Nitrogen Fertilizer Management and Recommendations for Wheat Production in Central Mexico

Search PMN

Agustin Limon-Ortega and Eduardo Villaseñor-Mir, INIFAP-CEVAMEX, AP 10, Km 17.5 Carretera Mexico-Lecheria, CP 56230, Chapingo, Mexico

Abstract

The current N fertilizer recommendation for wheat (Triticum aestivum L.) production in the central highlands of Mexico is to apply 71 lb N per acre. However, this N rate appears to exceed current crop needs. Therefore, research activities in farmers’ fields to estimate the optimum N rate and timing were initiated in 1999 and continued until 2002. A field trial was conducted at eleven location-years representing the major wheat production areas of central Mexico. Four N rates were applied (0, 36, 63, and 89 lb N per acre) under three N timing strategies; basal at planting, end of tillering-early jointing, between 30 and 60 days after planting, and split (one third at planting and the remaining two thirds at the end of tillering-early jointing). To estimate the critical N fertilization rates for optimum grain yield in each environment, yield data was fitted to a linear plateau model on N fertilization rates. Results indicated that the critical N rate for optimum wheat production is lower than current N fertilizer recommendation. According to the regression model used, predicted critical N fertilization, that explained 47% of the optimum grain yield variability, ranges from 28 to 66 lb N per acre. For grain yield goals below 45 bu/acre (± 3.1 bu/acre), farmers do not have to apply N fertilizer. For grain yield goals above this threshold, farmers have to apply 2.6 lb N per acre for every additional bushel expected. Nitrogen fertilizer application at the end of tillering-early jointing (Feekes scale 6) showed a tendency to increase grain yield by means of increasing the number of spikes per unit area.

Introduction

Wheat grain yields in farmers’ fields across the central highlands of Mexico average less than 45 bu/acre (8) and vary considerably throughout the region from 10.5 to 75 bu/acre. This is due primarily to a combination of variable rainfall conditions (15.8 to 39.4 inches) and variable management practices, mostly for weed control and N nutrition. Generally, the application of these inputs may exceed the recommended rates, contributing to lower potential yield and N use efficiency.

Soil and plant tests to develop optimum N recommendations are often not available, or their use is not economic for growers in many less-developed countries. Under such conditions, empirical field experiments with N fertilizer rates are an option that can generally be used to generate fertilizer recommendations. However, this approach may provide varying results as different statistical models can be used to fit the response curve to N application. For example, results from a study (1) using a quadratic regression model for a year-location study with wheat in the central highlands of Mexico showed that economic optima for responsive sites to N varied between 68 and 192 lb N per acre. However, the current N fertilizer recommendation from the National Institute for Forestry, Agricultural, and Livestock Research (INIFAP) is to apply a flat rate of 71 lb N per acre. This contrasting information could be attributable to differing models the researchers might have used, as well as to the genotypes planted, as the N use efficiency varies among them (13). This is especially true under rainfed conditions with variable water regimes (15). Subsequently, as more reliable models can be applied and more genotypes are released for the wheat production areas, current N management guidelines need re-examination due to the potential negative effects on the environment, crop development, and production costs.

Most wheat growers in the specific case of the central highlands of Mexico apply an arbitrary N fertilizer rate that in many cases may not meet the crop N needs. In the case of excessive applications, the extra N can ‘leach’ as nitrate from soils to ground water reservoirs giving rise to environmental concerns (10). Such losses represent higher production costs since the N use efficiency is greatly reduced. Therefore, research in farmers’ fields is needed to adjust N rates to grow wheat using statistical models that predict lower critical N rates without reducing grain yield.

As an alternative, the application of linear plateau models to N response curves developed from field experiments has been suggested to predict critical N fertilization rates. The advantage of this model over methodologies that use linear or quadratic regression procedures is that the predicted critical N rates for optimum yield are lower (3,18). A linear plateau model consists of two intersecting straight lines fitted to grain yield on N fertilizer rates. The transition zone where the two regression lines intercept represents the critical N rate that produces an optimum grain yield at an economically optimum rate of fertilization (6). The first regression line in a spline model is an upward line that goes from the intercept on grain yield (abscissa) at zero N application, to the point where it reaches the origin of the second regression line. Depending upon field conditions, the second regression line tends not to show major slope changes. This means that after the critical N rate has been reached, grain yield is not expected to change substantially with additional N fertilizer applications. However, even linear plateau models are sometimes criticized because the abrupt change in trend going from one segment to the next does not represent what would naturally occur (16); their application to N response curve in maize (3), barley (18), and wheat (12) has proved to be an adequate approach.

