE LAKE VICTORIA HAPLOCHROMIS SPECIES 43 Haplochromis maculipinna (Pellegrin), 1913 (Text-fig. 3) Paratilapia maculipinna Pellegrin, 1913 Bull. Soc. Zool. France, 37, 311 ; Idem, 1914, in Voyage de Ch. Alluaud et R. Jeannel en Afrique Occidental, 16, PI. i, fig. i, Paris. Paratilapia prognatha (part) : Boulenger, 1915, Cat. Afr. Fish., 3, 333. Haplochromis maculipinna : Regan, 1922, Proc. zool. Soc. Londn., 177, fig. 5. HOLOTYPE : a fish 122-0 mm. S.L. (Paris Museum No. 12-258) from Port Florence (Kavirondo Gulf) collected by Alluaud and Jeannel. This specimen differs from all others now included in the species by its much larger eye. In other characters, however, it agrees with these specimens and differs from the few other Victoria Haplochromis species characterized by large eyes. DESCRIPTION : based on 33 specimens (including the holotype), 91-5-166 mm. S.L. Depth of body 33-3-37-0 (M = 35-9) per cent of standard length, length of head 32-6-37-0 (M = 35-5) per cent. Dorsal head profile straight or slightly concave in those fishes with prominent premaxillary pedicels, sloping at 3O-35. Preorbital depth 16-4-20-4 (M = 18-2) per cent of head, least interorbital width 20-7-25-5 (M = 22-8) per cent. Snout as long as broad to i-i times broader than long, its length 30-3-37-0 (M = 33-7) per cent of head, eye diameter 24-0-29-2 (31-7 in the type), mean 26-3 per cent, ratio of eye/preorbital 1-3-1-6 (M = 1-5) but 1-9 in the type ; depth of cheek 23-2-29-8 (M = 25-3) per cent. Caudal peduncle 14-5-18-8 (M = 16-3) per cent of standard length, 1-2-1-8 (modal range 1-2-1-5) times as long as deep. Mouth moderately oblique, sloping upwards at 35-40, lower jaw projecting slightly to strongly, its length 43-3-52-8 (M = 48-0) per cent of head, 1-6-2-3 (modal 1 C M FIG. 3. Haplochromis maculipinna. Drawn by Barbara Williams. 44 P. H. GREENWOOD range 1-7-2-1) times as long as broad. Lips not noticeably thickened, premaxilla not expanded medially. Posteriorly tip of the maxilla not quite reaching the vertical through the anterior orbital margin. Gill rakers : moderately stout, the lower 1-3 reduced, the upper 2-5 flat and some- times divided ; 8-n (mode 10) on the lower part of the first arch. Scales : ctenoid ; lateral line with 31 (f.i), 32 (f.i5), 33 (f.i5) or 34 (f.2), cheek with 3-5 (mode 3) rows. Five and a half (rare) to 8 (modal range 6-7) scales between the upper lateral line and the dorsal fin origin, 5^ (rare) -9 (modal range 7-8) between the pectoral and pelvic fin bases. Fins : Dorsal with 24 (f.5), 25 (f.26) or 26 (f.2) rays, comprising 14 (f.i), 15 (f.24) or 16 (f.8) spinous and 9 (f.n) or 10 (f.22) branched rays. Anal with n (f.3), 12 (f.27) or 13 (f-3) rays, comprising 3 spines and n (f.3), 12 (^27) or 13 (f.3) branched elements. Pectoral 23-0-31-0 (M = 26-4) per cent of standard length. Pelvics with the first two branched rays produced in both sexes, but markedly elongate in adult males. Caudal truncate to subtruncate. Teeth. Except in the smallest specimen (91-5 mm. S.L.), the outer teeth are uni- cuspid ; in the exceptional fish most teeth are weakly bicuspid but a few are unicuspids. All outer teeth are slightly curved and slender, the curvature being most marked in teeth situated laterally and posterolaterally ; teeth situated posterolaterally in the upper jaw are the smallest and finest. There are 50-80 (M = 62) teeth in the outer row of the upper jaw. Teeth in the inner rows are more variable in form, and there is no clear-cut correla- tion between tooth form and the fish's size. The smallest specimen has only tricuspid inner teeth ; other and larger fishes may have only weakly tricuspids in both jaws, or tricuspid teeth predominating in both jaws but some unicuspids occurring in the lower jaw, or an admixture of tri- and unicuspids (the latter predominating) in the upper jaw and only unicuspids in the lower, or a mixture of tri- and unicuspids in both jaws, or unicuspids in the upper jaw and an admixture in the lower, or only unicuspids in both jaws. Some of the largest fishes fall in either the mixed uni- and tricuspid category or in the purely unicuspid one. The inner teeth are arranged in I (rare) 4 (modes 2 and 3) rows in the upper jaw, and 2 (less commonly I or 3) rows in the lower. Osteology. The neurocranium of H. maculipinna is virtually identical with that of H. serranus, differing only in having a relatively lower supraoccipital crest. The premaxilla, dentary and suspensorium are also like those of H. serranus, except that the dentary is somewhat deeper and shorter in H. maculipinna. The lower pharyngeal bone is triangular and rather fine ; the dentigerous surface is slightly broader than long. Lower pharyngeal teeth are slender and distinctly cuspidate, those of the two median rows are the coarsest ; the teeth are arranged in 20-24 rows. Vertebral counts for seven specimens are : 13+16 (f.i) ; 13+17 (f.3) ; 14 +16 (f.2) and 14 + 17 (f.i), giving totals of 29 to 31. Coloration : Live colours are unknown. Preserved specimens : Males (adult and sexually active) : ground colour very dark brown, almost black, with a golden under- lay on the flanks and operculum, and a sooty overlay on the chest. Head, including A REVISION OF THE LAKE VICTORIA HAPLOCHROMIS SPECIES 45 the lower jaw, dark but lips light brown. A very faint, but broad, lachrymal stripe runs from the antro ventral margin of the orbit to the angle of the lower jaw. The branchiostegal membrane is black. Dorsal fin almost uniformly dark sooty, the lappets black. Caudal dark on its proximal three-quarters, dusky distally. Anal dusky, but with a black band along its base, the band expanding anteriorly to cover most of the spinous part of the fin, which is black ; two or three large greyish ocelli are present. Pelvics black to dusky. Males (adult but sexually quiescent] : ground colour dark golden brown, lightest on the anterior flanks. Head dark brown, with a distinct, narrow lachrymal stripe from orbit to angle of lower jaw, and a narrow black vertical bar on the ascending preopercular limb ; branchiostegal membrane greyish-brown. Dorsal fin yellowish- grey, with black lappets, and the membrane between the branched rays dark grey- brown ; the pigment often broken into discrete maculae between the last three or four branched rays. Caudal dark yellowish-grey. Anal uniformly yellowish, with two or three faint, whiteish ocelli. Pelvics dark on the anterior third, otherwise yellowish to hyaline. Males (immature] : ground colour light brown on the flanks and belly, darker above the upper lateral line and on the dorsum of the head ; the branchiostegal membrane is light brown-grey, and a faint lachrymal stripe is visible. On the flanks there is a trace of an interrupted, dark midlateral band on the anterior half of the body, and a continuous band on the posterior half, extending to the caudal origin. Dorsal fin yellowish-brown, the lappets dark, as are the maculae between the branched rays. Caudal yellowish-brown, with dark elongate blotches between the rays. Anal uniformly yellowish-brown, with two or three, distinct and dusky-grey ocelli. Pelvics yellowish with a faint dusky overlay, especially over the anterior part of the fin. Females (adult and juvenile] : ground colour golden brown, darker on the upper half of the body, and the dorsal surface of the head ; faint traces of a rather broad lachrymal stripe are often visible, the stripe generally not extending to below the level of the maxilla, but reaching the angle of the lower jaw in some specimens. Faint traces of an interrupted midlateral band on the anterior half of the body, and a continuous band on the posterior half are often visible ; in some specimens no lateral band is visible, and in others the band is continuous except for a short break at about its midpoint. A few specimens show indications of a much interrupted band (really a series of 6 or 7 broad blotches) running slightly above the upper lateral line on the anterior half of the flanks, and on the lateral line posteriorly. Dorsal fin yellowish, usually darker between the posterior spines, and darkly maculate on the soft part, but uniformly yellowish with very faint maculations posteriorly in others. Anal, caudal and pelvic fins uniformly yellowish. Ecology. Although some individuals occur over sandy, exposed and wave- washed beaches, members of this species are commoner in sheltered gulfs and bays where the water is from 10-30 ft. deep and the substrate is either soft mud or sand and shingle ; a few specimens are from deeper water (35-40 ft.) near off-shore islands. Food. Eleven of the 30 specimens examined (from 16 localities) contained food ZOOL. 15, 2. 3 46 P. H. GREENWOOD in the stomach and intestines. Seven fishes yielded fragmentary fish remains (identified as a cyprinid in one, and as Haplochromis species in two others), three contained fragmentary insect remains (probably larval Ephemeroptera) , and one bottom debris. Breeding. Little information is available ; most specimens less than 140 mm. S.L. are immature, as is one slightly larger individual (145 mm.). Both sexes attain the same maximum adult size. Affinities. In both its gross and detailed morphology H. maculipinna shows affinity with the " serranus " species group, i.e. H. serranus, H. victorianus, H. spekii, and their deep water relatives H. pachycephalus, H. boops and H. thuragnathus. Criteria for distinguishing H. maculipinna from all but the first two species are considered under the descriptions of those species (see pp. 37, 42, 49 and 51 for the species respectively). From H. serranus, H. maculipinna is distinguished by its larger eye (24-0-3I-7, M = 26-3% head, cf. 20-4-26-0, M = 23-3%), shorter and more oblique lower jaw (43'3-52-8, M =48-3% head, cf. 47-7-60-0, M =54-3%), and higher eye/preorbital ratio (1-3-1-6, M = 1-5 cf. 1/1-1-5, M = 1-3). From H. victorianus, it differs in its larger eye (24-0-31-7, M = 26- 3% head cf. 21-7-25-5 M = 23-6%), higher eye/preorbital ratio (1-3-1-6, M = 1-5, cf. 1-1-1-3, M = 1-2), more oblique lower jaw (sloping at 30-35 cf. horizontal or very slightly oblique) and its fewer and finer outer teeth (50-80, M = 62, cf. 64-86, M = 74 teeth in the upper jaw). The close resemblance between H. maculipinna and these two species is obvious, and is greater than the resemblance between H. maculipinna and other members of the " serranus " group. There are two other species, H. nyanzae and H. bartoni which, at least superficially, resemble members of the " serranus " group although H. bartoni seems to belong to a different phyletic line (see p. 109) . Haplochromis maculipinna differs from H. nyanzae in its larger eye (24-0-31-7, M = 26-3% head, cf. 19-1-24-0, M = 22-1%) and higher eye/preorbital ratio (1-3-1-6, M 1-5, cf. 1-1-1-3, M = x * 2 )> an( i by its finer and less curved outer teeth. From H. bartoni, it differs in having a shorter head (32-6-37-0, M = 35-5% standard length, cf. 36-2-39-7, M=375%), broader interorbital (20-7-25-5, M = 22-8% head, cf. 17-0-21-0, M = 18-6%) and shorter, more oblique lower jaw (43-3-52-8, M = 48-3% head, cf. 50-8-57-0, M = 52-5%). Neurocranial form differs in the two species, that of H. maculipinna being of the " serranus " type, and that of H. bartoni being of the " prognathus " type (see p. no). Resemblances between H. maculipinna and H. acidens are discussed on p. 76. It seems unlikely that the species are closely related. Phyletically, H. maculipinna was probably derived from a H. serranus-like ancestor. A REVISION OF THE LAKE VICTORIA HAPLOCHROMIS SPECIES 47 STUDY MATERIAL AND DISTRIBUTION RECORDS Museum and Reg. No Locality B.M. (N.H.) 1906.5.30.263 B.M. (N.H.) 1966.3.9.145-151 B.M. (N.H.) 1966.3.9.132-134 B.M. (N.H.) 1966.3.9.135-144 B.M. (N.H.) 1966.3.9.129-131 B.M. (N.H.) 1966.3.9. 124-6 B.M. (N.H.) 1966.3.9.123 B.M. (N.H.) 1966.3.9.128 B.M. (N.H.) 1966.3.9.127 Paris Museum 12-258 (Holotype) Collector Degen E.A.F.R.O. E.A.F.R.O. E.A.F.R.O. UGANDA Bunjako ..... Napoleon Gulf, near Jinja . Beach near Nasu Point (Buvunia channel) Near Ramafuta Island (Buvuma Channel) Between Yempita and Busiri Island (Buvuma Channel) Buka Bay Fielding Bay Kazima Island (near Entebbe) Pilkington Bay . KENYA Port Florence (Kavirondo Gulf) . Alluaud & Jeannel E.A.F.R.O. E.A.F.R.O. E.A.F.R.O. Uganda Fisheries Dept. E.A.F.R.O. B.M. (N.H.) 1966.3.9.122 LAKE VICTORIA Locality unknown . . . E.A.F.R.O. Haplochromis boops sp. nov. (Text-fig. 4) HOLOTYPE : an adult male, 190 mm. standard length, from 120 ft. of water, off the southern tip of Buvuma island (Uganda). B.M. (N.H.) reg. no. 1966.2.21.7. DESCRIPTION : based on three specimens, 179-194 mm. standard length ; all are males. Depth of body 40-5-42-3 per cent of standard length, length of head 35-3-36-1 per cent. Dorsal head profile straight, sloping steeply at 4o-5o ; premaxillary pedicels not prominent. Cephalic lateral line pores enlarged, especially those on the preoperculum, preorbital and dentary. Preorbital depth 17-8-18-6 per cent of head, least interorbital width 21-7-25-7 per cent. Snout 1-2-1-3 times as broad as long, its length 32-6-32-8 per cent of head ; diameter of eye 23-9-25-7, depth of cheek 28-0-30-0 per cent. Caudal peduncle 14-8-15-6 per cent of standard length, 1-2 times as long as deep. Mouth somewhat oblique, sloping at 30-35 (a horizontal line drawn through the tip of the lower jaw passes below the orbit). Jaws equal anteriorly or the lower projecting slightly, its length 50-0-52-5 per cent of head, 1-5-1-8 times as long as broad. Posterior tip of the maxilla reaching to a point below the anterior part of the eye. Gill rakers : variable in form, from slender to stout, even in one individual ; the upper 3 rakers branched in one fish. Eight or 9 on the lower part of the first gill arch. 4 8 P. H. GREENWOOD FIG. 4. Haplochromis boops. Drawn by Barbara Williams Scales : ctenoid ; lateral line with 33 (f .2) or 34 scales, cheek with 4 (f .2) or 5 rows. Seven or 8 scales between the dorsal fin origin and the upper lateral line, 7 or 8 between the pectoral and pelvic fin bases. Fins : Dorsal with 15 spines and 9 (f.i) or 10 (f.2) branched rays, anal with 3 spines and 8 (f.i) or 9 (f.2) branched rays. Pectoral 25-2-32-6 per cent of standard length. Pelvics with the first branched ray produced. Caudal subtruncate. Teeth : In the outer row of both jaws, the teeth are small, curved and slender, with about 70 in the upper jaw. Inner teeth, arranged in three rows in both jaws, are unicuspid, small and slightly curved. Osteology. No complete skeleton is available, but radiographs have been studied. The lower pharyngeal bone is triangular, with its dentigerous surface broader than long. Lower pharyngeal teeth are relatively stout with cylindrical necks, and compressed, weakly cuspidate crowns. The teeth are arranged in 20-22 irregular rows. The vertebral counts in three specimens are : 13+16 (f.i) and 13 -f- 17 (f.2). Coloration : Live colours are unknown. Preserved material : Males (adult and sexually active) : ground colour almost uniformly black (including the head, snout, branchiostegal membrane and belly) but with a brownish tinge. Dorsal with the spinous part blotched black on dark grey (black predominating), soft part black but with a hyaline band originating at the level of the tip of the last spine thence passing slightly downwards to end at a point about one third of the distance from the tip of the last branched ray ; the dark band distal to the hyaline strip is less intense than that along the basal part of the fin. Anal black over the spines and along the basal third of the whole fin, remainder sooty ; the five grey-white ocelli are arranged in two rows. Pelvic fins black. A REVISION OF THE LAKE VICTORIA HAPLOCHROMIS SPECIES 49 No female specimens are available. Ecology. Habitat. The two localities from which H. boops is known are in deep water (about 120 ft.) near islands and over mud substrates. Food. Two specimens provided data on feeding habits. In both, the stomach had been everted, but fragments of macerated fish were collected from the pharynx and amongst the folds of the stomach wall. Judging from the scales and fin spines collected in this way, the fishes had fed on Haplochromis. Breeding. All three specimens are adult, sexually active males. Affinities. Haplochromis boops most closely resembles H. thuragnathus ; at present the species can only be distinguished by the more oblique jaw of the latter (see p. 51). Like H. thuragnathus, H. boops appears to be a derivative of the H. serranus species group, probably from an ancestor resembling H. maculipinna. From that species H. boops is immediately distinguished by its broader snout (broader than long, cf. as long as broad), enlarged cephalic lateral line pores, and deeper body. STUDY MATERIAL AND DISTRIBUTION RECORDS Museum and Reg. No. Locality Collector UGANDA B.M. (N.H.) 1966.3.9.182 Near Dagusi Island . . . E.A.F.R.O. B.M. (N.H.) 1966.2.21.7 Off southern tip of Buvuma (Holotype) Island E.A.F.R.O. B.M. (N.H.) 1966.3.9.181 Off southern tip of Buvuma Island E.A.F.R.O. Haplochromis thuragnathus sp. nov. HOLOTYPE : an adult male, 191 mm. standard length, from 120 ft. of water off the southern end of Buvuma island (Uganda) ; B.M. (N.H.) reg. no. 1966.2.21.8. DESCRIPTION : based on three specimens, 191 and 200 mm. standard length. Since so few specimens are available, only ranges for morphometric characters are given. Depth of body 39- 8-41*5 per cent of standard length, length of head 34-5-35-1 per cent. Dorsal head profile slightly concave, sloping at about 30. Cephalic lateral line pores are enlarged, especially those on the preorbital and preopercular bones. Preorbital depth 16-4-18-8 per cent of head, least interorbital width 23-2-24-5 per cent. Snout 1-2-1-3 times as broad as long, its length 30-9-31-8 per cent of head, eye diameter 24-6-26-8, depth of cheek 27-4-29-8 per cent. Caudal peduncle 16-0-17.7 P er cent of standard length, 1-3-1-4 times as long as deep. Mouth oblique, sloping at 40-45, the jaws equal anteriorly or the lower projecting slightly, length of lower jaw 53-6-56-5 per cent of head, 1-7-2-2 times as long as broad. Posterior tip of the maxilla extending to the vertical through the anterior orbital margin or to below the anterior part of the eye. A horizontal drawn antero- 5 o P. H. GREENWOOD posteriorly through the tip of the lower jaw passes through the lower part of the eye (cf. H. boops where the line passes below the orbit.) Gill rakers : stout, 9 on the lower part of the first gill arch. Scales : ctenoid ; lateral line with 32, 33 or 34 scales, cheek with 2-4 rows. Seven or 7^ scales between the upper lateral line and the dorsal fin origin, 7 between the pectoral and pelvic fin bases. Fins. Dorsal with 16 spines and 9 branched rays. Anal with 3 spines and 8 branched rays. Pectoral 26-o-30'0 per cent of standard length. Pelvics with the first branched ray produced, proportionately more so in males. Caudal subtruncate. Teeth. In the outer row of both jaws, the teeth are unicuspid, small and curved ; there are 70 teeth in the upper jaw. Inner teeth are unicuspid, small and slightly curved, and are arranged in 3 series in the upper jaw, and 2 or 3 series in the lower. Osteology. No complete skeleton is available, but radiographs of the three specimens were examined. The lower pharyngeal bone is triangular, its dentigerous surface broader than long. The pharyngeal teeth are relatively coarse, with cylindrical necks and compressed, weakly biscuspid crowns, and are arranged in 20-22 irregular rows. Vertebral counts for all three specimens are : 13 -f- 17. Coloration : Live colours are unknown. Preserved material : Males (adult and sexually active) : ground colour sooty over dark brown dorsally (including the head), silvery on the belly and midflank ; chest and belly darker (i.e. sootier), almost black There is a faint golden flush on the operculum, but it is confined to the centre of this bone and is outlined with a broad dark margin. On the flank of the lighter coloured fish are traces of a broad, dark midlateral stripe. The branchiostegal membrane is black except for its posterior and ventral margins which are greyish. Dorsal fin black except for the distal half of the soft part which is greyish. Caudal dark, but lighter towards the distal margin. Anal black on its basal half and over the spinous portion ; one large white ocellus is present in the fish with the smaller testes, but the other has 8 ocelli arranged in two irregular rows. Pelvic fins black. Female (quiescent) : ground colour brownish, darker on back, head and snout, lighter (with silvery background) on flanks and belly ; very faint traces of a broad (three scale rows deep) interrupted midlateral band on the flanks. Operculum silvery ; a faint, dark lachrymal stripe from the orbit to behind the posterior tip of the maxilla. All fins grey-brown, the anal with 3 small, whiteish spots in the position occupied by the ocelli in males ; pel vies more grey than brown. Ecology. Habitat. All three specimens came from nets set on the mud-bottom in water about 120 ft. deep off the southern tip of Buvuma island. Food. Two of the three specimens examined had fragments of small Haplochromis species in the stomach and intestines ; the guts of the third fish were empty. Breeding. The three specimens are adults, the two males sexually active, the female quiescent. Affinities. Haplochromis thuragnathus is most closely related to H. boops. Indeed, when more material is available it may be shown that the species are not distinct. Information on the live coloration of adult males of the two species would be extremely useful in establishing their status. From H. boops, H. thuragnathus A REVISION OF THE LAKE VICTORIA HAPLOCHROMIS SPECIES 51 is distinguished by its more oblique lower jaw. If a horizontal line is drawn poster- iorly from the tip of the lower jaw (when closed) it passes through the lower part of the eye in H. thuragnathus , but below the eye in H . boops. A third member of this group, H. pachycephalus, is compared with#. thuragnathus. on p. 42. It seems probable that H. thuragnathus was derived from an H. maculipinna-like ancestor, and more particularly from one like H. boops (assuming that the more oblique lower jaw is a derived condition). Like H. pachycephalus and H. boops, Haplochromis thuragnathus differs from other members of the " serranus " phyletic assemblage by its broad snout, and from individual members of the group by various combinations of morphometric characters (see descriptions of H. serranus and H. victorianus in Greenwood, 1962, and of H. maculipinna and H. spekii on pp. 46 and 37 above). STUDY MATERIAL AND DISTRIBUTION RECORDS Museum and Reg. No. Locality UGANDA B.M. (N.H.) 1966.2.21.8 B.M. (N.H.) 1966.3.9.179-180 Off S. tip of Ramafuta Island Off S. tip of Ramafuta Island Collector E.A.F.R.O. E.A.F.R.O. Haplochromis xenostoma Regan, 1922 (Text-figs. 5 and 6) Paratilapia prognatha (part) : Boulenger, 1915, Cat. Afr. Fish., 3, 333 (two specimens, one collected by Sir H. H. Johnston, the other from Entebbe and collected by Degen). Haplochromis xenostoma, Regan, 1922, Proc. zool. Soc. Londn., 185, fig. 10. LECTOTYPE : an immature fish 104-0 mm. S.L. (B.M. [N.H.]reg. no. 1901 . 6 . 24 . 90) collected by Sir H. H. Johnston from Lake Victoria (locality unspecified). FIG. 5. Haplochromis xenostoma, juvenile ; lectotype, about natural size. Regan, Proc. zool. Soc. From 52 P. H. GREENWOOD DESCRIPTION : based on 27 specimens (including the lectotype) 99-203 mm. S.L. but excluding the paralectotype. Only four specimens are less than 140 mm. S.L. (99, 104 [lectotype], 106 and 119 mm. respectively) ; on the basis of this material it appears that body depth and lower jaw length may show some slight positive allometry. Depth of body 27-0-40-0 (M = 36-5) per cent of standard length, head length 34-6-39-7 (M = 37-3) per cent. Dorsal head profile straight or slightly convex, sloping at an angle of 20-3O, its outline noticeably interrupted by the prominent premaxillary pedicels which give it a stepped appearance. Preorbital depth 17-5-22-7 (M = 20-3) per cent of head, least interorbital width 20-4-27-5 (M = 24-5) per cent. Snout 1-2-1-5 times as long as broad, its length 34-2-39-2 (M s= 37-4) per cent of head ; eye diameter 18-5-24-3 (M = 19*7), depth of cheek 24-3-30-8 (M = 28-1) per cent. Caudal peduncle 14-3-18-3 (M = 15-3) per cent of standard length, 1-1-1-7 (modal range 1-3-1-4) times as long as deep ; the lectotype has an unusually shallow peduncle (ratio 1-7). Mouth very oblique, sloping at an angle of 40-45, the lower jaw strongly pro- jecting beyond the upper, its length 50-0-62-0 (M = 57*o) per cent of head and 2-0 (rarely)-3-o times as long as broad (modal range 2-3-2-5). Posterior tip of maxilla generally not reaching the vertical through the anterior orbital margin, but reaching this point in a few specimens. Gill rakers : short and stout, or relatively slender and elongate, the lower one or two reduced ; 8-10 (mode 9) on the lower part of the first gill arch. Scales : ctenoid ; lateral line with 29 (f.i), 31 (f.io), 32 (f.i2), 33 (f.3) or 34 (f.i), cheek with 3 (mode) or 4 rows. Six or 7 (rarely 5 or 8) scales between the upper 1 CM FIG. 6. Haplochromis xenostoma adult. Drawn by Barbara Williams. A REVISION OF THE LAKE VICTORIA HAPLOCHROMIS SPECIES 53 lateral line and the dorsal fin origin, 5-7 (rarely 8) between the pectoral and pelvic fin bases. Fins : Dorsal with 24 (f.Q), 25 (f.17) or 26 (f.i) rays, comprising 14 (f.i), 15 (f.22) or 16 (f_4) spinous and 9 (f.n) or 10 (f.i6) branched rays. Anal with n (f.n) or 12 (f.i6) rays, comprising 3 spines and 8 or 9 branched elements. Pelvics with the first two branched rays produced in both sexes but proportionately more so in adult males. Pectoral 24-2-33-0 (M 28-0) per cent of standard length. Caudal trun- cate, scaled on its proximal half or slightly more. Teeth. In fishes 119 mm. S.L. and above, the outer teeth in both jaws are unicuspid and moderately stout (but occasionally slender) , those in the anterior part of the jaw with a slight inward curvature, and those situated laterally and posteriorly even less curved. The lectotype (104 mm. S.L.) has an outer dentition like that of larger fishes, but in the other small fishes (88-106 mm. S.L.) the outer teeth are distinctly bicuspid anteriorly, and weakly bicuspid laterally and posterolaterally. There are 56-94 (M = 82) teeth in the outer row of the upper jaw. Fishes less than 106 mm. S.L. have either only tricuspid teeth or a mixture of uni- and tricuspids (some weakly so) in the inner series. A specimen 119 mm. S.L. has predominantly bicuspid teeth in the upper jaw, but in the lower jaw the first row of inner teeth is composed of unicuspids, and the other rows of tri- and weakly tricuspids. In all other specimens the inner rows are composed entirely of uni- cuspids. Inner teeth may be implanted somewhat obliquely so as to be medially inclined. The teeth in the outermost row of the upper inner series are often notice- ably larger than their congeners. There are 2-5 rows of inner teeth in the upper jaw, and 2 or 3 (rarely 4) in the lower. Osteology. The neurocranium of H . xenostoma is similar to that of H. victorianus (see Greenwood, 1962) but has a longer preorbital face (30-3 per cent of neurocranial length cf. 26-0 per cent ; the preorbital face being measured from the anterior tip of the vomer to the lateral ethmoid) ; the neurocrania of the two species also differ in that the supraocciptal crest of H. xenostoma is relatively higher and more pointed than in H. victorianus. It differs from the neurocranium of H. serranus (as it does from that of H. victorianus) in its less curved dorsal profile, and its longer preorbital face. The very oblique and prognathous lower jaw is reflected in certain details of the suspensorium (text-fig. 7) ; all comparisons were made with H. serranus, a species 2 MM FIG. 7. Articulatory facet for the quadrate on the articular of (A) H. xenostoma, and (B) H. serranus. 54 P. H. GREENWOOD with a moderately oblique jaw angle of 20-30. In H. xenostoma the articulatory surface of the articular is deeper and more nearly " U " shaped in lateral view ; it lacks the posterior prolongation of its ventral border, but has a marked, near- conical eminence developed postero-medially. All these differences seem to be associated with the oblique angle of the jaw at rest, and the wide angle through which it can be abducted when the mouth is maximally protruded. The small eminence appears to function as a control for the degree of lateral movement of the dentary, particularly when that bone is dropped almost to the horizontal. The posterior vertical limb of the articular in H. xenostoma slopes forward at a much greater angle from the perpendicular, thus providing more space between this bone and the suspensorium. On the basis of my material (two skeletons of H. xenostoma, and one each of H. serranus and H. victorianus) it appears that the horizontal length of the suspen- sorium (as measured in a horizontal plane from the mid-point of the hyomandibular to the articular surface of the quadrate) is greater in H. xenostoma than in the other two species. This could account, at least partly, for the greater prognacity of this species, whose lower jaw has the same relative length as that of H. serranus and H. victorianus. The angle of the hyomandibular relative to the perpendicular is similar in all three species. The premaxilla has a slight median expansion of its dentigerous surface, and the pedicels are relatively shorter than those of H. serranus. As compared with the dentary of H. serranus and H. victorianus, that of H. xenostoma is deeper and stouter, and the dentigerous surface has a more pronounced upward sweep towards the coronoid region. The lower pharyngeal bone is triangular, and fairly stout ; the dentigerous area is as long as broad or very slightly broader than long. Lower pharyngeal teeth are variable in form, usually with coarse, cylindrical necks and compressed, weakly cuspidate crowns ; in some fishes, however, the crowns are distinctly cuspidate. Less commonly, the teeth are slender and compressed, with very weakly cuspidate crowns. The teeth are arranged in 22-24 rows. Vertebral counts in 7 specimens are : 13+15 (f.i), 13 + 16 (f.5) and 13 + 17 (f.i), giving totals of 28-30. Coloration : Live colours are known only for a single juvenile female ; ground coloration silvery, shading to yellowish-grey dorsally, the dorsal surface of the snout dark grey. Dorsal fin dark hyaline, pelvics hyaline, anal yellow, caudal yellowish- grey. Preserved coloration : Males (adult and sexually active) have the ground coloration generally dusky over dark brown dorsally, and silver on the flanks and belly, the latter region together with the chest often with a dusky overlay ; lower jaw and entire branchiostegal membrane also sooty. Snout and preoperculum sooty, but the upper lip is dark brown ; operculum with a faint golden flush. A faint but dark and broad midlateral stripe runs along the flank from the posterior margin of the preoperculum to the caudal fin origin. Dorsal fin yellow-brown, with a sooty overlay on the spinous part, and black lappets ; soft part with dark spots and dashes. Caudal dusky on its proximal threequarters, yellowish-brown distally. Anal light brownish-yellow, with a thin dark line along the base, and black lappets ; A REVISION OF THE LAKE VICTORIA HAPLOCHROMIS SPECIES 55 3 or 4 large, darker whitesh ocelli are present, each with a narrow black outline. The ocelli are arranged in one or two rows. Pel vies entirely dusky. Males (adult but quiescent] : have essentially the same coloration as active males, but some are lighter (that is, with more silvery flanks, and greyish branchiostegal membrane) . Females (adult, and at various degrees of sexual activity) . Greyish-brown above, shading through silvery-grey on the flanks to gold below ; snout and preorbital region dark. On the flanks, a faint dark midlateral band (as in males) may be visible, and in addition, a fainter upper band running just above the upper lateral line. Dorsal fin yellowish-grey with a sooty overlay on the spinous part, and with black lappets ; soft dorsal sometimes darkly spotted. Caudal dark on the proximal two-thirds, lighter distally. Anal yellowish with a faint sooty overlay, sometimes with ill- defined dark spots in the position of the ocelli in males. Pel vies hyaline, usually with a dusky overlay. Ecology. Habitat. The species is apparently confined to sheltered or relatively undisturbed water, being common in bays and gulfs where the water is less than 40 ft. deep, and the bottom is of soft mud, sand or shingle. Available records suggest that the species favours a mud substrate. Food. Of the 21 fishes examined (from 9 localities), only 6 contained food in the guts. In each case the food comprised fragmentary fish remains, unidentifiable except in one instance (a small Haplochromis species) . Breeding. All fishes less than 160 mm. S.L. are immature ; one larger individual (a female 163 mm. S.L.) is also immature. Males and females appear to reach the same maximum adult size. Affinities. The very oblique mouth, marked prognacity, and relatively deep body (at least in adults) serve to distinguish H. xenostoma from the majority of larger Haplochromis species in the lake. There is some resemblance between this species and H. macrognathus and H. plagiostoma, both species with an oblique mouth, and in the case of H . macrognathus, a prominent lower jaw. Haplochromis xenostoma is readily distinguished from H. macrognathus by its broader head (interorbital width 20-4-27-5, M = 24-5% head cf. 16-5-22-2, M = 18-6% ; snout 1-2-1-5 times as long as broad, cf. 1-5-2-2 times in H. macro- gnathus), and its more oblique mouth. The neurocranium also differs, that of H. macrognathus being of the " prognathus " type, whilst the skull of H. xenostoma is clearly of the "serranus" type (see p. in and discussion on pp. 109-113 ; also Greenwood, 1962). From H. plagiostoma, H. xenostoma differs in its larger adult size, longer and narrower snout (34-2-39-2, M = 37-4% head, cf. 28-2-35-5, M=32-5%), more prominent and longer lower jaw (50-0-62-0, M = 57-0% head, cf. 44-0-54-5, M = 49-2% ; lower jaw rarely projecting in H. plagiostoma) and the greater number of teeth in the outer row of the upper jaw (56-94, M =82, cf. 44-68, M =57). Neurocranial form in these two species is similar (see p. 113). Another species with an oblique mouth is H. cavifrons. It is distinguished from H. xenostoma by its unique mottled coloration, lack of prognacity, broader snout (as long as broad or slightly broader than long, cf. 1-2-1-5 times as long as broad), 56 P. H. GREENWOOD and fewer teeth (56-74, M = 63, cf. 56-94, M =82). The profile of the head also differs (compare fig. 6 above with fig. 20 in Greenwood, 1962). Phylogenetically, H. xenostoma could be derived from a species resembling H. plagiostoma ; its affinities seem to lie more with the " serranus " group than with the "prognathus" group to which H. macrognathus belongs. (See also discussion on PP. US)- STUDY MATERIAL AND DISTRIBUTION RECORDS Museum and Reg. No. Locality B.M. (N.H.) 1906.5.30.257 (Paralectotype) B.M. (N.H.) 1966.3.9.92-93 B.M. (N.H.) 1966.3.9.88, 96-106 B.M. (N.H.) 1966.3.9.87 B.M. (N.H.) 1966.3.9.90-91 B.M. (N.H.) 1966.3.9.89 B.M. (N.H.) 1966.3.9.86 B.M. (N.H.) 1966.9.9.85 B.M. (N.H.) 1901.6.24.90 (Lectotype) B.M. (N.H.) 1966.3.9-94-95 UGANDA Entebbe .... Napoleon Gulf, near Jinja . Ekunu Bay Pilkington Bay Off Ramafuta Island (Buvuma Channel) . KENYA Naia Bay (Kavirondo Gulf) . Nanga Bay (Kavirondo Gulf) Off mouth of Nzoia River . LAKE VICTORIA Locality unknown Locality unknown Collector Degen E.A.F.R.O. E.A.F.R.O. E.A.F.R.O. E.A.F.R.O. E.A.F.R.O. E.A.F.R.O. E.A.F.R.O. Sir H. H. Johnston E.A.F.R.O. Haplochromis pseudopellegrini sp. nov. (Text-fig. 8) HOLOTYPE : an adult male 139 mm. standard length, from Pilkington Bay (Uganda). B.M. (N.H.) no. 1966.2.21.2. Named pseudopellegrini because of its resemblance to H. pellegrini Regan. DESCRIPTION : based on 17 specimens (including the holotype) 98 to 150 mm. standard length. Depth of body 29-0-33-5 (M = 30-9) per cent of standard length, length of head 32-4-37-0 (M = 33-8) per cent. Dorsal head profile gently curved (rarely straight) but interrupted by the prominent premaxillary pedicels. Preorbital depth 19-1-22-4 (M = 20-4) per cent of head length, least interorbital width 20-4-27-1 (M = 23-9) per cent. Snout a little longer than broad (1-2-1-3 times), its length 35-1-39-0 (M =37-1) per cent of head; eye diameter 18-5-25-4 (M = 20-6), depth of cheek 24-0-29-6 (M = 26-8) per cent. Caudal peduncle 16-3-20-8 (M = 18-3) per cent of standard length, 1-3-2-0 (modal range 1-6-1-7) times as long as deep. A REVISION OF THE LAKE VICTORIA HAPLOCHROMIS SPECIES 57 1 CM Haplochromis pseudopellegrini ; this specimen shows extreme development of an obliquely truncate caudal fin. Drawn by Barbara Williams. Mouth slightly oblique, lower jaw projecting a little. Premaxilla slightly ex- panded medially. Lower jaw 46-8-53-4 (M = 49-1) per cent of head, 2-0-2-6 (modal range 2-1-2-4) times as long as broad. Posterior tip of the maxilla reaching, or almost reaching the vertical through the anterior orbital margin, extending a little beyond this point in a few specimens. Gill rakers : of variable form, from relatively slender to moderately stout (reduced to short knobs in one specimen), the upper four sometimes flat ; 9 (rarely 8 or 10) on the lower part of the first gill arch. Scales: ctenoid; lateral line with 32 (f.4), 33 (f.7), 34 (f.4) or 35 (f.i) scales. Cheek with 4 (less commonly 3, rarely 5) rows. Six or 7 (rarely 7^) scales between the dorsal fin origin and the upper lateral line, 7 (occasionally 6, rarely 8) between the pectoral and pelvic fin bases. Fins. Dorsal with 23 (f.i), 24 (f.4) or 25 (f.i2) rays, comprising 15 (f.g) or 16 (f.8) spinous and 8 (f.i), 9 (f.i2) or 10 (f-4) branched rays. Anal with n (f.io) or 12 (f.7) rays, comprising 3 spines and 8 (f.io) or 9 (f.7) branched elements. Pectoral 22-7-28-0 (M = 24-9) per cent of standard length. First branched ray of pelvic fin slightly produced in both sexes, proportionately more so in adult males. Caudal scaled over its proximal half ; in most specimens the fin is truncate or subtruncate but in a few the lower half, or the ventro-posterior corner, slopes obliquely forward and may also be rounded. Teeth. The outer row of teeth in both jaws is composed of slender and slightly recurved unicuspids ; there are 35-52 (M = 44) teeth in the outer row of the upper jaw. Inner teeth are also unicuspid, and are implanted at an angle varying from near vertical to almost horizontal, the latter condition being most common. There are 2 (rarely 3) rows of inner teeth in the upper jaw, and I or 2 rows in the lower. Osteology. The neurocranium of H. pseudopellegrini shows some similarity with 58 P. H. GREENWOOD that of H. mento and other species in the " prognathus " group (see Greenwood, 1962, and p. no), but at the same time it retains characteristics of the more generalized skull seen in H. serranus and its allies. In this respect it resembles the neurocranium of H. prognathus, but is slightly less " mento "-like. In general appearance and proportions it is similar to the neurocrania of H. bayoni and H. dentex but lacks the characteristically decurved ethmo-vomerine region of these species (see Green- wood, op. cit.