Rate and Timing of P and K Fertilization in Corn-Soybean Rotations in Minnesota

data summary

introduction

objectives

materials and methods

results and discussion

2011 Data

Summary statistics are given in Table 3 for the 2011 yield data. Phosphorus application increased corn yield at Morris and soybean yield at Lamberton (P data summaries are given in Figure 1). At Morris, maximum yield was achieved with 60 lbs P2O5. Yields were slightly depressed at this site due to excessive rainfall and wet soils more most of the early season. At Lamberton, the soybean yields increased in a linear fashion and did not appear to be maximized by the highest P rate. The actual yield difference between the high and low P rates was only 3 bu/ac. There did appear to be a linear increasing yield trend at St. Charles. However, within site variability prevented this trend from achieving significance.

For the potassium trials, corn yield was increased at Morris while soybean yields were increased at Delavan (Figure 2). At both Morris and Delavan the 0 lb K2O plots could be easily distinguished late in the season. The area between the rows could still be seen late into the season. However, this only affected yields at the one location. It took 80 lbs K2O to maximize corn yields at Morris and 90 lbs to maximize soybean yields at Delavan. We have not fully conducted an economic analysis on this so it is unclear if the yield increases would cover the cost of the fertilizer itself. It is likely that it will since the yield increases were significant enough to give an adequate return.

Early plant growth was measured but the data are not available at this time. Plant samples have been sent out to the lab for analysis but the data has not come back yet. In addition, we did collect grain samples to measure total P and K uptake. Those samples have just come back from the lab but have not been fully analyzed.

A summary of change in soil P and K over the first two years of the study are given in Figures 3 and 4. The two year application rates affected post-harvest soil test at the Lamberton and Saint Charles P locations and the Lamberton and Morris K locations. It is unclear why we did not see any clear trends for the Morris P and Delavan K sites other than the simple effect of soil test variability across the locations. Listed in Table 4 is the average amount of fertilizer P that was needed to raise each P soil test by 1 ppm. At the Lamberton site it took 13 lbs P2O5 to increase the Bray-P1 test by 1 ppm and 21 lbs for the Olsen test. The difference between the two tests is reflective of the fact that the two tests do not measure the same pools of P in the soil and that the scale for the Olsen test is much smaller. It took more P to raise the soil test at Saint Charles. However, this site was limed between the first and second years of the study so that likely had a large impact on the soil P test at this site. For K (Table 5), it took 15 lbs K2O to increase the K test by 1 ppm at Lamberton and 8 lbs at Morris. These numbers were fairly consistent between locations. We could not draw any clear data from the Delavan site. That site had the largest change in soil test K from the initial to sampling at the first cropping season. This site has been managed with no-tillage so there may be some stratification of K occurring in the upper soil surface.

2012 Data

Yield data significance for both P and K studies in 2012 is given in Table 6. For the P studies, yield was only significantly increased at the Morris soybean location. Yield increased up to about 120 lbs P2O5 and there was no difference between application timing either when fertilizer was applied before corn, before soybeans, or split applied both years (Figure 5). Yield did trend higher for the Lamberton and Morris corn locations but the effect of rate failed to reach significance. At this time we have not looked at outliers in the data thus plot variability may be preventing any significance at these locations. A yield response would be expected at Morris since soil tests have significantly trended downward in the four years of this study. Overall, soil test P was maintained with a 30 lb annual application or 60 lbs applied over a two year period. Even though starter fertilizer increased early plant growth there was no effect on yield.

For the K studies, yield was increased at the Delavan corn location and Morris soybean location. At Delavan, yield was increased up to the 60 lb rate (Figure 4). Again, time of application in the corn soybean rotation did not significantly affect yield. At Morris, soybean yield was increased up to 180 lbs K2O with no difference between times of application. There was no clear decrease in soil test K at any location (Figure 6). What is interesting is that the effects on yield seem to be significant for only one block out of the two at each location. If you compared data from Table 1 there were significant treatment effects in the same block but with other crops (for example the block with corn at Lamberton in 2012 would have been the block with soybean in 2011). This indicates some significant variability in the areas being studied at each location and will likely necessitate separate analysis over years when this is conducted.

