The best understanding of a fishery is obtained through the adoption of a two-pronged approach, one part of which concentrates on the fishery itself and the other which addresses the fish community. Because resource assessments can be lengthy and expensive, requiring large amounts of trained manpower, it is rarely possible to carry out detailed studies except in the most exceptional cases. Thus, simple, quick but “dirty” methods for evaluation have also been developed. This chapter will thus discuss the methods available for assessment of fish stocks under three headings:
(i) Rapid methods for assessing fishery potential
(ii) Assessment of the fishery
(iii) Assessment of the fish stock.
Assessments of river fisheries are required for a number of purposes, principal among which are:
(a) To make an inventory of the fisheries resources of a defined geographic area, such as a country or river basin, usually for economic planning at an overall level. Here, rapid assessment methods frequently suffice, although in certain cases information on the state of the fishery is also desired.
(b) To better develop and manage the resources on which the fishery is based and to answer logistic questions about how and where to deploy limited resources. This requires a certain preliminary assessment of the level of exploitation of the fishery relative to its potential by a simple method. Thereafter, more detailed studies depend on the findings of the preliminary survey. In fisheries which are under-exploited further surveys are needed to identify the constraints, usually socio-economic, to their further development. In fisheries which are approaching full exploitation details of the biological and ecological constraint to the resource are needed for effective management and monitoring.
(c) To evaluate impacts of alteration of the environment by natural or man-provoked processes. Here interest centres on loss of benefit which often requires considerable supplementary socio-economic information.
Sampling large river systems is known to be extremely difficult throughout the world, so many methods have been developed or modified from other environments for the assessment of fish stocks. The following presents a selection of those which have been used more or less successfully in rivers. Some of these methods may be directly applicable to sets of river systems other than those for which they were developed, others, while working well in their original environment, serve to illustrate types of approach.
Assessments made by methods in this category are usually rapidly arrived at and cheap, but sacrifice a good deal of accuracy to attain these ends. They are used for such purposes as inventories of large and diverse resources whose abundance would prohibit detailed individual study, for gross economic projections of the value of a resource, to evaluate preliminary needs for development, and for the quick assessment of environmental impacts. These methods assume that rivers, or similar sections of rivers, can be grouped into sets whose fish populations behave in similar ways. Although they are simple, they rely on data accrued over many years from a number of individual systems. Taking the information available for the previously studied members of any set, parameters such as biomass, productivity or potential catch may be extrapolated for unstudied members of the same set. Some care has to be exercised in the interpretation of modes based on this procedure as they lack the statistical validity of formally designed data collection.
During their studies of the fishery resources of the rivers of Belgium and Northern France, Leger and later Huet were able to formulate some valuable generalizations as to the zonification of the running waters of temperate Europe. Arising from this, Huet (1949 and 1964) proposed a simple model to assess the approximate ichthyomass available for exploitation of those waters belonging to the set of temperate European rivers. The basic formula for this method is:
K = BLk
Where K represents the annual productivity (or harvest) of the water in kilogrammes per kilometre of river, L represents the average width of the river in metres, B is the “biogenic capacity” and k is the coefficient of productivity.
Values of B are as follows:
1 – 3 Waters with little fish food
4 – 6 Waters with average amounts of food
7 – 10 Waters that are particularly rich.
The coefficient K is the product of k1 + k2 + k3 where k1 is an expression of annual average temperature whose value is determined as follows:
|Average annual temp. °C||7||10||16||22||28|
k2 depends on acidity or alkalinity of waters and its possible values are:
|k2 for non-calcareous waters||=||1.0|
|k2 for calcareous waters||=||1.5|
k3 summarizes the type of fish population with the following values:
|k3 value for cold water (rheophilic) species||=||1.0|
|k3 value for mixed communities||=||1.5|
|k3 value for warm water (limnophilic) species||=||2.0|
The original method was modified by Lassleben (1977) for the assessment of fish catch in the German stretch of the Danube (Kolbing, 1978). Here fish catch (C) for a section of river is determined as:
C = BK 10 kg ha-1 yr-1
where B is the above and the various components of K are as above, except for k3 whose values are determined as follows:
Assuming a sample catch from a section of river containing species (or species groups) A and B with different k3 values k3A and k3B in proportions PA of species A and PB of species B, then the percentage of the area of the region occupied by species A will be:
The total k3 value (k3AB) can then be calculated as follows:
This method has given good results in north temperate streams and extended to larger rivers such as the Danube using the following modifications (Holcik, 1979).
