Fish 1-UrV ' ' z>i LAKE TURMN/1 A REPORT ON THE FINDINGS OF THE LAKE TURKANA PROJECT 1972-1975 GOVERNMENT OF KENYA AND THE MINISTRY OF OVERSEAS DEVELOPMENT, LONDON. EDITED BY A .J. HOPSON Volume 4 of 6 Volumes M/V- , * NAT, Hist. A 6 MAY 1983 } PURCHASED * OVERSEAS DEVELOPMENT ADMINISTRATION LONDON 1982 Published 1982 Institute of Aquaculture University ol Stirling STIRLING FK9 4LA Scotland Copyright (C) 1982. Crown Copyright Reserved. ISBN 0 - 901636 - 41 - X Produced, Printed and Bound at the University ot Stirling CONTENTS Page Forward i Preface lii Acknowledgements vi Chapter 1 Geographical, physical and chemical aspects ol Lake Turkana. 1 2 Studies on algal dynamics and primary productivity in Lake Turkana 109 3 Studies on the zooplankton of Lake Turkana 163 4 A study of the prawns Caridina nilotica (Roux) and Macrobrachiurr l niloticum (Roux) in Lake Turkana 247 5 The fishes oi Lake Turkana 281 6 The commercial fishery of Lake Turkana 349 7 Shore survey 557 8 Openwater surveys 581 9 The biology of cypnniiorm fishes 751 10 The biology ot silurirorm fishes 1017 11 The biology of perciform fishes 1283 12 The food of zooplanktivorous fishes 1503 13 Conclusions and recommendations 1563 Main Appendix 1 An annotated checklist oi Invertebrates recorded in Lake Turkana 1589 Main AppendLx 2 Specification of the welded steel research vessel “Halcyon” 1603 Index 1605 VOLUME 4 Chapter 10 The Biology of Siluriform Fishes ill CHAPTER lO THE BIOLOGY OF SILURIFORM FISHES by J.M.Lock 1017 CONTENTS INTRODUCTION MATERIAL AND METHODS SAMPLING SECTORS BAGRIDAE Bagrus bayad Distribution Depth distribution Seasonal fluctuations in trawl catches Conclusions Age and Growth Methods Resul ts Mathematical expression of growth Length-weight relationship Discussion Breeding Maturity stages Size and age at maturity Seasonal changes in maturity Sex-ratio Site of spawning grounds Stimulation of breeding Condition Fecundity Discussion Food Method Food of juvenile fish Food of immature and adult fish Diet and predator size Diet with relation to depth Predator-prey size relationship Diel feeding pattern Seasonal feeding pattern Discussion Commercial fisheries Gillnets Longlines Trends in the commercial catch Fishing pressure in the Central Sector Bagrus docmac Chrysichthys auratus Distribution Breeding Feeding Fisheries 1018 Auchenoglanis occidentalis CLARIIDAE AND MALAPTERURIDAE Clarias lazera Malapterurus electricus SCHILBEIDAE Schilbe uranoscopus Distri bution Trawl catches Frame trawling Gillnetting Beach seines Discussion Growth Sexual differences in length frequency distribution Ana-lysis of length frequency data Mathematical expression of growth Growth rates in the Northern and Southern Sector Length-weight relationship Conclusions Breeding Maturity stages Sex-ratio Size and age at maturity Location and time of breeding Food Composition of food in relation to depth Diet with relation to size Diel feeding pattern Conclusions Commercial fishery MOCHOKIDAE Synodontis schall Distribution Bottom trawling Gillnetting Frame trawling Beach seining Conclusions Growth Me thod Sexual differences in length frequency distribution Seasonal changes in length frequency distribution 1019 Estimates of growth rate Mathematical expression of growth Age composition of trawl catches Length-weight relationship Discussion Breedi ng Maturity stages Size at maturity Seasonal changes in maturity Central sector Northern sector Southern sector Spawning sites Stimulation of breeding Sex-ratios Condition Fecundi ty Embryological and larval development Introduction Methods Development of embryos and larvae Egg stage Development of the embryo Larval development Discussion Food Method Food categories Food in relation to depth Diet with relation to size Diel feeding rhythms Discussion The fishery for Synodontis schall Synodontis frontosus Mochocus niloticus SUMMARY REFERENCES APPENDIX 111. 1021 10 • THE BIOLOGY OF SILURIFORM FISHES IN LAKE TURKANA J M Lock INTRODUCTION Catfish belonging to the suborder Siluroidea are widely distributed in Africa and are of importance in commercial and subsistence fisher- ies. Siluriform fish are readily distinguishable by a complete lack of true scales. They may also have complicated anterior spines on the dorsal and pectoral fins which are lockable into position and which frequently bear serrations or barbs. Representatives of this group are extremely common in Lake Turkana and form an important part of the ichthyofauna. A total of ten species were identified from the region of the lake during the present survey (Table 10.1). Three of these, Auchenoglanis occidentalis , Synodontis f rontosus and Mochocus niloticus were confined to the northern fringes of the area in the delta of the River Omo , while the remaining seven species occurred within the actual lake . Two further fishes Heterobranchus longifilis (C and V) and Andersonia leptura Boul . are known to occur in the region of the Omo delta. The former, which was collected by Zaphiro and Macmillan from the middle Omo, (BOULENGER, 1911) was almost certainly the fish known as Labe to Turkana fishermen and marketed in small quantities at the Todenyang Co-operative. Unfortunately no recognisable specimen was ever available for critical examination during the present study. Andersonia leptura, a small and easily overlooked species, was recorded from the Omo delta by PELLEGRIN (1935). Despite the widespread distribution of silurids in Africa rela- tively little is known about their biology. Of these species inhabit- ing Lake Turkana only Synodontis schall and Bagrus docmac have been studied elsewhere. BISHAI and GIDEIRI (1965 a and b, 1968) provided details of the growth, feeding and reproduction of Synodontis schall in the River Nile at Khartoum and WILLOUGHBY (1974) has examined the population inhabiting Lake Kainji, Nigeria. Bagrus docmac in Lake Victoria has been studied by a number of workers who give valuable information on its feeding and breeding (CORBET, 1955, 1958, 1960, 1961; CHILVERS, 1968; CHILVERS and GEE; ELDER, 1960, 1962, 1974). Little published work is available on the remaining species, although a few fish from the same genera have been studied in other parts of Africa. TWEDDLE (1975) aged Bagrus meridionalis Gunther from Lake Malawi. VAN DER WAAL (1974) has examined the breeding habits of Clarias gariepinus (Burchell) in South Africa and VAN DER WAAL and SCH00NBEE (1975) its age and growth. MUNRO (1967) gives details of the diet. FRANK (1974) has aged Synodontis nebulosus Peters, and Schilbe mystus (Linn) from Lake Kariba, and WHITEHEAD (1959) has studied the migratory behaviour of the latter fish in Lake Victoria. Previous studies of the Lake Turkana fish populations were carried out by WORTHINGTON and RICARDO (1936) and by' MANN (1964). Their observa- tions include information on the biology and distribution of several of the catfish species. 1022 The aims of the present study were as follows: 1. To complete an inventory of siluriform fish occurring in Lake Turkana . 2. To investigate the biology of species likely. to be of importance in a commercial fishery. 3. To determine the most appropriate means. of catching such species commercially . This study is based on field work carried out between May 1972 and July 1974. Of the six siluriform species occurring regularly in the actual lake, Bagrus bayad and Synodontis schall are the most com- mon. Details of their biology, including growth, breeding and diet were collected during the course of the field work and these results, together with information relevant to their commercial exploitation, are discussed at length in the following account. Information on the less important but widespread species, Schilbe uranoscopus and Chrysichthys auratus is also given. Clarias lazera and Bagrus docmac were less commonly caught and all available information is presented. It has proved impossible to obtain biological data on the remaining four species which were encountered only rarely during the course of the field work. Distributional records and other data concerning these species are included in the study where available. A detailed account of the physical and chemical environment is provided in Chapter 1. Rainfall is considered to be particularly im- portant in the breeding biology of the siluriform species under review. Additional data on monthly precipitation patterns at Kalokol, which is of special relevance to the present account, are presented in Table 10.2. MATERIAL AND METHODS A high proportion of the material on which the present study is based was caught by bottom trawling. Data from all areas of the lake have been incorporated but the most detailed information is derived from regular monthly samples made between May 1972 and July 1974 in the Longech area at depths ranging from 10 to 80 m. In the case of Siluri- form fishes bottom trawl catch per effort has been expressed in terms of the catching power of the "Victoria" trawl and not the 3-bridled trawl as has been the practice elsewhere in the report. A factor of 1.43 may be employed to convert values in the present chapter to the 3-bridled trawl standard (see Chapter 8, page 587 , Fig. 8.1). Details of the vertical zonation and diel migrations of siluriform fishes were investigated with various types of pelagic gear including metre townets, midwater trawls and particularly frame trawls (see Chapter 8). Bottom, midwater and surface-set gillnets also yielded valuable data (see Chapter 8, page 620 , Tables 8.23-8.25). Long- lining, which was undertaken in the deeper waters of the lake, provided supplementary information on Bagrus bayad . Relatively large hooks (No 6) were used with the snoods attached to the line at intervals of 11 m. Details of length, weight and gonad condition were recorded wher- ever possible on freshly caught fish. Fork lengths have been used throughout this study except for two species with rounded caudal fins, 1023 Clarias lazera and Malapterurus electricus. All stomachs, except those from larger Bagrus bayad were preserved in 5% formalin for subsequent examination in the laboratory. Juveniles were similarly treated. SAMPLING SECTORS To simplify the analysis of data collected in the field, Lake Turkana was divided into three sectors (Fig. 10.1) covering the following approximate areas (km“): North 3548 Central 1497 South 2515 The Central sector covered the smallest area but it was the most thoroughly investigated due to its close proximity to the research centre. Regular monthly bottom trawl samples were made in the Central sector enabling changes in gonad condition, fish numbers and diet to be monitored. The Northern sector included the Omo delta area where depths did not exceed 40 m and Allia Bay with extensive stretches of shallow water. The Southern sector comprised the Southern basin and the Rerio-Turkwell area. Much of the former consisted of water con- siderably deeper than 40 m, whilst the latter was generally shallower and lay within the influence of the Kerio and Turkwell rivers. BAGRIDAE BAGRUS BAYAD Bagrus bayad is widely distributed across the soudanian region of Africa, occurring in the Niger, Senegal, Volta and Cha basins. It also extends through much of the Nile system from the delta to Lake Mobutu Sese Seico (formerly Lake Albert) (GREENWOOD, 1966). Observations on the biology of this fish prior to the present study were limited to notes on the Lake Turkana population made by WORTHINGTON and RICARDO (1936) and MANN (1964). Since B. bayad is primarily a deep-water fish in this particular lake, numbers examined by previous workers were small and few details could be given. The recent introduction of long- lines in Lake Turkana has brought about a sharp rise in the catch of B_. bayad and an account of its biology is clearly needed to assist in fisheries management. DISTRIBUTION Bagrus bayad which grows to a maximum fork length of 105 cm is essen- tially a demersal fish. Bottom trawl catches prove that it is widely, though unevenly, distributed throughout Lake Turkana. Catches in mid- water were uniformly low, although results suggest that B. bayad moves off the bottom to feed at night. Details of the distribution of _B. bayad are based mainly on data obtained during the trawl survey of Lake Turkana. Catch per effort, expressed as kilograms per hour is considered to be an index of rela- tive abundance. The pattern of distribution has been investigated in relation to both lake depth and season. Changes in habitat correlated with increasing size have also been studied. Gill net and frame trawl data have provided additional data. 1024 Depth distribution The most important factor governing B. bayad distribution is depth. Table 10.3 gives mean bottom trawl catch per effort in nine depth zones for the Northern, Central and Southern sectors of the lake (see also Fig. 10.2). Catches increased steadily from 7 kg/hr in water of less than 10 m depth to 102 kg/hr in the deepest areas of the lake. Data analysed separately for the various areas, however, show some diver- gence from this general pattern. In the Northern sector low densities of _B. bayad were recorded in the Allia Bay area with less than 5 kg/hr in over 50% of the hauls. Elsewhere in the Northern sector catches were similar to the lakewide means, although between North Island and the Omo delta exceptionally high catches were sometimes recorded. The Central sector, for which most data is available, can be examined in greater detail. The mean catch rate (Table 10.3) shows the lakewide trend of increasing catches with depth. However, wide seasonal fluctuations in inshore areas were apparent and will be con- sidered below. In the Southern sector catches were generally lower than else- where. Above average catches were however recorded in the vicinity of the Kerio and Turkwell rivers. Also exceptional were the deeper waters of the Southern basin where catches below the 70 m contour were consistently high. To examine depth distribution in fish of different sizes, catch rates (numbers per hour) were analysed for eight length groups (Table 10.4 and Fig. 10.3). With the exception of the smallest fish, catch increased with depth indicating that higher population densities occur in deeper water. The 0-9 cm length group does not follow the general pattern and peak numbers of these small fish occurred in water of be- tween 30 and 40 m. Breeding in B. bayad is thought to occur in shallow water and the results in Figure 10.3 indicate that juvenile _B. bayad tend to remain near the spawning sites. It was observed that juvenile _B. bayad below 3 cm FL feed largely on copepods which are more common in shallow water. Seasonal fluctuations in trawl catches Seasonal fluctuations of catch in the Ferguson's Gulf area within the 40 m contour have been analysed in Figure 10.4. Above average catches were recorded between August and November, followed by a decline to a minimum in April. Peak catches coincide with the seasonal influx of turbid Omo flood water into the Central sector of the lake and the movement of B. bayad into shallower water is perhaps associated with this change in the water climate. One possible reason for the inshore migration is examined in detail in the section on feeding. Briefly the presence of turbid water may result in increased numbers of prawns inshore, either a result of their migration from deeper water or in- creased abundance in response to greater primary productivity. Since prawns form an important part of the diet of B. bayad the movement of this fish into shallower water is perhaps a feeding migration. An analysis of the stomach contents of B. bayad caught in shallower water has shown that from September to November the proportion of prawns in the diet is greater than at any other time of the year. 1025 There is also the possibility that the minimum depth at which B. bayad occurs during daylight hours is determined by light intensity. Increased turbidity during the flood period would result in a decrease in light penetration and the point at which B. bayad tends to avoid critically high light intensities would shift nearer the surface and closer inshore. The result would be a movement of B. bayad into shallower water such as was observed. Fluctuations in bottom trawl catches have also been noted in the Southern sector in the vicinity of the Kerio and Turkwell rivers, and in the Northern sector between the Omo river and North Island. Unfor- tunately regular sampling in these two areas was not possible but the higher catches appear also to be associated with the presence of turbid water. In the Rerio-Turkwell area catches equivalent to twice the annual mean were recorded in October 1973 in water of between 20 and 30 m when the two rivers were flowing and turbid water was present. Similarly in the north catches during the peak flood period were three times the annual mean in water between 10 and 20 m. There is no evi- dence that the concentration of B. bayad in the vicinity of flooding major rivers is associated with large scale breeding migrations. Diurnal changes in distribution are considered below (page 1038) in the section on feeding. Conclusions Bagrus bayad is a demersal fish. Bottom trawl catches show that densities increase progressively from inshore areas to deep water. The densities of B. bayad in shallow water however appear to depend largely on water turbidity. High turbidity often results in increased catch rates. In areas such as Allia Bay, which has relatively long periods each year free from turbid water, low catches of B. bayad are recorded. The inshore movement of B. bayad during the flood season is probably a response to reduced light penetration, and to seasonally high densities of prawns in shallow water. AGE AND GROWTH In the tropics the ageing of fish poses many problems. Scales and skeletal structures may show rings similar to those laid down by temperate fish but they cannot be assumed to be annual until the seasonal timing of their formation has been demonstrated. Growth checks may be formed for a number of reasons. External factors include adverse conditions in the dry season (DAGET, 1952; J0HNELS, 1952; LOWE, 1964), fluctuations in food supply (BISHAL and GIDIERI, 1965a) and temperature change (HOPSON, 1965) • Checks may also be associated with internal factors such as spawning (TWEDDLE, 1975) or loss in con- dition (HOLDEN, 1955; GARR0D, 1959). In some species checks may be laid down twice a year as a result of both external and internal factors (HOPSON, J, 1972). In many tropical species however no regular pattern or sequence of check formation occurs. Lake Turkana has a remarkably stable tem- perature regime with an annual variation of less than 2 C in deeper water, and temperature changes are unlikely to result in growth checks. Heavy flooding resulting in widespread hydrological changes is a regu- lar seasonal event in Lake Turkana but there is no evidence to show that it results in an interruption of the growth pattern of B. bayad. 1026 A second method of ageing fish which can sometimes be applied in the tropics is the interpretation of length frequency distributions (TESCH, 1968). Its application is limited to fish with breeding re- stricted to a short period of time each year. Such species are not uncommon in the tropics, especially in arid regions where spawning is frequently associated with seasonal rainfall. Because of the rapid growth of tropical fish, there is often little overlap in the size of individuals in adjacent year classes and this method of ageing may be particularly useful. Vertebrae have been successfully used by other workers in ageing various African silurids. BISHAI and GIDEIRI (1965a) and WILLOUGHBY (1974) aged several species of Synodontis using this method and TWEDDLE (1975) was similarly successful with Bagrus meridionalis . In B. bayad however no correlations between check formation and fish age could be established. Fortunately breeding was found to be confined to approxi- mately 4 months each year and an inspection of length frequency data revealed that Petersen's method for age and growth determination could be used. Methods Large samples of fish caught at regular intervals of time are essential for satisfactory estimates of growth to be made from length frequency data. All the B. bayad used in the present age analyses were caught by bottom trawling during the regular monthly programme. Data for length frequency analysis is ideally obtained by non-selective gear which samples the entire length range of the fish. Most fishing gear, however, is selective. The bottom trawl fitted with the standard cod end of 38 mm mesh did not catch fish less than approximately 10 cm. Smaller fish were obtained periodically by means of a bag of 19 mm mesh laced onto the cod end of the main net. Although fish of up to 85 cm were caught by the bottom trwl, the proportion of fish over 45 cm was low. It was not possible to determine whether this dearth of larger fish in the samples was the result of escapement. Altogether more than 16,000 B. bayad were measured between May 1972 and July 1974 of which 11,000 were caught in the Central sector of the lake. Wherever possible fish were sexed so that variations in growth rate between males and females could be analysed. Results Length frequencies of all B. bayad caught in the Longech area during January, February and March 1973 are presented in Figure 10.5. In each month two peaks are prominent and the progression of the two modes from January to March gives a rough estimate of the monthly growth rates. For a more accurate growth estimate it was necessary to examine all the length-frequency data collected between May 1972 and July 1974 and also to establish whether there was a significant difference in length distribution between the sexes. Figure 10.6 compares the length frequencies of males and females caught in March and July 1973 in the Ferguson's Gulf area. The histograms have been smoothed (moving aver- age a =3), using the method of TAYLOR (1965), and no great difference between the sexes is apparent. Means and variances were estimated for each of the size groups and females were compared with males using Students t-test (see Table 10.5). There was no significant difference ( P = < 0.05) between the sexes in either size groups and further 1027 analysis of other trawl samples showed that this was typical for the population. Data for the two sexes was therefore combined for growth analysis . Percentage length-frequency histograms of all _B. bayad caught at the monthly bottom trawl stations in the Longech area, between Sep- tember 1972 and May 1974, are presented in Figure 10.7. The 1972 and 1973 year class peaks have been shaded and their progression can be traced from month to month. The 1972 year clss first appeared in the trawl catches in September 1972, and could be followed through sub- sequent months until the completion of sampling. Fish from the 1973 year class were first caught by the trawl in August 1973 and August to September is evidently the annual period when 0-group fish are re- cruited into the trawlable stock. The 1971 year class was prominent in trawl catches until November 1973 when a mean length of approximately 39 cm was attained. At lengths above 40 cm the division of the population into definite length groups was no longer clear and age determination of fish exceeding 3 years old was not possible. The extent of each year class on the monthly length frequency histograms (Fig. 10) was determined by careful comparison with results from adjacent months. Mean monthly fork lengths of each year class were then calculated (Table 10.6). They show a steady increment with time, and observed lengths in any month generally compare well with the corresponding lengths in other years. Before plotting the growth curve it was necessary to establish a 'birthdate'" for _B. bayad . The histograms in Figure 10.7 show that the length distributions of each year class were centred round a prominent mode, indicating that a spawning peak was reached each breeding season. The analysis of sea- sonal changes in gonad condition (page 1031) show that the main breed- ing period was between April and July and that peak spawning occurred in May. Mid-May therefore has been taken as the 'birthdate' for the Longech population of B. bayad . Mean lengths of the four year classes in Table 10.6 have been plotted separately against age in Figure 10.8 and the combined data from Table 10.7 have been plotted in Figure 10.9 to demonstrate the observed pattern of growth over the first two and a half years of life. In both the 1972 and 1973 year classes growth appears to be almost linear over the first year and fish attain a length of approximately 17 cm at the end of the first year. Growth over the second year is* apparently erratic with a sudden acceleration in October-November . However, this anomaly is probably a result of sampling error, since at this time medium sized B. bayad moved into shallower water where they were caught in large numbers. By the end of the second year the 1970 year class had an observed length of 31 cm and the 1971 and 1972 year classes had attained a length of 34 cm. Since few fish larger than 40 cm were caught, direct observtions of growth have been restric- ted to the first 30 months of life. Fish from the 1971 year class had an observed length of 39 cm at an age of two and a half years. Mathematical expression of growth Table model 10.7 provides sufficient data to enable a von Bertalanffy growth to be fitted to the observed values, using the equation 1028 1 t L. CO (1 - e'K^-V) where is the asymptotic length, K is the rate at which L^ is ap- proached, lt is the length at any age t and tQ is a constant. Growth parameters were estimated using the graphical method des- cribed by BEVERTON and HOLT (1957) and RICKER (1958). The Walford plot showing the relationship between 1^ and lt+g months is presented in Figure 10.10. The following values were obtained for the growth para- meters of B. bayad . L =84 cm OO K = 0.2592 0.0672 years t o The estimate of agrees well with field observations. Only 0.3% of B. bayad in the trawl catches exceeded 80 cm FL and the largest caught by trawling was a female of 85 cm FL. A value of tQ = 0.0672 years indicates a birthdate in early June, a little later than would be suggested by the breeding data. There is close agreement between the lengths estimated from the von Bertalanffy growth equation and observed lengths over the first two and a half years of life as shown in Table 10.8. By extrapolation it is possible to estimate theoretical lengths of older fish, assuming that the growth pattern of B. bayad does not change. Growth in fish does not always follow the von Bertalanffy model. VAN DER WAAL (1975) found that in Clarias gariepinus growth followed an arithmetic pattern for the first ten years. HOPSON, A J (1972) observed that in Lates niloticus (L) the growth pattern changed after the first five years from an exponential to an arithmetic form. Studies of Bagrus docmac (CHILVERS, 1968) and Bagrus meridionalis (TWEDDLE, 1975) indicate how- ever that growth in Bagridae follows the same exponential pattern over the entire length range. It is reasonable therefore to estimate the lengths of older B. bayad by extrapolation. Estimates of length at age have been plotted in Figure 10.11 to- gether with the observed lengths from Table 10.8 (see also Table 10.11). Length-weight relationship Observations on the length-weight relationship are based on data ob- tained from B. bayad caught during the monthly trawl stations made in the Longech area. During the two years of sampling a total of 980 fish were weighed. Fish larger than 20 cm included 504 females and 374 males. The remaining fish were immature of between 10 and 20 cm FL. The length-weight relationship of fish can usually be expressed by the equation or logarithmically log1Qw = log1Qa + b (log1Ql) 1029 where a is the intercept of the regression line with the y axis and b is the regression coefficient. LE CREN (1951) showed that the length-weight relationship may vary with size, sex or maturity stage. To establish whether this occurred in EL bayd the length-weight rela- tionships were calculated separately for a number of size groups of each sex (Table 10.9). The slopes were subsequently compared by apply- ing d-tests. Results, given in Table 10.10, show that there was no significant difference (P = 0.05) between any of the regression co- efficients. The length-weight data for all JS. bayad can be combined therefore, giving' the following regression equation log^w = 3.032 (log^Ql) - 2.088 or w = 0.00817 l3*032 where w = weight in g and 1 = length in cm. A logarithmic plot of all fish caught in 1973 and 1974 during April and May is given in Figure 10.12 together with fitted regression line. Mean weights at one to ten years are presented in Table 10.11. Discussion Previous work on the ageing of catfish has been based mainly on the analysis of growth rings. In the present study however, although rings were found on the vertebrae of B. bayad no correlation between their formation and fish age could be established. Growth and age estimates up to two and a half years were possible however using length frequency data from trawl samples and growth was found to fit the von Bertalanffy growth method. It is of interest to compare the growth rate of B. bayad with two other African Bagrids, B_. docmac (CHILVERS, 1968) from Lake Victoria and _B. meridionalis (TWEDDLE, 1975) from Lake Malawi. In Table 10.12 growth for the three species is given for the first five years. Results show that the growth rate is higher in B. bayad than in the other species. More recent evidence, however, indicates that these species probably have a more rapid rate of growth than was first estimated. RINNE (pers. comm.) has reported that in Lake Victoria a single tagged j5. docmac grew from 26 to 40 cm in a year, a growth rate equivalent to that of _B. bayad . TWEDDLE (1975) assumed that two year classes of B_. meridionalis were absent from the trawl catches. However observations (TWEDDLE, pers. comm.) of juveniles with fork lengths of 7 to 9 cm in nests with an attendant female early in the breeding season indicates that growth in the first year is perhaps much more rapid than was first estimated and only one year class was absent from the trawl catches. This would bring the growth rate more in line with that of B_. bayad . BREEDING In the course of two years' fieldwork, 7,800 B. bayad were sexed in the Longech area. A further 600 were examined in the Northern sector and 1,100 in the Southern. A study of the changes in gonad condition has shown that in the Longech area an annual cycle of maturation 1030 occurred and a well defined breeding season, extending from April to July was apparent. Maturity stages The maturity stages of B. bayad detailed below were assessed according to the classification by NIKOLSKY (1963). Sexing was possible at a fork length as small as 6 cm when the testes were already slightly villose in appearance. Females Stage 2 Resting or recovering: found in all sizes of fish. Ovaries sac-shaped, translucent in maturing fish but pink in post-breeding fish due to greater vascularisa- tion . Stage 3 Mature: ovary larger and well vascularised . Develop- ing eggs whitish or yellow and less than 0.6 mm in diameter . Stage 4 Ripe: ovary greatly increased in size and weight. Ovary wall transparent, opaque eggs clearly visible varying in diameter from 1.0 to 1.4 mm. Eggs not free flowing when ovary wall cut. Stage 5 Ripe-running: ovary very turgid in appearance. Eggs translucent, varying in diameter from 1.3 to 1.6 mm. Slight pressure on belly causes extrusion of eggs. Stage 6 Spent: ovaries flabby in appearance and red in colour with residual eggs few in number. Males Stage 2 Resting or recovering: testes a pair of thin trans- parent strands. In virgin males a very slight scal- loped edge to testes visible. In recovering males larger flattened villiform protrusions on edge of testes . Stage 3 Maturing: villiform protrusions prominent, greyish white in colour. Stage 4 Ripe: testis white in colour, particularly the anter- ior villiform protrusions. Sperm released when cut and squeezed Stage 5 Ripe-running: testis white and villi turgid in appearance. Copious white milt extruded when cut. Size and age at maturity Changes in the proportion of _B. bayad at each maturity stage with in- creasing size are shown in Table 10.13 combining data collected during the breeding season April to July. The results show that males and females matured at approximately the same size. Fifty percent maturity in males was reached at a length of between 31 and 32 cm FL and in 1031 females at a length of between 33 and 34 cm FL. From the growth estimates presented above it would thus appear that B_. bayad matures towards the end of the second year of life. Seasonal changes in maturity The regular appearance of 0-group fish in trawl samples as demonstrated in length frequency histograms (Fig. 10 . 