Skip to main content
ARS Home » Pacific West Area » Pullman, Washington » Grain Legume Genetics Physiology Research » Research » Publications at this Location » Publication #354427

Research Project: Improving Genetic Resources and Disease Management for Cool Season Food Legumes

Location: Grain Legume Genetics Physiology Research

Title: Field phenotyping using multispectral imaging in pea (Pisum sativum L) and chickpea (Cicer arietinum L)

item QUIROS, JUAN - Washington State University
item McGee, Rebecca
item Vandemark, George
item ROMANELLI, THIAGO - Universidad De Sao Paulo
item SANKARAN, SINDHUJA - Washington State University

Submitted to: International Journal of Remote Sensing
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 6/10/2019
Publication Date: N/A
Citation: N/A

Interpretive Summary: Plant breeders typically develop new crop varieties by selecting parents with traits that the breeders would like to see in new varieties, such as improved disease resistance, higher yields, or better nutritional qualities. Many important traits, including yield and disease resistance, may be difficult to reliably measure or may require that plants be grown to maturity in order to be accurately measured. In this report we describe studies in which unmanned aerial systems (UAV), which are small flying vehicles that are remotely controlled, were used to take pictures of pea and chickpea field plots at several times during the growing season. The pictures were analyzed with computer software to estimate the amount of above ground biomass present in each plot, and relationships between plant biomass and plot seed yield were determined. We found that the closest relationship between plot biomass and seed yield was observed for both peas and chickpeas when pictures were taken of the plots when the plants were flowering. UAVs may allow us to identify high yielding lines without requiring us to wait until the plants are mature. These selected plants can immediately be used in the field to make crosses with plants that have other desirable traits. This will allow us to more rapidly develop improved varieties of peas and chickpeas that will contribute to the sustainable production of these nutritious grain legumes (pulses).

Technical Abstract: Dry pea (Pisum sativum) and chickpea (Cicer arietinum) are important grain legumes and the Palouse region of the Pacific Northwest U.S. is a major production area. Grain legumes are typically grown in rotation with cereals such as wheat, as they disrupt the lifecycle of pathogens and weeds, fix atmospheric nitrogen in symbiosis with rhizobium bacteria, and complement the nutritional qualities of small cereal grains. The USDA-ARS grain legume breeding program focuses on developing varieties of both spring and autumn-sown peas, and spring-sown chickpeas that are high yielding and have high levels of resistance to biotic and abiotic stresses. Yield potential is one of the critical traits influencing the selection of high performing varieties. In this study, low-altitude aerial multispectral imaging was performed to phenotype yield differences early in the season, with the objective of developing a reliable image processing protocol to estimate yield potential for breeding program in both crops. Five experiments (3 of dry pea and 2 of chickpea) with 10 to 25 genotypes, four of which were planted at two locations, were analyzed for this purpose. Images were acquired at about 60, 70 and 90 days after planting (DAP). Normalized difference vegetation index (NDVI), green normalized vegetation index (GNDVI), soil adjusted vegetation index (SAVI) and simple ratio (SR) image based features (SUM, MIN, MAX, MEAN) were extracted. In most cases, the MEAN was found to be consistently correlated with dry seed yield (p<0.05). MAX values had stronger correlation with yield when using SAVI and SR, since these indexes are not affected by saturation. Higher Pearson’s correlation coefficient values between yield and extracted features were obtained during the data collected during the flowering phenological growth stages. In addition, differences in correlation coefficient between image features from individual and mosaicked images with that of seed yield were not high. The developed methodology can be replicated as a reliable image processing protocol to estimate yield potential for dry pea and chickpea breeding experiments.