a license to display the preprint in perpetuity. It is made available under a CC-BY 4.0 International license. 540 Varifocal Lens 1080P USB Camera with H.264 High DeYinition Sony IMX323 Webcam. The 541 code for scanning ears, generating projections, and uploading those into cloud storage was 542 also updated and is available at https://github.com/fowler-lab-osu/EarScannerUtilities. 543 The set of ear projections used for the training set included 409 examples from the 544 summer Yield seasons of 2018, 2019 and 2022, encompassing images generated from three 545 different digital cameras and two different versions of the MES. For this training set, 546 projections were manually annotated using
Open resource ↗fowler-lab-osu/EarScannerUtilities · pdf-layout-page:20 lines:1-56Unverified paper record
Spatial inheritance patterns across maize ears are associated with alleles that reduce pollen fitness
bioRxiv · 20 Sept 2025 · 10.1101/2025.09.17.676879
Abstract
Significance Statement Early studies noting uneven spatial distribution of progeny genotypes after pollination support a hypothesis where differences in pollen tube growth rate can bias inheritance. We used computer vision and statistical analysis to show alleles reducing maize pollen fitness are likely to produce statistically significant increasing, decreasing, or curvilinear spatial patterns from the apex of the inflorescence to the base, suggesting that differential pollen tube growth is not the only mechanism at play. Summary Often, more pollen grains land on recipient flowers than there are ovules to fertilize. Consequently, the haploid male gametophyte engages in post-pollination competition, one way that pollen genotype can influence inheritance. The maize ( Zea mays subsp. mays L.) inflorescence (ear), with its elongated stigma and style structures (silks), has a conspicuous spatial heterogeneity, with longer silks at the base of the ear than those at the apex. To evaluate the hypothesis that alleles with reduced pollen fitness influence the spatial distribution of progeny genotypes along the ear, we developed an updated phenotyping platform that maps mutant Ds-GFP kernel phenotypes on the ear via an implementation of the Faster R-CNN machine vision model (EarVision.v2) and a statistical pipeline that evaluates the relationship between kernel position and transmission ratio (EarScape). In our dataset (1384 ears) representing 58 Ds-GFP alleles, none with Mendelian inheritance (0/48) showed any significant pollen-conditioned spatial trend. In contrast, 50% of alleles with a pollen-specific transmission defect (5/10) exhibited significant spatial effects. An insertion into a gene encoding a putative actin-binding protein, base-to-apex gradient1* ( bag1* ), conditions increased mutant transmission at the ear apex relative to the base. Surprisingly, mutant alleles of two other pollen-expressed genes can generate the opposite pattern, decreased mutant transmission toward the ear apex; and two mutant alleles of the sperm-cell attachment factor, gamete expressed2 ( gex2 ), can produce ears with transmission highest at both base and apex. We conclude that pollen fitness mutants have relatively common but heterogenous effects on the spatial distribution of progeny genotypes.
Plant phenotyping relevance
トウモロコシ穂上のカーネル表現型を画像認識でマッピングし、空間的位置と遺伝子型伝達比を解析するプラットフォームおよび統計パイプラインを開発しており、表現型取得・抽出法が研究の中心である。
abstractwe developed an updated phenotyping platform that maps mutant Ds-GFP kernel phenotypes on the ear via an implementation of the Faster R-CNN machine vision model (EarVision.v2) and a statistical pipeline that evaluates the relationship between kernel position and transmission ratio (EarScape).
Code and data availability
The paper's maize ear phenotyping assets are publicly available: the EarVision.v2 repo contains the training images with bounding-box annotations and the trained Faster R-CNN model, and the EarScannerUtilities repo contains the ear-scanning/projection code. The EarScape spatial-analysis repo (with coordinate .xml files
This is an automatically classified, unverified record. Curator approval is required before any resource enters the Catalog.