A Bayesian framework for assessing fish tumour occurrence across all the Canadian AOCs

Session: Poster session

Ariola Visha, University of Toronto Scarborough, [email protected]
Mark McMaster, Environment and Climate Change Canada, [email protected]
George Arhonditsis, University of Toronto Scarborough, [email protected]

Abstract

Fish tumors and other deformities are a Beneficial Use Impairment established by the International Joint Commission to identify Areas of Concern (AOCs) in the Great Lakes basin. Contaminated sediments have been connected with the development of liver cancer in bottom-dwelling species such as brown bullhead. There has been considerable discussion around the standardization of the methodological procedures associated with fish neoplasm sampling and histopathology in order to ensure objective comparisons between impacted and unimpacted sites. Given the current dialogue about refining delisting criteria, the emerging knowledge about the role of fish covariates, and the growing awareness about the need for probabilistic management frameworks as opposed to strictly deterministic ones, we propose a Bayesian methodological framework that can provide a possible aid for policy decisions across all AOCs in the Great Lakes. The purpose of our project is to develop a series of Bayesian statistical models that can be used to predict liver neoplasm and pre-neoplasm occurrence rates in fish. We illustrate three statistical formulations, Bernoulli, Zero-Inflated Poisson, and Binomial–Poisson, founded upon the causal linkages between the demographics/physical characteristics of the fish samples (e.g., age, fork length, liver/gonad weight, and total fish weight) and the likelihood of tumor incidence. 

1. Keyword
fish diseases

2. Keyword
fish populations

3. Keyword
Great Lakes basin