Modelling pesticides residues | EPFL Graph Search

This work is a contribution to the development of a specific method to assess the presence of residues in agricultural commodities. The following objectives are formulated: to identify and describe main processes in environment — plant exchanges, to build of a model to assess the residue concentration at harvest in agricultural commodities, to understand the functioning of the modelled system, to characterise pesticides used in field crops and identify optimisation potentials in phytosanitary measures. The frame for the methodological developments corresponds to the procedure for the evaluation of the toxicity provided for the Life Cycle Impact Assessment methodology and for the method Impact 2002+. In chapter 2, the methodological procedure for the assessment of human toxicity potential is introduced. First the factors of fate and exposure are described, including the notion of harvest fraction, the amount of substance found in harvest per unit of substance emitted initially in the system, the main result of the present study. Then the effect factors and the framework for impact evaluation are introduced. Chapter 3 describes the principles accounted for the building of the fate model. Wheat crop and a restricted list of substances are chosen for these methodological developments. The model is composed by compartments describing the environment and the plant. Its functioning is based on initial amounts of substance in the source compartments, on transfer rates linking the compartments and on a dynamic evolution as a function of time between the treatment and the harvest. Air, soil and formulation deposit on plant are the primer compartments receiving the treated substance. Each transport is described by a transfer rate accounting for the process and for the equilibrium partitioning between the two exchanging compartments. Degradation of substance and plant growth are additional processes considered. Each compartment is described by a linear differential equation for the variation of mass accumulating and dissipating. Their assembly builds the model solved as a function of time. This exact resolution is complemented by additional tools to better understand the system functioning and to provide further approximations of the results: the system is simplified into subsystems describing the source and the receiving plant compartment and analytically solved using interpretable equations. Chapter 4 describes and discusses all transport and dissipation processes determining the fate of the substance in the limits of the system. The recent publications concerning the understanding and the modelling of pesticide transfer from formulation deposit on plant through the cuticular membranes give new possibilities to model pesticide fate and to better account for the direct applications on the plant. In chapter 5, the core model is first applied and its functioning analysed. The low availability and partly unsatisfying quality of data for pesticides description is a main complication for the methodology application: the lack of data for the half-life of the substance in the plant especially leads to a strong extrapolation for this determinant factor. A large difference is observed between early and late applied pesticides with respectively a major release to soil or a release to formulation deposit on plant surface. The initial transport processes quickly distribute the substance in the system. Once each plant compartment has accumulated residues up to a maximum amount, a dissipation phase occurs. The duration of these periods is determinant for the level of residue in harvest. The soil is a determinant source for long term evolutions of the system, for soon applied substances with low degradation rate. The half-life of substance deposited on plant is equal to a few days, but the transfer is fast from formulation deposit to the inner plant, where degradation is generally much slower. The accumulation of substance from the air is mostly negligible. The sum of the subsystems gives an approximation of the total system, useful for interpretation. The possibility to simplify the subsystem by ignoring the transfer back from receiving to the source compartment underlines the low contribution of these transfers in the functioning of the model. An approximated resolution is based on the determination of the maximum accumulated substance and on the subsequent dissipation process. However, an important loss of precision is observed. This approximation is useful for interpretation and for extrapolations. In chapter 6, an evaluation of the model is conducted through a sensitivity and uncertainty analysis. The sensitivity analysis consists of evaluating the effect on the output of a change in an input, on the basis of three complementary approaches: the effect of a fixed change in the input of e.g. 0.1%, the effect of a change specific to the uncertainty of the input and the effect of a change in input value from a minimum to a maximum. The uncertainty of an output is evaluated according to the relative contribution of the confidence factors of the inputs. Results show that the half-lives and the time are the most important factors determining the sensitivity of the system and the propagation of uncertainty. The contribution of the half-life to the confidence factor of the harvest fraction reaches between 30% to 98% of the total uncertainty. The confidence factors of results increase exponentially with the time interval between application and harvest. The role of partition coefficients to the behaviour of substances is highly variable, may be determinant or negligible, with increasing or limiting effect on mobility. Sensitivity and uncertainty for parameters describing the agricultural or environmental system are very variable, but sometimes determinant and so confirmed as essential for the system functioning. Consequently, differences in harvest fractions between substances are only significant if they are high. A first comparison of the computed results with measures of residues obtained by an experiment and with references such as tolerance values lead to a pertinent verification of the overall methodology. Finally, the qualitative comparison with other models underlines the specificities and the originality of the present methodology in particular by comparison with environmental multi-media models running in steady state. In chapter 7, the model is finally applied for an ultimate interpretation. The harvest fractions for more than 100 substances are evaluated. Among all types of substances, low and high levels of residues per treatment are found, representative for the high variability of harvest fractions from 5E-16 for bromoxynil to 7E-03 for tebuconazole sprayed on wheat. The fate process represents the highest source of variation for the toxicity. If the application rate does not explain the high differences in residue level at harvest, the time of application may represent an optimisation potential particularly for late treatments. However, the toxicity needs to account for both fate and effect factors, as only their combination effectively allows to evaluate the toxicity. According to the available list of Human Damage Factors per treatment, problematic substances may be effectively identified and substituted. In chapter 8 answers to questions brought with the objectives bring a conclusion to the study. The appendices include notably the results of harvest fractions and toxicity per unit substance applied, per treatment and per unit cultivated crop area, for the main substances and field crops. A LCA is also presented on the intensity level of wheat production.

