Yields and Profits of a Low-External-Input Cropping System Compared to a Moderately-Low-External-Input Cropping System

Jon Baldock [1] and Josh Posner [2]

24 Apr 2001

INTRODUCTION

External crop inputs (such as fertilizer and pesticides) are a key factor in profitability as well as a major factor in the pollution potential of a cropping system. One of the original justifications for the WICST was to furnish reliable data for the debate between the high-external-input advocates (nozzle heads) and the low-external-input advocates (granola heads). However, there are many more options for input levels than those two extremes and the intermediate alternatives are often more viable and thus more interesting. Much of the discussion about the intermediate options centers on how to set the level of inputs low enough to keep input costs down and avoid pollution problems without severely reducing crop yield and income.

For example, the WICST data taken on the main plots in the first ten years could be interpreted as showing that the inputs in the low external input systems are too low. That is, corn in the moderate input level system, CS2, out-yielded corn in the low input system CS3 in 14 of 16 site-years by an average of 23 bu/acre. Also, the high input forage system, CS4, produced more corn than the relatively low external input forage system, CS5, in 12 of 14 site-years by an average of 17 bu/acre. Thus, even if corn prices are low, say $2.00/bushel, the low-external-input systems grossed $34 to $46/acre less than the higher-external-input systems.

On the other hand, the WISCT data from the primary experiment also show the merits of minimizing external inputs. That is, the low-external-input systems (CS3 and CS5) reduced costs sufficiently that their gross margins were almost as good as the higher yielding systems. They also had lower soil nitrate levels in the fall and use less pesticide.

Ideally, we would like to avoid the lower productivity of the low-external-input systems, but we would like to retain their low cost and low environmental risk features. This ideal begs the question "Is there a set of slightly greater external inputs than used in CS3 and CS5 that would capture the yield advantages of CS2 and CS4 while maintaining the low expense and low pollution hazard benefits of CS3 and CS5?" To help answer this question the WISCT has compared a moderately low-external-input approach (dubbed “ChemLite”) based on the Corn-Soybean-Wheat(red clover) cropping system of CS3 to CS3 itself at the Arlington site.

Possible Inputs for ChemLite

Opinions vary on what inputs are most important to convert a very low-input system such as CS3 to a ChemLite system. However, for corn there are three commonly recommended inputs. One, add sufficient starter fertilizer to help P and K uptake in cold, wet springs (usually 10 to 20 lb/a of N, P2O5, and K2O). Two, fertilize with 40 to 60 lbs N/a to fill in most of the difference between the legume N credit for the red clover cover crop (60 to 120 lbs N/a) and the Best Management Practice recommendation of 160 lbs N/acre on the Arlington soils. And three, apply grass and broadleaf herbicides at less than the full-labeled rate. There are several options for using reduced herbicide rates. The herbicide could be applied either as a preemergence application in a band over the row that covered approximately 50% of the area at the full labeled rate or at a reduced rate. Another alternative is to broadcast the herbicide at a reduced rate. With the addition of one or more herbicides, the number of mechanical weed control passes are generally reduced to one cultivation plus one rotary hoeing if conditions require it. Usually the combination of herbicide and mechanical weeding provide more consistent weed control than either alone. For example, using a preemergent herbicide controls weeds when it is too wet to rotary hoe. However, sometimes it is too dry for herbicides to work, but those are the best conditions for rotary hoeing.

For soybeans, only one additional input is needed to convert CS3 to a ChemLite System and that is a reduced-rate-herbicide program. The same herbicide application options discussed above for corn apply to soybeans. For wheat, there are four possible additional inputs to transform CS3 into a ChemLite System: 1) a broadleaf herbicide to control winter annual weeds, 2) a fungicide, 3) some starter fertilizer applied at planting, and 4) a small amount of N applied in the spring to carry the crop needs until the soil is warm enough to release N from the previous soybean crop. The first two options have the highest success rate and could be based on scouting or other integrated pest management techniques.

MATERIALS AND METHODS

Cunningham et al. (1992) and Posner, Casler, and Baldock (1995) describe the design and conduct of the main WISCT cropping systems, so those details will not be repeated here. The ChemLite versus CS3 comparison was started at Arlington in 1995 on the east end of Blocks 1 and 2. The Corn-Soybean-Wheat(red clover cover crop) sequence used from Cropping System 3 (CS3) was used in the ChemLite System so the effect of slightly increased input levels could be compared to CS3 itself. The actual inputs used in the comparison of the ChemLite System to CS3 are shown in Table 1.

