Predicting protein levels is part of the future toolkit

Figure 1: sorghum grain protein mapped using an on-the-go protein monitor over a large area In the 2004 crop. This prototype will be further trialled In winter crops.

The challenge is to know protein levels even before the grain is formed, write Rob Kelly, Wayne Strong, Troy Jensen, David Butler, Natasha Wells and Armando Apan.

Many Australian graingrowers could increase returns from cereal crops if they could guarantee production of grain within that window of protein that attracts premium market returns. Even higher returns might eventuate if grain within that protein window could be identified prior to crop harvest, enabling better segregation of grain during harvest or blending at receival to maximise the quantity of premium grain delivered.

Two applications of PA are being explored to enable graingrowers to increase capacity to produce grain within a desired window of grain protein; and to increase capacity to distinguish areas within the crop of similar or contrasting grain protein for improved segregation of grain during harvest.

"On-the-go" protein monitoring devices available for about US$11,000 should increase grower interest in recent research findings in southern Queensland and northern NSW.

On-the-go grain protein measurement during harvest, which now seems feasible, provides the opportunity to create, at grain harvest, a protein map similar to the yield map. Most harvesting machinery has been able to map yield for several years. On-the-go protein monitors have been progressively modified and evaluated since our research began in 1999. We trialled a recent prototype in sorghum crops in 2004-05 (see Figure 1), where grain samples were taken for laboratory analysis for comparison with field measurements. The monitor will be further trialled with winter cereal crops.

Figure 1: sorghum grain protein mapped using an on-the-go protein monitor over a large area In the 2004 crop. This prototype will be further trialled In winter crops.

At a recent international conference on PA, researchers in Australia, Europe and north America presented results obtained with similar protein monitors. The accuracy of these monitors is yet to be confirmed, but several applications for use in segregating grain, either at harvest or at grain receival, have been proposed. An operational issue for the use of a protein monitor is the requirement to check calibrations used within the instrument that affect accuracy of measurement.

Year-to-year variation in plant varieties, growing conditions and even climatic factors may create spectral measurements outside the range of the training set used to calibrate the instrument. Also, monitors may be of different design, which impacts on the manner by which the instrument"s calibration is achieved; transfer of calibrations between instruments may not always be feasible, making technical support essential. When a map showing variation in grain protein is used together with a yield map, an improved cost-effectiveness of managing nitrogen inputs for future cereal crops becomes feasible.

Coincidental monitoring of yield and protein will enable retrospective assessment of the crop"s nutritional status. Information gathered from yield and protein maps enables ‘N supply" (kg/ha) available during the growth of wheat, barley or sorghum crops to be estimated. However, because these maps are available only after crop harvest, it is the management of future cereal crops that would benefit.

Remote sensing using satellite images promises an opportunity to identify areas of nutritional or other crop stress, with the advantage of being able to make management decisions prior to harvest. Satellite and aerial spectral sensors that capture in season spectral information (reflected off crop canopies) could enable interventions to improve crop management.

Naturally, the spectral information would need to be a surrogate measure of crop yield, grain protein or of a crop stress responsible for creating variations within these crop attributes. Areas of high and low grain protein were identified, using spectral reflectance measurements for a wheat crop in 2002 and from satellite images of a barley crop in 1999 (Figure 2).

Figure 2: Protein map (top) derived from a barley field, near Dalby, In 1999. The landsat-5 image (bottom), obtained In mid-September, displayed a similar pattern to grain protein harvested In late November (r2=0.71). An earlier image obtained when the crop was severely water-stressed failed to identify the final pattern of grain protein in this crop.

Protein map derived from a barley field, near Dalby, In 1999

The landsat-5 image, obtained In mid-September, displayed a similar pattern to grain protein harvested In late November

Reliability of imagery to accurately identify highs and lows in final grain protein appears to depend on the growth status of the crop at the time the image is captured. Where the crop is water-stressed, for example, the image appears to be of little value for distinguishing highs and lows of grain protein.

The proposition that grain protein might be estimated in a cereal crop before grain is even formed seems extraordinary to many. Estimating other crop attributes from crop reflectance responses seems more plausible and some, like total above-ground crop production or grain yield, have been examined by researchers in Australia and elsewhere. Estimating grain protein from spectral reflectance response of the crop before flowering has received less attention, but recent research has shown it to have promising possibilities.

In 2003, reflectance measurements obtained with a hand-held instrument over the wheat canopy around flowering on a multi-rate long-term fertiliser experiment on the Darling Downs, confirmed potential usefulness for predicting final grain protein. High prediction accuracy between reflectance data available from inexpensive imagery, particularly In the NIR spectral region, and grain protein is being evaluated further for its reliability.

A biological link between the protein content of cereal grain and the crop"s spectral response around flowering is evident In the association between the crop"s internal nitrogen status and final grain protein content; spectral response at flowering may signal the crop"s internal nitrogen status and thus behave as a surrogate signal for final grain protein. Our recent research, However, offers an alternative explanation; reflectance response of the crop canopy at flowering can signal crop vigor, which under certain conditions may be inversely related to grain protein.

A 1.8m-diameter tethered helium balloon was used to position a multispectral sensor, consisting of two digital cameras (1 megapixel), above a wheat multi-rate nitrogen fertiliser experiment. Colour and near-infrared images of the entire experiment (3ha) were acquired from mid-tillering to around flowering. High correlation was found between reflectance In the infrared band and grain yield (r2 = 0.91) as well as moderate correlation with grain protein (r2 = 0.66). Other low-cost platforms such as satellites may provide necessary imagery to use crop reflectance information to its best advantage.

GRDC Precision Agriculture Initiative (SIP09)
GRDC Research Code: DAQ00067

For more information: troy Jensen, 07 4688 1307,

Region North