CHAPTER 7

THE FISHERY

THE FISHERMEN

All fisheries are to a great extent shaped by the nature of the environment and the characteristics of the fish stocks they exploit. The factors influencing fisheries in rivers are very different from those shaping lacustrine and marine fisheries. Three factors in particular give riverine fisheries their character, diffuseness in space, seasonality and diversity. Because these are common to the vast majority of the world's rivers there is a remarkable parallelism in the general form of riverine fisheries and the communities that exploit them. This is subject, of course, to certain modifications according to the general state of development of the countries in which fishery is practiced.

Rivers are lineaform and of limited width so the total area of water that can be reached from any point on the bank is limited by the capacity to move up- or downstream. In most temperate rivers commercial fisheries are usually motorized and the developed road networks make access relatively easy. In many of the tropical rivers movement is by hand propelled canoe, although outboard motors are also used. What ever the mode of propulsion it is rare for there to be an expanse of water available to one landing sufficiently large to support a capture and marketing operation of any size. In many regions the problems posed by the linear form of rivers are aggravated by the swampy and changeable nature of the terrain, which, together with the periodic submergence of any flanking floodplains, hinders the installation and maintenance of permanent roads and other forms of communication. In most areas of the world spatial dispersion and inaccessibility have combined to make fisheries labour intensive, artisanal operations which are located in a series of small settlements along the river channels or spread over the floodplain on islands of higher elevation. There are some obvious exceptions to this general principle in estuarine 10/12/0waters, large floodplains or lakes and reservoirs associated with rivers where the available area of water is greater.

In flood rivers the seasonal expansion and contraction of the water area can separate settlements from the main course of the river by many kilometers in the dry season even though they are at the waters edge during the flood. Coupled to the hydrological cycle are the migrations of fish within the system which causes the stock to undergo changes in abundance density and location. In response to the fluctuations in the ecosystem fishermen either have to alternate with other occupations or they themselves have to migrate. Because of this it is very difficult to define accurately who is, or is not, a fisherman. In some systems the situation is fairly clear, as fishing is the task of one particular ethnic group or tribe, and practically the whole of the active population of such a group may be interpreted as being involved in the fishery in some way. In such cases the fishing group often does not own land and coexists with other groups equally specialized in agriculture or pastoralism. In other rivers, the riparian population remains relatively unspecialized as a whole, although sub-groups may concentrate on one or more of the specialized activities including fishing. Within the fishing community there is usually a well defined division of labour. The men fish, construct and maintain the gear and build the boats, whilst the women collect, treat and market the produce. In such populations most able bodied male, and sometimes female, individuals fish at some time or other. Furthermore, the role of small pre-adolescent boys in many artisanal fisheries should not be underestimated as they act as fisheries aids. In this capacity they paddle canoes, bait and control long lines, lift traps and generally keep the fishery going when the adult members of the community may be occupied elsewhere.

For the purposes of convenience fishermen may be classified into three categories on the basis of the time they spend fishing (FAO/UN, 1962). These are:

(i)   occasional fishermen;

(ii)    part-time fishermen;

(iii)    full-time fishermen.

Such classification refers mainly to food fisheries in the tropics and sub-tropics and excludes such special categories as temperate zone recreational fishermen.

OCCASIONAL FISHERMEN

In contrast to lakes, where a boat is usually essential to reach fishing grounds and to operate gear, many river channels and floodplain waters can be reached on foot during part of the year at least and fished by simple apparatus from the bank or by wading. This relative ease of access for the inhabitants of floodplains coupled with a certain amount of free time between the sowing and the harvest of floodplain crops, means that casual fishing for subsistence is popular in floodplain communities. Furthermore, in many areas of Africa and Asia, certain depression lakes or sectors of the river and floodplain are traditionally reserved for the inhabitants of particular villages. These are fished during festivals or fish drives which take on all the aspects of a holiday and in which all members of the community participate. Most of the fish caught enters directly into the diet of the fishing community. The individual time spent is low and for the most part the gear used is simple and relatively unproductive. On the other hand the numbers participating are often very high. It is therefore difficult to assess the contribution of such efforts to the total catch of any particular system.

PART-TIME FISHERMEN

Many sedentary peoples living on floodplains fish during part of the year. This is an activity that is co-equal to or inferior to the alternative activities of such populations. The flood cycle, the biological cycle of the fish, and the seasonal needs of agriculture impose a cyclicity on such communities. During the floods there is very little activity in either domain, but as the waters drain from the plain, fishing increases. As the floodplain dries the preparation of the soil and sowing of the seeds take priority, to be followed with a second burst of fishing at low water. Harvesting of the crops follows and the cycle repeats itself year by year. It is perhaps not surprising that, while part-time fishermen use most of the types of gear used by professional fishermen, they also have a tendency to practice various types of extensive aquaculture techniques. For example, the development of drain-in fish ponds, is associated with a certain agricultural type of land and water tenure in countries as widely separated as Cambodia (Chevey and Le Poulain, 1940) and Benin (Hurault, 1965). Similarly, the association of drain in ponds with paddy fields is a feature of many rice growing communities (Tang Cheng Eng et al., 1973).

PROFESSIONAL FISHERMEN

In most aquatic systems there are groups of individuals that live entirely by fishing. The need for year round employment and the movements of the fish stocks often force such groups to be nomadic. Migratory fishermen have been noted from many systems and are particularly a feature of river fisheries. For example, Bhuiyan (1959) described the migrations of the Hilsa fishermen of the Indus who followed the movements of that species up and down the river. The fishermen of the Kafue similarly move around depending on the water level and abundance of the fish stocks (Everett, 1974). In the Magdalena fishery the fishermen move from the cienagas where they fish in the dry season, to intercept the “Subienda” migration in the river (Bazigos et al., 1977). Some of these migrations by fishermen can be of very large proportions. The historical upstream movements of the Haoussa in the Niger river (Daget in FAO, 1962), might reasonably be compared to the operations of a factory trawler in oceanic fisheries when the size of the investment relative to the per capita income of the community is taken into account. This movement, which may be taken as fairly typical, involved whole families of 20–30 persons who moved upstream in September and October to fish the northerly portions of the Central Delta on the falling flood. The distance moved was in excess of 1 000 km. The main vessels, which were often up to 15 m long and were equipped to support whole families on the journey, were accompanied by a flotilla of small craft which were used in the actual fishing. On the return journey, which was made on the next flood, the main vessels were loaded with many tons of fish for sale in the Nigerian markets. These northern parts of the Central Delta are far from the main centres of population, and have been exploited by wandering fishermen of the Haoussa and Bozo tribes for many centuries. The incidence of nomadism therefore increases from the southern parts of the basin where only two percent of the population are mobile to 52 percent in the heart of the lake district (Raimondo, 1975).

Not all nomadic fishermen undertake such long journeys, but most have to leave their native villages during at least part of the annual cycle. Many fishermen construct temporary 10/12/01fishing camps on high ground within the floodplains which they occupy during the fishing season and which they move following changes in the river level. To encourage this, it has been proposed that artificial islands be constructed on some floodplains, especially the Kafue and the Sudan Sudd, to enable the fishermen to exploit areas that have so far remained undeveloped for lack of suitable living space. In some flooded areas, such as the Mesopotamia area of the Tigris-Euphrates system or the Ganges delta in Bangladesh, the rural community including fishermen and their families already live on islands made of soil and domestic refuse accumulated over the years. As an alternative many fishing communities have developed houses on stilts which remain above all but the highest floods. These are typical of the lowland rivers of the West African coast and of Asia. In Asia, too, some fishermen live, at least temporarily, on boats or rafts which are associated with traps, stownets or lift nets in the river channel (see Fig. 7.1).

Figure 7.1  Weighted average monthly landings related to mean water level of the Yamuna river at Agra during 1968–69 (After Wishard, 1978)

Professional fishermen use a vast range of fishing gear, but in recent years have tended to concentrate on one or two of the fishing methods based on modern materials such as seine nets, gillnets or cast nets. Such fishermen, too, use powered craft to a great extent, if not for fishing then for the transport of fish from the fishing grounds to the markets.

In recent years long distance nomadism has been discouraged in many parts of the world as this practice has involved the movements of the nationals of one country into the waters of another. With increasing awareness of the potentialities of their inland fisheries most countries have felt the need to retain these benefits for their own nations and have trained local people accordingly. Many of the traditional movements, such as that described for the Niger, have therefore broken down and in their place shorter local mig rations have intensified.

Most artisanal fishing communities have developed elaborate traditions, legends and even religious systems which enable them to integrate culturally with the general ecology of the river on which they live. Parallel institutions are found, not only among tropical communities, but also in many early European and North American temperate peoples. Such studies as those of Smith (1981) or the many papers included in Gunda (1984) indicate the rich folklore associated with such communities, many elements of which serve genuinely conservationist ends. Such traditional mechanisms for regulating the fishery include the definition of closed seasons, the delimitation of refuge areas or limitation on access to the fishery. Such traditional institutions are fragile and disappear under pressure from more modern political and socio-economic systems. For example, fishing is often seen as a kind of relief occupation from the chronic unemployment and land hunger of much of the tropical world. The influx of new and inexperienced fishermen into the traditional fishing communities results in their disruption. Unfortunately no new institutions arise to replace the traditional management of the fishery and the resulting anarchy may harm not only the fishing communities but also the fishery resources on which they depend.

