Welcomme R. & Halls A.
Renewable Resources Assessment Group, Department of Environmental Science and Technology Imperial College, Prince Consort Road, London, SW7 2AZ, UK E-mail Welcomme@dial.pipex.com
Key Words: fisheries, rivers, floodplains
Much work has been done on the flow requirements for maintenance of fish populations and fisheries in temperate rivers but few equivalent studies are available upon which to base criteria for the management of water regimes for fishes and fisheries in large river systems, particularly in the tropics. Fish in such rivers are heavily influenced by flood regimes that fluctuate naturally from year to year. Recently increasing pressure on water for a wide range of uses other than fisheries has led to damming, river training, water abstractions and water transfers that have substantially altered flood regimes of rivers throughout the world. Such alterations usually have resulted in the loss of fish production and biodiversity. The current emphasis on sustainable development and biodiversity conservation is leading to efforts to mitigate negative impacts of these changes through planning for altered river flows and the release of artificial floods from dams or polder sluices. The typical flood regime contains several characteristics that may influence the recruitment, growth and survival and growth of the individual fish species. Understanding of these characteristics will help determine flow criteria for the maintenance of floodplain fish faunas and design appropriate flood curves that maximize benefits from the water available.
The growing human pressures for water, both as a resource in its own right and for the many other functions that it is called on to provide, impact strongly on the quantity and quality of water available in any river system. Water abstractions and transfers alter the amount of water in the system. Transversal damming of the channel, construction of longitudinal levees and river training structures and the poldering of river floodplains change the form and function of the river (World Commission on Dams 2000; Jackson and Marmulla, in press). Changes to the landscape such as de-forestation, land clearances for agriculture and wetland reclamation may also influence the nature and timing of the hydrograph. The function and structure of floodplain rivers are also conditioned by the pulses of nutrients and alluvial material that vary according to the type of hydrological regime (see for example the PULSO model for the Parana River - www.Neiff.com.ar).
Winemiller (2004) identifies three major types of river according to their hydrology, temperate with aseasonal (seemingly random) flood pulses, temperate with seasonal flood pulses and tropical with seasonal flood pulses. Fish have evolved physiological adaptations, life history strategies and spawning and feeding behaviour to cope with these differing types of fluctuating flow conditions in rivers (see Lowe McConnell 1975; Welcomme 1979 and Bunn and Arthington 2002 for reviews). Through these adaptations, different species are able to respond to changes in flow in different ways. As a result, the relative abundance of species forming riverine fish assemblages changes in response to natural variations in flood regimes between years. For example, rivers, particularly those with highly variable annual hydrographs, appear to have separate components that are adjusted to years of high flow and years of low flow. In years when the floodplains flood normally the high flow elements predominate and in years when the floodplains do not flood the low flow elements are more abundant. This variability may arise from separate species that are adapted to low flow and high flow, as in the Niger (Dansoko 1975; Dansoko, Breeman and Daget 1976; Welcomme 1979; Quensiere, Benech and Dansoko 1994). They may also possibly arise from variation within the genotype of one species, that has both migratory and static elements, as appears to be the case in several European cyprinid species such as roach (Rutilus rutilus L. or Bream Abramis brama (L.) that have become adapted to the static conditions of canals and regulated rivers but are migratory under more natural conditions (Stott 1967; Linfield 1985; Lucas and Baras 2001). Species, or genotypic variation within species, persist under natural variation in the short term but may be threatened by long-term alterations to flow regimes to which they are less well adapted.
The impacts of changing hydrological regimes on fish populations were early classified by Tennant (1976) using the Montana method and are regularly assessed in north temperate rivers using instream flow incremental methodologies and the related Physical Habitat Simulation System (PHABSIM) [see Bovee (1982) for methodology and Gibbins et al. (2001) for an example of the application of the methodology to a reservoir and water transfer system]. Instream flow methodologies have been used in many temperate zone countries to determine legal discharge requirements for the protection of fish and invertebrate faunas of rivers. These methods however are limited to relatively small systems and deal with instream or main channel processes and have little capacity to predict the impacts of changes in flow on floodplains. They have also been elaborated primarily for a relatively limited group of fishes, the salmonids and have not been developed to deal with the far more diverse and com- plex fish communities of large rivers. Some recent attempts have been made to link the productivity of larger, floodplain rivers to their flow characteristics, for example the DRIFT methodology used in South Africa (King, Brown and Sabet In Press 2003); Arthington et al. 2003) and the Benchmarking Methodology (Brizga 2000) and other holistic methodologies developed in Australia (Arthington, 1998; Arthington et al. 2003). None of the methods developed so far directly address the problems of the impacts of changing flow regimes on fish catch in the types of multi-species, multi-gear fisheries so common in large rivers, especially those of the tropics.
The biology and ecology of fish in large rivers are strongly linked to the annual hydrological regime in the main channel and the regular flooding of the associated floodplains (Welcomme 1985; Junk, Bayley and Sparks 1989). Current pressures on water from other users, notably agriculture, means that there is an increasing trend to control hydrological regimes. Such interventions almost inevitably act to the detriment of living aquatic resources and fisheries. Losses of fish catch below dams and other river regulating structures are now known to be significant and represent a considerable loss of food and income to the societies exploiting them (World Commission on Dams 2000). Water abstraction and transfer schemes can also induce changes in hydrological regimes that are potentially damaging to fish (Davies, Thoms and Meador 1992; Bunn and Arthington 2002). Impacts on fish and fisheries of schemes that change the form and function of the river and the hydrograph can be anticipated in project planning. For example, ensuring that adequate water is maintained in the system at all times to protect the species of major interest to the fishery or for conservation (environmental flows) can keep losses to a minimum. In some circumstances, releases of water from upstream dams or through the sluice gates of enclosing polders can simulate a flood and this approach is being increasingly advocated to compensate for the highly regulated state of some systems. For example the Phongolo River floodplain has been managed by artificial releases of water from the Phongolopoort dam (Heeg, Breen and Rogers 1980; Weldrick 1996), artificial releases have been tried in Thailands Pak Mun dam to encourage migration and breeding in the river downstream (Jutagate pers comm.), systematic releases are planned on the Colorado river to aid in the conservation of the native squaw fishes and releases are planned for the rehabilitation of the Dyje floodplain in the Czech republic (Lusk, Halaeka and Lusková 2003). However, detailed knowledge of the form and function of the river system and of the responses of the fish species are needed for such planning to be effective. Such detailed knowledge of individual systems is generally lacking. As a result, control of the amount of water in the system is often pursued uncritically according to the needs of the major user, usually agriculture, rather than according to the requirements of the fish population and the fishery. This paper is intended to review some of the aspects of hydrological regimes that influence fisheries and that need to be taken into consideration when recommending ecological flows or artificial flow releases. It addresses particularly temperate and tropical rivers with seasonal flood pulses although many of its conclusions apply to other types of flood pattern. It builds on the ideas expressed by Poff et al. 1997 and Welcomme and Halls (2001) by synthesising information on the impacts of various characteristics of flooding on the different aspects of fish ecology and fisheries. Of necessity some of the suppositions are based more on theoretical speculation than on hard facts as floodplain research is still at a very early stage and detailed information of the behaviour of most species is not available. However, enough knowledge exists to derive preliminary guidelines for the best way to conserve fish faunas through environmental flow scenarios or water releases to simulate floods.
