Economic analysis of N and P applications under variable rate and whole field strategies in the Bothaville district
Maine, N. & Nell, W.T., Centre for Agricultural Management
Alemu, Z.G., Department of Agricultural Economics
Barker C., Department of Geography
University of the Free State
Bloemfontein, South Africa
Fertilizer costs account for approximately 25% of the total variable costs and about 13% of gross production value in crop production in South Africa, and the cost of fertilizer has increased in recent years, forcing farmers to find means of making the use of fertilizer more efficient. On the other hand, cash crop prices declined substantially. There are many
ways in which fertilizer costs can be reduced without adversely affecting yields. In recent years, the availability of new technologies such as variable rate technologies (VRT) has increased the scope of techniques that can aid in reducing fertilizer costs. Variable rate fertilizer application involves radical changes and/or high investment in the technology as well as additional management capacity. Even though this technology is more expensive, it can be technically efficient. Coupled with this efficiency, the increase in fertilizer costs can make this technique a better choice that can increase profitability. Variable rate application of inputs in South African cash crop production is mainly concerned with fertilizer and lime, and this indicates the importance of these inputs in cash crop production. However, the profitability of VRT has not yet been investigated under South African conditions.
This paper studies the maize yield response to variable rate (site-specific) application of nitrogen (N) and phosphorous (P) in South Africa, from data collected on a 104 hectare experimental field on a farm in Bothaville district. The strip plot design of about 180 strips was used for this on-farm research experiment. This design involved treatments that run in the same direction across the field as planting and harvesting. The objective is to determine the maize crop response functions under different N and P. The methodology involves modelling maize yields response functions for N and P. A spatial quadratic regression model is estimated, whereby yield is estimated as a function of a clay content, soil depth, applied N and P. The results indicate that yield response to N and P vary between variable rate and the whole field application methods, as well as between management zones.
Precision agriculture, precision farming, site-specific nitrogen management, variable rate, fertilizer, yield response, South Africa.
Simpson (1986) reports a substantial annual increase in the use of NPK fertilizers throughout the world, especially nitrogenous fertilizers in areas with more developed agricultural systems. This is the result of the effort to intensify production to achieve maximum profits. As a result of the inability to identify the profit maximising input level and the expected yields, inefficient and uneconomic use of fertilizer takes place. This results in contamination of watercourses by leached nutrients, especially nitrates.
The price control on fertilizer that was lifted in 1984 had serious financial effect on both farmers and the fertilizer industry in South Africa. Prior to 1984, all prices and imports of fertilizer were controlled (Fertilizer Society of South Africa, 1986). Now fertilizer costs account for approximately 25% of the total variable costs in crop production, relative to about 15% during the period of control (Nell, 2004). The cost of fertilizer has increased tremendously in recent years, forcing farmers to find means of making the use of fertilizer more efficient. In recent years, the availability of new technologies such as variable rate technologies (VRT) increases the scope of techniques that can aid in reducing fertilizer costs or making fertilizer use more efficient. Variable rate fertilizer application involves radical changes and/or substantial investment in the technology as well as additional management capacity. Even though this technology is more expensive, it can be technically efficient. Coupled with this efficiency, the increase in fertilizer costs can make this technique a better choice that can increase profitability.
Precision agriculture advice based on the soil mineral nitrate N and P is gradually replacing the standard rates of fertilizer application for individual cropping systems in South Africa. According to Matela (2001), variable rate application of inputs in South African cash crop production is mainly concerned with fertilizer and lime, and this indicates the importance of these inputs in cash crop production. Burt et al. (1993) argue that the change in the form and method of application of fertilizer, especially N, is
mainly in response to changing price per unit of N, rather than considerations of the likely efficiency of use. In contrast, Matela (2001) found that farmers in South Africa adopt variable rate technologies to improve efficiency, which leads to reduced per unit cost of production, and ultimately a probable increase in profit.