FISHING GEAR

SEASONALITY

In stable water bodies, including reservoir rivers and the larger permanent water bodies associated with river systems, fisheries are pursued around the year. In cold temperate zones fishing may be intensified in such rivers during the otherwise unfavourable cold season because of the greater activity of salmonids at such times and in some areas fishing even continues under the ice cover. In flood rivers, on the other hand, fish are vulnerable to the fishery for only part of the year when their accessibility depends much on the ichthyomass per unit area as represented in Fig. 6.17 or during temporary concentrations of migrating fish. During high water fish are dispersed over the floodplain and concealed by mats of vegetation. The bulk of the population are juveniles, often too small to be captured by any practical gear. Physical conditions are also unfavourable to the fishery as strong currents sweep down the main channel and out over the plain carrying with them logs and floating masses of vegetation that carry away any fixed engines. As a consequence there is a slackening or complete absence of fishing during the flood season in most river systems, although some gears such as the ‘Atalla’ lift net of the lower Niger river, are operated specifically at such times to catch small species which have taken refuge in vegetation fringing the main channel or juvenile fish which are concentrated in the shallow margins of the lagoons (Awachie and Walson, 1978). Fishermen may also continue to be active in some of the larger floodplain lakes where conditions are more stable or in the flooded forest where specific fisheries for frugivorous fishes take place (Goulding, 1981). In older and more tradition bound fisheries this cessation of activity is often reinforced by laws or taboos which close the fishery to some or all gears. Many such traditional bans seem to have arisen from the recognition of the flood season as a period of breeding essential for the continuation of the stock.

Fish become more available for capture as they congregate in the channels and pools of the floodplain as the water begins draining off the main channel. At this time there is a very heavy and concentrated fishery. Intensive fishing continues throughout the dry season both in the standing waters of the floodplain and in the main river channel. Later as the water begins to rise again fairly specialized fisheries concentrate on the adult fish migrating to their breeding sites. In most river systems this sequence gives a pattern of capture whereby the efficientcy of capture is inversely proportional to the strength of the flood or is related to water level by more complex relationships such as those described by Wishard (1976) for the fisheries of the Yamuna river at Agra (Fig. 7.1).

CAPTURE METHODS

The types of fishing method chosen for use in rivers are conditioned by three factors; first, the nature of the fish stock, second, the form of the river and third the degree of development of the fishing community. As has been described the fish communities of tropical and sub-tropical rivers are particularly diverse containing a large number of species, most of which differ to some degree in their selection of habitat, diet, migration pattern and ease of capture. They are also represented by an age structure which is more than usually biased toward juvenile fish. The fishermen have, therefore, to establish priorities for their fishery with respect to which species and age classes they attempt to capture. Preference is often imposed by local food taboos or customs, although there are a few fisheries which have been founded on the export of species which are not locally accepted items of diet. In temperate climates the diversity of the community is reduced but, even in complex fisheries such as that of the Mississippi, a few species only are selected as forming the basis for capture. In some tropical fisheries too, particularly those at an early stage of development, only the larger species are captured and in such restricted fisheries the range of gear in use is quite limited. In many other fisheries, however, the fishery exploits all species that are catchable and so a great number of fishing methods have evolved. Until recently, when cotton and later nylon twine and nets were introduced to river fisheries, gear was constructed uniquely from local materials. Roots, vines, plant fibres, leaves, stems etc. have all been, and still are, used for much of the gear encountered in rivers and floodplains. The similarity in comportment of the different elements of the fish communities and the raw materials of which the gear is made, have led to a considerable parallelism in most methods, even though they have clearly been developed independently.

There are a number of local variations, and these are described in the many catalogues of fishing methods in various countries such as Chevey and Le Poulain (1940) for the Mekong, Ahmad (1956) for the waters of Bangladesh or Blache and Miton (1962) for the Chari/Lake Chad basin. A detailed description of all gear used in river fisheries is inappropriate here so this chapter will examine the general principles of the capture methods used in as far as they are relevant to the broad ecology of these systems. Clearer descriptions of artisanal and other fishing gears, classified by type are available elsewhere, for instance in Andreev's (1966) “Handbook of fishing gear and its rigging” or in Brandt's (1984) book on “Fish catching methods of the world”.

While general types of gear used are determined by the above there is also a seasonal selection depending on the flow regime. In flood rivers this tends to fluctuate during the year leading to the successive use of a variety of gears. Typical examples of this are the fisheries of the Oueme R. in Africa (Fig. 7.2) and the Mekong in Asia (Fig. 7.3). Similar examples of differential use of gears according to flood stage are given by Goulding (1981) and Smith (1981) for rivers of the Amazon system. In rivers with more constant flows the methods used are more restricted and generally correspond to the gear types used during the low water phase in flood rivers.

Figure 7.2  Fishing timetables for the Mekong fishery. Thin line = fishing possible; thick line = peak effort with method. Subsistance methods are not included in the tabulation. (Adapted from Fily and d'Aubenton, 1966)

Figure 7.3  Fish gear utilization in the lower Oueme river during the course of one year

Certain types of fishing gear relying on modern industrially manufactured twines are in widespread use. These are the seine net, the gillnet and the cast net. In many systems these methods are replacing much of the traditional gear, and they are especially favoured by professional (full-time) fishermen as their individual catching power is superior. Their care and maintenance requires an expertise often lacking amongst the more casual elements of the fishing community. The cast net is in fact one of the mainstays of these fisheries. Conditions in reservoir rivers and in flood rivers in the floodplain lakes and the main stream at low water are well suited to its use. The water is fairly shallow, the bottom unencumbered and the fish are sufficiently concentrated to give a good chance of capture. The mesh size can be varied according to the species and size of fish sought. Fishermen using cast nets may fish either individually or in combination with others whose manoevres serve to concentrate the fish still further. The seine net requires relatively large teams to operate it and is very expensive. It is, therefore, the gear of the professional “par excellence”. Its use is limited by current and especially by the availability of bottoms which are sufficiently free of obstructions which would otherwise cause the net to snag. The tenure of suitable seining beaches is often hotly disputed in the fishing community. As a gear it has several precursors among traditional methods and communal fish drives often have a seine like approach with lines of fishermen wielding baskets or clap nets. Alternatively a barrage fence, which is one of the basic items of equipment for the floodplain fisherman, is set in a large semi-circle and moved inshore in the same way as the seine. The gillnet is also common although it is sensitive to floating vegetation and will not operate in strong flows. On the other hand, it does come into its own in the larger floodplain lagoons and lakes where the depth, the large stretches of open water and lack of current make it one of the most effective gears.

These gears either singly or together dominate certain fisheries. In the Magdalena river the cast net in the main method, although seines, gillnets, traps and spears are also used to some extent. In recent years the seine net has become the principal gear in the Senegal river, to the exlusion of many traditional methods which are still current in the neighbouring Niger river (Reizer, 1974). The Kafue river fishery also relies almost entirely on the combination of gillnets and seine nets (Everett, 1974).

Catches by these methods vary considerably according to the construction of the gear, the habits of the fishermen and the time of year.

Low Water

During the low water period the majority of fish tend to remain relatively static. Riverine migrant species which ascend the main course of the river and its tributaries during low water are an exception to the more general behaviour pattern and form the basis of such fisheries as those of the “Subienda” and “Piracema” in Latin America, the Alestes fishery of the Chari river, the Khone Falls fishery as described by Chanthepha (1972), and the “Roak” fishery of the Yamuna river (Wishard, 1976). Fishing methods tend to be active although some static gear is also used during low water. This is often baited, for fish are still attracted to baited gear despite the general lowering of feeding intensity. At this time the use of gillnets, seine nets and cast nets is maximal both in the river and in the lakes and lagoons, but these gears on their own cannot reach all habitats or catch all species, consequently many other fishing methods are in use. Hook lines of various types are common in the main river channel. Longlines, baited with starch paste, offal, small fish, etc., are stretched across the river and catch mainly the larger predatory species which are often not vulnerable to other gears. Unbaited snagging and entangling lines are laid in the deeper portions of channels where big fish are accustomed to rest, and small boys are usually to be seen in most places with a rod or leger line with which they capture a seemingly endless succession of small fish.

Barrages are common throughout river systems, sometimes taking complex labyrinth-like forms, but more often simply dividing the main river channel or the smaller drainage canals into sections which limit fish movement and facilitate their capture by other means. Many types of trap, (Fig. 7.4), are associated with the barrages where they capture the fish that are milling about trying to pass the barrier. Traps are also set by themselves amongst vegetation where they attract fish seeking refuge there. The traps may be unbaited, but they may also be baited to select for certain species. In West African rivers for instance the same type of cylindric trap catches greater proportions of Clarias when baited with oil palm fruits, Tilapia/Oreochromis when baited with maize meal, and Macrobrachium when rotting meat is used.

The problem of extracting fish from under the vegetation which fringes lagoons and river channels is tackled in a number of ways. Special robust hand nets, which may be made out of netting or basket work (Fig. 7.5), are scraped along the under-surface of the vegetation, or the vegetation mass might be surrounded by a fish fence. In the latter case the plants are cut out piece by piece and the fence advanced inwards so as to enclose the fish within a small space from which they can be captured by hand nets or baskets. In the Oueme river such vegetation masses may be planted deliberately at the end of the flood, either attached to the bank or recessed into it at the mouth of the channels which drain the plain. They are left to collect fish for about two months after which they are fished and replaced to be emptied again toward the end of the dry season. Harvests from such “refuge traps” or “fish parks” can be quite high and in the Oueme river 15 installations of this type of mean area 440m² gave a mean harvest equivalent to 1.88 t/ha (of park) per fishing or 3.88 t/ha/yr between 1958 and 1968.

A variety of gears have been developed to fish the main channel, especially the deeper portions where the larger fish come to rest. Many of these are drawn or propelled by boats. A fairly widespread device is the frame trawl drawn by one or two boats but the most popular is probably the Vee-shaped net (Fig. 7.5) which can be mounted in a variety of ways and has great operational flexibility. Armed with small mesh netting, it can, for instance, skim the surface to catch the small pelagic clupeid, cyprinid and characin species found there, or be scraped along the bottom to capture many small bottom-living species. With larger mesh it can also be plunged near the bottom where it drags for bigger bottom-living species. With larger mesh it can also be plunged near the bottom where it drags for bigger bottom-living siluroids. Nets similar to this may also be mounted on the bank where they take migrating fish as they follow the shoreline or move inshore away from the current (Fig 7.6). In Asian rivers lift nets perform a similar function. They may be mounted on the bank, but are sometimes operated from rafts on which the fishermen live with their family. Such an apparatus is the “Sadung” of the middle Mekong, whose operation was described by Fraser (1972).