9) strongly suggests that _B. bayad has a well defined breeding season. This was confirmed by the analysis of monthly changes in gonad condition, for although some ripe fish were present in most months, peak gonad activity was confined to approximately four months each year. Monthly changes in the gonad con- dition of fish in the 35 to 39 cm length group from the Longech area are presented in Tables 10.14 and 10.15 for the period July 1972 to June 1974. In Tables 10.16 and 10.17 (see also Figs. 10.13 and 10.14) data for the two years have been combined for three length groups. Most ripe males and females in the Longech area were caught between April and July, with peak numbers occurring in May and June. In the 35 to 39 cm length group, only 3% of all male fish examined in May had quiet gonads, while 56% were ripe. In females of the same size group 47% were ripe in May. After the breeding season the proportion of ripe fish declined considerably, with a minimum being recorded between Sep- tember and January. Fewer samples are available from the Northern and Southern sectors of the lake and the breeding pattern of fish from these two sectors is not so well known. The seasonal changes in gonad condition of fish caught outside the Central lake sector are given in Tables 10.18 and 10.19. In the Southern sector of the lake fish appear to follow a similar pattern of gonadial development to that found in the Ferguson's Gulf area, with most ripe fish occurring in the June sample and fewest in October and December. The breeding season appears to be less well defined in the North- ern sector of the lake (Table 10.19). All samples from the north in- cluded a greater percentage of ripe females than was recorded in the corresponding monthly samples from the Longech area. During November 1972 over 40% of the females larger than 35 cm were ripe (stage 4). In the Longech area November is a post-spawning time and over 70% of females larger than 35 cm were resting (stage 2). Regular monthly sampling in the Northern sector would be needed to confirm that the breeding pattern differs from more southerly sectors. Sex ratio The proportion of males to females was not constant, but change pro- gressively with size. Table 10.20 presents the sex ratio of five length groups of fish examined in the Ferguson's Gulf area. In fish smaller than 40 cm the ratio of males to females was roughly 1:1 while in larger fish the proportion of males declined with increasing size. Seasonal variations in sex ratio were observed in fish larger than 30 cm. The proportion of male fish increased during the breeding period reaching a peak in May and subsequently declined in the post- spawning months. This is shown graphically in Figure 10.15 where monthly sex ratios for three length groups have been plotted for a full year. From October onwards relative numbers of males increased and by May, the peak spawning month, in the Ferguson's Gulf area, males exceeded females in the 30 to 34 and 35 to 40 cm length groups. Peak 1032 numbers of males were recorded in the over 40 cm length group during the same period. Relative numbers of males declined during the period following the spawning season to a minimum in August and September. Sex ratio data from the Northern and Southern sectors of the lake suggest that a similar decline in the proportion of males occurs throughout the entire lake during the post-spawning months. The lake- wide decline between June and September is probably therefore a result of the greater mortality of males during the early post-spawning phase, rather than a movement of males away from the sampling area. Site of spawning grounds Despite widespread sampling, the precise location of spawning grounds is still not known. Although occasional _B. bayad have been reported in the Kerio river, there is no evidence that this species breeds in any of the rivers flowing into Lake Turkana. During the present in- vestigations, few males or females were caught in the ripe-running state (stage 5) which suggests that the spawning beds are limited in extent. In an attempt to establish where breeding occurred a series of bottom trawls were made at a wide range of depths in the Longech area during May, the peak breeding time. Table 10.21 summarises the results and shows that whereas ripe females (stage 4) were widespread in their distribution relatively high numbers of stage 5 females were found only in the sample from 15 m. The results suggest that the spawning grounds are located in water of moderate depth. This hypothesis is supported by the depth distribution of juvenile B. bayad (Fig. 10.3) which are concentrated in water of 30-40 m. ” Stimulation of breeding The breeding season, which extends from the beginning of April to the end of July, coincides with the time of year when rainfall is most likely to occur in the Lake Turkana area. Records of the monthly rain- fall at Ferguson's Gulf over a 5 year period are given in Table 10.2. Although some variation occurs, heavy rain invariably falls between March and June, and in the five years nearly 60% fell during these months. It is possible that one or more factors associated with rain- fall stimulates breeding activity. In the Ferguson's Gulf area rain- fall is at times sufficient for the small rivers discharging into the lake to flow for a few hours or even days. Where these rivers enter the lake, turbidity increases and conductivity drops. However, there is no evidence that B. bayad breeds close inshore and the lowered conductivity and increased turbidity are generally ex- tremely localised and are unlikely to be factors directly stimulating breeding activity. LAKE (1967a) examined possible environmental factors involved in the stimulation of breeding in the Australian catfish Tandanus tandanus which spawns after flooding has occurred. He concluded that a volatile oil called petrichor, which is released when dry ground is inundated by flood water, may induce this fish to breed. The chemical nature of petrichor and its formation has been investigated by BEAR and THOMAS (1964, 1965). It is composed of a complex mixture of organic compounds containing basic, neutral and acid fractions, and it represents an ac- cumulation of these substances by adsorption onto materials such as 1033 silica. The oil possesses a characteristic odour and has been obtained by steam distillation of most silicate minerals and rocks. In nature it is formed particularly in dry arid regions after long periods of exposure. Following rainfall, petrichor is released and its presence in flood water may be an important factor in stimulating certain fish to breed. VAN DER WAAL (1974) reported that Clarias gariepinus spawned the night after they had been placed in shallow ponds that had been dry and filled the day before. He suggested that the release of petrichor from the bottom of the ponds when water was introduced, may have been the chief factor triggering off breeding. Petrichor may be responsible for the inception of spawning activity in other Claridae which breed during the flood season. Clarias mossambicus Peters reacted to flood waters flowing into Lake Victoria within eighteen hours (GREENWOOD, 1956) and Clarias gariepinus spawned in the early morning after heavy overnight rain in Rhodesia (HOLL, 1966 and 1968). It is possible that although B. bayad in Lake Turkana do not actually breed in flood waters following rainfall, they are able to detect the presence of petrichor, and this may promote breeding. Fish are known to have a remarkable sense of odour perception (HASLER, 1966; KLEEREKOPER, 1969). After several dry months the fall of rain in the arid region surrounding Lake Turkana may release sufficient petrichor into the lake for it to be detected by B. bayad at some distance from the point of inflow. The other environmental changes that take place after flooding, higher turbidity and lower conductivity seem unlikely to stimulate spawning activity in this fish. During September a sedi- ment plume from the River Omo extends southwards to include the Longech area of the lake. There is widespread, though slight reduction in con- ductivity and increase in suspended particles at this time in the Longech area, yet no surge of gonadial activity was observed. A high proportion of water entering Lake Turkana from the River Omo originated from the Ethiopian Highlands and may take several months to reach the- lake. It seems likely that during this period the amount of petrichor is much diminished by dilution, vaporisation etc. A second environmental factor which may be of importance in stimulating breeding is temperature. Records of temperature changes in water below the 80 m contour indicate that there is a slight fall in water temperature during March to a seasonal minimum of just below 25.5° C. This fall in temperature coincides with the onset of matura- tion in the gonads of B. bayad . Condition Values for the condition factor of all fish weighed during the course of routine data collection were estimated using the following equation k = w x 100 where w is the observed weight and w is the corresponding mean weight derived from the equation 3.032 w = 0.008171 (see page 1028) 1034 The mean condition factors of all mature _B. bayad weighed during the two years of sampling are given in Table 10.22 for four phases of the annual breeding cycle. The results show that there is a seasonal change in condition in both sexes. In males peak condition was recor- ded between October and January, the resting phase, possibly as a re- sult of increased fat deposits. In females the value of k similarly increased at this time, but peak condition was recorded in the spawning phase between April and July as a result of increased gonad weight. With the shedding of gonadial products, a loss of condition was noted in both males and females in the post-spawning phase. Fecundity In this study fecundity is defined as the number of maturing eggs found in a female just prior to spawning. During May 1974, which is the peak breeding month, all female _B. bayad with well developed gonads (stage 5) were measured to the nearest 0.1 cm. Pairs of ovaries were removed and a longitudinal slit made in each prior to preservation in Gilson's fluid. The preparation of preserved eggs for counting followed the method of BAGENAL (1968) and fecundity was estimated by the dry method, also described by BAGENAL (op. cit.). The results of the fecundity estimates are plotted logarithmically in Figure 10.16 and the points are distributed approximately on a straight line. A regression line has been fitted and the relationship between fork length and fecundity can be expressed as follows: loglO egg number = 2.373 (log^Q fork length cm) + 0.0857 The estimated length of a three year old B. bayad is 45 cm and using the above equation a fish of this length would have a fecundity of 10,200 eggs. An 80 cm fish would have an estimated fecundity of 40,000 eggss. Discussion Prior to the present study there was little information available on the breeding of B. bayad . Worthington found that in Lake Turkana the ripest fish occurred close inshore (WORTHINGTON and RICARDO, 1936). MANN (1964) suggested that B. bayad may come into shallower water to breed since the only ripe fish examined was caught at the entrance of Ferguson's Gulf. Evidence from the present study does not support this view. Ripe fish were caught in all depths of water. The occurrence of large numbers of ripe-running females in 15 m of water and the pre- sence of juveniles mainly in 30 to 40 m suggests that breeding occurred in water of moderate depth and not in very shallow water close inshore as was found to be the case in Synodontis schall (see page 1070 ). It is interesting to compare the breeding of B. bayad with two other African Bagridae, jj. meridionalis from Lake Malawi and B. docmac from Lake Victoria. JACKSON e_t al_ (1963) reported that in _B. meridionalis breeding coincided with the rainy season during the first quarter of each year. Spawning sites appeared to be in shallow, sheltered water where the sandy substrate was interspersed with rocks. A breeding migration occurred from the deeper water to these areas and circular or oval nests were built by male fish. Both JACKSON et al (op. cit.) and TWEDDLE (1975) caught a greater number of females than males. JACKSON et al (op . cit . ) suggested that this was a result of males remaining 1035 on the breeding grounds for longer periods of time, and thus being caught in fewer numbers in the deeper sampling areas- Tvweddle however considered that since males were proportionally fewer than females in the trawl catches at all times of the year, the difference in ratio was not due to any significant change in their behaviour. In Lake Turkana there was no evidence that males concentrated at any particular depth of the lake and the differences in sex ratio was probably the result of increased male mortality after spawning. Evidence from Lake Victoria was thought by CORBET (1960) to support the view that B. docmac breeds in the main lake. The large numbers of juveniles caught on exposed rocky shores suggested that breeding occurred in the vicinity but the exact spawning site was not known. CORBET (op. cit.) considered a rocky shore to be ecologically similar to that of the turbulent reaches in rivers. ELDER (1960) reported that B. docmac takes part in upstream migrations, but only in relatively small numbers. In Lake Turkana, although fishermen reported that B. bayad were caught in the River Kerio during the flood period, it would seem from all other evidence that spawning occurs in the main lake. If there was a large scale migration up the Kerio and Turkwell rivers during flooding, numbers of _B. bayad handled by the Kerio Co- operative would be expected to rise. Figure 10.26 shows that there was no apparent increase in numbers of 13. bayad sold per fisherman per week at this Co-operative during the flood season. Similarly there is no evidence from Todenyang Co-operative that a massive spawning migration of B. bayad occurs up the River Omo . FOOD The diet of B. bayad consisted primarily of prawns and fish. Variation in the proportions of these two food categories in the diet depended on predator size and on depth. Insects and ostracods were only rarely eaten and were of little importance as food items except in the shallow water of Ferguson's Gulf. Few post-larval B. bayad were caught, but those that were examined had fed entirely on copepods. Method During the two years of sampling the stomachs of over 2,000 B. bayad were examined. Except where otherwise indicated, stomachs were from fish caught by bottom trawling and hence the times and depths at which samples were made are known. Stomachs taken from fish of less than 15 cm were usually preserved in 5% formalin and scrutinised later in the laboratory with the aid of a microscope. Larger fish were examined on board the research vessel. Stomach fullness of fish larger than 10 cm FL was estimated using a points system. This method has been widely used by other workers (FROST, 1943; HYNES, 1950). Stomach contents are removed, sorted, identified and each food item is allotted points depending on size and abundance. Thus one large organism is considered equivalent to many small organisms and the method is essentially volumetric. In the present study a stomach that was considered to be full was allocated sixteen points. The relative volume of each food item was reflected in the number of points. For example, if a stomach was half full and judged to contain 75% prawns and 25% fish, prawns would re- ceive six points and fish two points. Using this method the percentage 1036 composition of the diet and frequency of occurrence of the various food items could be estimated. In fish smaller than 10 cm only the occur- rence of a food item was noted. Food of juvenile fish A sample of 70 juvenile B. bayad ranging from 20 to 30 mm FL was caught in May 1973 at a depth of 26 m in the Longech area. An analysis of stomach contents showed they had fed entirely on copepods. Unfortunate- ly further samples of similar sized fish were not caught and it is not known whether copepods are the only food of juveniles. It seems likely that other invertebrates, such as ostracods and cladocera are also eaten before the diet changes entirely to prawns. Food of immature and adult fish The diet of B. bayad larger than 3 cm FL consisted almost entirely of fish and prawns. The most common fish preyed on were Alestes spp., Haplochromis macconneli and B. bayad . The two prawns Caridina nilotica and Macrobrachium niloticum which occur widely in the lake formed the other major item in the diet. The relative proportion of fish and prawns eaten depended on two factors; predator size and the lake depth. Diet and predator size Figure 10.17 gives the change in diet with increasing predator size as illustrated by fish caught in the Longech area. The results show that there is a change in the diet of B. bayad with increasing length, from prawns to fish. Thus prawns which were the only food eaten by fish under 10 cm occurred as the sole stomach contents in only 2% of B. bayad larger than 70 cm. In all length groups few fish had fed on both prawns and fish, indicating that individual fish tend to prey exclusively on a particular type of food. Figure 10.18 illustrates the change in diet that occurs in mature. bayad . The number of points scored by fish in the diet has been ex- pressed as a percentage of the total points allotted in each 2 cm length group over the size range 30 to 74 cm. The results show that there is an abrupt change in the diet of B. bayad between 48 and 50 cm with the points attributable to fish increasing from 30% to 78% of the total. Diet with relation to depth The food of B. bayad also varies with depth. Fish tended to be more common in the diet of shallow water B. bayad of all sizes over 10 cm FL than was the case in deeper water. Table 10.23 presents a detailed analysis of the change in diet with predator size in relation to depth. The results have been combined on a lakewide basis for all bottom trawl stations . In the 10-19 cm group, fish were the major food item only between the 10 and 29 m contour. At all other depths prawns were more common. With increasing size fish were eaten to a greater extent in all depth zones, although B_. bayad of up to 50 cm still fed mainly on prawns in deep water. At lengths over 50 cm the feeding habits were principally piscivorous at all depths. 1037 The variation in the percentage of prawns in the diet of B. bayad at different depths is probably a reflection of the distribution of prawns in the lake. Densities of prawns are generally lower in shal- lower water except under turbid conditions during the flood period (see Chapter 4). Consequently B. bayad inhabiting inshore areas depend to a greater extent on fish as a food item. Once B. bayad exceed a length of 50 cm changes in diet with depth are less apparent since fish are the preferred food, even when prawns are present. Insects rarely occurred in the diet of adult B. bayad and were found only in offshore stomachs. Of the identifiable remains, locusts (Orthoptera) made up 70% of the total insect points, Anisopteran nymphs 19% and corixid nymphs and adults 11%. It was surprising to find locusts in the stomachs of fish caught in 80 m of water. These insects may alight on the water surface in error. They have a fairly thick exoskeleton, and once in water would quickly sink to the lake bottom where presumably they were scavenged by B. bayad . Within the confines of Ferguson's Gulf where depths do not exceed 4m the diet differed markedly from the open lake. The stomach contents of fish caught in the Gulf by beach seines are presented in Table 10.24. Prawns were noticeably absent from the diet of these inshore fish. They were replaced by insects which formed more than 25% of food points in the 20-49 cm length groups. Ostracods which were not recor- ded elsewhere were also eaten by B. bayad in Ferguson's Gulf. Mature B. bayad were found to feed on a variety of fish. The com- monest prey species were Haplochromis macconneli , smaller Alestes spp . and B. bayad , and their relative importance varied with depth. In Figure 10.19 points allotted to prey species have been expressed as a percentage of the total number of fish points recorded at each depth interval. In water of less than 30 m Alestes spp. accounted for more than half the total intake of fish. Species of Alestes in the diet included A. nurse, an inshore form, and A. ferox and A. minutus which occur mainly in midwater. They are the principal species of pelagic fish occurring in the midwater scattering layer which is situated in the open waters of the lake at a depth of between 3 and 30 m (see page 599 ). These fish therefore would be readily available to B. bayad in- habiting shallower water, within the 30 metre contour. In water deeper than 30 m Haplochromis macconneli become the dominant fish eaten. Trawling has shown that this species is restric- ted to water deeper than 20 m and becomes more abundant with depth. The importance of H. macconneli as a food item gradually increased off- shore. Bagrus bayad is also cannibalistic, feeding to a significant degree on its own young in deeper water. The remaining species are of less importance. Their occurrence in the diet reflects their distribution in the lake. Hydrocynus f orskalii (Cuvier), a surface living predator, Tilapia spp. and several silurid fish (other than _B. bayad) formed a minor proportion of the diet of B_. bayad living in shallow water. Lates longispinis which is caught by trawling in the scattering layer was occasionally found in stomachs from depths of between 10 and 40 m. Engraulicypris stellae has a widespread distribution in the lake and occurred in the diet in most depth zones. The deep water Barbus cf anema was present in the diet of B. bayad caught at depths between 20 and 60 m. 1038 Predator-prey size relationship The relationship between predator and prey size is given in Figure 10. 2Q for the three principal prey forms. Total lengths of Haplochro- mis macconneli and Alestes spp. and fork length of J3. bayad eaten have been plotted against predator fork length, and regression lines have been fitted. The results show that the mean lengths of Haplochromis and Alestes vary only slightly over the observed predator size range. Neither fish attain a size, much greater than 10 cm in Lake Turkana. Mean length of Haplochromis in the diet was 5.34+; 1.55 cm and of Alestes spp. 5.37,11.46 cm. Mean length of _B. bayad occurring in the diet increased with pred- ator size but the minimum prey size remained virtually unchanged in all size groups. Approximately one fifth of the B. bayad eaten were more than 40% of the predator length and in one instance the prey ex- ceeded 60% of this length. The results show that J3. bayad is capable of ingesting prey over a wide size range. Diel feeding pattern Catch rates with the bottom trawl varied considerably with the time of day. Results from a 24 hour trawling station maintained in 80 m of water off Central Island are given in Figure 10.21. Catch rates for B. bayad in kilograms per hour have been plotted against the times at which the trawls were made. The results show that between 17.00 and 18.00 hours catch rate fell by over 50%. The decline continued after nightfall and at 21.00 hours less than 1 kg of fish was caught per hour of trawling, equivalent to about 1% of the average daytime catch. With the approach of daylight the catch rate increased and at 10.00 hours it was similar to that of the previous day. The results from a series of frame trawl samples made over a 24 hour period at a station in 25 m of water confirm that during the night B. bayad moves off the bottom into the midwater layer. Table 10.25 gives the numbers of fish caught per ten minute trawl in seven depth zones at five different times during a 24 hour period. The results show that at dusk B. bayad move off the lake floor and occupy the mid- water layers, reaching a maximum distance above the lake bottom during the night. At dawn the fish return to their daytime position on or near the bottom. Results from stomach analyses are given in Figure 10.22 where mean percentage fullness of all fish caught is presented for each period during the 24 hour cycle. Peak fullness was recorded during the night and the vertical migration would appear to be an adap- tation to maintain B_. bayad in contact with its prey. Feeding is prob- ably not confined to the hours of darkness since food was also present during the day. The frame trawl results help to explain the marked decrease which occurred at night in bottom trawl catch rates in deep water shown in Figure 10.21. If fish behaviour follows the pattern demonstrated for shallower water, a vertical migration off the bottom at the approach of night would result in the low catches that were actually recorded. Seasonal feeding pattern Results from trawl catches have shown that in the Longech area there is a movement of B_. bayad into shallower water between August and November. Table 10.26 examines the seasonal variation of catch rates 1039 and diet of 30 to 39 cm B. bayad in water of 10 to 20 m. Although sam- ples from December and April are. absent, the results show that there is an increase in the proportion of prawns in the diet from September onwards, with prawns providing the entire food intake during November. This coincides with the time when bottom trawl catches reach an annual peak in water of less than 40 m (see Fig. 10.4). From February to July fish form over 50% of the food and during these months catch rates with the trawl in water of less than 40 m are below the annual mean. The reason for the increase in prawns is not clear. From June to October the Omo river is in flood and sediment plumes spread south reaching the vicinity of Ferguson's Gulf in early September. Associated with the flood water is a reduction in light penetration. The depth at which 99% light extinction occurs decreases from 7.5 m in March, April and May to only 2 m in September and October. The reduction in light pene- tration results in the ascent of the scattering layer into surface waters and the movement of open water pelagic forms, including prawns, inshore. The movement of B. bayad inshore may be a direct result of increased prawn density and is possibly a feeding migration. Discussion Although immature B. bayad of between 3 and 30 cm feed on prawns, adults are mainly piscivorous and are important predators on fish in Lake Turkana. The other major piscivorous species in the lake are Lates niloticus and Hydrocynus forskalii . The role of these two species and other Hydrocynus spp. as predators in African lakes has been discussed at length by several workers (HAMBLYN, 1960; JACKSON, 1961; FRYER, 1965; LEWIS, 1974). In Lake Turkana H. forskalii occurs in surface waters while Lates niloticus mainly inhabits water of less than 30 m though it is occasionally caught down to a depth of 80 m. Adult Hydrocynus forskalii feed chiefly on Alestes minutus and Engraulicypris stellae in Lake Turkana and therefore do not compete directly with B. bayad for food. Lates niloticus however preys on a wider variety of fish, including jB. bayad and Alestes spp. It is therefore in competition with B. bayad in the shallower areas of the lake . Schilbe uranoscopus , Lates longispinis and Synodontis schall may also be regarded as competitors, since they are partly piscivorous, feeding particularly on Alestes spp. Bagrus bayad has no competitors in the deeper waters of the lake, where it is the only major predator on Haplochromis macconneli . One group of fish rarely eaten by _B. bayad , but which forms an important part of the diet of Lates niloticus is Tilapia spp. In Lake Victoria CHILVERS and GEE (1974) reported that few Bagrus docmac had fed on Tilapia spp. They suggested that the behaviour pattern, warning systems or some other factors made these cichlids inaccessible to jB. docmac ♦ Tilapia spp. in Lake Turkana similarly occurred only rarely in the diet of J3. bayad but the reason would appear to be due to major ecological differences between the two forms. Tilapia spp. are rest- ricted to a narrow inshore zone within the 4 m contour where B. bayad occur only infrequently. A change in diet with increasing size was noted in B. docmac from Lake Victoria (CHILVERS and GEE, 1974) although this species became principally piscivorous at a fork length of 16 cm compared with 48 to 50 cm in B_. bayad . The change in diet in both B. bayad and B. docmac is not a result of change of habitat. It is possible that larger bag- rids have to switch to a diet of fish in order to obtain sufficient food . 