Yield predictive models for Sri Lankan reservoir fisheries

Tropical reservoirs are primarily constructed for irrigation, generation of hydroelectricity and water supply schemes. Development of inland fisheries is a secondary use of most reservoirs. In Sri Lanka, most reservoirs are s cattered in the rural areas of the country so that investigation of the fis heries of individual reservoirs with a view to developing management plans is prohibitive. The present study was instigated to explore the possibiliti es of developing suitable yield predictive models, which can be used in dev eloping management strategies for the Sri Lankan reservoirs. The study was carried out in 11 perennial reservoirs of Sri Lanka. Basic limnological par ameters (conductivity, dissolved phosphorus, total phosphorus, chlorophyll a [chl a] content and alkalinity) were determined in each of these reservoi rs. Daily data on fish catch and fishing effort were collected in each rese rvoir. Data on catchment areas (CA), reservoir area (RA) and reservoir capa city (RC) were obtained from the irrigation and survey departments. It is e vident that chi a is positively influenced by nutrients (dissolved phosphor us and total phosphorus), morphoedaphic indices derived as alkalinity to me an depth (MEIa) and conductivity to mean depth (MEIc) ratios and CA/RC rati os. MEIa and MEIc are also positively influenced by CA/RC ratios. All these morphological and edaphic parameters were found to positively influence fi sh yield in reservoirs. As fishing intensity (FI) is also a major determina nt of fish yields, fish yield was better accounted by multiple regression m odels in which FI and individual morphological and edaphic parameters were used as independent variables. Of these multiple regression relationships, the best predictive power for fish yield (Y in kg ha(-1) yr(-1)) was found by Y = 18.9 +/- 6.78 FI + 0.0073 CA/RC where FI is expressed as boat-days h a(-1) yr(-1) and CA and RC are in km(2) and km(3), respectively. In this re lationship, FI and CA/RC account for about 68% of the variation in fish yie ld.