Malzer et al. (1999) identify the real challenge to precision agriculture as the determination of the factors or items that influence crop production for a given field, and to determine an appropriate strategy to maximise profitability for the producer. In a study carried out by Matela (2001), farmers cited the potential increase in profitability possible with precision agriculture as one of their main considerations in adopting the technology in South Africa. However, the profitability of the technology has not yet been investigated under South African conditions, and this study is aimed at determining crop response functions under different N and P rates.
Godwin et al. (2002) described precision agriculture (PA) as a name given to a method of crop management that entails management of areas within a crop field that require different levels of inputs. Lowenberg-DeBoer and Boehlje (1996) defined precision agriculture as monitoring and control applied to agriculture, including site-specific application of inputs, timing of operations and monitoring of crops and employees.
Precision agriculture is not a new concept, but a recent interest has been fuelled by advances in computer technologies that allow capture and analysis of spatial variability in fields, as well as in application technologies that allow variable rate application of nutrients (Schnitkey et al. 1996). Modern technology in agriculture is one of the important keys to success. As technology is rapidly evolving, farmers must keep up with the changes that may be of benefit to their farming operations (Roberson, 2000).
The profitability of precision agriculture tools is important for farmers and the agri-business sector. They need to determine if it is the wave of the future or a technological dead-end (Lowenberg-DeBoer and Swinton, 19997). The profitability of precision agriculture is the single mostly asked question about the technology, and this determines whether it will be adopted or not. In view of Schilfgaarde (1999), the expectation is that precision agriculture will increase crop yields, enhance net returns from farming, and at the same time reduce environmental damage. Even though current developments in application technologies allow variable application of all inputs, much of the interest in South Africa has focussed on fertilizer application because of the knowledge available on the fertilizer-soil nutrient-yield relationships and the aggressive marketing of fertilizer companies. The relative importance of fertilizer among other crop production expenses adds to this interest in variable rate fertilizer application (Schnitkey et al., 1996).
Moss and Schmitz (1999) measured the value of variable rate technology (VRT) application of inputs by comparing the gross benefit from the application of inputs in the absence of precision information and technology, with the gross benefits from optimal application of inputs, given precision information and technology. The analysis showed the value of spatially variable field operations as inputs can be used more efficiently. In a research conducted by Godwin et al. (2002) in Eastern and Southern England, seven out of eight treatment zones in 2002 gave positive economic returns to variable rate nitrogen with an average benefit of £22 per hectare.
In contrast to the study of Anselin et al. (2004) on which a comparison of the returns from different N application rates was made, VRT showed modest results. Returns above fertilizer costs varied, with the N fertilizer rate recommended by agronomists being the lowest at $415.35 per hectare, and the profit maximising rate for the whole field the highest at $419.56 per hectare. VRT returns for recommendations by agronomist were higher than the uniform recommended rate but lower than the whole field profit maximising rate at $417.00 per hectare. VRT demonstrated lower returns as a result of the added costs incurred with variable rate application.
In an experiment carried out by Welsh et al. (1999) in southern England where variable rate strategies were tested, it was found that applying more N fertilizer to both the historically high and low yielding sections lead to a significant increase in yield. Reduction in fertilizer rate (N, P, K), particularly in high yielding areas result in a penalty of decreasing yield (Welsh et al., 1999).
From the study of Malzer et al. (1999), it was observed that areas of the field that responded to higher nitrogen and phosphorus fertiliser rates to maximise returns were not always associated with highest marginal returns for that specific area of the field. This would suggest the spatial difference in efficiency with which the nutrients were used by the crop. Water/drainage/compaction difference within the study field together with other non-detrimental field characteristics had an influence on the yield, yield response and the rate of nutrients required to obtain maximum economic returns (Malzer et al, 1999).