1040 It is concluded that B. bayad is most successful in the deeper waters of Lake Turkana where competition from other predatory species is minimal. In shallower water where interspecific competition from prawns and fish is greater, the species is relatively less abundant. COMMERCIAL FISHERIES Bagrus bayad is of very minor importance, in the inshore gillnet fisher- ies. However since the introduction of longlines in 1972 this species has been caught in significant quantities and in 1974, 245 tonnes of 13. bayad were exported from the lake. This represented 7% of the total commercial catch of 3,500 tonnes. The commercial exploitation of J3. bayad by gillnets and longlines will now be discussed in detail. Gillnets Trials to evaluate the commercial use of gillnets in a _B. bayad fishery were carried out in the Longech area during the period January- July 1974. Details of gear used in the trials are presented in a previous section (see page 1022 ). The bottom-set gillnets were fished for periods of 24 hrs at various depths from 9 m to 73 m. Mean catch per effort for B. bayad (kg/24hrs/100 yds headrope) is given in Table 10.27, together with the percentage of weight of the total catch. It is clear from the few results-in deeper water that gillnet catches of _3. bayad were relatively poor in comparison with trawling or long- lining. Figure 10.23 presents catch per effort data for trawling and gillnetting at various depths in the Longech area. Catch per effort for the gillnets is given as kilograms of B_. bayad per 24 hrs per fleet of gillnets, and for trawling in kilograms per hour. Whereas the trawl catch gradually increased with depth, results from gillnetting showed a peak catch rate between 25 and 30 m, with a decline in the deeper water. The diurnal migration pattern of _B. bayad may help to explain poor gillnet catches in deeper water. Bottom trawl and frame trawl catches over 24 hours (Fig. 10.21 and Table 10.25) have shown that at night B. bayad move off the bottom to feed. Although gillnets fished up to 4m off the lake bed, it is possible that B. bayad were feeding above this depth. After feeding in midwater, fish return to the bottom at dawn, and probably remain relatively immobile during the day. Such behaviour could result in little horizontal movement in the vicinity of the deepwater nets, thus reducing the likelihood of capture. In shallower waters the probability is greater that the gillnets were set in part of the midwater zone where B. bayad were feeding, resulting in the higher observed inshore catches. Observations on the behaviour of Bagrus docmac (ELDER, 1962) have shown that this species is inactive in the daylight hours. Work with gillnets is incomplete and more data is needed from the deeper areas of the lake. It would appear however at this stage that gillnetting is not a satisfactory method for catching B. bayad at depth. The selectivity of the nine nets used in this survey is given in Appendix 10B and mean retention lengths in Appendix 10A. 1041 Longlines Longlines have been used for several years by Turkana fishermen but they were mainly set in shallower water to catch Lates niloticus. Only since 1972 has this method been used in the deeper parts of the lake through the encouragement of NORAD* master fishermen. Under their guidance an experimental scheme was started in January 1974 at Namudak, 10 km north of Ferguson's Gulf. A 7 m fibre glass dinghy powered by an inboard engine was used t enabling fishermen to operate up to 20 km offshore in the centre of the lake. Data from this scheme was collec- ted and estimates of catch per effort are presented in Table 10.28 for the period January to September 1974. Lates niloticus was the only other fish caught by the longlines and catch per effort data is pre- sented in Appendix 10E. Figure 10.24 compares the length-frequency distribution of fish caught by longlining and by trawling during January 1974. The size of B. bayad was considerably greater in longlines, where a mean of 66 cm ( sd = 12 cm) was recorded, than in the bottom trawl with a mean of only 27 cm (sd = 11 cm). The relatively large size of fish caught selectively by the longlines were of a size which was represented by only a small proportion of fish in the bottom trawl catches. Catch per effort in the longlines varied during the experimental period (Table 10.28) with the mean monthly catch rate ranging from 20.7 kg/100 hooks/night in January to 8.7 kg/100 hooks/night in July. An analysis of the catch on a weekly basis, however, has revealed no ten- dency for catch per effort to decline over this initial period (Fig. 10.25). Catches tended to fall between January and March but there was a sharp rise in April. Similarly there was no tendency for a re- duction in mean fork length during these preliminary investigations. However since the completion of this trial fishing period in September 1974 the Norwegian master fisherman has reported that during late 1974 and 1975 longline catches of B. bayad have fallen to uneconomical levels in offshore areas in the Central sector of the lake (E RifKKE , pers. comm.). The implications are that the present level of fishing intensity is already having a detrimental effect on stocks of B. bayad. Longlining is probably the ideal method for catching _B. bayad in Lake Turkana since only larger individuals are selected. Recruitment into the longline fishery is delayed until after B. bayad has bred at least once, and usually twice. The reason for the bias in favour of large size may be due to the feeding habits of fish smaller than 50 cm which as has been shown above, feed principally on small fish and prawns and are less likely therefore to take bait. Trends in the commercial catch A detailed account of the present day commercial fishery on Lake Turkana appears in Chapter 6. This information enables trends in catches of _B. bayad between 1972 and 1974 to be examined. Table 10.29 gives the numbers of IS. bayad handled, expressed as a percentage of the total fish bought, at each of seven collecting centres in the last three months of 1972, 1973 and 1974. During October-December 1972 the numbers of J3. bayad exceeded 5% of the total catch only at Namudak and Kalokol co-operatives. These were the two centres where longlines were ♦Norwegian technical aid. 1042 first introduced to the Turkana fishermen in early 1972. Bagrus bayad quickly became an important fish in the commercial catch, accounting for up to 27% of the total fish handled at Namadak during 1972. Else- where on the lake, gillnetting was still the main fishing method and few B. bayad were caught. Between 1972 and 1974 the numbers of 3. bayad increased progressively at all co-operatives and by October- December 1974 lakewide catches were approximately three times greater than during the same months in 1972. Catches were still comparatively low to the north in the Todenyang and Lowarengak areas, and to the south in the Kerio area, where gillnetting remained the principal fish- ing method. The higher catches at the remaining co-operatives was largely due to the increased use of longlines. By the end of 1974 Kalokol co-operative had replaced Namadak as the main centre for _B. bayad. This was primarily the result of an expansion of longlining into the central part of the lake from a base on Central Island. The island is close to deep water and to high concentrations of B. bayad . The establishment of bases on the island overcame the problem of travelling long distances to set and haul gear. Fish caught from Central Island were sold mainly through the Kalokol co-operative. In December 1974 nearly 70% of all fish handled by this co-operative was _B. bayad, as compared with less than 5% in 1972. Much of the longline catch at these two co-operatives was caught in the vicinity of North Island where a fishing camp had been established during 1973. By the end of 1974 B. bayad accounted for an appreciable propor- tion of the fish handled by most co-operatives. Values for catch per effort of the commercial fisheries, expressed as numbers of 3. bayad sold per fisherman per week are given in Figure 10.26 for all co- operatives over a 30 month period, July 1972 to December 1974. At Kerio, Lowarengak and Todenyang where, as mentioned previously, long- lines were not in use, less than one B. bayad was sold per fisherman per week. Catch per effort was appreciably higher than this mean value at all other • co-operatives during 1974 and at Kalokol the commercial catch at times exceeded ten B. bayad per week. Longlining was expanding continually during the period under con- sideration and pronounced seasonal trends may well have been masked. However at Lowarengak, Nachukui and Kataboi in 1973 an increase is noted in the middle of the year perhaps as a result of the inshore migration of B^. bayad observed in the Longech area when turbid Omo flood water was present. An inshore migration would bring B. bayad into the shallow water gillnet areas. Fishing pressure in the Central area In the absence of reasonable estimates for either the small amount of _B. bayad eaten locally or the negligible quantity caught by gillnets it is considered that co-operative data at Kalokol and Namadak provide a good approximation of the total longline catch in the area. The mean catch rate observed in experimental longlines in the same area was one _B. bayad per 21 hooks per night. Since the weekly numbers of B. bayad purchased by the Namadak and Kalokol centres is available, an estimate of the numbers of hooks set per night during 1973 and 1974 in the Central lake area can be made as shown in Table 10.30. By the end of 1974 an estimated 5,000 hooks were set nightly in the Central lake area, with a weekly catch of 1,500 B. bayad. This area of the lake has the highest longline fishing intensity and this 1043 method now accounts for between 50% and 70% of all fish handled by the Kalokol and Namadak co-operatives. Future expansion In other areas of the lake, particularly in the north, fishing effort with longlines is still low and the potential catch rate is likely to be much higher than was recorded in 1974. Results from trawl catches show that in the northern sector population densities of _B. bayad are similar to those observed in the Central sector. This indicates that catch per effort in the north may increase to a level similar to that of the Central sector. Potential yield in the Southern sector is generally low. An exception is the small area of deep water in the south of South Island, where trawl catches were well above average. After the initial expansion and high catches however, indications are that catch rate rapidly declines and falls below economical levels. Figure 10.24 shows that the majority of fish caught by longlines in January 1974 were larger than 50 cm. In Table 10-31 the percentage length distribution of the total catch by longlines between January and September 1974 has been compared with the entire trawl catch from the two years of sampling. Longline catches were composed of large fish 85%, of which exceeded 50 cm in length. In trawl catches only 7% exceed this upper size limit. For the exercise which follows it is assumed that trawl catches provide a true estimate of population structure. The main longlining area of the lake encompasses an estimated 700 km to the north of Central Island. The depth of water is generally greater than 60 m and all the B. bayad caught here are sold through the Kalokol and Namadak collecting centres. Bottom trawl results in this area show that a mean of 12 _B . bayad larger than 50 cm FL were caught per hour. Since the trawl covers an area of 0.11 km each hour (see page 586 ), the total number of B. bayad in the main longline area larger than 50 cm is 76,400. During 1974 a total of 58,660 B. bayad were sold at Namadak and Kalokol, and it may be assumed that the majority of these were caught by longlining and were larger than 50 cm. Thus in 1974 at least 75% of the estimated total standing stock of fish larger than 50 cm were removed. With fishing intensity as high as this it is not sur- prising that the catch rate has declined and is no longer economical in the Longech area. Since fish between 30 and 40 cm which make up approximately 80% of the breeding population, are not susceptible to longlining, recruitment would continue and no permanent damage to stocks is likely. In other areas of the lake expansion of longlining will probably occur, with increased numbers of B. bayad entering the co-operatives. After the initial expansion however catch rates will certainly decline, eventually stabilising at a much lower level. BAGRUS DOCMAC Bagrus docmac is widely distributed throughout the soudanian region of Africa, occurring in the Chad, Niger and Volta systems. It extends through much of the Nile basin including Lake Victoria. The species has been studied in detail by a number of workers. ELDER (1960), CORBET (1961), CHILVER and GEE (1974) have investigated the population inhabiting Lake Victoria. Prior to the present survey B. docmac had 1044 not been recorded from the Lake Turkana basin and despite the extensive sampling programme carried out between May 1972 and July 1974, a total of only 33 specimens were caught by bottom trawling. Two further specimens were obtained from gillnets and two from beach seines. How- ever an intensive programme of gillnetting carried out in the Loiengalanr area of the Southern basin during early 1975 showed that B. docmac occurred frequently in rocky inshore areas. The material was insufficient for a detailed analysis of distri- bution. The position of stations where B. docmac were recorded are given in Figure 10.27. No capture was noted in water deeper than 30 m and it was absent from samples north of North Island, though a very small number have been recorded at the Todenyang co-operative. An association between the occurrence of B. docmac and a hard substrate was noted, since the largest single catch coincided with extensive net damage, caused by large rocks entering the trawl. It is interesting to note that the only Malapterurus electricus (see page 1048) trawled during the present survey occurred in the same haul. At a second station where two B. docmac were caught the trawl was again damaged, presumably by rocks. Bagrus docmac inhabiting Lake Victoria also shows a preference for a coarse substrate (ELDER, 1960) and BLACHE (1964) noted that in the Lagone and Chari rivers this species was principally caught in the vicinity of rocks. The majority of B. docmac examined had resting gonads and fish in a ripe or ripe/running condition were not caught (Table 10.32). ELDER (1960) reported that although some B_. docmac in Lake Victoria migrate upstream to spawn, the majority probably breed within the con- fines of the lake. There was no evidence of anadromous breeding move- ments in the Lake Turkana population and spawning is probably fully lacustrine . The food of B. docmac below 35 cm FL consisted chiefly of fish and prawns in equal proportions but insects were also frequently eaten (Table 10.33). The diet was thus more varied than was noted in . bayad of similar size, where prawns formed over 70% of the diet (Fig. 10.17). The fish most commonly preyed on were Haplochromis macconneli and Chrysichthys auratus . Bagrus docmac larger than 35 cm FL were entirely piscivorous, feeding predominantly on Synodontis schall and Tilapia spp. CHILVERS and GEE (1974) reached similar conclusions in Lake Victoria where B. docmac larger than 27 cm FL were principally piscivorous . The low incidence of _B. docmac in the bottom trawl samples sug- gests that this species will be of little commercial importance. Never- theless a small quantity of B. docmac passes through the co-operative each year and BAYLEY (1975) has estimated that a total of 5.9 tonnes (fresh weight) was exported in 1974. More than half of the catch came from the Kerio and Eliye co-operatives, probably from gillnets set on the more rocky east coast of the lake. Few B. docmac were recorded at Kalokol and Namadak, the main longlining centres indicating that in the Central sector no significant increase in the catch can be ex- pected from the development of this method. However, as mentioned pre- viously, trials with gillnets carried out by a group of Italian volun- teers in the Loiengalani area of the Southern basin during the period February to April 15 show that in rocky inshore areas the proportion of this species in the catch is much greater than in the northern areas of the lake. For example in El Molo Bay during March 1975, two 6 inch 1045 gillnets set for a total of four nights caught 6 large JJ. docmac forming 18% by weight of the total catch. CHRYSICHTHYS AURATUS Chrysichthys auratus occurs in the Niger, Chad and Nile basins. There are no previous records of this species from the basin of Lake Turkana, although the present studies show that it is widespread and at times common. The maximum size recorded during the survey was 29 cm FL but the species matures at a length of only 7-8 cm, and the great majority of fish were smaller than 15 cm (Fig. 10.28). As a result of its small size C. auratus is not of potential commercial importance. Distribution Mean bottom trawl catches, in kilograms per hour, are presented in Table 10.34 for 12 depth zones. All data collected between May 1972 and July 1974 have been combined. Chrysichthys auratus is distributed over a wide depth range, peak numbers occurring within the 10 to 25 m contour. The mean catch rate was less than 2 kg/hour in all depth zones, but occasional hauls greatly exceeded this level. Thus during September 1973 a catch of 48 kg/hour was recorded at a depth of 12 m off Koobi Fora and in January 1974, 35 kg were trawled in one hour off Longech Spit in 16 m of water. Each of these samples included more than 2,000 fish suggesting that at times shoaling may occur. J Evidence from frame trawling, carried out at a depth of 24 metres off Longech Spit over a 24 hour period shows that during the daytime C. auratus is confined to the region of the lake bed (Table 10.35). At dusk a vertical migration occurs, probably in response to falling light intensity, and fish swim into the midwater layers where they remain during the night. Peak stomach fullness was recorded at this time and the vertical migration thus coincided with an increase in feeding activity. Close inshore C. auratus was only occasionally caught by beach seines during the day. However results obtained over a 24 hour period in Ferguson's Gulf in October 1975 show that C. auratus migrates in- shore at night. Beach seine catches with a 45 m net of 19 mm mesh increased dramatically from 18.00 hours when no fish were caught, to approximately 500 per haul at 22.00 hours and 200 per haul at 02.00 hours. An analysis of stomach contents revealed that these fish had been feeding mainly on chironomid larvae and pupae. At dawn catch rate fell to only 3 per haul, and none were caught in the morning. Chrysichthys auratus therefore has a markedly different distribu- tion at night compared with the day. During the daylight hours it is principally demersal but at night it migrates into midwater areas and closer inshore to feed. Breeding A total of 444 males and 470 females were examined during the two years of sampling. The ratio of males to females over the length range 5 to 29 cm is presented in Table 10.36. Data has been grouped into 5 cm length classes and chi-square test revealed that the sex ratio was not significantly different from the hypothetical ratio of 1:1 (*\j2 = 4.08, with 4 degrees of freedom). 1046 Size at maturity results (Tabe 10.37) show that males and females both matured at the same length. The smallest mature fish in both sexes measured 6 cm and 50% maturity was reached at between 7 and 8 cm. Only two ripe males were caught throughout the study period but ripe females were recorded in all months of the year. Seasonal changes in the proportion of ripe females, larger than 6 cm, are presented in Table 10.38. Insufficient data was available for all months but the results indicate that peak gonadial activity occurs between February and May. This coincides with the rainy season and breeding probably occurs in response to rainfall and flooding. The fecundity-length relationship of £. auratus is presented in Figure 10.29 where egg number has been plotted against fork length. The diameter of preserved eggs ranged from 2.0 to 2.5 mm and in com- parison with Synodontis schall and Bagrus bayad fecundity was low. Feeding The diet of Chrysichthys auratus was comprised almost entirely of insects and Crustacea. The relative volume of each food item was esti- mated using the points system described previously for Bagrus bayad (see page 1036) and in Table 10.39 the stomach contents of all fish ex- amined have been analysed. Chironomid larvae and pupae and ostracods are the most important food items and together contribute two thirds of the diet. Table 10.40 analyses diet composition in relation to depth. In water shallower than 20 m chironomids and ostracods were the dominant food items, but in deeper water prawns were of greater importance. The migration of £. auratus into midwaters and inshore for food has already been commented on. Fisheries Chrysichthys auratus has no commercial or subsistence value due to its small size and relative scarcity. AUCHENOGLANIS OCCIDENTALIS MANN (1964) was the first to record this species from Lake Turkana. Ten specimens, ranging from 26 to 31 cm FL were caught by gillnets during his visit, all from within Ferguson's Gulf. Despite extensive sampling during the present survey only one individual measuring 26.2 cm FL was caught, at Arika Lubwangole , an old river channel at the north end of the lake (see Fig. 10.27). It is curious that further specimens were not caught within Ferguson's Gulf or elsewhere in the lake, considering the relatively large number examined by Mann during his brief visit. It appears that A. occidentalis is now confined to the northern boundary of the lake and is perhaps fairly common in the River Omo and its delta. CLARIIDAE AND MALAPTERURIDAE CLARIAS LAZERA This species has an extremely wide distribution in Africa, occurring in the Nile, Chad, Volta, Senegal and Lake Turkana basins (BOULENGER, 1911). Clarias lazera has also been recorded in Syria and Lake Galilee (GREENWOOD, 1966). In Lake Turkana C. lazera attains a maximum recor- ded length of 102 cm TL (Table 10.41). 1047 The species is confined Co shallow water close inshore and only one individual was caught in the course of several hundred trawl hauls. Clarias lazera was also rare in inshore gillnets and almost all speci- mens examined during the present survey were obtained by beach seining. Large samples were never obtained and data are insufficient for a de- tailed biological study. Small numbers occurred regularly in monthly seine net samples from Ferguson's Gulf and these, together with occa- sional specimens from seine net hauls in other parts of the lake, form the basis of the present account. Escapement from the shore seines was probably high since Clarias rarely became entangled and was fre- quently seen swimming through tears in the net or slipping under the foot rope. The habitat of this species ranged from soft mud in the Todenyang area to a stony substrate at Moite. Information on the breeding biology of Clarias in Lake Turkana is incomplete. The species probably matures at a length of over 50 cm. The smallest ripe male observed measured 36 cm and the smallest female was 66 cm TL. Data on the seasonal occurrence of ripe indivi- duals is inconclusive. Males in this condition were recorded during the months of September, October and January and the only ripe females were also recorded in January. Spawning however is thought to occur during the rains and under normal circumstances it is probably preceded by a migration into floodwater entering the lake. It seems likely that the River Omo is an important breeding area but spawning certainly occurs elsewhere as proved by a large sample of juvenile Clarias of 12-24 mm TL which were caught at the mouth of the River Kerio in July 1972. The species also migrates up smaller temporary rivers to breed and a juvenile of 35 mm was caught in a seasonal pool in the River Kalokol , approximately 12 km from the lake during July 1974. Breeding however is not confined to rivers. During the 1930-31 Cambridge Expedition, several specimens were collected from Crater Lake A on Central Island (WORTHINGTON and RICARDO, 1936) and this isolated group of fish were clearly breeding in a lacustrine environment. Prob- able courtship behaviour between two large adult Clarias was noted in June 1975 on the grassy margin of Low Island in Allia Bay, suggesting that breeding might be occurring in this locality away from the influ- ence of rivers. In Lake Victoria Clarias mossambicus breeds in small temporary streams which flow into the lake during the rainy season (GREENWOOD, 1955). Breeding occurs over a limited period of time and ripe fish migrate upstream on. one or two nights only, in response to flooding. In southern Africa Clarias gariepinus also breeds after flooding has occurred (VAN DER WAAL, 1974). The analysis of stomach contents showed that copepods were the principal food with occasional ostracods, chironomid larvae and corix- ids (Table 10.42). The remains of a Tilapia were found in the stomach of a 36 cm Clarias. These results agree with the findings of WORTHING- TON and RICARDO (1936). POLL (1939) noted that in Lake Mobutu Sese Seico (formerly Lake Albert) C. lazera is omnivorous, but mainly a predator on small fish (Haplochromis and young Tilapia). The small number of Clarias examined during the present survey gives an inaccurate picture of the importance of this fish in the pre- sent commercial catch. A total of 50 tonnes fresh weight (see Chapter 6) were exported in 1974, most of which came from the Todenyang and Kerio areas, and from the east shore between Allia Bay and Moite. 1048 MALAPTERURUS ELECTRICUS PELLEGIN (1935) recorded a single specimen of Malapterurus electricus , the electric catfish, from the Omo delta. During the present survey one fish, measuring 20 cm TL, was caught by the bottom trawl in 15 m of water, 1 mile off the west shore of Longech Spit (Fig. 10.27). This capture coincided with the largest sample of B. docmac obtained during the survey (see page 1044) and both these species appear to prefer a rocky substrate. For this reason Malpterurus electricus may be more plentiful in the Southern basin. A second specimen of 28 cm TL was caught in a shore seine on the east shore opposite the Kerio delta at Sandy Bay during January 1975. The substrate was muddy. It is re- ported to be fairly common in the River Omo in the vicinity of Kalom, just north of the delta, and has been noted in the permanent waters of the upper Turkwell river at Mamatte (J. RIENKS pers . comm.). SCHILBE1DAE SCH1LBE URANOSCOPUS Schilbe uranoscopus is a nilotic species restricted to the River Nile basin. WORTHINGTON and RICARDO (1936) recorded this fish frequently in the open waters of Lake Turkana but obtained only a single specimen from inshore waters within the confines of Ferguson's Gulf. In Lake Turkana Schilbe attains a maximum recorded fork length of 34 cm. DISTRIBUTION Results from the present survey show that although more plentiful in offshore waters Schilbe also occurs regularly inshore where it is caught in beach seines and gillnets. Although direct evidence is not available, larger numbers of mature fish probably migrate into the three main rivers flowing into Lake Turkana during the breeding season. A riverine population of Schilbe uranoscopus was discovered in 1974 in the upper reaches of the River Turkwell between Ketilo and Sigor. Schilbe was present in roughly 50% of all the bottom trawl samples but catch per effort in terms of kilograms per hour was relatively low. Trawl catches exceeded 10 kg/hour in only 5% of the hauls and thus al- though Schilbe is widely distributed, the biomass is not high. Trawl catches Catch per effort in kilograms per hour of bottom trawling are presented in Table 10.43 for the three sectors of Lake Turkana. In the Central sector Schilbe was most abundant at depths of be- tween 20 and 25 m where the mean catch rate reached 10 kg per hour. Catches in this depth zone however varied considerably. During April 1973 catches fell to 6 kg per hour whereas in April of the following year 193 kg of Schilbe were trawled in one hour. In water shallower than 20 m and deeper than 30 m the mean catch rate did not exceed 3 kg/hour. Schilbe were rarely recorded at depths greater than 50 m and it is possible that fish from deep water stations may have been caught in midwater as the net was being shot or hauled. Similar trends in distribution were noted in the Northern sector with catches very low at depths greater than 40 m. In the Southern 1049 sector no marked trend is evident and, although values varied, rela- tively low catch rates, not exceeding 2.5 kg/hour, were recorded from both shallow and deep water. Frame trawling Data from a series of daytime frame trawl hauls made at an inshore and an offshore station in the Longech area during September 1973 are pre- sented in Table 10.44. Results from the inshore station show that Schilbe was restricted to a depth zone of between 4 and 20 m, and that peak numbers occurred within the midwater scattering layer (see page 605 ). A similar pattern of distribution was also observed at the offshore station where the relatively small numbers of Schilbe were confined to a narrow depth zone between 8 and 16 m. Frame trawl re- sults thus indicate that in open water during the daylight hours Schilbe tended to concentrate in the vicinity of the midwater scatter- ing layer. Further frame trawl observations to be discussed below in the section on feeding (see page 1058) show that Schilbe migrates to- wards the surface at night. Gillnettting Four bottom set gillnet stations were established off Longech Spit and were regularly fished between January and July 1974. The depth of the four stations, A, B, C and D and their relationship to the scattering layer are given in Figure 10.39 and Table 10.63. Catch data for Schilbe in the 2 inch and 3 inch nets are presented in Table 10.45. The highest catch rate with the 3 inch net was observed at station A in 9 m of water, where an average of 8.3 kg of Schilbe were caught per night. The lowest catch was recorded at station C in 19 m of water within the midwater scattering layer. In the 2 inch nets catches varied less but a minimum was again recorded at station C. These re- sults at first suggest that Schilbe were least abundant within the scattering layer, and conflict with trawl results. However it is pos- sible that Schilbe rest during the daytime in the vicinity of the scattering layer, but at night swim into shallower water to feed. If this were the case station A would be the best placed to catch fish involved in such a movement. Results from bottom set gillnets in water deeper than 25 m support evidence from the bottom trawl which indicated that Schilbe was virtu- ally absent from deep water. Catch rates of Schilbe in 2 inch and 3 inch nets at depths of 40, 60 and 70 m were less than 0.5 kg/night. Gillnet catches were very variable within the sheltered waters of Ferguson's Gulf, and it appeared that Schilbe entered the Gulf in large shoals at certain times of the year. The highest catch was re- corded in July 1974 when a 4 inch net caught a total of 17 kg of Schilbe in one night. An analysis of the gonads showed that over 40% of the females were ripe suggesting that these fish had moved into Ferguson's Gulf in the course of a spawning migration. At other times of the year the catch rate fell to below 1 kg/night, indicating that Schilbe were normally not abundant within the Gulf. Beach seines Although Schilbe were occasionally caught in large numbers within Ferguson's Gulf by gillnets, the species was generally uncommon in beach seine catches on all shorelines. Records of seine net catches 1050 in Ferguson's Gulf indicate that this fish was present only between March and July. This coincides with the period when, as noted above, the July gillnet catches included a high proportion of ripe females. Discussion It is suggested in the section on breeding (see page 1055) that Schilbe spawns principally in the River Omo during the flood period, July to September. Ripe Schilbe from the lake probably migrate into the Omo delta but unfortunately it was not possible to sample this area, which lay within the boundary of Ethiopia. Schilbe definitely migrates up the Kerio and Turkwell rivers and fishermen confirmed that large num- bers were at times captured in these rivers during heavy floods. The migratory behaviour of Schilbe would account for the wide variation observed in trawl catches in the Ferguson's Gulf area, and in gillnets set within Ferguson’s Gulf. WHITEHEAD (1959) points out that some species of anadromous fish in Lake Victoria commence shoaling one or two months before the rivers flood. He suggested that pre- spawning assembly perhaps assists synchronisation of breeding in widely dispersed stocks. The indications are that shoaling also occurs in _S . uranoscopus in Lake Turkana. GROWTH Most of the material for biological investigations was caught by bottom trawling but the sample number was generally low. It has proved pos- sible, however, to estimate age and growth by reference to length fre- quency histograms. Schilbe of under 15 cm FL were rare in the bottom trawl but from August 1973 onwards frame trawl catches provided samples of juvenile fish. Sexual differences in length frequency distribution There was a marked difference in the length-frequency distribution of mature Schilbe, females attaining a maximum length of 34 cm, while males rarely grew to more than 21 cm. Figure 10.30 compares data for all males and females caught by bottom trawling during April 1973 in the Longech area. The difference between the means is highly signifi- cant (P = <0.001) and growth of mature males and females therefore has been, analysed separately. The limited data available indicates that there is no sexual dif- ference in the growth of immature fish. Length distributions of a sample of immature males and females caught by frame trawling in April 1974 are compared in Figure 10.31. Modal lengths are identical, and for growth analysis of immature fish the length data for both sexes have been combined. Analysis of length frequency data The length frequency distribution of Schilbe caught by bottom and frame trawling at a number of stations between July 1973 and June 1974 in the Central sector are presented in Figure 10.32. Males and females have been plotted separately and the length distribution of immature fish of both sexes combined is indicated by cross hatching. Histograms have been smoothed using Taylor's method (moving average, a = 3) (TAYLOR, 1965) . In all months the distribution of fish in trawl samples was either unimodal or bimodal. A regular increment in modal length is 1051 clearly apparent in immature fish and females, though not in males. 