The novel methods of quantitative classification of plant life cycle forms and weight collocation: taking classification of life cycle forms of Psophocarpus tetragonolobus as an example
With the development of theoretical study on plant life cycle forms, quantitative classification methods have become one of the main research approaches in this field. The current methods for classification of plant life cycle forms are based on the principal component analysis (PCA). However, these methods just assume that plant trait indicators areindependent variables and never make more serious with the interactions among target traits. Therefore, classification methods adapted for ' mesh ' structure of plant life cycle forms need to be developed as an new indicator system. According to hierarchical characteristics of the indicator system of plant life cycle forms, both the climbing and dwarf types of Psophocarpus tetragonolobus were used as the experimental models to allocate the weights for trait indicators by principal component analysis (PGA) , analytic hierarchy process (AHP) and analytic network process ( ANP) respectively. The composite scores and classification parameters of trait indicators had been calculated respectively. The results were as follows; compared with ANP, parameter values (x) of V type (vegetative growth type) was below 0. 39 and parameter values (z) of S type (sexual reproduction type) was over 0. 39 by PCA, parameter values (x) of V type was over 0. 614 and parameter values (2) of S type was below 0.088 by AHP. There were significant differences among these classification parameters of life cycle forms by three methods. For the PCA and AHP required independent trait indicators couldn't eliminate significant correlation among them, so there were deviations between the classification results based on PGA or AHP of plant life cycle forms on P. tetragonolobus. These results showed that the correlation of trait indicators affected classification results of life cycle forms. The indicator system of ANP is a * mesh' structure, and indicators for control and network layer are associated with each other. To build judgement matrices of control layer and network layer on ANP, the correlation information had been extracted from correlation matrices of trait indicators and the weight distribution really reflected the correlationship among trait indicators. The results based on ANP of life cycle forms in different types of climbing and dwarf for P. tetragonolobus were V0.517C0.327S0.156 and V0.416C0.43S0.154 respectively. If the correlation among trait indicators was not significant, both PCA and AHP can be used to calculate weights. Otherwise, the ANP is better for the case that there were significant correlation among them. To sum up, the classification method of plant life cycle forms based on ANP can be used not only to resolve the interdependence among each pair of indicators but also to provide new and effective ways for quantitative classification of plant life cycle forms.
life cycle formsweight collocationanalytic network process (ANP)Psophocarpus tetragonolobus
国家科技期刊平台
NSTL
万方数据
维普