A summary of change in soil P and K over the first two years of the study are given in Figures 5 and 6. The two year application rates affected post-harvest soil test at all P and K locations. Listed in Table 7 is the average amount of fertilizer P that was needed to raise each P soil test by 1 ppm. At the Lamberton site it took 6.4 lbs P2O5 to increase the Bray-P1 test by 1 ppm, 7.8 lbs at Morris, and 17.8 lbs at Saint Charles for the Bray-P1 test. For the Olsen test, it took 10.9 lbs at Lamberton, 11.3 lbs at Morris, and 27.5 lbs at Saint Charles. The difference between the two tests is reflective of the fact that the two tests do not measure the same pools of P in the soil and that the scale for the Olsen test is much smaller. It took more P to raise the soil test at Saint Charles. However, this site was limed between the first and second years of the study so that likely had a large impact on the soil P test at this site. For K (Table 8), it took 2.7 lbs K2O to increase the K test by 1 ppm at Delavan, 6.4 lbs at Lamberton, and 5.4 lbs at Morris. These numbers were fairly consistent between locations. Previously we had not seen any soil test increase after 2-years at the Morris P and Delavan K sites. However, after four years it is clear the treatments are starting to significantly impact P and K soil test values.

2013 Data

Table 9 summarizes the statistical significance of yield data collected in 2013 from both the P and K locations. Overall yield levels were low at some locations, especially Morris where lack of rainfall significantly reduced yields and potentially yield response. There was a clear response to P for corn at the Morris location while the corn K plots were not significant at the accepted level. However, examination of the data showed a sizeable difference between the no K treatments and the lowest applied rate (30 lbs K2O) per acre. A regression was run on the data and the relationship between yield and K rate was found to be significant. Thus this site would likely be considered responsive. The only other site that was responsive was the K soybean but examination of the yield data in Figure 10 shows very little yield response (~ 2 bu/ac). There was a significant corn grain yield response to K at Delavan but similar to Morris the low K rates supplied enough needed K for the crop. The greatest overall fertilizer requirement occurred for the corn P site at Morris where yield was increased to 143 lbs P2O5 per acre. One thing to remember is that fertilizer is applied and reapplied to the same plots on a yearly or bi-yearly basis thus fertilizer has been applied at least 2 to 3 times on some of the plots. Thus, one factor we are interested in studying is the long term feasibility of the rates as well as the timings within the rotation. For the first five years of the study timing has had relatively little importance in the corn-soybean rotation. Fertilizer rate has been the important factor which indicates that both crops would respond equally as well to a two-year applied ahead of either crop. We are starting to better analyze some of the timing data, in particular the starter fertilizer treatments, but at this time there only has been one site year out of 15 where starter increased yield seemingly further than the broadcast alone. We have not found there to be any advantage to the starter for corn plus broadcast for soybean treatments. Broadcast fertilizer appears to be adequate to supply the needs of the crop long term as long as the correct rate is applied.

2014 Data

Table 10 summarizes the statistical significance of yield data collected in 2014 from both the P and the K study. Fertilizer rate affected yield at all corn P locations and one soybean P location (Morris). Fertilizer K rate affected yield only for corn at Delavan and Morris. The greater impact of P is not surprising due to P soil tests gradually lowering more rapidly than K which has shown a greater degree of variation over the years the project has been in place at the three locations. Timing did not have a great impact on corn or soybean yield. The only site where timing significantly affected corn yield during 2014 was at Lamberton. There was no impact of fertilizer application timing in the soybean plots in 2014 and there was no instance where rate or timing of fertilizer application was significant and there was an indication that a significant interaction occurred. The sites were a significant interaction was shown, such as the K soybean site at Lamberton, could not be clearly identified as to the cause of the interaction. There were likely small differences due to timing that were indicated among some of the P or K rates. However, there was no general trend where timing of application within the two year rotation always resulted in differences in yield across P application rates. As with previous years, the main consideration for fertilization in a two year rotation should be the amount of fertilizer applied and it does not matter if the fertilizer is applied before the corn or soybean crop or as a split application to both crops.

2015 Data

Table 11 summarizes the statistical significance of yield data collected in 2015 from both the P and the K study. Fertilizer rate affected corn yield at Lamberton and Morris and Soybean yield at St Charles. Fertilizer K rate affected corn yield and Lamberton and soybean yield at Lamberton and Morris. What is interesting is that there was less response overall in 2015 than was seen in 2014. We theorized that the continued removal of P and K would increase a sites’ responsiveness to the nutrients over time. The greater than average yield at the locations in 2015 should have put increased pressure on soil P and K concentration. However, we are still encountering unpredictability in some of the locations response to P or K. Some of the unpredictability could be due to soil tests maintaining around the medium classification over the seven years of the study. The only exception to this maintenance has been the Morris P site where the soil P tests have declined to low for many of the low rate plots and response to P has been consistent across years.

