The major factor in the development of Sri Lankan inland fisheries was the introduction, in 1952, of the exotic cichlid Oreochromis mossambicus (Peters) which dominated the catch ever since. The most important water bodies for the inland fisheries are the larger (>100 ha) man-made reservoirs in the dry lowlands of north, east and south east Sri Lanka. The reservoirs cover a total surface area of around 100 000 ha. They are shallow with maximum depths often less than 5 in and exhibit considerable fluctuations in water level. Supply of irrigation water is the main purpose of these reservoirs and fish production is a secondary function of recent origin. The mean annual yield is estimated at 270 to 300 kg/ha/yr which is high for tropical lacustrine fisheries. The total catch from inland fisheries was low before the introduction of 0 . mossambicus, because a fishery for indigenous fresh water species hardly existed. The commercial fisheries still use only a small proportion of the available fish biomass. Although large stocks of indigenous minor cyprinids are present in the lowland reservoirs of Sri Lanka, these species form only a minor part of the catch.
An additional fishery for the unexploited cyprinid resources, with small-meshed gillnets, was recently proposed for the improvement of inland fisheries yield. While considering such additional small-meshed fisheries, however, the conservation of the presently successful fishery for 0 . mossambicus should remain a first priority for the reservoir fishery management. Formulation of a management strategy is therefore necessary, both for the existing fishery on 0.mossambicus and for optional subsidiary fisheries on presently unexploited species.
The fish community structure, fish population dynamics and characteristics of the reservoir fishery were studied in Tissawewa, a typical irrigation reservoir of 200 ha. in the south eastern lowland of Sri Lanka. The major part of the fish community in Tissawewa consists of a group of eleven common species, including two introduced tilapias: 0 . mossambicus and 0 . niloticus (L.). The other species are indigenous to Sri Lanka, including five riverine cyprinids: Amblypharyngodon melettinus (Valenciennes), Barbus chola (Hamilton), B. dorsalis (Jerdon), B. sarana (Hamilton) and Rasbora daniconius (Hamilton); two riverine catfishes: Mystus gulio (Hamilton) and M.vittatus (Bloch); one riverine goby: Glossogobius giuris (Hamilton) and one estuarine halfbeak: Hyporamphus gaimardi (Valenciennes). Several other species are present in the reservoir, but only in small quantities.
A major part of the fish biomass (54%) is formed by the small (max. length 10 cm) pelagic cyprinid A. melettinus, which feeds directly on phytoplankton and detritus. Detritus is the most important food item for 0. mossambicus (max. length 30 cm), which represents 7% of the fish biomass. The group of minor Barbus spp. (max. length 30 cm), representing 25% of the fish biomass, feeds mainly on small zoobenthos, which is also the main item in the diet of Mystus spp. (max. length 30 cm). These catfish are partly piscivorous at larger sizes. Zooplanktivorous fish, like R. daniconius (max. length 15 cm) and H.gaimardi (max. length 20 cm), represent less than 10% of the fish biomass and piscivorous species like, G. giuris (max. length 30 cm), even less than 5%. The total fish biomass is estimated at 1 829 kg/ha, with a total biological fish production of 5 400 kg/ha/yr. The biological fish production per species (-group) is 3 600 kg/ha/yr small pelagic cyprinids (90% A . melettinus and 10% R. daniconius), 1 100 kg/ha/yr Barbus spp., 400 kg/ha/yr 0 . mossambicus and 300 kg/ha/yr other species.
The commercial fishery in Tissawewa is typical for Sri Lankan lowland reservoirs, with gillnets fished from outrigger canoes as the most important gear. Daily effort remains around 9 fishing trips as long as water levels do not drop too low. A fishing trip normally includes one or two fishermen who set two gillnets of about 225 m long, usually during the night. The most common mesh sizes are 64, 70 and 76 mm stretched mesh, used in a ratio of 4:4:1 in terms of numbers of nets in operation. The mean catch per trip in the period 1991-1992 was around 14 kg and the annual yield in this period was 242 kg/ha, with 0.mossambicus accounting for about 70% of the catch. 0 . niloticus accounted for 5% and stocked Indian carps for 20% of the catch in this period. The relative importance of 0 . niloticus seems to be increasing during recent years. Indigenous species represented only 5% of the total catch.
