Water bodies may be classified on a scale ranging from unproductive to productive, depending on the rate of biomass production (Welch, 1980; Forsberg and Ryding, 1980). Various indices such as primary production (Grandberg, 1973), algal biomass (Dillon and Rigler, 1974), oxygen deficit (Cornett and Rigler, 1979; Welch, 1980), indicator species (Rawson, 1956), fish yield (Melack, 1976; Oglesby, 1977) aquatic macrophyte production (Canfield et al, 1983) and nutrient concentration (Vollenweider, 1976), or combination of the above (Carlson, 1977) can be used to assess productivity. However, all reflect and are related to changes in nutrient supply (see below).
The introduction of enclosure culture to a water body will alter the productivity, and although cage and pen culture can influence the flora and fauna directly through the introduction of novel species (parasites, exotic fishes etc) and the attraction of predatory birds and mammals to the area, most of the changes that occur are as a result of changes induced in the productivity. The direction of change is determined by the method of culture employed (extensive/semi-intensive/intensive), whereas the magnitude of the changes will depend upon the characteristics of the site, the type of enclosure structure used, the species cultured and the size of the venture. These relationships are summarised in Fig. 13.
In Fig. 14, the impact of enclosure culture is seen to range from a net uptake of nutrients, resulting in a decrease in productivity, to a net output of nutrients, resulting in an increase in productivity. Extensive culture relies on natural food, and therefore by stocking, culturing and harvesting fish, nutrients are removed from the system. During intensive culture all nutritional requirements are met through the external supply of high quality food, and thus there is a net increase in supply of nutrients to the system through waste feeds, faeces and urine. The slope of the line, and the points at which it intercepts the X and Y axes depend on a number of variables: species, site, season, stocking density, quality of food, management. At the point where the line crosses the X-axis, there is effectively no change in productivity and the quantity of supplementary feed used balances the impact of the fish culture operation.
In attempting to model how the environment changes when cage and pen culture methods are used, we need to know:-
We also require to know how much change is permissible in order to try to manage the system.
Aquatic ecosystems consist of large numbers of different types of organisms which depend upon the fixation of light into carbon-containing compounds by photosynthetic green plants, and the subsequent cycling of material through a complex web of food chains (Barnes, 1980). Most lakes and reservoirs derive their energy base of organic material principally from autochthonous (internal) production by algae, macrophytes and periphyton (Pomeroy, 1980). These plants require only light, a carbon source and a supply of nutrients. Plant material is consumed by planktivores and herbivores which in turn are preyed on by primary and secondary carnivores. Unconsumed plant and animal material plus faecal material form the non-living organic detritus which is utilised by a wide variety of organisms - suspension feeders, shredders, grazers, scrapers, deposit feeders and bacteria and fungi.
In some lakes and reservoirs fringed by swamps, such as Lake Chilwa in Africa and Bukit Merah in Malaysia, and in many lotic systems, the external supply of dead organic material (allochthonous) may be greater than autochonous tissue elaboration by macrophytes and algae (Cummins, 1974; Howard-Williams and Lenton, 1975; Townsend, 1980; Yap, 1982, 1983) (see also 2.2 above). Productivity data from a range of deep and shallow water bodies (reservoirs, lakes, rivers) from throughout the world are summarised in Fig. 15. The data comes principally from the IBP results (Le Cren and Lowe-McConnell, 1980), but also includes more recent data from tropical South America, Africa, and the Philippines (Hill and Rai, 1982; Marten and Polovina, 1982; Tundisi, 1983). It can be seen that in general there is an increase in gross primary production (gCm-2day-1) from high to low latitudes, and this corresponds to the findings of Brylinsky and Mann (1973) Schindler (1978), and Hill and Rai, (1982). Low latitudes have greater solar radiation and higher temperatures than high latitudes, and it is thought that the greater availability of light and the effects of temperature on growth kinetics underlie this trend. These findings are strongly supported by theoretical considerations, and their validity has been subject to much scrutiny in both the laboratory and the field (e.g. Schiff, 1964; Talling, 1957, 1971; Goldman and Carpenter, 1974).
From the IBP global data set a second correlation was found by Brylinsky and Mann (1973), between primary production and nutrient concentration, as measured by conductivity. A range of nutrients are required for growth by algae and macrophytes and these include several vitamins and a large number of inorganic salts (Stewart, 1974; Fogg, 1975, 1980). In theory any one or more of these essential nutrients could restrict growth. If the supply of a particular nutrient was less than the demand of the primary producers then this nutrient would be limiting. The search for which nutrient/nutrients limit primary production in inland waters has been of consuming interest to scientists and resource managers for years. The arguments hinge around the questions of supply and demand.
Although vitamins such as cobalamin, thiamine and biotin/coenzyme R have been identified from laboratory work as essential, they have rarely been found to be limiting under natural conditions (Welch, 1980). Thus research has been focussed on inorganic nutrients. In undisturbed ecosystems most of these nutrients are derived principally from the erosion of rocks and soil (the lithosphere), although C, N, B, S and Cl also have large atmospheric reserves. Recent studies have shown that many of these atmospheric nutrients are never or rarely limiting. The demand for B, S and Cl is much less than the supply (Moss, 1980). Carbon is the element required in largest quantities for primary production and most appears to be atmospheric in origin (Schindler, 1971, 1974). Although large pools of dissolved organic C compounds such as humic acids are found in some water bodies (e.g. Scottish Highlands), these are highly resistant to chemical oxidation (Shapiro, 1957). However, atmospheric CO2 is readily dissolved and easily enters aquatic ecosystems through diffusion or in rainwater (Hutchinson, 1957). Current opinion is that although C may be temporarily limiting, for most of the time supply exceeds demand by a factor of about 30 (Schindler, 1971, 1974; Moss, 1980; Welch, 1980).
Nitrogen constitutes almost 80% of the atmosphere, but despite this is relatively unreactive. Before it becomes available to plants, atmospheric N must be fixed either through electrical discharge (lightning) or through biological fixation by bacteria and blue-green algae. The relative importance of N-fixation to the total annual input of N to a water body varies greatly, from <1% in Lake Windermere, England, (Horne and Fogg, 1970) to almost 90% in Pyramid Lake, Nevada, USA, (Horne and Galant, in Goldman and Horne, 1983), and is primarily dependent on the density of blue-green algae. Other sources of N include rainwater, which contains both NH4 and NO3, although the relative amounts of each fraction differ between temperate and tropical areas (Hutchinson, 1957).
In the absence of O2, bacteria can denitrify NO3 to N2, which then may be lost to the atmosphere. However, this process is now thought to be of little importance in N-limitation (Welch, 1980). Nevertheless, there is evidence that N can be limiting to growth in a number of circumstances, and these will be discussed below.
Of those essential elements derived almost solely from the lithosphere, P is the scarcest with respect to algal and higher plant requirements (Vallentyne, 1974). In Table 8, the relative amounts of the other essential elements in the lithosphere compared to P are shown in column 2. Those elements with a ratio of < 1 are less common than P, and include Mn, Co, Cu, Pb and Mo. The third column shows the ratio of P to algal and higher plant requirements. Values > 1 show those elements which are in greater demand than P (e.g. Ca, K, Mg). In column 4, the ratio of supply (column 2) to demand (column 3) is shown. Values > 1 show those elements whose demands are more likely to be met by lithosphere supply than those of P, and it can be seen that all elements fall into this category.
