Case Studies

Agriculture Census: A new prospect for inland fishery information

"The sustainability of many government statistical programs in agriculture and fisheries is threatened by resource shortages."

Adele CRISPOLDI
FIDI, Fisheries Department
Food and Agriculture Organization
Rome, ITALY

While the collection of agricultural statistics has a long tradition in most countries, the importance of inland fisheries data collection may have been overlooked by line ministries. Inland fishery activities are often not monitored due to their dispersion and the large subsistence component in many communities, which makes their social and economic contribution less evident and the task of data collection more difficult than for other activities.

An agriculture census is a large scale, periodic, national statistical operation for collecting quantitative information on the structure of a country's food production sector. Acknowledging the interrelation of agriculture and fishing in mixed-type farms in areas endowed with water bodies, some census related statistical activities could be further exploited to collect more information. This would improve listings on which inland fisheries sample surveys may be based and increase the availability of socio-economic data related to households engaging in fishing.

The sustainability of many governmental statistical programs in agriculture and fisheries is threatened by resource shortages. The way forward is to promote an integrated approach to sector data collections, to identify ways to expand the scope of long established agricultural data collection programmes that are regularly supported by government budgets and that would optimize, with some additional input, the use of limited resources.

This paper advocates the expansion of the array of information collected through agricultural censuses and associated statistical activities to include inland fisheries. It complements methodological approaches outlined in other case studies for improving the quality and reliability of inland fishery production estimates, and increasing the availability of the type of socio-economic data required by policy makers.

Basic objectives

The basic objectives of undertaking an agriculture census are:

• To provide aggregate totals for fundamental agricultural data to use as benchmarks for inter-census estimates

• To provide a frame for other agricultural sample surveys

• To provide data for small administrative units (local-level estimates are necessary for making policy decisions at local level) and detailed cross-classifications of farm structural attributes

Countries with developed statistical systems and sufficient resources may add one or more of the following objectives:

• To obtain benchmark data for improving current production estimates

• To obtain detailed data on characteristics of the agricultural population and on various inputs used for agricultural production, particularly those relevant to the environment (such as type and quantity of fertilizers, pesticides, source of irrigation water and credit)

Defining holdings and households

An agricultural holding is an economic unit of agricultural production under single management, without regard to title, legal form, size or location. Single management may be exercised by an individual or household, jointly by two or more individuals or households, by a clan or tribe, or by a juridical person such as a corporation, cooperative or government agency.

The household concept is one of the basic elements of a national statistics system. According to the United Nations, the concept of 'household' is based on the arrangements made by persons, individually or in groups, for providing themselves with food or other essentials for living. The persons in the group may pool their incomes and have a common budget; they may be related or unrelated persons or a combination of both. In rural areas, particularly in developing countries, a one-to-one correspondence between a household and a holding is quite common. Thus, households (complex socio-economic units) serve to identify holdings (simple economic units).

For national accounts purposes, further clarification is needed of the economic activities of the agricultural production units, particularly in cases where the holdings are also engaged in secondary or ancillary non-agricultural activities. Ancillary activities are considered part of agriculture in national accounts. A typical example of an ancillary activity in an agriculture holding is the harvesting of fisheries products from waters on the holding or accessible to the holding. In rural areas, such waters are most likely freshwaters and contribute to inland fishery production. Criteria that may be used in differentiating between secondary and ancillary activities of agriculture production units are the size of the activity and its purpose. As a general rule, non-agricultural activities which are small in scale or which are generally for the use of the holding (for subsistence) rather than for sale in the market, are considered ancillary. Inland fishing is often a seasonal activity and even fishing households with fishing as the largest source of income may engage in non-fishing activities to integrate the income.

Conducting a census

A census is usually conducted every ten years. It is best suited to collecting data on characteristics relating to agricultural holdings that change slowly over time. Since inland fisheries are known to change relatively slowly, a census would be suitable to collect structural data associated to holdings and households engaging in them. A census aims to understand the structure of the agricultural production sector (e.g. number and size distribution of agricultural holdings by type of enterprise, the purpose of production and the factors of production). Other structural items may relate to the educational level of the holder and farm labour inputs, the legal status of the holder and other social and demographic characteristics of holders and households. Structural data do not allow for any analysis of the performance of the sector. This requires data on quantities of inputs and outputs, enterprise costs and returns and farm income, as well as complementary data on variables such as food prices, consumption and nutrition. These change rapidly over time and are best monitored through more frequent sample surveys.

A census of agriculture is based on an extensive or total^[5] coverage of holdings and potentially provides a sound frame for sample surveys to estimate performance. The census also provides structural data for individual small areas (e.g. communities, administrative units, agro-ecological zones), which are needed in the preparation of plans and policies for rural development. The agricultural census is also useful in identifying disadvantaged groups such as subsistence farmers and female holders who need to be assessed separately in policy formulation to ensure that their living standards have improved.

Enumerating all^[6], (or a large sample of) agricultural holdings in a country without omission or duplication is a critical step that requires several sources. These include maps, topographic charts, aerial photographs or satellite images. Where these are not available, the agricultural census is undertaken on a complete list of villages or other identifiable geographic units or even on a sample enumeration basis. The lists generally include identification or classifier variables such as size of villages or units such as agricultural population or people engaged in agriculture, population of ethnic groups, total area and agricultural land area, main agricultural practices and facilities including water availability for irrigation. These data are useful for stratification purposes to improve the efficiency of the sample design and are of particular significance for identifying households engaging also in inland fisheries in relevant areas.

Census frame

The census is generally undertaken in one of three ways: using a list of agricultural holdings, a list of households or an area frame.

Listing all the agricultural holdings to be enumerated requires screening the entire population within an area using a short questionnaire requesting information such as area cultivated, number of animals and responsible persons. Lists of holdings or holders available at administrative offices are frequently incomplete and out of date and unsuitable for census enumeration. A population census taken shortly before the agricultural census is an important source to provide a first draft listing. Countries without maps and other independent sources generally include a few screening questions on the population census questionnaire. Countries lacking these sources may have to prepare a new listing of households and holders within households to identify the holdings for each selected enumeration area.

Whereas inland fishing activities can be easily accommodated in listings of households and area frames, the identification in a listing of holdings would require refinements to take into consideration that a significant land area is not an indispensable input in inland fishery production.

Many countries use the annual-households approach for establishing listings of economic units and keep them up-to-date before conducting an agricultural census. The success of any census or survey depends to a large extent on the quality of the frame used to identify the statistical units in the population. The ideal situation would be to have a complete list of all statistical units, with prior information for each of them on particular characteristics of interest, before starting.

Based on emerging requirements for alternate aggregations, the splitting of existing class 0500 (Fishing, fish farming and related service activities) was recently approved^[7]:

0501 -Fishing
0502 -Fish farming

This distinction between capture fisheries and aquaculture (fish farming) is relevant in view of their different structure, resource basis and technology. However, neatly separating the two information for each of them on particular characteristics of interest, before starting the census or survey. Registers of statistical units (e.g. agricultural holdings, agricultural service establishments, or households, administrative records of fishers groups and licence holders) are not generally available in most countries. Consequently, many surveys are based on multi-stage sampling schemes due to insufficient prior information on sampling units.

Economic activities in agriculture and fishing

The scope of an agricultural census includes fishing and related activities if carried out on the holding, although economic units engaged solely in fishing are not considered agricultural holdings and are therefore generally excluded from the census^[8].

In connection with the FAO WCA 2000 programme, special guidelines were developed to expand the scope of the census to collect structural information on aquaculture and assist countries in improving their current aquaculture surveys or provide a framework for those countries intended to develop an information base on aquaculture. A similar approach may be used for inland fisheries.

The essential features of the economic activities carried out by agriculture production units can be outlined with reference to the UN International Standard Industrial Classification of All Economic Activities (ISIC), which also provides a framework for the international comparison of national statistics. ISIC Rev. 3 separates fishing from agriculture and forestry:

A -Agriculture, hunting and forestry

01 - Agriculture, hunting and related service activities
02 - Forestry, logging and related service activities

B - Fishing

05 - Fishing, fish farming and service activities incidental to fishing

activities may pose particular problems in inland fisheries where activities may be integrated and interrelated (e.g. stocking, fertilization, predator removal etc.).