INFLUENCE OF HYDROLOGICAL REGIMES ON FISHERIES
The flood is important for most species of fish because the flooding of the lateral plains increases the area of food rich habitat and shelter from predators and provides ideal sites for young fish to develop and grow (see Welcomme 1979 for review). Older fish too profit from the improved feeding opportunities to lay down sufficient fat to permit them to survive the stresses of the dry season and to complete reproduction. Some species, usually predators, complete their life histories within the main channel of the river and rarely, if ever, venture onto the floodplain. The abundance and biomass of floodplain dependant species fluctuates from year-to-year depending on the strength of flooding. This is believed to reflect greater reproductive success, survival of fry and growth of fry and adults during years of better flooding. The greater biomass of fish in the system is reflected in fisheries catches. Many authors have found correlations between catches in year y and the intensity of flooding (usually represented by HI1) in the same or in preceding years - y-1 or y-n (see Stankovic and Jankovic 1971) for the Serbian Danube; Krykhtin (1975) for the Amur R.; Holcik and Kmet (1986) and Holcik (1996) for the Slovakian Danube; Moses (1987) for the Cross R.; Novoa (1989) for the Orinoco R.; Quiros and Cutch (1989) for the la Plata system; Payne and Harvey (1989) for the Pilcomayo R.; Lambert de Brito Ribeiro and M. Petrere (1990) for the Amazon; Welcomme (1979) and Lae (1992) for the Niger; Christensen (1993) for the Mahakam R.; Baran, Van Zalinge, Bun et al. (2001) for the Dai fisheries of the Mekong R.; and de Graaf, 2003 for floodplain fisheries in Bangladesh. Similar responses are found in estuarine or even coastal marine systems. For example Loneragan and Bunn (1999) found close correlations between high river discharges and the production of coastal fisheries in a Queensland river. Initially many of the earlier authors such as Krykhtin (1975), Welcomme, (1979) and Holcik and Kmet (1986) found the strongest correlations between the catch in year y and the strength of flooding in years y-2 to y-5 reflecting the time taken for the large fish forming the bulk of the catch to recruit to the fishery. However, more recent workers (Lae 1992; Halls 1998) have found correlations to be generally with the floods of the same year. This shortening of response time between the flood event and the fish catch is due to the fishing-down process, whereby fish are recruited into the fishery in their first year in todays heavily exploited fisheries (Welcomme 1999; Albaret and Lae, 2003). Some authors have also found correlations between catches and the amount of water persisting in the system over the low water period, notably University of Michigan 1971 and Quiros and Cutch 1989, however, best correlations were usually with the indices of flooding. All of this argues that in normal, humid rivers the flood component of the hydrological regime is the most important. The situation in arid rivers has been less well described although, even during the drought years of the Niger River, when the system was in an arid phase, good correlations with the strength of flooding were still obtained (Lae 1992).
Welcomme and Hagborg (1977) as a generic model, Moreau (1980) for fluctuating lake/river systems of Madagascar, Morand and Bousquet (1994) for the Central Delta of the Niger, Halls (1998) and Halls (2001) for the Puntius sophore fishery of Bangladesh, have modelled the processes regulating within year and year-to-year abundance and biomass of fish populations. The models employ empirical relations between recruitment, growth, mortality and fish density based on basic parameters of river fish dynamics, the main driver being water height. These models simulate biomass in response to hydrological conditions (Figure 5). They are based primarily on the dynamics of black and grey fish species (sensu Regier et al. 1989) that spawn on the floodplain and whose fry are assumed to have survival and growth closely correlated with the intensity of flooding. White fish species, which migrate upstream to breed in the channel and whose fry drift downstream with the current and are eventually washed onto the floodplains, may well have a different dynamic, especially with regard to survival and growth during the earliest, drifting phases. Little is understood of the dynamics of fish larvae in the drift and how they are affected by differences in main channel discharge although they are assumed to behave as other species do once they have entered the floodplain.
THE HYDROLOGICAL REGIME
Large rivers generally have one or more pronounced flood events during a years cycle although in most temperate and some tropical rivers various types of flow regulating structure have modified these to a point where the natural regime has been largely suppressed. The hydrological regime can be described as a curve (Figure 1) that has a number of characteristics, each of which, potentially, has an effect on the various fish species that comprise a fishery. The phase of the regime that exceeds bankfull and either totally or partially inundates the floodplain is referred to as the flood.
Figure 1. Various parameters of a flood curve having biological significance
The curve has mostly been defined for fishery purposes by an index (Hydrological Index or HI) that serves to indicate the relative magnitude of the flood or the low water phase during any one event. University of Michigan et al. (1971) first proposed this series of indices and since then many fishery workers have used them. Hydrological indices have been formulated from a number of indicators, including rainfall over the basin upstream of the floodplain, water height or discharge at selected gauges or evaporation from the floodplain. See Stankovic and Jankovic 1971; Welcomme (1979); Moses (1987); Holcik and Kmet (1986); Novoa (1989); Payne and Harvey (1989); Lae (1992); Christensen (1993). Typically an index will sum:
the area of the flood curve above the bankfull line, as an indicator of the intensity of flooding (HI 1), or
the area below the bankfull level and the curve defined by the depth of the residual water, as an indicator of the amount of water remaining in the system during the dry period (HI 2) (Figure 2).
