Tasselyzer, a machine learning method to quantify anther extrusion in maize, based on PlantCV

Abstract

SummaryMale fertility in maize is controlled by development and genetic programming and is directly impacted by environmental factors such as light, temperature, water, and nutrient availability; the control of this trait has substantial agronomic utility. Maize anthers emerge from male florets, which are clustered to form the tassel at the top of the plant separated from the female ear. Quantification of anther extrusion is one important aspect in the determination of male fertility. To address the lack of an automated method to measure anther extrusion on a large scale, we developed ‘Tasselyzer’, a quantitative, image-based color trait analysis pipeline for tassel image segmentation, based on the existing PlantCV platform, and we applied it to determine the proportion of anther extrusion. We evaluated Tasselyzer in maize during the seven-day period of pollen shedding as well as in the temperature-sensitive male sterile mutant dcl5. With tassel images obtained with a smart phone camera, we show that the anther scores positively correlate with anther extrusion, and such methods can be used to measure environmental impacts on the dcl5 mutant. Altogether, this work establishes an automated and inexpensive method to quantify anther extrusion in maize, which would be useful for research and breeding.Significance StatementTasselyzer is a novel image-based segmentation tool, to address the lack of an automated method to measure anther extrusion, and the impact of genetic and environment variation on male fertility in maize.