Intra-leaf modeling of Cannabis leaflet shape produces leaf models that predict genetic and developmental identities.

The iconic, palmately compound leaves of Cannabis have attracted significant attention in the past. However, investigations into the genetic basis of leaf shape or its connections to phytochemical composition have yielded inconclusive results. This is partly due to prominent changes in leaflet number within a single plant during development, which has so far prevented the proper use of common morphometric techniques. Here, we present a new method that overcomes the challenge of nonhomologous landmarks in palmate, pinnate, and lobed leaves, using Cannabis as an example. We model corresponding pseudo-landmarks for each leaflet as angle-radius coordinates and model them as a function of leaflet to create continuous polynomial models, bypassing the problems associated with variable number of leaflets between leaves. We analyze 341 leaves from 24 individuals from nine Cannabis accessions. Using 3591 pseudo-landmarks in modeled leaves, we accurately predict accession identity, leaflet number, and relative node number. Intra-leaf modeling offers a rapid, cost-effective means of identifying Cannabis accessions, making it a valuable tool for future taxonomic studies, cultivar recognition, and possibly chemical content analysis and sex identification, in addition to permitting the morphometric analysis of leaves in any species with variable numbers of leaflets or lobes.

This research was supported by projects WECANN (CGL2017-84297-R, Ministerio de Ciencia, Innovación y Universidades), Generalitat de Catalunya (grant no. 2021SGR00315), and M. Balant FPI predoctoral contract of the Ministerio de Ciencia, Innovación y Universidades (PRE2018-083226). This work is also supported by NSF Plant Genome Research Program awards IOS-2310355, IOS-2310356, and IOS-2310357. This project was supported by the USDA National Institute of Food and Agriculture, and by Michigan State University AgBioResearch. We would like to thank Joan Uriach Marsal from Uriach Laboratories for additional financial support

Introduction Materials and Methods Plant material and growing conditions Leaf sampling and imaging Image analysis and landmarking Reconstruction of the new modeled leaves Validation of the leaf modeling approach Morphometric analysis of the central leaflet shape using previously established methodologies Data analysis of modeled leaves Results Heteroblastic changes in leaflet number along the main axis Validation of the leaf modeling approach Width : length ratio and central leaflet shape analysis Principal component analysis on modeled leaves Linear discriminant analysis and prediction of genetic and developmental identities on modeled leaves Discussion Conclusions Acknowledgements Competing interests Author contributions