THE INFLUENCE OF DAMS ON RIVER FISHERIES

by

Donald C. Jackson
Professor of Fisheries Management
Department of Wildlife and Fisheries
Mississippi State University
Box 9690, Mississippi State, USA
and
Gerd Marmulla
Fishery Resources Officer
Inland Water Resources and Aquaculture Service
Food and Agriculture Organization of the United Nations
Viale delle Terme di Caracalla, 00100 Rome, Italy

EXECUTIVE SUMMARY

The extent to which fisheries can be developed, sustained or protected along riverine ecosystems modified by dams reflects basin topography, geological features, watershed hydrology, and climate, as well as engineering features of the dam itself, and operational programmes for retention and release of water from the reservoir, through the dam and into the tailwaters. Compensation for loss in yield from river fisheries can be difficult to achieve through development of reservoir fisheries. The larger the river, and the more downstream the location of the dam, the less potential there is for a reservoir fishery to compensate in terms of yield for losses sustained by the river fishery. Compensation potentials apparently are higher in shallower reservoirs in tropical regions than they are in deeper reservoirs and in more northern latitudes. Even if compensation is achieved from a fishery perspective, specific needs of fish species not included in the fishery, and/or that may be threatened or endangered, must be considered to avoid negative impacts to these fishes.

There is considerable variability in fishery production among and within regions with respect to reservoir and river fisheries. River fishery production is dependent on length of river, catchment area and, for specific sections of rivers, the position of the segment along the river continuum. In tropical and temperate rivers, fish yields per unit surface area are considerably greater in rivers with flood pulses and floodplains than in nearby impoundments where flood pulses are reduced or absent. In the tropics, for example, large, slow-flowing rivers averaged 30-100 kg/ha/year and the floodplains averaged 200-2 000 kg/ha/year. Fish yields in floodplain river ecosystems are directly related to the height and duration of floods. If altered hydrology resulting from dams curtails or eliminates normal, historical downstream flooding, overall fisheries production throughout the system can be negatively impacted.

In Africa, large reservoirs subject to moderate to heavy fishing (e.g. Kariba, Nasser/Nubia, Volta), have yields ranging 27-65 kg/ha/year. In contrast, however, Lake Kainji, another large African reservoir has yields of only 3.5-4.7 kg/ha/year. For medium-sized African reservoirs, estimated mean yield was approximately 80 kg/ha/year. Mean yield from a variety of Sub-Saharan small water bodies was 329kg/ha/year. Substantial overall losses to overall fishery production in river basins have been reported as a result of dam construction in Africa. For example, an annual net loss of 11 250 t of fish per year were lost in the Senegal River system as a result of dam construction associated with Lake de Guiers.

Major concern throughout Asia is that movements of migratory fishes along river courses will be blocked by dams. Additionally, dewatering of stream channels immediately downstream from dams can be a serious problem. Reservoir yields in China are reported to range from 127 to 152 kg/ha/year, but these high values tend to be the result of intensive stocking programmes. In India reservoir fishery yields range from 11.4 (large reservoirs) to 49.5 (small reservoirs) kg/ha/year. Reservoir fishery yields in Southeast Asia (e.g. Malaysia) and Central Asia and Kazakhstan are reported to be much less than in other parts of Asia, with values typically around 15 kg/ha/year or less. Yields in Sri Lanka range from 40 to 650 kg/ha/year, but these yields are primarily the result of stocking reservoirs with exotic species.

In Australia, dams have generally resulted in negative impacts to native riverine fishes while encouraging exotic species. This has been attributed, in part, to disruption of seasonal flood cycles, and to dams acting as barriers to fish movements. The Murray River now has the lowest commercial fish yield per km2 of floodplain of any of the world's major rivers, although historical catches were comparable. In reservoirs constructed on Queensland rivers, fish stocks are maintained through stocking of native fish species.

In Latin America and the Caribbean, reservoir fishery yields tend to be higher for the Caribbean (Cuba 125 kg/ha/year; Dominican Republic 29-75 kg/ha/year) than is generally recorded for Central andSouth America reservoirs. Records available for Brazil (2.1-11.5 kg/ha/year) and Panama (4.8-63.2kg/ha/year), suggest that reservoirs can have quite variable yields, depending on flushing rates, elevation, and basin morphology. Higher yields throughout the region typically result from stocking of exotic species.

