中国1950s到2000s玉米产量-密度关系的Meta分析

【目的】明确中国玉米产量-密度试验结果的共性规律，评价Meta分析方法在产量-密度关系研究中的作用。【方法】按照Meta分析原理，汇集中国1950s到2000s玉米产量-密度的文献结果，利用约1 500个产量-密度数据对，进行梯度、边界线和核密度等统计分析。【结果】最适密度的变异系数高达20.4%，相应产量的变异为33.65%，单株产量的变异为30.8%。密度和单株产量对产量的相对重要性因二者组合所处的区域而异，产量上界与密度关系符合Y＝-0.0134x3＋3.15x，对应密度区间为0.99，15.0] plant/m2。产量上界与单株产量关系为两段直线，先沿Y1＝113.1x上升，然后沿Y2＝-69.84x＋33.87下降。以90%为累积频率，偏离最适密度15%的最大减产幅度为6.18%，偏离1 plant/m2的最大减产幅度为0.88 t•hm2。【结论】容易实现15 t•hm-2高产目标的单株产量在0.185 kg左右，密度区间为7.0，9.7] plant/m2，最适密度的变异远小于对应产量和单株产量的变异。Meta分析能够从多方面剖析玉米产量-密度关系。

Meta-Analysis of the Relationship Between Maize Crop Yield and Plant Density from 1950s to 2000s in China

【Objective】 The objective of the study is to reveal the general characteristics of yield-density relationships of maize plant density experiments in China, and to evaluate the suitability of meta-analysis as a tool to study maize yield response to plant density. 【Method】Based on requirements for meta-analysis, a collection of historical maize plant density experiments from 1950s to 2000s in China was built with some 1 500 pairs of plant densities and their crop yields. All eligible data sets were subject to statistical analysis such as gradients, boundary curves and 2-D kernel density. 【Result】Variation coefficient was 20.4% for optimum density, 33.65% for the highest yield and 30.8% for yield per plant under optimum density. Relative importance of yield components varied with combination zones of plant density and yield per plant. Upper boundary yields as a function of plant density followed the equation of Y＝-0.0134x3＋3.15x, within plant density range of 0.99, 15.0] plant/m2. Upper boundary yields as two stage functions of yield per plant were that increased along Y1＝113.1x, then declined along Y2＝-69.84x＋33.87 when yield per plant is larger than 0.185 kg. Plant density range within which there is the highest possibility for the yield of 15 mg?hm-2 fell in 7.0, 9.7] plant/m2. The 90th quantile for yield loss percentage was 6.18% due to 15% offset from the optimal plant density, and the 90th quantile due to 1 plant/m2 offset was 0.88 t?hm-2. 【Conclusion】Meta-analysis facilitates the distillation and abstraction of bulk data from maize yield-density experiments with multiple views, and it is applicable to other crop science areas.