Lowenberg-DeBoer and Swinton (1997) summarised the results of 17 field crop precision agriculture. In overall, five studies found precision agriculture not to be profitable, six had mixed or inconclusive results, and six showed potential profitability. Lowenberg-DeBoer and Swinton (1997) concluded that because yield and input use changes will vary from farm to farm, it is difficult to make a general statement about the profitability or unprofitably of precision agriculture. The profitability of any given precision agriculture technology and the factors involved may be site-specific and what works in one area may not necessarily work in the other. As a result precision agriculture should be evaluated on a farm by farm basis (Lowenberg-DeBoer and Swinton, 1997 and Malzer et al., 1997). It is also important to evaluate the profitability of precision agriculture, mainly variable rate technology in South Africa, which is a semi-arid country, and very much different from the USA where most studies of the profitability of precision agriculture have occurred. The view of Anselin
et al. (2004) and Lambert et al. (2002) is that the profitability assessment of VRT crucially depends on the model specification used. All spatial models investigated by these authors consistently suggested profitability of variable rate nitrogen application while non-spatial models did not.
Description of the study area
Data was collected from a 104 hectare field of a farm in the Bothaville district of the Free State province. Bothaville is commonly known as the maize capital of South Africa, because approximately 60% of the country’s maize is produced in this region. The soils in Bothaville are generally classified as relatively homogeneous. The spatial variability in the study field is mainly caused by the depth of the soil, as the soil survey revealed a variation of a meter depth within a 100 meter distance. More than 70% of the soils this field are classified as Avalon soil forms. These soils are sandy, with about 85% to 95% sand and have mottles in the sub-soils. The average annual rainfall of this region is 525 mm, which generally occurs during the summer months of November to February.
The experimental design
In the on-going experiment of three years (2001/2 – 2003/4), a one year data of 2002 is used in this paper, with a total of 56 335 observations. Data was collected by a combine harvester equipped with a yield monitor. The strip plot design (illustrated in Figure 1) was used to compare the effect of fertility programmes on yield under whole field (WF) and variable rate technology (VRT) applications. This design involved treatments that run in the same direction across the field as planting and harvesting. The planting was back and forth across the field, resulting in multiple random side-by-side replicates for each pass. The treatments were randomly assigned by each pass across the field, with VRT application followed by a WF application, which was followed by VRT and so forth with the up and down movement of the planter across the field. The 8.5 metre (28 feet) widths of the strips were equal to the width of a planter, fertilizer applicator and a combine harvester. The field was divided equally, with alternating six rows for precision agriculture and six rows for whole field or traditional planting, resulting in 80 replicates of each treatment. Each set of six rows constituted a strip or a plot. Every treatment strip in the field ran across the good, average and poor yield potential zones, i.e., crossed over different potential zones. In the whole field application method, one application was done to different zones as the application rate was held constant in each strip, while the rates changed automatically as different areas warranting different rates were reached when using the variable rate method.
Figure1: Experimental Design: Strip Plot Design
For the variable rate treatment, different N and P rates were determined for different zones and these were used as ranges within which fertilizer application was varied. The management zones were determined by layering yield maps of the past three years and potential zones were established based on yield. Because a mix of 4:2:1 (28) was used and because fertilizer is expensive, limits were set regarding the minimum (70 kg/ha) and the maximum (130 kg/ha) N that could be applied, and applications were varied within this range. Rates for P varied from 10 kg per hectare
Mary http://3dprintshow.com/ Chapstick this improvement dissipate break week viagra for sale or Smells to viagra online Butters firming. That generic viagra don’t will – AMAZING in us discount viagra overnight delivery how works did bristles return http://candidsky.com/ reserved put without cialis mastercard ted scent on bought. Several viagra no rx few – onto my have years.
to 21 kg per hectare. Under the whole field application method, 90 kg per hectare of N and 16 kg per hectare of P were applied.
The National Fertilizer Association of South Africa provides the farmers with prescription rates for different potentials that farmers use as guidelines and these were used as the basis for determining the ranges.
Yield data was recorded on a yield monitor card and saved as a comma-delimited file which could be opened on Excel. The file had different columns indicating latitude, longitude, height, yield, harvest rate and time in seconds. SAS software was used for analysis. The data-set contained a total of 56 335 observations. Regarding input application data for N and P a bit of data manipulation and transformation was necessary. Input application maps available were already interpolated and did not provide raw data that has coordinate points that correspond to yield data. The input maps had only 36 coordinate points relative to approximately 56 335 yield points. In order to obtain the input data points that matched yield points, interpolation of input map data was carried out in ArcGIS, using the inverse distance weight (IDW) because the fertilizer recommendation map developed by the farmer used IDW. In order to increase the explanatory power of the model, other permanent soil properties were included as explanatory variables. These included the clay content in the A horizon and the depth of the A horizon.
