Global uncertainty and sensitivity analysis of a food-web bioaccumulation model

Estimating n-octanol-water partition coefficients for neutral highly hydrophobic chemicals using measured n-butanol-water partition coefficients

Direct measurement of the n-octanol partition coefficients (KOW) for highly hydrophobic organic chemicals is extremely difficult because of the extremely low concentrations present in the water phase. n-Butanol/water partition coefficients (KBW) are generally much lower than KOW due to the increased solubility of solute in the alcohol saturated aqueous phase, and therefore become easier to measure. We measured the KBW for 25 neutral organic chemicals having measured log KOWs ranging from 2 to 9 and 4 additional highly hydrophobic chemicals, with unmeasured KOWs, having estimated log KOWs ranging from 6 to 18. The measured log KBW and log KOW values were linearly related, r2 = 0.978, and using the regression developed from the data, KOWs were predicted for the 4 highly hydrophobic chemicals with unmeasured KOWs. The resulting predictions were orders of magnitude lower than those predicted by a variety of computational models and suggests the estimates of KOW in the literature for highly hydrophobic chemicals (i.e., log KOW greater than 10) are likely incorrect by several orders of magnitude.

Modelling ecological and human exposure to POPs in Venice lagoon - Part II: Quantitative uncertainty and sensitivity analysis in coupled exposure models

The study is focused on applying uncertainty and sensitivity analysis to support the application and evaluation of large exposure models where a significant number of parameters and complex exposure scenarios might be involved. The recently developed MERLIN-Expo exposure modelling tool was applied to probabilistically assess the ecological and human exposure to PCB 126 and 2,3,7,8-TCDD in the Venice lagoon (Italy). The 'Phytoplankton', 'Aquatic Invertebrate', 'Fish', 'Human intake' and PBPK models available in MERLIN-Expo library were integrated to create a specific food web to dynamically simulate bioaccumulation in various aquatic species and in the human body over individual lifetimes from 1932 until 1998. MERLIN-Expo is a high tier exposure modelling tool allowing propagation of uncertainty on the model predictions through Monte Carlo simulation. Uncertainty in model output can be further apportioned between parameters by applying built-in sensitivity analysis tools. In this study, uncertainty has been extensively addressed in the distribution functions to describe the data input and the effect on model results by applying sensitivity analysis techniques (screening Morris method, regression analysis, and variance-based method EFAST). In the exposure scenario developed for the Lagoon of Venice, the concentrations of 2,3,7,8-TCDD and PCB 126 in human blood turned out to be mainly influenced by a combination of parameters (half-lives of the chemicals, body weight variability, lipid fraction, food assimilation efficiency), physiological processes (uptake/elimination rates), environmental exposure concentrations (sediment, water, food) and eating behaviours (amount of food eaten). In conclusion, this case study demonstrated feasibility of MERLIN-Expo to be successfully employed in integrated, high tier exposure assessment.

Modelling the Stoichiometric Regulation of C-Rich Toxins in Marine Dinoflagellates

Toxin production in marine microalgae was previously shown to be tightly coupled with cellular stoichiometry. The highest values of cellular toxin are in fact mainly associated with a high carbon to nutrient cellular ratio. In particular, the cellular accumulation of C-rich toxins (i.e., with C:N > 6.6) can be stimulated by both N and P deficiency. Dinoflagellates are the main producers of C-rich toxins and may represent a serious threat for human health and the marine ecosystem. As such, the development of a numerical model able to predict how toxin production is stimulated by nutrient supply/deficiency is of primary utility for both scientific and management purposes. In this work we have developed a mechanistic model describing the stoichiometric regulation of C-rich toxins in marine dinoflagellates. To this purpose, a new formulation describing toxin production and fate was embedded in the European Regional Seas Ecosystem Model (ERSEM), here simplified to describe a monospecific batch culture. Toxin production was assumed to be composed by two distinct additive terms; the first is a constant fraction of algal production and is assumed to take place at any physiological conditions. The second term is assumed to be dependent on algal biomass and to be stimulated by internal nutrient deficiency. By using these assumptions, the model reproduced the concentrations and temporal evolution of toxins observed in cultures of Ostreopsis cf. ovata, a benthic/epiphytic dinoflagellate producing C-rich toxins named ovatoxins. The analysis of simulations and their comparison with experimental data provided a conceptual model linking toxin production and nutritional status in this species. The model was also qualitatively validated by using independent literature data, and the results indicate that our formulation can be also used to simulate toxin dynamics in other dinoflagellates. Our model represents an important step towards the simulation and prediction of marine algal toxicity.

