7 •2.. 6*

u-

BULLETIN OF

THE BRITISH MUSEUM (NATURAL HISTORY)

ZOOLOGY VOL. V 1957-1960

PRINTED BY ORDER OF THE TRUSTEES OF

THE BRITISH MUSEUM (NATURAL HISTORY) LONDON: 1960

DATES OF PUBLICATION OF THE PARTS

No. i.	26 February 1957
No. 2.	21 March 1957
No. 3.	30 July 1957
No. 4.	12 October 1957
No. 5.	26 August 1958
No. 6.	24 January 1959
No. 7.	24 February 1959
No. 8.	24 February 1959
No. 9.	24 February 1959
No. 10.	22 January 1960

PRINTED IN GREAT BRITAIN

AT THE

BARTHOLOMEW PRESS DORKING

BY ADLARD AND SON, LTD.

CONTENTS

ZOOLOGY VOLUME 5 PAGE

No. i. Studies on the structure and taxonomy of Bulinus jousseaumei

(Dautzenberg). By C. A. WRIGHT (Pis. 1-2) i

No. 2. On Spelaeogriphus, a new cavernicolous crustacean from South Africa.

By ISABELLA GORDON 29

No. 3. The pelecaniform characters of the skeleton of the Shoe-bill Stork,

Balaeniceps rex. By PATRICIA A. COTTAM (PI. 3) 49

No. 4. A revision of the Lake Victoria Haplochromis species (Pisces, Cichlidae) Part II : H. sauvagei (Pfeffer), H. prodromus Trewavas, H. granti Blgr., and H. xenognathus, sp. n. By P. H. GREENWOOD (PI. 4) 76

No. 5. A revision of the genera Nidalia and Bellonella, with an emendation of nomenclature and taxonomic definitions for the family Nidaliidae (Octocorallia, Alcyonacea). By Huzio UTINOMI 99

No. 6. Ear plug laminations in relation to the age composition of a popula- tion of fin whales (Balaenoptem physalus). By P. E. PURVES and

M. D. MOUNTFORD (Pis. 5-6) 123

No. 7. The monotypic genera of cichlid fishes in Lake Victoria Part II and a revision of the Lake Victoria Haplochromis species (Pisces Cichlidae) Part III. By P. H. GREENWOOD 163

No. 8. The Rosaura Expedition 1937-38 Chaetognatha. By JOHN S. COLMAN 219

No. 9. A North Bornean pygmy squirrel, Glyphotes simus Thomas, and its

relationships. By J. E. HILL (Pis. 7-8) 255

No. lo. Revision of the world species of Aplysia (Gastropoda, Opistho-

branchia). By N. B. EALES 267

Index to Volume 5 405

STUDIES ON THE

STRUCTURE AND TAXONOMY

OF BULINUS JOUSSEAUMEI

(DAUTZENBERG)

C A. WRIGHT

BULLETIN OF THE BRITISH MUSEUM (NATURAL HISTORY)

ZOOLOGY Vol. 5 No. i

LONDON: 1957

STUDIES ON THE STRUCTURE AND

TAXONOMY OF BULINUS JOUSSEAUMEI

(DAUTZENBERG)

BY

C. A. WRIGHT

Pp. 1-28 ; Plates 1-2 ; 41 Text-figures

BULLETIN OF THE BRITISH MUSEUM (NATURAL HISTORY)

ZOOLOGY Vol. 5 No. i

LONDON: 195?

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 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.

This paper is Vol. 5, No. i of the Zoological series.

PRINTED BY ORDER OF THE TRUSTEES OF THE BRITISH MUSEUM

Issued February 1957 Price Ten Shillings

STUDIES ON THE STRUCTURE AND TAXONOMY OF BULINUS JOUSSEAUMEI (DAUTZENBERG)

By C. A. WRIGHT

SYNOPSIS

Many recent workers on the freshwater mollusca of Africa have tended to present synonymies based more on the literature and on personal opinions than on a study of the animals themselves. This is particularly true of those gastropods of medical importance which act as intermediate hosts for flukes parasitizing human beings. This paper is an attempt to establish the relation- ships of one such snail.

HISTORICAL

DAUTZENBERG (1890) described Bulinus jousseaumei as a member of the genus Isidora and stated that it belonged to the /. contorta group with a strong affinity to /. natalensis. Pilsbry & Bequaert (1927) list B. jousseaumei in the sub-genus Bulinus s. str. and Amberson and Schwarz (1953) place it in the synonomy of B. truncatus. I have shown that both on conchological and anatomical grounds the species is properly placed in the sub-genus Physopsis (Wright 1956). Smithers (1956) has shown that in parts of the Gambia Protectorate this snail is an important vector of the human blood-fluke, Schistosoma haematobium. From personal observations I am sure that it is also the vector implicated in the Casamance Province of Senegal by Deschiens (1951) under the name Bulinus trigonus, a species characteristic of some lakes in East Africa.

MATERIAL AND METHODS

The material of B. jousseaumei from the Gambia and Senegal used in this work was collected personally, in collaboration with S. R. Smithers. Material from other localities has mostly been seen in the collection of the British Museum (Natural History) or has been sent to me by other workers.

Shell measurements were made with an eyepiece micrometer in a binocular dissecting microscope, the shells were held steady in a horizontal position by means of a piece of plasticine on the microscope stage. Radula preparations were made following the rapid methods recommended by Meeuse (1950).

THE SHELL

The original description of the shell by Dautzenberg is very adequate and my re- description (Wright, 1956) adds little to the original, apart from noting a further colour variation from light yellowish-straw to dark reddish-brown, and amplifying

ZOOL. 5, i.

4 THE STRUCTURE AND TAXONOMY OF BULINUS JOUSSEAUME1

the details of the ornamentation. Since reference will be made later to the sculpture my previous observations are repeated here. There is a regular punctate pattern on the nuclear whorl and the rest of the shell is covered with fine growth lines superimposed on which is a delicate pattern of short, wavy, vertical lines. It is surprising that Dautzenberg, who must have been acquainted with Physopsis, was so definite about the affinities of B. jousseaumei in spite of his mention of the twisted base of the columella in his description.

One of the greatest difficulties in the description of gastropods is the definition of size. The old workers when describing a new species usually gave the dimensions of the specimen (often the largest) which they had selected as the holotype. Rarely, they gave the dimensions of the largest and the smallest specimens available to them. The dangers of this approach to the problem in an animal which grows steadily without a well-defined adult phase are obvious. Size ranges are valueless since they give no idea of distribution, and ranges given with a mean and standard deviation are not of much use unless the growth stages represented in the sample are known. The ratios of certain measurements of the shell are of more value but these may vary with shell length. For instance Peters (1938) has shown that the ratio shell length /aperture length increases with shell length in Lymnaea pahistris. Hubendick (1951) using smaller samples suggests that there is no significant change in the ratio aperture length/shell length x 100 in Lymnaea peregra.

In order to investigate the relationships of B. jousseaumei it has been found necessary to analyse the size composition of various samples. The largest of these samples was collected at a washing place in the Simoto Bolon near the village of Diabugu Basilla, Upper River Division, Gambia on the 4th March, 1955. Two- hundred and thirty-eight shells were measured and spirit material dissected to try to determine an approximate correlation between shell size and the attainment of the " adult " condition. The results obtained from a study of this sample appear to be characteristic for the other samples collected in the Gambia at the same time. The time of year at which a sample is collected is of great importance since it is almost certain that the population early in the wet season contains a much higher proportion of juveniles than it would some months after the end of the rains, the time when the material under consideration was collected. Text-fig, i is a size frequency histogram of this sample, based on shell length. It is obvious that the greatest frequency occurs in the 8-0-9-0 mm. shell length group and the mode of the graph is at the 8-5 mm. level. The graph is, however, asymmetrical due to an overweighting of the lower size groups, the mean shell length of the sample is therefore less than the modal length. The graph does not provide any indication of whether the mode is due to large numbers of adults or juveniles and this can only be determined by anatomical study. It will be shown later that it is in the size group 6-0-7-0 mm. shell length that the accessory reproductive glands reach a fully functional stage of their develop- ment and this is taken as the beginning of the adult phase of life. If the curve show- ing distribution on the upper side of the mode in Text-fig, i is drawn and its mirror image is reproduced on the low side a normal distribution curve is formed, the lower end of which coincides well with the known onset of the adult phase. If the frequency distribution is plotted on arithmetical probability paper following the

THE STRUCTURE AND TAXONOMY OF BULINUS JOUSSEAUMEI 5

SHELL LINGTH IN MILLIMITM*

SKILL UNGTH IN MILLIMETRE

FIG. i. Size-frequency histogram for a population of B. jousseaumei from

the Gambia.

FIG. 2. Graph of ratio shell length /aperture width against shell length for Gambian population of B. jousseaumei.

FIG. 3. Graph of ratio shell length /aperture length against shell length for the same population,

6 THE STRUCTURE AND TAXONOMY OF BULINUS JOUSSEAUMEI

method of Harding (1949) a curve is obtained which appears to be too complex to be analysed by the usual methods. This may in part be due to faulty sampling but is also undoubtedly caused by the complicated structure of a population of fresh water snails. There may be present the representatives of several successive generations, a few senescent survivors of previous generations, the adults of the present and the progeny of these adults. If the breeding season is continuous then a smooth curve might be expected but there is reason to believe that breeding is somewhat spasmodic in the bulinids and this might help to account for the complex composition of the population. Since the anatomical findings confirm that the mode of the histogram in Text-fig. I is in fact the mean of the adult part of the population it seems justifiable to accept 8-5 mm. as the mean shell length of the adults in this sample. An indication of the proportion of juveniles in the whole sample is given by the index of skewness, (mean-mode) /standard deviation. Having obtained a rough mean length for the adults in the population the adult means for the other dimensions may be calculated by increasing the population mean by the same percentage as that by which the mean population length differs from the adult mean length. The following table gives the mean shell dimensions of the population in millimetres and their standard deviations together with the approximate adult dimensions and the maximum sizes observed.

Aperture Aperture

Length Max. diam. length width

Mean S. D. Mean S. D. Mean S. D. Mean S. D.

Whole population . 7-84 ±1-52 . 6-00 +1-12 . 6-65 ±1-03 • 3'37 ±0-66 Adults . . -8-5 .6-50 .7-2 . 3-65

Maximum . . 1 1 • 8 . 8-0 . 9-0 . 4-5

Passing from the absolute dimensions of the shells to the ratios of some of these dimensions to one another, one is faced with a number of these proportions from which to choose the most useful. Hubendick (1951) pointed out that the selection of measurements in any particular case should be made with advance knowledge of the variation to be studied. In this study of B. jousseaumei the object of interest is not so much phenotypic or genotypic variation as changes in form during growth. With this in mind the two ratios chosen are those of shell length /aperture length (I/ml) and shell length /aperture width (1/mb). The first of these in this type of shell gives an indication of the exsertion of the spire for if the spire is strongly depressed the ratio will approach unity and will increase with the exsertion of the upper whorls, or, more properly, the descent of the body whorl. The second of the two ratios expresses the relationship between the increase in diameter of the body whorl with increasing shell length. A similar result might easily be obtained by using the ratio shell length /shell diameter but with the technique of measurement employed the width of the aperture was more accurately obtainable than the maximum diameter of the shell. Text-figs. 2 and 3 show these two ratios plotted against shell length ; in both it can be seen that there is an increase in the value of the ratio with increasing shell length. The change in the values is so slight that the means of the ratios provide a good index of shell form. The graph I/ml plotted against shell length

THE STRUCTURE AND TAXONOMY OF BULINUS JOUSSEAUMEI 7

(Text-fig. 3) shows an interesting feature. For small shell lengths the graph is practically a straight line parallel to the horizontal axis and the upward slope only becomes apparent after the 6-5 mm. shell length mark. As pointed out earlier, this is the approximate shell length at which adult anatomical characters become developed and the change in form of the shell at this point is, presumably, a reflexion of the anatomical changes. The actual change in form which the shell exhibits may be explained thus : in its early stages the spire is completely depressed, subsequent whorls being added around the preceding ones ; at the onset of maturity the accessory genital glands increase in size very rapidly necessitating an increase in shell volume to accommodate them and this is achieved by the body-whorl moving downward in relation to its predecessor, giving an increase in shell length without a corresponding increase in aperture length.

The means and standard deviations of these two ratios in the population under consideration are

I/ml mean 1-17 s.d. i 0-08

1/mb mean 2.31 s.d. ± 0-143

THE MANTLE

The mantle markings of B. joiisseaumei from the Gambia consist of patterns of small black spots and patches scattered irregularly over a light grey field. The spots often appear dark grey rather than black owing to the pigment granules of which they are composed being only loose aggregations rather than dense concentrations.

The nephridial ridge on the underside of the mantle may be well developed or almost absent. It is often present only on the distal end of the kidney. The inter- mediate ridge is almost always well developed and is about equal in length to the kidney. An examination of transverse sections of the mantle has failed to show the presence of ciliated epithelium on this intermediate ridge (described for B. africana by Hubendick, 1948) but this may be due to the method of fixation employed.

THE RADULA

The radula is in no way remarkable. The number of tricuspid laterals varies from six to eight in each transverse half-row. This number does not appear to change with the age of the snail, but the number of marginal teeth does appear to increase in older specimens. It is perhaps worthwhile recording here that the first lateral tooth is tricuspid (Text-fig. 4). Dupuis and Putzeys (1923) mention as the only real difference between Physopsis and Isidora that the first lateral in the former group is bicuspid while in the latter it is tricuspid. The endocone of the first lateral is often difficult to observe and it is doubtless this fact that gave rise to the erroneous statement of these authors.

CENTRAL NERVOUS SYSTEM

The central nervous system in the Planorbidae shows little variation between the various genera. The connectives between the ganglia are relatively long in B. jousseaumei, but since the material on which these anatomical observations are

8 THE STRUCTURE AND TAXONOMY OF BUL1NUS JOUSSEAUMEI

based was all narcotized and well extended such details may not be strictly compar- able with those of other workers (Text-fig. 5) .

The penial nerve appears to be a composite structure with its main source in the left cerebral ganglion ; a number of other fibres running with the main component arise from the left pedal ganglion. The cerebral part of the nerve branches off from a larger trunk arising near the origin of the left cerebro-buccal connective and passing forwards giving off finer branches to the sides of the head and lips. It is probable that the nerves arising from the cerebral ganglia have a primarily sensory function, while those from the pedal ganglia are mainly motor. This would mean that the composite nature of the penial nerve provides both sensory and motor innervation for the copulatory organ. The principal nerve arising from the dorsal surface of each of the cerebral ganglia runs to the tentacle, its associated lobe and the eye on the same side of the body as that from which it arises. The pedal ganglia send several large nerves down into the foot and running back from the two buccal ganglia is a pair of fine nerves, one on either side of the oesophagus. The three visceral ganglia send nerves to the organs in the visceral mass and one large trunk arising from the left visceral ganglion passes to the anal lobe and pseudo-branch, while a similar large trunk from the right visceral ganglion passes upwards to the mantle. Contributory evidence as to the sensory nature of the cerebral ganglia is obtained from the otocyst which, although partially embedded in the posterior side of the pedal ganglion appears to be innervated solely from the cerebral ganglion above (Text-fig. 6).

BLOOD CIRCULATORY SYSTEM

The heart lies within its extremely delicate pericardium on the mantle close to the proximal end of the kidney, above the point at which the oesophagus passes into the crop. The auricle receives blood from the very large vein running along the anterior edge of the kidney. The aorta leaving the ventricle is variable in length, it may pass right over the intestinal loop which curves round the gizzard before dividing (Text-fig. 7) or it may divide so far back that the ventricle has a bifid appearance. In either case, the lesser of the two branches follows the intestine on its course round the gizzard while the major branch passes upward over the posterior edge of the gizzard. As it passes over the space between the intestine and gizzard a very large artery passes down between these two organs and divides almost at once, one branch going to the crop and distal side of the gizzard, the other to the accessory genital glands and the head cavity. A short distance after this division of the main branch from the aorta a smaller branch is given off to supply the stomach and proximal part of the gizzard and the rest of the vessel continues up the side of the gizzard, past the point at which the digestive gland opens from the intestine, and then, giving off a number of side branches into the digestive gland, follows the path of the intestine as it loops up into the upper whorls of the body. The principal artery to the head follows beneath the oesophagus to the circum-oesophageal nerve ring where vessels supplying the ganglia are given off ; then, after passing through the nerve ring it divides. One branch passes vertically downwards as the pedal artery and the other continues forward to the underside of the buccal mass where the

THE STRUCTURE AND TAXONOMY OF BULINUS JOUSSEAUMEI g

I.QMM.

ZOOL. 5,

FIG. 4. Radula teeth of B. jousseaumei from Gambia.

FIG. 5. Central nervous system of B. jousseaumei.

Fig. 6. Right lateral view of central nervous system of B. jousseaumei to show

the otocyst and its innervation

FIG. 7. Heart and principal arteries of B. jousseaumei. (Figs. 5, 6 & 7 to same scale.)

io THE STRUCTURE AND TAXONOMY OF BULINUS JOUSSEAUMEI

vessel dilates before breaking up into fine branches supplying the muscles of the mass and the sides of the head and lips. A fine lateral branch which leaves the main trunk in the region of the nerve ring serves the penial complex. It runs parallel with the penial nerve and appears to enter the complex at the junction of the penis sheath and preputium.

ALIMENTARY SYSTEM

The digestive tract in the Planorbidae is so well known and subject to so little variation that it is not necessary to enter into a full description here.

REPRODUCTIVE SYSTEM

The gross morphology of the genital tract of B. jousseaumei has already been described (Wright, 1956). It is intended here to consider the histology of the tract and its development. Hubendick (1948 a & b) has described the anatomy and histology of the male copulatory organs of several species of Bulinus. Larambergue (1939) has described fully the reproductive anatomy and histology of Bulinus truncatus. Abdel-Malek (1954 a & b) has given detailed accounts of the morphology and histology of the genital organs of two Planorbids Helisoma trivolvis (Say) and Biomphalaria boissyi and comparisons between all of these and Bulinus jousseaumei will be made.

Histologically no real differences were found between Abdel-Malek's description of the ovotestis in Biomphalaria and Helisoma and that in the present species. The acini are enveloped in " Ancel's layer " of thin connective tissue and the germinal epithelium within this layer appears to line only the lower parts of the acini. Heavily pigmented connective tissue is largely confined to the layer covering the top of the organ. In the adult snail all stages of spermatogenesis and oogenesis can be observed in the same acinus. Young oocytes and spermatids are largely confined to the lower parts of the acini and the upper parts are occupied by maturing ova, as many as six having been seen in a single acinus as compared with two to three reported by Abdel-Malek in Helisoma trivolvis. The mature ova are enclosed within a follicular membrane made up of nurse cells and connective tissue. Mature spermatozoa are more or less ubiquitous in the acini, either free in the lumen or attached by their anterior ends to basal " Sertoli " cells. The motility of these basal cells is shown in that they may be found on the outer wall of the follicles of maturing ova, a position that could only be reached by their independent locomotion. The hermaphrodite duct is lined with an epithelium of cuboidal cells bearing short cilia and the duct is sheathed in a thin layer of connective tissue (Text-fig. 8). The epithelium lining the seminal vesicles is similar to that in the hermaphrodite duct but the median

FIG. 8. Transverse section of hermaphrodite duct of B. jousseaumei.

FIG. 9. Transverse section of first part of sperm duct of B. jousseaumei

FIG. io. Transverse section of sperm duct of B. jousseaumei.

FIG. ii. Epithelial lining of prostate tubule of B. jousseaumei.

FIG. 12. Transverse section of vas deferens within body wall of B. jousseaumei.

FIG. 13. Transverse section of vas deferens in head cavity of B. jousseaumei.

THE STRUCTURE AND TAXONOMY OF BULINUS JOUSSEAUMEI

13

12 THE STRUCTURE AND TAXONOMY OF BULINUS JOUSSEAUMEI

nuclei are a little larger and the cytoplasm is more granular. The separation of the male and female ducts from the common duct occurs at a point well embedded in the base of the albumen gland. The sperm duct at its source is narrow with a small lumen lined by short columnar epithelial cells, ciliated, and with median to basal nuclei and finely granular cytoplasm (Text-fig. 9). This part of the duct is quite short and it soon becomes considerably larger in diameter although the size of the lumen does not increase greatly. The epithelium lining the duct loses its cilia and the cells are of a much taller columnar type with median nuclei, larger than those in the earlier part of the duct, and the cytoplasm is filled with large, eosinophilic, retractile granules. A few wedge-shaped cells occur between the apical ends of the columnar type and these do not contain the eosinophilic refractile granules (Text-fig. 10). The sperm duct retains this histological structure right up to its entry into the prostate gland. Hubendick (1948(2) in discussing the structure of the prostate in Bulinus has shown that in B. inflatus, an Australian species, the tubules of the prostate open individually into the vas deferens which enters the gland at the proximal end and leaves it distally, a structure similar to that in Physa. In B. jousseaumei however the prostate is a discoidal structure, flattened on one side and convex on the other. The sperm duct enters the gland almost in the centre of the flattened surface and the vas deferens leaves it also almost in the centre. There is therefore a central point at which the tubules of the prostate discharge their secretion into the male duct and it is at this point that the histological structure of the male duct changes from the typical form of the sperm duct to that of the vas deferens. The tubules of the prostate gland are close-packed, each is ensheathed in a thin-layer of connective tissue and is lined by large secretory cells with large basal nuclei with granular contents. The cytoplasm of these cells is eosinophilic and not granular (Text-fig, n). There are some basophilic cells near the blind ends of the tubules. As the vas deferens leaves the prostate it is lined with sparsely ciliated epithelium of rather flattened cuboidal cells with central nuclei containing sparse chromatin granules. This epithelium is surrounded by a thin layer of circular muscle. As the duct proceeds on its course there is an increase both in the ciliation of the lining epithelium and in the thickness of the circular muscle layer. Where the duct passes through the body wall the lumen is large, the cells of the epithelial lining are wider than high, the cilia are plentiful but not long and the muscle layer is thick (Text-fig. 12). After leaving the body wall and entering the head cavity the structure of the vas deferens changes slightly. The lumen becomes more restricted, the epithelial lining more columnar, with basal nuclei, still with few chromatin granules in the clear nucleoplasm, the cilia become longer and the muscle layer thicker (Text-fig. 13). At the point of entry into the penis sheath the duct becomes the epiphallus which lies coiled in the upper part of the sheath and which has a very different appearance in section (Text-fig. 14). The overall diameter is reduced but that of the lumen remains more or less unchanged. The internal epithelium is of irregularly cuboidal cells with central nuclei and no cilia. The nuclei are densely packed with chromatin and stain deeply in haematoxylin. The epithelial layer is surrounded by a layer of circular muscle and outside this is a layer of transverse muscle and connective tissue with bundles of longitudinal muscle fibres embedded in it. The transition

THE STRUCTURE AND TAXONOMY OF BULINUS JOUSSEAUMEI 13

from epiphallus to penis proper is gradual. Due to the mode of operation of the copulatory organ in Bulinus [Larambergue (1939) and Hubendick (19486)] the part that is proximal when the organ is at rest is distal when it is erected and the distal resting part is proximal during copulation. The following description is based on resting specimens. Proximally, the lumen increases gradually, the epithelial lining becomes regularly cuboidal and the central nuclei are less deeply staining. At about the maximum diameter of the penis the lumen is partially occluded by profound folding of the lining. The epithelium is of cuboidal cells with central nuclei containing sparse chromatin granules and beneath this is a layer of connective tissue with transverse muscle fibres and blood spaces. Around this is a layer of muscle and the outer layer is of mixed muscle fibres with blood spaces and connective tissue (Text-fig. 15). The distal tip of the penis, just before the point where it unites with the penis sheath, has a tri-radiate lumen surrounded by the same sort of epithelium as before and outside this only a thin layer of circular muscle (Text- fig. 16). The penis sheath has an outer covering of a thin, flattened epithelium within which is a layer of mixed muscle fibres followed by an innermost layer of almost pure circular muscle. The lining of the preputium is thrown into a number of folds with two muscular pillars predominating. . Usually only one of these two pillars extends the whole length of the preputium, the other usually fades out before the junction with the penis sheath at the upper end. The position of the pillars within the preputium coincides with the point of attachment on the outside of two series of muscle fibres that connect the organ to the body wall of the head cavity. The epithelial lining of the preputium is fundamentally of columnar or cubical cells with more or less basal nuclei. In the most proximal parts of the organ a few ciliated cells are present. Distributed irregularly are gland cells, much larger than the other components of the epithelium, with basal nuclei often displaced to one side of the cell (Text-fig. 17). The cytoplasm of these gland cells may be either coarsely granular and eosinophilic or much more finely granular and staining deeply in haematoxylin. They are often clustered together in patches and may penetrate quite deeply into the muscular layers beneath. They are more common in the upper proximal parts of the organ but in one specimen examined they were confined almost exclusively to the epithelium covering one of the muscular pillars. Beneath the epithelium is a layer of circular muscle and outside this a thick layer of mixed muscles and connective tissue with blood spaces.