ISIC is activity-based and at present does not provide for a subdivision between marine and inland fishing. Whereas in landlocked countries all capture fisheries are from inland waters, in countries adjacent to marine waters, the expansion of one digit to provide for a sub-division between activities undertaken in marine and inland waters may be relevant if countries have important freshwater bodies, either entirely owned or shared, that sustain fishing activities.

The ILO International Standard Classification of Occupations (ISCO-88) was developed to serve as a model for countries revising their national classifications and to facilitate international comparisons of occupational statistics. It provides for reporting employment according to the following categories^[9]:

• Aquatic-life Cultivation

• Inland and Coastal Waters Fishing

• Marine Deep-sea Waters Fishing

National occupational classifications where fisheries are of significant economic importance should retain such categories and also consider the separation between inland and marine coastal fishing as appropriate^[10].

The FAO WCA 2000 recommends that the following information should be collected in identifying the economic activity of a holding and the main purpose of its production:

ECONOMIC ACTIVITIES

Whether holding is part of an enterprise engaged also in other economic activities

Other economic activities of enterprise

• Agricultural services

• Hunting, trapping and game propagation

• Forestry and logging

• Fishing

• Manufacturing

PURPOSE OF PRODUCTION

• Producing mainly for home consumption

• Producing mainly for sale

Fishing - as a producer of food commodities and as a provider of food and income to agricultural households - is relevant to food and agricultural decision-making of agricultural production units.

In eliciting the purpose of production, "mainly" means more than half of the production of the holding. Essential and desirable census items related to inland fisheries may be identified similarly to those for aquaculture in the category "other activities" as proposed below:

Other Activities: (identifies holdings carrying out forestry, fishery and other activities simultaneously with agricultural activities).

FISHERIES

Existence of Fisheries Activities On Holding

Whether fish or other aquatic animals and plants are taken from the waters within the holding or accessible to the holding

• Source of fishing

  Lake, rivers, canals
  Reservoirs
  Ponds
  Other (specify)

• Method of fishing (Boats/Gear)

• Kind of product

• Value of sales

Aquaculture Installation

Indication of type of aquaculture installation used for fisheries

• Pond

• Rice fields

• Other (specify)

• Kind of products

• Value of sales

Based on the environment definition, freshwater fisheries would include activities undertaken in water bodies of constantly low salinity^[11] and include, for example, reservoirs, rivers, canals, lakes, and paddy fields. With the inclusion of such items, after the census is taken, the tabulation programme would provide the number of holdings, which undertake inland fisheries activity and hold cultural installation, according to their type, kind of product (and as a desirable item, the annual value of sales). A programme of cross tabulation for holdings that carry out fishery activities may also include the area of the holdings, the purpose of production and may be cross tabulated with holders legal status, age and sex. Further reference may be with the use of freshwaters on the holding for irrigation.

Refining the household concept to accommodate inland fishery concerns

In most countries issues like taxes, subsidies, price control and programs related to poverty alleviation are being decided in isolation without studying their direct and indirect effects on different sectors of the economy. A more in-depth analysis of the economics of agricultural households, which are at the same time consumers and producers, is in order.

As fishing and agriculture are primary activities often seasonal in nature, households may undertake other economic activities to secure their income. It is important to understand the inter-relationship between farm activities and eventual non-agricultural activities and the data that need to be generated. Such data are essential for the successful implementation of plans and also to assess the impact of governmental policy decisions related to the levels of living of households dependent on agriculture and fishing. The required data can be collected through household surveys after appropriate consideration of concepts and definitions.

The need to consider all types of activities in an integrated frame has been recognised by the 1993 System of National Accounts (SNA), putting greater emphasis on the use of the accounting macro-framework for organising the database, rather than as a tool for compiling macro-economic aggregates. While studying economic aspects of an institutional unit or an establishment it is rarely feasible to find in real life a unit that is solely devoted to the farm activity as defined in the ISIC. Most likely, income of the unit will cover income derived from primary, secondary and ancillary farming activities as well as non-farming income. This is mainly because it is not often feasible to separate inputs, labour and assets according to the individual economic activity.

Where the focus is on agricultural and fishery households, it is necessary to go beyond the macro-framework given by the SNA and attempt sub-sectoring of the households. In this respect, ISIC Rev. 3 states that: "Ideally the principal product of the unit should be determined by reference to the value added of the goods sold or services rendered. In practice, it is generally not possible to obtain such information for individual products. It is therefore recommended that the principal kind of activity be determined by the gross output of the unit that is attributable to the goods or services associated with these kinds of activities. Where this method is not applicable, the principal kinds of activity should be determined from the proportion employed in these activities".

Interest should first be focussed on households whose income and resources are derived primarily from their own agricultural and fishing production. Thus, a fishery-dependent household should be defined as a household that derives its largest source of income from fisheries.

Using the approach of classifying households according to the prevailing product or prevailing source of income would permit a post-stratification of households where fishing activities are known to be prevalent or contribute significantly to the household livelihood. This economic concept should be cross-checked with spatial stratification based on the location of villages (e.g. proximity to lakes, large reservoirs, rivers etc.).

After clustering villages according to their location, a post-stratification of the production units is required, based on the relative number of people doing fishing activities and relaying on fishing for their livelihoods according to the relative proportion of fish sold to that of fish consumed. This listing would be the frame from which samples can be drawn to study in greater depth aspects of inland fishing activities through small cost-efficient surveys.

Main issues and considerations for future action

The information requirements of the inland fishery sector are often neglected by national statistics offices as well as by core statistics units of Ministries of Agriculture. Although logistical and operational problems may increase the cost of the systematic collection of inland fisheries data, it is recognized that biological, social and economic information is becoming increasingly critical for policy assessment and when environmental issues are locally emerging.

Since the sustainability of many government statistics programs is threatened by resource shortages, efforts should be made to better co-ordinate national statistics programmes to ensure the appropriate coverage of inland fisheries in agricultural censuses and rural household surveys between responsible units (typically the Department of Fisheries and the National Statistics Office). This will result in substantial improvements in the availability of data.

It is important to establish suitable frames for undertaking inland fisheries sample surveys. Where fishery censuses are conducted or where licensing and pond ownership is compulsory, the listing of economic units may originate from administrative registers (e.g. fishers co-operatives, license registers). However, even if listings can be obtained as by-products, their year-to-year maintenance may be costly. In countries where farmers extensively engage in fishing, population census and agriculture census frames are already integrating some fishery concerns, however not to the level of differentiating between marine and inland environments.

A population census may provide a frame of agricultural households for the agriculture census, and the latter provides an updated frame for other surveys whose statistical units are agricultural holdings. In mixed-type farms, agricultural censuses collect information on secondary and ancillary activities of the holding, including fishing. In countries where inland fishing is an important source of food the census may generate relevant socio-economic information on the activity of fishery households. There is a need for better exploiting census related activities that can be beneficial to inland fisheries and aquaculture surveys.

Annual household surveys and the analysis of household farm income permit post-stratification of villages where fishing is of economic relevance. Further work is required by agriculture and fishery statisticians to identify concepts and data requirements suitable to inland fisheries in ongoing surveys.

Due to the nature of inland fisheries there is no single method of data collection to be efficiently used, but a combination of methods according to specific needs. Inland fishery data collection methodology through objective area-measuring and yield-estimation would benefit from the agriculture experience.

Improvements to data quality and reliability may come from the systematic use of sample techniques in collecting commercial capture inland fisheries data and the use of occasional surveys for estimating the semi-commercial and subsistence components.

The validation of fishery production statistics may, from time to time, benefit from the conduct of consumption surveys. These surveys do not permit validation of data such as seasonality and method of production unknown to consumers. Market survey data are bound to produce reliable trend indications of production but under-coverage of total production due to self-consumption. Household socio-economic surveys may provide valuable information from the demand side.

In planning for the implementation of an improved information system it must be recognized that every country has a statistical system in place to begin with and which is generally considered adequate to meet perceived data needs (national and international) and commensurate to the resources available to implement it. Therefore, the implementation of a new system or a new component should not be a question of scrapping what already exists but rather a transition from one system to another.