The reasons for the scarcity of P are threefold. First of all, it is relatively rare, there being no gaseous reserves in the atmosphere (unlike C, H or N), and secondly it is relatively insoluble and readily complexes with a wide range of metals which includes Fe, Al, Mn and Ca, and is thus precipitated (Stumm and Leckie, 1971). Phosphorus also adsorbs on the surface of particulate organic matter and is absorbed by phytoplankton, both of which being comparatively heavy are prone to sedimentation and loss from the water column (Welch, 1980; Sonzogni et al, 1982).
On theoretical grounds, therefore, P and N are prime contenders as limiting nutrients. Algal requirements for N are about 16 times greater than for P, on a molecular basis, as calculated by Stumm (1963) (in Welch, 1980)
106CO2+ 90 H2O + 16NO3 + 1PO4 + Light C6H180O45N16P1 + 154½O2
Results from experiments by Chiandani and Vighi (1974) confirm this and show the range of algal N:P requirements to be 17–13:1 (8–6:1 on a weight:weight basis). Similar conclusions were reached by Palaheimo and Zimmerman (1983). However, the results from fertilisation experiments, where N, P, C, and trace metals have been added to lakes and reservoirs in various combinations and subsequent changes in productivity measured (Goldman, 1960; Schindler, 1971, 1974; Schindler and Fee, 1974; Robarts and Southall, 1977) have demonstrated that additions of N have little or no effect whereas even small amounts of P can stimulate production dramatically. Additions of C and trace metals have also been found to have limited effect.
Indirect evidence from analyses of large data sets compiled by a number of government bodies and researchers on productivity, plankton biomass and nutrient levels in lakes and reservoirs has confirmed that P is usually the limiting nutrient. These data are summarised in Table 9. First of all, almost all of the lakes investigated have N:P ratios above the critical 8:1 value, suggesting that there is adequate N available to supply algal requirements. Exceptions have been found in P-rich volcanic areas, such as in areas of Japan (Sakamoto, 1966), East and Central Africa (Moss, 1980) and South Island, New Zealand (White et al, 1982). Secondly, although weak correlations between plankton biomass and total N exist (e.g. OECD, 1982; Prepas and Trew, 1983; Hoyer and Jones, 1983), much stronger correlations are found with spring or summer total-P concentrations.
Most of these regressions have a considerable amount of variance associated with them, and some data sets show weaker total P-algal biomass (as measured by chlorophyll ‘a’ concentrations) correlations than others e.g. North American reservoir data of Canfield and Bachmann (1981), Jones and Novack (1981), and Walker (1982). Part of the variance may be explained by differences in methodology (Hoyer and Jones, 1983), and by inter- and intra-specific differences in algal chlorophyll ‘a’ content (Palaheimo and Zimmerman, 1983). Not only does the chlorophyll ‘a’ content vary from one species to another, but it also changes with age, cell volume, light intensity and nutrient concentration (Nicholls and Dillon, 1978; Grandberg and Harjula, 1982). Using multiple regresson techniques, N has been found to account for little or none of the variance in all but one data set (Clasen, 1981; Walker, 1982; OECD, 1982; Prepas and Trew, 1983; Hoyer and Jones, 1983), confirming that total N:P ratios are usually above the 8:1 level, and that few water bodies contain ideal N:P ratios for algal growth.
However, multiple regression analyses have identified one further important factor which can account for a large part of the variance in some instances: turbidity. Generally, water clarity, as measured by secchi disc, is a function of algal density (Bachmannand Jones, 1974; Dillon and Rigler, 1975). However, analyses of a large number of North American lakes and reservoirs by Canfield and Bachmann (1981) have shown this relationship to be weaker in artificial lakes, which often have high levels of inorganic suspended solids. Other recent studies have shown that algal biomass - total P relationships in shallow water bodies which are susceptible to wind-induced disturbance of bottom sediments and in those with high inputs of inorganic nutrients are weak (Pieterse and Toerien, 1978; Clasen, 1981; Nielsen, 1981; Walker, 1982; Hoyer and Jones, 1983). Inorganic suspended solids can reduce plankton biomass and adversely affect production in several ways. High inorganic turbidities reduce light penetration and restrict the depth of the euphotic zone (Marzolf and Osborne, 1972; Moss, 1980; Canfield and Bachman, 1981), and a number of studies have also shown that PO4-P is readily adsorbed onto inorganic particles thus decreasing the concentration of biologically available P (Fitzgerald, 1970; Stumm and Leckie, 1971; Edzwald et al, 1976; Furness and Breen, 1978; Hoyer and Jones, 1983).
Phosphate-P concentration is currently regarded as the primary limiting factor governing algal biomass and productivity for large parts of the year in both temperate and tropical inland waters. However, the generalised P-algal biomass relationships tend to treat all water bodies as a homogenous group despite the fact that the P-algal biomass/primary production correlations in Table 9 suggest that there is often a great deal of associated variance that can only be explained by taking into account other locally important factors such as geology, depth, exposure, climate and watershed size and use. As more empirical data is collected, refined models governing different categories of water body will evolve.
In summary, aquatic food chains function as shown in Fig. 15. In lentic water bodies, autochthonous production of organic material and its subsequent processing, fuels the system. In lotic water bodies, particularly fast-flowing erosive systems, plankton are often present in insignificant numbers and the main sources of energy are derived from periphyton, fringing macrophyte communities, or allochthonous detrital material. Productivity of the entire system may be controlled at any of the points illustrated in Fig. 15:- light, essential nutrients or, in the case of lotic systems, allochthonous detritus.
Phosphorus and, occasionally, light are the principal factors limiting production in both temperate and tropical freshwaters, and thus the net addition or uptake of P or materials which greatly influence the light climate will alter productivity. In this Section, however, the latter factor will be ignored but will be considered in Section 4.6.
Phosphorus is an essential element required by all fish for normal growth and bone development, maintenance of acid-base regulation, and lipid and carbohydrate metabolism (Ketola, 1975; Ogino and Takeda, 1976; Lovell, 1978; Cowey and Sargent, 1979; Lall, 1979; Sakamoto and Yone, 1980; Takeuchi and Nakazoe, 1981). Diets deficient in P can suppress appetite, normal food conversion and growth, and under extreme circumstances affect bone formation and lead to death (Murakami, 1967; Andrews et al, 1973; Lall, 1979). Although the uptake of labelled 32P by fish from water has been demonstrated many times (e.g. Tomiyama et al, 1956, in Lall, 1979), it is believed that the rate of absorbtion is generally very low, and that fish derive their P requirements principally from food (Phillips et al, 1957; Nose and Arai, 1979).