0. mossambicus in Tissawewa are growing to about 14 cm in one year and are caught in gillnets during their second year of life. The modal length in commercial catch length frequency distributions is invariably found at 15.5 cm. The mean instantaneous fishing mortality ( F ) for the length-range of 15.0 to 18.5 cm is high, with F = 4 .7 yr -1and F = 5 .5 yr -1for males and females respectively. The mean fishing mortality for the total exploited length-range of 10.0 to 19.0 cm is 1.6 yr -1The instantaneous natural mortality ( M ) for length classes from 10 cm upwards is also high, with M = 1.7 yr -1and M = 1 .9 yr -1for males and females respectively. Due to the high fishing pressure, specimen of 0. mossambicus larger then 19 cm are rare in Tissawewa, although estimates of potential growth indicate that lengths above 25 cm could be reached after 3 to 5 years (for males and females respectively). The size at maturity of female 0 . mossambicus has decreased from 16 to 13 cm during recent years, and a danger of growth overfishing is clearly present.
The management options for Sri Lankan reservoir fisheries can be divided in three major categories. The first category is formed by management measures related to the exploitation of introduced tilapias. The conservation of the fishery for 0 . mossambicus, in view of its present importance, should have the greatest priority, not only within in this category but in the overall context of reservoir fishery management. A large (>25%) increase in the catch is not expected from any management measure in this fishery, which is near the level of maximum sustainable yield (MSY). Conservation of the existing fishery, by reenforcement of the legal minimum mesh size of 76 mm stretched, is therefore recommended in the present study. This mesh size would reduce the chance of stunting in 0 . mossambicus populations since the peak in the catch length-frequency distributions would shift from 15 to 17-18 cm, leaving more fish of 16 cm a chance to reproduce. The results of this '76 mm option' should be monitored through catch and effort data recording. If catch levels would increase with more than 10%, following this measure, and if 0.mossambicus of 18 cm would become dominant in the catch, a legal minimum mesh size of 82 mm could be implemented. Larger mesh sizes would certainly be beneficial when 0 . niloticus would replace 0 . mossambicus as the most abundant tilapia species, but legal minimum mesh sizes above 88 mm stretched mesh are not recommended for the fishery on 0 . mossambicus. Any increase in fishing effort in this fishery should be prevented but forced reduction of the effort is not recommended either, since this would lead to a decrease in total yield. Since the costs of fishing are very low in this fishery, the income of the fishermen is determined by their total returns rather than by the catch per unit of effort. The total catch from inland fisheries is also important as a source of high-protein food for the rural population.
The second category of management options contains all measures related to the direct exploitation of indigenous species. Two groups of minor cyprinids, the Barbus spp. and A . melettinus, have been identified as candidates for proposed additional fisheries parallel to the existing fishery for tilapias. These two groups both represent a major part of the fish biomass and production in the reservoirs. These additional fisheries, however, would lead to the use of even smaller mesh sizes as presently operated, since the cyprinids are smaller than the exploited tilapias (Fig. A). Mesh sizes as small as 30 and 15 mm stretched mesh would be needed to exploit the Barbus spp. and A. melettinus respectively. Therefore these additional fisheries should only be considered if technical interactions with the existing fishery can be prevented. In case of the Barbus spp. this is not possible since these species show considerable habitat overlap with the juveniles of 0 . mossambicus. Since the smallest stages of 0. mossambicus, vulnerable to mesh sizes of 15 mm, are spatially segregated from A. melettinus, this additional fishery may be feasible without causing serious damage to the existing fishery. Based on the large biomass and production of A. melettinus, a subsidiary fishery with 15 mm stretched mesh gillnets and/or lift nets may at least equal the present yield of 0. mossambicus.
The third category of management options contains all measures related to culture-based fisheries with exotic species. A government enhanced stocking program with major Indian carps was ended in 1990, apparently for religious reasons. Poor returns left this program open to criticism and privatization has sofar not taken place. A culture-based fishery with controllable (non-reproducing) piscivorous fish is proposed in the present study, as a means of indirect exploitation of the large production of small indigenous species. The effect of piscivorous fish on the existing fishery must be carefully examined in isolated situations since such predators are presently rare in reservoir fish communities in Sri Lanka. The best candidate for such a culture-based fishery seems to be Lates calcarifer (Bloch), an estuarine predator, indigenous to Sri Lanka, which grows well in fresh water reservoirs but does not reproduce in land-locked situations. L. calcarifer is a prized consumption fish all over south east Asia. Due to food conversion, the resulting increase in catch from this option will be an order of size lower than in the case of direct exploitation of small cyprinids, but the product will be of a much higher quality.
The inland fishermen in Sri Lanka are relatively well organized in fishermen societies which exist around all the major reservoirs. These small units show a considerable amount of internal control and are able to practice a limited access to the fishery of each reservoir. The fishermen societies form a useful interface for fisheries management and can easily be contacted by extension workers. Fishermen societies are willing to accept management regulations, if these are unanimously agreed upon and if they are strictly enforced by the government. This situation should be taken advantage of when management measures are to be implemented.