Appropriate sampling strategies to estimate sea lice prevalence on salmon farms with low infestation levels

Effective sampling is essential to monitoring and controlling sea lice infestations on salmon farms. However, official sampling regimes are often inadequate, typically adopting a one-size-fits-all approach. Over the past decade, the thresholds at which mandatory treatment is required has also been reduced in many regions and, as such, the total infestation loads reported from salmon farms tend to be lower. Therefore, the use of prevalence, as opposed to the more conventionally used abundance, becomes a metric of interest and, from an analytical perspective, offers some beneficial characteristics. This paper explores a range of sampling scenarios and their impacts on the accuracy of sea lice estimation, particularly when prevalence is the adopted metric. Empirical sea lice count data demonstrated a good fit to the negative binomial distribution and provided a probable range of values that could be used to describe typical levels of over-dispersion. It was demonstrated that, when prevalence is low, it can be reliably used to predict abundance. Monte Carlo simulations of a hypothetical salmon farm were then used to test results for a variety of sample sizes and sea lice infestation scenarios. Different sea lice infestation levels between pens in a farm (i.e. a spatial clustering effect) and aggregations of sea lice on their hosts (i.e. the effect of over-dispersion) were simulated to explore a variety of conditions. The extent to which higher levels of clustering and/or over-dispersion necessitate the need for larger sample sizes to achieve similar levels of accuracy was explored. The level of accuracy that can be achieved in practice depends on many factors and what is considered to be an acceptable level of accuracy will, by definition, be subjective and vary according to the purpose for which the estimation is being carried out. This study includes a variety of possible situations to guide farm operators in choosing sample sizes according to their particular requirements. Furthermore, appropriate sample size determination can be expected to reduce overall sampling effort, achieve better overall control, help avoid unnecessary treatments, and reduce both associated costs and fish welfare impacts.