The N fertilizer is often split applied to ensure maximum tiller development (11,17,19). It has also been reported that delaying the application of N fertilizer during the cropping season generally increases the content of grain protein (7,20).

The objective of this study was to re-examine the critical N fertilization rate for wheat production by means of a linear plateau model that optimized grain yield and the effect of N timing on grain yield.

Field Studies

Field plots were established at eleven location-years in farmers’ fields from 1999 to 2002 in the states of Tlaxcala and Mexico of the central highlands of Mexico. A location-year combination is referred to as an environment. A brief description of each environment is shown in Table 1. All plots were established under rainfed conditions with the farmers controlling all of the production practices except for seeding rates, N application rates, and weed control.

Table 1. Summarized description of 11 environments in the states of Tlaxcala and Mexico.

Plots were sown to the bread wheat (Triticum aestivum L.) variety ‘Nahuatl F2000,’ released in the year 2000 by INIFAP, at a seeding rate of 89 lb/acre. Plots were 4 m by 7 m and harvested by hand. Grain yield components were determined in four environments. Nitrogen treatments were 0, 36, 63, and 89 lb/acre. The N was applied with three different timing treatments, all at planting, all at the end of tillering-early jointing (Feekes scale 6), and a split treatment consisting of one third at planting and two thirds at the end of tillering-early jointing. A factorial arrangement with a RCB design and three replications was used. Urea was the N source used. At planting, fertilizer was broadcast and incorporated by disking along with the seed. At the end of tillering-early jointing growth stage, between 30 and 60 days after crop emergence, N fertilizer applications were surface applied without regard for soil moisture conditions.

Analysis of variance using SAS software (SAS Institue, Inc., Cary, NC)was used to estimate the effect of N rates and timing on grain yield. Treatment main effects were considered as fixed and locations as random (5). To determine the N rate that produced the optimum grain yield, a linear plateau model consisting of two intersecting straight lines was fitted to grain yield from each environment on N rates as shown by (9). This model is defined by Eq. [1] and [2]:

Grain Yield (bu/acre) = a₁ + b₁*N, if N fertilizer < critical N rate [1]

Grain Yield (bu/acre) = a₂ + b₂*N, if N fertilizer ≥ critical N rate [2]

where N is the rate of N application in lb/acre; and a’s (intercepts), b’s (linear coefficients, grain yield response to N fertilizer application), and critical rate of N fertilization (occurs at the intersection of the two linear regressions) are constants obtained by fitting the model to the data and are estimated using SAS software with the PROC NLIN option (16).

Nitrogen Rate

The analysis of variance showed that most of the grain yield variability can be attributed to the effect of the environment-by-N rate interaction (Table 2). Therefore, results from each environment were analyzed individually to explain the effect of N rate. Nitrogen timing and all of its interactions were not significant for grain yield.

Table 2. Statistical significance from the analysis of variance for grain yield and yield components.

Sources of variation^x	DF	Grain yield	DF	Spikes per ft²	Kernels per spike	Kernel weight
Environment (Env)	10	**	3	**	**	**
N rate	2	ns	2	ns	ns	ns
Env*N rate	20	*	6	ns	ns	ns
N timing	2	ns	2	ns	**	**
Env*N timing	20	ns	6	ns	ns	ns
N rate*timing	4	ns	4	ns	ns	ns
C. V. (%)		17		22	12	7.4
Mean		3718		428	36	30

^x The zero-N plots were excluded for purposes of the analysis of variance.

The linear plateau approach applied individually to each environment indicated that the critical N rates varied from 28 to 66 lb N per acre (Fig. 1) averaging 43 lb N per acre (± 12.5 lb N per acre) from the pooled data. This indicates that the predicted critical N fertilizer rate for optimum grain yield can be lower than the current N fertilizer recommendation (71 lb N per acre) for wheat production.

Fig. 1. Spline regression model applied to grain yield on N rates in 11 environments of the central highlands of Mexico.

It is interesting to note that the critical N rate in the state of Tlaxcala for environment 4 and 2 varied from 30 lb N per acre (17.9 inches of rain) to 66 lb N per acre (15.9 inches of rain), respectively, and in the state of Mexico for environment 7 and 9, from 28 lb N per acre (28.4 inches of rain) to 59 lb N per acre (23.3 inches of rain), respectively. However, the lowest critical N rates in those environments (4,7) corresponded to the highest total amount of rainfall, and the highest critical N rate to the lowest total amount of rainfall.