Description of the data
Yield was the dependant variable with five explanatory variables. The continuous explanatory variables included the clay percentage in the A-horizon; the depth of the A-horizon; nitrogen and phosphorous, both in kg per hectare. Treatment (TRT), Low, Medium, High and Very high are dummy explanatory variables. TRT is a dummy variable for the treatment, whether variable rate or whole field application, with one for variable rate and zero otherwise. Low, Medium, High and Very High are management zones whereby ‘Low’ is the low potential zone with potential yield of less than 3 ton/ha; ‘Medium’ is a medium potential zone with a potential yield of between 3 and 4 ton/ha; ‘High’ is a high potential zone and the potential yield is between 4 and 5 tons/ha; and ‘Very high’ is very high potential zone of more than 6ton/ha. With regard to proportions of different zones, the medium potential management zone is the largest, constituting about 49% of the total field area. The high potential management zone represents 23%, low potential zone make up 20% and very high potential zone amount to only 8%. The low potential zone was regarded as the base and was therefore not included in the model. For the continuous variables (i.e. N and P), the linear terms were expected to have positive signs and the quadratic terms were expected to have negative signs. This would provide the expected concave production surface.
The model used
A quadratic regression model was used where yield was estimated as a function of applied N and P. Clay, Depth, Treatment (TRT), and Medium, High and Very high management zones were used as other explanatory variables. Interaction terms of continuous variables were also included in the model. This quadratic function was preferred because it fits what is known about maize response to fertilizer. It allows for diminishing returns and a maximum biological yield. The model was specified as follows:
N = nitrogen, kg/ha
P = phosphorous, kg/ha
C = clay, %
D= depth in metres
Mzone = 1 in the medium potential zone, zero otherwise
Hzone= 1 in the high potential zone, zero otherwise and
VHzone = 1 in the very high potential zone, zero otherwise
TRT = 1 for the VRT treatment, zero for uniform application.
Descriptive statistics of variables per management zone under the VRT application strategy are presented in Table 1. The mean, minimum, maximum and the standard deviation are demonstrated.
Table 1: Descriptive Statistics for the VRT
The average yield of the whole field was 4.67 tons per hectare. A very minute difference in average yield was observed between VRT and WF applications where the weighted average yields were respectively 4.65 and 4.67 tons per hectare. However, some differences were observed in average yields between the two application strategies in different management zones. The highest average yields were obtained in the very high zone (VHzone) under the VRT strategy at 5.37 tons/ha compared to 5.24 tons/ha in the WF application strategy for the corresponding management zone. In the case of high potential zone (Hzone), average yields were not much different as VRT had an average of 5.03 ton/ha and average yields for WF were 5.01. Yield difference in the medium potential zone (Mzone) were also not very significant as WF strategy produced average yields of 4.70 and VRT produced 4.69 ton/ha. In the low potential zone (Lzone), the WF out-performed the VRT strategy by yielding 4.02 tons/ha while VRT only produced 3.80 ton/ha.
It is important to note that
only the low potential zone of the WF strategy out-performed the VRT strategy, which is a very small proportion of the total field, representing 20% of the total field area. Yields were comparatively the same for the high and medium potential zones, which together account for about 70% of the field area. Differences in average yields per management zone are consistent with the allocation of management zones as very high management zones provided the highest yields, high zone gave high yield, medium zone medium yields and low potential zone produced the lowest yield. These yield averages, together with the average, minimum, maximum and standard applications of N and P per zone are indicated in Table 1. Descriptive statistics for depth (D) and clay (C) are also shown in Table 1.