Similar patterns exist with regard to fishery yields from reservoirs in temperate zones. The average yield from North American reservoirs is only 24 kg/ha/year. In Europe, records indicate reservoir fishery yields ranging from 21 to 76 kg/ha/year.

Dams can enhance some riverine fisheries, particularly tailwater fisheries immediately below dams that result from discharge of seston (primarily plankton) from the upstream reservoir. However, discharge of seston is typically attenuated quickly downstream from the dam, with corresponding attenuation of the associated fisheries. If discharge is from the hypolimnion of the reservoir, lowered temperatures in the receiving tailwater can curtail or eliminate warmwater river fisheries and require stocking of exotic coldwater species, e.g. salmonids (assuming that the water is sufficiently oxygenated). Productive tailwater fisheries targeting these coldwater fishes can result but generally require supplemental hatchery programmes and introduction of coldwater invertebrates to serve as forage items for these fish. In North America, yield from coldwater tailwater fisheries have been recorded for up to 339 kg/ha/year with fishing effort seven times higher than the respective upstream reservoir. This high amount of effort reflects high standing stock of salmonids in these environments. Some of these coldwater tailwater fisheries can extend considerable distances (e.g. > 150 km downstream from dams in Arkansas, USA).

Reservoirs resulting from construction of dams can in some situations result in productive fisheries. This is particularly true for locations where river fisheries contribute little to overall national fishery yields. Beneficial reservoir fisheries also exist in drier regions where dams are constructed for agricultural irrigation, and fisheries are secondary considerations. Benefits seem more pronounced for smaller, shallower reservoirs that have reasonably high concentrations of dissolved solids and that are located in the upper reaches of their respective river ecosystem. Stocking of exotic species (both in reservoirs and in tailwaters ) can enhance yields, as long as the exotic fishes are environmentally sound and culturally acceptable to the surrounding human population. In this regard, caution is warranted in cultures where fishing and fish consumption are non-traditional activities. Building reservoirs in the context of such cultures may not achieve projected fishery benefits even though exploitable fish stocks may exist.

Development of reservoir plankton reflects nutrients captured by the reservoir. This plankton generally relates directly to fisheries production of the respective reservoir. However, when several dams are constructed on upstream tributaries of a river ecosystem, the cumulative effects of these dams can be that of blocking the flow of nutrients originating from the catchment basin from the lower reaches of the ecosystem, thereby negatively affecting fisheries production in downstream portions of the ecosystem (including estuary and marine environments). Dams also can block the flow of nutrients from ocean environments upstream into riverine environments by preventing anadromous fishes that die after spawning (e.g. Pacific salmonids) from depositing these nutrients via carcass decay in upstream reaches.

Furthermore, and if the riverine fishery is sustained by stocks of migrating fishes that become blocked by a dam, the riverine fishery can be severely impacted. If the migrating fishes are anadromous or catadromous species, linked to ocean fisheries, or those of inland seas or large lakes, the negative impacts to these stocks and their associated fisheries can be catastrophic.

Because dams tend to be constructed to enhance socio-economic development activities, they tend to attract people and industry. Subsequently, river ecosystems containing dams must contend with secondary environmental pressures such as increases in pollution as well as increased exploitation and extraction of their resources (primarily water, fish, and substrates), that are independent from and in addition to the direct influences of dams and reservoirs on the physical and biological dimensions of the system.

Determining the impact of dams on river ecosystems and their associated fisheries depends on spatial and temporal scales of interest. If spatial scales are sufficiently large (planetary, continental, perhaps regional and biome), and temporal scales are sufficiently long (decades, centuries, millennia), placing a dam on a river does little more than increase atmospheric water vapour (through evaporation from the reservoir), reduce long-term streamflows downstream, desiccate terrestrial environments, salinate surrounding areas, and shift bio-energetic processes (some of which can lead to floral and faunal extinction at various scales of resolution). We cannot assign the terms "good" or "bad" to any of these phenomena. They simply reflect anthropogenic activity on this planet. However, if we look at smaller spatial and shorter temporal scales, (which we obviously cannot neglect since we have to make decisions that have bearings on the present and future human generations and also on present and future living aquatic resources) we have to keep in mind that dams and their reservoirs (which can under certain circumstances help to better nourish people and make their livelihoods more sustainable) can - if wrongly placed - also lead to significant declines of fisheries and to extinction of aquatic species.

