Date of Award

Spring 5-8-2020

Level of Access Assigned by Author

Open-Access Thesis

Degree Name

Doctor of Philosophy (PhD)


Ecology and Environmental Sciences


Lakesh Sharma

Second Committee Member

Andrei Alyokhin

Third Committee Member

Richard Brzozowski

Additional Committee Members

Lily Calderwood

Sukhwinder K. Bali


Nitrogen (N) fertilizer rate is important for high yield and good quality of potato tubers. In this dissertation, I seek to study the response of different potato cultivars under different N fertilizer rates and how that can impact tuber quality, examine the performance of active optical sensors in improving a potato yield prediction algorithm, and evaluate the ability of active optical sensors (GreenSeeker (GS) and Crop Circle (CC)) to optimize a N recommendation algorithm that can be used by potato growers in Maine. This research was conducted at 11 sites over a period of two years (2018–2019) in Aroostook County, Maine; all sites depended on a rainfed system. Three potato cultivars, Russet Burbank, Superior, and Shepody, were planted under six rates of N (0-280 kg ha-1), ammonium sulfate and ammonium nitrate, and were applied in a randomized complete block design (RCBD) with four replications. Active optical sensor readings (normalized difference vegetation index (NDVI)) were collected weekly after the fourth leaf stage began. The coefficient of determination (R2) between soil organic matter (OM) content and total tuber yield for all sites combined was 0.78**. Sites with ≥ 30 g kg-1 of soil OM produced higher total tuber yield, marketable yield, and tuber weight per plant (39.45%, 45.22%, and 54.94%, respectively) than sites with ≤ 30 g kg-1 of OM. Specific gravity increased by 0.18% in the sites with ≥ 30 g kg-1 of OM. The total tuber yield for the three cultivars was maximized at 168 kg N ha-1. Vegetation indices measurements obtained at stages of 16 or 20 fully expanded leaves were significantly correlated with tuber yield, which can be used in the yield prediction model. Sensor measurements obtained at the 20th leaf stage were significantly correlated with tuber yield, with the exponential model showing the best fit for the regression curve. The recommended N rate calculated based on in-season sensor readings was reduced by approximately 12–14% compared to the total N rate that growers currently apply based on the conventional approach.