Monitoring parasitic abundance in cage-based aquaculture: the effects of clustering

Most discussions of sampling protocols within the literature on monitoring aquatic parasites are based on the assumptions of simple random sampling. There has been a growing recognition within the fields of human and terrestrial veterinary epidemiology that data are often collected from individuals within clusters where such assumptions are not valid. These circumstances arise when monitoring ectoparasitic sea lice on Scottish salmon farms. In previous work the authors have demonstrated that significant intraclass correlation coefficients (ICC) values are associated with cage-level abundance of sea lice, particularly when the parasite reaches its adult stage of development. In this paper two sets of data from Scottish farms with ICC values for adult L. salmonis of 0.35 [0.08-0.72, 95%CI] and for adult C. elongatus of 0.42 [0.14-0.66, 95%CI] are used to investigate the implications of clustering. A Monte Carlo simulation approach is used to illustrate the effect of various sampling approaches. The protocols simulated reflect those typically used across a range of countries and production environments in which salmon are currently reared. By illustrating clearly from empirical data sets what is known by theoretical argument it is hoped that guidelines for sampling parasites, and disease monitoring more generally, within aquaculture will in future incorporate appropriate consideration of issues related to the clustering that is typically present in cage-based production systems.