Given sufficient time, geophysical and climatic forces will override and erode the physical influences of dams, and evolutionary forces will alter how life forms interact with the resulting environments. Caution is warranted to avert potential negative impacts from dams with respect to fisheries and associated human interactions with these and other river resources. Such caution underscores the reality that people are depending on us, the scientists, the resource managers, the decision-makers, to be right.

1. INTRODUCTION

Dams interrupt streamflow, and generate hydrological changes along the integrated continuum of river ecosystems (Vannote et al., 1980; Junk et al., 1989) that ultimately can be reflected in their associated fisheries. The most obvious effects from placing dams on rivers result from formation of new lentic or semi-lentic environments upstream from the dam, and tailwater environments downstream from the dam. Both environments can be conducive to the establishment and maintenance of fish stocks appropriate for exploitation by fisheries.

Photo 1: Through the effect of this hydroelectric dam, the fishery of the White River (Arkansas, USA) was converted from a naturally sustained warmwater fishery into an artificially sustained trout fishery dependant on periodic stocking by Government hatcheries. (Photo: D.C. Jackson, Dept. of Wildlife and Fisheries, Mississippi State University, USA)

The extent to which fisheries can be developed, sustained or protected along these modified riverine ecosystems reflects basin topography, geological features, watershed hydrology, and climate, as well as engineering features of the dam itself, and operational programmes for retention and release of water from the reservoir, through the dam and into the tailwaters. Fundamental considerations must include establishment and maintenance of habitat for spawning, recruitment and maturation of the fish stocks, and provisions for passage by fishes that during certain phases of their life cycles, depend on longitudinal movements along the stream continuum (FAO, 1998).

In this regard, Bernacsek (1984) provided an excellent summary of design and operational features for dams to address fisheries concerns. Although the emphasis of Bernacsek's paper focused on African reservoirs, the general orientation has applicability to many situations on a global scale. He suggested: (i) maximum possible crest elevation; (ii) discharge structure intakes positioned at highest possible elevation; (iii) discharge water into tailwaters be sufficiently oxygenated to support aquatic fauna; (iv) annual water level fluctuation in the reservoir to be within the range of 2.5-4.0 m; (v) drawdown rate not to exceed 0.6 m/month; and (vi) downstream discharge to include an annual artificial flood event.

Photo 2: A section of the River Perak (Malaysia) has been dewatered downstream of a dam; only few fish are caught in the dewatered stream stretch. (Photo: D.C. Jackson, Dept. of Wildlife and Fisheries, Mississippi State University, USA)

Along the stream continuum, dams and their associated upstream reservoirs have downstream effects on riverine environments and, subsequently, diverse influences on downstream fisheries, even beyond the lotic ecosystem. Cumulative effects of dams in catchment basins and tributary streams can significantly block nutrient flow throughout the ecosystem, affecting fisheries production in downstream reservoirs (Welcomme, 1985), river channels (Hess et al., 1982) and estuary and marine environments (Ryder, 1978). Tolmazin (1979) related reduced fish yields in the Black Sea and the Sea of Azov to impoundments on the Danube, Dnieper, and Dniester rivers in Europe and, in line with the orientations of Welcomme (1985), Hess et al. (1982), and Ryder (1978), suggested that such patterns reflected dams acting as nutrient traps.

Dams also block the flow of nutrients from ocean environments upstream into riverine environments. This is particularly true of anadromous fishes such as Pacific salmon (Oncorhynchus spp.) that die in the rivers after spawning one time. Cederholm et al. (1999) give an account of the essential contributions of nutrients and energy of Pacific salmon carcasses to the ecosystem. Post spawning mortality of these adult fish introduces nutrients back into the stream in proportion to the number of carcasses deposited. Blockage of this allochthonous organic material from the sea can severely restrict subsequent recruitment of young salmonids in these rivers, directly by limiting their consumption of flesh from dead adults, and indirectly by reducing primary production of plankton and secondary production of benthic macroinvertebrates (Piorkowski, 1995).