0-group fish of the 1973 year class were first caught by the frame trawl in November 1973 when at a mean length of 5 cm, and were easily distinguishable as a separate size group ranging from 6 to 8 cm in the December 1973 trawl samples (Fig. 10.32). Fish from this year class were caught in increasing numbers in the following months (Fig. 10.32) and by June 1974 when the sampling programme was completed females of this year class had reached a modal length of 17 cm. Although the sub- sequent growth of these 0-group fish was not studied, results from the previous year provide data for the sequence of growth in late 0-group and 1-group fish. A modal length of 19 cm recorded in 0-group females of the 1972 year class during July 1973 (Fig. 10.32) corresponds close- ly with the value quoted above for 0-group to 1-group in this month. Females of the 1972 year class, which became 1-group fish in September 1973 dominated trawl catches from July 1973 until May 1974. The histograms (Fig. 10.32) strongly suggest that late 1-group and 2-group fish are absent from the trawl samples. Evidence on breed- ing behaviour (see page 1055) suggests that Schilbe migrates away from the trawling grounds in the open lake at the end of the second year of life to breed in rivers. High mortality in the spawning areas would account for the dearth of the older fish. Length-frequency histograms for males indicate that the course of growth is different to females. Definite year classes are distin- guishable (Fig. 10.32) and these show the same pattern of recruitment in 0-group fish and migration in late 1-group fish. However the age classes of mature males show little progressive increase in size from month to month. This is probably a result of lower growth rate in males and they may also commence their spawning migration earlier than females . Estimates of monthly mean lengths of all immature and female Schilbe in the 1972 and 1973 year classes caught by the bottom and frame trawls during the sampling period are presented in Table 10.46. The observed growth of females up to age 2 is shown in Figure 10.33 where mean lengths from Table 10.46 have been plotted against time. The points fall on a line which cuts the time axis in late August or early September. Since seasonal changes in gonad condition indicate that breeding probably coincides with the Omo flood season (i.e. June to October), early September has been taken as the 'birthdate' of Schilbe . Growth over the first year is rapid and a length of approxi- mately 22 cm is reached at an age of 12 months. Over the second year growth increments decline but with the movement of mature individuals out of the population to spawn it was not possible to observe growth beyond 20 months of age. At an age of 24 months, however, the fork length of female Schilbe is estimated to be approximately 30 cm (see also mathematical expression of growth below) . The mean lengths of immature and male Schilbe were similarly esti- mated and are presented in Table 10.47. The observed growth of males up to an age of 24 months is shown in Figure 10.36 where mean lengths from Table 10.47 have been plotted against time. Growth in males is clearly slower than in females, with an increment of only 3 cm during the second year. 1052 Mathematical expression of growth Females A von Bertalanffy growth curve has been fitted to the observed growth pattern given in Figure 10.33 using the following equation: lt = (1 ~ where 1 is the length of fish at age t, is the asymptotic length, K is a constant and t is the theoretical time when the fish has zero length. In the Walford graph (Fig. 10.34) 1^+3 months has been plot- ted against lt months. The following values were obtained for the growth paramters of female Schilbe = 36 cm % K = 1.026 tQ = -0.0208 years The largest female caught by trawling measured 34 cm FL and thus agrees well with observed values. Likewise the value for tQ agrees with the observed time of peak spawning. Substituting the above para- meters in the growth equation theoretical lengths at age can be estima- ted (Table 10.48) and a growth curve fitted to the data is given in Figure 10.33. Theoretical growth agrees well with observed values and in female Schilbe growth clearly follows the von Bertalanffy growth model . Males Values for , K and tQ were similarly estimated for males. In the Walford graph (Fig. 10.35) length at 1 +5 months has been plotted against 1^ using data from Table 10.47. The following values were obtained for the growth parameters of male Schilbe L = 19.5cm OO K = 2.1696 tQ = -0.0175 years The estimate of Leo is rather less than that indicated by field observations and 5% of males caught by the bottom trawl exceeded 20 cm FL. The value for t agrees with the expected 'birthdate'. Sub- stituting the above paramters in the growth equation, theoretical lengths at age can be estimated (Table 10.49) and a growth curve fitted to the data is given in Figure 10.36. Theoretical growth agrees well with observed values and in male Schilbe growth clearly follows the von Bertalanffy growth model. Growth rates in the Northern and Southern Sectors Insufficient data were available from areas outside the Central sector to enable direct growth estimates to be made. However by comparing the length frequencies of trawl samples from the Northern and Southern sectors with samples from the Central sector it was hoped to be able to establish whether growth rates were different. Length frequency 1053 histograms of samples from outside the Central sector were plotted and lengths were found to fall into definite groups as in the Central sec- tor. Mean lengths of females from the Northern and Southern sectors have been compared with those of the equivalent monthly samples from the Central sector by applying t-tests and results are given in Table 10.50. In the Northern sector during all months when samples of suf- ficient size were available, there was no significant difference be- tween the means (P = 0.05). Only two samples of female Schilbe are available from the Southern sector for similar comparisons (Table 10.50). The December 1973 sample was not significantly different to the corresponding Central sector sample, but in May 1973 the mean length of Schilbe sampled from 2 km off the mouth of the River Kerio were significantly larger. The River Kerio was in flood at the time the latter sample was taken and it is possible that these females had recently left the main population in preparation for a spawning migration. Under such circumstances it is likely that they were among the largest individuals and thus unrepre- sentative of Schilbe in the Southern sector as a whole. The similarity of the length-frequency distribution given in Table 10.50 suggest that there is a single population of Schilbe in Lake Turkana. However it is known that relatively small numbers of Schilbe do breed in the lower reaches of the River Kerio and it is possible that these form a breeding population distinct from the main population which spawns in the River Omo . Length-weight relationship A total of 95 Schilbe uranoscopus were weighed, and the length-weight relationship of males and females has been examined using the following equation : 1 og 1 0 wei§ht = lo§l0 a + b (lo8l0 length) where a is the intercept of the regression line of log weight on log length, and b is the regression coefficient. The slopes of the regres- sions for males and femals were compared and found not to be signifi- cantly different (d = 1.273). Data for both sexes were therefore com- bined and the relationship between length and weight for all Schilbe may be represented by the following equation: 1 og i 0 w = 2.9947 (log1Q 1) - 2.0848 or w = 0.00821 l2*9947 where w = weight in g and 1 = length in cm. Estimated values of mean weight for age are given in Table 10.51 for the first 2 years growth in male and female Schilbe. Conclusions The analysis of length-frequency histograms has enabled estimates of growth rate to be made and observed values agree well with theoretical growth based on the von Bertalanffy equation. Schilbe is a migratory fish and the evidence suggests that during the breeding season, large 1054 numbers of mature individuals leave the population of the main lake to spawn in major rivers. As a consequence growth during the second year may be underestimated, especially in males. The length frequency histograms gave no indication of a return of post breeding fish into the trawlable population and it would appear that the majority of Schilbe do not live longer than 2 years. BREEDING Throughout the study period relatively few ripe Schilbe uranoscopus were caught and there is little direct evidence to indicate the loca- tion and time of breeding. The virtual absence of sexually active fish in trawl samples suggests that spawning does not occur in the lake. Previous workers on Lake Turkana (WORTHINGTON and RICARDO, 1936; MANN, 1964) also noted a dearth of ripe fish. In Lake Victoria Schilbe mystus ascends rivers to breed (WHITEHEAD, 1959; CORBET, 1961) and in the River Niger spawning occurs during the flood season (DAGET, 1954). Circumstantial evidence suggests that in the Turkana basin S. uranoscopus also migrates up rivers to spawn. Maturity stages The classification of NIKOLSKY (1963) was used to determine the matur- ity stages of Schilbe. Females Stage _2 Resting or recovering; gonads thin, sac-shaped, pale yellow in colour. Stage 3 Mature: ovaries greatly increased in size, and opaque yellow eggs distinguishable with the naked eye. Stage 4 Ripe: eggs of uniform size but do not separate when ovary wall cut. Stage _5 Ripe-running: not observed. Males Stage 2 Resting or recovering: testes thin with a very slight scalloped edge, just visible to the naked eye. Stage 3 Mature: scalloped edge clearly visible. Testes white or yellow. Stage 4 Ripe: small quantity of sperm released when testes cut and squeezed . Stage 5 Ripe-running: not observed. Sex-ratio Both WORTHINGTON and RICARDO (1936) and MANN (1964) noted that female Schilbe attained a much larger size than males. This was confirmed during the present study as is evident from Table 10.52 where the sex 1055 ratio of fish in the 10 to 31 cm length range from the Ferguson’s Gulf area is given. Males and females smaller than 17 cm were caught in roughly the same numbers. In the size range 17 to 19 cm males outnumb- ered females by over 3 to 1 , but in larger fish the proportion of males declined. The relatively greater abundance of males in the 17 to 19 cm length range is explained by the slower growth rate of males (see page 1050 ). The largest male measured in the Ferguson's Gulf area had a fork length of 24 cm, compared with a female of 34 cm. Size and age at maturity Data relating to length at maturity show that males matured at a slightly smaller size than females (Table 10.52). The smallest ripe male measured 16 cm compared with a minimum length of 19 cm in females. Males reach a length of approximately 17 cm and females 23 cm at an age of 12 months and therefore the majority of Schilbe mature in the second year of life. Location and time of breeding Between May 1972 and July 1974 less than 2% of the 1,300 females and 1% of the 500 males caught by trawling in the Longech area had ripe gonads and the analysis of seasonal changes in gonad condition of fish caught by trawling (Table 10.54) gave no clear indication of a peak of gonadial activity. Ripe females occurred in the open water of the Longech area between March and June and between September and November. In June 14% of females were ripe and it is possible that some breeding occurs in the Central sector at this time. In July 1974 nearly half of a sample of 88 females caught by gillnetting within Ferguson's Gulf were ripe and this sheltered area is perhaps a site of limited breeding activity. Regular monthly samples were not available from the remainder of the lake. In the Southern sector the highest proportion of ripe females occurred in June 1974, when in a sample of 43 females 26% were ripe. In the Northern sector most ripe females were also caught in June, when 14% of a total of 37 females were ripe. The small numbers of ripe Schilbe caught suggest that breeding probably does not occur in the main lake. Even if the major spawning sites were close inshore, where trawling was not possible for technical reasons, appreciably more Schilbe would have been caught. The major breeding sites are probably in the Omo , Kerio and Turkwell rivers, mature fish ascending these rivers to spawn. As mentioned earlier there is ample evidence that Schilbe mystus inhabiting Lake Victoria migrates upstream to breed. CORBET (1955) noted that this fish was caught in an advanced state of maturity in affluent rivers of Lake Victoria at a time when river levels were fall- ing after flooding. WHITEHEAD (1959) however, noted that in another part of Lake Victoria S. mystus asccended rivers at a time when flood- ing began. Migration was of a medium duration, fish ascending a few miles upstream and then passing laterally to spawn in floodwater pools. During July 1972 the Kerio river was investigated as a possible spawning site. It proved impossible to obtain samples in the fast flowing river but basket net catches of the local fishermen confirmed that this species was present, though in small numbers, in the river 20 kilometres upstream from the lake. Three Schilbe were examined and 1056 all were mature females (Stage 3). Fishermen reported that large catches could be made at certain times of the year, and it is probable that the movement of Schilbe into this river is a spawning migration. Unfortunately samples were not obtained from the River Omo which is situated in Ethiopia. This river, on account of its large size, is probably the major spawning area of the Lake Turkana population. Further evidence that Schilbe spawns in rivers was provided by the distribution of juveniles which were only observed in the vicinity of the Omo and Kerio rivers. A sample of 12 fish ranging in length from 27 to 31 mm were caught near the mouth of one of the main distri- butary channels of the River Kerio in October 1974. The river was not flowing at the time of sampling but it is highly probable that these small individuals had been swept down by river flow some time previous- ly,and that the spawning sites were located upriver. Further samples of small Schilbe were obtained by fine shore seines in the Omo delta area. In August 1975 five individuals, ranging in length from 18 to 57 mm were caught at stations 2 to 8 km north of Todenyang and in October 1975 a further two Schilbe of 27 and 29 mm were caught in the same area. On the east shore only one juvenile was caught, at a station on the southern fringe of the Omo delta, during October 1974. Despite extensive sampling with the metre townet in the open waters of the lake, a total of only 8 juvenile Schilbe were caught. Lengths of these individuals ranged from 30 to 44 mm and all were caught during October 1974 within 10 km of the Omo delta. The small numbers of juvenile fish caught in the open lake, and their occurrence in the vicinity of river mouths clearly indicates that breeding takes place in rivers and not in the actual lake. Their time of capture in the north shows that breeding coincides with the flood period, between July and October. Breeding in the two major rivers further south is probably less well defined since flooding in the Kerio and Turkwell may occur between March and October. Results from growth studies (page 1053 ) similarly indicate that the 'birthdate' of Schilbe is in late August - early September, coinciding with peak flooding of the River Omo . FOOD WORTHINGTON and RICARDO (1936) noted that Schilbe uranoscopus appeared to be primarily a predator feeding on Alestes spp, Tilapia spp. and Engraulicypris stellae . The present study shows that Schilbe also feeds on prawns, and that the proportion of fish and prawns in the diet depends on lake depth and on the size of the individual predator. Schilbe has a well defined stomach. During the course of routine field data collection, stomachs were removed, preseved in 5% formalin, and examined later at the laboratory. The composition of the food was evaluated by means of a points system similar to that employed for B. bayad (see page 1036). A total of 690 stomachs were examined, most of them from fish caught by bottom trawling. Gillnetting within Ferguson's Gulf provided information on the diet of fish from shallow water and data from frame trawling enabled the diurnal feeding pattern in the open lake to be examined. In Table 10.55 the number of points scored by each food type has been expressed as a percentage of the total points, combined for all samples during the two year period. 1057 Fish and prawns contributed 98% of the food, and insects appear to be of little importance in the diet of Schilbe . Alestes spp. and Engraulicypris stellae provided over 90% of the total fish intake and other prey species including Barbus cf anema , Haplochromis macconneli and Bagrus bayad formed only a small part of the diet. The proportion of- the two main food items, prawns and fish, was found to vary with depth and predator size. Composition of food in relation to depth Prawns were most important in the diet of Schilbe in the 10 to 20 m depth zone, where they contributed 67% by volume of the total food intake (Table 10.56). Schilbe from water deeper than 20 m had fed almost equally on prawns and fish. The diet of inshore Schilbe both from Ferguson's Gulf and from shallow water within the 10 m contour was markedly different with fish constituting the dominant food, and prawns forming only a minor proportion of the diet. This was corre- lated with a general scarcity of prawns inshore. Diet with relation to size Individuals under 20 cm had a varied diet feeding mainly on prawns with a minor proportion of fish (Table 10.57). A few ostracods were also included in the diet of the under 15 cm length group. Higher propor- tions of fish were observed in the food of 20-24 cm Schilbe but prawns still formed three quarters of the total intake. The relative import- ance of fish and prawns was reversed in the largest Schilbe (over 25 cm) with fish now contributing three quarters of the food. A tendency for fish prey to change from Engraulicypris stellae in smaller Schilbe to Alestes spp. in larger specimens was also noted. Diel feeding pattern The diurnal feeding pattern of Schilbe was examined both in the shallow water of Ferguson's Gulf and in the open lake. Investigations in Ferguson's Gulf were carried out concurrently with studies on Synodon- tis schall and the methods used will be discussed below (see page 1078 Schilbe inhabiting the Gulf were mainly piscivorous and results from the 24 hour study showed that feeding occurred primarily during the hours of darkness. Changes in diet and stomach fullness, together with catch rate are given in Table 10.58. The number of points for each food type has been expressed as a percentage of the total number allot- ted during each time phase, and the catch rate is given as numbers of Schilbe caught per hour in the 4 inch gillnet. The results show that the diet varied little through the 24 hour period, with fish forming over 90% of the food intake throughout. However, Alestes spp. were more commonly eaten at night than during the day. The diurnal feeding pattern contrasted with Synodontis schall where there was a marked change in the diet from day to night (see page 1078). Catch rate and stomach fullness, however, fluctuated considerably in Schilbe with peak catches, of 10 fish per hour, occurring at dusk and the lowest catches at dawn and during the morning. The results in- dicate that Schilbe was most active at sunset and during the early part of the night. Feeding followed a similar cycle with maximum stomach fullness being recorded at night. Schilbe inhabiting Ferguson's Gulf thus appear to have a definite feeding rhythm with peak activity between sunset and dawn. The increase in catches at dusk and during the night may also be partly due to the migration of Schilbe inshore. 1058 It is of interest to compare catch rates of Schilbe with Hydro- cynus f orskalii, another common predator, over the same 24 hour period (Fig. 10.37). The observations show that the two species were caught at entirely different times. Hydrocynus f orskalii is an active 'chasing' predator relying on sight for the capture of prey. The high- est H. f orskalii catches were made during the hours of daylight, indi- cating peak activity during the period when Schilbe was least active. Hydrocynus forskalii was virtually absent from gillnets during the night. Although temporarily separated and with different feeding methods, the two predators competed for the same, food resources. Engraulicypris stellae and small Alestes spp. were the dominant prey species in each case. To investigate the vertical migration and diurnal feeding pattern of Schilbe in the open lake frame trawl samples were made over a 24 hour period at a lake depth of approximately 25 m, 2 km east of Longech Spit in April 1974. A series of hauls was repeated at the following intervals during the 24 hour period; afternoon, dusk, night, dawn and morning. Each series consisted of seven 10 minute hauls covering seven depth zones from the surface to the bottom. The catch rates (Table 10.59) prove that Schilbe undertakes a vertical migration, leaving the lake floor at dusk and occurring mainly in the midwater layers during the night. At dawn Schilbe returns to the lake bottom remaining there during the daylight hours. The migration probably occurs in response to falling light intensity. Evidence from frame trawl hauls further offshore (see page 1049 ) suggests that at greater depths Schilbe re- mains in the midwater layers during daylight hours, where optimum light conditions prevail, and do not descend to the bottom. The results of stomach analysis for the 24 hour study are given in Table 10.60. The composition of the food again showed little vari- ation over the 24 hour period, but unlike the inshore Schilbe consider- ed above, prawns formed the major part of the diet. Peak stomach full- ness was recorded during the night and dawn phases, coinciding with the time when most Schilbe were in the midwater layers. The vertical movement that commenced at dusk thus coincided with an increase in feeding activity. Conclusions Results from the present study agree to some extent with the findings of earlier workers (WORTHINGTON and RICARDO, 1936; MANN, 1964). Larger individuals were found to prey on other fish, in particular Engraulicypris stellae and Alestes spp. However, prawns were of equal or greater importance in the diet of fish smaller than 25 cm, an obser- vation not previously recorded. Diet was also found to be dependent on depth, fish being more commonly eaten in the shallow inshore water. Schilbe appeared to feed primarily at night. In the open lake a vertical migration commenced at dusk, and fish occupied the midwater layers throughout the night. By dawn stomachs were relatively full and fish returned either to the bottom or in deeper water to the lower limits of the scattering layer, where they remained during the day. In Ferguson’s Gulf gillnet catches showed that a similar pattern of feeding activity occurred with few Schilbe being caught during daylight hours . The closely related Schilbe mystus , a widely distributed nilotic species feeds principally on small fish (Haplochromis spp) in the case 1059 of the Lake Victoria population, although insect larvae and prawns were also eaten (GRAHAM, 1929; WORTHINGTON, 1929; CORBET, 1961). Most in- sects were taken at the surface and consisted of terrestrial forms. In Lake Turkana there is no evidence to suggest that Schilbe uranoscopus fed regularly at the surface of the lake. Insects made up less than 2% of the diet and were mostly derived from on or near the lake bed. COMMERCIAL FISHERY Schilbe does not feature in the commercial catch and results from the present study show that this species is likely to be of only slight importance in any futute development of the Lake Turkana fisheries. Although Schilbe is a widespread species it is generally not abundant and catches in both bottom trawl and gillnet are generally low. The present catch rate of Schilbe could be increased by fishing with gillnets of smaller mesh size. Experimental results have shown that with a graded fleet of 2 to 9 inch nets highest catches are gener- ally made in 3 inch nets. Under the present conditions however the introduction of 3 inch nets would be most undesirable since the immature of numerous commercially valuable species would be destroyed. The mesh selectivity of nets from 2 to 4 inches, stretched mesh, and mean retention lengths, are given in Appendices 10A and 10D. Since males rarely grow larger than 24 cm, virtually the entire catch of Schilbe in the 2 and 3 inch nets would be females. Fifty percent maturation (Stage 3) in female Schilbe occurs at a length of 24 cm and the 3 inch net with a mean retention length of 30 cm would catch mainly mature fish. M0CH0KIDAE SYNODONTIS SCHALL Over fifty species of Synodontis have been recorded from Africa but only two occur in Lake Turkana. One of these, Synodontis schall is very common and widespread throughout the lake whereas the second species, Synodontis f rontosus was only rarely caught in the Omo Delta during the present study, and appears to be restricted to a riverine environment. Several species of Synodontis have been studied in detail in other parts of Africa. BISHAI and GIDEIRI (1965 a and b, 1968) in- vestigated the biology of certain species, including S. schall , occur- ring in the River Nile at Khartoum, and WILLOUGHBY (1974) has studied the biology of several Synodontis spp, again including S. schall, from Lake Kainji. Both WORTHINGTON and RICARDO (1936) and MANN (1964) made valuable observations on the Lake Turkana population of S. schall . However, a more detailed study of the biology of this species was re- quired because of its importance in subsistence fisheries. S . schall is probably the species most commonly eaten by the fisherman of Lake Turkana and their families. Moreover , since the latter half o’f 1974, increasing quantities have been smoked in the Longech area by Luo fish traders for export to markets elsewhere in Kenya. DISTRIBUTION Synodontis schall is an important species occurring in many of the major river systems in the soudanian region including the following 1060 river basins: Nile, Niger, Volta and Senegal (BOULENGER, 1911). The species attains a maximum fork length of 37 cm in Lake Turkanan where it is widely distributed, occurring in 85% of all bottom trawl samples made between May 1972 and July 1974. Greatest numbers were generally caught by bottom trawling and by gillnetting in inshore areas, where the midwater scattering layer was well marked and in close proximity to the lake bed. It was surprising to find that S. schall was not ex- clusively demersal in distribution, but was caught regularly in mid- water trawls and in gillnets set at the surface even in the deepest areas of the lake, far offshore'. Bottom trawling The mean bottom trawl catch per effort (corrected to the catching effi- ciency of the two-seam trawl, see page 587 ) in twelve depth zones was estimated for the Northern, Southern and Central sectors of the lake as shown in Table 10.61. The Southern sector had relatively lower densities of S. schall than elsewhere. Only in the Kerio-Turkwell area, between 20 and 25 m, was the catch rate higher