It must be equally recognized that a fishery information system is but a subsystem of a national information system. Statistical problems exist in different countries with differing degrees of severity and with emphasis on different causes and effects. The roots of the problem lie with inadequate national efforts of an interdisciplinary, inter-ministerial and continuing nature in the development and operation of an information system to support effective government interventions in the sector.

Future challenges include the maintenance of an information system to sustain decisions for improvement of rural livelihoods, farm incomes and food security (especially in food-deficit and marginal areas) to address increasing relative poverty of the resource-poor areas which are more favourably endowed with fish resources and an improved information basis for conservation and management decisions concerning inland fisheries and their interactions with the ecosystem.

References

FAO 1986, Food and Agriculture Statistics in the context of a national information system, FAO Statistical Development Series, Vol. 1.

UN 1990, International Standard Industrial Classification of All Economic Activities (ISIC), Series M No.4 Rev. 3 ILO 1990, International Standard Classification of Occupations: ISCO-88.

FAO 1992, World Census of Agriculture 1980, Methodological Review.

UN et al. 1993, System of National Accounts, Series F No. 2 Review 4.

FAO 1995, Programme for the World Census of Agriculture 2000 (WCA 2000), FAO Statistical Development Series Vol. 5.

FAO 1996, Multiple Frame Agricultural Surveys, Vol. 1, Current surveys based on area and list sampling methods, FAO Statistical Development Series Vol. 7.

FAO 1996, A System of Economic Accounts for Food and Agriculture, FAO Statistical Development Series Vol. 8.

FAO 1997, Guidelines on the collection of structural aquaculture statistics, FAO Statistical Development Series Vol. 5b.

FAO 1999, Guidelines for the routine collection of capture fishery data, (prepared at the FAO/DANIDA Expert Consultation, Bangkok 18 - 30 May 1998), FAO Fisheries Technical Paper 382, Rome 1999.

Narain, P., 1999, How to integrate economic aspects of agriculture with other economic activities of households: farm plus non-farm income, Development and improvement of economic statistics for food and agriculture, Joint ECE-OECD-Eurostat-FAO Meeting on Food and Agricultural Statistics in Europe (Geneva, 23 - 25 June 1999).

FAO/FIDI, 2000, Inland fishery and freshwater aquaculture production statistics in Asia/Pacific- Some suggestions for their improvement, Asia Pacific Commission for Agricultural Statistics (APCAS) 18th Session, Bali, Indonesia, 6 - 10 November 2000.

C. Stamatopoulos, 2002, Sample based fishery surveys, A technical handbook, FAO Fisheries Technical Paper 425, Rome 2002.

Improving Fisheries Catch Statistics in Lake Victoria

"The current top-down approaches to manageing fisheries resources in the three countries have met with great difficulties."

I.G. COWX
University of Hull
International Fisheries Institute
Hull U.K.

Lake Victoria is one of the African Great Lakes and the second largest lake in the world covering 68 000 km2. The lake is shared by Kenya (6% by area), Uganda (43%) and Tanzania (51%) (Fig.1). It has a mean depth of 40 m, maximum depth of 84 m, shoreline of 3 450 km, a water retention time of 140 years and a catchment area of 193 000 km2 and extends into Rwanda and Burundi.

Figure 1. Lake Victoria showing international boundaries

Over 30 million people live in the Lake Victoria Basin and depend directly or indirectly on the lake's resources. Fisheries contribute up to 3% to the GDP of the riparian states and they are major sources of income, food, employment and foreign exchange earnings. Fish from Lake Victoria is the most important source of affordable protein in East Africa and the most important source of freshwater fish on the African continent. The fishery is diverse and highly dispersed and fragmented with about 1 500 landing sites and more than 120 000 fishers. The lake is also important in conservation terms because of the great biodiversity of endemic fish species. Additionally, the lake is an important moderator of regional climate.

The lake fisheries are diverse, dispersed and catch information is inadequate for supporting management. The Lake Victoria Fisheries Research Project (LVFRP) was established in 1997 to assess the status of the fisheries and the strategies employed provide a good case history for comparison with the situation in the Mekong River. This paper describes the status of the fishery and data recording systems prior to the LVFRP programme and the strategies adopted to improve the monitoring of the fishery.

Production trends

Until the 1970s, Lake Victoria supported a multi-species fishery dominated by tilapiine and haplochromine cichlids. There were important subsidiary fisheries for more than 20 genera of non-cichlid fishes, including catfish (Bagrus docmak (Forskåll), Clarias gariepinus (Burchell), Synodontis spp. and Schilbe intermedius (Rüppell), the lungfish (Protopterus aethiopicus (Heckel) and Labeo victorianus Boulenger (Kudhongania and Cordone, 1974).

Signs of overfishing were reported as early as the 1970s when catch rates for the native tilapiine fish of Lake Victoria, Oreochromis esculentus and Oreochromis variabilis were reduced by selective fishing and the failure to control fishing effort (Jackson, 1971). These fishes were originally the backbone of the commercial fishery,

Stocks of most of these species further declined and others disappeared following the introduction of four tilapiines during the 1950s (Oreochromis niloticus (L.), O. leucostictus (Trewavas), Tilapia rendalli Boulenger and T. zillii (Gervais)) and Nile perch (Lates niloticus (L.), the contribution of haplochromines (cichlids) to fish biomass decreased rapidly from 83% during the 1970s to less than 1% by the mid-1980s (Fig. 2). This was due in part to predation by Nile perch. Oreochromis niloticus, on the other hand, hybridised and competed for food and space with O. variabilis and O. esculentus, leading to the decline of endemic tilapiines. It is believed that more than 60% of Lake Victoria's endemic fish species became extinct between 1970 and 1986, with the remaining species reduced to insignificant levels (Fig. 2). The establishment of the Dutch Government sponsored Fish Meal Plant in Mwanza in the 1970s also contributed substantially to the decline of the haplochromines in the lake since the factory targeted this fish group.

The Lake Victoria fishery has changed from the complex multi-species fishery of the late 1970s to one dominated by three species, namely the introduced L. niloticus and O. niloticus and the native cyprinid species, Rastrineobola argentea (Pellegrin) (Fig. 2). In Kenya, total fish landings increased from about 19 000 tonnes in 1977 to approximately 220 000 tonnes in 1992 due to increases in the contribution of Nile perch. Catches have now fallen to around 160 000 tonnes as a result of a fall in catches of Nile perch. In Uganda, total fishery yield increased from 11 000 tonnes in 1977 to 120 000 tonnes in the early 1990s. This was again due to an increase in the contribution of Nile perch. The data for the 1990s are fragmented and no discernible trends are possible, except landings in the year 2000 were in the order of 141 000 tonnes. In Tanzania, the quantity of fish landed increased from 72 000 tonnes in 1983 to 231 000 tonnes in 1990, again due to landings of Nile perch increasing from 274 tonnes in 1981 to 175 000 tonnes in 1990. Poor quality catch assessment data have prevented any evaluation of trends in yield in recent years.

Figure 2. Trends in landings (tonnes) of major fish species and species groups in countries of Lake Victoria

Lates niloticus

Rastrineobola argentea

Tilapias

Haplochromine cichlids

Other species

It was the Nile perch fishery that created a remarkable stream of economic benefits. These benefits included an expansion of the artisanal fishing industry and availability of Nile perch to consumers in the region to the development of a multi-million dollar export industry for chilled and frozen fillets. In 1994, revenue from fish landings in Uganda was US$ 77.13 million, whilst in Kenya in 1998 it was US$ 80 million. For the same period in Tanzania, revenue was US $200 million (SEDAWOG, 1999).

However, since the mid-1990s, the dominant Nile perch has shown signs of decline. Changes in the efficiency of fishing gear, motorisation of canoes and an increase in total fishing effort to maintain production were observed. The intensity of the fishing pressure is evident from the results of a frame survey carried out in the year 2000 (Table 1). Extension of fishing grounds was also evident, but all against a continued decrease in catch per unit effort and mean size of fish caught (Mkumbo and Cowx, 1999).

