Bioconcentration factor (BCF) and bioaccumulation factor (BAF) are key metrics for measuring the accumulative potential of a chemical in an organism.  BCF considers chemical uptake via aqueous exposure only, while BAF considers chemical uptake from both water and diet.  In this thesis, BCF models of neutral organic compounds (neutral fraction ≥ 0.9) in amphipod and midge were developed based on first-order kinetics of relevant uptake and elimination processes.  A multiple linear regression model based on Abraham parameters, temperature, organism weight and lipid content was used to predict bioaccumulation of neutral organic compounds in shrimp and prawn.

With negligible contribution of growth dilution (k_G) and egestion (k_E), amphipod BCF was modeled as the respiratory uptake (k₁) over the sum of respiratory elimination (k₂) and biotransformation (k_M).  The model generally performs well within ±1 log unit (RMSE = 0.68).  Approximately 12% of BCFs are underpredicted and over 60% of k₂ estimates exceed the total depuration (k_T) by up to around 2 log units.  The underpredicted BCFs and the excessive k₂ values cannot be fully explained by the error in k₁ or the uncertainties in calculation of organism to water partitioning (K_biow).  Analyses indicate the underestimation in K_biow, which may be improved by incorporating exoskeleton as a relevant partitioning component and / or refining the membrane-water partitioning model.

Midge BCF was modeled as k₁ / k_T where k_T = k₂ + k_M + k_G + k_E.  Over 80% of midge BCFs can be predicted within ±1 log unit (RMSE = 0.67) by the proposed model.  The radiolabeling data quality criterion was found to be peripheral as both radiolabelled and non-radiolabeled BCF data were well predicted by the proposed model.  Establishing a midge-specific biotransformation model and / or refining the current egestion and growth model may further improve the BCF prediction accuracy.

In shrimp and prawn, BAFs measured under either laboratory (lab BAFs) or field condition (field BAFs) are generally higher than BCFs as expected.  Field BAFs were found to vary by as much as 5 log units (at log K_OW = 7.6).  Possible sources including lipid content, organism weight, temperature, pH and organic carbon in water may explain this variability by 0.4, 0.7, 1, 0.4 and 2.8 log units at most, respectively.  A multiple linear regression with 8 predictors including 5 Abraham parameters, temperature, lipid content and organism weight was used to model field BAF.  The model can predict approximate 88% of field BAFs within ±1 log unit (RMSE = 0.71).  Statistics suggests the regression model is reasonable as all 8 parameters contribute significantly in the model.

This work advances the establishment of the invertebrate bioaccumulation database and the construction of the comprehensive bioaccumulation model for organic chemicals in invertebrates.