than the lake average of this depth. In the Northern sector catches in the shallow water of Allia Bay were high in comparison with the remainder of the sector, with peak catches in water of less than 10 m. The environment of Allia Bay, with extensive beds of Potamogeton spp. is atypical of shallow water elsewhere in the lake. Under these conditions S. schall were found to feed largely on aquatic insects and ostracods. Copepods, important food items in other shallow water areas, were absent from the diet of Allia Bay fish (see page 1077) The distribution of S. schall in the remainder of the Northern sector and in the Central sector were basically similar. Catches de- clined from inshore areas to the deeper water and few fish were caught below the 40 m contour. Abnormally high concentrations of S. schall were occasionally found in the Omo delta area in the vicinity of the lake water-river water interface. For example, in February 1973 a catch of 240 kg/hr was recorded in 30 m of water at the interface, a value seven times the annual mean for this depth. Exceptionally large numbers of S. schall , with catches as high as 581 kg/hr were sometimes recorded in the Ferguson's Gulf area. Table 10.62 represents monthly data for the 15 to 20 m depth zone of the Ferguson's Gulf area over a period of two years. Peak catches were recorded between January and March in both 1973 and 1974. Bottom trawling has shown that in all three sectors of the lake S' schall was most abundant in water shallower than 25 m. In the Northern sector (excluding Allia Bay) peak catch rate was recorded in water of 10 to 15 m. Further south, in the Central sector the fish was most abundant in water of 15 to 20 m whilst in the Southern sector maximum numbers were found between the" 20 and 25 m contours. The results show a correlation between the distribution of S. schall in relation to depth and the postion of the midwater scattering layer (see page 597 for details of this layer). In peripheral areas of the lake where the bottom rises above the 20 m contour the scatter- ing layer lies in close proximity to the lake bed and catches of S. schall were generally high. The stomach contents of those fish caught in or just below the scattering layer show that feeding has occurred on copepods and ostracods. Results from plankton surveys (A J D FERGUSON, pers. comm.) however, show that zooplankters are not particu- 1061 larly concentrated in the scattering layer, although densities tend to increase in shallow water. As mentioned previously (see page 597 ) the depth of the midwater scattering layer depends to a large extend on light penetration. In the north where water is generally turbid , the upper limit of the layer may be within 2 m of the surface. In the south, water is clearer and the layer may descend to a maximum of 30-35 m. Light penetration has been measured at several stations in the three sectors and the mean depth at which 99% light extinction occurs is given in Figure 10.38 for each of the sectors. The depth at which maximum catch rate occur- red for S. schall in the bottom trawl is also included. Peak catch rates occurred at approximately 10 m below the 99% light extinction depth in all three sectors . These results indicate that the distribu- tion of S_. schall in inshore areas depends on the depth of light pene- tration, perhaps through the influence of light intensity on the depth at which the scattering layer occurs. Allia Bay, where S. schall were caught in large numbers above the scattering layer, was exceptional, possibly as a result of the abundance of food other than copepods in this area. Gillnetting To examine the association between the scattering layer and the distri- bution of _S. schall , a programme of gillnetting was established in the open lake off Ferguson's Spit between January and June 1974. Fleets of bottom-set gillnets were fished at four stations, chosen for their position relative to the scattering layer. The layer was well marked in this area and its depth was determined by reference to the echo sounder on board R V HALCYON. The approximate position of the scatter- ing layer and depths of the four stations are shown in Figure 10.39 (see also Table 10.63). Mean weights of fish caught in the 2, 3 and 4 inch nets per 24 hour period are given in Table 10.64. Since the gill- nets were relatively unselective for the size range of S. schall occu- rring in the samples, the combined results for the three nets will be considered below. Station C which sampled within the scattering layer had the high- est catch rate of S. schall with a combined mean of 39-4 kg per night for the three nets. The catch was lowest at station A, above the scattering layer, with only slightly higher catches at station B. Below the scattering layer, at station D, catch rates were nearly as high as in station C. These results reinforce evidence suggesting that inshore, S. schall is found in greatest numbers in close association with the midwater scattering layer. Bottom trawl results (Table 10.61) have shown that below the 20 m contour the catch rate of _S. schall steadily declined in the Central sector of the lake. However, evidence from surface-set gillnets indi- cates that S. schall is not confined to the bottom. Table 10.65 com- pares the catch rate (kg S. schall/50 yards net/24 hours) of 4 inch (124 mm) gillnets set at the surface, in midwater and at a depth of 73 m on the bottom at an offshore station. The results show that in the deeper water of the lake, S. schall were more common at the surface than in midwater or on the bottom. Since the trawl data proves that S. schall were rare on the bottom in deep water these fish had either migrated away from the inshore areas on feeding forays or were perma- nent inhabitants of the midwater layer. 1062 Frame trawling Results from frame trawl catches were in general agreement with gillnet data and indicated that S. schall was not numerous in midwater, except in shallower water where the scattering layer approached the bottom. Data from a series of daytime frame trawl hauls carried out in Septem- ber 1973 at an inshore and at an offshore station in the Ferguson's Gulf area are presented in Table 10.67. Inshore, _S. schall was well distributed from a depth of about 4 m to the bottom. Peak numbers occurred within the midwater scattering layer. At the offshore station the species occurred only in small numbers and was restricted in verti- cal distribution to the region of the scattering layer at a depth of 8 to 16 m. Further results from frame trawling carried out over a 24 hour period at a station in 24 m of water are presented in Table 10.68. Catch rate is given as numbers of fish caught per 10 minute haul for each depth zone at 5 intervals of time. _S. schall were caught in rela- tively small numbers and the results show that there are no marked trends. However, it may be significant that the only period when S. schall occurred in the top 12 m was during the night, suggesting that a vertical feeding migration may occur. Beach seining Larger numbers of S. schall were regularly caught by beach seining within the confines of Ferguson's Gulf. In other areas of the lake the species was caught over a variety of substrates by this method, including gravel and sand. No beach seining was possible on rocky shores but occasional S. schall were caught on rod and line in such areas. These data suggest that S. schall has no preference for a particular substrate although the species was not common in exposed situations on the western shore. Conclusions Synodontis schall has a fairly complex distribution in Lake Turkana and, although predominantly a demersal species during the hours of day- light, it also occurs regularly in the midwater scattering layer of the lake. It is found mainly in water less than 25 m deep, with particu- larly dense concentrations occurring where the midwater scattering layer meets the lake bottom. The depth at which maximum densities of S. schall occur thus varies with the position of the scattering layer which is in turn dependent on light penetration. GROWTH Growth studies on Synodontis spp in other areas of Africa have been based on the analysis of vertebral growth checks. BISHAI and GIDEIRI (1965a) found that growth rings in S. schall from the River Nile were laid down annually as a result of breeding and reduced feeding during the flood season. WILLOUGHBY (1974) working on Lake Kainji concluded that vertebral growth checks observed in several species of Synodontis were formed during the breeding season. In Lake Karibia, CHITRAVADI- VELLI (1974) working on Synodontis zambezensis and FRANK (1974) working on Synodontis nebulosus , both estimated age with the aid of vertebral growth checks. Attempts to age the Lake Turkana population of S. schall using vertebrae failed since, although checks were present, no correlation 1063 between their formation and age could be established. However, an inspection of length-frequency data revealed that the method of Petersen (TESCH, 1968) for age and growth estimation could be used. Large samples of mature S. schall were regularly obtained by the bottom trawl but unfortunately fish smaller than 15 cm were only rarely caught. As a result growth estimates of immature fish are less reli- able than for adults. Method Material for age and growth studies was obtained over a 26 month period from regular monthly trawl samples made in the Ferguson's Gulf area at a depth of between 20 and 30 m. Although the bottom trawl caught few S' schall smaller than 15 cm FL it is considered that the gear was probably non-selective down to an appreciably smaller size. This was in fact proved by exceptional samples containing significant numbers of fish as small as 6 cm as demonstrated in Figure 10.40. Thus S. schall probably migrated into the trawl sampling areas on attaining a length of between 15 and 20 cm. It is unlikely that there was any size- dependant escapement of the largest S. schall from the trawl and random samples of the open lake population were probably obtained. All fish were measured to the nearest cm below fork length and a considerable number were also sexed to enable possible differences in growth rate between males and females to be assessed. Fish caught outside the Longech area have not been included in the growth analysis since the time of peak spawning may have been slightly later in more northerly areas of the lake. Data from other sectors was too scanty to enable separate analysis of growth to be made. Sexual differences in length frequency distribution Monthly mean lengths of all males and females caught by bottom trawling in the Longech area between July 1972 and July 1973 have been compared by applying Students t-test, and the results are given in Table 10.74. In most months there was no significant difference, but during March, April and May 1973, and in November 1972 females were significantly larger than males. The discrepancy noted in March, April and May coincided with the main breeding season. As shown later (page 1069) the proportion of males increased at inshore spawning sites during this time, indicating that they remained on the breeding grounds for longer periods than the females. It is reasonable to assume that the ripest fish were also the largest and their departure from the main population would result in the observed fall in relative size in the regular trawling area. The second discrepancy noted in November 1972 may also be a result of spawning movements since during this period further breeding activity was observed. Seasonal changes in length-frequency distribution Since mean lengths of males and females were not significantly differ- ent for most of the year, data for the two sexes, covering the period June 1972 to February 1974, have been combined for growth analysis. Length-frequency histograms were plotted for the monthly samples (Fig. 10.41) and examined in sequence. Histograms have been smoothed (moving average a = 3) using Taylor's method (TAYLOR, 1965). Pronounced modes were found to progress from month to month and a general pattern 1064 through the year was apparent: 1. From November to May, the length-frequency distribution of fish in the trawl samples was unimodal. Throughout this period the modal length increased regularly each month. 2. In June each year samples were augmented by the influx of smaller fish and the length-frequency distribution became bimodal. This was obviously the result of recruitment of fish smaller than 22 cm into the trawlable position. The length-frequencies of samples obtained between June and November exhibited one of the following distributional patterns : (a) Unimodal with the mean length usually greater than 25 cm (b) Unimodal with the mean length usually less than 22 cm (c) Bimodal in which both length groups of fish were present These features are highlighted in Figure 10.42 where length- frequency distributions from three separate samples made in the Ferguson's Gulf area during August 1973 are shown. 3. After September the larger fish became relatively fewer and by October-November the smaller length group was dominant in the trawl catches. A similar sequence was repeated both in 1972 and 1973 indicating clearly that the recruitment of young fish into the adult population was a regular annual event. The pattern of recruitment suggested that breeding was restricted to a definite season each year. This was con- firmed by an anlysis of seasonal changes in maturity to be discussed below (see page 1068). The main breeding season in the Ferguson's Gulf area extended from March to July with a peak in May and mid May is re- garded as the 'birthdate' of the population under investigation. Analyses of growth (see below) prove that the influx of small _S . schall into the trawling areas during June occurs when the fish are approximately 13 months old. The bottom trawl catches are therefore composed chiefly of 1-group and 2-group fish. The length-frequency histograms suggest heavy mortality in 2-group fish during the period August-November each year which corresponds to the end of the breeding season . The whereabouts of the main population of 0-group fish is at present unknown. Exceptionally, during October and November, they may occur in trawl samples. Figure 10.40 represents a sample taken in November 1973 from a depth of 16 m where in addition to the usual size group of fish over 20 cm in length, a group of smaller fish with a modal length of 10 cm appears. Small numbers of S. schall of 2-15 cm also occurred in shore seine catches particularly in more northerly areas of the lake. It is possible that 0-group S. schall live mainly inshore within the 5 m contour. 1065 Estimates of growth rate As a preliminary to the calculation of monthly growth increments, the mean lengths of fish in each year class were calculated for all stations of sufficient sample size. Since the majority of samples had a unimodal size distribution this presented no problem. In samples where the length distribution was bimodal the peaks of the two year classes were separated by eye and the means of each length group were then calculated. Table 10.70 provides additional information on the 1971 year class. Stations which provided the routine samples shown in the monthly histograms (Fig. 10.41) are marked with an asterisk. Each of the routine samples has been compared with samples from other trawl stations in the Longech area made during the same month. Results of t-tests showed that only at 4 out of 15 stations were mean lengths of the additional samples significantly different to the routine monthly samples (P = < 0.05). All monthly data from the 1971 year class was therefore combined to give the mean lengths presented in Table 10.71. Results for the 1970, 1972 and 1973 year classes were treated in a similar way and are aslo included in Table 10.71. The mean lengths (Table 10.71) compare well with values obtained for the corresponding age group during the appropriate month in other years indicating that a regular seasonal pattern of spawning, recruit- ment and growth occurred. In Table 10.72 data for the four year clas- ses has been combined to give mean observed lengths up to an age of 29 months. Mathematical expression of growth A von Rertalanffy growth curve has been fitted to the length for age data shown in Table 10.72 using the following expression: 1 = L (1 - e-K(t-t0) t CO where L is the asymtotic length, K is a measure of the rate at which L is approached, 1.. the length at any age t and t_ is a constant. co t. O The Walford plot showing the relationship between lt and ]$ + 2 months is presented in Figure 10.43. The following values were ob- tained for the growth parameters of S. schall : L = 3 2 cm 00 K = 0.735 t = -0.175 years The estimate of agrees well with field observations where the largest fish caught by trawling was a female of 35 cm and only 0.4% of all fish in the trawl samples exceeded 32 cm. The value of -0.175 years for t indicates a birthdate in mid-March, two months earlier than expected. The growth parameters were used in conjunction with the von Bertalanffy growth equation to estimate length for age and observed lengths are compared with estimated lengths in Table 10.73 for the first three years (see also Fig. 10.44). 1066 There is close agreement between the estimated and observed lengths indicating that the application of the von Bertalanffy growth t equation is appropriate. Age composition of trawl catches The relative abundance of four year classes of S. schall in the monthly trawl samples between July 1971 and February 1974 are shown in Figure 10.45. 0-group fish were generally absent from the trawl catches. Re- cruitment into the trawlable population occurred at an age of approxi- mately 13 months and fish of 1-group dominated the trawl catches. Rela- tive numbers of 2-group fish declined, probably as a result of heavy mortality during the spawning and post spawning period. Individuals older than about 30 months were tco few in number to be distinguished in length frequency histograms. Older fish almost certainly occurred in trawl samples but it proved impossible to separate them from 2-group fish. Trawl catches between December and May each year consisted of a single year class, the 1-group fish. Length-weight relationship Observations on the length-weight relationship in S. schall are based on data from fish caught in routine monthly trawl samples from the Longech area. During the two years sampling period approximately 1,000 S. schall were weighed. The relationship between length and weight can be represented by the equation: 1 og 1 0 wei§ht 3 lo§l0 a + b (lo§io lenSch) where a is the intercept of the regression line of log weight on log length, and b is the regression coefficient. Using the above express- ion, the length-weight relationship for mature and immature fish were estimated and the slopes of the regression lines were compared (Table 10.74). The results show that in S. schall there is no significant sexual difference in length-weight relationship and no difference be- tween mature and immature fish. Data was therefore combined and the relationship between length and weight for all S. schall may be repre- sented by the following equation: log10 w = 3.001 (log10 1) - 1.679 or w = 0.0209 l3'001 Estimated values of mean weight for age are given in Table 10.75 for the first three years of growth. Discussion Age determination during the present investigation was based on the analysis of length-frequency distributions in the absence of seasonal growth rings on skeletal structures. The Lake Turkana population ap- pears to have a considerably higher rate of growth compared with S. schall studied elsewhere in Africa (Table 10.76). Thus in Lake Kainji WILLOUGHBY (1974) estimated that fish grew to a length of only 14.5 cm in 3 years, and matured at a length of only 12-14 cm (cf 22 cm in Lake Turkana). 1067 BISRAI and GIDIERI (1965a) estimated different growth rates for males and females in the River Nile and found that growth was more rapid during the third than in the second year. WILLOUGHBY (1974) sug- gested that an incorrect interpretation of growth rings might explain this anomaly. High, relatively uniform temperatures may explain the enhanced rate of growth in the Lake Turkana population compared with elsewhere. BREEDING Routine data on sex and gonad condition were collected at each trawl station. Additional information was obtained from gillnets and beach seines. Seasonal changes in maturity and sex-ratio have been studied and the time and place of breeding has been established. Maturity stages The sexes in S. schall are easily distinguishable since mature and maturing males have noticeably villose testes. The presence in males of a muscular protruberance or papilla anterior to the urinogenital opening enabled sexes to be separated externally. The papilla, gener- ally less than 1 cm in length, becomes suffused with blood during spawning and on males caught at the spawning grounds it was well vascularised and pinkish in colour. WILLOUGHBY (1974) noted that the papilla on Synodontis spp. in Lake Kainji enlarged slightly during the breeding season. Nikolsky’s classification of gonad condition (NIKOLSKY, 1963) was used to describe maturity stages in S. schall . Females Stage 2 Resting or recovering: ovaries sac shaped, usually less than 5 cm in length. Translucent in maturing fish, but pink in post-breeding individuals. Stage 3 Maturing: ovaries yellow in colour, extending more anteriorly and of greater cross-sectional area. Ovary wall semi-transparent and slightly vascularised with small blood vessels visible. Developing eggs 0.4 to 0.8 mm in d i ame ter. Stage 4 Ripe: ovaries up to 8 cm in length, filling most of body cavity. Eggs clearly visible through well vascularised ovary wall. Eggs opaque white or yellow, occasionally greenish, and up to 1.0 im in diameter. Stage 5 Ripe-running: eggs 1.2 to 1.4 mm in diameter, trans- lucent grey in colour. Slight pressure on belly causes extrusion of eggs. Stage 6 Spent: ovaries flaccid in appearance and pink in colour with residual eggs few in number. 1068 Males Stage 2 Resting or recovering: testes a pair of thin strands, translucent grey in colour. In virgin males lateral edge of the testes slightly scalloped. In recovering males testes flattened, villose protrusions clearly visible. Anterior villi up to 2 mm in length, larger and more vascularised than posterior villi. Stage 3 Maturing: villi more prominent, up to 15 mm in length and 2 to 3 mm wide. Anterior villi translucent yellow, posterior opaque pink. Stage 4 Ripe: anterior villi whitish in colour, sperm released when cut and squeezed. Stage 5 Ripe-running: anterior villi turgid in appearance. When cut copious milt produced. Posterior villi pinkish grey in colour and less turgid. Size at maturity Data relating to size at maturity, based on material from trawl samples in the Longech area are presented in Table 10.77. Males started matur- ing at 16 cm and 50% had reached maturity at a length of 19 to 20 cm. The smallest observed ripe male (stage 4) had a fork length of 19 cm. Maturation in females commenced at 17 cm, slightly larger than was observed in males, and 50% had reached maturity (stage 3) at a length of between 20 and 21 cm. The smallest ripe female measured 19 cm. The length for age estimates presented above indicate that 25% of males and 5% of females mature at the end of their first year and the remaining fish mature soon afterwards, early in their second year of life. Table 10.78 compares the size at first maturity and breeding of S. schall in Lake Turkana with populations in the River Nile (BISHAI and GIDEIRI, 1968) and in Lake Kainji (WILLOUGHBY, 1974). Maturation and breeding clearly occurs at a much larger size in Lake Turkana, probably a result of their more rapid growth rate (see page 1065). Relatively larger size at maturity has been noted in other species in- habiting Lake Turkana. HOPSON, J (1975) showed that Alestes baremose grow more rapidly and matured at a larger size in Lake Turkana than in Lake Chad and HOPSON, A J (1975a) made similar observations on popu- lations of Lates niloticus in the two lakes. Seasonal changes in maturity Central Sector Seasonal changes in maturity were examined with the aim of establishing the time and duration of spawning. The majority of data came from the Central sector and results are presented in Tables 10.79 and 10.80. Ripe males and females were caught during all months of the year, and the proportion of ripe females over 20 cm did not fall below 10%. Fewer ripe males were caught and gonadial activity was confined mainly to two periods, May-June and September. A peak spawning season was not apparent in samples from the main lake. 1069 However an investigation of the changes in maturity stages of fish caught inshore in the vicinity of the spawning grounds showed that S. schall had a definite breeding season. WORTHINGTON and RICARDO (1936) suggested that Ferguson's Gulf may have been one of the breeding sites of S. schall since the majority of ripe fish examined during the Cambridge Expedition came from here. The following evidence from the present study confirms that Ferguson's Gulf is a major spawning area: (1) The occurrence of large numbers of ripe-running fish in each seine at certain times of the year. (2) An increase in the proportion of males during this time. (3) The occurrence of S. schall eggs on the floor of the Gulf. (4) The capture of females in a spent condition. Spawning grounds were located in water of less than 1 m depth, in areas where freshwater run-off entered the lake after heavy rain. Seasonal changes in conductivity and turbidity, associated with the flooding, were noted in these areas and both factors may be important in inducing S. schall to breed. Flooding brought about a fall in con- ductivity, from the usual level of 3,700 yS/cm to as low as 2,000 uS/cm at the mouths of temporary rivers entering the gulf. Turbidity result- ing from the suspension of plant debris and soil particles derived from the land was also much greater. Male and female _S. schall with ripe-running gonads (stage 5) were caught in large numbers with beach seines under such conditions. Table 10.81 examines the seasonal changes in gonad condition of all males and females larger than 20 cm caught in Ferguson's Gulf. Peak maturity was reached in May, with 74% of females, and 66% of males in a ripe or ripe-running condition. In the main lake ripe and ripe-running fish never reached such high proportions. A second but less pronounced peak of spawning activity also occurred in October when 25% of females were ripe. Both these peaks of activity coincided approximately with the two periods of the year when rainfall and hence flooding was most likely. As discussed earlier (see page 31 and Table 10-2) 60% of the rain recorded at Ferguson's Gulf over a period of five years fell be- tween March and June, with October and November accounting for a fur- ther 15%. Results from the main lake do not conflict with the conclusions drawn from the spawning area in Ferguson's Gulf, although differences are evident. As noted above, in the Central lake sector an appreciable percentage of the population were ripe in all months of the year. It is perhaps an advantage in an area where rainfall is rather unpredict- able that a certain percentage of the population should be ripe at all times so that spawning can take place whenever the necessary conditions arise. During the principal rainy season, March to June, ripening fish probably leave the main population and move onto the spawning grounds, thus accounting for the absence of enhanced gonadial activity between March and July in samples from the main lake. Northern Sector Results from the Northern and Southern sectors of the lake are incom- plete since it was not possible to sample these areas every month. Limited data from the north (Table 10.82) indicate that spawning was 1070 more continuous over the period April to November than was noted in the Ferguson's Gulf area. In the large samples obtained in April 1973, over 80% of the females larger than 20 cm were ripe. The proportion of ripe male fish was low, probably due to their concentration on the actual spawning sites elsewhere. However data from metre townet sam- ples taken in shallow open water within 10 km of the Omo delta (Table 10.83) showed that large numbers of S. schall post larvae were present during the period June to October. No ripe males and only a few ripe females were noted in September 1973 (Table 10.82) but over 50% of both males and females were ripe in catches made during November 1972, indi- cating that circumstances varied from year to year. Southern Sector Data from the south is presented in Table 10.84. The results indicate that as in the other sectors, spawning occurred in May and June. In June 1972 and May 1973 more than 40% of the fish examined had ripe gonads. Results from October 1973 however indicate that unlike more northerly areas, breeding activity was at a minimum in the Kerio- Turkwell area. The Kerio river flooded frequently between April and October and peak spawning in S. schall populations near the river mouth probably occurred during the earlier months of the flood season. It is concluded from gonadial data that peak spawning in S. schall occurs between late March and June in the Longech area, during the main rainy season. In the Northern sector breeding activity is more prolonged and may continue until October and November. Spawning sites There is ample evidence to show that S. schall not only breeds in inshore areas of the lake near the mouths of rivers, but also enters rivers to spawn upstream. Post larvae of S. schall were caught at the mouth of the River Kerio in July 1972. The small fish ranged from 11 to 15 mm in length, and from laboratory observations on larval growth (see below) they were approximately two weeks old. After hatching some distance upstream, the larvae probably remained in the numerous chan- nels of the delta, gradually drifting downstream in the current. R R McCONNEL (pers. comm.) of the Kenyan Fisheries Department has obser- ved S. schall several kilometres up the River Kalokol , a minor river which usually flows into Ferguson's Gulf for a period of only two or three days after heavy rain. Small numbers of S_. schall may therefore enter many of the similar small rivers feeding into the lake during temporary flooding and under favourable conditions, breeding probably occurs. Unfortunately it was not possible to obtain samples from the valley of the River Omo which is probably of major importance as a spawning area. Synodontis schall is not entirely dependent on flood conditions for spawning as shown by the breeding population in Crater Lake A on Central Island. Nets set in this lake caught several S. schall, the smallest measuring 10.6 cm FL. Since this lake has been separated from the main lake for at least 70 to 80 years spawning must have occurred within the crater lake itself. Here conditions resembling flooding were unlikely, since rain water run-off would have come from a rela- tively small area. Changes in conductivity and turbidity would have been minimal and yet breeding had evidently occurred. 