The standing waters of the floodplain are often the property of a particular village or group, which exploits them communally using clap nets or plunge baskets to virtually empty the water of any living thing. But many waters are exploited by individuals.

Figure 7.4  Various types of fish trap from tropical rivers: (A) cylindrical drum trap (worldwide); (B) vertical slit trap (Asia, Bangladesh and Mekong river); (C) folded woven trap (Niver river); (D) funnel trap (worldwide); (E) spring trap (Africa, Niger, Chari and Zaire rivers)

Larger lakes, of course, are fished in much the same manner as the main channel with a full variety of gear. However, smaller floodplain lakes are usually either poisoned or fished out with fences in much the same manner as the brush parks. There are many plant toxins which are used to capture fish both in the lagoons and in pools of the main river channel. Table 7.1 gives some indication of the number of bio-active plants in the Benue river system alone; although many of the shrubs listed are also found in other parts of the tropical world, and Gunda (1984) describes an equally large range of plants used for poisoning fish in European rivers. In recent years, even more powerful poisons have become available in the form of insecticides such as dieldrin or endrin, which are used with apparent abandon. Fishing with poisons is one of the main methods used by occasional fishermen and has the unfortunate side-effect that it selects against the young fish which are particularly susceptible to these substances. Its effect on the fish stock is of course minimal where temporary pools are fished, for there the fish would die anyway before they could rejoin the stock but in permanent standing waters the annihilation of the population is more serious, as here the fish take refuge during the dry season to form a reconstituted stock in the floods. Poisons are also used to eradicate unwanted species, for example, the piranhas (Serrasalmus spp.) in Brazil (Braga, 1976).

Figure 7.5  (A) Basket dip-net (Ouemem river); (B) “Vee” shaped dip-net (worldwide); (C) “Vee” shaped net mounted on a canoe (Chari river) (adpated from Blache, Miton and Stauch, 1962)

Figure 7.6  (A) Bank mounted lift-net (widespread, example from Niger river); (B) handheld life-net (widespread, example from Bangladesh); (C) raft mounted lift-net “Sadung” (middle Mekong)

As has been shown in the section on standing stocks, many lagoons contain a very high ichthyomass, most of which is removed during intensive fishing of these waters. This fact has apparently been recognized and, as we have seen in some areas, fishermen attempt to retain the maximum volume of water in the depressions by damming the outlet channel. In more developed floodplains, the shape of the lagoons may be regularized and eventually drain-in fish ponds may be dug into the surface of the plain.

Fishing during the Rising and Receding Floods

As the floods start to rise there is a burst of activity by the fish as the adults move preparatory to breeding. Locally high concentrations are present giving rise to very heavy but localized fisheries. When the floods recede from the plain the water becomes confined increasingly into depressions and channels. The fish follow these flow patterns to reach either the main channel or what will become the standing waters of the plain itself. Fishing methods take advantage of these movements and are mainly aimed at either directing the fish into places where they are more easily captured, or to retaining the fish in floodplain depressions from which they may be more easily removed later. Such gear may be based on bamboo or palm frond fish fences which are installed across the plain or the channels through which the water enters or leaves. They can reach a considerable length and can be arranged in complex forms giving a labyrinth-like effect. Capture is either in trap shaped chambers (Fig. 7.7) or in special cylindrical traps or nets which are placed in openings in the barrage. Durand (1970), for instance, has described a cross river barrage from El Beid river which drains the Yaeres floodplains of the Logone river. Here the fences are arranged in a series of Vee's at the apexes of which are held large hand nets. Both upstream and downstream migrants may be caught in the same type of trap.

Figure 7.7  Barrage trap from coastal floodplain of Benin; note heart-shaped capture chamber

Thus in the Nzoia river the ‘kek’ type of barrier trap (Fig. 7.8) intercepts breeding fish moving upstream, as well as adult fish returning from the spawning grounds (Whitehead, 1959a). Such fences can contribute a large proportion of the total annual catch in some systems. In the Lubuk Lampam (Indonesia) guide fences accounted for about 50 percent of the fish caught in 1975 (Arifin and Arifin, 1976) and on the African Barotse plain the “maalelo” fishery produced about 25 percent (631 t) of the total catch in 1969. The “maalelo” of the Barotse floodplain is perhaps typical of the more open plain type of barrier fishery. Here the fish guides are usually earth bunds some 75 cm high and between 3 to 40 m long, which deflect fish into traps placed at intervals along their length, although reed, wire mesh or brushwood fences may also be used. Alternatively the bunds may join two areas of high ground so as to retain a pool behind them. When eventually the dam thus formed is breached, the fish are caught in traps and baskets at the outflow giving a yield equivalent to about 33.6 kg/ha of area impounded (Bell-Cross, 1971). A few “maalelo” are fished during rising flood. Weiss (FAO/UN, 1970a), estimated that there were 10 000 of these weirs operating each year on the Barotse plain.

Figure 7.8  Fishing basket used in Kenya (A), and arrangement in a “kek” barrage (B). Fish enter through the vertical funnel. (After Whitehead, 1958)

Complete blocking of small channels leading water out of depressions is common in other systems; it has been noted from the Gambia river by Svensson (1933) and by Chevey and Le Poulain (1940) from the Mekong. The pool thus created is fished, often considerably later in the dry season, either by breaching the dam and catching the fish in traps, baskets or nets or by bailing the water out until the enclosed section is dry. This principle has been suggested as a means for improving the fishery productivity of floodplains as an extensive form of aquaculture.

In large channels which are not easily blocked with dams, stownets and wing traps form an alternative to the barrage trap. The stownet is a conical fixed gear which operates rather as a static trawl, the water passing through the net rather than vice versa. In the River Rhine these are operated using a single anchored otter board, in the Oueme such nets are slung from poles securely stuck into the bottom, but in the Tonle Sap, which drains the Grand Lac into the Mekong, they reach complex proportions in the form of the “day” (Fig. 7.9). Such gears give very high yields during the main migrations into and out of the major flood depression and in the main channel. Fily and D'Aubenton (1966) recorded a mean catch of 33 t per unit for two lunar catching periods for 11 installations in 1962–63. Earlier results by Chevey and Le Poulain for 1938-39 gave about twice this figure (64.5 t). In the rivers and canals of the Chao Phrya and Mekong deltas, stationary wing traps of the type illustrated in Fig. 7.10, also take large quantities of migrating fish as they moved up and down the channels during the dry season. According to Tongsanga and Kessunchai (1966) catch rates ranged between 1.2 and 3.8 t/day in various canals.

Figure 7.9  Contribution of a “day” stow-net from the Tonle Sap: (A) cylindrical trap (“day”); (B) disposition of individual traps in stow-nets mounted in a barrage; (C) method of use (i) with gear lowered, and (ii) with gear raised (after Chevey and LePoulain, 1940)

Fishing at Peak Flood

Although in most rivers fishing is highly seasonal with only minimum effort being expended at high water, some fisheries are pursued during the floods particularly in forested rivers such as the Amazon. Such fisheries concentrate on frugivorous species which congregate in areas rich in fruit bearing trees and shrubs. Here fish are captured mostly by hooklines, although bows and harpoons are also employed for larger species (Goulding, 1981; Smith, 1981). In savanna rivers fishing continues in some systems using traps, lines or gill nets set in areas of slack water or in the deeper parts of floodplain lakes. However, the risk of loss of gear, the low return on effort and the difficulty of working on a vegetation encumbered floodplain tend to restrict the practice.

Figure 7.10   Stationary wing-trap from the Chao-Phrya delta. (After Tongsanga and Kessunchai, 1966)

BOATS

A detailed description of the various kinds of fishing craft is not appropriate here, but the availability of suitable means of water transport is crucial in a fishery which depends much on the mobility of the fishermen. Furthermore, in flood rivers the plains are inundated for several months of the year, and the communities inhabiting them are forced to adopt a semi-aquatic way of life. During the floods water-born transport is the only means of communication and even the markets are conducted from canoes. As a simplification, two main types of boats are used in tropical and sub-tropcical fisheries. The first are the fishing craft themselves which are usually dug out canoes or planked craft between 4–6 m long. Motorization of the smaller fishing boats is comparatively rare, especially among part-time fishermen, and is in many cases of no great advantage. The second class of boats are longer, often up to 10 m long and are used for transport of fish from the landing to the market. These are more frequently motorized. In the main equatorial rivers, the construction of boats presents few problems as wood is plentiful and dug-outs are easily made. In the savanna rivers, however, the lack of good wood is limiting and canoes are often scarce and expensive as they have to be imported from elsewhere. The development of suitable substitutes for traditional types of craft, which are cheap enough for the fishermen, is one of the main preoccupations of fisheries administrators in such areas.

PRESERVATION OF FISH

TYPES OF PRODUCTS

A certain amount of the fish caught in river systems is consumed fresh by the fishermen themselves and by communities with a limited radius of the fish landings, but in the more important fisheries, a surplus to local requirements is produced which is sold for transport elsewhere. To improve the quality of their product, fishermen as far apart as the Mekong river and the Magdalena river, keep their catch in live chambers and even transport it in special boats with wet holds. Some species survive and keep better than others in the fresh state after landing, and air-breathing fish particularly are sought after because of the time they can be kept alive after capture. In India and part of Africa, murrels, cat fish and some anabantids are transported for considerable distances in baskets lined with damp weeds or moss. However, because of the dispersion and inaccessibility of fishing sites in the river-floodplain system, and the rapidity with which fish deteriorate under tropical conditions, most fish have to be preserved by one means or another for it to arrive in the markets in an acceptable condition. Furthermore, the seasonal nature of the fishery means that a period of excess fish production is followed by one of scarcity. Preservation techniques are thus needed to prevent fish that are in excess of demand being lost and to even the supply of the year. Several types of treatment are used, depending on local conditions and preferences.

