1007 Environmental Toxicology and Chemistry, Vol. 24, No. 4, pp. 1007–1017, 2005  2005 SETAC Printed in the USA 0730-7268/05 $12.00 + .00 APPLICATION OF THE RICEWQ–VADOFT MODEL FOR SIMULATING THE ENVIRONMENTAL FATE OF PRETILACHLOR IN RICE PADDIES DIMITRIOS GEORGIOS KARPOUZAS,*² A LDO FERRERO,‡ FRANCESCO VIDOTTO,‡ and ETTORE CAPRI² ²Universita ´ Cattolica del Sacro Cuore, Istituto di Chimica Agraria ed Ambientale, 29100 Piacenza, Italy ‡Dipartimento Agronomia, Selvicoltura e Gestione del Territorio, Universita degli Studi, Torino, Italy ( Received 6 April 2004; Accepted 16 September 2004) Abstract—No validated models in Europe are capable of simulating the environmental fate of pesticides under the specific conditions of rice fields. Rice water quality—vadose zone flow and transport (RICEWQ–VADOFT) is a model developed from the coupling of a surface runoff model (RICEWQ) and a vadose zone flow and transport model (VADOFT) for determining predicted environ- mental concentrations in paddy water and sediment, runoff, and groundwater. This study is intended to evaluate the capability of this model to simulate effectively the environmental fate of the herbicide pretilachlor in the paddy environment. A two-year field study conducted in a representative rice-cultivated area of northern Italy provided measured concentrations of pretilachlor in paddy water and sediment and also a limited number of observations on runoff losses. The model successfully predicted the water balance in the paddy field in both years. After limited calibration, the model predicted the fate of pretilachlor in paddy water and sediment with high accuracy. Agreement between predicted and measured concentrations of pretilachlor in both years was assessed statistically using several statistical indicators. For example, modeling efficiency (EF) values of 0.867 to 0.935 and 0.702 to 0.718 in paddy water and sediment, respectively, document the strong agreement between predicted and measured pesticide concentrations. The model predictions showed high agreement with the limited amount of measured runoff data in 2002. The model predicted that no significant amounts of pretilachlor would leach below the top 25 cm of the soil, although no measured data were available to evaluate the predicted results. A sensitivity analysis of the model to variables controlling pesticide partitioning to paddy sediment (VBIND, depth for direct partitioning of pesticide to bed sediment; VMIX, mixing velocity by molecular diffusion) revealed that the predictions of pesticide leaching were influenced strongly by those variables. Generally the RICEWQ–VADOFT model is a useful modeling tool for pesticide risk assessment in rice paddies. Keywords—Risk assessment Pretilachlor Rice paddies Sensitivity analysis INTRODUCTION Rice is considered a high-value crop that is cultivated in Europe under submerged conditions. The requirements of high volumes of water for maintaining flooding conditions through- out the cultivating season have led to the establishment of rice growing areas within river basins. Such rice-cultivating areas are located close to large urban areas, including the Axios river basin in northern Greece and the Po valley basin in north- ern Italy, whereas others are situated within zones of high ecological value such the National Park of Donana and the Lagoon of Valencia in Spain [1]. Weeds have been recognized as the most significant crop protection constraint in rice cultivation in Europe, and their control mainly relies on selective herbicides. However, their intensive use in rice cultivation has resulted in the frequent detection of herbicides in adjacent surface and groundwater systems. A monitoring study employed in the Albufera Lake in Spain, which receives and provides water for rice irrigation, revealed maximum concentrations of 1.78, 10.3, and 0.31 for the herbicides molinate, bensulfuron-methyl, and benthiocarb, respectively [2]. Similar results have been reported in the main rice cultivation areas in Greece [3], Italy [4,5], and Portugal [6], raising concerns about the risk of environmental and hu- man exposure. According to the directive 91/414/EC (available at http:// europa.eu.int/comm/food/fs//phps/pro/legal/indexen.htm), * To whom correspondence may be addressed (dimitrios.karpouzas@unicatt.it). environmental exposure analysis using validated mathematical models should be provided by all companies seeking registra- tion for their compounds. Although a uniform approach for regulatory modeling has been adopted within Europe, this is not applicable to rice. The particular conditions of rice culti- vation would require the development of specific guidelines and the use of appropriate modeling tools. However, few math- ematical models currently are available for simulating the en- vironmental fate of pesticides in rice paddies. The pesticide paddy field model was developed in Japan for predicting pes- ticide concentrations in paddy fields and the runoff amount of the pesticide in nearby surface water bodies [7]. In the United States of America, a pesticide runoff model for rice cultivation, rice water quality (RICEWQ), was developed for pesticide- exposure assessment [8]. Both models can calculate chemical dissipation in the paddy and surface water releases but do not consider pesticide losses through leaching [7,9], which could be a significant dissipation path for certain pesticides under field conditions. In order to describe adequately both leaching and runoff processes, an improved version of the RICEWQ 1.6.2v was developed where an interface between RICEWQ and vadose zone flow and transport (VADOFT) model was built. Preliminary evaluations of the RICEWQ 1.6.2v model using monitoring data from a field study in northern Italy with the low-input herbicide cinosulfuron illustrated that it could be an effective tool for exposure assessments under the par- ticular conditions of rice cultivation [10]. However, further validation of the model using pesticides with varying envi- ronmental and agricultural properties is necessary to allow for