Multimedia and food chain modelling of toxics for comparative risk and life cycle impact assessment

Ralph Rosenbaum

The overall goal of this dissertation is to contribute to the development of best available practice in environmental multimedia fate and multipathway exposure modelling for human health and ecosystem impacts, addressing several important aspects. Environmental multimedia models relate environmental emissions to impacts (or risk factors) combining multimedia fate and multipathway exposure estimates with effect assessment data. Many organic chemicals formed in combustion and released to air as well as organic chemicals released to air, water, and soil from industrial activities enter humans primarily through food – in particular through meat and dairy products. Therefore, food chain modelling is of utmost importance for human exposure assessment. Model comparison and quantification of uncertainties of model results are vital for their correct interpretation. This leads to the following questions: How would a model framework be structured favouring flexibility, transparency, comparability, interpretability, without limiting the capabilities of the model parts? How can carry-over modelling of chemical biotransfer into biota such as meat or milk be improved considering limited data availability and high uncertainties? How can environmental multimedia multipathway models be compared in a systematic and detailed way, and what are important sources of differences? How to identify important sources of uncertainty in a model result, including regression models used for estimation and when modelling large sets of chemicals? In order to answer that, the following specific goals are addressed: Develop a consistent matrix algebra framework for multimedia fate, multipathway exposure and toxicity effects models. Review existing biotransfer models for chemical transfer into meat and milk and develop, compare and evaluate an improved carry-over model for meat and milk. Compare selected multimedia fate, multipathway exposure and toxicity effects models on the level of final characterisation factors as well as for intermediate results. Develop and test approaches to assess the uncertainty of model results accounting for regression model uncertainty and main parameters contributing to overall uncertainty. Chapter 1 puts this dissertation into its context and introduces multimedia fate, multipathway exposure and human health and ecosystem effects modelling with a focus on the food chain exposure in the frame of comparative risk and Life Cycle Impact Assessment. Chapter 2 presents a framework for multimedia multipathway models entirely based on matrix algebra. When assessing human health or ecosystem impacts of chemicals several calculation steps need to be addressed. We suggest a matrix algebra framework which includes the fate, exposure, effect, and damage assessment for human health and ecosystems, also applicable to spatial modelling. Special emphasis is laid upon interpretation of the physical meaning of different elements within the matrices. Chapter 3 describes the development, comparison and evaluation of an improved carry-over model for chemical transfer into meat and milk. The objective of this study is to increase the reliability and transparency of chemical biotransfer modelling into meat and milk and explicitly confront the uncertainties in exposure assessments of chemicals that require such estimates. To address this we developed a dynamic three compartment cow model (called CKow), able to distinguish lactating and non-lactating cows, to more accurately track the transport of a chemical into meat fat and milk within the cow. Using the CKow model as base model we derived a simplified model proposing an improved, less uncertain regression model to predict COR for meat and milk based on Kow. Chapter 4 is a detailed comparison of four characterisation methods used in LCIA. The main objective was to investigate whether or not the models' results agree with each other in terms of ranking as well as in absolute values and identify where differences come from. A comparison of characterisation factors and intermediate results of IMPACT 2002, CalTOX, USES-LCA, and EDIP97 has been performed for generic organic compounds. The comparison procedure was undertaken in three tiers: 1) comparison of available published factors, 2) calculated factors based on the harmonised chemical test set and 3) an in depth comparison revealing differences of the models on the level of processes and algorithms. Chapter 5 explores possibilities to estimate the uncertainty of results from multimedia fate and multipathway exposure models and to account for regression model uncertainty. Quantification and communication of uncertainties related to model results is vital for their correct interpretation and use. We explore and compare approaches to identify and quantify main sources of uncertainty, enabling to focus the data acquisition for the uncertainty assessment, especially when modelling large sets of chemicals. Chapter 6 presents general conclusions, wrapping up the achievements of this dissertation and answering the scientific questions raised above.

