Statistical methods for water quality predictions in shellfish farms to support management decisions

Events of poor water quality due to increase of E. coli or marine toxin concentrations can result in temporary closures of shellfish aquaculture harvesting. This can lead to substantial financial loss for the aquaculture... [ view full abstract ]

Events of poor water quality due to increase of E. coli or marine toxin concentrations can result in temporary closures of shellfish aquaculture harvesting. This can lead to substantial financial loss for the aquaculture business and reduced consumer confidence in shellfish products. The ‘ShellEye’ project aims to provide shellfish farmers with early warning of the occurrence of microbiological and toxic phytoplankton events.
We first identified the environmental drivers for E. coli and marine toxin (Okadaic acid/ Dinophysis toxins/ Pectenotoxins) levels for a test site in the Celtic Sea. Near-real time versions of these variables (in situ measurements, satellite Earth observations and meteorological forecasts) were then used to develop statistical prediction models. Over a three-month trial period, the accuracy of a suite of prediction models were evaluated through comparisons of predicted and observed levels of E. coli and marine toxin concentrations at the test site. Whilst the models were unable to predict the absolute levels to an acceptable accuracy, they were able to predict the weekly variation for both groups. The results obtained to date support the use of these prediction models for indicating how water quality changes within a farm. We present the method and show initial results of including these predictions in a trial bulletin service for a shellfish farmer to support management decisions. We will also discuss how further work will focus on extending these models to additional shellfish waters to ensure wider exploitation.