1071 It is clear from the present study that S. schall has a wide range of tolerance to environmental conditions at the spawning site. Breed- ing may occur during the flood period either in the actual rivers or in shallow inshore areas influenced by flood water entering the lake. There is also evidence to show that isolated populations of S_. schall may breed successfully in small lakes away from the influence of large- scale flooding. Stimulation of breed i ng Breeding in S. schall is clearly related to the influx of freshwater into the lake under flood conditions. The specific factor or combina- tion of factors which stimulate breeding is uncertain, but it may in- clude a fall in conductivity, an increase in turbidity or the inflow of petrichor (see page 50 ) into the lake. The commencement of spawn- ing in the Ferguson's Gulf area coincided with rainfall and breeding sites were situated close inshore in the vicinity of fresh water run- off entering the lake after heavy rain. Elsewhere in Africa S. schall also breeds during the rainy or flood season. BISHAI and GIDEIRI (1968) reported that in the Nile S. schall bred between July and Octo- ber while the river was flooding, and in Lake Kainji the initial ap- pearance of maturing specimens of S. schall could be linked with the first local rainfall (WILLOUGHBY, 1974). Sex-ratios The ratio between numbers of male and female S_. schall caught by bottom trawling in the Central sector is compared in Table 10.85. The sex- ratio, expressed as numbers of males per 100 fish is given for each cm group in the length range 15 to 34 cm. The results show that below 21 cm there was a slightly greater number of males than females. The proportion of males declined progressively at lengths over 21 cm. In Table 10.86 the ratio has been examined on a monthly basis in the 20-24 cm and 25-29 cm length groups. Variation was noted from month to month but no definite relationship between the peak spawning period (March to June) and relative numbers of males was observed. Within Ferguson's Gulf the proportion of males was somewhat higher than in the main lake. This is demonstrated for mature fish in Table 10.87. Although there was no marked increase in the proportion of males in Ferguson's Gulf as a whole, during the breeding season the ratio was observed to rise to as high as 11:1 on the actual spawning sites. The higher proportion of females in samples from the main lake may indicate that males spend a greater period of time in the vicinity of the breeding grounds. However a higher male mortality is also likely since there was no apparent increase in the proportion of males in the main lake after the breeding season. These results are in agreement with the observations of WILLOUGHBY (1974) and BISHAI and GIDEIRI (1968) who found significantly more females than males in Lake Kainji and the River Nile. Condition Values for the expected weight of all weighed fish were calculated using the regression equation for the length-weight relationship derived previously (see page 1066). w = 0.029 1 3.001 1072 These results were subsequently used to obtain a condition factor (k) for each fish by substituting in the following equation: k = wQ x 100 w where wQ is the observed weight and w is the expected weight derived from the above length-weight regression equation. Mean monthly condi- tion factors for all fish larger than 20 cm caught in the Longech area are given in Table 10.88 and presented graphically in Figure 10.46. Although the condition factor was almost always slightly higher in females than in males the seasonal pattern is similar in the two sexes with the exception of samples for May where a marked fall in condition was noted only in males. Condition was at a seasonal peak between September and November. The peak corresponded to the recovery period at the end of the breeding season. Relatively low values between December and March may be due to the replacement of 2-group fish in the trawl samples by 1-group fish which had not yet built up reserves of fat. The increase in the condition of females during the breeding season between April and June is probably the result of the additional weight of the ovaries. WILLOUGHBY (1974) found that in the population of S_. schall in Lake Kainji no significant variation occurred in the mean condition factors of two consecutive months or between the maximum and minimum levels of the condition factors. However, he did note that a reduction in fat content occurred in adult fish during the course of the breeding season . Fecundity Fecundity studies are based on data from fish caught in March 1974 by beach seining in Ferguson's Gulf. Gonads from ripe S. schall were col- lected and preserved in Gilson's fluid. Methods for the estimation of egg numbers were similar to those used for B. bayad (see page 1034) and described by BAGENAL (1968). In Figure 10.47 egg number has been plotted against fork length. There is a wide scatter of points and considerable variability in fecundity between fish of the same length. A logarithmic transforma- tion of the data was carried out and regression values estimated. How- ever, the relationship between log length and log egg number showed very little correlation (correlation coefficient = 0.195) and regres- sion values are not given. Assuming that the subsampling method used for estimating egg numbers was statistically sound, which is a reason- able assumption judging by results from Bagrus bayad (see above), the number of eggs appears to vary considerably at a given length. It was possible that since fish for this study were collected on or near the breeding grounds, partial release of eggs had already occurred. There was no evidence, however, to suggest that eggs were released in batches. Eggs in each fish appeared to reach roughly the same degree of ripeness at the same time and partially spent fish were never caught. On the contrary spent females caught near the spawning grounds had flaccid gonads with only few ripe eggs remaining in the ovary. Once the release of eggs commences it appears that all eggs are shed and gonads used for the present fecundity study probably had their full complement of eggs. 1073 Synodontis schall is therefore moderately fecund, with fish larger than 24 cm FL producing more than 16,000 eggs. The highest number of eggs recorded in the present study was nearly 24,000 for a fish of 25.8 cm. WILLOUGHBY (1974) also found a wide range in the fecundity at a given length in S. schall from Lake Kainji, and was unable to establish a fecundity-length relationship. EMBRY0L0GXCAL AND LARVAL DEVELOPMENT Introduction Although siluriform species are common throughout Africa , comparatively few details of the embryological and larval development of this impor- tant group of fish are known. GREENWOOD (1955) has briefly described the larval development of Clarias mossambicus Peters from Lake Victoria. More details are given by LAKE (1967b) on the larval de- velopment of the Australian freshwater catfish Tandus tandanus (Mitchell). Larval development in Synodontis schall proved to be quite similar to this fish, but there were some important differences which may be attributed to the adaptation in Synodontis of laying eggs in fast flowing but short lived rivers. Artificial fertilisation of S_. schall eggs was successfully accom- plished during the present investigations and larval development was followed up to five days after hatching. Methods Live, ripe fish from the spawning areas in Ferguson's Gulf were taken to the laboratory and eggs were stripped from the females. The eggs were placed, in batches of 20, into 10 cm diameter petri dishes con- taining tap water. A portion of the anterior section of the testes of a ripe male was then agitated in the water. After one hour most of the water in the petri dishes was replaced with fresh tap water and any remaining pieces of testes removed. The tap water was pumped from the River Kalokol and contained no chlorine. A variety of treatments for rearing were tried. Mortality was lowest in petri dishes where the embryos were covered with approxi- mately 1 cm of water. The water was changed every 12 hours and any infertile or moribund eggs were removed. Temperatures varied from 25 to 28.5°C. Measurements of embryos and larvae have been made on living material removed with the aid of a pipette and anaesthetised in a di- lute solution of MS 222 Sandoz . Development was rapid and the eggs hatched within 24 hours of fer- tilisation. Larvae were actively swimming and feeding within 3 days of hatching. Development of embryos and larvae Table 10.89 gives the approximate time taken from fertilisation to attain each developmental stage. A selection of stages is illustrated in Plate 10.1. 1074 Egg stage The ripe eggs of S. schall vary in diameter from 1.2 to 1.4 mm and are translucent yellow in colour. Contact with water causes the outer gelatinous covering to swell and in 20 minutes the diameter of the eggs has increased to about 2.2 mm. The outer surface is initially smooth, but within 30 seconds of contact with water it assumes a sculptured appearance (Plate 10. 1A). Eggs are slightly adhesive and cannot be dislodged from the bottom of the petri dish by gentle agitation. After approximately 2 hours the adhesive quality is lost but eggs are denser than water and remain in contact with the bottom of the dish. The yolk mass, 1.0 to 1.2 mm in diameter, fills most of the peri vitelline space. Development of the embryo Within one and a half hours of fertilisation the first blastomere has formed, enclosing one third of the yolk, as shown in Plate 10. IB. The blastomere divides rapidly (Plates 10. 1C and 10. ID) and a blastodermal cap is formed within four hours of fertilisation. During the blastula stage the chorion becomes elliptical in shape measuring 1.3 mm by 1.0 mm. Gastrulation commences an hour later and the head and tail folds are visible eight hours after fertilisation (Plate 10. IE). Over the next 12 hours head and tail folds become prominent and somites develop. Plate 10. 1G shows an embryo 20 hours after fertilisation. The caudal region is free from the yolk sac and the auditory capsule, heart and somites are clearly visible. Occasional lateral movements of the tail occur. Immediately before hatching the embryo is 3.00 mm in length and lies with the caudal region curving ventrally below the body. In the late embryonic stages, two or three hours before hatching, the chorion increases in size to a mean diameter of 1.6 mm. Larval development Hatching takes place approximately 22 to 24 hours after fertilisation. Just before hatching the embryo becomes extremely active and appears to be pushing against the perivi telline membrane and outer sheath, through which it eventually breaks (Plate 10. 1H). Newly hatched larvae have a yolk sac approximately 1 mm in diameter. Blood flow is visible in. the caudal region and the heart beats approximately twice per second. There is at first no pigmentation (Plate 10.11) but during the next 24 hours melanophores develop in the optic cup and both dor- sally and ventrally on the fin fold (Plates 10. 1J and K) . Young larvae swim actively generally, with the yolk in contact with the bottom of the dish. Two days after fertilisation the yolk sac is no longer rounded and the buds of developing barbels are visible (Plate 10. 1L). In the following 24 hours the barbels develop rapidly, the anterior margin of the maxillary barbel becoming crenellated. Pigment forms in the head region and the blood supply to the gills is visible. Approxi- mately 93 hours after fertilisation the yolk sac appears to be empty, and four and a half days after fertilisation the first signs of food are visible in the gut (Plate 10. 1M). The post larvae swim both in contact and away from the bottom. Six days after fertilisation the larvae measure approximately 6.8 mm in length (Plate 10. IN) and caudal and pectoral fin rays are present. 1075 Discussion Although S. schall breeds in Lake Turkana this is probably a relatively recent adaptation. The spawning runs observed in the Kerio river are perhaps more typical of the breeding behaviour of this fish. In Lake Turkana S. schall may take advantage of any river flow of sufficient volume to leave the lake and to migrate upstream for the purpose of breeding. With the exception of the River Omo , Lake Turkana is fed by temporary rivers which rarely flow for more than a week at a time and more typically for only days or hours. In most cases the current is strong and the substrate generally coarse with little or no vegeta- tion. In such ephemeral rivers there is an obvious danger of eggs and larvae being trapped in pools upstream from the lake. The adhesive quality of the eggs may help to overcome the problem of eggs being swept away before being fertilised. BISHAI and GIDEIRI (1958) have observed Synodontis schall spawning in aquaria. Just prior to egg release fish were seen to be digging the mud with their pectoral spines after which eggs and milt were released. The slight adhesive- ness of the eggs might enable them to stick to the bottom thus delaying their flow downstream and give them a better chance of being ferti- lised. The urino-genital papilla on males may help with the more accurate placing of sperm onto the eggs rather than broadcasting them over a wide area. With river flow usually restricted to a few days or hours it is clearly not an advantage for the eggs to remain in one place too long, because of the danger of stranding. This is probably why only light, temporary adhesion is present, so that a short time after fertilisation eggs are free to move downstream with the river flow. Various features of the embryological and larval development of Synodontis schall enable the successful downstream movement of ferti- lised eggs and larvae possible. The swelling of the outer egg covering may have two functions. Firstly the larger size may prevent eggs from becoming trapped between sand grains, thus lowering the risk of strand- ing. Secondly the outer covering may offer physical protection as the egg is swept downstream. The rapid development of the embryos and early hatching are further possible adaptations which enable successful downstream move- ment. Larvae were observed to swim actively in the laboratory and although movement was usually in contact with the bottom, the young fish occasionally rose towards the surface. In a fast flowing river such a vertical movement would bring larvae into stronger currents which would result in more rapid movement downstream and reduce the possibility of stranding. Synodontis schall is therefore adapted to an early life spent in rapid flowing but short lived streams and rivers. The rate of develop- ment is different to that of Tandanus tandanus , an Australian catfish which also breeds in response to flood conditions (LAKE, 1967b) but in water of a less ephemeral nature. Parent fish build nests and a rapid rate of development is less important. Noticeably absent from the eggs of T. tandanus is the protective outer covering which is so prominent in S. schall. Larval development is slower and hatching occurs 151 hours after fertilisation. At this age larvae are at approximately the same stage of development as newly hatched S. schall only 24 hours old. 1076 Features of the egg and larval behaviour of the cyprinid Labeo victorianus (Boulenger) are in some ways similar to those of S. schall . Labeo victorianus migrates up the affluent rivers and streams of Lake Victoria (FRYER and WHITEHEAD, 1959) and both fish have similar spawn- ing habits. The developing cyprinid embryo is protected by the swell- ing of the vitelline membrane, not by a gelatinous sheath as in S_. schall, but the adaptive value is probably the same. Larval behaviour in the two species is also similar, both swimming actively on the bottom with occasional sorties into the midwater layers. In the Turkana basin, Synodontis schall has become more lacust- rine, with breeding no longer confined to rivers (see page 1070). Early development which seems to be well adapted to ephemeral riverine conditions might not be so advantageous in the lake. However, jS. schall is extremely abundant and the change to spawning in the lake has not apparently created any major problems for egg development. The inshore area where breeding now occurs is to some extent similar to a riverine environment since there is water movement, especially when onshore winds are strong. FOOD The food and feeding habits of Synodontis schall have been investigated in detail elsewhere in Africa. BISHAI and GIDEIRI (1965b) reported that the population inhabiting the Nile at Khartoum fed on a variety of food items ranging from insect larvae to diatoms. WILLOUGHBY (1974) provided information on the Lake Kainji population which also has an extremely varied diet. In Lake Turkana WORTHINGTON and RICARDO (1936) found that S. schall was omnivorous and results from the present survey confirm that the diet includes a large number of food categories. Method Observations on the feeding habits of S. schall are based on data col- lected from 1,200 fish caught in all areas of the lake. The majority were from bottom trawl samples, but gillnetting and beach seining pro- vided data from inshore areas, such as Ferguson's Gulf, where trawling was not possible. All stomachs were preserved in 5% formalin and exam- ined in the laboratory with the aid of a microscope. The relative volume of each food item was estimated using the points system des- cribed previously for B. bayad (see page 1035). Food categories Crustaceans formed the major part of the diet. Ostracods (including Megalocypris brevis Bars, Darwinula stevensoni (Brady and Robertson) and Limnocythere af ricana Klie) and copepods (Tropodiaptomus banforanus Kiefer, Mesocyclops leuckarti Claus and Thermocyclops hyalinus Rehberg) were the most common and occurred in the stomachs of fish from all depths of the lake. Less widespread were prawns (Caridina nilotica and Macrobrachium niloticum) recorded in deeper water and Cladocera (including Diaphanosoma excisum Sars and Hyalodaphnia barbata Weltner) confined to more inshore areas. Insects were the second most common food category. Corixids (Micronecta ras Hugh), notonectids (including Anisops worthingtonia Jaczewiki) and Ephemeroptra ( Povilla sp A and Caenis sp) occurred mainly in water of less than 20 m. Chironoraids, of which 20 to 30 1077 species have been identified in the lake, were recorded from all depths. Trichoptera and various terrestrial insects (including Coleop- tera) were occasionally present in shallow water. Gasteropoda were frequently found in the stomachs of fish inhabit- ing water deeper than 40 m. The most common was Cleopatra pnothi fol- lowed by Gabiella rosea and Melanoides tuberculata. Fish, especially Alestes spp., were preyed on by larger S_. schall . Barbus cf anema, Tilapia spp., Lates longispinis and Engraulicypris stellae were also included in the diet. Scales of Lates niloticus and Barbus bynni were common in the stomachs of fish from deep water. Hydracarina, vegetable matter, algae and diatoms were only occa- sionally eaten. Food in relation to depth Diet was found to vary most markedly with depth. Results based on material obtained by bottom trwling in the main lake are analysed in Table 10.90 for six depth zones. Data collected in Ferguson's Gulf and Allia Bay by beach seining and gillnetting are also shown. From these results the following generalisations may be made concerning the food of S_. schall in the main lake. (a) Inshore, in water of less than 5 m, insects, particularly chironomid larvae and corixids , and Crustacea (ostracods and copepods) were of roughly equal importance and together con- tributed over 75% of the food eaten. Alestes spp. made up most of the remainder of the diet. (b) In water of moderate depth, between the 5 and 40 m contours, Crustacea contributed between 60% and 83% of the food. Ostra- cods and copepods were still the principal crustaceans, but Cladocera and Caridea were also included. Few chironomids were eaten in this depth zone, and insects were of less importance than inshore. Alestes spp. continued to be a common food item- (c) In water deeper than 40 m where relatively few S. schall were caught, gasteropods were the most important food, fol- lowed by Crustacea. The eating of fish scales was confined to water deeper than 60 m. Whilst the diet of S. schall inhabiting the sheltered water of Ferguson's Gulf showed affinities with those caught in the inshore area of the lake, fish from Allia Bay had a markedly different diet with insects , principally corixids and chironomids, forming 51% of the food eaten. Allia Bay was the only area of the lake where copepods were not recorded in the diet. Diet with relation to size In all size groups of _S . schall ostracods and copepods together contri- buted at least 60% of the food eaten. The proportion of chironomids and fish however was dependent on predator size and changes in diet with increasing size are examined in Figure 10.48. Results are based on data collected from fish caught by bottom trawling in the main lake. In the 15 to 19 cm length group fish were virtually absent from the 1078 diet whilst in the largest length group 25% of the points were allo- cated to this food category. Chironomids showed a reverse trend and were common in the smaller length group and absent from fish larger than 30 cm. BISHAI- and GIDEIRI (1965b) similarly found that food consumed by the Nile population to some extent depended on predator size and the larger S. schall ate a greater proportion of fish. Diel feeding rhythms Feeding and activity in S. schall was found to follow a well defined daily pattern. This is demonstrated in Table 10.91 where results from a 24 hour gillnet station maintained in Ferguson's Gulf are presented. Four inch gillnets, which fished the entire water column from surface to bottom, were lifted and fish removed at the end of six time phases; dusk, late evening, early morning, dawn and morning. Both stomach fullness and catch rate varied considerably during the 24 hour period (Fig. 10.49). Peak stomach fullness was observed in the dawn and morn- ing sample when a mean of 6 points per fish were recorded. In the afternoon little feeding apparently took place and stomachs were more than three-quarters empty. Following a short burst of feeding at dusk, stomach fullness declined further to a minimum in the late evening. Catch rate followed an identical pattern and it is evident that in- creased feeding resulted in enhanced catches. Over 80 S. schall were caught per hour at dawn, coinciding with the time when stomach full- ness also reached a peak. During the morning, although stomachs re- mained relatively full, catches declined dramatically which indicates that this was a rest period. Catch rates increased slightly at dusk, coinciding again with the rise in stomach fullness, but subsequently declined during the night. Variation in the proportion of the three most common food items eaten during the six time phases are given in Figure 10.50. While the proportion of ostracods remained fairly constant, copepods and chirono- mids were clearly preyed on at different times during the 24 hour period. At night chironomids accounted for between 45% and 75% of all food, while copepods were preyed on mainly at dawn and during the morning. Changes in the diet during the 24 hours are probably asso- ciated both with the relative availability of food items and changes in the behaviour pattern of schall . During the night the high propor- tion of chironomid larvae in the food indicates that S. schall is feed- ing on the bottom. The marked change in diet to copepods which occurs at dawn suggests that at this time there is an increase in feeding activity in the midwater layers and at the surface. Discussion BISHAI and GIDEIRI (1965b) reported that in the River Nile S. schall feeds principally on bottom fauna including insect larvae and pupae, annelids, nematodes, crustaceans and lamellibranchs . Fish were also eaten, especially by the larger length groups, indicating that feeding was not solely confined to the bottom. WILL0UGHBY-( 1974) has examined the relative buoyancy of Synodontis spp. inhabiting Lake Kainji, and found that not all species were confined to bottom living, there being some better adapted to surface and midwater feeding. He found that Synodontis schall occurred mainly on the bottom and had a diet of humus, insect larvae and nymphs, and plant matter. 1079 In Lake Turkana, where there is no competition from other Synodontis spp. which are more specialised in surface and midwater feeding, S. schall was not restricted to a diet of benthic forms. It fed actively on Alestes spp. and on copepods and it was also commonly caught in surface setgillnets. The Lake Turkana population is obviously very versatile with regard to feeding habits and is consequently able to live in a wide variety of habitats at all depths. THE FISHERY FOR Synodontis schall Prior to 1973, Synodontis schall was of no importance in the commercial fisheries of Lake Turkana, since on account of its small size it was unacceptable in a salted and sun dried form at the co-operatives. How- ever the species has probably always been important in the subsistence fisheries of the lake. BAYLEY (1975) estimated that S_. schall formed 25% of the fish eaten by the inhabitants of the Kalokol area. In 1973 Luo fish traders started smoking fish including _S. schall for sale in markets elsewhere in Kenya and the species is now of com- mercial value. No precise records are available for quantities of smoked fish, since this branch of the fishery is not controlled by the co-operatives. However, BAYLEY (1975) estimated that in 1974 approxi- mately 122 tonnes (fresh weight equivalent) of smoked S. schall were exported. This constitutes 20% of the total weight of smoked fish and an increase from 8% in 1973. The steady increase in the importance of S. schall as part of the trade in smoked fish may be limited in the future by shortage of fuel for the smoking kilns. The trawl survey (see page 590) has shown that S. schall occurs mainly inshore in areas easily reached by light fishing craft. Gill- nets of a smaller size than are used at present have proved to be the ideal fishing method, and catch rate is highest at depths where the midwater scattering layer meets the lake floor (see page 590). In the Longech area this lies between the 15 and 25 m contour for much of the year, slightly deeper than the inshore gillnetting areas of the Turkana fishermen. Catches of _S . schall could be increased, therefore, by set- ting nets of smaller mesh size slightly further offshore. Appendix 10C gives the gillnet selectivities of the fleet of nets used during the present survey. Since the majority of S_. schall mature at sizes greater than 19 to 20 cm (see page 1068) it would be inadvisable to use nets of smaller than approximately 4 inch stretched mesh. In 4 inch nets 17% of S. schall were 20 cm or smaller, compared with 37% in 3 inch nets. However under present conditions the encouragement of a fine net fishery for S. schall would be contrary to conservation measures urgently required to protect other commercially important species such as Labeo horie and Barbus bynni . Nets of 4 inch mesh would catch juveniles of these species in considerable quantity. BAYLEY (1975) has recommended that regulations should limit the minimum mesh size for gillnets to 6 inches. Experimental fishing has revealed that _S. schall is also caught in large numbers in surface gillnets in water of 25 m depth (see page 623 ). Labeo horie and Barbus bynni were virtually absent from these nets, the catches being composed almost entirely of Alestes baremose, Hydrocynus f orskalii and _S. schall (see Chapter 8, page 621 ) . Nets of 3-| inch mesh caught S. schall with a mean length of 25.2 +2.3 cm and would be commercially successful for this species. Before fishing could commence in the surface waters of the lake with 3i inch nets 1080 however legislation would have to be implemented to prevent them being set on the bottom in water shallower than 25 m. This would be a diffi- cult law to enforce for a number of reasons, and at the present time the development of a surface fishery for S. schall should not be con- sidered . Thus although the trawl survey shows that schall is potentially one of the most important fish in the lake, a large scale expansion of the exploitation of this species would conflict with policies for the conservation of stocks of large fish. SYNODONTIS FRONTOSUS MANN (1964) reported that four specimens of Synodontis frontosus rang- ing from 18.5 to 29 cm FL were caught in small mesh gill nets at the mouth of Ferguson's Gulf. The only previous record of this fish in the lake was from the ME end at Galeba collected by ZAPHIRO and McMILLAN (BOULENGER, 1911). During the present survey all S_. frontosus examined were caught in the River Omo . Thirteen specimens ranging from 18 to 27 cm FL were netted in a fine shore seine at Lochui in the River Omo, 3 km east of Namurupath on the Ethiopian frontier (Fig. 10.27). No examples were obtained from the main lake, and the earlier record by MANN (1964) from Ferguson's Gulf suggests that exceptional condi- tions may have prevailed at the time of his visit. MOCHOCHUS NILOTICUS BOULENGER (1911) recorded that M. niloticus had previously been collec- ted from Lake Turkana though the specimen(s) had apparently been lost after collection. Neither the Cambridge Expedition (WORTHINGTON and RICARDO, 1936) nor MANN (1964) were successful in catching further examples of this fish. During the present survey a single specimen of 33 mm FL was caught in September 1972 in the Typha beds at Todenyang (see Fig. 10.27). This species is probably more common in the River Omo and perhaps only occasionally enters the lake after flooding has occurred . SUMMARY Siluriform catfish form an important part of the ichthyofauna of Lake Turkana (formerly Lake Rudolf) and the present investigations, which extended from May 1972 to July 1974, examine the biology and commercial potential of this group of fish. A total of ten species have been identified from the region of the lake, of which Bagrus bayad , Syno- dontis schall and Schilbe uranoscopus were most commonly caught, thus providing adequate material for detailed biological studies. Each of the three species studied in detail have been aged by examining length-frequencies of samples caught by bottom trawling. Growth is relatively rapid in each species, and follows the von Bertalanffy growth model. No sexual differences in growth pattern were noted, except in Schilbe uranoscopus where females attain an appre- ciably greater maximum length than males. The commencement of spawning activity in all three species is associated with rainfall and flooding. Bagrus bayad breeds within the 1081 lake probably in coastal waters of moderate depth after local rainfall, while Synodontis schall spawns both close inshore and in seasonal rivers. Schilbe uranoscopus is an anadromous form and undergoes a major breeding migration up the River Omo . Artificial fertilisation of Synodontis schall was successfully carried out. Larval development proved to be rapid, and showed adaptations to an early life in tempor- ary rivers. The diet of Bagrus bayad changes with increasing size, from prawns to fish, and the species is an important predator on populations of small fish in Lake Turkanan. Synodontis schall feeds on a wide range of food items, and the species is widely distributed at all depths, though it is most common inshore. The diet of Schilbe uranoscopus is restricted mainly to prawns and fish. Since the introduction of longlines , the Bagrus bayad fishery has increased in commercial value and in areas such as the East Shore where deep water fishing is important, the species may form up to 70% of the total number of fish sold to the co-operative. There are indications that stocks of catchable fish are being depleted rapidly. However fish smaller than 50 cm were rarely caught by longlining and since Bagrus bayad matures at a length of approximately 35 cm, a relatively high proportion of the breeding population is conserved. Synodontis schall is of major importance in the subsistence fisheries of the lake and was caught in larger numbers by gillnets. Yields of both Synodontis schall and Schilbe uranoscopus are potentially greater than at present, in gillnets of 2 to 4 inch stretched mesh. However the use of such fine mesh sizes would be detrimental to the juvenile populations of the larger commercially important fish. The remaining seven species of silurids were less commonly caught and all available information has been included. I* •ii* 4SIH- ’ wiiiii 1083 REFERENCES BAGENAL, T H (1968). Eggs and early life history. Part I. Fecundity in: Methods for assessment of fish production in fresh waters, IBP handbook No 3 (Ed) RICKER, W E, Oxford, Blackwell. BAYLEY, P B (1975). The commercial fishery of Lake Rudolf, Kenya. Unpublished report to Ministry of Overseas Development, London. BEAR, I J and THOMAS, R G (1964). Nature of argillaceous odour. Nature, Lond . 