Figure 2. Flashiness of rivers as a function of basin area exemplified by three rivers from the Chari-Logone River system
Flood indices describe the total amount of water during the period in question but in their existing form say little about the duration and amplitude of the flood. Thus, the same HI could describe a short but deep flood or a long but shallow one. Equally they say nothing about the other parameters of smoothness, rapidity of change or timing. The information needed to derive these parameters is often available from the water height record but is difficult to integrate into a descriptive model that permits comparison between years.
ROLE OF ASPECTS OF THE HYDROLOGICAL REGIME ON RIVER FISH ECOLOGY
This section examines in more detail the significance of the various parameters of the hydrological regime identified in Figure 1.
The timing of the flood is important to many river fish species because of the synchronisation between physiological readiness to spawn and the flood phase. Most species of river fish have defined breeding seasons centred on a particular flood phase (see Lucas and Baras 2001 for a detailed review of migration). Migratory, whitefish species (sensu Regier et al. 1989), such as Prochilodus (Bonetto and Pignalberi 1964; Bonetto et al. 1971) or the cyprinids of the Mekong are especially sensitive to the timeliness of the flood because they begin their migration from their downstream feeding habitat during the dry season and so time their migration as to arrive at the upstream spawning site before, or contemporaneously with, the rising flood (Fuentes and Espinach Ros 1998). Such species may mature during migration or at the upstream site, postponing the last stages of maturation until the waters begin to rise. By contrast (Humphries and Lake 2000) found that the species present in the Murray Darling system were unlikely to rely on discharge as a cue for final maturation and spawning so the generalisation that all river spawning behaviour is linked to the hydrological regime may be incorrect.
Total spawners (sensu Lowe-McConnell 1975), such as many characin, cyprinid and siluroid species, tend to have semi-pelagic eggs and larvae that enter the drift. Some hint of the complexity of the drift process is given by the work of Fuentes (1998) who showed that predatory species such as Salminus and Pimelodus migrate further upstream than the prey species so that, in drifting downstream from upstream, their larvae achieve a size at which they are able to feed upon the prey species as they enter the drift in their turn. Little is known as to the flexibility of such behaviour and its tolerance to substantial temporal displacement of the rising flood phase. Equally important, but little understood, are the population dynamics of the drifting fry with respect to survival, growth and distribution under different flood regimes. However, it is clear that accelerated flows may result in the drifting fry being swept past their destination and that flooding failure in floodplain nurseries will result in the loss of a whole year class of fish Gaygalas and Blatneve 1971; Fuentes 1998). In such species, modified hydrographs and artificial flood regimes must fulfil two requirements. First they must be sufficient and timely enough to induce spawning up-river. Second they must be extensive enough to ensure the flooding of the nursery floodplains downstream.
Timing is also important over shorter periods. In the Mekong and possibly other systems, migration and reproduction are closely linked to the lunar cycle (Sao Leang and Dom Saveun 1955). Inappropriate manipulations or modifications to the hydrological conditions during this cycle may not favour breeding in such species, especially in the flashier regimes of upstream reaches.
Grey fish (sensu Regier et al. 1989) including many small cyprinid and characin species may also be influenced by the flood as many of these are total spawners with a defined breeding season during the rising flood that so times the release of eggs as to enable the fry to be washed onto the floodplain by the advancing waters. Other grey and blackfish species such as the cichlids and smaller siluroids are partial spawners and small-brood spawners (sensu Lowe- McConnell 1975) that may breed on several occasions throughout the flood season or even into the dry season and are thus better adapted to changes in the timing of flow regimes.
Timing of floods is also important for climatic reasons. In many temperate and sub-tropical regions the flood coincides with rising temperatures of spring and summer (thermal coupling). This favours the growth of young fish by increasing the amount of food available and the rate at which it can be metabolised. Delay of flooding until late summer or early autumn in most rivers would result either in failure of the fish to spawn or in poor growth and low survival of the young fish due to the lower floodplain productivity in the cooler season. In some rivers such as the Murray- Darling in Australia and the Okavango in Botswana the downstream flood occurs during winter (thermally decoupled) and floodplain dependent production is rel- atively low in both these cases. There is little evidence for floodplain use by any life history stage of any fish species in the Murray-Darling system (Humphries, King and Koehn 1999), which may be due to the limited advantage of occupation of the floodplain during the cooler time of year. In some temperate systems occupation of the floodplain or the anabranches and backwaters of the main channel in winter appears more as a refuge from high flow than a feeding and breeding migration (Holcik 1988).
In natural systems floods may be interrupted by one or more drought periods. Discontinuities are also induced in regulated systems when the primary user places demands on the water that interrupt the smooth progression of flooding. Such discontinuities may be particularly damaging to white and grey fish total spawners, which may spawn during the first flooding but whose eggs and larvae are then unable to colonise the floodplains because of the temporary recession of the waters. Black and grey fish multiple spawners are less likely to be affected by such discontinuities but may lose one or more broods when the floodplain dries during the recession.
The smoothness of the flood is a measure of the steadiness of the rise and fall of the waters. It is the inverse of flashiness, which is the rapidity with which the river responds to local flood events. As smaller streams respond only to rainfall on their immediate basin they are extremely flashy. As the basin area increases the river tends to average out the rainfall over its surface and thus becomes less and less conditioned by local events (Figure 3).
Fish faunas of smaller rivers and low order streams must have reproductive and shelter seeking behaviours that are adapted to sudden changes in the discharge if they are to survive. However, species living in higher order systems are usually better adapted to smoother flood curves. The smoothness of the flood curve is particularly critical for total spawning white fish, as temporary recessions can interfere with larval drift in the same way as discontinuities in flooding. For example Nikonorov, Maltsev and Morgunov (2001) found that there are no important spawning grounds for sturgeons left downstream of the Volgograd reservoir in the Volga River due to the sharp fluctuations in water level resulting from the operation of the power station. The fluctuations cause mass destruction of sturgeon eggs and oocytes were resorbed in 30 percent of female sturgeons. Severe fluctuations in level also pose potential difficulties for marginal spawners and some classes of nest builders such as T. zillii, which can repeatedly move its eggs to new nest sites as water levels rise. Excessive, rapid variation in level can strand attached egg masses of the marginal spawning phytophils resulting in the failure of that batch of spawn. Equally retreating waters could expose nests leaving the eggs and fry to desiccate. Similar arguments apply to many of the invertebrates that serve as one of the major food sources for the growing fish.