There was very little interaction between P or K rate and time of application. For P the interaction with rate and timing was only significant for soybean grain yield at Lamberton. The reason for the interaction was not clear as soybean grain yield at Lamberton generally increased with rate of P applied. The only discrepancy was that soybean grain yield for the 0 P control for the timing before corn yielded greater than the 0 P control for the before soybean treatment. Data were checked for outliers but none could be clearly identified. The yield for the 0 P treatments for the before corn treatments was 7 bu/ac greater and was not different than plots that received at least 20 lbs P2O5.

The rate by timing interaction was significant for K for both corn and soybean yield at Lamberton. For corn yield at Lamberton, there was a gradual increase in yield with applied K rate for the plots where K was applied before the soybean crop and there was more plot to plot variation where K was applied before the corn or where K was split applied. The variability is insignificant and has no bearing on wheat would be recommended for rate of application at Lamberton for corn. For soybean at Lamberton the effect of the interaction on soybean yield was clear. There was no impact on soybean grain yield for rates of K applied before corn. Soybean yield was decreased by rates of K applied directly before the soybean crop and for K that was split applied. A decrease in soybean yield due to application of KCl is not unheard of and has happened previously and the KCl is 50% K (not K2O) and 50% Cl. There have been reports of Cl toxicity in soybean thus an application of enough KCl could induce a toxicity. I took around 200 lbs of KCl to decrease soybean grain yield at Lamberton. If high rates of KCl are applied they should be applied before corn to reduce any negative impact of the fertilizer source on soybean grain yield.

A summary of the change in soil test P and K is given in Tables 12 and 13. Seven cropping years have been completed as of the end of 2015. Previously, the incline/decline rates of P and K were calculated based on the amount of fertilizer applied in a 2-year rotation. Data was recalculated following the 2015 crop using the total amount of P or K applied over the seven crop years. It took between 14 to 25 lbs P2O5 to increase soil test P by 1 ppm according to the Bray P1 test and 25 to 50 lb P2O5 to change the Olsen test by 1 ppm over the seven years of the study. The 14 lb P2O5 value fits with typical rules of thumb for increasing soil test P in the northern Corn Belt. For K the amount needed to increase the soil test ranted from 5 to 8 lb K2O. A typical rule of thumb is 7 lb K2O.

2016 Data

Table 14 summarizes the statistical significance of yield data collected in 2016 from both the P and the K study. Phosphorus fertilizer rate affected corn yield at St Charles and Soybean yield at Morris. Fertilizer K rate affected did not affect corn grain yield at the accepted probability level (P<0.10) but soybean grain yield was affected at Lamberton and Morris. There was less response in 2016 than 2015. It was anticipated that as time progressed more of the sites would respond to both P and K. There were a few instances, such as both the corn and soybean crops at Delavan, where a response model could be fit to the yield data by using the 2-year application rates (data are shown in Figure 26). The large number of treatments included in the study makes it difficult to fully assess the impact of P or K rate. In addition, the design used in unbalanced which also results in significant interaction effects that are due to a lack of response within the split application treatments. Since there is no 0 control for the split there tends to be little difference among the yield levels produced for the various rates. Interactions shown to be significant in Table 14 are a result of no response to P within the split application rate treatments and not a difference in response based on timing. There has been, and still is no indication that the corn and soybean crop are impacted by the timing of fertilizer application. The primary consideration should the selecting the appropriate application rate and the timing of application should not be considered for P. For K there may be some benefit to application before the corn crop which will be discussed below.

Soybean yield again decreased at 1 location (Morris) with increasing K rate. A similar response occurred in 2015 at Lamberton. The exact cause of the decrease is unclear and there was no indication of a interaction similar to which occurred at Lamberton where the decrease was greatest when K was applied directly before the soybean crop. The impact of Cl on the crop should be studied to determine if there is an issue and if application prior to corn can mitigate potential yield reductions in soybean.

A summary of the change in soil test P and K is given in Tables 15 and 16. Eight cropping years have been completed as of the end of 2016. t took between 12 to 33 lbs P2O5 to increase soil test P by 1 ppm according to the Bray P1 test and 20 to 50 lb P2O5 to change the Olsen test by 1 ppm over the seven years of the study. The wide range in values are consistent with previously conducted research in Minnesota. Soil test K was increased 1 ppm by 6 to 10 lbs K2O per acre.

