Source-linked AI summary
Agronomic Characterization and Time-Series UAV Phenotyping to Support Field Pea Breeding in Iowa
Yanben Shen, Joscif Raigne, Asheesh Singh, Arti Singh
TL;DR
Evidence on field pea adaptation in Iowa is limited, and yield-based selection may be less reliable amid genotype-by-environment effects. This study combined agronomic measurements with time-series UAV phenotyping; canopy area was the most useful yield-relevant UAV trait, with 60–65 DAP the best yield-prediction window (r=0.67).
Problem
Iowa lacks field pea adaptation-yield tests, while limited Midwest-adapted advanced lines and genotype-by-environment effects constrain direct selection based on yield alone.
Source: §1 · p. 2, §4 · p. 11Method
The study combined conventional agronomic evaluation with time-series UAV phenotyping to assess field pea performance and trait relationships in Iowa.
Source: §1 · p. 2Results
60–65 DAP was the best window for predicting yield (r=0.67), and canopy area was the most useful UAV-derived trait for pea yield.
Source: §4 · p. 10Takeaways & Limitations
Indirect selection using UAV-derived traits may improve selection efficiency and accelerate genetic gain in the Midwest breeding context described.
Source: §4 · p. 11, §4.1 · p. 11Because the CDC and PI trials shared no genotypes or checks, individual genotypes could not be compared directly across the two sets.
Source: §2.4 · p. 6
Abstract
from arXiv · showhide
Field pea has gained interest in the Midwest U.S. and has the potential to benefit Iowa cropping systems. This study evaluated 17 accessions and 7 Canadian varieties for agronomic and seed yield performance. This study found that PI accession trials exhibit genetic diversity in phenology and canopy traits, whereas the Canadian elite varieties demonstrated great yield potential under Iowa conditions. The accession genotypes matured between 75 DAP (late June) and 95 DAP (late July) across six site-years. The top-yielding Canadian varieties averaged 1961 kg/ha across two site-years. To comprehensively assess field pea performance, this study extracted time-series Unmanned Aerial Vehicle(UAV)-derived traits, assessed UAV-derived traits, and evaluated their correlations with multiple agronomic traits and yield. The results found that UAV-derived time-series canopy area was the most yield-relevant trait (r = 0.63) among all UAV-derived canopy features, vegetation indices, and manually collected measurements. Compared with yield, UAV-derived canopy area showed higher selection reliability (H2 = 0.87, CV = 13.46%), making it suitable for phenotypic selection and accelerating Midwest field pea breeding programs. This study identifies useful UAV-derived features for pea breeding and demonstrates the value of integrating diverse germplasm, elite commercial varieties, and time-series UAV phenotyping. Combining genetic variation from the accession trial with the yield potential of high-performing commercial varieties provides valuable breeding materials and phenotyping strategies for developing locally adapted, high-yielding field pea varieties for future production in the Midwest.
1 Introduction
Field pea could diversify Iowa cropping systems, but its adaptation and management there are poorly documented. This study pairs agronomic evaluation with time-series UAV phenotyping to identify useful traits for Midwest breeding.
1 Introduction
- Iowa lacks field pea adaptation-yield tests and key agronomic and management information, despite potential crop-rotation benefits.Prior pea and sorghum rotations and pea yield after other crops suggest regional potential, but the cited studies do not establish Iowa adaptation.Source: §1 · p. 2
- Time-series UAV phenotyping captures canopy development across the season and can provide more informative growth indicators than single-date measurements.Prior work applied UAV imagery to crop traits including biomass, flowering, height, lodging, maturity, and yield prediction.Source: §1 · p. 2
- The study combines time-series UAV phenotyping and conventional agronomic evaluation to identify desirable field pea genotypes and yield-relevant canopy features in Iowa.It also evaluates relationships between UAV-derived and agronomic traits and compares time-series with single-time-point phenotyping.Source: §1 · p. 2
- The study’s UAV phenotyping approach is presented as a way to improve field pea evaluation efficiency and support breeding for Midwest adaptation and yield potential.The section positions this approach alongside the need for Iowa-specific performance information.Source: §1 · pp. 2–3
2 Method
The study evaluated diverse pea germplasm across Iowa site-years, measured agronomic traits and time-series UAV features, and used mixed models and trait analyses to compare performance and measurement reliability.
