Establishment And Verification Of Forest Pest Based Index Model

The population density of forest diseases and insect pests is determined by both biotic and abiotic factors.When forest diseases and insect pests are disturbed by the factors above,their population density,species structure and spatial pattern will change due to the maladjustment of relevant regulatory factors,which will further cause the loss of regulatory power of forest ecosystem and eventually lead to the outbreak and epidemic of pests.Based on the basic principles of forest diseases and insect pests,this study proposes indicator of pest based index?PBI?.The key stand factors affecting the occurrence of diseases and insects in the stands were screened out and the functional relationships between stand factors and disease and insect pest severity indexes were established,then the optimal model was selected as the guide curve.The curve group of pest based index was obtained by stretching the guide curve in equal proportion.On the basis of PBI curve group,the PBI-site factor model was established.Then the range of the index?0?100?was divided into 5 intervals on average.Finally,the study selected pure forest of Japanese red pine?JRP?in Kunyushan Mountains to set up a sample plot,and studied the effects of stand and site factors on pine needle blight and Kunyushan web-spinning sawfly,respectively,the purpose of which was to establish a related model to verify the reliability and reasonableness of the PBI curve group and PBI-site factor model mentioned above.The main research results are as follows:?1?Based on the basic principles of forest diseases and insect pests,the difference of specific insect and disease severity in the pure forest in the same study area was attributed to the comprehensive effect of stand factor and site factor.In order to evaluate quantitatively the effect level of a pure forest site factors on the potential occurrence of specific forest pests after the joint action of site factors and stand factors,the concept of pest based index was proposed.It is also an indicator to evaluate the potential occurrence degree of pests and diseases in woodland.?2?The quantitative method of pest based index was determined,and the guide curve and curve group of PBI were established,which can quantitatively evaluate the effect level of site condition on the potential occurrence of specific pests in forest.The five levels in the curve group were denoted from bottom to top:?level-light pest grade,?level-mild pest grade,?level-moderate pest grade,?level-severe pest grade,?level-especially severe pest grade,respectively.If the disease and pest severity index of the sample point fell within the interval of[0,20),then the potential level of disease and insect pest on the site was?,and plant diseases and insect pests is very mild in this kind of site,and so on.?3?The PBI-site factor model was established,which can provide theoretical guidance for predicting whether no-forest land was suitable for planting specific artificial pure forest.By substituting the value of the key site factor of a no-forest land into the equation,the PBI value of the woodless land can be calculated.If the PBI value was in the interval of[0,20),the magnitude role of non forest land for specific disease and pest severity was level?,and so on.It,together with PBI curve group,constituted a complete set of quantitative evaluation of the potential degree of diseases and pests in different sites.?4?A sample plot was set up in Japanese red pine pure forest in Kunyushan Mountains,and the effects of stand and site factors on Japanese red pine and Kunyushan web-spinning sawfly were studied.Stand density and canopy coverage had highly significant and positive correlation with the occurrence of JRP needle blight?P<0.01?,while height under branch,tree height,crown width and DBH had significant and negative correlation with the occurrence of the disease?P<0.01?.The occurrence degree of JRP needle blight in plots with high altitude,flat slope,sunny and concave slope as well as thin humus layer was lower than other types of plots.The values of soil bulk density,maximum moisture capacity,organic matter,total nitrogen and total potassium were significantly different between healthy plots and susceptible ones?P<0.01?,while other soil physical and chemical properties were not significantly different in six types of sample plots.The stand density and herb coverage in sawfly non-infested plots were highly significantly higher than that of infested ones?P<0.01?,but the values of tree height,crown width,DBH and height under branch were lower than that in sawfly infested plots.The Kunyushan web-spinning sawfly occurred seriously in the JRP forest with high altitude and middle slope,but less in the JRP forest with shady slope.?5?Stepwise regression and partial correlation analysis showed that the key stand factors affecting the occurrence of JRP needle blight and Kunyushan web-spinning sawfly were stand density and crown width,respectively.According to the establishment method of PBI model,the guide curve and curve group of the disease based index of JRP needle blight and the insect based index of Kunyushan web-spinning sawfly were established respectively.Guide curve expressions were:Q=65.61/(1+e^{-0.0015x+2.32})and Q=75.53/(1+e^-0.84x+3.40).The model field test showed that the fitted equation met the average predicted accuracy?94.65%and 89.28%?and the actual law of pine needle blight and web-spinning sawfly with stand density and crown width,which indicated that the PBI model proposed in this study was reasonable and reliable.?6?The corresponding disease or insect based index of each study plot was obtained by consulting the figure of disease based index curve group for pine needle blight or the figure of insect based index curve group for Kunyushan web-spinning sawfly,then key site factors were selected and two PBI-site factor multiple linear regression models were established by using the theory of quantification?.Their expressions were:y=77.055+9.081x₂₁+6.425x₂₂+1.689x₂₃-15.031x₄₁-14.868x₄₂-14.125x₄₃-0.052x₅-1.605x₈ andy=5.187+27.236x₄₁+29.914x₄₂+7.816x₄₃+0.047x₅-0.497x₆+2.157x₈.Both models reached the extremely significant level statistically?P<0.01?,and their coefficient of determination?R²?were above 0.6500,indicating that the model fitted well.The off-site test of the models showed that their values of mean predicted error?MPE?were 8.73%and 5.87%,respectively,indicating that the accuracy of prediction was above 90%and their TRE values were close to 0.