}. The premaxilla is moderately beaked, that is, the dentigerous part of the bone is somewhat expanded medially ; the pedicels are short, being about two-thirds the length of the dentigerous arm. The lower pharyngeal bone is triangular, its dental surface is as long as broad or slightly longer than broad. Lower pharyngeal teeth are slender and compressed, those in the two median rows are slightly coarser than their lateral congeners ; the teeth are arranged in 22-24 rows. Vertebral counts in 14 specimens are : 13+17 (f.4), 13 + 18 (f.i), 14 + 17 (f.g), giving totals of 30 and 31. Coloration : Live colours are known only for adult and sexually active males, which have the dorsal surface of the head and body dark brown, the flanks and belly golden-yellow overlain by an orange-red flush on the chest and anterior flanks, and also on the operculum. Dorsal fin dark neutral with a slight orange flush. Caudal dark neutral, with a reddish flush at the base and over the ventral quarter of the fin. Anal neutral, with yellowish-red ocelli. Pelvics are sooty. Preserved material : Males (adult and active] : ground colour brownish overlying silver, silvery-yellow on the chest and belly, and on the operculum ; rest of head (including the branchiostegal membrane) brownish with very faint traces of a lachry- mal blotch. Dorsal fin hyaline, with a faint, narrow black band running from the tip of the eleventh spine to about the middle of the last branched ray (i.e. curving gently downwards). Caudal brownish to hyaline, dark on the proximal half. Anal hyaline, with faint traces of one or two whiteish-grey ocelli. Pelvics sooty, darkest on the anterior half. Adult but quiescent males are light brown dorsally, shading to silvery below, some showing a faint but broad and dark midlateral stripe, and a fainter upper lateral band above the upper lateral line. The two lateral bands are connected by 4-6 vertical bars, which extend ventrally a little below the midlateral band ; where the lines intersect, the lateral one is diffusely expanded. A faint lachrymal blotch is present below the anterior part of the orbit. All fins are yellowish-brown, the soft dorsal and the proximal part of the caudal are often darkly maculate. Anal with 2 or 3 faint, whiteish ocelli. Pelvics variable, from yellowish-brown to sooty. Ecology. Habitat. The species is recorded from four localities only. Two of these are shallow, sheltered bays, one is a fairly exposed, offshore and deep (90 ft.) channel, and the fourth is not fully documented except for a note that the nets were set in water about 20 ft. deep. In all, the substrate is of organic mud. Food. Fourteen specimens were examined, and of these only two contained food, very fragmentary and generically unidentifiable fish remains. Breeding. Little information is available on the reproductive biology of this species. The sex of the smallest specimen (98 mm. S.L.) is indeterminable ; the A REVISION OF THE LAKE VICTORIA H A PLOCHRO M I S SPECIES 59 others (132-153 mm.) are all adults, and only two are females. The two largest fishes are males. Affinities. In general appearance H. pseudopellegrini closely resembles H. pellegrini ; however, it reaches a much larger adult size, and the preserved coloration of adult males is much lighter (uniformly dark brown, nearly black in H. Pellegrini, light brown over silver in H. pseudopellegrini}. The species also differ in certain morphometric characters. Haplochromis pseudopellegrini has a longer snout (35-1-39-0, M = 37-1% head, cf. 30-8-36-0, M = 34-0%), a longer and more slender caudal peduncle (16-3-20-8, M = 18-3% standard length, cf. 13-2-17-8, M = 15-4% ; length/depth ratio 1-3-2-0 [modal range 1-6-1-7], c f- I M=34'4%), slightly narrower interorbital region (20-4-27-1, M = 23-9% head, cf. 23-4-30-2, M = 27-4%), longer and narrower lower jaw (48-6-53-4, M =49-1% head, cf. 39-2-48-2, M = 44-4% ; length/breadth ratios 2-0-2-6 [modal range 2-1-2-4] cf. 1-5-2-3 [mode 2-0]), fewer teeth in the outer row of the upper jaw (35-52, M = 44, cf. 48-74, M = 62). From H. gowersi it differs in head shape (and neurocranial form), and in having a much shorter head (34-2-37-0, M = 33-8% standard length, cf. 35-8-38-4, M = 37-0%), a broader snout, a larger eye (18-5-23-4, M = 20-6% head, cf. I5'5-I9'3, M = 17-5%), a slightly shallower cheek (24-0-29-6, M = 26-8% head, cf. 27-8-33-3, M = 29-5%), and a longer caudal peduncle (16-3-20-8, M = 18-3% standard length, cf. I3-3-I7-6, M = i4-8%). From H. dentex, H. pseudopellegrini differs, superficially, by its less strongly de- curved dorsal head profile. At a deeper level, there are differences in the shape of the neurocranium, that of H. dentex having a sharply decurved ethmoid-vomer region (see Greenwood [1962], p. 168 and fig. 25) ; but in other respects, the neurocrania of the two species are similar. Other interspecific differences lie in the more numerous and closely set teeth of H. pseudopellegrini (35-52, M =44, cf. 32-48, M =36), its deeper body (29-0-33-5, M = 30-9% of standard length, cf. 24-6-29-5, M = 26-7), and the greater posterior extension of the maxilla in this species (posterior maxillary tip reaching anterior orbital margin or to below the eye, cf. not reaching the orbital margin) . From H. estor, H. pseudopellegrini differs, principally, in having a shorter head 60 P. H. GREENWOOD (32-4-37-0, M = 33'8% standard length, cf. 37-2-38-5, M = 37'8%), a shorter lower jaw (46-8-53-4, M =49-1% head, cf. 54-2-57-5, M =55-5%), and a longer caudal peduncle (16-3-20-8, M = 18-3% standard length, cf. 14-3-16-1, M = 15-8%). Neurocranial form in the two species differs, that of H. estor belonging to the " prognathus "group, whilst that of H. pseudopellegrini has stronger affinities with the " altigenis "-type (see p. no). The resemblance between H. pseudopellegrini and H . mento is probably the most distant of all. Osteologically, there is a clear-cut difference in neurocranial form (like that distinguishing H. pseudopellegrini and H. estor}, and in most specimens the external head shape is distinctive (cf. fig. 12 in Greenwood [1962] with Text-fig. 8 above) . Nevertheless, most cephalic morphometric characters are similar in the two species, although the snout of H . pseudopellegrini is broader (length/breadth ratio I-I-I-3 cf. 1-5-1-8 in H. mento). The outer teeth in H. mento are stouter and more strongly curved than those of H. pseudopellegrini ; the range for the number of outer upper jaw teeth overlaps in the two species, but the mean for H. pseudopellegrini is lower (44 cf. 52). Phyletically, H. pseudopellegrini appears to be a derivative from an H, guiarti- like stem, and thus shows relationship with H. bayoni and H. dentex. However, unlike these species it also shows relationship with both H . altigenis and H. pellegrini. Structurally, H. pseudopellegrini could represent an ancestral level in the evolution of H. altigenis. Museum and Reg. No. Locality Collector UGANDA B.M. (N.H.) 1966.3.9.299-310 Ekunu Bay .... E.A.F.R.O. B.M. (N.H.) 1966.3.9.296 Trawl in Buvuma Channel off Nasu Point . . . .- E.A.F.R.O. B.M. (N.H.) 1966.3.9.297 Pilkington Bay . . . . ~ E.A.F.R.O. B.M. (N.H.) 1966.2.21.2 Pilkington Bay . . . E.A.F.R.O. (Holotype) B.M. (N.H.) 1966.3.9.298 Sesse Islands . . . .-..-:; E.A.F.R.O. Haplochromis altigenis Regan, 1922 (Text-figs. 9 and 10) Paratilapia longirostris (part) : Boulenger, Cat. Afr. Fish., 3, 332. Pelmatochromis spekii (part) : Boulenger, 1915, op. cit., torn, cit., 417. Haplochromis altigenis Regan, 1922, Proc. zool. Soc. Londn., 175, PI. i. LECTOTYPE : a male, 186 mm. standard length from Bunjako (Uganda), collected by Degen. B.M. (N.H.) reg. no. 1906.5.30.294. DESCRIPTION : based on 25 specimens (including the lectotype and paralectotype), 100-202 mm. standard length. Depth of body 28-4-34-7 (M = 31-5) per cent of standard length, length of head 36-7-39-5 (M = 38-2) per cent. Dorsal head profile gently curved, rather variable in its shape, tending to slope more steeply in large fishes which therefore have more rounded profiles ; the two type specimens have the most strongly sloping head A REVISION OF THE LAKE VICTORIA HAPLOCHROMIS SPECIES 61 FIG. 9. Haplochromis altigenis ; lectotype, about -57 times natural size. From Regan, Proc. zool. Soc. profiles (ca 40) ; most other specimens lie in the range 30-35. The premaxillary pedicels are prominent, and clearly break the outline of the profile. Altogether, one is left with the impression of a heavy-headed fish. Preorbital depth 17-6-23-7 (M = 19-9) per cent of head length, least interorbital width 17-4-25-0 (M 20-2) per cent. Snout 1-1-1-3 times as long as broad, its length 36-6-42-2 (M =39-3) per cent of head, eye diameter 16-9-21-7 (M = 19-2), depth of cheek 25-6-34-5 (M = 30-5) per cent. Cheek depth may show positive allometry with standard length in fishes > 180 mm. ; the three largest fishes have the deepest cheeks. Caudal peduncle 12-4-16-0 (M = 14-7) per cent of standard length, 1-2-1-4 (mode 1-3) times as long as deep. FIG. 10. Haplochromis altigenio, to show the usual head profile in fishes less than ca. 1 80 mm. S.L. Drawn by Barbara Williams. ZOOL. 15, 2. 4 62 P. H. GREENWOOD Mouth slope variable, from near horizontal to moderately oblique (20-30). Lips slightly thickened, the premaxilla expanded medially. Lower jaw generally projecting (but jaws equal anteriorly in some specimens), its length 51-0-55-8 (M = 52-7) per cent of head, 1-8-2-4 (modal range 1-9-2-0) times as long as broad. Posterior tip of the maxilla almost reaching the vertical through the anterior orbital margin (reaching this point in a few specimens) . Gill rakers : stout (finer in the two smallest fishes), the upper 1-4 sometimes flattened and anvil-shaped ; 8 or 9 (rarely 7 or 10) on the lower part of the first arch. Scales : ctenoid ; lateral line with 31 (f.7), 32 (13) or 33 (.4), cheek with 4-6 (mode 5) rows. Six to 8 scales between the upper lateral line and the dorsal fin origin ; chest and anterior belly scales small, 7 or 8 (rarely 9) between the pectoral and pelvic fin bases. Fins : Dorsal with 24 (f.8), 25 (f.i6) or 26 (f.i) rays, comprising 15 (f.22) or 16 (f.3) spinous and 9 (f.io) or 10 (f.15) branched rays. Anal with n (f.2), 12 (f.22) or 13 (f.i) rays, comprising 3 spines and 8 (f.2), 9 (f.22) or 10 (f.i) branched elements. Pectoral 20-1-26-2 (M = 22-6) per cent of standard length. Pelvics with the first branched ray produced, usually more so in males but almost as elongate in some females. Caudal subtruncate, scaled on its proximal half. Teeth. In the size range of fishes studied, all teeth in the outer row of both jaws are stout, unicuspid and curved. There are 40-60 (M = 50) teeth in the outer row of the upper jaw. The inner teeth of the two smallest fishes (90 and 119 mm. S.L.) are predominantly unicuspids, but some weakly tricuspid teeth occur in both jaws. In all other speci- mens, only unicuspids are found ; these are moderately large, curved and implanted obliquely. The inner tooth rows are often irregularly arranged, with 3 (rarely 4) rows in the upper jaw, and 2 or 3 in the lower. Osteology. The neurocranium of H. altigenis resembles that of H. bayoni, but is relatively broader in the otic region. Thus, although it shows some of the characters associated with the H. mento-type skull (see Greenwood, [1962], fig. 25) it still retains the curved preorbital profile, greater preorbital skull depth, and broad otic region of the more generalized neurocranium. In these characters it also resembles the neurocranium of H. pseudopellegrini. The dentary and premaxillary show no outstanding characters. The former is a stout bone, and the dentigerous surface of the latter is but moderately expanded medially (less so, for example, than in H. bayoni). The dentigerous surface of the lower pharyngeal bone is as broad as long but is broader than long in large fishes. Anteriorly this surface narrows rather abruptly so that the apex of the dentigerous triangle is produced into a narrow " stem ". The lower pharyngeal teeth are rather coarse, somewhat compressed and clearly but weakly cuspidate. Vertebral counts in 7 specimens are : 13+16 (.5) and 13 + 17 (f.2), giving totals of 29 and 30. Coloration. Live colours are unknown. Preserved material : Males (adult and sexually active): ground coloration a very dark brown, almost black on the dorsal A REVISION OF THE LAKE VICTORIA HAPLOCHROMIS SPECIES 63 surface and snout, and with a sooty overlay on the chest. Lower jaw and branchio- stegal membrane pinkish-brown ; margin of the preoperculum outlined in dark brown. Dorsal fin yellow-brown, lappets black, and with a faint black outline to the margin of the soft part. Caudal darker yellow-brown (darkest proximally). Anal yellow-brown, with 2 large, dead-white ocelli. Pelvics very dark brown, appearing black when folded. Pectorals dark brown (dark pectorals are unusual). Males (adult but sexually quiescent) have a light brown ground coloration, darker on the dorsal surface of the snout, head and body. A prominent, but narrow, dark lachrymal stripe runs from the lower anterior border of the eye, passes almost vertically downwards behind the posterior tip of the maxilla to end on the dentary ; a slightly broader dark bar lies immediately anterior to the vertical limb of the preoperculum. The lower jaw and branchiostegal membrane are very light brown ; the chest is somewhat dusky. The flanks and caudal peduncle are crossed by seven very faint, moderately broad dark bars ; these extend (on the flanks) from the dorsal fin base almost to the ventral body outline. Dorsal fin yellow-brown, the lappets black and the soft part darkly maculate. Caudal yellowish-brown, darkest on the proximal half. Anal dark yellow-brown. Pelvics dark brown, nearly black, on the anterior half, otherwise light yellow-brown. Male (immature) : essentially as for females (see below), but with black lappets on the dorsal fin, the caudal uniformly light but with some dark spots on the upper fifth, 3 faint whiteish ocelli on the anal, and a dusky overlay on the pelvics. Note on the coloration of the type specimens. Both these specimens are males, but both are now a pale silver, shading to white. One specimen (that illustrated by Regan, see fig. 9 above) has a very faint midlateral stripe, which is now much fainter than is shown in the figure. Both fishes have dusky pelvics. In one, there are five, large, dead- white ocelli (arranged in three rows) on the anal fin, and in the other there are four (in two rows). The difference in coloration between these specimens and those described above is, presumably, due both to time and to the fact that the types were not fixed in formol but in alcohol. Females (immature and adult) : ground coloration light brown, darkest dorsally, and shading to yellowish-silver on the chest and belly. Lachrymal and preopercular stripes are as described for males, but in addition there are two, faint, parallel dark bands across the snout. The lower jaw and branchiostegal membrane are light yellow-brown. A faint dark midlateral band runs along the flank from slightly behind the opercular margin to the base of the caudal fin (sometimes extending onto the fin itself), and is crossed by 8 to 10, moderately broad vertical bars which extend from the dorsal fin base to about the level of the pectoral fin. Dorsal fin light greenish-brown, the soft part darkly maculate. Caudal yellow-brown on its distal quarter to third, dark brown basally. Anal greenish-brown. Pelvics light yellow-brown, somewhat darker along the anterior margin. Three fishes (140, 148 and 149 mm. S.L.) caught on two occasions off Kisigala; Point, North Rome Island, show typical piebald coloration of black on silver; all three are immature females. These are the first known examples of piebald polychromatism amongst the piscivorous species-groups of Lake Victoria Haplo- chromis, although the phenomenon is recorded from other trophic and phyletic lines 64 P. H. GREENWOOD (see Greenwood, 1956, 1957, 1959, and p. 95 below). Ecology. Habitat. The species occurs over both hard and soft substrates in sheltered bays and gulfs, as well as off-shore in places where the water is relatively undisturbed ; apparently it does not occur at depths of over 50 ft. Food. Eleven of the 26 specimens examined (from 15 localities) contained food in the gut. In all cases this consisted of finely macerated fish remains, identifiable as cichlids in three specimens, and as Haplochromis species in six others. Breeding. Haplochromis altigenis is a female mouth breeder. Fishes less than 145 mm. S.L. are immature, or, in the upper levels of the range, show early signs of maturation. The largest specimen (202 mm. S.L.) is a female, but both sexes occur in the size range 170-194 mm. Affinities. At least superficially, four species, H. pseudopellegrini, H. estor, H. dichrourus and H. gowersi, closely resemble H. altigenis. A fifth species, H. squamulatus shows a more distant resemblance. Similarities between H. altigenis and H. dichrourus are considered elsewhere (see p. 68) ; the resemblances between these species may indicate a fairly close phyletic relationship. From H. pseudopellegrini, H. altigenis is distinguished by its longer head (36-7- 39-5, M = 37-2% of standard length, cf. 32-4-37-0, M=33-8%), deeper cheek (25-6-34-5, M =30-5% head, cf. 24-0-29-0, M = 26-8%), longer lower jaw (51-0- 55-8, M = 52-7% head, cf. 46-8-53-4, M = 49-1%) and its shorter and deeper caudal peduncle (12-4-16-0, M = 14-7% standard length, cf. 16-3-20-8, M = 18-3%, length/depth ratio 1-2-1-4 [mode 1-3] cf. 1-3-2-0 [modal range 1-6-1-7]). The neurocranium in the two species is generally similar. It seems that H. altigenis could have evolved from an H. pseudopellegrini-\ike ancestor. From H. estor, H. altigenis differs in its longer snout (36-6-42-2, M = 39'3% head, cf. 34-5-37-0, M =36-2%), smaller eye (16-9-21-7, M = 19-2% head, cf. 22-8-25-4, M = 24'5%), much deeper cheek (25-6-34-5, M = 30-5% head, cf. 22-8-25-4, M = 24-5%) and slightly shorter lower jaw (51-0-55-8, M = 52-7% head, cf. 54-2-57-5, M =55'5%). There are fairly marked differences between the neuro- crania of the two species, that of H. estor being of the " prognathus " type. Superficially, the resemblances between H. gowersi and H . altigenis are great, and some difficulty may be experienced in separating certain specimens of the two species. However, there are distinct differences between modal specimens of the two species, and there is a clear difference in neurocranial form ; the skull of H. gowersi is of the " prognathus " type. This difference in neurocranial shape probably accounts for the more declivous snout and dorsal head profile of H. altigenis, and the greater prominence of the premaxillary pedicels in this species. In addition to these qualitative differences, H. altigenis differs from H. gowersi in having the interorbital width equal to or slightly greater than the preorbital depth (interorbital less than preorbital in H. gowersi), a broader head (greatest width, measured at about the middle of the operculum, 41-5-45-5, M = 42-7% head, cf. 35-6-39-0, M = 36-8) ; the greater head breadth is also reflected in the broader lower jaw of H. altigenis. Haplochromis squamulatus and H. altigenis both have noticeably rounded head profiles, and very small chest and anterior belly scales. However, they differ in A REVISION OF THE LAKE VICTORIA HAPLOCHROMIS SPECIES 65 several morphometric characters. Haplochromis altigenis has a longer head, narrower interorbital, longer snout, deeper cheek and a longer lower jaw (see p. 87). The neurocrania show several interspecific differences (that of H. squamulatus showing affinities with the neurocrania of H. martini and H. michaeli) and there are marked differences in preserved coloration. Considering the evidence available, both morphometric and osteological, it seems H. altigenis could have evolved from an H. pseudopellegrini-like. stem ; certainly its neurocranial form does not favour a closer association with H. estor and H. gowersi than with H. pseudopellegrini. STUDY MATERIAL AND DISTRIBUTION RECORDS Museum and Reg. No. Locality Collector UGANDA B.M. (N.H.) 1906.5.30.294 Bunjako Degen (Lectotype) B.M. (N.H.) 1906.5.30.295 Bunjako ..... Degen (Paralectotype) B.M. (N.H.) 1966.3.9.220-227 Napoleon Gulf, near Jinja . . E.A.F.R.O. B.M. (N.H.) 1966.3.9.228-235 Karenia Beach, near Jinja . . E.A.F.R.O. B.M. (N.H.) 1966.3.9.215-216 Near Ramafuta Island (Buvuma Channel) . . E.A.F.R.O. B.M. (N.H.) 1966.3.9.219 Pilkington Bay . . . . E.A.F.R.O. B.M. (N.H.) 1966.3.9.218 Manadu Island . . . . E.A.F.R.O. B.M. (N.H) 1966.3.9.213-214 Off Entebbe harbour . . . E.A.F.R.O. TANZANIA B.M. (N.H.) 1966.3.9.217 Mwanza Harbour . . . E.A.F.R.O. LAKE VICTORIA B.M. (N.H.) 1966.3.9.211-212 Locality unknown . . . E.A.F.R.O. Haplochromis dichrourus Regan, 1922 (Text-fig, n) Paratilapia serranus (part) : Boulenger, 1915, Cat. Afr. Fish., 3, 334. Haplochromis dichrourus Regan, 1922, Proc zool. Soc. Londn., 178, fig. 6. HOLOTYPE : a juvenile female, 113-0 mm. S.L., from Buganga ; B.M. (N.H.) reg. no. 1906.5.30.265. DESCRIPTION : based on eight specimens (including the holotype), 84-186 mm. S.L. Depth of body 28-6-35-5 (M = 32-1) per cent of standard length, length of head 35.3-37-7 (M = 36-6) per cent. Dorsal head profile convex, sloping steeply (ca 40) especially in the snout region ; premaxillary pedicels prominent and breaking the outline of the profile. Preorbital depth 16-4-21-0 (M = 18-7) per cent of head, least interorbital width 18-0-24-6 (M = 21-3) per cent. Snout 1-10-1-25 times as long as broad, its length 66 P. H. GREENWOOD 34-1-41-0 (M 37-4) per cent of head, eye diameter 19-4-24-6 (M = 21-1), depth of cheek 23-0-30-6 (M = 27-6) per cent. Caudal peduncle 12-4-17-2 (M = 15-2) per cent of standard length, 1-1-1-5 times as long as deep. Mouth slightly oblique (20-30), lips thickened, premaxilla slightly expanded medially. Lower jaw projecting, its length 49-0-54-0 (M = 52-7) per cent of head, 1-8-2-6 (no distinct mode) times as long as broad. Posterior tip of maxilla reaching the vertical through the anterior margin of the orbit, or to below the anterior part of the eye. FIG. ii. Haplochromis dichrourus ; holotype, about -93 times natural size. From Regan, Proc. zool. Soc. Gill rakers : stout, the lower 2 or 3 reduced, the upper 3 or 4 flat and expanded in some specimens ; 8 or 9 (mode) on the lower part of the first gill arch. Scales : ctenoid ; lateral line with 32 (f.2), 33 (f.4) or 34 (f.2), cheek with 3 (f.i), 4 (f.i), 5 (f.5), or 6 (f.i) rows. Seven to 9 scales between the upper lateral line and the dorsal fin origin, 6-9 (mode 8) between the pectoral and pelvic fin bases ; scales on the chest and belly small. Fins : Dorsal with 25 rays, comprising 15 (f.i) or 16 (f-7) spinous and 9 (f-7) or 10 (f.i) branched rays. Anal with 11 (f.2), 12 (f.5) or 13 (f.i) rays, comprising 3 spines and 8 (f.2), 9 (f.5) or 10 (f.i) branched elements. Pectoral fin 21-2-27-3 (M = 23-7) per cent of standard length. Pelvics with the first, and to a lesser degree, the second branched rays produced, slightly so in females and markedly elongate in males. Caudal truncate or subtruncate. Teeth. In all specimens the outer teeth (numbering 48-70, M = 58 in the upper jaw) are unicuspid, slender and very strongly curved (the tips of the anterolateral premaxillary teeth are not visible when the specimen is viewed laterally) . The inner teeth in fishes > 149 mm. S.L. are all unicuspid, curved and obliquely implanted. In the four smaller specimens, the 113 mm. fish (holotype) has mostly tricuspids in the upper jaw and an admixture of tri- and unicuspids (the latter A REVISION OF THE LAKE VICTORIA HAPLOCHROMIS SPECIES 67 predominating) in the lower jaw ; the 84, 101 and 121 mm. specimens have a mixture of unicuspids and weakly bicuspids in both jaws ; as in the larger fishes, these teeth are curved and obliquely implanted. The inner teeth are arranged in 1-3 series in the upper jaw, and in 2 or 3 in the lower (in a single row in one fish). Osteology. With so few specimens available, it has not been possible to prepare a complete skeleton, but radiographs have been studied. The lower pharyngeal bone is fine, with a triangular dentigerous surface which is equilateral or slightly broader than long. The lower pharyngeal teeth are slender, with cylindrical necks and slightly compressed weakly cuspidate crowns, and are arranged in 20-22 rows. Vertebral counts in 5 specimens are : 13+16 (f.i) and 13 + 17 (f. 4), giving totals of 29 and 30. Ecology. Habitat. The species has a wide depth range, from 10-90 feet. It is found in both sheltered and exposed places (including beaches) over sand and shingle substrata. Food. Of the five specimens examined (from four localities) four contained food in the gut. In three fishes this consisted solely of fish remains (Haplochromis sp.), and in the fourth fish-remains (a cichlid) and fragments of larval insects (one probably a dragon-fly, the other Ephemeroptera) . Breeding. Little information is available ; three of the smallest fishes (101, 113 and 121 mm. S.L.) are immature females. An unusual feature of this species is the coloration of females, which, at least in preserved specimens, seems to be as polychromatic as that of males, and certainly more complex than the female coloration of other species (except, perhaps, H. chromogynos) . Coloration. The only information on live colours is provided by brief field notes made on a sexually active male caught in deep water near Soswa island. In this fish the median fins and back were described (by Mr. J. D. Kelsall) as vivid flame red, the belly as jet black. Preserved Colours : Males. The most extreme pattern is shown by a fish which, although adult, is in an early stage of sexual activity. The upper part of the head (above the level of the lower orbital rim) brownish, snout yellowish and crossed by a thin, well-defined and dusky bar ; running parallel with the premaxillary pedicels are a pair of short, dark bars, each of which (at about the level of the pedicel tips) broadens somewhat and curves sharply at right angles to meet the orbit. The lower part of the head (cheeks, preorbital and the entire operculum) jet black. This colour extends onto the chest but does not reach above the level of the pectoral fin base. Immediately behind the pectoral fin, the margin of this black area curves ventrally but rises again above the vent so that the posterior half of the body is black on its lower half. On the caudal peduncle the dark area rises again to cover about the entire lateral aspect. Above the black areas the body is brownish. Dorsal and caudal fins are dark yellow-orange, with a narrow black crescent at the caudal base. The anal fin is a similar colour, with two ocelli, each outlined by a narrow black margin. Pelvics jet black, except for the yellowish innermost ray of each side. In two other males (both sexually active, with convoluted testes and therefore thought to be more mature than the specimen above) the head and anterior half of 68 P. H. GREENWOOD the body are light orange-brown the posterior half and the caudal peduncle dusky brown. A well-defined black band crosses the snout (just above the upper lip) to the anterior margin of the preorbital bone. Another transverse black band at the level of the pedicel tips, runs from orbit to orbit. An interrupted band extends from the upper posterior margin of the orbit across the nape. A dusky, nearly vertical lachrymal stripe is present. The branchiostegal membrane and chest are black, but the belly is brown. Dorsal fin yellowish, the soft part is maculate, the spots being clear. Anal yellowish but dusky along its base, and with two, hyaline ocelli (set, in one specimen, in the dark basal zone). Caudal yellowish-grey, but with a broad, dusky band at its base. The pelvics are black except for the distal half of the membrane between the last two rays. Females (adult and immature] : body and head dark grey-brown with a blueish hue. Branchiostegal membrane and chest dusky as are the lower parts of the operculum, preoperculum and interoperculum. Faint traces of snout and lachrymal bands (like those of males) are visible, as is a nuchal bar. Dorsal fin greyish to light sooty, the lappets dark. Anal pale orange with a single, well-defined, dead- white ocellus. Caudal with a dark base, the upper half grey and the lower orange. Pelvics light sooty, the pigment most concentrated on the anterior half of the fin. Affinities. The anomalous coloration of female H. dichrourus was commented upon above. Apart from the relatively infrequent piebald females in certain species, and the occurrence of a piebald coloration as the usual one in females of H. chromo- gynos (see Greenwood, 1959), I know of no other species in Lake Victoria with colourful females. In its gross morphology, H. dichrourus closely resembles H. altigenis although there are marked differences in the preserved coloration of the females. Both species have very small chest scales. They differ in that the maxilla of H. dichrourus reaches further posteriorly (rarely reaching the orbit in H. altigenis) and the pre- maxilla is not markedly expanded medially. Also, the teeth in H. dichrourus are more strongly curved, so that when viewed laterally, their tips are hidden ; in H. altigenis, the tips can always be seen. The two smallest fishes resemble specimens of H. Pellegrini of a similar size. They are, however, distinguished by their more slender and much more strongly recurved outer jaw teeth (in H. pellegrini these teeth have a gentle curvature confined to the distal part), and by having the maxilla extending further posteriorly (to below the anterior orbital margin). Also, in H. dichrourus the premaxillary pedicels are more prominent, and there are differences in preserved coloration. Phyletically, H. dichrourus could be related to H. altigenis and H. pellegrini. STUDY MATERIAL AND DISTRIBUTION RECORDS Museum and Reg. No. Locality Collector UGANDA B.M. (N.H.) 1906.5.30.265 Buganga ..... Degen (type) B.M. (N.H.) 1966.3.9.186 J in j a, off golf course . . . E.A.F.R.O. B.M. (N.H.) 1966.3.9.187 Katebo . ;.:* - . gj$* . E.A.F.R.O. B.M. (N.H.) 1966.3.9.188-9 Karenia, Napoleon Gulf . . E.A.F.R.O. A REVISION OF THE LAKE VICTORIA HAPLOCHROMIS SPECIES 69 TANZANIA B.M. (N.H.) 1966.3.9.185 B.M. (N.H.) 1966.3.9.183-4 Off Soswa Island Beach N. of Majita J. D. Kelsall E.A.F.R.O. Haplochromis paraguiarti sp. nov. (Text-fig. 12) HOLOTYPE : an adult male 130 mm. standard length, from a beach near Nasu Point, Buvuma Channel (Uganda); B.M. (N.H.) reg. no. 1966.2.21.6. Named paraguiarti because of its close superficial resemblance to H. guiarti. DESCRIPTION: Based on 31 specimens (including the holotype), 70-156 mm. standard length. Depth of body 31-7-36-6 (M = 34-1) per cent of standard length, length of head 33-0-36-8 (M = 34*8) per cent. Dorsal head profile straight or very slightly curved, the premaxillary pedicels forming a slight prominence. Preorbital depth 14-5-19-5 (M = 16-9) per cent of head length, least interorbital width 22-9-27-7 (M = 25-3) per cent. Snout as long as broad or slightly longer (1-2 times), its length 31-0-36-2 (M = 33-5) per cent of head, eye diameter 23-8-28-6 (M = 26-2), depth of cheek 20-0-26-1 (M = 23-8) per cent. Caudal peduncle 14-7-18-6 (M = 16-6) per cent of standard length, 1-3-1-9 (modal range 1-3-1-5) times as long as deep. Mouth horizontal or slightly oblique, the jaws equal anteriorly. Lower jaw 41-0-48-6 (M = 45-0) per cent of head, 1-7-2-4 (modal range 1-9-2-1) times as long as broad. Posterior tip of the maxilla reaching the vertical through the anterior orbital margin, occasionally extending to below the eye. Premaxilla slightly expanded medially. 1 CM FIG. 12. Haplochromis paraguiarti. Drawn by Barbara Williams. 70 P. H. GREENWOOD Gill rakers : moderately stout (relatively slender in fishes less than 90 mm. S.L.), the upper 4-6 flat and sometimes branched in fishes over 115 mm. S.L., the lower i or 2 reduced ; 8-10 (mode 9), rarely ir, on the lower part of the first gill arch. Scales : ctenoid ; lateral line with 31 (f.4), 32 (f.i2), 33 (f.i2) or 34 (3) scales, cheek with 3 or 4 (rarely 2 or 5) rows. Five or 6 (rarely 6|) scales between the upper lateral line and the dorsal fin origin, 6 or 7 between the pectoral and pelvic fin bases. (Because many specimens were damaged during capture by gill-nets, the last two counts are based on only 17 specimens). Fins : Dorsal with 23 (f.i), 24 (f.8), 25 (f.ig) or 26 (.3) rays, comprising 15 (f.2i) or 16 (f.io) spinous and 8 (f.i), 9 (f.i-5) or 10 (f.i5) branched rays. Anal with n (f.6), 12 (f.23) or 13 (f.2) rays, comprising 3 spines and 8 (f.6), 9 (f.23) or 10 (f.2) branched elements. Pelvic fins with the first unbranched ray slightly produced in the adults of both sexes, but proportionately more so in males. Pectoral 24'3~30'0 (M=26-3) per cent of standard length. Caudal truncate or subtruncate, scaled on its basal half to two-thirds. Teeth : The form of the teeth in the outer row of both jaws shows some correlation with standard length. Only bicuspids (with a few tricuspids posterolaterally in the upper jaw) are found in fishes 70-75 mm. S.L. Unicuspids anteriorly, with bicuspids laterally are characteristic of most fishes 84-113 mm. S.L. (and in one exceptional individual 144 mm. S.L.). In most specimens over 112 mm. S.L., only unicuspids occur, but this is also the condition in a few fishes between 99 and 112 mm. S.L. The unicuspids are moderately stout teeth, and are only slightly curved. There are 42-62 (M = 54) outer teeth in the upper jaw. Teeth in the inner rows are relatively large ; tricuspids predominate in fishes of all sizes, but in specimens over 120 mm. S.L. weakly tricuspids are commoner than the distinctly tricuspid ones, and some unicuspid teeth also occur (especially in the outer- most row of the inner series) . There are 3 or 4 inner rows in the upper jaw, and 2 or 3 (less frequently I row) in the lower. Osteology : The neurocranium of H . paraguiarti is identical with that of H. acidens (see p. 75). It differs somewhat from the presumed generalized piscivore skull of H. guiarti, and shows some of the characters found in the more specialized type of H. prognathus. It is, in fact, almost intermediate between the two types. The premaxilla has a less pronounced medial expansion than in H. acidens, but as in that species it has long pedicels (as long as the horizontal dentigerous arms) which meet the horizontal arms at an appreciable angle. In these characters the premaxilla of H, acidens differs from that bone in the generalized species like H. guiarti and H. serranus. The lower pharyngeal bone is slender, its dentigerous surface slightly broader than long. The teeth are fine, slender, compressed and clearly bicuspid, only those in the posterior one or two rows of the median series are enlarged. There are 26-30 (mode 28) rows of lower pharyneal teeth. Vertebral counts for 6 specimens are : 13+17 (f-4) and 14 + 16 (f.2), giving a total of 30. Coloration : live colours unknown. Preserved material : Males (adult and sexually active) : ground colour dark sooty, almost uniformly so except for a lighter A REVISION OF THE LAKE VICTORIA HAPLOCHROMIS SPECIES 71 (dusky gold) patch on the midflank, crossed by four, faint and narrow vertical stripes. Branchiostegal membrane dark (darker below the opercular series than between the jaws). A very faint, near vertical lachrymal stripe runs from the anterior orbital margin to behind the posterior tip of the maxilla. Dorsal fin dusky, lappets black, the soft dorsal maculate. Caudal dark. Anal dusky, but lighter than the dorsal except for a narrow band along the base, and over the spinous part ; two or three greyish ocelli present. Pel vies dusky, the proximal threequarters almost black, the distal quarter lighter. Males (adult but quiescent] : ground colour dusky silver-grey, darkest dorsally ; cheek bright silver. A faint lachrymal stripe present. Dorsal fin greyish-dusky, darkly maculate between the last four spines and all the branched rays. Caudal dark. Anal dark hyaline, with two or three greyish ocelli. Pel vies sooty, darkest on the leading edge. Females (adult) : ground colour silver grey, darker dorsally : cheek bright silver (i.e. like quiet males but lighter and brighter). Dorsal fin dark hyaline, with dark spots on the soft part. Caudal dark hyaline, weakly to distinctly maculate. Anal and pelvic fins hyaline. Females (juvenile] : Two small (75 mm. S.L.) specimens have colours like those of adult females, but with traces of 7 or 8 vertical bars on the flanks and caudal ped- uncle ; these bars reach the dorsal outline but do not extend ventrally below the level of the pectoral fin base. Ecology. Habitat. Most records of H. paraguiarti are from exposed, wave- washed beaches where the substrate is of sand, rock or shingle. There are, however, a few records from more sheltered beaches and areas, but with one exception (a mud bottom) the substrate was hard, and the depth invariably never more than 30 ft. Food. Twenty specimens (from 12 localities) were examined. Of these, 17 con- tained ingested material. One individual contained fish remains (very fragmentary and unidentifiable, even to family) together with a little macerated phanerogam, tissue. Eleven others yielded either macerated plant tissue (f-4), or a mixture of plant tissue and insect fragments (larval Ephemeroptera) ; five specimens contained only fragments of larval Ephemeroptera. Breeding. Little information is available on reproduction in this species. Fishes less than 95 mm. S.L. are immature, as is one larger specimen (a male, 100 mm. S.L.). The largest fish (156 mm. S.L.) is a male. Affinities. Superficially, H. paraguiarti resembles H. guiarti both morphologically and trophically, although H. guiarti apparently includes a greater proportion of fish in its diet. Morphologically the species may be distinguished by the straight dorsal head profile of H. paraguiarti (gently curved in H. guiarti], its longer head (33-0-36-8, M =34-8 per cent standard length cf. 29-5-33-8, M =31-4 per cent), somewhat broader snout, and fewer, coarser outer teeth (42-60, M = 53, cf. 48-74, M = 62 is the upper jaw). Live colours are unknown for H. paraguiarti, but the coloration of preserved, sexually active males is noticeably darker than that of H. guiarti males. Osteological differences, especially in neurocranial shape, are discussed above (p. 70). 