Risk assessment framework for ballast water introductions – Volume II

Hayes, K.R. and Hewitt, C.L. (2000) Risk assessment framework for ballast water introductions – Volume II. CSIRO Marine Research

Abstract

This report provides a detailed description of the ballast-water risk assessment framework developed by the Centre for Research on Introduced Marine Pests (CRIMP), on behalf of the Australian Quarantine and Inspection Service (AQIS). The report also includes the preliminary results of a demonstration project designed to estimate the ballast water risk posed by Asterias amurensis and Gymnodinium catenatum for vessels arriving in Newcastle from selected ports in Japan.

The risk assessment framework is both modular and hierarchical, allowing increasingly accurate estimates of risk as more data is made available to the analyst. Risk estimates are made on a per vessel, per species basis, for the month in which the vessel intends to de-ballast in the recipient port. Ballast water risk is defined as

Risk = p(ω).p(φ).p(ψ).p(υ) species ,

where p(ω) is the probability that the donor port is infected with the species, p(φ) is the probability that the vessel becomes infected with this species, p(ψ) is the probability that the species survives the vessel’s journey and p(υ) is the probability that the species will survive in the recipient port.

The probability that the donor port is infected p(ω) should be determined via a survey – ideally one designed to allow an objective estimate of the probability of Type II error (ie the species is present but undetected). As an interim measure, the infection status of the donor port bioregion can be used as a surrogate for international ports that have not been surveyed.

A fault tree analysis identifies ten infection scenarios that are mutually exclusive for most species. The assessment framework uses these infection scenarios to quantify the probability of vessel infection p(φ). For large complex ports it will be difficult to accurately quantify the probability of infection because third party vessels will influence the vertical and horizontal distribution of target species, and the ballast withdrawal envelope described by the target vessel. For species that exhibit resistant or diapause life-stages, however, this is a very important component of the assessment because substantial risk reductions may not be achieved elsewhere in the assessment framework.

The probability of journey survival p(ψ) is estimated by comparing the species life expectancy in the ballast tank with the vessel’s journey duration. Uncertainty regarding the species life expectancy is expressed through a probability distribution. Birth-death models were avoided in this context because it is very difficult to estimate the initial inoculum size on any given ballast event. By contrast it is much easier to measure the life expectancy using on-board sampling. The probability of survival in the recipient port p(υ) is estimated by comparing the species temperature and salinity tolerances with the probability distribution of salinity and temperature in the recipient port. The recipient port is divided in environmental sub-units for the purposes of this analysis. Ideally the temperature and salinity extremes of each environmental sub-unit are characterised by monthly extreme value distributions. The risk assessment framework allows for kernel density estimates and sample distribution functions, however, if there is insufficient data to fit an extreme value model.

Item Type:	Report
Series Name:	Centre for Research on Introduced Marine Pests. Technical Report No. 21
Publisher:	CSIRO Marine Research
URI:	http://researchrepository.murdoch.edu.au/id/eprint/64834

Item Control Page

Downloads