Figure 3. Flashiness of rivers as a function of basin area exemplified by three rivers from the Chari-Logone River system
RAPIDITY OF CHANGE
The rate of the rise and fall of the water level is potentially critically important for many organisms. Overly rapid changes in level can affect fish more directly. During the rising waters rapid increases in level can submerge nests of bottom breeding species to too great a depth. Tilapias (Oreochromis, Sarotherodon and Tilapia species), for example, will tolerate only a narrow range of depths and substrate types for their nests. If the water is too deep, turbidity and low light levels do not permit them to complete their breeding. The rapid currents associated with such transitions in water current can sweep larvae and eggs of phytophilous species that deposit their eggs on the margins of floodplain and species with pelagic and semi-pelagic larvae in the main channel past their appropriate destination. During falling waters an overly rapid retreat of the flood is commonly assumed to increase the risks of stranding of fish in the temporary pools and channels of the floodplain resulting in unduly high mortality at this critical season.
The amplitude of the flood reflects the difference between the water level at low water and the maximum level reached during the flood. The higher the flood the greater the area of floodplain submerged. This means that the area available for nutrient recycling according to the flood pulse concept is greater (Junk et al. 1989; Junk and Wantzen 2004). Deeper (higher amplitude) floods produce greater flooded areas that can provide spawning sites, food and shelter for the fish.
The influence of amplitude on fish with drifting larvae is less easy to speculate upon, as the factors affecting survival and growth during the earliest drifting phases is generally unknown.
In some species such a Prochilodus and Semaprochilodus adult fish may be stranded in floodplain lagoons that are isolated from the river. Those closest to the river are connected yearly during flooding but lagoons at greater distances are connected less frequently and only during floods of greater amplitude. This accounts for the correlation between Prochilodus abundance and flood intensity in the Orinoco found by Novoa (1989) and in the La Plata system by Quiros and Cuch (1989). Periodic higher floods would therefore renew the fish and other faunas of lagoons that are more separated from the main channel.
The duration of flooding (measured from bankfull on the rising flood to bankfull at drawdown) influences the time available for fish to grow and for them to shelter from predators. As such, longer duration of flooding extends the growing season resulting in heavier fish that have a greater potential to survive the following dry season and an improved reproductive potential.
Duration of flooding may also affect the floodplain vegetation. In the Mekong system, flooded vegetation is adapted to a normal flood cycle and substantially longer floods lead to die-off and rotting. This in turn contributes to de-oxygenated conditions in the system. Similarly changes in flooding patterns can alter the viability and composition of floodplain forests as with the disappearance of the red gum in parts of Australia (Bren 1988) and Acacia species in the Pongolo system in South Africa (Furness 1978).
RELATIONSHIP OF AMPLITUDE TO DURATION
Because both amplitude and duration can have positive and negative effects on the dynamics of the various fish species, the optimal flood for any group of species probably lies in a compromise between the two. This can be expressed as a ratio FR = Amplitude/Duration. Any volume of water available for environmental flows or constructed floods can have a number of ratios depending on the way in which the water is released (Figure 4). Models of the dynamics of floodplain fish communities (Welcomme and Hagborg 1977; Halls, Kirkwood and Payne 2001) as well as Weldricks 1996 specific model for the Phongolo R., can shed some light on how the two com- ponents of the flood interact. Because different species respond differently to different types of flood regime, a correct balance between these various factors for all fish species may be difficult to achieve through a standardised flood repeated annually and a range of flood types over a number of years may be more suitable.
Figure 4. Configuration of different flood regimes having the same Hydrological Index (HI) but different Flood Ratios (FR) where FR = Ax/dx
Environmental flows and the constructed floods associated with them call for a manipulation of the amounts of water in the river. Very often this will be a determined volume negotiated with other users of the resource as reserved for the needs of the living aquatic organism. It is then essential to make the best use of this water. Given that some degree of flooding of the floodplain is needed to secure the survival of many of the species comprising river fish communities, the relationship between the amplitude and duration of the inundation is critical. In these circumstances a long flood of low amplitude will produce a smaller flooded area for a greater duration, which means that reproductive success and fry survival may be lessened but that growth may be enhanced. However, if the flood is of high amplitude but too short a duration reproductive success may be higher but the young fish may not have sufficient time to grow and store sufficient fat. This would increase later losses through predation, as the smaller fish are more vulnerable and would also lower survival through the prolonged dry season as energy reserves may prove insufficient. Density-dependent mortality might also rise as the larger fish compete for reduced trophic resources.
Baran et al. (2001) suggest that amplitude may be more important than duration, at least for the growth of the floodplain spawning Henicorhynchus spp. in the Great Lake area of the Mekong, mainly because of the improved influx of nutrient rich silt bought in by the greater volume of water. On the other hand the model of Halls et al. (2001) suggests that duration may be more important because of the improved growth of the fish stock (Halls and Welcomme, in press). Unfortunately there are very few analyses of catches by floodplain fisheries that have been carried to the level of detail needed to resolve this question.
Extreme flood events
At intervals flood patterns can deliver extreme events that may challenge the capacity of the physical and living components of the ecosystem. Such extreme floods have tragic consequences for human populations whose occupation of the riparian zone of the river is adapted to more normal events. Living aquatic organisms can be severely affected by both abnormally high and low discharges. High discharges can wash away adult and juvenile fish, especially in rivers that have been hard engineered to contain flow in the main channel. Similarly, drifting eggs and larvae can be washed past suitable floodplain nurseries and lost to the population. Extremely low flows may operate mainly on water quality. They can lead to deoxygenation of the water through natural processes or through the failure of self-purifying mechanisms to correct human induced euthrophication (see articles by Szmes and Leibman and Riechenbach Klinke in Leipolt 1967). In extreme circumstances low flows can lead to desiccation of much of the riverbed and of an increased percentage of floodplain water bodies.
