Modelling of Taste-Taint in Fish ... 55leverage significant (surprise) change in product or plant (Davey et al., 2015).
They used probability distributions for the input parameters to mimic naturally
occurring fluctuations in parameters. The output therefore was also a
distribution of scenarios, with the probability of each occurring (e.g., Davey et
al., 2015; Abdul-Halim and Davey, 2015; Davey, 2015; 2011; Davey et al.,
2013).
Hathurusingha and Davey (2016 c) recently applied this risk methodology
to the model of Hathurusingha and Davey (2014) to demonstrate quantitatively
the impact of naturally occurring fluctuations in RAS water (growth
environment) on taste-taint accumulation in fish-flesh. A significant new
insight was that taste-taint chemical as GSM and MIB in RAS farming of
barramundi can be significantly impacted by naturally occurring, small
fluctuations in key process parameters. The accumulation of taste-taint as a
result of these fluctuations could unexpectedly lead to the consumer threshold
being exceeded i.e., taste taint failure - with consequent economic losses to
farmers.
The approach was based on the taint chemicals entering the fish-flesh via
the gills and dilution through metabolism and growth, together with a chemical
taint risk factor (p) such that for all p > 0 the taint chemical was above a
desired threshold concentration (which includes a practical tolerance of 10 %)
respectively, 0.814 and 0.77, μg kg-^1 for GSM and MIB. Monte Carlo (with
Latin Hypercube) sampling of chemical in the growth water, water
temperature and growth time were used to simulate practical RAS farmed
barramundi for up to 260 days growth.
It was concluded that some 10.10 % of all harvests over the long term
could result in fish with taste-taint as GSM above the threshold concentration
due to natural fluctuations in the RAS environment. For MIB this failure rate
was 10.56 %. The vulnerability to taste-taint failure was shown to be impacted
highly significantly by the time to harvest, and to a lesser extent by
concentration and fluctuation of the taint chemicals in the RAS water. They
showed further the time to harvest could be readily monitored and practically
controlled by farmers to limit taste-taint accumulation as GSM and MIB. A
time to harvest of > 240 day would result in rapidly increasing vulnerability to
taint accumulation. Findings appeared to be of immediate benefit to RAS
farmers and to risk researchers in foods processing.
Pointedly, this insight could not be obtained from traditional single value
assessment (SVA) computations or with traditional risk and hazard
assessments. This is because this random element is not explicit in these. It is