Traditional fishing in Uganda's waters was limited to subsistence methods using traps, baskets, spears, handlines, and local netting materials (Graham, 1929; Worthington, 1929; Balarin, 1985). The introduction of netting materials made of cotton and flax during the first decade of this century stimulated the development of a commercial fishery, most notably for tilapia in Lake Victoria (Garrod, 1961). The first documented fisheries surveys were carried out by Graham (1929) for Lake Victoria and Worthington (1929) for Lakes Albert and Kyoga (see summary chronology of events, Appendix III). Graham recommended, inter alia, the establishment of a management Authority for Lake Victoria, whose remit would combine both administration and research, including the collection of statistics. Worthington endorsed this recommendation, but suggested that a Lake Fishery Authority should cover all the lakes of Uganda.
Both Graham and Worthington pointed out the difficulties of characterising the state of the fisheries in the absence of statistical data. Each urged that efforts be undertaken to collect production statistics from the major fisheries through systematic recording of fish consignments shipped by road, railway, or steamer, and of market transactions in fish at major population centres. Data on place of origin and weight by species were deemed to be of particular importance. Worthington foresaw that:
Statistics, if collected in this way from the first, will be of the greatest possible assistance in following the progress of the lake fisheries, and will make any future work, which is sure to become necessary if the fisheries are to be exploited to their best advantage, a matter of much greater value [1929:32].
Although a certain amount of information on fishing methods, commercial effort, catch per net, and distribution and marketing patterns was available during the 1920s, the actual collection of catch statistics for Lake Victoria did not commence until 1933 (Mann, 1970). And it was not until 1936 that the first official fisheries statistics for Uganda were published (Uganda Game Department, 1936). Even then, the returns were not very comprehensive in their coverage, nor very systematic in their formulation.
It was only after the creation of the Lake Victoria Fisheries Service (LVFS) of the East African High Commission in 1947; that regular and more useful records began to be collected for Lake Victoria (Beauchamp, 1955). Working together, the LVFS and East African Fisheries Research Organisation (EAFRO), established in 1949, fulfilled Graham's recommendation for a Lake Victoria Fisheries Authority combining both administration and research. The LVFS had the responsibility for the collection of statistics around the whole of the lake until 1960, when its role was taken over by the governments of the three riparian territories of Kenya, Tanganyika, and Uganda (LVFS 1959; EAFRO 1959). In Uganda, other water bodies with the exception of Lake George were covered during this period by the Fisheries Section of the Game Department, which became the Game and Fisheries Department in 1950 (Sastry, 1957; EAFFRO, 1966 suppl. 3). The Fisheries Section became a Department in its own right in 1961, and since that time has assumed the primary responsibility for the collection of fisheries statistics in the country.
Various observations have been made about the reporting procedures which were in force on Lake Victoria and other Uganda water bodies during the 1950s and 1960s. The 1950 Annual Report of the LVFS provides a brief description of the data collection system which the Service implemented during that year for Lake Victoria. The system involved daily recording of various net catches on specified beaches around the lake. For Uganda, the same report states that “it has been more difficult to train the staff…as they had previously been trained in other methods”. Unfortunately, no descriptions have been found relating to these previous “other methods.” Garrod (1957), in his review of the Lake Victoria Fishery Service records for 1951–56, observes that the collected data do not give a complete record of catches landed and should therefore only be used for general indications of trends and fluctuations.
The most detailed of the earlier accounts of fisheries data collection in Uganda is that provided by Sastry (1957). His report comprises the first comprehensive assessment of the situation on a region-by-region basis, and is here reviewed at some length in order to set the background against which the development of fisheries monitoring and reporting in Uganda can be understood.
Sastry categorised the various data essential for proper monitoring and planning activities in the fisheries under five headings. These included: production factors; production; utilisation, demand, and supply; socio-economic data; and biometric data.
Sastry identified multiple weaknesses in the system of data collection in effect at the time of his evaluation. In terms of production factors, he noted that whereas the locations and geographic extent of fishable waters in the country were relatively well known, information on exploitation patterns was highly inadequate. Numbers of licensed vs. unlicensed fishermen, non-operating owners of craft and tackle, and those people employed in side activities like boatbuilding, etc., were incompletely known. Even for places where such information was being reported, Sastry judged that it was largely based on guesswork and rough approximation, and that it was therefore unreliable. He reckoned that the records covering the licensing of canoes were somewhat better organised, expecially for the western zone covering Lakes George, Edward, and Albert. Canoe licensing during that time was vested in African Local Government bodies or, in the case of Lake Victoria, the local Sleeping Sickness Inspectorates. Sastry noted however that problems existed with the transmission of craft licensing information between these authorities and the Fisheries administration, and commented that:
As long as the respective fisheries authorities are not either exclusively vested or at least directly associated with the licensing responsibilities, the information is not available to users either in sufficient detail and accuracy, or at the appropriate time [ibid: 13].
The situation with regard to gear enumeration was deemed equally unsatisfactory. The LVFS was responsible for issuing permits for beach seines, and these gear were the only ones subject to licensing requirements. Approximations of the numbers of seines in operation could be drawn from the licensing records, but of course did not reflect the numbers which were being operated illegally. Gill net information based on import figures from custom authorities could be compiled, but was lacking in both detail and accuracy. Other gear and equipment (hooks, traps, twines, floats, engines, etc.), was not being systematically enumerated at the time of Sastry's evaluation. Such information as was being compiled was thought to be based mostly on guesswork.
