Effects Of Cnaphalocrocis Medinalis On Rice Physiological And Ecological Parameters And Yield Under Condition Of Field Experiments

Cnaphalocrocis medinalis is one of the important pests that endanger rice development and yield. The changes of SPAD (Soil and Plant Analyzer Development)value, LAI (Soil and Plant Analyzer Development) and leaf-rolled rate, as well as the canopy hyper-spectral reflectance at different growth stages were measured after the larvaes of C. medinalis with different quantities were set in different treatments. The characteristics of hyper-spectral reflectance, first derivative and red edge parameters were also analyzed. The correlation between the rice hyperspectral and physiological and ecological parameters in different bands were studied by the means of correlation analysis,sensitive spectral bands and 13 key vegetation indexes were selected,and monitoring models of physiological and ecological parameters were established. The aim of the research is to provide a basic theory for monitoring rice damage from C.medinalis using hyper-spectral remote sensing measure. The single factor variance method was also used to estimate the loss of rice yield in each plot, especially for rice natural compensation ability. The correlation between rice hyper-spectral parameters and yield was investigated by using correlation analysis method.Vegetation indexes such as DVI (Difference Vegetation Index), RVI (Ratio Vegetation Index), PVI (Perpendicular Vegetation Index), NDVI(Normalized Difference Vegetation Index) and the edge peak Dx (Red Edge Slope) parameters were selected to establish optimal yield estimation models under different growth stages. Finally, the Seinhorst Model was used to estimate the relationship between the initial pest population density of C. medinalis and rice yield. Accuracy of two models was compared as well. The main conclusions were showed as follows:(1) Physiological and ecological parameters of rice were correlated with meteorological factors (included the effective temperature higher than 10℃, the accumulated rainfall, the accumulated sunshine duration) and the days of rice growth without C. medinalis in the field, and three main influencing factors were extracted by PCA (Principal Component Analysis) such as the duration of the growth period, solar-thermal conditions and moisture conditions.LAI and SPAD regression models were constructed by using the principal components respectively. The results of the LAI estimation model were good,R2 was 0.59, and the R2 of the SPAD model was only 0.12. The correlations among physiological and ecological parameters, meteorological factors and the growth days under the C. medinalis treatments were much less than CK. LAI is closer to the meteorological factors compared with SPAD and leaf-rolled rate.The correlation between physiological and ecological parameters and meteorological factors is lower with the increase of pest population in the plot.In other words,the amount of the pest is a key factor to affect LAI in the treatment fields. The meteorological factors and the days of growth had no obvious effect on the SPAD and leaf-rolled rate in the treatment plots.(2) The canopy hyper-spectral reflectance of rice decreases in the visible and near infrared bands along with the deepening of damage during the growth period.The "Bimodal" phenomenon of the first derivative spectrum was visible from 680nm to 770nm. The red edge parameter which increased at the beginning, and then decreased later shifted to shortwave direction with the deepening of damage. SPAD and LAI increased first, and then decreased during the whole growth period.They reached the peak at the flowering stage. SPAD and LAI decreased with the increase of pest amount gradually. While the leaf-rolled rate decreased first and then increased. It reached the maximum value at the jointing stage, which was consistent with the change of Cmedinalis generations. 13 vegetation indexes selected from sensitive bands had the strongest correlation with LAI, followed by the SPAD value and leaf-rolled rate. For vegetation index estimation model of physiological and ecological parameters, the best fitting was LAI regression model. R2 reached 0.762, which is constructed by a multiple regression analysis. The polynomial model of single factor constructed by single factor vegetation index PVI was the best for SPAD, R2 was 0.546. The vegetation index was less suitable for the leaf-rolled rate, R2 was only 0.492 at most.(3) The results showed that rice spike length was not affected by the increase of initial population density. But the increase of pest population could influence the growth of rice, and resulted in the reduction of plant height and tillers after comparing with different rice yield indicators under initial population density in the field. The higher population density was, the lower rice seed setting rate and grain weight were, and finally these caused large yield loss. The regression equation constructed by NDVI during the grain filling was the best for yield estimation, R2 reached 0.698. The results of the Seinhorst Model indicated that C. medinalis began to cause damage to rice. The minimum relative yield is 0.879 when the damage was the most serious. It indicated the population reached more than 17 per 100 strains. Finally, the results of two models were compared and it was found that the regression equation calculated by hyperspectral vegetation indexes was more realistic when the initial population density was from 45 to 80 per 100 strains. The results of the Seinhorst Model were more consistent with rice biological characteristics and the actual yield loss when population was more than 80 per 100 strains.