Returning to the point of separation of the hermaphrodite duct, the female tract begins with a short duct with a narrow lumen lined by columnar glandular cells with darkly-staining basal nuclei. Between the free ends of these columnar cells are wedge cells with nuclei containing few chromatin granules. These wedge cells bear short cilia (Text-fig. 18). This short duct opens into the carrefour which has a wide lumen and a folded wall. The epithelial lining (Text-fig. 19) is of tall columnar cells and wedge cells. The glandular columnar forms have dark staining basal nuclei. There are patches of ciliated cells particularly in the folds of the wall. Opening into the carrefour is the duct from the albumen gland. This gland is made up of numbers of tubules with loose connective tissue between them. The lining of the tubules is quite characteristic in section ; it consists of roughly cuboidal cells

i4 THE STRUCTURE AND TAXONOMY OF BULINUS JOUSSEAUMEI

with basal granular nuclei and with numbers of large droplets in the cytoplasm. These droplets stain deeply in haematoxylin (Text-fig. 20). Distally to the carrefour the oviduct is lined with a columnar epithelium containing a few wedge cells and a few gland cells but without cilia (Text-fig. 21). The oviduct leads into the uterus which is elongate transversely in cross section. The epithelial lining is similar to that of the oviduct but there are more gland cells, some with acidophilic and some with basophilic granules in the cytoplasm. There are also a few ciliated cells (Text- fig. 22). For a considerable part of its course the uterus is surrounded by the muci- parous gland. Macroscopically this gland is smooth, colourless and rather translucent in appearance. In section it is seen to be made up of close-packed tubules with little intervening connective tissue. The tubules are lined with large, irregularly cubical cells with darkly staining basal nuclei. The cytoplasm is entirely without granules and is completely unstained by either haematoxylin or eosin (Text-figs. 23). The tubules open individually into the dorsal side of the uterus. Following immediately after the muciparous gland the uterus passes into the oothecal gland, distinguished macroscopically by its opaque white appearance in contrast to the translucent colourlessness of the previous gland. Internally the uterine wall is deeply folded in this region. The epithelium (Text-fig. 24) consists almost entirely of tall columnar glandular cells with basal nuclei. Both acidophils and basophils are present but the coarsely granular or eosinophilic type of cytoplasm predominates. A few patches of ciliated cells are present and the glandular cells may be several layers thick in places, penetrating deeply into the underlying connective tissue. The transition from the folded, glandular wall of the oothecal gland to the tubular, ciliated vagina is quite abrupt. The epithelial cells in this part are of a short columnar type with median to basal nuclei containing sparse chromatin granules. The cytoplasm of these ciliated cells is very finely granular and eosinophilic (Text-fig. 25). The seminal receptacle duct which opens from the vagina is lined in its distal part (that nearest to the vagina) with a columnar epithelium with basal nuclei and finely granular acidophilic cytoplasm but without cilia. The rest of the duct lining is similar but the columnar cells are taller and ciliated. The distended sac of the

FIG. 14. Transverse section of epiphallus of B. jousseaumei. FIG. 15. Part of transverse section of penis of B. jousseaumei. FIG. 1 6. Transverse section of tip of penis of B. jousseaumei.

FIG. 17. Part of epithelial lining of the preputium of B. jousseaumei showing

basophilic and eosinophilic gland cells.

FIG. 18. Epithelial lining of proximal part of oviduct of B. jousseaumei.

FIG. 19. Epithelial lining of carrefour of B. jousseaumei.

Fig. 20. Epithelial lining of albumen gland tubule of B. jousseaumei.

FIG. 21. Epithelial lining of distal part of oviduct of B. jousseaumei.

FIG. 22. Epithelial lining of uterus of B. jousseaumei.

FIG. 23. Epithelial lining of muciparous gland tubule of B. jousseaumei.

FIG. 24. Epithelial lining of oothecal gland of B. jousseaumei.

FIG. 25. Epithelial lining of vagina of B. jousseaumei.

THE STRUCTURE AND TAXONOMY OF BULINUS JOUSSEAUMEI 15

16 THE STRUCTURE AND TAXONOMY OF BULINUS JOUSSEAUMEI

receptaculum is thin walled, a thin layer of connective tissue lying outside a tall columnar epithelium with basal nuclei and indistinct cell boundaries.

The foregoing description is based on a study of sections of the genital organs of several adult snails in the 8-o-9'5 mm. shell length range. For comparison series of sections were cut through the reproductive organs of a juvenile snail (4*5 mm. shell length) and an " adolescent " specimen 5-7 mm. long.

The gonad of the juvenile showed neither mature ova nor spermatozoa. Active cell division appeared to be in progress in the zone of the germinal epithelium near the bases of the acini but the state of fixation of the specimen made definite observa- tions difficult. No sections of the haemaphrodite duct were obtained. The male tract was not clearly defined until the level of the prostate gland. This organ was present but in section the tubules although defined were lined with an undifferentiated cuboidal epithelium with central nuclei. The vas deferens leading from the prostate was lined with a similar unciliated epithelium. The sheath of circular muscle present in the adult snail was represented by close-packed undifferentiated connective tissue cells. The duct remained more or less unchanged in this condition throughout its course. The epiphallus was again ensheathed in undifferentiated tissue and the epithelial lining was of more columnar cells with basal nuclei. The penis and penis sheath at this stage also showed no clear definition of muscle tissue and the separation between the muscle layers of the two parts was just becoming apparent. The folding of the inner wall of the penis was already in evidence. The preputium also was sur- rounded by undifferentiated muscular tissue and the lumen, already S-shaped owing to the development of the two main muscular pillars, was lined with a cuboidal epithelium with central nuclei. Of the female system at this stage little can be said. The uterus and vagina are present as tubes lined by a uniform epithelium. The albumen gland is entirely undeveloped, likewise the muciparous and oothecal glands. The receptaculum seminis and its duct are present but the receptacle sac is scarcely more than a slight dilatation of the duct. The only evidence of differentiation in the female tract in this specimen is that the nuclei of the epithelial cells in the regions that will become glandular stain more deeply in haematoxylin than do those of the other regions.

The gonad of the " adolescent " snail (5-7 mm. shell length) showed clusters of mature spermatozoa in the acini with a few maturing oocytes in the upper parts. Some of the largest of these oocytes showed dividing nuclei and were, presumably, undergoing maturation division. Sections of the seminal vesicle and hermaphrodite duct were packed solidly with spermatozoa. The sperm duct after its separation from the common duct shows the same short, ciliated part followed by the long, glandular part with refractile eosinophil granules in the cytoplasm as is found in the adult. The prostate gland also is identical histologically to the adult. Through- out its course the vas deferens corresponds in histological detail to the form already described in the adult. The transition to the epiphallus is similar as are the structures of the penisand penis sheath. The preputium, however, is, in its distal part, similar to that in the juvenile in that the lumen is S-shaped and only two muscular pillars are present. More proximally additional folds in the wall do occur. The epithelium lining the lumen is more ciliated than in the adult and few gland cells have been

THE STRUCTURE AND TAXONOMY OF BULINUS JOUSSEAUMEI 17

observed. The female tract at this stage is far less well developed. The albumen gland is represented by a fairly compact mass of tissue of undoubtedly glandular nature but scarcely organized into tubules as in the adult. The few tubules which are present are not lined with the characteristic glandular epithelium of the adult but with cells which probably later develop into this form. The carrefour is present as a dilatation of the oviduct but its epithelial lining is only differentiated into glandular areas in parts, some cilia are also present. The oviduct is similar histo- logically to that in the adult and it in turn passes into the ciliated part of the uterus. Neither the muciparous nor the oothecal glands is represented by more than a slight thickening of the uterine wall. The receptaculum seminis is well developed and contains spermatozoa and the vagina is strongly ciliated.

These histological observations support the opinion already formed from gross anatomical studies that the male genital tract develops slightly earlier than the female system. Morton (1954) has suggested that a protandrous sexual cycle is the primitive condition in gastropods and that simultaneous hermaphroditism has been later developed in the higher pulmonates and opisthobranchs. Larambergue (1939) states that in Bulinus contortus there is no protandry since spermatozoa and ova are produced simultaneously throughout life even though spermatozoa do appear first. It seems probable that the earlier development of the male tract is a relic of the primitive protandrous condition which has become almost completely obscured, particularly where the length of the breeding season is limited by adverse environmental conditions.

To complete this histological study of the genital organs of B. jousseaumei a few comparisons with similar studies on related forms should be considered. It has already been said that there are no differences between the fundamental structure of the gonad in this species and in Biomphalaria boissyi and Helisoma trivolvis as described by Abdel-Malek (loc. cit.) and it is probable that this structure is fairly uniform throughout the Planorbidae. The hermaphrodite duct corresponds to the description by Larambergue (loc. cit.) for that in B. contortus, also the point of separation of the male and female tracts. The sperm duct differs from that in B. contortus in that no ciliation has been observed in its lumen apart from the very short part immediately after its separation from the common duct. Abdel-Malek mentions no ciliation of the sperm duct in B. boissyi and only near the prostate in H. trivolois. In both of these species he mentions the refractile, eosinophilic, cytoplasmic granules but also records the presence of scattered basophils not seen in B. jousseaumei. The prostate corresponds well with that described by Larambergue for B. contortus and this author stresses the fact that the gland is not traversed by the male duct but that its tubules open into a central chamber into which the sperm duct opens at one side and from which the vas deferens leaves at the other. The vas deferens in B. jousseaumei differs from that in B. contortus in that it is ciliated throughout its length while Larambergue reports that in the second species the epithelium is unciliated and of a mucous-secreting type. In both of the species described by Abdel-Malek the vas deferens is ciliated throughout its length although, due to the different form of the prostate in these species, there is only a gradual transition from the form of the sperm duct to that of the vas deferens instead of a

i8 THE STRUCTURE AND TAXONOMY OF BULINUS JOUSSEAUMEI

clear-cut demarcation of the two as in B. jousseaumei. The epiphallus which is so well differentiated from the vas deferens in the present species was also noted (although not under this name) by Larambergue to be histologically different. Since the structure of the copulatory apparatus in the species described by Abdel- Malek differs so markedly from that in the Bulinids it is not possible to draw com- parisons between the histology of the two but it is interesting to note that in his species the seminal canal within the penis (and therefore that part of the male duct actually inside the penis sheath) is ciliated and that the epithelial lining does not differ from that of the vas deferens. The histological structure of the penis, penis sheath and preputium in the present species show no significant differences from those described by Hubendick (19486) and Larambergue. Larambergue does not describe the histology of the female genital tract of B. truncatus apart from mention- ing the ciliation of the receptaculum seminis duct. The epithelial linings of the first part of the oviduct and of the carrefour are similar to those of B. boissyi and H. trivolvis although not so heavily ciliated. Beyond the carrefour the epithelium of the oviduct in the present species almost entirely lacks cilia while these are present in Biomphalaria. The epithelium of the uterus is similar in the two species, and the muciparous gland merges gradually with the uterus rather than abruptly as in Helisoma. The remainder of the female tract is very similar to Biomphalaria except that the part of the receptaculum seminis duct nearest to the vagina is unciliated, a condition similar to that in Helisoma.

RELATIONSHIPS OF B. JOUSSEAUMEI

In considering the affinities of B. jousseaumei it is necessary to review the species of the sub-genus Physopsis known from West Africa. For this purpose it is often necessary to refer to the medical literature as it is largely in works on the epidemiology and transmission of human schistosomiasis that records of intermediate hosts are to be found. The only species of Physopsis actually described from West Africa is P. globosa (Morelet, 1866) collected in Angola. Of the other twenty or so species, all were described originally from East or South Africa. The question of the identity of P. globosa with P. africana Krauss, 1848, will not be dealt with here since opinions are still divided on this point and much detailed work must be carried out before a solution of the problem can be reached. It must, however, be pointed out that many authors now treat P. globosa as an absolute synonym of P. africana others consider it to be a variety or sub-species of the latter while still others recognize it as a separate species. In the following account it will be pointed out (where it is known) which of these three courses was adopted by the workers concerned. Since a chronological account of these records would undoubtedly prove confusing, they are presented in a geographical sequence from north to south.

McCullough & Duke (1954) were the first to record Bulinus africanus from the Gambia and they noted that this was probably the northernmost record for the species in West Africa. They were following the classification of Amberson & Schwarz (1953) who treated all species of Physopsis as forms of B. africanus and there is no doubt that their record refers to B. jousseaumei. I later recorded (Wright,

THE STRUCTURE AND TAXONOMY OF BULINUS JOUSSEAUMEI 19

1956) a form of B. (P.) globosus from one locally in Upper River Division, Gambia, as well as B. jousseaumei from the Casamance Province of Senegal, the next territory to the south of the Gambia. No published records of Physopsis from Portuguese Guinea have been found, and Pinto (1949) in a recent survey of vesicle schistosomiasis in that territory found only Bulinus forskali and B. dautzenbergi. Vogel (1932) in a similar survey in French Guinea and Liberia found (and illustrated) Physopsis globosa. More recently in a rather confusing account of schistosomiasis in French West Africa, Le Gall (1944) mentions as a probable vector of urinary schistosomiasis in French Guinea " Physopsis ovoides and boissyi " from Kissidougon. This presumably refers to P. ovoideus and Biomphalaria boissyi. The next territory southwards from the French Guinea coast, Sierra Leone, is perhaps one of the best documented areas in tropical Africa with respect to the epidemiology of schistosomiasis. It has been the subject of three major surveys. Blacklock (1924), Blacklock & Thompson (1924), Blacklock (1925), Gordon, Davey & Peaston (1934) and Gerber (1952). Connolly (1928) published an account of the freshwater molluscan material collected by Blacklock and considered the Physopsis to be P. globosa with strong affinities to P. didieri. He mentioned the presence of a well- marked spiral sculpture on the material from Sierra Leone, a feature which he had not observed on P. globosa from Angola and Mozambique. Connolly also identified material submitted to him by Gordon and his co-workers and identified the Physopsis as P. globosa. Gerber submitted his material to Dr. W. J. Rees who identified one fully-grown and three smaller specimens out of a batch of 204 shells as Bulinus globosus and referred the remaining 200 to B. africanus. The same sample of shells was shown to Berry who pronounced them all to be Physopsis africana. To the south of Sierra Leone is Liberia and here Vogel (loc. cit.) records Physopsis globosa and Veatch (1946) mentions Physopsis africana var. globosa as the intermediate host of Schistosoma haematobium in the Western Province. No records of identified Physopsis from the Ivory Coast have been seen but Ingram (1924) incriminated P. globosa as the possible vector of urinary schistosomiasis in the Gold Coast ; more recently Edwards & McCullough (1954) have demonstrated that the parasite is carried by P. africana but they also mention that they consider P. globosa to be either a race or an absolute synonym of this species. The only further record of an identified Physopsis between the Gold Coast and the Belgian Congo is that of P. globosa collected at Kano in Northern Nigeria (Gordon, 1932). In the excellent work of Pilsbry £ Bequaert (1927) a variety of P. africana is recorded from a number of localities in the Belgian Congo, also P. africana globosa. These authors refer the other recorded species of Physopsis (apart from P. tanganyicae von Martens) from the Belgian Congo either to P. africana var. or to P. africana globosa. Finally, the territory to the south of the Belgian Congo, Angola, is the type locality for P. globosa Morelet.

Examination of material from French Guinea, Sierra Leone, Liberia, Belgian Congo and Angola has shown that without doubt the same species of Physopsis is present in all these territories and that B. jousseaumei is a form of that species. The table below gives the mean dimensions in millimetres of the populations examined from the territories mentioned above ;

20 THE STRUCTURE AND TAXONOMY OF BULINUS JOUSSEAUMEI

Territory and number of specimens	Shell length	Shell max. Aperture diam. length	Aperture	Shell length Shell length
width Aperture width Aperture length
Gambia . 283	- 7-84 - - ±1-52 •	6-00 . ±1-12 .	6-65 . ±1-03 •	3-37 • ±0-66 .	2-31 ±0-14	1-17 ±0-08
Casamance 20 .	8-69 . . ±0-96 .	6-32 . ±0-76 .	7-21 . ±0-70 .	3-60 . ±o-33 -	2-41 ±0-13	1-20 ±0-04
French Guinea 20 .	- 10-33 .	7-OI . ±0-42 .	8-31 - ±0-51 •	4-04 . ±0-20 .	2-55 ±0-11	1-24 ±0-05
Sierra Leone . IOO	10-07 • . Jt 2 • O6 .	7-18 . ±i-75 •	8-43 -	3-97 - ±0-93 -	2-55 ±0-19	±0-08
Liberia . 6 .	10-56 . . ±2-81 .	±1-63 -	8-8 . ±1-98 .	4-28 . ±1-06 .	2-47 ±0-15	±0-07
Belgian Congo IOO	. ±1-64 .	8-50 . ±1-22 .	9-51 - ±1-52 .	4-54 • ±0-63 .	2-48 ±0-11	I-I9 ±0-065

Angola (Type series) 13-43 . 9-51 . 9-60 . 5-13 . 2-62 . 1-40

26 . . . ±2-19 . ±1-63 . ±1-48 . ±0-85 . ±0-15 . ±0-085

Text-fig. 26 shows the means for the ratio Shell length /Aperture length plotted against the mean lengths for each of these samples. Text-figs. 27-30 show the ratios shell length /aperture length and the frequency distributions of shell length in the samples from Sierra Leone and the Belgian Congo.

Reference to the table shows a gradual decrease in the dimensions of the shell from south to north. The composition of the samples is not wholly comparable, those from the Gambia, Sierra Leone and Belgian Congo are more or less random popula- tion samples with a proportion of juveniles and those from Casamance, French Guinea and the type series from Angola contain mostly adult specimens. The small sample from Liberia consisted of four adults and two juveniles, hence the rather large standard deviations. The means of the type series differ from those given by Mozeley (1939) because his figures were based on ten specimens of the series only. In addition to the discrepancies in sampling certain differences due to ecological conditions also occur. The most obvious of these is the " still water " effect. Schwetz (1954) has described the effect of changed environment on shell form in several planorbid snails and the exsertion of the spire in forms which develop in static water compared to those in gently moving streams is well known. This accounts largely for the difference in the mean values of the ratio shell length/ aperture length in the type series and in the Belgian Congo sample. The latter came from a pool in a recently dried stream bed while Morelet (1868) notes that the Angolan material was collected in a lake. Similarly this probably accounts for some of the differences between Gerber's and Blacklock's specimens from Sierra Leone. Gerber's material from the still waters of a rice swamp has a more exserted spire than Blacklock's from a stream,

THE STRUCTURE AND TAXONOMY OF BULINUS JOUSSEAUMEI 21

27

I/ML U

28

29

30

FIG. 26. Graph of mean ratio shell length /aperture length against mean shell length for populations of B . jousseau mei from Gambia (G), and Casamance (C) and B. globosus from French Guinea (FG), Sierra Leone (SL), Liberia (L), Belgian Congo (BC) and the type series from Angola (A) .

FIG. 27. Size-frequency histogram for a population of B. globosus from Belgian Congo.

FIG. 28. Graph of ratio shell length /aperture length against shell length for the same population.

FIG. 29. Size-frequency histogram for sample of B. globosits from Sierra Leone.

FIG. 30. Graph of ratio shell length /aperture length against shell length for the same population.

22 THE STRUCTURE AND TAXONOMY OF BULINUS JOUSSEAUMEI

The general form of the shell is similar throughout the whole range under considera- tion (Plate I). In the same population shells may be seen with the columella margin fused entirely to the body whorl and others with a space between the two. This was stressed by Connolly (1934) as being the only constant conchological difference between Physopsis africana and P. globosa. There is also a range of variation in the angle at which the outer lip meets the body whorl at the top of the aperture, a character used by Pilsbry & Bequaert (1927) for the separation of the same two species. The sculpture pattern described earlier for the Gambian material is found also throughout the range, being most marked in the specimens from Sierra Leone (Plate 2,) and, as noted by Connolly (1928) practically absent from the Angolan material. This is of interest since Mandahl-Barth (1954) erected a sub- species Bulinus globosus ugandae which he separated from the nominate sub-species because it completely lacked the spiral sculpture of the typical form. In passing it seems appropriate to note that amongst the shells collected in Northern Rhodesia by Buckley (1946) were a number of specimens labelled by Connolly as juvenile Bulinus natalensis but which were also referrable to B. hemprichii depressus Haas, 1936. These specimens were indistinguishable from many of the juvenile Phy- sopsis globosa seen during this work. Unfortunately the type specimens of Haas' subspecies were destroyed or lost during the war but from a study of the photo- graphs of these specimens there can be little doubt that the sub-species should be placed in the synonymy of Bulinus (Physopsis) globosus.

It can be seen from Text-fig. 30 that the pattern of the graph of shell length/ aperture length against shell length for the sample from Sierra Leone is similar to that for the Gambian population (Text-fig. 3). For the lower values of shell length the graph is almost parallel to the horizontal axis and the gradual upward slope does not begin until a shell length of about 8-0 mm. is reached. It has already been shown that this change is brought about by the onset of the adult phase and it should be noted that it occurs at a greater shell length in the Sierra Leone population than

FIGS. 31-35. Stages in the development of the accessory genital glands and male copulatory organ of B. jousseaumei from the Gambia.

Fig. 31 at 4-5 mm. shell length. Fig. 32 at 5-2 mm. shell length. Fig. 33 at 6-3 mm. shell length. Fig. 34 at 7-5 mm. shell length. Fig. 35 at 8-8 mm. shell length.

FIGS. 36-38. Stages in the development of the accessory genital glands and male copulatory organ of B. globosus from Sierra Leone.

Fig. 36 at 6- 1 mm. shell length. Fig. 37 at 8-5 mm. shell length. Fig. 38 at 9-2 mm. shell length.

FIGS. 39-41. Similar stages in B. globosus from Angola. Fig. 39 at 8-9 mm. shell length. Fig. 50 at 9-5 mm. shell length. Fig. 41 at 9-75 mm. shell length.

THE

STRUCTURE AND TAXONOMY OF BULINUS JOUSSEAUMEI 23

36

39

41

24 THE STRUCTURE AND TAXONOMY OF BULINUS JOUSSEAUMEI

in that from the Gambia. That the same characteristic is present in the correspond- ing graph for the Belgian Congo material is almost certain ; the distribution of individuals in the sample is, however, such that the feature is not well marked al- though it seems probable that it occurs at an even greater size than in the Sierra Leone population.

Although no anatomical differences in structure have been found in the limited material examined (Sierra Leone, Belgian Congo and Angola) there is a variation in degree of development of the genital organs at corresponding shell sizes. Text-figs. 31-35 show stages in the development of the accessory genital glands and male copulatory organ at various shell lengths for B. jousseaumei in the Gambia. Text- figs. 36-38 are of three stages in B. globosus from Sierra Leone and Text-figs. 39-41 for material from Angola. The degree of development of the Gambian specimen of 5-2 mm. shell length (Text-fig. 32) compares well with that of 8-5 mm. from Sierra Leone (Text-fig. 37) and 9-5 mm. from Angola (Text-fig. 40). The protandrous development of the male copulatory organ and prostate is even better marked in these last two groups of specimens than it is in B. jousseaumei.

Of possible significance, but at present insufficiently investigated, is the increased pigmentation of the mantle in the more southern forms. Specimens from Sierra Leone and the Belgian Congo have the mantle heavily blotched with black in contrast to the more diffuse spotting of that in B. jousseaumei. Three of the four specimens dissected from Angola had almost the entire mantle black with one or two lighter patches.

Differences in the radula do not seem significant ; one or two more laterals in each transverse half row may be found in the larger forms but this is to be expected.

DISCUSSION

Evidence has been presented to show that Bulinus jousseaumei from the Sene- gambian region is closely related to the other species of the sub-genus Physopsis from neighbouring West African territories and that these in turn are related to the typical Bulinus globosus. In spite of a number of distributional gaps, probably due to an absence of collectors rather than of snails, it seems clear that there is a well- marked cline grading from the typical globosus form in the south to the small jousseaumei at the extreme northern limit of the range. Not only is this cline represented by a gradation in size but also by a gradation in the degree of protandric development, possibly also by differences in the intensity of mantle markings. The change does not become really well marked until (moving northward) Sierra Leone is reached. It seems very probable that the cline is correlated with the length of the rainy season. In the Senegambian region there is a single wet season of about four months duration while to the south the season is prolonged and may be duplicated. The short single wet season will limit the time during which the streams and bolons are suitable for the development of snails, hence the telescoping of the sexual phases and the reaching of sexual maturity at a smaller shell size. In regions of more continuous rainfall the need for rapid development is less, resulting in the more marked protandrous development (a primitive character) and the later

THE STRUCTURE AND TAXONOMY OF BULINUS JOUSSEAUMEI 25

onset of sexual maturity, the latter being of course closely connected with the larger shell size.

Although the use of a trinomial system of nomenclature is of doubtful value unless it is well documented it seems justifiable to retain the name jousseaumei as a sub- species in this case, for the Senegambian form. It differs from the typical form of globosus in its considerably smaller size and its apparently more contracted life cycle. It is at present geographically isolated from the typical form by the distribu- tional gap in Portuguese Guinea but subsequent work may well show that this is not actually so. The name Bulinus (Physopsis) globosus jousseaumei (Dautzenberg) is therefore proposed.