INFLUENCE OFWATER LEVELS IN THE DRY SEASON
The dry season is a period of great stress to the majority of river fish species. At this time most species are confined to the main channels of the river although some specialists can survive in permanent floodplain waterbodies. Variations in water level at this time can have a great impact on the extent and nature of various habitats for a range of organisms including fish, Puckridge et al. (1998) and can influence the amount of and access to spawning substrates and dry-season refugia such as riparian vegetation. Flow may cease in the main channel and deoxygenated conditions may appear both in parts of the river channels and in the floodplain waterbodies. The numerous individuals generated during the flood have to find space in the much-reduced environment (on the floodplain itself the water volume during the dry season may be less than 5 percent of the volume during the flood). Many species seek refuge in tributaries and in deep pools within the main channel, thus conservation efforts have to be directed at maintaining adequate water in such habitats. On the floodplain, insufficient channel flooding can result in the permanent waterbodies becoming desiccated and their fish populations defunct. Many species feed little during the dry season, an effect that Lowe-McConnell (1985) termed the physiological winter.
Conservation measures should seek to ensure that adequate water is provided so that a number of floodplain water bodies and the refuge areas within them are maintained with adequate water in them throughout the dry season. Fish are at their most vulnerable to the fishery and other predators during the low water period, so both main channel and floodplain refuges should be protected by law against illegal and excessive fishing. The models of Welcomme and Hagborg (1977) and Halls et al. (2001) indicate that the dry phase is limiting to population densities in most unregulated systems, acting as a sort of filter through which the population has to pass to survive into the following year. However, the fact that the better population densities created by good flood years are still detectable in catches as much as five years later implies that the fish assemblages have some type of memory that enable years of good recruitment and growth to persist for a period despite the intervening dry seasons.
Stabilizing river flows to an almost constant discharge throughout the year may appear more efficient than retaining a pronounced flood pulse in that it would avoid much of the drawdown mortality and apparently lead to more stable fish stocks. It would favour fish species that are repeat spawners and are able to survive in the main channel alone (Lae 1995). The alternation between dry and wet phases confers an advantage in terms of overall aquatic productivity in fluctuating systems, such as flood rivers and lakes, as compared to more stable systems (Junk et al. 1989; Junk and Wantzen 2004). The advantage of the flood cycle to activities other than fishing, such as drawdown and irrigated agriculture, cattle grazing, wildlife is such that it cannot be ignored in planning for sustainable use of such land-water interface zones.
The question of what comprises the optimal relationship between the duration of the flood and the period when the river is separated from the floodplain during the drawdown remains unresolved. Models provide information on the dynamics of fish populations under different regimes of low and high water (Figure 5) but assume the flood as a feature of the model. Generally the longer the flood-phase the shorter the period of low water with its attendant high mortality.
Figure 5. Contour plots of equilibrium yield (t) for P. sophore generated by the FPFMODEL of Halls, (1998) for different combinations of mean flood and dry season water height and for a range of fishing mortality
Rivers are used for a number of human functions other than fisheries and the needs of high economic profile activities, such as power generation, frequently cause conflicts between abstractive industries and the water requirements of the fish and of fisher communities. Many of the current provisions for flow and flood regime control are inappropriate to fisheries in that the flood is managed for the conflicting objectives of fisheries, animal grazing and agriculture (particularly drawdown agriculture and rice culture). In such conflicts the agricultural interest invariably prevails. One reason for this, apart from the greater financial and political power of the agricultural lobbies, is that the flood conditions required for agriculture are relatively well understood, whereas the requirements of fisheries are less clearly defined. It is part of the purpose of this paper to draw attention to the need to better refine fisheries models in order to represent fishery interests more effectively in negotiations for the allocation of water to fish.
Four main tools exist for predicting the responses of fish species to differences in flooding in large rivers produced by human agencies.
Knowledge of the biology of individual species can be used to predict the reaction of the species to some characteristics of the flood curve such as timing, smoothness and rapidity of change.
Modelling of fish community responses to differences in flood regime are more appropriate when looking at dynamic issues such as amplitude, duration and the relationship between them.
Evidence can be derived from practical experiences of artificial flood releases such as those carried out on the Pongolo River in South Africa. See for example Bruwer et al. (1996) and Heeg and Breen (1994).
Application of best professional judgement systems such as DRIFT and the Instream Flow Incremental Methodology (IFIM).
Because of the considerable diversity of river fish species in their migratory and spawning behaviour, it is unlikely that any one set of flood conditions will affect all species equally. A good flood for one species may be detrimental to another. The most obvious example of this is in arid zone rivers such as the Sahelian Niger River. Here a group of species that breed preferentially in the main channel assumed dominance over similar species that spawn on the floodplains during the failed floods of the 1970-1980 drought (Quensiere et al. 1994; Lae 1995). Indeed much of the year-to-year variation in relative abundance of species in rivers, as reflected in the catch, may be explained by differences in the quality of the floods between years.
Similarly the ability of many European and North American species to adapt to the regulation of temperate zone rivers probably lies in the inherent behavioural and genetic variability within species that first arose as an adaptation to extreme year- to-year variation in flooding intensity. There are indications, for example, that many of the cyprinids that were originally semi-migrant, grey fish species had lotically and lentically oriented genetic components. Evidently the regulation of most modern rivers and canals has favoured the lentic component although there is evidence that, given the opportunity, migratory elements re-emerge (Linfield 1985).
In general, however, Arrington and Winemillerss (2003) analysis of the literature on fish diversity in floodplain rivers indicates that the loss of the flood pulse not only will impact biological production but impoverish regional species pools. Furthermore, the reduction of landscape heterogeneity associated with lowered flows may impair the resilience typically observed in flood river systems. Strategies for the conservation of floodplain rivers must, therefore, protect the hydrological variability characteristic of the river. Likewise strategies for the restoration of such rivers must seek to restore the hydrological regime as a primary objective.
Four main types of flow can be listed depending on how they interact with the fish fauna:
Population flows influence biomass through density dependent interactions with individual population parameters such as growth and mortality. Major criteria here are the magnitudes of the high and low season flows.
Critical flows trigger events such as migration and reproduction. Here the main criteria are timing and quantity.
Stress flows endanger fish because of excess velocity at high water or through desiccation at low water. These are typically extreme flows occurring as isolated peaks in an irregular hydrograph.
Habitat flows are needed for the maintenance of environmental quality including temperature, dissolved oxygen levels or sediment transport (see Bun and Arthington 2002).
Management of environmental flows for the sustainability of fish stocks and fisheries requires an understanding of all four types of flow. In this regard, it is already possible to derive some principles that can serve as guidelines in planning flow requirements and releases.
