A Strategic Soil Nitrogen Test For Flooded Rice

Abstract

From 1998 until 2002 a project to develop a soil nitrogen (N) test for flooded rice was conducted in the Rice CRC. The reason for wanting such a test for the Australian rice industry is that N fertiliser is used more efficiently when applied before sowing so it is economically and environmentally preferable for as much as possible of the optimum amount of N fertiliser to be applied at that time. However excessive N applied before sowing leads to a high risk of yield loss due to cold damage. The aim was to develop a system to forecast the optimum N supply for pre-flood application and minimize the amount being topdressed which has been a safe, but inefficient system. The method of developing the test was first to compare the near infrared reflectance (NIR) spectra with crop productivity and N mineralisation measured by wet chemistry. These measurements were made with soil from 22 previous experiments measuring yield response to N applied at sowing. There were close relationships of the NIR spectra with crop productivity and N mineralisation but because of the small data set the relationships had little predictive value. However the close relationships found between NIRS, N mineralisation measured in the laboratory and crop performance encouraged us to proceed with further studies. A more detailed study related soil mineralisation across farms to crop performance. Seventeen methods of mineralisation were tested and the most reliable was found to be anaerobic incubation at 40°C for 21 days. This method predicted the optimum N requirement with a standard error of about 75 kgN/ha, which is clearly unsatisfactory for an industry where the average amount of N fertiliser applied is 145 kgN/ha. A possible reason for the low correlation between mineralisation and crop performance was that other factors were limiting N response. There was some evidence that sowing date and deficiencies of other nutrients were partly responsible for the variability of the N response. However it is unlikely that including information about these factors would lift the soil-N test to acceptable accuracy for commercial use. The most likely reason for the low correlation was that the soil depth used for mineralisation measurements was poorly defined because of the widespread levelling of rice fields which led to different depths of topsoil. Two options are proposed for more reliable application of N fertiliser at the time of sowing. Both require further research. One is to use the existing soil test only to identify soils with large amounts of potentially mineralisable N. Such a test could be the basis of a recommendation to apply little or no N fertiliser before sowing. Rice growers would still have the option of topdressing N fertiliser at the panicle initiation stage. The advantage of using a test in this way is that it is most unlikely to result in ‘false positives’, i.e. - recommendations for excessive N fertiliser leading to yield reductions. The other option is to set up a system of zone management for N fertiliser based on the likely N mineralisation in different parts of a rice field. The results in this project suggest that yield responses are more accurately predicted by sodicity than by the soil N test. It is likely that sodicity is a good indication of the depth of topsoil cut in the process of levelling. If this result is shown to be general, maps of ‘cut and fill’ areas may help in deciding the optimum amount of N fertiliser. Evidence from the Ricecheck database shows that about 10% of rice paddocks receive too much N fertiliser at sowing and suffer a large yield reduction. This leads to an annual loss of about $18 m. While this project has not led to a solution to this problem, the two suggestions arising from the project offer methods to reduce the problem.