2.1 Experimental design and management
- Trials covered six central Iowa site-years with two replications; 14 selected accessions, three commercial checks, and seven Canadian varieties were evaluated, with the Canadian set grown in three adjacent site-years.The Canadian variety side trial was intended to quantify a regional yield gap and set Midwest yield baselines; entries shared planting dates, field conditions, and measurement protocols.Source: §2.1 · p. 3
- The target seeding rate was 300,000 seeds acre−1; plot row configuration varied by year, and corn residue was managed before planting.Herbicides and hand weeding controlled weeds; germination exceeded 75% without significant treatment differences.Source: §2.1 · pp. 3–4
2.2 Data collection
- Agronomic measurements included early seedling vigor, flowering and maturity dates, plant and canopy height, and harvested seed yield.Vigor was scored at 17–30 days after planting on a 1–5 scale; flowering and maturity were assessed every two days after the first genotype reached each stage.Source: §2.2 · pp. 4–5
2.3 High throughput UAV phenotyping
- RGB UAV imagery was processed into orthomosaics and elevation maps, from which canopy area, canopy height, and vegetation indices were extracted.The flights used an Inspire 2 with an RGB camera, ground control points, and 0.2 cm/pixel image resolution.Source: §2.3 · p. 5
- Weekly UAV flights were fitted with a generalized additive model using at least four time points to derive canopy growth features and timing.Extracted features included mean and peak canopy area, area under the curve, maximum growth rate, and days after planting at maximum growth rate.Source: §2.3 · p. 5
2.4 Statistical analysis
- Pearson correlations and principal component analysis characterized trait relationships, while mixed-effects models and genotype BLUEs supported agronomic comparisons.Model assumptions were checked with residual diagnostics, and least-squares means and multiple-comparison procedures were used for genotype comparisons.Source: §2.4 · pp. 5–6
- Because the CDC and PI trials shared no genotypes or checks, their yield sets were compared by average performance in a mixed model, with genotype nested within experiment.The CDC–PI inference was restricted to the three shared environments; site-years were weighted equally for estimated means.Source: §2.4 · p. 6
- Broad-sense heritability was estimated from genotype, environment, and interaction variance components, and coefficient of variation assessed consistency across site-years.The heritability calculation accounts for the number of site-years and replications; CV uses residual standard deviation relative to the trait mean.Source: §2.4 · pp. 6–7
3 Result
Across the evaluated traits, mean UAV-derived canopy area was the strongest correlate of seed yield and showed greater selection reliability than yield. The results also show variation in accession performance and maturity, high yields among Canadian varieties in selected Iowa site-years, and time-series canopy differences that increased as the season progressed.
3.1 Agronomic performances
- Across six site-years, the 17 genotypes averaged 973.89 kg/ha; PI505144 was the highest-yielding accession at 1429 kg/ha.Nine accessions had yields similar to the commercial check Delta.Source: §3.1 · p. 7
- The 17 genotypes separated into early-, intermediate-, and late-maturing groups, with flowering and maturity timing correlated at r = 0.75 (P <0.001).Early, intermediate, and late groups matured within 75, 75–85, and more than 85 DAP, respectively.Source: §3.1 · p. 7
- PI117264 and PI164548 had long stems but low canopy heights, which the authors interpret as susceptibility to lodging in these lines.Source: §3.1 · p. 7
3.2 Agronomic performance in Canadian elite varieties
- The seven Canadian varieties averaged 1961 kg/ha across 24-Bruner and 25-Bruner; CDC Lewochko yielded 2566 kg/ha.The 24-Burkey site-year was excluded from this yield-potential comparison because severe early-season waterlogging affected growth; CDC-set yield advantages over the PI set were environment-dependent.Source: §3.2 · pp. 7–8
3.3 UAV-derived traits and agronomic traits
- Canopy area correlated with seedling vigor across four environments, while the strongest UAV trait for other agronomic traits varied by site-year.For grain yield, greenness 10th quantile (Gp10) was the most highly correlated UAV trait (r = 0.5); PCA also aligned mean and peak canopy area and canopy height with yield.Source: §3.3 · p. 8
- Mean canopy area correlated most strongly with seed yield among evaluated parameters (r = 0.63***), and was selected for subsequent time-series growth analysis.The analysis compared all agronomic and UAV-derived metrics.Source: §3.3 · p. 8
- Canopy area increased through the growing season, with greater genetic variation at later DAPs; a single-flight yield-prediction window of 60–65 DAP was recommended.The best single time point outperformed the time-series mean in five of six site-years, while measurements at 62 DAP performed comparably to that mean.Source: §3.3 · pp. 8–9
- Mean canopy area (H2 = 0.87, CV = 13.46%) and peak canopy (H2 = 0.91, CV = 14.83%) were more heritable and less variable than grain yield (H2 = 0.72, CV = 25.69%).The authors identify mean canopy area and peak canopy as selection criteria for field pea breeding.Source: §3.3 · p. 9
4 Discussion
The discussion identifies canopy area and season timing as promising phenotyping signals, while describing constraints from germplasm adaptation, breeding-stage, and Iowa field conditions. The conclusion frames time-series UAV canopy area as a high-throughput tool to support Midwest pea breeding.