Photo 3: A dewatered stream reach downstream of a dam in the Dominican Republic. (Photo: D.C.Jackson, Dept. of Wildlife and Fisheries, Mississippi State University, USA)

Dams also can enhance some riverine fisheries, and particularly with respect to tailwater fisheries immediately below dams. Fishes can become concentrated below dams as a result of the attractive foraging opportunities there as well as from seasonal congregations of migratory fishes (Jackson, 1985a). On a per unit area basis, tailwater fisheries can be better than those of the reservoirs themselves (Bennett, 1970). Fry (1965) reported that the tailwater fisheries below Table Rock and Taneycomo dams on the White River (Missouri, USA) and Clearwater Dam on the Black River (Missouri, USA) received 7, 10 and 16 times, respectively, more fishing effort per unit area (angler hours per hectare per year) than their associated upstream reservoir. Table Rock tailwater is a coldwater tailwater dependant on stocking of exotic rainbow trout (Oncorhynchus mykiss), a species that does not reproduce naturally in the system. The other two tailwaters are warmwater tailwaters with native species that reproduce naturally in the rivers. Yields were 339 kg/ha, 364 kg/ha and 753 kg/ha, respectively for the Table Rock, Taneycomo and Clearwater tailwaters. Reservoir yields during the same period (1950s-1960s) for these reservoirs were: Table Rock, 21.4 kg/ha (SE 4.23, N = 10 years); Taneycomo, 71.2 kg/ha (SE 9.93, N = 9 years); Clearwater, 31.35 kg/ha (SE 8.58, N = 4 years) (Turner and Cornelius 1989). For the two warmwater systems (Taneycomo and Clearwater), the tailwaters (Taneycomo 129.5 ha; Clearwater 32.4 ha) also provided greater overall total harvests by weight than did their respective reservoirs (Taneycomo 570.6 ha; Clearwater 805 ha) (Fry, 1965). This high level of production can be related to the transport of seston (primarily plankton) from the upstream reservoir to the receiving tailwater (Jackson et al., 1991).

Photo 4: A blue catfish captured in the tailwater of Coosa River below Jordan Dam (Alabama, USA). (Photo: D.C. Jackson, Dept. of Wildlife and Fisheries, Mississippi State University, USA)

Fisheries benefits from most tailwater fisheries typically encompass relatively short sections of streams below their respective dams. For example, in navigation channels of the Tennessee-Tombigbee Waterway (Mississippi, USA), tailwater influences extend approximately 4 km below Aberdeen and Columbus dams (Jackson and Dillard, 1993). In the Coosa River tailwater below Jordan Dam (a hydropeaking facility in Alabama, USA), tailwater influences on the fisheries extended approximately 4 km downstream under low flow regimes and nearly 15 km downstream under high flow regimes (Jackson and Davies, 1988a, 1988b; Jackson et al., 1991). Maintaining instream flows to address fisheries concerns in the tailwater below Jordan Dam has been a subject of intense debate in the bio-political arena of re-licensing this hydroelectric facility (Jackson, 1985a). Two examples of tailwater fisheries in Africa are the ones on the Volta River below the Akosombo Dam (Ghana) and below the Kainji Reservoir (Nigeria).

Eschmeyer and Miller (1949) and Miller and Chance (1954) estimated that 35% of the angling in Tennessee Valley Authority waters (USA) occurred below dams, and that tailwater fisheries accounted for 52% of the total harvest. Jackson (1985a), Jackson and Davies (1988b), and Jackson and Dillard (1993) recorded highly productive fisheries in the warmwater tailwaters of the Alabama-Coosa and Tennessee-Tombigbee Waterway systems in the southeastern USA. Hess et al. (1982) noted that reservoirs on the Missouri River produced plankton that upon discharge through the dams was beneficial to the respective downstream fisheries. However, Jackson (1985a), Jackson et al. (1991) and Sarnita (1991) demonstrated that plankton transport in a tailwater is rapidly attenuated downstream. Maintaining higher discharges from dams can extend the beneficial influences of plankton discharged from the dam to lower stream reaches; and thereby lengthen tailwater fisheries in the respective system. Caution, however, is warranted, because excess flushing rates from the upstream reservoir can result in reduced residence time for water in the reservoir, which in turn can preclude development and production potentials for plankton in the reservoir. This would undermine the plankton foundation that supports both the reservoir fishery and its respective downstream tailwater.

