Scholarlink

<p>Quantifying the genetic variability and diversity of <i>Passiflora</i> spp. species is essential to determine their fruit, ornamental and medicinal value. The objective of this study was to evaluate the phenotypic diversity of 114 spp. genotypes based on 74 morphological descriptors related to the leaves, flowers and whole plant, as well as to eliminate redundancy through a list of minimum descriptors for characterization of the germplasm. For this purpose, we submitted 18 quantitative (morphometric) traits to principal component analysis, discarding the traits with low contribution (≤ 6.0%) to the first two components (PC1 and PC2), but which presented correlation above 0.70. The 56 qualitative (multicategorical) descriptors were analyzed according to Shannon's entropy. The analysis was performed using the Gower distance and clustering by UPGMA (Unweighted Pair Group Method Using Arithmetic Averages). After selecting the most informative descriptors, we used nonparametric analysis (random forest - RF) to identify the most important variables within each group formed, as well as the reliability of the clustering. Based on the results obtained, it was possible to identify and discard 29 (51.79%) qualitative descriptors and seven quantitative ones (38.89%). The joint analyses with and without discarding the qualitative and quantitative descriptors led to similar clusters, especially among genotypes of the same species. The RF analysis achieved consistently satisfactory results, enabling ranking the predictor variables within the groups formed with a Kappa index of 87% and accuracy of 88%. The information generated can guide future work both in the area of genetic resources and genetic improvement of <i>Passiflora</i>. </p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Optimizing germplasm characterization via phenotypic diversity assessment and descriptor selection in Passiflora spp.

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年3月9日

查看全文
  • Onildo Nunes de Jesus,
  • Taliane Leila Soares,
  • Idalia Souza dos Santos,
  • Jeovani Silva Almeida,
  • Mateus Araujo dos Santos,
  • Lucas Kennedy Silva Lima

摘要

Quantifying the genetic variability and diversity of Passiflora spp. species is essential to determine their fruit, ornamental and medicinal value. The objective of this study was to evaluate the phenotypic diversity of 114 spp. genotypes based on 74 morphological descriptors related to the leaves, flowers and whole plant, as well as to eliminate redundancy through a list of minimum descriptors for characterization of the germplasm. For this purpose, we submitted 18 quantitative (morphometric) traits to principal component analysis, discarding the traits with low contribution (≤ 6.0%) to the first two components (PC1 and PC2), but which presented correlation above 0.70. The 56 qualitative (multicategorical) descriptors were analyzed according to Shannon's entropy. The analysis was performed using the Gower distance and clustering by UPGMA (Unweighted Pair Group Method Using Arithmetic Averages). After selecting the most informative descriptors, we used nonparametric analysis (random forest - RF) to identify the most important variables within each group formed, as well as the reliability of the clustering. Based on the results obtained, it was possible to identify and discard 29 (51.79%) qualitative descriptors and seven quantitative ones (38.89%). The joint analyses with and without discarding the qualitative and quantitative descriptors led to similar clusters, especially among genotypes of the same species. The RF analysis achieved consistently satisfactory results, enabling ranking the predictor variables within the groups formed with a Kappa index of 87% and accuracy of 88%. The information generated can guide future work both in the area of genetic resources and genetic improvement of Passiflora.