201 : 993-995. BEAR, I J and THOMAS, R G (1965). Petrichor and plant growth. Nature, Lond. 207 : 1415-6. BERGSTRAND, E and C0RD0NE, A J (1971). Exploratory bottom trawling in Lake Victoria. Afr. J. Trop. Hydrobiol . Fish. 1 : 13-23. BEVERTON, R J H and HOLT, S J (1957). On the dynamics of exploited fish populations. Fish, invest. Lond. Ser 2 19 : 1-533. BISHAI, H M and GIDEIRI, Y B A (1965a). Studies on the biology of the genus Synodontis at Khartoum. 1. Age and growth. Hydrobiologia 26 : ’ BISHAI, H M and GIDEIRI, Y B A (1965b). Studies on the biology of the genus Synodontis at Khartoum. 2. Food and feeding habits. Hydrobiologia 26 : 98-113. BISHAI, H M and GIDEIRI, Y B A (1968). Studies on the biology of the genus Synodontis at Khartoum. 3. Reproduction. Hydrobiologia 31 : 193-202 . BLACHE, J (1964). Les poissons du basin du Tchad et du bassin adjacent du Mayo Kebbi . Paris, 0RST0M. BOULENGER, G A (1911). Catalogue of the fresh-water fishes of Africa in the British Museum (Natural History) 2_ : London, British Museum (NH) . CHILVERS, R M (1968). A method of age estimation from length frequency data for larger fish with reference to Bagrus docmac (Forsk) . Ann. Rep. E. Afr. Freshwat. Fish. Res. Org . 1968 : 32-38. CHILVERS, R M and GEE, J M (1974). The food of Bagrus docmac (Forsk). (Pisces Siluriformes ) and its relationship with Haplochromis hilgendorf (Pisces: Cichlidae) in Lake Victoria, East Africa. J. Fish. Biol. 6 : 483-505. CHITRAVADIVELLI , K (1974). The plain squeaker Synodontis zambezensis Peters, 1852, in Lake Kariba: a man-made tropical ecosystem in Central Africa: 298-310. Balon, E K_ and Coche , A G (Eds), The Hague, Junk. CORBET, P S (1955). The reproductive cycle of non-cichlids fishes, Ann. Rep. E. Afr. Freshwat. Fish. Res. Org. 1954/55: 21p. 1084 CORBET, P S (1958). The food of non-cichlid fishes in Lake Victoria, Ann. Rep. E. Afr. Freshwat. Fish. Res. Org . 1958 : 21-37. CORBET, P S (1960). Breeding sites of non-cichlid fishes in Lake Victoria, Nature, Lond. 187 : 616-617. CORBET, P S (1961). The food of non-cichlid fishes in the Lake Victoria basin with remarks on their evolution and adaptation to lacustrine conditions. Proc. zool . Soc. Lond. 136 : 1-101. DAGET, J (1952). Memoires sur la biologie des poissons due Niger Moyen. I. Biologie et croissance des especes du genre Alestes . Bull . Inst, fr. Afr. noire 141 : 191-225. DAGET, J (1954). Les poissons due Niger superieur. Mem. Inst. Franc. Afr. noir. No. 36 : 1-391. ELDER, H Y (1960). Bagrus docmac investigation. Ann. Rep. E. Afr. Freshwat . Fish. Res. Org. 1960 : 19-20. ELDER, H Y (1962). Preliminary notes on the relationship between feeding and growth rate in the siluroid fish Bagrus docmac (Forsk). Ann. Rep. E. Afr. Freshwat. Fish Res. Org. 1961 : 19- 22. — - — - - ■ FRANK, S (1974). The spotted squeaker Synodontis nebulosus , the butter catfish Schilbe mystus , the vundu Heterobranchus longif ilis and the electric catfish Malapterurus electricus in Lake Karibia: a man-made tropical ecosystem in Central Africa : 325-332. Balon, E K and Coche, A G (Eds) The Hague, Junk. FROST, W E (1943). The natural history of the minnow, Phoxinus phoxinus . J. Anim. Ecol . 12 : 139-162. FRYER, G and WHITEHEAD, P J P (1959). The breeding habits, embryology and larval development of Labeo victorianus Boulenger (Pisces: Cyprinidae). Rev. Zool. Bot . afr. 59 : 33-49. FRYER, G (1965). Predation and its effects on migration and speciation in African fishes: A comment. With further comments by Greenwood, P H and a reply by Jackson, P 3 N. Proc. zool. Soc. Lond. 144 : 301-322. FUCHS, V E (1934). The geological work of the Cambridge Expedition to the East African lakes, 1930-31. Geol . Mag. 7 1 : 98-112. FUCHS, V E (1935). The Lake Rudolf Rift Valley expedition 1934. Geogrl . J. 86 : 114-142. FUCHS, V E (1959). The geological history of the Lake Rudolf basin, Kenya Colony. Phil. Trans. Roy. Soc. B. 229 : 293-273. GARR0D, D J (1959). The growth of Tilapia esculenta Graham in Lake Victoria. Hydrobiologia 12 : 268-298. GRAHAM, M (1920. The Victoria Nyanza and its fisheries. A report on the fishing survey of Lake Victoria 1927-1928, London. Crown Agents for the Colonies. 1085 GREENWOOD, P H (1955). Reproduction in the catfish, Clarias mossambicus Peters. Nature, Lond . 176 : 516-519. GREENWOOD, P H (1956). Breeding of Clarias mossambicus (Peters) in Lake Victoria. Second symposium on African hydrobiology and inland fisheries, Brazzaville. Publ . No. 25 : 77p. GREENWOOD, P H (1966). The fishes of Uganda. Kampala, Uganda Society. HAMBLYN, E L (1960). The food and feeding behaviour of nile perch. Ann. Rep. E. Afr. Freshwat. Fish Res. Org . 1960 : 33-34. RASTER, A E (1966). Underwater guideposts: homing of salmon. Madison, University of Wisconsin. HOLDEN, M J (1955). Ring formation in the scales of Tilapia variabilis Boulenger and T. esculenta Graham from Lake Victoria. Ann. Rep. E. Afr. Freshwat. Fish Res. Org. 1954/55 : 36-40. ROLL, E A (1966). Some notes on the breeding of barbel Clarias gariepinus (Burchell) in Rhodesia. News. Lett. Limnol. Soc . S. Afr~. TT" 38-41. HOLL, E A (1968). Notes of spawning behaviour of barbel Clarias gariepinus (Burchell) in Rhodesia. Zoologia Africana 3 : 185-188 HOPSON, A J (1965). Winter scale rings in Lates niloticus (Pisces: Centropomidae) from Lake Chad. Nature, Lond. 208 : 1013-1041. HOPSON, A J (1972a). A study of the Nile Perch (Lates niloticus (L) , Pisces: Centropomidae) in Lake Chad. Overseas Res. Publ. 19 : 99pp . HOPSON, A J (1972). Breeding and growth in two populations of Alestes baremose (Joannis) (Pisces: Characidae) from the northern basin of Lake Chad. Overseas Res. Publ. 20 : 48pp. HOPSON, J (1975a). Preliminary observations on the biology of Lates niloticus in Lake Rudolf. In: Symposium on the hydrobiology and fisheries of Lake Rudolf. Unpublished. HOPSON, J (1975). Preliminary observations on the biology of Alestes baremose (Joannis) in Lake Rudolf). In : Symposium on the hydro- biology and fisheries of Lake Rudolf. Unpublished. HYNES, H B N (1950. The food of freshwater sticklebacks ( Gasteros teus aculeatus and Pygosteus pungitius) with a review of methods used in studies of the food of fishes,. J.Anim. Ecol. 19 : 35-48. JACKSON, P B N (1961). The impact of predation especially by the tigerfish (Hydrocyon vi ttatus Cast) on African freshwater fishes. Proc. zool. Soc. Lond. 136 : 603-622. JACKSON, P B N, ILES , T D, HARDING, D and FRYER, G (1963). Report on the survey of northern Lake Nyasa. 1954-1955. Zomba, Government Printers . 1086 JOHNELS, A G (1952). Notes on scale rings and growth of tropical fishes from the Gambia River. Ark. Zool . Ser . 3 : 363-366. KLEEREKOPER, H (1969). Olfaction in Fishes. Bloomington, Indiana University Press. LAKE, J S (1967a). Rearing experiments with five species of Australian freshwater fishes. I. Inducement to spawning. Aust. J. mar. Freshwat .Res . 18 : 137-153. LAKE, J S (1976b). Rearing experiments with five species of Australian freshwater fishes. II. Morphogenesis and ontogeny. Aust. J. mar. Freshwat. Res. 18 : 155-173. LE CREN, E D (1951). The length-weight relationship and seasonal cycle in gonad weight and condition in the perch (Perea f luviatilis ) . J. Anim. Ecol. 20 : 201-219. ~ LEWIS, DSC (1974). The food and feeding habits of Hydrocynus f orskalii Cuvier and Hydrocynus brevis Gunther in Lake Kainji, Nigeria. J. Fish. Biol. _6 : 349-363. LOWE-McCONNELL, R H (1964). The fishes of the Rupununi savanna district of British Guiana, South America. J. Linn. Soc. Zool. 45 : 103-144. MANN, M J (1964). Report on a fisheries survey of Lake Rudolf, Kenya. Ann. Rep. E. Afr. Freshwat. Fish Res. Org. 1962/ 3 : 53-61. MUNRO, J L (1967). The food of a community of East African freshwater fishes. J. Zool. Lond . 151 : 389-415. NIKOLSKY, G V (1963). The ecology of fishes. New York, Academic Press . PELLEGRIN, J (1935). Pisces, Mission Scientifique de l'Omo 2 : 131— 139. POLL, M (1939). Poissons, Exploration du Parc National Albert, Mission G F de Witte (1933-1935). Fasc. 24 : 1-81. RICKER, W E (1958). Handbook of computations for biological statistics of fish populations. Bull. Fish Res. Bd . Can. No 119 : 1-300. TAYLOR, J R (1965). The Analysis of Polymodal Frequency Distributions. J. Anim. Ecol. 34 : 445-452. TESCH, F W (1968). Age and growth in: Methods for assessment of fish production in fresh waters. IBP Handbook No. 3 (Ed) Ricker, W E Oxford, Blackwell. TWEDDLE, D (1975). Age and growth of the catfish Bagrus meridionalis Gunther in southern Lake Malawi. J. Fish. Biol. 7 : 677-685. VAN DER WAAL, B C W (1974). Observations on the breeding habits of Clarias gariepinus (Burchell). J. Fish. Biol. 6 : 23-27. 1087 VAN DER WAAL, B C W and SCH00NBEE, H J (1975). Age and growth studies of Clarias gariepinus (Burchell) (Clariidae) in the Transvaal, South Africa. J. Fish. Biol. _7 : 227-233. WHITEHEAD, P J P (1959). The anadromous fishes of Lake Victoria. Rev . Zool. Bot. afr. 59 : 329-363. WILLOUGHBY, N G (1974). The ecology of the genus Synodontis (Pisces: Siluroidei) in Lake Kain ji , Nigeria. Unpublished PhD thesis, University of Southampton. WORTHINGTON, E B (1929). Report on the fishing survey of Lakes Albert and Kyoga . London, Crown Agents. WORTHINGTON, E B (1932a). Scientific results of the Cambridge Expedition to the East African Lakes 1930-1. 1. General introduction and station list. J. Linn. Soc . Zool. 38 : 99-119. WORTHINGTON, E B (1932b). Scientific results of the Cambridge Expedition to the East African Lakes 1930-1. 2. Fishes other than Cichlidae. J. Linn. Soc. Zool. 38 : 121-134. WORTHINGTON, E B, RICARDO, C K (1936). Scientific results of the Cambridge Expedition to the East African Lkes, 1930-1. No. 15. The fish of Lake Rudolf and Lake Baringo. J. Linn. Soc. Zool. 40 : 33-69. 1089 TABLE 10.1 The ten species of catfish inhabiting Lake Turkana SPECIES FIRST COLLECTED BY: Bagridae Bagrus bayad (Forsk. ) Cambridge Expedition, 1930-31 (Worthington and Ricardo, 1936) Bagrus docmac (Forsk.) Zaphiro and McMillan, 1908 (Boulenger, 1911) Chrvsichthvs auratus (Geoffr.) Present survey Auchenoglanis occidentails (C. and V. ) Clariidae Reported by Boulenger (1911) Clarias lazera (C. and V.) Schilbeidae Cambridge Expedition, 1930-31 Schilbe uranosconus Rudd. Mochokidae Cambridge Expedition, 1930-31 Synodontis schall (Bl.-Schn) Zaphiro and McMillan, 1908 Svnodontis frontosus Vaill . Zaphiro and McMillan, 1908 Mochocus niloticus Joannls Malapteruridae Reported by Eoulenger (1911) MalaDterurus electricus (Gmelin) Mission Scientifique de L'Omo, 1932-33, (Pellegrin , 1935) 1091 TABLE 10.2 Annual rainfall (mm) recorded at the Fisheries Department, Ferguson's Gulf, 'over 5 years. The number of days with rain is given in brackets Month Year 1968 1969 1970 1972 1973 J anuary : mm 25.7 30.2 0 . 5 days (2) C4) Cl) February : mm 23.9 1.8 - 45.0 - days (4) (1) (1) March : mm 82.1 20.5 5.1 - - days (5) (1) (1) April : mm 137.4 _ 92.2 - 46 . 4 days (6) (2) 03) May : mm - 42.0 11.6 8.9 36.7 days (.4) (4) (2) 02) June : mm 18.0 - - 70.3 - days (1) (2) July : mm - - - - 10.6 days (2) August : mm - - 16.6 - - days (1) September : mm - - - - 41.2 days (2) October : mm 3. 9 6.9 - 92.0 10.2 days (2) (1) 12) ci) November : mm 0.2 9.6 - 0.5 24.5 days (2) (1) (1) (2) December : mm 42.3 _ - 17.0 - days (4) — (1) 1093 TABLE 10 . 3 Bagrus ba/£ad_: mean bottom trawl catch per effort, in kilograms per hour in the Northern > Southern and Central sectors of Lake Turkana for the period May 1972 to July 1974. Depth zones in metres 0-10 10-20 20-30 30-40 40-50 50-60 60-70 70-80 804 Northern sector. Allia Bay 3.5 14.5 3.5 - - n/a n/a n/a n/a other areas 8.3 33.4 23.9 62.9 39. 8 30.0 50.0 92.0 n/a Combined 4.7 31.4 21.4 62.9 39.8 30.0 50.0 92.0 n/a Central sector 18.5 33.2 33.2 30.4 53.2 43.8 64.7 80. 5 99.3 Southern sector 2.6 5.1 35.1 21 .7 6.0 31.0 21.2 152.0 113.0 Overall mean 7.1 29.5 30.0 40. 3 47 .7 41.6 50.2 86.4 102.0 n/s - no sample n/a not applicable ; 1095 TABLE 10.4 Bagrus bayad: catch per effort (mean numbers of fish caught per hour of trawling) in eight length groups. All data from the Longech area collected between May 1972 and July 1974 has been combined, and is given for nine depth zones. Depth zones (metres) length groups (cm) 0-9 10-19 20-29 30-39 40-49 50-59 60-69 over 70 0 - 10 0 0 4 7 4 + 0 0 10 - 20 1 5 14 35 9 I 1 ■+* 20 - 30 3 11 16 25 8 3 3 + 30 - 40 25 5 12 18 8 4 2 + 40 - 50 10 38 24 25 9 5 3 1 SO - 60 9 46 18 18 6 3 2 2 60 - 70 2 44 44 42 8 6 5 3 70 - 80 3 71 34 70 6 7 4 2 Over 80 1 55 34 109 5 5 3 2 ( + - less than 1 fish/hour) 1097 TABLE 10.5 Bagrus bayad: Comparison of mean lengths of males and females for each of the two main size groups caught by bottom trawling in the Longech area in March and July 1973 (see Fig. 9). Results of t- tests are given Mean fork length Variance (cm) Sample Size Degrees of freedom t value March: Females 13.05 8.36 20 48 0.72 Males 12.37 12.61 30 Females 34. 11 17 . 99 186 323 0.86 Males 33.70 19.20 139 July: Females 15.61 9.52 72 130 0.13 Males 15.53 15.86 60 Females 35.27 30.25 113 216 0.51 Males 35.65 31 . 10 105 *> if 1099 TABLE 10.6 Observed mean monthly lengths of four year classes of Bagus bayad caught by bottom trawling between May 1972 and July 1974 in the Longech area. Sample sizes are given in brackets , '1 Year class . ; Sample date 1970 1971 1972 1973 j 1972: May 31.1 C 57) 18.8 (167) Jun 33.4 ( 26) 22.9 ( 26) Jul 34.2 (197) ' . Aug 36.6 (137) Sep 35.1 ( 67) 5.0 (222) 9 Oct 36.5 ( 25) 23.8 (174) 7 . 3 C 36) Nov 39.3 ( 17) 27.4 ( 20) 8.0 ( 64) Dec 1 9.3 ( 24) §[■■ 1973: Jan 31 . 4 C 87) 10.9 (139) Feb 32.9 (165) 13.3 (101) Mar 33.3 (180) 14.4 (338) Apr 32.8 ( 83) 14.1 (174) 'v: . . May 34.2 (288) 15.1 (252) i Jun 34.5 ( 40) 17 . 5 ( 18) , Jul 15.5 (332) Aug 35.8 ( 39) 19.6 ( 62) 5.1 ( 28) ' if Sep 36.6 (162) 21 . 0 ( 17) ■ , Oct 33.2 ( 65) 24.1 ( 86) 8.9 (125) : : *' Nov 28.9 (176) 10.4 ( 90) Dec 11.7 (389) : : 1974: Jan 31 . 0 (197) 11.3 ( 64) Feb 29.7 ( 52) 13 . 6 ( 88) *i Mar 33 . 5 (283) 14.2 (268) Apr 32 . 0 (258) 14.9 (241) May 34.4 (289) 17.2 (107) Jun 32 . 3 ( 53) 18.6 ( 25) Jul 33.7 (267) 19.4 ( 17) \ tf *lll* 1101 TABLE 10.7 Growth of Bagrus bay ad: Data from four year classes (Table 6) have been combined to give the observed monthly increment in to 2j years of age. Approx age (months ) Mean length cm. Sample size May Jun 1 - - Jul 2 - - Aug 3 5.1 28 Sep 4 5.0 222 Oct 5 8.5 161 Nov 6 9.4 154 Dec 7 11.6 413 Jan 8 11.0 203 Feb 9 13.4 182 Mar 10 14.3 606 Apr 11 14.6 415 May 12 16.7 526 Jun 13 19.9 69 Jul 14 15.7 349 Aug 15 19.6 62 Sep 16 21.0 17 Oct 17 23.9 260 Nov 18 28.7 196 Dec 19 - - Jan 20 31.1 284 Feb 21 32.6 214 Mar 22 33.4 463 Apr 23 32.2 341 May 24 34.0 634 Jun 25 33. 3 119 Jul 26 33.9 464 Aug 27 36.4 176 Sep 28 3G.2 229 Oct 29 34.1 90 Nov 30 39.3 17 1103 TABLE 10 . 8 Bagrus bavad observed and theoretical lengths in Age (months) Observed fork Theoretical fork length cm length cm 3 5 . 1 3.9 6 9.4 9.0 9 13.4 13. 7 12 16.7 18.2 15 19.6 22.4 18 28.7 26.3 21 32.6 29.9 24 34.0 33.3 27 36.4 36.5 30 39.3 39.5 1105 TABLE 10.9 Bagrus bayad. Length : weight regression coefficients, calculated from log"1Q weight (g) = log^a + h log.^ length (cm) for a number of size groups of each sex. correlation coefficient b l°g10a No. of fish Females larger than 20cm. Gonad stage 2 0. 899 3.066 -2. 149 421 3 0. 992 2. 815 -1. 703 39 4 0.991 2.995 -2.006 44 All 0.905 3.053 -2. 122 504 Males larger than 20cm. Gonad stage 2 0.993 3.067 -2.159 128 3 0.993 2.998 -1.908 215 4 0.994 2.945 -1.917 31 All 0.994 3.023 -2.073 374 Females less than 20cm. 0.980 2. 749 -1.751 36 Males less than 20cm. 0.974 2. 826 -1. 846 66 All less than 20cm. 0.975 2. 801 -1. 815 102 All B. bayad 0.968 3.032 -2.088 980 1107 TABLE 10.10 Bagrus bay ad : comparing the length-weight relationship of grouped data (Table 10.9). The slopes of the regressions have been compared by applying d- tests. Comparison between : Degrees of freedom d value S t age 2F cf stage 3F 453 1.505 Stage 2F cf stage 4F 463 1. 434 Stage 2M cf stage 3M 341 1.253 Stage 2M cf stage 4M 16 7 0.604 S t age 2M cf stage 2F 547 0.005 Stage 3M cf stage 3F 252 0.653 S t age 4F cf stage 4M 73 0.123 M of F under 20 crnFL. 100 0.395 F < 20cm cf F > 20cm. 538 1.019 M C 20cm cf M > 20cm. 438 1.598 All fish 20cm. OSOcm cf all> 978 1.580 F : females M : males 1109 TABLE 10.11 Bagrus bay ad : showing estimated mean lengths and weights at ages 1 to 10 years. Age (years) Mean fork length (cm) Mean weight (g) 1 18.2 54 2 33.3 337 3 44. 7 824 4 53. 7 1436 5 60.6 2072 6 66.0 2684 7 70. 1 3223 8 73. 3 36 89 9 75.7 406 8 10 77.6 4385 1113 TABLE 10.13 Changes in the maturity of male and female Bagrus bay ad with size during the spawning season, April to July, in the Longech area. Data from 1972, 1973 and 1974 have been combined, and numbers of fish at each stage are expressed as a percentage of the total in each 1cm length group. Fork length cm. Number of Maturity stage % fish 2 3 4 5 Males 25 16 100 . . . 6 18 89 11 - - 7 13 85 15 - - 8 22 95 5 - - 9 42 69 27 2 - 30 41 68 32 - - 1 36 56 36 8 - 2 36 28 50 22 - 3 54 20 59 20 - 4 59 17 47 36 - 5 59 22 47 31 - 6 62 14 44 40 2 7 56 14 41 45 - 8 40 7 38 55 - 9 17 6 65 29 - 40 15 20 40 40 * Females 25 22 100 6 15 93 7 - - 7 17 94 6 - - 8 30 90 7 3 - 9 31 94 3 3 - 30 37 81 16 3 - 1 38 68 16 16 - 2 43 72 23 5 - 3 51 51 31 14 4 4 68 44 18 35 3 5 60 47 28 23 2 6 48 31 44 23 2 7 47 40 30 28 2 8 26 50 27 19 4 9 29 69 21 7 3 40 25 48 16 32 4 1115 TABLE 10.14 Changes in the gonad condition of male Bagrus. bayad in the 35 to 39cm length group in the Longech area between July 1972 and June 1974. Numbers of fish at each maturity stage have been expressed as a percentage of the monthly total. Mon th Total Gonad stage % number 2 3 4 5 1972 Jul 23 35 48 17 - Aug 40 20 60 20 - Sep 17 71 29 - - Oct 24 42 58 - - Nov 37 22 73 5 - Dec 11 64 36 - - 1973 J an 18 11 89 - - Feb 2 7 52 33 15 - Mar 37 35 49 16 - Apr 30 57 33 10 - May 113 2 35 62 1 Jun 12 75 25 - - Jul 61 10 85 5 - Aug 64 31 63 6 - Sep 54 24 76 - - Oct 27 22 78 - - Nov 27 30 67 3 - Dec 21 24 76 - - 1974 J an 20 25 75 - - Feb 50 16 76 8 - Mar 59 - 86 14 - Apr 33 3 88 9 - May 47 6 51 40 - Jun 12 17 75 8 - 1117 TABLE 10.15 Changes in the gonad condition of female Bagrus bay ad in the 35 to 39 cm length group in the Longech area between July 1972 and June 1974. Numbers of fish at each maturity stage have been expressed as a percentage of the monthly total. Mon th Total Gonad stage % number 2 3 4 5 6 1972 Jul 33 30 70 - - - Aug 27 41 29 30 - - Sep 22 100 - - - - Oct 61 84 8 6 - 2 Nov 56 70 20 9 1 - Dec 7 100 - - - - 1973 Jan 27 78 18 4 - - Feb 26 69 23 8 - - Mar 81 61 5 34 - - Apr 56 86 5 9 - - May 29 17 17 66 - - Jun 15 80 20 - - - Jul 55 2 29 64 4 1 Aug 85 75 18 5 2 - Sep 98 63 31 6 - - Oct 49 74 25 1 - - Nov 38 87 13 - - - Dec 35 80 20 - - - 1974 Jan 24 67 29 4 - - Feb 30 73 23 4 - - Mar 20 70 20 10 - - Apr 40 18 80 2 - - May 56 28 34 25 13 - Jun 10 40 10 50 - - 1119 a Si © O © © a SuO S © 0 > J 3 s: © S3 © -P CuO 3 a -P 0 CQ p >> -P T5 •H c8 P >i 3 a3 ■P j2 3 a © 3 a P o SC 3 as © CQ ■P © 3 rH Cfl s: a CQ •H , CQ 3 rH 0) © S3 bfl © -P 3 S3 3 ■P 0 CQ © a > >> 3 © •P S3 Si •H -p p CO 3 p -P 3 0 3 © a >1 © SC a; O 3 J3 -P -p -P 3 © C © U •H J3 P -P © cn a © P Sfl O 3 C . 3 T3 o (H Q © rH 3 CQ C © w o • © a CQ 3 P CQ 3 © 3. < © P X b< cq 3 © O CQ CQ o 1 1 03 00 03 1 1 00 HP CQ CQ 1 © a CQ CO CQ CQ CQ CO CQ CQ CO > CO 03 rH 1 1 HP ID o IQ 1 03 HP 03 0- 1 o CO IQ HP CO 03 O’ 03 CO 2 03 CQ CO rH | rH rH 03 1 1 CO o o 1 1 Q 0- CQ CO IQ CQ CO HP H 03 o a IQ 0* CQ i i H IQ iQ 1 1 03 00 O 03 1 © CO IQ HP CQ CO CQ CQ LQ rH CQ SO 03 00 CO CO 1 HP 0- 03 H 1 m CO 03 IQ 1 3 CQ HP tH < rH rH CO C3 Hp 1 HP O’ IQ 00 1 03 t> O CQ 1 3 (X) CQ IQ 00 H I> CQ HP HP rH >-3 3 m o CQ 1 03 O IQ 00 1 O H rH 00 1 3 rH HP rH H rH O rH HP HP rH - rH HP o CO 1 o CQ 03 CO 03 03 O IQ CO 03 a 03 rH IQ CQ CO CQ IQ HP tO 03 rH § rH H p O’ rH 03 1 1 CQ 03 03 03 1 CO CQ O i a 00 IQ HP CO 03 CO Hi H CO < p CQ l> 03 1 CD CQ 03 IQ 1 o 03 o HP i 3 03 HP HP 03 rH O tH CO H CO tH a £3 03 ■O' CO 1 1 O'* 03 H o 1 O 00 00 HP i © Cp O IQ HP 0* 03 CO rH to HP HP 3 00 03 00 1 1 00 00 03 i I 03 CO 1 i cd 0- HP IQ CQ H 00 HP 03 o ^3 03 CQ HP IQ 03 CQ HP IQ 03 CQ HP IQ P P P © © © © © © S3 SC S3 SO a S3 bO £ a £ 3 o cd a d 4-> a 9 +-> o -M © 2 CQ o 2 CQ HP 2 CQ 03 CQ CQ P 1 SS 1 © ®5 o IQ - > CQ CQ o I t- it 1121 si u a © 33 SO 33 a JQ o j © > © 3 JS s: -P 03 s SO 0 3 p -P CO T3 os -p >. •H as p X! 3 ■p CO 3 3 S P tJ3 3J as O CQ 3 © 0) rH -P 3 3 S 0) JQ CO •H O CO 0 a 3 CO 0 P u © bfi 32 a 33 3 ■P S SC fl T3 3) EJ rH 3 3) ■a 3) © P c rH si •rH 3 ■P 32 •P a 0 0 -p 0 33 CO 3 rH 3) 33 si bfl © -p OS 32 c +-> 0 33 © a > >> 3 © ■P 33 -C •H •p p CO 3 P «w ■P 3 0 OS © S >. © so 33 0 3 si ■P -p -P 3 © 3 © U •H 3! P •P © CO a © P bfl O 3 C 03 Q > CO CO rH 1 1 O' 0- ID rH 1 ao CO CO o rH 1 1 » 0 CO 00 rH 03 C'* rH O r~ rH rH 1 1 ‘ 2 rH I'j \ P O' 03 ID CO 1 1 O 03 CO O' i rH i> CO 03 O' i 00 O 03 03 rH rH rH fr o rH rH a O cn rH O' 1 1 o O ID ID i i 03 n CO 03 i 1 lb i © 00 00 rH 03 0- 03 o rH IN CO rH rH so C3 CO CO CO 1 rH 03 O' O rH i 03 rH i> ID rH CO 3 03 00 rH rH CO 03 pH (T C- rH < rH rH i rH 0* 03 oo 03 1 rH 00 CO o' o 03 rH n CO S1' rH i CO 3 03 CO 03 CO 00 rH O' o' O' O' rH f ■; r 1 k’ S3 03 03 oo CO 1 i m o o o 1 1 rH O' o C0 CO 1 3 03 us rH 03 03 rH n CO 00 rH O' 03 o O' O' i ID CD oo 03 00 1 O CO 03 CD 00 03 cd o O' 03 CO OO 03 03 CO ID CD rH 03 a rH p rH o oo 03 1 i CO I> fr CO 1 1 ID CO 03 CD 1 1 i a o i> 03 03 ID CO CO CO rH rH oo CD 1 i CO rH CO CO 1 1 CO 03 CO rH O' 1 •- © CO rH ID i> 03 t> CO rH rH Ph 3 oo t> CO 1 1 ! rH 03 O' O' I 1 O' CO CO O' 1 | • 3 C" 1> 03 ID C'" 03 CO 00 1 , *“D 03 CO O' to co 03 CO O' ID CO 03 CO O' ID CO p P H © © © © © © JQ sc 32 SC £3 32 SC cd s a Q cd a 3 ■p a 3 P o z$ ■P o 2 C/3 y 2 CO O' 2 CD O 03 O' CO P 1 1 S3 © 5® o ID — > CO CO O i n i 1123 TABLE 10.18 Maturity stages of female and male Bagrus bay ad from the Southern sector. Numbers of fish at each maturity stage have been expressed as a percentage of the total examined in two length groups Samp le size Gonad stage % 2 3 4 5 6 Females 35-39 cm 1972 Jun 8 75 25 - - - Dec 45 78 22 - - - 1973 Oct 42 67 31 2 - - 1974 Jun 19 68 16 15 - - Over 40 cm 1972 Jun 29 41 28 14 14 3 Dec 48 75 12 12 - - 1973 Oct 49 78 18 4 - - 1974 Jun 32 56 22 22 - - Males 35-39 cm 1972 Jun 8 50 25 25 - Dec 46 11 89 - - 1973 Oct 27 44 56 - - 1974 Jun 8 50 25 25 - Over 40 cm 1972 Jun 24 29 38 29 4 Dec 35 6 88 3 3 1973 Oct 35 34 63 3 - 1974 Jun 19 89 11 - - o 1125 TABLE 10.19 Maturity stages of female and male Bagrus bay ad from the Northern sector. Numbers of fish at each maturity stage have been expressed as a percentage of the total examined in two length groups . Sample Gonad stage % size 2 3 4 5 6 Females 35-39 cm 1972 Nov 39 21 33 46 - - 1973 Feb 42 86 7 7 - - Apr 73 78 6 16 - - Jul 48 77 13 10 - - 1974 Jun 24 50 33 17 - - Over 40cm 1972 Nov 20 30 20 40 - 10 1973 Feb 55 64 18 16 2 - Apr 43 70 5 23 - 2 Jul 29 90 7 3 - - 1974 Jun 27 22 33 22 22 - Males 35-39 cm 1972 Nov 35 14 66 20 - 1973 Feb 19 53 47 - - Apr 49 31 69 - - Jul 24 58 42 - - 1974 Jun 13 62 23 - - Over 40cm 1972 Nov 8 25 50 25 - 1973 Feb 28 18 75 4 3 Apr 25 4 88 8 - Jul 21 38 62 - - 1974 Jun 15 60 7 26 7 1127 TABLE 10.20 Changes in the sex-ratio of Bagus bay ad from the Longech area, with size. The sex-ratio is expressed as numbers of males per 100 fish in each length group. All data collected between May 1972 and July 1974 has been included. Length group Number of males Number of females Sex-ratio under 19cm 464 466 50 20-2 9 cm 918 868 51 30- 34cm 864 944 48 35- 39 cm 864 980 47 over 40cm 531 972 35 1129 TABLE 10.21 Bagrus bay ad : changes in the proportion of ripe- rumfing^Tstage 5) females , larger than" 30cm with depth in the Longech area during May 1974. expressed as a percentage of the total caught at each depth. Sample depth 7 Number of females Gonad stage % 2 3 4 5 6 15m 38 39 10 16 35 - 20m 25 36 28 28 8 - 30m 12 50 25 25 - - 50m 7 86 - 14 - - 60m 33 33 49 18 - - 80m 36 42 19 36 3 - 1131 TABLE 10.22 Bagrus bay ad : seasonal changes in mean condition factor (k) of fish larger than 30cm. k is given for four breeding phases and all data from the Longech area has been included. Months Feb Mar Apr to Aug Sept Oct to Jul Jan Breeding Pre-spawning Spawning Post-spawning Resting phase Males > 30cmFL k Sample size 101.9 (82) 101.6 (95) 98. 3 (46) 104. 3 (102) Females > 30cmFL k 99.2 103.9 99.1 102.1 Sample size (107) (119) (82) (163) 1133 i TABLE IQ. 23 Bag ms bay ad : variation in diet composition with depth and length. The~lvumber of points scored by each food category has been- ~ expressed as a percentage of the total points allotted in each length group and depth zone. All B. bay ad caught during the two years sampling by the bottom trawl have been included. Depth zone (m) Length Food groups items % 0-9 10-19 20-29 30-39 40-49 50-59 60-69 70-79 °° o 10- 19 cm Prawns 100 33 50 81 98 95 100 99 93 Fish - 67 50 14 2 5 - 1 7 Insects - - - 5 - - - - - 20“ 29 cm Prawns 83 73 88 67 92 77 92 73 94 Fish 13 27 12 33 7 23 8 27 6 Insects 4 - - - 1 - - - - 30- 39 cm Prawns 31 62 71 80 86 75 77 84 80 Fish 69 36 28 19 13 25 23 16 20 Insects - 2 1 1 1 - - - - 40- 49 cm Prawns 8 41 59 61 67 67 64 65 83 Fish 87 56 40 38 28 33 35 34 17 Insects 5 3 1 1 5 - 1 1 - 50- 59 cm Prawns 3 8 19 28 26 30 13 32 10 Fish 97 92 79 71 22 60 86 66 77 Insects - - 2 1 2 10 1 2 13 6 0-6 9 cm Prawns 9 2 8 7 21 - 17 4 — Fish 88 96 91 90 76 96 80 94 100 Insects 3 2 1 3 3 4 3 2 - 70- 79 cm Prawns 3 8 9 2 Fish n/s 100 97 88 91 97 92 98 100 Insects 4 3 6 2 Ove r 80 cm P r awns n/s 100 n/s 100 J.00 n/s n/s 100 100 n/s = no sample L 1135 TABLE 10.24 The food of Bagrus bay ad inhabiting Ferguson’s Gulf. The tatai~~number "of points for each food type have been each length group. All fish caught by beach seines within the Gulf have been included. Length group cm 10-19 20-29 30-39 40-49 50-59 Food categories % Alestes spp . 25 27 30 - - Engraulicypris stellae — 15 * Fish remains 75 23 17 63 100 Total fish 100 50 62 63 100 Chironomid larvae - 10 9 13 - Corixids - 13 7 - - Other Insects - 15 13 12 - Total insects - 38 29 25 - Ostracods - 12 - 12 - Prawns - - 8 - - Total Crustacea - 12 8 12 - Hydracarina - - 1 - - 1137 TABLE 10.25 Frame trawl catches of Bagrus bay ad over a twenty-four hour period at a station in 25m of water in the Longech area, during April 1974. Catch/effort is given as numbers of fish/10 minute haul . Day p.m. Dusk Night Dusk Day a,m. Surface 4m 8m 12m 16m 20m 25m (bottom) 1 13 1 9 12 4 - 3 3 7 4 4 1139 TABLE 10.26 Seasonal variation of bottom trawl catch rates and diet of 30 to 39cm Bagrus bay ad in water of 10 - 20m depth in the Longech area. Changes in the depth of light penetration in the sampling area are also given. All data for the two years has been combined. Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Number of B . bay ad caught per hour of trawling 30 34 56 79 94 n/s 38 30 45 n/s 54 14 Diet % Prawns - 21 79 97 100 47 17 34 42 20 Fish 100 79 21 3 - 53 83 66 57 80 Water state G G B BB GB G G G G G G G Light com- pensation point (metres) 6.0 4.5 2.0 2.0 3.0 4.0 5.0 5 . 5 7.5 7.5 7.5 7.0 G = green B = brown BB = very brown GB = green brown n/s = no sample 1141 TABLE 10.27 Mean catch, per effort of Bagrus bay ad Ckg/100 yard net/night) in a fleet of experimental gillnets fished at various depths in the Longech area of Lake" Turkana~ The percentage B. bayad by weight of the total" catch, is shown in brackets. Mesh size (inches) Depth m. No . of nights 2 3 4 5 6 7 8 9 10 Mean catch/ effort % of total 9 4 - - 4° - - - - - - catch (i) Mean catch/ effort % of total 14 6 1 1 + - - - - - 1 catch (2) (1) (i) (19) Mean catch/ effort % of total 19 5 - 1 4 2 4 - - -• - catch (2) (13) (ID (63) Mean catch/ effort % of total 25 5 15 2 3 1 2 4 + 4 1 catch (36) (2) (6) (9) (56) (31) (8) (100) (8) Mean catch/ effort % of total 40 1 -- - - 8 - 10 - - - catch (40) (100) Mean catch/ effort % of total 60 1 - - - 12 12 - - - - catch (46) (60) Mean catch/ effort % of total 73 3 - 3 2 1 5 4 - - - catch (33) (29) (29) (78) (52) Results from an experimental longline fishery in the Central lake sector. Catches of Bagrus bayad are given for the period January to September 1974. 1143 CO 03 04 CO 03 00 o rH a rH UO 04 * • • 43 Hp 00 CO Hp 04 CO cn rH co rH Hp OS 04 o HP HP HP 00 bn rH CO CO O e • 3 CO HP HP 04 04 04 < rH CO rH rH co Hp C— o rH m rH 04 >. HP n o « • c < 3 CO HP HP n CO CO 3 rH co rH 03 03 03 o HP o o CO U rH CO CO o • e a o HP CO o CO CO < rH rH IP- 04 n Hp 03 o HP 04 HP CO u 03 HP CO o « • « • 3 CO HP HP CO HP CO s rH rH CO rH 03 CO in o rH o CO rH .Q rH 04 co o • • . • 43 CO Hp CO CO 00 CO fa rH rH co rH Hp CO 04 o in HP CO 04 a CO 03 © « • • • 3 CO Hp CO CO o CO rH CO 04 03 03 X cn 3 o x X fH o c3 0 bn X CS o cO 43 2 CQ o fa o fa T3 x O fa rH 0 4) bn O 0 \ ■3 fa ■H rH fa fa •h -3 2 \ X 2 x: 0 bn \ cn fa bn bn O 3 fa 03 3 bn fa - •H fa* •h 4) 3 X Sh 3 03 43 cn 43 U o bn O tn 43 s S bn 5 fa- •H 0 3 rH CJ 03 cn cn 2 2 CQ '-r 2 w x bn +■> 3 X cn 03 X X P • • • u 43 s-r bn 0 0 O 0 p H *H 3 2 2 2 fa fa fa 03 s CQ 3 3 3 JH 3 3 3 P • • 3 oj co cO 3 3 bn 0 0 4) 43 43 43 43 43 3 2 2 § s 3 3 3 3 CQ Numbers of Bagrus bay ad handled at seven co-operatives expressed as a percentage of the total number of fish bought in October, November and December 1972, 1973 and 1974. Actual number of B. bayad is given irT brackets. 