Summary of the amount of P and K required to maximize corn and soybean grain yield

The amount of fertilizer required for a two year rotation to maximize yield within a single year is summarized in Tables 17 and 18. Since there were two block on each site in order to have both crops each year, each block was studied separately in order to determine if blocks at each location differed in their responsiveness to P or K. The starting soil test for each block is given in Tables 14 and 15. Average soil test values were somewhat similar but did differ between blocks. For instance, the P site a Morris tested 6 ppm for the North block while the south block averaged 10 ppm. The difference between the soil test values for each block was evident when studying responses at the North block responded to P application in five of the six years studied. For K, the North block at Delavan five out of six years. The area with the second highest rate of response was the East block at Morris which responded three out of six years.

The amount of fertilizer needed to maximize agronomic yield is far different that the amount required achieving the greatest economic benefit. Figures in the appendix show the relative response of yield at the given locations to fertilizer rate. Soil test should decline in some of the low rate treatments over time to which all sites should eventually respond to P or K. What is surprising is the lack of response at both the P and K sites at Lamberton and P site near Saint Charles. Soil test P was High at the start of the Lamberton study so the soils still could be supplying adequate amounts of P at this site. At Saint Charles soils tested low but there has been no indication of a yield response at this site. We have not studied any soil test levels below 6” at any site. High subsoil P levels could explain some of the lack of response at the locations. We are planning on taking deeper samples in upcoming years to determine the effect of subsoil fertility. For the K sites, the soil test can vary considerably depending on the time of year. Thus, a true measure of availability can be difficult to achieve. We are running different soil analyses for K but the data are not yet analyzed. It should be noted that samples were collected in the fall. Thus, any potassium still within any crop stover may not have been accounted for in the soil test. We are not currently studying the amount of K in the stover. However, other research has shown that stover has a great capacity to take up K in amounts over what the crop needs. This make getting an accurate soil test for K difficult unless samples are collected always at consistent time frames within the year.

A full economic analysis was conducted for each block within each P or K site (tables 19 and 20, respectively) over the eight years of data collected through the end of the 2016 cropping season. The eight years included four years of corn and four years of soybean where model was fit to the average response based on the 2-year application rate of P or K. Timing was not considered due to the effect of timing not being significant for any locations when analyzing the 8-year data. The economic optimum rates of P or K application are summarized assuming a fixed price ration of 0.10 according a cost of $0.40 per pound of P2O5 or K2O and a corn value of $4 per bushel. The price ratio was not calculated for soybean. The EOPR and EOKR values in Tables 19 and 20 represent the 2-year application rate that produced the maximum net return to P or K based on the data in Figures 29 and 30 in the Appendix.

Application of P over the eight cropping years did provide a positive economic response at Lamberton and Morris and never produced an economic response at Saint Charles in spite of low P soil tests. In general, there was little yearly response to P at Saint Charles until the 7th and 8th crop year where only a small rate, 20 lbs P2O5, was enough to achieve maximum yield (Table 17). At Lamberton the maximum economic rate of P ranged from 30-40 lbs P2O5 which the rate was greater at Morris ranting from 50-90 lbs. There has been a clearer response to P for Morris in Block 2. Block 1 has responded but there is little difference in the net return to P for rates ranging from 30-100 lbs P2O5.

Net return per lb of P applied has typically been greater for P sites than for K sites. However, the analysis by block did indicate that some K is warranted for most sites (except for Lamberton Block 1). The relatively higher return at response sites to P is likely a result of soil P tests which were in the low classification for sites while K tests were medium for all sites. At Morris and Lamberton a rate of K from 20-30 lbs provided the maximum net return using the 0.10 price ratio. The response was general greater at Delavan ranging from 30-50 lbs K2O which is surprising since the soil test was the greatest at trial initiation at Delavan compared to the other locations. The air dried test could be overestimating available K at Delavan. Block 1 at Delavan had a greater response to K than Block 2 which had a higher starting average soil test. More work is being conducted studying the impact of drying on soil samples from these sites. Soil sample data from 2016 has not been analyzed at the time of this report. In addition, the economic analysis for both P and K were conducted over eight years. Funding is being sought to carry each site through year ten before the studies are terminated. If ten years of data can be obtained then the economic analysis will be broken down to early, middle, and the final cropping years to see how the economic optimum P or K rates may change over time. We would expect that the required rate should change based on crop removal of nutrients over the rotation.