Chilled Fish

The preservation of fish by icing or freezing is a comparative innovation to the river fisheries. Even now its use is limited to areas where sufficient fish is caught to justify the expense of ice-making plants, and where communications are sufficiently easy for the fish to be collected, iced and removed rapidly. This method of keeping fish has grown up on major fisheries located in the main river channel such as that for Hilsa ilisha in the Indus (Husain, 1973) where the fish are collected at certain landings for icing prior to transport by rail to the major towns. Icing is especially popular in Latin America, where the comparatively recent development of the river fisheries has left little time for traditional methods of preservation to have arisen. On the Magdalena river and the Amazon, for instance, fish traders come to the fishing sites in motorized boats equipped with ice to preserve the fish. In recent years, preservation in ice has tended to replace other methods of treatment. In the Mekong, Fraser (1972) recorded the addition of ice preserved fish to the customary range of products, and in the Kafue, Williams (1960) commented on the growing proportion of fish which is carried in this form.

Dried Fish

Sun drying of fish without salting is not practical in many of the world's river systems because of the high atmospheric humidity. But in desertic or Sahelian savanna rivers the practice is common, especially for smaller species. In the Senegal basin, for example, simple drying is the usual form of treating the fish after they have been eviscerated, scaled and, in the case of larger species, cut into strips. On the Niger, only the smaller species, such as Alestes, are sun-dried and in the Chad basin one of the traditional fish products, “Salanga”, consists of Alestes dentex and A. baremoze which are split open ventrally and laid on mats to dry in the sun. In the Mekong, and other Asian rivers, sun-drying is also common, although this is sometimes combined with salting.

There is a noticeable loss of weight as fish dry. This varies much with the species of fish, for instance, Clarias with their massive head bones tend to lose less than do characins or cyprinids. It is generally agreed that the ratio wet fish: dry fish is about 3:1.

Salting

Preserving fish with salt is not common in inland waters, largely because of the high cost of the salt, and secondly because in the more humid areas the deliquescence of the salt shortens the life of the product rather than increasing it. In the Indus valley, salting is used only when there is a very heavy catch of Hilsa which exceeds the capacity of the ice plants (Husain, 1973). Certain species are prepared in this way in the Parana river (Vidal, 1969), where Lycengraulis olidus and more recently Hoplias malabaricus and Pseudoplatystoma spp. are treated with salt prior to sun-drying.

Smoking

Smoke-drying is perhaps the most widespread way of preserving fish in Africa and is practised in nearly all river systems. Techniques vary somewhat and the type of oven used changes from place to place. In Malawi, fish are smoked in a special thatched smoke house, whereas in other regions ovens constructed of clay are open to the air (Fig. 7.11). Small fish are usually smoked whole. Medium sized fish are scaled, eviscerated, and sometimes split open or slashed down the sides. The largest fish are cut into pieces before treatment. As they are smoked fish lose weight in about the same proportion (3 kg fresh fish = 1 kg smoked fish) as when they are dried. One treatment is rarely sufficent, especially in more humid areas, consequently the fish have to be retreated at intervals of between a week and ten days to keep them in acceptable condition. The smoke-dried product is often stockpiled over a considerable period of time, especially in those temporary camps furthest from major centres of communication.

Figure 7.11   Different types of smoking oven: (A) circular oven (Haussa); (B) rectangular oven made of baked earth; (C) pit oven; (D) sections of rectangular and pit ovens (adapted from Blache, Miton and Stauch, 1962)

In some areas the lack of wood for smoking is causing problems for the fishery. In the Sahel, for instance, the availability of domestic firewood is becoming one of the factors limiting human occupation. Studies carried out in Mali by the Operation Peche, indicate that about 1 kg of wood is necessary to produce 1 kg of smoked fish, using traditional ovens. An improved oven is being introduced in the Central Delta of the Niger, which will reduce the demand to 0.5 kg of wood for every kg of smoked fish produced (Operation Peche, 1976). A similar type of oven is being used in the development of the Elephant marsh fishery in Malawi where it is also intended to plant copses of trees to supply them with wood; a copse of about 1 hectare is thought adequate to serve one oven (Tweddle et al., 1977).

Fish Meal

The only river fishery whose catch is used for making fish meal is the Prochilodus fishery of the Parana. According to Vidal (1967) these “Sabelerias” accounted for nearly 70 percent of the total inland catch of Argentina, although these have declined considerably from about 1965 onwards.

Fish Oil

Several species of fish are particularly rich in fats and are traditionally used for extracting oil. Oil is, for instance, a by-product of the “Sabelerias” fish meal factories. Brycinus leuciscus is exploited early in the fishing season in the Niger for its oil, and several species of the Mekong, principally Cirrhinus and Dangila are taken for the same purpose. The extraction process involves boiling for a certain time to release the oil which floats to the surface and may be skimmed off. Dangila spp. can produce up to 15 percent of their weight in oil (Chevey and Le Poulain, 1940).

Fermented Products

Fermentation of the cleaned and gutted fish in water for 12 hours is a common preliminary to salting in the Mekong, or to sun-drying in the Senegal river. In Asia a number of fermented products including fish pastes and sauces with a high salt content are also produced.

PROTECTION AGAINST INSECT INFESTATION

One of the major curses of the fishing industry in many parts of the world is the infestation of fish with insects. Moist fish is susceptible to be damaged by blowflies and their larvae in particular. A blow-fly larvae consumes the fish flesh so it is only when the fish is sufficiently dried that it becomes unattractive to the adult fly for egg laying.

Dried or smoked fish are subject to be attacked by beetles, mainly Dermestes spp. The level of losses due to Dermestes spp is directly related to the length of storage of the fish.

Prevention measures include the re-heating or re-smoking of the infested fish in a temperature above 50°. The use of insecticides is not advisable as there is no safe insecticide officially recognized yet that can be used on cured fish.

However, keeping the fish off the ground and improving the hygienic conditions of the processing and storage areas as well as their surroundings can significantly reduce the problem of insect infestation.

In West Africa early attempts to control Dermestes infestations included scorching, whereby during the hot smoking process the outer skin of the flesh was blackened and hardened to lessen the successful penetration of the flesh by the beetle. Other techniques have included packing the fish in bales wrapped with matting as soon as it is smoked, but such attempts have largely failed. The Operation Pêche in Mali has claimed a certain success by treating the fish with natural insecticides (Bioresmetrine) after drying or smoking but improved hygiene and heat treatment using a polythene tent and solar energy and simply keeping fish off the ground probably provide more satisfactory alternatives.

SPECIES CAUGHT

Because of the large number of species involved in river fisheries, detailed species lists serve little purpose. Analysis of catch data from most gears shows them to conform to the canonical distribution of species abundance already described. Distributions of this type predict that only a few species will be dominant in the catch of any gear, and a knowledge of these is essential, both for the management of the stocks and for establishing priorities for research.

LATIN AMERICA

The river fisheries of Latin America have concentrated mainly on the low water “Piracema” or “Subienda” migrations. The fisheries are still relatively undeveloped and have selected for species of large size and consumer appeal. Consequently catches are composed of the larger characins or siluroids. In the Magdalena river well over half the catch consists of Prochilodus reticulatus (Granados, 1975) although Pseudoplatystoma fasciatum, Pimelodus clarias, P. grosskopfii, Brycon moorei, Sorubim lima, Plagioscion surinamensis and Ageneiosus caucanus also contribute significantly. The preoccupation with the subienda species in Columbia led to a neglect of other potential food fishes and Bazigos et al. (1977) drew attention to an unexploited stock of Hemiancistrus and Pterygoplichthys in the cienagas which could substantially increase the yield from this system. Fish of the genus Prochilodus are the mainstay of other South American fisheries. P. reticulatus forms the basis of a heavy fishery in the southwestern portion of Lake Maracaibo and the inflowing floodplain river Catatumbo (Espinosa Giminez, 1974). In the Apure and Upper Orinoco systems, the fishery concentrates on the larger Pimelodidae: Pseudoplatyptoma fasciatum, P. tigrinum, Brachyplatystoma filamentosum and B. vaillantii (Canestri, 1972) but a greater diversity of genera including Prochilodontids are found in the Lower Orinoco and Delta (Novoa, 1982) (Table 7.2).

In the Parana, Vidal (1969) listed 18 species as being the principal ones of commercial value (Table 7.3). Further upriver, in the Mogi Guassu, P. scrofa makes up 0–60 percent of the catch which also contains 16 other species of characin (Godoy, 1975). Although the Amazon basin contains some 2 000 species, only a small proportion of these are captured by the fishery. Meschkat (1975) listed the major commercial species, based on the catch statistics for Amazonas state (Table 7.4).

The heavy and selective fishing pressure on the largest species A. gigas and C. bidens, had already led to a sharp decline in the populations of these fishes by 1975 and Bayley (1981) posit that the catch of all large species can not be maintained at present fishing pressure. In exchange the share of the catch contributed by species of such genera as Semaprochilodus, Prochilodus and Triportheus has, and will, increase until they dominate the fishery.

AFRICA

In Africa, the river fisheries are exploited at a much greater intensity and a very broad spectrum of species are caught, especially in the basins of the western side of the continent. In the Niger river gill net and cast net catches can contain over 50 species and it is difficult to identify the major elements of the catch. However, Raimondo (1975) listed the nine most important species of the Upper Niger (Table 7.5).

Other species commonly represented are Auchenoglanis occidentalis, Clarias anguillaris, and particularly Brycinus leuciscus which forms the basis of a specialized fishery for fish oil production.

