Patterns of variability in quantitative morpho-agronomic characteristics of Philippine traditional corn from selected provinces

Two hundred recently collected traditional corn populations were characterized based on quantitative morpho-agronomic descriptors, aimed to assess the distribution and frequency of quantitative variation, determine the principal component of variation and similarity groupings of the collection. Data showed wide spread statistical range and high variance and standard deviation for kernel, tassel, plant and ear descriptors. Nine descriptors however, showed skewness, indicating aggregation of values towards either end of the range. Means indicated values closer to the unimproved corn types such as shorter plant heights, higher number of leaves, narrower stem diameter, lighter kernel weight, shorter ear length, early tasseling and silking, smaller kernel dimensions, smaller cob diameter, and non-synchronous flowering. As expected, collection were distributed over many classes ranging from 6 to all 10 frequency classes but generally falling within the 4 or 5 central classes. In general, values obtained indicated high variability of the collection having wide dispersion distributed to multi class ranges. A cumulative 74.6% of sample variability was contributed by 11 PCs with the first 3 principal components explaining about 44.48% of the sample variation. PC1 was composed of 22 variables representing plant, tassel and ear characteristics. PC2 included tassel peduncle length and number of kernel rows while PC3 identified 100-kernel weight and kernel width. Number of primary branches of tassel and rachis diameter where not found significant in all primary axes. The collection clearly scattered over the biplot space but no clear pattern of provenance association can be identified. Weak groupings can be observed for some collections from Masbate, Oriental Mindoro and Agusan del Sur [Philippines]. Cluster analysis further confirmed inherent morpho-agronomic variability of the collection set where maximum clustering was achieved at 15.75 Euclidean distance coefficient. The collection can be grouped into two clusters with 9 outliers. Cluster 2 was the larger group with 3 sub-clusters while Cluster I was composed of 18 collections. Cluster 2 maybe sub-grouped into 3 smaller clusters 2A, 2B, 2C. No clear association to provenances was established but 21 of the Masbate collection clustered in IIA and 8 of 14 Aklan and 7 of 10 Agusan del Sur collections both clustered in 2C. Cluster analysis validated high degree of variability of the collection set. The present study therefore confirmed the presence of considerable genetic diversity, outlined distribution patterns of variations observed and identified the principal components among the local corn germplasm. Recommendations were forwarded.