Assis Kamu (2016) Modelling the yield loss of oil palm due to Ganoderma Basal Stem Rot disease. UNSPECIFIED thesis, Universiti Malaysia Sabah.
|
Text
Modelling the yield loss.pdf Download (1MB) | Preview |
Abstract
Oil palm or scientifically known as Elaeis guineensis Jacq. is the most efficient oilseed crop in the world. This commodity crop is considered as the golden crop in Malaysia. This is due to the contribution of the oil palm industry to the country's overall economy, providing both employment and income from exports. The efforts of the country to strengthen the industry are being interrupted by a fatal disease which is called as Ganoderma Basal Stem Rot (BSR) disease. This disease can cause a significant economic loss to the industry. To date, there is still no effective control of the disease at the commercial fields' level. The existing control measures can only prolong the productive life of the infected palms. It is very crucial to the planters to estimate the yield loss due to the disease. Currently, there is no existing mathematical model that can be used for that purpose. Therefore, this empirical study was conducted to build a mathematical model which can be used for yield loss estimation due to the disease. For the purpose of data collection, three commercial oil palm plots with different production phase (i. e. step ascent phase, plateau phase, and declining phase) were selected as the study sites. The yield and disease severity of the selected palms in the three study sites were recorded for the duration of twelve months. Before building the yield loss model, a data screening was performed in order to remove palms with extreme yield values. The identification of the main sources of multicollinearity was also performed based on correlation-based test and also variance-based test. All the remaining data set was splitted into model building data set and validation data set. Two model building approaches were applied, which are estimation-post-selection and Bayesian model averaging (BMA). For estimation-post-selection approach, there were two subset selection algorithms were applied, namely backward stepwise subset selection and best-subset selection. The best single model from the best-subset selection algorithm was chosen based on eight criteria, namely Akaike Information Criterion (AIC), Finite Prediction Error (FPE), Generalised Cross Validation (GCV), Hannan- Quinn (HQ), RICE, SCHWARZ, sigma square (SGMASQ) and SHIBATA. The predictive performance of the three best models which represent three different model building algorithms were assessed and compared. Based on mean square error (MSE), root mean square error (RMSE), and mean absolute error (MAE), BMA model has the lowest values, thus selected as the best model for oil palm yield loss. This best model (i. e. estimated loss of total bunch weight in 12 months = -24.632 + (-18.307*R2) + (13.456*R3) + (21.531*R4) + (2.346*AUDPC) + (0.551*NEIGHBOUR) + (35.113*PT) + (0.014*AUDPC*NEIGH BOUR) + (- 0.011*AUDPC*PT)) revealed that planting technique as the most important predictors of oil palm yield loss and followed by disease progress (AUDPC), disease severity (mild, medium, and severe), number of infected neighbouring palms, and two interaction variables. The economic loss was then estimated by using the best model. The estimated economic loss showed that the loss can be up to 68 percent as compared to the attainable yields of all the infected palms. In conclusion, the yield loss model built in this study can potentially be used by the oil palm planters in helping them to estimate the yield loss as well as economic loss due to Ganoderma BSR disease if no treatment is applied.
| Item Type: | Thesis (UNSPECIFIED) |
|---|---|
| Keyword: | oil palm, Ganoderma Basal Stem Rot, mathematical model, yield loss estimation |
| Subjects: | S Agriculture > SB Plant culture |
| Department: | FACULTY > Faculty of Science and Natural Resources |
| Depositing User: | MUNIRA BINTI MARASAN - |
| Date Deposited: | 06 Dec 2017 13:49 |
| Last Modified: | 06 Dec 2017 13:49 |
| URI: | https://eprints.ums.edu.my/id/eprint/17733 |
Actions (login required)
 |
View Item |