Temperature can greatly influence riverine fishes, and particularly warmwater fishes. Ye (1996) and Jackson and Ye (2000) related hydrological and climatological factors to principal fish stocks of the Yalobusha River (Yazoo River ecosystem, Mississippi, USA) and identified water temperature (R2 = 0.99) as the most important factor influencing stock structure of channel catfish (Ictalurus punctatus). Cooler water apparently curtailed reproduction and subsequent recruitment, and resulted in stocks dominated by larger fish. Rutherford et al. (1995) reported that growth increments of channel catfish in the lower Mississippi River were positively related only to length of the growing season (number of days > 15°C) and attributed this to favourable production of fish food items (primarily invertebrates) during extended warm environmental conditions. Subsequently, cool and cold water releases from dams can curtail or eliminate warmwater fisheries in the tailwaters below the dams (Pasch et al., 1980). However, oxygenated hypolimnetic discharges of cold water can sustain stocks of salmonids where normally waters are too warm during the summer for these fishes (Cadwallader, 1978). Unlike warmwater tailwater fisheries, many coldwater tailwater fisheries require supplemental stocking for their maintenance, primarily because variable flow regimes from hydropower facilities preclude availability of seasonally-stable spawning environments.

Dams purposefully or inadvertently alter downstream hydrology, including flooding. If the altered hydrology curtails or eliminates normal, historical downstream flooding, overall fisheries productivity throughout the system can be impacted negatively (Holcik and Bastl, 1977; Welcomme, 1976, 1985; 1986; Junk et al., 1989). In both tropical and temperate rivers, fish yields per unit surface area are considerably greater in rivers with flood pulses and floodplains than in nearby impoundments where flood pulses are reduced or absent (Sparks, 1995). Flooding sets into motion incorporation of extra-channel allochthonous organic material as well as nutrients of terrestrial origin into aquatic dimensions of the riverine ecosystem (Vannote et al., 1980; Junk et al., 1989; Bayley, 1989; 1995; Thorp and Delong, 1994; Sparks, 1995).

Once flooding occurs, invertebrates and fishes colonize the inundated areas to take advantage of these allochthonous resources and their products (e.g. invertebrates) on the floodplain (Flotemersch, 1996). Floodplains thus serve as important spawning and nursery grounds, as well as important sources of food for fish of all sizes. The moving interface between the aquatic and terrestrial dimensions of the ecosystem is particularly important because this environment, which is limited in time, promotes faunal interactions biotically as well as abiotically, and rapid nutrient exchanges (Goulding, 1980; Bayley, 1989). Fish yields from floodplain river ecosystems are directly related to the height and duration of floods (Holcik and Bastl, 1977; Goulding, 1980; Welcomme, 1985; Jackson and Ye, 2000; Jackson, in press).

Photos 5a and 5b: Tibbie Creek, a tributary of Tombigbee River (Mississippi, USA), during (a) the dry season and (b) the wet season. Both photos show exactly the same location. (Photos: D.C. Jackson, Dept. of Wildlife and Fisheries, Mississippi State University, USA)

It is essential that a fishery be understood as a composite of three interactive components: (i) fish stocks; (ii) habitat; and (iii) people (Nielsen, 1993). If one of these components is missing, there is no fishery. The presence of fisheries resources (e.g. a reservoir stocked with fish appropriate for exploitation) does not necessarily mean that a fishery exists. People must be exploiting the resources consumptively or otherwise for there to be a fishery. This exploitation can be curtailed or rendered void or non-existent by factors such as access, culture and tradition, social disturbance, and economics. Modifications to or loss of the natural river environment supporting fish stocks, and human interactions with these stocks, can challenge or eliminate traditional, and culturally-important fisheries (Jackson, 1991). River fisheries are non-portable. Persons with individual, community and/or sub-cultural identities linked to river fisheries can suffer profound social and economic stress if the foundation for their identities (i.e., the river and its resources) is taken from them (Baird, 1994; Brown et al., 1996). Shifting focus and techniques from those appropriate for the seasonal dynamics of river ecosystems to those appropriate for reservoirs and tailwaters can require training and experience. However, both the training and the gain in experience require time elements that persons living and working on the social, economic and nutritional margins of a given society may have difficulties in coping with. This does not exclude that in some areas fishers can adapt readily to the new situation, as seems to be the case for Lake Volta in Ghana (Petr; pers. comm.). Elsewhere in Africa (e.g. on the reservoir Nyumbaya-Mungu in Tanzania), skilled fishers have immigrated from other countries to exploit new resources; however, this is yet another strategy and does not contradict the above statement on training needs of the local population to cope with the new conditions.