7 2 P. H. GREENWOOD Haplochromis paraguiarti also resembles, rather closely, H. acidens ; the species are compared on p. 76. Some specimens of H. prognathus, a rather variable species, resemble H. para- guiarti but there are several differences which serve to distinguish even these superficially similar individuals ; H. paraguiarti has a shorter head (33-0-36-8, M = 34-8 per cent of standard length, cf. 35-5-38-4, M 36-9 per cent), a shallower preorbital (14-5-19-5, M = 16-9 per cent head, cf. 18-8-23-1, M =20-8 per cent), shorter snout (31-0-36-2, M = 33-5 per cent head, cf. 33-4-39-0, M = 37-0 per cent), and a larger eye (23-8-28-6, M = 26-2 per cent head, cf. 20-0-25-0, M = 22-6 per cent). Differences in neurocranial form between the species are also distinctive, although the neurocranium of H. paraguiarti is intermediate between the more generalized H. guiarti type and the relatively elongate skull of H. prognathus (see p. in). When all characters are considered, H. paraguiarti, despite its superficial resem- blance to H . guiarti, is probably more closely related to H. prognathus. It appears to possess the structural characters of the ancestral species or species group from which piscivorous predators like H. prognathus, H. bartoni and H. mandibularis were derived. Outside the piscivorous predator group, it shows very close relation- ship with H. acidens (see p. 76). STUDY MATERIAL AND DISTRIBUTION RECORDS Museum and Reg. No. B.M. (N.H.) 1966.3 B.M. (N.H.) 1966.2 (Holotype) B.M. (N.H.) 1966.3 B.M. (N.H.) 1966.3 B.M. (N.H.) 1966.3 B.M. (B.H.) 1966.3. B.M. (N.H.) 1966.3 B.M. (N.H.) 1966.3 B.M. (N.H.) 1966.3 9-324 21.6 9-338-344 9.313-314 9.320-323 9-333-337 9.317-319 9.312 9-325-330 Locality UGANDA Beach near Jinja Beach near Nasu Point (Buvuma Channel) . Beach near Nasu Point Fisherman's point near Jinja Near Grant Bay Entebbe Harbour Buka Bay Katebo .... Beach near Hannington Bay Collector E.A.F.R.O. E.A.F.R.O. E.A.F.R.O. E.A.F.R.O. E.A.F.R.O. E.A.F.R.O. E.A.F.R.O. E.A.F.R.O. E.A.F.R.O. B.M. (N.H.) 1966.3.9.316 KENYA Kamaringa (Kavirondo Gulf) E.A.F.R.O. B.M. (N.H.) 1966.3.9.311 B.M. (N.H.) 1966.3.9.315 TANZANIA Beach N. of Majita Majita Beach E.A.F.R.O. E.A.F.R.O. B.M. (N.H.) 1966.3.9.331-332 LAKE VICTORIA Locality unknown E.A.F.R.O. A REVISION OF THE LAKE VICTORIA HAPLOCHROMIS SPECIES 73 Haplochromis acidens sp. nov. (Text-fig. 13) Haplochromis percoides (part) : Boulenger, 1915, Cat. Afr. Fish., 3, 296 (specimens from Kakindu, Victoria Nile, B.M. [N.H.] reg. no. 1911.3.3.82-3). Paratilapia serranus (part) : Boulenger, 1915, op. cit., torn, cit., 334 (specimens from Ripon Falls, nr. Jinja, B.M. [N.H.] reg. no. 1911.3.3.20-1). Haplochromis nigrescens (part) : Regan, 1922, Proc. zool. Soc. Londn., 172 (excluding his tentative synonymy of Astatotilapia roberti Pellegrin, and the synonymy of Paratilapia parvidens Blgr., for which see Greenwood, 1959, p. 194). NOTES ON SYNONYMY. It is obvious that Regan's (1922) redescription of H. nigrescens (Pellegrin), 1909 was influenced by Pellegrin's figure of that species (Pellegrin, 1910, Mem. soc. Zool. France, 32, pi. 14, fig. 3). The fish depicted certainly does resemble a specimen of the species here described as H. acidens, but the type specimen of Pellegrin's Astatotilapia nigrescens more closely resembles Haplochromis flavipinnis (personal observations). Boulenger (1915) actually synonymized A . nigrescens Pellegrin with H. percoides Blgr., 1906 but this action was unacceptable to Regan, who resurrected the species as Haplochromis nigrescens. Certainly there are similarities between A . nigrescens type specimen and H. percoides, but my study of the type was revealed several differences (see appendix for a re- description of this specimen and further comments on its affinities). In earlier papers I had accepted Regan's resurrection of H. nigrescens, and used specimens identified by him as a basis for comparing other Lake Victoria species with H. nigrescens ; thus, where H . nigrescens is mentioned in those papers, the name should now be corrected to read H. acidens. Two of the three Lake Victoria specimens examined by Regan undoubtedly can be referred to H. acidens, but I am uncertain about the identity of the third fish. It is the smallest specimen, and because so little is known about small fishes of this and related groups, diagnostic characters applicable to larger individuals cannot be used with confidence. In many respects, this specimen resembles larger individuals of H. prognathus. HOLOTYPE : an adult male, 108 mm. standard length, from Kisumu Kavirondo Gulf (Kenya), B.M. (N.H.) reg. no. 1966.2.21.1. Named acidens from its sharp, needle-like teeth. DESCRIPTION : based on 17 specimens 67-128 mm. S.L., including the holotype and two of the specimens examined by Regan (B.M. [N.H.] reg. no. 1911.3.3.20-1). Depth of body 34-6-40-3 (M 38-1) per cent of standard length, length of head 34-0-36-7 (M = 35-4) per cent. Dorsal head profile gently concave (clearly so in most specimens but weakly concave in a few), sloping fairly steeply ; premaxillary pedicels not prominent. Preorbital depth 15-2-19-2 (M = 17-1) per cent of head, least interorbital width 21-4-28-2 (M = 24-5) per cent. Snout as long as broad or 1-1-1-2 times longer, its length 31-8-37-2 (M =34-2) per cent of head, eye diameter 23-0-29-0 (M =26-4), depth of cheek 23-0-28-2 (M = 25-7) per cent. 74 P. H. GREENWOOD Caudal peduncle 14-8-17-9 (M 16-6) per cent of standard length, 1-1-1-5 (mode 1-3) times as long as deep. Mouth slightly oblique, lips somewhat thickened, premaxilla with a slight median expansion. Jaws equal anteriorly, the lower 44-0-50-2 (M = 46-9) per cent of head, 1-9-2-4 (rare), mode 2-1, times as long as broad. Posterior tip of the maxilla reaching the vertical through the anterior orbital margin, occasionally not quite reaching this point. FIG. 13. Haplochromis acidens. Drawn by Barbara Williams. Gill rakers : moderately slender to slender, the lower 1-4 reduced, the upper 3 or 4 flattened and lobed in some fishes ; 10 (rarely 9 or n) on the lower part of the first gill arch. Scales : ctenoid ; lateral line with 31 (f.i), 32 (f.8), 33 (f-7) or 34 (f.i), cheek with 3 (rarely 4) rows. Five and a half to 7 (mode 6) scales between the dorsal fin origin and the upper lateral line, 5 or 6 (mode), rarely 6| or 7 between the pectoral and pelvic fin bases. Fins : Dorsal with 23 (f.i), 24 (f.7) or 25 (f.g) rays, comprising 15 (f.12) or 16 (f.5) spinous and 7 (f.i), 9 (f.n) or 10 (f.5) branched rays. Anal with n (f.7) or 12 (f.io) rays comprising 3 spines and 8 or 9 branched elements. Pectoral 22-7-28-0 (M = 24-9) per cent of standard length. Pelvics with the first two branched rays slightly produced, relatively more so in adult males. Caudal truncate to subtruncate, scaled on its proximal half to two-thirds. Teeth. In the outer row of both jaws the teeth are tall, slender and slightly curved. Fishes between 67 and 90 mm. S.L. have unicuspid teeth anteriorly, and weakly bicuspid teeth laterally and posterolaterally. Larger fishes have only unicuspids in the outer row. The anterior teeth in the upper jaw of the largest specimens may be relatively enlarged (i.e. stouter and longer than the lateral teeth). There are 40-64 (M = 48) teeth in the outer, upper row. A REVISION OF THE LAKE VICTORIA HAPLOCHROMIS SPECIES 75 The inner rows of the upper jaw are composed of tricuspid teeth in fishes < 90 mm. S.L. and a mixture of uni- and tricuspids in larger fishes ; there is an increased proportion of unicuspids in specimens > 115 mm. S.L. In fishes over TOO mm. S.L. the outermost row of the inner series is often composed of unicuspids and the re- maining rows of tricuspids. The inner rows are widely spaced in most fishes so that the teeth form a broad band anteriorly and anterolaterally. There are 3 or 4 rows of teeth in the upper jaw. In the lower jaw of most specimens less than 120 mm. S.L., only tricuspid teeth are found, but some unicuspids also occur in fishes of this size group. Larger in- dividuals have a mixture of tri- and unicuspid or tri- and weakly tricuspid teeth ; only rarely are all the lower, inner teeth unicuspid. There are 2 (mode) or 3 rows of inner teeth in this jaw. Osteology. The neurocranium of H. acidens departs from that of H, guiarti towards the H. mento type (see Greenwood, 1962) . The slope of its anterior dorsal profile (from vomer tip to the origin of the supraoccipital crest) is less steep than in H. guiarti, and the maximum width (across the otic region) is relatively less, as is its greatest depth (exclusive of the supraoccipital crest). In all these characters, H. acidens is intermediate between H. guiarti and H. prognathus, a species whose neurocranium could provide a basic " bauplan " from which the more extreme " mento "-type evolved. The premaxilla is somewhat beaked, with the dentigerous surface expanded and protracted medially ; the pedicels are elongate, being almost as long as the horizontal arms of the bone. The lower pharyngeal bone is fine, and has the dentigerous surface broader than long. Lower pharyngeal teeth are arranged in 24-28 rows, and are slender, com- pressed and distinctly cuspidate, with those in the two median rows somewhat coarser than the others. Vertebral counts from 12 specimens are : 13+16 (f-4), 13 + 17 (f-7) and 14 + 16 (f.i), giving totals of 29 and 30. Coloration. Live colours : Males (sexually active) have a slatey-blue ground colour, darkest dorsally. Dorsal fin smokey-grey with a pale red margin and a deep red flush on the soft part. Caudal dark grey with deep red streaks between the rays. Anal dark hyaline with a dull maroon flush, and orange-red ocelli. Pel vies black. Females (adult and quiescent) : dorsal surface yellowish-grey shading to silver on the flanks and belly. Dorsal fin hyaline but with a yellowish base. Anal and caudal dark hyaline. Pelvics hyaline. Preserved material : Males (adult and sexually active) : ground colour brown (darker than in females and juveniles), darkest dorsally and on chest, belly and ventral half of the caudal peduncle ; six to eight vertical bands cross the flanks and caudal peduncle, each band broadening slightly below the level of the upper lateral line, and merging ventrally with the dark body coloration, but remaining discrete dorsally. A narrow, dark lachrymal stripe runs from the anterior border of the eye, sloping backwards to pass behind the posterior tip of the maxilla ; a dark narrow stripe outlines the vertical limb of the preoperculum. The spinous dorsal is greyish, the soft part darker on its basal two-thirds, maculate over the posterior half. Caudal 76 P. H. GREENWOOD maculate on its proximal half to two-thirds. Pelvics dusky, particularly on the basal half ; membrane covering the spine colourless. Anal greyish to hyaline, with two, ill-defined and dark grey to dusky ocelli. Males (sexually quiescent) : as above but all stripes and bars are much fainter and less well-defined, and the ventral body surface is not noticeably darker than the dorsum, although still darker than the mid-flank region. Only the basal half of each pelvic fin is dusky, and the anal ocelli are almost invisible. Immature fishes have the same coloration as females. Ecology. Habitat. The majority of specimens is from a sheltered habitat in the Kavirondo Gulf near Kisumu. At this place the water was about 10 ft. deep, and the substrate of mud ; fairly dense stands of submerged plants were common in the area. The other specimens are recorded as being from " Jinja, Ripon Falls ". I am presuming that this locality is in the Napoleon Gulf, that is, above the falls. The area, before the submergence of the Falls, was sheltered, with a hard substrate and, at least close inshore, fairly dense plant stands. Food. All sixteen specimens examined came from the Kisumu locality, but were caught on different occasions ; fifteen specimens contained ingested material in the stomach and intestines. Every gut had, as its major content, finely macerated phanerogam tissue and varying amounts of epiphytic diatoms. Some specimens also contained a flocculent, grey-green mass (as seen in preserved material) thought to be bottom debris, principally blue-green algae. In addition, one gut yielded a few fish bones, one a number of insect eggs (apparently undigested), one a mass of tissue (thought to be the foot of a gastropod) and some fragments of larval Ephemeroptera, and another, fragments of unidentifiable insects. The phanerogam tissue is very finely divided, and a larger proportion seems to be digested than is the case in other phytophagous species examined. The apparently vegetarian diet of H. acidens is at variance with the dentition and general facies which are essentially those of a piscivorous predator. The length of the gut (i \ to 2 times standard length) is greater than in the piscivores, but is not as long as the gut in purely phytophagous species. Breeding. Specimens < 90 mm. S.L. are immature ; both sexes reach the same maximum adult size. Affinities. In its general appearance, H. acidens resembles a number of the deeper- bodied piscivorous species, and one of the specialized predators on larval and embryo fishes, H. parvidens (Greenwood, 1959). Perhaps the greatest resemblance is between H. acidens and H. paraguiarti. Both species have similar neurocrania and jaw structure, but H. acidens has finer teeth, a deeper body (34-6-40-3, M = 38-1 per cent of standard length, cf. 31-7-36-6, M = 34-8%) and a concave as opposed to a straight dorsal head profile. Also showing an overall similarity with H. acidens is H. maculipinna. The con- cavity of the dorsal head profile in H. acidens again serves as one differentiating character. Others are its less oblique jaws, fewer (40-64, M = 48 cf. 50-80, M =62), more slender and longer teeth, and its larger chest scales (5-6 between pectoral and pelvic fin bases, cf. 6-9, mode 7 or 8, in H . maculipinna) . The neuro- crania of the two species also differ, that of H . maculipinna being deeper and having A REVISION OF THE LAKE VICTORIA HAPLOCHROMIS SPECIES 77 a steeper dorsal profile ; the premaxilla in this species lacks a pronounced median expansion, and the pedicels meet the horizontal arms almost at right angles (and not at the more acute angle found in H. acidens). From other species of the H . serranus group (H. victorianus, H. serranus, H. spekii and H. nyanzae], H. acidens is distinguished by its concave dorsal head profile, smaller eye, finer, longer and fewer teeth (at least as shown by the mean number, the ranges overlap), less oblique and non-prognathous lower jaw, and by the absence of a prominent mental bump at the symphysis of the dentaries. Other differences also serve to distinguish H. acidens from individual species of the H. serranus complex ; for these see the species descriptions in Greenwood (1962) and p. 32 above for H. spekii. At about the same level of similarity with H. acidens is H. prognathus. This species differs from H. acidens chiefly in its shallower body (30-1-37-1, M = 33-3% of standard length, cf. 34-6-40-3, M=38-i%), deeper preorbital (18-8-23-1, M = 20-8% of head, cf. 15-2-19-2, M = 17-1%) and smaller eye (20-0-25-0, M = 22-6% of head, cf. 23-0-29-0, M = 26-4%). The dentition and neurocrania of the species differ quite markedly, with H. prognathus having a skull nearer the H. mento type (see Greenwood, 1962, and p. 81) and stouter teeth. It is difficult to assess the phyletic position of H. acidens. The level of anatomical specialization, especially of the syncranium and dentition, suggests affinity both with the H. serranus species group and with H. prognathus, a species not far removed from that group but probably representative of the ancestral type from which such specialized forms as H. mento and H. macrognathus evolved. Its plant diet suggests a possible trophic specialization paralleling that achieved by a species (H. phyto- phagus) related to the small, anatomically generalized, and usually insectivorous species (Greenwood, 1966). In an earlier paper (Greenwood, 1959), I suggested that H. parvidens might have evolved from a species anatomically like H. acidens (called H. nigrescens in that paper) ; this relationship is discussed further on page 114. Phyletically, H. acidens was probably derived from the same stem as H. para- guiarti ; this stem could have been related to either an H. guiarti-\ike lineage, or to an H. serranus-like one. STUDY MATERIAL AND DISTRIBUTION RECORDS Museum and Reg. No. Locality Collector UGANDA B.M. (N.H.) 1911.3.3.20-21 Jinja, Ripon Falls . . . Bayon KENYA B.M. (N.H.) 1966.3.9.107-121 Kavirondo Gulf, near Kisumu . E.A.F.R.O. B.M. (N.H.) 1966.2.21.1 Kavirondo Gulf, near Kisumu . E.A.F.R.O. (Holotype) ZOOL. 15, 2. 78 P. H. GREENWOOD Haplochrotnis prognathus (Pellegrin), 1904 (Text-figs. 14 and 15) Paratilapia prognatha Pellegrin, 1904 (probably in part ; the identity of the three smallest syntypes is still in doubt), Bull. Soc. Zool. France, 29, 185 ; Idem, 1905, Mem. Soc. Zool. France, 17, 181, pi. 16, fig. 4. Boulenger, 1915, Cat. Afr. Fish., 3, 333, fig. 224 (in part, the syntype B.M. [N.H.] reg. no. 1905.2.28.1, ex Kavirondo Gulf). Pelmatochromis spekii (part) : Boulenger, 1915, Cat. Afr. Fish., 3, 417. Haplochromis prognathus: Regan, 1922, Proc. zool. Soc. London, 177. Haplochromis taeniatus Regan, 1922, op. cit., 170, text-fig. 3. Haplochromis macrodon Regan, 1922, op. cit., 176, text-fig. 4. Haplochromis lamprogenys Fowler, 1936 Proc. Acad. Nat. Sci. Philad, 88, 330, fig 137. Haplochromis steindachneri Lohberger, 1929. Anz. Akad. Wiss. Wein, no. 17, 207. Haplochromis versluysi Lohberger, 1929, Anz. Akad. Wiss. Wein no. 17, 206. Haplochromis rebeli Lohberger, 1929, Anz. Akad. Wiss. Wein., no. n, 94 (probably in part; the identity of one paratype, a female 116 mm. S.L., reg. no. 18768 of the Vienna Museum, is still doubtful). LECTOTYPE : a fish 136-0 mm. standard length, B.M. (N.H.) reg. no. 1905.2.28.1 (presented by the Paris Museum), collected by C. Alluaud from Kavirondo Bay, Kenya. This specimen appears to be the fish figured by Pellegrin (1905), and is chosen as lectotype principally for that reason. Note : Defining this species has proved particularly difficult, mainly because of the condsiderable intraspecific variability in head shape (see text-fig. 15). This variability, although apparent to the eye, is not readily quantifiable. Intergrades exist between the most outstanding variants, and now that a large series of specimens is available it appears that the variation is, to a large extent, size correlated. In fishes less than no mm. S.L., the snout seems protracted and the dorsal head profile has a marked and extended concavity above the eye. In larger fishes the elongate " face " is less noticeable, and the dorsal profile is straighter and slopes steeply (but still with a slight supraorbital concavity). DESCRIPTION : based on 43 specimens, 70-141 mm. S.L., and including the lectotype, the syntypes of H. taeniatus and H. macrodon and the holotypes of H . steindachneri and H. lamprogenys. The types of H. rebeli and H. versluysi were examined but are not included in this redescription. Depth of body 30-1-37-1 (M = 33*3) per cent of standard length, length of head 35-3-38-4 (M = 36-9) per cent ; head profile variable (see note above), usually with a supraorbital concavity, the dorsal outline sloping at an angle of 30-40 with the horizontal. Preorbital depth 18-8-23-1 (M = 20-8) per cent of head, least interorbital width 18-1-23-7 (M = 21-0) per cent. Snout length 33-4-39-6 (M = 37'0) per cent of head, i to i^ times as long as broad (rarely if times). Eye diameter with slight negative allometry, 20-0-25-0 (M = 22-6) per cent of head ; in fishes 70-90 mm. S.L., the eye is clearly larger than the cheek is deep but in specimens 90-110 mm. it is equal to or slightly smaller than the cheek, a relationship that holds for some larger in- dividuals (110-125 mm. S.L.) although in most individuals in this size range the eye diameter is manifestly less than the cheek depth. Depth of cheek 20-6-28-6 (M = 24-5) per cent of head (18-9% in the smallest specimen). A REVISION OF THE LAKE VICTORIA HAPLOCHROMIS SPECIES 79 1-1 . bo II > -1- M ei 8o P. H. GREENWOOD 1 CM FIG. 15. Haplochromis prognathus, showing variability of head profile. Caudal peduncle 14-6-19-0 (M = 16-0) per cent of standard length, 1-2 (rarely)-i-7 times is long as deep (modal range 1-3-1-5). Mouth moderately oblique, the jaws equal anteriorly or the lower projecting slightly ; lower jaw with a slight mental protuberance, its length 42-5-51-1 (M = 46-2) per cent of head, 2-0-2-8 (mode 2-4) times as long as broad. Posterior tip of the maxilla not reaching the vertical through the anterior orbital margin (except in one specimen), but usually reaching a point nearer this line than one through the nostril. Gill rakers : 8-10 (mode 9) on the lower part of the first gill arch, the lower 1-3 rakers reduced, the upper 3 or 4 often flat or flat and lobed, the remainder slender. Scales : ctenoid. Lateral line with 30 (f.2), 31 (f.12), 32 (f.2i) or 33 (f.8) scales ; some individuals lack pores in the scales of this series. Cheek with 3 (occasionally 4) rows. Five to 7 (mode 6) scales between the dorsal fin origin and the upper lateral line, 5-7 (mode 6) between the pectoral and pelvic fin bases. Fins. Dorsal with 23 (f.6), 24 (f.26) or 25 (f.io) rays, comprising 14 (f.n), 15 (f.28) or 16 (f.3) spines and 8 (f.i), 9 (f.28) or 10 (f.i3) branched rays. Anal with n (f.2), 12 (f.2i) or 13 (f.n) rays, comprising 3 spines and 8 (f.2), 9 (f.2i) or 10 (f.n) branched elements. Pectoral fin shorter than head, 22-0-29-3 (M = 25-7) per cent of standard length . First pelvic branched ray slightly produced in both sexes but proportionately more so in males. Caudal truncate or subtruncate, scaled on its basal half to two- thirds. Teeth. In the smallest fish examined (70 mm. S.L.) the outer teeth in both jaws are mostly unequally bicuspid, but a few unicuspids are present. Fishes between this size and 105 mm. S.L. have mostly unicuspids in the outer row, but some bi- cuspids occur posterolaterally in the upper jaw. Fishes > 105 mm. have only uni- cuspid outer teeth, a condition found occasionally in individuals as small as 90 mm. S.L. The unicuspids are slightly curved inwards, and vary in form from relatively slender to moderately stout. There are 30-60 (M = 45) teeth in the outer row of the upper jaw, the number not showing any correlation with the fish's size. Regan's (1922) " key " character separating H. taeniatus from H. macrodon (outer teeth numerous, close together cf. outer teeth rather strong, set well apart) is hardly A REVISION OF THE LAKE VICTORIA HAPLOCHROMIS SPECIES 81 trenchant even when only the type specimens are compared, and is inapplicable when a large series is examined. The inner teeth are mostly tricuspids in fishes < no mm. S.L., but an admixture of tri- and unicuspids (or weakly tricuspids) is found in fishes at the upper end of this size range. Some tricuspids also occur in large fishes, but in the majority of these all inner rows are composed of unicuspids. When tri- and unicuspids occur together in fishes > 120 mm. S.L. the tricuspids are confined to the innermost rows. Inner teeth are arranged in 2 (mode) or 3, rarely I, rows in the upper jaw, and in I (mode) or 2 (rarely 3) rows in the lower. Osteology. The neurocranium of H. prognathus is approximately intermediate between that of H. paraguiarti and H. mento (see p. 70, and Greenwood, 1962). It thus closely resembles the neurocrania of H. bartoni, H. mandibularis , H. argenteus and H. longirostris (see p. 112 and Greenwood, op. cit), in Lake Victoria, and H. venator in Lake Nabugabo (Greenwood, 1965). From the four former species it differs most noticeably in its deeper supraoccipital crest, and from H. argenteus also by its being somewhat broader anterior to the orbits. There are also resemblances in neuro- cranial form between H. prognathus and H. xenostoma (see p. 53) but the differences here are somewhat more pronounced than in the other species mentioned. The premaxilla has an expanded median dentigerous surface which gives the bone a beaked appearance ; this is a feature of those piscivorous species which I consider to be more specialized in body-form than the members of the H. guiarti and H. serranus species complexes (see Greenwood, 1962). In contrast, the dentary of H. prognathus is of the stout, deep-bodied type found in the latter groups. The lower pharyngeal bone is triangular, with the dentigerous surface as long as broad or slightly broader (both conditions are of equal frequency). The lower pharyngeal teeth are fairly coarse, with cylindrical necks and compressed, weakly bicuspid crowns ; these teeth are well-spaced on the bone in 18-24 rows. Vertebral counts in 8 specimens are : 12+17 (f.2), 13 + 15 (f.i), 13 -f 16 (.4) and 13 + 17 (f.i) giving totals of 28-30. Coloration. Live colours are unknown. Preserved material : Males (adult and sexually active). Body and head brownish above the midline, silvery-yellow below ; two faint lateral bands are visible, one running midlaterally from behind the eye to the origin of the caudal fin, the other along the upper lateral line. There is a faint, ill-defined but broad lachrymal band, but it does not extend ventrally to the margin of the preorbital. Dorsal fin greyish, the dark pigment most intense basally on the spinous region, lappets dusky ; soft dorsal maculate. Caudal dark grey, darkest between the rays on the proximal two-thirds of the fin. Anal greyish, lappets black, as may be the basal part of the membrane between the spines ; 2 or 3 large but faint, greyish ocelli. Pel vies black. Females (adult and juvenile) . Light grey-brown above, silvery below ; very faint traces of longitudinal bands as in males, the lower often visible on the posterior third of the body. All fins hyaline, but the caudal is dark on its proximal two- thirds. Ecology. Habitat. Haplochromis prognathus is apparently confined to water less than 20 ft. deep, and to hard substrates ; it occurs in both sheltered and exposed 82 P. H. GREENWOOD localities, including wave-washed beaches. Food. Of the 33 specimens examined (from n localities), 22 contained food. Ten specimens contained only remains of fishes (predominantly small Cyprinidae, but Haplochromis were also identified), 3 yielded fish and insect remains (the fish unidentifiable, the insects adult Isoptera and larval Ephemeroptera) , 7 contained only insects (adult Isoptera and larval Ephemeroptera), i contained unidentifiable fish remains and some plant fragments (undigested), and 2 specimens contained unidentifiable material (? bottom debris) and some plant remains. Breeding. Little is known about the breeding habits of this species ; no brooding individuals have been recorded. Sexual maturity is attained at a length of about 100 mm., but larger juvenile individuals (up to 120 mm.) are known. Both sexes reach the same maximum adult size. Affinities. There is nothing particularly outstanding about the general appearance of H. prognathus when it is compared with the other relatively deep-bodied predatory species of Lake Victoria. Thus, it bears a superficial resemblance to several pisci- vorous species, especially H. serranus, H. bartoni, H. nyanzae and H. mandibularis . From H. serranus, H. prognathus differs in its slightly shallower body (30-1-37-1, M = 33'3% standard length cf. 32-7-39-2, M = 36-0%), deeper preorbital (18-8-23-1, M = 20-8% head, cf. 14-6-20-0, M = 17-7%), slightly longer snout (33-4-39-0, M=37-o% head, cf. 30-8-37-0, M=34'O%), shorter lower jaw (42-5-51-1, M = 46-2% head, cf. 47-0-60-0, M = 54-3%) and slightly larger chest scales (5-7, mode 6, between the pelvic and pectoral fin bases, cf. 7 or 8 [rarely 6]). The pre- served coloration of the two species suggests a fairly marked difference in live colours, and there are clear-cut differences in neurocranial form (see p. no). From H. bartoni (probably its nearest extant relative), H. prognathus differs in having a broader and shorter lower jaw (42-5-51-1, M = 46-2% head, cf. 50-8-57-0, M = 52-5%), fewer outer teeth in the upper jaw (30-60, M = 45, cf. 50-80, M = 62), the maxilla not extending so far posteriorly (reaching or almost reaching the orbit in H. bartoni), and its smaller chest scales. Neurocranial form in the two species is similar. The characters separating H. prognathus from H. nyanzae are its longer snout (33'4-39' 6 M =37'0% head, cf. 33'4-35'8, M =34-5%). more oblique lower jaw, its maxilla not reaching the orbit, and its larger chest scales. Unfortunately it has not been possible to compare directly the neurocrania of the two species. But, from radiographs it seems probable that the skull of H. nyanzae is like that of H. serranus (see Greenwood, 1962). Haplochromis prognathus differs from H. mandibularis in having a broader snout, shorter lower jaw (42-5-51-1, M=46-2% head, cf. 47-3-56-8, M=5i-5%), and longer caudal peduncle (14-6-19-0, M = 16-6% standard length, cf. 12-2-15-2, M = 14-2%). Neurocranial form in the two species is similar. Haplochromis prognathus is, apparently, closely related to H. argenteus. At first sight, the great variability in head shape of H. prognathus obscures the relationship. But, if smaller specimens of H. prognathus are compared with larger individuals of H. argenteus the likeness is striking. Morphometrically, even superficially dissimilar specimens of the two species are not readily separable. Characteristics of the lower A REVISION OF THE LAKE VICTORIA HAPLOCHROMIS SPECIES 83 jaw serve to distinguish the species. The lower jaw of H. prognathus is broader (length/breadth ratio 2-0-2-8 [modal range 2-0-2-4], c f- 2 '3~3' 1 [modal range 2-8- 3-0]), shorter (42-5-51-1, M = 46-2% of head, cf. 45-0-59-0, M 50-2%), less oblique and less prognathous. The possible phyletic relationship of H. prognathus within the Lake Victoria species-flock will be discussed later (p. no) ; outside Lake Victoria, H. prognathus is probably related to H. venator of Lake Nabugabo (Greenwood, 1965). Indeed H. venator could well have been derived from populations of H. prognathus cut off when the sand bar which isolated Lake Nabugabo was formed about 4,000 years ago. STUDY MATERIAL AND DISTRIBUTION RECORDS Museum and Reg. No. B.M. (N.H.) 1906.5.30.304 (Syntype H. macrodon) B.M. (N.H.) 1906.5.30.305-6 (Syntypes H . macrodon) B.M. (N.H.) 1906.5.30.250 (Syntype H. taeniatus) B.M. (N.H.) 1966.3.9.50 B.M. (N.H.) 1966.3.9.64-65 B.M. (N.H.) I966-3-9.53-59, 66-67, 7 2 ~75 B.M. (N.H.) 1966.3.9.68-71 B.M. (N.H.) 1966.3.9.79-83 B.M. (N.H.) 1966.3.9.51-52 B.M. (N.H.) 1966.3.9.60-63 Vienna Museum 17872 (Type of H . steindachneri) Academy of Nat. Sciences Philadelphia ANSP 66131 (Holotype of H . lamprogenys) B.M. (N.H.) 1905.2.28.1 (Lectotype) B.M. (N.H.) 1905.12.11.4 (Syntype H. taeniatus) B.M. (N.H.) 1966.3.9.76-78 B.M. (N.H.) 1966.3.9.84 B.M. (N.H.) 1906.5.30.251-2, 256 Vienna Museum 17876-7 (Types of H. versluysi) Vienna Museum 18830 (Holotype of H. rebeli) Vienna Museum 18769-71 (Paratypes of H. rebeli) Locality UGANDA Entebbe . Bugonga Entebbe . Entebbe, Airport Beach Napoleon Gulf, near Jinja . Beach near Nasu Point (Buvuma Channel) . Katebo . Bugonga, near Entebbe Kagera Bay, near mouth of Kagera River Bukafu Bay Sesse Islands Kitala KEYNA Kavirondo Gulf Kavirondo Gulf TANZANIA Mwanza Harbour Capri Bay, Mwanza . LAKE VICTORIA Locality unknown Locality unknown Locality unknown Locality unknown Collector Degen Degen Degen E.A.F.R.O. E.A.F.R.O. E.A.F.R.O. E.A.F.R.O. E.A.F.R.O. E.A.F.R.O. E.A.F.R.O. Rolle Alluaud Alluaud E.A.F.R.O. E.A.F.R.O. Degen Rolle Rolle Rolle 84 P. H. GREENWOOD Haplochromis argenteus Regan, 1922 (Text-fig. 16) Paratilapia longirostris (part): Boulenger, 1915, Cat. Afr. Fish., 3, 332. Haplochromis argenteus Regan, 1922, Proc. zool. Soc. Londn., 186, fig. 12. LECTOTYPE : a juvenile female 114-0 mm. standard length, from Bunjako, Uganda (B.M. [N.H.] reg. no. 1906.5.30.266 ; the specimen figured by Regan). FIG. 1 6. Haplochromis argenteus ; the lectotype, a juvenile, about -75 times natural size. From Regan, Proc. zool. Soc. DESCRIPTION : based on 23 specimens (including the lecto- and paralectotype) 93-0-202-0 mm. standard length. Depth of body 26-8-34-8 (M = 31-2) per cent of standard length, length of head 35-5-39-6 (M = 36-8) per cent. Dorsal head profile slightly concave, with prominent premaxillary pedicels, and sloping at an angle of 20-25. Preorbital depth 19-6-24-0 (M = 21-7) per cent of head, least interorbital width 17-7-22-3 (M = 20-0) per cent. Snout 1-25-1-40 times as long as broad, its length 34-8-41 '6 (M = 37*8) per cent of head, diameter of eye 19-4-23-5 (M = 21-5), depth of cheek 22-3-28-7 (M = 24-7) per cent. Caudal peduncle I3*5-I7'7 (M = 16-1) per cent of standard length, 1-3-1-7 (modal range 1-3-1-5) times as long as deep. Mouth oblique (35-40), the lower jaw projecting moderately in some specimens, its length 45-0-59-0 (M = 50-2) per cent of head, 2-3-3-1 (modal range 2-8-3-0) times as long as broad. Lips slightly thickened. Premaxilla expanded slightly in the midline. Posterior tip of the maxilla reaching a point about midway between the orbit and the nostril, or almost reaching the orbit. Gill rakers : generally slender but moderately stout in a few fishes ; lower i or 2 reduced, the upper 3 or 4 (in one fish, the upper 6) flat and lobed. Eight to 10 (mode 9) on the lower part of the first arch. Scales : ctenoid ; lateral line with 29 (f.2), 30 (1.2), 31 (f-5), 32 (f.n), 33 (f.i), 34 (f.i) or 35 (f.i) scales. Cheek with 3 or 4 (rarely 5) rows. Five or 6 (rarely 7) scales between the dorsal fin origin and the upper lateral line, 6 or 7 (less frequently 5) between the pectoral and pelvic fin bases. A REVISION OF THE LAKE VICTORIA HAPLOCHROMIS SPECIES 85 FIG. i6A. Haplochromis argenteus, adult. Drawn by Lavinia Beard. Fins. Dorsal with 23 (f.2), 24 (f.i2) or 25 (f.g) rays, comprising 13 (f.i), 14 (f.i), 15 (f.ig) or 16 (f.2) spines and 8 (f.2), 9 (f.i2) or 10 (f.g) branched rays. Anal with II (f.3), 12 (f.ig) or 13 (f.i) rays, comprising 3 spines and 8, 9 or 10 rays. Pectoral 24-1-29-7 (M = 25-0) per cent of standard length. Pelvics with the first branched ray somewhat elongate, proportionately more so in adult males. Caudal truncate. Teeth. Except for the smallest fish examined (the paralectotype 93 mm. S.L.) the outer teeth in both jaws are unicuspid, slender and slightly curved (those situated posterolaterally in the upper j aw more so than the others) . In the smallest individual most outer teeth are like those described above but there are also some weakly bicuspid and slender teeth. There are 44-80 (M = 58) teeth in the outer row of the upper jaw, the number showing slight positive correlation with standard length. In all except four specimens (the two types, 93 and 114 mm. S.L., and two others 88 and 112 mm. S.L.) the inner teeth are unicuspid. Of the exceptional specimens, the largest (lectotype) has an admixture of tri- and unicuspids, the 93 mm. fish (paralectotype) only tricuspids, while the 88 and 112 mm. individuals have tricuspids in the upper jaw and a mixture of tri- and some unicuspids in the lower. Inner teeth are arranged in 2-4 (rarely 5) rows in the upper jaw, and 1-3 in the lower. One exceptional fish (182 mm. S.L.) has a single, irregular row of widely separated teeth in both jaws. Osteology. The neurocranium of H. argenteus is virtually identical with that of H. longirostris (see Greenwood, 1962), and also resembles the skull of H. prognathus. The premaxilla is more beaked than that of H. longirostris. The lower pharyngeal bone is narrow, but its dentigerous surface is generally broader than long ; less frequently it is equilateral. The teeth are rather coarse, although they are finer and more compressed in a few fishes. These teeth are arranged in 20-24 (usually 22) rows. 86 P. H. GREENWOOD Vertebral counts in 8 specimens are : 12+17 (f.3) and 13 -f- 16 (f.5) giving a total of 29. Coloration. Live colours are unknown. Preserved material : Males (adult and sexually active) : ground colour grey-silver, darker (almost brown) on back and dorsal head surfaces ; tip of lower jaw and anterior part of each ramus dusky. Branchiostegal membrane sooty in the opercular region, dark (but not black) anteriorly. Sub- and interopercula with a golden-yellow flush. An ill-defined to distinct lachrymal stripe runs from the anterior orbital margin to behind the posterior tip of the maxilla. Dorsal fin greyish-yellow, with black lappets, and in some specimens a dusky, irregular banding or marbling on the entire spinous and anterior soft parts. Caudal greyish-yellow, dark grey-brown on the proximal half. Anal greyish with black lappets, and 2 or 3 dead- white ocelli. Pelvics dusky. Males (sexually quiescent, and immature) : Ground colours like those of active individuals but somewhat lighter, often with very faint traces of 4 or 5 bars mid- laterally on the flanks ; the lachrymal stripe is of variable intensity. Dorsal fin as above or without the dark marbling. Other fins also as above, but the anal ocelli may be weakly defined and small, and the pelvics lighter. Females (adult and immature). Ground coloration as in males, but with a faint, interrupted and narrow midlateral band running from the posterior opercular margin to the basal part of the caudal fin. Even fainter traces of a more dorsal longitudinal band (following the course of the upper lateral line) are seen in some specimens ; the two bands may be linked by 3 or 4 extremely faint, short and narrow vertical bars. All fins are yellowish, the caudal dark basally. Ecology. Habitat. Most specimens are from sheltered localities where the bottom is composed of organic mud, and at depths of less than 40 ft. Other localities are, however, more exposed and the substrate is of rock, sand or shingle, but only a few (and juvenile) fishes were obtained from nets operated over exposed beaches. Food. Of the 20 individuals examined (from n localities) 8 had food in the gut. Four fishes contained only insect remains (in 2 thought to be terrestrial species, and in the others identified as larval Ephemeroptera) , and 4 had fragmentary fish remains (identified as the cyprinid Engraulicypris argenteus) . Breeding. Fishes < 114 mm. S.L. are immature, as is one exceptional specimen of 146 mm. Both sexes reach the same maximum adult size. Affinities. The species most like H. argenteus is H. longirostris. There are, however, a number of morphological differences which serve to separate them, in- cluding the larger adult size reached by H. argenteus. From H. longirostris, H. argenteus also differs in its less oblique and longer lower jaw (45*0-59-0, M = 50-2% of head, cf. 42-2-51-4, M =46-0%), shorter and deeper caudal peduncle (13-5-17-7, M = 16-1% of standard length, cf. 17-2-22-2, M = 19-2% ; length/depth ratio I '3~ 1 '7 [modal range 1-3-1-5], cf. 1-7-2-3 [modal range 1-9-2-0]), somewhat deeper body (26-8-34-8, M = 31-2% of standard length, cf. 24-6-30-4, M = 27-2%) and slightly longer head (35-5-39-6, M = 36-8% standard length, cf. 29-2-36-2, M = 33'0%) ; also, the premaxilla of H. argenteus is somewhat more beaked. In an earlier paper (Greenwood, 1962) I used the ratio of eye diameter to interorbital A REVISION OF THE LAKE VICTORIA HAPLOCHROMIS SPECIES 87 width as a diagnostic character ; more material of H. argenteus has shown, however, that this difference is too slight to be of value. The similarities between H. argenteus and H. prognathus are discussed elsewhere (p. 82). When these two species and H. longirostris are considered together, it seems probable that H. argenteus and H. longirostris were derived from an H. prognathus-like ancestor, probably as distinct lines developing almost in parallel. STUDY MATERIAL AND DISTRIBUTION RECORDS Museum and Reg. No. B.M. (N.H.) 1906.5.30.266 (Lectotype) B.M. (N.H.) 1906.5.30.267 (Paralectotype) B.M. (N.H.) 1966.3.9.194-195 B.M. (N.H.) 1966.3.9.201-202 B.M. (N.H.) 1966.3.9.193 B.M. (N.H.) 1966.3.9.196-198 B.M. (N.H.) 1966.3.9.192 B.M. (N.H.) 1966.3.9.208-210 B.M. (N.H.) 1966.3.9.191 B.M. (N.H.) 1966.3.9.199-200 B.M. (N.H.) 1966.3.9.203-204 B.M. (N.H.) 1966.3.9.207 B.M. (N.H.) 1966.3.9.205-206 B.M. (N.H.) 1901.6.24.90 B.M. (N.H.) 1966.3.9.190 Locality UGANDA Bunjako .... Bunjako .... Ekunu Bay Pilkington Bay Beach near Nasu Point Ramafuta Island (Buvuma Channel) Off S. end of Buvuma Island Beach near Grant Bay (Buvuma Channel) . Channel between Yempita and Busiri Islands Kazima Island Off Entebbe Harbour Entebbe Harbour Bukakata Bay . LAKE VICTORIA Locality unknown Locality unknown Collector Degen Degen E.A.F.R.O. E.A.F.R.O. E.A.F.R.O. E.A.F.R.O. E.A.F.R.O. E.A.F.R.O. E.A.F.R.O. Uganda Fisheries Dept. E.A.F.R.O. E.A.F.R.O. E.A.F.R.O. Sir. H. Johnston E.A.F.R.O. Haplochromis squamulatus Regan, 1922 (Text-fig. 17) Paratilapia pectoralis (non Ctenochromis pectoralis Pfeffer) Boulenger, 1911, Ann. Mus. Genova (3), 5, 66, pi. i, fig. 2. Idem, 1915, Cat. Afr. Fish, 3, 339, fig. 229. Paratilapia serranus (part) : Boulenger, 1915, op.cit., torn, cit., 334. Haplochromis guiarti (part): Regan, 1922, Proc. zool. Soc. Londn., 174. Haplochromis squamulatus Regan, 1922, op. cit., 175 (nom. nov. for Paratilapia pectoralis Blgr., 1911). HOLOTYPE : a male (probably adult), 149 mm. standard length, from Jinja (Uganda). Genoa Museum reg. no. C.E. 12977. DESCRIPTION : based on 54 specimens (including the holotype), 