College of Agriculture & Natural Resources Department of Plant, Soil and Microbial Sciences

Announcing Rabin KC: Final Thesis Defense Crop and Soil Science Ph.D. Degree seminar

December 1, 2025 1:00PM - 2:00PM

---

NITROGEN CYCLING IN AGROECOSYSTEMS:

FROM FIELD ASSESSMENTS TO PREDICTIVE MODELING

Contact Lauren Andring for ZOOM information

Members of the Examining Committee and their Department:

1. Dr. Sieglinde Snapp, Plant, Soil and Microbial Sciences, MSU
2. Dr. Kimberly Cassida, Plant, Soil and Microbial Sciences, MSU

1. Dr. Jennifer Blesh, School for Environment and Sustainability, UMich
2. Dr. G Philip Robertson, Plant, Soil, and Microbial Sciences, MSU
3. Dr. Sasha Kravchenko, Plant, Soil, and Microbial Sciences, MSU

 ABSTRACT

Managing nitrogen (N) in agroecosystems that rely on biological sources of N is challenging because of the fundamental need to synchronize N supply with peak crop demand. My dissertation addresses knowledge gaps regarding timing of soil mineralization processes and cover crop supply of N to cash crops, how weather and management practices influence N release, and how to quantify N-supplying capacity of soils across diverse landscapes.

In my first study, I examined how cover crops affect corn performance across five diverse Michigan organic farms. I found that cover crop biomass and Carbon (C):N ratio were more influential than background soil context in determining nitrogen dynamics and corn performance. High quality crimson clover (Trifolium incarnatum) residue (C:N ratio 15:1) was associated with higher soil inorganic nitrogen status, corn chlorophyll content, tissue N content, and grain yields relative to low quality cereal rye residue (C:N ratio 25:1). In addition, I found that corn nitrogen uptake efficiency (NUE, N in corn biomass relative to N supply) was less than 1 in farms with high soil organic matter (SOM), along with a positive ΔN (N input – N output) at the end of the season, which suggests oversupply of N relative to demand and highlights N synchrony challenges and the risk of N offsite movement even in N-limited production systems.

In my second study, I utilized 30 years of bi-weekly soil N data in two corn-soybean-wheat rotation systems with cover crops to identify key drivers of soil inorganic N release using random forest modeling. Models explained 35% of the variability in soil NO₃^--N and 15-32% variability in soil NH₄⁺-N across independent years in the two systems. I found that days after cover crop termination (DAT) was the most important driver affecting soil NO₃^--N and NH₄⁺-N availability, with air temperature as a close second. Soil NO₃^--N increased rapidly following cover crop termination and peaked at approximately 50 DAT. Two-dimensional partial dependence plots revealed interactions among DAT, temperature, and cover crop biomass in affecting soil NO₃^--N. Temperature greater than 12°C and cover crop biomass above 4000 kg ha^-1 were associated with high soil nitrate NO₃^--N levels. Although there were productivity differences between the management systems studied, both systems showed similar N dynamics, suggesting this approach was robust for understanding underlying drivers concerning N mineralization.

In my third study, I explored the opportunity of rapidly assessing soil properties such as potential mineralizable N (PMN) using soil spectroscopy across a plot scale (Michigan) and landscape scale (Malawi) study. I found that the standard 10-fold cross-validation produced optimistic predictions for PMN but using spatially independent validation withholding entire farms (Michigan) or Extension Planning Areas (EPAs) revealed substantial collapse in performance with R²dropping from 0.538 to 0.218 for Michigan, and from 0.359 to -0.014 for Malawi. This performance collapse occurred because spatially close samples share management history and weather conditions, allowing 10-fold CV to interpolate within sites rather than extrapolate to new locations. In contrast, clay content predictions at the landscape scale (Malawi) produced comparable performance between validation strategies at the landscape scale (R²= 0.729 vs 0.523). My third study definitively showed that standard 10-fold cross validation is appropriate for interpolation within farms while spatially independent validation provides more realistic performance estimates and is appropriate for generalizing to new locations.

Overall, my work contributed to understanding N dynamics in cover cropped agroecosystems as influenced by weather and management practices. My dissertation revealed opportunities and constraints of rapid soil spectroscopic methods to predict biologically dynamic soil properties to make informed decisions on N management

 

---