Photo 6: A good catch of river catfish, the production of which is the result of the healthy and intact floodplain river ecosystem of the Kapuas River (Kalimantan, Indonesia). (Photo: D.C. Jackson, Dept. of Wildlife and Fisheries, Mississippi State University, USA)

2. YIELD MODELS

From a global perspective, large river ecosystems are the critical lotic resources with respect to fisheries (see Dodge, 1989, and references therein). Welcomme (1985) developed yield models for large rivers relating river basin area and length of the main channel to catches.

For river basin area the relationship is:

C = 0.03A0.97

( r = 0.91 )

where C = annual yield in tons, and A = river basin area in km2.

For length of the main channel, the relationship is:

C = 0.0032L1.98

( r = 0.90 )

where C = annual yield in tons, and A = channel length in km.

Photo 7: Due to the fact that there are no dams, the Pahang River (Malaysia) still has an intact floodplain river ecosystem and a productive fishery. (Photo: D.C. Jackson, Dept. of Wildlife and Fisheries, Mississippi State University, USA)

Photo 8: Fishing for giant catfish in the Mekong River shared by Thailand and Laos. (Photo: D.C. Jackson, Dept. of Wildlife and Fisheries, Mississippi State University, USA)

Figure 1 depicts yield estimates for different channel lengths from a hypothetical large river. Exponential increases in yield as segments lengthen relate to the connectivity and cumulative influences of upstream processes within the system ("River Continuum Concept": Vannote et al., 1980), and lateral processes associated with riparian, watershed and floodplain dimensions of the stream ecosystem ("Flood Pulse Concept": Junk et al., 1989).

Figure 1. Predicted fish yields for reservoirs from the global yield model developed by Schlesinger and Regier (1982): Log10 Yield = 0.044 Temperature + 0.482 Log10 MEI + 0.021

With respect to lacustrine systems, Ryder (1965) developed the morpho-edaphic index (MEI) to assist in making first order estimates of fish yield from moderately fished north-temperate lakes. The MEI is calculated by dividing the value of total dissolved solids (mg/l) by the mean depth (m) of the water body. Jenkins (1982) successfully applied this methodology to North American reservoirs. However, Jenkins (1982) tracked reservoir fishery yield over time and noted substantial within-system variability in yield. This variance should be included as a characteristic of a given fishery alongside its respective magnitude parameter (i.e. mean yield). Models addressing variation in yield have not received the attention enjoyed by those addressing means.

The simplicity of the MEI, and its generally good predictive capabilities have resulted in its application worldwide, subject to regional modifications. Generally, the MEI demonstrates that as nutrients in the water increase and depth decreases, fish production increases. Jenkins (1982) emphasized that relationships can be curvilinear, with greatest predictability within intermediate ranges for MEI and less fit at extremes. Schlesinger and Regier (1982) expanded the model to incorporate temperature effects, and subsequently enhanced its global applicability. Figure 2 depicts theoretical MEI yield estimates for lakes having annual mean temperatures ranging from 5°C to 25°C using the model developed by Schlesinger and Regier (1982). Table 1 estimates fish yield for hypothetical reservoirs having surface areas of 100 to 10 000 ha, depths of 5-15 m, total dissolved solid (TDS) concentrations of 50 to 200 mg/l, and temperatures of 5°C, 10°C and 25°C.

Figure 2. Predicted fish yields for rivers of different lengths (values calculated from Welcomme 1985: C = 3.2L1.98, where C = yield (kg/year) and L = length of river (km).

Table 1. Theoretical fish yield estimates for reservoirs having different surface areas (hectares), average depths (m), total dissolved solids (TDS, mg/l) and average annual temperatures (°C), based on the global, temperature-adapted morpho-edaphic index by Schlesinger and Regier (1982)1.