If B. jousseaumei is a sub-species of B. globosus then the problem of its affinities with the strongly umbilicate form of B. globosus described from the Gambia (Wright, 1956) arises since two geographical races of the same species are not to be expected in the same geographical region. In this form the columellar margin of the aperture is greatly developed and not reflected, giving rise to a wide umbilicus with a slight keel around its opening. The columellar truncation is also thus suppressed leaving only a thickened line on the inner surface of the columella. There appears to be no record in the literature that the type series of B. globosus includes several specimens which show this character in a very limited degree. Material in a collection made in Northern Rhodesia by Dr. P. Le Roux contains a number of specimens which show this character even further developed. Intermediate forms between the Rhodesian and Gambian specimens have been seen in Blacklock's collection from Sierra Leone. In these the umbilicus is well developed but not quite so wide as in the Gambian material. Although the evidence is incomplete it appears that this variety of B. globosus also shows a clinal distribution parallel to that described for the typical form and B. jousseaumei. In this instance the main character in which gradation has been observed is in the degree of overgrowth of the columellar margin with con- sequent suppression of the columellar truncation and increase in the size of the umbilicus. Insufficient spirit material has been available for a study of associated anatomical variation. The only locality from which this umbilicate form was obtained in the Gambia was at Badja Kunda, Upper River Division where the ecological conditions differed slightly from those in the typical B. jousseaumei habitats. A single specimen of B. jousseaumei showing this character slightly developed was obtained with normal specimens at Sudowol bridge over the Simoto bolon, Upper River Division. The evidence suggests that this is possibly a recessive genetic character of the normal form which, when it occurs as a pure homozygous population is better adapted to slightly different habitats and appears to behave as a separate species.

SUMMARY

1. A brief biometrical study of the shell of B. jousseaumei is made.

2. The histology and course of development of the genital tract of this snail is described and compared with related planorbids.

3. The records of Physopsis spp. from West Africa are briefly surveyed. The possibly fallacious premises on which Bulinus globosus ugandae Mandahl-Barth

26 THE STRUCTURE AND TAXONOMY OF BULINUS JOUSSEAUMEI

was described are mentioned and Bulinus hemprichii depressus Haas is referred to the synonymy of B. globosus.

4. The relationship of B. jousseaumei to B. globosus is discussed and the former is reduced to a sub-species of the latter as B. g. jousseaumei, the northernmost representative of a cline of the typical form.

5. The relationship of B. g. jousseaumei to the umbilicate form of B. globosus is discussed and this form is related through a graded series to the typical form.

ACKNOWLEDGMENTS

In addition to acknowledgments made in the text I am greatly indebted to Dr. T. P. Eddy, Director of Medical Services, Sierra Leone, and to Dr. J. Schwetz of Brussels, both of whom have sent to me useful material from Sierra Leone and the Belgian Congo respectively ; also to Dr. John Morton who has been kind enough to read the manuscript and make a number of useful suggestions. My thanks are also due to the Director of the Institut Royal des Sciences Naturelles de Belgique both for the photograph of the type specimens of Isidora jousseaumei and for permission to publish this photograph.

REFERENCES

ABDEL-MALEK, E. T. 19540. Morphological studies on the family Planorbidae (Mollusca : Pulmonata). I. Genital organs of Helisoma trivolvis (Say) (Subfamily Helisomatinae F. C. Baker, 1945). Trans. American Microscopical Soc. 73 (2) : 103-123.

19546. Morphological studies on the family Planorbidae (Mollusca : Pulmonata). II.

The genital organs of Biomphalaria boissyi (Subfamily Planorbinae, H. A. Pilsbry 1934), Ibid. 73 (3) : 285-296.

AMBERSON, J. M. & SCHWARZ, E. 1953. On African Schistosomiasis. Trans. Roy. Soc. Trop.

Med. Hyg. 47 (6) : 451-502. BLACKLOCK, D. B. 1924. Report of an investigation into the prevalence and transmission

of human schistosomiasis in Sierra Leone. Sierra Leone Ann. Med. San. Rep. for 1923 :

80-87.

1925. Endemic goitre and schistosomiasis in Sierra Leone. Trans. Roy. Soc. Trop. Med.

Hyg. 18 (8) : 395~427-

& THOMPSON, M. G. 1924. Human schistosomiasis due to S. haematobium in Sierra

Leone. Ann. Trop. Med. dx Parasit. 18 (2) : 211-234.

BUCKLEY, J. J. C. 1946. A helminthological survey in Northern Rhodesia. /. Helminthology,

21 (4) : 111-174. CONNOLLY, M. 1928. The non-marine mollusca of Sierra Leone. Ann. Mag. Nat. Hist.

Series 10, 1 (4) : 529-551.

1934. On the planorbid hosts of bilharziasis in South and West Africa. Ann. Trop. Med.

& Parasit. 28 (3) : 439-443.

DAUTZENBERG, P. H. 1890. R6coltes malacologiques de M. le capitaine Em. Dorr, dans le Haut-Senegal et le Soudan Francais de 1886-1889. Mem. Soc. Zool. France, 3 : 132-134.

DESCHIENS, R. 1951. Le probleme sanitaire des bilharzioses dans les territoires de 1'Union Fran9aise.jBwW. Soc. Path. Exot. 44 : 631-638.

DUPUIS & PUTZEYS. 1923. Deuxieme note concernant le Faune Malacologique Africaine. Ann. Soc. Zool. et Malac. Belgique, 53 : 69-79.

EDWARDS, E. E. & McCuLLOUGH, F. S. 1954. Studies on the life cycles of Schistosoma haema- tobium and S. mansoni in the Gold Coast. Ann. Trop. Med. and Parasit. 47 : 164-177.

GERBER, J. H. 1952. Bilharzia in Boajibu. Part i. /. Trop. Med. <&• Hyg. 55 (3) : 52-58.

• 1952. Bilharzia in Boajibu. Part 2. Ibid. 55 (4) : 79-93.

THE STRUCTURE AND TAXONOMY OF BULINUS JOUSSEAUMEI 27

GORDON, R. M. 1932. The molluscan host of Schistosoma haematobium in Northern Nigeria. Ann. Trop. Med. & Parasit. 26 : 117-118.

DAVEY, T. H. & PEASTON, H. 1934. The transmission of human bilharziasis in Sierra

Leone, with an account of the life-cycle of the schistosomes concerned, S. mansoni and S. haematobium. Ibid. 28 (3) : 323-418.

HAAS, F. 1936. Binnen-Mollusken aus Inner Afrika. Abh. Senckenberg. Naturf. Ges. 431 :

1-156. Frankfurt A.M. HARDING, J. P. 1949. The use of probability paper for the graphical analysis of polymodal

frequency distributions. J.M.B.A. 28 : 141-153. HUBENDICK, B. 19480. The Anatomy of Bulinus, with a discussion of the term prostate and

its sense in the Basomatophora. Proc. Malac. Soc. Land. 27 (5) : 186-196.

19486. Studies on Bulinus. Ark. for zoologi, 40A (16) : 1-63.

1951- Recent Lymnaeidae. Kungl. Svenska Vetenskap. Hand. Fjarde Serien, 3 (i) :

1-223.

INGRAM, A. 1924. Note on a possible intermediate host of Schistosoma haematobium in the

Gold Coast. Ann. Trop. Med. &> Parasit. 18 (3) : 265-266. LARAMBERGUE, M. DE. 1939. Etude de 1'autofecondation chez les gasteropodes pulmones.

Recherches sur 1'aphallie et la fecondation chez Bulinus (Isidora) contortus Michaud. Bull.

Biol. Paris, 73 : 1-231. LE GALL, R. 1944. Les bilharzioses en Afrique Occidentale Fra^aise au Togo et a Madagascar

de 19390 1941. Bull. Off. Int. d'Hyg. Pub. 36 : 116-126. MANDAHL-BARTH, G. 1954. The freshwater mollusks of Uganda and Adjacent territories.

Ann. Mus. Roy. Congo Beige Tervuren, Serie in 8°, Sciences zoologiques 32. McCuLLOUGH, F. S. & DUKE, B. O. L. 1954. Schistosomiasis in the Gambia, i. Observa- tions on the potential snail vectors of Schistosoma haematobium and S. mansoni. Ann.

Trop. Med. & Parasit. 48 (3) : 277-286. MEEUSE, A. D J. 1950. Rapid methods for obtaining permanent mounts of radulae. Basteria,

14 (2 & 3) : 28-43. MORELET, A. 1866. Coquilles nouvelles recueilles par le Dr. Fr. Welwitsch dans 1'Afrique

equatoriale et particulierement dans les provinces portugaises d'Angola et de Benguela.

/. de Conchy Ho logie, 14 : 153-163.

— • — 1868. Mollusques terrestres et fluviatiles. Voyage du Dr. Friederick Welwitsch, Paris. MORTON, J. E. 1954. The pelagic mollusca of the Benguela Current. With an account of

the reproductive system and sexual succession of Limacina bulimoides. Discovery Reports,

27 : 163-200. MOZELEY, A. 1939. The freshwater mollusca of the Tanganyika Territory and Zanzibar

Protectorate and their relation to human schistosomiasis. Trans. Roy. Soc. Edinburgh, 59 (3) :

687-744. PETERS, B. G. 1938. Biometrical observations on shells of Limnaea Species. /. Helminth.

16 (4) : 181-212. PILSBRY, H. A. & BEQUAERT, J. 1927. The aquatic molluscs of the Belgian Congo. Bull.

Amer. Mus. Nat. Hist. 53 (2) : 69-602. PINTO, A. R. 1949. Os primeiras dados sobre a existencia da Schistosomiase vesical na

Guin6 Portuguese e importancia da contagem de ovas do parasito no sedimento urinario.

Anais. Inst. Med. Trop. Lisboa 6 : 75-114. SCHWETZ, J. 1954. L'influence du milieu sur la taille et la forme du meme Planorbe ou du

meme Bulinus. Ann. Soc. Roy. Zool. Beige, 85 (i) : 23-34. SMITHERS, S. R. 1956. On the ecology of schistosome vectors in the Gambia with evidence of

their r61e in transmission. Trans. Roy. Soc. Trop. Med. &• Hyg. 50 (4) : 354-365. VEATCH, E. P. 1946. Human trypanosomiasis in Liberia 1941-44. Supplement to Amer. J.

Trop. Med. 26 (5) : 1-56. VOGEL, H. 1932. Beitrage zur epidemiologie der schistosomiasis in Liberia und Franzosisch-

Guinea. Arch. Schiffs-u. Tropenhyg. 36 (3) : 108-135.

28 THE STRUCTURE AND TAXONOMY OF BULINUS JOUSSEAUMEI

WATSON, H. (In CONNOLLY, M.). 1925. The non-marine mollusca of Portuguese East Africa.

Trans. Roy. Soc. S. Africa, 12 (3) : 105-220.

WRIGHT, C. A. 1956. The anatomy of six species of the Molluscan genus Bulinus (Planorbidae) from Senegambia. Proc. Malac. Soc. Land. In press.

FIGS. 5 &6.—

ABBREVIATIONS USED IN FIGURES

FIG. 7.-

FIG. 35.—

BUG	= buccal ganglion.
CER	= cerebral ganglion.
OT	= otocyst.
OTN	= otocyst nerve.
FED	= pedal ganglion.
PEN N	= penial nerve.
PL	— pleural ganglion.
VIS	= visceral ganglion.
CAE	— caecum.
CRO	== crop.
GIZ	— gizzard.
INT	= intestine.
STO	= stomach.
ALBG	= albumen gland.
HD	= hermaphrodite duct.
MUCG	= muciparous gland.
OOTG	= oothecal gland.
PR	— preputium.
PROS	= prostate.
PS	= penis sheath.
RS	— receptaculum seminis.
SD	= sperm duct.
VAG	= vagina.
VAS DEF	= vas deferens.

PLATE i

Top row :

X 2.)

2nd row : 3rd row : 4th row : 5th row : 6th row : yth row :

Bulinus (Physopsis) globosus jousseaumei from Upper River Division, Gambia

B. (P.) globosus jousseaumei from Casamance Province, Senegal, (x 2.)

B. (P.) globosus from French Guinea. ( x 2.)

B. (P.) globosus from Kailahun District, Sierra Leone. ( x 2.)

B. (P.) globosus from Bolahun, Liberia, (x 2.)

B. (P.) globosus from Kongola, Belgian Congo, (x 2.)

B. (P.) globosus from Angola, specimens from type series, (x 2.)

Bull. B.M. (N.H.) Zool. 5, i.

PLATE i.

PLATE 2

Top row : Figured type specimens of Isidora jousseaumei Dautzenberg. ( x 1-5.) 2nd row : Umbilicate form of Bulinus (Physopsis) globosus from the Gambia. ( x 2.) Umbilicate form of B. (P.] globosus from Sierra Leone. ( x 2.) Umbilicate form of B. (P.) globosus from Northern Rhodesia, (x 2.) Umbilicate form of B. (P.) globosus from Angola, specimens from type series.

3rd row 4th row 5th row

X 2.)

Bottom

High-power view of micro-sculpture on shell of B. (P.) globosus from Sierra Leone

Bull. B.M. (A7.//.) Zool. 5, i.

PLATE 2.

°- ON SPELAEOGRIPHUS, A NEW CAVERNICOLOUS CRUSTACEAN FROM SOUTH AFRICA

ISABELLA GORDON

BULLETIN OF

THE BRITISH MUSEUM (NATURAL HISTORY) ZOOLOGY Vol. 5 No. 2

LONDON : 1957

ON SPELAEOGRIPHUS, A NEW

CAVERNICOLOUS CRUSTACEAN FROM SOUTH AFRICA

BY

ISABELLA GORDON

Pp. 29-47 > 26 Text-figures

BULLETIN OF

THE BRITISH MUSEUM (NATURAL HISTORY)

ZOOLOGY Vol. 5 No. 2

LONDON : 1957

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.

This paper is Vol. 5, No. 2 of the Zoological series.

PRINTED BY ORDER OF THE TRUSTEES OF THE BRITISH MUSEUM

Issued March 1957 Price Six Shillings

ON SPELAEOGRIPHUS, A NEW CAVERNICOLOUS CRUSTACEAN FROM SOUTH AFRICA

By ISABELLA GORDON, D.Sc., Ph.D.

SYNOPSIS

A new cavernicolous Malacostracan from a pool in a cave on Table Mountain, Spelaeogriphus lepidops n.g. and sp., is described and figured. The affinities of the genus are discussed. In some respects it resembles Monodella (Thermosbaenacea), in others it approaches the Anaspidacea (Syncarida), or the Tanaidacea (Peracarida), but it is not referable to any of these Orders. Specimens received later included an ovigerous female carrying ten to twelve large ova in a characteristic Peracaridan brood-pouch composed of five pairs of oostegites. The genus, therefore, belongs to the Division Peracarida and, as it is not referable to any of the existing Orders of that Division, is placed in a new Order Spelaeogriphacea and a new family Spelaeogriphidae each with the characters of the genus. Nothing is known of the internal anatomy or of the embryology.

INTRODUCTION

RECENTLY members of the South African Spelaeological Association obtained some specimens of a small blind Malacostracan from a pool at a depth of no ft. in a cave on Table Mountain, South Africa. The animals were said to " swim swiftly with rapid undulations of the body ". The temperature of the water in which they lived was 50° F. (February, 1956).

The specimens were submitted to Dr. K. H. Barnard who found that they represented a new genus and species of Crustacea Malacostraca which at first sight seemed referable to the Division Syncarida but on closer examination seemed rather to belong to the Peracarida — its " affinities seem to be with the Tanaidacea, especially Apseudes " to quote from Dr. Barnard's letter. In April 1956 Barnard sent to the British Museum six of the specimens, together with notes and sketches, and suggested that I might like to describe this interesting new species and discuss its affinities in more detail. I wish to express my thanks to Dr. Barnard for presenting these specimens to the Museum and also for the privilege of studying them.

The specimens were sent to London in two small vials. One vial contained two almost-perfect specimens which are quite opaque and much better preserved than the other four ; these have been selected as the holotype, a male measuring 7-2 mm. in length (from tip of rostrum to posterior margin of telson) and the allotype, a female measuring 5-6 mm., respectively. The four paratypes are very delicate, almost transparent, and more or less imperfect as to their appendages ; they range from 4-9 to 6-8 mm. in length and comprise two females and two males (one rather immature). The holotype and allotype have been handled with care and, for the

32 A NEW CAVERNICOLOUS CRUSTACEAN FROM SOUTH AFRICA

necessary dissection, the more incomplete paratypes have been used. Some specimens have been retained for the South African Museum collection and according to Barnard the largest measures 7-5 mm. in length.

The name suggested by Barnard for this new genus was most appropriate but, unfortunately, it is now preoccupied by Spelaeocaris Matjasic (1956, p. 65) a new genus of the family Atyidae. I therefore propose the name Spelaeogriphus from " griphos " meaning something complicated, a puzzle or riddle, not " gryphos " meaning a griffin (Jaeger, 1955).

Spelaeogriphus n. g.

DIAGNOSIS. Body elongated, subcylindrical (somewhat depressed). Carapace short, deep, coalescing dorsally with first thoracic somite and overhanging on each side to largely conceal the mouthparts and to enclose a branchial cavity within which lies the large, pedunculate, cup-like epipodite of the first thoracic limb (maxilliped). Each lateral flap of carapace deeply separated anteriorly from the dorsal part ; cervical furrow visible in the better-preserved specimens. Thoracic somites 2-8 free (although the second is almost entirely overlapped by the carapace) , deepening progressively posteriorly. Abdomen long, exceeding half the total length of body ; telson distinct from the last somite. Ocular lobe oval, plate-like, movably articulated to side of rostrum, without visual elements or pigment. Antennulae almost contiguous basally, long, with two unequal or subequal flagella and a 3- segmented protopodite without statocyst but modified in the adult male. Antenna longer and more robust than antennula ; protopodite 2-segmented (peduncle therefore 4-segmented), exopodite small, scale-like, multiarticulate flagellum nearly as long as body. Mandible with lacinia mobilis, a series of 16-20 spines and a slender, i-segmented palp. Maxillula with a slender palp near distal end of outer margin of endite 3 (broad outer lobe). Maxilla well developed ; endite 3 deeply bifurcate, each lobe with long curved apical setae ; a few stout penicillate processes on endite 2 (inner lobe). Maxilliped " Isopodan " in form, with a few retinacula on inner margin of inner plate (endite of basipodite) a 5-segmented palp, no exopodite, a large respiratory cup-like epipodite. Lower lip without movably articulated apical lappets. Peraeopods simple, ambulatory, none markedly modified ; epipodites absent ; exopodites present on 1-6 (a rudiment on 7 is exceptional). Three anterior pairs of exopodites 2 (3)-segmented, setose, natatory ; three posterior pairs i (2)-segmented, non-setose, respiratory (gills). Pleopods alike in both sexes ; first four pairs well-developed, biramous, natatory, fifth pair vestigial. Uropods broad, biramous ; exopodite 2-, endopodite i-segmented. A simple penial process on coxopodite of peraeopod 7 in male ; incipient oostegites on peraeopods 2-5 in female (not mature). In a more mature female sent after this paper was completed oostegites are present on peraeopods 1-5 inclusive (see p. 44).

Nothing is known of the internal anatomy or of the development.

Gender of genus : masculine. Genotype : Spelaeogriphus lepidops n. sp.

Holotype, allotype and the paratypes described below will be incorporated in the British Museum Collection.

A NEW CAVERNICOLOUS CRUSTACEAN FROM SOUTH AFRICA 33

Spelaeogriphus lepidops n. sp.

DESCRIPTION. The slender elongate body recalls that of many small cavernicolous Malacostraca ; it is subcylindrical being slightly depressed especially in the posterior third. In dorsal aspect the sides are almost parallel throughout, except posteriorly since the free telson is narrower than the abdominal and thoracic somites. In lateral aspect the body is as represented in Text-fig, i but the delicate side plates (pleura or epimera) of abdominal somites 1-4 are not at first glance apparent and therefore these somites appear to be less deep.

The small, distinct carapace is smooth except for the cervical furrow which is distinct in the holotype, but only faintly indicated in the more transparent specimens. It is produced anteriorly, between the pair of oval ocular lobes or scales, to form a somewhat depressed, broadly triangular rostrum (Text-fig. 2). In lateral aspect, the carapace is as deep as long and, in all the preserved specimens (which may be somewhat contracted), it overlaps the first free thoracic somite (number 2) leaving only a small portion exposed dorsally, and part of the third somite laterally (Text-fig, i). Each lateral part of the anterior margin of the carapace is continued backwards, on a level with the outer rim of the ocular scale, for a considerable distance before fusing with the dorsal portion at the cervical furrow. Thus these antero-lateral flaps are doubtless capable of considerable lateral movement. Near the postero- lateral margin there is a conspicuous oval patch, above thoracic somite 2, represented by stippling in Text-fig, i. This area, whose significance is unknown, is part of the carapace wall and can also be seen from the inside, as shown in Text-fig. la, where the ventral rim of the carapace is indicated, slightly posterior to the respiratory cup-like epipodite of the maxilliped. This large " gill " is visible through the thin wall of the carapace, but the oval patch behind it is not equally well marked in all the specimens ; for example, it is rather faint in the immature male paratype.

Thorax. The first somite is completely fused with the head region ; the second somite is free from, but almost entirely overlapped by, the carapace. Somites 3-8 become progressively deeper as represented in Text-fig, i although their dorsal margins are approximately equal in length.

The abdomen exceeds half the total length of the body. Somites i, 2 and 5 are subequal in length and shorter than the remaining three. Somites 1-4 decrease gradually in depth but, as already mentioned, their pleura are delicate and not easy to discern. The small epimeral portion of somite 5, however, is distinct (Text-fig, i).

The telson is free from, and narrower than, the sixth abdominal somite (Text-fig. 16). The median length is nearly equal to the basal width and the rounded apex bears a number of spines of varying length.

The antennulae, which are not widely separated from each other, are shorter and much less robust than the antennae. The proximal part of the right antennula of a female paratype is represented, in dorsal aspect, in Text-fig. 2, that of the holotype in latero-ventral aspect, hi Text-fig. 3. The first segment of the protopodite, which is equal to the sum of the second and third segments, has no statocyst. In the adult male the second segment of the peduncle is modified, the distal half of the inner margin being expanded to form a lobe which is richly beset with rows of

34 A NEW CAVERNICOLOUS CRUSTACEAN FROM SOUTH AFRICA

FIGS. 1-5. — -Spelaeogriphus lepidops n.g. and sp. Fig. i. <J paratype, in lateral aspect, peraeopods omitted except for the exopodites or " gills " on 4-6 respectively. ia. Ventral margin of carapace and cup-like epipodite at base of the maxilliped. Fig. 2. Lower figure — rostrum, right ocular lobe and proximal segments of antennula and antenna of a $ paratype, in dorsal aspect ; upper figure — rostrum and right ocular lobe of the smallest paratype, an immature <$, in dorsal aspect. Fig. 3. Proximal segments of antennula and antenna of the holotype, in ventro-lateral aspect (the segments of the protopodite are broader than shown in figure) and, at higher magnification, a few distal

A NEW CAVERNICOLOUS CRUSTACEAN FROM SOUTH AFRICA 35

conical papillae. Ventrally, this lobe is bounded by a series of long setae as shown in Text-figs. 3 and 4. There is also a small patch of the special, sharply pointed, conical papillae on the inner distal margin of the basal segment of the male peduncle. The longer flagellum comprises 40 or more segments, the smaller up to 36 segments ; in some specimens the two flagella are nearly equal in length. The distal three- fourths of the shorter (outer) flagellum has a series of short aesthetascs as shown in the enlarged portion (Text-fig. 3).

The antennal peduncle of a female is represented in dorsal aspect in Text-fig. 2, that of the male holotype in latero-ventral aspect and at a lower magnification, in Text-fig. 3. There is no obvious sexual dimorphism of the peduncle or protopodite. The small, scale-like exopodite bears some 10 marginal spines. The flagellum when complete is almost as long as the body and comprises about 70 segments.

The ocular lobe is a thin oval scale, movably articulated by means of a short peduncle to the side of the rostrum (Text-fig. 2) ; on the posterior outer edge is a tiny papilla. If viewed obliquely, the ocular scale seems to be decidedly club- shaped. There is no trace of pigment or of visual elements.

The mouthparts represented in Text-figs. 8-14 are, with the exception of the maxilliped (Text-fig. 13), all from the same female paratype. The upper lip, as indicated in Text-fig, i, is deep ; the distal portion is represented in lateral aspect in Text-fig. 8. The distal portion of the left mandible is shown (slightly distorted by pressure of the coverslip of the micropreparation) in Text-fig. 9. The palp is reduced to one elongate segment, the molar process is well developed and the spine row comprises 16-20 graded spines. There is only very slight asymmetry of incisor process and lacinia mobilis respectively (Text-fig. 10). The robust outer lobe (endite 3) of the maxillula bears about 16 short apical spines and, on the distal outer margin, a slender plumose seta-like palp (Text-fig, n) ; the rather slender inner lobe (endite i) has three short plumose terminal papillae. The maxilla, together with various details at a larger magnification, is represented in Text-fig. 12 ; endite 3 is deeply bifurcate, the long curved spines on the outer lobe consist of a shaft and a slightly curved distal portion whereas those of the inner lobe have the apex bifurcate. The short inner spines of each series are somewhat conical with striate or serrulate concave (? outer) margins. Endite 2 has a few stout penicillate setae, three short and one long, on the inner margin. The maxilliped represented in Text-fig. 13 is from a considerably smaller specimen and, as it was not completely removed, the proximal portion (indicated by a broken line in the figure) may not be quite exact although the relative size and position of the epipodite is correct, the concavity facing inwards and forwards. The two maxillipeds are held firmly together by means of a few small retinacula on the inner margin of the inner plate (endite of the basipodite). The palp appears to be 5-segmented; the wide second segment bears 6 stout plumose setae on the inner margin ; the third segment, which

segments of the shorter antennular flagellum. Fig. 4. Second segment of antennular protopodite of <$ paratype, in ventral aspect, to show specifically modified area beset with rows of conical papillae. Fig. 5. Vestigial pleopod 5 of 6* paratype, highly magni- fied. Scale of Fig. i and ia = i mm. Scale of Figs. 2 (lower) and 3 respectively

36 A NEW CAVERNICOLOUS CRUSTACEAN FROM SOUTH AFRICA

FIGS. 6-14. Spelaeogriphus lepidops n.g. and sp. ? paratypes. Fig. 6. Distal segments of peraeopod i . Fig. 7. Exopodite of peraeopod i of a larger specimen. Fig. 8. Upper lip, in lateral aspect. Fig. 9. Distal part of left mandible, slightly distorted by the coverslip of microslide. Fig. 10. Portion of left and right mandibles, at higher magni- fication, to show slight asymmetry of incisor process and lacinia mobilis. Fig. n. Maxillula, and apex of inner lobe more highly magnified. Fig. 12. Maxilla, with enlarged details of setae and spines. Fig. 13. Maxilliped of a smaller specimen, in ventral aspect (appendage broke just below the palp ; position of epipodite relative to the maxilliped is correct. Fig. 14. Lower lip. Figs. 8-12 and 14 from the same specimen. All except Fig. 10 drawn to scale = 0-5 mm.