Catches from other West African rivers have a similar combination of species. In the Senegal river, Reizer (1974) investigated the number of species captured as a function of mesh in gillnets (Fig. 8.7) and showed that the number of species increased as mesh size decreased. On the basis of these experimental fishings it appeared that the ten most important species to the fishery were Schilbe mystus, Lates niloticus, Alestes dentex and A. baremoze, Hydrocynus brevis, Labeo senegalensis, Eutropius niloticus, Citharinus citharus, Heterotis niloticus and Hepsetus odoe. In some systems, such is the diversity of gear that it is almost impossible to establish the true weighting of the various species in the total catch. In these instances, studies of the abundance of fish in the markets adjacent to the fishery give some idea. For instance, in the Oueme system the order of abundance of the various species was as shown in Table 7.6 although as many as 40 species figured in the fishery as a whole. This abundance is of course, biased by the food preferences of the fishermen themselves as some of the species caught rarely reach the market.

In the Chari-Logone system the elements of the catch are very difficult to separate from those produced in the lake. However, Blache and Miton (1962) listed the principal elements of the catch from a number of fishing methods in the Chari and Logone rivers. From these it appeared that the migratory Alestes dentex and A. baremoze were by far the most important species to the fishery. Several larger species were also of major importance including Citharinus citharus and C. latus, Distichodus rostratus and D. brevipinnis, Labeo senegalensis, Hydrocynus brevis and H. forskahlii, and Lates niloticus. Some smaller species were also important including Oreochromis galilaeus and O. niloticus, Schilbe mystus, Brycinus nurse and divers Synodontis spp. In the swamps Clarias lazera and C. anguillaris were particularly abundant. The drying of the Lake Chad basin induced by the Sahelian drought has, however, radically changed this composition and the catch is now dominated by tilapiine cichlids.

In contrast to the specific richness of West African fisheries, the rivers of East Africa produce only a restricted variety. The Shire river fishery was found to have five species, Clarias gariepinus, Clarias ngamensis, Oreochromis mossambicus, Marcusenius macrolepidotus and Eutropius depressirostris which made up about 90 percent of the catch (Willoughby and Tweddle, 1977) despite the fact that there are 39 species in the system. The Kafue fishery takes a greater number of species (18) but of these, 6 contributed about 90 percent of the catch (Everett, 1974). These were: Oreochromis andersoni, O. macrochir, T. rendalli, Serranochromis angusticeps, Schilbe mystus and Clarias gariepinus.

ASIA

In the rivers of the Indian sub-continent there is one migratory species, Hilsa ilisha, which, in the Indus, Ganges and Godavari systems, is the subject of specialized fisheries. Apart from this the fisheries are based largely on a series of large cyprinids known as the major carps, as well as some siluroids, ophicephalids and notopterids. The major Gangetic carps are Labeo rohita, L. calbasu, Catla Catla and Cirrhinus mrigala, and the principal siluroids Mystus aor, M. seenghala and Wallago attu. In addition to these, Jhingran (1975) listed a further 12 species which contribute significantly to the catch. A similar species complex occurred in the Brahmaputra river, with the addition of Labeo gonius, Puntius sarana and Notopterus notopterus. In the Cauvery river a somewhat different group of species dominated the fishery: Acrossocheilus hexagonolepis, Tor putitora, Barbus carnaticus, Labeo kontius, Cirrhinus cirrhosa and Osteocheilus brevidorsalis amongst the cyprinids, Glyptothorax madraspatanus, Mystus aor, M. seenghala, Pangasius pangasius, Wallago attu and Silonia silondia amongst the siluroids, together with Channa (= Ophicephalus) marulius a murrel and Notopterus notopterus.

Husain (1973) listed the 13 principal species of the lower Indus fishery as shown in Table 7.7 although these were drawn from a pool of 66 species. From the list it will be seen that there are many elements in common with the Indian rivers, although O. mossambicus has been introduced from Africa.

Fishing in the Chao Phrya river is largely done by stationary wing traps which block most of the main channel (Tongsanga and Kessunchai, 1966). Of the 77 species commonly captured, Crossocheilus reba made up about 60 percent of the catch. Other important species were Wallago attu, Macrognathus aculeatus, Ophicephalus micropeltes, Ophicephalus striatus, Puntius gonionotus, Pangasius sutchi and Cirrhinus microlepis.

More than 150 of the total 800 species that inhabit the Mekong make up the bulk of the catch in that system. Of these a few may be singled out as being particularly conspicuous or sought after. These vary with the region of the river fished, and separate authors have identified different dominant components. Pangasianodon gigas particularly is distinguished by its size but has diminised in importance due to overfishing. Adopting the summary of Fily and D'Aubenton (1966) the species listed in Table 7.8 made up over one percent of the catch in Cambodian waters.

EUROPE AND NORTH AMERICA

Catches in North American and European highland streams are both dominated by salmonid species but richer faunas are to be found in the lowland reaches. In Europe these are based mainly on cyprinids and of the 20 species listed from the middle of Danube by Liepolt (1972) and Holcik et al. (1981) over half belong to this family (Table 7.9). In the lower Danube, as with other Black Sea and Caspian rivers, migratory species, such as Huso huso, Acipenser ruthensis, A. stellatus, A. guldenstaedtii and Alosa pontica make up a considerable part of the catch.

The Vistula in Poland contains many less species although the fishery does have some of the elements present in the Danube. According to Backiel (1983) the following species appeared in order of importance: Abramis brama, Vimba vimba, Anquilla anquilla, Esox lucius, Stizostedion lucioperca, Chondrostomus nasus, Rutilus rutilus and Blicca bjoerkna. This relative poverty in species is exaggerated by environmental degradation which has already eliminated some of the previously most important anadromous fishes from the river.

The range of families appearing in the catches of North American lowland rivers is greater than in Europe and the diversity of the catch is equally high as is illustrated by the commercial and recreational catches of the Middle Mississippi (Rasmussen, 1979) (Table 7.10).

These examples show that even fisheries in rivers from the temperate zone can be based on a large number of species.

FISHERIES FOR JUVENILE FISH

The very heavy exploitation of juvenile fish, in the form of fish of the year moving to the dry season habitats at the end of the flood, is a particular feature of floodplain fisheries. In the Oueme, as in many African and Asian fisheries, small mesh nets of various types are used intensively in the canals draining the plain. Cross channel dams and barrages, such as those of the El Beid or the Barotse plain are also designed for the capture of young fish. Durand (1970) estimated that up to 90% of the catch by number and weight of the El Beid river was made up of juvenile fish moving from the Yaeres floodplain towards Lake Chad. The “maalelo” fishery of the Barotse removed about 3.7 percent of the juveniles of the 15 most important species each year (Bell-Cross, 1971). Regression analysis also indicates that considerable proportions of the catch from African rivers (ranging from 90% in the Shire and drought affected Niger to 50% in the Niger under more normal regimes) are repeatedly drawn from 0+ fish. Many millions of fingerlings and fry of the major carps are withdrawn annually from Indian rivers to stock reservoirs in their basins, and the capture of fry is common throughout Asia for the stocking of floodplain depressions, rice paddies and culture ponds.

It is common prejudice that the removal of large quantities of juvenile fish will prove harmful to the stock. The persistence of many of the fisheries themselves, indicates that there is little danger so long as the practice is kept within reasonable limits. Both Reed et al. (1967) in his defence of the “atalla” fishery for juveniles on the Niger river, and Bell-Cross 1971) in his analysis of the “maalelo” fishery, made the point that, with the high mortality rates current among river fishes, the loss of a proportion of year class 0+ fish is hardly liable to affect the final population at all. A theoretical analysis of such a fishery (Welcomme and Hagborg, 1977) indicated that a high proportion of the juveniles can be removed during the period of drain-off without damaging the fishery and in simulated fisheries, where juveniles were exploited at the same time as adults, the combined catch exceeded the maximum catch of either juvenile or adult fisheries on their own. Careful control of these fisheries is, however, essential and further studies are needed on actual situations where the juveniles are heavily exploited.

CATCH

ANALYSIS OF CATCH IN DIFFERENT RIVERS

Catch as a Function of River Form

Catch statistics from rivers are often of low quality because of the difficulties inherent in collecting data from fisheries which operate from many landings dispersed along a system that may traverse several countries. As in most fields of fisheries, there is a need to improve the quality and quantity of the catch statistics and the collection of biological and other data as a prerequisite for the proper management of the riverine stocks. However, because of the lack of concise information on many of the individual systems, attempts have been made to extrapolate general principles from the small group of water bodies about which something is known. Some approaches to this have already been discussed in the section which describes methods used to establish either standing stock, production or yield in temperate rivers from a wide variety of ecological parameters. The information needed for such excercises in tropical rivers and an alternative means of analysing catch patterns has been sought. Despite the inadequacies of the data, an analysis of fish yield patterns from African rivers has given a fairly coherent picture of the factors involved in determining the catch that can be expected from any particular system (Welcomme, 1976). In the rivers used in this analysis, see Table 7.11 all of which were moderately to heavily fished, there was a good correlation between the drainage basin area of the river system in km² (A) and the catch in tons obtained from it (C). Excluding catches from exceptionally large flooded areas, the sample conformed to the relationship:

C = 0.03A^0.97 (r = 0.91)

Because the basin areas and the total length of the longest channel of the river are also simply related.

(Main channel length = 4.95 basin area^0.45, in African rivers) this equation transformed into a relationship for yield in t as a function of the main channel length in km (L):

C = 0.0032L^1.98 (r = 0.90)

or approximately one three hundreth of the square of the length of the stream (L²/300).