The extrapolation of k1 (t°C) can be made according to the equation:
k1 = -0.6671 + 0.1667T°C
The simple percentage of rheophilic and limnophilic species is sufficient to establish k3 if all species present are lumped into two groups according to whether they are rheophilic (PA) or limnophilic (PB) thus:
but as PA + PB = 1
The biogenic capacity B of the river may be assessed using the biomass of benthic invertebrates instead of the quantity of aquatic vegetation. According to Albrecht (1953 and 1959) streams with a biomass of zoobenthos less than 60 kg/ha can be considered poor, those with a biomass from 60 to 300 kg/ha medium and streams with 300–700 kg/ha good. From these values the following regression equations:
(i) B = 0.00 + 0.05 Bb
(ii) B = 0.35158 + 0.45469 log Bb
have been obtained to calculate B where values of benthic biomass are: (i) 60 kg/ha or less and (ii) 60–700 kg/ha.
Binns and Eiserman (1979) follow an essentially similar approach for developing a Habitat Quality Index (HQI) to predict trout standing crops in streams in Wyoming. They derived two models using a range of parameters which were moderately easily estimated. Their most successful model was the expression:
log10 Y + 1 = (-0.903) + (0.807) log10 (X1 + 1) + (0.877) log10 (X2 + 1) + (1.233)
log10 (X3 + 1) + (0.631) log10 (F + 1) + (0.182) log10 (S + 1) 1.12085
|Y||=||Predicted trout standing crop|
|X2||=||Annual stream flow variation|
|X3||=||Maximum summer stream temperature|
|F||=||Food index consisting of X3 (X4) (X9) (X10)|
|X4||=||Nitrate nitrogen (mg/1)|
|X10||=||Water velocity m³/second|
|S||=||Shelter index consisting of X7 (X8) (X11)|
|X8||=||Eroding stream banks|
The various parameters are rated as in Table 1.
This model, which works well for the trout streams for which it was developed, illustrates the complexity that such models can attain. Their use is based on a thorough understanding of the ecological requirements of a limited number of high value species as well as the availability of detailed basic information on a number of parameters.
Methods for assessment of fish stocks or catch which involve indices based on a number of factors are of little application to larger tropical rivers. These ecosystems are generally very poorly understood and there is a great lack of data on most aspects of the fishery and the fish community. Furthermore, the systems concerned are highly complex, housing equally complex fish faunas. In the search for the simplest of parameters, Welcomme (1976) sought to directly relate fish catch with simple geographical and morphological features of the rivers. The simplest of these is the relationship between catch and river length, where, for the set of African rivers:
Catch = 0.0033 river length1.9539
or more approximately,
This relationship adequately describes the catch from rivers with normal floodplain development. It may be further modified by applying a correction for the conductivity of the water, but the effects of variation of this parameter on river fish populations are insufficiently understood for such to be applied with confidence.
In rivers where the floodplains are highly developed, such as the Niger, an even more general extrapolation describes the catch to be expected from the plains. This shows that mean catch from a number of floodplain rivers is 38.3 kg/ha of flooded area. As several of these floodplains are very lightly fished it may be concluded that normally exploited flood plains can be expected to produce between 40 and 60 kg/ha/yr. Although these formulae are very general and are unlikely to have a precision of more than say ± 50%, they have been applied with success to river systems outside the set from which they were originally derived both in Africa, Asia and South America. The derivation of similar relationships from other continental areas is no doubt feasible, as is the increase in precision by adding further terms, such as conductivity, to the equation.