Estimating trophic levels and trophic magnification factors using Bayesian inference

Food web biomagnification is increasingly assessed by estimating trophic magnification factors (TMF) where solvent (often lipid) normalized contaminant concentration is regressed onto the trophic level, and TMFs are represented by the slope of the relationship. In TMF regressions, the uncertainty in the contaminant concentrations is appreciated, whereas the trophic levels are assumed independent and not associated with variability or uncertainty pertaining to e.g. quantification. In reality, the trophic levels may vary due to measurement error in stable isotopes of nitrogen (δ(15)N) of each sample, in δ(15)N in selected reference baseline trophic level, and in the enrichment factor of δ(15)N between two trophic levels (ΔN), which are all needed to calculate trophic levels. The present study used a Markov Chain Monte Carlo method, with knowledge about the food web structure, which resulted in a dramatic increase in the precision in the TMF estimates. This also lead to a better understanding of the uncertainties in bioaccumulation measures; instead of using point estimates of TMF, the uncertainty can be quantified (i.e., TMF >1, namely positive biomagnification, with an estimated X % probability).

Uncertainties in a Baltic sea food-web model reveal challenges for future projections

Models that can project ecosystem dynamics under changing environmental conditions are in high demand. The application of such models, however, requires model validation together with analyses of model uncertainties, which are both often overlooked. We carried out a simplified model uncertainty and sensitivity analysis on an Ecopath with Ecosim food-web model of the Baltic Proper (BaltProWeb) and found the model sensitive to both variations in the input data of pre-identified key groups and environmental forcing. Model uncertainties grew particularly high in future climate change scenarios. For example, cod fishery recommendations that resulted in viable stocks in the original model failed after data uncertainties were introduced. In addition, addressing the trophic control dynamics produced by the food-web model proved as a useful tool for both model validation, and for studying the food-web function. These results indicate that presenting model uncertainties is necessary to alleviate ecological surprises in marine ecosystem management.

Estimating chemical biotransformation rates from food web concentrations

Biotransformation is widely recognized as the most important and most uncertain determinant of bioaccumulation. A step-wise method for estimating organism-specific biotransformation half-lives from field observations and using established food web modeling is developed. As a proof of concept, the method is applied to the case of nine polycyclic aromatic hydrocarbons (PAHs) in a well-studied food web in Bohai Bay, China. The estimated half-lives are in good agreement with the existing literature. The proposed biotransformation estimation method, through data mining, for sufficiently defined ecosystems, may greatly reduce the necessary animal testing involved in chemical assessments by providing useful guidance to experimentalists and regulators.

Explaining differences between bioaccumulation measurements in laboratory and field data through use of a probabilistic modeling approach

In the regulatory context, bioaccumulation assessment is often hampered by substantial data uncertainty as well as by the poorly understood differences often observed between results from laboratory and field bioaccumulation studies. Bioaccumulation is a complex, multifaceted process, which calls for accurate error analysis. Yet, attempts to quantify and compare propagation of error in bioaccumulation metrics across species and chemicals are rare. Here, we quantitatively assessed the combined influence of physicochemical, physiological, ecological, and environmental parameters known to affect bioaccumulation for 4 species and 2 chemicals, to assess whether uncertainty in these factors can explain the observed differences among laboratory and field studies. The organisms evaluated in simulations including mayfly larvae, deposit-feeding polychaetes, yellow perch, and little owl represented a range of ecological conditions and biotransformation capacity. The chemicals, pyrene and the polychlorinated biphenyl congener PCB-153, represented medium and highly hydrophobic chemicals with different susceptibilities to biotransformation. An existing state of the art probabilistic bioaccumulation model was improved by accounting for bioavailability and absorption efficiency limitations, due to the presence of black carbon in sediment, and was used for probabilistic modeling of variability and propagation of error. Results showed that at lower trophic levels (mayfly and polychaete), variability in bioaccumulation was mainly driven by sediment exposure, sediment composition and chemical partitioning to sediment components, which was in turn dominated by the influence of black carbon. At higher trophic levels (yellow perch and the little owl), food web structure (i.e., diet composition and abundance) and chemical concentration in the diet became more important particularly for the most persistent compound, PCB-153. These results suggest that variation in bioaccumulation assessment is reduced most by improved identification of food sources as well as by accounting for the chemical bioavailability in food components. Improvements in the accuracy of aqueous exposure appear to be less relevant when applied to moderate to highly hydrophobic compounds, because this route contributes only marginally to total uptake. The determination of chemical bioavailability and the increase in understanding and qualifying the role of sediment components (black carbon, labile organic matter, and the like) on chemical absorption efficiencies has been identified as a key next steps.

