浙江双栉蝠蛾发生与土壤关系的层次递进判别分析

为了揭示浙江双栉蝠蛾发生与土壤化学指标、土壤深度的关系,在福建省邵武市拿口镇三丰村选取春笋被浙江栉蝠蛾重度危害和轻度危害各2个样地进行调查,土壤由表及里依次分成3个层次,对每个层次测定pH值、全N含量等7个指标,采用判别分析法分别对每一层及不同层次的组合进行分析。结果表明: (1)各层独立分析,第1层单个指标中具有显著性的是有机质(X₇),判别最有效的组合是全N含量(X₂)和速效K(X₆);第2层中单个指标都不具有显著性,判别最有效的组合仍为全N含量 (X₂)和速效K(X₆);第3层单个指标中具有显著性的是有机质(X₇),判别最有效的组合是速效K(X₆)和有机质(X₇)。 (2)依据第1层判别函数能够完全正确判别分组第2层和第3层,第2层判别函数能够完全正确判别分组第3层,但第2层判别函数不能够完全正确判别分组第1层,第3层判别函数不能够完全正确判别分组第1层和第2层,判别函数具有随土壤层递进而正确判别的递进性,而不具有逆土壤层进行正确判别的逆向性。 (3)对第1层与第2层的组合、第2层与第3层的组合的判别分析,发现单个指标的显著性有所变化,但判别组合的指标没有变化,再次说明了土壤理化性状是由表及里的影响特性。 (4)与方差分析和多重比较的结果相比较,判别分析不仅能反映单个指标在分组中的显著性,而且反映出指标组合的重要性,表明单个指标的显著性不能构成对分组区别的决定性。在层次递进分析过程中,可以清晰地表明各指标在各层以及层次组合中的作用,反映出了层与层间的关系。 由此可以得出:土壤的理化性状是由表及里地影响,与浙江双栉蝠蛾发生危害程度关系密切的是全N含量 (X₂)和速效K(X₆),即全N含量 (X₂)的增加和速效K(X₆)的减少会使危害程度加重,其次为全P含量(X₃),全P含量的增加可使危害程度加重。判别分析法较方差分析和多重比较更全面深刻地反映了浙江双栉蝠蛾发生与土壤的关系。

Analysis of layer progressive discriminant relationsbetween the occurrence of Bipectilus zhejiangensis and soil

In order to explore relationships betweenthe occurrence of Bipectilus zhejiangensis moth in the bamboo shoot and soil chemical characteristics and depth. Two plots of bamboo forest damaged by the moth were selected to perform this study in 2007 in Shaowu, Fujian.Soil beneath the damaged bamboo shoots were divided into 3 depth levels (0 20cm, 20 40cm and 40 60cm) in each 60 cm × 60 cm plot, and were labeled as Layer 1, 2 and 3 from the surface. Soil samplings from each level were composited to test for pH, total N content and seven other soil parameters. Discriminant analysis technique was used to analyze each layer, 1,2-layer combination, 2,3-layer combination and 1,2,3-layer composition. The results indicated that: (1) Layer 1 had significant organic matter with the most effective combination discriminant of the total nitrogen content and instant potassium; Layer 2was not significant in single index, but the most effective combination discriminant were still total nitrogen content and instant potassium. Layer 3 had significant single parameter of organic matter, the most effective combination discriminant were instant potassium and organic matter. Levels of importance on the target groups were not identical. The individual indicator was significant in the portfolio which had great impact on the group, but not completely decisive. Combination indicator played a more important role than the individual indicator. (2)Layer 1 discriminant function can be used to correctly classify Layer 2 and Layer 3. Layer 2 discriminant function can be used to correctly classify Layer 3, but not to Layer1. Layer 3 can not be used to correctly classify Layer1 and Layer2. Discriminant function had forward classifying to the lower layer with a certain degree of accuracy, but hadn't reverse classifying to the upper layer. From the surface to bottom layer, the damage level increased with the total nitrogen content and decreased with instant potassium content. (3)Discriminant analysis of 1,2-layer combination and 2,3-layer combination showed that Significance in single indicator changed, but the combination of the index didn't. The most effective indicators of 1,2-layer combination were still total nitrogen content and instant potassium. 2,3-layer combination had the same indicator in single layer which were still total N content and instant potassium, which illustrates again that the physical and chemical properties of soil affect the damage level from surface to bottom layer. (4) Compared with the results of variance analysis and multiple comparisons, discriminant analysis reflects not only the single indicator statistically significant in a group, but also the importance of index portfolio. This result indicates that significance of single indicator could not differentiate decisively between groups. Therefore, we conclude that the physical and chemical properties of soil could affect the occurrence of B. zhejiangensis from surface to bottom layer, and the damage level increases with the increase of total nitrogen and the decrease of instant potassium Using discriminant analysis method to identify the relations between the occurrence of B. zhejiangensis and soil was superior to variance analysis and multiple comparisons.