A NEW CAVERNICOLOUS CRUSTACEAN FROM SOUTH AFRICA 3y

is contracted distally, also has marginal plumose setae ; the fourth segment is somewhat expanded distalty and the fifth is inserted at an angle to the apex. The lower Up, represented in Text-fig. 14, shows no trace of the movably articulated lappet on each lobe which is characteristic of the family Apseudidae.

Peraeopods. The peraeopods, which are simple and ambulatory (pediform), are arranged in two series, an anterior one of three, and a posterior one of four, pairs as shown in Dr. Barnard's sketch (Text-fig. 26). This division is emphasized by the fact that the exopodites on each of the three anterior pairs are natatory whereas those on peraeopods 4-6 respectively are reduced and modified to serve as gills (Text-figs, i and 18-20). In the allotype there is also a minute gill or exopodite on the right peraeopod 7, but this is exceptional. Each natatory exopodite consists of an elongate proximal, non-setose or sparsely setose, segment and a distal segment bearing marginal setae, the apical ones being especially long and often curled ; in addition a very short basal segment may be more or less clearly demarcated (Text-figs. 7, 18, 21, 24 and 25). The respiratory exopodites, the posterior of which is quite small, lack the setose terminal segment ; the proximal segment is swollen and specially modified and again there is a more or less distinct small basal segment (Text-figs, i, 19, 20 and 22).

In the adult male peraeopods 1-3 are considerably more robust than in the female or young male, but otherwise there is no sexual dimorphism. The first peraeopod is the only one which may be regarded as slightly modified ; it is rather shorter than either peraeopod 2 or 3, with a smaller exopodite, a broader carpus, and more numerous spinules on the ventral margin of carpus and propodus respectively (Text- figs. 21 and 24). Text-figs. 24 and 25 represent peraeopods i and 3 respectively, drawn to the same scale, of the male paratype shown in Text-fig, i ; peraeopod 2 is very similar to, though a trifle shorter than, peraeopod 3. In the holotype, which is rather larger, the spine-setae on the carpus are more numerous, especially in one of the two rows (3 and 5-6, as against 2 and 2 in the paratype represented in Text- fig. 25). When the distal segments are flexed these two rows lie one on either side of the propodus. Peraeopods 4-7 are relatively more slender and increase gradually in length owing chiefly to the progressive elongation of the carpus and especially of the propodus (cf. Text-figs. 22 and 23, peraeopods 4 and 7 respectively, of a male in the South African Museum). Peraeopod 7 as a rule lacks the exopodite ; a simple penial process is present in all three males on the coxopodite. This process is short in the smaller male paratype with the as yet unmodified antennular protopodite (and therefore rather immature) ; in the adult male it is long so that the two processes almost meet in the median line (Text-fig. 23).

The peraeopod represented in Text-fig. 6 was already detached from one of the paratypes but it probably came from one of the females ; in the allotype the first peraeopod has some spinules on the ventral margin of carpus and propodus and the exopodite is shorter, especially as regards the terminal segment, than that on either peraeopod 2 or 3 (cf. Text-figs. 7 and 18). Peraeopod 4 is more slender than peraeo- pod 3, with a longer more slender propodus ; the exopodite shows a hint of the short basal segment which, however, is distinct in that of peraeopod 5. What I consider to be incipient oostegites are present at the bases of peraeopods 2-5 but they are

38 A NEW CAVERNICOLOUS CRUSTACEAN FROM SOUTH AFRICA

17

SMM.

FIGS. 15-20. Spelaeogriphus lepidops n.g. and sp. Fig. 15. Uropod of allotype. Fig. 1 6. Telson and posterior margin of abdominal somite 6, in dorsal aspect. Fig. 17. Fourth pleopod, with grappling hook more highly magnified. Fig. 18. Third right peraeopod of $ with well developed natatory exopodite and incipient oostegite o. Fig. 19. Fourth peraeopod with respiratory exopodite or " gill " and oostegite. Fig. 20. Detached exopodite of peraeopod 5. Figs. 16-20 from same $ paratype. Figs. 15 and 17 at smaller scale — 0-5 mm. ; rest at larger scale = 0-5 mm.

A NEW CAVERX1COLOUS CRUSTACEAN FROM SOUTH AFRICA 39

not easy to make out (Text-figs. 18 and 19, o). As in the male, peraeopods 5-7 lengthen progressively and the respiratory exopodite on peraeopod 6 is smaller than that on peraeopod 5. Perhaps none of the female specimens is fully mature since the oostegites are so small and inconspicuous.

Pleopods. These are alike in both sexes. The first four pairs are well developed and natatory. Each comprises a broad protopodite with two retinacula at the distal end of the inner, and a series of spines on the outer, margin and a paddle-shaped, heavily setose exopodite and endopodite (Text-fig. 17). The fifth pair is very much reduced and concealed by the small epimera of the fifth abdominal somite. In Text-fig. 5 the fifth pleopod has been displaced so that the drawing could be made, it is reduced to a slender club-shaped segment with 4 terminal setae.

The uropods appear to be very fragile since in none of the specimens are they now complete ; either the exopodite, or both exopodite and endopodite are as a rule missing. In the vial containing the holotype and allotype two detached but complete uropods were found. These must belong to the allotype since in the holotype the peduncles are still attached to the specimen. One of these uropods is represented in Text-fig. 15 ; the short stout protopodite is furnished with an outer and an inner cluster of apical spines ; the exopodite is 2-segmented, the proximal segment is armed with spines along the inner margin and apically — the latter spines being especially long, the distal segment is very similar in shape and size to the exopodite and both are heavily setose. In the case of the male paratype represented in Text- fig, i the exopodite and endopodite had apparently undergone regeneration and are smaller than is normal for the species, so in the illustration I have indicated the normal length of the endopodite by a broken line and have omitted the exopodite (which in any case is most frequently missing).

REMARKS. The material available is insufficient for a study of the differences due to age and sex. Apart from that exhibited by the protopodite of the antennula (Text-figs. 2 and 3-4), sexual dimorphism appears to be slight. The segmentation of the antennular flagella is indistinct proximally, where the segments appear to be crowded, becoming more distinct distally. Dr. Barnard gives the number of seg- ments of the flagella as about 36 and 48 respectively but this is presumably the maximum number found in the largest specimen (/ = 7-5 mm.). In the holotype the two flagella are not very unequal but the number of segments seems to be about 32-33 and 40-42 respectively. In the allotype the flagella are markedly unequal yet each appears to have less than 30 segments (about 24 and 28 respectively). These differences may be due to sex. In the other specimens the flagella are not always complete.

In the smallest male there is as yet no trace of the sexual modification of the first two segments of the antennular protopodite. Indeed, the antennula is relatively much shorter than in the adults, especially as regards the basal segment of the protopodite ; the smaller flagellum comprises only 15-16 segments. The larger flagellum is missing, as also are the antennae beyond the short protopodites (the two basal segments). The rostrum is also appreciably shorter than in the adults, and the ocular lobes are much more circular in outline (upper figure in Text-fig. 2).

4<> A NEW CAVERNICOLOUS CRUSTACEAN FROM SOUTH AFRICA

O*5 MM.

24

FIGS. 21-26. Spelaeogriphus lepidops n.g. sp. Figs. 21, 22 and 23. Peraeopods i, 4 and 7 respectively of a $ in South African Museum — sketches by Dr. K. H. Barnard ; magnification not stated. Specimen probably larger than the holotype. Figs. 24 and 25. Left peraeopods i and 3 respectively, of 6* paratype represented in Fig. i. Scale — 0-5 mm. Fig. 26. Sketch of whole animal, in lateral aspect, by Dr. K. H. Barnard, showing arrangement of peraeopods in two groups, pleopods and uropod.

A NEW CAVERNICOLOUS CRUSTACEAN FROM SOUTH AFRICA 41

The first peraeopod is very similar to that of the female (Text-fig. 6) ; the others are all incomplete although in most instances the proximal segments with the well developed exopodites remain.

In a female paratype of approximately the same size as the male represented in Text-fig, i (/ = 6-5 mm.) the antennular protopodite is more slender and, especially as regards the first segment, shorter than in the male and there are fewer segments in the shorter flagellum ; the longer one is incomplete. The protopodite of the antenna is also relatively shorter and more slender (the male antennular and antennal protopodites are considerably broader than I have indicated in Text-fig. 3 where the appendages are lying obliquely, but I did not wish to risk damage to the holotype). The first three peraeopods of the male become more robust with increase in size ; in the holotype they are somewhat more robust than in the paratype from which Text-fig. 24 was obtained and Dr. Barnard's sketch is probably from a male exceeding 7 mm. in length. (Text-fig. 21).

AFFINITIES OF THE GENUS SPELA EOGR IPH US

In recent years many new cavernicolous and interstitial Crustacea Malacostraca have been discovered ; these are referable for the most part to the Orders Thermos- baenacea, Bathynellacea (Syncarida), Isopoda and Amphipoda (Peracarida) .

Spelaeogriphus, with its slender, elongate body, bears a striking resemblance to one of these cavernicolous forms namely, Monodella argentarii Stella (1951^, p. 2, fig. i). But, the general similarity of telson, uropods, mandible (the palp excepted) and the exopodites1 of the peraeopods notwithstanding, Spelaeogriphus is most certainly not referable to the Thermosbaenacea. This Order is unique amongst Malacostraca in the possession of a dorsal marsupium or brood-pouch in the female, " a chamber formed by the posterior portion of the carapace, which covers the first three somites of the body " (Stella, 19516, fig. 3 of plate ; 1953, pi. i, fig. 2). Barker exhibited some ovigerous or larvigerous females of Thermosbaena mirabilis Monod, with a similar dorsal marsupium, at the XIV International Congress of Zoology held in Copenhagen in 1953, but his description and figures have not so far been published (Barker, 1953). Fertilization in the Thermosbaenacea must, therefore, be internal. In Monodella, according to Stella (1955, p. 464), from each ovary a short duct, the vagina, leads to the base of the sixth peraeopod and another one, the oviduct, goes dorsally to the brood-pouch. The position of the vaginal openings on the seventh thoracic somite (bases of peraeopods 6) is unusual ; in Malacostraca the female genital openings are, as a general rule, on the sixth somite. Another unusual feature in Monodella is the presence in the male of an additional coupling organ on the maxilliped (Stella, 1955, p. 464 ; Karaman, 1953, figs. 7 and 10). These characters, together with a study of the embryology of Monodella, led Taramelli (1954) to exclude the Order Thermosbaenacea from the Division Peracarida and with this Siewing agrees (1956, p. 168, Diagram 3). The Thermosbaenacea have certain characters of the Syncarida, others of the Peracarida, and still others which are unique.

When Barnard first examined the specimens of Spelaeogriphus he thought that

1 None of these exopodites are respiratory in Monodella.

42 A NEW CAVERNICOLOUS CRUSTACEAN FROM SOUTH AFRICA

they belonged to the Division Syncarida. But, he writes, " further consideration shows the impossibility of including this Crustacean in that Division. Barring a superficial resemblance in having exopods on six of the peraeopods, it has none of the special features found in the Syncarida. On the contrary, its affinities seem to be with the Tanaidacea, especially Apseudes. The mouthparts are Isopodan in character, and the cup-like epipod on the maxilliped is clearly analogous to that found in Apseudes ' ' (Barnard, in letter received 14 . iv . 56) . After I had described Spelaeogriphus and had considered its possible relationships, I sent some notes and sketches to Dr. K. Lang, Director of the Stockholm Museum, since he has for some years past been engaged on a revision of the Order Tanaidacea. He replied in the following few words : " The animal you picture does not belong to the Tanaidacea but to the Anaspidacea " (letter dated 2o.vii.56). Thus two eminent authorities on the lower Eumalacostraca disagree as to the systematic position of Spelaeogriphus.

In my opinion Spelaeogriphus does not agree with either the Anaspidacea or the Tanaidacea as at present defined. In fact, like the Thermosbaenacea, it does not quite fit into either the Syncarida or the Peracarida. In Kiikenthal & Krumbach's Handbuch der Zoologie, Zimmer (1927, p. 566) defines the Divisions of the Eumalaco- straca and, as regards the external characters, Spelaeogriphus differs from the Syncarida and agrees with the Peracarida in having : (i) a carapace which encloses gill chambers but leaves most of the thoracic somites free ; (ii) a lacinia mobilis on the mandible ; (iii) two, not three, segments distal to the " knee " of the peraeopods ; (iv) oostegites in the female (but see later, p. 44). The antennal protopodite consists of two, not three, segments so that the peduncle comprises four segments ; this holds for some Syncarida and also (though not mentioned by Zimmer in his diagnosis on p. 566) for the Tanaidacea alone amongst the Peracarida (Caiman, 1909, p. 191 ; Zimmer, 1927, p. 686). Thus there is something to be said in favour of Barnard's view that Spelaeogriphus is a primitive Apseudid.

On the other hand, the genus differs in quite a number of respects from Zimmer 's (1927, p. 685) diagnosis of the Tanaidacea. There are seven, not six, free thoracic somites. The sides of the carapace are deep and separated anteriorly for a long distance from the dorsal or median part. The telson is distinct from, not fused with, the last abdominal somite. The abdomen itself is far longer than that of the Tanaids; but this difference may not be significant since in the Thermosbaenacea Monodella has a long, Thermosbaena a short, abdomen. The exopodites are more numerous and well developed, three pairs being natatory, three pairs respiratory ; in the Tanaidacea vestigial exopodites are sometimes present on the first two pairs of peraeopods only. While the form of the ocular lobe and of the epipodite on the maxilliped strongly recall the Apseudidae, Spelaeogriphus differs from that family in other respects, namely : The first peraeopods are not chelate or subchelate, nor is the second pair modified and fossorial. The antennulae are set closer together and are decidely smaller than the antennae. The mandibular palp is reduced to one segment, not " triarticulate " ; but this is probably of slight importance since the palp is absent in the family Tanaidae. The palp of the maxillula is not large and refiexed into the gill chamber, but small and placed near the distal end of endite 3 (in Anaspides the palp is even smaller, though more proximally placed, see

A NEW CAVERN1COLOUS CRUSTACEAN FROM SOUTH AFRICA 43

Chappuis, 1927, p. 596, fig. 584). There are no apical lappets on the lower lip. The uropods are broad, natatory whereas in Apseudids exopodite and endopodite are, as a rule, slender and miiltiarticulate. The Tanaidacea are entirely marine. I do not think that Spelaeogriphus is referable to the Order Tanaidacea, nor can it be placed in any of the other Peracaridan Orders — Cumacea, Mysidacea, Isopoda or Amphipoda.

In addition to the characters already mentioned (p. 42), Spelaeogriphiis differs from the Anaspidacea in other respects : Epipodites are absent from all the peraeo- pods, whereas in Anaspidacea there are one or two on each, with the exception of the last pair. The pleopods are alike in both sexes, and the endopodite is well developed in the anterior four pairs ; in the Syncarida the endopodite is rudimentary or absent, with the exception of the first two pairs in Anaspididae and Koonungidae, in which the endopodites are modified as copulatory organs in the male (Chappuis, 1927, p. 594 ; Smith, 1909, figs 29 and 52 ; Nicholls, 1931, pi. 32, figs. 12 and 13). A thelycum or spermatheca appears to be absent in Spelaeogriphus but is present in Anaspidacea (Smith, 1909, fig. 27 ; Nicholls, 1931, p. 476, figs A and B). There is no statocyst in Spelaeogriphus such as occurs in e.g. Koonunga (Smith, 1909, p. 502, fig. 5). Certain characters of Spelaeogriphus, on the other hand, do recall those of some Anaspidacea. For example, both Spelaeogriphus -and Koonunga exhibit sexual dimorphism of the antennulae although the modified area in the male differs in position and in form in the two genera (c.f. Zimmer, 1927, p. 595, fig. 580 with Text-fig. 3 of the present paper). The three pairs of respiratory exopodites on peraeopods 4 to 6 in Spelaeogriphus are unusual — they resemble epipodites but from their position on the limbs both Barnard and I think they must be exopodites. The only other Eumalacostraca with exopodites of this type are Syncarida ; in the Anaspididae peraeopod 6 (thoracic limb 7) bears, in addition to the two epipodites, a reduced respiratory exopodite, whereas those on the anterior peraeopods are long and multiarticulate (Smith, 1909, p. 516, fig. 24 and p. 513, fig, 21). The free second thoracic somite, free telson and broad uropod recall the Anaspidacea and in Koonunga there is a distinct V-shaped notch above the attachment of the antenna in the frontal margin of the cephalon (Sayce, 1908, pi. i, figs, i and 3) ; in Spelaeo- griphus there is a long slit in this position (Text-figs, i, 2). It is possible to imagine a Syncarid with a carapace since, in the Division Peracarida, the carapace is present or absent and, when present, varies greatly in relative size. According to Barnard the mouthparts are Isopodan in character but the maxillula is not unlike that of Anaspidacea, especially the position and direction of the palp (Sayce, 1908, pi. i, fig. 12 ; Smith, 1909, p. 508, figs. 13, 14). In the Anaspididae the mandible shows a hint of bifurcating although there is no lacinia mobilis, and the proximal epipodite on the maxilliped is large although not cup-like (Smith, 1909, figs. 9, 10 and 19). If Spelaeogriphus is a Syncarid it certainly is not referable to either the Anaspididae or the Koonungidae. Nor can it be placed with the minute rather degenerate members of the Bathynellidae although, if Ueno's observations are correct, some species of this family would seem to possess oostegites. Dr. Chappuis, whom I consulted on this point, writes " No! there is no brood pouch in Bothy nella or Parabathynella ; the eggs are laid one after the other just where the animal happens

44 A NEW CAVERNICOLOUS CRUSTACEAN FROM SOUTH AFRICA

to be" (letter dated 22.V.26). Yet Miuri and Morimoto (1953, p. 239) say of Bathynella morimotoi Ueno " Adult females carrying eggs and newly-hatched larvae are obtainable at all seasons of the year ". In the following year Ueno (1954, p. 525, fig. 36) figures a long elliptical lamella on the coxopodite of the second peraeopod of Bathynella inlandica n. sp. and says that these structures, which are also present on the first pair of peraeopods, are presumably oostegites (marsupium). Here then is a Peracaridan character in certain species of the Bathynellacea.

Like the Thermosbaenacea, Spelaeogriphus possesses certain characters of the Syncarida, others of the Peracarida. For the present it seems advisable to refer it to a new family, Spelaeogriphidae, with the characters of the genus, and to leave the systematic position of the family as uncertain. Perhaps when the internal anatomy and the embryology of Spelaeogriphus are known the systematic position of the family will be elucidated. As new forms of primitive Eumalacostraca come to light it may be necessary to revise the classification and even to redefine the major Divisions.

ADDITIONAL NOTE

After the manuscript was finished I received from Dr. Barnard two further specimens accompanied by the following note : " New species of shrimp ; pair found copulating. Bats Cave, stream at bottom. Collected by S.A.S.A. 29.7.56." The male and female were thought to be copulating when caught, and each should therefore be sexually mature. Unfortunately, uropods, antennulae and antennae are incomplete in both and in the male the posterior two or three peraeopods are broken and most of the gill-like epipodites are missing.

In the male, which measures 6 • 5 mm. in length, the modified lobe on segment 2 of the antennular protopodite is more pronounced distally than that represented in Text-fig. 4 and the conical papillae extend almost to the proximal articulation of the segment ; the patch on the inner distal margin of segment i is conspicuous. There seem to be a few papillae at the inner distal angle of segment 2, and a row of 5 blunt cones on the inner margin of segment 3, of the antennal peduncle. The body of the female is slightly bent, but it appears to be rather shorter and is more slender than that of the male. The oostegites are quite unmistakable in this specimen although they are narrower than one might expect in a breeding female. In addition to the four pairs which I detected in the type specimens, a small pair is present on peraeopods 1. The first four pairs meet or even overlap medially ; each member of the fifth pair is only about as long as wide and does not quite reach the median line.

In both specimens the peraeopods are rather bunched together and each is flexed towards its partner. Barnard sketches the peraeopods as arranged in two series, 1-3 directed forwards and 4-7 directed backwards (Text-fig. 26) and in life this may be the case.

There can now be no doubt as to the presence of a ventral thoracic marsupium such as is characteristic of the Division Peracarida. Spelaeogriphus, therefore, seems referable to that Division and, as far as the external characters are concerned, it agrees with the definition of the Peracarida given by Caiman (1909, p. 149) and

A NEW CAVERN1COLOUS CRUSTACEAN FROM SOUTH AFRICA 45

also that given by Zimmer (1927, p. 566) if very slightly modified to read "... Anten- nenstamm 2- oder 3-gliederig." (As already mentioned, Zimmer failed to recall that the antennal protopodite of the Tanaidacea is only 2-segmented, although he does mention this in his treatment of the Order on p. 686) . However, the family Spelae- ogriphidae cannot be placed in the Order Tanaidacea for reasons which I have already given (p. 42). Nor can it be placed in any of the other Peracaridan Orders although the elongated abdomen and free telson, the large number of exopodites, and the sexual modification of the antennula in the male are characters which it shares with the Mysidacea. The only alternative, therefore, is to establish a new Order Spelaeogriphacea, with the characters of the family, to receive it.

It is to be hoped that ovigerous females and larval stages may soon be collected and also that specimens fixed in Bouin or another suitable fixative will be available for sectioning.

POSTSCRIPT i8.xii.56.

After the manuscript had gone to press I received five additional specimens collected in Bats Cave, on g.ix.56 by the S.A.S.A. One of these is an ovigerous female with a relatively large brood-pouch containing about 10-12 large ova ; the outlines of the separate oostegites are not clearly distinguishable but it is a normal Peracaridan brood-pouch composed, as already stated, of five pairs of oostegites on peraeopods 1-5 (somites 2-6).

I also sent tracings of the illustrations to Dr. Rolf Siewing of Kiel, who has done some excellent work on the comparative morphology of the Crustacean Malacostraca. He replied as follows : " Mit grosser Freude habe ich Ihre Zeichnungen von Spelae- ogriphus lepidops studiert. Der neue Fund hat mich sehr interessiert . . . Meine Meinung nun zu der Neuentdeckung ist, dass es sich nicht um einen Vertreter der Syncarida handelt. Es fehlen bei Spelaeogriphus Epipodite, die bei den Syncarida wenigstens an einigen Thorakalextremitaten ausgebildet sind. Ein freier Carapax ist bei den Syncarida ebenfalls niemals ausgebildet. Die Oostegite und die Lacinia mobilis sind aber ganz typische Charakteristica der Division Peracarida. Ich halte es nicht fiir wahrscheinlich, dass sich diesse Organe unabgehangig in einer anderen Kategorie der Malacostraca noch einmal entwickelt haben. Auffallig sind aber manche Ubereinstimmungen mit den Thermosbaenacea : Bau der Extremitaten des Thorax, Carapax, und Lacininia mobilis der Mandibel. Moglicherweisse ist Spelaeogriphus mit ihnen naher verwandt und stellt ein primitives Bindeglied dar. Sicher wird die Untersuchung der inneren Anatomic weitere Aufschliisse geben."

I too had been much impressed by the resemblances between Spelaeogriphus and Monodella — apart from the position of the marsupiuin, which is ventral in Spelaeogriphus and all the Peracarida, dorsal in the Thermosbaenacea. Dr. Siewing's comments have been most helpful and give me more confidence in proposing the new Order Spelaeogriphacea. The position of this primitive Peracaridan Order in Siewing's Diagram 3 (1956, p. 168) would appear to be within the Division Peracarida, near the suggested position of the Thermosbaenacea, thus :—

46 A NEW CAVERNICOLOUS CRUSTACEAN FROM SOUTH AFRICA

AMPHirODA

SPELAEOGRIPHACEA

Exopodite- gilis.

CUMACEA

DIVISION PERACARIDA.

THERMOSBAENACEA

Marsupium dorsal.

With lacinia mobilis; marsupium thoracic.

Adapted from Slewing, 1956, p. 168. Upper left-hand portion of Diagram 3.