Sastry observed that the statistics on fish production, which were for the most part collected in the field by fish guards, were deficient on several grounds. The most fundamental weakness recognised was the lack of standardised procedures which would ensure uniformity in the quality of data within and between localities and across time.
Reporting on the situation within different fisheries areas, Sastry noted that for Lake Victoria, the LVFS had posted some twenty fishguards at various stations along the Ugandan shoreline. These personnel were assigned the task of keeping daily records on the number of active canoes together with their crew, type of gear used, and number of fish landed (both fresh and smoked) by species. These data from the various enumeration centres were used as the basis for annual reports, along with the information made available from the Sleeping Sickness Inspectorates on canoe registration, and from the local African Government authorities on the number of licensed fishermen. LVFS records on beach seine permits and results of occasional sample weighings of different species of fish provided additional information. Figures on annual catch tonnages along the Uganda shoreline of the lake were compiled through these various sources, although no breakdown by months or regions or gear was provided. Figures on average number per unit of gear of Tilapia esculenta and of all species together were computed on the basis of fishguards' reports on number of gill nets and other gear at the selected landings.
While this approach to information collection must have generated a considerable body of material, it was critically deficient in that the fishguard stations selected did not necessarily represent a fair cross-section of the shoreline. As Sastry commented, the Uganda shoreline of Lake Victoria included “… a large but unknown number of landings with unknown magnitude of activity” (ibid: 15). Further deficiencies with the Lake Victoria reporting system during the mid-1950s included the way some fishguards would rely on the word of boat crews or fishmongers for information pertaining to the quantity and species composition of catches. Doubts as to the accuracy of canoe registration, seine permit, and species average weight records were also expressed, raising questions about their validity as a means to determine total production, and their usefulness in helping to provide breakdowns of catches according to different seasons, regions, or gear types.
With regard to Lake Kyoga, the Uganda Game and Fisheries Department had some eight fishguards posted at various strategic points around the complex during the time of Sastry's evaluation. Sastry remarked that “Each fishguard records statistics of fish landed (number and weight) by species and type of gear at his main duty station and a few surrounding landings, the number and frequency of his visits to the latter largely depending on his own discretion” (ibid). In those areas where wholesale or primary markets were held, fishguards were “…obliged to visit the markets occasionally and make a record of the arrivals of smoked and fresh fish” (ibid). End-of-the-month returns compiled by each of the Kyoga fishguards were based on information obtained from a limited number of landings visited on a limited number of days, with no particular schedule being used to standardise procedures. The monthly returns were used to determine annual production levels from the Kyoga complex:
On the basis of an assumed total number of fishing days at each place, estimates of the annual catch at these centres are worked out. Estimates of annual arrivals of fresh and smoked fish (converted to fresh at an assumed equivalent rate) at the weekly and fortnightly markets visited by the fishguards are worked out similarly on the basis of reports submitted to them. To these are added ad hoc estimates of catch from areas not covered by the fishguards to determine the annual catch from the lake [ibid: 16].
In Sastry's view, this monitoring and reporting scheme was ill-equipped to cope with the complicated realities of the fishery of Lake Kyoga. These realities included numerous remote and secluded landings situated about the swampy and indented shorelines of the Kyoga complex, shifts in access channels to fishing grounds and landing points, and the marked seasonal variations in fishing activity due to migration of fishermen both between different landings and between the fishery and other economic pursuits. Sastry characterised the Kyoga fishery as one which was in a stage of gradual development, and for which precise information on all landing site locations and activity levels was lacking. He went on to say that:
In such a context, any guess about the contribution from areas not covered by the fishguards is bound to be crude and of uncertain accuracy. The fishguards are not guided by any objective programming of their visits, even to the centres included under the present coverage. The present recording arrangements can hardly provide a realistic guidance for estimating the total number of fishing days or the seasonal variations at any given centre. There is also the possible danger of double counting when arrivals at the primary markets are added to catches determined from records at the landing points. At landing points that are located close to primary markets, there is a possibility of some landing activity even on non-market days which is not presently accounted for. The present arrangements do not provide for breakdown of total catches by seasons, regions or type of gear [ibid: 16].
For Lake Albert, the fishguard staffing situation at the time of Sastry's study was extremely thin, with the Game and Fisheries Department personnel operating from only five central points. This arrangement supposedly allowed access to the estimated 35 to 50 landings scattered along the shoreline. However, recognising the inadequacy of geographical coverage, the Department chose not to use fishguards' reports as a basis for “… extrapolation in time or space in evaluating total production” (ibid: 17). Instead, the reports were only being used:
(i) to determine the percentage composition of different species, treating the catch recorded by fishguards as a sample of the total catch, and (ii) in determining the selectivity of the different types of gear used, by working out the catch of different species per unit of each gear type [ibid].
Total production was however estimated in a different way, which entailed reference to custom returns on salted fish exports through the ports of Mahagi (on the then Congo side) and Vurra (on the Uganda side):
A ratio of 1:3 is used to determine the fresh equivalent, and to the quantity so determined is added estimated local consumption, quantities sold to inland markets and contributions from subsistence fishing, to work out the total production from the lake [ibid].
