New application of meta-analysis method to analysis of non-controlled field experiment data for the effects of drip-irrigation on cotton yield in Xinjiang Region
New application of meta-analysis method to analysis of non-controlled field experiment data for the effects of drip-irrigation on cotton yield in Xinjiang Region
复制标题
元分析方法在新疆地区滴灌对棉花产量影响非对照田间试验数据分析中的新应用
DOI：
--
复制
发表时间：
--
期刊：
影响因子：
--
通讯作者：
Liu Wang
中科院分区：
文献类型：
--
作者：
Aiqin Zhao;Xiuju Wei;Yingkuan Wang;Liu Wang
荟萃分析方法全面量化了来自明确对照组和治疗组的现场数据。然而，在实践中，大量数据缺乏对照组，这限制了Meta分析对它们的应用。以新疆不同灌溉量下的棉花产量和灌溉水利用效率为例，探讨将非受控数据转化为受控数据的方法，以便对非受控数据进行Meta分析。从52个实验中收集了185个观察结果。以最大灌溉量处理为对照，将185个观测值转化为133个受控数据组。采用Meta分析方法计算产量和IWUE的响应比，分析其随相对灌溉量（灌溉量与最大灌溉量的比值）的变化。结果表明，随着灌水量的增加，产量增加，但IWUE降低。灌水量大于378 mm时，产量最高，但IWUE最低，两者间产量差异不显著灌水量为338- 378 mm时，IWUE显著提高，但差异不显著（p>0.05当灌水量减少到294-338 mm（平均316 mm）时，平均产量没有明显下降，但平均IWUE提高了17%（p<0.05）。294 mm以下，继续减少灌水量，产量显著下降（p<0.05）。结合产量和IWUE，全疆适宜灌溉量为294-378 mm：水资源相对充足区为338-378 mm，平均358 mm;水资源相对缺乏区为294-338 mm，平均316 mm。研究证明了Meta分析应用中非控制数据向控制数据转化方法的可行性，扩展了Meta分析的应用范围，为不同水资源区的水资源规划决策提供了有价值的信息。
Meta-analysis method comprehensively quantifies data from field with clear control and treatment groups. However, a large number of data lack control groups in practice, which limits meta-analysis application to them. This study took an example of cotton yield and irrigated water use efficiency (IWUE) under different irrigation amounts in Xinjiang to explore the method to transform uncontrolled data into controlled data, so that meta-analysis can be applied to them from non-controlled data. A total of 185 observations were collected from 52 experiments. Taking the treatment with maximum irrigation amount as control, the 185 observations were converted into 133 controlled data groups. The Response Ratio of yield and IWUE were calculated by meta-analysis method to analyze their change with relative irrigation amount (the ratio of irrigation amount to the maximum irrigation amount). A stair-step line graph of yield and IWUE change with irrigation amount was obtained, which showed that the yield increased but the IWUE decreased with the irrigation amount. The yield was the highest but IWUE was the lowest when the irrigation amount was higher than 378 mm. Compared to that, the yield had no significant difference (p>0.05) but IWUE significantly increased by 6% when the irrigation amount was 338-378mm (average 358 mm), and the average yield was not greatly decreased while the average IWUE increased by 17% (p<0.05) when the irrigation amount was reduced to 294-338 mm (average 316 mm). Decreasing continuously the irrigation amount below 294 mm, the yield would be significantly decreased (p<0.05). Combining the yield and IWUE, the suitable irrigation amount was 294-378 mm in whole Xinjiang: 338-378 mm (average 358 mm) and 294-338 mm (average 316 mm) respectively for the area with relatively sufficient and relatively lack of water resources. The study proves the feasibility of transformation method of non-controlled data to controlled data for application of meta-analysis, extends meta-analysis application scope and provides valuable information for decision of water resources planning in different water resources areas.