66-198 mm. S.L. Depth of body 23-5-35-1 (M = 30-3) per cent of standard length, length of head 29.4-39.5 (M == 34-3) per cent. Dorsal head profile moderately to strongly sloping, 88 P. H. GREENWOOD the premaxillary pedicels prominent and breaking the outline of the profile, snout region sloping at an angle of 40-5o with the horizontal. Preorbital depth in fishes < 100 mm. S.L. (N = n), 12-5-16-5 (M = 15-5) per cent of head, in larger fishes (N = 43) 15-6-21-6 (M = 18-9) per cent, least inter- orbital width 21-2-28-3 (M = 25-0) per cent. Snout as long as broad to 1-3 times longer, its length in fishes < 100 mm. S.L., 26-0-32-1 (M = 29-1) per cent of head, in larger individuals 20-6-25-9, M = 23-3 per cent. Cheek depth 19-5-29-3 (M = 25-7) per cent of head. Caudal peduncle 14-3-22-3 (M = 17-9) per cent of standard length, 1-2 (rare)-2-o (mode 1-7) times as long as deep. Mouth horizontal or very slightly oblique, jaws equal anteriorly, or infrequently, lower projecting slightly. Length of lower jaw in fishes < 170 mm. S.L. (N = 46), 39-0-48-0 (M =44-1) percent of head, 44-2-51-0 (M = 47-5) per cent in larger fishes (N = 8) ; 1-5-2-5 (mode 1-8, modal range 1-5-2-0) times as long as broad in fishes of all sizes. Posterior tip of the maxilla generally reaching the vertical through the anterior orbital margin or to below the eye, but not reaching either point in a few individuals. Lips slightly thickened ; premaxilla not expanded medially. Gill rakers : short and stout, often flat, in most specimens ; in fishes < 120 mm. S.L., but also in a few larger individuals, most of the rakers are relatively slender. Eight-io (mode 9), rarely n, on the lower part of the first arch, the lower 1-3 rakers reduced. Scales : ctenoid. Lateral line with 32 (f.2), 33 (f.12), 34 (f.8), 35 (f.15), 36 (f.io), 37 (f.6) or 38 (f.i). Pore distribution in this series is irregular, with some pore scales occurring in the horizontal row above that in which the majority lies ; also, some scales are without pores and are often smaller than those before and behind them. Cheek with 3 (f.8), 4 (32), 5 (f.12) or 6 (f.i) rows of scales. Scales on the nape, chest and belly are small ; 6 (.3) 6| (3), 7 (f.n), 7i (1.5), 8 (14) or 9 (f.i6) between the FIG. 17. Haplochromis squamulatus, holotype, about -72 times natural size. Cat. Afr. Fish. From Boulenger, A REVISION OF THE LAKE VICTORIA HAPLOCHROMIS SPECIES 89 dorsal fin origin and the upper lateral line, 6 (f.i), 7 (f.8), 8 (f.i6), 9 (.25) or 10 (.3) between the pectoral and pelvic fin bases. Fins. Dorsal with 24 (f.g), 25 (f.37) or 26 (f.6) rays, comprising 14 (f.2), 15 (f.26) or 16 (f.24) spines, and 8 (f.2), 9 (23), 10 (.25) or n (f.2) branched rays. Anal with II (f.3), 12 (f.4o) or 13 (f.5) rays, comprising 3 spines (except for one specimen with 2), and 8 (f.3), 9 (f-4o) or 10 (f.5) branched elements. Pectoral 21-8-27-0 (M =24-7) per cent of standard length. Pelvics with the first branched ray produced in both sexes but proportionately more so in adult males. Caudal subtruncate to truncate. Teeth. Most fishes < 105 mm. S.L. have either an admixture of uni- and bicuspid teeth in the outer row of both jaws, or only bicuspids. One specimen (89 mm. S.L.) has predominantly bicuspids and a few tricuspids in the upper jaw, but mostly uni- cuspids in the lower. In larger fishes of the < 105 mm. group, unicuspid teeth predominate in both jaws, and in fishes > 105 mm. S.L. only unicuspids are found. All outer teeth are relatively slender and gently curved, the unicuspids more so than the bicuspids. The mean number of teeth in the outer row of the upper jaw shows some positive correlation with standard length although there is little difference in the ranges for the various groups ; fishes < 120 mm. S.L. have 50-72, M = 57 teeth in this row, while larger specimens have 52-80, M = 68. There is less obvious correlation between tooth form and the fish's size when the inner rows of teeth are considered. All fishes < 105 mm. S.L. have only tricuspids in the inner rows. Above this length many fishes have only unicuspids, but a mixture of uni- and tricuspids (or weakly tricuspids) is common even in the largest individuals. Unicuspids occur most frequently in the outer rows of the inner series. Inner teeth are arranged in 2-6 rows (usually 3 or 4) in the upper jaw, and in 1-6 (usually 2 or 3) in the lower jaw. There is perhaps some correlation between the number of rows and the fish's length because the largest specimens (179-198 mm. S.L.) have the greatest number of inner rows. Also, in larger fishes the inner teeth are implanted very obliquely but are almost vertical in fishes < 105 mm. S.L. Osteology. The neurocranium of H. squamulatus combines characteristics of both the H. guiarti and the H. serranus types (see Greenwood, 1962). The dorsal pre- orbital profile rises steeply and is gently curved ; its line is continuous with that of the supraoccipital crest whose anterior point lies further forward than in the skulls of H. guiarti and H. serranus. In general there is a great similarity between the neurocrania of H. squamulatus, H. michaeli and H. martini (see Greenwood, op. cit). The premaxilla is of the generalized type, and thus lacks a pronounced anterior extension of its medial dentigerous surface. The lower pharyngeal bone has the dentigerous surface broader than long, from slightly to markedly so. The lower pharyngeal teeth are relatively slender, com- pressed and distinctly cuspidate ; the teeth are arranged in 22-26 (mode 24) rows. Vertebral counts in 10 specimens are : 13+16 (f.2), 13 + 17 (f.2), 13 + 18 (f.2), 14 + 16 (f.i), 14 + 17 (f.3) giving totals of 29-31. Coloration. Live colours are known for females (adult) and quiescent males. Females : ground colour yellow-silver above shading to yellowish-white ventrally, the dorsal body and head surfaces are darkest. Two distinct longitudinal bands are 90 P. H. GREENWOOD invariably present, the lower running from the operculum to the caudal fin origin along the level of the lower lateral line, the upper extending from the nape to the end of the dorsal fin at a level about two scale rows below the dorsal fin insertion. Both bands have a finely zig-zagged outline ; the upper band may be faint. Dorsal fin dark neutral. Caudal dark neutral with a yellow flush. Pelvic and anal fins yellow. Quiescent males have a coloration like that of the females described above. Preserved material : Males (adult, sexually active but not ripe). The general im- pression is one of dusky greyness, with small areas of light brown on the flanks. The entire head is dark but is lighter on the operculum and lower lip (the lower jaw is dark). A broad black band (faint in some specimens) runs from the posterior opercular margin to the caudal fin origin ; it is crossed by 3 or 4 broad but faint vertical bars in the zone of lighter flank coloration. These bars merge with the dark dorsum and sooty-grey chest and belly. The ventral aspects of the caudal peduncle are also sooty-grey. Dorsal fin yellow-brown with a sooty overlay, the lappets are black, and a few ill-defined dark maculae may be visible on the posterior part of the soft dorsal fin. Caudal dark grey-brown on its proximal three-quarters, yellowish-brown distally. Anal coloured like the dorsal but greyer over the spinous part ; ocelli either not clearly defined, or white with a sooty surround. Pelvic fins dusky over a yellowish ground. Males (adult but quiescent}. Ground colour light brown, shading to golden on the ventral surfaces which are, however, overlaid with greyish-black. Dorsal surface of the head and body dark brown. A fairly broad, distinct and almost vertical lachrymal stripe runs from the anteroventral orbital margin to the dentary. A very distinct, broad lateral stripe runs from the opercular margin to the caudal fin base ; the upper margin of this band is irregularly serrate. There are 3 or 4 in- complete but broad and distinct vertical blotches on the ventral half of the flanks, each blotch originating from the midlateral band but not extending to a point more than half way between the lateral band and the ventral body outline. A very faint dark upper lateral band may be distinguished running at a level about midway between the dorsal fin base and the upper lateral line ; this band is often in- distinguishable from the generally dark coloration of the dorsum. A third band at the base of the dorsal fin may be visible. Dorsal fin pale brownish-yellow, with a faint sooty overlay ; lappets black, dark blotches present on the soft fin. Anal pale yellowish, with a very faint sooty overlay. Pelvics yellowish with an intense sooty overlay on the anterior half. Caudal dark greyish-yellow. Immature males are coloured like females (see below), but have a darker ground coloration (almost grey on the chest and belly) and faintly sooty pel vies. Females (adult and immature}. Female coloration is very distinctive because of the well-marked lateral bands. The ground coloration is a yellowish-silver, some- what greyish above the upper lateral line and on the dorsal head surface ; the cheek and operculum are yellow-silver. The midlateral band is black and has character- istically serrate upper and lower margins ; the depth of this band is rather variable in any one fish, and is generally deepest above the anal fin. The upper lateral band also has irregularly serrate margins but it is generally less obvious because of its A REVISION OF THE LAKE VICTORIA HAPLOCHROMIS SPECIES 91 position on the dark colour of the dorsum. A third longitudinal band, narrower and interrupted, lies along the base of the dorsal fin as a series of narrow, elongate blotches. Dorsal fin yellowish with a faint sooty overlay, and sooty lappets ; the soft part often has dark spots and blotches between the rays. Caudal yellowish with dark maculae on the proximal quarter to third. Anal and pelvic fins yellow. Ecology. Habitat. The species is known from a wide variety of habitats, in- cluding sheltered bays and gulfs, exposed beaches, and from certain off-shore localities near small islands. In most places the substrate is hard (rock, sand or shingle) but H . squamulatus is known to occur over mud bottoms. Food. Thirty-six specimens were examined (covering the entire size-range, and from 18 localities) ; of these, 19 contained food in the gut. In 16 fishes the food consisted entirely of fishes, the remains so macerated that certain identification was difficult ; in two cases the remains were identified as Haplochromis species, in one as Engraulicypris , and in another as a cichlid. In two female fishes the stomach con- tained many small embryos of cichlid fishes ; these could well be the fishes' own broods swallowed at the time of capture, although it is not known whether H. squamulatus is a mouth-brooder. A third female fish had several small post-larval cichlids in the stomach ; judging from the gonad state of the predator it seems un- likely that the larvae were its own brood. In addition to the sixteen piscivorous fishes, one other yielded only crustacean remains, one several insect egg-masses, and a third the remains of boring mayfly larvae (Povilla adusta). Breeding. All specimens, except one, below 135 mm. S.L. are immature ; the exceptional fish (93 mm.) is a ripening female. The largest fishes (180-198 mm. S.L.) are females but there is one male of 179 mm. S.L. Affinities. The colour pattern, small chest and nuchal scales and the sharply decurved head profile of H. squamulatus make at least large specimens readily identifiable. But, smaller individuals are less easily distinguished from specimens of H . martini (although the latter are adult at a size when most H. squamulatus are still immature). Unfortunately, little is known about the live coloration of sexually active male H. squamulatus but there are several similarities in the coloration of female H. martini and H. squamulatus, particularly in the striping and the yellowish coloration. In fishes of all sizes, H. squamulatus differ from H. martini in their less strongly decurved head profile, maxilla not extending so far posteriorly (never reaching a point below the pupil as is general in H. martini], in having a higher modal number of lateral line scales (35 cf. 33) and of scales between the pectoral and pelvic fin bases (9 cf. 7 or 8), and in the somewhat shallower body (23-5-35-1, M = 30-3% standard length, cf. 30-8-38-0, M =34-4%). When specimens over 100 mm. S.L. are compared, a number of additional mor- phometric differences are apparent : H. squamulatus has a longer snout, smaller eye and a longer lower jaw. Specimens less than 100 mm. S.L. have only one trenchant morphometric difference, the size of the eye (eye diameter in H. squamulatus is 25-9-30-4 [M = 28-3]% of head, cf. 29-4-37-5 [M = 317]% in H. martini}. Also resembling H. squamulatus (and H. martini, see Greenwood, 1962) is Haplo- chromis michaeli. The species differ in their preserved coloration (see above, and 92 P. H. GREENWOOD Greenwood, op. cit., p. 205), in the straighter, slightly concave, dorsal head profile of H . michaeli, and the greater posterior extension of the maxilla in that species (gener- ally to below the pupil). Since specimens of H. michaeli over 100 mm. S.L. only are known, interspecific morphometric comparisons are restricted. These, however, show that H. squamulatus has a slightly shallower body (23-5-35-1, M=30-3% of standard length, cf. 30-8-37-6, M=343%), and a smaller eye (20-6-25-9, M = 23-3% head, cf. 24-0-29-1, M =27-0% in H. michaeli). The nuchal scales of H. squamulatus are somewhat smaller (modal number of scales between the dorsal fin origin and the upper lateral line 8 or 9, cf. 6 or 7 in H. michaeli) as are those of the lateral line series (mode 35 cf. 33) ; the upper limit of the range for the lateral line scale count is also higher in H. squamulatus (38 cf. 35) . All in all, the resemblances between H. squamulatus and H. martini are greater than than those between H. squamulatus and H. michaeli, but the three species seem to form a closely related group within the more generalized piscivorous predators of Lake Victoria. There is a certain similarity, albeit superficial, between H. squamulatus and H. altigenis. On closer examination, however, it does not seem likely that the re- semblance can be construed as implying a close phyletic connection between the species. Haplochromis altigenis is probably a derivative of the H . guiarti stem. In its general facies, H. squamulatus shows some Similarity with H. dichrourus, but there is a very pronounced interspecific difference in coloration (cf. p. 90 with p. 67), as well as differences in some morphometric characters. Haplochromis squamulatus has a larger eye (20-6-25-9, M = 23-3% of head, cf. 19-4-22-6, M = 21-1) a shorter lower jaw (44-2-51-0, M = 47-5% of head, cf. 51-3-54-0, M =52-7%) and a higher modal number of lateral line scales (35 cf. 33). Phyletically, H. squamulatus and H. michaeli may represent slightly divergent developments (less so morphologically and ecologically in the former species) from an H. martini-like stem, an increase in adult size being a common factor in the two lines. Museum and Reg. No. B.M. (N.H.) 1966.3.9.262-272 B.M. (N.H.) 1966.3.9.282-284 B.M. (N.H.) 1966.3.9.285-288 B.M. (N.H.) 1966.3.9.289-295 B.M. (N.H.) 1966.3.9.259-260 B.M. (N.H.) 1966.3.9.261 Genoa Museum C.E. 12977 (Holotype) B.M. (N.H.) 1906.5.30.233-9 B.M. (N.H.) 1909.5.4.4-5 B.M. (N.H.) 1966.3.9.278-281 Locality Collector UGANDA Napoleon Gulf near Jinja . . E.A.F.R.O. Fisherman's Point near Jinja . E.A.F.R.O. Beach near Nasu Point (Buvuma Channel) . . . E.A.F.R.O. Ramafuta Island (Buvuma Channel) . . " . . E.A.F.R.O. Off Buvuma Island (Buvuma Channel) . . : 3,,=, .... E.A.F.R.O. Grant Bay . . . . E.A.F.R.O. Jinja (Napoleon Gulf) . . . Bayon Nsonga . . . . Degen Sesse Islands .... Bayon Beach in Entebbe Harbour E.A.F.R.O. A REVISION OF THE LAKE VICTORIA HAPLOCHROMIS SPECIES 93 B.M. (N.H.) 1966.3.9.258 B.M. (N.H.) 1966.3.9.273-277 B.M. (N.H.) 1966.3.9.255 B.M. (N.H.) 1966.3.9.257 B.M. (N.H.) 1966.3.9.256 B.M. (N.H.) 1901.6.24.86 Entebbe, Airport Beach Kazima Island near Entebbe Old Bukakata Bay . TANZANIA Capri Bay, Mwanza . Mwanza Harbour LAKE VICTORIA Locality unknown E.A.F.R.O. Uganda Fisheries Dept. E.A.F.R.O. E.A.F.R.O. E.A.F.R.O. Sir H. Johnston Haplochromis barbarae sp. nov. (Text-figs. 18 and 19) HOLOTYPE : an adult female, 97 mm. standard length, from a beach near Nasu Point (Buvuma Channel) ; B.M. (N.H.) reg. no. 1966.2.21.5. Named in honour of Mrs. Barbara Williams, whose drawings illustrate this and others of my papers. DESCRIPTION : based on 9 specimens (including the holotype) 89-0-106-0 mm. standard length. Depth of body 32-1-35-5 (M = 34-0) per cent of standard length, length of head 31-4-35-0 (M = 33-5) per cent. Dorsal head profile very slightly decurved or straight. Preorbital depth 15-1-18-0 (M = 16-9) per cent of head length, least interorbital width 24-2-26-7 (M = 25-5) per cent. Snout as long as broad, its length 30-0-32-4 (M=3i-5) per cent of head, eye diameter 22-9-27-4 (M=258), cheek depth 20-0-25-0 (M = 22-2) per cent. Caudal peduncle 15-7-18-0 (M = 16-8) per cent of standard length, 1-4-1-7 times as long as deep (no well-defined mode). Mouth very slightly oblique, the jaws equal anteriorly ; lower jaw 36-6-41-2 (M = 38-0) per cent of head, 1-5-1-7 (in one fish 2-0) times as long as broad. Posterior tip of the maxilla reaching a point near the vertical through the anterior orbital margin. FIG. 18. Haplochromis barbarae. Drawn by Barbara Williams. ZOOL. 16, 2. 94 P. H. GREENWOOD Gill rakers : of variable form, from moderately short and stout to slender and relatively elongate, but of constant form (except for the reduced lower rakers) in any one individual ; 8 or 9 rakers on the lower part of the first gill arch. Scales : ctenoid ; lateral line with 33 (f.3), 34 (f.2) or 35 (f.4), cheek with 3 or 4 (rarely 2) rows. Seven (rarely 6^ or 8) scales between the upper lateral line and the dorsal fin origin, 7 or 8 (rarely 6 or 9) between the pectoral and pelvic fin bases. Scales on the ventral aspects of the chest are noticeably smaller than those situated laterally. Fins. Dorsal with 25 (f.7) or 26 (f.2) rays, comprising 16 spinous and 9 (f.7) or 10 (f.2) branched rays. Anal with n (f.2), 12 (f.6) or 13 (f.i) rays, comprising 3 spines and 8 (f.2), 9 (f.6) or 10 (f.i) branched elements. Pectoral 25-0-29-2 (M = 26-7) per cent of standard length. Caudal truncate, scaled on its proximal half. Teeth. In the outer row of both jaws, the teeth are relatively stout, bicuspid and slightly curved ; in a few specimens some posterolateral upper teeth are unicuspid and enlarged. There are 40-65 (M = 50) teeth in the outer row of the upper jaw. Teeth forming the inner rows are relatively large and tricuspid, and are implanted somewhat obliquely so that the crowns point inwards. There are 2 or 3 inner rows in the upper jaw, and 2 (3 in one specimen) in the lower. Osteology. No complete skeleton is available. The lower pharyngeal bone is, compared with that of similar species, small. Its dentigerous surface is broader than long, and narrows rapidly at about the midpoint so that not only is the whole bone relatively small, but so is the area of pharyngeal teeth. The teeth are fine, com- pressed and distinctly cuspidate ; they are rather sparsely distributed in from 22-24 rows. Vertebral counts in 9 specimens are : 13+17 (f.8) and 13 -f 18 (f.i) giving totals of 30 and 31. Coloration : Live colours are unknown. Preserved colours : Males (adult but quiescent). Ground colour dark yellowish-brown. A well-defined lachrymal stripe is present, as are faint traces of two transverse, parallel stripes across the snout. There is some darkening over the preoperculum but this is not concentrated into a FIG. 19. Lower pharyngeal bones (in occlusal view, with dentition shown on one side only) of (A) H. barbarae, and (B) H. chromogynos. A REVISION OF THE LAKE VICTORIA HAPLOCHROMIS SPECIES 95 bar or stripe. The branchiostegal membrane is dark, most intensely so below the opercular bones. On the flank is a faint midlateral stripe, and four even fainter and incomplete vertical bars. Dorsal and anal fins are yellowish with a sooty overlay, the anal being darkest along its proximal third ; anal ocelli very faint. Pelvics sooty over a yellowish ground. Caudal yellowish on the margins and distal third, dark brown proximally. Females (adult.} Three distinct colour patterns occur ; two of these are probably identical with the polychrome patterns described for live Hoplotilapia retrodens (see Greenwood, 1956). The colour pattern corresponding with the modal coloration in other polychromatic species has a yellow-brown ground colour with a distinct midlateral dark stripe running from the opercular margin to the caudal origin, and extending onto the caudal fin itself for about half its length ; in two specimens there are four, narrow vertical bars on the flanks, extending from the dorsal fin base almost to the ventral outline. In all specimens a very faint upper line is situated slightly above the upper lateral line. All fins are yellowish-brown. Three out of the eight female specimens have this type of coloration. The second pattern is a piebald, black on silvery-yellow (to yellow brown), the black pigment arranged in blotch-like bars of variable width and dorso-ventral extent. Some blotches extend onto the dorsal, caudal, anal and pelvic fins. The cheeks and snout may be blotched or clear. All fins are yellowish. This pattern does not differ from that shown by other species with a piebald coloration in females. Four of the eight H. barbarae females are piebald. The third pattern, found only in one specimen, has an orange-yellow background peppered with fine melanophores which are, in places, aggregated into blotches (especially on the dorsum) with an irregular distribution. Some dark blotches occur on all fins (but especially the dorsal and caudal) ; these spots are smaller than those on the body. The ground colour of all fins is light orange-yellow. Ecology. Habitat. The nine known specimens came from three different sites. Two of these are shallow, exposed and sandy beaches, the third, shallow water (ca. 10 ft. deep) over a hard substrate a short distance off-shore and near the water- lily zone fringing a papyrus swamp. Food. Seven of the 9 specimens examined contained ingested matter in the guts. In each case this consisted of from 1-8 (mode 5) recently fertilized cichlid ova. In two specimens these could have been part of the fish's own brood swallowed during capture, because the fishes have recently spent ovaries. But, judging from the advanced stages of oogenesis shown by four other specimens, and the fact that the fifth is a male, it seems reasonable to conclude that the embryos were taken as food (see Greenwood, 1959, for a discussion of other paedophagous Haplochromis species). Breeding. Nothing is known about the breeding habits of H. barbarae. All nine specimens are sexually mature ; the sole male (102 mm. S.L.) is the second largest specimen. Affinities. Anatomically, H. barbarae is very like H. brownae, a generalized species exhibiting many characters suggestive of affinity with anatomically generalized predators such as H. guiarti (Greenwood, 1962). 96 P. H. GREENWOOD Haplochromis barbarae differs from H . brownae in having a narrower interorbital (24-2-26-7, M = 25-5% of head, cf. 26-0-34-0, M = 29-8%), a smaller eye (22-9-27-4, M = 25-8% of head, cf. 26-0-31-3, M = 28-6%), fewer gill rakers (8 or 9, cf. 9-12 [modes 10 and n]), and smaller scales, especially on the ventral aspects of the chest. There is also a difference in dention in that the outer teeth of this species are stout bicuspids whereas in H. brownae of a similar size the teeth are slender unicuspids. It differs from H. guiarti in having a shorter snout (30-0-32-4, M = 31*5% of head, cf. 31-7-37-5, M = 34-4) and lower jaw (36-6-41-2, M = 38-0% of head, cf. 39-2-48-2, M = 44-4%), fewer gill rakers (8 or 9, cf. 9-11, mode 10) and, at com- parable sizes, the presence of bicuspid teeth anteriorly in both jaws (unicuspid in H. guiarti} . Because of its generalized anatomy and unspecialized dentition, H. barbarae closely resembles a number of other species in the Lake Victoria flock, viz. H. cinereus, H. macrops, H. lacrimosus and H. chromogynos (see Greenwood, 1959 for the latter species and Greenwood, 1960 for the three former). In addition to their anatomical similarities, H. barbarae and H. chromogynos also share (with several other and structurally unrelated species) the piebald coloration in females. However, H. barbarae differs from H. chromogynos in having a longer lower jaw (36-6-41-2, M = 38-0% of head, cf. 30-0-34-4, M = 32-5%), and, when specimens of equal size are compared, bicuspid instead of slender unicuspid teeth (specimens of H . chromo- gynos < 95 mm. S.L. have a mixed bi- and unicuspid dentition but larger individuals have only unicuspid teeth). The shape of the pharyngeal bone differs in the two species, and it is this character (see p. 94 and fig. 19) which most readily distinguishes H. barbarae from the other three species mentioned above, none of which is known to have piebald females or paedophagous habits. The peculiar feeding habits of H. barbarae immediately suggest some affinity with the other paedophagous Haplochromis species, and because H. barbarae is anatomic- ally unspecialized, particularly with the more " generalized " paedophages, H . obesus and H. cronus. However, in many morphological details these two differ consider- ably from H. barbarae. For instance, the peculiar dental morphology of H. obesus (with the small teeth deeply embedded in the gums), and the stout unicuspid teeth of H. cronus, serve as immediately diagnostic characters, as do the several mor- phometric differences between the species. It is of interest, however, to recall that piebald females occur in all three species ; but, since this character appears in other and widely different species, it is not thought to have any phyletic significance. Any supposed close phyletic relationship between H. barbarae and the other members of the paedophagous species group would also be difficult to substantiate. At present all that can be suggested is that H. barbarae might represent an isolated line, derived from a generalized and probably insectivorous stem, paralleling trophically the paedophagous species group (itself probably of diphyletic origin). Alternatively, H. barbarae might represent a survivor of the stem from which such species as H. cryptodon and H. microdon arose ; more will have to be learned about the anatomy of H. barbarae before this possibility can be substantiated. A REVISION OF THE LAKE VICTORIA HAPLOCHROMIS SPECIES 97 STUDY MATERIAL AND DISTRIBUTION RECORDS Museum and Reg. No. Locality Collector UGANDA B.M. (N.H.) 1966.3.9.243 Napoleon Gulf, off Jinj a . . E.A.F.R.O. B.M. (N.H.) 1966.2.21.5 Beach near Nasu Point (Holotype) (Buvuma Channel) . . . E.A.F.R.O. TANZANIA B.M. (N.H.) 1966.3.9.244-251 Beach near Majita . . . E.A.F.R.O. Haplochromis tridens Regan and Trewavas, 1928 (Text-fig. 20) Haplochromis tridens Regan & Trewavas, 1928, Ann. Mag. nat. Hist, (io), 2, 226. LECTOTYPE : a fish 116-0 mm. standard length (caudal fin damaged), B.M. (N.H.) reg. no. 1928.6.2.41, collected in Tanzanian waters at Michael Graham's station 234 (i 4' S, 32 13' E), at a depth of over 100 ft. (Graham, 1929). DESCRIPTION : based on 16 specimens (including the lectotype and the paralecto- type) 72-119 mm. standard length. Depth of body 30-1-36-2 (M = 33-5) per cent of standard length, length of head 32-0-37-2 (M = 35*4) per cent. Dorsal head profile straight, but interrupted by the prominent premaxillary pedicels, sloping steeply at 4o-45. The cephalic lateral line system with prominent pores, especially those of the preopercular and preorbital canals. These pores are probably larger than in any Lake Victoria Haplochromis (including H . pachycephalus and H. hoops] . Preorbital depth 16-0-20-9 (M = 17*2) per cent of head, least interorbital width 15-0-19-5 (M = 16-7) per cent. Snout 1-0-1-3 (mode i-i) times as long as broad, its length 28-0-34-8 (M = 30-4) per cent of head. Eye with an oval, horizontally aligned pupil, eye diameter 25-6-34-0 (M = 30-7) per cent of head, depth of cheek 17-5-22-8 (M = 19-2) per cent. Caudal peduncle 16-7-19-8 (M = 17-9) per cent of standard length, 1-4-1-9 (mode 1-6) times as long as deep. Mouth horizontal or slightly oblique, lower jaw projecting slightly and with a distinct mental projection ; length of lower jaw 43-3-51-8 (M = 47-5) per cent of head, 2-0-2-8 (mode) times as long as broad. Premaxilla slightly expanded in the midline. Posterior tip of the maxilla generally reaching to below the anterior part of the eye, and occasionally to below the pupil ; rarely reaching only to the vertical through the anterior margin of the orbit. Gill rakers : relatively slender, except for the lower 1-3 which are reduced, and the upper 2-4 which are usually flat ; 8-n (mode 9) on the lower part of the first gill arch. Scales : strongly ctenoid, especially those on the chest region. Lateral line with 31 (f.4), 32 (f.5), 33 (f.4) or 34 (f.2) scales. Cheek with 3 (mode) or 4 rows. Five and a half to 6^ (modes 6 and 6|) scales between the upper lateral line and the dorsal fin origin, 5-6^ (mode) between the pectoral and pelvic fin bases. 9 8 P. H. GREENWOOD FIG. 20. Haplochromis tridens, paralectotype, about -73 times natural size. Drawn by Miss M. Fasken. Fins. Dorsal with 23 (f.y), 24 (f.y) or 25 (f.2) rays, comprising 15 (.4) or 16 (f.12) spinous, and 7 (f.6), 8 (f.6), 9 (f.3) or 10 (f.i) branched rays. Anal with 10 (f.io), II (f.5) or 12 (f.i) rays, comprising 3 spines and 7 (f.io), 8 (f.5) or 9 (f.i) branched elements. (N.B. More than usual difficulty was experienced in deciding if the last dorsal and anal ray was a single, deeply divided element, or two distinct rays ; this may account for the number of specimens with low (7) branched ray counts). Pectoral 27-0-33-3 (M = 29-3) per cent of standard length. Pelvics with the first branched ray produced, proportionately more so in adult males. Caudal truncate. Teeth. The trivial name " tridens " was given to the species because the larger of the two syntypes had only tricuspid teeth in the outer tooth row of both jaws, an unusual (probably unique) feature for a Haplochromis species. This condition is, however, rare in the species. Uni-, bi- and tricuspid outer teeth may all occur in a single specimen. Tricuspid teeth, at least in the upper jaw, are usually found posterolaterally, the bi- and unicuspids laterally and anteriorly. This arrangement is by no means constant ; in some specimens tricuspids occur anteriorly and antero- laterally. One fish (72 mm. S.L.) has only bicuspids in the upper jaw, and another (81 mm.) only unicuspids. There is no apparent correlation between the predomin- ance of any one tooth form and the size of the fish. In three out of the sixteen specimens examined, no tricuspid outer teeth were found in either jaw. There are 58-80 (M = 66) teeth in the outer row of the upper jaw. Variation in the type of outer teeth occurring in the lower jaw follows the same pattern as in the upper jaw. Teeth forming the inner rows in both jaws are predominantly tricuspids. These teeth are relatively large, are stout, and have the median cusp larger than the lateral ones. There are usually 2 inner rows (occasionally 3, rarely 4) in the upper jaw, and 2 (rarely i) in the lower. Osteology. No entire skeleton is available. The lower pharyngeal bone has its dentigerous surface equilateral or slightly broader than long. The lower pharyngeal teeth have cuspidate, compressed crowns, and cylindrical necks ; the teeth are arranged in 22-26 rows. A REVISION OF THE LAKE VICTORIA HAPLOCHROMIS SPECIES 99 Vertebral counts in 10 specimens are : 13+16 (f.y) and 13 -f 17 (3), giving totals of 29 and 30. Coloration. Live colours are unknown. Preserved specimens : Males (adult and sexually active) : ground colour grey-blue (gun-metal) above the upper lateral line, greyish silver on the flanks, and light dusky silver on the chest and belly. Dorsal and caudal fins hyaline. Anal hyaline except for a dusky area between the spines, and black lappets ; the dusky area extends as a fairly distinct line at the base of the spinous part, and may be expanded basally onto the soft fin. There are two, large, grey- white anal ocelli. Pelvics dusky, darkest on the lateral four-fifths of the fin. Males (adult but quiescent} have a ground coloration more like that of females. All fins are hyaline except for a faint darkening between the anal spines, and on the pelvic fins. Females (juvenile and adult) : greyish silver above, shading to silver below. One specimen (ovaries ripening) is darker dorsally (almost brown), and has a broad, interrupted midlateral stripe running from behind the operculum to about half the length of the caudal fin (on which the stripe narrows) ; the band is interrupted at about the middle of its length. All fins yellowish to hyaline. Ecology. Habitat. Excepting the two types, all the material came from one trawl haul at an unknown locality (thought to be off the Kenya coast). The types came from deep water (more than 100 ft.) some distance off-shore, and from over a soft bottom. The large eyes and hypertrophied pores of the cephalic lateral line system certainly suggest adaptations to a deep water habitat. Food. Only 6 of the 10 specimens examined (all from the same, and unknown, locality) contained ingested material in the gut. In each, the predominant contents are undigested blue-green algae, and empty diatom frustules. Two specimens have, in addition, a few fragments of Crustacea, and some unidentifiable insect remains. Breeding. All fourteen of the non-typical specimens are adults. The types are not well-preserved internally, but both appear to be females. Affinities. The admixture of tri-, uni- and bicuspid outer teeth together with the hypertrophy of the cephalic lateral line pores, provide a trenchant means of distin- guishing H. tridens from other species in the lake. Indeed, it is difficult to suggest any close relationship between this species and any other so far considered. Perhaps some relationship will become apparent when more is known about the numerous and small species of Haplochromis which make up the bulk of fishes caught by trawling in the deeper waters of Lake Victoria (Greenwood, unpublished). Anticipating these results, it is possible to say that the elongate body and general " predatory " facies (especially the large mouth) of H . tridens are not common amongst these species, and nor is the H. tridens dental type. Amongst the larger species with a predatory facies, H. tridens most closely resembles H . victorianus. There are, however, many differences between the species, not least of which are their differences in ecology and the much smaller adult size of H. tridens. Haplochromis victorianus differs also in its dentition, broader interorbital region (21-5-24-5, M = 22-6% of head, cf. 15-0-19-5, M = 16-7%) smaller and rounder eye (21-7-25-5, M = 23-6% head, cf. 25-6-34-0, M =30-7%), and deeper P. H. GREENWOOD cheek (22-5-26-2, M = 24-6% head, cf. 17-5-22-8, M = i9'9%). It should be remembered, however, that the data for H. victorianus are derived from larger speci- mens than are available for H. tridens, and that the most trenchant morphometric differences are in characters most affected by allometric growth. Nevertheless, it seems very unlikely that H. victorianus and H . tridens are close relatives. STUDY MATERIAL AND DISTRIBUTION RECORDS Museum and Reg. No. B.M. (N.H.) 1928.6.2.41 (Lectotype) B.M. (N.H.) 1928.6.2.42 (Paralectotype) B.M. (N.H.) 1966.3.9.152-165 Locality TANZANIA 104' 8,32 13' E 104' S, 32 13' E LAKE VICTORIA Locality unknown (? Kenya coast) Collector M. Graham M. Graham E.A.F.R.O. Haplochromis orthostoma Regan, 1922 (Text-fig. 21) Pelmatochromis spekii (part): Boulenger, 1915, Cat. Afr. Fish., 3, 417. Haplochromis orthostoma Regan, 1922, Proc. zool. Soc. Londn., 184, fig. 9. HOLOTYPE : a male (probably adult), 91 mm. standard length (B.M. [N.H.] reg. no. 1912.10.15.67) from Lake Salisbury (Kyoga system). The specimen is now in very poor condition, and has lost most of its scales. This species, possibly one of the two most distinctive looking members of the Victoria-Kyoga flock, was described from a single specimen. Since then, two further specimens have been collected from Lake Salisbury, and three superficially similar specimens have been caught in Lake Victoria. The latter specimens differ from the Salisbury fishes in their dentition (and some other characters) and are thought to represent a distinct species. It is for this reason that I am redescribing the Lake Salisbury species in this paper. The unusual physiognomy of H. orthostoma is well shown in text-figure 21 ; the snout and nuchal region meet at a distinct angle, the nuchal musculature not bulging anteriorly and laterally. Since there are only three specimens available, morphometric characters are tabulated below. The holotype is indicated with an asterisk. S.L. (mm.) 83-0 *9i-o Depth t 36-3 Head Preorb. Interorb. Snout Eye Cheek Caudal t /o /o /o /o /o Peduncle f 36-0 19-6 20-4 32-7 24-5 24-5 19-3 36-1 2O-O 23-3 30-0 21-7 30-0 J 7'5 35'5 18-5 21-5 30-8 22-8 30-8 16-5 t Per cent of standard length. % Per cent of head length. A REVISION OF THE LAKE VICTORIA HAPLOCHROMIS SPECIES 101 Mouth very oblique, sloping at ca 5o-7o, the lower jaw projecting, its length 51-0-56-6 per cent of head length, and 2-5-3-4 times as long as broad. Lips not thickened, the lower jaw with a distinct mental protuberance. Posterior tip of the maxilla reaching a point about midway between the vertical through the anterior orbital margin and that through the nostril. Snout 1-1-1-3 times as long as broad, its dorsal surface slightly rounded, the premaxillary pedicels not prominent. Caudal peduncle 1-5-1-6 times as long as deep. Gill rakers : moderately stout, the lower 1-3 reduced, the upper 2 or 3 flattened (anvil-shaped in one fish) ; 9 or 10 on the lower part of the first gill arch (n on one arch of a fish with 9 rakers on the other arch) . FIG. 21. Haplochromis orthostoma, holotype. From Regan, Proc. zool. Soc., about 86 times natural size. Scales : ctenoid ; lateral line with 30 or 31 scales (Regan gives 33 for the now scaleless type), cheek with 3 or 4 rows (4 or 5 in the type according to Regan). Six or 6| between the upper lateral line and the dorsal fin origin (6 or 7 in the type ace. Regan), 6 or 7 between the pectoral and pelvic fin bases, the chest scales small. Fins. Dorsal with 14 (f.i) or 15 (f.2) spines and 9 branched rays, anal with 3 spines and 8 (f.2) or 9 branched rays. First branched pelvic ray not produced in the type, but elongate in the two other specimens (male and female) . Caudal subtrucate. Teeth. In the two larger specimens (83 and 91 mm. S.L.), the outer teeth in both jaws are a most distinctive feature. These teeth are slender and unicuspid with sharply curved tips directed medially ; those situated anteriorly are somewhat larger than the others. There are 66 and 61 teeth in this row for the two specimens respectively. In the lower jaw, the outer teeth, at least anteriorly, are somewhat stouter than their opposites in the upper jaw. The smallest specimen (67-5 mm. S.L.) has outer teeth quite unlike those of the larger individuals. In the upper jaw, the anterior and lateral teeth are stout and bicuspid, those situated laterally and posteriorly are stout and tricuspid. All outer 102 P. H. GREENWOOD teeth in the lower jaw are bicuspid and stout, and are a little stouter than the upper jaw teeth. There are 68 teeth in the outer row of the upper jaw. The inner teeth in the two larger fishes are unicuspid, small, slender and implanted obliquely ; in the smallest fish they are small and tricuspid. The dental arcade in all specimens is V shaped with a rounded apex ; there are 2 inner tooth rows in the upper jaw, and I or 2 rows in the lower jaw. Osteology. No complete skeleton is available. The lower pharyngeal bone has a triangular and equilateral dentigerous area. The teeth are relatively slender, with bicuspid and weakly compressed crowns, and are arranged in 20-22 rows ; except in the smallest fish the teeth of the two median rows are coarser than the lateral rows. Vertebral counts for 2 specimens are : 13+16 (type) and 12 + 16. Coloration : Live colours are unknown. The type is now completely colourless ; originally it was described as greyish, with a dark lachrymal stripe, and blackish pelvic fins. A juvenile female has a brown ground coloration, with very faint traces of four broad vertical bars on the flanks running from the dorsal fin origin to about the level of the ventral margin of the pectoral fin. No lachrymal stripe is visible. The lower jaw (especially over its anterior half) is rather dusky. The dorsal and caudal fins are yellowish, the former with black lappets. Anal dusky yellow. Pel vies faintly dusky, especially at their tips. Pectorals yellowish-grey. Adult Male (probably fixed in alcohol) : light brown above, shading to silver on the mid-flanks, and silvery white on the belly. Snout dark, as are the lips ; branchio- stegal membrane pale. A faint lachrymal stripe is visible. Dorsal fin greyish, with black lappets. Anal greyish, with two white ocelli. Caudal grey, the melanophores most concentrated along its midline. Pelvics dark along the anterior quarter, hyaline elsewhere. Ecology. No information is available on the habitat or food of H. orthostoma, nor is there any information on the breeding habits of the species. The two males (91 and 83 mm. S.L.) are adult, and the female (67-5 mm.) is apparently immature. Affinities. The peculiar head shape, large and very oblique mouth, coupled with the peculiar tooth form (at least in the larger fishes), serve to distinguish the species from all others in the Lake Victoria-Kyoga flock. The nearest relative is H. par- orthostoma from Lake Victoria (see below) . Further material, and field observations, may yet show that the two species are not distinct at that level. STUDY MATERIAL AND DISTRIBUTION RECORDS Museum and Reg. No. Locality Collector B.M. (N.H.) 1912.10.15.67 Lake Salisbury . . . . Presented by (Holotype) F. J. Jackson B.M. (N.H.) 1958.12.5. 