Raising serious reservations about the adequacy of the Lake Albert method of production estimation, Sastry commented as follows:
The ad hoc estimates are of unknown accuracy. Even estimates of the large bulk of exports to the Congo are of unknown accuracy because they are based on the voluntary declarations of buyers at the customs ports, and because of possible inaccuracy in conversion factors used to work out the fresh weight equivalent. Ideally, these figures should only be used as an external check on production, the principal determination of which should be based on records collected at the production points [ibid].
With regard to Lake Edward and its Kazinga Channel tributary, the situation as reported by Sastry was again one of a limited field staff establishment, though not perhaps as serious as in the cases of Lake Albert and Lake Kyoga. A fishguard was posted at each of the four authorised landings on Lake Edward and two at Katunguru (one on either side of the Kazinga Channel), to supply daily reports on the catches landed, by species and gear. In view of the large size of Katwe landing, the daily routine of the fishguard posted there was to provide information only on the fishing canoe traffic in and out of the beach. More detailed statistics on fishing activity were then collected by a small team of extra fishguards, who would visit Katwe for two or three consecutive days each month. Monthly estimates of production were based on the information collected during these intervals, appropriately multiplied to cover the entire 30-day reporting periods. Sastry cautioned that the use of this approach at such a large centre like Katwe “… may not be precise enough and also not economical in view of the similarity one would expect in the activity of successive days” (ibid: 18).
The minor lakes of the Lake Edward region and Kigezi District were given irregular coverage by fishguards during the mid-1950s. Lakes Nakivali and Kachira, being larger bodies of water, each had one fishguard assigned to them. Another fishguard was posted to cover some half dozen of the minor lakes in Kigezi, following a rotation scheme wherein each lake would receive a visit for a few days each month. Ad hoc estimates were made for the production of all the remaining minor lakes not covered by fishguard visits. Sastry did not regard the coverage of the minor lakes as being any more effective than that found for the major fisheries. He concluded that: “The estimates for the principal minor lakes on which fishguards are available suffer from the drawbacks arising from the personal discretion allowed to the fishguards in compiling data from the several centres comprising their jurisdictions” (ibid).
A special reporting system was in force for the fisheries of Lake George, in that information on catch returns was provided by The Uganda Fish Marketing Company, or TUFMAC. TUFMAC operated a processing plant on the lake at Kasenyi, and had a monopoly right to fish collection and purchasing from all the landings around the shoreline. The TUFMAC records were regarded to be of quite good quality and detail, as far as they went. Sastry described the reporting procedures as follows:
A receipt is issued in the name of each canoe contributing to the catch, bearing the numbers and weights of catch of different species delivered to TUFMAC. When fish are brought over to Kasenyi from each landing centre, there is a second check on the total quantity…. Daily statements of incoming quantities, disposal and recoveries of dry salted fish and fillets are made out. Monthly statements are prepared under the headings of landing and allocation by different species along with recovery percentages and stock position. Statements are also made at the end of each month showing the number of participating canoes, the number of fish, the average weights per fish, the total poundage, and the actual number of canoe days by species. These statements form the basis of statistics reported for Lake George in the annual report of the Game and Fisheries Department [ibid].
With regard to utilisation, demand, and supply, Sastry stressed that the monitoring of these areas was crucial to the task of ensuring efficient and balanced performance of the fisheries for the benefit of the country. Successful monitoring would require reliable data on total consumption of fish, total quantities handled by type of cure, internal and external distribution patterns and volumes, and prices at different levels of the distribution and marketing network. Such information was generally found to be lacking or only available in unsystematic records. For Lake Victoria, Sastry remarked that fishguards were only collecting data on “… quantities of smoked fish arriving at the established recording stations” (ibid: 19). No information was being collected on source or inland destination of these products. TUFMAC records for Lake George were found to be providing fairly complete accounts of the proportions of fish handled according to different modes of disposal, but this was an exceptional case. Statistics on imported and exported fish products, available through returns provided by customs posts and the East African Railways and Harbours Authority were judged to be unreliable and unsystematically collected.
Qualitative and quantitative information on socio-economic characteristics of the fishing population was recognised by Sastry as being a fundamental requirement to development planning for the sector. Data were needed on such characteristics as the size and distribution of the fishing population, family composition and dependency patterns, alternative sources of income, housing and sanitation conditions, literacy rates, and general levels of income and wealth. Sastry found, however, that such information was almost totally lacking.
With reference to biometric data, Sastry listed such areas as population dynamics, gear selectivity, and seasonality as important foci of research by fisheries biologists. He noted that work through the East African Fisheries Research Organisation had been concentrated especially on the biology of tilapia species.
It was Sastry's general observation that fisheries statistics in Uganda were in need of substantial improvement and enhancement in terms of their geographical coverage, accuracy, and scope of items enumerated. While periodic comprehensive inventories could be carried out to determine the state of production factors and other characteristics of the fishing population, to serve as a kind of baseline measurement which could be recalibrated every five or ten years, Sastry did not favour the technique of complete enumeration as a routine part of the information collection system. He regarded the latter as “prohibitively expensive” and as something “…bound to result in organizational imperfections leading to errors of a much more complicated and difficult nature than those noticed at present” (ibid: 23).