The Thermosbaenacea and the Peracarida have a lacinia mobilis on the mandible and a thoracic brood-pouch or marsupium. The Spelaeogriphacea have three pairs of exopodites modified as gills ; this may be a secondary specialization, perhaps an adaption to the freshwater habitat, although no other cavernicolous Malacostracan possesses such gills. The relationship of the Spelaeogriphacea to the other Peracaridan Orders must, for the present, remain uncertain.

REFERENCES

BARKER, D. 1956. The morphology, reproduction and behaviour of Thermosbaena mirabilis

Monod. Proc. XIV Intern. Zool. Congr. Copenhagen : 503-504. CALMAN, W. T. 1909. Crustacea in Lankester's A Treatise on Zoology VII. London. 346

pp., 194 text-figs. CHAPPUIS, P. A. 1927. Crustacea Malacostraca — Syncarida in : Kiikenthal & Krumbach's

Handbuch der Zoologie. Bd. 3, Hft. i : 593-606, with text-figs. JAEGER, E. G. 1955. ^ Source Book of Biological Names and Terms. 3rd Editn. Springfield,

Illinois. KARAMAN, S. 1953. Ueber einen Vertreter der Ordnung Thermosbaenacea . . . Mono-

della halophila n. sp. Acta Adriatica. Split, 5 (3) : 1-22, 24 figs. MATJA§IC, J. 1956. Ein neuer Hohlendecapode aus Jugoslawien. Zool. Anz. 157 (3-4) :

65-68, 2 text-figs. MIURI, Y. & MORIMOTO, Y. 1953. Larval development of Bathynella rnorimotoi Ueno. Annot.

Zool. Jap. 26 (4) : 238-245, 4 text-figs., 2 tables.

A NEW CAVERXICOLOUS CRUSTACEAN FROM SOUTH AFRICA 47

NICHOLLS, G. E. 1931. Micraspides calmani, a new Syncaridan from the West Coast of

Tasmania. Journ. Linn. Soc. Zool. 37 (no. 254) : 473-488, figs. A, B and pis. 31 and 32. SAYCE, O. A. 1908. On Koonunga cursor, a remarkable new type of Malacotracous Crustaceans.

Trans. Linn. Soc. London, Zool. (2) xi, i : 1-15, 2 pis. SIEWING, R. 1956. Untersuchungen zur Morphologic der Malacostraca (Crustacea). Zool.

Jahrb. Anat. Ontog. 75 (i) : 39-176, 72 text-figs. SMITH, G. 1909. On the Anaspidacea., living and fossil. Q. J. Microsc. Sci., London, 53

(3) : 489-578, 62 text-figs., pis. ii and 12. STELLA, E. 19510. Monodella argentarii n. sp. di Thermosbaenacea (Crustacea Peracarida)

Arch. Zool. Ital. Torino, 36 : 1-15, 22 figs. (Reprint.) I95I&. Notizie biologiche su Monodella argentarii Stella . . . Boll. Zool. Torino, 18

(4-6) : 227-233, 4 figs. — • — • 1953. Sur Monodella argentarii Stella, espece de Crustace Thermosbenac6 . . . Hydro-

biologia, 5 (1-2) : 226-232, i pi. — — - 1955. Behaviour and development of Monodella argentarii Stella, a Thermosbenacean

from an Italian cave. Proc. Intern. Assocn. Limnol. 12 : 464-466. TARAMELLI, E. 1954- La posizione sistematica dei Thermosbenacei quale risulta dallo studio

anatomico di Monodella argentarii Stella. Monitore zool. ital. Firenze, 62 (i) : 9-24, 3 pis. UENO, M. 1954. The Bathynellidae of Japan (Syncarida-Bathynellacea) . Arch. f. Hydrobiol.

49 (4) : 519-538, 9 text-figs., 2 tables. ZIMMER, C. 1927. Parts on Crustacea Malacostraca in : Kiikenthal & Krumbach's Handbuch

der Zoologie. Bd. 3, Hft. i : 553-566 ; 683-696, with text-figs.

PRINTED IN GREAT BRITAIN BY ADLARD AND SON. LIMITED, BARTHOLOMEW PRESS, DORKING

51/7. A.

THE PELECANIFORM CHARACTERS

OF THE SKELETON OF THE

SHOE-BILL STORK,

BALAENICEPS REX

PATRICIA A. COTTAM

BULLETIN OF

THE BRITISH MUSEUM (NATURAL HISTORY) ZOOLOGY Vol. s No. 3

LONDON: 195?

THE PELECANIFORM CHARACTERS OF THE

SKELETON OF THE SHOE-BILL STORK,

BALAENICEPS REX

BY

PATRICIA A. COTTAM

Pp. 49-72 ; Plate 3 ; 4 Text-figures

BULLETIN OF

THE BRITISH MUSEUM (NATURAL HISTORY) ZOOLOGY Vol. 5 No. 3

LONDON: 1957

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.

This paper is Vol. 5, No. 3 of the Zoological series.

PRINTED BY ORDER OF THE TRUSTEES OF THE BRITISH MUSEUM

Issued July, 1957 Price Eight Shillings

THE PELECANIFORM CHARACTERS OF THE

SKELETON OF THE SHOE-BILL STORK,

BALAENICEPS REX

By PATRICIA A. COTTAM

CONTENTS

Page

INTRODUCTION ........... 51

HISTORICAL NOTE . . . . . . . . . .51

METHODS ............ 54

OSTEOLOGICAL CHARACTERS ......... 55

A. Skull 55

B. Sternum and pectoral girdle ........ 62

c. Pelvic girdle .......... 64

D. Hind limb ........... 64

SOME NON-SKELETAL PELECANIFORM CHARACTERS ..... 66

DISCUSSION ........... 66

SUMMARY ............ 70

REFERENCES ........... 71

PLATES ............ 72

INTRODUCTION

IN the course of rearranging the bird skeletons in the collections of the British Museum (Natural History) it seemed to me that the skeleton of Balaeniceps rex had more pelecaniform than ciconiiform characters. The position of Balaeniceps in orthodox classifications is, and nearly always has been, near the storks and herons, so that this anomalous impression of its affinities seemed to require detailed investigation. The results of this study are presented here. The skeletal characters of Balaeniceps rex have been reassessed in relation to stork-like and heron-like characters on the one hand, and pelican-like characters on the other.

I am grateful to Dr. H. W. Parker and to Dr. F. C. Fraser for reading the manuscript and making helpful comments, and also to Mr. J. D. Macdonald for advice and encouragement at all stages of the investigation.

HISTORICAL NOTE

There are few important original contributions to the knowledge of Balaeniceps s affinities. Gould, who described the bird in 1850, called it the " Grallatorial type of the Pelecanidae ", although he also noted that its external features resembled " in general contour " those of Grus, Ardea and Cochlearius. Jardine (1851) noted likenesses to herons in the plumage. He considered that differences from the " true " pelicans, in the nostrils, pouch, position of the laryngeal opening and the absence of webs on the feet, were sufficient to show that Balaeniceps was not closely related

ZOOL. 5, 3. 3

52 THE PELECANIFORM CHARACTERS OF THE SHOE-BILL STORK

to them. Von Heuglin (1856 : 60) placed it between Anastomus and Dromas in his systematic list. Bonaparte (1855 : 143) put it in the same subfamily as Coch- learius, describing it as intermediate between the pelicans and the Boat-bill. Des Murs (1859 : 48°) considered that the egg was like that of Phoenicopterus.

These were the conflicting opinions in 1860 when Parker examined the skeleton of Balaeniceps. He was impressed by its similarities to Scopus and Cochlearius, especially the latter, and indicated many characters which it had in common with the " Ardeine " birds. Although the storks were included in his term " Ardeine ", he seemed to stress the heron-like characters of Balaeniceps because he considered it to be a large edition of Cochlearius. He noted some similarities to the Pelecaniformes but attributed them to convergence. Bartlett's discovery (1861) of powder-down on Balaeniceps seemed to add weight to Parker's conclusions. Reinhardt (1860), unaware of Parker's work, found more similarities with Scopus than Cochlearius in the external characters of Balaeniceps, and considered that Balaeniceps and Scopus were nearer the storks than the herons. In 1861, after reading an abstract of Parker's paper, he compared a skull of Balaeniceps with those of Scopus and Cochlearius, but still maintained that Balaeniceps was related to Scopus and the storks. The similarities between the skulls of Balaeniceps and Cochlearius he attributed to convergence. Parker admitted (1862) that he knew nothing of the anatomy of Scopus when he wrote his paper, but, having seen a live Balaeniceps, he remained convinced of its likeness to the herons. He regarded Ardea as the " central type " of the storks and herons, linked to Cochlearius and Scopus by Balaeniceps.

These two opinions became established. Some authors agreed with Reinhardt's conclusions and placed Balaeniceps with Scopus and the storks, but most of them agreed with Parker and placed it with Cochlearius and the herons. Giebel (1873) showed that the pectination of the middle claw and the pterylosis of Balaeniceps are similar to those of Scopus and different from Cochlearius. Beddard (1888) compared the visceral anatomy with that of the storks, herons and Scopus and, because of the alimentary tract, powder-down patches and syrinx, concluded that Balaeniceps was allied to the herons. Like Beddard, Furbringer (1888) and Gadow (1893 : 137) agreed with Parker. So did Shufeldt (1901), who wrote a paper on the osteology of Scopus and Balaeniceps without having seen a skeleton of the latter.

The next important contribution was made by Chalmers Mitchell (1913) who dissected a specimen and described many more anatomical details. It is interesting that he could find no outstanding characters which indicated affinities with the herons rather than the storks, or vice-versa. He noted that Scopus and Balaeniceps had many similarities, and that they had characters common to both herons and storks. When he took each character at its face value he found Balaeniceps had more in common with the storks than the herons, so he decided to put it in the same suborder as Scopus, storks and ibises. He acknowledged that this was an arbitrary rather than a phylogenetic arrangement. He noted several similarities to the pelicans, but thought they occurred because the pelicans were related to the storks and herons.

THE PELECANIFORM CHARACTERS OF THE SHOE-BILL STORK 53

Bohm (1930) studied the structure of the skulls of juvenile and adult Balaeniceps. After a comprehensive investigation he concluded that Reinhardt's, Parker's, Giebel's and his own researches showed Balaeniceps to be a typical stork, linking the storks to the herons. He mentioned the " outstanding relationship " between Balaeniceps and Pelecanus, but did not seem to think it significant because he thought Pelecanus itself was so different from the other Pelecaniformes. The only other investigations of Balaeniceps' s anatomy were made by Technau (1937 : 567) during his studies of the nasal cavity of birds, and by Glenny (1955 : 560) in his work on the aortic arches of birds. The former drew no conclusions as to Balaeniceps's affinities, though Glenny thought it less like the Ciconiidae than is usually supposed.

After Chalmers Mitchell's contribution most authors placed Balaeniceps by itself in a group of equal rank with the herons and storks (e.g. Streseman, 1927-34 : 809 ; Wetmore, 1930 : 3). Mayr & Amadon (1951 : 6), however, followed Bohm's suggestion and placed it with the typical storks in the family Ciconiidae. Wetmore's classification (1951 revision) shows the generally accepted taxonomic position of Balaeniceps in relation to the orders Pelecaniformes and Ciconiiformes.

Order Pelecaniformes.

Suborder Phaethontes.

Family Phaethontidae. Suborder Pelecani.

Superfamily Pelecanoidea.

Family Pelecanidae. Superfamily Suloidea. Family Sulidae.

Phalacrocoracidae. Anhingidae. Suborder Fregatae.

Family Fregatidae.

Order Ciconiiformes. Suborder Ardeae.

Family Ardeidae.

Cochlearidae. Suborder Balaenicipites.

Family Balaenicipitidae. Suborder Ciconiae.

Superfamily Scopoidea.

Family Scopidae. Superfamily Ciconioidea.

Family Ciconiidae. Superfamily Threskiornithoidea.

Family Threskiornithidae. Suborder Phoenicopteri.

Family Phoenicopteridae.

ZOOL, 5, 3. 3§

54 THE PELECANIFORM CHARACTERS OF THE SHOE-BILL STORK

METHODS

As the purpose of this study is to examine the pelecaniform characters of Balaeniceps's skeleton in relation to its ciconiiform characters, the composition of the Pelecaniformes and Ciconiiformes will be discussed to decide what Balaeniceps ought to be compared with.

The living members of the Pelecaniformes are apparently not very alike. Nearly every genus is placed in a separate family. From a comparison of their osteology it seems that the differences are mainly due to adaptive radiation, and that there is a well-defined basic similarity. For instance, superficially, pelicans and cormorants look less alike than storks and herons, but their skeletons have more characters in common. A possible exception is Phaethon, which is peculiar in many respects and may not be closely related to the rest of the Pelecaniformes. Wetmore (1951 : 5) thinks that " the Phaethontes possibly may have separated earlier than the Fregatae " from the pelecaniform stock. Therefore, as Phaethon is atypical the Phaethontes will not be referred to in this investigation. The Fregatae are also considered aberrant by some authors, but they have so many of the osteological characters typical of the Pelecani that they are probably fairly closely related to them.

The Ciconiiformes is basically a less uniform group than the Pelecaniformes. Osteologically, it seems to be a collection of unrelated groups which, superficially, only have long beaks, long necks and long legs in common.

The genera of the Ardeae are very alike, their outstanding variation being in size. Cochlearius is the most aberrant genus but, apart from its skull, it has all the characters of the typical herons. Even in its skull the heron-like characters are not completely obscured. For the present purpose, therefore, the Ardeidae and Cochlearidae will be considered together, as a monophyletic group representing the herons.

The families in the Ciconiae are not so closely related. The Scopidae, with its single monotypic genus Scopus, is as enigmatic in its relationships as Balaeniceps. The skeleton of Scopus is like that of a small stork in some characters, but very unlike it in others. It has often been compared with Balaeniceps, and most authors consider the two related. However, there is no point in comparing one genus of doubtful affinities with another, so Scopus will not be referred to. The Ciconiidae is probably a monophyletic group ; its genera are fairly alike, although they vary more than those of the Ardeidae. This variation mainly seems to be due to different adaptations of the beak, correlated with differences in the size and shape of the head. The third family, Threskiornithidae, appears to have much in common with the Ciconiidae, but it also has certain resemblances to the Phoenicopteri. As the affinities of the Phoenicopteri themselves are controversial it is advisable not to discuss either group until their relationships have been more fully investigated.

For the purposes of this investigation, therefore, Balaeniceps is compared with the suborders Pelecani and Fregatae, representing the Pelecaniformes ; the ciconi- iform suborder Ardeae, representing the typical herons ; and the family Ciconiidae, representing the typical storks.

THE PELECANIFORM CHARACTERS OF THE SHOE-BIL STORK 55

Three complete skeletons of Balaeniceps and three skulls were available. There was also adequate material of pelicans and their allies, frigate birds, herons and storks. The skeleton of Balaeniceps was systematically compared with those of Pelecanus, Ardea amd Ciconia, but other genera, especially in the Pelecaniformes, were consulted to determine the range of variation in each group.

For convenience, in the following description Nannopterum, Halietor and Anhinga will not be mentioned unless they differ from Phalacrocorax.

OSTEOLOGICAL CHARACTERS

A. Skull, see Plate 3

(1) Premaxilla

Of the Pelecaniformes considered here, Pelecanus, Phalacrocorax and Fregata each have a well developed hook at the tip of the premaxilla. The newly hatched chick of Sula also has this hook, but it decreases with age, and in the adult the tip of the premaxilla is only slightly decurved. Anhinga has no hook in chick or adult, but this may be an adaptation to its habit of spearing fish.

In the Ciconiidae there is no suggestion of a hook to the premaxilla in any of the genera. The nearest approach is the decurved bill of Ibis and Mycteria, but in these the distal fifth of the mandible is involved in the curvature.

The Ardeidae have straight bills. Parker (1862 : 299) argues that in Cochlearius a hook " certainly does exist, although feebly " but, although the tip of the rhamphotheca is decurved, it is not hooked, and the premaxilla is quite straight ventrally.

Balaeniceps has a prominent hook at the tip of the premaxilla, like the typical Pelecaniformes.

(2) Nasal groove

In the Pelecani and Fregatae there is a conspicuous groove running along each side of the culmen from the anterior edge of the nostril to the cutting edge of the premaxilla beside the terminal hook. This relationship of the nasal groves to the premaxillary hook is constant in Pelecanus, Sula, Phalacrocorax and Fregata. In Anhinga the groves are only faintly indicated.

In the Ciconiidae the nasal groves are either absent, or represented by very shallow depressions which extend from the nostrils to, at most, half-way along the beak. Both conditions are often found in the same species.

The Ardeidae have shallow depressions like those of the Ciconiidae instead of nasal grooves. In Cochlearius these depressions are expanded to form broad, shallow troughs, each with a ridge along the mesial border.

Balaeniceps has conspicuous nasal groves which extend from the nostril to the cutting edge of the premaxilla beside the terminal hook, exactly as they do in the Pelecani and Fregatae. The grooves are not shallow, like those of the Ardeidae and Ciconiidae, or broad like those of Cochlearius, but deep like those of Pelecanus.

56 THE PELECANIFORM CHARACTERS OF THE SHOE-BILL STORK

(3) Nasal septum

In the Pelecani and Fregatae there is an ossified nasal septum. The nasal septum ot the Ciconiidae and Ardeidae is not ossified, and it is perforated in the region of the external nares. Cochlearius has a complete, unossified nasal septum. In Balaeniceps the nasal septum is ossified, as it is in the Pelecani and Fregatae.

(4) Nasal passage

In Pelecanus the external nares are vertically above, or even slightly posterior to the internal nares, and the nasal cavity lies almost vertically between them. In the other Pelecani the external nares are only slightly anterior to the internal nares. In the Ciconiidae and Ardeae the external nares are an appreciable distance anterior to the internal nares, and the nasal cavity lies obliquely between them. In Balaeniceps the relative positions of the nares and nasal cavity are exactly the same as they are in Pelecanus.

(5) Palate (See Fig. i)

In the Pelecani and Fregatae the palatines are always ankylosed along the mid- line posterior to the internal nares. There is usually a median ventral ridge, more or less well developed, along the suture, with a depression for the pterygoid muscle on either side of it. These depressions extend forwards past the posterior edge of the inter- nal narial opening only in Fregata. In the region of the internal nares the mesial edges of the palatines are parallel in the Pelecani, and nearly so in the Fregatae. In Pelecanus the ventral part of the nasal passage is divided along the mid-line by a membranous septum. There is no trace of an ossified prevomer in association with this septum, and in Sula the septum itself is weakly developed. The septum is better developed in Phalacrocorax, and in at least one species, P. urile, there is a thorn-like cartilaginous prevomer associated with it (unless it is carefully dissected out the prevomer is easily lost in prepared skeletons of Phalacrocorax). In Fregata the prevomer is also thorn-like, though longer and definitely ossified. The maxillopalatines vary in size throughout the Pelecani and Fregatae. In Sula and Phalacrocorax they are small and do not project beyond the palatines mesially. They are slightly larger in Fregata, and can be seen, in ventral view, bordering the anterior half of the internal nasal opening. In Pelecanus they are very large and meet in the mid- ventral line between the anterior ends of the palatines. Also in Pelecanus, they nearly fill the inside of the skull in the nasal region, and their posterior edges slant backwards in a straight line from the internal nares to the cranio-facial hinge. Posteriorly, the maxillopalatines do not extend past the cranio-facial hinge-line in any of the Pelecani or Fregatae.

The palatines of the Ciconiidae are not fused along the mid-line except at one point. Instead of the median ventral ridge found in the pelicans, there is a ventral crest along the mesial edge of each palatine where it borders the internal narial opening. The depressions for the pterygoid muscles extend further forward on either side of the narial opening than in Fregata. Immediately anterior to the internal nares the

THE PELECANIFORM CHARACTERS OF THE SHOE-BILL STORK 57

FIG. i. Diagrams of ventral views of palatine regions of (a) Balaeniceps rex, (b) Pele- canus crispus, (c) Phalacrocorax urile, (d) Ciconia ciconia, (e) Ardea goliath, (f) Coch- learius cochlearius.

i = maxillopalatine, 2 = prevomer, or position of unossified septum, 3 = palatine, 4 = depression for pterygoid muscle.

58 THE PELECANIFORM CHARACTERS OF THE SHOE-BILL STORK

palatines approach the mid-line, and in the larger genera, such as Leptoptilos and Jabiru, they may even touch. This divides the space between them into anterior and posterior parts, the nasal passage being confined to the posterior part. The prevomer arises at the posterior end of the internal narial opening, and the palatines are ankylosed at this point. The prevomer varies from a narrow strip of bone in Ibis, to a substantial triangular plate, drawn out into a thin filament anteriorly, in Leptoptilos and the larger genera. The maxillopalatines are well developed when compared with those of most Pelecani and Fregatae. They always meet in the mid- ventral line, where they occupy most of the space between the palatines anterior to the internal nares. Each maxillopalatine is extended posteriorly into a convex projection which usually reaches beyond the cranio-facial hinge ; in the Pelecani there is no such projection.

The palatines of the Ardeidae are like those of the Ciconiidae except that they are separate along the mid-line, even where the prevomer arises. In Cochlearius, however, they are ankylosed at this point, as they are in the Ciconiidae. The depressions for the pterygoid muscles extend as far forwards on either side of the internal narial opening as they do in the Ciconiidae. In the Ardeae, unlike the two previous groups, the vomer is V-shaped in cross-section, though this is less obvious in Cochlearius. The maxillopalatines meet in the mid-ventral line anterior to the nasal opening, much as they do in the Ciconiidae. Their posterior edges are convex like those of the Ciconiidae, and extend well beyond the level of the cranio-facial hinge. The maxillopalatines are smaller in Cochlearius, but otherwise they are very like those of the Ardeidae.

In Balaeniceps the palatines are ankylosed along the mid-line, posterior to the internal nares, with a broad ventral ridge along the suture. The depressions for the pterygoid muscles lie on either side of this ridge ; they extend forward to the level of the nasal aperture, but no further. The condition of the palatines is thus very like that of Pelecanus. The prevomer is weakly developed and its degree of ossification varies in the specimens examined. It is a thin, triangular plate, often perforated, lying in a vertical plane. In some specimens the apex does not reach the anterior end of the nasal opening. This weak development of the prevomer is reminiscent of the Pelecani, in which the prevomer is reduced and sometimes missing. The maxillopalatines are strikingly like those of Pelecanus. Their posterior faces are flat, even concave, and not convex like those of the Ciconiidae and Ardeae (Pycraft, 1898 : 83).

(6) Lachrymal (See Fig. 2)

The lachrymal in the Pelecani and Fregatae descends from the frontal to the quadrat ojugal bar, to which it is usually attached by a ligament. Viewed posteriorly, it is a column of bone with a lateral groove, of varying depth, to accommodate the lachrymal duct. In Phalacrocorax a narrow lateral process of the interorbital septum meets and fuses with the ventral end of the lachyrymal. This process is larger in Anhinga, lying beside the lachrymal throughout its length without touch- ing it. In lateral view, the lachrymal is more or less pillar-shaped in Pelecanus,

THE PELECANIFORM CHARACTERS OF THE SHOE-BILL STORK 59

Fregata and Phalacrocorax, but in Sula it is expanded anteriorly into the antorbital vacuity. There is a tendency, in the Pelecani and Fregatae, for this vacuity to be reduced. In Pelecanus it is comparatively large. In Fregata the maxilla grows back into it a little posteriorly. In Phalacrocorax there is a splint of bone resting on the quadrate jugal bar. This bone fills most of the antorbital vacuity in Anhinga, in which the maxilla is produced posteriorly as well. The large lachrymal itself fills most of the antorbital vacuity in Sula, though the maxilla and the quadratojugal also expand into it.

Although the lachrymal is well developed in the larger Ciconiidae, it never reaches the quadratojugal bar. In posterior view it is roughly triangular, with the apex of the triangle downwards. In some genera, including the four largest, the lachrymal

' «m.

/ cm.

Diagrammatic transverse sections of lachrymals of (a) Balaeniceps, (b) Sula,

(c) Ciconia, (d) Leptoptilos, (e) Ardea, (f) Cochlearius. i = lachrymal duct, 2 = lachrymal bone, 3 == quadratojugal bar.

duct is wholly or partly enclosed in bone, giving a flat surface to the outer face of the lachrymal bone. Unlike the Pelecani or Fregatae, the ciconiid lachrymal has a mesial projection extending towards the interorbital septum and passing ventral to the duct of the nasal gland. The lachrymal is triangular in cross-section, and it never extends into the antorbital vacuity. This vacuity is large in the Ciconiidae, and there is no obvious tendency for the surrounding bones to expand into it.

In the Ardeidae the lachrymal nearly reaches the quadratojugal bar. Its shape seems to be peculiar to the Ardeidae and is quite different from the Pelecani, Fregati and Ciconiidae. In Cochlearius the lachrymal is reduced, and in lateral view looks different from that of the Ardeidae ; but in cross-section it is almost identical. The antorbital vacuity is large in the Ardeidae and in Cochlearius.

Balaeniceps has a large lachrymal. In posterior view it is like the lachrymal of the Pelecani and Fregatae, a column of bone which meets the quadratojugal bar ventrally. The lacrymal duct lies in a large foramen through the lachrymal bone, as it does in some Ciconiidae. Anteriorly, the lachrymal fuses with the maxilla, so that the antorbital vacuity is obliterated. There is a slight groove which may

60 THE PELECANIFORM CHARACTERS OF THE SHOE-BILL STORK

represent the suture between the lachrymal and the maxilla. If it does, the lachrymal is pillar-shaped in lateral view as it is in the Pelecani and Fregatae. The complete occlusion of the antorbital vacuity, which occurs in Balaeniceps, is not found in any of the other groups considered here, but the Pelecani and Fregatae have a tendency towards reduction of the size of the antorbital vacuity.

