Fig 1.
Methodological procedure for the phenome characterization of Phaseolus and Arachis accessions.
(1) Digital phenotyping during seed regeneration: A) Seed photographic capture using photobox and classic descriptors. Color and binarize image extraction; B) Field photographic capture of flowers and classic color descriptors. Color image extraction for digital colorimetry; C) Scanning of trifoliate leaves and classical descriptors. Binarized image extraction for geometric morphometrics analysis; D) Image capture of pods in the photo box and classic descriptors. Binarized image is extracted for geometric morphometrics analysis. (2) Extraction of phenomic descriptors and selection: E) Contour analysis using geometric morphometry of seed, leaves and pods. The morphospaces (principal components -PC) are determined; F) Extraction of RGB color spaces from each colorimetric group using kmeans in seeds and flowers; G) Selection of descriptors using random forests from the mean minimum depth. (3) Data analysis of functional phenome: H) Three databases are generated that represent the selected functional descriptors, the class of the descriptors (Character, numerical and ordinal) and the associated groups that are the functional entities (Species, ecological group and genetic group); I) Vertices accessions of the functional spaces. These accessions represent the greatest phenotypic variation; J) Functional diversity indexes that relate accessions to functional entities. Functional specialization (FSpe), functional originality (FOri) and functional independence (FIde) are calculated. (4) Icons relating the stages of digital phenotyping. Image capture and image preprocessing are observed.
Fig 2.
Functional phenomic descriptors that describe the phenotypic diversity of Arachis spp. accessions.
(A) Principal component 1 of leaflet morphometrics. (B) Principal component 1 of pods morphometrics. (C) Principal component 1 of seeds morphometrics. (D) Flower and seed colorimetric clusters of accession 18745 A. pintoi. (E) Flower and seed colorimetric clusters of accession 22162 A. repens. (F) Flower and seed colorimetric clusters of accession 22226 A. archeri. (G) Flower and seed colorimetric clusters of accession 22636 A. paraguariensis. (H) Flower and seed colorimetric clusters of accession 22242 Arachis sp.
Fig 3.
Functional phenomic descriptors that describe the phenotypic diversity of Phaseolus spp. accessions.
(A) Principal component 1 of leaflet morphometrics. B) Principal components 1 and 2 of seeds morphometrics. (C) Flower and seed colorimetric clusters of accession G24764B P. dumosus x P. vulgaris. (D) Flower and seed colorimetric clusters of accession G26457 P. lunatus. (E) Flower and seed colorimetric clusters of accession G35271 P. coccineus. (F) Flower and seed colorimetric clusters of accession G35586 P. dumosus. (G) Flower and seed colorimetric clusters of accession G40503 x (P. lunatus x P. polystachyus. (H) Flower and seed colorimetric clusters of accession G40885 P. tuerckheimii.
Fig 4.
Selection of classical descriptors, phenomic descriptors and their combination used a minimal depth distribution on Arachis spp accessions and its classification.
(A) Distribution of minimal depth of the phenomic descriptors (B) Classification of accessions using confusion matrix from phenomic descriptors. (C) Distribution of minimal depth of classical descriptors. (D) Classification of accessions using confusion matrix from classical descriptors. (E) Distribution of minimal depth of the combination of phenomic and classical. (F) Classification of species using confusion matrix from combining phenomic descriptors with classical descriptors.
Fig 5.
Selection of classical descriptors, phenomic descriptors and their combination evaluated via a minimal depth distribution on Phaseolus spp accessions and its classification.
(A) Distribution of minimal depth of the phenomic descriptors (B) Classification of accessions via confusion matrix from phenomic descriptors. (C) Distribution of minimal depth of classical descriptors. (D) Classification of accessions via confusion matrix from classical descriptors. (E) Distribution of minimal depth of the combination of phenomic and classical descriptors. (F) Classification of species via confusion matrix from combining phenomic descriptors with classical descriptors.
Table 1.
Indices of functional diversity for the accessions of Phaseolus and Arachis species evaluated with phenomic (Phe), classic (Cla) and their combined (Comb) descriptors.
The indices of functional specialization (FSpe) and functional originality (FOri).
Fig 6.
Vertices accessions of the functional space for each of the descriptor types used for bean and forage peanut accessions.
(A-B) Bean and forage peanut vertices accessions with phenomic descriptors. (C-D) bean and forage peanut vertices accessions with classical descriptors. (E—F) bean and forage peanut vertices accessions with the combination of descriptors. The value of the area of the convex hull and the number of vertex accessions for each of the descriptor types for both beans and forage peanuts.
Fig 7.
Functional diversity indices of the accession grouping by Phaseolus spp. for each type of descriptor.
(A-F) Functional diversity indices for phenomic descriptors in the accession grouping of P. dumosus, P. coccineus and P. lunatus. (G-L) functional diversity indices for classical descriptors in the accession grouping of P. dumosus, P. coccineus and P. lunatus. (M-R) functional diversity indices for combined descriptors in the accession grouping of P. dumosus, P. coccineus and P. lunatus. For the indices of functional specialization (FSpe), functional identity (FIde), and functional originality (FOri), the relationship between the first two primary coordinates (PC1, PC2) is observed.
Fig 8.
Functional diversity indices of the accession grouping by Arachis species for each type of descriptor.
(A-C) Functional diversity indices for phenomic descriptors in the accession grouping of A. pintoi and A. paraguariensis. (D-F) functional diversity indices for classical descriptors in the accession grouping of A. pintoi and A. paraguariensis. G-I) functional diversity indices for combined descriptors in the accession grouping of A. pintoi and A. paraguariensis. For the indices of functional specialization (FSpe), functional identity (FIde), and functional originality (FOri), the relationship between the first two primary coordinates (PC1, PC2) is observed.
Fig 9.
CIAT genebank workflow suggestion resulting from the proposed methodology.
(A) Capture of images of flower, pods, leaves, whole plant and herbarium sampling and digitization during the collection of new accessions. (B) Capture of images of flower, pods, leaves, whole plant and herbarium sampling and digitization during the seed multiplication and regeneration. (C) Image processing, morphometric and colorimetric data extraction, descriptor selection and data analysis with AI to classify accessions. (D) Capture images of pods and seeds, to feed the database and in the seed purity verification process.