He advocated instead the technique of representative sampling, pointing out its advantages in terms of economies of operation and potential for yielding information of both greater scope and reliability. As a means of exploring possibilities for improvements in statistical collection, particularly in the area of production, he organised pilot sample surveys to be mounted at different centres on Lakes Victoria, Kyoga, and Edward, alongside the existing official system of enumeration. The approach used was broadly the same for all the lakes, with components being varied from place to place depending on local circumstances and to assess the utility of alternative procedures.
The way the investigations were conducted on Lake Victoria can serve as an illustration of Sastry's basic approach. Three centres (Masese, Bugonga, and Bukakata) were selected as the pilot sites, and several survey forms were designed to be administered by specially trained fishguards. Certain information on each canoe departing from the chosen landing sites was collected on a daily basis, including registration number, whether motorised or not, purpose of trip (fishing, carrier boat, in transit), type and number of gear, time of departure, and size of crew.
Further information on catches landed was collected through a basic sample enumeration technique wherein five survey days were randomly selected each fortnight. On each survey day, interval sampling was carried out on every fifth canoe, based on a random start. Details noted included size of crew, number of nets, fishing grounds worked, state of fish landed (fresh or smoked; if smoked: when, where, and by whom), intended disposal (intermediary involved or proposed utilisation), destination of catch and transport involved, and number of fish by species.
A randomly selected two day sub-sample of the five day sample of each fortnightly interval was used to gather supplementary information for the enumerated canoes in respect of weight of catch by species and unit prices of each species when sold fresh at the landing site. At one of the sites, a further “tracer” investigation was made to determine actual (as distinct from proposed) particulars of disposal of catch, final market destination, and final unit price of catch by species.
Estimates of effort, production, species composition, and other parameters could be obtained by averaging various results for sampling days and expanding or raising them in accordance with the time interval monitored (e.g., total fishing days in a given month, or monthly rates for some measure within a given quarter).
Comparing the preliminary monthly results of his pilot survey investigations of fishing effort factors with the official figures for the same landings, Sastry found that there was a general pattern of under-estimation in the latter. However, it was not Sastry's primary concern to make strict comparisons between the pilot sample method and the existing official recording system in terms of measurement of effort factors, prices, end-product disposal, and so on. The pilot exercises were neither of sufficient duration nor representativeness to be taken as absolutely conclusive. The main objective was rather to demonstrate the usefulness of the sampling approach to upgrade the statistics on production and related aspects of the fisheries. As Sastry put it,
Monthly estimates of catch and fishing effort on a regional basis and for individual major centres can be obtained with a useful degree of accuracy. Estimates of catch for individually important major species can be determined on a quarterly basis with useful precision. The investigation has also shown the feasibility of obtaining reliable and useful information on aspects such as landed prices, disposal and breakdown of catches according to the main fishing grounds. It is therefore desirable to initiate arrangements aimed at expanding the present investigation over the whole lake, and eventually to replace the present arrangements for collection of statistics with an approach which is more systematic, objective and wider in scope [ibid: 36].
A number of major proposals were made in the course of Sastry's evaluation report and in his final recommendations section. These may be paraphrased as follows:
Following the lines developed in conjunction with the pilot sample surveys, a revised scheme for the collection of statistics on the country's fisheries should involve a combination of:
random sampling and sub-sampling of canoes at production points/landing sites and markets to determine catches by species, numbers and weights, gear in use, product disposal and prices;
the use of standardised collection forms and visiting schedules; and
the initial compilation of complete inventories of major and minor production points/landing sites and market centres so as to establish a basic frame for planning geographical coverage and sample stratification.
Any revised scheme of collecting statistics should be designed to operate through the existing fishguard field force, as part of their normal duties. This arrangement would be far more efficient, economical, and sustainable than the alternative of posting a special field force solely for the purpose of statistical collection.
At the same time, efforts should be made to strengthen the fishguard agency through additional postings in some areas and redistributions and replacements in others, and through adequate provision of training in collection procedures.
Of equal importance, support for primary enumeration must be provided to ensure the quality of data collected. Supervision could normally be carried out by the existing local fisheries officers and their assistants. Where necessary, extra fishery assistants could be appointed. Supervisory work should involve the training of fishguards and the programming and co-ordination of their work. It should also entail preliminary checking, compilation, and tabulation of returns from the field.
Overall responsibility for planning and co-ordination of data collection should be vested in an appropriate central agency, which would deal with such tasks as:
ongoing research in methodology to improve the system.
In light of cost and efficiency considerations, either the East African Statistical Department or the East African Fisheries Research Organization (EAFRO), both of the East African High Commission, should be designated to oversee planning and co-ordination of fisheries statistics. Such a task would prove very expensive for the Uganda Game and Fisheries Department to look after on its own, and wider regional concerns need to be taken into account as well.
Of the two East African Commission organizations, it would be most suitable to assign the task of statistical planning and co-ordination to EAFRO. The Statistical Department is not so directly involved with fisheries statistics for the member territories, and there are compelling reasons for statistical work to be carried out in close conjunction with the fisheries research work of EAFRO.
Finally, there is a need for co-ordinating machinery at a higher level, to serve the interests of all the concerned departments of the East African High Commission and the member territories. The composition of such an inter-territorial board might involve: the Secretary of the East African High Commission; the Director, EAFRO; the Chief Fishery Officer, LVFS; the Chief Fishery Officers of Member Governments; and the Director, East African Statistical Department.