Icm.

Icm.

Icm.

FIG. 3. Diagrams of articulating surfaces of quadrates and lower jaws of (a) Balaeniceps vex, (b) Sula bassanus, (c) Ciconia ciconia, (d) Cochlearius cochlearius.

i = left quadrate, 2 = left ramus of lower jaw, 3 = mesial articulating facets, 4 = lateral articulating facets, 5 = lateral process.

(7) Lower jaw articulation (See Fig. 3)

Each of the three groups being compared with Balaeniceps has a different arrange- ment of the articulating surfaces of the quadrate and lower jaw. The arrangement is constant within each group, except that the one typical of the Ardeidae is found mainly in the larger species.

In the Pelecani and Fregatae there are two articulating facets. On the quadrate, the mesial articulation has a broad ridge, which slides in a corresponding trough in the lower jaw. The long axis of the articulation lies at an angle of about 45° to the long axis of the skull, and is in line with the long axis of the pterygoid. This is

THE PELECANIFORM CHARACTERS OF THE SHOE-BILL STORK 61

especially noticeable in Pelecanus. In Sula, Fregata and Phalacrocorax the lateral edge of this ridge on the quadrate is undercut, and the corresponding side of the groove in the lower jaw is overhanging. The result is a locking mechanism which, in the dried skull, is efficient enough to hold the lower jaw in place when the beak is closed. The lateral articulation is usually S-shaped, though in Sula it may be L-shaped. Its long axis lies approximately at right angles to that of the other articulation. In Pelecanus, possibly because of its wide gape, the lateral part of this articulation is modified. In the lower jaw, instead of a sigmoid articular surface there is a groove, running anteriorly, parallel to the mid-line. Along this groove slides part of the lateral articulating surface of the quadrate. This groove in the lower jaw is represented in Sula by a shallow transverse trough, the anterior side of which projects laterally and dorsally and lies anterior to the lateral process of the quadrate when the beak is closed. The lateral process on the lower jaw is reduced in Fregata and Phalacrocorax.

The Ciconiidae also have two articulating facets. Unlike the Pelecani and Fregatae the long axis of the mesial facet is at right angles to the long axis of the skull, and at an angle of about 120° to the pterygoid. There is no locking mechanism. The lateral articulation is curved, so that while its lateral end is at right angles to the long axis of the mesial articulation, its mesial end is parallel to it. The lateral process on the lower jaw is well developed. The relationship between the two facets is quite different from that found in the pelicans or herons, and it is very alike in all the Ciconiidae examined, whatever the relative proportions of bill and skull.

In most of the larger Ardeidae there are four articulating facets, as each of those occurring in the pelicans and storks is in two parts. The lateral part of the mesial facet and the mesial part of the lateral facet lie on a plane nearly parallel to the pterygoid. On the quadrate the mesial facet, although it is in two parts, is undercut laterally to give a locking mechanism, as it is in most of the Pelecani and Fregatae. The lateral process on the lower jaw is more prominent than in the other groups described. In Cochlearius the articulating facets are essentially the same as in the larger herons, but the mesial facet on the quadrate is undercut mesially as well as laterally, apparently increasing the efficiency of the locking mechanism. The lateral process on the lower jaw is even better developed than in the Ardeidae, and, with the lateral part of the lateral articulation, seems to function as an auxiliary locking device.

Balaeniceps has two undivided articular facets, like the Pelecani, Fregate and Ciconiidae. On the quadrate, the mesial facet consists of a broad ridge, undercut laterally, which, in the lower jaw, slides in a trough with an overhanging lateral edge, much as it does in Sula, Phalacrocorax and Fregata. The mesial side of the trough also overhangs slightly, but not as much as in Cochlearius. The axis of the mesial articulation on the quadrate is in line with the pterygoid, as it is in the Pelecani and Fregatae, and is in contrast to the condition in the Ciconiidae. The lateral articula- tion is L-shaped, as it is in Sula ; its long axis is nearly at right angles to that of the mesial facet, like the Pelecani and Fregatae, and unlike the Ciconiidae. In the lower jaw, the lateral process is insignificant and the lateral articulation takes no

62 THE PELECANIFORM CHARACTERS OF THE SHOE-BILL STORK

part in the locking mechanism as it does in Cochlearius. Balaeniceps has none of the well defined ciconiid characters in its jaw articulation ; it resembles the Ardeae in some ways, but differs in others ; it is like the Pelecani and Fregatae in all the characters in which they differ from the Ciconiidae and Ardeae.

B. Pectoral Girdle (See Fig. 4)

(1) Furculum

There is a tendency in the Pelecani and Fregatae for the hypocleideum of the furculum to be fused to the keel of the sternum. The joint is ligamentous in Phalacrocorax ; sometimes ankylozed in Sula ; usually ankylosed in adults of Pelecanus ; and so ossified in adults of Fregata that the suture is obliterated. Except in Fregata each arm of the furculum forms an arc, convex anteriorly, between each coracoid and the carina sterni. Characteristic of the typical pelecaniform pectoral girdle is the well developed acrocoracoid flange, which forms a flat transverse surface on the clavicle for articulating with the coracoid. In Fregata the clavicle is completely fused to the coracoid in this region, but in young specimens the presence of the acrocoracoid flange can be inferred from the sutures. Although an acro- coracoid flange is present in several other apparently unrelated groups, it is never as well developed as it is in the Pelecani, Fregatae, Balaeniceps and Scopus.

In the Ciconiidae, although the hypocleideum of the furculum joins the carina sterni, it forms a bony fusion with it only in some specimens of one genus, Leptoptilos, and the suture is always obvious. Unlike the Pelecani, each clavicle forms a sigmoid curve in lateral view. The dorsal part of the curve is convex anteriorly, and the ventral part, where the clavicles unite in the mid-line is concave anteriorly. There is no indication of an acrocoracoid flange in any of the Ciconiidae.

The furculum of the larger Ardeae is mainly like that of the Ciconiidae. The joint between the hypocleideum and the carina sterni is always ligamentous. The presence of a small interclavicle is characteristic of the Ardeae, and it is not found in the other groups considered here.

The hypocleideum of Balaeniceps is fused to the carina sterni as it is in Fregata and most adult specimens of Pelecanus. The suture is obliterated by ossification in all the British Museum specimens. The clavicle of Balaeniceps forms a continuous arc from the coracoid to the carina sterni, as it does in the Pelecani. This character may not be significant, as the clavicle of Fregata is in a slightly sigmoid curve and that of Cochlearius is almost in a continuous curve. The acrocoracoid flange is well developed in Balaeniceps, a character typical of the Pelecaniformes.

(2) Sternum

In Balaeniceps the sternal keel extends along the whole length of the sternum to the posterior border as it does in the Ciconiidae, Ardeae and most other birds. In Pelecanus, Sula and Phalacrocorax it only reaches half-way back from the anterior end of the sternum. Parker (1860 : 329) considered this a significant difference between Balaeniceps and the Pelecani, but apparently was not aware of the condition

THE PELECANIFORM CHARACTERS OF THE SHOE-BILL STORK 63

£ 3

U->

N ,

a I

Q 'S

64 THE PELECANIFORM CHARACTERS OF THE SHOE-BILL STORK

in Fregata, in which the keel extends almost to the posterior end of the sternum. Thus within the Pelecaniformes both types of keel occur.

c. Pelvic Girdle

Parker (1861 : 336) considered the pelvis of Balaeniceps to be " typically ardeine " because it was narrow like that of the Ardeae. Chalmers Mitchell (1913 : 696) thought it more like the ciconiid pelvis because it had a notch in the posterior border, like the Ciconiidae, and lacked the ridge on the ilium which is present in the Ardeae. However, the shape and details of the pelvis in birds seem to depend mainly on the function and relative size of the legs and leg muscles, and the pelvis is probably a very adaptable part of the skeleton. In the Pelecaniformes, for example, the pelvis of Pelecanus, a bird with strong legs, seems to have very little in common with that of Phalacrocorax, in which the legs are weaker and used mainly for swimming, or of Fregata, in which the legs are very weak and only used for perching. In groups in which there is less adaptive radiation, such as the Ciconiidae, or Ardeae, the function of the legs is more uniform and the shape of the pelvis varies little within the group, the main differences being in size. In Balaeniceps the pelvis is roughly the same shape as it is in the Ardeae and some Ciconiidae, but it differs from both in details. It seems even less like that of any of the Pelecaniformes, but as there is already a good deal of variation of the pelvis in this group Balaeniceps would perhaps be less out of place with them than with the Ciconiidae or Ardeae.

D. Hind Limb (i) Tibio-tarsus

There are two forms of the distal condyles and the inter-condylar sulcus of the tibio-tarsus in the groups considered here. One is found in the Pelecani, Fregatae and Ardeae. In it the distal condyles are roughly semicircular in lateral view and the distal border of the outer condyle has no notch. The anterior aspect of the inter-condylar sulcus is fairly shallow, and the knob on the tarso-metatarsus which fits into it is not well developed. This type of articular surface is probably un- specialized, as it occurs throughout the Pelecani and Fregatae, in which there is considerable variation in the function of the legs, and in the Ardeae, in which the legs are long and unlike those of any of the pelican groups.

A second form occurs in the Ciconiidae. In it the distal borders of the condyles are flattened, and the condyles themselves are elongated posteriorly, so that they are oval in lateral view. There is a notch in the distal border of the outer condyle. The anterior aspect of the inter-condylar sulcus is deep, and proximal to it there is a hemispherical depression with a prominent knob immediately beside it. The knob on the tarso-metatarsus is much larger than in the first type, and articulates with the hemispherical depression when the leg is bent. The second condition apparently only occurs in long-legged birds, such as the Threskiornithidae and Phoenicopteridae, and to a lesser extent in the Gruidae and long-legged Charadrii.

The form in Balaeniceps is similar to that of the Pelecani, Fregatae and Ardeae.

THE PELECANIFORM CHARACTERS OF THE SHOE-BILL STORK 65

(2) Tarso-metatarsus

In most groups of birds the hypotarsus is well ossified to form a varying number of " bridges " which enclose the flexor tendons in bony tubes. Of the Pelecani, Phalacrocorax has one tube and Sula and Pelecanus two.

In contrast, the Ciconiidae have a so-called " simple " hypotarsus. It consists of two parallel bony ridges with a groove between. The flexor tendons lie in this groove and are supported by unossified ligaments instead of bony bridges.

The Ardeae have a more ossified hypotarsus, rather like that of the Pelecani. In most genera there is only one tube, but the smaller genera sometimes have more.

Balaeniceps has two complete bony tubes through the hypotarsus. Their arrangement is strikingly like that of Pelecanus, and quite unlike the Ciconiidae.

(3) First metatarsal

In the Ciconiidae and Ardeae the first toe points backwards. In the Pelecani it is joined to the second toe by a web and is restricted to a lateral position, although it is mobile enough to be able to be pointed forwards. Parker (1861 : 344) says that in Balaeniceps the first toe is " very mobile " and is turned " very far inwards " when walking. Photographs show that it is directed backwards when the bird is standing still.

The position of the first toe influences the form of the first metatarsal. When the toe normally points backwards the metatarsal, if straight, would lie in the same plane as the flexor tendons of the other digits and interfere with their functioning. But the shape of the metatarsal is modified, usually giving it the appearance of bending round to one side of the tendons, and it often has a diagonal groove in which the tendons run freely.

In Pelecanus, in which the first toe does not point backwards, the metatarsal is straight, with only a shallow depression, mid-way along its length, where it touches the flexor tendons. In the other Pelecani this depression varies in size and depth, but it is never so marked as it is in the Ciconiidae and Ardeae. In Sula and Phala- crocorax the metatarsal is slightly bent round the flexor tendons.

In the Ciconiidae there is a broad, deep, diagonal trough for the flexor tendons, and the metatarsal appears twisted through an angle of about 90°. In the Ardeae the first metatarsal does not press against the flexor tendons as closely as it does in the Ciconiidae, because of the way in which it articulates with the first phalanx. As a result the diagonal groove in which the tendons lie is less marked than it is in the Ciconiidae and narrower than it is in the Pelecani.

In Balaeniceps the metatarsal has a depression for the flexor tendons which is very little deeper than that of Pelecanus. It is shallower than that of the Ciconiidae and broader than that of the Ardeae. The metatarsal appears slightly twisted, though less so than it is in the Ciconiidae. The form of the first metatarsal and the function of the first toe of Balaeniceps therefore seem to be more like those of the Pelecani than the Ciconiidae or Ardeae.

66 THE PELECANIFORM CHARACTERS OF THE SHOE-BIL STORK

(4) Toe articulations

The proximal articulating surfaces of the basal phalanges of the second, third and fourth digits are fairly alike in the Pelecani and Ciconiidae, being roughly square in shape. In the Ardeae each of these articulating surfaces has a characteristic, irregular shape. Balaeniceps is like the first two groups, with the articulations almost square in proximal view.

SOME NON-SKELETAL PELECANIFORM CHARACTERS OF BALAENICEPS

(a) von Heuglin (1873 : 1095)

The egg is white with chalky lumps. Similar chalky lumps and nodules occur on eggs of Phalacrocorax and Sula.

Birds join up in parties to herd shoals of fish into corners. This communal fishing is characteristic of some Phalacrocorax and Pelecanus species.

(6) Chalmers Mitchell (1913)

The rhamphotheca is compound, as it is in the Pelecani and Fregatae.

A pyloric chamber is present in the stomach, as in Pelecanus.

The dermo-temporalis, biventer maxillae, temporal and pterygoid muscles are similar in Pelecanus.

There are no intrinsic muscles of the syrinx in Balaeniceps and Pelecanus.

The hyoid muscles are " excessively like those of Pelecanus."

The division of the pectoral muscle is similar in Pelecanus.

The arrangement of the wing tendons is the same in Pelecanus.

(c) Technau (1936 : 567)

The secondary nostrils can be closed, as in Pelecanus.

(d) Glenny (1955)

The right carotid is absent in Balaeniceps and some Pelecaniformes. When one carotid is missing in the Ciconiiformes it is the left one.

DISCUSSION

Those who have studied Balaeniceps's affinities from its skeleton seem to have been mainly concerned with its heron-like or stork-like features, and have neglected to consider its likeness to the pelicans. Jardine (1852) may have been responsible for this when he noted what he thought were significant differences from the " true pelicans ". Earlier impressions of Balaeniceps however were that it was near the pelicans. For example, its first mention in literature was by Ferdinand Werne (1848 : 143) who recorded that on I5th December 1840, " During my siesta someone saw a water bird that seemed to be as big as a young camel, which actually had a beak just like a pelican's, only without the pouch beneath it ". Even Gould (1852) referred to it as a kind of pelican, and Chalmers Mitchell (1913 : 701) considered that this opinion was " at least as happy as the more confident statements of later writers". None of the non-pelican characters given by Jardine are skeletal. No

THE PELECANIFORM CHARACTERS OF THE SHOE-BILL STORK 67

evidence from comparative osteology has been given as a reason for not putting Balaeniceps with the Pelecaniformes, although this has usually been implied on the few occasions when differences between Balaeniceps and the Pelecaniformes have been described. For example, Parker (1860 : 329) when describing the sternum, mentions that the keel extends to the posterior end of the sternum in Balaeniceps, as it does in the storks and herons, " whereas in the Pelicans, Gannets and Cormorants it scarcely continues beyond the middle of that bone ".

On the other hand, skeletal characters common to Balaeniceps and the Pelecani- formes have often been referred to. Sometimes they have been attributed to conver- gence (Parker 1861 : 308), or to the " common inheritance " of the Pelecaniformes and Ciconiiformes (Chalmers Mitchell 1913 : 699), but more often they are mentioned without comment or even without reference to the fact that they occur in both Balaeniceps and the Pelecaniformes. These characters are summarized below. They are arranged under three headings, and when a character is mentioned by more than one author, or under more than one heading, it is only referred to the first time it occurs on the list. Only original works are referred to. An asterisk is placed against the characters considered in the present investigation.

CHARACTERS OF BALAENICEPS'S SKELETON THAT ALSO OCCUR IN THE PELECANIFORMES.

A. Noted and commented on by authors

Parker (1861 : 308)

* The palatines have the " same essential structure " in other fish eating birds, such as the Pelican, Cormorant and Gannet, because the " motions of the upper jaw on the cranium " are the same.

Chalmers Mitchell (1913 : 699)

* Long lachrymals.

* Mesial ankylosis of the palatines. Shell-like paroccipital processes.

* Clavicle ankylosed to carina sterni. Shape of the head of the humerus.

He says these are either due to " convergent modifications between birds which, after all, are not very far apart in the system " or to the " common inheritance " of the Pelecaniformes and " their immediate allies ".

Bohm (1930 : 700)

Balaeniceps resembles Pelecanus in its closed palate.

* Hook to premaxilla.

* Bony nasal septum. Complete interorbital septum. Well developed postorbital process.

Lack of a postangular process on the lower jaw.

68 THE PELECANIFORM CHARACTERS OF THE SHOE-BILL STORK

He says the Pelecaniformes are more like the herons than the storks, except for Pelecanus which is atypical, and more like the storks and Balaeniceps.

B. Noted by authors, without comment

Parker (1860)

Cervical vertebrae have haemal arches (p. 328).

* Furculum articulates with acrocoracoid flange (p. 329). Tongue is small (p. 330).

Parker (1861)

" General class resemblance " in occipital region (p. 275). Sudden bend in furculum (p. 340).

Chalmers Mitchell (1913)

* Nasal groove (p. 690).

c. Mentioned by authors, without reference to the Pelecaniformes

Parker (1861)

* Arrangement of the articulating surfaces of the quadrate and lower jaw (p. 310).

Ischium is longer than ilium posteriorly (p. 337).

No prepubic process (p. 337).

Wing skeleton — Parker says this is like that of the herons (p. 342) ; I found

it as much like that of Pelecanus as Ardea.

Well developed " cnemial ridges " in tibio- tarsus (p. 343).

Slight sigmoid curve at distal end of tibio-tarsus (p. 343).

" Anterior cavity " at proximal end of tarso-metatarsus is deeper than

in Ardea (p. 343).

* Complex hypotarsus (p. 344).

* Mobile hallux (p. 344).

Chalmers Mitchell (1913)

A rounded notch separates the metasternum from the posterior lateral

processes of the sternum (p. 694). Bases of coracoids do not meet in the mid-line (p. 696). A notch separates the posterior ends of the ilium and ischium (p. 696). No horizontal ridge formed by the " dorso-lateral edge of the post-acetabular

ilium " (p. 696). A tibial bridge is present (p. 697).

Of these skeletal characters common to Balaeniceps and the Pelecaniformes, some have been described as being due to " convergence " and others to " common inheritance ". It is one of the problems of taxonomy to distinguish between these two causes. In this instance the problem is to determine what are significant taxonomic characters in the pelecaniform skeleton. At the same time it might not

THE PELECANIFORM CHARACTERS OF THE SHOE-BILL STORK 69

be out of place to refer to the wider problem of the taxonomic significance of osteological characters One point of view is that bone is not an easily adapted substance, and that therefore phylogeny is readily determined from an examination of the skeleton. Verheyen (1953 : 480), for example, says that " systematics based on comparative osteology is perfectly realizable " since osteological characters are " practically invariable " and are " sheltered from the adaptations and modifica- tions imposed by frequent habit ". There is clearly a good deal of truth in this, but it is a point of view that should be regarded with caution. There is evidence to show that bone is a plastic substance readily moulded by any change in the forces exerted by the muscles attached to it. This opinion is expressed, for example, by Weinmann & Sicher (1947 : 120) who say, " if it be true that functional stresses shape the bone, then it is equally true that a change of strength or direction of forces will lead to changes in the form and structure of bones ". Changes in muscle function related to changes in habit are therefore reflected in the skeleton. Similarities in habit of unrelated species and differences in habit of related species can produce a crop of adaptive osteological characters which may obscure phylogeny. Phylogeny may be apparent only in a number of small characters which have been relatively unaffected by adaptive changes. The sum of these characters may be peculiar to a particular group. Although the members of such a group vary in appearance and habit, and show convergence with other groups, they will have most of the small characters typical of their group. These " non-adaptive " characters differ from group to group and may occur in different parts of the skeleton, so that each group must be studied separately to get the " feel " of its typical characters.

In the light of these observations it will be understood that a " pelicaniform character " is hard to define precisely. The skeletal characters considered here are mainly those which distinguish the Pelecani and Fregatae from the Ciconiidae and Ardeae. Some of them, for example the acrocoracoid flange, also occur in other groups of birds. For this reason authors have not regarded it as important that Balaeniceps has them (Chalmers Mitchell, 1913 : 695). However, they have been included here because it is now understood that any given character may be taxonomically significant in one group, but not necessarily in another (e.g. see Cain, 1954 : 268). The acrocoracoid flange distinguishes the Pelecani, Fregatae and Balaeniceps from the Ciconiidae and Ardeae, but not from the Scopidae, Falconiformes, most Charadriiformes, Columbidae, Strigiformes, some Procel- lariiformes, and many other groups. It is not intended to imply that all the groups with an acrocoracoid flange are related, or that any of them are necessarily more closely related to the pelicans than to the storks.

Not all the " pelicaniform characters " considered here occur throughout the Pelecani and Fregatae. Sometimes one genus, or more, may differ in one feature from the others. For example, adults of Anhinga and Sula have no hook at the tip of the premaxilla, but they are typical in most other respects.

In other cases there may be a general trend, or tendency within the group, towards a certain condition, though all the genera are not necessarily concerned in it. One example of this is the tendency for reduction of the antorbital vacuity. In

70 THE PELECANIFORM CHARACTERS OF THE SHOE-BILL STORK

Pelecanus, the antorbital vacuity is large, as it is in most birds ; in Fregata, Phala- crocorax, Anhinga and Sula it becomes progressively reduced. Balaeniceps, there- fore, with no antorbital vacuity, would complete the series.

Sometimes an underlying pattern can be traced in a structure, with differences in details in each genus. A good example of this is the arrangement of the quadrate condyles in the jaw articulation (see Text-figures). The general form of the nasal cavity and of the palate possibly come into this category. In each of these instances Balaeniceps has the same underlying pattern as the Pelecaniformes, but the storks and herons do not.

To sum up, the general skeletal features which can be described as " pelecaniform " and which occur in Balaeniceps but not in the storks and herons are as follows :

(1) Position of nasal groove along upper mandible, and strong terminal hook (see A(I) and A(Z)).

(2) Arrangment of nasal cavity (see A(3) and A(4)).

(3) Relationship of bones of palate and maxillopalatines (see A (5)).

(4) Size of lachrymal and antorbital vacuity (see A (6)).

(5) Type of jaw articulation (see A (7)).

(6) Some features of the pectoral girdle and sternum (see B).

(7) Shape of first metacarpal (see 0(3)).

In my opinion the skeleton of Balaeniceps has many points of similarity, due to convergence, with the Ciconiidae and Ardeae, but, in spite of its difference in out- ward appearance from any of the Pelecaniforms, it shares several apparently non- adaptive features with them. I find it difficult to account for this unless Balaeniceps is more closely related to the Pelecaniformes than it is usually considered to be. Therefore, from a consideration of the skeletal characters of Balaeniceps rex, it seems that this species could occupy a monotypic family in the order Pelecaniformes, possibly near the Pelecanidae.

SUMMARY

1. A number of features of the skeleton of Balaeniceps rex were found to be more like the pelicans than either the storks or herons, with which Balaeniceps is usually grouped.

2. A study of the literature showed that the pelican-like characters of Balaeniceps had never been fully investigated.

3. The skeleton of Balaeniceps was compared with those of all the families of the Pelecaniformes, except the Phaethontidae, and with the Ardeidae, Cochlearidae and Ciconiidae of the Ciconiiformes. Reasons are given for limiting comparison to these groups.

4. The characters common to Balaeniceps and the Pelecaniformes are described in detail.

5. The osteological evidence suggests that Balaeniceps is more closely related to the Pelecaniformes than to the Ciconiiformes, and the family Balaenicipitidae may reasonably be placed in the Pelecaniformes, possibly near the Pelecanidae.

THE PELECANIFORM CHARACTERS OF THE SHOE-BILL STORK 71

REFERENCES BARTLETT, A. D. 1861. On the affinities of Balaeniceps. Proc. Zool. Soc. London, 1861 :

131-134- BEDDARD, F. E. 1888. On certain points in the visceral anatomy of Balaeniceps vex bearing

on its affinities. Ibid., 1888 : 284-290. BOHM, M. 1930. tJber den Bau des jugendlichen Schadels von Balaeniceps vex nebst Bemer-

kungen iiber dessen systematische Stellung und iiber das Gaumenskelett der Vogel. Z.

Morph. Okol. Tiere, Berlin, 17 : 677-718.

BONAPARTE, C. L. J. L. 1855. Conspectus generum avium, 1849-57. Lugduni Batavorum. CAIN, A. J. 1954. Subdivisions of the genus Ptilonopus. Bull. Brit. Mus. (nat. hist] Zool

2 (8) : 267-284. DES MURS, M. O. 1859. Considerations oologiques sur 1'oiseau type du genre Bal6niceps.

