Characteristics of a Sri Lankan reservoir fishery and consequences for the estimation of annual yield

Abstract

A quantitative analysis of dynamics in catch, effort and catch per unit of effort (CpUE) was applied to indicate how a system for the collection of catch and effort data could be optimised and how the accuracy of fisheries statistics could be improved. Temporal and spatial patterns in the commercial fisheries were studied in Tissawewa, a representative irrigation reservoir in Sri Lanka. Considerable fluctuations of water level in shallow waters are important characteristics of the irrigation reservoirs in Sri lanka. Reservoir water level and surface area in Tissawewa fluctuated around means of 2.30 m and 175 ha respectively. The reservoir almost ran dry halfway through the study period. After 2 months the reservoir filled up again. The total annual catch from Tissawewa was estimated at 42 431 kg year −1 over the first year, and 11 011 kg year −1 over the second. The annual yield was estimated to range from 150 to 250 kg ha −1 year −1 depending on the occurrence of dry periods. The main fishing methods were normal (stationary) gillnetting, cast netting, seining with gillnets and ‘water beating’. Normal gillnets, usually with 64 or 70 mm stretched mesh, were by far the most important gear. The CpUE was defined as the catch of a single fishing trip by an outrigger canoe. Strata in the CpUE data set were identified to enable the development of an efficient catch and effort data recording system (CEDRS). The overall mean catch per trip was not significantly different between fishing methods or between fishing areas. Significant differences, explaining 30% of the total variance in CpUE, were observed between sampling months. A CEDRS with an accuracy of 28% for the estimated annual catch, 17% for the catch per unit of effort, and 11% for the effort was proposed for the reservoir fishery. Not more than one sampling day per month is needed to achieve this accuracy in a reservoir like Tissawewa. This is a 50% reduction in sampling investments compared with the former official CEDRS for Sri Lankan reservoirs. Application of the proposed CEDRS will remove the bias which presently occurs in the official data, which are based on the number of registered canoes and not on actual recordings of effort.