173 Ongino, Lake Salisbury . . Pitman B.M. (N.H.) 1966.3.9.252 Lake Salisbury . . . . E.A.F.R.O. A REVISION OF THE LAKE VICTORIA HAPLOCHROMIS SPECIES 103 Haplochromis parorthostoma sp. nov. (Text-fig. 22) HOLOTYPE : an adult male, 117 mm. standard length, from near Zero Island (Buvuma Channel), Uganda. B.M. (N.H.) reg. no. 1966.2.21.4. DESCRIPTION. The overall similarity between this species and H. orthostoma is great, particularly since both share a peculiar head profile not seen in any other Haplochromis species from Lakes Victoria or Kyoga. The dorsal head profile is strongly concave, with the nuchal region meeting the snout at a noticeable but rounded angle, the junction emphasized by an anterior bulge of the cephalic epaxial body musculature. This muscular protuberance gives the fish a pronounced " forehead ", especially in the frontal plane. Since only three specimens are available, morphometric data are tabulated below ; the holotype is marked with an asterisk. Preorb. 16-7 19-5 t Per cent of standard length. % Per cent of head length. Mouth oblique, sloping upwards at ca. 40-5o, jaws equal anteriorly, or the lower projecting slightly. Lower jaw length 48-3-53-5 per cent of head, 2-3-2-4 times as long as broad ; chin with a distinct protuberance. Posterior tip of the maxilla reaching a point nearer the vertical through the anterior orbital margin than one through the nostril. Snout 1-2 times as long as broad, with a convex dorsal surface ; premaxillary pedicels not prominent. Lips moderately thickened. Caudal peduncle 1-1-1-3 times as long as deep. S.L. Depth Head (mm.) t t 86-0 38-4 35-o IIO-O 42-3 36-4 117-0 41-0 35' Interorb. Snout Eye Cheek Caudal o/ /o % o/ /o /o Peduncle f 2O-O 31-6 27-7 25-0 17-5 17-5 33-8 25-0 27-5 12-7 19-5 34-9 24-4 26-8 14-5 FIG. 22. Haplochromis parorthostoma. Drawn by Lavinia Beard. 104 p - H - GREENWOOD Gill rakers : moderately stout (relatively stouter in one specimen) , the lower 1-3 reduced, the upper 2 or 3 expanded and anvil-shaped ; 9 (f.2) or 10 rakers on the lower part of the first gill arch. Scales : ctenoid, lateral line with 30 or 32 (f.2) scales, cheek with 2 or 3 rows (in the former specimen, the scales not covering the ventral aspects of the cheek). Six or 7 scales between the upper lateral line and the dorsal fin origin, 6 or 7 between the pectoral and pelvic fin bases. Fins. Dorsal with 15 (f.2) or 16 spines and 8, 9 or 10 branched rays. Anal with 3 spines and 8 or 9 (f.2) branched rays, the spines short and stout. Pectoral 24-5-26-2 per cent of standard length. First branched pelvic ray not or slightly produced (all specimens are males) . Caudal almost rounded, scaled on its basal two-thirds. Teeth. The outer row in both jaws is composed of unicuspid, slender and slightly curved teeth, the curvature being gentle and not confined to the distal part of the tooth (cf. H. orthostoma, p. 101) ; in one fish, the last three teeth in the upper jaw are larger, stouter and straighter than the anterior ones. A few bicuspid, moderately stout teeth occur posteriorly in the lower jaw of the smallest (86 mm.) fish. In no specimen are the lower jaw teeth stouter than those in the upper jaw. There are 38, 40 and 52 teeth in the outer row of the upper jaw. Teeth forming the inner series are small, tricuspid and weakly tricuspid, and are arranged in 3 or 4 rows in the upper jaws and in 2 rows in the lower. The dental arcade is V shaped, with the apex broadly rounded. Osteology. No complete skeleton is available. The lower pharyngeal bone has its triangular dentigerous surface as long as broad, or slightly broader than long. The teeth, arranged in 24-30 rows, are fine, with weakly compressed bicuspid crowns ; those in the two median rows are but slightly larger than the lateral teeth. Vertebral counts in 3 specimens are : 13+15 and 13 -j- 16 (f.2). Coloration : Live colours are known for a sexually active (but not ripe) male. Ground colour dark slate-grey, with faint vertical bars of a darker shade. Dorsal fin dark grey with crimson lappets on the posterior two-thirds of the spinous part, and a dark crimson margin to the entire soft part. Caudal dark grey with a crimson flush, particularly intense on its lower half. Entire anal fin, except for a dark base, crimson. Pel vies black on the anterior third, remainder dull crimson. Preserved material : Males (adult) . Ground colour light brown (including the branchiostegal membrane in two fishes ; this membrane blackish in the third specimen) ; flank crossed by 5 or 6 dark but incomplete bars, each bar originating just above the upper lateral line and extending to about the level of the ventral margin of the pectoral fin. A fairly distinct vertical lachrymal stripe runs from the antero ventral margin of the orbit to the angle of the lower jaw. Dorsal fin yellowish- brown along its margin, but dark brown between the rays. Anal yellowish, sooty or dark brown between the branched rays ; very faint indications of 2 or 3 whiteish ocelli. Caudal light to dark brown (almost black). Pelvics dusky on the anterior third to half, otherwise hyaline. Ecology. One specimen came from an exposed, sandy beach, another from a rocky outcrop in about 20 ft. of water near an off-shore island, and the third from over a rocky shelf in about 40 ft. of water, also near an island. A REVISION OF THE LAKE VICTORIA HAPLOCHROMIS SPECIES 105 No information is available on the food of H. parorthostoma. The three specimens are adult males. Affinities. Haplochromis parorthostoma seems to be very closely related to H. orthostoma of the Kyoga system, at least in its peculiar head-shape. There is, however, a marked interspecific difference in the shape of the outer teeth. In H. parorthostoma these teeth are gently curved, whereas in H. orthostoma they have sharply recurved crowns but relatively straight necks. Furthermore, the inner teeth of H. orthostoma are unicuspid and arranged in one or two rows, but in H. parorthostoma are tricuspid and arranged in three or four rows in the upper jaw (two rows in the lower). An exception (at least with regard to outer tooth shape) is provided by the smallest specimen of H. orthostoma whose stout, erect and bicuspid teeth do not resemble those of its larger congeners (or, for that matter, H. parortho- stoma}. But, since the cranial morphology of this small fish is so like that of larger H. orthostoma individuals it is included in that species. Haplochromis parorthostoma and H . orthostoma differ in characters other than dental ones. The bulging cranial epaxial muscles of H. parorthostoma impart a different shape to the dorsal head profile, which is further modified by the less oblique mouth. The available samples also suggest that H. parorthostoma has a bigger eye than H. orthostoma (and this despite the fact that the specimens of H. parorthostoma are larger) and probably a shorter and broader lower jaw. Clearly, much more material is required before it will be possible to reach more definite conclusion about the status of the two species. Even then the decision will be complicated by the fact that they are allopatric, and their areas of distribution are physically isolated (now by the virtually impenetrable Owen Falls dam, and previously by the Ripon Falls, perhaps not a complete barrier to migration from Lake Victoria to the Kyoga system). For the moment it is not possible to speculate on the wider relationship of these two species. STUDY MATERIAL AND DISTRIBUTION RECORDS Museum and Reg. No. Locality Collector Uganda B.M. (N.H.) 1966 .2.21 Near Zero Island (Holotype) (Buvuma Channel) . . . E.A.F.R.O. B.M. (N.H.) 1966.3.9.253 Near Zero Island (Buvuma Channel) . . . E.A.F.R.O. B.M. (N.H.) 1966.3.9.254 Near Kazima Island . . . Uganda Fish. Dept. Haplochromis apogonoides sp. nov. (Text-fig. 23) HOLOTYPE an adult male, 118 mm. standard length, from Ekunu Bay, Uganda. B.M. (N.H.) reg. no. 1966.2.21.3. Named " apogonoides " because of its fancied resemblance to the genus Apogon. io6 P. H. GREENWOOD DESCRIPTION : based on eight specimens (including the hole-type) 112-132 mm. standard length ; with one exception (a quiescent female) all are males. Depth of body 36-0-39-6 (M = 38-1) per cent of standard length, length of head 35-0-37-2 (M = 36-1) per cent. Dorsal head profile curved (strongly so in some specimens), sloping fairly steeply (ca. 40), the premaxillary pedicels not prominent. Preorbital depth 13-6-16-7 (M = 15*2) per cent of head length, least interorbital width 27-5-31-0 (M = 29-5) per cent. Snout 1-3-1-5 (mode 1-4) times as broad as long, its length 29-3-33-4 (M =31-2) per cent of head, eye 25-3-27-5 (M = 26-6), depth of cheek 27-8-31-0 (M 29-3) per cent. Caudal peduncle 15-3-19-8 (M = 18-0) per cent of standard length, 1-5-1-7 times as long as deep. Mouth slightly to moderately oblique, lips somewhat thickened, the jaws equal anteriorly. Lower jaw 45-0-51-0 (M = 47-9) per cent of head, 1-2-1-5 (modal range 1 CM. FIG. 23. Haplochromis apogonoides. Drawn by Barbara Williams. 1-4-1-5) times as long as broad. Posterior tip of the maxilla reaching to below the pupil. Premaxilla not expanded medially. Gill rakers : stout, the lower 1-3 reduced, the upper 3 or 4 flat ; 8-10 on the lower part of the first gill arch. Scales : ctenoid ; lateral line with 32 (f.4), 33 (f.3) or 34 (f.i) scales. Cheek with 3 or 4 (mode) rows. Six to 8 (mode 6) scales between the upper lateral line and the dorsal fin origin, 6-8 (modes 6 and 7) between the pectoral and pelvic fin bases. Fins. Dorsal with 22 (f.i), 24 (f.6), or 25 (f.i) rays, comprising 14 (f.2) or 15 (f.6) spinous and 8 (f.i), 9 (f.5J or 10 (f.2) branched rays. Anal with 10 (f.i), n (f.5) or 12 (f.2) rays, comprising 3 spines and 7 (f.i), 8 (f.5) or 9 (f.2) branched elements. Pectoral 29-0-34-1 (M = 30-8) per cent of standard length. Pelvics with the first ray slightly produced. Caudal truncate to subtruncate, scaled on its basal half to two-thirds. Teeth. The outer teeth in both jaws are a characteristic feature of the species, being unicuspid, moderately stout and with very strongly recurved tips. Such teeth A REVISION OF THE LAKE VICTORIA HAPLOCHROMIS SPECIES 107 are otherwise found only in the H. sauvagei species group (see Greenwood, 1957). There are 50-60 (M = 58) teeth in the outer row of the upper jaw. The inner teeth are also unicuspid, large and recurved, and are implanted obliquely. There are 2 rows in the upper jaw, and I or 2 rows in the lower jaw. A distinct space separates the inner series from the outer row. Osteology. No complete skeleton is available. The lower pharyngeal bone is relatively stout, the dentigerous surface noticeably broader than long (i^ to ij times) . The teeth are stout and cuspidate, and are arranged in from 16-20 rows, those of the two median rows being slightly coarser than the others. Vertebral counts in 7 specimens are : 13+15 (f.i), 13 + 16 (f.5) and 14 + 16 (f.i), giving totals of 28-30. Coloration. Live colours are unknown. Preserved coloration : Males (adult and sexually active, but probably not ripe). Ground colour light yellow-brown, with a silvery underlay on the flanks. Belly, chest and branchiostegal membrane dusky, the branchiostegal membrane darkest below the opercular region. On the flanks there are faint traces of 4 or 5 fairly broad vertical bars which become very faint dorsally and ventrally ; the anterior 3 or 4 bars merge ventrally with the dark belly coloration. A dark lachrymal stripe is present ; it does not reach the ventral margin of the preorbital, but appears to pass upwards across the eye. Dorsal fin yellowish, the soft part maculate in some specimens ; also in some fishes the lappets are black. Caudal yellowish marginally and on the distal third to quarter, dark yellow-brown proximally. Anal yellowish, with one large greyish to whiteish ocellus faintly outlined in black. Pel vies sooty, the colour less intense between the last two or three rays. Female (adult and quiescent) . Ground colour greyish-silver. A very faint lachry- mal bar extends from below the orbit to below the posterior tip of the maxilla ; a dark spot on the upper part of the eye suggests that this bar may pass across the eye. All fins are yellowish, the dorsal with dusky lappets, and the pelvics with a faint duskiness over the anterior rays. Ecology. Habitat. The species has been caught in two localities only. In one, a sheltered bay, the water was between 20 and 30 feet deep, and the substrate of organic mud. The second locality was at a depth of about 80 ft in the Buvuma Channel, near Buvuma Island ; again the substrate (on which the nets were set) was organic mud. Food. Regrettably, the gut was empty in all except one specimen ; this fish contained only a little, unidentifiable sludge. Breeding. Apart from the sexually inactive female (132 mm. S.L.), all the speci- mens are adult and active males. Affinities. The stout, unicuspid outer teeth with sharply recurved crowns immediately suggest affinity with H. sauvagei, H.prodromusa.ndH.granti (Greenwood, 1957). Like these species, the dorsal head profile of H. apogonoides is strongly rounded. However, unlike these species, there are fewer rows of inner teeth in H. apogonoides, there is a distinct interspace between the inner and outer tooth rows, and the lower jaw is much longer (45-0-51-0, M =47-9% head cf. 30-6-37-7, M = 34-5% for H. sauvagei, 30-5-37-8, M = 34-3% for H. prodromus, and 22-2-30-6, io8 P. H. GREENWOOD M = 26-8% for H. granti). Haplochromis apogonoides also differs from these species in its broader snout, and from H. granti in the anatomy of the jaws (see Greenwood, 1957). Superficially, H. apogonoides resembles H. cronus (Greenwood, 1959) and, to a lesser degree, H. empodisma and H. michaeli (see Greenwood, 1960 and 1962 for the species respectively). From H . cronus, it is differentiated by its dental morphology and its longer lower jaw. From H. empodisma, H. apogonoides differs in its longer lower jaw (45-0-51-0, M = 47-9 per cent of head, cf. 39-1-48-7, M = 43-9%), dental morphology (strongly cf. gently curved teeth), broader interorbital (27-5-31-0, M = 29-5% head, cf. 20-6-28-6, M =24-3), broader snout, and noticeably wider dentigerous surface on the lower pharyngeal bone( if-ij times as broad as long, cf. longer than broad). Haplochromis apogonoides differs from H. michaeli in dental morphology (see above), in its slightly deeper body, shallower preorbital, broader interorbita , markedly broader snout, and deeper cheek (27-8-31-0, M = 29-2% of head, cf. 22-9-27-7, M = 25-8%). The lower pharyngeal bone is similar in both species, but the teeth in H. apogonoides are stouter. The marked similarity between the outer jaw teeth of H. apognoides and those in species of the H. sauvagei complex (which includes H. xenognathus, a species not mentioned above because of its distinctive jaw morphology, tooth pattern, and head shape) suggests that H. apogonoides might be an off-shoot from this species group. Unfortunately, nothing is known about the feeding habits of this species ; all members of the H. sauvagei group are snail-eaters with the unusual habit of removing the snail from its shell before ingesting it. The large mouth and not especially strong jaws of H. apogonoides, do not, however, suggest similar feeding methods ; rather, these characters indicate piscivorous habits. STUDY MATERIAL AND DISTRIBUTION RECORDS Museum and Reg. No. Locality Collector UGANDA B.M. (N.H.) 1966.2.21.3 Ekunu Bay .... E.A.F.R.O. (Holotype) B.M. (N.H.) 1966.3.9.238-242 Ekunu Bay .... E.A.F.R.O. LAKE VICTORIA B.M. (N.H.) 1966.3.9.236-237 Locality unknown . . . E.A.F.R.O. DISCUSSION Phytogeny In an earlier paper (Greenwood, 1962) I outlined the various morphological trends shown by the piscivorous species, and suggested a possible ancestral morphotype (represented today by H. brownae) from which the different lines could have evolved. Also in that paper I indicated two major, and two minor, possibly phyletic groups of fish-eating predators. Additional information provided by the species described A REVISION OF THE LAKE VICTORIA HAPLOCHROMIS SPECIES 109 above does not affect the suggested trends, but does alter the phyletic picture. It now seems likely that there are three major phyletic lines, and probably three minor ones as well. As mentioned before (Greenwood, op. cit.}, possible phyletic lines amongst pisci- vorous species are less readily detected and defined than those of other trophic groups. In these latter there are dental as well as somatic characters which may be used for this purpose, but amongst the piscivores (at least when adult) the teeth are invariably unicuspid and of a very similar form. The principal " group " characters in these fishes are neurocranial shape, and body form ; the latter character often shows a greater or lesser degree of intergroup convergence. Using these two character complexes, I at first recognized two species aggregates, the " sen anus " group, and the " mento-macrognathus " group. The former comprises the supposedly more generalized, broad-headed and deep-bodied species, and the latter group the more specialized, slender-bodied forms. Information obtained from the species described in this paper suggests that my " mento-macrognathus " group consists of two groups, neither of which is as readily defined as the " serranus " group, but both being easily distinguished from that group. One of the minor groups (that of H. percoides, H. flavipinnis and H. cavifrons) is now thought to be polyphyletic (see p. 113). For convenience the groups will be referred to by the trivial epithet of a con- stituent species, which species, however, is not necessarily to be considered a " typical " member of the group. Indeed it is often difficult to determine just what a " typical " species would be ; at the most, the nominate species of a group is representative of a structural type found in three or four species of the group. The three major groups are : (i) The " serranus " group, consisting of H. serranus, H. victorianus, H. nyanzae, H. spekii, H. maculipinna, H. boops, H. thuragnathus and H. pachycephalus : morphologically, this is a relatively homogeneous group (but, see also p. no). (ii) The " altigenis " group, comprising H. guiarti, H. bayoni, H. dentex, H. pseudo- pellegrini, H. altigenis, H. pellegrini and H. dichrourus ; a number of subgroups (some monotypic) can be recognized, and are discussed later. (iii) The " prognathus " group, comprising H. paraguiarti, H. acidens, H. prog- nathus, H. bartoni, H. estor, H. gowersi, H. mento, H. mandibularis , H. macrognathus , H. longirostris and H. argenteus. This, the largest species aggregate shows several intragroup trends of which the most distinctive are the H. longirostris H . argenteus, and the H. mandibularis H. macrognathus subgroups. Members of the "serranus" group differ from those of the other two groups in having shorter snouts* (one species out of seven with the snout > 36 per cent of head length, cf. fourteen species out of eighteen), deeper bodies (one species out of seven with the body depth < 36 per cent of standard length, cf. seventeen out of eighteen), and broader heads (no species with the interorbital width < 22-6 per cent of head, cf. thirteen out of eighteen species.) Osteologically, the group is character- * The figures given are derived from those for the mean value of a particular character in species of the groups under consideration. ZOOL. 16, 2. no P. H. GREENWOOD ized by having a neurocranial shape nearest that of the generalized skull (see Greenwood, 1962), but with the preotic part elongate. The dorsal skull roof is straight and slopes fairly steeply, is broad both interorbitally and across the otic region, and the supraoccipital crest is high and presents a substantial area for muscle insertion (see fig. 25 in Greenwood, op. cit.}. The " serranus " group could have evolved directly from an H. brownae-like ancestor, the principal morphological changes being an increase in adult size, and those alterations in neurocranial proportions already mentioned. Within the group, the most differentiated species are H. hoops, H. thuragnathus and H. pachycephalus, all three being confined to deep water (see pp. 49, 50 and 41) . Haplochromis boops and H . thuragnathus were apparently derived from an H. maculipinna-like. ancestor, whilst H. pachycephalus seems to show greater affinity with the H. serranus H. spekii level of organization. Haplochromis maculipinna is also essentially of this affinity, but has markedly larger eyes. In turn, H. serranus is clearly derived from an H. brownae-like stem. The " altigenis " and " prognathus " groups probably evolved from an H. guiarti- like ancestor or ancestors, the latter species also showing affinities with H. brownae. Although both the "altigenis" and "prognathus" groups have included species with a relatively deep body, the main trend shown by both groups is towards a slender, somewhat compressed body-form and a correlated head shape. It is difficult to characterize these two groups, particularly their more basic members. The most trenchant group characters are probably in neurocranial form. In lateral view there is little to differentiate the skull form in the two groups ; both have the preorbital part relatively more elongate than in the " serranus "-type skull, the dorsal profile slopes upward at a rather slight angle (as compared with the angle in skulls of the " serranus " group), and the supraoccipital crest is relatively low. (These generalizations must, however, be modified somewhat for those species which seem to be structurally basal for the groups [H. guiarti and H. pseudopellegrini for the " altigenis " group, and H. paraguiarti and H. acidens for the "prognathus " group]. In these species the preorbital face is less protracted and consequently [since relative neurocranial depth varies little amongst all members of both groups] the dorsal skull roof slopes more steeply and the supraoccipital crest is higher and has a fairly extensive area). However, when the neurocranium is viewed dorsally, a difference between the groups (including their basal species) is apparent (see text-fig. 24). In members of the " altigenis " group, the otic region is relatively broader than in those of the " prognathus " group so that the outline narrows more rapidly (from a point immediately behind the orbit) than in " prognathus " skulls. In these the outline is that of a narrow wedge with the margins closing gradually from a point further behind the orbits. In supposedly basal members of both groups, the otic region is of about equal relative breadth but basal " prognathus " members nevertheless have a more gradual medial inclination of the lateral margins. Further- more, in these species the dorsal skull profile is straighter than in the most basic " altigenis " group member, H. guiarti. Indeed, in most members of the " altigenis " group the profile is more curved than in species of the " prognathus " group. Reasons for considering H. guiarti as a basic morphotype in the radiations of A REVISION OF THE LAKE VICTORIA HAPLOCHROMIS SPECIES in U FIG. 24. Outlines of the dorsicranium in species of : line A, the " serranus-group ", line B, 4-1, , " ^//^^Vc-o-rrmr. " anrl line C the " broE'nathus-erou'p" . the " altigenis-group ", and line C, the " prognathus-group ZOOL. 15, 2. 7 H2 P. H. GREENWOOD predatory piscivores have been discussed already (see Greenwood, 1962). Starting from a species similar to H. guiarti, the principal morphological changes seen amongst species of the " altigenis " group involve relative elongation of the pre- orbital face, a slight decrease in skull height (especially in the otic region) and a consequent flattening in the slope of the dorsal skull profile. In these respects H. bayoni represents a fairly marked departure from the basic " guiarti " skull form, but one less marked than that shown by H. dentex, H. altigenis or H. pellegrini, particularly the latter (see fig. 25 in Greenwood, 1962). A link between these forms is provided by the skull of H. pseudopellegrini which is intermediate between the " bayoni " and " altigenis " types. There is not a great deal of difference in body form or jaw morphology among members of the " altigenis " group. Perhaps the most extreme member is H. altigenis itself, a species with the deepest head and con- sequently the largest mouth. Haplochromis pellegrini is, because of its small adult size and relatively small mouth, atypical for the group. There is far greater variation in body form and cranial morphology within the " prognathus " group. Here the basic species, H. paraguiarti and H. acidens, are morphologically similar to H. bayoni of the " altigenis " group and could be derived either from a " bayoni " or a " guiarti "-like ancestor. Haplochromis acidens is, of course, peculiar in that despite its predatory facies, it is apparently a phytophage (see p. 76). Haplochromis prognathus and H. bartoni are, in most respects, very similar to one another and represent the next morphological step in the evolution of such species as H. longirostris and H. mandibularis from a " paraguiarti "-like stem. That is to say, the neurocranium clearly shows narrowing and elongation, and there is a related refinement of body proportions. Haplochromis estor and H. gowersi continue this trend with, in addition, the development of a more oblique angle to the jaws, and in H. gowersi a deepening of the cheek which contributes to a larger buccal cavity. Haplochromis mento probably evolved from a "prognathus "-like ancestor, and shows many of the trends exhibited by H. gowersi and H. estor. However, in this species the mouth is almost horizontal, and the neurocranium is somewhat nearer that of H . prognathus. Also apparently stemming from a " prognathus "-like ancestor is H. mandibularis. Here the trend is towards further narrowing of the skull, deepening of the cheek, increased obliquity of the jaws and lengthening of the lower jaw. This trend culminates in H. macrognathus (see Greenwood, 1962, pp. 180 and 186), a species which could well be a direct descendant of H. mandibularis. A third derivative from a "prognathus " or " bartoni "-like ancestor is H. longi- rostris. In this line there has been little change in neurocranial shape but there is a marked increase in jaw obliquity (much greater, too, than in the " mandibularis "- " macrognathus " line), and a marked refinement in body proportions which results in one of the most slender bodies found amongst Lake Victoria Haplochromis species. These characters are shared by H. argenteus, although the elongate body-form seems less obvious in that species. Haplochromis argenteus could be derived either from a " longirostris "-like stem, or perhaps directly from a " prognathus "-like ancestor. Thus, the " prognathus " group seems to show at least three radiations from a basal " prognathus "-" bartoni " stem, viz. the H. estor, H. gowersi, H. mento sub-group, A REVISION OF THE LAKE VICTORIA HAPLOCHROMIS SPECIES 113 and the H. mandibularis-H . macrognathus , and H. longirostris-H . argenteus species pairs (but see above for possible reservations about the last named pair) . It seems reasonable to assume that the H. prognathus level of organization was derived from a level similar to that shown by the extant species H. paraguiarti. The relationships of that species are, however, less clear-cut. As was mentioned above, H. para- guiarti shows several structural affinities with H. bayoni, a species probably derived from an H. guiarti-like ancestor. But, it is difficult to overrule the possibility that H. paraguiarti evolved independently from an H. brownae-\ike stem. (The status of H. brownae in relation to the piscivorous predators is discussed in Greenwood, 1962). Two species, H . plagiostoma and H. xenostoma, have not been included in the dis- cussion so far. Both are characterized by having the mouth set at a very steep angle to the horizontal (ca. 40-5o) ; H. plagiostoma is further characterized by its obliquely truncate caudal fin, and H. xenostoma by its pronounced prognathism. Because of these characters, especially the oblique jaws, neither species shows any obvious superficial similarity with members of the groups discussed above. The neurocranium of H. plagiostoma is of the " serranus " type (see also Greenwood, 1962) but also shows certain " guiarti "-like features. Thus, on this character complex H. plagiostoma could either be associated with the " serranus " group or be looked upon as an isolated offshoot of the " altigenis " group arising from an ancestor near the stem of that complex. In either eventuality, H. plagiostoma is not linked with the basal group by any extant, structurally intermediate species. Haplochromis xenostoma, both in its overall organization and in its neurocranial form, seems to represent a further development from a " plagiostoma " stem. The chief trend involves an increase in adult size, and a relative enlargement of the mouth and jaws. The neurocranium of H. xenostoma reflects these changes, especially in the longer preorbital region ; it is thus essentially a " serranus " group neurocranium. From the available evidence it is impossible to determine whether H. plagiostoma and H. xenostoma are part of the same phyletic lineage or whether the two species are end-points of parallel evolution from " guiarti " and " serranus "-like stems respectively. The two smaller species complexes, the H. michaeli, H. martini, H. squamulatus group and the H.percoides, H . flavipinnis , H. cavifrons group, will now be considered. The relationships of H. michaeli and H. martini are discussed in my paper of 1962, and there is nothing further to add. Haplochromis squamulatus is included with these species because of its several similarities with H. martini, similarities which include a basically similar coloration of a type not otherwise found in species of the Lake Victoria Haplochromis flock. In the "michaeli" group, the relationship between H. martini and H. squamulatus seems to be closer and more direct than that between either species and H. michaeli (see also Greenwood, 1960, pp. 245-8 ; idem, 1962, p. 206, and p. 91 above). When discussing the relationships of the H. percoides-H. cavifrons group (Green- wood, 1962), I suggested that H. percoides could have evolved from an H. pellegrini- like stem, and that H. flavipinnis was a derivative of an H. percoides-like ancestor. Also in that paper I noted the less certain relationships of H. cavifrons, but considered it to be part of the H. percoides phyletic line. On revising this complex, I began to H4 P. H. GREENWOOD doubt my earlier conclusions about the affinities of H. cavifrons, which now seems to have greater relationship with the " serranus " group. This review provided no evidence to negative my conclusions about the interrelationships of H. percoides and H. flavipinnis, or the association of H. percoides with an H. pellegrini-\ike ancestor. However, I am not at all certain about the validity of my previous suggestion that the neurocrania of H. percoides and H. flavipinnis show affinity with those of H. bartoni and H. longirostris (i.e. with the " prognathus " group as it is now denned). With a better appreciation of neurocranial form in the piscivores as a whole, I now think that the skull of H. percoides is basically an " altigenis " group one, as is the skull of H. pellegrini. Similar re-evaluation leads me to think that the syncranial organization and form shown by H. cavifrons links that species more closely with the " serranus " group than with the " altigenis " group and its H. percoides-like off-shoots. The freckled coloration of H. cavifrons remains unique (Greenwood, 1962), and nothing ap- proaching it is seen in the " serranus " group. If H. cavifrons is a " serranus "-group derivative, then it stands in the same relationship to that group as does H. plagio- stoma, namely as an isolated off-shoot without any extant intermediates bridging the gap. Thus, the H. percoides-H. cavifrons " group ", although a valid one on grounds of general similarity between the constituent species, is probably polyphyletic in origin. Because so little material or information is available on three species described above (H. tridens, H. orthostoma, and H. parorthostoma) , they cannot as yet be in- cluded in a discussion on phylogeny. Haplochromis tridens does not seem to be re- lated to any of the piscivorous species groups ; its affinities probably lie with the complex of small, bottom-living species which form the bulk of trawl catches in sheltered, mud-bottomed areas of the lake. Haplochromis orthostoma and H. paro- thostoma are closely related to one another but cannot readily be associated with any other species. Leaving for the moment those species which as adults feed on post-larval fishes, consideration will be given to species which prey on larval and embryo fishes, the paedophagous species. Only one paedophage, H. barbarae, is described in this paper ; the others are dealt with in an earlier publication (Greenwood, 1959) . Haplochromis barbarae resembles small specimens of H. guiarti and adults of H. brownae, and does not show any close similarity in body form, jaw morphology or dentition with other paedophagous species ; unfortunately it has not been possible to study its osteology in detail. The probable phylogeny of the larval and embryo fish eating species was discussed in the paper cited above ; the conclusion reached was that the group had a polyphyletic origin. Little more can be added to these thoughts, except to reconsider the possible relationships existing between H. par- videns and H. acidens (previously misidentified as H. nigrescens; see above, p. 73). The morphology of the lower jaw in H. parvidens differs considerably from that of H. acidens (as does the diet, paedophage cf. herbivore) ; but, in other syncranial characters, and in body-form, the species are very similar. In these latter characters the two species are more similar than are H. microdon and H . cryptodon, the only A REVISION OF THE LAKE VICTORIA HAPLOCHROMIS SPECIES 115 known paedophages showing a morphotype which could be ancestral to that of the " parvidens " level. Haplochromis cryptodon could have evolved from an H . brownae- like stem (possibly one like H. barbarae which had already adopted paedophagous habits?), the chief morphological changes involving the dentition (reduction), jaw form (to give greater distensibility) and a differential growth of the preorbital neuro- cranium (also leading to greater jaw motility). The changes in neurocranial form would lead to a skull essentially like that found in basal species of the " altigenis " and " prognathus " groups, in other words one like that in H. acidens. Thus, the origin of the " parvidens " structural grade is equivocal. Reconsidering the relationship of this grade (as represented by H. cryptodon, H. parvidens and H. microdon) with the other distensibly-mouthed grade of paedo- phages (represented by H. obesus and H. maxillaris), I can find, as before, few reasons to support a close phyletic linkage between them. Osteologically neither H. obesus nor H. maxillaris has a dentary like that occurring in the H. parvidens group, nor in the case of H. obesus is its form one from which a " parvidens " type might evolve. The neurocranium of H. obesus is most unlike that of the "parvidens " group, but that of H. maxillaris is virtually identical with the skull of H. cryptodon. The character which most clearly distinguishes these two species from any member of the " parvidens " group is the occurrence of teeth in which the crowns are curved labially (and not buccally as is usual in unicuspid teeth). This dental character is so marked, and restricted to these two species, that I am inclined to give it con- siderable weight when speculating on phylogenies, particularly since this tooth-form seems to have no adaptive significance. If the peculiar teeth in H. maxillaris and H . obesus do indicate a fundamental relationship between the species, then their syncranial differences would suggest an independent origin from a common stem, possibly a form like H. cronus (see Greenwood, 1959). Dentally, and also in its general level of organization, H. cronus is like those generalized Haplochromis species that attain a larger adult size than most members of that group ; it could be derived from an H. empodisma-like stem (see Greenwood, 1960). Natural History Only broad generalizations can be made about the natural history of the pisci- vorous predators. To date, seventy-nine species of Lake Victoria Haplochromis (and related monotypic genera) have been revised. Of these, forty-two species can be classed as piscivores, thirty-four species preying on free-swimming fishes, and eight species on cichlid embryos and larvae presumably taken from the mouths of parent fishes. Anticipating results still to be obtained from those species as yet unanalyzed, it seems probable that the number of non-piscivorous species will be increased sub- stantially. I suspect that ultimately the number of piscivorous species will be about forty per cent of the total. These figures apply to sub-adult and adult