Sustainability indices from nutrient budgets of agro-ecosystems

基于产量、养分浓度、比率、养分吸收、养分添加和养分收支的植物和小区尺度养分利用效率的经典测量被虚假相关性和冗余信息扭曲，不考虑高水平的相互作用。一个整体的方法安排到平衡可以纠正这种扭曲使用的原则之间的正交子集的组成部分。养分可持续性指数（NSI），然后计算距离参考组合物使用组织养分分析在植物尺度和养分预算在地块尺度。此后，可将维持农业生态系统生产力的养分利用效率综合成指数，以便利作出正确的决策和改进养分管理做法。在工厂规模上，参考化合物是使用魁北克数据集在高产水平上相互作用的营养物质。在小区尺度上，需要进行长期试验，以说明耕作制度的残留效应。使用等距对数比或标准正交平衡（由具有在支点处诊断的平衡的移动的和在桶中理解的分量示出），可以将NSI计算为与参考组合物的多变量距离。我们的目标是监测偏离农业生态系统的参考组合物在植物和地块尺度。通过肥料试验和长期试验中养分失衡程度的校正，诊断叶片养分。2008年建立的长期试验允许监测玉米（M）-大豆（S）-小麦（W）序列和连续饲料（MMM，MSM，MSWFFF，FFF）对作物产量和土壤质量的影响。该设置还包括两种土壤耕作方法（免耕和常规），两种类型的残留物管理（残留物留在土壤上或清除）和两种施肥制度（根据当地建议的矿物或有机;不施肥的控制），重复4次（156块）。在2013年和2015年，对最初在5、10、20、40、60和100厘米处取样的土壤进行了重新取样，每年收集生物量和谷物样品用于营养预算。土壤质量元数据包括团聚度和紧实度指数、导水率、pH值、抗渗性和土壤有效养分。这种新颖的管理工具将使决策者和利益相关者能够全面地看待营养管理策略（“可以明智地管理测量的内容”），避免通常难以解释的跨越无数比率和单个成分的整体放射图。这些模型将支持专家系统在当地规模和施肥策略背后的4 R方法的加拿大化肥行业（正确的速度，来源，时间，地点）的发展。四名研究生（硕士，博士）将构建一个从设计到创新的可解释平衡的新系统。

Classical measures of nutrient use efficiency at plant and plot scales based on yield, nutrient concentration, ratios, nutrient uptake, nutrient additions and nutrient budgets are distorted by spurious correlations and redundant information by not accounting for high-level interactions. A holistic approach arranged into balances can correct such distortions using the principle of orthogonality between subsets of components. Nutrient sustainability indices (NSI) are then computed as distances from reference compositions using tissue nutrient analyses at plant scale and nutrient budgets at plot scale. Nutrient use efficiency sustaining agroecosystem productivity could be synthesized thereafter into indices to facilitate making sound decisions and improve nutrient management practices. At plant scale, reference copmpositions are those of nutrients interacting at high-yield level using a Québec data set. At plot scale, long-term experiments are required to account for residual effects of cropping systems. Using isometric log ratios or orthonormal balances illustrated by a mobile with balances diagnosed at fulcrums and components appreciated in buckets, NSIs can be computed as multivariate distances from reference compositions. Our objective is to monitor the departure of agroecosystems from reference compositions at plant and plot scales. Foliar nutrients are diagnosed by calibrating the degree of nutrient imbalance in fertilizer trials and the long-term experiment. The long-term experiment established in 2008 allows monitoring the effect of maize (M)-soybean (S)-wheat (W) sequences and continuous forage (MMM, MSM, MSWFFF, FFF) on crop yield and soil quality started with M) or F. The setup also comprises two soil tillage methods (no-till and conventional), two types of residue management (residues left on soil or removed) and two fertilization regimes (mineral or organic based on local recommendations; control without fertilization), replicated 4 times (156 plots). Soils sampled initially at 5, 10, 20, 40, 60 and 100 cm were resampled in 2013 and 2015, and biomass and grain samples collected annually for nutrient budgeting. Soil quality metadata are aggregation and compactness indices, hydraulic conductivity, pH, and resistance to penetration, soil available nutrients. This novel management tool will allow decision makers and stake holders look holistically at nutrient management strategies ("what is measured can be managed wisely"), avoiding the usual hard-to-interpret holistically radial diagrams across myriads of ratios and individual components. These models will support the development of expert systems at local scale and fertilization strategies behind the 4R approach of the Canadian fertilizer industry (right rate, source, time, place). Four graduate students (M.Sc., Ph.D.) will frame novel systems of interpretable balances from design to innovation.