Predicting grain yield and designing density-tolerant maize ideotypes through 3D architectural phenotyping at silking

Abstract

Accurate prediction of maize ( Zea mays L.) yield and design of density-tolerant ideotypes are crucial for crop management and yield improvement. Three-dimensional (3D) phenotypic traits are closely related to light interception efficiency and yield formation, theoretically serving as important indicators for yield prediction and plant architecture optimization. However, studies utilizing 3D phenotypic traits for yield prediction and ideotype design remain limited. In this study, a two-year (2023–2024) field experiment was conducted with 10 maize hybrids grown under three planting densities (37,500 (low density, LD), 67,500 (medium density, MD), and 97,500 (high density, HD)) plants ha -1 . Plant 3D phenotypic traits at the silking stage were captured using the MVS-Pheno platform. The results showed that increasing planting density led to more compact plant architecture and significant changes in 3D phenotypic traits. A partial least squares regression model integrating 3D phenotypic traits with canopy light interception data achieved high prediction accuracy for yield (R 2 = 0.91, RMSE = 0.49 Mg ha -1 ). Feature sensitivity and correlation analyses further identified projected area (PJA) and plant side width (PSW) as critical indicators for designing varieties tolerant to high density. Furthermore, a strategy is proposed to match plant ideotype to different planting densities: under MD, the leaf area per plant (LAP) and PJA increased, whereas the PSW and leaf orientation value (LOV) decreased; under HD, the LAP, PJA, and PSW decreased, whereas the LOV increased. These findings provide an effective model for yield prediction and a valuable reference for breeding maize with optimal architecture for high-density cultivation. • Partial least squares regression model combining point cloud parameters with canopy light interception predicted yield well. • Plant side width and projected area stand out as pivotal traits influencing maize grain yield. • Plant type optimization under 37,500 and 67,500 plants ha -1 aimed to enhance population light interception. • Plant type optimization focused on improving canopy light distribution under 97,500 plants ha -1 .

Plant phenotyping relevance

MVS-Phenoによる3D形態形質の取得と、収量予測・品種設計への解析が研究の中心であり、形質抽出および予測手法を実質的に評価している。

Code and data availability

公開論文であることは確認できましたが、現在の公式API・許可済み取得経路では本文を自動取得できませんでした。

No evidence-backed public reproduction asset is currently recorded.