Impact of Crop P and K Removal on Soil Test Maintenance

One item of interest in this study was to determine the effect of crop removal of P or K on change in soil test. Many growers are interested in removal based application in order to maintain fertility within their fields. Since soils can vary in their chemical properties there can be some differences in how P or K levels build over time. Within most maintenance systems it is not recommended to use the past year’s yields to determine how much fertilizer to apply. Even though it is seems logical, variation in yield and nutrient uptake in grain create great uncertainty on the exact numbers to use to where over-application of nutrients is highly likely if removal is based on previous yields. Typically long term averages work better in order to determine include and decline of soil test.

The net amount of P or K removed was calculated following seven cropping years at all locations and is summarized in Figures 23 and 24. Table 16 summarizes the annual rates of change based on if exact crop removal is targeted as well as the amount of fertilizer needed on an annual basis to maintain soil test levels. For P, a removal based approach resulted in an increase in soil test P in the top 6” at all sites. The increase was the greatest at Lamberton at 2.4 ppm per year and least at Morris averaging 0.9 ppm per year. In order to maintain soil test 30 lbs P2O5 per acre per year could be removed at Lamberton, followed by 26 lbs at Saint Charles, and 22 lbs at Morris.

Data generated over five cropping seasons is summarized in Figures 11 and 12. In order to maintain soil test 59 lbs P2O5 per acre per year could be removed at Lamberton, followed by 42 lbs at Saint Charles, and 17 lbs at Morris. Comparing the data after seven years shows that more P could be removed up to year five which maintaining soil test P. The decrease in the amount of P that could be removed is most likely a result of soil tests lowering to a concentration where deficiencies are more likely. It is reasonable to assume that high removals of P will eventually result in mining of the soil and an increase in the need for fertilizer P to be applied. Factoring in crop removal, applying the data suggests that applying up to 50% of crop removal could maintain soil test P in the top six inches.

Since we are only measuring the top 6”, and given the fact that plants typically use 30% or less of fertilizer P applied, an increase in surface P could be happening at the expense of subsoil P. This will need to be studied to determine if subsoil levels are dropping in the plots.

Since the K soil test levels changed and were relatively higher than the initial tests, the annual rate to maintain the K soil test is difficult to determine. The average soil test change if the exact removal could be targeted ranged from 7.9 to 12.0 ppm based on the air dried ammonium acetate K test. However, we do not know if the change would be the same had the samples been taken in the spring rather than the fall following the crop being harvested. In addition, the point at which we are measuring change against are the preliminary samples taken in spring of 2009. At this time caution should be observed when trying to utilize this data to determine whether removal rates of K should are warranted.

Comparison of Field Moist versus Air Dried Soil Test Values

Multiple soil samples were collected from selected plots starting in 2013 to gauge the effect of air drying on soil test K values from the three LTK locations. Plots were sampled that did not receive a fresh application of K the fall when the samples were collected. Soil test values for K for the moist test were plotted against values for the air dried samples and we found no relationship between the two tests (data are not shown). This lack of relationship is consistent with past findings. We thus looked at the difference between the two (moist – dry test) versus the soil test level for either test. There again was no clear relationship between the different between the tests and the amount of K extracted by the air dried samples. When compared, there was a positive linear correlation between the difference from the moist to the dry and the amount of K extracted by the moist test across all locations and for both years (Delavan data for 2014 are not yet available). What is interesting is that both tests appear to give identical values near 200 to 250 ppm K in solution. When soils tests less than 200 ppm then the moist test tends to be lower than the dry while the opposite is true at levels higher than 200 ppm. We expected some differences among the sites but the 200 ppm threshold was consistent.

The only major difference among the sites is that the difference between the two tests is greater at Delavan with the moist samples testing nearly 100 ppm lower than the dry test at low K application rates. We have seen consistent yield response to K at Delavan in-spite of the air dry test indicating that there should be little chance for a response. The soils at the Delavan site appear to be such that drying the sample releases a significant amount of K and thus would over-estimate K supply to a crop. For Morris and Lamberton there is less of an impact of drying on the lowest K test plots in the study. It is more likely that the air dry test will underestimate availability at Lamberton and Morris.

The collection of samples for moist K analysis was switched to June in 2015. The change was made to try to obtain a better correlation between soil test and yield response to K from the locations. Figure 22 summarizes data collected from selected plots in June 2015. There was a strong correlation between the moist soil test for K and the difference between the moist and dry tests at Delavan and Morris. There was more variability in the data at Lamberton. However, there was a general relationship between the moist test and difference between the moist and dry test at Lamberton. The difference between the tests was 0 when the moist test was around 200 ppm at Delavan and Lamberton while the moist test always resulted in a higher soil test value at Morris regardless of the soil test which is different than past soil samples collected in the Fall.

conclusions

acknowledgements