Shifting focus from quantity to quality: Mapping, understanding, and managing within-field variability in crop quality in cotton and grain systems

Abstract

Australian cotton and grain growers are renowned for producing high-yielding, high-quality fibre and grains. However, there is considerable variation in both yield and quality within and between fields, farms, and seasons. Grain quality, namely grain protein content (GPC), and cotton fibre quality, including length and micronaire (a measure of fibre fineness and maturity), are key determinants of price recieved by growers. Thus, growers must manage for both quality and quantity to attain premium prices. Site-Specific Crop Management (SSCM), the practical application of precision agriculture (PA), allocates resources and practices to match spatiotemporal variability. However, uncertainty about how much variation justifies PA investment, and limited understanding of its drivers to inform decisions, hinders adoption. More on-farm and industry data is being collected than ever (e.g. yield data, variable-rate inputs), and public data (e.g. remote sensing imagery) is freely available to represent variability in GPC and fibre quality. Understanding how and why quality varies within fields can equip growers and advisors to make better decisions for profitable, sustainable systems. This thesis explores how on-farm and public spatial data can describe and quantify within-field variability in cotton yield, fibre quality (length, micronaire), and GPC, and understand its drivers. Chapter 1 provides background on the cotton and grains industries and PA’s role in managing variability. Chapter 2 presents a geostatistical approach using area-to-point kriging to map areal crop data. Chapter 3 shows how yield, agronomic, and public data can model GPC in the absence of sensors. Chapter 4 investigates the yield–GPC relationship using machine learning to identify drivers of spatial variation. Chapter 5 compares SSCM opportunities for yield and GPC using the Opportunity Index. Overall, this thesis explores how PA can support better decisions that optimise both yield and quality.