Because of the great difficulty in obtaining reliable estimates of fish catch and fishery yield potentials from the main channels of large rivers, several authors have sought correlations between ichthyomass and other more easily sampled organisms in the river system. For example, Russev (1973) and Holcik (in press) have attempted to relate benthos to ichthyomass in the Danube. Holcik and other workers established that in minor rivers of the Danube Basin the ratio between benthic biomass and ichthyomass was a mean of 1.6:1. This factor was used to convert benthic biomasses to ichthyomass in the main channel of the river.
Russev's method seeks to convert benthic biomass to ichthyomass through a series of conversion factors which presumably introduce some flexibility into the process. Here, benthic biomass (Bb) is converted to production of benthic organisms (Pb) by:
Utilization of Pb by fish ranges between 15 and 30% depending on the richness of the waters. The weight of benthos utilized by fish is converted to fish production (Pf) by the application of a feeding ratio whereby 5 kg benthic production = 1 kg fish production.
Finally, the production of fish (Pf) is converted to ichthyomass (Bf) by:
Subsequently a correction within the fish biomass may be applied for forage fish and predatory species of
This method, however, introduces many assumptions into the estimation and is probably no more accurate in the long run than a simple conversion such as that used by Holcik.
Of the basic information needed to manage fisheries the most important is the actual amount of fish being caught in any defined period. Further information on composition of catch, seasonality of landings, price structure, etc., is also of considerable interest but is less vital. Two main approaches are usually adopted to evaluate the state of the fishery: (i) the landing approach, and (ii) the market approach. These are described by Bazigos (1974) who analyses in detail the statistical basis of the various sampling methods. What follows here is a broad outline of the principal features of the different approaches.
Fisheries generally need to be assessed at two levels: firstly, at a level general enough to give an overall knowledge of the quantity and species compositions of fish being landed and to give some indication of any seasonal patterns within the fishery; secondly, at a more detailed level to determine the characteristics of the catch, the types of gear being used and their catch characteristics. The landing approach has the flexibility necessary to gather information at both levels.
General assessments of fish catch (C) may be obtained from:
C = cpue × f
where cpue = catch per unit of effort, and f = effort.
In restricted fisheries such as those of small lakes, determination of catch per unit effort and effort are relatively easy. However, in larger lakes and especially in rivers, the numbers of fishermen involved in the fishery rise and the diversity and extent of the system increase to unmanageable proportions. For this reason, assessments have to be based on two types of statistically-based surveys used together. These are:
(i) The frame survey to determine the number of fishing units and their component operating in the system as well as the seasonality of their fishing activity;
(ii) The catch assessment survey which samples catch and gathers information needed to estimate total landings and the fishing effort involved.
Both of these components depend for their effectiveness on the stratification of the system to be covered by the surveys.
The purpose of stratification is to subdivide the whole system to be sampled into a number of manageable units which are homogeneous within with respect to survey characteristics. Good stratification requires a thorough examination of the topography of the region under study from maps and where available from satellite imagery. One of the ways in which a system can be stratified is illustrated in Fig. 11 for the Magdalena River, Colombia (see also Bazigos, Kapetsky and Granados, 1975). Here the system was divided first into four strata corresponding to the main channel of the Magdalena River (01) and three major tributaries, the Canal del Dique (02), the Rio San Jorge (03) and the Rio Cauca (04). The main river was divided into four zones. Within each zone and stratum the lagoons were counted, measured and classified into size groups to provide the third level of stratification upon which the final sampling frame was based.
The objectives of the frame survey are to establish a basic inventory of the fishery resource in the given river system. It aims at collecting information on: (i) the size and distribution by area of the fishing or landing sites along the river, (ii) the number and distribution by area of fishing craft, (iii) the number, type and distribution of fishermen and/or fishing communities. In the survey, information is also collected on the temporal movement of the fishermen and of the processing, transporting and marketing of fish catches. There are two basic ways of collecting this information. Firstly, by water/road approach, whereby a mile-by-mile survey is carried out during which time the observer has to record all fishing sites, boats, etc., seen. The observer has to obtain information at each landing site on the number of fishermen and the types of boats and gear in use. This procedure has advantages in that it gives more detailed and possibly more accurate information than its alternative - the aerial count - but is slow and costly in manpower and, in long rivers with complex systems, is virtually impracticable. Secondly by air, whereby the observer flies slowly along the river course counting fishing sites, boats, etc. This method has the advantage that it is quick and that the view from the air reveals the distribution of fishing sites in the floodplains flanking the river, which are frequently undetectable from the river. The method provides less complete information on factors other than number and type of fishing boats seen or house counts.