Table 1. Summary of Frame Survey 2000 showing distribution of landing sites, crafts and gear (values in brackets are density by country per km2)

This decline in the Nile perch fishery has been mirrored by an expansion in the less profitable Rastrinbeobola fishery. Recent studies have revealed that some species feared extinct (e.g. zooplankton-feeding haplochromines) are reappearing in the lake and posing a threat to R. argentea whose food requirements are similar.

Attempts to manage Lake Victoria's fisheries date from 1927 when Graham (1929) conducted the first fishery survey. At that time it was noted that the gill net fishery was negatively affecting the stocks. Thus a minimum mesh size of 5 inches was set in 1933. In 1947, management and research of the lake's fisheries were placed under the Lake Victoria Fisheries Service (LVFS). LVFS was dissolved in the early 1960s. With the collapse of the East African Community (EAC) in 1977, the Food and Agriculture Organization of the United Nations (through the CIFA sub-committee for Lake Victoria) continued to co-ordinate the activities of the riparian states on Lake Victoria's fisheries. FAO also assisted the three riparian states to establish the Lake Victoria Fisheries Organization (LVFO) in 1995.

Strengths and weaknesses of data collection system

At first glance, the trend analysis provided in the previous section suggests that the fishery is well monitored and adequate information is available on catch statistics. Closer examination of the data reveals numerous weaknesses with the output, most notably the lack of realistic statistical data for Uganda and Tanzania, the two countries that represent 94% of the lake surface area. The situation in Kenya is slightly different because all landing beaches are monitored by KMFRI and total estimates of catch are available, although the quality of the data is variable (see later). The weaknesses in data collection and their root causes are as follows.

Catch assessment surveys

It is impossible to make a total count of the fish catches in a highly diverse and dispersed fisheries such as found in Lake Victoria, and indeed the Mekong. The very nature of a fishery with many landing sites and use of semi-commercial and subsistence fishing prevents such an enumeration. The traditional way of assessing a fishery in these circumstances is to promote either a representative or random sampling strategy of the landing sites (catch assessment surveys) to obtain estimates of catch per unit effort and then raise the sampled catches by a value of overall fishing effort (frame survey data). In Lake Victoria, until recently, this has failed for a number of reasons and no reliable catch statistics have been available in Uganda or Tanzania since the mid 1990s.

Catch recording

With the exception of Kenya, site recording has been inadequate or non-existent. In Uganda, decentralization of the local fisheries staff to the districts in 1995 resulted in loss of control of their duties by the central Fisheries Department. As a consequence, no beach recordings have taken place since that time. The only records are for fish that pass through the larger beaches and are usually destined for the processing factories. These records remain at the district offices and are not collated nationally. In Tanzania the situation is somewhat different but with the same outcome. Here a beach recording is operational and records are centrally collected by the Fisheries Department. However, the quality of the records is dubious, the overall output is unreliable and no national report is produced. This is somewhat surprising because the Tanzanian Government imposes a 6% levy on catches so they are losing revenue through poor recording. However, the imposition of the levy means fishers tend to avoid traditional landing beaches. In Kenya, a comprehensive beach recording system is carried out by the Kenya Marine Fisheries Research Institute (KMFRI) and duplicated by the Fisheries Department, although not on the same scale. The results from the two sources are conflicting (Fig. 3), instilling little confidence in the results. In all three countries the main problems with the beach recording systems are lack of financial resources and poor motivation of staff. The staff have no incentive to record information accurately because they are poorly paid, if they are paid at all.

Figure 3. Comparison of Kenyan catch statistics from the Fisheries Department () and KMFRI ()

Nile perch

Nile tilapia

R. Argentea

Total catch

Estimation of fishing effort

One of the key elements for assessing a complex fishery such as found on Lake Victoria is an estimation of the effort expended. This is usually done through a frame survey, which must be updated at regular intervals to ensure changes in fishing effort are known (Sparre and Venema, 1998). For a dynamic and rapidly changing fishery such as in Lake Victoria, a biennial frame survey is deemed necessary. Prior to the start of the LVFRP the last frame survey in Uganda was in 1990 (Tumwebaze and Coenen, 1991) and in Tanzania partial surveys were carried out in 1990 and again in 1995 (Mkumbo and Cowx, 1995).

These surveys were poorly conducted and the results were never fully analysed because of the lack of resources or suitably qualified personnel. In Kenya, the need for a frame survey is less prominent because of the total coverage of the beaches. The frame surveys carried out are also considered weak because they were not harmonised between countries and are conducted at completely different times. This is an important issue because the fisheries in each country do not abide by national boundaries. Kenyan fishers in particular fish extensively in both Ugandan and Tanzanian waters. This is patently obvious from stock assessment surveys carried out under the LVFRP where the annual catches for Kenya exceed the total fish standing stock by some 50% (Cowx et al. 2002). The main reasons for regular frame surveys not being undertaken or reported were lack of financial and human resources, inadequately trained staff and poor motivation. The upshot is that estimates of catches in the late 1990s were based on outdated frame survey data (Fig. 4) and do not account for the major shifts in the fishery that have evolved because of overcapacity within the fishery. This would include changes in types of gear used (types as well as mesh sizes of gill nets) and their modes of operation (e.g. active versus passive use of gill nets; see for example).

Figure 4. Changes in the numbers and mesh sizes of gill nets in Ugandan waters between 1990 and 2000 to illustrate the increase in fishing effort with time and change in gear usage.

Export market chain

Export figures are one source of information that can provide potentially reliable capture statistics. In the past this has created problems because the processing factories are reluctant to reveal their revenues or indeed give an accurate picture of their export volume. Since the majority of the export is freighted out of the region by air as chilled fillets, volume can be determined from shipping company records (Fig. 5). The problem arises from conversion factors from fillets to whole fish and the records only refer to the most valuable species, Nile perch. However, a percentage take of 41% of the flesh was determined from work in the factories so the volume exported can be converted to wet weight of fish. Notwithstanding these problems, export figures are potentially a valuable source of accurate data.

Figure 5. Total value of Uganda annual fish exports from 1991 to 2001. The grey boxes indicate the major reason for the EU ban of fish imports from Uganda in 1997 to 2000.

Illegal fishing

In recent years the catches of Nile perch in Lake Victoria have declined. This is coupled with a declining catch per unit effort (Cowx et al. 2002). One of the traditional responses in a poorly regulated fishery is for the fishers to move towards smaller mesh nets and use illegal gear which exploits smaller fish. Such a response is in progress on Lake Victoria. There is now a high proportion of illegal gear types being used on the lake, both in form of illegal sized meshed nets and banned gear types. Some 17% of gill nets are below the legal mesh size of 5 inches and some 30 000 seine nets, a prohibited type of gear, are being used around the lake. Until recently, some 15 trawls were also being illegally operated in Kenyan waters, but these have been outlawed. Part of the problem stems from the processing factories attempting to meet demands from export markets. These markets are highly lucrative but are now demanding fillets from smaller fish because they are less fatty. As a consequence, the factories are supplying smaller mesh sized nets to fishers tied into financial and supply agreements. Much of the very small fish caught in these nets is not recorded and is siphoned off to other markets (see below).

Unreported or unrecorded catch

International demand for Nile perch is the engine driving the fishery (Fig. 1). The huge quantities being exported from the region have undoubtedly pushed up the beach price for fresh fish and made the product too expensive for the local populace. Waste and small sized fish rejected by the processing factories find their way into the local markets, but all too often this supply chain is unrecorded. In addition, a large proportion of the fish not accepted by the processors is being exported to the DR Congo, Rwanda and other countries. These fish are usually small and caught with illegal gear so bypass any recording system. It has been estimated that this component could represent up to 20% of the total catch. These fish are transported by road and do not pass major urban centres where some control could be made.