The data from 1995 through 1999 were available for this analysis. Thus, the data set included 5 years × 2 cropping systems × 2 replications × 3 crops for a total of 60 cases. In 1995 the ChemLite soybean and wheat data were apparently recorded as a mean over the two replications instead of as two separate plots. Also, one replication of CS3 soybeans was missing in 1999. These three missing values caused some complications in the analyses, but are unlikely to have caused any changes in the results.

Although the ChemLite crops were not randomly interspersed with the CS3 crops, the plots used for ChemLite were adjacent to the main trial and deemed homogeneous enough to apply statistical tests. I used only Blocks 1 and 2 from the main trial in the analyses because those were the blocks that were adjacent to the ChemLite plots and that achieved the balance needed for the statistical tests (2 reps for each system). The model equation shown below provides the basis for the analysis of variance.

Z_jkl=m+Y_j+B_k+YB_jk+T_l+YT_jl+e_jkl

where

• Z_jkl = observed yield for the jkl th case,
• m = overall mean yield
• Y_j = year,
• B_k = block k
• YB_jk = interaction of year j with block k
• T_l = cropping system,
• YT_jl = interaction of year j with cropping system l
• e_jkl = residual error j = 1, ..., y k = 1, ..., b l = 1, ..., c.

Table MM1 outlines the analysis of variance that corresponds to this model. Because there were too few sites and years to have a good sampling of environments, we regarded years as a fixed factor, not a random factor, which made the bottom line the appropriate error for testing the significance of both the Year × Cropping System interaction and Croppping System (see the Expected Mean Squares in Table MM1). We calculated the analyses of variance in JMP 4.0 (Lehman, 2000). Because there was some nonhomogeneity, we also analyzed these data with the sign test using STATISTIX7 (Analytical Software, 2000).

Table MM1. Outline of the analysis of variance.

RESULTS AND DISCUSSION

Crop Yields

Table 2 shows the analysis of variance for the crop yields in the two systems. Observe that there are a number of significant effects in spite of the low number of degrees of freedom for error. Table 3 presents the yearly mean crop yields and the mean crop yields averaged over years for both systems. This table shows that the ChemLite system out yielded CS3 in winter wheat grain every year and for the other commodities in four of five years.

The largest difference in mean yields and the highest level of significant differences occurred in corn. These results are consistent with one of the primary reasons the ChemLite system was studied, that is, low CS3 corn yields in some years. Eighty percent of the 19-bu/acre advantage for the ChemLite system occurred in 1996, which was a cool, wet spring. Even though these conditions only happened once in the ChemLite comparison, they should not be regarded a rare event because similar conditions with similarly low CS3 corn yields occurred in 1992 and 1993 in the main WICST trial. It is not clear which input was most responsible for the improved yield in ChemLite or if more than one of the three added inputs contributed. In years other than 1996, the CS3 corn yields were close to, or even slightly above those in the ChemLite system. As a result the Year × System interaction was highly significant. In some cases, a significant interaction means that the difference between the systems was so variable over years that a recommendation cannot be made. However, in this case, the results favor the ChemLite system four out of five years with a large advantage in one of five years. Thus based on corn yields, the ChemLite system is clearly recommended over the CS3 system.

The ChemLite system also produced an average of 4 bu/acre more soybeans than the CS3 system. Because herbicide use was the main additional input, it probably was the main cause of the difference, but residual effects of the starter fertilizer applied to corn may have also contributed to the increased yield. The Year × System interaction was not significant, which was due at least in part to the consistent advantage of the ChemLite system (the low number of degrees of freedom for error was also likely a contributing factor). Thus, although the advantage is not as dramatic as for corn, the ChemLite system is recommended over the CS3 system based on soybean yields.

The ChemLite wheat grain yields were greater than those for CS3 in all 5 years, but the difference between systems was not quite significant (Table 2). That is not too surprising because the ChemLite and CS3 inputs levels for wheat were virtually the same (Table 1) and there were only 4 df for error (Table 2). However, the small advantage for ChemLite in straw yield was significant (Table 2). Because the wheat was managed the same, the ChemLite advantage in straw yield (Table 3) was likely due to residual effects of fertilizer or weed control from the other phases. Thus, the differences in wheat yields would probably have been greater and significant for both grain and straw, if small amounts of those inputs had been applied directly to the wheat.