Rev. et Mag. Zool. Ser. 2, 11 : 477-481. FURBRINGER, M., 1888 Untersuchungen zur Morphologie und Systematik der Vogel. 2 : 1 187-1 196.

Amsterdam. GADOW, H. 1893. Vogel : Aves in Bronn's Klass. n. Ord. des. Thier-Reichs. Bd. 6, Abth 4.

Leipzig.

GIEBEL, C. G. 1873. Balaeniceps rex. Z. ges. Naturw. Berlin, 41 : 350-354. GLENNY, F. H. 1955. Modifications of pattern in the aortic arch system of birds and their

phylogenetic significance. Proc. U.S. nat. Mus. No. 3346, 104 : 525-621. GOULD, J. 1852. On a new and most remarkable form in ornithology. Proc. Zool. Soc.

London, 1851 (published 1852) : 1-2. HEUGLIN, M. T. VON. 1856. Systematische Uebersicht der Vogel Nord-Ost-Afrika's. Wien.

1873. Ornithologie Nordost-Afrika's, 3 : 1095-1099. Cassel.

JARDINE, W. 1852. Ornithology in 1850 ; Balaeniceps rex. Contr. Orn. 1851 (published 1852) :

11-14.

MAYR, E. & AMADON, D. 1951. A classification of recent birds. Amer. Mus. Novit. 1496. MITCHELL, P. C. 1913- Observations on the anatomy of the Shoe-bill (Balaeniceps rex) and

allied birds. Proc. Zool. Soc. London, 1913 : 644-703. PARKER, W. K. 1860. Abstract of notes on the osteology of Balaeniceps rex. Ibid., 1860 :

324-330.

1861. On the osteology of Balaeniceps rex. Trans. Zool. Soc. London, 4 : 269-352.

1862. [Letter on Reinhardt's 1861 paper.] Ibis, London : 297-299.

PYCRAFT, W. P. 1898. Contributions to the osteology of birds. I. — Steganopodes. Proc.

Zool. Soc. London, 1898 : 83. REINHARDT, J. 1860. On the affinities of Balaeniceps. Ibid., 1860 : 377-380.

1861. Some remarks on the genus Balaeniceps. Ibis, London, 1862 : 158-175. [Trans- lation by A. Newton of article in K. danske Vidensk. Selsk. Keberih. 1861 : 135-154.]

SHUFELDT, R. W. 1901. Notes on the osteology of Scopus umbretta and Balaeniceps rex.

J. Anat. Physiol. 2 : 405-412. STRESEMANN, E. 1927-34. Aves in Kukenthal & Krumbach's Handb. d. Zool. Bd. 7 Hft. 2.

Berlin and Leipzig.

TECHNAU, G. 1936. Die Nasendriise der Vogel. /. Orn. Leipzig. 84 : 511-617. VERHEYEN, R. 1953. Bijdrage tot de osteologie en de systematick der Anseriformes. Gerfaut,

Bruxelles, 43. Supplement : 373-500.

WEINMANN, J. P. & SICKER, H. 1947. Bone and Bones. London.

WERNE, F. 1848. Expedition zur Entdeckung der Quellen des Weissen Nil (1840-41). Berlin. WETMORE, A. 1930. A systematic classification for the birds of the world. Proc. U.S. nat.

Mus. 76, 24 : 1-8.

1951. A revised classification for the birds of the world. Smithson. misc. Coll. 117,

4 : 1-22.

PLATE

[Scale : The skulls have been variously reduced, so that the crania are of approximately the same size. The actual total length of each skull is given below in brackets.] Lateral views of skulls of :

(1) Balaeniceps vex (265 mm.).

(2) Pelecanus onocrotalus (410 mm.).

(3) Sula bassanus (180 mm.).

E = external naris.

G — nasal groove.

H = premaxillary hook.

i = position of internal nares.

(4) Ciconia ciconia (255 mm.).

(5) Ardea goliath (235 mm.).

(6) Cochlearius cochlearius (125 mm.

L = lachrymal.

M =maxillopalatine.

p = palatine.

v = antorbital vacuity.

Bull. B.M. (N.H.) Zoo/. 5, 3.

PLATE 3

PRINTED IN GREAT BRITAIN BY ADLARD AND SON, LIMITED, BARTHOLOMEW PRESS, DORKING

REVISION OF THE LAKE VICTORIA HAPLOCHROMIS SPECIES (PISCES, CICHLIDAE)

PART II: H. SAUVAGEI (PFEFFER), H. PRODROMUS TREWAVAS,

H. GRANTI BLGR., AND H. XENOGNATHUS SP. N.

P. H. GREENWOOD

BULLETIN OF

THE BRITISH MUSEUM (NATURAL HISTORY) ZOOLOGY Vol. 5 No. 4

LONDON: 195?

A REVISION OF THE LAKE VICTORIA

HAPLOCHROMIS SPECIES (PISCES, CICHLIDAE)

PART II: H. SAUVAGEI (PFEFFER),

H. PRODROMUS TREWAVAS,

H. GRANTI BLGR, AND H. XENOGNATHUS, SP.N.

BY

P. H. GREENWOOD

East African Fisheries Research Organization, Jinja, Uganda.

Pp. 76-97 ; Plate 4 ; 8 Text-figs.

BULLETIN OF

THE BRITISH MUSEUM (NATURAL HISTORY) ZOOLOGY Vol. 5 No. 4

LONDON: 1957

THE BULLETIN OF THE BRITISH MUSEUM (NATURAL HISTORY), instituted in 1949, is issued in Jive 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.

This paper is Vol. 5, No. 4 of the Zoological series.

PRINTED BY ORDER OF THE TRUSTEES OF THE BRITISH MUSEUM

Issued October, 1957 Price Eight Shillings

A REVISION OF THE LAKE VICTORIA HAPLOCHROMIS SPECIES (PISCES, CICHLIDAE)

PART II1: H. SAUVAGEI (PFEFFER), H. PRODROMUS TREWAVAS,

H. GRANTI BLGR. AND H. XENOGNATHUS, SP. N.

By P. H. GREENWOOD

CONTENTS

INTRODUCTION .... Haplochromis sauvagei (Pfeffer) Synonymy and description Distribution . Ecology

Diagnosis and affinities Study material and distribution records

Haplochromis prodromus Trewavas . Synonymy and description Distribution ..... Ecology .....

Study material and distribution records Diagnosis and affinities

Haplochromis granti Boulenger Synonymy and description Distribution .....

Ecology ......

Diagnosis and affinities

Study material and distribution records

Haplochromis xenognathus sp. nov. Synonymy and description Distribution . Ecology

Diagnosis and affinities Study material and distribution records

DISCUSSION ......

SUMMARY .......

ACKNOWLEDGMENTS .....

REFERENCES ......

1 Part I, see Greenwood, 19566. ZOOL. 5, 4.

Page 76 76

77 80 80 80 81

82 82 85 85 85 86

86 86 89

90

90 9i 94 94 94 95

95 97 97 97

4§

76 REVISION OF LAKE VICTORIA HAPLOCHROMIS SPECIES

INTRODUCTION

MORPHOLOGICALLY, the H. sauvagei complex stands apart from any other species- group in Lake Victoria. The principal group character is that of the dentition which combines recurved outer teeth with multiseriate inner tooth-bands (Text-

fig. 3).

Furthermore, the shape of the neurocranium, although differing intra-specifically within the group, is unlike that of other Haplochromis. This character is probably associated with the multiseriate dentition and relatively powerful jaw musculature. Indeed, amongst the non-piscivorous predators such marked divergence in cranial anatomy is otherwise only found in mollusc-eating species with hyper-developed pharyngeal bones and musculature. Two specifically constant forms of neurocranium occur in the " sauvagei " group, but neither can be correlated with the type of dental pattern present.

Trophically, members of the group may be classed as mollusc eaters, although available data indicate that other food organisms do contribute to their diet, usually in a subsidiary capacity. Unlike other mollusc-eating Haplochromis in this lake, species of the " sauvagei " group do not swallow the shells of their prey, but remove the soft parts before ingestion takes place. In this respect the feeding method is like that of Macropleurodus bicolor (Blgr.), a monotypic genus apparently derived from this group.

Haplochromis sauvagei (Pfeffer), 1896 (Text-fig, i and PI. i upper fig.)

Ctenochromis sauvagei Pfeffer, 1896, Thier. Afr. Fische, 15.

Haplochromis nuchisquamulatus (part), Boulenger, 1915, Cat. Afr. Fish., 3, 290.

Paratilapia granti (part), Boulenger, 1915, op. cit., 342.

Paratilapia bicolor (part) Boulenger, 1915, op. cit., 346.

Paratilapia retrodens (part), Boulenger, 1915, op. cit., 235.

Haplochromis sauvagei (part), Regan, 1922, Proc. zool. Soc., Land., 167.

? Paratilapia crassilabris (part), Boulenger, 1915, op. cit., 345.

I was unable to examine the holotype of H. sauvagei which was mislaid during the 1939-45 war ; at present the Berlin Museum authorities cannot confirm whether this specimen has been lost. Pending more definite information, no neotype can be selected, but, should such a step become necessary, I suggest that the specimen B.M. (N.H.) Reg. No. 1956.9.17.1, a male from Entebbe (Text-fig, i) be given neotypical status.

Fortunately, Pfeffer's original description of Ctenochromis sauvagei is compre- hensive, and, when coupled with a photograph of the type, clearly indicates to which Haplochromis species his specimen should be referred. The photograph, preserved in the British Museum (Natural History), is reproduced in Plate i.

Additional material discloses only one important discrepancy with the original description, in which the mouth and lower- jaw profile are described as rising steeply : " . . . ; das untere Profil der Unterkinnlade steigt viel starker. Die von dicken und breiten Lippen umgebene kurze Mundspalte steigt nach vom sehr steil auf."

REVISION OF LAKE VICTORIA HAPLOCHROMIS SPECIES 77

In most specimens the ventral head profile is almost horizontal, or, at most, slightly oblique. A possible explanation for this discrepancy may lie in the fact that Pfeffer's description was taken from a fish preserved with its mouth open. From the photo- graph it is clear that, if the jaws were restored to their natural position, the cleft and lower jaw profile would be slightly oblique.

The present synonomy for H. sauvagei is essentially that prepared by Regan (1922), but some of the specimens which he referred to this species are now placed in others. In this connection, reference should be made to the list of study material.

Paratilapia crassilabris part (Boulenger, 1915) is tentatively retained in the synonomy on the basis of a single specimen (B.M. (N.H.) Reg. no. 1911.3.3.32). This individual cannot be identified with certainty, but it is nearer H. sauvagei than any other species with thickened lips and dentition not of the generalized type.

FIG. i. Haplochromis sauvagei, £, B.M. (N.H.) 1956.9.17.1. Drawn by Miss L. Buswell.

Description. Based on 85 specimens, 58-105 mm. standard length. Of the measurements made, only cheek depth clearly shows allometry with standard length.

Depth of body 30-4-41-8, mean (M) = 35-6 ; length of head 29-6-34-5 (M = 31-9) per cent of standard length. Dorsal head profile varying from decurved to straight, but strongly sloping, the former shape occurring more frequently.

Preorbital depth 15-4-20-2 (M = 17-3) per cent of head length, least interorbital width 23-0-31-2 (M == 27-0) per cent. Snout as broad as long, its length 27-2-35-5 (M = 30-8) per cent of head ; eye diameter 25-7-33-4 (M = 28-9) per cent. Cheek becoming relatively deeper with increasing standard length ; four size-groups are recognized, 58-69 mm. S.L. (N = 13), 70-80 (N = 21), 81-90 (N = 27) and 91- 105 (N = 24), for which the cheek depth is 21-0-25-0 (M = 23-2), 20-4-26-0 (M = 23-7), 22-0-26-9 (M = 24-4) and 24-1-26-6 (M = 25-1).

78 REVISION OF LAKE VICTORIA HAPLOCHROMIS SPECIES

Caudal peduncle 13-9-19-3 (M = 16-4) per cent of standard length ; its length 1-1-1-6, times its depth.

Mouth horizontal or slightly oblique ; posterior maxillary tip reaching or almost reaching the vertical to the anterior orbital margin. Lips thickened ; the depth of the upper lip, measured mid-laterally, is contained 4-4^ times in the eye-diameter. Jaws equal anteriorly, the lower 30-6-37-7 (M = 34-5) per cent of head length and 1-0-1-5 (mode 1-3) times as long as broad.

Gill rakers short, 7-9 (rarely 10) on the lower limb of the first arch.

Scales ctenoid ; lateral line with 31 (1.7), 32 (f.29), 33 (f-39), 34 (f.g) or 35 (f.i) scales ; cheek with 3-4 (rarely 2) series. 7-9 (less frequently 6) scales between dorsal fin origin and upper lateral line ; 7 or 8 (less commonly 6 or 9) between pectoral and pelvic fin-bases.

Fins. Dorsal with 24 (f.n), 25 (f.56) or 26 (f.i8) rays, anal 10 (f.i), n (f.io), 12 (f.7i) or 13 (f.3), comprising XV-XVII, 8-10, and III, 7-10 spinous and branched rays for the fins respectively. Pectoral slightly shorter than the head, or occasionally of equal length. Pelvic with the first ray produced, variable in its posterior extension but longer in adult males than females. Caudal sub-truncate.

Lower pharyngeal bone triangular, its dentigerous area 1^-1$ times as broad as long ; pharyngeal teeth slender and cuspidate. In some specimens, teeth in the median rows are slightly enlarged, but retain their bicuspid crowns.

Teeth. In the outermost series of both jaws, the teeth have strongly recurved tips and are unequally bicuspid or unicuspid. The predominant tooth form is apparently correlated with length. Fishes less than 80 mm. S.L. have mainly bicuspid teeth, those in the range 80-90 mm. have either unicuspids or an admixture of uni- and bicuspids, whilst larger individuals possess mainly unicuspid teeth. When both types of teeth are present, the unicuspid form usually occurs anteriorly and laterally. There are 32-56 (mode 42) outer teeth in the upper jaw.

Inner teeth are either tri- or unicuspid ; as in the outer series, unicuspid teeth are commoner in fishes above 80 mm. S.L. Antero-medially, the teeth are arranged in a broad band comprising 3-8 (mode 4) and 2-6 (modes 3 and 4) rows in the upper and lower jaws respectively. Laterally, the band narrows to a single series. A distinct inter-space separates the inner and outer series.

Syncranium and associated musculature. Neurocranial form in H. sauvagei departs quite considerably from the generalized Haplochromis type, and approaches that of Macropleurodus bicolor (Greenwood, 19560). Essentially the same points of dif- ference with the generalized type occur in both species. The skull has a fore- shortened appearance due to the strongly decurved and almost vertically disposed ethmovomerine region. This curvature affects the morphology of the entire pre- orbital skull which is less gently curved than in the generalized neurocranium.

On the other hand, the jaws do not exhibit such radical departure from the generalized condition. The premaxilla, apart from a slight broadening of its denti- gerous area, compares closely with that of other Haplochromis ; the dentary is somewhat shorter, more massive and has a wider median dentigerous area than is common in generalized species. Consequent upon these modifications slight differ- ences are apparent in the suspensorium.

REVISION OF LAKE VICTORIA HAPLOCHROMIS SPECIES

79

Muscle disposition and form are similar to those of basic Haplochromis species. However, the adductor mandibulae I is slightly longer (38-44 per cent head length compared with 36-39 per cent) and broader (length /breadth ratio 3-0-3-7 cf.

4-4-5-5)-

The syncranium, its musculature and the dentition all foreshadow the condition found in H. prodromus, and therefore that which reaches its ultimate expression in the genus Macropleurodus Regan (Greenwood, op. cit.). It is perhaps significant that variability in the degree to which the syncranium departs from the basic type is greater in H. sauvagei than in H. prodromus.

Causal factors responsible for the characteristic preorbital face in both H. sauvagei and H . prodromus are not readily determined. From an examination of larval fishes it is manifest that, as in M. bicolor, this form develops during post-larval ontogeny. It is probably effected by differential growth of various syncranial parts, especially since the ethmovomerine region is not directly affected by the moulding influence of muscle insertions. However, the premaxilla, which is closely associated both anato- mically and functionally with the dentary, could exert considerable influence over this region. In H. sauvagei the dentary is short in relation to the head and also in comparison with other Haplochromis species of comparable size. If, during post- larval ontogeny, this bone increased in length more slowly than the neurocranium, and if it is to remain functionally integrated with the upper jaw, then there can be two morphological results : either a skull of the H. sauvagei type, or one in which the upper jaw projects anteriorly beyond the lower. A third possibility, that the sus- pensorium be rotated anteriorly, cannot be considered in this case, since its almost vertical alignment in H. sauvagei is typically that of the basic type.

H. sauvagei includes Gastropoda as a substantial part of its diet. As in H. prodromus and M. bicolor the soft parts alone are ingested. The feeding habits of M. bicolor have been described elsewhere (Greenwood op. cit.) : when feeding on snails, aquarium-kept H. sauvagei follow the same general pattern, except that after grasping the foot of the snail between its jaws the fish then uses the shell as a fulcrum to lever out the soft parts. Only rarely is the shell crushed by the jaws.

Coloration in life : Breeding males. Ground colour dark grey-green or blue-grey, lighter or yellowish ventrally ; a suffused coppery sheen on the flanks and ventral aspects of the operculum. Dorsal fin black basally, becoming slate-coloured distally ; lappets red ; red spots, often coalesced, between the soft rays. Anal dark with a red flush ; ocelli yellow. Caudal dark grey proximally, lighter distally, and with an overall orange-red flush. Pelvics black laterally, orange-red medially. Non- breeding males have similar coloration except that the copper flush is absent and other bright colours are less intense. Females and immature males. Ground col- our golden-green, shading to pearly- white ventrally. All fins yellow-green.

In both sexes there may develop after death a dark longitudinal band running mid-laterally from the eye to the dorsal fin base, a second band running dorso- laterally approximately along the upper lateral-line, and 6-10 narrow transverse bars across the flanks. In life these markings are rarely discernible.

Amongst females a second type of coloration is known. This takes the form of irregular black blotches on a yellow ground and is identical with the bicolor pattern

ZOOL. 5, 4. 4§§

8o REVISION OF LAKE VICTORIA HAPLOCHROMIS SPECIES

described for certain female Macropleurodus bicolor, Hoplotilapia retrodens, and Haplochromis nigricans (Greenwood, 1956^ and b).

Since collectors show some predilection for fishes with a striking colour pattern, it is difficult to obtain accurate frequency estimates for the bicolor pattern. In the present sample, 25 per cent of females are bicolor. As most specimens were obtained by collectors aware of possible biasing factors, this figure may be accepted as fairly reliable. No male bicolor variants have yet been recorded. Thus, the incidence of bicolor variants seems sufficiently high to recognize the phenomenon as sex-limited polychromatism, and not merely the maintenance of an atypical genotype by re- current mutation. Aberrantly coloured females were found in most localities. None exhibits a pattern intergrading with that usual for females.

Sex-limited polychromatism involving the same phenotypic expression was observed in M. bicolor and Hoplotilapia retrodens (Greenwood, 19560). It seems probable that hypotheses regarding its genie basis and evolutionary significance in these species are also applicable to H. sauvagei. The possible significance of bicolor females as indicating phyletic relationship amongst the various species in which they occur has also been discussed (Greenwood, op. cit.}. It was concluded that, in general, no reliability could be placed on this character, and that its repeated appear- ance was probably attributable to the oligophyletic origin of the Lake Victoria species-flock. Nevertheless, it is suggestive that both H. sauvagei and M. bicolor exhibit " bicolor " polychromatism as well as an apparent similarity in fundamental syncranial morphology.

Colour in preserved material : Adult males. Slate-grey to sooty, the longitudinal and transverse banding often obscured. Spinous dorsal fin grey, soft part hyaline but maculate. Anal and caudal hyaline. Pelvics black on the outer half, hyaline mesially. A dark lachrymal stripe and two bars across the snout are often present. Females and immature males. Ground colour variable, from silver-grey to brownish. Banding, as described above, usually developed. All fins hyaline, the soft dorsal and upper half of the caudal, maculate.

Distribution. Known only from Lake Victoria.

Ecology : Habitat. Restricted to littoral zones where the bottom is hard (sand or shingle) ; the species is especially common over exposed sandy beaches.

Food. The gut contents of forty-five fishes from various localities indicate that H. sauvagei feed mainly on Gastropoda (f.19), bottom deposits, which included insect larvae, Copepoda and diatoms (1.19), and Insecta (chiefly larval boring may-flies, Povilla adusta Navas) (f.4). No fragments of snail shell were observed, although opercula occurred frequently in the stomach and intestine (see also p. 79).

Breeding. Spawning sites and behaviour are unknown. In many localities, sexually active and quiescent fishes, and brooding females occur together.

The smallest adult fish was a female 72 mm. S.L. All specimens over 80 mm. were adult. No difference was detected in the sizes of adult males and females.

Diagnosis. H. sauvagei is distinguished from other Haplochromis in Lake Victoria by combinations of the following characters : lips thickened ; outer teeth with strongly recurved tips ; usually more than three inner rows of teeth in the upper jaw (mode 4). The species closely resembles H. prodromus, from which it may be

REVISION OF LAKE VICTORIA HAPLOCHROMIS SPECIES

81

separated by its slightly thinner lips and smaller adult size. In life, male breeding coloration serves to separate the two species.

Affinities. Similarity in the skull architecture and the dentition of H. sauvagei and H. prodromus suggest a phyletic relationship between the species. Consequent upon these anatomical similarities, the species show a close parallel in their feeding habits and food preferences, although in this respect H. sauvagei may be considered less specialized than H. prodromus.

Study material and distribution records Museum and Reg. No.

Locality.

Uganda

B.M. (N.H.) 1908.5.30.365-366 (as Paratilapia Bunjako

granti)

B.M. (N.H.) 1906.5.30.371-372 (as P. granti) B.M. (N.H.) 1906.5.30.413 „ 1911-3-3-27 •

1909-3-29-9 .... (all as P. bicolor) B.M. (N.H.) 1909.5.11.11

„ 1906.5.30.374-377 ,, ,, 1956.9.17.1 (See Text-fig, i )

1956.10.9.1-25 „ „ „ ,, « 26-30

„ „ „ 31-34 •

Bugonga (Entebbe) Sesse Is.

Bunjako

Entebbe, Airport beach

Collector. Degen. Bayon.

Degen. E.A.F.R.O.

35-36, 2OI 37-40 -

41 42

43

44

45-72

73

74-75

76 77-80

81 82

83

Entebbe, harbour Bugungu (Napoleon

Gulf) Jinja pier Shore opposite Kirinya

Point (Napoleon

Gulf)

Kirinya Point Old Bukakata Katebo

Tanganyika Territory

Mwanza Majita Ukerewe Is. Bukoba

Kenya

Kisumu

Kamaringa (Kavirondo

Gulf) Kach Bay (Kavirondo

Gulf) Open water 5 miles N.

of Kendu (Kavirondo

Gulf) Rusinga Island

82 REVISION OF LAKE VICTORIA HAPLOCHROMIS SPECIES

Haplochromis prodromus Trewavas, 1935 (Text-figs. 2 and 3)

Paratilapia retrodens (part), Boulenger, 1915, Cat. Afr. Fish., 3, 235. Haplochromis ishmaeli (part), Boulenger, 1915, op. cit., 293.

Haplochromis annectens Regan 1922 (nee. Cyrtocara annectens Regan, 1921), Proc. zool. Soc. Lond., 167, fig. 2.

Description. Based on sixty-two specimens (including the holotype), 68-130 mm. S.L. None of the morphometric characters studied shows allometry with standard length.

In its general appearance H. prodromus closely resembles H. sauvagei, from which species it is distinguished by its thicker lips, slightly deeper cheek and larger adult size.

FIG. 2. Haplochromis prodromus, g, holotype (from Regan, the cichlid fishes of Lake Victoria,

Proc. Zool. Soc., 1922, 168, fig. 2).

Depth of body 32-8-40-0 (M = 36.2) ; length of head 29-4-33-6 (M = 31-5) per cent of standard length. Dorsal head profile somewhat variable, but always curved ; strongly decurved in some large individuals, less so in smaller fishes (70-75 mm. S.L.).

Preorbital depth 14-0-19-1 (M = 15-8) per cent head length ; least interorbital width 24-0-31-3 (M = 28-1) per cent. Snout as broad as or slightly broader than long, rarely longer than broad, its length 27-5-36-8 (M = 32-7) per cent of head ; eye diameter 25-8-33.3 (M = 27-8) ; cheek 22-0-30-5 (M = 26-7) per cent.

Caudal peduncle 12-6-18-1 per cent of standard length, its length 1-0-1-7 (mode 1-3) times its depth.

Mouth horizontal ; posterior maxillary tip reaching or almost reaching the vertical to the anterior orbital margin. Lips thickened ; the depth of the upper lip, measured mid-laterally, contained 3-3^ times in eye diameter. Jaws equal anteriorly, or infrequently the lower very slightly shorter ; lower jaw 30-5-37-8 (M = 34-3) per cent of head length, up to 1-3 (mode i-i) times as long as broad.

Gill rakers short, 7-9 on the lower limb of the anterior arch.

REVISION OF LAKE VICTORIA HAPLOCHROMIS SPECIES 83

Scales ctenoid; lateral line with 30 (f.i), 31 (£.7), 32 (f.i6), 33 (f-35) or 34 (f.2) scales ; cheek with 3 or 4 series. 7 or 8 (rarely 6| or 9) scales between origin of dorsal fin and the lateral line, 7 or 8 (less frequently 9) between pectoral and pelvic fin bases.