members of the species only since no data are available on the feeding habits of younger stages. The same restrictions apply to considerations on habitat preferences among the piscivores. Piscivorous species have been found in all the localities so far sampled, but much of the deeper water in Lake Victoria remains unexplored, especially at the levels n6 P. H. GREENWOOD occupied by bathypelagic species. In order to give a general picture of habitats in the lake, these can be divided, rather crudely, into three types : (i) sheltered areas such as bays and the smaller gulfs, (ii) Exposed areas, especially wave-washed beaches, (iii) Relatively undisturbed, open, off-shore waters (undisturbed that is, relative to the rather turbulent conditions prevailing over exposed beaches), not enclosed in gulfs or bays. Such a subdivision is, in many respects, unsatisfactory because conditions like those in habitat (iii) do occur in the larger gulfs and bays. More species are found in sheltered areas (habitat [i]) than in the other two habitats ; 23 species have been recorded frequently in habitat (i), 14 from habitat (ii), and 15 from habitat (iii), with, in the habitats respectively, 2, 3, and i species occurring infrequently. However, these various species are by no means confined to a particular habitat. For example, ii species were found in both sheltered and exposed areas, 9 in sheltered and off-shore habitats (3 of these species also occurring over beaches) ; only 6 species are apparently confined to sheltered areas. Two species occur both over beaches and in the quieter off-shore areas, and only 4 are apparently confined to the latter habitat (but 3 of these species are known only from deep water) . The nature of the substrate seems to exert a less restricting influence on the piscivores than on members of other trophic groups. Twenty-four species are recorded as occurring over hard substrata (sand, shingle or rock), and 20 over a soft substrate (organic mud) ; 10 of these species are found over both kinds of substrate, 6 are thought to be confined to a soft bottom, and 13 to a hard substrate. The eight strictly paedophagous species (see Greenwood, 1959, and p. 114 above) are not included in the analysis above. These species appear to be rather more re- stricted in their distribution, particularly from the viewpoint of depth range. It seems that the paedophages are confined to the littoral and immediately sublittoral zone, and are probably restricted to sheltered bays and exposed beaches. The depth range of the other 34 piscivores is, in general, confined to water less than 60 ft. deep (with of course, the exception of H. boops, H. pachycephalus , H. thuragnathus and H. dichrourus which have been caught in water about 120 ft. deep). Most species have a wide range within these depth limits, and few if any are restricted to purely littoral areas. However, it must be emphasized that this picture may be unduly biased by sampling limitations. The horizontal distribution of the species has not been studied critically ; in water less than 20 feet deep, the spatial distribu- tion of piscivores caught in nets set to cover about the first five feet of water below the surface, and the five feet above the bottom, does not show any obvious horizontal stratification of the particular species. But, these observations were not tested statistically. From the relatively few gut analyses available it would seem that the majority of piscivores prey on other Haplochromis species, and to a much lesser extent on small cyprinid fishes (especially Engraulicypris argenteus). Some species have a mixed insect-fish diet, and there are several records of otherwise exclusively piscivorous species eating insects when these are periodically and suddenly super-abundant, as for instance after a heavy termite hatch. The paedophagous species appear to feed mostly on cichlid embryos and larvae (it is presumed those of Haplochromis species) , but insects are also recorded from the gut contents of these species. A REVISION OF THE LAKE VICTORIA HA PLOCHROMIS SPECIES 117 Information on the breeding habits of piscivores is almost non-existent ; in those cases where some data are available, the species are known to be female mouth brooders. The turbidity of the water in most parts of Lake Victoria has precluded field observations on the actual spawning sites of both predators and prey alike. There has not yet been enough collecting on a lake-wide basis to establish whether or not any of the fish-eating species has a geographically restricted distribution. For eight species this possibility can definitely be overruled, (and in ten others it seems very unlikely) but for sixteen species there are suggestions of geographical restriction because they have not been caught in some regions where, on the basis of habitat and niche suitability, they should be present. SUMMARY (1) Ten species are redescribed on the basis of new material. (2) Nine new species (H. pachycephalus, H. boops, H. thuragnathus, H. pseudopelle- grini, H. paraguiarti, H. acidens, H. barbarae, H. parorthostoma and H. apogo- noides) are described. (3) Although all these species have a general facies and dentition usually associated with piscivorous habits, some do not belong to this trophic group ; one species (H, acidens] is apparently herbivorous. (4) The possible phyletic interrelationships of the piscivorous species are discussed. Three major morphological groups can be detected, at least on the basis of their more extreme members, but the boundaries are ill-defined. Some minor groups are also considered, as are the larval and embryo fish-eating species groups. (5) Broad summaries of the natural history of piscivorous species are given. (6) The holotype of Astatotilapia nigrescens Pellegrin is redescribed and its possible synonymy discussed. ACKNOWLEDGEMENTS I am deeply indebted to many people for their assistance in preparing this paper. To my colleagues Mr. A. C. Wheeler and Dr. E. Trewavas are due my thanks for, respectively, providing numerous and excellent radiographs, and for many profitable discussions. Through the cooperation of Dr. M. Blanc of the Paris Museum, I have been able to examine Pellegrin's type specimens and thus to settle several problems. Dr. Paul Kahsbauer of the Vienna Museum has graciously lent me Lohberger's type material ; I am deeply indebted to him for this privilege. In east Africa, Dr. M. Gee and Mr. R. Welcomme of E.A.F.F.R.O. aided my studies by proving additional material and field observations ; material collected by the Uganda Fisheries Department has provided many specimens from localities which I was unable to sample personally. The cooperation of E.A.F.F.R.O. and the Uganda Fisheries Department is warmly appreciated. n8 P. H. GREENWOOD APPENDIX The disputed identity of Astatotilapia nigrescens Pellegrin, 1909, (Bull. Soc. Zool. France, 34, 157) was mentioned on page 73. Boulenger (1915) synonymized this species with Haplochromis percoides Blgr. 1906. Regan (1922), however, resurrected the species as Haplochromis nigrescens, and included in his redescription a number of specimens which I have placed in a new species, H. acidens. Pellegrin's figure of A. nigrescens is misleading and does not convey an accurate impression of the holotype and unique specimen (Paris Museum, number 09-508). Recently, I examined this specimen, an immature male 71 mm. standard length, collected by Alluaud from the Kavirondo Gulf, Kenya. It does not agree closely with any other specimens I have handled, but is does show affinity with both H. percoides Blgr., 1906 and H . flavipinnis (Big.), 1906, especially the latter. Before considering its identity further, a redescription of the holotype will be given. mm. Proportional percentage Standard length . . . . .71-0 Depth of body ..... 24-0 33-8 standard length Length of head ..... 26-0 36-6 standard length Depth of preorbital . . . . 4-3 16-5 head length Width of interorbital . . . . 6-0 23-1 head length Length of snout ..... 8-0 30-8 head length Diameter of eye . . . . . 7-0 27-0 head length Depth of cheek ..... 6-5 25-0 head length Length of lower jaw . . . .12-0 46-2 head length Length of caudal peduncle . . . 13-0 18-3 standard length Length of pectoral fin . . . . 18-0 25-3 standard length Caudal peduncle 1-4 times as long as deep. Lower jaw slightly oblique and very slightly projecting ; twice as long as broad. Posterior tip of the maxilla almost reaching the vertical through the anterior orbital margin. Gill rakers : moderately stout, 9 on the lower part of the first gill arch. Scales : ctenoid ; lateral line with 31 scales, cheek with 4 rows (the rows short, so that the anterior part of the cheek is naked). Seven scales between the upper lateral line and the dorsal fin origin ; 7 between the pectoral and pelvic fin bases. Chest, belly and nuchal scales small. Teeth : in the outer row of both jaws relatively slender, slightly curved ; about 50 in the upper jaw. Inner teeth tricuspid or weakly tricuspid, arranged in 2 and I rows in the upper and lower jaw respectively. Coloration. The preserved colour pattern (in shades of brown) closely resembles that of H. percoides and H. flavipinnis (see Greenwood, 1962). The dorsal fin is marbled, and the caudal both marbled and maculate ; the pelvics are dark. DISCUSSION : In its general facies, and particularly its head shape, the holotype of A. nigrescens resembles both H. percoides and H '. flavipinnis , especially the former because the angle of the mouth is less oblique than in most specimens of H. flavi- pinnis. However, when morphometric characters are considered a number of A REVISION OF THE LAKE VICTORIA H A.PLOCHRO MI S SPECIES 119 differences between H. percoides and A.nigrescens holotype are apparent. In six characters (body depth, preorbital depth, interorbital width, snout length, eye dia- meter, and caudal peduncle length) the values fall outside the known range for H. percoides, and the lower jaw is relatively longer than in specimens of H. percoides of a comparable size. There is greater correspondence between these characters in A. nigrescens and H . flavipinnis, since only two (the shallower preorbital and larger eye) fall outside the range for H . flavipinnis, but there is a greater difference in head shape. Thus, A . nigrescens could be an aberrant specimen of either H. flavipinnis (differing especially in head shape) or H. percoides (numerous morphometric differences) . For the time being, however, I do not think that A. nigrescens can be formally synonymized with either species. I would prefer to recognize it as the purely nominal species Haplochromis nigrescens (Pellegrin) until more is known about the range of variation in small specimens of H. flavipinnis, the species I think it most closely resembles. REFERENCES GILCHRIST, J. D. F., & THOMPSON, W. W. 1917. The freshwater fishes of South Africa. Ann. S. Afr. Mus., 11 : pt. 6, 465-575. GRAHAM, M. 1929. A Report on the Fishing Survey of Lake Victoria, 1927-1928, and Appendices. Crown Agents, London. GREENWOOD, P.H. 1956. The monotypic genera of cichlid fishes in Lake Victoria. Bull. BY. Mus. nat. Hist., Zool. 3 : No. 7, 295-333. - 1957. A revision of the Lake Victoria Haplochromis species (Pisces, Cichlidae) Part II. Bull. Br. Mus. nat. Hist., Zool. 5 : No 4, 76-97. - 1959. A revision of the Lake Victoria Haplochromis species (Pisces, Cichlidae) Part III. Bull. Br. Mus. nat. Hist., Zool. 5 : No. 7, 179-218. 1960. A revision of the Lake Victoria Haplochromis species (Pisces, Cichlidae) Part IV. Bull. Br. Mus. nat. Hist., Zool. 6 : No. 4, 227-81. - 1962. A revision of the Lake Victoria Haplochromis species (Pisces, Cichlidae) Part V. Bull. Br. Mus. nat. Hist., Zool. 9 : No. 4, 139-214. 1965. The cichlid fishes of Lake Nabugabo, Uganda. Bull. Br. Mus. nat. Hist., Zool. 12 : No. 9, 313-57. - 1966. Two new species of Haplochromis (Pisces, Cichlidae) from Lake Victoria. Ann. Mag. nat. Hist., (13), 8 : 303-318. TREWAVAS, E. 1964. A revision of the genus Serranochromis Regan (Pisces, Cichlidae). Ann. Mus. Roy. Afr. Cent. Tervuren, ser in 8, Zool. No. 125, 1-58. PRINTED IN GREAT BRITAIN BY ADLARD & SON LIMITED BARTHOLOMEW PRESS, DORKING / 27 FEB1967 THE VACHELL COLLECTION OF CHINESE FISHES IN CAMBRIDGE P. J. P. WHITEHEAD & K. A. JOYSEY BULLETIN OF THE BRITISH MUSEUM (NATURAL HISTORY) ZOOLOGY Vol. 15 No. 3 LONDON: 1967 27 FEBH THE VACHELI^ COLLECTION OF CHINESE \^ FISHES IN CAMBRIDGE BY P. J. P. WHITEHEAD (British Museum, Natural History) "\ and K. A. JOYSEY ( University Museum of Zoology, Cambridge) Pp. 121-165 ; 3 Plates BULLETIN OF THE BRITISH MUSEUM (NATURAL HISTORY) ZOOLOGY Vol. 1 5 No. 3 LONDON: 1967 THE BULLETIN OF THE BRITISH MUSEUM (NATURAL HISTORY), instituted in 1949, is issued in five series corresponding to the Departments of the Museum, and an Historical series. Parts will appear at irregular intervals as they become ready. Volumes will contain about three or four hundred pages, and will not necessarily be completed within one calendar year. In 1965 a separate supplementary series of longer papers was instituted, numbered serially for each Department. This paper is Vol. 15, No. 3 of the Zoological series. The abbreviated titles of the periodicals cited follow those of the World List of Scientific Periodicals. World List abbreviation : Bull. Br. Mus. nat. Hist. (Zool.) Trustees of the British Museum (Natural History) 1967 TRUSTEES OF THE BRITISH MUSEUM (NATURAL HISTORY) Issued 24 February, 1967 Price i is. THE VACHELL COLLECTION OF CHINESE FISHES IN CAMBRIDGE By P. J. P. WHITEHEAD & K. A. JOYSEY CONTENTS Page 1. INTRODUCTION .......... 123 2. RICHARDSON'S " REPORT "........ 123 3. THE VACHELL COLLECTION ........ 124 4. SYSTEMATIC LIST ......... 127 5. EXTANT VACHELL SPECIMENS ....... 147 6. REFERENCES .......... 158 7. INDEX ........... 160 ABSTRACT The Rev. G. Vachell of Macao presented about a hundred Chinese fishes to the Cambridge Philosophical Society in the early part of the last century. In his Report on the Ichthyology of the Seas of China and Japan, Richardson based 22 new species on Vachell material. Although 80 species are listed here, the greater part of the Vachell collection is no longer extant and only 15 specimens survive ; these include the types of Anguilla clathrata Rich., Pelor tigrinum Rich., and Seserinus vachellii Rich. i. INTRODUCTION SOME i8th and igth century fish collections are well-known and their contents well documented. Others have lapsed into obscurity and provide considerable difficulties when the need arises to establish types or validate old names. One of these is the Vachell collection of Chinese fishes, the remains of which are housed in the University Museum of Zoology in Cambridge. The Vachell collection, which appears to have contained about 80 species of fish from Macao, formed a significant part of the material on which Sir John Richardson based his " Report on the Ichthyology of the Seas of China and Japan " published in 1846. As often happens with old collections, the Vachell material included a large proportion of types. Thus 22 of the new species described by Richardson in the " Report " were based on Vachell fishes. Unfortunately, only a fraction of the Vachell collection is now extant. But this collection is of sufficient importance for a complete list to be given, particularly in order to record those specimens which survive (including the types of 3 Richardson species) those which are missing (including 7 types), and those specimens which are now known with certainty to have been destroyed (including 12 types). The principal author, P.J.P.W., is responsible for all the systematic zoology, and K.A.J. undertook the investigation of the records in Cambridge. 2. RICHARDSON'S "REPORT" Richardson (1846) listed 665 species of fishes from the seas of China and Japan, an enormous increase on any previous list ; 142 of these were described as new species or varieties. Over three hundred of the species listed were represented by an original ZOOL. 15, 3 8 124 P. J- P- WHITEHEAD & K. A. JOYSEY coloured drawing from a collection of illustrations of fishes from Macao and Canton compiled by John Reeves (1774-1856), an Inspector of Tea for the East India Company at Canton ; 83 new species introduced by Richardson were based solely on a Reeves illustration. Three sets of these illustrations are now in the Zoology Library of the British Museum (Natural History). Since half (40 out of 80) of the species listed here (including n of the types) are illustrated by a Reeves drawing, the Reeves illustration number is cited in the synony- mies. This may assist in identifying some of the missing Vachell specimens. The illustrations of the types are reproduced here, apparently for the first time (except Chatoessus maculatus see Whitehead, 1966). John Reeves, and his son J. R. Reeves, also sent a collection of fishes from Macao and Canton to the British Museum. These were examined by Richardson and Giinther, and since they came from the same area as the Vachell fishes, reference is made to them in the text. Richardson evidently examined the Vachell collection in the years up to 1845. Four of the species listed here (Batistes vachellii, Anguilla ctathrata, Aploactis breviceps, and Congrus lepturus) had already been described as new by Richardson in the " Zoology of the Voyage of H.M.S. Sulphur " published in three parts between 1844 and 1845 (see " Report " p. 316 for dating). In general, however, Richardson left description of the Vachell material until the " Report ". 3. THE VACHELL COLLECTION According to Richardson's " Report " (1846, p. 189), the Rev. George Vachell was Chaplain to the India Company at Macao in about 1830. He made a collection of about a hundred fishes from that region, and these were presented to the Philosophical Institution in Cambridge, preserved in spirits, and mostly in good condition. None- theless, Richardson himself described at least one fish as " not in very good condition " (p. 204), another that " the colours have suffered from long maceration in spirits " (p. 208), and another that " it is flaccid and may have lost its exact shape " (p. 267). In 1865 the collections of the Cambridge Philosophical Society were transferred to the University of Cambridge and housed in the newly built Museum of Comparative Anatomy and Zoology. Indeed, the Philosophical Society Collection formed the nucleus of the Museum of Zoology, and although under the same roof as the Museum of Comparative Anatomy, the two collections were kept in separate rooms and remained distinct for several years. In 1867 J. W. Clark, who was Superintendant of the Museums, reported that, " Two presses have been provided to contain the collections of Fishes presented by the Philosophical Society. These consist of the following : the series of Fishes of Madeira, collected by Mr. Lowe, many of which are unique, and nearly all the types of his descriptions in the transactions of the Cam- bridge Philosophical Society : a considerable number of specimens procured by Mr. Darwin during the voyage of the ' Beagle ', also the types of the descriptions published by the late Sir John Richardson : a collection sent from China by Mr. Vachell ; and a collection of Fishes of Great Britain formed by the late Professor Henslow and Mr. L. Jenyns. These will at no distant period be named and catalogued by Dr. Giinther, of the British Museum, who has most generously offered to undertake VACHELL COLLECTION OF CHINESE FISHES 125 this work ". (It should be noted that the punctuation of this passage suggests that Clark did not realize that the Richardson types were included within the Vachell collection.) The following year Clark (1868) reported that during 1866-67 " Dr. Gunther, of the British Museum, one of the first ichthyologists in Europe, has been so kind as to examine the collections of Fish in spirits, and to determine those that were unnamed." The major part of Gunther 's " Catalogue " had by then been published, but Gunther makes only rare reference to specimens in Cambridge in the subsequent volumes (7 and 8), and no mention at all of having seen Vachell fishes. It is probably for this reason that the Vachell collection has been generally ignored. A few years later, Clark (1871) again referred to the identification of the spirit collection of Fish, Amphibia and Reptiles and reported that, " I had hoped to have announced the completion of an arrangement by which the services of Dr. A. Gunther, of the British Museum, the best living authority on the subject, might have been secured for this work ; but though the proposal made by the Museums' Syndicate to the Trustees of the British Museum was a most liberal one, that body did not think proper to accede to it. It will therefore be necessary to select some other competent person ". Apparently this statement had the desired effect and only a year later Clark (1872) reported that these collections had been examined and determined by Dr. Gunther! It is clear that in both 1866-67 an d 1870-71, Gunther worked on the fish collections in Cambridge. Shipley (1913) even claims (p. 265) that the fishes were catalogued by Dr. Gunther, but we have been unable to trace any other record of such a catalogue, either in London or in Cambridge. In 1893, S. F. Manner catalogued the entire fish collection at Cambridge, indexing the species according to the volume and page numbers of Giinther's " Catalogue". The names used by Harmer are generally identical to those of Gunther, but in the case of the Vachell specimens the names often differ from those given in Richardson's " Report ". This suggests that Gunther re-identified and relabelled many of the Vachell specimens, despite the fact that he did not refer to them in the " Catalogue ". Altogether, 80 species (plus 3 names here considered synonyms) are listed as appearing either in Richardson's " Report " or in Harmer's catalogue, and often in both. The vast majority are now either missing or known to have been destroyed. A total of 32 species which were listed by Richardson as including Vachell speci- mens are not so listed by Harmer (including the types of 10 Richardson species). These specimens may have been lost, exchanged or destroyed without record either in the Cambridge Philosophical Institute between 1845 and 1865, or in the Museum of Comparative Anatomy and Zoology between 1865 and 1893. It is possible that Richardson worked on some of the Vachell material at the British Museum, that it was never returned to the Philosophical Institution, and that it became incorporated into the British Museum collections without ever being registered. There are in fact a number of unregistered bottles, including some labelled as presented (or collected) by Richardson. In some cases these involve Vachell species which were never included in Harmer's Catalogue of Cambridge material, and each of these have been carefully checked. Unfortunately in no case is there supplementary evidence which would prove the case either way. 126 P. J. P. WHITEHEAD & K. A. JOYSEY It is also known that J. W. Clark spent a good deal of the summer of 1866 over- hauling the Physiological Series, which had just been moved into the newly built Museum in Cambridge, and it is recorded that useless and decayed specimens were thrown away (Clark, 1867 ; Shipley, 1913). Although this activity does not refer explicitly to the fish collection it does give a clear indication of the policy that was being implemented in the Museum during the same period that Gunther was working on the fish, and this might account for some of the 32 species which were missing before Harmer's 1893 catalogue. The possibility that some of the Vachell material was acquired by the British Museum between 1865 and 1893 has also been checked. Such an acquisition seems most likely to have occurred during or after Giinther's visit in 1866-67, or as a P ar t of the arrangement made in 1870-71, the terms of which are not known. But the British Museum registers from 1866-93 show no gifts, purchases or exchanges in- volving the Cambridge Museum. It is certain that no Vachell material went to London officially, and in some doubtful cases the British Museum material has been checked and no evidence found that any arrived unofficially. Conversely, there are 16 species listed by Harmer as being Vachell material which are not so mentioned by Richardson. Either Richardson missed these, or perhaps they were not true Vachell specimens but were erroneously included by Harmer. Among these, one Cambridge specimen of Boleophthalmus campylostomus (see p. 147) was transferred to the British Museum in 1917, but although Harmer lists it as a Vachell fish, Richardson (Report, p. 209) states that he had seen no specimens. Harmer's 1893 catalogue is in two parts, one listing the stored material and the other listing those specimens which were on exhibition at that time, the latter being given separate registration numbers. A large number of Harmer's entries in the catalogue of stored material have since been crossed out, and these specimens are no longer extant. A note in the front of the catalogue states : " The collection of fishes in store was overhauled during May, 1939, when specimens thought to be of little or no value were discarded." We understand from the present Director, Dr. F. R. Parrington, that this clearance included many dried-up, rotten and dis- integrated specimens, and he recalls that his predecessor, Sir Clive Forster-Cooper, once recounted that some of the fishes which had matured beyond repair were found to have been pickled in rum! It seems certain that those specimens which are crossed off Harmer's catalogue were destroyed, in contrast to those which are just missing insofar as they do not appear in Harmer's list. It is right and proper to ask why so little of the Vachell collection survives today, despite the fact that it was presented to an Institution which could well have been expected to provide security. Only by such enquiry can it be hoped to avoid similar disappointments in the future. In this case, there seems to have been a combination of circumstances. Richardson provides some evidence of poor fixation and preserva- tion, which continued to take its toll more than a century later. Clark apparently did not appreciate that the Vachell collection included the Richardson types, and for some inexplicable reason Giinther drew no attention to them. Hence, in the majority of cases, Harmer was unaware of the type specimens, and did not indicate them as such when he prepared his catalogue. In consequence, when the collection VACHELL COLLECTION OF CHINESE FISHES 127 was overhauled nearly 50 years later, many types were discarded. Until relatively recently it has been accepted practice in most museums that material which had suffered beyond repair was destroyed and crossed off the cata- logue, unless, of course, it was recorded as type material. In retrospect this policy is to be regretted in the present case, and this has caused us to give some thought to the problem. For one reason or another, whether it be faulty fixation, poor storage conditions, sheer neglect or genuine accident, nearly all Museums possess some important material that has suffered, apparently beyond repair. We now hold the view that those responsible for such collections must accept this as almost inevitable, and without shame should retain such material, rather than totally destroying it. Advances in technology are now providing new methods of obtaining information from such material, and although apparently useless to one generation, it may be capable of yielding information to the next. Methods now exist for reconstituting dried up specimens, and for " developing " labels which have become illegible in ordinary light. In the present instance, X-ray photography of even the rotten material might have yielded sufficient information about the skeleton to establish the identity of the species, but unfortunately none of these techniques can now be applied, because most of the material has been destroyed. 4. SYSTEMATIC LIST In the following list of Vachell species, the synonyms of Richardson and Giinthei have been given, headed by the modern name for the species. The latter have posed considerable difficulties, particularly when no specimens survive and where the identification must be made on a brief description by Richardson, sometimes supplemented by a Reeves illustration. Richardson's specimens were evidently not always consistent with the Reeves drawing (e.g. in the case of Sebastodes vachellii p. 140), a fact not always appreciated by later authors. In many cases, therefore, it has been possible to do no more than to follow the comprehensive synonymies of Chu (1931), Herre (1953), Chu, Tchang & Chen (1964) and Fowler (Synopsis of fishes of China, from 1930 onwards) ; only Chu and Fowler have listed all or almost all Richardson names. Sometimes, a recent study of a particular group has helped to identify the Vachell material, but even then the Vachell specimens themselves do not appear to have been examined, perhaps since the time of Giinther's visits to Cambridge nearly a century ago. In addition, very few authors seem to have examined the Reeves illustrations. We have listed here fifteen specimens representing eleven species which appear to be part of the original Vachell collection. Even these numbers are by no means certain since several are listed only by Harmer and not by Richardson as being Vachell material. Notes on these extant specimens and their identifications are given separately in Section 5 (p. 147). Those of Richardson's species which were based wholly or partly on Vachell material are marked with an asterisk and are listed separately in Table I. We are particularly grateful to Mr. W. L. Chan, of the Fisheries Research Station in Hong Kong, for his help in identifying the fishes shown in the Reeves illustrations reproduced here and his comments on the Chinese names applied to these fishes. 128 P. J. P. WHITEHEAD & K. A. JOYSEY Class CHONDRICHTHYES Family ORECTOLOBIDAE i. Chiloscyllium plagiosum (Bennett) Chiloscyllium plagiosum: Richardson, 1846: 194 (Vachell material) (Reeves illustr. No. 252). Chiloscyllium indicum: Giinther, 1870, 8 : 413 (material listed under six varieties). As Chiloscyllium indicum " (several) China Rev. G. Vachell " in Harmer's list destroyed, 1912 according to catalogue. A half-grown Richardson specimen and a Reeves juvenile in British Museum. Giinther does not mention any specimens in Cambridge, although he had inspected the Vachell material by this time. The identification of the three sharks listed here is based on the review of Taiwan species by Chen (1963). Family CARCHARHINIDAE 2. Carcharhinus melanopterus Quoy & Gaim. Carcharias (Prionodon) melanopterus: Richardson 1846: 194 (not stated as Vachell specimen) (Reeves illustr. No. 23). Carcharias melanopterus: Giinther, 1870, 8 : 369 (South Africa and Amboyna material only). As Carcharias (Prionodon) melanopterus, juvenile " China, Rev. G. Vachell " in Harmer's list destroyed. No Cambridge material mentioned by Giinther. Family TRIAKIDAE 3. Triakis scyllia Miiller & Henle Triakis scyllium: Richardson, 1846: 195 (no specimens). Triads scyllium: Giinther, 1870, 8 : 384 (no specimens). As Triads ? scyllium, juvenile, " China Rev. G. Vachell " in Harmer's list destroyed. Vachell specimen presumably overlooked by Richardson, if indeed it did belong to the Vachell collection. Again, Giinther makes no mention of Cambridge material. Class OSTEICHTHYES Family CLUPEIDAE 4. Clupanodon thrissa (Linnaeus) *Chatoessus maculatus Richardson, 1846 : 308 (a single Vachell specimen, TYPE) (Reeves illustr. No. 109). Chatoessus maculatus: Giinther, 1868, 7 : 409 (Formosa specimens only). Harmer lists " Clupea thrissa. (several). No histy.", but these were subsequently destroyed. Since Chatoessus maculatus was the only clupeid represented in the Vachell collection, it is possible that one of the Harmer specimens was the type. Giinther mentions only three Formosan specimens but not the Vachell material in Cambridge. For identification of C. maculatus, see Whitehead (1966). Reeves illustration reproduced here (Plate I, fig. i). VACHELL COLLECTION OF CHINESE FISHES 129 Family ENGRAULIDAE 5. Coilia playfairii (McClelland) Coilia playfairii: Richardson, 1846 : 309 (see note below). Coilia clupeoides: Giinther, 1868, 7 : 404 (" Chinese Seas ", no mention of Cambridge material). Listed as Coilia (mystus), with the name clupeoides written above, " ' Fishes from China ' Canton ", not in Harmer's hand EXTANT. A single specimen now exists, but with only a modern label indicating that it is a Canton fish. However, Richard- son states " specimens exist in all the collections of Chinese fishes that we have seen ". Giinther does not mention having seen any Cambridge material. Measurements and notes on the extant specimen are given in Section 5 (p. 149), and further notes on Richardson material are given by Whitehead (1966) . Family SALANGIDAE 6. Salanx chinensis (Osbeck) Leucosoma chinensis: Richardson, 1846 : 303 (Reeves and Vachell specimens). Salanx chinensis: Giinther, 1866, 6 : 205 (China, including type of Leucosoma reevesii Gray). Listed by Harmer as " Salanx chinensis (2) ? History "destroyed. The British Museum has a specimen from the Haslar collection, as well as the type of Leucosoma reevesii. Identification based on Chu (1931). Family ANGUILLIDAE 7. Anguilla japonica (Temminck & Schlegel) *Anguilla clathrata Richardson, 1844 : 104 (one Vachell specimen, TYPE) ; Idem, 1846 : 312. Anguilla clathrata: Giinther, 1870, 8 : 23 (doubtful species No. 4, name only). A single specimen listed by Harmer as A. vulgaris " China. Rev. G. Vachell. "- EXTANT. This specimen was subsequently transferred as F.2002 to the Exhibited series. Giinther had already examined the Vachell collection but did not acknow- ledge having seen this fish. Richardson's description (Zoology of the Sulphur, p. 104) was based on a Vachell fish of 8-8 inches from Canton. See Section 5 for description and notes on this HOLOTYPE. Family MURAENESOCIDAE 8. Muraenesox cinereus (Forsskal) Congrus tricuspidatus : Richardson, 1846 : 312 (Vachell specimens). Muraenesox cinereus: Giinther, 1870, 8 : 46 (Reeves specimens and type of C. tricuspidatus}. Listed as M, cinereus " China Rev. G. Vachell " by Harmer destroyed. Cam- bridge material not mentioned by Giinther. Identification based on Chu, Tchang & Chen (1963). 13 P. J. P. WHITEHEAD & K. A. JOYSEY Family CONGRIDAE 9. Uroconger lepturus (Richardson) Congrus lepturus Richardson, 1844 : 106, and 1846 : 132 (Canton, but no reference to Vachell specimens) . Uroconger lepturus: Giinther, 1870, 8 : 44 (Reeves specimens, the TYPE, and another Chinese specimen) . Listed as Uroconger lepturus, 2 specimens " China Rev. G. Vachell " by Harmer destroyed. Giinther does not refer to Cambridge material. The species was des- cribed by Richardson (Zoology of the Sulphur, p. 106, PL 56, figs. 1-6) with a single set of measurements relating to a specimen of 9 inches from Canton, but no indication of the collector (presumably Reeves). There is a specimen of 160 mm. S.L. (io inches) labelled as type in the British Museum (unregistered but with a metal tag " 821 ") ; the second Chinese specimen is even larger (322 mm.). The first is labelled "Reeves " and the second was purchased from a Mr. Warwick. Present identifica- tion based on Chu, Tchang & Chen (1963). Family OPHICHTHIDAE 10. Pisodonophis boro (Ham. Buch.) Ophisurus harancha: Richardson, 1846 : 313 (one fish of 14^ inches in Camb. Phil. Inst., a Reeves specimen and an Indian specimen). Ophichthys pattens: Giinther, 1870, 8 : 61 (the Reeves type only). Harmer listed 2 specimens of Ophichthys cancrivorus " China Rev. G. Vachell " destroyed. Although Richardson (1846) believed the Vachell, Reeves, and Indian specimens to be conspecific, he elsewhere (Erebus & Terror, p. 10) proposed the name pattens for the Reeves fish, while still claiming the Vachell specimen to be true harancha. Giinther (1868, p. 61) considered Richardson's harancha to be pattens, as also did Chu (1931). Fowler (19320, p. 126) agreed, but placed both under an earlier Hamilton-Buchanan name, Pisodonophis boro, and this course has been adopted here. Family BAGRIDAE IT. Pseudobagrus vachellii (Richardson) *Bagrus vachellii Richardson, 1846 : 284 (one Vachell specimen of 5 inches, the TYPE). Pseudobagrus vachellii: Giinther, 1864, 5 : 85 (Chinese specimens). Listed as P. aurantiacus " China. Rev. G. Vachell " in Harmer's catalogue destroyed. Giinther had kept the closely related P. vachellii and P. aurantiacus separate, but he may have reconsidered this when he came to identify the Cambridge material. Since Harmer listed two Vachell Pseudobagrus species, and Richardson two Vachell Bagrus species, it must be assumed that the two species correspond. Present identifications follow Chu (1931). VACHELL COLLECTION OF CHINESE FISHES 131 12. Pelteobagrus fulvidraco (Richardson) *Bagrus limbatus Richardson, 1846 : 283 (one Vachell specimen, the TYPE). Pseudobagrus fulvi-draco: Giinther, 1864, 5 : 85 (one Chinese fish presented by Giinther himself). Listed as Pseudobagrus fulvi-draco " China. Rev. G. Vachell " by Harmer destroyed. Fowler (19326) identified Richardson's species as Plotosus anguillaris (Bloch), but curiously did not list Richardson's B. vachellii. Family PLOTOSIDAE 13. Plotosus anguillaris (Bloch) Plotosus lineatus: Richardson, 1846 : 286 (specimens in the Camb. Phil. Inst.) (Reeves illustr. No. /Sn). Plotosus anguillaris: Giinther, 1864, 5 : 24 (Chinese specimens). Listed as Plotosus anguillaris " (several) Phistory " in Harmer's Catalogue des- troyed. Possibly the Cambridge Philosophical Institution had other Chinese material in addition to the Vachell specimens. Present identification based on Chu (1931) and Fowler (19326). Family BELONIDAE 14. Strongylura strongylura (van Hasselt) Belone caudimaculata: Richardson, 1846 : 264 (no Vachell material) (Reeves illustr. No. ^33). Belone caudimaculata: Giinther, 1866, 6 : 245 (no China specimens). Listed as B. strongylurus, " ? China " by Harmer destroyed. Richardson gave no description but mentioned specimens from Canton (Reeves) and from Port Essington. Giinther (1866) placed the former in B. strongylurus and the latter in B. caudimaculata. Present identification based on Fowler (19326). Family SOLEIDAE 15. Microbuglossus ovatus (Richardson) *Solea ovata Richardson, 1846 : 279 (a single Vachell specimen, 3^ inches, the TYPE) (no Reeves illustr.). Solea ovata: Giinther, 1862, 4 : 472 (Chinese specimens). Listed by Harmer under Solea ovata, 4 specimens " China Rev. G. Vachell " destroyed. Giinther (1862) lists an Amoy and a Haslar Collection specimen, and 4 fishes presented by Belcher. The latter are labelled as types, but they are too small and Richardson mentions only Vachell material. The single Haslar fish, however, may well have been 3^ inches in length (caudal now damaged) but there are 59 dorsal rays (65 described). Identification based on Chu, Tchang & Chen (1963). 132 P. J. P. WHITEHEAD & K. A. JOYSEY 16. Zebrias zebra (Bloch) *Solea ommatura Richardson, 1846 : 279 (two Vachell specimens, the TYPES) (Reeves illustr. No. 13). Synaptura zebra: Giinther, 1862, 4 : 484 (Chinese specimens). Listed by Harmer, as Synaptura zebra, four fishes, " No histy." destroyed. No British Museum material which could be the lost Vachell types. Reeves illustration reproduced here (Plate I, fig. 4). Identification follows Chu, Tchang & Chen (1963). Family CYNOGLOSSIDAE 17. Cynoglossus grammicus (Richardson) *Plagiusa grammica Richardson, 1846 : 280 (two Vachell specimens 3! inches, the TYPES) (no Reeves illustr.). Plagiusa grammica: Giinther, 1862, 4 : 492 (doubtful species No. 5, name and reference to the " typical specimen " in Camb. Phil. Soc. collection). Listed as Cynoglossus trigrammus " (several). China. Rev. G. Vachell " by Harmer (presumably having been re-identified by Giinther) destroyed. No evidence of Vachell material in British Museum. Richardson's species is ignored by Chu, Tchang & Chen (1963), and the present identification follows Fowler (1934). Family HOLOCENTRIDAE 18. Holocentrus ruber (Forsskal) Holocentrum albo-rubrum: Richardson, 1846 : 223 (Vachell specimens) (Reeves illustr. 