The information gathered during the frame survey must be supplemented by more detailed data on individual catches to be of advantage in determining the amount of fish caught in the whole system. The catch assessment survey is designed to provide such data. The first requirement of the catch assessment survey is to select a series of fishing or landing sites within the established sampling frame for closer study. Ideally, selection of the sites should be randomized but in certain cases the choice is limited by purely practical considerations such as accessibility and manpower requirements. Attempts should be made, however, to ensure that a representative sample of sites be selected in each zone or stratum. Selection should also take into account mobility patterns among the fishermen who are apt to move seasonally in response to difference in the flood regime and migrations within the fish community. The actual survey at each sampling site may take into account all fishermen or in the case of larger fishing communities sub-sampling of the fishermen may be required. In either case, the survey may be conducted either by the interview method, whereby the required information is obtained by questioning individual fishermen, or by the actual measurement of the catch of the selected landings. Data on many other characteristics of the fishery can be acquired at the same time, including the types of gear available and their manner of use. More detailed socio-economic information capable of identifying external constraints on the fishery can also be obtained at very little additional cost.
Fig. 11 Sampling stratification of River Magdalena, Colombia
In some rivers the fishermen are so dispersed and landing points so numerous that adequate samples are almost impossible to obtain without a massive input of manpower. In these cases, the market approach may be feasible. This depends on two main components:
(i) The market survey to determine the quantity, type and seasonal variations of fish entering the primary markets of the river fishery;
(ii) The auto-consumption survey to estimate the quantity and type of fish consumed by the fishermen, their families and other related riparian communities.
This approach can obviously be pursued with value only where there is a distinct and well established marketing network and where the sociological structure of the fishery is well understood.
The market survey: This relies on counts of fish entering the primary markets from the producing fishing sites and the processing sites. Where fishermen and fish processors operate independently of major landings, this point is often the first occasion that fish is assembled in appreciable quantities. There are, however, many pitfalls to be avoided in surveying markets, not least the presentation of the same fish for sale on separate occasions or in several places, which obviously leads to overestimation.
The auto-consumption survey: Because the market survey measures only the production of the fishery that is surplus to local needs, full information on total weight of catch is only obtainable by joining to it a survey to evaluate the consumption of fish or the fishing areas. This process is extremely difficult as it requires, firstly, a knowledge of the number of persons involved in the fishery or associated with it in such a way that they participate in the consumption of the fish caught; secondly, some estimate of the average per caput consumption of fish across such a community. The position is sometimes complicated by the influx of fish from elsewhere, for instance an adjacent marine fishery, which permits the target fishermen to market more expensive elements of their own catch.
In conclusion, therefore, the market approach is generally unreliable as a means of estimating total catch in large inland fisheries, and if it is adopted, relies on experienced survey staff for its design and execution. Its other disadvantage is that some important aspects of the fishery are neglected such as cpue, gears and their characteristics, numbers of fishermen, etc.
The term “fish stock” is used here somewhat loosely as, in general, the fish communities inhabiting rivers are highly complex, consisting of a number of species populations, each of which may not be composed of discrete stocks. Assessments may, therefore, be directed at various levels, from that of the whole community down to a detailed analysis of individual species, or, where necessary, even to component stocks of species of great importance. The objective of this type of assessment is basically to gain some idea of the potential for production of the chosen level in the community in order to relate it to the existing state of exploitation by the fishery. Other items of information are also sought, such as seasonal variations in the community, composition of the fish stock and such details of biology of individual species as may be required for their management. Clearly, the more complete the information required the more intensive and costly the sampling programme. In this section we will consider only those methods which yield quantitative information on various levels of the fish community such as potential catch or standing stock.