¹Log10 Yield = 0.044 Temperature + 0.482 Log10 MEI + 0.021

From Figure 1 and Table 1 note that if 25 km of river channel is converted to reservoir environment, an estimated 1875.5 kg/year of fish yield is lost from the river. This loss is theoretically compensated by a 100-ha reservoir in the tropics (annual temperature 25°C), regardless of depth (range 5-15 m) or TDS concentration (range 50-200 mg/l), but not by reservoirs of similar size, depth or TDS concentration in temperate (10°C) and higher latitude (5°C) regions. If the reservoir has 1 000 or more hectares of surface area associated with the 25 km of river channel, then compensation occurs throughout the entire ranges of TDS and mean depths in all the regions. Similar exercises can be conducted for longer sections of river channel and for reservoirs having greater surface area. For example, loss of fish yield from 100 km of river channel (29 184 kg/year) can be compensated by a tropical reservoir with 1000ha of surface area, and mean depth of 5 m with TDS of 50 mg/l, and by deeper tropical reservoirs with higher TDS concentrations; but larger reservoirs are required in temperate and higher latitude regions to compensate for the loss.

From this simple exercise, it would appear that, at least theoretically and hypothetically, reservoir fisheries can reasonably mitigate losses sustained from impacts to river fisheries. However, a potential, fundamental weakness in the comparison is that the river model does not have the same temperature adjustment factors as does the reservoir model. Additionally, the actual size (width; depth) and gradient of the river may have some bearing on yield, not only in terms of biological production, but also with respect to exploitation by persons engaged with the fishery. Finally, there can be cumulative (positive) influences on the fishery as one proceeds from headwaters to downstream reaches of the river (e.g. higher nutrient loads; substrates with higher organic content; greater diversity in terms of size and content of allochthonous organic material inputs; more stable thermal regimes) (Vannote et al., 1980).

In this last regard, consider again the 25-km segment of river utilized for the exercise above. The general model estimates that fishery yield for a 25-km segment is 1 875.5 kg/year. However, considering cumulative influences upstream to downstream (Table 2) and using the model developed by Welcomme (1985, p. 213), we note that at a distance of 50 km from the river's source, a 25-km section of river yields 9 113 kg/year and at a distance of 250 km downstream from the source, a 25-km section of the river yields 37 197 kg/year. If a dam were constructed at a distance of 400 km from the river's source, and resulted in loss of a 25-km section of the river at that point, the reservoir would need to compensate for 57 925 kg/year. This could be accomplished, for example, with tropical reservoirs having mean depth of 5 m, TDS of 100 mg/l and a surface area of somewhat more than 1000 ha. Temperate reservoirs with the same mean depth and TDS could compensate for this loss of river fishery yield with a surface area of 4 728 ha. Although the distance compensating model for this exercise was developed for African rivers, it has been used successfully for rivers in other regions (e.g. the Mekong, Danube and Magdalena rivers) (Welcomme, 1985).

Table 2. Estimated fish yields for a 25-km section of river at different distances from the river's source along the river continuum.¹

¹ Calculated from the model developed by Welcomme (1985, p.213): xCy = Cx+y - Cy where x = the length of the stream se.g.ment (25 km in the example), y = the distance of the stream se.g.ment from the river's source and C = yield in kg/year. Catches for Cx+y and Cy are calculated by the equation C = 3.2L1.98, where L values are lengths for x + y and y, respectively.

Much depends on the specific nature of the riverine fishery in question. If the riverine fishery is sustained by stocks of migrating fishes that become blocked by a dam, the riverine fishery can be severely impacted. If the migrating fishes are anadromous or catadromous species, linked to ocean fisheries, or those of inland seas or large lakes, the negative impacts to these stocks and their associated fisheries can be catastrophic. Carried to the extreme, stock reductions can reach levels where the stocks become threatened beyond fishery concerns (i.e. no longer economically feasible to exploit), and enter the arena of bio-ethical concerns regarding their extirpation from this planet (see Cederholm et al., 1999).