4 Discussion
- The CDC trial out-yielded the PI trial in two of three environments, but absent shared genotypes or checks, this is a trial-level rather than genotype-level comparison.The trials did not differ significantly in the third environment.Source: §4 · p. 9
- The earliest pea genotype averaged 75 days to maturity, fitting a potential April-planted pea–millet sequence, although Central Iowa spring rainfall can delay planting and pea is susceptible to waterlogging.The passage describes an 80–90-day pea season through July followed by millet for a mid-October harvest.Source: §4 · pp. 9–10
- Canopy area was the most useful UAV trait and consistently correlated with seedling vigor, but may discriminate less well once elite populations reach full canopy and tolerate lodging.The discussion suggests canopy height, canopy volume, and vegetation indices may better distinguish such populations.Source: §4 · p. 10
- 60–65 DAP was the best yield-prediction window (r=0.67), while early canopy area related to seedling vigor and mid-season traits aligned with phenology and height.The time-series approach tracks changing traits across development, unlike single-time-point measurements.Source: §4 · p. 10
- Prior crop studies report high broad-sense heritability for UAV canopy traits and strong agreement between UAV height estimates and ground measurements, supporting UAV phenotyping's applicability to pea breeding.The cited studies also found UAV heritability estimates comparable to or sometimes better than ground-based measurements across crop cycles in wheat.Source: §4 · p. 11
- Yield-based selection is less reliable in the early Midwest breeding population because few advanced lines are locally adapted and yield is strongly influenced by genotype-by-environment interaction.The discussion proposes indirect selection using UAV-derived traits as a possible way to improve selection efficiency and genetic gain.Source: §4 · p. 11
4.1 Conclusion
- The conclusion presents time-series UAV canopy area as a high-throughput, reliable, yield-relevant trait for phenotypic selection in Midwest field pea breeding.It also calls for long-term breeding to develop locally adapted varieties.Source: §4.1 · p. 11
7 Funding
The study was supported by the Raymond F. Baker Center for Plant Breeding at Iowa State University.
7 Funding
- The Raymond F. Baker Center for Plant Breeding at Iowa State University supported this research.Source: §7 · p. 11
8 Figures
Figures 1–7 map the study’s UAV extraction and time-series analyses alongside agronomic and yield comparisons. They show phenotypes, correlations, PCA patterns, and comparisons across site-years.
8 Figures
- Figure 1 outlines time-series RGB imagery, canopy area and height extraction, vegetation indices, and seasonal canopy-area summaries and distributions.The overview also depicts traits extracted from a fitted canopy-area curve.Source: Figure 1 · p. 13
- Figure 2 compares grain yield and five manual agronomic traits across pea genotypes.The traits are seedling vigor, days to first flower, days to maturity, canopy height, and plant height.Source: Figure 2 · p. 14
- Figure 3 presents Canadian elite varieties’ seed yield averaged across two site-years in 2024 and 2025.Figure 4 plots correlations between UAV-derived traits and agronomic traits by days after planting, with darker blue dots indicating higher correlations.Source: Figure 3 · p. 15, Figure 4 · p. 15
- Figure 5 presents phenotype PCA for time-series canopy area, agronomic, and yield traits, plus genotype PCA across site-year distributions.Figure 6 compares canopy-area and yield correlations across the growing season in each site-year, including the strongest flight date and a 60–65 DAP flight.Source: Figure 5 · p. 16, Figure 6 · p. 17
- Figure 7 compares grain yield, canopy height, peak canopy, and averaged canopy area across site-years.The figure captions identify the traits and comparisons but do not report the plotted outcomes.Source: Figure 7 · p. 18
1 Supplementary Figures
The supplementary figures show pea cropping-system yield comparisons, genotype yields across site-years, Canadian genotype performance, and correlations between traits and seed yield.
1 Supplementary Figures
- Supplementary Figure S1 compares average yields for pea–milo, pea–soybean, soybean monoculture, and corn monoculture using data from [1].Source: Figure S1 · p. 24
- Supplementary Figure S2 shows mean seed yield by genotype across site-years and notes that genotype yield trends varied among site-years.Source: Figure S2 · p. 25
- Supplementary Figure S3 presents means for grain yield and five agronomic traits across seven Canadian genotypes in two site-years.Tukey-test letters indicate significant differences for canopy height, plant height, flowering, and maturity.Source: Figure S3 · p. 26
- Supplementary Figure S4 presents Pearson correlations of UAV-derived and agronomic traits with seed yield.Source: Figure S4 · p. 27
2 Supplementary Tables
The supplementary tables list the RGB vegetation indices used and statistical summaries for agronomic and UAV-derived traits.
2 Supplementary Tables
- Supplementary Table S1 lists the RGB-based vegetation indices used in the study.Source: Table S1 · p. 28
- Supplementary Table S2 covers genetic variance components, heritability (H2), and ANOVA statistics for agronomic and UAV-derived traits.Source: Table S2 · p. 28