TOWARDS GUIDELINES FOR ENVIRONMENTAL FLOWS AND ARTIFICIAL RELEASES OF WATER AIMED AT FLOODING FLOODPLAIN RIVERS
In general, projects and interventions in river basins that are likely to alter the amount of water available to the river and the timing of the delivery of the water should make arrangements to release the flows necessary for the maintenance of healthy fish populations. It is insufficient that these flows be calculated only in terms of the total amount of water available to the system. In general, they should be as close to the natural flows as defined by Poff et al. (1997) and Bunn and Arthington (2002) as is possible given the resources available and as such should respect certain norms with regard to timing and to the shape of the flood curve that results from planned discharges and releases.
A flood must be induced, preferably every year but if not every year then at least with sufficient frequency as to allow all species to reproduce within their life spans.
Flood releases should be timed to arrive after the wetting of the floodplains by local rainfall. This means that the water volume is used to maximum efficiency in flooding rather than in saturating the desiccated soils of the floodplain.
Flood releases should correspond to the needs of fish for hydrological stimuli that induce migration and spawning.
Flood curves should be as smooth as possible to avoid repeated advances and withdrawals of the water that strand and desiccate eggs adhering to marginal vegetation and expose nests.
Rises and falls in level should be relatively slow. This should avoid over-rapid submergence of nesting sites and excessive stranding of biota during the falling phase.
High short floods should be alternated with lower but longer ones to favour all groups of species.
Extreme flow events that result in washout of adults, juveniles and drifting fry should be avoided.
Adequate dry season flows should be assured. The amount of water remaining in the river is as critical to the survival of the fish population as the flood. Prolonged periods when no water is released, that desiccate the channel of the river and allow it to dry out into a series of de-oxygenated pools should be avoided.
The supply of water for ecological flows and artificial floods in regulated rivers does not come cheap. For example, recent public debate of plans to secure artificial flows in some US rivers are estimated to cost around $2M each in lost revenue from power generation. The quantities of water involved are impressive. Heeg et al. 1980 estimated that the Phongolo floodplain (South Africa) [10 265 ha at peak flood and 2700 ha of river and lakes at mean retention level] required 26 x 106 m3 yr-1 to maintain mean retention level of its floodplain lakes and a further 100 x 106 m3 yr-1 to flood the whole plain. However, a model developed by Weldrick (1996), showed that part of the floodplain could be submerged and the lakes could be filled by a discharge of 100 m3 s-1 for 5 days (equivalent to a total discharge volume of 2.16 x 108 m3). That such interventions are successful in large rivers is, however, attested to by the benefits of floodplain restoration along the Rhine River in the Netherlands (Grift 2001).
This paper could not have been written without the early comments of Tad Backiel and M. Zalewski and the later contributions by Angela Arthington and Brad Pusey, who carefully reviewed the manuscript and contributed much to the ideas about environmental flows.
Albaret J.J. & Lae R. 2003. Impact of fishing on fish assemblages in tropical lagoons: The example of Ebrie Laoon, West Africa. Aquatic Living Resources, 16: 1-46.
Arrington D.A. & Winemiller K.O. 2003. Organization and maintenance of fish diversity in shallow waters of tropical floodplain rivers. In Welcomme R.L. & Petr T. eds. Proceedings of the Second International Symposium on the Management of Large Rivers for Fisheries: Volume 2. Food and Agriculture Organization of the United Nations & Mekong River Commission. FAO Regional Office for Asia and the Pacific, Bangkok. RAP Publication 2004/17. pp 25-36.
Arthington A.H. 1998. Comparative evaluation of environmental flow assessment techniques: Review of holistic methodologies. LWRRDC Occasional Paper 26/98. Canberra, Land and Water Resources Research and Development Corporation.
Arthington A.H. Tharme R.E., Brizga S.O., Pusey B.J. & Kennard M.J. 2003. Environmental flow assessment with emphasis on holistic methodologies. In R.L. Welcomme & T. Petr, eds. Proceedings of the Second International Symposium on the Management of Large Rivers for Fisheries Volume 2. Food and Agriculture Organization & Mekong River Commission. FAO Regional Office for Asia and the Pacific, Bangkok. RAP Publication 2004/17. pp. 37-66.
Arthington A.H., Rall J.L., Kennard M.J & Pusey B.J. 2003. Environmental flow requirements of fish in Lesotho rivers using the DRIFT methodology. River Research and Applications, 19: 641- 666.
Baran E., van Zalinge N., Bun N.P., Baird I. & Coates D. 2001. Fish resources and biological modelling approaches in the Mekong Basin. Penang, Malaysia, ICLARM & Mekong River Commission Secretariat. 60 pp.
Bonetto A.A. & Pignalberi C. 1964. Nuevos aportes al conocimiento de las migraciones de los peces en los rios mesopotamicos de la Republica Argentina. Comm. Inst Nacional Limnologia Santo Tome, 1: 1-14
Bonetto AA., Pignalberi C., Cordiviola E. & Oliveros O. 1971. Informaciones complementarias sobre migracion de peces en la cuenca de la Plata. Physis Buenos Aires, 30: 505-520
Bovee K.D. 1982. A guide to stream habitat analysis using Instream Flow Incremental Methodology. US Fish and Wildlife Service, Instream Flow Information Paper No. 12, FWS/OBS - 82/26. 248 pp.
Bren L.J. 1988. Effects of river regulation on flooding of a riparian red gum forest on the River Murray, Australia. Regulated Rivers, 2: 65:77.
Brizga S.O. 2000. Burnett Basin water allocation and management plan: Proposed environmental flow performance measures, Vol. 1&2. Brisbane, Australia, Department of Natural Resources.
Bruwer C., Poultney C. & Nyathi Z. 1996. Community based hydrological management on the Phongolo floodplain. In M.C. Acreman, G.E. Hollis eds. Water management and wetlands in subsaharan Africa. Gland, Switzerland, IUCN. pp. 199-211.
Bunn S.E. & Arthington A.A. 2002. Basic principles and ecological consequences of altered flow regimes for aquatic biodiversity. Environmental Management, 30: 492-507.
Christensen M.S. 1993. Artisanal fishery of the Mahakam River floodplain in East Kalimantan, Indonesia. Actual and estimated yields, their relationship to water levels and management option. J. Appl. Ichyol., 9: 202-209.
