Prediction Model of Basal Stem Rot (BSR) Based on Ganoderma Boninense Using UV/Vis Diffuse Reflectance Spectroscopy

Abstract

Basal Stem Rot (BSR) disease was considered as the most destructive disease in oil palm tree. Ganoderma boninense. fungi causing BSR on oil palm tree, could release cell wall degrading enzymes (CWDE) which responsible to degrade polysaccharide on oil palm tree cell wall such as cellulose and lignin into reducing sugar. In this research optical measurement using UV/Vis Diffuse Reflectance Spectroscopy (DRS) was utilized to qualify and quantify BSR level based on its reducing sugar. Several stages were conducted to qualify and quantify reducing sugar comprising sample preparation, spectral data acquisition and chemometrics analysis. Health Stem (HS) and Infected Stem (IS) were prepared. The two stem condition were dried and powdered separately then mixed into 5 mixtures from both stems (100% HS, 75% HS + 25% IS + 50% HS + 50% IS, 25% HS +75% IS and 100% IS) and duplicated to produce 10 total sample. Reducing sugar was measured for each sample. Other than that, spectrum acquisition data was conducted using UV/Vis DRS. In the final stage, chemometrics analysis was performed where reducing sugar was set as response and spectral data was set as predictor. The result showed that Principal Component Analysis (PCA) could well classify 4 group of mixtures. While for quantitative analysis, Partial Least Square (PLS) and Support Vector Machine Regresion (SVR) were used to develop prediction model of reducing sugar. PLS showed low performance with the highest R2 val and R2 cal accounting for 0.218 and 0.019 and RMSEC and RMSECV were 0.019 and 0.023, respectively. Besides, SVMR using first derivative savitzky-golay (DG1) preprocessing showed high R2 cal and R2 val accounting for 0.823 and 0.701 with RMSEC and RMSECV were 0.012 and 0.028, respectively.