Qualitative and multivariate analyses of clonal cocoa resistance to cocoa pod borer (Lepidoptera : Gracillaridae)

Identification of potentially resistant cocoa clones to the cocoa pod borer (CPB) is one of the objectives of the MARDI's research on CPB host plant resistance research programme. Using morphological features such as the husk thickness ratio, hardness and thickness of the sclerotic layer, average damage severity index (ADSI), and the number of eggs per pod, a resistance index (RI) was calculated for each of the 53 clones studied. The clones were ranked and categorised into resistant, moderately resistant and susceptible classes. the qualitative grouping of cocoa clones into three different droups based on RI was then verified using the multivariate techniques of linear discriminant analysis and classification and the cluster analysis. A total of 11, 24 and 18 clones were qualitatively categorised as resistant, moderately resistant, and susceptible, respectively. The RI for the resistant category ranged from 4.2 for LAF17 to 19.4 for NGK28; 20.2 for 1466 tp 30.0 for NGK16 in the moderately resistant category; 32.4 for NGK13 to 45.6 for ICS39 in the susceptible category. two linear discriminant functions explained the total variation and correctly classified about 92.5 percent of the clones into their appropriate categories. Natural grouping using cluster analysis resulted in 22, 21 and 10 clones grouped into resistant, moderately resistant and susceptible categories, respectively. This classification resulted ca. 52.8 percent being appropriately categorised into respective groups, qualitatively. There seemed to be slight discrepancies between the classification developed through the RI compared to that using the multivariate techniques. The classification of clones using the RI into various resistant categories was justified as apparent error rate (APER) of reclassification using discriminant analysis and classification was very small (7.55 percent). Although correct classification of clones into respective categories using cluster analysis was low, this multivariate technique might be suitable for exploring the potential number of resistant groupings prior to the analysis using RI. Pertinent to the implementation of CPB resistance breeding programme, it would be a serious error to consider susceptible clones as resistant clones. On the other hand, loss of a potentially resistant clone may be incurred when resistant clones were misclassified as susceptible