Dansoko F.D. 1975. Contirbution a letude de la biologie dhydrocyon dans le Delta Centrale du Niger. Mali, Bamako, Ministere de lEducation Nationale. (Masters thesis)
Dansoko D.D., Breeman H. & Daget J. 1976. Influence de la secheresse sur les populations de dhydrocynus dans le Delta Centrale du Niger. Cahier ORSTOM Hydrobiol., 10: 71-76.
Davies B.R., Thoms M. & Meador M. 1992. An assessment of the ecological impacts of inter-basin water transfers and their threat to river basin integrity and management. Aquatic Conservation: Marine and Freshwater Systems, 2: 325-349.
De Graaf G. 2003. Dynamics of floodplain fisheries in Bangladesh, results of 8 years fisheries monitoring in the compartmentalization pilot project. Fisheries Management and Ecology, 10: 191-199.
Fuentes C.M. 1998. Deriva de larvas de sabalo Prochilodus lineatus Valeniciennes, 1847 y otras especias de peces de interes comercial en el Rio Parana Inferior. [Drift of larvae of Sabalo Prochilodus lineatus Valenciennes, 1847 and other species of commercial interest in the lower Parana River] Universidad de Buenos Aires. (Masters thesis)
Fuentes F. & Espinach Ros A. 1998. Distribucion espacial y temporal del ichthyoplncton en un punto del bajo delta del rio Parana. Revista del Museo Argentino de Ciencias Naturales Bernardino Rivadavia, 8: 51-61.
Furness H.D. 1978. Ecological studies on the Pongolo River floodplain. Working Doc. IV. Workshop on Man and the Pongolo floodplain 14/106/7c. Pitermaritzberg. SA. CSIR.
Gaygalas K.S. & Blatneve D.P. 1971. Growth characteristics and the structure and abundance of the bream Abramis brama L. stock in the water courses of the summer polders of the Nyumunas River delts. J. Ichthyology, 11: 682-692.
Gibbins C.N., Soulsby C., Jeffries M.J. & Acornley R. 2001. Developing ecologically acceptable river flow regimes: A case study of Kielder reservoir and the Kielder water transfer system. Fisheries Management and Ecology, 8: 463-485.
Grift R. 2001. How fish benefit from floodplain restoration along the lower River Rhine. Wageningen, The Netherlands, Wageningen University. 205 pp. (Doctoral dissertation)
Halls A.S. 1998. An assessment of the impact of hydraulic engineering on floodplain fisheries and species assemblages in Bangladesh. London, Imperial College of Science, Technology and Medicine. 532 pp. (Masters thesis)
Halls A.S., Kirkwood G.P. & Payne A.I. 2001. ADynamic pool model for floodplain river fisheries. Ecohydrology and Hydrobiology, 1: 323-339.
Heeg J. & Breen C. M. 1994. Resolution of conflicting values of the Pongolo River and floodplain South Africa. In Patten et al. eds. Wetlands and shallow continental water bodies, volume 2. The Hague, SPB Academic Publishing. pp. 303-359.
Heeg J., Breen C.M. & Rogers K.H. 1980. The Phongolo floodplain: A unique system threatened. In M.N. Bruton, K.H. Cooper eds. Studies on the ecology of Maputaland. Cape Town, SA, Rhodes University and the Ntal Branch of the Wildlife Society of Southern Africa.
Holcik J. 1988. Influence of hydrological regime and water temperature on the activity and population density of fishes in the anabranches of the Danube. Prace Ust. Rybar. Hydrobiol., 6: 3358.
Holcik J., & Kmet T. 1986. Simple models of the population dynamics of some fish species from the lower reaches of the Danube. Folia Zoologica, 352: 183191
Humphries P. & Lake P.S. 2000. Fish larvae and the management of regulated rivers. Regulated Rivers: Research & Management. 16: 421-432.
Humphries P., King A.J. & Koehn J.D. 1999. Fish, flows and floodplains: Links between freshwater fishes and their environment in the Murray- Darling River system, Australia. Environmental Biology of Fishes, 56: 129-151.
Jackson D.C, & Marmulla G. The influence of dams on river fisheries. Review prepared for the World Dams Commission. (forthcoming)
Junk W. J., Bayley P.B. & Sparks R.E. 1989. The flood pulse concept in river-floodplain systems. In D.P. Dodge ed. Proceedings of the international large rivers symposium. Can. J. Fish. Aquat. Sci. Spec. Publ., 106: 110-127.
Junk W.J. & Wantzon K.M. 2004. The Food Pulse concept: New aspects, Appaooches, and applications- an update. In R.L. Welcomme & T. Petr, eds. Proceedings of the Second International Symposium on the Management of Large Rivers for Fisheries Volume 2. Food and Agriculture Organization & Mekong River Commission. FAO Regional Office for Asia and the Pacific, Bangkok. RAP Publication 2004/17. pp. 117-140.
King J., Brown C. & Sabet H. 2003. A scenario based holistic approach to environmental flow assessments in rivers. River Research and Applications, 19: 619-639.
Krykhtin K.L. 1975. Causes of periodic fluctuations in the abundance of the non-anadromous fishes of the Amur River. J. Ichthyology, 15: 826- 829.
Lae R. 1992. Influence de lhydrobiologie sur levolution des pecheries du delta Centrale du Niger, de 1966 a 1989. Aquat. Lvg. Res., 5: 115-126.
Lae R. 1995. Climatic and anthropogenic effects on fish diversity and fish yields in the Central Delta of the Niger River. Aquatic Living Resources, 8: 43-58.
Lambert de Brito Ribeiro M.C. & Petrere M. 1990. Fisheries ecology and managment of the jaraqui Semaprochilodus taeniurus, S. insignis in Central Amazonia. Regulated Rivers, 5: 195-215
Leipolt R. ed.1967. Limnologie der Donau. Stuttgart, E. Schweizerbartsche Verlagsbuchhandlung pp. var.
Linfield R.S.J. 1985. An alternative concept to home range theory with respect to populations of cyprinids in major river systems. J. Fish. Biol., 27 Supp. A: 18796.
Loneragan N.R. & Bunn S.E. 1999. River flows and estuarine ecosystems: Implications for coastal fisheries from a review and case study of the Logan River, Southeast Queensland. Australian Journal of Ecology, 24: 431-440.
Lowe-McConnell R.H. 1975. Fish communities in tropical freshwaters. London, Longman. 337 pp.