The catch of any reach of river of length x km at distance y from its source can be calculated from _xC_y = C_y+x - C_y, where values of C_y can be obtained from the preceding equation. In its most extreme form, where x = 1 km, this yields a theoretical equation for the catch that might be expected for any kilometre or river at different distances from the source (i.e., catch at km_y = 0.0064y^0.95). When catches from rivers with extensive floodplains (those whose flood area exceeds 2% of the total basin) are also included in the analysis, a second relationship emerges whereby C = 0.44A^0.90. Values predicted by these relationships are of course averages over a number of systems from which actual values from individual systems deviate quite widely. Nevertheless these formulae have been used to predict catch in a number of systems lying outside the original set. The Mekong, Danube and Magdalena rivers, with floodplain areas of between 3 and 8 percent of their respective basins, are distributed around the extensive floodplain line. Rivers such as the Indus, Mogi Guassu and Pongolo with no extraordinary development of their floodplains fit the ‘normal’ relationship nearly exactly. Furthermore some independant estimates of potential, such as that obtained by Paiva (1973) from the Parnaiba river, Brazil, of 7370 t/yr for a basin area of 362 000 km², agree completely with the predictions of the formulae. However, caution should be exercised in their application of rivers outside Africa owing to the different relationships between basin area and channel length for the various continents.

Deviations from the mean regression line may arise for two main natural reasons. First, there is a possibility that edaphic factors may influence the basic productivity of a system in such a manner as to determine yield. There is so far no strong evidence for this in the data set available on savanna rivers, although there is more than a strong suspicion that certain types of poor river, such as the tropical black waters, are indeed less capable of supporting sustained fishing than is normal. Second, morphological factors have been shown to determine the yield characteristics of river systems. The very fact that it has proved necessary to derive two different relationships for yield against basin area indicates the predominant role of floodplain area in determining the productivity of a system. This factor alone accounts for 70% of the differences between actual and predicted catch in different reaches of the same river. From the data in Tables 7.12 and 7.13 mean catch levels of 17 fully exploited African floodplains were 54.7 ± 36.5 kg/ha/yr. Six Asian floodplains had mean catches of 44.02 ± 17.9 kg/ha/yr and, given the means and standard deviations of the two populations they can be considered one set which can be merged. When data from the Orinoco and Magdalena rivers, which also lie within the limits of confidence, are added, the mean yield for the 25 rivers is 51.55 ± 32.1 kg/ha/yr. With the present data it is not possible to detect any differences between the tropical continents but at 18.7 kg/ha catches from the temperate Danube are slightly lower. However, the catches from 1481 km of the Upper Mississippi river (Rasmussen, 1979) show that the 171 136 ha of pooled river, produced 6499 t of fish (mean of 25 yrs) equivalent to 37.8 kg/ha. In this section of the river there is a succession of 26 pools in which the lower ends have lacustrine characteristics and the upstream portions have retained their floodplain. Catches obtained from the three blackwater rivers for which data are available indicate catch levels of between 11 and 19 kg/ha/yr which are considerably less than the mean for savanna rivers. When catch is plotted against floodplain area for the 25 tropical floodplains which are exploited at a reasonably intense level, a relationship:

C = 4.23A^1.005

is obtained. Although the best fit is a power curve the exponent is sufficiently close to 1 as to make the relationship almost linear. It does, however, indicate that large floodplains may be marginally more productive per unit area than smaller ones, and values range from 42.8 kg/ha for a 1 hectare plain to 44.6 kg/ha for a 5 million hectare plain. These figures all confirm the earlier estimates of between 40 and 60 kg/ha/yr that may be expected from tropical floodplains.

One question that frequently arises is the degree to which the catch of reservoir rivers or modified rivers which have no annual flood, compare to the levels of production of flood rivers. The only two fully fished rivers of this type for which data are available, the Mississippi and the Nile, lie within the same range of values appropriate to their level of exploitation as do flood rivers. More data are needed, however, to clarify this as the dynamic processes underlying the fisheries of such rivers are likely to differ considerably from those in rivers having a pronounced seasonal flood.

Catch as a Function of Fishing Intensity

Standard models relating catch and effort, such as those of Schaefer and Fox, have been used successfully in some fisheries of small rivers or in the analysis of fisheries based on single species, for example those for Colossoma (Petrere, 1983) or Plagioscion spp. (Annibal, 1983) both of the Amazon basin. Examination of catch and effort data in large rivers, however, generally illustrates the difficulties of applying such models to complex fisheries which are pursued with a variety of methods on the whole fish community in any river or river reach. Because of the scarcity and poor quality of most catch and effort data from river fisheries any analysis of the factors regulating catch must be regarded as somewhat speculative. Two approaches can be adopted for the analysis of the data available. First, data from individual rivers can be studied to identify any relationships arising from changes in temporal fishing patterns. Second, data from a number of rivers can be grouped for treatment as one set with a range of differing regimes. Data sets from individual rivers are rare and only two sets, one from the Mississippi and one from the Nile, have so far been traced. Similar power curves fit the data for catch per boat for 30 years from the Mississippi:

(CPUE = 29731E^-1.24: data from Rasmussen, 1979)

and for catch per boat for 10 years from the Nile:

(CPUE = 11117E^-1.03: data from Borhan, 1981) (Fig. 7.12).

Curves of this type have been described previously for individual fisheries, for instance by Beverton (1959) who used one to describe the response of Tilapia to increased fishing in Lake Victoria. When 22 observations from 16 of the worlds rivers are compared (Fig. 7.13, Table 7.13) the single best fit regression line is also an inverse power curve of the form CPUE = a E-b although the value of the relationship is limited by a lack of points in the middle range (5 – 20 fishermen/ha). The inclusion of data from various continents and climatic zones appears justified at this time as the points for the Magdalena River, Ganges and Misissippi all lie well within the confidence limits of the African data. The relationship between effort and catch per unit effort is perhaps better described by a curve within the log-log plot which may be interpreted as two different regression lines calculated from all points between 0.05 and 3 fishermen/km2, where CPUE = 2.72E^-0.47, and between 3 and 30 fishermen/km2, where CPUE = 10.2E^-1.09. Separate plots of total catch against catch per unit effort extrapolated from the two regression lines cross over at about 8 fishermen/km2.

The form of these relationships may be explained in three ways:

(a)   That there is an interference effect between fishermen as densities increase, whereby gears compete with one another for an increasingly limited resource. This would mean that fishing mortality is no longer linearly related to nominal effort.

(b)   That the measures of effort customarily adopted for ease of recording, i.e., numbers of fishermen or numbers of canoes, is unsatisfactory. This may be particularly significant in artisanal and subsistence river fisheries where fishermen have alternative occupations. They may then allocate the time spent on fishing according to its rewards relative to such ocupations, in which case fishing mortality is again no longer linearly dependent on the variable selected to represent effort. Here some other measure, for instance fisherman/days, might be a more appropriate index. Such measures demand an increased effort in surveying the fishery which exceeds the financial possibilities of most developing countries. In many river systems the switch between farming and fishing is in any case dictated by the exigencies of the agricultural cycle, which may leave unstructured time which can only be filled by fishing. Further verification of these questions rests upon detailed socio-economic studies in such communities but the possibility signals the need for caution in adopting such measures of effort for management.

(c)  There is a real effect at the fish community level, whereby the yield curve concept as applied to individual species is not applicable. Some of the changes occurring in exploited fish communities described below indicate that such an effect does occur.

Figure 7.12   Catch per unit effort as a function of effort in (A) the Nile and (B) the Mississippi

Figure 7.13   Catch per unit effort as a function of effort for 17 rivers: Solid line = regression for all points; dashed lines = best fit lines for catches from 0.05–3 fishermen/km² and 3–30 fishermen/km²

The form of these relationships also implies that the relationship between effort and total catch shows an initial rise followed by a somewhat flat line during which there is little change in catch over a considerable range of effort (Fig. 7.14). The Nile downstream of Aswan, for instance, produced 8 410 ± 542 t for ten years despite a threefold increase in effort. Plainly such a plateau can be extended indefinitely neither to the right nor the left of the relationship, and it is suggested that curves of catch and catch per unit effort have three phases as set out in Fig. 7.15.

Figure 7.14   Plots of total catch (---•---) against effort for the Nile. Also shown is catch per unit effort against effort (---o---)

Figure 7.15   Theoretical diagram of changes in catch and catch per unit effort as a function of effort, letters explained in text

(1)  Initiation phase: Limitations on catch per unit effort, and hence total catch, are set independently off the relationship by the physical capacity of the fisherman to handle his catch. Observations from all over the world on both lacustrine and riverine fisheries indicate that this capacity rarely exceeds 1 – 4 t per year per fishing unit although exceptionally up to 13 t/yr may be handled. The attainment of maximal levels at low fisherman densities depends much on the state of development of the society in which the fishery is pursued. Thus, whereas in advanced economies sufficient support exists to permit the fisherman to land and market large catches, in developing economies individual efficiency may be poor for lack of efficient gear and the absence of infrastructure for landing, marketing and transportation of the catch. Fishing therefore tends to remain a subsistance activity. As the numbers of fishermen increase the creation of markets will give rise to a rapid expansion of the fishery.

(2)   Sustained exploitation phase: During this phase, catch levels are maintained in the face of increasing effort. At the same time changes occur in the composition and abundance of the exploited fish community which lead to a progressive shift in catch composition from large to small species.

(3)   Collapse phase: Few cases have been documented where the fishery has been intensified to the point of total collapse of the fish community. It is to be supposed that the fishery becomes uneconomic long before and most reported overfishing has occurred where larger and highly favoured species have declined while the rest of the fish community remains intact or underexploited.

It is conceivable that real collapse at the community level can occur under exceptionally heavy exploitation. This may be impelled by socio-economic factors, as in the case of the Oueme, where the disappearance of nearly all the larger species, the very small size at capture and the loss of catch over the decade 1957–68 would suggest a community at the limits of its tolerance. The exceptionally high fisherman densities of 30/km2 in this river result from a populous ethnic group which is confined to a floodplain area by demographic pressures. Other drastic declines in fisheries have been recorded in the African Sahel where the failure of the floods over a number of years has resulted in a decrease in natural productivity in rivers such as the Niger, Senegal and Chari/Logone. The fish stocks, thus diminished, have been called upon to support an increased fishing pressure as other food sources have disappeared.