Dried, smoked and fried fish products provide the basis of an extensive trade in low cost fish protein that find their way into low income households. There is also an extensive subsistence fishery around the lake especially for households living close to the shore. These fish are not counted in traditional recording systems. Finally, one component of the catch that is not considered is for bait for the extensive longline fishery for Nile perch. There are an estimated 3.5 million longline hooks (Table 1) continuously being used, which need baiting on a regular basis. The modes of operation and bait species used vary between countries (haplochromines in Tanzania and Uganda, and Clarias in Kenya) but it was estimated by Cowx et al. (2002) that some 5 700 tonnes of fish are caught by hook and line or in seine nets to support this extensive fishery. This catch is not recorded.

Catch recording and administration

Until recently it has been the responsibility of national governments to coordinate data collation and reporting. As reported above, this has proved ineffective and the recently established LVFO (2000) has been charged with coordinating the data collection on a regional scale. However, without the raw material it is difficult to see how this could be achieved. The problem is made worse because there is no national or regional database in which to store and process the data. To date, all data are held in paper form and only secondary processing is carried out on computers. This inevitably leads to transcription and calculating errors. Again this is evident from the Kenyan problem where the Fisheries Department and KMFRI report different catches.

Role of the fishing communities

One underlying factor that previously has been ignored with respect to the management of the fisheries of Lake Victoria is the role of the fishing communities. Management has been command driven from central government departments or agencies and this has in part led to many of the problems within the fisheries, including the poor quality statistical data collection. The lack of involvement of the fishing communities can only be seen as a retrograde step because they are the source of the information. The central control driven management has led to general distrust and non-cooperation with the fisheries departments, and therefore the communities provide no support for the statistical collection procedures. Overfishing and the use of damaging or illegal fishing gear is only in part a reflection of the failure of centralised management strategies on the lake and the lack of feedback from 'research results'. The communities themselves recognise that the fishery is overexploited but unless they are informed of the status of the stocks in relation to catch statistics they cannot be expected to respond to vague calls for them to reduce the amount of fish harvested. There is thus a clear need to address problems in data and research dissemination, and consideration must be given to exploring ways of relaying such information to stakeholder groups. The trends described above represent a grave threat to the sustainability of Lake Victoria's fisheries. It will only be with the support of the fishing communities that sustainability of the fishery is an achievable objective.

The way forward

As can be seen from the above description, fisheries statistics for Lake Victoria are both inadequate and unreliable. It is imperative that these statistics are improved because the resources represent a valuable commodity to the riparian states both in terms of export earnings from the Nile perch trade and as a source of income, employment and protein for the local people. If the current decline in the fishery is allowed to continue it could lead to considerable social hardship for the people dependent on the fishery for their livelihoods and source of protein. The need to manage the fishery on a sustainable basis is therefore paramount but this cannot be achieved if information on the exploitation patterns is not forthcoming. To resolve this issue the LVFRP has put into place, coordinated, or collaborated in a number of actions. These include improved fisheries data collection systems, a regional fisheries and environmental database and co-management initiatives to manage the fishery.

Fisheries data collection systems

The institutions in the region charged with management and research on fisheries are all under-funded. In most cases the funds received from the central government pay only the salaries of employees. This leaves no funds to undertake research and management activities including monitoring, control and surveillance (MCS). There are no funds to purchase equipment or to employ additional research and management personnel. Consequently, any fisheries monitoring programme has to function on limited resources and be cost effective. Under the LVFRP, simple cost effective data collection systems were designed that provide the minimum of data to support management initiatives and meet the statistical reporting requirements of the countries. The programme is multifaceted to allow cross validation of the outputs.

Fisheries dependent survey programme

With the exception of Kenya, the existing catch assessment surveys are woefully inadequate. Consequently, one of the primary objectives was to promote efficient catch assessment data surveys linked to regular frame surveys. This was achieved on two fronts.

A regional task force was established to set up a frame survey of the lake to assess patterns in fishing effort. This involved setting up an appropriate questionnaire that was sufficiently comprehensive to provide the information required but not too complex to make it unmanageable to complete in a short time or extract the data.

Several workshops were conducted to train regional representatives and then enumerators, who were often senior members of the local fishing communities. The frame survey was conducted lake-wide over two days in March 2000 and repeated in June 2002. A summary of the output of the 2000 frame survey is given in Table 1 and was considered to be the most comprehensive survey of the fishery carried out to date. The lake-wide coverage over a very short time reduces possible double counting of boats and gear types, especially because there is considerable cross-border fishing. The biggest problem was accounting for the fishers and gear that stays permanently on the lake and is not returned to the landing beaches. This was partially overcome by involving the local communities who were able make best estimates. Notwithstanding the success of the frame surveys they were carried out at considerable cost. The initial set up costs were high and funded from regional development projects, although the cost of training staff and implementation of the 2002 survey was much lower.

The frame survey was carried out in conjunction with a dedicated catch assessment survey. This was set up from scratch in Uganda and Tanzania because the existing surveys were defunct, whilst the Kenyan survey was formalised to improve the reporting procedures. In Tanzania and Uganda, a stratified random survey was difficult to implement because smaller landing beaches change over time. The surveys were therefore based on a set number of fixed beaches in each country, which could be surveyed appropriately on a regular basis. Each country was divided in three zones (Fig. 1) and a number of landing beaches were surveyed in each zone each month. In total, 18 beaches were surveyed on a three-monthly basis in Tanzania and 25 in Uganda (Fig. 1). Thus six beaches were surveyed per month in Tanzania and eight in Uganda. Although the number of beaches was small, this was the minimum that would provide coverage of the fishing patterns. The beaches were selected to represent:

• Fishing in the major ecological zones, from shallow inshore waters to deep offshore waters

• Landing sites specialised in either of the major commercial fisheries, i.e. Nile perch, Nile tilapia or R. argentea

• Landing sites with a wide variety of fishing gear and methods including beach seining, long lining, hand lining, cast netting, mosquito seines for R. argentea, and gill netting by both large motorised boats and small paddled boats

Neyman allocation was not used to select the beaches as this proved unrealistic. During each survey the catch per species was recorded in relation to the boat type and gear and number of boats operating. At each beach, approximately 100 fish of each of the major commercial species were measured to assess the population dynamics. The surveys took about 10 days each month. The output for Nile perch in Tanzanian waters is given in Table 2. Similar data were available from the Tanzanian Fisheries Department, the official agency responsible for data collection.

Table 2. Nile perch catch statistics from Tanzanian waters in the year 2000

In addition to the regular CAS, it is proposed that observers are placed in the 27 processing factories situated around the lake. These persons can record both the volumes of fish entering and leaving the factory. These data will provide valuable insight into production trends and help validate the outputs from the catch assessment studies. They will also be able to collect basic biological information on the fish populations (e.g. length distributions, reproduction characteristics), which can be used to support management decision-making. They will also monitor whether the factories are complying with regulations on harvestable sized fish. Recently a regulation was passed whereby only fish of a slot size between 50 and 85 cm could be processed in an effort to reduce fishing pressure on juvenile fish and large mature adults. It is recommended persons enforcing the regulation should be changed regularly so they cannot be corrupted in their duties.

The efficiency of the catch assessment surveys was tested by comparing the outputs from the surveys against estimates derived from virtual population analysis and processing factory outputs adjusted for fish passing through other marketing channels in Uganda (Table 3). The similarity between the CAS and VPA outputs suggests that the former is a viable, cost effective approach, but it must be recognised that the work was carried out by a dedicated, highly motivated researcher. It is likely that less motivated, poorly-paid enumerators will not carry out the surveys with the same level of dedication and thereby compromise the accuracy of the results.

Table 3. Comparison of the estimates of total annual catches of Nile perch and Nile tilapia from catch assessment data and length structured VPA in the Ugandan partof Lake Victoria in 2000

This is what happened in Tanzania, where the researcher lacked motivation and the quality of the output was weaker, despite intense supervision. Involvement of local fishing communities could help resolve this problem. The poor conformity of the processing factory data were because the EU imposed an export ban on Nile perch into Europe for most of 2000 because of problems with fish poisoning and the factories were operating at very low throughput (Fig. 5).