EPFL2006

Propagation Models for Biochemical Reaction Networks

Maria-Emanuela-Canini Mateescu

In this thesis we investigate different ways of approximating the solution of the chemical master equation (CME). The CME is a system of differential equations that models the stochastic transient behaviour of biochemical reaction networks. It does so by describing the time evolution of probability distribution over the states of a Markov chain that represents a biological network, and thus its stochasticity is only implicit. The transient solution of a CME is the vector of probabilities over the states of the corresponding Markov chain at a certain time point t, and it has traditionally been obtained by applying methods that are general to continuous-time Markov chains: uniformization, Krylov subspace methods, and general ordinary differential equation (ODE) solvers such as the fourth order Runge-Kutta method. Even though biochemical reaction networks are the main application of our work, some of our results are presented in the more general framework of propagation models (PM), a computational formalism that we introduce in the first part of this thesis. Each propagation model N has two associated propagation processes, one in discrete-time and a second one in continuous-time. These propagation processes propagate a generic mass through a discrete state space. For example, in order to model a CME, N propagates probability mass. In the discrete-time case the propagation is done step-wise, while in the continuous-time case it is done in a continuous flow defined by a differential equation. Again, in the case of the chemical master equation, this differential equation is the equivalent of the chemical master equation itself where probability mass is propagated through a discrete state space. Discrete-time propagation processes can encode methods such as the uniformization method and the fourth order Runge-Kutta integration method that we have mentioned above, and thus by optimizing propagation algorithms we optimize both of these methods simultaneously. In the second part of our thesis, we define stochastic hybrid models that approximate the stochastic behaviour of biochemical reaction networks by treating some variables of the system deterministically. This deterministic approximation is done for species with large populations, for which stochasticity does not play an important role. We propose three such hybrid models, which we introduce from the coarsest to the most refined one: (i) the first one replaces some variables of the system with their overall expectations, (ii) the second one replaces some variables of the system with their expectations conditioned on the values of the stochastic variables, (iii) and finally, the third one, splits each variable into a stochastic part (for low valuations) and a deterministic part (for high valuations), while tracking the conditional expectation of the deterministic part. For each of these algorithms we give the corresponding propagation models that propagate not only probabilities but also the respective continuous approximations for the deterministic variables.

EPFL2011

Multimedia and food chain modelling of toxics for comparative risk and life cycle impact assessment

Ralph Rosenbaum

EPFL2006

Propagation Models for Biochemical Reaction Networks

Maria-Emanuela-Canini Mateescu

EPFL2011