Yield response of an ensemble of potato crop models to rising CO2 concentration in continental Europe

Authors: Fleisher, David; CONDORI, B - International Potato Center; BARREDA, C - International Potato Center; BERGUIIS, H - University Of Wageningen; BINDI, M - University Of Florence; BOOTE, K - University Of Florida; CRAIGON, J - University Of Nottingham; VAN EVERT, F - University Of Wageningen; FANGMEIER, A - University Of Hohenheim; FERRISE, R - University Of Florence; GAYLER, S - University Of Hohenheim; HOOGENBOOM, G - University Of Florida; KREMER, P - University Of Hohenheim; MERANTE, P - University Of Gottingen; NENDEL, C - Leibniz Centre; NINANYA, J - International Potato Center; PLEIJEL, H - University Of Gothenburg; RAES, D - Catholic University Of Leuven; RAMIREZ, D - International Potato Center; REIDSMA, P - University Of Wageningen; SILVA, J - University Of Wageningen; STOCKLE, C - Washington State University; SUPIT, I - University Of Wageningen; STELLA, T - Leibniz Centre; VANDERMEIREN, K - Veterinary And Agrochemical Research Center, Coda-Cerva; VAN OORT, P - University Of Wageningen; VANUYTRECHT, E - Catholic University Of Leuven; VORNE, V - Natural Resources Institute Finland (LUKE); WOLF, J - University Of Wageningen

Submitted to: European Journal of Agronomy
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 2/28/2021
Publication Date: 3/3/2021
Citation: Fleisher, D.H., Condori, B., Barreda, C., Berguiis, H., Bindi, M., Boote, K., Craigon, J., Van Evert, F., Fangmeier, A., Ferrise, R., Gayler, S., Hoogenboom, G., Kremer, P., Merante, P., Nendel, C., Ninanya, J., Pleijel, H., Raes, D., Ramirez, D., Reidsma, P., Silva, J.V., Stockle, C.O., Supit, I., Stella, T., Vandermeiren, K., Van Oort, P., Vanuytrecht, E., Vorne, V., Wolf, J. 2021. Yield response of an ensemble of potato crop models to rising CO2 concentration in continental Europe. European Journal of Agronomy. https://doi.org/10.1016/j.eja.2021.126265.
DOI: https://doi.org/10.1016/j.eja.2021.126265

Interpretive Summary: Climate impacts on many important crops need to be understood so that appropriate management measures can be put in place to maintain food security needs. Rising atmospheric carbon dioxide concentration (CO2) is one major climate factor that is predicted to continue to increase over the next several decades. Mathematical tools, called crop models, are the primary method scientists use to estimate effects rising CO2 will have on future crop yields. More evaluation of the accuracy of these tools is needed. Experimental data from seven European countries were used to test ten different potato models for their estimate of CO2 response on tuber yields. Results showed the models were able to accurately estimate the relative yield increase in response to higher CO2 values. Accuracy was not as good as expected when predicting the exact yield values. Differences due to model structure on model accuracy, or the way in which individual models incorporate environment, management, and soil factors into making yield predictions, were not observed, but appropriate use of calibration data was shown to be very important. These results show that these models can be used with confidence to estimate relative response to certain climate factors once properly calibrated. Scientists and policy planners concerned about food security will benefit from these research findings.

Technical Abstract: A multi-model inter-comparison study was conducted to evaluate performance of ten potato crop models to accurately predict potato yields in response to elevated CO2 (Ce) when calibrated with ambient CO2 data (Ca). Experimental data from the same cultivar, obtained from seven open-top chambers and two free-air-CO2 enrichment systems, across continental Europe were used. Model performance metrics for yield predictions were lower than expected, even for calibration responses. Percent errors averaged over all datasets were 26.5% for Ca and 27.2% Ce data. Other metrics such as Wilmott's index of agreement (IA) and root mean square relative error (RMSRE) ranged broadly among individual models and locations, such that four of the ten models out-performed the median or mean of the ensemble for about half of the Ce datasets. These top performing models were representative of three different model structural groups such that no one particular group was necessarily the best performing. Relative response to rising CO2 was more accurate than absolute yield responses when averaged across all locations, and within 3.3 kg ppm-1 (or 5%) of observed values. Specific targets in the model structure needed for improvement were not identified due to large, and inconsistent, variation in yield predictions across locations. However, models with the lowest calibration errors tended to be top performers for Ce predictions as well. Where possible, scientists using potato models to estimating climate change responses should use Ce calibration data to improve confidence in such predictions.