Phosphorus requirements for different species of fish range from 0.29% to 0.90% of the diet (Table 10). However, these figures refer to available phosphorus which varies greatly with species depending on the dietary source. The majority of currently available intensive fish feeds are largely of animal origin, such as fish meal, meat meal and bone meal, where most of the P is present in inorganic form, and the remainder is in the form of P-complexes in proteins, lipids and carbohydrates (Lall, 1979). Nearly all of this P is readily available to carnivorous fishes such as rainbow trout (Ogino et al, 1979). However, the availability of P in fish meal diets to omnivores and herbivores is highly variable. Whilst O. niloticus can utilise 65% (as much as rainbow trout) of P in fish meal based diets (Watanabe et al, 1980a), the availability to common carp is almost zero (Ogino et al, 1979) due to the absence of acidic gastric juices (pepsins) (Yone and Toshima, 1979). On the other hand, 60–80% of the total P in plant materials exists as the Ca or Mg salt of phytic acid, known as phytin, and is unavailable to fish as they don't possess the necessary enzyme, phytase, to break down the compound. (Ogino et al, 1979; Lall, 1979). There is some evidence that omnivores/herbivores such as the carps are better at utilising the non-phytin fraction in plant-based diets than the carnivorous rainbow trout (Ogino and Takeda, 1976).
The availability and utilisation of P has also been shown to be influenced by the amount ingested, body P reserves, other elements in the gut and body tissues, and the remaining dietary ingredients (Nakamura, 1982; Tacon and De Silva, 1983). Thus, depending on the digestibility of the source, a significant proportion of the P intake may be egested. Absorbtion efficiency and (at levels above dietary requirements) growth rate, however, are independent of the level of dietary P, and hence excretion is positively related to intake (Nakashima and Leggett, 1980). Phosphorus surplus to dietary requirements is largely excreted through the kidneys (Forster and Goldstein, 1969). These relationships between intake, excretion, growth and absorbtion efficiency are illustrated in Fig. 16.
Most feeds used for intensive culture in temperate countries are commercially made and are in dry, pelleted form. Some farms in Europe still use trash fish as a diet, although this practice is now being restricted in many countries (see Alabaster, 1982a). A summary of Tacon and De Silva's (1983) survey of the P content of commercially available European salmonid diets is given in Table 11. Mean values suggest that P content of trout diets is ∼ 1.49%, and that of salmon ∼ 1.47%. In Poland, Penczak et al (1982) used feeds (dry pellet, and fresh fish/wheat bran/yeast moist pellets) with an average P content of 1.45%, for cage culture of trout, whilst Ketola (1982) in the USA used a European “low-pollution” diet (1.40%) and a commercially available North American diet (2.2%) in his trout culture studies.
Intensive culture of carps and tilapias still relies largely on the manufacture of diets from locally available materials, the exception being the commercially produced tilapia diets available in Taiwan. The P content of raw materials and diets in use in various countries is given in Table 12. For tilapias, the P content of diets varies between 1.30 and 2.52%, whilst those compounded for carps vary between 0.93 and 3.06%.
Feed losses are inevitable during fish culture for a number of reasons. Many near-surface feeding fishes, such as the salmonids, are visual feeders (Blaxter, 1980) and only ingest food items within a particular size range which is positively related to some function of fish biomass (Wankowski and Thorpe, 1978). Pellet sizes for salmonids, based on manufacturers' recommendations, are given in Table 13. Food items which are outside the particular recommended size category for a given size range of trout, will not be eaten, but instead contribute to the wastes from the operation. Manufacturers estimate that 2% of feed is ‘dust’, due largely to the crumbling of pellets during packing and transport. Thus at least 2% of commercial trout feeds will be uneaten.
Particle size in the diet of tilapias seems at first glance to be less important. Many of the cultured species, such as O. niloticus and O. aureus are microphagous feeders (Bowen, 1982), and according to Miller (1979) and Coche and Lovshin (Pullin and Lowe-McConnell, 1982) powdered feeds produce as high yields from pond culture as pelleted feeds, without the added expense of pelleting.
Whilst the above findings may apply to tilapia culture in ponds and pens, they do not apply to cage culture. Losses of feed from cages have frequently been observed (Collins, 1971; Loyacano and Smith, 1976; Hoelzl and Vens Cappell, 1980; Penczak et al, 1982; Phillips et al, 1983), and are due both to passive water currents as well as to currents induced by the fish during feeding. Thus pelleted feeds for tilapia cage culture have been recommended by many authors (Guerrero, 1980; Coche, 1982; Santiago, 1983). Jauncey and Ross (1982) have observed that in general tilapias prefer smaller pellet sizes than most other cultured species, and recommended sizes are given in Table 13.
In summary, P is an essential mineral which fish obtain almost exclusively from their diet. Most diets developed for intensive culture contain P surplus to requirements or in a form which is partially unavailable to the fish. Surplus P is excreted, whilst unavailable P is passed out in the faeces. In fishes such as the salmonids which have size-specific preferences for food, damaged pellets may not be ingested, but instead contribute to the P supply of the water body. Other sources of P to the environment are derived from food which is washed out of the cage by both natural currents and turbulence caused by the fishes during feeding.
The principal P losses to the environment associated with intensive enclosure culture are summarised in Fig. 17. There are several methods which can be used to quantify these losses:-
Direct measurement of inputs from pens and cages
Theoretical calculations based on available information on P content of feeds, etc.
Extrapolation of data from intensive pond and raceway culture to cage and pen production.
Although there are a number of studies where wastes from intensive cage trout farms are being measured (see Section 3.3.2), only that of Penczak et al (1982) has been completed. In this study, waste production from cage trout culture at Glebokie Lake, Poland, was determined by measurement of C, P and N inputs and outputs. Total nutrient losses to the environment, Nutenv' were computed as being equivalent to the difference between the nutrients added in the food, Nutfood, and those assimilated by the fish which were subsequently harvested, Nutfish:-
Nutenv = Nutfood - Nutfish
A combination of trash fish and pellets were used as feeds, and the C, P and N composition, as well as the quantities used were recorded. The weights and C, P and N content of the trout harvested were measured and nutrient loads to the lake computed on a per-kg-cage-fish-production basis.
The results are summarised in Table 14, and show that for every kg of fish harvested, the lake was enriched by 0.75 kg C, 0.023 kg P and 0.10 kg N.
A similar method was used by Beveridge et al (1982), based on published data on P content of feeds, FCR (Food Conversion Ratio) values, and P content of fish carcasses. In Table 16, total-P loads associated with intensive trout and tilapia culture have been calculated, using the feed formulations detailed in Table 12, and their associated FCR values (Table 15). The total-P content of trout and tilapia carcasses is taken from Ogino and Takeda (1978), Penczak et al (1982) and Meske and Manthey (1983).
The total-P load to the environment is variable, depending on the P content and the digestibility of the feed used. For trout, the most common FCR values for cage culture are 1.5–2.0:1, and thus total-P loads per tonne fish produced are 17–25 kg. For intensive tilapia production the usual FCR values are in the range 2.0–2.5:1. The exceptionally high FCR value for the Central African Republic diet is believed to have been due to poor O2 conditions (see Coche, 1982) and will not be considered here. Thus 23–29 kg total-P are added to the environment for every tonne of cage tilapia production. Total-P losses are therefore approximately the same for both intensive trout and tilapia production.