These divergent results suggest that to estimate the optimal N rate in each environment, rainfall distribution during the cropping season is more important than the total amount of rainfall. For example, the scarcity of rains at the onset of the season in an environment with high yield potential (environment 4) until booting stage (Fig. 2) produced a poor plant establishment with low final grain yield and a relatively low critical N rate (Fig. 1). By way of contrast, the significant amounts of rain at the time of N application in environment 2 produced greater yields and required a higher critical N rate. It should be noted that the total amount of rain in environment 2 was lower than in environment 4. Previous reports (14) have indicated that dryland wheat production is highly dependent on rainfall after planting.

Fig. 2. Precipitation (inches per 10 days) during the cropping season from June to October at 11 environments in central Mexico. Solid bars indicate the period when N treatments were applied; except for environment 11, the crop was planted the last week of June when there was no rainfall.

Then, to decide the amount of N fertilizer to apply within the range of 28 to 66 lb N per acre, farmers should consider factors such as their yield goal (20) and the anticipated rainfall distribution for the cropping season. However, the flat application of 43 lb N per acre, as estimated from pooled data, would be enough to obtain an average final grain yield of about 62 bu/acre, which is 27% greater than the mean wheat grain yield for central Mexico (45 bu/acre). On the other hand, for grain yield goals below 45 bu/acre (±3.1 bu/acre), mainly in low and medium yield potential environments, farmers may not have to apply N fertilizer (Fig. 3) as there will not be a substantial response, but a negative effect on environment and farmers’ economy. For grain yield goals above 45 bu/acre (±3.1 bu/acre), farmers should apply 2.6 lb N per acre for every additional bushel expected (Fig. 3).

Fig. 3. Relationship between predicted grain yield and critical N rates from the linear plateau model (environments 1 and 11 excluded). [Yield = 45 + 0.39(N), r² = 0.47]

Grain yield measured on the zero-N plots and average yield response to N fertilizer application showed variable results (Fig. 1). This differential result could be attributed to a higher N mineralization rate resulting from warmer temperatures and higher soil organic mater contents in environments at relatively low altitudes (2). These results are similar to those (1) reporting that the soil’s ability to mineralize the soil organic matter and annual weather conditions have a strong influence in determining annual variability in wheat responses to N fertilizer applications. This variation, according to b₁ parameter from Eq. 1 in Fig. 1, indicated that wheat yield response across environments ranged from 0.30 to 0.63 bu/lb of N fertilizer irrespective of the grain yield potential in each environment. The negative response to the N fertilizer application for the first segment of the spline regression in environment 8 (Fig. 1) could be attributed to the amount of rain that was unusually low compared to the long term (Table 1).

Nitrogen Timing

Data analysis across environments for the N timing treatments did not affect grain yields or spike numbers per unit area (Table 2). However, N timing affected kernels per spike and kernel weights in the environments where these yield components were measured. Table 3 indicates that basal N applications at planting increased both kernel weigh and number of kernels per spike. Even the number of spikes per unit area was not affected by N timing treatments; the application of N at the end of tillering-early jointing (Feekes scale 6) showed a tendency to increase this yield component. Thus, this result is an indication that N applications at the end of tillering-early jointing stage, as part of the split N timing strategy, promote the growth of fertile tillers. It has been shown that the number of heads varies according to available water and N management (4). The split-N application is recommended to ensure maximum tiller development (11), in addition to the improvement of grain quality (7,20).

Table 3. Grain yield components as affected by N timing applications.

N timing treatments	Thousand kernel weight (ounces)	Kernels per spike	Spikes per ft²
Planting	1.08 a	36.2 a	38.0 b
Split	1.06 ab	35.9 a	39.2 b
End of tillering	1.03 b	33.2 b	44.2 a
Mean	1.06	35.1	40.5
LSD (0.05%)	0.043	2.3	4.39

Recommendations

In general, the application of the linear plateau model to estimate the critical N rate for wheat production showed that the pooled critical N rate is 43 lb N per acre (± 12.5 lb N per acre). However, if grain yield goal is considered in estimating the N rate, farmers should apply 2.6 lb N per acre for every additional bushel expected above 45 bu/acre. For grain yield goals below this, farmers may not apply N fertilizer.