There are several capture techniques aimed at the removal of all fish from the portion of water to be studied. Two methods particularly are favoured: electric fishing and poisoning, although the two methods have also been used in combination.
Total removal methods are most commonly used on small rhithron-type streams where the reach to be sampled can be completely isolated from the rest of the stream course with block nets. Once blocked, the reach is fished systematically, using either electric gear, which is particularly valuable where there is dense submerged aquatic vegetation, or by a poison such as Rotenone, which is drifted downstream with the current. In the case of poisons, the toxin is neutralized with a neutralizing agent such as potassium permanganate as it passes through the lower stop net.
Both methods have also been used in various parts of the potamon river -floodplain complex. In floodplain water bodies, the open waters have been studied by enclosing sample areas with block nets and in the inshore waters embayments have been enclosed prior to dissemination of the poison. The main river channel, by reason of its size and strong current is less accessible to these, or indeed any other method. Nevertheless, several attempts have been made to sample pools, backwaters and even the open waters during the dry season using similar techniques to those employed in lagoons.
Not only do these methods provide good information on the standing stock present at any one time in any one portion of the river system, but they also give large samples of fish for the study of any biological parameters as may be of interest.
Although in theory fishing with either electricity or with poisons should result in nearly all fish in the selected reach being caught, there is, in fact, considerable selectivity which can result in samples which are low in total number and weight and are biased by species. With electricity there is a tendency to capture the larger fish more easily and with poisons some species are more resistant than others. Furthermore, both methods are time-consuming and require a large labour force. Nevertheless, they remain the most reliable and informative of all sampling systems.
A further method involving total sampling of the fish community is the use of barrage traps to close off the canals draining individual floodplain depressions. Here, all fish can be captured and counted as they leave the floodplain depression for the main channel of the river. Similar approaches, whereby migrating fish are counted individually as they pass through fish passes or traverse weirs, are current in the high-value salmonid fisheries of north temperate rivers. In these cases, a range of counting devices have been used including visual observation, automatic electric counters and sonar systems with integrators. Sonar systems have also been used to locate and estimate density of fish shoals in waters other than the main channel and new advances in equipment hold promise for the most extensive application of sonar in the future.
In large river systems where the distribution of the fish stock fluctuates considerably from season to season, the use that can be made of mark-recapture methods is limited. The method may be used to estimate stock size in water bodies of the floodplain or in back waters of the main channel when they are inundated at low water. Otherwise, in the main channel marking is of use mainly to delineate migratory pathways.
Mark-recapture methods depend on the capture and marking of a sample of fish which are then returned to mingle with the original population. The subsequent treatment of this population under study differentiates between three basic types of method -the single census, multiple census and multiple recapture methods. All methods rely on certain assumptions that:
(i) marked fish are randomly distributed among the total population;
(ii) that marked fish and unmarked fish are equally accessible to the fishery;
(iii) that marked fish have the same mortality rate as unmarked fish;
(iv) that marks are not lost;
(v) that there is no recruitment, immigration or emigration into the populations between markings.
Single census method: Here the mixed population of marked and unmarked fish is sampled once. This catch will contain a number of marked fish (r) among the total number of fish caught (C). This proportion is assumed to be the same as the ratio between the total number of fish marked (m) and the total population (N). Thus:
Multiple census method: Here the population is sampled on several occasions at which time additional fish are marked and added to the population and already marked fish are recorded. The total number of fish in the population (N) is thus estimated on several consecutive occasions according to the formula:
Where Ct = total sample taken on day t; Mt = total number of fish marked prior to day t and rt = number of recaptures in sample Ct.
