This work is dedicated to the turfgrass managers of Delaware…you know who you are and I am indebted to you! Thank you!
This was the original manuscript for the USGA Greens Section Record July 2025 article, titled “Indexing the Soil Nutrient Status of Turfgrass Systems to Develop Regional MLSN Guidelines: A Delaware Case Study“
Original title: Indexing the Soil Nutrient Status of Turfgrass Systems in Delaware – An MLSN Comparison
Background/Introduction
Throughout my entire turfgrass management career, until recently, I have always followed the traditional nutrient recommendations. The wonderful collection of superintendents (and academics) that I worked for over the years followed the same nutrient recommendations, and by my standard, they were all successful. By default, that meant that in order for me to be successful I should follow their methods without question. It now appears to me to be a classic case of the Blind Authority Fallacy.
I began working on a golf course in the summer of 2000. Since then, it is nearly impossible to estimate the number of times or the quantity of phosphorus (P), potassium (K), calcium (Ca), and magnesium (Mg) I have applied to turfgrasses. What’s even more interesting is the number of times I have seen turfgrass respond to those applied nutrients. That number is one. One instance where a P deficiency was observed and symptoms were quickly alleviated with a fertilizer application. I have never seen a K, Ca, or Mg deficiency in any turfgrass system except on a research putting green or a greenhouse, which was by design due to years of clipping removal and withholding of fertilizer. To my knowledge, there has never been a documented case of a Ca deficiency except for turfgrass grown in hydroponics. To quickly recap, tons of fertilizer applied over 25 years using traditional recommendations and only a single unique case where my nutrient applications resulted in a positive response to the grass.
On a golf course there is always something to be done, yet there I was religiously applying these nutrients, especially K and Ca, multiple times per year simply because the recommendations said so. Each time I applied these nutrients I felt as if I should be doing something else more useful like calibrating a sprayer, or fixing that leaky irrigation head, or catching up on paperwork. I can vividly remember when these fertilizer applications began to feel unnecessary and a waste of my time because I never saw an improvement in the turf. Once I became a superintendent in charge of my own budget, it felt like I was wasting time and money. As I reflect back, it seems to me that the recommendations I was following made me a less efficient turfgrass manager. I am confident that my personal anecdotes are not unique. Nevertheless, the point of my story is to highlight how traditional nutrient recommendations have been handed down decade after decade and we continue to follow them even when there is little or no evidence to justify their use. Are traditional recommendations wrong? Not necessarily, because applying fertilizer by way of the traditional interpretation methods undoubtedly supplies the grass with ample amounts of nutrients. Are these recommendations broken? Broken may be too strong of a descriptor, but they definitely aren’t very efficient or accurate, and as every turfgrass manager knows, operational efficiency is the key. A great example of this conundrum is when Gelertner et al. (2016) and Shaddox et al. (2023) showed that those who soil test and use traditional recommendations apply more nutrients than those who don’t. The conclusion I draw is that if the recommendations that many turfgrass managers follow result in applying more nutrients than necessary, which is wasteful in my opinion, then it seems obvious to me that we desperately need to rethink how we interpret soil tests and give nutrient recommendations.
The Minimal Levels for Sustainable Nutrition (MLSN) was first introduced to me on a Turfnet webinar in late 2018. My take-home message was that I could apply much less fertilizer without compromise to turfgrass quality (TQ) and that stuck with me. I must admit that I didn’t fully understand how to practically use MLSN at first, but I was intrigued. Later, I found myself constantly thinking about the MLSN idea as it was so different from the way I was taught in my undergraduate studies and how my mentors applied fertilizer. In 2020 I left the golf course management industry and moved into a turfgrass extension role at the University of Delaware. My new position afforded me the opportunity to pursue graduate studies. After diving into the scientific literature on turfgrass nutrient management, it became clear to me the body of work around MLSN needed to be expanded. When it came time to choose a topic for my thesis I knew exactly what I wanted to study and research. I secured funding from the USGA to develop Delaware-specific nutrient guidelines using similar methods from Woods et al. (2016) and compare them to the current MLSN guidelines.