The Sastry Report occasioned a good deal of discussion and debate on the most appropriate ways of recording commercial catches. During 1958, a comparison was made by Fisheries Officers in Uganda between the results obtained through the random sample method advocated by Sastry and those obtained through the established total enumeration method. This comparison was made at several landing stations on Lake Victoria and marked differences in outcome were confirmed. As it was felt that total enumeration was much less subject to error, serious doubt was cast upon the reliablity of the Sastry approach. In January 1959 a Committee meeting was convened at Jinja under the Chairmanship of the Director, E.A. Statistical Department in order to discuss this question. The meeting was attended by all interested parties, including Sastry, the Director of EAFRO, officers of the E.A. Statistical Department, LVFS Fisheries Officers, and representatives of the three East African Territorial Fisheries Departments. According to contemporary accounts,
There was very full discussion of the problems which had arisen, and also some straight speaking, especially by the [LVFS] Fisheries Officers. It was eventually decided that, in general, the [Lake Victoria Fisheries] Service should return to the old total count method of recording landings, except where the intensity of landings rendered some form of sample recording indispensible. Inevitably this reversion to the original method of recording has had its disadvantages -- from the point of view of yet another change -- but at the same time it has been possible to include in the information obtained about landings, certain additional information which was not collected previously when total enumeration was being employed. Arrangements are now in force whereby the month's records for each territory are summarized in a particular form, and a copy of the summary is sent to the Director, EAFRO, each month to enable his research staff to keep in immediate touch with trends [LVFS 1959:8].
The Committee in its review of the situation took into account the proposals contained in the Beverton Report (Beverton 1959), which emphasised the importance of maintaining the statistical record and called for a survey of the spatial distribution of fishing activities around Lake Victoria. It was recommended that as much survey work as possible be mounted during 1959 within the three territorial sections of the Lake in order to furnish supplementary information to that already available. The idea of establishing a regional “minimal programme”, using a standardised reporting form and of sending returns to EAFRO for processing and analysis, anticipated the disbandment of the LVFS in June 1960. From then on, the task of primary data collection for the commercial catch records for the Lake fell to the three Territorial Fishery Departments.
In the event, the “miminal programme” did not function very effectively. Catch records furnished by the three territorial Governments under the revised approach were questioned as to their accuracy and the methods used for their collection (Garrod, 1960). During the period 1962–67, returns to EAFFRO (the Organization was renamed the East African Freshwater Fisheries Research Organization in 1960) became irregular and even stopped in the case of the Kenya Fisheries Department. This made the proper analysis of fisheries statistics quite difficult (EAFFRO, 1964, 1966, 1967). The situation was made worse in that EAFFRO did not have additional staff to undertake specific responsibility for the processing of available data. Garrod had carried out extensive processing and analysis of the LVFS records during his tour with EAFFRO, but apparently there was no one available to replace him in this role after he left the Organization (EAFFRO, 1966; 1967).
In recognition of the need to strengthen biological and statistical research capabilities in freshwater fisheries for East Africa, efforts were started in late 1963 to draw up a project for submission to the United Nations. After some delay, the FAO/UNDP Lake Victoria Fisheries Research Project (LVFRP) was launched in 1967. A principal objective was to put into place a lakewide sampling system for the collection of resource and catch and effort data (UNSF, 1967; EAFFRO, 1967).
The need to develop a uniform system of recording catch and other fisheries statistics for the Lake Victoria Region was again stressed during the Sixth Annual East African Technical meeting in Mombasa, during August 1971. As a result, an International Fisheries Statistics Seminar was organised in Kisumu (Kenya) in November of the same year. This seminar was attended by some 40 delegates of the Partner States, EAFFRO, and the FAO/UNDP LVFRP (Okorie, 1971). During the Seminar, Mr. Bazigos, an FAO Statistician working with LVFRP, offered training to participants to introduce them to his newly designed programme of catch assessment surveys (CAS) for Lake Victoria.
The CAS programme was based on results of the aerial and coverage check surveys carried out by the Sociological Unit of LVFRP during mid-1971 (Bazigos, 1971; Butcher and Colaris, 1975). It involved a system of stratified sampling wherein a nested set of sampling units were defined. These included the following: country, limnological zone, area, fishing site, and fishing economic unit. Sampling was to be carried out on a quarterly basis. A detailed description of the survey design and procedures is given by Bazigos (1971; 1972); reviews of the CAS system are also provided by Wetherall (1972, 1974).
Shortly after the Kisumu seminar, survey teams were formed in each country and the first period of CAS data collection began during the beginning of 1972 (Walker, 1971; Wetherall, 1972). In the event, the second phase of the LVFRP, scheduled to start in 1972, did not materialise (EAFFRO, 1973). This was most unfortunate, since the opportunity was lost of effecting final implementation of a uniform, lakewide system of fisheries data collection.
In other developments, Stoneman had in 1969 undertaken a separate evaluation of the country's fisheries data collection system. He noted that “The collection of statistics in various regions of Uganda varies, and the methods of analysis vary as well” (1969:25). “However”, he went on to observe,
the basic parameters are the same, and it will be desirable to eventually standardise both the collection and presentation of the data over the whole country. What we need to know is the production of fish from the various bodies of water, broken down in various ways [ibid].
