Addy John W G, Ellis Richard H, Macdonald Andy J, Semenov Mikhail A, Mead Andrew
Computational and Analytical Sciences Department, Rothamsted Research, UK.
School of Agriculture, Policy and Development, University of Reading, UK.
Agric For Meteorol. 2020 Apr 15;284:107898. doi: 10.1016/j.agrformet.2019.107898.
The effect of weather on inter-annual variation in the crop yield response to nitrogen (N) fertilizer for winter wheat () and spring barley () was investigated using yield data from the Broadbalk Wheat and Hoosfield Spring Barley long-term experiments at Rothamsted Research. Grain yields of crops from 1968 to 2016 were modelled as a function of N rates using a linear-plus-exponential (LEXP) function. The extent to which inter-annual variation in the parameters of these responses was explained by variations in weather (monthly summarized temperatures and rainfall), and by changes in the cultivar grown, was assessed. The inter-annual variability in rainfall and underlying temperature influenced the crop N response and hence grain yields in both crops. Asymptotic yields in wheat were particularly sensitive to mean temperature in November, April and May, and to total rainfall in October, February and June. In spring barley asymptotic yields were sensitive to mean temperature in February and June, and to total rainfall in April to July inclusive and September. The method presented here explores the separation of agronomic and environmental (weather) influences on crop yield over time. Fitting N response curves across multiple treatments can support an informative analysis of the influence of weather variation on the yield variability. Whilst there are issues of the confounding and collinearity of explanatory variables within such models, and that other factors also influence yields over time, our study confirms the considerable impact of weather variables at certain times of the year. This emphasizes the importance of including weather temporal variation when evaluating the impacts of climate change on crops.
利用位于洛桑试验站的Broadbalk小麦和Hoosfield春大麦长期试验的产量数据，研究了天气对冬小麦（）和春大麦（）氮肥作物产量年际变化的影响。使用线性加指数（LEXP）函数，将1968年至2016年作物的谷物产量模拟为施氮量的函数。评估了天气变化（按月汇总的温度和降雨量）以及种植品种的变化对这些响应参数年际变化的解释程度。降雨和潜在温度的年际变化影响了作物对氮的响应，进而影响了两种作物的谷物产量。小麦的渐近产量对11月、4月和5月的平均温度以及10月、2月和6月的总降雨量特别敏感。在春大麦中，渐近产量对2月和6月的平均温度以及4月至7月（含）和9月的总降雨量敏感。本文提出的方法探讨了随着时间推移农艺和环境（天气）对作物产量影响的分离。在多个处理中拟合氮响应曲线有助于对天气变化对产量变异性的影响进行有益分析。虽然此类模型中存在解释变量的混杂和共线性问题，且其他因素也会随时间影响产量，但我们的研究证实了一年中某些时候天气变量的重大影响。这强调了在评估气候变化对作物的影响时纳入天气时间变化的重要性。
