Investigating the Relationship Between Site-specific Yield and Protein of Cereal Crops S. NORNG AND A. N. PETTITT Sorn.Norng@dpi.vic.gov.au School of Mathematical Sciences, Queensland University of Technology, 2 George Street, Brisbane 4000, Australia R. M. KELLY, D. G. BUTLER AND W. M. STRONG Department of Primary Industries (DPI), PO Box 102, Toowoomba 4350, Queensland Australia Abstract. Agronomists use overlaying protein and yield maps to identify factors limiting cereal crop growth and development. Management decisions can be derived from knowing what and where these limiting factors are. In using protein and yield in this manner, there is an assumption that a physiologically or biologically significant relationship exists between grain protein and grain yield at the local level. In this paper, we investigate whether within-field yield and protein data support this relationship. The protein- yield relationship was modelled using weighted regression with global and local neighbourhoods in both 1-D and 2-D spatial location frameworks. The results from both the 1-D and 2-D analyses showed that the relationships between protein and yield are significant at both the macro (field level) (r 2 ¼ 0.25) and the micro-scale (local within field level) (r 2 ¼ 0.69). The assumption of a significant local relationship between protein and yield is supported by these data, suggesting that management decisions may be determined using such a relationship. Keywords: yield maps, grain protein, protein-yield relationship, local neighbourhood, weighted regression Introduction Within-field variation in grain protein has been observed in North America (Long et al., 1998, 2000, 2001; Engel et al., 1999) and Europe (Reyns et al., 1999, 2000; Stafford, 1999) in grain crops such as wheat and barley. Similar observations have been made in southern (Low et al. 1997) and northern grain-growing regions of Australia (Kelly et al., 2001a; Stewart et al., 2002), with the range in protein content varying from 2% to as much as 5%. For Australian grain farmers, these findings are significant for two reasons. Firstly, payment for wheat and barley commodities is based on the protein content found within the grain at harvest. In 2001 for example, malting-grade barley, with a protein content from 9% to 11.5%, fetched a premium of nearly 20% above the price paid for feed barley (i.e. A$200 versus $235; Grainco, 2001). A similar quality payment system, although with a greater degree of segregation, exists for hard-red wheat thus providing incentives to maximise grain protein content. An under- standing of the within-field variation in protein content for these crops may advantage farmers who can segregate high quality grain from low quality grain (Kelly et al., 2001b; Stewart et al., 2002). Precision Agriculture, 6, 41–51, 2005 Ó 2005 Springer Scienceþ Business Media, Inc. Manufactured in The Netherlands.