Location: Floral and Nursery Plants Research
2024 Annual Report
Objectives
Objective 1: Characterize turf germplasm diversity to enhance breeding for biotic and abiotic stress resistance. (NP215 C1, PS1D)
Sub-objective 1A: Develop a SNP-based genetic linkage map of a creeping bentgrass x colonial bentgrass interspecific hybrid mapping population.
Sub-objective 1B: Identify unique SNP markers in bentgrass suitable for use in cultivar identification strategies.
Sub-objective 1C: Identify effective methods to utilize unique SNP markers to monitor/quantify bentgrass population transitioning under field conditions.
Objective 2: Determine genetic mechanisms influencing turfgrass properties that would allow reduced management inputs. (NP215 C1, PS1D)
Sub-objective 2A: Identify new sources of germplasm and associated alleles that result in enhanced drought tolerance and post-drought recovery.
Sub-objective 2B: Identify new sources of germplasm and associated alleles that enhance dollar spot resistance.
Sub-objective 2C: Identify new sources of germplasm and associated alleles controlling yield potential.
Objective 3: Evaluate the impact of diverse plant germplasm (improved turfgrasses and alternative species) on urban ecosystem services. (NP215 C1, PS1D)
Sub-objective 3 A: Determine turfgrass species suitable for the mid-Atlantic transition zone under different mowing frequencies.
Approach
Our project plan explores Genetic (G) x Environment (E) x Management (M) interactions by leveraging high-throughput genotyping and phenotyping technologies, including the development of an AI-enabled graphical user interface to identify genes/SNPs associated with traits of interest (abiotic and biotic stress, yield potential) for developing low input turf germplasm (Objective 1 and 2); and determining the impact of existing and improved turf germplasm on urban ecosystem services (Objective 3).
Progress Report
In Objective 1a, Genotype by Sequencing (GBS) was utilized to generate Single Nucleotide Polymorphisms from a creeping x colonial bentgrass mapping population. A total of 4068 segregating markers were identified and utilized to develop a linkage map of the mapping population. The generated linkage has 21 linkage groups covering approximately 2000 cM. The linkage map has low levels of marker duplication with only 5 loci having duplicate markers. This level of duplication is far lower than other bentgrass linkage maps that have been developed.
In Objective 1b, GBS was also utilized to generate data from 12 plants from each of 20 bentgrass cultivars. DNA extraction is completed and samples are currently being processed through the GBS pipeline. The resulting marker data will be utilized to identify SNP markers that are the most useful for bentgrass cultivar identification/discrimination.
In Objective 1c, leaf clippings were collected every week over the summer from four bentgrass cultivars in the NTEP fairway trial, and two quantification methods using a droplet digital PCR (dd-PCR) and a quantitative PCR (q-PCR) with the newly developed unique SNP markers are being tested to see whether early disease detection (low fungal dose) is possible using this sensitive quantification method in combination with the unique SNP markers.
In Objective 2 a-c, an interspecific bentgrass hybrid population was utilized to study the genetics of important characteristics such as drought tolerance, high yield, and different morphological traits, such as stolon vs rhizome production.
In Objective 2a, seasonal drought studies during the winter and the summer, respectively, were conducted using an interspecific bentgrass hybrid population comprising 300 hybrids along with two parents - creeping (drought susceptible) and colonial bentgrass (drought tolerance). During dehydration treatments, the percentage of greenness was monitored daily using an autonomous raspberry pi system with RGB sensors. This system allowed effective and precise evaluation of drought progression of the whole population. Quantitative Trait Locus (QTL) mapping was performed based on % greenness results, and we found two genomic regions associated with mild drought stage, six regions with intermediate drought stage, and three regions with severe drought stage. The concept of daily-basis QTL mapping has never been done as it is time-, labor-, and resource-intensive. The newly developed AI-based imaging and image processing system allowed us to identify several QTLs associated with drought progression. The second year of the summer and winter drought studies, respectively, will be performed in FY25. Furthermore, the growth patterns (i.e. stolon vs rhizome), and root biomass of the population were evaluated. There were no significant correlations between drought tolerance and growth pattern or root biomass, indicating that the hybrid population includes many hybrids displaying creeping type growth habit that are also drought tolerant. Some genotypes were intermediate types displaying both creeping and colonial types. Those drought tolerant hybrids displaying creeping or intermediate type growth habits are especially valuable because creeping type (stolon) plants produce large volumes of seeds (high yield) compared to colonial type (rhizome).
Also in Objective 2a, a smartset of the F1 interspecific hybrid population was selected based on their drought performance from the summer and winter drought studies. Temporal profiling of physiological responses (i.e. leaf temperature, relative soil water content) as well as metabolic responses (i.e. carbohydrates, organic acids, and amino acids) of 15 hybrid lines displaying drought tolerance, intermediate tolerance, or susceptibility, along with two parents were investigated to understand the mechanisms of drought tolerance in this F1 interspecific hybrid bentgrass population.
In Objective 2a, we investigated seasonal changes in turfgrass quality in response to changes in temperatures, daylength, and species-specific dormancy over the summer by monitoring monthly the color and overall quality of an F1 interspecific bentgrass hybrid population along with two parents from June to December. We found that rhizomatous colonial-type hybrids were generally darker, and stayed green longer over the summer while stoloniferous creeping-type were lighter and went dormant earlier.
In Objective 2b, disease experimental plots were established in the field. Six replicates of a total of 290 hybrid lines along with 30 replicates of two parents were plugged in the field.
In Objective 2c, a study to measure potential yield traits such as tiller number, seed weight, and germination rate was conducted. All the measured traits, i.e. drought performance, growth pattern, dormancy, yield potential, of the interspecific bentgrass population are being processed, and will be compared to select and develop superior turf germplasm. The second-year study will be performed in FY25.
Also under Objective 2c, we developed a method to count tillers in individual Agrostis (bentgrass) hybrid lines quickly and accurately using deep learning models (i.e. convolutional neural network (CNN)). Methods used YOLOv8 and Faster R-CNN, and these developed models were compared to a traditional edge segmentation algorithm-based method. When three methods were compared, the CNN-based YOLOv8 model (R2 = 0.97) outperformed in estimating tiller numbers compared to the Faster R-CNN model (R2 = 0.85) and the edge segmentation algorithm (R2 = 0.87). The use of neural network technology is increasing in the realm of agricultural research, and developments into new methodologies will drive advances in phenotype categorization and quantification.
Accomplishments
1. Development of AI-based image processing system for turfgrass. Precise phenotypic assessment of large mapping populations, comprising a few thousand plants, is time-consuming and labor-intensive, and highly dependent on individual observers. In addition, results may be subjectively biased and can vary due to environmental effects. To make this process more efficient, ARS researchers in Beltsville, Maryland, built a cost-effective system to automatically capture hourly images of 1,000 turf genotypes growing in the greenhouse on a daily basis. An AI-based image processing pipeline was developed using simple algorithms that can quantify drought symptoms and drought progression patterns. This imaging system and image processing pipeline can perform the same type and amount of work at significantly less cost than high-priced robotic gantry systems, and can be done in a few hours vs months of manual observations. This system is of potential value for plant breeders of any crop that can be evaluated under greenhouse conditions, as it allows effective and accurate temporal evaluation of large populations.