In the turfgrass management industry there are three methods laboratories may use to interpret the results of a soil test. Sufficiency Levels of Available Nutrients (SLAN), Base Cation Saturation Ratios (BCSR), and as of late, MLSN. I will only cover SLAN and MLSN as the BCSR method is wasteful and has no scientific merit to support its use (Culman et. al 2021, Kopittke & Menzies, 2007). Saturated paste extracts (SPE) have had some popularity as well, but this method should only be used for assessing soil salinity and sodium hazard. SPE is not a valid testing method to determine macronutrient needs.
MLSN
First introduced in 2012, MLSN was a joint effort between Larry Stowell and Wendy Gelertner, formerly of PACE Turf, and Micah Woods of the Asian Turfgrass Center (ATC). Both the ATC and PACE Turf had large databases of soil test results from all over the world. To develop the first MLSN guidelines, they combined the two data sets together (n=16,163) then filtered the data by pH (5.5 to 8.5) and cation exchange capacity (CEC) (≤ 6.0 cmol/kg⁻¹). They then graphed the distribution of each element and identified the concentration that corresponded to the 10% probability on the curve. The ensuing values became the MLSN guideline. For an example, let’s look at the MLSN guideline for P, which is 21 parts per million (ppm). 90% of the samples in the data set are producing good performing turf with a concentration of 21 ppm (Mehlich 3) or greater. These guidelines are essentially a concentration of nutrients in the soil that you do not want to go below. Although, there is a 10% buffer built into the guidelines because 10% of the samples in the data set are still producing good performing turf below the MLSN guideline of 21 ppm. The MLSN data set is so large and contains many different species of grass and soil types from around the world that it is logical to expect that MLSN will work for any and all turfgrass systems. While MLSN as an interpretation method is generally accepted as sound, there has been industry and academic push back to the assertion of MLSN working for all turfgrasses on all soils. Soil testing is not, and likely never will be, perfect. It’s not realistic and we must concede to the fact that we cannot perform traditional soil test calibrations for every soil type, every species, muchless every cultivar. Thompson et al. (2023) eloquently stated that “because calibration data does not exist for all turfgrass species on all soils…concessions are inevitable”. Quotes of this nature can be found littered throughout our most popular turfgrass textbooks and the scientific literature over the past 50+ years.
SLAN
The SLAN interpretation method assumes that there is an “ideal” or “sufficient” range for the concentration of each nutrient, whereas MLSN simply says “don’t allow the soil concentration of the nutrient to drop below X”. Soil test reports using SLAN will categorize the concentration of each nutrient. The categories will be some iteration of “high”, “medium”, or “low” and may vary slightly from lab to lab. For example, the University of Delaware Soil Testing Lab (UDSTL) uses the categories “low”, “medium”, “optimum”, or “excessive” where “low” is interpreted as having a high probability of a favorable plant response to applied fertilizer, “medium” has a low to moderate probability of a favorable plant response, “optimum” is unlikely to respond to applied fertilizer, and “excessive” is when the soil nutrient concentration is more than adequate for plant growth and further applications will be uneconomical and may have undesirable effects on plant growth or the environment (Sims & Gartley, 2001). Fertilizer recommendations will decrease as the soil test nutrient concentration approaches the “optimum” category. No fertilizer will typically be recommended if categorized as “optimum” and will never be recommended if levels are considered “excessive”. The basic idea of SLAN is to apply fertilizer until the soil concentration is considered “optimum”. There are three major flaws I see with SLAN. First, the SLAN interpretation method has slowly evolved from the agricultural sector where yield and economics are the driving parameters for determining fertilizer needs. Applying fertilizer to achieve maximum yield is wasteful, potentially detrimental, and mostly irrelevant in turfgrass systems. Second, turfgrass systems are judged on quality and performance. The MLSN data set along with the data presented here are contemporary refutations concerning the relative level of nutrients needed to produce high quality, good performing turf. Third, the soil test calibration data that supposedly supports the SLAN method does not exist or was done on crops other than turfgrasses. Frankly speaking, SLAN grossly overestimates the amount of fertilizer required to grow high quality turfgrass.