Changes in Community Structure with Increasing Fishing Pressure

The form of the relationship between catch and effort, whereby there is an initial rise followed by a more or less prolonged plateau where catch remains stable in the face of increasing effort, masks a series of adjustments in the fish community. Unfortunately systematic monitoring is still not carried out in most river fisheries. As a consequence theories on the impacts of fisheries in rivers with complex multi-species stocks must be derived from a number of river and lake fisheries. Experience has shown that a similar succession of events, which has come to be termed the fishing up process, occurs as fishing pressure is applied and increased. It is a common observation that there is a progressive disappearance of the larger species as pressure is applied. Similarly, increases in fishing intensity and total catch were correlated with a reduction both in mean size and as a proportion of catch of the large pimelodid catfishes of the Orinoco river (Novoa, 1982), whereas the proportion of smaller Semaprochilodus in the catch increased to more than compensate for the deficit. In the Amazon at Manaus larger species such as Arapaima gigas and Colossoma bidens have progressively disappeared from the catch and have been replaced by increased quantities of smaller prochilodontids and clupeids as well as new species such as the pimelodid catfishes. Lates niloticus, Heterobranchus longifilis and Bagrus docmac have completely disappeared from the heavily fished lower reaches of the Oueme river, and other large species such as Citherinus latus, Distochodus rostratus, Labeo senegalensis and Heterotis niloticus diminished in abundance to be replaced by small catfishes, characins and mormyrids. Sritingsook and Yoovetwatana (1977) remarked on the decline of the larger migratory species such as Probarbus jullieni from the Mekong river and its tributaries.

It is useful to examine the history of the Oueme fishery in more detail as one of the best examples of community overfishing in rivers. The total catch of the Oueme floodplain was estimated at 10 400 t in 1955–57 (CTFT, 1957). By 1968–69 this had fallen to an estimated 6 484 t (FAO/UN, 1971). The fall was accompanied not only by the shifts in species composition noted above, but by a fall in catch per unit effort in most gears as summarized in Table 7.14. There was some increase in total effort during the period of study (about 19%) but this was insufficient to account for the drop in CPUE. Larger mesh fishing gears were eliminated from the fishery by 1968 and the mean weight of the species caught decreased in most of the gears that remained in use. For example, the medium-mesh cast net caught fish of 124.9 g in 1955–57 but the mean weight had dropped to 26.5 g by 1968–69. Similarly the small-mesh net caught fish of 25.2 g in 1955–57 but only 9.8 g in 1968–69. The only gear in which catches increased was a trap for Macrobrachium, a detritus eating freshwater prawn that was previously heavily preyed upon by the fishes.

Other changes have been noted from a number of systems. In Lake Tanganyika, for instance, increased fishing led to the elimination of the major predator (Lates) with a considerable increase in total catch from the small pelagic clupeids and smaller predator (Luciolates) which form the basis of the fishery (Coulter, 1970). In Lake Victoria the fishery was being pushed from a dominance of large tilapiine cichlids towards a preponderance of small haplochromine cichlids and even smaller Rastrineobola before the spread of introduced Nile perch reversed the trend. Further changes occurred within the haplochromine population under fishing pressure from trawls with the disappearance of larger piscivorous and molluscivorous species and a progressive reduction in size in the remaining planktivores and limnivores (Witte and Goudswaard, 1984). The establishment of a trawl fishery in Lake Malawi also resulted in the successive displacement of the species caught to smaller and smaller lengths (Turner, 1981). Certain types of environmental pressure would appear to mimic or to reinforce the effects of increased fishing pressure, which makes it difficult to fully interpret the cause of observed changes. Thus in the Caspian Sea there has been a long standing tendency toward smaller species in the fishery in the face of sustained effort and environmental stress although the level of catch has remained the same (Carre, 1978). Similarly, Regier and Loftus (1972) were able to trace a succession of alterations in the composition of the fish communities of the N. American Great Lakes arising from a combination of fishing, ecological and environmental pressures.

From this evidence, it may be supposed that individual species within the community pass through a typical production curve of catch against effort. The succession of such curves agrees with the observations that exploitation eliminates first the larger individuals of the larger species then the larger species themselves (a and b in Fig. 7.16). This selective disappearance tends to be due to a preference, both by fishermen and consumers, for larger individuals, as well as the lower capacity of the long-lived larger species to support high levels of fishing mortality. In an effort to sustain catch the fishing methods tend to be replaced by those which capture smaller fish. As the few largest species are removed from the fishery larger numbers of intermediate-size species (c-j in Fig. 7.16). In other words, the community is pushed successively from K dominance to r dominance. A feature of r selected species with short life spans, high fecundity and high productivity is that they fluctuate considerably in abundance. It is, therefore, reasonable to suppose that as a fishery becomes more and more dependent on only a few such species it may become destabilized with large fluctuations in catch with time. Indeed, trajectories of this sort are characteristic of fisheries for pelagic clupeids throughout the world.

Figure 7.16   Theoretical changes in a fish community when subjected to increasing fishing pressure: (A) of certain population and fishery parameters; (B) of total catch showing schematic evolution of individual species ‘a’ through ‘o’

Effects on species dominance may also be attributable to environmental disturbances outside the fishery, or to other changes within the fish community, such as the introduction of a new element to the fauna. In some cases, native or introduced species which have been held at a low level in the community are able to expand to occupy a dominant position when indiginous species are overfished or when the habitat is changed in their favour by human action.

In river communities where there is as paucity of species at lower trophic levels, as shown by high Predator/Forage fish ratios, the capacity of the system to benefit by such changes is limited. Similar limitations have been observed in nutrient poor systems such as tropical blackwaters where the amount of nutrients accumulated within the fish community represents a significant proportion of the total nutrient pool. Here removal of the larger predators, whether fish or crocodiles, has even been held accountable for the declining long term productivity of the environment (Fittkau, 1973). In such systems rapid declines in catch may follow the onset of exploitation as happened in the Kamulondo depression, where catch fell from 9 063 to 4 810 t in the space of four years after the introduction of an intensive fishery based on nylon gillnets (Poll and Renson, 1948). More recently similar rapid declines on individual fishing grounds have been observed in the blackwater tributaries of the Amazon. Eventually even in the most productive of systems the capacity of the fish community to absorb further increases in exploitation pressure will be exceeded, recruitment will fail and numbers and biomass will drop sharply.

During the period when yield changes relatively little with increasing exploitation, the ichthyomass will tend to remain relatively constant or to decrease as the species composition favours those with higher turnover rates. At the same time biological fish production will increase. The tendency for ichthyomass to remain constant may be reinforced in flood rivers where only a certain proportion of it is able to survive through the dry season anyway. The number of species which make up the catch will initially increase, although in some regions market preferences may be restricted so that only a few stocks may be fished. When these are reduced, it may take some time before the market can be reoriented to replacement species. Until this happens catches may fall and fishing be reduced, as happened in the Parana river with the decline of the ‘sabalo’ (Prochilodus platensis) fisheries and in temperate rivers where salmonid stocks were reduced by eutrophication. Where the market demand is highly diverse, as it is in most African and Asian countries, the number of species in the catch will first rise rapidly and then diminish as the more susceptile forms succumb to fishing and other pressures.

These changes are most noticeable in fisheries using mesh selective gears, such as gill nets, cast nets or seine nets, where changes in the size and species composition of the stock are accompanied by the classic pattern of the successive replacement of large mesh gear by smaller and smaller mesh sizes. Such changes are more difficult to detect in traditional fisheries which from the outset exploit a broad spectrum of species and sizes with a variety of gear. The available evidence from rivers such as the Oueme indicates that the use of gears designed to capture larger species and individuals progressively declines as the fishery becomes heavily exploited.

FLUCTUATION IN CATCH BETWEEN YEARS

The available evidence indicates that the year-to-year variations in fish stock abundance are directly linked to the degree of variability of the hydrological regime. Thus in stable systems such as reservoir rivers or those with flood control the magnitude of the standing stock varies little from year to year, whereas it is apparent that the catches from flood systems fluctuate in a manner that is in some way dependent on changes in the flood cycle. Such changes in yield were noted early in the study of river fisheries when Antipa (1910) concluded that the fisheries production of the Danube delta was directly proportional to the extent and duration of the floods. This sentiment was echoed by subsequent workers on that river (Botnariuc, 1968; Holcik and Bastl, 1977). That a similar state of affairs existed in tropical rivers was suspected by many workers including Wimpenny (1943) in the Nile prior to its complete control.

Models of production as a function of water regime would suggest that year-to-year variation in catch arises from two main sources. First, the amount of water in the system at low water can affect catches of the same year by altering the ease with which fish are captured or can influence catches in succeeding years through dry season mortality. Second, the intensity of the flood determines the magnitude of the stock in future years through differences in recruitment, survival and growth. In certain rivers it can also alter the ease of access between the channel and the floodplain water bodies thus altering catch within the same year. In combination these effects render analysis difficult and obscure trends over time but sufficient examples of each exist to illustrate these various cases.

Improved catches during exceptionally low waters are well known from most fisheries as the shallower waters favour the capture of fish. Usually such comments take the form of qualitative statements such as those of Vidy (1983) when describing the Logomatia fishery of the Yaeres floodplain. The relationship has been quantified in the Magdalena river as:

Catch _y(t) = 171779.36 - 23706.17 Water Level _y

for the whole fishery (Arias pers.comm.) and by Annibal (1983) as:

Catch _y(kg) = 56937 DDF_y - 164206: (r = 0.84)

for the Plagioscion fishery of the Lago do Rei of the Amazon basin, where DDF is an index of the amount of water remaining in the system during the dry season.

