Study design considerations for assessing the health of fish populations impacted by agriculture in developing countries: a Sri Lankan case study.

Studies on the use of indigenous or endemic fish species for the assessment of agricultural impacts on fish populations are lacking in tropical South and Southeast Asia. This paper describes the steps involved in developing an agricultural impacts assessment program focused on river health, using recent studies on wild fish in Sri Lanka. The assessment methodology includes the selection of fish species for monitoring, and development of a strategy for sample timing, sample size requirements, and selection of appropriate reference site(s). Preliminary fish sampling data from several tributaries of the Uma-oya and the Amban-ganga (Knuckles streams) from the Mahaweli River basin were evaluated and temporal patterns of gonadal recruitment were investigated for three common species: Garra ceylonensis, Devario malabaricus, and Rasbora daniconius. The data on reproductive development were statistically incorporated to evaluate appropriate sample timing and sample size requirements. For this study, we proposed a cluster gradient design with a range of assessment endpoints and suitable statistical methods; an alternate assessment in different agricultural catchments would facilitate verification. The review and preliminary data support provide a template for study design considerations for agricultural impact assessments in South and SE Asian countries.