Materials and Methods
A soil sampling campaign of the turfgrass systems in Delaware began in May of 2023 and was completed in June of 2024. All soil samples were taken from various turfgrass stands throughout the state of Delaware (Figure 1.). Sites that were sampled included golf courses, athletic fields, lawns, and parks (Figure 10.). All soil cores were taken from a depth of 10.2 cm (4.0 inches) using an industry standard soil probe (Turftec TSS1-S) with an inside diameter of 1.9 cm (0.75 inches). For each defined turfgrass sward (green, fairway, lawn etc.), regardless of the total area, 12 cores were taken at random, mixed in a plastic pail, and composited to form one sample to be submitted to the lab. The entire core was left intact, including the verdure, and combined with the composited sample. The composited sample was labeled and placed in a brown paper bag. All soil samples were left to sit in an open paper bag to air dry until arrival at the soil testing lab.
All soil samples were taken from good performing turf. Good performing turf can be defined here as “a turfgrass sward at the time of sampling which was, by any trained eye, healthy, vigorous, had no visual symptoms of nutrient deficiencies, and would be expected to perform well under its intended use and management regiment”. The majority of samples were given a TQ score at the time of sampling of either an A, B, or C using a modified rating scale similar to what is used by the National Turfgrass Evaluation Program (NTEP) (Figure 12.) The NTEP TQ scale ranges from “1 to 9 where 9 is outstanding or ideal turf and 1 being poorest or dead. A quality rating value of 9 is reserved for a perfect or ideal grass, but it also can reflect an outstanding treatment plot. A rating of 6 or above is generally considered acceptable” (Morris n.d.). TQ scores used in this research can be further clarified using the following scale: A = turfgrass quality cannot be improved and is considered exceptional for the corresponding use and would be equivalent to an NTEP rating of 8 or 9. B = turfgrass quality is very good, but the turf could be improved to an A with minimal effort and would be equivalent to an NTEP rating of 7 or 8. C = turfgrass quality is still good and could be easily improved and would be equivalent to an NTEP score of 6 or 7.
NDVI
A Normalized Difference Vegetation Index (NDVI) reading was taken for many of the samples using a handheld sensor (GreenSeeker Handheld Crop Sensor – Trimble Agriculture). The GreenSeeker provides instant NDVI readings and will display an average score for the time that the device’s trigger was held. The manner in which NDVI readings were taken differed depending on the site, but the height of the gun above the surface for all readings was waist height which is approximately 91.4 cm (3 feet). For all golf course greens and tees, the largest or widest portion of the sample area was identified, and the reading was taken by traversing from one end to the other. For the fairways, the approximate middle point between the two farthest sub-samples was identified and then the measurement was taken by walking from one edge of the fairway to the other. NDVI was measured on football, soccer, and lacrosse fields by walking from one sideline to the other, 10 paces from the centerline. For baseball fields, the measurement was determined by walking foul line to foul line in the outfield and the infield. These two numbers were then averaged. The NDVI of lawns, parks, and golf course roughs were measured by traversing between the approximate locations of the where the 12 random samples were taken. Some samples in the data set do not have NDVI readings. There were three reasons why NDVI readings were omitted for a particular sample. 1.) A sand topdressing had recently been applied, 2.) Time was limited due to golfer, athlete, or landowner use, or 3.) I did not feel particularly welcomed and was rushing to finalize sampling as quickly as possible. NDVI is a way to quantify how green and dense a stand of turf is. The unitless scale ranges from 0-1. A reading of 0.6 or greater is commonly considered to be a healthy stand of turf and often correlates closely with TQ. However, many samples had NDVI readings that fell below 0.6, yet when I rated them visually they were quite good. There were several samples that were rated an “A” on the TQ scale and the NDVI was 0.4 or 0.5 (Figure 8.). The problem with NDVI for this particular study is that green and lush turf is not always ideal for golf and other sports that rely on smooth and predictable ball roll. In this study there was little correlation (R² = 0.01) between NDVI and TQ (Figures 7.).
Sample Analysis
Soil samples were submitted to the UDSTL for analysis. Once at the lab, soil samples were dried in a forced air drying cabinet. Samples were then ground and sieved in preparation for nutrient extraction. Organic matter was analyzed by the loss on ignition method using a muffle furnace temperature of 360℉. The soil pH was determined by using a 1:1 soil to H₂O. The concentrations of P, K, Ca, Mg, and S were measured by inductively coupled atomic plasma emission spectroscopy (ICP-AES) following extraction using the Mehlich 3 method.