Stoneman regarded as essential points of enumeration:
the catch of different species;
the quantity of different gear used, and the proportion of the catch landed by each type of gear; and
the catch per unit of fishing gear, whether it be in net, trap, canoe, long-line, etc.
With reliable data covering such points for each fisheries region, it would then become possible to:
assess total annual production;
trace changes in the abundance of one species against another from year to year;
follow the weight of different species landed from year to year, and hence observe the decline or increase in catch of any given species;
determine, through the catch per fishing vessel, the differences in efficiency of types of vessel, and changes in the return to fishermen in overall catch from year to year; and
gauge, by following the varying catch in different gear, the fishing pressure on various species and long-term trends in species' abundance.
Stoneman stressed that in the absence of such information, “it would become impossible to plan control and management measures on a rational basis” (ibid). It was clear to him however that the Fisheries Department still had a long way to go in this regard:
All regions in Uganda produce annual figures of total production, production of the individual species, fishing effort and the catch per unit effort, and from these we compile overall national figures which purport to give general trends. In all regions, the statistics produced are at best intelligent guesses and the sources of error are numerous.
Nowhere in Uganda do we attempt to make a total count of all fish landed. In every instance, we sample a portion of the catch, and multiply this by some factor to give the final figure. It is important that the sampling be done on a preconceived and known basis. It must be truly random, and not be susceptible to individual preferences by the sampler. The units used to collect data must be the same throughout any given sampling programme, and preferably throughout the country [ibid: 25–26].
Sources of error identified by Stoneman as common included several which could be traced to the way Assistants carried out their primary work in the field. Inaccurate and unreliable information resulted from such practices as:
partial and preferential selection of landings to visit;
failure to turn up for work, and subsequent invention of figures in order to fill reports;
preferential selection of canoes for enumeration;
taking the word of fishing operators about gear units and/or catch quantities, without direct checking; and
failure to report gear assemblages in standard units, resulting in the equation of units which may in fact have widely varying effective fishing sizes.
Errors at the primary level were in turn thought to be seriously compounded by practices of supervisory staff in the regional offices, owing to:
failure to scrutinise returns for obvious mistakes.
Stoneman's suggestions for improvements on the mode of data collection were wide ranging, and quite reminiscent of those submitted by Sastry earlier on. They included the following major items:
an up-to-date list of all fish landings should be compiled for each region.
Assistants should be allocated responsibility to cover particular sets of landings and monthly programmes should be drawn up specifying visiting days for each, based on random selection; spot checks should be carried out from time to time to ensure that the programmes are being followed.
Each visit day for each landing should be programmed so that Assistants will know the number and order of canoes to be sampled; the order should be determined randomly.
On each sampling day, the number of canoes actually fishing from a particular landing should be carefully noted down in standard units.
Units of gear must be accurately described and standardised; return forms should also include provision to fill in the date, the name of the enumerator, the landing, type of fishing vessel and its order of return from the fishing grounds.
At the regional office, each month's returns should be checked to make sure that the work programme has been followed; the collected information should be assembled in tabular form (e.g. number of sampling days, number of canoes working the landing, and number of working days during the year) in order to facilitate calculations on production, catch composition, etc.
It is clear from experience in subsequent years that Stoneman's suggestions were not followed up very satisfactorily. Karuhanga (1974), in a later evaluation of the data collection and reporting mechanisms operating for Lake Kyoga and Lake Albert, cited a list of problems that virtually replicated those catalogued by Stoneman. Karuhanga did however give somewhat greater emphasis to the very real practical difficulties faced by UDF staff working in the field, such as their thin distribution on the ground and their lack of transport to cover widely dispersed enumeration sites, which made it exceedingly hard to carry out duties successfully.
On another front, the UFD in 1972 organised a countrywide frame survey exercise along lines similar to that of Bazigos for Lake Victoria. The objective was to promote a “uniform and accurate method of Catch Assessment” for all the waters of Uganda (Walker 1972; Dhatemwa & Walker 1972). An analysis of the results of the 1972 countrywide aerial survey was made by Walker (1972), who also conducted follow-up ground checks in order further to verify the findings.
The UFD had been involved with aerial fishing craft survey exercises before, in 1960, 1962, 1965, 1966, and 1970. These surveys varied in the scope of their coverage and reliablity. The 1970 survey by Wildlife Services Ltd. of Nairobi was the first to be conducted by “professional, trained observers,” and was thought to be the most accurate up to that time (Cadwalladr 1970; also see A. Graham 1970). Both the 1966 and the 1970 surveys covered Lakes Victoria, Kyoga, and Albert, as well as the Victoria and Albert stretches of the River Nile. Neither included the Lake Edward-George complex or the Kigezi minor lakes.
The 1972 exercise was unique in that it was based on highly systematic techniques and provided total geographical coverage. It proved extremely useful in providing a baseline of information on fishing factors which served the UFD for many years to come. No similar countrywide survey has been mounted from that time. However, localised surveys have since been carried out for Lake Albert and the Albert Nile (A. Graham, 1986) and the Lakes Edward-George/Kazinga Channel complex (Dunn, 1989; Enfield, 1989).