Fisheries independent survey programme

The research carried out has been mostly concerned with ecology and biology of fish species including limited stock assessment and limnology, which provides information only on trends in stock size and composition. There has been very little attention to socio-economic criteria or methodologies in developing strategies proposed to tackle the issues of declining stock size and adverse species compositional changes. This shortcoming has contributed to managers being ill equipped and exacerbating problems associated with the failure to regulate and manage the lake fisheries. To overcome these problems, a set of research projects dealing with assessment of stock abundance and fish population characteristics as well as the socio-economic dimensions of the fishery were undertaken under the auspices of the LVFRP and in conjunction with the fishery dependent surveys. The biological surveys included regular trawl surveys in the riparian countries to estimate standing stock biomass and population characteristics such as population size structure, growth rates, mortality rates, size at maturity, plus six-monthly lake-wide hydroacoustic surveys to assess stock biomass and distribution. These surveys provided valuable support information on which to base policy decisions on fishery regulations. The socio-economic studies concentrated on marketing, poverty, nutritional status of the lakeside communities and the feasibility of introducing co-management initiatives for the lake fisheries. The latter studies were fundamental to establishing future management initiatives for the lake and the role the fishing communities could make to support assessment of the status of the fisheries.

Data dissemination and database management

Recent research programmes on Lake Victoria (LVFRP and LVEMP) have considerably improved the knowledge of fish stocks. It is essential that this information flow is maintained and continually upgraded if the resources are to be managed on a sustainable basis. Financial and human resources must therefore be made available to continually monitor the status of the stocks and to allow management processes to respond to changes in a timely and appropriate manner. Consequently, fish stock assessment, including analysis and timely reporting, is now a programmed activity by the research institutions in collaboration with the Fisheries Department and answerable to the Lake Victoria Fisheries Organisation (LVFO). The LVFO is charged with producing reports to collate all available information on the status of the stocks, exploitation patterns and socio-economic indicators to aid formulation of policy. Reports are made accessible to all stakeholders and written in a language that both the layman and professional stakeholder can understand.

One of the key problems was the lack of an appropriate database management system. Consequently, a Database Management System for the Lake Victoria fisheries (SAMAKI) was developed under the auspices of the LVFRP. The system contains the following items:

• Publications database

• Frame survey database

• Fish-processing database

• Socio-economic database

• Catch assessment database

• Additional tables facilitating data entry and security

The core software is implemented in Access 2000. The Access platform was adopted because the computer facilities available in the region would not support a more complex system such as UNIX and it was recognized that continuous donor funds would be required to update a more complex system. The development of the system follows the common Windows approach for design of database systems. Attention has been paid to user friendliness of the system. All features and capabilities of the system are put under one user interface and there are no hidden or misleading facilities. The decision support system is designed in such a way that it gives full access to the entire dataset to the lowest level. Data mining is part of the same software that supports data entry and the user does not need to explore the data using separate software. There are several export facilities available to facilitate data transfer from the database to other popular applications like Excel and Word. One characteristic of the system is the spatial component of all the data entered into the system. The design enables the exploitation of the data using Geographic Information Systems. The system is designed to support the national level of the Lake Victoria Database Management System and is currently being extended to a Regional Database Management System.

Partnerships: Co-management

The current top-down approaches to managing fisheries resources in the three countries have met with great difficulties. These have included understaffing and poor motivation among others. Relationships between the lakeside communities and the fisheries departments also need to be improved. In an effort to address the problem, the riparian governments are looking to empower local communities to actively enter the management process, especially in the areas of the monitoring (data collection), surveillance and control of all activities associated with the fisheries economy. There is also interest both at the centre of government and the lake communities to take on the challenge of security and the fisheries management process. The proposed institutional framework to address this scenario is given in Fig. 6. The government of Tanzania has set up Beach Management Units (BMUs) empowered to take on management functions at a local level. Similarly in Uganda there is interest in devolving powers to Landing Management Committees (LMCs). These interventions are at an early stage of development. So far in Kenya there has not been any measurable progress in either decentralising or devolving power to the lakeside fishing communities. It is important to note the government still remains central within any system of co-operative fisheries management since it is an effective source of legitimacy in rule making and enforcing. The co-management approach is expected to lead to lower transaction costs at the planning and implementation phase because fishers can provide information on fishing patterns, catches and the status of the resources (Sen & Nielsen 1996). The success of co-management will depend on political commitment on the part of the governments to fisheries management. This commitment would require support by appropriate legislation and adequate technical and financial resources. Under co-management, new institutions would have to be developed and this is a long-term process.

Figure 6. Proposed institutional framework for management of the Lake Victoria fisheries

Conclusions

Lake Victoria is a valuable case study for assisting and improving inland capture fisheries statistics in the Lower Mekong Basin because the fishery characteristics are similar. Fisheries are diverse and dispersed and both regions face similar problems collecting data. Comparisons with the situation on Lake Victoria could provide valuable lessons for resolving some of the problems faced in the LMB and elsewhere. Ultimately, accurate fisheries catch statistics are fundamental to the sustainable management of any fishery.

References

Cowx I.G., Muhoozi L., Mkumbo O., Getabu A. & Okaronon J. (2002) Summary evidence for the overexploitation of fisheries resources in Lake Victoria with special reference to Ugandan waters, Strategic exports for Uganda: Fisheries. EU Delegation, Kampala.

Kudhongania A. W. & Cordone, A.J. (1974) Batho-spatial distribution pattern and biomass estimate of the major demersal fishes in Lake Victoria. Afr. J. Trop. Hydrobiol. Fish. 3, 15-31.

Fryer G. (1993) The Lake Victoria fisheries: some facts and fallacies, Biological Conservation 5, 304-308.

Graham M. (1929) The Victoria Nyanza and its Fisheries. A report on the fishing survey of Lake Victoria 1927-28. London: Crown Agents, 255 pp.

Jackson P.B.N. (1971) The African Great lakes: food source and world treasure. Biological Conservation 5, 302- 304.

LVFO (2000) Results of the first regional frame survey on Lake Victoria conducted in 2000. Lake Victoria Fisheries Organization, Jinja, Uganda.

Mkumbo O.C & Cowx I.G. (1999) Catch trends from Lake Victoria - Tanzanian waters. In: Cowx I.G. & Tweddle D. (eds) (1999) Report on fourth FIDAWOG workshop held at Kisumu, 16-20 August 1999. LVFRP/TECH/99/07, The Lake Victoria Fisheries Research Project Technical Document No. 7, Jinja, Uganda, pp. 99-107.

SEDAWOG (1999) The survey of Lake Victoria's Fisheries. LVFRP/TECH/99/05. The Lake Victoria Fisheries Research Project Technical Document No. 5, Jinja, Uganda, 37 pp.

Sen S. & Neilsen J.R. (1996) The fisheries co-managment: A comparative analysis. Marine Policy 20, 405-418.

Sparre P. & Venema S. C. (1998) Introduction to tropical fish stock assessment, Part I: Manual. FAO Fish. Tech. Pap. 306:1 Rev. 2. 407 pp.

New Approaches to Inland Capture Fisheries Statistics In Sri Lanka

"Inland fisheries in most countries of tropical Asia are not managed scientifically."

Upali S. AMARASINGHE
Department of Zoology
University of Kelaniya
Kelaniya, Sri Lanka

Fisheries statistics in most inland reservoirs of Sri Lanka are not accurate. However, some reliable data are accumulated in a handful of reservoirs through various research activities. Potentially these data could be used to revise official fisheries statistics in some reservoirs. In reservoirs where middlemen are involved in fish marketing, logbooks can be consulted to improve data on catch and effort. However, over 90% of the total fish landings in Sri Lankan reservoirs are comprised of O. mossambicus and O. niloticus, which usually do not have price differences and these logbook records are not maintained species-wise. Despite this limitation, reasonable estimates on total fish production in reservoirs and catch per fisher can be collected from these books. Also, fisheries co-operative societies are functioning effectively in some reservoirs. It might be possible to obtain participation of these fisher communities in scientific data collection. One effective method is to use G.C.E. (Advanced Level) qualified youth in each reservoir to collect data on fish production and fishing effort.

Empirical yield predictive models based on catchment features of reservoirs quantified by Geographical Information Systems (GIS) have high predictive power. In these models, the ratio of catchment land use patterns to the reservoir area or reservoir capacity is used as a predictor variable. Using these models, it might be possible to predict fish yields of individual reservoirs with some accuracy. As fish yield is linearly related to fishing intensity expressed as boat-days/ha/year, fishing intensity corresponding to fish yield predicted by GIS-based empirical models can be determined.