Metadata
Other metadata that was collected for each sampling site consisted of the site name, GPS coordinates, physical address, sampling date, site use (green, tee, fairway, rough, lawn, or athletic field), soil root zone type (native soil, sand root zone, or pushup)(Figures 9 & 11.), predominant species(s) of grass, the predominant soil series as identified by the Natural Resources Conservation Service WebSoilSurvey (NRCS, 2024), and if clippings were collected or not (yes or no). At this time, it is not clear how to utilize or analyze much of the metadata. Further work is needed, but some of the graphs will be shown and discussed here.
Statistical Analysis
The methods of statistical analysis used to calculate the Delaware-MLSN (DEMLSN) guidelines are the same as those used in Woods et al. (2016). To briefly summarize, this data set of samples (n = 435) was filtered by pH and CEC. The pH filter includes the samples with a value ranging from 5.5 ≥ 8.0 and the CEC filter includes all samples with a CEC of 6.0 ≤ 6.0 cmol/kg⁻¹ or less. Filtering the data set by pH essentially avoids including samples that could experience aluminum or alkalinity toxicities. As for the CEC filter, the idea is that if we identify the soils that have a low nutrient supplying power while supporting good performing turf, it is logical to expect a soil with a CEC above 6.0 cmol/kg⁻¹ to be equally capable of supplying the grass with sufficient nutrients. Once the data set was filtered by pH and CEC (n = 288), the distribution of the samples for each element was graphed using the Tidyverse package (ggplot) in Rstudio. The value designated as the DEMLSN guideline was the individual nutrient concentration of each element that corresponded to the 10% probability on the distribution curve (Tables 2-6.).
Results/Discussion
A consideration for the practical use of the DEMLSN guidelines is that they are only valid for the state of Delaware. Although, the Delmarva Peninsula soil type is predominantly Atlantic Coastal Plain and these guidelines would likely be valid for use on the entire peninsula. More regional sampling would be required to justify this claim.
The DEMLSN guidelines for P, K, Ca, Mg, S are 20, 22, 235, 34, and 5 ppm, respectively (Table 1.). After hosting a couple seminars on the topic of MLSN and the preliminary results of my work, I have had turfgrass managers approach me and state that “MLSN is too low!”. Yet, if you grow turfgrass in Delaware the data presented here suggests that MLSN is too high. Although the DEMLSN guidelines are less than the MLSN guidelines, they are between 185% to 409% lower than the traditional SLAN recommendations of the UDSTL (Table 1.). The UDSTL uses SLAN guidelines that are fairly consistent with those of other reputable labs around the nation.
The data set contained a fairly even mix of turfgrasses grown in sand root zones and native soil. (Figure 11.). While 236 samples came from putting greens, 199 came from all other site uses (Figure 12.). Putting greens dominated the portion of samples from a sand root zone while the remaining site uses came from a soil with significant clay content (Figure 9.). One aspect of the data set that may be of concern is the disparity in number of samples per site (data not shown). There were some locations that contained 20+ samples and other sites with 2 or 3. There are statistical analysis methods to address sample weighting and it may be dealt with at a later date. I am not sure if it will have a significant impact on the DEMLSN guidelines, but it seems worth exploring.
Another aspect of the data that really jumped out at me was the DEMLSN values for Ca and Mg. These two elements were 71% and 72% respectively less than the MLSN guidelines. It is not clear on why this is the case for the samples, but it is possible that the native soils of Delaware have some amount of Ca and Mg bearing minerals that provide a slow-release source of nutrients.
To me, it is evident that traditional recommendations using the SLAN methods are antiquated and inefficient. MLSN is a great starting point for anyone looking to use contemporary recommendations. DEMLSN is tailor-made for use by the turfgrass managers of Delaware. Also, in order for us to move forward as an industry we need more local/regional soil testing data and it needs to be collected and statistically analyzed in a similar manner. In today’s world, I feel that it is important that the work to improve nutrient recommendations doesn’t stop here. The more robust data set we have, the more accurate and useful our nutrient recommendations become.