Multiple recapture method: This method, which relies on several catchings during which fish caught are given differing marks by fishing, may be used in cases where the system is less isolated and there is some movement in or out of the original population. This makes the method suitable for back waters which remain connected to the main channel. The population must be fished at least three times. At the first fishing (t1) m1 are marked and returned. At the second fishing (t2) the catch C2 is recorded, the recaptures from the first marking (r12) are counted, unmarked fish are given a different mark and all are returned. At the third fishing (t3) the total catch C3 is recorded and the numbers of fish marked on the first fishing (r13) and the second fishing (r23) are recorded separately to form a table as follows:
|Time period||Fish newly marked||Fish in catch||Recapture from||Recapture from|
|first marking||second marking|
The estimate of abundance at time 2 can be obtained from:
The catch from certain types of active gear such as trawls or seines can be used to estimate the standing stock per unit area in certain types of water body. The main pre-requisite for the use of such methods is a clear and unencumbered bottom suitable for dragged gear, although pelagic trawls may be used to sample surface-living fish in a wider variety of habitats. These are most commonly found in the main channel of large rivers or in the open waters of large floodplain lakes. Estimates rely simply on:
Where Bf = ichthyomass per unit area, C = catch per trawl, A = areas swept by the net and E = a measure of the efficiency of the gear. In the case of trawls: A is further defined as Dgh where D = distance trawled, g = gape of net and h = the distance separating the footrope and headrope.
Ideally, all fish within the swept area should be caught but this is rarely the case and it is common for a correction factor (E) to be applied to compensate for the loss of a certain percentage of the fish. Furthermore, as both trawls and seines are mesh selective, at least in respect of fish greater than that which will pass through the mesh, samples have to be adjusted for this too.
Seber-Le Cren two-catch method: This method relies on two successive catches being taken from an area whose fish population is isolated from the rest of the system. This may be either a small isolated water body or a portion of a larger water body or the main channel where a known area is enclosed by a block net. Fishing may be carried out with several types of gear but an active gear such as a seine net is perhaps most reliable. The basic assumptions of this method as set out by Seber and Le Cren (1967) are:
(i) That the first catch is large enough to have a significant effect on the population and therefore that the second catch is smaller than the first;
(ii) That the fishing effort is the same for the two catches and that fish remaining after the first fishing are as vulnerable to capture as those caught during the first fishing;
(iii) That there is no recruitment, mortality, immigration or emigration between the times of the two fishings;
(iv) That the first catch is removed from the population or the individual fish are marked so as to be recognizable in the second catch;
(v) That the population is totally available to the fishery.
In practice, the two samples can be taken in rapid succession, minimizing any population changes between the two fishings.
Assuming the above conditions, the size of the population N can be estimated from two successive catches, C1 and C2 as follows:
the variance is calculated from:
from which the standard error is:
However, in rivers the mobility of the fish is such that they rapidly fill in the space being sampled, producing radical changes in species composition and also invalidating assumption (iii). This method is thus of little use in the main channel but may have application in the standing waters of the flood plain.
Where the fishery is already well developed and there are adequate data on effort and catch over a number of years, these data can be used instead of experimental fishing to estimate various quantitative parameters of the population.
Leslie or DeLury method: This method departs from similar basic assumptions to those of the Seber-Le Cren two-catch method except the assumption (i) where the population must be fished heavily enough to significantly reduce catch per unit effort (cpue). The method follows from the principle that the decrease in cpue which follows from the depletion of the population is directly related to the extent of that depletion. It can only be applied where fishing is heavy relative to the population and commercial catch statistics are often called upon to provide the raw data, although in smaller systems repeated catching with experimental gear will also give the necessary information.
Estimates of fish population can be derived graphically by plotting cumulative catch by number against catch per unit effort (see Fig. 12) in which case the extension of the plotted line to cut the x axis gives at the intercept an estimate of total abundance.
Fig. 12 Regression line of relationship between catch-per-unit effort and cumulative catch. The intercept on the X axis gives estimate of total abundance.
Backiel's method: This method, suggested by Backiel (1971) is based on catch statistics and also requires enough biological data to make some estimate of natural mortality and the ratio between production and biomass. Here production (P) is defined as:
where Bc = annual catch, Bm = biomass removed by natural mortality and k = P/B.
Where insufficient data are available to calculate natural mortality, it may be assumed that (i) the catch and the biomass removed by natural mortality are of the same order of magnitude, thus total mortality (Bc + Bm) = 2Bc; (ii) mortality is balanced by recruitment and growth so that the population is in a steady state, and (iii) the P/B coefficient varies within a narrow range. P/B values have been estimated at between 0.6 and 0.9 in rivers in the temperate zone and between 0.7 and 1.3 in the tropics. However, the P/B ratio is very sensitive to the effects of the early stages of the life history and when eggs and fry are included in the calculation, values for P/B are raised by as much as a factor of 2. Nevertheless, the assessment of production by fish stocks made by this method may be very close to those obtained by more time-consuming procedures.
