双重标准差法在昆虫科阶元分类学上的应用

在科分类阶元上对半翅目、鳞翅目和鞘翅目8个科的23种昆虫图像中提取的昆虫面积、周长等11项数学形态特征进行了双重标准差法分析，以评估该方法在昆虫科阶元分类上的应用有效性。结果表明，在科的阶元上11项特征可靠性大小依次为: （似圆度、偏心率、 圆形性>（横轴长、形状参数、叶状性>（面积、周长、球状性> (纵轴长、亮斑数)。科的亲缘关系远近结果显示： 蝽科和缘蝽科关系较近>丽金龟科、天牛科与鳃金龟科关系较近>夜蛾科、大蚕蛾科和粉蝶科关系较近。所得结果与统计假设检验分析所得结果非常相似。

Use of double standard deviation at family level of insect taxonomy

Double standard deviation analysis with 11 math-morphological features (MMFs) ( such as area, perimeter, etc.) from the images of 23 species of insects of the Pentatomidae, Coreidae, Noctuoidea, Saturnijdae, Pieridae, Melolonthidae, Rulelidae and Cerarnbycidae families was made to assess its use in insect taxonomy at family level. The results indicate that the ranked reliability of MMF in the identification of insect families is, from high to low: roundness-likelihood, eccentricity, circularity >X-length, form parameter, lobation >area, perimeter, sphericity>Y-length, hole number. From the perspective of mathematical morphology, the kinship can be ranked as follows: Pentatomidae and Coreidae>Rulelidae, Cerarnbycidae and Melolonthidae>Noctuoidea, Saturnijdae and Pieridae. The results are similar to those of statistical hypothesis testing analysis.