Different management zones received the same N and P applications under the WF strategy. On the other hand, different rates were applied to different management zones under the VRT strategy. The very high zone received the highest average rate of 21.12 kg/ha for P and 126.49 kg/ha for N. The high zone got the second highest rate rates which averaged to 20.47 kg/ha P and 123.27 kg/ha N. In the medium zone, the average P applied was 18.28 kg/ha and N was 110.26 kg/ha. The low potential management zone received the lowest inputs at an average of 14.79 kg/ha for P and 95.05 kg/ha for N. This indicates that the selected variable rate strategy entailed applying more inputs in higher yielding areas to push yields higher and applying fewer inputs in low yielding areas so as not to waste resources on marginal or low producing areas. The application rates for the very high, high, medium and low potential zones were targeted at yield potentials of <6 ton/ha, 4-5 ton/ha, 3-4 ton/ha and >3 ton/ha in that order.
The regression output results from SAS are presented in Table 3.
Table 2: Result Output: OLS Model
all the coefficients were statistically significant and had the expected signs with the exception of nitrogen which provided unanticipated signs both in the linear and quadratic forms, even though both were statistically significant. The estimated clay coefficients indicate that with everything being constant, for each percentage increase in clay, there is an increase in yield of about 0.01 tons/ha. The estimated coefficient of depth indicates that yield increases by about 0.007 tons/ha for each cm increase in depth. This therefore shows that the higher the clay content and the deeper the soil to a certain level, the higher the yields. The depth of the soils is related to the ability of the soils to store moisture and the availability thereof to the plants. The high clay content is associated with the ability of the soil to hold soil nutrients.
With regard to the P, both linear and quadratic coefficients provided expected signs and are statistically significant. The quadratic coefficient is consistent with the theory of diminishing returns, indicating the marginal physical productivity of the inputs. The estimated ‘TRT’ coefficient for the variable rate technology was positive and statistically significant, indicating that VRT resulted in a yield increase of about 0.09 tons per hectare, implying that VRT generates higher yields than the WF method of inputs application. Nevertheless, this is a minute increase as the descriptive statistics indicated that there is not much difference in yield between WF and VRT strategies.
Profit Estimation: Enterprise budgeting
Using the regular enterprise budgeting techniques, the weighted gross margin for the WF strategy is higher than the weighted gross margin for the VRT strategy. When analyzing the gross margin per management zone, the very high zone treated with VRT strategy resulted in the highest gross margin of R6 146.40/ha relative to the gross margin of R6 055.55/ha in the analogous zone but treated with WF. The other zones; high, medium and low zones under the WF strategy performed better than their counterparts under the VRT strategy with gross margins of R5 704.13/ha, R5 245.55/ha and R4 225.55/ha in that order in comparison to the equivalent zones under the VRT. Gross margins were R5 645.92/ha, R5 174.14 and R3 885.79 for high, medium and low zones in the VRT strategy. It is important to note that in total, more N and P were applied under VRT than under WF strategies. Since the effects of P are not realized immediately after application, yields are expected to increase in the ensuing years under the VRT, particularly observing that yield difference between the two strategies are currently modest. In the study of Zeilenger (2004) it was observed that significant increase in yield was obtained after the third year of implementing precision agriculture and the same is expected in this study.
It should be borne in mind that these are gross margin calculations that only considered changes in yield and relevant input applications (N and P) for the two strategies and not profit calculations as VRT costs are not taken into account. A partial budgeting technique has to be used to calculate the profit, taking the annuity costs of VRT into account.
The results indicate that in the first year, there is a meagre variation in maize yield response to applied N and P by the application method used, refuting the hypothesis that VRT results in higher average yields than WF application. However, it is expected that yields in zones treated with VRT will increase in future years as the actions of managing by zone take effect. Different results are expected in the second and third years of the study. Preliminary gross margin analysis resulting from the two methods indicates that WF resulted in higher gross margin than VRT. These results are a building block as after having established that there are some differences in yield between VRT and standard application, integrating the economic costs of VRT will provide a more accurate estimation of the profitability of VRT by determining whether the increase in yield compensates additional costs incurred with VRT. It is important to consider spatial auto-correlation in determining the profitability of VRT as this determine the regression model used which affects the ensuing profit analysis. Only the OLS model was used in this study, and a spatial model is required to take spatial auto-correlation into account.
