Contents lists available at ScienceDirect Field Crops Research journal homepage: www.elsevier.com/locate/fcr Development and evaluation of HUME-OSR: A dynamic crop growth model for winter oilseed rape Ulf Böttcher, Wiebke Weymann, Jeroen W.M. Pullens, Jørgen E. Olesen, Henning Kage Institute of Crop Science and Plant Breeding, Christian-Albrechts University at Kiel, Hermann-Rodewald-Str. 9, Kiel, 24118, Germany ARTICLE INFO Keywords: Oilseed rape Crop Growth model Dry matter partitioning Nitrogen uptake Nitrogen partitioning ABSTRACT A new dynamic crop growth model based on empirically derived allometric partitioning rules was developed for winter oilseed rape. The model simulates dry matter production, nitrogen uptake and distribution, leaf, stem and pod area expansion and yield formation under optimal and water- and nitrogen-limited conditions. The model includes hibernation, senescence due to self-shading, freezing and aging, translocation of assim- ilates and nitrogen, light absorption and reflection by flower layer and oil synthesis. It was parameterized with two data sets from Hohenschulen, northern Germany, and validated with datasets from Germany, France, Great Britain, and the Czech Republic. Model performance in terms of prediction of total aboveground dry matter production gave an RMSE of 180 g m -2 and the linear regression between measured and simulated root/shoot ratios gave an r 2 of 0.64. In addition, nitrogen uptake (RMSE 4.26 g m -2 ) and distribution (r 2 simulated/measured leaf N/stem N = 0.44) are quite well represented. In contrast, year-to-year variability of final seed yield was less correctly estimated, probably due to variation in the harvest index, which is not included in the model prediction. Relative differences in seed yield due to sowing date and nitrogen application were well reproduced in most cases. Therefore the model has potential to be used for supporting optimization of management strategies, climate change scenario studies and future breeding progress. 1. Introduction Winter oilseed rape (WOSR, Brassica napus L.) has become an im- portant crop for food and biodiesel production, especially in Germany, France and UK with 1228, 1615 and 601 kha cropped area in 2018 (Eurostat, 2019). It is therefore of high importance that the crop growth is well understood. Crop growth and yield are influenced by the com- plex interaction of management (e.g. sowing date, fertilization strategy) with environmental factors, including weather, soil texture, water and nutrient availability (Mendham et al., 1981; Asare and Scarisbrick, 1995). During recent decades the average yield increased due to breeding progress and management optimization, but this has not im- proved yield stability (Rondanini et al., 2012) and yield variability of WOSR is still high (Berry and Spink, 2006). Simulation models can be effective tools to analyze and optimize the development and yield formation of WOSR under different environ- mental conditions. Most plant growth models available for WOSR were developed and published during the 1990s (Habekotté, 1997b; Gabrielle et al., 1998a, b). During recent years, new effort has been given to developing and adapting rapeseed models (Deligios et al., 2013; Jing et al., 2016, Hoffman et al. 2015, Robertson and Lilley, 2016; He et al., 2017), because of the increased importance of the crop for food production and biodiesel. However, none of the published si- mulation models for WOSR is able to sufficiently simulate crop growth and yield in different environments under nitrogen (N) and water limitation (Yin et al., 2017). Also the data base for parameterization and evaluation of the models has often been scarce and the validity of model predictions for conditions differing from parameterization data set is sometimes doubtful. In contrast to other major field crops a structured model comparison for OSR-models is still lacking. The LINTUL-BRASNAP model, published by Habekotté (1997a) is only parameterized for optimum growth conditions and does not si- mulate autumn and winter growth. The CERES-RAPE model by Gabrielle et al. (1998b) is parameterized for fully irrigated crops. Petersen et al. (1995) adapted the DAISY model to WOSR data, but important processes as leaf loss due to senescence and partitioning of dry matter (DM) between pods and vegetative parts after flowering are not included. The newly adapted model CSM-CROPGROW, which is based on the DSSAT model, was used for data from Mediterranean environments (Deligios et al., 2013), Canada (Jing et al., 2016; Qian et al., 2018) and California (George and Kaffka, 2017). The APSIM model family includes a rapeseed model (Hoffmann et al., 2015, https://doi.org/10.1016/j.fcr.2019.107679 Received 4 April 2019; Received in revised form 8 November 2019; Accepted 8 November 2019 E-mail address: boettcher@pflanzenbau.uni-kiel.de (U. Böttcher). Field Crops Research 246 (2020) 107679 0378-4290/ © 2019 Elsevier B.V. All rights reserved. T