Introduction

Inland fisheries in most countries of tropical Asia are not managed scientifically. One of the greatest problems in the development of inland fisheries in Asia is the lack of sufficient knowledge of sustainable use of fisheries resources, possibly due to lack of reliable data (De Silva, 1987). Inaccuracy in fisheries statistics is a common problem in developing countries (Marr, 1982). It has been suggested that fabricated returns are important to show that the policy pursued by the government for the development of the fishery is a success. Unfortunately, some statistical returns seem to be produced in this manner. Accurate catch and effort statistics are important for fish stock assessment and for planning social welfare programmes, economic analysis and human nutritional studies (Caddy and Bazigos, 1985). In this paper, published information on the inland fisheries of Sri Lanka is synthesized with a view to identifying new approaches for the improvement of inland capture fisheries statistics.

Brief review of the inland fishery of Sri Lanka

The inland fishery in Sri Lanka is essentially a capture fishery based on reservoirs. This is a relatively new development since the introduction of exotic cichlid species Oreochromis mossambicus into Sri Lankan freshwaters in 1952. The growth of the fishery and its recent trends have been detailed by Fernando and Indrasena (1969), Fernando and De Silva (1984), De Silva (1983, 1988) and Amarasinghe (1992, 1994, 1998). Reservoir fisheries are characterized by: (a) the use of non-mechanized fibreglass canoes, (b) use of gill nets and (c) the predominant catch is exotic cichlid species, O. mossambicus and O. niloticus.

From 1979 to 1989, the government developed the capture fisheries in reservoirs by providing fishers with fibreglass canoes and gill nets under a subsidy scheme. De Silva (1988) has shown that as a result, fishing effort considerably increased. The fisheries authorities have imposed regulations to control fishing effort and size of fish landed. Use of mechanized boats and any kind of shore seine nets is forbidden in perennial reservoirs and the minimum permissible mesh size for the gill net fishery is 8.4 cm. However, gill nets of mesh sizes smaller than minimum and beach seines are operated in some reservoirs sporadically (Amarasinghe and De Silva, 1992).

The dramatic increase of inland fish production from negligible levels before 1952 to very high levels (about 283 kg/ha/year in the 1980s (Fernando, 1984; De Silva, 1988) is said to be due to the ability of exotic cichlid species to colonize lacustrine habitats of reservoirs. Indigenous fish are riverine and marsh-dwelling fish species and cannot sustain dense populations in lacustrine habitats (Fernando and Holik, 1991). During the early 1980s, fisheries cooperative societies (FCS) functioned effectively for the simple reason that fishers had to be members to be eligible to receive boats and gill nets under the state-sponsored subsidy scheme. Under well-functioning FCSs, fishers tended to arrive at collective agreements regarding a complete stop in beach seining and an increase in the minimum mesh size of gill nets. These community based management strategies brought about considerable increase in fish production. The highest annual production, 39,300 tonnes, was reported in 1989 (Amarasinghe and De Silva, 1999). Production declined markedly after 1990 when the state discontinued patronage for a four-year period. During this period, government funding for monitoring and stocking programs was interrupted. In the absence of state monitoring programs, fishers began using smaller mesh gill nets that resulted in "growth overfishing". However, the fisheries have nearly fully recovered since the state renewed its support to these fisheries after the mid-1990s. Trends in inland fish production in Sri Lanka from 1978-1999 are shown in Fig. 1. Overall, fish production in most reservoirs has stabilized close to the optimal level or zero net-economic-revenue-level due to the open access nature of the reservoir fishery.

Figure 1. Trends in the inland fish production in Sri Lanka from 1978-1999. Percent contribution of the inland fishery to the total fish production is also indicated here (After Nissanka, 2001).

Reservoir fish yield (FY in kg/ha/year) and fishing intensity (FI expressed as boat-days ha/year) are linearly related according to the following equation (Fig. 2).

Figure 2. Relationship between fish yiedd and fishing intensity in reservoirs of Sri Lanka

FY = 11.036 FI + 24.867 (r = 0.775; p< 0.01)

(Source: Nissanka, 2001).

Methods of inland fisheries statistics collection

In Sri Lanka, inland fisheries statistics are collected by Aquaculture Extension Officers (AEOs) employed by the National Aquaculture Development Authority of Sri Lanka. AEOs are required to visit fish landing sites in the areas assigned and collect catch and effort data and information on species composition of landings. This procedure is unsatisfactory due to the lack of transport facilities and lack of incentives for field staff. Amarasinghe and Pitcher (1986), Amarasinghe (1992) and Pet et al., (1995) have shown that the pattern of overestimated yield in official statistics is a general trend in reservoir fisheries.

More reliable data on inland fisheries production are accumulated in a handful of reservoirs through various research activities (Amarasinghe and Pitcher 1986; Amarasinghe et al.;1987, 2002; Amarasinghe et al. 1989; Amarasinghe and De Silva, 1992; Pet et al., 1995). Of course, these data include comprehensive information on catch.

Various research teams have collected these data monthly (about 5 days a month at each landing site). Potentially, these data can be used to revise official fisheries statistics in some reservoirs. For this purpose, there would need to be a national level scheme to develop databases. These databases could be developed through the existing institutional mechanisms in research coordinating and monitoring agencies such as the National Science Foundation of Sri Lanka, Council for Agricultural Research Policy and National Aquatic Resources Research & Development Agency.

In some reservoirs, middlemen play a major role in the fish marketing process. Amarasinghe (1988) observed that over 95% of the daily landings in Pimburettewa reservoir (830 ha) had been purchased by the Secretary of the Fisheries Cooperative Society (FCS). These were taken to urban areas for retail and wholesale marketing. As the middleman maintain logbook records of daily catches of individual fishers, daily fish production data could be extracted for fish stock assessment (Amarasinghe, 1987). However, the total fish landings in Sri Lankan reservoirs consist of O. mossambicus and O. niloticus, which usually do not have price differences and these logbook records are not maintained species-wise. Despite this limitation, reasonable estimates on total fish production in reservoirs and catch per fisherman can be collected from these logbooks. Through a survey of fishing gear in each fishing household it is possible to gather information on the variations of fishing methods (i.e. number of net pieces and mesh sizes used), which in turn can be used to standardize the fishing effort.

In the Muthukandiya reservoir, there is a well-functioning FCS (Amarasinghe and De Silva, 1999) and fishers arrive at collective agreements on fisheries management and environmental protection. This co-management procedure in which the centralized administration authority of the Ministry of Fisheries and the fishing community share responsibility has been useful for preventing over-exploitation even during the period of non-state-sponsored monitoring procedures from 1990 to 1994 (Berkes 1994; Pomeroy 1995; Sen and Raajaer-Nielsen 1996; Amarasinghe and De Silva 1999). In this reservoir, the FCS collects one Rupee per kilogram of fish landed from those fishing to provide a welfare fund for the society. The FCS maintains a receipt book and issues a proper receipt to each fisher every day for the money collected. Using the records of the welfare fund, daily data on weight of fish landed by individual fishers can be collected. Through this procedure, total enumeration of fish production is possible. However, data on species composition are not available in these records.

Nissanka et al. (2000) and Amarasinghe et al. (2002) adopted a completely different procedure to collect reliable data in 11 other reservoirs. They assigned G.C.E. (Advanced Level) qualified youth in each reservoir to collect data on catch and effort from June 1997 to May 1999.

Detailed identification guides were provided to all data collectors and there were regular meetings with project personnel. G.C.E data collectors visited landing sites at least 20 days a month to record information on total fish catch in each boat and species composition of the landings. Length frequency data of the most abundant species (O. mossambicus and O. niloticus) were also recorded by these data collectors. This procedure sets a new norm in reservoir fishery statistics collection.

