Martin Guillaume, Magne Marie-Angélina, Cristobal Magali San
AGIR, Université de Toulouse, INRA, INPT, INP-EI PURPAN, ENSFEACastanet-Tolosan, France.
GenPhySE, Université de Toulouse, INRA, INPT, INP-ENVTCastanet-Tolosan, France.
Front Plant Sci. 2017 Aug 29;8:1483. doi: 10.3389/fpls.2017.01483. eCollection 2017.
The need to adapt to decrease farm vulnerability to adverse contextual events has been extensively discussed on a theoretical basis. We developed an integrated and operational method to assess farm vulnerability to multiple and interacting contextual changes and explain how this vulnerability can best be reduced according to farm configurations and farmers' technical adaptations over time. Our method considers farm vulnerability as a function of the raw measurements of vulnerability variables (e.g., economic efficiency of production), the slope of the linear regression of these measurements over time, and the residuals of this linear regression. The last two are extracted from linear mixed models considering a random regression coefficient (an intercept common to all farms), a global trend (a slope common to all farms), a random deviation from the general mean for each farm, and a random deviation from the general trend for each farm. Among all possible combinations, the lowest farm vulnerability is obtained through a combination of high values of measurements, a stable or increasing trend and low variability for all vulnerability variables considered. Our method enables relating the measurements, trends and residuals of vulnerability variables to explanatory variables that illustrate farm exposure to climatic and economic variability, initial farm configurations and farmers' technical adaptations over time. We applied our method to 19 cattle (beef, dairy, and mixed) farms over the period 2008-2013. Selected vulnerability variables, i.e., farm productivity and economic efficiency, varied greatly among cattle farms and across years, with means ranging from 43.0 to 270.0 kg protein/ha and 29.4-66.0% efficiency, respectively. No farm had a high level, stable or increasing trend and low residuals for both farm productivity and economic efficiency of production. Thus, the least vulnerable farms represented a compromise among measurement value, trend, and variability of both performances. No specific combination of farmers' practices emerged for reducing cattle farm vulnerability to climatic and economic variability. In the least vulnerable farms, the practices implemented (stocking rate, input use…) were more consistent with the objective of developing the properties targeted (efficiency, robustness…). Our method can be used to support farmers with sector-specific and local insights about most promising farm adaptations.
在理论基础上，人们广泛讨论了适应农业以降低其对不利环境事件脆弱性的必要性。我们开发了一种综合且可操作的方法，以评估农场对多种相互作用的环境变化的脆弱性，并解释如何根据农场配置和农民随时间的技术适应措施，最好地降低这种脆弱性。我们的方法将农场脆弱性视为脆弱性变量原始测量值（例如生产的经济效率）、这些测量值随时间的线性回归斜率以及该线性回归残差的函数。后两者是从考虑随机回归系数（所有农场共有的截距）、全局趋势（所有农场共有的斜率）、每个农场相对于总体均值的随机偏差以及每个农场相对于总体趋势的随机偏差的线性混合模型中提取的。在所有可能的组合中，通过所有考虑的脆弱性变量具有高测量值、稳定或上升趋势以及低变异性的组合，可获得最低的农场脆弱性。我们的方法能够将脆弱性变量的测量值、趋势和残差与解释变量相关联，这些解释变量说明了农场对气候和经济变异性的暴露、初始农场配置以及农民随时间的技术适应措施。我们在2008 - 2013年期间将我们的方法应用于19个养牛（肉牛、奶牛和混合养殖）农场。选定的脆弱性变量，即农场生产力和经济效率，在养牛场之间以及不同年份差异很大，均值分别为43.0至270.0千克蛋白质/公顷和29.4 - 66.0%的效率。没有一个农场在农场生产力和生产经济效率方面都具有高水平、稳定或上升趋势以及低残差。因此，最不易受影响的农场代表了两种表现的测量值、趋势和变异性之间的一种折衷。没有出现用于降低养牛场对气候和经济变异性脆弱性的农民实践的特定组合。在最不易受影响的农场中，实施的实践（载畜率、投入使用……）与发展目标属性（效率、稳健性……）的目标更为一致。我们的方法可用于为农民提供有关最有前景的农场适应措施的特定部门和本地见解的支持。