Unfortunately, there really is no simple formula for addressing the impact of dams (positively or negatively) on riverine fisheries. There are, obviously, situations where reservoir fisheries are great assets. For example, Sugunan (1997) reviewed fisheries in small impoundments for seven countries representing Africa, Asia and Latin America/Caribbean. He reported that tilapia (exotic fishes), especially in island nations (e.g. Cuba and Sri Lanka) increased fish production in reservoirs. This is particularly important in countries such as these where contributions by river fisheries to overall national fishery yields are naturally relatively minor. Success with regard to enhancing fish yield via reservoir fisheries also was noted for situations where there were few competing species, and few predators. Even in countries with substantial river fisheries, reservoir fisheries development has been beneficial. For example, Sugunan (1997) noted that tilapia stocked into small Brazilian reservoirs in the northeast region of the country resulted in higher fish yields than those from reservoirs without tilapia. Reservoir construction in drier zones of India, Thailand, northeast Brazil, Sri Lanka and Mexico is primarily for agriculture irrigation, but these reservoirs provide secondary benefits via fisheries. Yields from small impoundments located in the seven countries addressed by Sugunan (1997) averaged 165 + 16.3 kg/ha/year.

Marshall and Maes (1994) compared yields from various types of water bodies in the tropics (Table 3). Shallow, managed, reservoirs averaged 30-150 kg/ha/year; deep reservoirs averaged 10-50 kg/ha/year; floodplains averaged 200-2 000 kg/ha/year; and large, slow-flowing rivers averaged 30-100 kg/ha/year. From these data, it seems that fisheries in large rivers hold their own quite well with respect to yield when compared to those of reservoir fisheries. Additionally, and if floodplains are included as components of the river (which they are), then reservoir fisheries, even in the tropics where they are most productive, are far less productive than river fisheries on a per unit area basis. Permanently inundating a floodplain by an impoundment (thereby restricting the moving littoral zone, sensu Bayley, 1989; Junk et al., 1989), or regulating water release from dams so that downstream floodplains are not sufficiently inundated in terms of depth, duration and seasonality of flooding, can potentially result in significant overall fishery losses.

Table 3. Estimated annual fish productivity from diverse aquatic systems in the tropics (compiled by Marshall and Maes 1994).

Table 4. Fish yield from selected reservoirs worldwide.

Region	Yield (kg/ha/year)	Reference
Africa:
Small Reservoirs	329	Marshall & Maes (1994)
Medium Reservoirs	80 - 90	Kapetsky (1986); van der Knapp (1994)
Large Reservoirs	27 - 65	Kapetsky (1986); Machena (1995);
		Rashid (1995); Braimah (1995)
Asia:
China	127 - 152	Lu (1986)
India
Small Reservoirs	49.5	Sugunan (1995)
Medium Reservoirs	12.3	Sugunan (1995)
Large Reservoirs	11.4	Sugunan (1995)
Kazakhstan	15	Petr & Mitrofanov (1998)
Malaysia	3 - 12	Ali & Lee (1995); Ali (1996)
Sri Lanka	40 - 650	Sugunan (1997)
Latin America & Caribbean:
Brazil	2.1 - 11.5	Dos Santos & de Oliveira (1999)
		Sugunan (1997)
Cuba	125	Sugunan (1997)
Dominican Republic	29 - 75	Jackson (1985)
Panama
Gatun Lake (Panama Canal)	4.8 - 5.3	Bayley (1986); Maturell & Bravo (1994)
Bayano Lake	63.2	Candanedo & D'Croz (1983)
North America & Europe:
Austria (Danube system)	32	Bacalbasa-Dobrovici (1989)
Commonwealth of Independent States	0.1 - 48.1	Karpova et al. (1996)
Germany
Danube system	65 - 76	Bacalbasa-Dobrovici (1989)
Rhine system	21 - 62	Lelek (1989)
Poland (Vistula system)	26	Backiel & Penczak (1989)
United States	24	Jenkins (1982)

Table 4 provides yield data from selected reservoirs worldwide. Although there is considerable variance in this data, the general trend seems to be that tropical and subtropical reservoirs tend to be more productive than temperate reservoirs with similar morphoedaphic characteristics. Additionally, smaller reservoirs are generally more productive on a per unit area basis than are larger reservoirs. Smaller impoundments usually have greater surface area to volume ratios than do larger impoundments. As a result, smaller impoundments tend to have overall higher primary production than do larger impoundments. Higher primary production typically enhances fishery yields. However, these size-related benefits can be lost if the impoundment is subject to excessive flushing or desiccation.

The variation noted above suggests that partitioning the world into general regions may clarify relationships between dams and reservoirs and their respective river fisheries. For partitioning purposes the following regions were delineated: Africa, Asia, Australia, Latin America and the Caribbean, North America, Europe and the Commonwealth of Independent States.