The next major event (though “non-event” is perhaps the better term) bearing on fisheries data system development in Uganda was the establishment of the Lake Victoria Fisheries Commission (LVFC) in 1973. LVFC terms of reference included the standardisation of fishery statistics (EAFFRO, 1973). Towards this end, LVFC created in 1974 a “Committee on Harmonisation and Standardisation of Methods of Collection, Processing and Compilation of Fisheries Data,” to be co-ordinated by EAFFRO statisticians (EAFFRO, 1974).
The LVFC failed to meet regularly in 1975. As a result, the recommendations of the Statistics Committee were not considered by the Commission before 1976, when they were finally forwarded to the partner States for consideration (EAFFRO 1975; 1976). By then it was becoming too late. Following the collapse of the East African Community in June 1977, EAFFRO was abolished. As a result, Uganda, Kenya and Tanzania founded their own national fisheries research organisations. These have continued to function to the present time and are known, respectively, as UFFRO (Uganda Freshwater Fisheries Research Organisation), KMFRI (Kenya Marine and Freshwater Research Institute) and TAFIRI (Tanzanian Fisheries Research Institute) (UFFRO, 1977).
The split up of the East African Community and its common fisheries research establishment put an effective damper on all efforts to promote a standardised system of fisheries data collection for Lake Victoria and the wider East African region. The situation with regard to Uganda was particularly acute, for a prolonged period of maladministration, civil strife, economic disruption, and general instability ensued from around the early 1970s, lasting through the mid-1980s. The ability of the UFD to carry on with routine data collection and administrative procedures was progressively compromised during this period (TDRI 1983; Reynolds and Greboval 1988).
Detailed observations on the state of fisheries data collection in Uganda and recommendations for its rehabilitation were contained in the reports of various missions and projects carried out in the country over the last decade. The Tropical Development and Research Institute (TDRI) carried out a full review of the situation in 1983 and submitted a lengthy set of findings under the title of the “Fisheries Rehabilitation Study -- Uganda.” This included strong recommendations for re-equipping the UFD and upgrading its staff establishment through training, so that it could be restored to a position of effective operation (TDRI 1983):
FAO organised a mission to assess the fisheries statistics picture for all Lake Victoria in 1986. In relation to Uganda, it was found that marked variation existed between the methods of data collection used by UFD staff at different points around the shoreline (Bernacsek 1987). It was noted that complete enumeration was still practiced in some cases, and a sampling approach in others. At certain sites a combination of the two was being employed. Aside from a lack of uniformity in collection methods, it was found that generally poor levels of field staff performance could be attributed to low morale due to meager salaries, and having to serve for extended periods at postings without going on refresher courses. It was further observed that the data quality was being adversely affected because of various practical problems such as inadequate equipment, supplies, transport, etc.
Very similar observations on the shortcomings of statistics collection were made in the report of the Agricultural Development Project team which implemented an extensive two-year survey of the Lake Kyoga fisheries between 1986 and 1988 (ADP 1988).
Indeed, it is quite striking that the various reports and commentaries on Uganda fisheries statistics and information issues that have appeared both in the recent and more distant past share so much in common in terms of their diagnoses of problems and weaknesses, and of their prescriptions for correcting the situation. Suggestions for improving the monitoring and reporting mechanisms have certainly not been lacking over the years. As a corollary, effective measures to implement suggestions on a sustained basis have been less in evidence. The disruptive events of recent years have been particularly damaging, since they not only precluded attempts to strengthen and upgrade fisheries information capabilities, but actually caused them to deteriorate and collapse. Under the calmer atmosphere now prevailing in the country, it is hoped that serious efforts can be mounted to correct matters.
It is important to point out in this review that although the thrust of past work pertaining to fisheries statistics in Uganda has been directed towards the harvest sector (fishing factors and production levels), some attention has been paid to the post-harvest sector as well. It will be recalled that early on, Graham (1929) mentioned the importance of recording the movement of fish and transactions at major markets. Sastry (1957) stressed the necessity of obtaining reliable data on quantities of fish by type of product (fresh, cured), distribution networks, price margins, and patterns of utilisation. He also pointed out the generally weak to practically nil provision within existing statistical recording methods for the collection of such data, and indeed for socio-economic information overall.
A study that represents something of a landmark for fisheries socio-economics in Uganda is the one conducted during the late 1950s by Crutchfield (1959), on marketing. As set out in the final report (ibid: 8), the principal objectives of the study were to:
carry out field surveys leading to fuller knowledge of the existing fish marketing structure of the country, including sources of supply, methods of transport, characteristics of the agents involved, seasonal variations, and prices and margins;
evaluate the export and import situation with regard to fish products, including the effectiveness of mechanisms for allocating products to domestic versus outside markets, and any bottlenecks to efficient operation; and
submit recommendations to Government in relation to the general development of the fishing industry and its markets, and specific measures which could be taken to alleviate bottlenecks in fish product distribution.
Although three decades have passed since Crutchfield's report on fish marketing was presented, it may still be regarded as the most comprehensive single piece of work on the subject (but also see Ford 1955; Garrod 1961; and Nyakaana 1984). Much of what he wrote did not deal directly with questions of statistical monitoring, and need not be reviewed at length here. Like Sastry, Crutchfield noted the essential weakness of recording and monitoring procedures with regard to distribution, marketing, and consumption of fish. However, he did not make any detailed recommendations as to how data collection procedures ought to be modified in order to correct this deficiency, since he thought that it was beyond the physical and staff capabilities of the (then) Game and Fisheries Department to maintain a continuous statistical series on fish marketing. Crutchfield did suggest, though, that spot surveys of the type used to collect information for his study might be used on occasion to keep abreast of developments in smaller markets, and that an overall marketing survey of the East African fishing industry be mounted (ibid: 146–7).
