Files
Abstract
Abstract
Global food production needs a substantial increase in yield to meet the food demand of the growing world population. The beneficial interaction between plants and microbes plays a vital role in plant growth and health and has been intensively investigated for improving crop yield for decades. However, specific beneficial plant-microbe interaction under laboratory or greenhouse conditions often fails to benefit agriculture production, suggesting a lack of comprehensive understanding of plant and microbe interaction in the natural environment.
In this dissertation, I explored how plants, microbiome communities, and the environment interact under complex field conditions. First, I investigated the effect of perennial legume cover crops on the soil microbiome community in maize fields and compared the microbiome community to those under annual cover crop systems and no cover crop control. This work reveals that soil microbiome communities under cover crop systems are significantly different from soil microbiome communities under no cover control, and the shift in microbiome composition is associated with soil nitrogen and lime buffer capacity, which are significantly different between cover crop systems and no cover control.
Second, I investigated the effect of maize genotype, growing environment, and genotype-by-environment interaction on the maize stalk endophyte community by sampling 20 specific maize hybrids across 18 locations in the United States during the summer of 2019. I found that growing environment and genotype-by-environment interaction drive the overall endophyte community composition and individual microbial taxa abundances, while maize genotype had little to no consistent effects across environments. We found that the difference in microbiome community between growing environments is associated with differences in soil and weather conditions between growing environments, such as relative humidity, soil pH value, and soil potassium value.
Third, I developed an automatic plant phenotyping pipeline based on machine-learning models to extract maize traits, which are related to maize-microbe nitrogen-fixing symbiosis and maize lodging resistance. The plant phenotyping pipeline quantifies important maize phenotypes such as stalk width, number of whorls with brace roots, and total number of brace roots by performing computer vision analysis on field images of maize plants. This pipeline provides a time-efficient, noninvasive, and non-destructive method to generate comparable phenotyping data to the labor-intensive manual phenotyping process.
Index words: Plant-microbe interaction, soil microbiome, machine learning.
Global food production needs a substantial increase in yield to meet the food demand of the growing world population. The beneficial interaction between plants and microbes plays a vital role in plant growth and health and has been intensively investigated for improving crop yield for decades. However, specific beneficial plant-microbe interaction under laboratory or greenhouse conditions often fails to benefit agriculture production, suggesting a lack of comprehensive understanding of plant and microbe interaction in the natural environment.
In this dissertation, I explored how plants, microbiome communities, and the environment interact under complex field conditions. First, I investigated the effect of perennial legume cover crops on the soil microbiome community in maize fields and compared the microbiome community to those under annual cover crop systems and no cover crop control. This work reveals that soil microbiome communities under cover crop systems are significantly different from soil microbiome communities under no cover control, and the shift in microbiome composition is associated with soil nitrogen and lime buffer capacity, which are significantly different between cover crop systems and no cover control.
Second, I investigated the effect of maize genotype, growing environment, and genotype-by-environment interaction on the maize stalk endophyte community by sampling 20 specific maize hybrids across 18 locations in the United States during the summer of 2019. I found that growing environment and genotype-by-environment interaction drive the overall endophyte community composition and individual microbial taxa abundances, while maize genotype had little to no consistent effects across environments. We found that the difference in microbiome community between growing environments is associated with differences in soil and weather conditions between growing environments, such as relative humidity, soil pH value, and soil potassium value.
Third, I developed an automatic plant phenotyping pipeline based on machine-learning models to extract maize traits, which are related to maize-microbe nitrogen-fixing symbiosis and maize lodging resistance. The plant phenotyping pipeline quantifies important maize phenotypes such as stalk width, number of whorls with brace roots, and total number of brace roots by performing computer vision analysis on field images of maize plants. This pipeline provides a time-efficient, noninvasive, and non-destructive method to generate comparable phenotyping data to the labor-intensive manual phenotyping process.
Index words: Plant-microbe interaction, soil microbiome, machine learning.