1145 cn pH OS o H 0 •rH !h © o X o fH 3 ■a ctf -ps o 05 ICO CO CM CO CM 00 c* CM w iH V w 05 CO HP rH cm ao co' rH co CM o a Ch 00 rH rH w Hp rH rH CM CO 00 o' d rH o O o 05 CO 05 HP rH rH CM CO rH in ffi H f o o o CO 05 o CO CO n rH w ^ — o rH LO CO 05 rH o CM rH CM •H 0 n as +» as •rH 3 ■3 £S O aJ Z l> CO Hp CO O 05 /-S rr. rr. /*>. p. o Ch CO o 05 CM I> CO CO HP CO o o f» 05 Hp CO HP 00 '•r s.' w 05 CO CM CO CO CO in in CM o • . c c • Hp CO CO HP O rH rH rr. r. PS rr. PS o o o o o fH CO fr rH n CO CO CM HP rH w vw Q ["« HP HP CO CO hp rH HP CO rH CM CO CO CO ps o rH a o o CO in 00 o O 05 CO CM CM CO rH rH CM in CO 'W w sp w t> CO CO CM t> HP o O rH no CM CO rH Hp rH rH CM CO n r-v PN Ps r. rs o o o o o HP o CO CM 05 00 CO 05 CO 05 s^ CM O’ Cr 00 CM 00 rH CO in o Hp CM rH CM n rH rH CO HP HP / — . PS rs P. o o o o o 00 rH 00 rH HP o CO rH rH CD s-/ w rH rH o CM CO CM o o CO 05 HP • • « • • 00 c-» CM o rH CM CM PS Ps r\ PS o o in o o 05 CO m HP o rH rH S— ' rH HP 'w' ' — ' SP 00 rH m rH CO CO CM CO o CM CM rH o CM 00 rH CM CM t> CM 05 CO rH SO s >» 3 © TO o Eh CO CO O CM l> 05 +■> O o > o z u © Q © © a .18 (280) 1.97 (330) 21.83 (390) 23.79 (1810) 57.35 (1240) 68.59 (5740) 3.49 (220) 13.82 (10,010) 1147 TABLE 10.30 Estimate of mean number of : books set per night in the principle longlining are a of the Central lake sector between January 1973 and December 1974 Total number Estimated mean of Bagrus nightly number of hooks 1973 Jan 620 423 Feb 440 332 Mar 540 36 8 Apr 320 226 May 620 423 Jun 510 359 Jul 490 334 Aug 1120 764 Sep 2064 1454 Oct 2820 1924 Nov 2250 1586 Dec 3240 2210 1974 Jan 3460 2360 Feb 3520 2658 Mar 3000 2040 Apr 3340 2354 May 4750 3240 Jun 6100 3348 Jul 4530 4160 Aug 4860 3090 Sep 7550 5 399 Oct 5 830 3976 Nov 4590 3235 Dec 69 80 4761 1149 TABLE 10.31 Comparison between length distribution of Bagrus bayad catches by trawling and longlining in the Longeck area. ~ All trawl data collected between May 1972 and July 1974 and longline data from January to July 1974 have been included. Length groups cm 30 30-39 40-49 50-59 60-69 70-79 / , _ 80-89 90 Trawl catch % distribution 55 . 8 30.5 6.6 3.6 2.4 1.1 Longline catch 0 . 4 4.9 9.4 12.5 21.5 29.7 16.2 5.3 % distribution 1151 TABLE 10,32 Bagrus docmac : Length, sex and gonad condition of all. fish caught by bottom trawling between May 1972 and July 1974. Depth metres Length Number Gonad condition range cm. Immature Male Female Male Female 1972 Jun 20- 30m 64,56 1 1 resting resting Tf 5 48 1 maturing Tf 14-18 91 1 resting f? 10 74 1 resting Aug 4-8 80 1 resting Sep 4 81 1 maturing 1973 J an 4 27 1 resting Oct 16 82,89 1 1 maturing resting tt 15 14-75 2 10 13 resting Si maturing resting 1153 TABLE 10.33 Diet of Bagrus docmac. Points for each food item are expTessed as percentages of the total points in two size groups. All data from May 1972 to July 1974 are included. Number of fish Number empty Length group under 35cm over 35cm FL. FL. 10 7 1 2 Food items % Fish Lates 1c Alestes spp . f orskalii Synodontis schall auratus Fish remains 1 7 18 16 All fish 42 Insects Hemiptera Ephemeroptera Insect remains All insects 14 Crustacea Prawns Ostracods 43 1 All Crustacea 44 Total points 77 13 24 50 13 100 32 1155 TABLE 10.34 Chrysi chthys auratus : mean bottom trawl catch per effort , in kilograms per hour. Results are based on all trawl samples made throughout the lake between May 1972 and July 1974. Depth zone metres % samples with C. auratus Mean catch kg. 0-5 8 + 5-10 17 + 10-15 32 1. 8 15-20 46 1.6 20-25 47 0. 8 25-30 50 + 30-40 41 0. 3 40-50 17 + 50-60 19 0.3 60-70 15 + 70- 80 6 + over 80 0 0 + : less than 0,1 kg/hour J 1157 TABLE 10.35 Vertical migration and diel feeding pattern of Ch.rysichtb.ys au rat us over a twenty-four hour period as shown- by" frame trawling results at a station in 25 metres of water off Longech Spit. Catch rates*7 in number of Chrysichthys per time phases. Percentage stomach fullness is also presented for each time phase . Time phase Day a . m . Dusk Night Dawn Day p.m. Surface - - - - - 4m - - - - - 8m - - 1 - - 12m - - 1 - - 16m 1 ~ 7 - - below 20m 1 1 1 1 - % fullness 12 0 19 0 0 • 1159 TABLE 10.36 Ratio of males to females in Chrysichthys auratus . The sex ratio is expressed as numbers of males per 100 fish in five length groups. All data collected between May 1972 and July 1974 have been included. Length group Number of males Number of females Sex-ratio under 9cm 152 211 42 10-1 4 cm. 147 145 50 15-19 cm 63 35 64 20-2 4 cm 76 75 50 25-29 cm 6 4 60 TOTAL 472 446 1161 TABLE 10.37 Changes in maturity of Chrysichthys auratus over the size range 5 to 15 cm. Data obtained between May 1972 and July 1974 have been combined and numbers of resting, mature and ripe fish are expressed as a percentage of the total in each 1cm length group. Gonad condition Number of fish Resting Mature Ripe Fork length cm Males 5 1 100 - - 6 13 92 8 - 7 22 73 27 - 8 52 31 69 - 9 64 52 48 - 10 63 52 48 - 11 42 74 26 - 12 22 59 41 - 13 15 53 47 - 14 6 50 50 - 15 12 58 42 - Females 5 6 100 - - 6 9 89 11 - 7 37 46 32 22 8 80 39 41 20 9 75 40 31 29 10 56 38 41 21 11 51 39 35 26 12 17 59 18 23 13 17 59 18 23 14 2 100 - - 15 7 40 - - — 1163 V, TABLE 10.38 Seasonal changes in the proportion of ripe female Chrysichthys auratus . All females larger than 6cm examined between May 1972 and July 1974 have been combined and the percentage of ripe females is given for each month. Number of fish % ripe Jan 49 20 Feb 34 47 Mar 13 31 Apr 7 57 May 2 50 Jun 7 29 Jul 68 9 Aug 72 16 Sep 36 11 Oct 45 38 Nov 42 24 Dec 72 32 1165 TABLE 10.39 Diet of Chrysichthys auratus . The number of points for each food item has been expressed as a percentage of the total allotted. Results are based on all fish examined throughout the study period. Number of stomachs with food 247 Total points allotted 1596 Food items % Insects : Chironomid larvae 34 and pupae Corixids 5 Other Insects 2 Total Insects 41 Crustacea : Ostracods 31 Copepods 4 Prawns 22 Total Crustacea 57 Hydracarina 1 Fish 1 1167 TABLE 10.40 Variation in the diet of Chrysichthys auratus with depth. Number of points scored for each food item have been expressed as a % of the total allotted in each depth zone. Ferguson ' s 0-10 10-20 Over 20 Gulf metres metres metres Total points 130 308 384 774 Food Items % Insects : Chironomid larvae and 62 41 44 21 pup ae Corixids 2 11 8 1 Other insects 8 4 3 + Total insects 72 56 55 22 Crustacea : Ostracods 27 37 36 28 Copepods - 6 5 4 Prawns - 1 1 44 Total Crustacea 27 43 42 76 Hy dracarina 1 1 1 + Fish - - 2 1 + : less than 1% 1169 TABLE 10.41 Combined length frequency distribution of all Clari as lazera caught by shore seines between May 1972 and July 19747 Total length Number 30 - 39 40 - 49 50 - 59 60 - 69 70 - 79 80 - 89 90 - 99 over 100 2 1171 TABLE 10.42 Diet of Clarias lazera. Points for each food item are expressed as a percentage of the total allotted. All data collected between May 1972 and July 1974 are included. Number of stomachs with food 13 Total points allotted 132 Food items % Copepods 73 Ostracods 4 Cladocera 3 Corixid 6 Chironomid 3 Tilapia spp 11 Schilbe uranoscopus : mean bottom trawl catch per effort, in kilograms per hour, in the Northern, Central and Southern sectors of Lake Turkana for the period May 1972 to July 1974. 1173 cn o 0= as CO CO rH 1 \ + + • • • o S O 03 rH CD 09 a Ht* o O’ l \ i 1 * ■ • o a o rH o irt o m CO CO 03 i 00 1 o 1 d d 03 1 rH -40 in CO 00 c- o 1 O 1 rH rH d rH rH CO o CO 00 3D r- in as 1 m 1 CO CO* d 03 rH ca in 03 1 1 © <3 o 03 in 1 o 1 0*3 05 o 03 o CN3 rH O ea 1 CO CO CO 1 co i m 1 CO* CO rH 1 rH rH in rH CO rH CO CO CO ■o* 1 O CO* 03 03 d o rH rH -10 CO O CO m I> i m I CO 05 d 03 rH m cn i 1 1 1 \ 1 1 o a a a 0 a o -p o •p o -p o a >. P o ® 3 CQ 03 ■a a> 85 ® CQ •a ® cn e S3 S3 a a a c8 ■H •H rH a rH z o CO o S) S3 «s o •H rH rH a a g a 3 as cn •p o 0 a a ii ii cn « \ a a 1175 TABLE 10.44 Frame trawl catches, of Schilbe uranoscopus at two stations in the Longech area in September 1973. Catches are expressed as mean number of fish per 15 minute haul. Inshore Depth station 30m Offshore station Depth 82m Number of Schilbe per 15 mins . Number of hauls Number of Schilbe per 15 mins. Number of hauls Sampling depth metres 0-4 0 3 0 1 4-8 1 3 0 2 8- 12* 3 2 2 3 12- 16* 11 1 2 3 16-20 3 2 0 1 20-2 4 0 1 0 1 24-30 0 1 0 1 30-40 n/a 0 1 40-60 n/a 0 2 60-80 n/a 0 1 * : appropriate depth of midwater scattering layer n/a : not applicable 1177 TABLE 10.45 Mean catch, rate of Schilbe uranos copus in kg/50 yard net/ night in two inch and three inch nets at four stations off Longech Spit. The percentage of Schilbe uranos copus by weight of the total catch is shown in brackets. Station B D Depth (metres) 9 14 19 25 No. of nights 4 764 2" Gill net Mean catch rate kg/night 2.5 2.5 0.1 2.4 % of total catch (32) (14) (3) (14) 3" Gill net Mean catch rate kg/night 8.3 3.2 0.6 3.7 % of total catch (22) (8) (2) (14) TABLE 10.46 1179 s (D rH TJ1 00 y*N © 3 e CO • r- £ 3 t> rH as CM P rH s-/ CM © JS CO as as >» • rH e 3* © as l> CM lO CO p s rH w CM W £ * Q p H CO 13 as a « 00 • OS 33 < CO CO rH c rH W' CM w 0 * 0) p rH /-S 3 3 • as • CO p s CO rH 00 rH rH CM s •H * Cfl 3 cm «r*\ rH as • 13 5 fe CM w as r-t p p « T3 as 4) t> 3 •a as 3 3 rH in rH P y 3 3 as •a •H >“3 3 as 0 W as a rH y • CO 0 CQ © CO 3< P Q CM P 4) 0 3 3 3 # /— \ > CO CM /-S >. 0 * CM • 00 3 * 2 13 w/ 13 t> CM S-^ -P 33 33 OS P 3 •H y 3 «H o y a as 3 CO 3 0 a • CO a p © Tf o 0 33 CO CM w o I m o 0 00 /■s 3 4) bn • 31 cd 3 3 CO CM p 3 < CM W 3 -P cd 4) a CO J3 s rH ►H as O /“N •H rH O £ 6 CM a • rH CM W w 3 i> 4) as rH rH 3 /«V s as s a 4) a y y P 3 cn W © W cn © > 3 3 >> 3 3 pH rH s OS a CO © 3 CM © 3 g >> t> s 2 1> s 2 as rH as OS © 3 rH rH s Mean lengths of male S chi Ibe uranoscopus caught by bottom and frame trawls in the Longech area between July 1973 and June 1974. Immature 0- group fish (*) are also included. 1181 1183 TABLE 10.48 Comparison between observed and theoretical lengths in female Schilbe uranos copus Age (months) Observed fork length (cm) Estimated fork length (cm) 3 n/s 8. 7 6 13.4 14.9 9 17. 1 19.7 12 24. 3 23. 4 15 26.9 26.2 18 28. 1 28. 4 21 29. 8 30. 1 24 n/s 31.5 n/s = no sample 1185 TABLE 10.49 Comparison between observed and theoretical lengths in male Scfailbe uranos copus , Age (mon ths) Observed fork length (cm) Estimated fork length (cm) 2 5.4 6.4 7 14. 1 14. 2 12 17. 7 17.4 17 18. 7 18.6 22 n/s 19.1 24 n/s 19. 3 n/s = no sample 1187 TABLE 10,50 Schilbe uranoscopus : results of t-tests comparing mean lengths of females in the Central sector with Northern and Southern populations . Mean length Numbers variance t value (cm) Comparison between Northern and Central populations 1972 Nov N 22.31 91 11.44 -1.5056 C 23.13 55 7. 30 1973 Feb N 23.91 33 4.59 0.0391 C 23. 89 36 4.67 Apr N 24. 81 417 5.74 1. 7397 C 25.44 48 4. 29 May N 25.09 47 4. 86 0.7229 C 25.76 87 6.95 Comparison between Southern and Central populations 19 72 Dec S 24.50 52 7.22 1. 5806 c 23.61 41 7. 34 1973 May s 27. 14 99 2.22 7.6999 c 24. 76 87 6.95 N C s Northern population Central population Southern population 1189 TABLE 10.51 Mean length and weight in male and female Schilbe uranoscopus at ages 1 and 2 years. Age (years) Mean fork length (cm) Mean weight (g) Males 1 17. 4 42. 7 2 19.3 58.2 Females 1 23. 4 103.7 2 31.5 252.5 1191 TABLE 10.52 Schilbe uranos copus : ratio of males to females in the Longecfa area. The sex-ratio is expressed as number of males per 100 fish. All data from May 1972 to July 1974 are include d . Fork length cm. Number of females Number of Sex-ratio males 10 1 1 50 1 4 5 56 2 7 8 53 3 16 13 45 4 13 8 38 5 14 13 48 6 29 31 52 7 24 60 71 8 34 87 72 9 42 111 73 20 69 79 53 1 82 41 33 2 114 14 11 3 154 8 5 4 140 2 1 5 147 - 0 6 115 - 0 7 140 - 0 8 85 - 0 9 61 - 0 30 28 - 0 1 8 - 0 1193 TABLE 10.53 Schilbe uranoscomis : size at maturi ty results for all females and males caught between May 1972 and July 1974 in the Longech area. Numbers of fish at each gonad stage ara given as a percentage of the total numbers in each cm length group . Fork length Number of Gonad stage (%) cm fish 2 3 4 5 Females 10 1 100 - - - 11 4 100 - - - 12 7 100 - - - 13 16 100 - - - 4 13 100 - - - 5 14 100 - - - 6 29 100 - - - 7 24 100 - - - 8 34 88 12 - - 9 42 90 7 3 - 20 69 87 13 - - 1 82 89 11 - - 2 114 69 30 1 - 3 154 53 47 - - 4 140 48 51 1 - 5 147 37 61 2 - 6 115 31 66 3 - 7 140 21 76 3 - 8 85 25 72 3 - 9 61 18 80 2 - 30 28 7 85 8 - 1 8 - 100 - - 2 3 4 5 Males 10 1 100 - - - 1 5 100 - - - 2 8 100 _ - - 3 13 100 - - - 4 8 100 - - - 5 13 100 - - - 6 31 87 10 3 - 7 60 77 23 - - 8 87 71 28 1 - 9 111 75 25 - - 20 79 72 26 2 - 1 41 54 46 - - 2 14 43 57 - - 3 8 13 87 - - 4 2 - 100 - - 1195 TABLE 10.54 Seasonal changes in the maturity stages of Schilbe uranos copus from trawl catches in the Longech area. Data collected between May 1972 and July 19~74 have been combined and numbers of a) females larger than 18emFL. and b) males larger than 16cniFL. at each maturity stage have been expressed as a percentage of the monthly total. Month Number of Gonad stage (%) fish 2 3 4 5 Females Jan 22 36 64 - - Feb 49 76 24 - - Mar 132 42 56 2 - Apr 249 26 72 2 - May 131 38 61 1 - Jun 28 57 29 14 - Jul 36 97 3 - - Aug 28 50 50 - - Sep 181 72 26 2 - Oct 151 67 32 1 - Nov 131 34 ' 66 1 - Dec 89 38 15 - - Males J an 13 85 15 - - Feb 8 100 - - - Mar 60 88 12 - - Apr 59 69 31 - - May 57 81 19 - - Jun 30 73 27 - - Jul 21 86 14 - - Aug 9 67 33 - - Sep 74 57 41 3 - Oct 60 70 28 2 - Nov 41 61 39 - - Dec 25 76 24 - - 1197 TABLE 10.55 Food of Sciiilbe uranoscopus. Data collected throughout the lake from May 1972 to July 1974 have been combined and points for each food item are expressed as a percentage of the total points recorded. Stomachs with food 520 Empty stomachs 160 Total points 2148 Food items % Fish Alestes sp 20.0 Engraulicypris stellae 13.1 Haplochromis macconneli 1. 4 Barbus cf. anema 0.7 Bagrus bay ad 0.7 Fish remains 21. 4 Total fish 57.3 Crustacea Prawns 40.6 Copepods 0. 1 Ostracods 0.3 Total Crustacea 41.0 Insects Corixids 0. 4 Locusts 1.2 Chironomid larvae 0. 1 Total Insects 1. 7 1199 V> TABLE 10.56 Schilbe uranos copus : diet with relation to depth. The number of points scored by each food type has been expressed as a percentage of the total points allotted in each depth zone. Results are based on all data collected throughout the lake between May 1972 and July 1974. 1201 TABLE 10.57 Schilbe uranoscopus : diet with relation to size. The number of points scored by each food item has been expressed as a percentage of the total points allotted in each of four length groups. Results are based on all data collected throughout the lake between May 1972 and July 1974. Length groups cm. Under 15 15 to 19 20 to 24 25 to 29 Stomachs with food 22 26 40 33 Empty stomachs 5 7 11 8 Total points 42 69 153 174 Food items % Fish Alestes spp - - 1.3 25.3 Engrauli cypris stellae 4.8 - 7. 8 11.5 Barbus cf anema - - 2.6 1.2 Fish remains - 2.9 13.1 40.2 Total fish 4. 8 2.9 24. 8 78.2 Crustacea Prawns 85.7 97.1 75.2 21.8 Ostracods 4. 8 - - - Total Crustacea 90.5 97.1 75.2 21.8 Insects 4. 7 - - - Changes in the diet of Schilbe uranos copus over a twenty-four hour period at a gillnet station within Ferguson's Gulf in July 1974. Points gained by each food item are expressed as a percentage of the total points recorded in each of six time phases. Catch rate (numbers of fish caught per 50 yard net) is also given . 1203 03 o fi bO •p SS s •1"} o o s o o fi « 0 oo 04 s o ft 03 0 fi fi JS o o fi o o 3 e • S3 CO 00 a o o o I I I I I I in o * o O cn ft 0> r-4 O o fi • • S3 Tp CO W 04 o co n i 04 lO 04 04 bfi S3 •H • S3 03 D o fi > fi Jfi in cn CD rH 04 04 cn 04 CO o o cq • rH 04 04 • 04 Tp 04 CO m t> m 04 N "3 fi O si 0 Si 00 fi 03 3 03 fi D fi r< -fi fi 3 O fi O 03 \ fi fi fi •3 fi 03 fi ■fi S s D 3 fi D 03 0 3 fi 3 fi fi 03 fi fi O •h 03 fi s: 03 a ft 0 fi o 3 ft 03 fi 3 fi >. X fi fi O 0 £ fi D 0 fi fi 03 < 0 a 0 • ft 3 fi fi s fi fi £5 ft >> fi 0) fi fi 03 CD D D fi 03 O 3 a 03 t a 03 V 3 rH 03 e rH 03 rH rH •H G 3 -P -P ►O G 0 O -p GS •a G 0 03 G G JG >> -p CM co tr. «w C3 0) O rH rH G 03 >i a bfl G G s -P 0 G G 03 03 03 y 03 a P > 3 03 -P 0 0. 03 P St bO G •P J3 03 J3 -P G bQ bD 3 C T3 G 0) 03 y rH 03 03 J5 03 03 cn 03 P •H P a

> O ■p 03 •H &C T3 p G 0) 3 -P 03 P 03 G G J3 s rH -P 03 • G •H •H p •H P 0 3 G -p 03 ■P -P y 03 G G y bJO a Q 03 § rC y 03 r+ y • as !> O G JS G • y -P rH o rH G rH G -p 0 rH G G a G w O P rj 03 JG .C -P ca G 03 y G < 03 •H G y H 02 «H y o O CM CO CD 03 Q l> CO CM CM CO CM I> co n w in t- 03 CO in > o o O 00 CM 2 m o4 rj< cm rH 03 CM 03 m CO CO CM O ih co •p in m O O CM o co O O O i o4 i> co rH cm t» H O4 03 m a io ^ y os CO co co o) n i h i h o n o CM H in N 03 t> CM c- bfi O4 CM ^ 3 t> CM < I> o H CO H O O4 CM O4 CM CM f* CM CM 03 CO th n t» 3 CO H |-o O cm co m i o cnot^i m cn m4 m co rH oo ’ o4 S rH O4 O4 3 H H •-J i> co » i co cm in H CM t> CM I 03 CO cm Sh H CO « CO CO H s co co n co i C" G4 rH fr rH 00 03 00 rH rH co CM 00 p o4 CO a ■o* in < o4 co cf n i ® co co o4 CO 03 03 00 rH CO l> p CM 00 | 1 rH I> m 00 CM 03 00 1 rH G in G4 CM co G4 00 CM § rH rH ja CM 00 1 1 CO m CO CM CO rH CO CO 1 G4 y i> CM CO t-H 00 G4 rH 00 C3 Ch a t"» co G CO co >-3 l> CO CO CO H ® 03 C>- G4 I CO cm h co co h in cm m 2 3 4 in CM CO G4 m 2 3 4 5 y y y bo be bO G G G •p -P P CO CO CO P p p a y a y a y y S y S y J2 03 a G4 g 03 a rH 3 CM 3 CM 3 1 2 1 2 1 2 m o m rH CM CM Seasonal changes in maturity of three length groups of female Synodontis schall . N of fish at each maturity stage are expressed as a percentage of the total number ex each month. Data is based on all fish caught between May 1972 and July 1974 in the Central lake sector. 1247 03 u CD -Q 05 in co 03 m h o m > CO I c* 0 03 2 O'1 CO CO rH H O' CO CM CO CO o o H H O' CM CM (N CM CO ■M o o o o CO CO CO 03 H O' CO rH CM CO o CO O' CO CM O' O CO a 03 rH 1 1 1 00 m rH CO CM CM 03 c- o 00 + O' rH ® 00 rH CM co CO CO CM CM CO CO o co rH CM 60 00 CM i 1 i in CM 03 o« O' rH o* O t- CO rH 00 CM 3 CO rH CM CO CO CM O O’ CM CM CO < rH rH rH a CO O 1 i O 00 rH co 1 O' 03 CM 03 rH 1. 00 3 03 CO CM CO CO 00 CO rH O' <3 1-3 s 3 '■o o o t> m cm co rH CO CM co m o m i i co cm cm in CM >. o 01 cs a s CO CO 03 CO CM O' O' CM CM CO rH CO O' CM in rH CO CM CO CO CO 1 1 1 t> t> m C**9 1 rH CO rH CO CO 1 I o 00 rH rH CO O' rH O rH rH CM 00 rH O I I 1 1 rH O CO O I 1 CO o I> CO 1 1 co o rH ^ CM rH CO rH rH m rH rH CM 8 8 4 1 O' rH 03 03 1 iH CO O' co t> C +-> ■P CO co CO u Sh Sh E E cn a O 3 o 2 be o Sh CM 03 fci c 3 P G X2 C3 *H P o 03 Ih 03 03 G bfl •H u a 1) 3 0) 03 rH cn JS cn o 03 3 rH 03 3 bfl JS a 3 03 < >. J3 •a P a 3 H bD 3 3 cn 3 03 O H 3 rH a H G 3 3 x: O ’“5 C3 3 cn Sh 03 03 ■H JS >» P a 3 a 3 s o a "3 0 03 c bn >. 3 Sh CO P • a S3 X! < 03 P 0 CJ S3 Sh O a 03 a o CU Sh •H JS 3 P C3 a •H • 3 •o 3< 03 3 f» O 03 Sh u H 0 X5 , Cm 3 H G c 3 03 o l-3 > bD H T3 bfl G a G 3 •H 3 03 3) Sh cn CM 3 a> t> bfl C3 03 G H on 3 3 XO >> P O 3 cn cn S 03 H >> H 3 c p 3 G 03 •H a O 03 Sh 03 cn 3 3 Sh 3 p P (1) 03 3 CO JS a oo 03 03 3* 03 00 co CM OO m 3< m oo 03 3< cn co 00 CM <13 bfl 3 p co in t> CM CO CM in 3< cn CM CO CO CO 03 ih ih i— i in CM 03 m i i cn cm c-* I I I 03 I I I 03 CM 03 I H CO H 03 CO ® n » rH rH CO co 00 rf in n n m 'f I I o 3* CM cn h i co CO t> I I I CM I CO o cn 3< m co u 03 JS cn a 0) - 3 rH 2 cd o1 co i i 3< H 00 H 03 O H I CO H CO 03 CM CO CM I O' CO CO 31 l> o CO I CO h m co 03 t- 3* I CM CM 3* CM CO oo m o* h 03 3* 31 co H 3< [- 00 O’ CM in H CM co m co O oo cm m h o1 I I CO CO 00 o CM I CM CM CO H 00 CM CO | I CM m h m ^ h i 03 cm co if in 03 bj 3 3 +> CO Number 127 1251 TABLE 10.82 Seasonal changes in maturity of male (M) and female (F) Synodontis s chall larger than 20cm in the northern sector. Numbers of fish at each maturity stage are expressed as a percentage of the total number examined each month Gonad stage Number 2 3 4 5 6 1972 Nov F 121 4 22 66 2 5 M 131 5 36 54 5 1973 Feb F 41 46 24 M 28 - 100 - - Apr F 188 10 3 86 1 - M 142 - 77 23 - Jul F 27 33 19 48 - - M 22 18 82 - - Sep F 164 84 10 6 - - M 133 71 29 - - Nov F 99 87 10 3 - - M 49 37 65 - - 1974 Jun F 28 7 18 39 36 M 12 17 17 67 - 1253 TABLE 10,83 Occurrence of post- larval Synodontis schall in meter townet catches within 10 kilometres of the Omo delta. Catch rate "(mean number of post larvae/haul) has been combined for the period November 1972 to September 1975. Month Number of hauls Mean number of post- larvae per haul J an n/s Feb 2 0 Mar n/s Apr n/s May n/s Jun 6 60 Jul 4 7 Aug 10 14 Sept 4 134 Oct 25 7 Nov 5 1 Dec n/s n/s no sample 1255 TABLE 10.84 Seasonal changes in maturity of male (M) and female (F) Synodontis schall larger than 20cm in the Southern sector. Numbers of fish at each maturity stage are expressed as a percentage of the total number examined each month. — Gonad stage Number 2 3 4 5 6 1972 Jun F 39 10 21 40 8 — M 25 16 32 48 - Dec F 12 92 8 - - - M 9 78 12 1973 May F 31 6 16 68 10 — M 31 - 29 68 3 Oct F 112 39 48 10 - 1 M 84 18 82 - 1974 Jun F 58 36 19 43 2 M 31 23 67 10 — 1257 TABLE 10.85 Ratio of male to female Synodontis schall in the Central lake area. The sex-ratio is expressed as number of males per 100 fish, and all fish caught by the bottom trawl between May 1972 and July 1974 have been included. Length cm 15 6 7 8 9 20 1 2 3 4 5 6 7 8 9 30 1 2 3 4 Number of males Number of females 22 30 38 65 126 189 263 317 319 323 355 394 339 236 145 60 23 8 5 4 Sex ratio 50 54 55 50 53 53 52 49 46 43 40 38 30 26 22 17 12 33 17 0 22 35 47 66 142 211 281 302 277 241 236 242 145 83 42 12 3 4 1 0 1259 TABLE 10.86 Ratio of male to female Synodontis s ch a 1 1 in the Central lake area. Sex-rati6 is expressed as numbers of males per 100 fish for two length groups in each month and includes all fish caught by the bottom trawl between May 1972 and July 1974. Sample size 20-2 4cm Sex ratio Sample size 25-29 cm Sex r Jan 260 50 173 34 Feb 289 49 290 32 Mar 311 48 374 32 Apr 375 53 256 30 May 213 41 217 37 Jun 49 47 17 53 Jul 149 40 113 40 Aug 224 54 221 40 Sep 220 41 271 26 Oct 87 51 45 33 Nov 205 50 94 34 Dec 363 48 167 34 1261 TABLE 10 . 87 Gulf. The sex-ratio is expressed as numbers of males per 100 fish in two length groups for each month. All data collected between May 1972 and July 1974 has been included. S amp le size 20- 24cm Sex ratio Sample size 25-29cm Sex ratio Jan 44 50 194 49 Feb 6 67 18 44 Mar 82 50 73 55 Apr 28 5 7 32 53 May 107 59 49 49 Jun 34 68 29 38 Jul 111 46 51 33 Aug 65 71 32 41 Sep 10 40 4 25 Oct 74 30 40 53 Nov 66 52 118 47 Dec 10 40 19 42 1263 TABLE 10.88 Synodontis schall : seasonal changes in mean condition factor (k) of males and females larger than 20cmFL. in the Longech area. Data from the two years fieldwork has been combined and k is given for each mon th . Month N umbe r k Standard Devi ation a) Females Jan 60 101 11. 7 Feb 29 97 9.8 Mar 32 97 9.0 Apr 82 105 9.3 May 74 103 11.3 Jun 72 100 10.5 Jul 8 93 10. 4 Aug 42 99 10. 2 Sept 66 108 9. 8 Oct 59 107 9.9 Nov 13 106 11. 7 Dec 15 101 10.0 b) Males Jan 16 100 6.9 Feb 60 95 9.0 Mar 40 97 9.6 Apr 67 99 9.4 May 79 92 8.9 Jun 34 95 9.7 Jul 5 89 7.6 Aug 47 96 10. 8 Sep 40 103 7. 8 Oct 30 103 11.2 Nov 17 107 8.9 Dec 12 94 7. 7 1265 TABLE 10 . 89 Embryological and larval development of Synodontis schall Approximate time after fertilisation Plate Stage of development 00 mins Egg diameter 1.2 to 1.4mm. Translucent yellow colour. 20 mins * 1A Outer covering swollen with a sculptured surface. Egg diameter 2.2mm. Yolk diameter 1.0 to 1 . 2mm . 1 hr. 30 mins. IB First blastomere formed. Approximately 0.5mm in length. 2 hrs. 1C and ID Division proceeding rapidly. 16-32 blastula. 4 hrs . Blastodermal cap 1.0mm in length covers ^ of yolk. Maximum cell diameter 0.15mm. 5 hrs . Gastrulation commences. Marginal ridge visible. 5 hrs. 30 mins . Blastopore closes. 8 hrs. IE Head and tail folds of embryo just visible. Embryo encircles approximately half of yolk. 10 hrs. First somites visible. 18 hrs. 20 somites present. Posterior end of embryo lifted slightly from yolk. 20 hrs. Head fold well developed, caudal region free from yolk sac. Auditory capsule visible. Somites extending to caudal region. Slight lateral movement. 23 hrs. Embryo 3.0mm in length. Fin fold forming, more developed ventrally than dorsally. Two pairs of otiliths now visible. 24 hrs. 1H Hatching after vigorous movement of embryo. Optic vesicle present but not pigmented. Contd. / 1267 27 hrs . 33 hrs. 45 hrs. 47 hrs. 48 hrs. 56 hrs . 6 8 hrs. 74 hrs. 93 hrs. 102 hrs . 116 hrs. 129 hrs. 11 1J and IK 1L 1M IN Larvae 3.8 to 4.0mm in length, swimming with elliptical yolk sac in contact with dish. Eye pigmented and slight pigment- ation, both dorsally and ventrally, in caudal region at base of fin fold. Fore and mid-brain prominent. Approximately 30 somites present. Three pairs of barbel buds developing in buccal region. Occasional flicks of body enables larva to move into midwater. Larval length 4.5mm. Pectoral fin bud visible. Nostril forming. Larvae 4.7 to 5.0mm in length. Lens forming in eye. Gill rudiments present . Anterior margin of maxillary barbel crenelated. Pectoral fin bud well formed. Larvae swim vigorously. Mouth opens and closes. Peristalsis visible in gut. A few melanophores present in head region. Blood supply to gills. Yolk sac empty. Maxillary and mandibular barbels lengthened and crenelated on anterior margin. Three pairs of otoliths visible. Food in gut. During feeding barbels splayed out and mouth closely applied to food. Larvae 6.4mm in length. Increased head pigmentation. Gut folded. Mandibular teeth visible. Larvae 6.8mm in length. Caudal and pectoral fin rays present. 141 hrs. TABLE 10.90 1269 Synodontis schall : Food in relation to depth. Points for each food item are expressed as a percentage of the total allotted in each depth zone. All stomachs examined between May 1972 and July 1974 have been included. Depth zones, metres. 0-5 5-10 10 -20 20-40 40-60 Over Ferguson 1 s Allia 60 Gulf Bay Number of stomachs with food 39 108 270 390 32 40 383 43 Total points 337 730 1505 2321 120 151 2792 275 Food items % Crustacea Ostracods 8 27 23 31 20 10 28 23 Copepods 28 10 48 37 7 5 49 - Cari dea 1 + 6 13 11 7 + - Cladocera - 23 + 2 - - - - All Crustacea 37 60 77 83 37 22 77 23 Insects Corixids 23 5 + 1 - - 4 20 Chironomids 15 2 1 2 11 5 13 24 Notonectids + - + - - - - - Ephemeroptera - 2 1 - - - - 3 Tri choptera - 1 1 - - - 4 Terrestrial insects 4 - + - 2 - + - All Insects 42 10 4 3 13 5 17 51 Gasteropods Cleopatra pnothi - - 1 - 33 21 - - Gabiella rosea - 1 - - - - - Melanoides tubucul ata - + - - - - - Unidentified - + 1 l 17 19 - 2 All Gasteropods - 3 l 50 40 - 2. Fish Engrauli cypris stellae - - - + - 1 2 • - Lates longis- pinus - - + + - - - 4“ Ti lapia spp . - - - + - - - 3 Barbus of anema - - - + - - - - Alestes spp. 14 22 11 + - - - - Scales (B.bynni L. ni loticus) - - - - - 32 - - Unidentified 2 3 3 2 - - 2 6 All Fish 16 25 14 12 - 33 4 9 Vegetable matter 5 2 - - - - - 4 Algae - 3 1 - - - + 4 Di atoms - + - - + 7 + = less than 1% 1271 TABLE 10.91 Changes in the diet of Synodontis schall through a twenty four hour period at a gillnet station in Ferguson's Gulf in July 1974. Points gained by each food type are expressed as a percentage of the total points recorded during six time phases. Catch rate (numbers of fish caught per hour per 50 yard net) is also given. Afternoon 12.00 - 18. OOhrs . Dusk 18.00 - 20. OOhrs. Late evening 20.00- 24.00 hrs . Early morning 24.00- 05.00 hrs . Dawn 06 . 00- 08.00 hrs . Morning 08.00- 12.00 hrs . Total number of fish caught/ 50yd net 69 40 17 61 15 7 10 Mean number/ hour 11. 5 20.0 4. 25 10.25 78.25 2.5 Number of stomachs with food 17 21 14 14 19 10 Number of empty stomachs 4 1 3 1 1 Total number examined 21 22 17 15 20 10 Total number of points 76 86 52 34 138 67 Mean number of points/ fish 3.6 3.9 3.1 2.3 6.9 6.7 Food items % Chironomi d 28 72 77 44 24 4 Corixid 3 3 8 - 2 2 Ostracod 35 19 15 44 9 16 Copepod 33 6 - 3 49 72 Hydracarina 1 - - 3 1 - Fish remains - - - 6 5 Diatoms and Algae - - - - 10 6 Mean retention lengths in cm (1) of Bagrus bay ad, Synodontis schall and Schilbe uranos copus for nine mesh sizes of gillnets. All data collected between May 1972 and July 1974 have been included. Standard deviations (sd) and sample sizes (n) are also given. 1273 Appendix 10. B 1275 Bagrus bay ad : gillnet selection. Length frequencies of fish caught between May 1972 and July 1974 have been accumulated for each net size. . 2 3 4 Stretched 5 mesh 6 inches 7 8 9 10 20 1 - - - - - - - - 21 - - - - - - - - - 22 2 - - - - - - - - 23 - - - - - - - - - 24 1 - - - - - - - - 25 1 - - - - - - - - 26 2 - - - - - - - - 27 - - - 1 - - - - - 28 1 - 1 - - - - - - 29 - - - - - - - - - 30 1 1 - 1 - - - - - 31 - 1 - - - - - - - 32 - 1 - 1 - 1 1 - - 33 - 1 2 1 1 - 1 1 - 34 - 1 - - - - - - - 35 - 2 - 1 - - 2 - - 36 - 1 - 1 - - 2 - - 37 - 3 - 1 - 1 2 - - 38 - - - - - 1 2 - - 39 - 1 - - 1 - - - - 40 - 3 3 2 - - - - - 41 - - — 1 - - 2 — _ s 42 - 2 1 2 - - - - - w 43 - - 1 1 1 1 - - - JS 44 - 2 - 1 - - - - - 45 - 2 6 2 - - - - - 46 - 2 2 1 - - - - - 47 «r. 3 2 _ X u 0 48 - - 6 1 - - 2 - - 49 - - 3 3 - - - - - 50 - - 4 2 - - - - - 51 - - 6 4 - - - - - 52 - - 3 1 - - - - - 53 - - - - - - - - - 54 - - - 1 - 1 - - - 55 - - 2 - - 1 - - 56 - - - - - - - - - 57 - - 3 3 - - - - - 58 - - 2 1 2 1 - - - 59 - - 1 - 1 1 1 - - 60 - - 1 - 1 - - - - 61 - - 3 - 2 - - - - 62 - - 1 - 1 1 - - - 63 - - - - - - - - - 64 - - 1 - 5 - - - - 65 - - 2 - - 1 - - - 66 - - - 1 - - 1 - 2 67 - - - 1 2 2 1 - - 68 - 1 - 1 1 1 - - - 69 - - - - - 2 2 - - 70 - - 1 - - 1 2 - - 71 - - 1 - - - - - - 72 - - - - 1 - 1 - - 73 - - 1 - 1 - - - - 74 - - - - - 1 - - - 75 - - - - - 1 - - - 76 - - 1 1 1 1 - - - 77 - - - - - - - - - 78 - - 1 - - - - - - 79 - 1 - - - 1 - 80 - - - - 1 - - - - 1277 Appendix 10. C Synodontis schall : gillnet selection. Length frequencies of fish caught between February and June 1974 at four stations off Longech Spit have been accumulated for each net size. 1279 Appendix 10. D Schilbe uranoscopus : gillnet selection. Length frequencies of fish caught between February and June 1974 at four stations off Longech. Spit have been accumulated for each net size. 7 1281 Appendix 10. E Results from an experimental longline fishery in the Central lake sector. Catches of Lates niloticus are given for the period January to June 1974. J an . Feb . Mar . Apr. May . Jun . No. of nights recorded 40 19 25 12 6 18 No, Lates caught 12 69 41 13 7 37 Mean No. /night 3 3.63 1.64 1.08 1.16 2.05 Mean No. hooks/ night - 1400 1400 1400 1400 1400 1400 Mean No. Lates/ 100 hooks 0.21 0.26 0.12 0,08 0.08 0.15 Mean length Lates 122.3 120.9 121.13 111.31 117.29 121. 76 Total fresh- weight of Lates /month 303 1560 1208 25 4 145 846 Mean fresh weight/100 hooks 5.41 5 .86 3,45 1.51 1.73 3.36 .J/V" - « v# '\CkQ%