Assessment of phenotypic variability in Simarouba amara Aubl., through qualitative and quantitative descriptors

Abstract

Phenotyping is one of the strategies used by Corporación Colombiana de Investigación Agropecuaria (AGROSAVIA) for evaluating native forest species. One hundred and twenty-one 6.2 year-old mountain damson/stavewood/bitterwood/paradise trees (Simarouba amara Aubl. commonly known as machaco in Colombia) were evaluated using 34 plant, leaf and fruit descriptors to validate the use of morphological descriptors and determine the phenotype variability of AGROSAVIA’s La Libertad Research Centre’s working collection. The centre is located in the foothills of Colombia’s eastern plains
(Villavicencio, Meta). Multiple correspondence analysis (MCA) was used for analysing qualitative data and principal component analysis (PCA) for +data to reduce dataset dimensionality; this was followed by cluster analysis, using Ward's method (minimum variance method or Ward’s minimum - agglomerative algorithm) for hierarchical caraccluster analysis for grouping the trees. The results led to identifying that the wood volume descriptor had the greatest variability (31.13%) and the
quantitative variables associated with crown size and diameter, stem diameter, trunk volume, leaf length and width, total and crown height had the greatest correlation with the first three components (57.82%). Nine clusters were obtained (accounting for 95.73% of original variability) and in which trees were found for timber, arboriculture and agroforestry system purposes. The trees’ stem shape, branch height and bifurcation, the type of bark and fruiting habits had minimum variation, contrary to the that found regarding stem straightness and crown shape.