Yield variability and the associated risk can be an important factor apart from mean yields. In this time period, which was characterized by generally good growing conditions and high yield in four of the five years, the ChemLite system had less variability in corn yields than the Low External Input System. The range in ChemLite corn yields was 53 bu/acre (standard deviation of 21.0 bu/acre), but for CS3 corn yields it was 119 bu/acre (standard deviation of 41.8 bu/acre). This difference is significant at the 5% level. The small number of years is probably not a factor that exaggerates the variability of CS3 because the one poor yield in five years for the CS3 in the ChemLite study is actually a lower percentage than CS3 experienced in the main WICST study (three poor years in eight).

The differences in variability were small for soybeans and wheat. The range in ChemLite soybean yields was 22.2 bu/acre versus 20.6 for CS3 (standard deviations 8.59 v. 11.0). The range in ChemLite wheat grain yields was 29.7 bu/acre compared to 21.1 for CS3 (standard deviations of 9.36 v. 8.15). The range in ChemLite wheat straw yields was 1.02 t/a compared to 0.93 for CS3 (standard deviations of 0.43 v. 0.37).

If we view this study as a comparison of two systems using the same three crops over five years, then there are fifteen chances for CS3 to out yield ChemLite and vice versa. This formulation allows use of the sign test, which has the advantage of not depending on assuming the errors are from the normal distribution or homogeneous. The sign test indicates the ChemLite yields were greater than the CS3 yields in 13 of the 15 chances if the grain yields were used for wheat, which was significant at the 1% level. If the straw yields were used for wheat, then ChemLite produced more than CS3 in 12 of the 15 crop-years, which was significant at the 2% level.

Thus, even though the ChemLite inputs were not applied every year or to every crop, the ChemLite systems out yielded CS3 for corn, soybeans, and wheat straw. The yield advantage for ChemLite would likely have been greater, if small amounts of fertilizer and herbicides had been used more consistently across all three crops. Also, the differences in yields in favor of ChemLite may increase as the soil P and K levels in CS3 decrease. That is because the CS3 soil P and K levels began at a very high level and although they are decreasing (Alt et al., 1998 Appendix III-A), they are still above the optimum for these crops (Kelling et al., 1998).

Economic Analysis

The ChemLite system out yielded CS3 for corn, soybeans, and wheat straw. Therefore, there is a set of modest inputs that can produce higher yields, which answers one of the questions that this study tried to answer. However, the questions remains “Are those yields high enough to pay for the inputs, so that profitability is improved?”

We used a modified version of the Crop Rotation Options Program (CROP; Baldock et al., 1998) to analyze the latter question. The main modification consisted of bypassing CROP’s formula for cropping district average yields and agronomic factors, so that the actual yields obtained in this study could be used. The economic analysis compared two 1000-acre cash crop operations: one running ChemLite, the other CS3. The machinery compliments for the two farms are shown in Table 4. Note that the only difference in equipment is that the ChemLite farm has a sprayer. The major input prices and commodity prices used in the comparison are shown in Table 5. The crop prices were set at the loan rate because harvest-time market prices have been below the loan rate for several years, but most farmers could easily obtain the loan price using the government loan program itself or the loan deficiency payment option. The N price is typical of the price that existed during the conduct of the trial. However, N prices have increased sharply in the last several months, so a sensitivity analysis for N prices will be discussed below. Similarly, some of the WICST team members believe the CS3 is an organic system and that the commodity prices should reflect an organic crop premium, so a sensitivity analysis for those premiums will also be discussed. Finally, the cost of fertilizers, herbicides, and mechanic weed control for the two systems are shown in Table 6. Unfortunately, the inputs were not as consistent as desired, so in many cases the average or the most common level was used.

Table 7 summarizes the income, expenses and net return for the two systems. With the standard price scenario the ChemLite Farm realized a pretax profit of $34092 and the CS3 Farm realized $20043. That is a 70% improvement in net return for a modest set of inputs. In other words, the ChemLite Farm’s income increased $3.10 for every dollar invested in additional inputs. If the herbicides and fertilizer had really been applied more consistently in the ChemLite system and the mechanical weed control passes reduced as discussed in the introduction, then ChemLite may have performed even better.

Recently, some people have been forecasting the demise of any system that applies N from external sources because fertilizer N prices have increased by 25 to 200%. We ran a sensitivity analysis to examine how the price of N affects the net return on the ChemLite Farm (the CS3 farm does not use any N, so the price of N has no bearing on its net return). Obviously, as the N price increases, the net return decreases, assuming everything else stays the same (Table 8). However, it would take a drastic increase in N price to reduce the ChemLite net return to the level of CS3. In fact, the N price would have to increase nearly 6-fold, to near $1.38/lb N, before the ChemLite income would be as low as CS3.