Estimates of total-P loadings from intensive land-based trout culture systems are given in Table 17. Most of the results are based on national surveys commissioned in European countries by EIFAC, and the enormous variation in the results (11–157 kg P tonne fish produced-1) is due to differences in system (pond/raceway/tank; hatchery/grow-out operation), feeding (floating/sinking; dry/wet; hand-fed/automatic feeders) and management practices (treatment/no treatment, prior to discharge), as well as sampling (daily/weekly/monthly) and analysis of effluents (dissolved/dissolved + particulate; total-P/ortho-P) (see Summary in Alabaster, 1982a). It is thus difficult to compare estimates for cage culture with these data. However, Ketola's (1982) results are based on careful measurements of inputs and outputs from one system, and show that trout fed on a standard, commercially available diet in the USA produce a load of 22.77 kg total-P per tonne production, which is within the range calculated above for cage trout production.
Unfortunately, there have been no similar studies of intensive land-based tilapia culture systems.
In estimating total-P loads from intensive cage culture, the feed fish wastes system has been treated as a black box with information restricted to inputs and outputs. However, no attempts have been made to quantitatively or qualitatively analyse the processes within the system which are involved in waste P-production. This is a necessary step prior to modelling, as recent reviews have shown that the form of P-wastes determines their impact on the environment (Lee et al, 1980; Sonzogni et al, 1982).
The various sources of P wastes in intensive cage culture are summarised in Fig. 17. Many of the parameters or processes involved can be quantified from empirical data, whilst others can be derived theoretically.
For intensive trout culture, total feed losses (dust and uneaten food) are estimated to be 20%, based on manufacturers' figures for dust (2% of feed) and estimates from various studies of 10–30% uneaten food (Collins, 1971; Hoelzl and Vens Cappell, 1980; Penczak et al, 1982). When FCR values for pond and cage culture are compared (Table 18), those for cage culture are at least 20% greater, thus supporting the argument that feed losses from cages are comparatively high.
Some P is leached from the food prior to ingestion. However, if we assume that feed is ingested within 3 minutes of being given to the fish, and if we assume ‘worst possible conditions’ (small pellet, high temperatures), then only 1% of the P in the feed would be leached out (Beveridge et al, unpublished data).
Using data from Penczak et al (1982) it seems that only 32% of P ingested (23% of feed given) is assimilated and utilised, the rest being either passed out in the faeces, or excreted in the urine.
The response of aquatic ecosystems to increases in P loadings has been the subject of intense debate for a number of years, and a wide spectrum of predictive models has been developed. The models are basically of two types: “dynamic” models, which may be defined as “mathematical representations of the key physical, chemical and biological processes governing algal growth” (Jones and Lee, 1982) or statistical models derived from large-scale surveys of lakes and reservoirs. The choice of appropriate model depends primarily on what it is to be used for, and the quality of the available data (Jørgensen, 1980). As stated in the Introduction, we wish to be able to predict the impact of intensive cage and pen culture on water quality (particularly phytoplankton numbers) so that comprehensive guidelines for the development of the industry can be established which take into account not only the effects of changes in water quality on fish production, but also other uses. The model (or models) must be readily useable without recourse to expensive and time-consuming data collection by highly-trained technicians. A simple model with few variables would therefore seem best.
The dynamic models range in complexity from simple 2 or 3 parameter type to the more complex models, such as CLEANER, developed by Massacheussets Institute of Technology, which has 40 variables. A recent study by Straskraba (1982) has shown that the simple predictive models are as accurate as the much more complex data-hungry models, since for every additional parameter considered, a further source of error is introduced. However, despite the fact that they give a great deal of insight into how aquatic ecosystems function, at this stage in their development they have been found to have limited predictive capabilities (Jones and Lee, 1982; OECD, 1982). Their development has also been restricted to temperate water bodies.
Statistical models based on empirical data were first described by Vollenweider (1968, 1975, 1976), and later developed by Dillon and Rigler (1974), Kirchner and Dillon (1975) and Jones and Bachman (1976) among others. All attempted to predict P concentrations in lakes and reservoirs through various mass balance equations, and to relate these to trophic state (productivity). These models have been calibrated and tested, verified and modified using a number of data bases: the United States Environmental Protection Agency's National Eutrophication Survey (USEPA 1978); the Organisation for Economic Cooperation and Development's survey of water bodies in 18 North American and European countries (OECD, 1982); the IBP global survey (Le Cren and Lowe-McConnell, 1980); and a survey of Southern Africa's lakes and reservoirs (Thornton and Walmsley, 1982; Walmsley and Thornton, 1984). The information is summarised in Table 19. Based on the predictive abilities of the various models, Dillon and Rigler's (1974) model has been chosen as the best available at the present moment. It has been widely tested using shallow and deep lakes and reservoirs in both temperate and tropical regions, and seems to perform best of all (Mueller, 1982; Thornton and Walmsley, 1982.)
Dillon and Rigler's modification of Vollenweider's original model states that the concentration of total P in a water body, [P], is determined by the P loading, the size of the lake (area, mean depth), the flushing rate (i.e. the fraction of the water volume lost annually through the outflow) and the fraction of P permanently lost to the sediments. At steady state,
where [P] is in gm-3 total P, L is the total P loading in gm-2 yr-1, z is the mean depth in m, R is the fraction of total P retained by the sediments, and ρ is the flushing rate in volumes per year.
A step by step approach to using the model has been adopted here.
Step 1: In order to determine the potential of a lake or reservoir for intensive enclosure, the productivity of the water body prior to exploitation must be assessed through measurement of the steady-state total-P concentration, [P]. With the exception of very shallow water bodies, temperate lakes and reservoirs are often stratified for much of the year and only mix twice during spring and autumn when there is little difference in temperature and thus density between surface (epilimnion) and deep (hypolimnion) waters, and when there is sufficient wind energy to induce mixing. During stratification, differences in [P] develop between the epilimnion, where P is utilised by algae, and the hypolimnion, where [P] is determined by sediment/water interactions rather than by the algal community. According to Dillon and Rigler (1974), Vollenweider (1976), and OECD (1982), the steady state [P] in northern temperate waters is therefore best determined at the time of spring overturn.
By contrast, tropical inland waters are either warm monomictic (mix once per year) or polymictic (cycle frequently) (Ruttner, 1963; Wetzel, 1975; Hill and Rai, 1982), and according to Thornton and Walmsley (1982), [P] should be taken as the measured mean annual total P concentration, [P] of surface waters.
Step 2: The development capacity of a lake or reservoir for intensive
cage and pen culture is the difference between the productivity of the
water body prior to exploitation, and the final desired level of productivity.
As stated above, [P] can be used as a productivity indicator.
However, it must be decided whether it is then mean annual algal biomass,
or the peak annual algal biomass, as measured by chlorophyll levels [ch1]
and 
respectively, that we wish to predict. Since fish are usually
held in cages throughout the year, it is the latter parameter which should
be considered.
The desired peak algal biomass is determined by a number of criteria, the most important being whether the water body is multi-purpose or single purpose (i.e. for fish culture alone). The multi-purpose nature of inland waters is impaired with increasing productivity - particularly if already highly productive (OECD, 1982) - and thus limits should be carefully set. However, it is difficult to find hard and fast guidelines as to recommended levels, since water resources vary in quantity and quality from country to country. For example, water used for drinking purposes should be as clean (i.e. free from toxic or noxious substances) as possible, and this is easiest to achieve when unproductive, unpolluted sources are used. However, in areas of high soil fertility highly productive water bodies may dominate and may have to be used for domestic supplies.