Surplus yield method: Where the fishery is well developed and there is adequate data on effort and catch over a number of years, the Schaeffer surplus yield model can be used to predict maximum catch. The model posits that the yield from a fish population is a function of population abundance and total fishing effort. The relationship between effort and yield is a logistic curve of the form:
Y = af - bf²
where Y = yield, f = fishing effort and a and b are constants (see Ricker (1975) for derivation and discussion).
From this it follows that the relationship between effort and catch per unit effort is:
Following from this equation the optimum level of fishing effort (fs) is:
and the maximum sustainable yield Ys is
The method has the great advantage that maximum sustainable yields (Ys) and optimum rate of fishing (fs) can be estimated from as few as two equilibrium levels of catch and fishing effort. The main disadvantages of the application of the method in rivers and particularly in flood rivers is that the population and hence the fishery rarely reaches any state resembling equilibrium. Fisheries in reservoir rivers, with their more stable regimes may be more suitable subjects for this kind of treatment but there is also some doubt as to the applicability of a single surplus-yield model to a multi-species fish community. Here responses to fishing will vary from species to species and changes will occur in the species composition within the community invalidating the assumption of equilibrium level. Furthermore, as shown in Fig. 8, fisheries based on multi-species stocks do not necessarily conform to the logistic model. However, should changes in species composition be compensated for within the fishery, the use of a single model to describe the overall behaviour of a community under fishing pressure in stable water regimes may be valid. The method is also of use where the fishery in such waters concentrates on a limited number of high values species.
In certain circumstances the yield from the fishery is limited more by environmental variables than by fishing pressure. Such cases are relatively poorly understood but the best studied are those rivers with pronounced seasonal flood regimes. Here, because flood regimes vary from year to year and because the fish stock responds to such variations with corresponding fluctuations in growth and mortality which are reflected in the ichthyomass, it may be assumed that the flood conditions in any year or group of years may be used to predict catch in the following year. In fact, very clear relationships emerge when catch is plotted against flood regime, although a considerable series of data on both parameters is required for the calculation of the appropriate regressions. This approach has been tried on several occasions and in several parts of the world, and regression equations of various forms have been derived. The conclusion that can be drawn from these attempts is that a regression of the form:
Cy = a + b (P0HIy + P1HIy-1 + P2HIy-2 + … PxHIy-x)
or some transformation of it can be used to predict catch (C) in year y from some index of the intensity of the flood (HI) in the same or in preceding years. The lapse of time between the year in which the flood occurs and the year when its effects are reflected in the catch, appears to be related to the growth rate of the fish and the time taken for a year-class to enter the fishery. Thus, in the tropics, where there is heavy fishing on young fish and growth is rapid, the best correlations of Cy are with HIy and HIy-1, whereas in the colder Amur River, with slower growth and a fishery aimed at larger individuals the best correlations are with HIy-3 and HIy-4 (Krykhten, 1975).
The methods described above cover a wide range of inputs in terms of cost and time, and a range of outputs in terms of accuracy and applicability. Some require very little data, at least from the system under study, but have a correspondingly low level of accuracy, whereas others presuppose the existence of an intensive and well documented commercial fishery. Perhaps the best way to view the application of these methods is as a succession, starting with the simplest generalized method for initial prediction of yield, progressing through experimental fishing to further define the stock, to commercial statistical methods to monitor and manage the fishery. However, things are rarely so simple and on most inland waters fisheries are usually already well established and the pressures to find quick answers along with financial and other constraints may not allow the time for this leisurely approach. Assessment methods, therefore, have to be chosen with care to match the information requirements as well as the financial and manpower resources available. Fig. 13 relates the methods discussed above to each other and to the types of output required.
Fig. 13 Possible combinations of methods for stock assessment and yield prediction