In Sri Lankan reservoirs, small-sized indigenous cyprinid species such as Amblypharyngodon melettinus, Puntius chola, P. dorsalis and P. filamentosus are abundant and can be differentially exploited by using small-mesh gill nets (Amarasinghe, 1985; De Silva and Sirisena, 1987). Due to the mesh restrictions these species are not exploited on a commercial scale. As the use of fishing gear other than gill nets is virtually impossible in most reservoirs due to the presence of impediments such as decaying tree stumps, these small cyprinids are not caught as a by-catch. Low consumer preference is another reason for not exploiting this resource. Amarasinghe (1990) reported small-scale fisheries operations for indigenous small cyprinids in some reservoirs that remain unreported in official fisheries statistics. Harmful fishing methods such as dynamiting and use of plant-derived poisons are negligible.

Use of fish yield predictive models

Amarasinghe et al. (2002) have shown the robustness of yield predictive models based on catchment features of reservoirs that were quantified by Geographical Information Systems (GIS). In these models, the ratio of catchment land-use patterns to the reservoir area or reservoir capacity is used as a predictor variable of fish yield. Of the various reservoir catchment land-use patterns, forest cover and shrub cover either singly or in combination had significant influences on yield. These relationships are shown in Fig. 3.

Figure 3. Relationships between fish yield (FYFSL) and ratios of different catchment land-use types to reservoir area and reservoir capacity.

(A)

(B)

(C)

(D)

Relationships between fish yield (FY) and ratios of different catchments land-uses to reservoir area (RA) and reservoir capacity (RC). FC - Extent of forest cover; SL- Extent of shrub land. All extents are expressed in km2. R2 = Coefficient of determination. (Sources: Nissanka 2001; Amarasinghe et al. 2002).

Amarasinghe et al. (2002) have shown by comparing actual yields with the yields predicted that the predictive power of these models is very high. As indicated by Meaden and Kapetsky (1991), GIS can be an effective means for data gathering and processing for a wide range of planning and management procedures. It might be possible to predict fish yields of individual reservoirs with considerable accuracy using the models, extents of different land-use types in catchment areas of reservoirs that can be determined from GIS methodologies and area and capacity of individual reservoirs. Also using the relationship between FI and FY presented in Fig. 2, it is possible to determine the FI corresponding to fish yield predicted by the above models. When the information on the fishery in question is hard to determine for standard stock assessment procedures, empirical yield predictive models provide an alternative method (Troadec, 1978).

Figure 3 shows the relationships between fish yield (FY_FSL) and ratios of different catchment land-use types to reservoir area and reservoir capacity.

References

Amarasinghe, U.S., 1985. Studies on the exploitation of minor cyprinids in Parakrama Samudra, a man-made lake in Sri Lanka, using gill nets. Journal of the National Aquatic Resources Agency (Sri Lanka), 32: 11 -23.

Amarasinghe, U.S., 1987. Status of the fishery of Pimburettewa wewa, a man-made lake in Sri Lanka. Aquaculture and Fisheries Management, 18: 375-385.

Amarasinghe, U.S., 1988. The role of fishers in implementing management strategies in reservoirs of Sri Lanka. In: Reservoir Fishery Management and Development in Asia (S.S. De Silva, ed.). 158-163 pp. International Development Research Centre, Ottawa, Canada.

Amarasinghe, U.S., 1990. Minor cyprinid resources in a man-made lake in Sri Lanka: a potential supplementary source of income for fishers. Fisheries Research 9: 81-89.

Amarasinghe, U.S., 1992. Recent trends in the inland fishery of Sri Lanka. pp. 84-105. In: Baluyut, E. A. (ed.) FAO Fisheries Report No. 458, Suppl. Rome, FAO, 281 p.

Amarasinghe, U.S., 1994. A synthesis on the management of the capture fisheries of Sri Lankan reservoirs. Vidyodaya Journal of Science, 5(1): 23-40.

Amarasinghe, U.S., 1998. Reservoir fisheries management in Sri Lanka: Achievements, mistakes and lessons for future. International Review of Hydrobiology, 83 (Special Issue): 523-530.

Amarasinghe, U.S. and S.S. De Silva, 1992. Empirical approaches for evaluating the efficiencies of different fishing methods in tropical, shallow reservoirs: a Sri Lankan case study. In: Reservoir Fisheries of Asia (S.S. De Silva, ed.). pp. 217-227. International Development Research Centre, Ottawa, Canada.

Amarasinghe. U.S. and S.S. De Silva, 1999. The Sri Lankan reservoir fishery: a case for introduction of a co-management strategy. Fisheries Management and Ecology, 6: 1-13.

Amarasinghe, U.S. and T.J. Pitcher, 1986. Assessment of fishing effort in Parakrama Samudra, an ancient man-made lake in Sri Lanka. Fisheries Research, 4: 271-282.

Amarasinghe, U.S., S.S. De Silva and C. Nissanka, 2002. Evaluation of the robustness of predictive yield models based on catchment characteristics using GIS for reservoir fisheries in Sri Lanka. Fisheries Management and Ecology 0: 000-000 (in press).

Berkes, F. 1994. Co-management: Bridging the two solitudes. Northern Perspectives 22: 18-20.

Caddy, J.F. and G.P. Bazigos, 1985. Practical guidelines for statistical monitoring of fisheries in manpower limited situations. FAO Fisheries Technical Paper 257: 86 p.

De Silva, S.S. 1983. The reservoir fishery: Present status and future strategies. Journal of Inland Fisheries, Sri Lanka 2: 3-13.

De Silva, S. S. 1987. Reservoir and lake fisheries. Archiv für Hydrobiologie 28: 269-271.

De Silva, S.S. 1988. Reservoirs of Sri Lanka and their fisheries. FAO Fisheries Technical Paper 298: 128 p.

De Silva, S.S. and H.K.G. Sirisena, 1987. New fish resources of reservoirs in Sri Lanka: Feasibility of introduction of a subsidiary gill net fishery for minor cyprinids. Fisheries Research 6: 17-34.

Fernando, C.H. and S.S. De Silva, 1984. Man-made lakes: Ancient heritage and modern biological resources. In: Ecology and biogeography of Sri Lanka (C.H. Fernando, ed.). pp. 431-451. Dr. W. Junk Publishers, The Hague, The Netherlands.

Fernando, C.H. and J. Holik, 1991. Fish in reservoirs. Internationale Revue gesamten Hydrobiologie 76: 149-167.

Fernando, C.H. and H.H.A. Indrasena, 1969. The freshwater fisheries of Ceylon. Bulletin of Fisheries Research Station, Ceylon 20: 101-134.

Marr, J.C. 1982. The realities of fishery management in the South East Asian region. In: Theory and management of tropical fisheries (D. Pauly and G.I. Murphy, eds). pp. 299-307. ICLARM Conference Proceedings 9, 360 p. International Centre for Living Aquatic Resources Management, Manila, Philippines and Division of Fisheries Research, Commonwelath Scientific and Industrial Research Organization, Cronuulla, Australia.

Meaden, G.J. and J.M. Kapetsky, 1991. Geographical information systems and remote sensing in inland fisheries and aquaculture. FAO Fisheries Technical Paper 318: 262 p.

Nissanka, C. 2001. Empirical models for fish yield prediction in selected reservoirs in the dry zone of Sri Lanka. Unpublished Ph.D. Thesis, University of Kelaniya, Sri Lanka, 257 p.

Nissanka, C., U.S. Amarasinghe and S.S. De Silva, 2000. Potential yield predictive models for Sri Lankan reservoir fishery. Fisheries Management and Ecology 7: 425-436.

Pomeroy, R.S. 1995. Community-based and co-management institutions for sustainable coastal fisheries management in Southeast Asia. Ocean and Coastal Management 27: 143-162.

Sen, S. and J. Raajaer-Nielsen, 1996. Fisheries co-management: a comparative analysis. Marine Policy 20: 405-418.

Troadec, J.P. 1978. Semi-quantitative methods of assessment. In: Methods for fish stock assessment. FAO Fisheries Circular 701: pp 99-109. FAO, Rome, Italy.

Pet, J.S., J.W.M. Wijsman, P.J. Mous and M.A.M. Machiels, 1995. Characteristics of a Sri Lankan reservoir fishery and consequences for the estimation of annual yield. Fisheries Research 24: 9-33.