019.) Holocentrum rubrum: Giinther, 1859, 1 : 35 (Reeves and other Chinese material). Two specimens are given in Harmer's list as Holocentrum rubrum, " China. Rev. G. Vachell." EXTANT. There are also two dry specimens listed by Harmer but stated to have " no history ". These are also extant and were presumably identified by Giinther. See Section 5 (p. 151) for notes on the two extant spirit specimens. Family CHANNIDAE 19. Channa maculata (Lacepede) Ophicephalus maculatus: Richardson, 1846 : 251 (two Vachell specimens) (Reeves illustr. Nos. 148 and /3ig). Ophiocephalus maculatus: Giinther, 1861, 3 : 480 (Reeves and China specimens). Listed as 0. argus in Harmer's list, 2 + I specimens, " China. Rev. G. Vachell " destroyed. Giinther considered 0. maculatus to be close to 0. argus, but seems to have re-identified the Cambridge material as the latter. Present identification based on Chu (1931). VACHELL COLLECTION OF CHINESE FISHES 133 Family ATHERINIDAE 20. Atherina bleekeri (Gunther) Atherina bleekeri Giinther, 1861, 3 : 398 (Reeves specimens). Listed as A. bleekeri "China. Rev. G. Vachell " by Harmer destroyed. No species of Atherina are listed by Richardson, so Manner's specimens were either overlooked or were not in fact Vachell fishes. Family MUGILIDAE 21. Mugil cephalus (Linnaeus) Mugil japonicus: Richardson, 1846 : 247 (no Vachell material). Mugil cephalus: Giinther, 1861, 3 : 419 (Reeves and Chinese specimens). Listed as M. cephalotus by Harmer, 2 specimens " China. Rev. G. Vachell " destroyed. Richardson included seven species of Mugil in the " Report ", none based on Vachell specimens. Richardson's M. macrolepidotus is another synonym of M. cephalus, according to Fowler (1935), who has been followed here. Family SPHYRAENIDAE 22. Sphyraena obtusata (Cuvier) Sphyraena chinensis: Richardson, 1846 : 266 (one Vachell specimen) (Reeves illustr. No. 62). Sphyraena chinensis: Giinther, 1860, 2 : 334 (doubtful species No. i). Not listed by Harmer. No British Museum specimens labelled S. chinensis. Present identification follows Chu (1931). Family TRICHIURIDAE 23. Leptur -acanthus savala (Cuvier) Trichiurus intermedius: Richardson, 1846 : 268 (one British Museum specimen, but no Vachell material) (Reeves illustr. No. ^56) . Trichiurus savala: Giinther, 1860, 2 : 347 (East Indian and Chinese specimens). One specimen listed by Harmer, as T. muticus " China. Rev. G. Vachell ", now transferred to Exhibited Series No. F.2685 EXTANT. Richardson placed T. muticus Gray in his synonymy of " T. lepturus, japonicus " (" Report ", p. 268), a record which he based solely on T. lepturus Temm. & Schl. from Japan. He included Trichiurus savala Cuvier in his synonymy of T. armatus Gray, but did not list any material at Cambridge. The present specimen may have been one that he overlooked. See Section 5 (p. 152) for discussion of this specimen. 134 P- J- P. WHITEHEAD & K. A. JOYSEY Family NOMEIDAE 24. Psenopsis anomala (Temm. & Schl.) *Trachinotus melo Richardson, 1846 : 270 (one Vachell specimen, the TYPE) (Reeves illustr. No. 97). Trachinotus melo: Giinther, 1860, 2 : 485 (on Richardson's description, no specimens). Not included in Harmer's list, presumed lost before 1893. The specimen is not in the British Museum collections. Reeves illustration reproduced here (Plate i, fig. 3). Identification follows Chu (1931) and Fowler (1936). Family FORMIONIDAE 25. Parastromateus niger (Bloch) *Seserinus vachellii Richardson, 1846 : 273 (two Vachell specimens, the larger 3-75 inches, the TYPES). Stromateus niger: Giinther, 1860, 2 : 401 (one Reeves specimen). Harmer lists two specimens under the name Platax teira " China. Rev. G. Vachell." EXTANT. These specimens have been identified as the types of Seserinus vachellii. See Section 5 (p. 153) for description and notes. Family CARANGIDAE 26. Alectis indica (Riippell) Gallichthys major: Richardson, 1846 : 271 (one Vachell specimen) (Reeves illustr. No. 189). Caranx gallus: Giinther, 1860, 2 : 455 (one Reeves specimen and one other Chinese specimen). As Caranx gallus in Harmer's catalogue, two fishes " China. Rev. G. Vachell " destroyed. Identification follows Fowler (1936, p. 297), who used Cuvier's generic name Scyris. 27. ? Alectis ciliaris (Bloch) Blepharis fasciatus: Richardson, 1846:271 (one Vachell specimen) (Reeves illustr. No. 269). Blepharis fasciatus (non Riipp.): Giinther, 1860, 2 : 422 (doubtful species No. 19, typical specimen in Camb. Phil. Inst.). Not listed by Harmer, presumed lost before 1893. Richardson identified his specimen with B. fasciatus Riippell, a synonym of A. ciliaris (Bloch), but Giinther disagreed (1860, p. 454), presumably on Richardson's description. Not in British Museum collections. Tentative identification of Fowler (1936, p. 295) followed here. 28. Caranx (Atule) kalla (Cuvier) * Caranx cancroides Richardson, 1846 : 274 (one Vachell specimen, the TYPE) (Reeves illustr. No. 30). Caranx cancroides: Giinther, 1860, 2 : 422 (doubtful species No. 12). Not mentioned in Harmer's list, presumed lost or destroyed before 1893. Reeves VACHELL COLLECTION OF CHINESE FISHES 135 illustration reproduced here (Plate I, fig. 2). W. L. Chan (in litt.} states that in Hong Kong the vernacular name Ha-tsee on the Reeves illustration (Hwa tsze in the " Report ") refers (prefix) to shrimp, and (suffix) to " a carangid fish either typically of the genus Decapterus, or of the subgenus Atule Jordan." Decapterus can be ruled out as being too slender (depth about 5 or more times in total length ; about 3 in Richardson's description and in the drawing). Caranx (Atule) kalla Cuv. is the most likely species and one that is caught in fair numbers by the Hong Kong shrimp trawlers. Richardson states " No spots are shown on the operculum " (present in C. kalla), but the figure shows a fish of only 4^ inches. 29. Caranx malabaricus (Bloch & Schneider) Caranx malabaricus: Richardson, 1846 : 275 (two Vachell specimens) (Reeves illustr. No. /?2i). Caranx malabaricus: Gunther, 1860, 2 : 436 (one Reeves and one other Chinese specimen ; BMNH. 1851.12.27.118). Not listed by Harmer, presumed destroyed or lost before 1893. An unregistered specimen merely labelled " Caranx malabaricus " in British Museum collection, but no indication of donor. Identification based on Fowler (1936, p. 293), who placed the species in Carangoides. 30. Citula armata (Forsskal) Caranx ciliaris: Richardson, 1846 : 276 (" spec. C. Ph. Inst."). Caranx armatus: Gunther, 1860, 2:453 (one Chinese specimen, BMNH. 1851.12.27.129, no donor given). Listed by Harmer as Caranx armatus " China. Cambridge Philosoph. Society's Collection " EXTANT, Exhibited Series No. F.2755. There is also an unregistered British Museum specimen with an old label " Caranx ciliaris " amended to " arma- tus ". The jar is unusual, being oval in cross-section and sealed with parchment ; such jars were initially suspected of being part of the Vachell collection, but the present case shows this to be incorrect. The Cambridge specimen is discussed further in Section 5 (p. 154). Family LEIOGNATHIDAE 31. Leiognathus brevirostris (Valenciennes) Equula nuchalis: Richardson, 1846 : 276 (two Vachell specimens) (Reeves illustr. Nos. Ggo and 85). Equula nuchalis: Gunther, 1860, 2 : 500 (Reeves and other Chinese specimens). Not included in Harmer's list. Reeves illustration No. /? 85 is not a leiognathid, but No. G 90 is almost certainly L. brevirostris, having a distinct black pre-dorsal blotch, a body depth 21/5 times in S.L., and the lateral line not reaching the base of the caudal. Richardson gives no description. 136 P. J. P. WHITEHEAD & K. A. JOYSEY Family GERRIDAE 32. Gerres erythrourus (Bloch) Gerres equula: Richardson, 1846 : 239 (one Vachell fish) (Reeves illustr. No. 215). Gerres oyena: Giinther, 1859, 1 : 352 and 1862, 4 : 261 (no Chinese specimens). Not included in Harmer's list. Identification follows Chu (1931). Family AMBASSIDAE 33. Ambassis commersonii (Cuvier) *Ambassis vachellii Richardson, 1846 : 221 (one Vachell specimen, the TYPE) (no Reeves illustr.). Ambassis vachellii: Giinther, 1859, 1 : 227 (no British Museum specimens). Listed as Ambassis commersonii, " ? China Rev. G. Vachell " in Harmer's catalogue, two specimens destroyed. Fowler (1937) tentatively recognized Richardson's species because of its deep body, but without specimens or illustration the species must remain doubtful. Family SERRANIDAE 34. Epinephelus fario (Thunberg) Serranus trimaculatus : Richardson, 1846 : 232 (Vachell specimens). Serranus trimaculatus: Giinther, 1859, 1 : 109 (Reeves fish and another Chinese specimen). As S. trimaculatus " China. Rev. G. Vachell " in Harmer's catalogue destroyed, with date 21.4.1939. Identification based on Chu (1931) and Fowler (19380, P- 273)- 35. Epinephelus akaara (Temm. & Schl.) * Serranus shihpan Richardson, 1846 : 231 (Vachell specimens, also Reeves and Hyde Park collections all part of TYPE SERIES) (Reeves illustr. No. 71). Serranus diacanthus: Giinther, 1859, 1 : no (Chinese specimens, presented by Reeves, Richard- son and the East India Co.). Not included in Harmer's list. Giinther placed Richardson's species in the synonymy of 5. diacanthus. There are several unregistered Chinese specimens in the British Museum (including a Reeves fish) labelled 5. diacanthus, but none labelled 5. shihpan. Reeves illustration reproduced here (Plate 2, fig. i). Richardson was strongly inclined to refer this species to E. akaara, but decided to keep it separate because of the " dark bars which cross the body ". Matayama (1960), Chu, Tchang & Chen (1963), as well as earlier authors, have placed Richard- son's species in the synonymy of E. akaara, but Fowler & Bean (1930) placed it in the synonymy of E. malabaricus (Bloch & Schn.). W. L. Chan (in litt.) has pointed out that the vernacular name Sek-dang-paan is used in Hong Kong for one of the commonest species, E. fasciatomaculatus (Peters), whereas E. akaara is known as Hung-pan. Also, the absence of definite spots on the dorsal fin in the Reeves figure, and the inclination of the vertical bars on che flanks and their tendency to VACHELL COLLECTION OF CHINESE FISHES 137 fork ventrally, all point to E. fasciatomaculatus. However, Matayama (loc. cit.} placed E. fasciatomaculatus (Peters) as a synonym of E. fario (Thunberg), but con- sidered E. fasciatomaculatus of Fowler & Bean to have been E. diacanthus (Val.). Because of such uncertainties in the nomenclature we have preferred to let Richard- son's species remain in the synonymy of E. akaara for the time being. Family PRIACANTHIDAE 36. Priacanthus tayenus (Richardson) *Priacanthus tayenus Richardson, 1846 : 237 (one Vachell specimen and one Reeves specimen, the TYPES) (Reeves illustr. No. /3i4). Priacanthus tayenus: Giinther, 1859, 1 : 221 (a single specimen, collected by Reeves). Listed as P. japonicus " China. Rev. G. Vachell " by Harmer destroyed. A single Reeves specimen in British Museum (BMNH. 1965.8.12.50) labelled as type. No evidence that the Vachell specimen is present. Reeves illustration reproduced here (Plate 2, fig. 2). Identification follows Chu (1931) and Fowler (19386, p. 67). Family PEMPHERIDAE 37. Pempheris otaitensis (Cuvier) Pempheris otaitensis: Giinther, 1860, 2 : 508 (one Chinese specimen). Listed by Harmer as P. otaitensis " China. Rev. G. Vachell " destroyed. Richardson (p. 244) only mentions the related P. moluca Cuvier without reference to any specimens at all. Family LUTJANIDAE 38. Lut -janus erythropterus (Bloch) Mesoprion annularis: Richardson, 1846 : 229 (one Vachell specimen). Mesoprion annularis: Giinther, 1859, 1 : 204 (Reeves specimens). Two specimens included in Harmer's list as M. annularis " ? History "destroyed. Identification follows Chu (1931) and Fowler (19386, p. 83). Family POMADASYIDAE 39. Pomadasys argenteus (Forsskal) Pristipoma nageb: Richardson, 1846 : 227 (one Vachell specimen) (Reeves illustr. No. 244). Pristipoma hasta: Giinther, 1859, 1 : 289 (Reeves and other Chinese material). Listed as P. hasta " China. Rev. G. Vachell " in Harmer's catalogue destroyed. Giinther retained P. nageb as a separate species (1859, p. 290), but mentioned no Cambridge material and only tentatively placed Richardson's record in the synonymy. Fowler (19390) considered Richardson's record to refer to P. argenteus (Forssk.) and the Reeves illustration is consistent with this. ZOOL. 15, 3. 9 138 P. J. P. WHITEHEAD & K. A. JOYSEY Family THERAPONIDAE 40. Therapon jarbua (Forsskal) Therapon servus: Richardson, 1846 : 238 (no Vachell specimens mentioned) (Reeves illustr. No. 044). Therapon servus: Gunther, 1859, 1 : 278 (Reeves and other Chinese specimens). Listed as T. servus " China. Rev. G. Vachell " by Harmer, several specimens destroyed. Richardson included three other species of Therapon (T. theraps, T. oxyrhynchus and T. quadrilineatus] , none of which were represented by Vachell material. Identification based on Fowler (19396, p. 204). Family SPARIDAE 41. Chrysophrys major (Temminck & Schlegel) Pagrus unicolor: Richardson, 1846 : 242 (Vachell specimen) (Reeves illustr. No. 160). Pagrus unicolor: Gunther, 1859, 1 : 468 (One Hong Kong fish presented by Richardson). Listed as P. major by Harmer " [A specimen from China, Rev. G. Vachell, was lost by evaporation of spirit] " Identification based on Fowler (1940, p. 53). 42. Spams latus (Houttuyn) ^Chrysophrys berda: Richardson, 1846 : 240 (no Vachell material) (Reeves illustr. No. 223). Chrysophrys hasta: Gunther, 1859, 1 : 491 (the Reeves type and other Chinese specimens). Listed as C. hasta, " China (? Rev. G. Vachell) " by Harmer destroyed. Richard- son gives nine species of Chrysophrys, none of which included Vachell material. He placed Sparus hasta Bloch & Schn. in his synonymy of C. berda. Identification follows Chu (1931). Family MULLIDAE 43a. Upenoides subvittatus (Temm. & Schl.) Upeneus subvittatus: Richardson, 1846 : 219 (one Vachell fish, 4 inches). Mullus subvittatus: Gunther, 1859, 1 : 397 (doubtful species No. 2). Not listed in Harmer's catalogue. Fowler (1941) listed this Richardson record under Upeneus subvittatus. 43b. Upenoides subvittatus(Temm. & Schl.) Upeneus russelii: Richardson, 1846 : 220 (one injured Vachell specimen) (Reeves illustr. 036). Upeneus indicus: Gunther, 1859, 1 : 406 (two Reeves stuffed specimens, no others). Not listed by Harmer. Fowler (1941) identified Richardson's record as Pseudu- peneus indicus (Shaw). VACHELL COLLECTION OF CHINESE FISHES 139 Family SCIAENIDAE 44. Otolithes argenteus (Cuvier) Otolithus argenteus: Richardson, 1846 : 225 (one Vachell specimen, 6-55 inches). Otolithus argenteus: Giinther, 1860, 2 : 310 (Reeves and East India Co. specimens). Not included in Manner's list. The identifications of the three sciaenids listed here are based on the recent work by Chu, Lo & Wu (1963). 45. Collichthys lucidus (Richardson) Sciaena lucida Richardson, 1844 : 87, and 1846 : 224 (see below). Collichthys lucida: Giinther, 1860, 2 : 312 (Reeves and other Chinese specimens). Listed as Collichthys lucida " China. Rev. G. Vachell " in Harmer's catalogue destroyed. Richardson made no direct reference to Vachell material but states " forms part of all the collections of Chinese fish that we have examined ". Richard- son described the species (Zoology of Sulphur, p. 87) on material in the British Museum, Hasler Museum and Camb. Phil. Soc. He measured a single specimen, 6-45 inches in length. There are two British Museum fishes labelled as types (BMNH. 1848.3.18.107-8) ; the larger of the two (163 mm. tot. 1., 6-4 inches) is most likely the specimen measured by Richardson. There are also two Haslar specimens (BMNH. 1855.9.19.195-7). 46. Dendrophysa russelii (Cuvier) Umbrina russelii: Richardson, 1846 : 226 (one Vachell fish) (Reeves illustr. No. /?37). Umbrina russellii: Giinther, 1860, 2 : 278 (Malaya and " China Seas? "). Listed by Harmer as Umbrina russellii " ? E. Indies " destroyed. Family SILLAGINIDAE 47. Sillago japonica (Temm. & Schl.) Sillago japonica: Richardson, 1846 : 223 (Vachell and Reeves specimens) (Reeves illustr. No. 40). Sillago sihama: Giinther, 1860, 2 : 243 (Reeves and China specimens). Listed as 5. japonica " China. Rev. G. Vachell " by Harmer, three specimens destroyed. Giinther (1860) referred all his Chinese material to S. sihama and not S. japonica ; presumably he changed his mind by the time he came to re-identify the Cambridge specimens. Identification based on Chu (1931). Family EPHIPPIDAE 48. Platax orbicularis (Forsskal) Platax teira: Giinther, 1860, 2 : 492 (China specimens). Harmer lists Platax teira, 2 specimens " China. Rev. G. Vachell." but, as shown earlier, the actual specimens are the types of Seserinus vachellii. Richardson (1846, I 4 o P. J. P. WHITEHEAD & K. A. JOYSEY p. 245) lists P. ehrenbergii Cuvier and P. vespertilis Bloch, both of which he identified from Reeves drawings ; in neither case does he mention specimens. Although the present specimens were mislabelled, the name Platax teira is included in this list of possible Vachell material. 49. Drepane punctata (Linnaeus) Drepane longimana: Richardson, 1846 : 245 (one Vachell fish) (Reeves illustr. No. 241). Drepane punctata: Gunther, 1860, 2 : 62 (two China specimens). Not listed by Harmer. Identification follows Fowler (1953, p. 12). Family SGATOPHAGIDAE 50. Scatophagus argus (Linnaeus) Scatophagus argus: Richardson, 1846 : 245 (two Vachell fishes) (Reeves illustr. No. " 272? "). Scatophagus argus: Gunther, 1860, 2 : 58 (one Reeves and other Chinese specimens). Listed by Harmer as " ? Fam. ? Scatophagus argus (2) PHistory " destroyed. Identification follows Fowler (1953, p. 14). Family AGANTHURIDAE 51. Teuthis fuscescens (Houttuyn) Amphacanthus margaritiferus : Richardson, 1846 : 243 (one Vachell specimen, 7 inches) (Reeves illustr. No. 259). Teuthis albopunctata: Gunther, 1861, 3 : 318 (Reeves and other Chinese specimens). Not listed by Harmer. Identification based on Chu (1931). Family SCORPAENIDAE 52. Vespicula trachinoides (Cuvier) Apistes trachinoides: Richardson, 1846 : 213 (one Vachell fish). Prosopodasys trachinoides: Gunther, 1860, 2 : 139 (one Richardson specimen). Listed by Harmer as Prosopodasys trachinoides " ? East Indies " destroyed. Richardson (p. 213) also lists the British Museum specimen, collected by Sir Edward Belcher. Identification follows Chu (1931). 53. Sebastodes vachellii (Richardson) *Sebastes vachellii Richardson, 1846 : 214 (one Vachell specimen, the TYPE) (Reeves illustr. No. 69?). Sebastes vachellii: Gunther, 1860, 2 : 95 (doubtful species No. i). Not listed by Harmer ; not amongst type material in British Museum. Reeves illustration reproduced here (Plate 2, fig. 3). W. L. Chan (in lift.) states that the caption to the Reeves illustration, pronounced in Hong Kong as Sek-gwau- VACHELL COLLECTION OF CHINESE FISHES 141 kwung (Shih kow kung in the " Report "), refers to one of the commonest of Hong Kong inshore fishes, Sebastiscus marmoratus (Cuv.), and that the drawing agrees with this. However, Richardson listed Sebastes marmoratus (" Report ", p. 215), based on a Biirger specimen in the British Museum, but was not able to identify a Reeves drawing with it. Further, he speaks of the " uncertainty of the drawing " of S. vachellii, citing it as " Icon. Reeves, 69? ". Since there are two intergrading colour forms of 5. marmoratus (Matsubara, 1943, p. 256), it is possible that Richardson did not recognize the Reeves drawing as that species. Neither Matsubara (loc. cit.) nor Chu, Tchang & Chen (1963) mention Richardson's 5. vachellii, and in the absence of the single Vachell specimen we have merely followed the synonymy of Chu (1931), one of the few authors to list this name. Family SYNANCEJIDAE 54 . Inimicus japonicus (Cuvier) *Pelor tigrinum Richardson, 1846 : 212 (one Vachell specimen, the TYPE) (Reeves illustr. No. M- Pelor japonicum: Giinther, 1860, 2 : 151 (two Reeves specimens). Listed as P. japonicum " China. Rev. G. Vachell " by Harmer EXTANT. Reeves illustration reproduced here (Plate 3, fig. i). See Section 5 (p. 154) for notes on extant specimen. 55. Minous monodactylus (Bloch) Minous woora: Richardson, 1846 : 213 (one Vachell fish). Minous monodactylus: Giinther, 1860, 2 : 148 (Chinese specimens). Listed as M. monodactylus, " = Apistus minous, Cuv. (2) China. Rev. G. Vachell." by Harmer destroyed. Identification follows Chu, Tchang & Chen (1963). 56. Polycaulus uranoscopus (Bloch & Schneider) Synanceia breviceps Richardson, 1844 : 71 (one Reeves and three Vachell fishes). Aploactis breviceps: Richardson, 1846 : 212 (three Vachell fishes). Polycaulus elongatus: Giinther, 1860, 2 : 175 (one Reeves fish, later cancelled, presumed des- troyed) . Listed as " Polycaulus elongatus (3) = Synanceia breviceps, Richards. China Rev. G. Vachell." by Harmer destroyed. Identification follows Chu (1931). Family PLATYCEPHALIDAE 57. Platycephalus indicus (Linnaeus) Platycephalus insidiator: Richardson, 1846 : 216 (one Vachell specimen). Platycephalus insidiator: Giinther, 1860, 2 : 177 (Reeves and other China specimens). Listed as P. insidiator " China. Rev. G. Vachell " by Harmer, three specimens- destroyed. Identification based on Chu (1931). 142 P. J. P. WHITEHEAD & K. A. JOYSEY 58. Thysanophrys crocodilus (Tilesius) Platycephalus guttatus: Richardson, 1846 : 217 (one Vachell fish). Platycephalus guttatus: Gunther, 1860, 2 : 183 (no specimens doubtful species). Not listed by Harmer. Chu, Tchang & Chen (1963) place Richardson's record in Inegocia guttata (Cuvier). The synonymy of Chu (1931) is followed here. Family POMACENTRIDAE 59. Amphiprion bicinctus (Riippell) Amphiprion chrysargyrus Richardson, 1846 : 254 (on Reeves drawing only) (Reeves illustr. No. L26). Amphiprion clarkii: Gunther, 1862, 4 : 5 (Reeves and other China material, including one stuffed fish collected by Reeves, the TYPE of A. chrysargyrus). A. chrysogaster " China. Rev. G. Vachell " is listed by Harmer (possibly an error since that is an Indian Ocean species) 2 fishes destroyed. Richardson mentions no specimens but based the species on a Reeves drawing ; in fact a Reeves fish (stuffed) is present in the British Museum collections. Fowler (1954) placed Richardson's A. chrysargyrus in the synonymy of A. xanthurus Cuvier, but the pattern of white bands in the Reeves drawing corresponds much more nearly with Fowler's figure I for A . bicinctus. Family LABRIDAE 60. Halichoeres nigrescens (Bloch & Schneider) *Julis exornatus Richardson, 1846 : 258 (Bankier specimens from Hong Kong, and Vachell and Reeves material TYPES) . Platyglossus dussumieri: Gunther, 1862, 4 : 143 (Reeves and other China specimens, including TYPES of Julis exornatus). Not listed by Harmer. Gunther listed China specimens as types of /. exornatus (BMNH. 1851.12.27.1634). The Vachell fishes can be considered to have been syntypical material. Two Richardson specimens in British Museum from Hong Kong (BMNH. 1848.3.8.6 and 1868.3.19.833). It is possible that these are Vachell specimens, but there is no evidence to show it. Reeves illustration re- produced here (Plate 3, fig. 4). Identification based on Fowler (1956, p. 230). Family GOBIIDAE 61. Glossogobius giuris (Ham. Buch.) Gobiusfasciato-punctatus: Richardson, 1846 : 204 (Vachell fish or fishes) (Reeves illustr. No. 146). Gobius giuris: Gunther, 1861, 8:21 (Reeves type of G. fasciato-punctatus in British Museum and other China specimens). Not listed by Harmer. Identifications and synonymies for the eleven gobioid species listed here have been based on Koumans (1931, 1953). VACHELL COLLECTION OF CHINESE FISHES 143 62. Glossogobius brunneus (Tern. & Schl.) *Gobius platycephalus Richardson, 1846 : 204 (one Vachell fish, the TYPE, " not in very good condition ") (Reeves illustr. No. LQ4). Gobius platycephalus: Giinther, 1861, 3 : 5 (doubtful species No. 39, related to G. brunneus). Not listed by Harmer. In the addenda to the " Report " Richardson states (p. 318) that he had just seen the description of G. brunneus by Temminck and Schlegel and he considered his G. platycephalus probably identical to that species. Vachell fish not amongst types in British Museum. Reeves illustration reproduced here (Plate 3, fig. 3). Fowler (1960) considered G. brunneus a synonym of G. giuris. 63. Cryptocentrus filifer (Valenciennes) Gobius filifer: Richardson, 1846:205 (Vachell material). Gobius knutteli: Giinther, 1861, 3 : 73 (Reeves and other China specimens). Listed as G. knutteli " China. Rev. G. Vachell." by Harmer destroyed. A second Vachell specimen listed by Harmer as " [Origl. label = G. filifer] (probably the speci- men alluded to by Richardson, in Brit. Assn. Report, 1845, p. 205) " destroyed also. 64. Rhinogobius margariturus (Richardson) *Gobius margariturus Richardson, 1846 : 205 (one, perhaps more, Vachell specimens TYPE) (no Reeves illustr.). Gobius margariturus: Giinther, 1861, 3 : 47 (on Richardson, no specimens listed). As G. margariturus " China. Rev. G. Vachell " in Harmer's list destroyed. Vachell specimen not amongst types in British Museum. Fowler (1960) placed this species in Ctenogobius and commented " A rare species, apparently not seen since described in 1846." 65. Drombus ripilepis (Richardson) *Gobius ripilepis Richardson, 1846 : 205 (one Vachell fish, the TYPE, 3^ inches) (no Reeves illustr.). Gobius ripilepis: Giinther, 1861, 3 : 5 (doubtful species No. 28, apparently allied to G. kokius "). Not included in Harmer's list. Placed in Ctenogobius by Fowler (1960). 66. Acanthogobius stigmothonus (Richardson) *Gobius stigmothonus Richardson, 1844 : 147, and 1846 : 205 (two Vachell fishes of 5 or 6 inches the TYPES) (no Reeves illustr.). Gobius stigmothonus: Giinther, 1861, 3 : 76 (no specimens). Listed as G. stigmothonus (3 fishes) " China. Rev. G. Vachell " by Harmer destroyed. The species was described by Richardson in the Zoology of the Sulphur (1844, p. 147) and was based on "... several specimens . . . brought from Canton by the Rev. George Vachell ". The Vachell specimens are not amongst the types in the British Museum. M4 p - J- P- WHITEHEAD & K. A. JOYSEY 67. Apocryptes serperaster (Richardson) *Apocryptes serperaster Richardson, 1846 : 206 (two Vachell specimens, 6 inches in length the TYPES) (Reeves illustr. No. 55) . Apocryptes serperaster: Giinther, 1861, 3 : 82 (Reeves and other China specimens). Listed by Harmer as A. serperaster " ? China Rev. G. Vachell [labelled " Reeves "]" destroyed. The words " Rev. G. Vachell " are crossed out, perhaps when the reference to Reeves was added. A Reeves specimen of 6 inches is in the British Museum, now labelled as a type, BMNH. 1965.8.12.51. It has a metal tag tied to the jaw "19.2.3.". Reeves illustration reproduced here (Plate 3, fig. 2). Family PERIOPHTHALMIDAE 68a. Boleophthalmus pectinirostris (Linnaeus) Boleophthalmus boddaerti: Richardson, 1846 : 208 (Vachell material) (Reeves illustr. No. ^38). Boleophthalmus boddaerti: Giinther, 1861, 3 : 102 (Indian Ocean specimens only). Not included in Banner's list. Fowler (1962) distinguished B. boddaerti (Pallas) from B. pectinirostris mainly on the presence in the former of dark transverse badns on the body. The Reeves drawing does not show these bands. 68b. Boleophthalmus pectinirostris (Linnaeus) Boleophthalmus pectinirostris: Richardson, 1846 : 208 (one Vachell fish, 2-80 inches). Boleophthalmus pectinirostris: Giinther, 1861, 3 : 102 (Reeves specimen). Not included in Harmer's list. Unregistered Reeves specimen in British Museum of 132 mm. tot. 1., and an unregistered Haslar specimen of 106-5 min - "tot. 1., labelled merely " 158 ". Both are too large to have been the Vachell fish. 6ga. Scartelaos histophorus (Valenciennes) Boleophthalmus aucupatorius Richardson, 1844 : 148, and 1846 : 208 (Vachell material). Boleophthalmus viridis: Giinther, 1861, 3 : 104 (" b,c. Half-grown. China. Presented by Sir J. Richardson Types of B. aucupatorius."). Not included in Harmer's list. Described by Richardson in the Zoology of the Sulphur (1844, P- T 4&}> based on fishes collected by Sir Everard Home, John Reeves and the Rev. George Vachell. Two Chinese specimens in the British Museum (BMNH. 1965.8.12.52-3) presented by Richardson; the larger of the two (2-65 inches total length) is undoubtedly Richardson's measured and figured specimen. There is no indication, however, that these are Vachell specimens. 6, 134, 139, 153 vespertilis, 140 PLATYCEPHALIDAE, 141 Platycephalus guttatus, 142 indicus, 141 insidiator, 141 platy cephalus, Gobius, 143 Platyglossus dussumieri, 142 playjairii, Coilia, 129, 149 PLOTOSIDAE, 131 Plotosus anguillaris, 131 lineatus, 131 Polycaulus elongatus, 141 uranoscopus, 141 POMACENTRIDAE, 142 POMADASYIDAE, 137 Pomadasys argenteus, 137 PRIACANTHIDAE, 137 Priacanthus japonicus, 137 tayenus, 137 Pristipoma hasta, 137 nageb, 137 Prosopodasys trachinoides , 140 Psenopsis anomala, 134 Pseudobagrus aurantiacus, 130 fulvi-draco, 131 vachellii, 130 Pseudupeneus indicus, 138 pulchella, Parapercis, 146 Percis, 146 punctata, Drepane, 140 quadrilineatus , Therapon, 138 reevesii, Callionymus, 145 Leucosoma, 129 Rhinogobius margariturus, 143 164 P. J. P. WHITEHEAD & K. A. JOYSEY ripilepis, Drombus, 143 Gobius, 143 ruber, Holocentrus, 132, 151 rubrum, Holocentrum, 132 rugosus, Amblyopus, 145 russelii, Dendrophysa, 139 Umbrina, 139 Upeneus, 138 russellii, Umbrina, 139 SALANGIDAE, 129 Salanx chinensis, 129 savala, Leptur acanthus, 133, 152 Trichiurus, 133 Scartelaos histophorus, 144, 155 viridis, 156 SCATOPHAGIDAE, 140 Scatophagus argus, 140 Sciaena lucida, 139 SCIAENIDAE, 139 scripta, Alutera, 147 scriptus, Monacanthus, 147 SCORPAENIDAE, 140 scyllia, Triakis, 128 scyllium, Triakis, 128 Seym, 134 Sebastes marmoratus, 141 vachellii, 140 Sebastiscus marmoratus, 141 Sebastodes vachellii, 140 serper aster, Apocryptes, 144 SERRANIDAE, 136 S err anus diacanthus, 136 shihpan, 136 trimaculatus, 136 servus, Therapon, 138 Seserinus vachellii, 134, 139, 153 shihpan, Serranus, 136 sihama, Sillago, 139 SILLAGINIDAE, 139 Sillago japonica, 139 sihama, 139 sinense, Pelor, 155 Solea, ommatura, 132 ovata, 131 SOLEIDAE, 131 SPARIDAE, 138 Spams hasta, 138 latus, 138 Sphaeroides, 157 Sphyraena chinensis, 133 obtusata, 133 SPHYRAENIDAE, 133 spinosissimus, Holocentrus, 151 stellatus, Abalistes, 146 Balistes, 146 stigmothonus, Acanthogobius, 143 Gobius, 143 Stromateus niger, 134 Strongylura strongylura, 131 strongylura, Strongylura, 131 Strongylurus , B clone, 131 subvittatus, Mullus, 138 Upeneus, 138 Upenoides, 138 Synanceia breviceps, 141 SYNANCEJIDAE, 141 Synaptura zebra, 132 Taenioides anguillaris, 145 tayenus, Priacanthus, 137 fern*, Platax, 134, 139, 153 TETRAODONTIDAE, 147 Tetrodon ocellatus, 147, 157 Teuthis albopunctata, 140 fuscescens, 140 Therapon jarbua, 138 oxyrhynchus, 138 quadrilineatus , 138 servus, 138 theraps, 138 THERAPONIDAE, 138 theraps, Therapon, 138 thrissa, Clupanodon, 128 Clupea, 128 Thysanophrys crocodilus, 142 tigrinum, Pelor, 141, 154 Torquigener, 157 trachinoides, Apistes, 140 Prosopodasys, 140 Vespicula, 140 Trachinotus melo, 134 TRIACANTHIDAE, 146 Triacanthus biaculeatus, 146, 156 brevirostris, 146, 156 weberi, 157 Triacanthus (Trixiphichthys), 157 TRIAKIDAE, 128 Triakis scyllia, 128 scyllium, 128 TRICHIURIDAE, 133 Trichiurus armatus, 133, 152 intermedius, 133, 152 lepturus, 133, 152 muticus, 133, 152 savala, 133, 152 INDEX 165 tricuspidatus , Congrus, 129 trigrammus, Cynoglossus, 132 trimaculatus , Serranus, 136 Trypauchen vagina, 145 TRYPAUCHENIDAE, 145 unicolor, Pagrus, 138 Umbrina russelii, 139 russellii, 139 Upenoides subvittatus, 138 Upeneus indicus, 138 russelii, 138 subvittatus, 138 URANOSCOPIDAE, 146 Uranoscopus asper, 146 japonicus, 146 uranoscopus, Polycaulus, 141 Uroconger lepturus, 130 vachellii, Ambassis, 136 Bagrus, 130 Batistes, 124, 146 Pseudobagrus, 130 Sebastes, 140 Sebastodes, 140 Seserinus, 134, 139, 153 vagina, Trypauchen, 145 vespertilis, Platax, 140 Vespicula trachinoides, 140 viridis, Boleophthalmus, 144 Gobius, 156 Scartelaos, 156 vulgaris, Anguilla, 129, 150 weberi, Triacanthus , 157 woora, Minous, 141 xanthurus, Amphiprion, 142 zebra, Synaptura, 132 Zebrias, 132 Zebrias zebra, 132 PLATE i FIG. i. Chatoessus maculatus ( Clupanodon thrissa). Reeves p. 91, No. 109. FIG. 2. Caranx cancroides ( = Caranx (Atule) hallo). Reeves p. 35, No. ^30. FIG. 3. Tmchinotus melo (=Psenopsis anomala). Reeves p. 68, No. 97. FIG. 4. Solea ommatura (=Zebrias zebra). Reeves p. 71, No. ^13. Bull. Br. Mus. nat. Hist. (Zool.) 15, 3 PLATE i ZOOL. 15, 3 PLATE FIG. i. Serranus shihpan (=Epinephelus akaara). Reeves p. 14, No. 71. FIG. 2. Priacanthus tayenus (=Priacanthus tayenus). Reeves p. 71, No. ^1 FIG. 3. Sebastes vachellii ( = Sebastodes vachellii). Reeves p. 14, No. 69. Bull. Br. Mus. nut. Hist. (Zool.) 15, 3 PLATE 2 PLATE 3 FIG. i. Pelor tigrinum ( = Inimicus japonicus). Reeves p. 28, No. ^42. FIG. 2. Apocryptes serperaster (=Apocryptes serperaster). Reeves p. 96, No. /?55. FIG. 3. Gobius platycephalus (=Glossogobius brunneus). Reeves p. 124, No. 194. FIG. 4. Julis exornatus (=Halichoeres nigrescens). Reeves p. 56, No. /Sio. Bull. By. Mus. nat. Hist. (Zool.) 15, 3 PLATE 3 PRINTED IN GREAT BRITAIN BY ADLARD & SON LIMITED BARTHOLOMEW PRESS, DORKING 27 FEB1967 k NOTES ON ASTEROIDS IN THE BRITISH MUSEUM (NATURAL HISTORY) V. NARDOA AND SOME OTHER OPHIDIASTERIDS AILSA MCGOWN CLARK BULLETIN OF THE BRITISH MUSEUM (NATURAL HISTORY) ZOOLOGY Vol. 15 No. 4 LONDON: 1967 NOTES ON ASTEROIDS IN THE BRITISH MUSEUM (NATURAL HISTORY) V. NARDOA AND SOME OTHER OPHIDIASTERIDS BY AILSA MCGOWN CLARK British Museum (Natural History) >c Pp. 167-198 ; 6 Plates BULLETIN OF THE BRITISH MUSEUM (NATURAL HISTORY) ZOOLOGY Vol. 15 No. 4 LONDON: 1967 THE BULLETIN OF THE BRITISH MUSEUM (NATURAL HISTORY), instituted, in 1949, is issued in five series corresponding to the Departments of the Museum, and an Historical series. Parts will appear at irregular intervals as they become ready. Volumes will contain about three or four hundred pages, and will not necessarily be completed within one calendar year. In 1965 a separate supplementary series of longer papers was instituted, numbered serially for each Department. This paper is Vol. 15, No. 4 of the Zoological series. The abbreviated titles of periodicals cited follow those of the World List of Scientific Periodicals. World List abbreviation : Bull. Br. Mus. not. Hist. (Zool.) Trustees of the British Museum (Natural History) 1967 TRUSTEES OF THE BRITISH MUSEUM (NATURAL HISTORY) Issued 24 February, 1967 Price NOTES ON ASTEROIDS IN THE BRITISH MUSEUM (NATURAL HISTORY) V. NARDOA AND SOME OTHER OPHIDIASTERIDS By AILSA McGOWN CLARK THE present study includes notes on the provenance of Nardoa variolata (Lamarck) the type-species of Nardoa, details of the true holotype of N. gomophia (Perrier), description of a new species from northern Australia and a discussion of the remain- ing species of the genus coupled with notes on the validity of the monotypic Gomo- phia, a new subspecies of which is described. In the course of this, the following conclusions are reached : Ophidiaster watsoni Livingstone (1936) is referred to Gomophia and is probably conspecific with G. egyptiaca Nardoa mamillifera Livingstone (1930) and N. tumulosa Fisher (1917) are prob- ably synonyms of N, frianti Koehler Nardoa obtusa (Perrier) is conspecific with N. tuberculata Gray Nardoa lemonnieriKoehler is revived from the synonymy of N. mollis de Loriol and Nardoa faouzii Macan is designated as type-species of a new subgenus. A key to the species of Nardoa is included. Other Ophidiasterids dealt with include : Fromia indica (Perrier), with F. tumida Bell newly referred to synonymy but F. elegans H. L. Clark revived as a valid species A new species of Fromia from the Indian Ocean . Ferdina Gray, re-diagnosed and restricted to a monotypic genus by removal of F. heffernani Livingstone to a new genus ; Ferdina flavescens Gray partially redescribed Tamaria tumescens (Koehler), with T. ajax Livingstone, 19320, newly referred to synonymy and Ophidiaster helicostichus Sladen now referred back to Ophidiaster from Hacelia with possible synonym 0. astridae Engel, 1938. Gomophia egyptiaca egeriae 1 subsp. nov. (PI. i, figs. 1-5) Nardoa tuberculata (pt.) Bell, 1894 : 396. ? Nardoa aegyptiaca : Koehler, 1910 : 157 [Non Gomophia egyptiaca Gray]. MATERIAL. Macclesfield Bank, South China Sea ; Admiralty : 23 metres, 2 specimens the holotype and paratype, B.M. reg. no. 92.8.22.52 and 53 ; 66-76 metres, 3 specimens, nos. 92.8.22.74, 259 and 260 ; 58 metres, I specimen, no. 92.8.22.29 ; 76-84 metres, I specimen, no. 93.8.25.203. 1 Named after H.M.S. " Egeria " which surveyed Macclesfield Bank. ZOOL. 15, 4 12 iyo A. M. CLARK DESCRIPTION OF HOLOTYPE. R/r = 70 mm./n-5 mm. = 6-1/1. The arms taper fairly evenly to attenuated tips ; br at the base = 10-5 mm., at half R = 9-0 mm. and at 10 mm. from the tip = 4-0 mm. The aboral skeleton is reticulate but the oblong secondary plates linking the five- or six-lobed primary abactinal ones are relatively inconspicuous and the reticulum is more compact than in the holotype of G. egyptiaca egyptiaca, the pore-areas being relatively small, mostly with only 3-6 pores in each. Three plates around the centre of the disc and fifteen to twenty other primary ones scattered along the upper side of each arm are abruptly elevated, the larger ones 2-4-2-6 mm. in basal diameter and 2-0-2-4 mm. in height. Their shape is conical but some have the apex blunted. The supero-marginal series of plates similarly includes both flat and markedly convex plates, the former being mostly equal in size to the largest of the flat abactinal plates. The convex plates arise at intervals of one to four flat ones, and are more widely-spaced on the distal half of the arm. The entire surface is coated with fine granules concealing the underlying plates. On the flatter parts of the aboral skeleton these number c. 6o/sq. mm. but on the convexities they become distinctly coarser, especially towards the tips of the plates. Most of these tips have been rubbed but on those that remain intact the apex is crowned by several enlarged granules among which a single central one may reach twice the diameter of the granules encircling it. Though slightly higher than the other granules, this single one is throughout too small to evoke comparison with a nipple. It is more often developed on the supero-marginal plates than the abactinal ones. The infero-marginal plates form a much more regular series than the supero- marginals, all being similar and gradually reducing in size distally ; the proximal ones are conspicuous by their size and measure almost 3 mm. in diameter. The number in one series partially denuded is c. 30. Between the two main marginal series proximally is a row of intermarginal plates. Beyond the basal third of the arm length these are concealed by the granulation but spaced ones are revealed on the denuded arm to c. half R. Small pore-areas are present both above and below the intermarginal plates continuing distally beyond them but there are none on the lower surface of the body below the infero- marginals. The actinal series of plates is poorly developed, the one cleared consisting of 18 plates, about three corresponding to each infero-marginal where they lie adjacent ; the distalmost actinal plate lies interstitially between the seventh and eighth infero- marginals and the adjoining adambulacrals, i.e. at just over one-third R from the mouth. The adambulacral plates for much of the arm each bear four, rarely five, prismatic furrow spines c. I mm. long, the fans of which are slightly oblique so that consecutive ones tend to overlap. Behind these are series of usually three shorter subambulacral spines, the middle one becoming progressively stouter and relatively longer on the distal part of the arm, while the proximal one disappears or becomes granuliform. The outer part of the plate bears polygonal granules indistinguishable from those of the adjacent actinal and infero-marginal plates so that the two rows of spines project NOTES ON ASTEROIDS: OPHIDI ASTERIDS 171 abruptly. In comparison with the species of Nardoa, the adambulacral spines are more slender and spaced farther apart. PARATYPES. [Treated in the same order as in table i on p. 177]. As indicated in the table, the raised abactinal and supero-marginal plates are more or less rounded at the tips and may lack the single enlarged granule altogether. In the specimen from the same depth as the holotype (R 65 mm., PL i, fig. 4) the enlargement of the supero-marginal plates is more frequent, there being a marked tendency for every second one to be convex. In this specimen the furrow spines number five on most of the plates of the proximal half of the arm ; also the granules adjacent to the subambulacral spines are slightly elongated. No. 93.8.25.203 (R7omm.) has alternate proximal supero-marginal plates tubercular but distally the frequency of these is less. Again there are five furrow spines on most of the proximal plates. No. 92.8.22.29 (R6omm.) has very regular alternation of convex and flat supero-marginals for almost the entire length of all the arms. There are four, occasionally five, furrow spines. In no. 92.8.22.74 (R 52 mm.) the single apical granule of some of the supero- marginal plates at least (most of the aboral ones being rubbed) approaches nipple- like proportions. The furrow spines usually number four. Finally, nos. 92 . 