Lowe-McConnell R.H. 1985. Ecological studies in tropical fish communities. Cambridge, Cambridge University Press. 382 pp.
Lucas M.C. & Baras E. 2001. Migration I Freshwater Fishes. London, Blackwell Science. 419 pp.
Lusk S., Halaeka K. & Lusková V. 2003. Rehabilitating the floodplain of the lower River Dyje for fish. River Research and Applications, 19: 281-288.
Morand P. & Bousquet F. 1994. Relations entre leffort de peche la dynamique du peuplement ichthyologique et le niveau des captures dans un systeme fleuve-plaine. In J. Quensiere ed. La Peche dans le Delta Central du Niger. pp. 267- 281. Paris, ORSTOM.
Moreau J. 1980. Essaie dapplication au lac Alaotra Madagascar dun modele detude de pecheries pour les plaines inondanles intertropicales. Chier ORSTOM (Hydrobiologi),13: 83-91.
Moses B.S. 1987. The influence of flood regime on fish catch and fish communities of the Cross River floodplain ecosystem, Nigeria. Environmental Biology of Fishes, 18(1): 51-65.
Nikonorov S.I., Maltsev S.A.& Morgunov S.V. 2001. The problems of natural spawning of sturgeons and the ways of their decision. Fish Economy, No. 5: 42-44.
Novoa D.F. 1989. The multispecies fisheries of the Orinoco River: Development, present status, and management strategies. In D.P. Dodge ed. Proceedings of the international large rivers symposium. Can. J. Fish. Aquat. Sci. Spec. Publ., 106: 422-428.
Payne A.I. & Harvey M.J. 1989. An assessment of the Prochilodus platensis Holmberg population in the Pilcomayo River fishery, Bolivia using scalebased and computerassisted methods. Aquaculture and Fisheries Management, 203: 23348.
Poff N.L., Allan J.D., Bain M.B., Karr J.R., Prestegaard K.L., Richter B.D., Sparks R.E. & Stromberg J.C. 1997. The natural flow regime. BioScience, 47: 769-784.
Puckridge J.T., Sheldon F., Walker K.F. & Boulton A.J. 1998. Flow variability and the ecology of large rivers. Marine and Freshwater Research, 49: 55-72.
Quensiere J., Benech V. & Dansoko D.F. 1994. Evolution de la composition des peuplements de poissons, pp. 105-121. In J. Quensiere ed. La Peche dans le Delta Central du Niger. Paris, ORSTOM.
Quiros R. & Cutch S. 1989. The fisheries and limnology of the lower Plata Basin. In D.P. Dodge ed. Proceedings of the international large rivers symposium. Can. J. Fish. Aquat. Sci. Spec. Publ., 106: 429-443.
Regier H.R., Welcomme R.L., Steedman R.J. & Henderson H.F. 1989. Rehabilitation of degraded river systems. In D.P. Dodge ed. Proceedings of the international large rivers symposium. Can. J. Fish. Aquat. Sci. Spec. Publ., 106: 86-97.
Sao Leang & Dom Saveun. 1955. Apercu general sur la migration et la reproduction des poisons deau douce du Cambodge. Proc. IPFC, 5: 138-62.
Sanchez L., Vazquez E. & Blanco L. 1985. Limnological investigations of the rivers Uracoa, Yabo, Morichal Largo and Claro in the Eastern Plains of Venezuela. Verh. Internat. Verein. Limnol. 22: 215360.
von Stankovic S. & Jankovic D. 1971. Mechainismus der fischproduktion im gebeit des mittleren Donaulaufes. Arch. hydrobiol., Suppl. XXXVI, 4: 299-305.
Stott B. 1967. The movements and population densities of roach Rutilus rutilus L. and gudgeon Cottus gobio L. in the River Mole. J. Animal Ecology., 36: 407-23.
Tennant D.L. 1976. Instream flow regimens for fish, wildlife, recreation and related environmental resources. Fisheries, 1: 6-10.
University of Michigan. 1971. The fisheries of the Kafue River flats, Zambia, in relation to the Kafue Gorge Dam. Report prepared for the UN/FAO. Ann Arbor, Michigan, University of Michigan Press. FI:SF/ZAM 11: Tech Report, 1: 161 pp.
Van Zalinge N. Degen P. Pongsri C. Nuor S. Jensen J.G. Nguyen V.H & Choulamany X. 2004. The Mekong River System. In R.L. Welcomme & T. Petr, eds. Proceedings of the Second International Symposium on the Management of Large Rivers for Fisheries Volume 1. Food and Agriculture Organization & Mekong River Commission. FAO Regional Office for Asia and the Pacific, Bangkok. RAP Publication 2004/16. pp. 335-358.
Welcomme R.L. 1979. Fisheries ecology of floodplain rivers. London, Longman. 317 pp.
Welcomme R.L. 1985. River fisheries. Food and Agriculture Organization Fisheries Technical Paper 262. 330 pp.
Welcomme R.L. 1995. Relationships between fisheries and the integrity of river systems. Regulated Rivers: Research & Management, 11: 121- 136.
Welcomme R.L. 1999. A review of a model for qualitative evaluation of exploitation levels in multispecies fisheries. Journal of Fisheries Ecology and Management, 6: 1-20.
Welcomme R.L. & Hagborg, D. 1977. Towards a model of a floodplain fish population and its fishery. Env. Biol. Fish., 21: 724
Welcomme R.L. & Halls A. 2001. Some considerations of the effects of differences in flood patterns on fish populations. Ecohydrology and Hydrobiology, 1: 313-321
Weldrick S.K. 1996. The development of an ecological model to determine flood release options for the management of the Phongolo floodplain in Kwazulu/Natal South Africa. Grahamstown, SA, Rhodes University. 81 pp. (Masters thesis)
Winemiller K.O. 2004. Floodplain river food webs: Generalizations and implications for fisheries management. In R.L. Welcomme & T. Petr, eds. Proceedings of the Second International Symposium on the Management of Large Rivers for Fisheries Volume 2. Food and Agriculture Organization of the United Nations & Mekong River Commission. FAO Regional Office for Asia and the Pacific, Bangkok. RAP Publication 2004/17. pp. 285- 310.
World Commission on Dams 2000. Dams and Development: A new framework for decision making. London, Earthscan Publications.