Source: AUBURN UNIVERSITY submitted to NRP
PLANT BREEDING-CIN: ADVANCING INTERCROP BREEDING WITH GENOMIC AND PHENOMIC SELECTION
Sponsoring Institution
National Institute of Food and Agriculture
Project Status
ACTIVE
Funding Source
Reporting Frequency
Annual
Accession No.
1032486
Grant No.
2024-67013-42667
Cumulative Award Amt.
$893,178.00
Proposal No.
2023-11038
Multistate No.
(N/A)
Project Start Date
Jun 1, 2024
Project End Date
Sep 30, 2028
Grant Year
2024
Program Code
[A1141]- Plant Health and Production and Plant Products: Plant Breeding for Agricultural Production
Recipient Organization
AUBURN UNIVERSITY
108 M. WHITE SMITH HALL
AUBURN,AL 36849
Performing Department
(N/A)
Non Technical Summary
Intercropping is the practice of growing two or more species in the same place and/or time. In best-case scenarios, intercrops yield more per acre than monocrops, with fewer inputs, and offer long-term yield stability. Breeding for intercropping (B4I) is an underexploited strategy to improve intercropping and our proposal's focus. Purpose-bred intercrop-adapted cultivars would complement agronomic optimization and help overcome barriers to adoption.Breeding better intercrops means testing lots of diversity in intercropping situations. For best results, breeders need to do this early on the selection process, when 100's to 1000's of potential varieties need to be tested for each species. In this situation, it isn't practical to test all combinations of both crops. This limits selection for intercrop performance until the end of the breeding process when most candidates are already removed.The purpose of our Coordinated Innovation Network is to demonstrate solutions to these problems, advancing B4I by example. We focus on oat-pea (Avena sativa, Pisum sativum) as a model of cereal-legume grain intercrops.We will demonstrate an approach wherein only a tractable sample of oat-pea combinations needs to be grown in the field, but all oats and all peas are observed with at least one partner of the other species. Though we limit field testing to only a subset of oat-pea combinations, we will obtain DNA fingerprints on all oats and all peas. Our multi-year, multi-location pilot trial will evaluate thousands of intercrop plots while only sparsely testing the possible oat-pea combinations. Based on the genomic and plant trait data, we will train a machine learning model to predict which peas, oats, and pea-oat combinations yield best together. If successful, our approach will enable selection for intercrop-ability to happen even in the early stages of breeding, when most candidate varieties haven't yet been field tested.Our results will enable the design of selection strategies that will accelerate the improvement of oat-pea and other intercropping systems through breeding.We would like to see our network grow into a B4I community, addressing many species and intercrops.
Animal Health Component
25%
Research Effort Categories
Basic
75%
Applied
25%
Developmental
(N/A)
Classification

Knowledge Area (KA)Subject of Investigation (SOI)Field of Science (FOS)Percent
2011560108135%
2011412108135%
2011560102015%
2011412102015%
Goals / Objectives
Our overall goal is to provide proof-of-concept that genomic selection and phenomics can overcome the otherwise intractable combinatorial problem that B4I represents.Obj. 1: Pilot genomic prediction as a decision-support tool for assessing the intercrop-ability of oats with peas, peas with oats, and their specific combinations.Obj. 2: Develop non-destructive, remote, and proximal-sensing-based predictions of intercropped grain yield both overall and partitioned by species.Obj. 3: Develop and validate a crop growth model suitable for cereal-legume intercrops like our focal oat-pea system.Obj. 4: Design selection strategies for accelerated improvement of oat-pea intercrops using stochastic simulation and multi-objective optimization.
Project Methods
Field Trials: Conduct a multi-year, multi-location field trial evaluating the performance of 4,000 total oat-pea intercrop plots (400 plots each over 10 location-years). Sparse co-testing using a p-rep p-loc experimental design of 480 oat and 440 pea genotypes.Phenotyping: Remote and proximal sensing with UAVs and ground-based camera systems. Biomass sampling, grain yields, and more.Genome-wide genotyping of all oats and all peas.Genomic prediction coupling genotype and phenotype data. Novelty: first attempt at genomic prediction of intercrop performance and the estimation of "general and specific mixing ability" parameters.Training ML models to predict intercropped grain yield both overall and partitioned by species based on non-destructive, remote and proximal-sensing.Adapting crop growth models to model intercrops.Multi-objective optimization and stochastic simulation to design future selection strategies for intercrops, especially for oat-pea.

Progress 06/01/24 to 05/31/25

Outputs
Target Audience:Our target audience of academic/scientific professionals has heard about this project at a few conferences during the summer of2025. Changes/Problems:One major opportunity led to a change from the narrative: instead of in-house developing a training dataset and ML for Sub Obj. 2.1-2.4, we partnered with Steve Mirsky's USDA Beltsville Team and are using his "Plant Map 3D" (PM3D) system. We didn't know about the system during the grant proposal stage. PM3D team provided us with the sensor systems and we are adding our image and biomass data to their training model. In return, we expect biomass predictions to accomplish Sub-obj. 2.5 in particular. What opportunities for training and professional development has the project provided?Several graduate students, a post-doc and student workers are gaining experience through participation at various levels of the project. How have the results been disseminated to communities of interest? Nothing Reported What do you plan to do during the next reporting period to accomplish the goals?The second year of the field experiment and data collection covering Obj. 1 and Obj. 2 will be completed. We will analyze year 1 data to accomplish the first round of model training and genomic prediction validation (Sub-obj. 1.4-1.5) By the end of period two we may be able to start analysis for Obj. 3 and Obj. 4. We will hold another scientific advisory board meeting.

Impacts
What was accomplished under these goals? Obj. 1: Pilot genomic prediction as a decision-support tool for assessing the intercrop-ability of oats with peas, peas with oats and their specific combinations. Sub-objective 1.1: Develop The Triticeae Toolbox (T3) to better handle data from intercropping. Underway. Almost all datatypes are to be housed in the database. Some data are uploaded already, more in progress. Sub-objective 1.2: Increase seed for all downstream experiments. Successfully accomplished during Summer of 2024. Sub-objective 1.3: Conduct a multi-year, multi-location field trial evaluating the performance of 4,000 total oat-pea intercrop plots. The first year's field trial was conducted during Spring+Summer of 2025. All trials have harvested and seed processing and data prep are underway. Sub-objective 1.4: Train oat-pea intercrop genomic prediction models. Soon. Genotyping was completed during Summer of 2024 for both oat and pea germplasm sets. Obj. 2: Develop non-destructive, remote and proximal-sensing-based predictions of intercropped grain yield both overall and partitioned by species. Sub-objective 2.1: Capture seasonal timing of magnitude of growth using aerial multispectral imagery, up to seven times per growth period. This was accomplished for as many times, at as many locations as possible. Not all locations. Sub-objective 2.2: Collect ground-based non-destructive (proximal-sensing) data on intercropped canopy cover as a complement to aerial imaging. Completed during the Spring 2025 trial. Data are being captured with the PM3D system by the PSA group at USDA-ARS Beltsville. Sub-objective 2.3: Generate validation data for remote and proximal sensing by pairing up to four imaging timepoints with ground-based destructive harvests. Completed paired with each proximal-sensing mission. Sub-objective 2.4: Train supervised machine learning models to predict species-specific canopy traits and grain yield based on remote and/or proximal-sensing data. Nothing to report, yet. Sub-objective 2.5: Assess the benefit of integrating phenomic time series data into (genomic) predictions of growth curves, biomass, yield of oat-pea intercrops. Nothing to report, yet. Nothing to report yet forObj. 3andObj. 4. A project advisory board meeting was held in May 2025.

Publications