8 . 22 . 259 and 260 are smaller, R 34 and 32 mm. (The latter has only two arms intact and is not included in the Table). The former (PI. i, fig. 5) has markedly conical tubercular plates, of which the two midradial ones near the base of each arm are so large as almost to meet the supero-marginal plates each side, only one small flat plate intervening. This is in marked contrast to the specimens of G. egyptiaca egyptiaca of similar size from Christmas Island, where the abactinal plates are relatively smaller and more numerous (PL i, fig. 6). Another difference is that the intermarginal plates are very narrow in the Macclesfield Bank specimen and indeed are only visible when the granulation is removed, being sandwiched closely between the two main marginal series. The Christmas Island specimens have conspicuous broad series of intermarginal plates extending for about half R. However, the two from Macclesfield Bank cannot be referred to Nardoa frianti since they both have unusually well-developed, almost nipple-sized and sharp apical tubercles on the conical abactinal and supero-marginal plates. It may be noted that a few of the supero-marginals are crowned by two or even three tubercles, which may also occur in G. egyptiaca egyptiaca, as shown by a specimen from Samoa (poorly preserved and not included in Table i) though here the tubercular plates tend to have a double apex like a cow-bell and the tubercles are separated at the ends of the ridge. Comparative remarks between egeriae and egyptiaca as well as with Nardoa are included in the general discussion of Nardoa and Gomophia. Nardoa variolata (Retzius) Tetractis cruciata variolata Linck, 1733 : 19, pi. i, no. i. Pentadactylosaster variolatus Linck, 1733 : 34, pi. viii, no. 10. Hexactis variolata Linck, 1733 : 37, pi. xiv, no. 24. 172 A. M. CLARK [Un-named] Encyclopedie Methodique, 1792 : pi. cxix, figs. 4, 5 [" Asterias variolata : Lamarck " in captions, 1827]. Asterias variolata Retzius, 1805 : 19 ; Lamarck, 1816 : 565. Linkia variolosa (sic) : Nardo, 1834 : 717. Nardoa variolata : Gray, 1840 : 286 ; 1866 : 15 ; Fisher, 1919 : 379, pi. cviii, fig. 4 ; H. L. Clark, 1921 : 51, pi. xxviii, figs. 3, 4. [Non Domantay & Roxas, 1938 : 224-225, probably = N. novaecaledoniae]. Nardoa Agassizii Gray, 1840 : 287 ; 1866 : 15. Scy 'taster variolatus : Miiller & Troschel, 1842 : 34-35 ; Michelin, 1845 : 21 ; Peters, 1852 : 178 ; Liitken, 1865 : 41 ; Perrier, 1875 : 159-162 (423-426) ; Smith, 1879 : 566 ; Mobius, 1880 : 50 ; de Loriol, 1885 : 43-44. [Non Bell, 1882 : 220 ; 1887 : 647 ; ?i9O2 : 226 ; Pigog : 19]. This species was formally designated as type-species of Nardoa by H. L. Clark (1921). Unfortunately, although there has long been agreement that the common Nardoa of Mauritius should be called variolata, the basis for this is not perfectly secure since no holotype is extant. One solution would be for a neotype to be chosen. Alternatively a lectotype could be selected from among Linck's figures (1733) which were cited by Retzius, the first post-Linnean author to use and thereby validate the name. As Linck's specimens are not preserved at the present museum in Trier the figures alone will have to suffice. Those of the normal five-armed specimen (Linck, pi. viii, no. io)reproducedin the" Encyclopedie Methodique", are to be preferred; they certainly have a strong resemblance to the common Nardoa of Mauritius. Linck, Retzius and Nardo all failed to give a locality for the species but Lamarck put " mers d'Europe " from which Gray appears to have derived " Mediterranean Sea " at the same time as he gave " Isle of France " (Mauritius) for Nardoa Agassizii, subsequently referred to the synonymy of variolata by Miiller and Troschel, who also record specimens from Mauritius. Some remarks about the geographical and morphological limits of the species are included in the general discussion of Nardoa. Nardoa gomophia (Perrier) (PI. 2, figs. 1-4) Scy taster gomophia Perrier, 1875:167-169 (431-433). [Non Nardoa " gamophia " : Fisher, 1919 : 380, which is based on a specimen of N. novaecaledoniae]. MATERIAL. New Caledonia ; Cuming ; the holotype, B.M. reg. no. 59.4.25.3. As Fisher noted in 1919, the specimen (PL 5, fig. 2) in the British Museum col- lections hitherto labelled as the holotype of Perrier's Scytaster gomophia does not particularly resemble Gomophia egyptiaca with which Perrier allied and compared his new species but is close to Nardoa novaecaledoniae. Its number is 59.4.25.4 and a search through the dry collections yielded a second specimen not conspecific with the first but also labelled as from New Caledonia and having the consecutive number 59.4.25.3 (PI. 2, figs. 1-4). This specimen bears a label" Scytaster n. sp. ", though not in Perrier's handwriting. A later printed label names it Nardoa perrieri, which I think must be an MS name given it by Bell prior to placing it on exhibition. This second specimen does have a considerable resemblance to Gomophia egyptiaca NOTES ON ASTEROIDS: OPHIDI ASTERIDS 173 owing to the presence of some spaced, abruptly projecting, tubercular plates on the upper side. As in Perrier's description, these are hemispherical rather than conical, unlike G. egyptiaca, and tend to form three alternating series along each arm. No such projections are present in the first specimen, 59.4.25.4, but this does agree in diameter with Perrier's measurement of 135 mm., the mean of its five possible diameters being 131 mm., whereas the same mean in no. 59.4.25.3 is 167 mm. However, the R/r ratio of 7-7/1 in the latter compared with 5-5/1 in no. 4 agrees so closely with Perrier's estimate of " nearly 8/1 " that I am convinced the wrong specimen has been labelled as the holotype of N. gomophia. Accordingly the assessment of the species must be reviewed since H. L. Clark referred gomophia to the synonymy of N. novaecaledoniae following Fisher's comments. On the one hand it is clearly related to those species of Nardoa which have some of the abactinal plates distinctly tubercular and on the other to Gomophia egyptiaca, as Perrier stated. To avoid repetition, the comparison is included in the general discussion of Nardoa and Gomophia. Nardoa sphenisci 1 sp. nov. (PI. 3, figs. 1-3) MATERIAL. Holothuria Bank, N.W. Australia, 27 metres ; H.M.S. " Penguin ". i specimen, the holotype, B.M. reg. no. 92.1.14.27. Parry Shoal (N.W. of Darwin), 22 metres ; H.M.S. " Penguin ". 2 specimens, no. 92.4.4.6 and 7. DESCRIPTION OF THE HOLOTYPE. R 115-120 mm. ; r c. 13 mm. ; R/r = 9-0/1. The arms are slender and attenuated, br at the base = 13-14 mm., at half R = c. 10 mm., at 10 mm. from the tip = 4-5-5-0 mm. and at the very tip = c. 2 mm. One arm has been broken near the base and is regenerating. Most of the abactinal plates are more or less convex, especially the midradial ones near the bases of the arms, which are 2-5-3-5 mm. in transverse diameter (slightly less longitudinally) and up to 1-5 mm. in height ; their tops are rounded. Beyond the bases of the arms the more convex plates are scattered over the whole upper side. Across the base of each arm between the two supero-marginal series the abactinal plates number five to seven ; when there are seven the two outermost plates are small, for most of the proximal part of the arm there are about five plates across the width though the arrangement is irregular. Distally the plates gradually decrease in size and become elongated, the more convex ones measuring 2-0-2-5 mm. in length and 1-0-1-5 mm - m breadth. Only on the regenerating arm is there an abrupt reduction in the size of the plates at the point of cleavage. The entire aboral surface is coated with polygonal granules which gradually increase in size on the convexities of the skeleton. The smaller granules between and towards the edges of the plates number 25-30 per square mm. ; they are flat-topped and each granule bears a small boss in the centre except for the coarser ones near the apices of the more conical plates, which tend to have their whole free surface 1 Of the " Penguin ". 174 A. M. CLARK convex and smooth. The smaller distal convex plates particularly have the few enlarged central granules distinctly projecting and on some of these plates there is one central granule which is markedly larger than the seven or eight immediately encircling it. This gives a rough texture to the distal parts of the arms although the central granules are barely visible with the naked eye and never so abruptly different from the other granules as to recall the nipple-like tubercles found on the swollen plates of most specimens of Gomophia egyptiaca. The supero-marginal plates also are more or less convex in form though never to the same extent as the proximal mid-radial plates. There is a strong tendency for alternate supero-marginals beyond the basal five to ten to be larger and more convex than the rest so that the outline of the arms in perpendicular view is distinctly and fairly regularly scalloped. The larger plates of the basal half of the arm are 2-0-2-5 mm. in diameter, the alternate plates being only c. i-o mm. long though still c. 2 mm. wide. In two supero-marginal series counted the total number of plates is 58. In contrast the infero-marginal plates are fairly equal in size, gradually diminishing distally. They number c. 54. Both series of marginal plates are covered with granulation similar to that on the abactinal plates ; this again becomes increasingly coarse on the more convex surfaces. On some of the distalmost plates of both series when only a single granule is enlarged the height of this may exceed its basal breadth though the tip is rounded. In each interradius the two marginal series are separated by a progressively narrowing series of irregularly-placed and irregularly-shaped intermarginal plates. On one area cleared these number about 40 and extend for c. 40 mm. along the arm, although when the granule-covering was intact the plates of the distal half of the series were concealed. On the lower side, the main series of actinal plates is visible through the covering granulation for about half the arm length, (i.e. to about the twenty-fifth infero- marginal) ; further plates extend interstitially concealed by the granulation to about the thirty-fifth infero-marginal or two-thirds of the arm length. The actinal plates are twice as numerous as the adjoining infero-marginals. On the disc there is a second series consisting of only two or three plates. The granulation is again coarser on the centres of the plates than around their edges. The adambulacral plates correspond in number to the actinal plates. Each bears four or five blunt angular furrow spines, the distal plates more often four, though the number may be only three on the terminal fifth of the arm. When there are five furrow spines, usually the proximalmost one is slightly smaller than the rest and inset from the furrow so that the series is curved and adjacent ones tend to overlap. The subambulacral row consists usually of three spines, shorter but somewhat stouter than the furrow spines and often with a smaller fourth proxi- mal spine almost in series with them. The outer part of the plate bears polygonal granules some of which, close behind the subambulacral spines are slightly elongated so that there is no abrupt transition from spines to granules. The pore-areas are small in comparison to the size of the plates. The number of pores in each area is difficult to estimate, perhaps owing to the dried condition ; NOTES ON ASTEROIDS: OPHIDI ASTERIDS 175 on the upper side it is probably 4-8 in most areas. Smaller pore-areas are also present intermarginally and below the infero-marginals on the proximal half of the arm. PARATYPES. Both have R/r 75-85 mm./io mm., approximately 8/1. The arms are slender and attenuated as in the holotype, which they closely resemble. Both have the intermarginal series of plates well-developed on most arms and only lacking on those arms which appear to have regenerated from the base. This can be assumed from the abrupt break in all three marginal series close to the interradius and to coincidence in the occurrence of intermarginal series on both sides of a single arm rather than on the two continuous series of adjacent arms across an interradius. The actinal plates are more extensive than in the holotype, being visible for about five- eighths R with further plates extending to three-quarters R concealed to the naked eye beneath the granulation. The furrow spines number four on most plates of the proximal half of the arm, then three. REMARKS. All three specimens were identified as Nardoa tuber culata by Bell. In 1953 I re-named them Gomophia egyptiaca on account of the intermarginal series of plates and the attenuated arms, disregarding the rounded rather than conical tubercular abactinal plates lacking in central conical nipples. However, a recent study of material of these two genera convinces me that the three specimens are referable to neither species. Their relationships are dealt with below in the general discussion of Nardoa and Gomophia. NARDOA Gray Gray, 1840 : 268 ; 1866 : 15 ; Fisher, 1919 : 378-382 ; H. L. Clark, 1921 : 49-50. The species of Nardoa are usually divided into two groups, those in which some of the abactinal plates are distinctly more convex than the rest and those with similar low plates. The relationship between the first group and the monotypic genus Gomophia should be reviewed, the latter having been referred to the synonymy of Nardoa by several specialists including Sladen (1889), Koehler (1910) and Fisher (1919) but restored to generic rank by H. L. Clark (1921) and following workers. Unfortunately, most of the species involved seem to be solitary and the available samples are very inadequate, rarely more than one or two specimens being taken at a time so that the range of variation is poorly known. Following the restoration here of N. gomophia as a valid species and the addition of N. sphenisci as a new one, the nominal species to be included in such a survey are : Gomophia egyptiaca Gray, 1840 Ophidiaster " watsoni Livingstone, 1936 Nardoa frianti Koehler, 1910 Nardoa tumulosa Fisher, 1917 Nardoa mamillifera Livingstone, 1930 Nardoa gomophia (Perrier), 1875 Nardoa sphenisci sp. nov. 176 A. M. CLARK Nardoa tuberculata Gray, 1840 Nardoa oUusa (Perrier), 1875 and possibly also Nardoa rosea H. L. Clark, 1921. Some of the distinguishing characters between these are brought out by the following table. In this the skeletal structure is indicated as either open, when the papular areas are sufficiently large to give a reticular appearance, as opposed to compact, with the spaces small and irregular in arrangement. A comparison of PL i figs. I and 3 with 6 and 8 may serve to illustrate this point). Negative entries for the abactinal tubercles signify that they are conical. The degree of fineness of the covering granulation is shown by an assessment of the number per square mm. taken on the flatter parts of the aboral skeleton where the granulation is at its finest. The other columns should be self evident. Intermediate conditions are indicated by " +/ " and a tendency one way or infrequent occurrence by brackets. H. L. Clark distinguished Gomophia from Nardoa by the unspecialized supero- marginal plates resembling the abactinal ones (this being enhanced by the series of intermarginal plates separating the supero-marginals from the markedly enlarged infero-marginals proximally) and by the absence of actinal papulae below the infero- marginals (this character implicit in his key) coupled with the conical form and apical " nipples " of the tubercular abactinal plates ; an additional distinction is provided by the rather open, apparently reticulate, aboral skeleton. However, although in the holotype of Gomophia egyptiaca and in a few other specimens closely resembling it, notably the one from Samoa shown in PL I, figs. 7 and 8, none of the supero-marginal plates are enlarged or projecting, this is not always the case ; for instance the fine large specimen from the Gulf of Aqaba (see A. M. Clark, 1952), the second one in the table, has a conspicuous conical tubercular plate in the position of every third or fourth supero-marginal. Similarly some much smaller specimens from Christmas Island in the Indian Ocean have some elevated supero-marginals; in this case alternate plates are so modified, as in the holotype of Livingstone's Ophidiaster watsoni to which the Christmas Island specimens bear great resemblance. In fact, despite the extent of some of the actinal series of plates beyond half the arm length and the slight enlargement of the granules adjoining the subambulacral spines, I have no hesitation in declaring 0. watsoni to be congeneric, possibly even conspecific, with Gomophia egyptiaca. It is significant that Endean (1965) has recorded G. egyptiaca from Queensland. Seven specimens from Macclesfield Bank, South China Sea, which I have referred to Gomophia, also have some of the supero-marginals, often alternate ones, tuber- cular ; such a frequency of the tubercular plates coupled with their linear arrange- ment gives a greater appearance of specialization to the supero-marginals than is evident in the holotype of G. egyptiaca. Bell included these China Sea specimens, together with three others which I believe are Nardoa frianti, in N. tuberculata. However, the presence of intermarginal series of plates, the much more prominent tubercular plates of the aboral side often crowned with enlarged granules or even a single small tubercle, the absence of actinal papulae, the reduction of the actinal series of plates and the abrupt transition from the granulation of the lower side to NOTES ON ASTEROIDS: OPHIDI ASTERIDS 177 ' t; ^3 P G 4) ll its 1111 s~ a i . i i i i+ i i . i Interstitial granulation no./sq. mm. (approx.) u 4-J OOOiOOOgoOOuivOu"! h-1 H M *d i t CD 4J 5 rt G PI < pn ft 1 1 1 1 1 1 1 1 1 1 1 I 1 1 S G , .5 - 3 j -,- H-H-H-H-H- 2S -*.-*. H-H-H-H- ^ "ftS o V V dri< V i G g C Ifll + + + + + + + + + + + + + Nil! CO M U5 1 + 1 1 + + + + + + + + + ps! + - H 8 ,l il++ ^^ +++++++ useum collec s photograp] string-along- i ,., G - s ^-, ^-N. ^-^s rv <"*" s-~* ^ CD * U &++++ 1 ++ 1 1 1 1 1 1 1 1 1 J^^J^^ 1 1 1 1 * 1 1 1 CD O $ 35 5 ~~~^ 60 to ^ '^.S* "s jV-S lOOOmrONOO-O N -* >0^0 fs| u"j O *O *O CO 00 u"> O ^- M N 00 a 'E " g ONinrOTt-oo.r> t^vo C^ vO u~l CO G^ N !>* CO O\ "O 1 O vO 00 >O Oi >O O> 00 M M ni h-3 rj O G .5 o 1 S >, ^-d ^ . ; ^ t ^ _rt "3 - - >"% en S n3 ^ ^ 8 3 "^ '^ T3 **" en g ; d -J 4^ 1 "^ W i_] T3 G P f>,'C -a . . fx, . , G lilts ** *H G 4^ 0^ O W o ** "Cfl k 3 < - > 'G' 3 tH " " " "3 P ^ O * P j P ^ J^ J3 <& co ^ FU H rQ O Gj f^3 S (U ^ rv- 1 "S 1 S i * s r U CD 1 > >, Mill M ^ -w - S (H " H ^ G "ft CD PH ^ -H o H o Ci ^ a ' S 3 W i- CD -'-' H* O rf Oj G^ Q e* " ^ * - * t H^, S w 2"S o &( ^ '^* *** f^J g ^ *g <0 CO ^ ^ ** ^ 13 * xs "a s< . - : : : ;^ o 'i <3 e "s : : T|J o o -^x. i| I f s o" I " *: ". . . *; "! * 4J ++00 /^ 178 A. M. CLARK the two rows of adambulacral spines all agree with G. egyptiaca. Nevertheless, when the fine coat of granules is removed from the upper side, the underlying skeleton appears to be much more compact in these specimens from Macclesfield Bank than in the others included in the table from the Red Sea, Mauritius, Samoa and Christmas Island. Possibly this can be attributed in some degree to contraction in preservation since an unnaturally flattened specimen from Eilat in the Red Sea similarly shows compaction of the abactinal skeleton. The more or less rounded form of the swollen abactinal and supero-marginal plates in most of the specimens from Macclesfield Bank, coupled with the poor development of the apical nipples (these often being confined to the supero-marginals) provides another distinction and three of the specimens have a slight transition from the granulation of the lower side to the subambulacral spines. Also the granulation of the abactinal plates seems to be slightly coarser than in those specimens of G. egyptiaca which most resemble the holotype. I do not consider that these differences carry sufficient weight to merit more than a subspecific distinction for the Macclesfield Bank specimens which are accordingly described here as Gomophia egyptiaca egeriae. The relative specialization of their supero-marginal plates weakens the generic distinction of Gomophia on the basis of this character. Of the other supposedly diagnostic characters, the absence of actinal papulae is certainly constant in G. egyptiaca but unfortunately is matched by individuals of several species of Nardoa, for instance the types of both Nardoa frianti and Nardoa tumulosa, also the specimen from the Ogasawara (Benin) Islands referred to frianti by Hayashi (19386) ; certainly the actinal papulae are few and inconspicuous in the specimens from Macclesfield Bank which I have attributed to frianti. As for the occurrence of intermarginal plates, these are not restricted to G. egypti- aca. Fisher notes that there are a few intermarginal plates at the base of each arm of the holotype of N. tumulosa, while Koehler writes to the effect that in the two syntypes of N. frianti " sometimes the two marginal series are fully contiguous but sometimes there are small plates intercalated between them, though these rarely form a regular and distinct series but rather are inserted in the intervals between the larger plates ". No such intercalary intermarginals are developed in the Macclesfield Bank specimens of Nardoa, while Hayashi (1938 and 19380) mentions only intermarginal papulae, not plates, in the specimens from the Caroline and Ryu Kyu Islands which he referred to N. tumulosa. However, the specimen from the Ogasawara Islands (Hayashi, 19386), figured under the name of N. frianti, not only has well-developed series of intermarginal plates but also has the tubercular abactinal plates restricted to the proximal parts of the arms (like Hayasaka's speci- men from Taiwan [Formosa], 1949, PI. ii, fig. 4, named tumulosa), combining to give it a strong resemblance to the holotype of N. tumulosa. Hayashi gave no reasons for naming his specimen frianti rather than tumulosa. Intermarginal series of plates are also developed in the type material of N. sphenisci. It should be noted that Nardoa frianti and Gomophia egyptiaca are sympatric for at least part of their ranges ; both have been taken in the Andaman Islands (Koehler), on Macclesfield Bank and in the Loyalty Islands. Koehler gave insuffi- cient information about his material of Gomophia from the Andamans for me to be NOTES ON ASTEROIDS: OPHIDI ASTERIDS 179 certain whether it should be left in egyptiaca or referred to subspecies egeriae, though his photograph suggests that the aboral skeleton is fairly open giving a reticulate appearance and thus linking it with egyptiaca sensu stricto. Following from the variation in shape of the abactinal tubercular plates in G. egyptiaca egeriae, I cannot help suspecting that Koehler's unfigured syntype (or paratype) of N. frianti should have been referred to Gomophia. This would account for the discrepancies between his description and the Macclesfield Bank specimens which I have referred to frianti with regard to the occasional presence of intermarginal plates and the absence of actinal papulae (though the latter could well evade detection if they are few in number). The generic status of Gomophia is therefore in some doubt but, since the range of variation and specific limits of most of the species involved in the comparison have yet to be properly determined, I hesitate at present to refer it yet again to the synonymy of Nardoa. [In this it must be confessed that I am influenced to some extent by respect for Gray's eye for a genus, so often vindicated]. Included in the table is a column dealing with the shape of the arm tips, whether blunt or attenuated. However, since several of the species of Nardoa appear to be particularly susceptible to loss of parts of the arms, the subsequent regeneration, when incomplete, may give a false impression of the true shape. However, the mutilated holotype of Nardoa mamillifera Livingstone has all three of there main- ing arms ending abruptly, R/r being 5-7/1. Similarly abbreviated arms are found throughout in the two specimens I have seen from the Loyalty Islands (R/r 5-7/1 and 5-5/1), which otherwise agree closely with the three of N. frianti from Macclesfield Bank. The latter all have markedly attenuated arms with R/r 7-75-8-2/1. In the syntypes of N. frianti the ratio is 7-3 and 7-1/1, while Domantay and Roxas (1938) recorded a specimen from the Philippines with a value as high as 8/1. Doder- lein (1926) recorded as N. frianti one from the Caroline Islands with R/r 6-4/1 but Hayashi (1938) identified as N. tumulosa his tuberculated Carolinian specimen ; this has relatively short though tapering arms, R/r being only 5/1. A specimen from the Ryu Kyu Islands which Hayashi (19380;) similarly named tumulosa has R/r 5-3/1. In Fisher's holotype of N. tumulosa the ratio is 6-4/1. There is con- siderable resemblance between the Ryu Kyu specimen and the individual taken by the Barrier Reef Expedition which Livingstone (1932) referred to Nardoa rosea. Despite Livingstone's assurance that the latter agrees perfectly with the original description of rosea and shows no variation, a direct comparison with a paratype of N. rosea (PL 5, figs. 3 and 4) reveals a marked disparity in the size and convexity of the larger abactinal plates, many of which in the Barrier Reef specimen are c. 3 mm. in transverse diameter and c. 1-5 mm. in height (compared with c. 2 mm. and 50 mm. ........ tumulosa Fisher, 1917 3' Tubercular plates relatively small, rarely much exceeding 2 mm. in diameter, even at R > 70 mm., often present distally as well as (or rather than) proximally . 4 4 Arms not attenuated but more or less blunt at the tip, R/r 5-0-7-0/1 ; the main actinal row of plates extending the whole length of the arm ; tubercular abactinal plates inconspicuous, tending to be fewer proximally and restricted to the lateral areas of the distal parts of the arms ; supero-marginal plates all similar in size, not alternating . tuberculata Gray, 1840 4' Arms very attenuated distally, R/r 7-5-9-2/1 (in the only four specimens known) ; the main actinal row of plates terminating at half to two-thirds R from the base of the arm ; tubercular abactinal plates well-developed, proximally as well as distally ; some of the supero-marginals, often alternate ones, reduced in size . 5 5 Intermarginal plates present in the arm angles ; the majority of abactinal plates markedly convex, particularly the midradial ones . . . sphenisci sp. nov. 5' No intermarginal plates ; tubercular abactinal plates widely separated, as well- developed laterally as mid-radially . . . gotnophia (Perrier), 1875 6 Supero-marginal plates tending to alternate large and small, especially distally rosea H. L. Clark, 1921 * An anagram of Nardoa. 1 If the abruptly terminating arms of the mutilated holotype of N. mamillifera are shared by other specimens from the type-locality this might provide a valid distinction from N. frianti, in which case specimens from the Loyalty Islands could also be referred to mamillifera. :88 A. M. CLARK 6' No regular alternation in size of the supero-marginals ..... 7 7 Abactinal plates of the proximal third to half of the arm forming distinct regular longitudinal and transverse series, though this breaks down distally Nardoa (Andora) fouazii Macan, 1938 7' Abactinal plates irregular in arrangement throughout ..... 8 8 Adambulacral plates each bearing only four furrow spines and four subambulacral ones (though in my interpretation of Liitken's danish the granulation backing on to these spines is transitional) ..... galatheae (Lutken), 1865 8' Granulation of the outer part of each adambulacral plate usually modified to form a third row of spines, numbering three like the two inner rows, though the furrow series may consist of four spines ......... 9 9 Many of the abactinal plates more or less conspicuously broadened and alternating to form an almost continuous pavement, the interstitial pore-areas being small ; the larger plates markedly bigger than the supero-marginals, exceeding 3 mm. in diameter, even at R only c. 40 mm., numbering only about five across the width of the arm basally at R < 50 mm., seven at R 50-70 mm. or up to nine in the rare specimens that exceed 70 mm. ; the distal abactinal plates similar to the proxi- mal ones though gradually becoming smaller . . variolata (Retzius), 1805 9' Few if any of the abactinal plates conspicuously larger than the supero-marginals and rarely broadened transversely to exceed 2-5 mm. in diameter ; if some large plates are present proximally then the distal plates are relatively much smaller, often with an abrupt change in magnitude at about two-thirds R . . . 10 10 All the abactinal plates, both proximally and distally, small in comparison with the marginal plates, numbering about 13 across the arm width basally, at least at R >go mm. ; granuliform pedicellariae often present between the aboral plates and distinctly coarser than the granules ; ventral papulae few and inconspicuous pauciforis (von Martens), 1866 10' The proximal abactinal plates, especially midradially, distinctly larger than the distal plates and sometimes equalling or even slightly exceeding the supero- marginals in diameter, rarely more than eleven plates across the width, more often seven or nine ; pedicellariae not recorded ; usually a number of pore-areas developed between infero-marginal and actinal plates . . . . . n 11 R/r usually 5-0-6-0/1 (see Table 2) . . novaecaledoniae (Perrier), 1875 n' Arms more attenuated, R/r usually 5*5-8-5/1 . . . . . . . 12 1 2 Distal abactinal plates conspicuously smaller than the proximal ones, not exceeding i mm. in transverse diameter, their positions shown by rounded groups of coarse granules among the fine interstitial granulation . . mollis de Loriol, 1891 12' Distal abactinal plates becoming gradually somewhat smaller than the proximal ones and usually 1-2-1-5 rnm. in diameter but markedly elongated in form with ovate groups of coarse granules ..... lernonnieri Koehler, 1910 Fromia indica (Perrier) Scytaster indicus Perrier, 1869 : 255 (?63). Fromia indica : Perrier, 1875 : 177-178 (441-442) ; Koehler, 1910 : 140, pi. xvii, figs. 7, 8 ; H. L. Clark, 1921 : 42 ; Hayashi, 19380 : 207 ; 19386 : 59-62, figs, i, 2, pi. iv, figs. 1-4 ; 1938^ : 279-280. Fromia tumida Bell, 18820 : 124, pi. vi, fig. 4 ; H. L. Clark, 1921 : 41. Fromia andamanensis Koehler, 1909 : 105-108, pi. vii, figs. 5, 6 ; H. L. Clark, 1921 : 41. Fromia elegans : Engel, 1938 : n, fig. i, pi. iii, fig. i [Non F. elegans H. L. Clark, 1921]. Fromia indica forma andamanensis ; Hayashi, 1938 : 428-430, pi. ii, figs. 7, 8, NOTES ON ASTEROIDS: OPHIDI ASTERI DS 189 MATERIAL. Ceylon ; Mr. Kelaart ; two syntypes of Fromia tumida Bell, B.M. reg. no. 52.5.15.62 and 63. Ceylon ; Dr. Ondaatje ; I specimen, no. 88.4.13.3. Andaman Islands ; Dr. Anderson ; 5 specimens, no. 86.6.26.20-22 and 79. Fiji Islands ; Admiralty ; I specimen, no. 56.9.24.26. SYNONYMY. I agree with Hayashi (1938 and 19386) that Fromia andamanensis Koehler is better ranked as no more than a forma of F. indica distinguished by the similar dimensions of the abactinal plates as opposed to forma indica with two series of enlarged plates along each arm. The material in the British Museum collections from the type-locality of andamanensis supports this synonymy. However, I do not agree that F. elegans is also a synonym, since Mr. James F. Clark has re- examined the holotype in the Museum of Comparative Zoology at my request and found that the granules of the actinal plates number seven to nine on a plate and are no higher than wide, differing from the very characteristic elongated granules or short blunt spines, usually in groups of about five, found in F. indica. My thanks are due to Mr. Clark for his observations. Both Domantay and Roxas (1938) and A. H. Clark (1949) have extended the geographical range of F. elegans from Torres Strait to the Philippines. Judging from the photographs given by the former, their specimens could well be elegans but the specimen from Banda Neira referred to elegans by Engel (1938) has the distinctive elongated armament of the actinal plates characteristic of F. indica. To the synonymy of F. indica must also be added Fromia tumida Bell. The syntypes are dried and have suffered some distortion so that the lower surface has become concave and the arms are probably unnaturally wide. The hollowed under side probably evoked Bell's erroneous statement that the adambulacral spines are smaller than the spines of the actinal plates. In fact the armament is just like that of F. indica with short spines on the actinal plates and slightly longer ones on the adambulacrals. Fromia nodosa sp. nov. (PI. 6, figs. 1-3) Ferdina Offreti (pt.) Koehler, 1910 : 143, 147, pi. xvi, figs. 4, 5 only. MATERIAL. Amirante Islands, W. Indian Ocean ; Gardiner collection ; the holo- type and one paratype. B.M. reg. no. 1907.7.1.95 and 96. South Nilandu, Maldive Islands ; Gardiner collection ; I specimen. Cambridge Museum. DIAGNOSIS. A species of Fromia with very conspicuous alternation in size and shape of the supero-marginal plates, the larger ones being abruptly convex and separated by less than their own width from the series of similarly enlarged and convex spaced carinal plates ; no crystal-bodies embedded in the surface of the plates. DESCRIPTION OF THE HOLOTYPE. R/r = 32 mm./7'5 mm - = 4'3/i ', br at base 7-5 mm. and at arm tip 3-5 mm. The body is flattened above and the arms taper evenly to blunt tips. i go A.M. CLARK The abactinal plates are of two distinct kinds, the majority being flat and poly- gonal while nine to eleven spaced carinal plates on each arm, starting with the primary radial plates, are enlarged, up to 2*75 mm. in transverse diameter (slightly less longitudinally), rounded in outline and markedly convex, standing out sharply from the surface though not exceeding I mm. in height. When denuded, the plates between the convex carinal ones are revealed as not part of the series, some being adradial in position though others may be midradial. All the plates are slightly notched at intervals peripherally to accommodate the single interstitial papular pores. There are four to six pores around each of the flatter plates but seven to nine round most of the enlarged ones. The surface of the denuded plates is almost smooth, there being no embedded crystal-bodies. The supero-marginals are similarly of two forms with almost perfect alternation, the larger ones equal to or even slightly exceeding the diameter of the enlarged carinal plates. The swollen supero-marginals alternate in position obliquely with the swollen carinals, the three series on each arm being separated by only single small flat plates for most of the arm length while from the tenth supero-marginal all three rows of swollen plates are partially in contact laterally or obliquely. There are fifteen supero-marginal plates in one series denuded, the first, second, fourth, sixth, eighth, tenth, twelfth, fourteenth and fifteenth being convex and dwarfing the intervening plates. The infero-marginals number sixteen and are all similar in form though gradually decreasing in size from a proximal breadth of c. 2 mm. There are two series of actinal plates, the outer consisting of ten plates and extend- ing to the fifth infero-marginal, two actinal plates corresponding almost exactly to one marginal. The inner row consists of 24 plates (in the series denuded) extending to the middle of the eleventh infero-marginal or 25 mm. distance from the mouth. All these plates are covered with a continuous coat of polygonal granules which are slightly larger on the swollen abactinal and supero-marginal plates, numbering c. 35/sq. mm. as opposed to c. 55/sq. mm. on the flatter plates. Adjoining many of the papular pores one or sometimes several of the granules are distinctly enlarged. There are about 46 adambulacral plates in each series. Four or five of the basal ones bear four furrow spines, the adoral one of these usually smaller and inset on the plate, like a thumb. The remaining plates, except at the very tip of the arm, each have three furrow spines. These are flattened and spatulate, measuring up to 0-75 mm. in length. Separated from the furrow spines by a distinct gap come two subambulacral spines, which are about two-thirds as long as the furrow spines but slightly stouter. Between the twenty-fifth and thirtieth adambulacral plates the number of subambulacral spines falls to one. The outer part of each plate is covered with small low polygonal granules similar to those on the actinal plates and abruptly contrasting with the projecting subambulacral spines. There are two series of single actinal papulae corresponding in extent to the actinal plates. PARATYPES. The second specimen from the Amirantes has R/r = 19 mm./5 mm, = 3-8/1 and closely resembles the holotype except that the arms maintain the NOTES ON ASTEROIDS: OPHIDI ASTERIDS 191 same width for about two-thirds of their length. Also the enlarged supero-marginal plates are distinctly bigger than the enlarged carinals. The second series of actinal plates consists of only one or two plates. There are only about five actinal papulae between the proximal infero-marginals and the main series of actinal plates. The Maldive specimen (PI. 6, fig. 3) has R/r 34 mm./8-5 mm. = 4-0/1 ; br is 9-5 mm. basally but only 2-5 mm. at the tip, the arms tapering markedly. The swollen carinal and supero-marginal plates, particularly the latter, are even larger than in the holotype, producing a very mosaic-like effect, the low interstitial plates occupying little area. The large