Fins. Dorsal with 24 (£.7), 25 (£.40) or 26 (£.15) rays, anal with n (£.7), 12 (f.47) or 13 (f.8), comprising XV-XVII, 8-10 and III, 8-10 spinous and branched rays for the fins respectively. Pectoral shorter than the head. Pelvic fins with the first ray produced and of variable posterior extension, but reaching the anal fin in most adult fishes. Caudal sub-truncate.

Lower pharyngeal bone triangular, its dentigerous surface about i| times as broad as long ; pharyngeal teeth slender and cuspidate ; those of the median series some- times enlarged.

Teeth. The dental pattern and tooth form in H . prodromus closely resemble those of H. sauvagei.

In the outer series of both jaws the teeth have strongly recurved tips and are unequally bicuspid or unicuspid. Bicuspid and weakly bicuspid teeth are the pre- dominating forms in fishes less than 100 mm. S.L. Above this size most teeth are unicuspid. 26-56 (mode 40) outer teeth occur in the upper jaw.

Inner teeth are either tri- or unicuspid, the tricuspid form occurring most fre- quently in fishes less than 90 mm. S.L. Antero-medially the teeth are arranged in 3-7 (modes 4 and 5) and 3-6 (modes 3 and 4) series in the upper and lower jaws respectively. The posterior medial margin of the upper tooth-band is straight or slightly curved, that of the lower band is distinctly curved (Text-fig. 3).

The dental pattern of the holotype must be considered aberrant ; it is not re- peated in any of the sixty-one additional specimens. In the type, some postero- lateral inner teeth are displaced medially from their series, thereby giving a spurious impression of a tooth band widened at that point. There is no increase in the width of the underlying premaxillary alveolar surface, nor is there an increase in the number of tooth rows (see fig. 14 in Regan, 1922). In all other respects the dentition of this specimen agrees closely with those described above.

Syncranium and associated musculature. The neurocranium and premaxilla of H. prodromus are virtually identical with those of H. sauvagei. The dentary, however, is relatively more massive and the mental profile is almost vertical.

Likewise, the jaw musculature compares closely with that of H. sauvagei, except that the adductor mandibulae I is somewhat shorter (36-39 per cent head length).

Observations made on the feeding methods of H. prodromus kept in aquaria, indicate that snails are removed from their shells in a manner similar to that employed by Macropleurodus bicolor. That is, the shell is crushed free by the jaws before ingestion takes place. The species was not seen to lever out the soft parts as is usual with H. sauvagei.

Coloration in life : Adult males. Ground colour slatey blue-grey ; a peacock-blue sheen on the belly and ventral flanks. Chest and branchiostegal membrane black, operculum with a golden flush. Very faint indications of a dark mid-lateral stripe and seven transverse bars ; also a faint lachrymal stripe. Dorsal dark, with a deep red flush between both spinous and soft rays ; lappets orange. Anal sooty, ocelli

84 REVISION OF LAKE VICTORIA HAPLOCHROMIS SPECIES

deep yellow. Caudal sooty, with a faint orange flush along its posterior margin. Pelvics black. Females and immature males. Ground colour silver-grey above the mid-lateral stripe and silver below, with a faint peacock-blue flush on the flanks. Transverse barring is indistinct. Dorsal fin dark. Caudal and anal hyaline. Pelvics faintly yellow.

I cm.

FIG. 3. The premaxillary and mandibular tooth bands in H. prodromus.

Colour in preserved material : Adult males. Ground colour dark grey ; in some, faint traces of transverse and longitudinal banding. Chest and branchiostegal mem- brane black. Dorsal, caudal and anal dark, the soft dorsal maculate. Pelvics black. Females and immature males. Pale, banding variable but usually a distinct mid- lateral stripe and a faint, more dorsal band running along the upper lateral line ; five to nine transverse bars across the flank. All fins hyaline.

REVISION OF LAKE VICTORIA HAPLOCHROMIS SPECIES 85

Distribution. Known only from Lake Victoria.

Ecology : Habitat. Restricted to littoral zones, particularly where the substrate is hard (sand or shingle) and occurring less frequently over mud. Thus, the habitat of H. prodromus broadly overlaps that of H. sauvagei. Nevertheless, although biasing factors are introduced by the size selectivity of sampling gear and the limitations imposed by the habitat on the use of certain gear, it seems that H. sauvagei occur most frequently over shallow exposed beaches — where H. prodromus are less common — and that H. prodromus are more abundant in off-shore to deeper waters. This assumption is supported by results obtained when such non-selective collecting methods as explosives were used in both habitats.

Study material and distribution records

Museum and Reg. No. Locality. Collector.

Uganda

B.M. (N.H.) 1907.5.7.78 (holotype H. prodromus) Buddu coast . Simon.

„ „ 1906.5.30.379 (as P. retrodens) . Bunjako . Degen.

B.M. (N.H.) 1956.10.9 84-97 .... Jinja . E.A.F.R.O.

„ ,, ,, ,, ,, 98-99 .... Shore opposite Kirinya . ,,

Point (Napoleon Gulf)

„ „ „ „ „ 100-105 • • • Beach near Nasu Point . „

(Buvuma Channel)

,, ,, ,, ,, 106 .... Pilkington Bay . ,,

„ ,, ,, ,, 107 .... Hannington Bay . „

,, ,, ,, ,, 108-125 • • • Entebbe harbour . ,,

,, ,, ,, ,, 126 .... Katebo . ,,

,, ,, ,, ,, 127-129 . . . BusungweBay (Kagera . ,,

river mouth)

„ „ „ „ „ 130 . . . . Dagusi Island . „

Tanganyika Territory

,, ,, ,, ,, ,, 131-133 . . . Mwanza, Capri Bay . ,,

,, ,, ,, ,, ,, 134-135 . . . Godziba Island . ,,

„ ,, „ „ „ 197-199 . . . Majita

Kenya

,, ,, ,, „ „ 136-137, 196 . . . Kamaringa (Kaviron- .

do Gulf) „ „ ,, ,, „ 138 .... Kisumu . ,,

Food. Stomach and intestinal contents of seventy-four fishes were examined. Of these, eleven were empty, fifty-seven contained only the remains of Gastropoda, three contained Gastropoda and Insecta, and three yielded unidentifiable sludge. Due to their very fragmentary nature the specific identification of molluscan remains was difficult ; where identification was possible the genus Bellamya predominated. As many as twenty-two snail opercula were recorded from the intestine of a single fish, although the modal estimated number of snails per individual was about four.

86 REVISION OF LAKE VICTORIA H APLOCH ROMI S SPECIES

Breeding. Sexually active and quiescent individuals were associated in all localities, but no data were collected on breeding sites or spawning behaviour. Only one female was found carrying larvae in the buccal cavity. There is apparently no sex-correlated adult size difference in this species ; the smallest sexually active individual was a male 102 mm. long.

Diagnosis. The same character complex serves to separate H. sauvagei and H. prodromus from the other Haplochromis of Lake Victoria. H. prodromus is distin- guished from H. sauvagei by its larger adult size, thicker lips, slightly deeper cheek, and, in life, by male breeding coloration.

Affinities. The apparent phyletic relationship between H. prodromus and H. sauvagei on the one hand, and the more specialized Macropleurodus bicolor on the other, has been discussed above and elsewhere (Greenwood 19560). In the latter paper, it was shown that Regan's suggested relationship between H. prodromus and Platytaeniodus degeni Blgr. can no longer be considered valid. Regan's views were based on the type and then unique specimen of H. prodromus whose dental pattern is aberrant. In any case, the posterior widening of the premaxillary dental surface is apparent and not actual in this fish, whereas in P. degeni the premaxilla undergoes a localized but distinct broadening during post-larval ontogeny.

Haplochromis granti Boulenger, 1906 (Text-figs. 4 and 5)

Paratilapia granti (part), Boulenger, 1915, Cat. Afr. Fish., 3, 342, Fig. 231. Haplochromis sauvagei (part), Regan, 1922, Proc. zool. Soc., Lond., 167.

In Regan's revision of the Lake Victoria Cichlidae (ibid., 1922), H. granti was treated as a synonym of H. sauvagei. After comparing the type with other specimens now available, I conclude that the two species should be regarded as distinct. This conclusion is supported by field observations. Both species have in common the " sauvagei " group characters of broad inner tooth bands, outer teeth with strongly recurved tips, and thickened lips. But they differ considerably in gross morphology and in certain details of dental pattern. The holotype of H. granti (figured in Boulenger, 1915) does not present a specifically typical appearance. However, its dental pattern indicates conspecificity with the specimens here described as H. granti. Furthermore, in the type, characters which contribute to gross morphology, for instance the form of the dentary and the head shape, intergrade with those of other specimens possessing a more typical facies.

One rather damaged specimen (B.M. (N.H.) Reg. No. 1911.3.3.28), identified by Boulenger as Paratilapia retrodens and later by Regan as H. sauvagei, should probably be referred to H. granti. Because of this uncertainty P. retrodens is not included in the revised synonomy of H. granti.

Description. — Based on the type, two paratypes and twenty-six additional speci- mens in the size range 70-122 mm. S.L. No clear-cut allometry with standard length was observed in the morphometric characters listed below.

Depth of body 32-7-39-3 (M = 35-4) ; length of head 28-8-33-3 (M = 31-5) per

REVISION OF LAKE VICTORIA HAPLOCHROMIS SPECIES 87

cent of standard length. Dorsal head profile slightly curved, or, less frequently, straight and gently to steeply sloping.

Preorbital depth 15-3-19-0 (13-3 in the smallest specimen) (M = 17-1) per cent of head length ; least interorbital width 25-0-32-8 (M = 28-6) per cent. Snout as broad as or slightly broader than long, its length 29-0-36-0 (M = 31-6) per cent of head ; eye diameter 25-0-31-0 (M = 27-5) ; depth of cheek 22-0-30-6 (M = 26-8) per cent.

Caudal peduncle 13-6-19-0 per cent of standard length, 1-2-1-7 times as long as deep.

FIG. 4. Haplochromis granti, $, B.M. (N.H.) 1956.9.17.2. Drawn by Miss L. Buswell.

Mouth usually somewhat oblique ; posterior maxillary tip almost reaching the vertical to the anterior orbital margin, or occasionally reaching this line. Lips thick, sub-equally developed in a few specimens (e.g. the type), but the upper lip clearly thicker than the lower in most. Jaws equal anteriorly, or the lower jaw slightly projecting, its length 22-2-30-6 (M = 26-8) per cent of head length and 1-0-1-5 (mode 1-3) times its width.

The oblique mouth and unequally thickened lips give an appearance of deformity to many specimens. This impression is apparently misleading since there is no indication of any impairment to the efficiency of the jaw mechanism, either as a mechanical unit or in relation to feeding habits.

Gill-rakers short, 7-9 on the lower limb of the first arch.

Scales ctenoid, lateral line with 32 (1.9), 33 (f.n), or 34 (f.g) scales ; cheek with 3 or 4 (in one specimen 2) series ; 7 or 8 scales between origin of dorsal fin and lateral line ; 7 or 8 (rarely 9) between pectoral and pelvic fin bases.

88 REVISION OF LAKE VICTORIA HAPLOCHROMIS SPECIES

Fins. Dorsal with 25 (f.n), 26 (1.17) or 27 (f.i) rays ; anal n (f.y) or 12 (f.2i), comprising XV-XVII, 9 or 10 and III, 8 or 9 spinous and branched rays for the fins respectively. In one specimen the anal fin had been damaged and subsequently healed irregularly, giving II, 10 rays. Pectoral shorter than the head, except in two specimens where it is of the same length. Pelvic fins extending to the vent in im- mature fishes and to the anal fin in adults ; the first ray is proportionately more produced in sexually active males. Caudal fin truncate or sub-truncate.

Lower pharyngeal bone triangular, its dentigerous surface iy-ij times as broad as long ; pharyngeal teeth similar to those of H. prodromus.

Teeth. In the outer series of both jaws, the teeth are similar to those of H. prodromus and H. sauvagei ; that is, unicuspid with strongly recurved tips. A few specimens- all below 90 mm. S.L. — have some bicuspid teeth situated postero-laterally in both jaws. There are 28-46 (mode, ill defined : 36) outer teeth in the upper jaw.

I cm.

FIG. 5. Mandibular tooth band in H. granti.

The inner series are composed of tricuspid teeth in most fishes below 90 mm. and of unicuspid teeth in larger specimens. An admixture of both types is known from three fishes. The teeth are arranged in 2-6 (mode 4) rows in both jaws, but narrow to single series laterally. In many specimens the lower tooth band is wider than the upper ; antero-medially, the posterior margin of this band is straight or very gently curved, thus contrasting with the lower series in H. sauvagei and H. prodromus, where the margin is clearly curved (Text-fig. 5).

Syncranium and associated musculature. The preorbital face of the neurocranium is intermediate in form between that of H. sauvagei and the generalized Haplo- chromis type. Greatest departure from the condition observed in H. sauvagei and H. prodromus is seen in the maxilla, which in H. granti is shorter and more bowed in its long axis. Also, the inner face of the posterior limb is markedly concave, which results in the outer face appearing more bullate than in other members of the " sauvagei " group. The dentary resembles that of H. sauvagei but differs in its less rounded, more angular, anterior outline.

REVISION OF LAKE VICTORIA HAPLOCHROMIS SPECIES 89

Shortage of material allowed only two dissections of head musculature to be made. The major muscles are distributed as in H. prodromus and H. sauvagei but the origin of the adductor mandibulae I is deeper and more fan-shaped in H. granti. In the two specimens dissected, the length of this muscle (42 and 43 per cent of head) is somewhat greater than in H. prodromus but equal to that in H . sauvagei.

Coloration in life : Adult males. Ground colour light blue-grey ; branchiostegal membrane dusky, especially between the rami of the lower jaw. Dorsal fin blue- grey, darkest on the proximal third ; lappets orange-red, as are the spots and streaks between the soft rays. Caudal blue-grey, darker on the proximal half ; margin out- lined in red ; orange-red spots between the rays. Anal dusky blue-grey, with an overall pink flush ; ocelli yellow. Pelvics black, faint pink mesially. Females and immature males. Coloration in life unknown.

Colour in preserved material : Adult males. Ground colour grey or brown ; branchiostegal membrane dark grey. An intense black mid-lateral stripe and often traces of a lachrymal stripe and 5-7 vertical bars across the flanks. Dorsal, caudal and anal fins hyaline or dusky ; pelvics black. Females and immature males. Ground colour silver-white, darkest dorsally. An intense mid-lateral stripe and often faint indications of an interrupted upper band running between the dorsal fin base and the upper lateral line. Seven to nine faint transverse bars are usually present on the flanks and caudal peduncle ; no lachrymal stripe. All fins hyaline.

Distribution. Confined to Lake Victoria.

Ecology : Habitat. Too few records are available to permit generalization on the habitat preferences of H. granti. The twenty-six specimens whose habitat had been recorded were caught in littoral zones and in water less than forty feet deep. Most localities represented in the collection can be classified either as sandy beaches on exposed shores or as exposed coastlines with a hard substrate. The few remaining localities are sheltered bays where the bottom is composed of organic mud.

Food. Twelve of the twenty-six fishes examined contained ingested material in the stomach or intestine. In each case only the soft parts of Gastropoda were found, except for some Lamellibranchiata shell fragments in one individual. From these admittedly few observations it is inferred that H. granti feed in a manner similar to that observed for H. prodromus and H. sauvagei.

Breeding. There is no information on any aspect of the breeding behaviour in this species ; all specimens below 90 mm. S.L. were immature.

Diagnosis. Haplochromis granti differs from other Lake Victoria Haplochromis in possessing broad bands of inner teeth (2-6, mode 4, series) in both jaws and by its unequally thickened lips, the upper usually thicker than the lower. This latter character, together with the oblique mouth and straight posterior margin to the inner tooth band of the lower jaw, serves to distinguish H. granti from H. prodromus and H. sauvagei.

Affinities. By virtue of its dentition, H. granti must be included in the H. sauvagei- H. prodromus species-group. Other characters probably associated with dentition, such as the shape of the premaxilla and dentary, are closely similar in all three species. But, despite resemblances in these dental and osteological characters, and in the associated musculature, the neurocranial morphology of H. granti has not

REVISION OF LAKE VICTORIA HAPLOCHROMIS SPECIES

departed so radically from the generalized Haplochromis type. Morphologically speaking, the relationship between H. sauvagei and H. prodromus is directly linear, whilst that of H. granti is somewhat divergent but with a parallel trophic trend.

Study material and distribution records

Museum and Reg. No.

Uganda

B.M. (N.H.) 1903.5.30.367 (holotype P. granti) .

B.M. (N.H.) 1903.5.30.368-369 (paratypes P.

granti) ........

B.M. (N.H.) 1956.10.9.139 ....

,, ,, ,, ,, ,, 140-141 „ „ ,, 142-144

E. African Fisheries Res. Lab. Jinja B.M. (N.H.) 1956.10.9.145-147

1956.9.17-2 • ,, ,, 1956.10.9.148-152

„ „ „ 153 „ „ „ 154

155-157 200

158

159 160-163

Kenya

Locality.

Bunjako

Bay opposite Kirinya

Point (Napoleon

Gulf) Bugungu (Napoleon

Gulf) Beach nr. Nasu Point

(Buvuma Channel) Pilkington Bay Thruston Bay Ekunu Bay Entebbe, harbour Near Busungwe Is. Busungwe Bay (Kagera

river mouth) Beach near Grant Bay Buka Bay

Collector. Degen. E.A.F.R.O.

Kisumu

Tanganyika

Ukerewe Is. Majita

Haplochromis xenognathus sp. nov. (Text-figs. 6 and 7)

The high intra-specific variability of H. xenognathus makes this species of par- ticular interest when considering the evolution of monotypic cichlid genera. Some of the more aberrant specimens, if studied in isolation, might well be given a status equal with the monotypic genera recognized at present. Less extreme individuals, on the other hand are not immediately distinguishable from H. sauvagei.

The modal type tooth-pattern and the usual arrangement of the jaws are, however, unlike those of other species in the " sauvagei " group (Text fig. 7). I am led to include H. xenognathus in this group because of its " sauvagei "-like tooth form and the multiseriate dental pattern.

The sample provides sufficient intra-specific variation to indicate morphological

REVISION OF LAKE VICTORIA HAPLOCHROMIS SPECIES 91

stages through which the typical specific facies may have passed in its evolution from a form similar to the extant H. sauvagei.

Type specimen. A male, 91 + 19 mm. ; from Entebbe harbour.

Description. Based on thirty-five specimens 80-113 mm. S.L.

Depth of body 31-2-38-0 (M = 34-8) ; length of head 29-2-35-4 (M = 33-1) per cent of standard length. Dorsal head profile usually straight and somewhat steeply sloping ; curved in a few specimens.

Preorbital depth 16-0-20-7 (M == 17-7) per cent head length ; least interorbital width 23-5-29-0 (M = 26-8) per cent. Snout from i^-i-j longer than broad, its length 31-8-37-8 (M =35.2) per cent of head ; eye 23-2-28-7 (M = 26-0) ; depth of cheek 23-2-28-7 (M = 26-0) per cent.

FIG. 6. Haplochromis xenognathus, <J, holotype, B.M. (N.H.) 1956.9.17.3. Drawn by Miss L. Buswell.

Caudal peduncle 13-8-19-0 (M = 15-9) per cent of standard length ; 1-1-1-7 (mode 1-4) times as long as deep.

Mouth horizontal, the posterior maxillary tip reaching, or almost reaching, the vertical to the anterior orbital margin ; lips slightly thickened. The lower jar is clearly shorter than the upper in 74 per cent of the specimens examined and sub- equal to the upper in 26 per cent. Even in this latter group the outermost teeth of the lower jaw occlude behind the equivalent upper jaw series. Lower jaw 32-0-38-0 (M = 34-5) per cent of head, and 1-1-1-8 (mode 1-4) times as long as broad.

Gill rakers short, 7-9 (rarely 10) on the lower limb of the first arch.

Scales ctenoid ; lateral line with 31 (£.4), 32 (£.14), 33 (f-11)* 34 (f-4) or 35 (£.2) scales ; cheek with 3 or 4 (rarely 2 or 5) series ; 6 or 7 (less frequently 8) scales between origin of dorsal fin and lateral line ; 7 or 8 (rarely 6 or 9) between pectoral and pelvic fin bases.

92 REVISION OF LAKE VICTORIA HAPLOCHROMIS SPECIES

Fins. Dorsal with 24 (£.13), 25 (f.ig), or 26 (£.3) rays ; anal with n (f.g), 12 (f.24) or 13 (f.i). One specimen has only two spines, giving a total count of 10 rays. The spinous and branched ray counts for the fins are XV-XVII, 8-10 and III, 8-10. Pectoral fins shorter than the head. Pelvic fins with first ray produced, extending to the anterior part of the anal fin in females and more posteriorly in adult males. Caudal truncate.

Lower pharyngeal bone triangular, its dentigerous surface i^-ij times as broad as long. The pharyngeal teeth are slender and bicuspid, those of the median series not noticeably enlarged. In one specimen, the lower pharyngeal bone is stout, the toothed surface slightly longer than broad and the median teeth enlarged and molariform ; this fish also shows a somewhat atypical oral dentition, in that the teeth are bluntly cuspidate.

Teeth. Except in the smallest specimen, unicuspid teeth predominate in the outer series, but some weakly bicuspid teeth do occur postero-laterally in both jaws of large fishes. In the smallest specimen, the entire outermost series is composed of bicuspid teeth.

The outer teeth, like those of H. sauvagei and other species of the group, are relatively stout and have strongly recurved tips. In the lower jaw, the anterior teeth are implanted at an acute angle, so that their necks lie almost horizontally ; but recurvature of the crown is such that the tip points almost vertically upwards (Text-fig. 8). There are from 32-52 (mode 44) outer teeth in the upper jaw.

Teeth forming the inner series are small and either unicuspid or weakly tricuspid. Considerable variation exists in the number of inner rows. In general, teeth are disposed in a broad crescent which narrows abruptly at a point almost mid-way along the premaxillary limb. Thereafter, there is a single inner row. The antero- medial depth of this band varies with the number of tooth rows, of which there are from 3-9 in both jaws (modes 7 and 5 for the upper and lower jaws respectively). In fishes with markedly disparate jaws, the most posterior inner teeth of the dentary do not occlude with the upper series.

The toothed surface of the dentary is often slightly convex, so that when viewed laterally several points on the inner band are higher than the crowns of the outer teeth (Text-fig. 8).

Fishes with narrow tooth bands in both jaws have a dental pattern closely resem- bling that of H. sauvagei ; the resemblance to this species is enhanced by the sub- equal jaws of these specimens. In contradistinction, other H. xenognathus with sub-equal jaws have a broad and specifically typical tooth pattern.

Syncranium and associated musculature. The neurocranium of H. xenognathus is identical with that of H. granti. The shape of the premaxilla varies slightly in relation to the number of inner rooth rows, but is otherwise comparable with the premaxilla of H. sauvagei. Likewise the dentary is similar in the two species, except that the dentigerous surface is inclined forwards and downwards in H. xenognathus. Perhaps the most characteristic appearance of this bone is imparted by the almost hori- zontally implanted anterior teeth.

A syncranial skeleton prepared from a specimen with H. sauvagei-like facies did

REVISION OF LAKE VICTORIA HAPLOCHROMIS SPECIES 93

O'Scm.

FIG. 7. Premaxillary and mandibular tooth bands in H. xenognathus.

I cm.

FIG. 8. Lateral view of the anterior part of the dentary of H. xenognathus.

94 REVISION OF LAKE VICTORIA HAPLOCHROMIS SPECIES

not differ, beyond the limits of individual variability, from that of a typical specimen and was clearly distinguishable from that of H. sauvagei.

Head musculature in H. xenognathus is similar to that of H. sauvagei and H. prodromus. The adductor mandibulae I is shorter than in the former species, but is equal to that of the latter (33 • 2-39-4 per cent of head) .

Coloration in life : Adult males. Ground colour dark bronze dorsally, shading to grey-bronze ventrally : cheek and operculum with a distinct bronze sheen ; chest and branchiostegal membrane bluish-grey. Dorsal fin sooty, with red streaks between the soft rays ; lappets red. Caudal dark, upper half with red spots, lower half flushed with red. Anal clouded ; ocelli yellow. Ventrals sooty, the first ray bluish-white. Live coloration of immature males is unknown. Females. Ground colour as in males, but the branchiostegal membrane greyish. Dorsal fin with red lappets but lacking the red streaks. Caudal dark yellow ventrally, lighter and maculate above. Anal and ventral fins dark olive-yellow, the first pelvic ray bluish-white.

Colour in preserved material : Breeding males. Dark grey, the flank with a faint coppery sheen ; a distinct lachrymal stripe. In some specimens five transverse bars may be discerned on the flanks. Dorsal, caudal and anal fins grey, the upper third of the caudal maculate. Pelvic fins black. Non-breeding and immature males. Ground colour silver-grey ; 7-9 distinct transverse bars. Fins as above. Females. As for non-breeding males, except that the pelvics are colourless. In some specimens there is a fairly distinct mid-lateral stripe.

Distribution. Known only from Lake Victoria.

Ecology : Habitat. The few and scattered records indicate