What if the CS3 Farm was able to obtain an organic premium for its grain? To answer that question, we examined how large such a premium would have to be across all three grains before the CS3 net return was as large as for ChemLite (Table 9). The answer is that the CS3 Farm would have to get a little better than a 6% premium to be as profitable as ChemLite. That is, instead of $1.80/bu, $2.47/bu, and $5.18/bu for corn, wheat, and soybeans; the CS3 farm would have to get more than $1.91/bu, $2.62/bu, and $5.49/bu. At these low price levels those premiums equate to 11¢/bu on corn, 15¢/bu on wheat, and 31¢/bu on soybeans. An exercise for another time would be to determine how much premium the CS3 Farm would need on one or two crops to be as profitable as the ChemLite Farm.

CONCLUSIONS

The ChemLite system out yielded CS3 for corn, soybeans, and wheat straw. There was no significant difference between the systems for wheat grain yield, but ChemLite had higher grain yields than CS3 in all five years. Therefore, there is a set of modest inputs that can produce higher crop yields. Were the ChemLite yields sufficiently more than the CS3 yields to pay for the additional inputs? An economic analysis showed that they were. In fact, the net return to labor, management, and capital for a 1000 acre ChemLite Farm was over $14,000 more than for the same size CS3 Farm ($34100/year versus $20000/year). In other words, income was increased $3.10 for every dollar invested in additional inputs. Sensitivity analyses showed that fertilizer N prices would have to increase nearly 6-fold to reduce the ChemLite return to the level of the CS3 Farm. A similar analysis for the effect of organic premiums on CS3 demonstrated that the premium would have to be over 6% for all three grains in order for the CS3 Farm net return to reach the level of the ChemLite Farm.

Table 1: Additional inputs for ChemLite system

* The rates shown are the amount per total acre. The applications were made in a 15" band on 30" rows, thus the rate in the band itself was twice that shown.

Table 2. Analysis of variance for crop yields.

† Soybean MSError has 3 df, Wheat grain and straw MSErrors have 4 df.
**,*,# Significant at the 1%, 5%, and 10% level respectively.
ns Not significant at the 10% level.

Table 3: Mean crop yields †

†n=2 for year means, unless noted otherwise
‡n=1

Table 4. Equipment compliment for the 1000-acre cash grain farms.

Table 5. Input and commodity prices used in the economic analysis.

Table 6. Cost of fertilizer, herbicides, and mechanical weed control for the two farms.

† $2/acre per hoeing (includes only variable costs)
‡ $5/acre per cultivation (includes only variable costs)

Table 7. Net return to labor, management, and capital on a 1000-acre farm.

Table 8. Effect of increasing price of N on net return for ChemLite farm.

Table 9. Effect of organic premiums on net return for CS3 farm.

LITERATURE CITED

Alt, Scott, et al.. 1997, 1998. The Wisconsin Integrated Cropping Systems Trial. Seventh Report. Agronomy Dept. University of Wis-Madison. Madison, WI

Analytical Software. 2000. Statistix 7: Users Manual. Analytical Software. Tallahassee. FL

Baldock, J.O., J.L. Posner, and D.R. Fisher. 1998. Users Manual: Crop Rotation Options Program. Ver 1.0. Agronomy Dept., University of Wis-Madison. Madison, WI

Cunningham, Lee, et al.. 1992, 1993. The Wisconsin Integrated Cropping Systems Trial. First, Second Report. Agronomy Dept., University of Wis-Madison. Madison, WI

Kelling, K.A., L.G. Bundy, S.M. Combs, and J.B. Peters. 1998. Soil test recommendations for field, vegetable, and fruit crops. Coop Ext Bul. A2809. University of Wis-Extension. Madison, WI.

Lehman, A., J. Sall, B. Jones, and E. Vang. 2000. JMP Version 4. SAS Institute. Cary, NC.

Posner, J.L., M.D. Casler, and J.O. Baldock. 1995. Wisconsin integrated cropping system trial: combining agro-ecology with production agronomy. J. Alternative Agric. 10:98-107.

Acknowledgements

This paper would not have been possible without Janet Hedtcke’s assistance with the yield and input tables. And, there would have been no data at all without the efforts of the WICST team, Janet, and the crew at the Arlington Research Station.

1. Email: agstat@aol.com
2. Email:jlposner@facstaff.wisc.edu

Home | Who we are | Research | Publications | Links | Site map | Contact us

University of Wisconsin-Madison
College of Agricultural and Life Sciences
If you have any difficulty accessing this page, please contact our web master.

Site hosted by UW-Madison Center for Integrated Agricultural Systems