Straw and Cover Crop Management in Rice Production
S. Pettygrove1, J. Williams, J. Hill, R. Webster,
K. Scow1LAWR Dept, University of California, Davis
CA 95616
Estimating the N fertilizer replacement value of legume green manure crops to the following crop is fairly easy to do with conventional randomized complete block or split plot field experiments. But there are severe limits to the accuracy that can be achieved, and there are situations where N fertilizer replacement values cannot be estimated unambiguously.
The standard method involves establishing replicated main plots with and without a cover crop. Following incorporation of the green manure cover crop (and using similar tillage on the non-cover cropped control plot), fertilizer N rate subplots are established within each main plot. It is important to have several rates around the yield plateau point. Five to eight N fertilizer rates are often required. Yields are measured. The N fertilizer value of the cover crop is considered to be the difference in the fertilizer N rates that were required to achieve the maximum yields on the green-manured and control plots.
One practical limitation is that with typical field plot experimental error (coefficients of variation >10%), confidence limits for the "maximum yield point" are so wide as to make comparison of the +cover crop and control useless. In Figure 1, it is not possible to pinpoint the N fertilizer rate that was required to achieve maximum rice yield on the green-manured plot. The actual value could be anywhere between 30 and perhaps 120 lb N/acre.
The conventional practice of fitting the data to a quadratic polynomial to determine the yield maximum point is to be discouraged. Fertilizer response functions are often asymmetrical. Use of the quadratic polynomial will in many cases significantly overestimate the N rate required to achieve maximum yield. A plateau model (linear-plateau or quadratic-plateau) may fit as well and not overestimate the maximum point.
To reduce experimental error, one should conduct the experiment and data collection very carefully, use small plots and arrange them as compactly as possible, have 5 to 7 replicates, and use data analysis techniques that incorporate the variation in cover crop biomass or N content among individual plots. These techniques are often not compatible with the requirements of on-farm research plots, which require large main plots, a small number of replicates, and cover crop biomass measurement only in a few "representative" microplots.
Even if the maximum yield point can be more precisely identified, in some cases, the maximum yields will be different for the control and green-manured plots. This suggests a non-N effect of cover-cropping. In Fig. 1, it appears that the maximum rice grain yield achieved on the non-green manured plots (approximately 150 lb N/acre) is slightly lower than that measured on the cover-cropped plot. If the effect of the cover crop were merely to substitute for the ammonium sulfate applied, the maximum yields achieved on the +cover crop and the control plots would be the same, and the response curve maxima would be shifted only in the direction of the x-axis.
In Fig. 2, results are shown for the Sills Farm experiment for a five-year period. The cover crop and control plots were repeated each year on the same main plot, but N rate subplots were shifted to a different location within a mainplot each year. Rice grain yields and N rates are shown as annual averages. Even more so than in Fig. 1, it is obvious that the maximum yield achieved on the green-manured plot is higher than on the control. In other words, on the non-green manured plot, no amount of N fertilizer could achieve the yield obtained with a purple vetch green manure and 60 lb N/acre of fertilizer. This appears to be an example of a non-N cover crop effect. Unless the non-N effect can be identified and quantified, it is not possible to unambiguously determine the N fertilizer replacement value.
One other measurement can help to determine the magnitude of the N fertility value of the cover crop, although it does not determine the N fertilizer replacement value. That measurement is the quantity of N taken up by the crop (in kg/ha), usually measured only in the above-ground portion. That can be done on the unfertilized N subplots or at the yield maximum points. On the unfertilized (zero N) plots, this will allow an estimate of the quantity of N supplied by the soil in the green-manured vs the control plots. Crop N uptake measurements on the maximum yield subplots can be used to more clearly identify a "rotation" or non-N effect. If at equal yield points, the crop (straw + grain in this case) N content of the two treatments differs, a non-N effect exists. In the real world, experimental error (variability) makes these measurements of limited use.
The examples used here are from the Sills Farm cover-crop/rice
straw incorporation experiment near Pleasant Grove in Sutter Co.
What was the cause of the apparent non-N related effect of the
purple vetch cover cropping? We did not find any obvious cause.
Disease, insects, crop stand uniformity or density effects were
not evident. Effects that might be important in a non-flooded
crop such as changes in water infiltration, tendency of soil to
crust, or soil water-holding capacity would seem not relevant
in the rice system. The cover crop mainplots usually received
one extra disking, but it appeared that spring tillage on the
whole experimental area eliminated that difference. Possibly the
green manure supplies N to the crop at a different rate, working
more as a slow release fertilizer. Perhaps on the control plots
late in the season, N runs out even on subplots that earlier in
the season were oversupplied with N.
Cover Crop Research and Education Summaries