A Bayesian synthesis of predictions from different models for setting water quality criteria

► Bayesian model averaging is used to synthesize forecasts from two mechanistic models. ► We use a weighting scheme that penalizes the likelihood of calibration bias. ► Ensemble ecological forecasts integrate across different system conceptualizations. ► The water quality criteria setting process should accommodate natural variability. ► Internal loading can be a regulatory factor of the restoration pace of the Harbour.