Dr. Steven Anderson
Dr. Steven Anderson

In today's Plant Phenomics First Author Insights, we invited Dr. Steven Anderson to share his insights as the first author of Prediction of Maize Grain Yield before Maturity Using Improved Temporal Height Estimates of Unmanned Aerial Systems published in The Plant Phenome Journal.

Tell us about yourself a little bit, what's your current position, your education experience and how did you get into science.

I am a Postdoctoral Researcher within Dr. Brian Pearson's lab at the University of Florida Institute of Food and Agricultural Science’s Mid-Florida Research and Education Center leading the Industrial Hemp Pilot Project research in Apopka, FL.
I received my B.S. in Biology, with a minor in Chemistry, from the University of Central Florida. I attended Texas A&M for my graduate studies earning my M.S. and Ph.D. in Plant Breeding under Dr. Seth Murray within the Maize Breeding and Quantitative Genetics Program. My M.S. thesis was focused on improving breeding efficiencies through in vitro, tissue culture methodologies and studying the effects of advanced matting designs (specifically multiparent advanced generation intercrossing) on linkage disequilibrium and quantitative trait loci (QTL) mapping power and resolution. My Ph.D. dissertation was focused on the implementation of unoccupied aerial systems (UAS; a.k.a. drones) withing a plant breeding program, specifically focused on improving phenomic prediction of grain yield through temporal growth analysis and accessing the use of UAS imaged derived phenotypes for temporal mapping of QTL.
Outside of science, I am still a major plant nerd. I spend a great deal of my time tending to my garden, growing vegetables, and collecting tropic fruit species with syntrophic and perma- agriculture in mind. I enjoy rock climbing, traveling, hiking, fishing, craft beer, video games, and spending my free time with family and friends.

What was the significant issue(s) in your paper? Why did you and your team care about it?

Maize is one of the most important cereal crops in the world. The majority of efforts to improve maize genetics are focused within the corn belt of the US, resulting in Texas grain yields half of that found in Iowa. Our group is interested in discovering novel techniques to drive genetic gain (faster genetic improvement of crops) of Texas germplasm and provide high-quality genetics to the farmers of the South Eastern US. Our manuscript demonstrates a low-cost, high throughput method to temporally estimate maize height throughout the growing season using UAS surveying, rather than the traditional manual, terminal height measurement which has been highly correlated/predictive of hybrid grain yield in TX. This study demonstrates that the use of temporal growth parameters to model the growth curve of maize can increase the explainable grain yield variation by four-fold. Improvement in predictive models, such as our finding, will enable breeders to make actionable decisions on breeding advancements prior to yield data availability or in the absence of yield data (e.g. biological destruction of a crop).

What was the problem(s) to be solved and your proposed solution?

In the past, genomic data has long been looked towards as the answer to deriving solutions to improving our crop genetics for present and future food production. Unfortunately, a critical requirement to utilizing genetic information is having highly informative phenotypic data to identify genomic regions of interest. Collecting phenotypic data is labor/time intensive and prone to human bias. There is an unknown quantity of phenotypic variation not visible/measurable by the human eye. Our study acts as a case study, using plant height, to demonstrate the possibilities of UAS versus manual phenotyping. Can we capture additional genotypic variation using UAS phenotypes? Can we identify actionable image-based phenotypes that may aide in selection accuracy? Can we create tools/pipelines so that others can utilize emerging UAS technologies? Looking forward, how do we incorporate phenomic and genomic/pedigree data to improve our selection accuracy?
Utilization of UAS to survey fields significantly reduces labor and time requirements in the extreme Texas summer heat. A group of three can collect height measurements within a maize trial of 500 plots in ~4 hrs, a drone can image 1500 plots in ~1 hr with a single individual involved. This demonstrates the potential of UAS phenotyping; we can reduce labor and time, allowing for increase population sizes which drives genetic gain. UAS phenotyping was shown to increase genotypic variation explained compared to terminal, manual height measurement which drives genetic gain. Additionally, we demonstrated that utilizing temporal UAS datasets and using functional parameters to model temporal growth improved the ability to explain grain yield variation (i.e. model prediction accuracy) by four-fold compared to conventional height measurements. In all, we demonstrate the use of UAS in collect a critically, important agronomic phenotype and demonstrated methods of which UAS could facilitate driving genetic gain.

What was the contribution(s) of this study and who could benefit from it?

Our research demonstrated the utilization of UAS image phenotypes beyond accuracy validation to manual measurements. We showed that UAS phenotyping provides improved (i) partitioning of variation explained by genotypes, (ii) correlation to grain yield, and (iii) ability to explain grain yield variation compared to manual phenotypes. All in all, our findings reach beyond height prediction of grain yield, demonstrating opportunities of UAS phenotyping (specifically temporal trends) to improve our understanding of plant biology, genotype-by-environment interaction, phenotypic plasticity, and phenomic prediction. We spent great effort to develop an example for publicly available UAS/phenomic datasets and have developed a publicly available software tool and pipeline scripts to improve the adoption of UAS phenotyping withing crop improvement/research programs (https://github.com/andersst91).

Are there any interesting stories behind the paper?

During manual data collection for validation purposes, several of UAS estimates we were out after several days of heavy rain collecting height measurements. Our newest graduate student (also working on the project) joined us to take height notes. Many of you are not familiar with the soil at the Texas A&M farm, but it is heavy, red clay and when it is saturated it is very tactile and you walk through the fields with 20 lbs of clay on each boot when you are not stuck halfway to your knees. We slogged through 500 plots of mud and rain everyone completely disheveled, soaked wet, and covered in mud. I remember her measuring stick was so covered in mud that she could not read the numbers and would have to wash it in standing water in the furrow, with no betterment. Needless to say, we were all miserable. But I remember her walking with her socks in her hand and we asked: “Why are you carrying your wet, muddy, destroyed socks? Throw them on the ground and we will get them later.” With hesitation, she left the sock. Several months later we learned that those socks were a gift from her mother in Brazil. Thinking to myself, "well now I feel terrible…" because we never went back to get the socks! Needless to say, it was an experience, although quite miserable, but we all look back and laugh at these moments.