Parameter uncertainty in modeling bioaccumulation factors of fish

We quantified the uncertainty due to biota-related parameters in estimated bioaccumulation factors (BAFs) of persistent organic pollutants for fish through Monte Carlo simulations. For this purpose, the bioaccumulation model OMEGA (Optimal Modeling for EcotoxicoloGical Applications) was parameterized based on data from the existing literature, analysis of allometric data, and maximum likelihood estimation. Lipid contents, fractions of food assimilated, the allometric rate exponent, normalized food intakes, respiration and growth dilution rates, and partial mass transfer resistances in water and lipid layers were included as uncertain parameters. The uncertainty in partial resistances was particularly important in the estimation of the rate constants for chemical intake from water by fish. Uncertainties in the fractions of food assimilated and partial water layer resistances from and to food were particularly important in the estimation of the rate constants of chemical intake from food. The uncertainty in the model outcomes for the bioaccumulation factors for fish was a factor of 10 (ratio of 95th and fifth percentile estimates), which was mainly caused by the uncertainty in the lipid fraction. For chemicals with a K(OW) of 10(3) to 10(6), the uncertainty in the lipid contents of fish accounted for more than 50% of the uncertainty in the estimated bioaccumulation factor. For chemicals with a high K(OW) (10(7) and higher), the fractions of food assimilated and partial resistances also contributed to uncertainty in the estimated bioaccumulation factor (up to 60%). A case study showed that uncertainty in estimated BAF for nonpersistent substances can be dominated by uncertainty in the rate constants for metabolic transformation.

Uncertainties in ecological, chemical and physiological parameters of a bioaccumulation model: implications for internal concentrations and tissue based risk quotients

Bioaccumulation models predict internal contaminant concentrations (c(i)) using ecological, chemical and physiological parameters. Here we analyse the effect of uncertainties on these parameters on bioaccumulation model predictions. Simultaneously considering the uncertainties on all these parameters in a bioaccumulation model resulted in uncertainty ranges of c(i) that increased with the octanol water partition coefficient K(ow) and reached maxima of up to 1.25 log units for mesozooplankton and up to 1.45 log units fish at logK(ow)=8. A global sensitivity analysis (SA) was performed to rank the contribution of different parameters to the observed uncertainty. The SA demonstrated that this interspecies difference resulted predominantly from uncertain production rates of fish. The K(ow), the water concentration and organic carbon-octanol proportionality constant were important drivers of uncertainty on c(i) for both species. A tissue based risk quotient (RQ(tissue)) combining uncertainty on c(i) with realistic tissue based effect thresholds indicated that fish were up to 10 times more probable to have RQ(tissue)>1 than mesozooplankton, depending on the considered threshold value. Conventional exposure based risk quotients were up to 5 times less probable to exceed one than were corresponding RQ(tissue), and this for both species.

The influence of diet on the assimilation efficiency of 47 polychlorinated biphenyl congeners in Japanese koi (Cyprinus carpio)

The influence of diet on polychlorinated biphenyl (PCB) assimilation was investigated by measuring dietary assimilation efficiencies (AEs) for 47 PCB congeners in juvenile koi (Cyprinus carpio) fed five experimental diets. Two of the diets were naturally contaminated and were obtained by collecting mayflies from Lake Erie (ON, Canada) and emerald shiners from the Detroit River (MI, USA). The remaining diets consisted of commercial fish pellets (lipid contents from 6.7 to 24%) that were contaminated by spiking with a PCB mixture. Experimental fish were held individually to quantify the amount of food consumed per fish and, following a 48-h fasting period to facilitate food digestion and assimilation; AEs were determined by mass balance. Fish fed the benthic invertebrate food exhibited the highest PCB AEs (70-101%) and were significantly elevated compared to the other diet treatments (AEs ranging from 23 to 87%). The PCB AEs for fish fed emerald shiners did not differ from those fed pellet formulations. Variation among PCB AEs was not related to diet lipid content. For all diet treatments, PCB AEs were significantly related to chemical hydrophobicity. The relationship between chemical AE and n-octanol/water partition coefficient (K(OW)) was best explained by a linear model compared to a two-phase resistance model. Overall, PCB AEs were observed to be dependent on both diet type and chemical hydrophobicity, with both factors contributing nearly equally to the variation measured in this toxicokinetic parameter.