These data show what happens to potassium (K) in the soil when different rates of K fertilizer are applied. Over two years, I made 25 applications of K to a plot of L-93 creeping bentgrass in Ithaca, New York. In 2002, I made 13 applications, and in 2003, I made 12 applications. K was applied at 6 different rates, and N was supplied in equal amounts to each plot.
This chart shows the starting soil test K, before any of the 25 treatments had been applied, and also the final soil test, two years after the first one, and after all those 25 fertilizer applications had been made. I'm showing data here from the 0.01 M SrCl2 soil test (that is "hundredth molar strontium chloride") because that test has high accuracy and sensitivity in sand rootzones. These data are proportional to Mehlich 3 data, but lower in this sand by about 50 ppm. So 30 ppm by 0.01 M SrCl2 would be about 80 ppm by Mehlich 3.
Before any of the treatments were applied, the soil test K was about 29 ppm. When no K was applied, the soil test K went down. When more K than N was applied, the soil test K went up.
This next chart shows those same data, with the 25 application dates when K was applied marked in green.
I'd like to point out that on the final date of sampling shown here -- 30 May 2004 -- it was 7 months after the final K fertilizer application of 2003. And you'll notice that there is a big difference in soil test K, with less than 20 ppm in the plots to which no K fertilizer was applied, and more than 50 ppm in the plots to which 4.6 grams of K were applied for every gram of N applied.
What about precipitation? Shouldn't heavy precipitation cause the K to leach? That's not the way it works. From the first soil test date of 4 June 2002, when the soil test K was 29 ppm, to the last date, there were 20 days during which the precipitation was greater than 25 mm. This chart adds on those dates, marked as blue asterisks. The asterisks are jittered up and down, to avoid overplotting.
Each of those 20 days had more than 25 mm of precipitation, for a grand total of 719 mm (28.3 inches) just on those 20 heavy precipitation days. There were 4 such days between the last K application and the soil testing on 30 May 2004. But the amount of K in the soil looks like it was controlled by the quantity of K fertilizer applied, not by the amount of precipitation.
I wrote about this in "I'd be applying potassium all the time" parts 1, 2, and 3. Adding K based on rainfall is a sure way to apply way more potassium than the grass can use or the soil can hold. For that matter, so is adding more K than N.
What is even more important than all the soil test numbers is the performance of the grass. And all the K added in this experiment, all 25 applications of K at different rates over 2 years, didn't cause any improvement in turf performance. Here, in the flagged rectangle, are those L-93 plots to which the K treatments were applied. This photo was taken on 19 August 2003.
At the soil test levels of K in this experiment, there was enough K to meet all the grass requirements, across the range of adding no K for every 1 gram of N (a 1:0 ratio of N:K) all the way to the highest rate of 4.6 grams of K for every 1 gram of N (a 1:4.6 ratio of N:K). All the more reason not to worry about replenishing soil K after a rain.
What one should do is look at the soil test K, make sure it will stay above the MLSN guideline for K, and then don't worry about K.