Recommended acceptable ranges and maximum permissible values of [P]
for water bodies with different uses are suggested in Figure 18 and Table
20. The [P] (mg m-3) values can be related to both (mg m-3) and
[ch1] (mg m-3), using the correlations derived by OECD (OECD, 1982) for
temperate waters, and these relationships are summarised in Table 21.
Note that three equations relate both [ch1] and [P] and and [P], and
that two equations relate to [ch1] and . The size of the data base
used also varies. The first equation in each case utilises unscreened
data. However, for the second equation data from lakes where light is the
limiting factor, due to heavy natural silt loads, and from lakes where
artificial aeration is used, are omitted. In all cases, the correlation,
r, is improved. The third equation uses data which has been further
screened, and from which lakes where N might be a limiting factor (i.e.
N:P ratios <10) are not included, and this further improves correlations.
Annual gross primary production Σ PP(gC m-2 yr-1) is related to both [P]
and to algal biomass by linear equations:-
∑PP = 31.1[P]0.54; r = 0.71; S.E. = 0.265; n = 49
∑PP = 56.5[ch1]0.61; r = 0.79; S.E. = 0.242; n = 49,
despite evidence of self shading and thus reduced algal biomass levels at high [P] (OECD, 1982).
Unfortunately, there are few data relating [P], algal biomass and productivity in tropical waters. However, in a recent paper Walmsley and Thornton (1984, in press) show that most southern African impoundments exhibit similar relationships to the North American and European OECD study lakes and reservoirs between [chl], [P] and orthophosphate [P]0:-
[chl] = 2.06[P]00.387 r = 0.81; n = 29
[chl] = 0.416[P]0.675 r = 0.84; n = 16
According to Melack (1979), three temporal patterns of algal biomass
and productivity exist in the tropics. Most water bodies show pronounced
seasonal fluctuations corresponding to variations in rainfall, river
discharges or mixing. Yet other water bodies exhibit little seasonal
variation, whilst a third category shows periodic abrupt changes from one
persistent (>10 generations) species assemblage and level of photosynthetic
activity to another persistent condition. However, there are
insufficient data to relate to either [P] or [chl].
The few available data relating [chl] to mean photosynthetic rate are summarised in Table 22, although due to the paucity of data and the range of units used, no relationship could be derived.
Step 3: The capacity of a water body for intensive cage and pen fish culture is the difference, Δ [P], between [P] prior to exploitation, [P]i, and the desired/acceptable [P] once fish culture is established, [P]f.
i.e. Δ [P] = [P]f - [P]i
Δ[P] is related to P loadings from fish enclosures, Lfish, the size of the lake, A, its flushing rate, ρ, and the ability of the water body to handle the loadings (i.e. the fraction of Lfish retained by the sediments, Rfish):-
The acceptable/desirable change in [P], Δ [P] (mg m-3), is determined as described above, and z can be calculated from hydrographic data obtained either from literature or survey work:-
where V = volume of water body (m3) and A = surface area
(m2) the flushing rate, (y-1) is equal to Qo/V, where Qo is the average
total volume outflowing each year. Qo can be calculated by direct measurement
of outflows, or in some circumstances can be determined from
published data on total long-term average inflows from catchment area
surface runoff (Ad.r), precipitation (Pr) and evaporation (Ev), such that
Qo = Ad.r + A(Pr - Ev) (see Dillon and Rigler, 1975, for further details).
The retention coefficient, R, can be determined experimentally by measuring the mean annual inflow and outflow [P], [P]i; and [P]o respectively:-
Using multiple regression analysis of data from temperate water bodies, Kirchner and Dillon (1975) found R to be highly correlated to the annual hydraulic loading, Q/A, such that:-
R = 0.426 exp (-0.271 Q/A) + 0.574 exp (0.00949Q/A), r = 0.94,
where Q = annual hydraulic loading (m3). Various other models have been developed for specific types of water body, such as oligotrophic or fast-flushing temperate lakes (Larsen and Mercier, 1976; Ostrofsky, 1978) and many of these have been critically evaluated by Canfield and Bachmann (1981). In view of their conclusions, it seems best to use different computations for R, depending on the type of water body being assessed, although of course choice will also depend on available information. Models are summarised in Table 23.
A similar, though less precisely defined relationship between R and Q/Aseems to hold for tropical lakes and reservoirs (Thornton and Walmsley, 1982) (Fig. 19). However, until the data necessary to define the relationship have been collected, temperate models must be used.
Lfish is largely in particulate form, and the proportion of the waste faecal and food P which contributes to the pool of dissolved P depends on many factors; the P content of the feed, diet composition, pellet shape, temperature, depth of water under the cages, presence/ absence of scavenging fish, etc. (Bienfang, 1980; Collins, 1983; Merican, 1983). Modelling of these wastes is in progress, and preliminary data suggests that Rfish > R. However, until such models are available, Rfish must be assumed to be the same as R in the first instance, and, subsequent to the introduction of cages and the steady state [P] having been reached (see below) it must be recalibrated:-
The response time of a water body to increases in P loading is a nonlinear function of the water residence time, t(M) [t(M) =1/ρ], and mean depth, z. The expected 95% response time, t(M)95, which is used as an approximation to the full response time, can be calculated from Fig. 20.
Step 4: Once the permissible/acceptable total P loading, Lfish, has been calculated, then the intensive cage fish production (tonnes y -1) can be estimated by dividing Lfish by the average total P wastes per tonne fish production (Table 16). A worked example is given in Appendix 1.
Before considering how to model the impact of extensive cage fish culture on the environment, the rationale behind using this method to increase fish production must be examined.
As discussed in Section 4.2, the rate of primary production in inland waters is dependent upon the availability of essential nutrients and light. Production in all other communities within the ecosystem is to some extent dependent upon primary production, and thus it is not surprising that Σ PP and annual fish yields, Fy, are related (Hrbacek, 1969; Henderson et al, 1973; Melack, 1976; Oglesby, 1977, 1982; McConnell et al, 1977; Hecky et al, 1981; Marten and Polovina, 1982; Adams et al, 1983).
Information on fish yields and productivity in tropical lakes and reservoirs are summarised in Fig. 21. The line which best fits the data is curvilinear, of the form Y = AeBx, and the correlation coefficient, r, is 0.64. There is a large amount of scatter in the data, which accounts for the low correlation value, and much of this variance is undoubtedly due to how the data were collected. However, there are several additional factors which must be considered. First of all we don't know the relative importance of other autochthonous sources of energy, such as periphyton or macrophytes, or the allochthonous inputs to the water bodies concerned. Both macrophytes and periphyton can make significant contributions to the total energy fixed in lentic water bodies (Moss, 1980), and although allochthonous inputs seem to be relatively unimportant in most lentic water bodies (Adams et al, 1983), they can be important in the energy budgets of small aquatic systems with low retention times, or those surrounded by swamps (Oglesby, 1977). In Bukit Merah reservoir, Malaysia, for example, more than 90% of the C cycled through the system is derived from allochthonous sources (Yap, 1983) thus leading to high production of detrivorous fishes and higher than expected yields per unit primary production.
Secondly, we don't know at what intensity the fisheries are being managed, or what gears are being used. A lightly exploited fishery (i.e. one operating well below maximum sustainable yield) would give low yields per unit primary production (Marten and Polovina, 1982). Finally, this plot does not take into account the types of fish being harvested.
