Long‐Term Nitrogen Fertilization in Short‐Season Cotton: Interpretation of Agronomic Characteristics Using Stability Analysis

Lint yield and many other agronomic characteristics of cotton ( Gossypium hirsutum L.) are greatly influenced by environment. The objectives of this work were to evaluate the use of stability analysis on various agronomic characteristics of cotton and to assess some of the environmental factors that influence their response. Lint yield, gin turnout, picked lint percentage, and micronaire data from a single‐site, long‐term, soil fertility experiment on cotton conducted in the field under irrigated, short‐season conditions were analyzed using conventional split‐plot in time analyses of variance. Stability analyses (linear regression of a response variable (e.g., lint yield) for a specific treatment on the location/year environment mean for the same response variable) were also used to investigate the treatment by environment interactions. Stability analyses suggested that applied N had no effect on lint yield in low‐yielding environments (defined as <500 lb/acre). Applied N resulted in significant increases in lint yield in high‐yielding environments (>500 lb/acre). No significant yield advantage was detected for applying more than 40 lb N/acre in low or high yielding environments. Applied N at 80 and 120 lb/acre reduced gin turnout in all environments. Higher N rates resulted in greater reductions in gin turnout in high gin‐turnout environments. Micronaire was reduced by applied N in low‐micronaire environments, but was increased by applied N in high‐micronaire environments. Only in 2 of 11 yr did N fertilization influence micronaire of the cultivars tested. Stability analyses were useful in detecting subtle treatment effects over multiple environments in agronomic characteristics of cotton and for partitioning treatment by year interactions. This information was enhanced by having a minimum of 10 yr of data for all characteristics evaluated. Research Question Nitrogen fertilizer is generally required for optimum yield in irrigated, short‐season cotton in the Rolling Plains of Oklahoma and Texas. Because most of the cotton produced in the region is stripper harvested, excessive vegetative plant debris can affect lint yield, grades, and ginning costs. Long‐term fertilizer experiments are usually initiated by researchers to better define the fertilizer requirements of a crop produced under localized conditions and to gain a better understanding of annual average nutrient needs. A problem associated with long‐term experiments is the interpretation of data collected over many seasons. Seasonal variability from year to year (which results in significant treatment × year interaction) restricts the by‐year and long‐term interpretation of applied N rates in cotton production. Use of stability analysis (the linear regression of treatment means on the environmental mean) may be usehl in understanding differential responses of agronomic characteristics as environments change. Literature Summary Most economic characteristics of cotton are seasonally influenced, and N management can significantly affect the yield and quality of the lint and seed produced. The N requirement for cotton is contingent upon expected yield level; but actual yield is a fhction of many factors, including the weather encountered within a particular season. Long‐term experiments have been initiated to help address N fertilizer recommendations. Split‐plot in time analysis of variance is commonly used to statistically analyze data accumulated at a single site over many years; but due to treatment × year interactions, interpretation of main effects may be restricted. Stability analysis has been used to partition genotype x environment interactions in plant breeding and more recently to partition treatment x year interactions in wheat and corn experiments. Because many important agronomic characteristics of irrigated, short‐season cotton are dependent on environment, stability analysis can be used to determine which environments producers could anticipate yield increases due to N fertilization. Study Description A long‐term, continuous soil fertility experiment containing multiple N, P, and K rates was established at the Oklahoma Agricultural Experiment Station Irrigation Research Station near Altus in 1972. The objectives of this experiment were to determine proper fertilizer inputs for cotton in southwestern Oklahoma. Nitrogen rates included 0, 40, 80, and 120 lb/acre. Forty lb/acre of P 2 O 5 and 80 lb/acre of K 2 O were applied to the plots in all years. The experiment emphasized short‐season management including late‐May planting of determinate stripper‐type cultivars, with a 1 September cutoff date for irrigation. Irrigation water was furrow applied, and application was variable among years. Twenty to 40 in. of total water (rainfall plus irrigation) were encountered in each growing season during the 13‐yr time period reported in this study. Since 1982, plots have been stripper harvested; and grab samples of the stripper harvested material from each plot have been processed on small gins to approximate relative gin turnout. Boll samples have been taken from each plot and ginned on a 10‐saw lab‐type gin to determine picked lint percentage and to obtain samples for micronaire analysis. Applied Questions What was the optimum long‐term, N fertilizer rate for Lint yield? Lint yield increased with applied N in high‐yielding environments, but no benefit from N fertilization was realized in low‐yielding environments (‐400 lb/acre). Stability analysis indicated that there was little lint yield benefit from applying greater than 40 lb N/acre per year over the long‐term. Even in high‐yielding environments, the 40 lb N/acre rate was sufficient for optimum yields. Adequate residual N (due to long‐term application of 40 lb N/acre) from the soil N pool was available to result in high yield. When low‐yielding environments were experienced, residual fertilizer N accumulated in the soil. This residual N was later available for crop use when high‐yielding environments were encountered. Cotton produced under hll‐season conditions with lessdeterminate cultivars and higher yield potential would presumably have a greater N requirement. What was the optimum long‐term N fertilizer rate for gin turnout and picked lint percentage? Gin turnout was reduced by N rates in excess of 40 lb/acre in all environments. This would result in higher ginning costs when higher rates of N were applied. When seasons were encountered that resulted in high gin turnout, N fertilization reduced gin turnout more than in low gin‐turnout environments. When N was applied at 40 lb/acre, gin turnout was not significantly reduced compared with the check. The 40 lb N/acre rate reduced picked lint percentage across all environments when compared with the check; whereas, higher N rates reduced it even more. What was the optimum long‐term N fertilizer rate for micronaire? Micronaire was significantly affected by N fertilization, but the response was environmentally dependent. In low‐micronaire environments (e.g., short seasons, excessive rain in the fall, low temperatures) high N, which tends to delay maturity, probably would result in more immature fibers. In high‐micronaire environments (i.e., under the opposite conditions fiom those listed above), fibers would generally be more mature and thus more fully express the genetic potential for micronaire in that particular cultivar. Although micronaire differences in this experiment were statistically significant among years, they were of limited practical importance for these cultivars as micronaire was in the base range (3.5 to 4.9 units) in most years (9 out of 11). Different cultivars with different inherent levels of micronaire probably would be affected to a greater degree.