Before I sign off, there is an elephant in the room that I want to address. I want to talk about nutrient deficiencies. There has been a lot of fear-mongering in the turfgrass industry on this topic since, well, forever. Let me be crystal clear. You will not wake up tomorrow, arrive at work, and all the turf is dead due to a nutrient deficiency. This is not how it works! Nutrient deficiency symptoms come on slowly over days and weeks. If you see your turf regularly, as golf course managers do, you will have plenty of time to respond before a decline in TQ. But, If you are armed with soil test reports you can make some basic calculations to ensure that you apply enough fertilizer to avoid nutrient deficiencies. This is accomplished by maintaining the soil nutrient concentrations above the DEMLSN or MLSN guidelines. What I just described is, in my opinion, the modern method of turfgrass nutrient management. Applying only what the grass needs, nothing more. Anything more is a waste of your time, your money, your energy, and your mental health. Superintendents and other industry turfgrass managers have enough to worry about and nutrient deficiencies shouldn’t be one of them.
P.S. – I would like to thank the USGA and the turfgrass managers of Delaware. You all know who you are and I am indebted to you.
Figure 1. Map of Delaware sample site
Table 1: Comparison of MLSN, DEMLSN, and UDSTL SLAN guidelines (ppm) (*Difference between UDSTL and DEMLSN)
| Element | MLSN | DEMLSN | UDSTL | *Difference |
|---|---|---|---|---|
| P | 21 | 20 | 50 | 250% |
| K | 37 | 22 | 90 | 409% |
| Ca | 331 | 235 | 500 | 212% |
| Mg | 47 | 34 | 65 | 185% |
| S | 7 | 5 | No Rec’s | N/A |
Figure 2. Distribution of P filtered by pH (5.5-8.0) and CEC (≤6.0 cmol/kg⁻¹)
Figure 3. Distribution of S filtered by pH (5.5-8.0) and CEC (≤6.0 cmol/kg⁻¹)
Figure 4. Distribution of K samples filtered by pH (5.5-8.0) and CEC (≤6.0 cmol/kg⁻¹)
Figure 5. Distribution of Mg filtered by pH (5.5-8.0) and CEC (≤6.0 cmol/kg⁻¹)
Figure 6. Distribution of Ca filtered by pH (5.5-8.0) and CEC (≤6.0 cmol/kg⁻¹)
Figure 7. Relationship between NDVI and turfgrass quality (TQ)
Figure 8. Counts of NDVI readings
Figure 9. Counts of site use by soil type
Figure 10. Counts of samples by site use
Figure 11. Counts of samples by soil type
Figure 12. Counts of TQ ratings
Citations
Culman S.W., Brock C, Doohan D., Jackson-Smith D., Herms C., Chaganti V.N., Kleinhenz M., Sprunger C.D., Spargo J. (2021) Base cation saturation ratios vs. sufficiency level of nutrients: A false dichotomy in practice. Agronomy Journal. 113: 5623–5634. https://doi.org/10.1002/agj2.20787
Gelernter, W.D., Stowell, L.J., Johnson, M.E. and Brown, C.D. (2016), Documenting trends in nutrient use and conservation practices on US golf courses. Crop, Forage & Turfgrass Management, 2: 1-10 cftm2015.0225. https://doi.org/10.2134/cftm2015.0225
Kopittke, P. M., & Menzies, N. W. (2007). A review of the use of the basic cation saturation ratio and the “ideal” soil. Soil Science Society of America Journal, 71(1), 259–265.
Morris, Kevin (n.d.). A Guide to NTEP Turfgrass Ratings.
NRCS, Soil Survey Staff, Natural Resources Conservation Service, United States Department of Agriculture. Web Soil Survey. Available online. Accessed [2024].
Sims, J.T. & Gartley, K.L. (2001). University of Delaware Soil Test Note 1.
Shaddox, T. W., Unruh, J. B., Johnson, M. E., Brown, C. D., & Stacey, G. (2023). Nutrient use and management practices on United States golf courses. HortTechnology, 33(1), 79-97. Retrieved May 22, 2025, https://doi.org/10.21273/HORTTECH05118-22
Thompson, C., Guertal, E., McGroary, P., Soldat, D., Hopkins, B. (2023). Considerations with soil testing in turfgrass. In Fidanza. M. (ed.), Achieving sustainable turfgrass management
Woods, M. S., Stowell, L. J., & Gelernter, W. D. (2016). Minimum soil nutrient guidelines for turfgrass developed from Mehlich 3 soil test results. PeerJ Preprints, 4, e2144v1.