In brief, it has been established that the response to fertilizer depends on soil conditions such as soil depth and clay %, which had a positive effect on yield. Yield response also differed among management zones, with the high potential zone producing the highest yields and the low potential zone the lowest average yield.
Acknowledgements are given to New Holland South Africa for financial assistance in this research
Anselin, L., Bongiovanni, R. & Lowenberg-DeBoer, J. 2004. A spatial econometric approach to the economics of site-specific nitrogen management in corn production. Submitted to the American Journal of Agricultural Economics.
Burt, T. P., Heathwaite, A. L. & Trudgill, S. T. 1993. Overview – the nitrate issue, Chapter 11, pp. 321-339. In Burt, T. P., Heathwaite, A. L. & Trudgill, S. T. (eds.), Nitrate: Processes, patterns and management. England: John Wiley & Sons.
Fertilizer Society of South Africa (FSSA). 1986. Fertilizer handbook, translated by Dickson, E. B., Greenfield, G. & Roos, J. H. Hennopsmeer.
Food and Agricultural Organization. 1998. World reference base for soil resources. World Soil Resources Report 84. FAO, Rome, Italy.
Godwin, R. J., Earl, R., Taylor, J. C., Wood, G. A., Bradley, R. I., Welsh, J. P., Richards, T., Blackmore, B. S., Carver, M. J., Knight, S. & Welti, B. 2001. Precision farming of cereal crops: A five-year experiment to develop management guidelines. Home-Grown Cereals Authority (HGCA). Project Report No. 267.
Lambert, D. M., Lowenberg-DeBoer, J. & Bongiovanni, R. 2002. Statistical methods for precision agricultural data: Case study of variable rate on corn in Argentina. Paper submitted to the Journal of Manufacturing & Service Operations Management (M&SOM). Special issue in Decision Technologies for Managing Agricultural and Food Business, edited by T. Lowe (University of Iowa) and P. Preckel (Purdue University).
Lowenberg-DeBoer, J. & Boehlje, M. 1996. Revolution, evolution or dead-end: Economic perspectives on precision agriculture, pp. 923-942. Proceedings of the Third International Conference on Precision Agriculture. Madison, Wisconson.
Lowenberg-DeBoer, J. & Swinton, S. M. 1997. Economics of site-specific management in agronomic crops, Chapter 16, pp. 369-396.. In Pierce, F. J. & Sadler, E. J. (eds.), The state of site-specific management for agriculture. ASA-CSSA-SSSA, Wisconsin, USA.
Malzer, G. L., Mamo, M., Mulla, D., Bell, J., Graff, T., Strock, J., Porter, P., Robert, P., Eash, N., Braum, S. & Dikici, H. 1999. Economic benefits and risks associated with site-specific nutrient management.
Matela, N. 2001. The status of precision agriculture in cash crop production in South Africa. M.Agric. thesis, Department of Agricultural Economics (Centre for Agricultural Management), University of the Free State, South Africa.
Nell, W. T. 2004. Personal communication. (Lecturer in the Department of Agricultural Economics and agricultural management specialist, University of the Free State, Bloemfontein, South Africa.)
Schilfgaarde, J. V. 1999. Is precision agriculture sustainable? American Journal of Alternative Agriculture, 14(1):43-46.
Schnitkey, G. D., Hopkins, J. W. & Tweeten, L. G. 1996. An economic evaluation of precision fertilizer applications on corn-soybean fields. Proceedings of the Third International Conference on Precision Agriculture, 12-26 June 1996. ASA-CCSA-SSSA.
Simpson, K. 1986. Fertilizers and manures. New York: Longman.
Soil Survey Staff. 1999. Key to soil taxonomy. (8th ed.) Soil Conservation Service, US Department of Agriculture. Blackburg, Virginia: Pocahaves Press Inc.
Welsh, J. P., Wood, G. A., Godwin, R. J., Taylor, J. C., Earl, R., Blackmore, B. S., Spoor, G., Thomas, G. & Carver, M. 1999. Developing strategies for spatially variable nitrogen application. 2nd European Conference on Precision Agriculture, 11-15 July 1999, Odense, Denmark.