The general shape of the curve is interesting, and suggests that at low levels of primary production, trophic transfer efficiences (production at tropic level 1/production at trophic level n-1) are low, whilst in highly productive water bodies transfer efficiencies are much higher. However, this is likely to be an artefact of the data pool used. Not only must the data vary qualitatively in terms of how primary production and fish yields were estimated, but also few highly productive water bodies were included in the analysis. Liang et al (1981) suggest that in fact the relationship is sigmoid, and that the data used here relates only to the lower portion of the curve. It is thus suggested that trophic transfer efficiencies are lower in highly productive waters.
On theoretical grounds, Slobodkin (1960) and others have suggested that ecosystem trophic transfer efficiencies should be around 10–15%. However, comparatively low transfer efficiencies of between 4 and 10% are common in freshwaters (Wright, 1958; Gulati, 1975; Rey and Capblancq, 1975; Coveney et al, 1975; Lewis, 1979). The efficiency of herbivore grazing is in part dependent upon phytoplankton quality - size, species, etc. (Zaret, 1980). However, in many instances the herbivorous zooplankton populations are heavily suppressed by predation, thus accounting for their failure to crop the major portion of primary production (Rigler et al, 1974; Jassby and Goldman, 1974; Kalff et al, 1975; Coveney et al, 1977; Lewis, 1979). Although there is a positive relationship between zooplankton and phytoplankton biomass, the ratio decreases with increasing productivity (McCauley and Kalff, 1981). Recent studies of zooplankton populations in temperate and sub-tropical lakes show that as productivity increases, zooplankton community composition shifts to dominance by microzooplankton (cilicates, rotifers, nauplii) which feed principally on bacteria (Gannon and Stemberger, 1978; Bays and Crisman, 1983). Thus in highly productive systems relatively more of the carbon fixed is diverted to the detrital pathways (Gliwicz, 1969; Pedersen et al, 1976; Wissmar and Wetzel, 1978), and by comparing observed with expected transfer efficiencies, it seems that as little as 30% of the phytoplankton production in lentic water bodies is grazed by herbivores.
By increasing grazing pressure through the stocking of microphyte-feeding fishes, part of the detrital supply could be converted directly into fish production, thus avoiding the energy losses associated with long food chains. Although fishes such as the tilapias and carps which feed at the base of the aquatic food web may have low transfer efficiencies when compared with organisms feeding at higher trophic levels (Borgmann, 1982), nevertheless at each successive step along the food web there are energy losses, so that fisheries which concentrate on capturing fishes at the end of long food chains have comparatively low yields (Jones, 1982).
An increase in herbivore grazing pressure will tend to reduce the average size of individual phytoplankters, whilst causing an increase in the turnover rate (Cooper, 1973) or relative production as it is generally called (Production/Biomass = P/B). Within limits this will stimulate the overall productivity of the system (Opuszynski, 1980).
A further reason for stocking inland water bodies with fishes is that in many tropical freshwaters, particularly in Asia and South America, not all trophic levels may be utilised (Fernando and Holcik, 1982). In such systems where cichlids or clupeids have not been introduced, the fish communities are of riverine origin and are not well adapted to the lacustrine areas of lakes and reservoirs.
The reasons for manipulating aquatic ecosystems through extensive aquaculture (fisheries, ranching, cage and pen culture) are summarised in Figure 22. By stocking with the appropriate species of fish which feed at the base of the food web, vacant niches in the system may be utilised and phytoplankton grazing encouraged, thus increasing the phytoplankton P/B, and reducing energy losses between autochthonous energy inputs and fish yields. Possible adverse effects will be considered in the discussion.
The principal species used in extensive enclosure culture are the tilapias (O. niloticus, O. mossambicus), although carps (H. molitrix, A. nobilis) and milkfish are also grown in this manner in some countries. Cages are more commonly used than pens. Since little research has yet been carried out into the diets of these fishes under extensive enclosure conditions, data from studies of food consumption under natural conditions and in fish ponds, and results from nutritional studies must be used to determine what foods are likely to be consumed, and the relationship between food intake and fish production.
The diets of the tilapias and carps are summarised in Table 24. O. niloticus like all other tilapias is principally a herbivore (Jauncey and Ross, 1982) and its diet under natural conditions is largely restricted to phytoplankton (Moriarty, 1973, Moriarty and Moriarty, 1973, 1973a). However, in highly stocked organically fertilised ponds, where the principal flow of energy is through the detritus pathways and where intraspecific competition for food can be severe, O. niloticus feeds and grows well on organic manures (Wohlfarth & Schroeder, 1979), although the principal nutritive value is not derived from the detritus itself, but from the micro-organisms which cover the surface of the particles (Kerns and Roelofs, 1977; Schroeder, 1978). O. mossambicus is more omnivorous, and it has been found to ingest a wide range of plant materials, as well as zooplankton, fish larvae and eggs, and detritus (Bowen, 1982). However, in cage conditions both species probably feed largely on phytoplankton, supplemented by detritus.
Studies on the diets of caged carps show that silver carp feed primarily on phytoplankton (8–100 um), whilst bighead carp consume phytoplankton, zooplankton and detritus in the range 17–3000 um (Cremer and Smitherman, 1980).
In the following Section, which deals with the potential production from extensive culture, most of the emphasis will be placed on cage tilapia culture which is the most common form of extensive culture. The use of pens and the culture of carps will be discussed in Section 4.6.
The food consumption of fishes can be summarised in the following equation:- C = P + R + F + U, where C = food consumption in energy terms (joules); P = energy used for tissue growth (including fat deposition, egg and sperm development); R = energy used for work (including body maintenance, digestion, activity); F and U = energy losses in faeces and urine respectively (Klekowski and Duncan, 1975). The amount of useful energy available to the animal, or assimilation (A) as it is generally termed, can be derived from
A = C - (F + U)
= P + R
Assimilation is often quantified in terms of assimilation efficiency (A):-
The A values have been found to vary in tilapias, depending on food source and temperature, from 45 to 55% (Table 25).
Many tilapia populations undergo diurnal migrations from the warm littoral regions they inhabit during the day, to the deeper, cooler offshore waters at night (Fryer and Iles, 1972; Bruton and Boltt, 1975; Caulton, 1975). Such behaviour has been shown by Caulton (1978) to have a considerable effect on the A in T. rendalli. At 18°C (average night-time temperature), A =∼48%, whilst at 30°C (average day-time temperature), A = 58%. However, fish held in floating cages are subject to little diurnal temperature fluctuation (± 1–2°C) and thus in a 25°–30°C annual temperature fluctuation we would expect no more than a 5% variation in A (from Caulton, 1982).
Only a portion of the energy assimilated is available for growth. Work done by Caulton (1982) has shown that T. rendalli can utilise approximately 0.5 A for growth, providing it can reduce its metabolic energy requirements by migration to colder waters at night. However, at a constant 28°C, only∼ 0.2 A is partitioned into growth, giving an overall food conversion efficiency (energy value of plant tissue consumed/energy value of fish tissue elaboration; P/C) of ∼ 10%. O. mossambicus fed on an algal diet, showed a higher food conversion efficiency of 16–22% at 25°C (Mironowa, 1974; in Fischer, 1979), and in the absence of any data, a food conversion efficiency which lies somewhere between the values for other species (15%) has been assumed for O. niloticus reared in cages. (N.B. This value is an estimate and ignores the effects of food quality, age, reproductive condition, etc. Fischer, 1979).
