GOSSYM/COMAX has been under development for several years and has been evaluated and used in pilot programs around the Cotton Belt. The system utilizes a computer model simulation of cotton plant growth (GOSSYM) under different weather conditions and management practices. COMAX is an expert system component which acts to make several GOSSYM simulations under slightly changing conditions to attempt to maximize cotton production by systematically adjusting nitrogen or irrigation.

Several factors are utilized in the growth simulation and tend to be very specific for a given location. These factors include 1) the soil type which has varied physical properties such as bulk density, moisture retention, saturated hydraulic conductivity, and the percent sand, silt, and clay; 2) the weather patterns for a given area; 3) agronomic cultural practices such as row spacing, planting patterns, and the variety maturity; and 4) the initial soil fertility status. For this presentation, all factors were held constant except the Initial Soil Fertility (ISF) file which provides the information on the residual soil N status, organic matter, and moisture status of the soil. Only the nitrate nitrogen was varied for the simulations. No NH₄-H was included in the soil profile and organic matter was held constant. The nitrate-N was added at 10 lb/A increments to the 0-611 zone up to 50 lb N/A. Additional nitrate was added either as 10 or 20 lb N/A to the subsoil zones and moved through the profile from the 6-1211 zone through to the 30-3611 zone. All simulated crops were grown without irrigation or mid-season adjustments and with 100 lb N/A applied two weeks before planting as urea-ammonium nitrate solution (75 lb N/A as ammonium-N and 25 lb N/A as nitrate-N). The soil type for the simulations was a Basket loam which was characterized near Greenville, Mississippi. Each GOSSYM run was summarized for lint yield, first square, first bloom, first open boll, date of fruit nitrogen stress (NF < 1.0), and days to maximum yield.

There were no differences detected in first square, first bloom, or first open boll regardless of the amount of profile nitrogen or the location of the nitrogen. The major factors which changed with the changes in the initial soil fertility were lint yield and days to nitrogen stress in the fruit (NF). The days to maximum yield were also affected by the N rate but the effect was small. The simulated lint yields increased with an increasing N rate in the 0-6" soil zone. The simulated lint yield ranged from 877 lb/A where no residual inorganic nitrogen was included in the profile to 1197 lb/A where there was 50 lb N/A in the 0-6" soil zone. When either 10 or 20 lb N/A was added to the other five 6-in soil zones, the simulated yields increased. When the 10 or 20 lb N/A was moved deeper into the profile the lint yields decreased in most cases. For this particular soil (Bosket loam) and conditions, the simulated lint yields were highest when the extra 10 to 20 lb N/A was located in the 12-18m soil zone. In most cases, the efficiency of the nitrogen in a particular zone was less than the efficiency of the nitrogen in the 0-6" zone.

It would appear from this limited data base, that more emphasis could be placed on adequate sampling of the top 0-18" for this soil rather than extensive soil sampling to 36". Field research is needed before this type of recommendation could be considered. Much more investigation is needed to determine whether the patterns observed for this particular soil type and weather scenario are similar for other soils and weather conditions. For most soils, a deeper profile may be needed at some time to determine if there are build-ups of nitrates in the lower zones of the soil profiles. It may be that more extensive sampling of the top soil zones will better serve the producer in terms of evaluating the effect of residual N on the total nitrogen pool.