A number of additional studies touching on socio-economic themes have been mounted since the time of Crutchfield's pioneering work. For instance, investigations conducted under the auspices of the UFD, EAFFRO and the Lake Victoria Fisheries Research Project during the late 1960s and early 1970s covered such concerns as: general characteristics of fisherfolk communities around the lakeshore; the profitability of various types of craft on Lake Albert; the implications of proposed trawling activities for local fisherfolk and consumers around Lake Victoria; cost and earning performances of fishing units operating in Lake Victoria, Lake Kyoga, and Lake Albert; market surveys at various sites; and test market experiments with smoked Haplochromis in and around places like Kampala, Entebbe, Jinja, Tororo, and Mbale (EAFFRO 1968, 1969, 1970, 1974; Butcher 1972; Butcher and Colaris 1975; Jiwani and Dhatemwa 1972). Also during this period, Martin (1974) conducted a study of Lake Victoria fishing camps under the auspices of the Makerere Institute of Social Research.
The report on findings of the in-depth survey of Lake Kyoga carried out during 1986–1988 as part of the Agricultural Development Project included extensive description and analysis of the artisanal fishery and its socio-economic structure, and post-harvest and marketing aspects. Although the recommendations called for a thorough overhaul of present statistical collection procedures, nothing specific was suggested in relation to improvements for gathering socio-economic data (ADP 1988).
For the western lakes fisheries, significant socio-economic field studies in recent years include those of Karuhanga (1974) for Lake Albert, and of Enfield (1989) and Etoori (1986) for the fishing villages located in and around Queen Elizabeth National Park (QENP) in the Kichwamba Region. Aerial censuses of fishing craft and community settlements were included as part of the EEC-sponsored Conservation of Natural Resources Project based at Mweya, in the QENP (Dunn 1989; Enfield 1989).
The latest work with regard to socio-economics and the Uganda-sector fisheries of Lake Victoria is that of Reynolds and Greboval (1988). An overall appraisal is given of the consequences of the Nile perch succession for fisherfolk communities and the wider population of fish consumers, along with cost-earning analyses of fishing units, and remarks on the shortcomings of statistical records. In connection with the latter, proposals for upgrading existing systems and the scope of data coverage are presented.
In terms of larger surveys or evaluations bearing on socio-economic issues across different fisheries regions, there have been at least five undertakings within the last decade. In 1983, the extensive study undertaken by TDRI examined issues of fish processing, distribution, marketing, and demand across different regions of the country. It was pointed out however that the limited availability of data did not allow for a comprehensive assessment of the situation (TDRI 1983).
A fishery investigation team from the People's Republic of China attempted an assessment of the country's principal fisheries regions in 1988, concentrating especially on marketing and distribution of products. The team's report provides interesting reading, but unfortunately is of somewhat limited utility owing largely to the short duration of the study and translation problems (MAIF 1988).
The report on “Capital and Conditions of Fisher-Labourers of Lakes Kyoga and Victoria Canoe Fisheries” by Asowa-Okwe (1989), contrary to its title, does not involve a comprehensive analysis of the fisheries of those two water bodies. It deals instead with a comparison of conditions within two fishing communities, one on each lake. Written from a neo-Marxian analytical perspective, the Asowa-Okwe's report is a rather unique contribution to the literature on Uganda's fisheries.
Again in 1988, the Planning Department in conjunction with the Fisheries Department of MAIF conducted a national fisheries survey with technical and financial support from the EEC-funded Artisanal Fisheries Rehabilitation Project (MAIF 1989). Frame survey coverage included all major fish producing areas of the country. Although certain types of information on processing, distribution, and markets were gathered in all the areas, only the markets in Kampala District, with their distinct wholesale and retail dimensions, were given detailed treatment. Although its results must be regarded with some reserve, owing to practical problems of implementation, the exercise represents a good start towards building a full appreciation of the significance of the fishing industry as a provider of employment and source of food for the country.
The MAIF Planning Department's frame survey was used as the basis of a stratified sample “Joint Fisheries Survey” conducted by the Artisanal Fisheries Rehabilitation Project (AFRP) and UFFRO between January and March 1989. Its objectives were to investigate present catch and effort levels, assess the impact of the AFRP, and characterise the socio-economic position of the country's fisherfolk. The Joint Survey was organised in two parts, with UFFRO assuming responsibility for catch assessment questions and the AFRP for wider socio-economic issues within the fishing industry. Results of the Joint Survey are expected to be published in early 1990 (Frielink, pers.comm. 1989).
Strong recommendations were also made in the final MAIF Planning Department's report in relation to follow-up surveys, to be mounted at regular intervals, which would yield more detailed information on patterns of processing and trade.
The latter point deserves emphasis, and will be addressed again presently. While most of the socio-economic studies cited above have provided useful insights into the structure, dynamics, and general significance of the national fisheries sector, they nevertheless represent a collection of episodic, ad hoc efforts. An underlying continuity in the collection and evaluation of basic socio-economic data is lacking, and this is something that only a routine statistical measurement system can provide.