In theory, therefore, 10–15% of primary production could be converted into fish (tilapia) tissue. In Fig. 23, fish production is plotted against primary production. A water body with primary production of 1000g C m-2 y-1 would yield 1000–1500 g fish tissue m-2 y-1 or 1000–1500 tonnes km-2 y-1, assuming a food conversion efficiency of 10–15%, and a fresh fish carbon content = 10% wet weight (Gulland, 1970). By comparison, the fish yield of a typical tropical inland water fishery with a similar rate of primary production is around 6 tonnes (Fig. 21).
As discussed above, the difference between actual fish yields, and theoretically possible yields is huge, and there is a great deal of scope for improvement through ecosystem manipulation. According to the classical fisheries theories of Russell (1931) and Beverton and Holt (1957), the size of the exploitable fish stock is determined by four factors -recruitment rate, growth rate, fishing mortality rate and natural mortality rate - which operate as illustrated diagrammatically in Fig. 24. It can be seen that by (i) excluding predators and minimising the effects of disease on the natural mortality rate, by (ii) bypassing the factors that govern recruitment rate, through artificial stocking, by (iii) stimulating the P/B of primary producers and maximising the conversion efficiency of the system through the appropriate choice of species, by (iv) harvesting prior to the food conversion efficiency being adversely affected by age or reproduction, by (v) minimising energy losses through foraging, and by (vi) maximising the fishing mortality rate, fish yields per unit primary production could be maximised.
There are two principal methods by which the above policies can be achieved. Using conventional methods of stocking and fisheries management, it is possible to fulfill criteria (ii) and (iii), and to have a degree of influence on others. The rate of natural mortality can be influenced by an eradication programme of piscivorous birds and mammals (see FAO, 1983, for details of management of Chinese lakes) and by intensification of fishing pressure, which would eliminate losses through age. Increased fishing mortality would increase fish P/B. Fish which cannot breed in lentic systems, such as Chinese carps, could also be stocked, thus minimising the energy losses associated with gonad development, and egg and sperm production. Such management practices are most practicable in small water bodies.
In China, up to 15-fold increases in yields have been achieved through these methods (Tapiador et al, 1977; Liang et al, 1981; FAO, 1983) (Table 26). Assuming 0.04–0.06% average transfer efficiencies from primary production to fish yield prior to stocking (from Fig. 21), this would result in an increase to 0.6 – 0.9%. Transformation of Liang et al's (1981) data for intensively managed lakes near Wuhan, China, shows a range of conversion efficiencies from areal primary production to areal fish yields of 0.5 – 2.3% (gross), or 0.2 – 2.2% (net) (Table 27). However, these figures are probably overestimates, since organic fertilisers and supplementary feeds were used in most of the lakes.
By contrast, most of the criteria for maximisation of yields from primary production can be met using extensive cage culture and consequently, yields should be higher. As an approximation of the conversion efficiencies attainable, data from Almazan and Boyd (1978) for tilapia (O. aureus) yields vs primary production in inorganically fertilised fish ponds has been replotted in Fig. 25. The uppermost curves relate to fish yields assuming 10 and 15% conversion efficiencies, whilst the middle plot is Almazan and Boyd's data. The lowest curve represents fish yields from tropical lakes and reservoirs (Fig. 21 replotted). It can be seen that the tilapia yield curve is of the same form (Y = AeBx) as that calculated for the tropical lakes and reservoirs, but that the correlation (r = 0.91) is much better. For any given value of Σ PP within the range examined, yields are ∼ 20 times (18–24) better from extensively managed ponds than from average lake or reservoir fisheries. The conversion efficiency of primary production to fish yields varies from 1.4% in highly productive ponds, to 1.3% in relatively unproductive ponds, which is similar to estimates for extensively managed fish ponds in Malaysia (Prowse, 1972) and India (Sreenivasan, 1972).
However, it must be borne in mind that only ponds with Σ PP over a small range (420–1640 g C m-2 y-1) were examined. Fish yields will not continue to increase exponentially with increasing productivity, since at high productivity algae is inefficiently grazed (see Section 4.4.1 above). The turning point in the curve, where increases in Σ PP would begin to result in smaller increases in Fy probably occurs at Σ PP levels of > 2500 g C m-2 y-1, since Liang et al (1981) found that an expotential curve best described their data set which included Σ PP levels greater than this. The comparatively high yields at low levels of primary production in Fig. 25 are likely to be misleading, since no ponds with Σ PP 420 g C m-2 y-1 were studied. A logistic curve passing through the origin, as suggested by Liang et al (1981) is thus likely to best describe the relationship.
The yields from extensive ponds serve as a guide to the conversion efficiencies we might expect from extensively managed cages. Nevertheless, the two methods of extensive culture differ in several respects. Yields per unit primary production might be expected to be greater in cages, since predation and respiratory energy losses through foraging are likely to be higher in ponds. However, fish in deep ponds can move to cooler waters at night, thus conserving energy (Caulton, 1982). The ability of caged fish to graze algae may also be restricted, through reliance on a largely passive food supply. In view of this, and in the absence of any hard supportive data, conservative estimates of annual fish yields from extensive cage culture are probably between 1.0 and 3.5% of primary production (Table 28) which are higher than yields from managed reservoirs and lakes (Table 27). However, these values apply to ideal conditions (i.e. taking into account species and quality of fish stocked, stocking rate, mesh size, siting of cages, etc; see below) and must be used with caution.
Step 1 Determine the annual gross primary production, Σ PP, of the site. Since many tropical inland water bodies exhibit seasonality in the pattern of primary production (Melack, 1979), regular measurements may have to be made.
Step 2 Convert Σ PP to potential annual fish yields, using Table 28 and Figure 25.
Step 3 The actual organisation of planned production depends on a number of variables. The number of crops per year and the size of the fish at harvest should be decided on. If, for example, tilapia are being farmed, then two crops per year of 160g fish (6 fish kilo-1) may be desirable. However, seasonality of primary production may mean that one crop takes longer to grow. In order to reach target harvest size, the sum of primary production during the crop l growth period, Σ PPcl, should approximate that of crop 2, Σ PPc2, although this ignores possible change in the cropping efficiency of the fish at different algal densities, and may have to be adjusted in practice.
Semi-intensive cage and pen culture are the most common methods of enclosure culture and also, sadly, the most difficult to evaluate and plan. The principle of semi-intensive culture is that low quality feeds are given to the fish to supplement their intake of natural food. However, as recent work carried out in the Philippines by Escover and Claveria (1984, in press) shows, at any particular site management practices vary enormously, depending on size of farm, availability of feedstuffs, and costs (Table 29).
The carrying capacity of inland waters for semi-intensive culture depends on (i) the productivity of the water body and the amount of natural food available, and (ii) the quantity and quality of supplementary food used.
Step 1 Determine the annual primary production, Σ PP, of the site being considered, as described in Section 4.4.4 above.
Step 2 Calculate the potential annual fish yield, Fy, from the site using the information in Table 28.