Nitrogen fertilizer application at the end of tillering-early jointing stage (Feekes scale 6) has the potential to increase the number of spikes per unit area overall if rainfall conditions are considered.

Even results from this study showed a differential response to N as to give a generalized N fertilizer recommendation, it is clear that the total amount to apply in a given environment at a given crop stage is associated with a degree of risk. However, since optimum N rate across environments varies within a small range (28 to 66 lb N per acre), the risk is negligible if growers experience and predicted yield level are applied in combination with rainfall forecast to obtain adequate grain yields.

Acknowledgments

The financial assistance of the Fundacion Produce Tlaxcala, A. C., the support of cooperating farmers, and the help of Adrian Ramírez-Vega in the field work are gratefully acknowledged.

Literature Cited

1. Barreto, H. J., and M. A. Bell. 1995. Assessing the risk associated with N fertilizer recommendations in the absence of soil test. Fert. Res. 40:175-183.

2. Bell, M. A. 1993. Organic matter, soil properties, and wheat production in the high valley of Mexico. Soil Sci. 156:86-93.

3. Bullock, D. G., and Bullock, D. S. 1994. Quadratic and quadratic-plus-plateau models for predicting optimal nitrogen rate of corn: A comparison. Agr. J. 86:191-195.

4. Campbell, C. A., Selles, F., Zentner, R. P., and McConkey, B. G. 1993. Available water and nitrogen effects on yield components and grain nitrogen of zero-till spring wheat. Agr. J. 85:114-120.

5. Carmer, S. G., Nyquist, W. E., and Walker, W. M. 1989. Least significant differences for combined analyses of experiments with two- or three factor treatment designs. Agr. J. 81:665-672.

6. Cerrato, M. E., and Blackmer, A. M. 1990. Comparison of models for describing corn yield in response to nitrogen fertilizer. Agr. J. 82:138-143.

7. Doyle, A. D., and Shapland, R. A. 1991. Effect of split nitrogen applications on the yield and protein content of dryland wheat in northern New South Wales. Aus. J. Exp. Agr. 31:85-92.

8. Fischer, R. A., Santiveri, F., and Vidal, I. R. 2002. Crop rotation, tillage and crop residue management for wheat and maize in the sub-humid tropical highlands. I. Wheat and legume performance. Field Crops Res. 79:107-122.

9. Mahler, R. L., and McDole, R. E. 1987. Effect fo soil pH on crop yield in Northern Idaho. Agr. J. 79:751-755.

10. Newbould, P. 1989. The use of nitrogen fertilizer in agriculture. Where do we go practically and ecologically?. Plant and Soil 115:297-311.

11. Phillips, S. B., Keahey, D. A., Warren, J. G., and Mullins, G. L. 2004. Estimating winter wheat tiller density using spectral reflectance sensors for early-spring, variable-rate nitrogen applications. Agr. J. 96:591-600.

12. Raun, W. R., and Johnhson, G. V. 1995. Soil-plant buffering of inorganic nitrogen in continuous winter wheat. Agr. J. 87:827-834.

13. Raun, W. R., and Johnhson, G. V. 1999. Improving nitrogen use efficiency for cereal production. Agr. J. 91:357-363.

14. Raun, W. R., Solie, J. B., Johnson, G. V, Stone, M. L., Mullen, R. W., Freeman, K. W., Thomason, W. E., and Lukina, E. V. 2002. Improving nitrogen use efficiency in cereal grain production with optical sensing and variable rate application. Agr. J. 94:815-820.

15. Sabata, R. J., and Mason, S. C. 1992. Corn hybrid interactions with soil nitrogen level and water regime. J. Prod. Agric. 5:137-142.

16. SAS Institute Inc., 1991. System for regression, Second edition Cary, NC.

17. Simon, M. R., Perello, A. E., Cordo, C. A., and Struik, P. C. 2002. Influence of Septoria tritici on yield, yield components, and test weight of wheat under two nitrogen fertilization condictions. Crop Sci. 42:1974-1981.

18. Sparrow, P. E. 1979. Nitrogen response curve of spring barley. J. Agr. Sci. Camb. 92:307-317.

19. Weisz, R., Crozier, C. R., and Heiniger, R. W. 2001. Optimizing nitrogen application timing in no-till soft red winter wheat. Agr. J. 93:435-442.

20. Wuest, S. B., and Cassman, K. G. 1992. Fertilizer-nitrogen use efficiency of irrigated wheat: II. Partitioning efficiency of preplant versus late-season application. Agr. J. 84:689-694.