The amount of water remaining in the system during the dry season may also influence the catches in subsequent years as was shown by the University of Michigan et al. (1971) for the Kafue Flats fishery. Here the relationship:

C_y = - 6630 + log_eDDF_n-1: (r = 0.77)

which was originally formulated for the years 1956–71, has continued valid through 1983 (Hayward, 1984) despite the construction of two sets of flow control dams in the intervening period.

Sufficient data are available from the Kafue and Shire floodplains and from the Central Delta of the Niger for an assessment of the effects of the intensity of the flood on fish catch. Unfortunately none of these sets of data include reliable estimates of changes in fishing effort, which may themselves be responsible for some year-to-year variations in catch.

In the Kafue river records are available from 1954 until the present. However, the fishery was not judged to have reached maximum expansion until 1958 (Muncy, 1978), and the flood regime was changed following the closure of the Kafue Gorge dam in 1972. This was followed by the construction of an upstream dam at Itezhitezhi and the two dams, acting together substantially altered the duration but not the periodicity of flooding. Calculations based on the years 1958–71 gave good correlations between catch in year y and the flood of the preceding year (y - 1) (r = 0.72) or the flood two years previously (y - 2) (r = 0.71). Muncy (1973 and 1978) presented a complete analysis of the various factors also finding good positive correlations between catch in year y and flood intensity in the preceding years (y - 1 and y - 2). Although detailed analyses were not continued after 1971 the predictions of these regressions are consistent with the greater catches that have been obtained because of the increased flooding following the installation of the two dams.

Sets of data for water level and catch are also available for the Shire river (1969–73) and the Central Delta of the Niger (1966–74) which were analysed by Welcomme (1975). This analysis used a simple sum of all weekly mean water levels which exceeded the bankfull stage. The results showed a highly significant correlation to exist between catch in year y and the flood regime in y - 1 in all three systems. Correlations of catch with flooding in the same year were not so satisfactory. Because the fisheries of most tropical rivers are based on fish that are one or two years old, it might be expected that the flood regime in the both preceding years might exert an effect on catch of any year. When the regressions of combined and weighted hydrological indices from both receding years were tested some improvements in correlation were noted. As a result it was finally concluded that catch in year y is best explained by a combination of the flood histories of the two preceding years. Thus the best fit linear regression lines which are plotted in Fig. 7.17 were as follows:

Kafue: C_y = 2962 + 70.54 (0.7 HI_y-1 + 0.3 HI_y-2)
Shire: C_y = 5857 + 38.11 (0.9 HI_y-1 + 0.1 HI_y-2)
Niger: C_y = 3239 + 32.10 (0.5 HI_y-1 + 0.5 HI_y-2)

Differences in Hydrological Index accounted for 57 percent of the variation in catch between years in the Kafue, 82 percent in the Shire and 92 percent in the Niger. Subsequently the Niger fishery has continued to show a strong correlation with flood regime and a data set for the 19 years from 1966 to 1984 yields a linear correlation:

C_y = 19.172 + 0.027 (0.7 HI_y-1 + 0.3 HI_y-2) (R² = 0.84)

Figure 7.17   Best-fit regression lines for the relationship between catch and flood regime for: (A) the Kafue Flats; (B) the Shire river; (C) the Central Delta of the Niger at Mopti

which HI is calculated from the mean six monthly flood discharge on the Kaulikoro gauge. A linea relationship was not at best felt, however, and an improvement of correlation to R² = 0.87 was obtained with the regression:

C_y = 151.73 log (0.7 HI_y-1 + 0.3 HI^y-2) - 428.26

The log-normal nature of this regression indicate a slight lessening of the response of fish communities as flood densities increase over a certain level. Other rivers of the Sahel also showed decline in catch over its period of the drought (Table 7.15)

There is considerable interest in identifying the relative role of the two components of the hydrological regime in determining productivity although it is sometimes difficult to separate them statistically. For example, where the intensity of flooding is highly correlated with the severity of drawdown, as in the Kafue (r = -0.78), it is difficult to determine whether the apparent dependence of catches on the preceding drawdown is a statistical artefact. Some indication of the relative roles might be deduced the Shire river, which shows a lesser correlation between drawdown and flood (r = -0.45) and also shows a lower correlation between drawdown and catch. However, this should not be taken to imply that the high water phase is the most influential of the two phases in all systems. The difference in response could equally be due to differences in the amount of water remaining in the sytem at low water relative to the amount at peak flood. In the Kafue the dry season area was about 27 percent of that during the floods, whereas in the Shire the equivalent figure was 48 percent. It may be surmised that the more stringent the drawdown, as reflected by the lessened percentage of residual water during the dry season, the greater the influence of the low water regime on catch in the next year. This is supported by the predictions of Welcomme and Hagborg's (1977) simulation of a floodplain and it's fishery which predicted that the more water remaining in a system at low water, the more the differences induced by variations in the high water regime are transmitted to subsequent years.

Analysis of catch relative to the flood history of a river has been carried out independently by several workers. Catches in the Orinoco gave two relationships depending on species (Novoa, 1982). Thus the total catch over 15 years was:

C_y = -634.8 + 0.53 (0.5 HI_y + 0.5 HI_y-1) : (r=0.62)

and the catch of Prochilodus mariae over 10 years was:

C_y = -389.4 + 0.19 (0.5 HI_y + 0.5 HI_y-1):(r=0.73)

In this river a second relationship beween catch and flooding was noted for Semaprochilodus laticeps where, because a greater number of water bodies was in contact with the river channel in higher floods, a logistic relationship was found between an index of flood intensity (HI) and catch during the same year:

Catch was also found to be related to flood height in the Madeira river by Goulding (1981). Here catch at low water was improved when low water flows were less than average. Heavier floods are reported by the fishermen to have the effect of releasing fish imprisoned within floodplain water bodies and increasing the recruitment, the fishery thus leading to better catches in the source as subsequent years.

Flood strength has also been found to influence catches in temperate rivers. Ivanov (quoted by Chitravadivelu, 1974) also noted similar effects in the Zofin arm of the Danube, but with a lag time of one year.

Because of the number of factors involved, for instance changes in fishing effort or the reduction in the age and length of first capture associated with the fishing up process, it is frequently difficult to interpret these relationships by direct regression analysis, although visual comparison of plots of flood regime and catch indicate some relationship to exist. Krykhtin (1975), for example, obtained excellent relationships between flood strength and catch in the Amur River where the best correlation was obtained with the flood regime three to four years previously. Dunn (1982) provided graphical evidence for a correlation between an index of catch in year y and of the flood five years previously for the Hilsa fishery of Bangladesh. Hilsa, which is a marine anadrome, is thought to enter the fishery in its 4th or 5th years. In analysing these results fluctuations between 1966 and 1976 are best explained by a combination of y-5 and y-6 hydrological indices (r=0.76) but between 1976 and 1981 there is little clear correlation. Similarly, graphical inspection of Rasmussen's (1979) catch data as a function of the Upper Mississippi floods shows there to be parallel trends but this data set is less consistent and did not yield a conclusive regression.

The lag between the year of flooding and the time when its effects are reflected in the catch are probably dependent on the time taken for fish to grow to the size range captured by the fishery. In tropical rivers this is very short, often less than a year, because of the small size and rapid growth of many of the species caught, and also because of the heavy fishing for fish of the year as they leave the floodplain. An example of this was provided by Benech and Quensiere (1984) in their studies of the fishery of the El Beid river which drains the Yaeres of Cameroun. Here there was a strong positive correlation between the quantity of juvenile fish moving down river and the floods of the same year. The same study showed the level of production to be independant of the species composition when a regular sequence of floods exists. Thus in the El Beid the relation of production to the intensity of floods compared well between a series of readings from 1968 and another series between 1974 and 1978, although the population structure was very different. By contrast, when the regular series of floods is disturbed, aberrant patterns may be detected in the following years. Ivanov also remarked on the high proportion of young year classes in the Danube fishery one year after high floods and a corresponding drop in catch a year after particularly poor flow. In some rivers where growth is slower and the larger species are favoured by the fishery, fish may take two or more years before they are susceptible to the gear. In the Amur river, for instance, Krykhtin proposed that the effect on catch is only felt after the incorporation of 20–30 percent of the new year class into the target stock.

The fisheries of some reservoirs with highly fluctuating water levels seem to behave like floodplains. In the Parakrama Sumadra of Sri Lanka a 20 year study by Scheimer (1983) has shown the effects of fluctuating water levels on fish catch to be described by a relationship:

C_y = 232.2 + 16.2 HI_y-3

In the lake fish take from 2–3 years to enter the fishery. This data together with that of Moreau (1980 and 1982) indicate that reservoirs and shallow lakes with highly fluctuating water levels are ecologically homologous to floodplains.

C_y = 232.2 + 16.2 HI_y-3

On the basis of the obvious correlation between the intensity of flooding and catch it is tempting to try to predict catches in future years from the flood of the year using regression formulae similar to the ones above. To test this it may be assumed that if one can indeed predict the catch in any one year from its flood or the flood of preceding years the accuracy of the prediction would improve as an increasing number of years of data are added to the regression. When the data for the Niger were treated in this way the accuracy of forecast did improve, from which it was concluded that it is possible to validly predict catches in river systems from regression analyses of the past performance of the fishery provided enough terms are available. In the case of the Niger, at least 14 years data were needed to accurately predict future trends, such predictions assume that there are no major changes to the resources base through overfishing, environmental change, etc. Likewise, correlations between water levels in years y and y+1 and catch in the Czechoslovak reaches of the Danube, were used by Holcik and Bastl (1977) to predict future catches from hydrological flow data. Because of the importance of such predictions both for the regulation of the fishery and for the establishment of criteria for mitigation of falling catches due to upstream flood control structures further work on this topic is desirable.