Step 3 Calculate the average annual amount of the various feedstuffs being used, and the FCR, in order to determine the fish yield attributable to the supplementary food. The quantities of feedstuffs can be determined from survey work, whilst the FCR can be derived from the literature. The FCR values of some of the more common feedstuffs used in tilapia culture are given in Table 30.
Step 4 Calculate the total-P loadings associated with the use of supplementary feedstuffs, Lfish, and using Dillon and Rigler's (1974) model, calculate the increase in total [P] (see Step 3, Section 4.3). The increase in total [P] can be used to calculate increases in primary production, Σ PPfish attributable to fish culture, although this is likely to be < 10% total fish production (see Appendix 3).
Step 5 Estimate the fish yields due to Σ PPfish, using the conversion efficiencies detailed in Table 28. Calculate total fish yields from semi-intensive culture, ΣFy, as:-
Σ Fy = (a ΣPP) + (ΣFood × FCR) + (b ΣPPfish),
where a and b are expected conversion efficiencies of primary production to fish biomass (see Table 27) and ΣFood is total amount of feedstuffs added. A worked example is shown in Appendix 3.
The models detailed above for use in estimating the environmental impact and thus the carrying capacity of inland water bodies for various methods of cage and pen culture are at the initial stages in their development. Emphasis has been placed on cage culture, and the more commonly farmed species, such as the salmonids and tilapias.
The main problem areas associated with each model are summarised in Table 31. For intensive culture, the setting of desirable/acceptable water quality criteria is a major area of concern. Although the USEPA (1976), OECD (1982), and others have set management objectives - albeit tentative ones - these have been primarily concerned with minimising nuisance blooms in multi-use water bodies. However, a major, and as yet unresolved, area of difficulty lies in setting management objectives for water bodies where fish culture is the primary or sole activity, and where fish health is the most important consideration. As intensive fish production at a site increases, the overall water quality (turbidity, O2, free NH4, NO2 levels, etc) deteriorates, and the risk of fish mortalities increases. The relationship between production and risk must be exponential, since an increasing number of mortality factors come into play with decreasing water quality, and their combined effects are synergistic rather than additive (Figure 26).
The model ignores changes in plankton species composition which can be important, since some of the blue-green algae which thrive in intensive cage culture situations can cause off-flavours (see Section 3.3.2), although farms may be willing to endure periodic problems, which they can treat (providing they have access to seawater/clean running water) in return for higher production. Similarly, some mortality due to poor water quality and disease may be acceptable from an economic standpoint. Management objectives in water bodies used solely for fish culture are thus likely to be geared towards predicting acceptable, rather than desirable, water quality standards.
The exact nature of the relationship between water quality and risk is likely to be site specific, since many local risk factors require to be considered, including species being cultured, quality of stock, timing of stocking and the prevailing water quality conditions at the site, the distance between nursery and on-growing site, management methods (e.g. frequency of grading), algal community composition, etc. All of these factors can greatly influence stock mortality, but are in practice extremely difficult to quantify. Thus the setting of acceptable water quality objectives for fish culture is still a highly contentious area. In view of this, the values in Table 20 must be used with caution to set management objectives, and these should be amended through experience and in the light of information collected from environmental monitoring.
Estimates of P-loading from intensive cage operations, Lfish, are likely to be revised in the near future as data on the nature of the wastes and bioavailability is published.
The model is restricted in use to P-limited water bodies, although most lakes and reservoirs fall into this category. For other types of water body, correction factors or modified P-algal biomass/primary production relationships may have been derived (e.g. Hoyer and Jones, 1983, for light limitations). The model is most applicable to small, well mixed water bodies, or to sites where the cages are widely dispersed. Cages sited near a lake or reservoir outflow may have much less impact on the water body than predicted by the model.
The overall predictive error associated with the type of model used
above is large (see Reckow, 1983, for review) and seems to be principally
due to the prediction of [chl] and from [P] (OECD, 1982). According
to Reckhow (1983) estimation of [P] from watershed characteristics and
hydrological variables often involves errors of ± 30%, whilst the OECD
(1982) data suggests the errors to be nearer ± 20%. Estimation of
or [ch1] from [P] involves further errors of around ± 35% (calculated from
OECD, 1982). The total error involved in predicting [chl] or is
thus around ± 55–65%. Although the magnitude of error involved seems
enormous, predictions should still be good enough to act as a management
guide to permissible levels of intensive fish production, which can be
adjusted in the light of water quality data collected when the farm is in
operation. The importance of instigating a water quality monitoring
scheme cannot be stressed too highly.
The model used for extensive culture is also based on a number of untested assumptions, and therefore the conversion figures of primary production to fish biomass must be treated with care. The conclusions are based on tilapias, although data for other phytoplankton feeders, such as the silver carp, are similar (Opuzynski, 1980). Zooplankton feeders, such as bighead carp, probably convert primary production into fish biomass more inefficiently, and for this reason have been used in attempts to control eutrophication (Yang, 1982). However, further knowledge on the effects of increased predation on particular trophic levels is required, since it seems that uncontrolled zooplanktivory can lead to increases in phytoplankton biomass (Elliott et al, 1983).
Efforts to estimate optimum stocking conditions from oxygen and food supply data are woefully inadequate at present (Appendix 4), and even assuming worst possible conditions (high temperatures, low flow conditions, small fish, increased metabolic demands following meals, etc) give stocking densities which are 5–20 times greater than used in practice. In the Philippines, stocking densities in extensive cages are around 1–10 kg m-3, depending on the productivity of the site. Appropriate stocking levels therefore must still be determined on a trial and error basis.
The preliminary stocking models illustrate the importance of water flow through the cages in maintaining food and oxygen supplies, and suggest that mesh sizes should be kept as large as possible, and that cages should be sited as far apart as possible in order to minimise the effects of the structures on current flow (see Fig. 27). In Selatar Reservoir, where extensive cage culture of bighead carp is carried out, cages are sited in this manner (Fig. 28).
Not surprisingly, since it is a hybrid of both the extensive and semi-intensive models, the model suggested for semi-intensive cage culture involves the errors associated with both. It is also difficult to collect information on quantities and qualities of feed being used, and in the absence of hard data, even more difficult to assess their dietary importance when being used as supplementary feeds. Nevertheless, even using the existing model, overexploitation should be reduced, and the typical pattern of lake and reservoir development (Fig. 10) changed to one which minimises financial risk to those who are most vulnerable (Fig. 29).
All the above models are concerned with cage rather than with pen culture. At present, pen culture is of much less importance and is largely restricted to a few countries in Southeast Asia (see Section 1.3). It is also only used for extensive and semi-intensive culture and may not be suitable for all fish species. Because fish kept in pens have access to the benthos, the conversion of primary production to fish biomass is likely to be higher, although it is difficult to estimate by how much until comparative studies are carried out. Preliminary data from the Philippines suggests that production of tilapias in pens may be as high as 800 g m-2 month-1 without supplementary feeding (Guerrero, 1983), which is six times greater than production of tilapias grown in cages with some supplementary feeding in the same area, during the same period (Table 32). Stocking densities were, however, different. The major drawback of this method seems to be in harvesting, and Guerrero (1983) recounts how only 15% of the fish stocked in the pens were recovered. Nevertheless, in view of these preliminary figures, a great deal more research is warranted.
