Content uploaded by William Reid
Author content
All content in this area was uploaded by William Reid on Sep 10, 2014
Content may be subject to copyright.
Do Modified Morgan and Mehlich-III P Have a Morgan P Equivalent?
Q.M. Ketterings
1
, B.C. Bellows
2
, K.J. Czymmek
3
, W.S. Reid
1
, and R.F. Wildman
4
Introduction
Cornell University publishes the “Cornell Guide” which includes recommendations
for N, P, K, Ca, and Mg and micronutrients for a large number of field crops in New York.
The recommendations are based on decades of field research in NY showing soil nutrients
extracted by Morgan solution are correlated well with nutrient response for the vast array
of soil types in NY.
Several private soil-testing laboratories that serve NY producers use the Mehlich-
III and/or modified Morgan extraction solution. In the past, Cornell’s fertilizer
recommendation software did not allow for the use of extractants other than Morgan's
solution because prior research conducted in NY demonstrated a poor relationship between
Morgan and Mehlich-III extractable P (Klausner and Reid, 1996). However, comparisons
within similar soil types (Pote et al., 1996), pH and textural class (McIntosh, 1969) or Al
content (Magdoff et al., 1999) suggested it might be possible to derive better conversion
equations (models) by including specific soil (chemical) characteristics in the equations. In
1999, Cornell University faculty and staff, agribusiness and state and federal agencies
joined resources in a statewide study aimed at deriving such conversion equations for NY
soils. In this article, we focus on P conversions. In a future issue of “What’s Cropping
Up?” we will address conversions for K, Ca, and Mg.
Field Sampling and Analyses
Personnel from Agway Inc., Agricultural Consulting Services Inc., ConsulAgr Inc.,
Cooks Consulting Services and the Miner Institute collected 235 soil samples (0-6 or 8
inches) in NY. These samples represented 27 soil types and eight major agricultural soil
groups from across NY (Table 1).
The soil samples were analyzed at Cornell’s Nutrient Analysis Laboratory, A&L
Laboratories Inc., Brookside Laboratories Inc., and Spectrum Analytic Laboratories. A&L
analyzed the samples for Modified Morgan and Mehlich-III P. Brookside determined
Mehlich-III P, K, Mg, Ca and Al while Spectrum generated pH and Mehlich-III P data. At
Cornell, soils were analyzed for pH and Morgan extractable P, K, Ca, Mg, and Al.
In early 2000, Agricultural Consulting Services Inc. added the modified Morgan P
extraction to its standard soil-sampling package. This generated a dataset of a 10,331
samples taken throughout NY with soil test P (STP) ranging from 1 to 559 ppm P
(modified Morgan). This dataset, referred to as the ACS 2000/2001 dataset, was used to
study the implications of using modified Morgan and/or Mehlich-III soil tests and a
conversion equation on P fertilizer recommendations generated with Cornell nutrient
management software.
1
Department of Crop and Soil Sciences, Cornell University.
2
Department of Agricultural and Biological Engineering, Cornell University.
3
Pro-Dairy, Cornell University.
4
Agricultural Consulting Services Inc.
Results
5
The original 235 soil sample dataset covered an extensive range of soil chemical
characteristics: 17-593 ppm Mehlich-III extractable P, 1-97 ppm Morgan STP, 380-1576
ppm Mehlich-III Al, 473-6025 ppm Mehlich-III Ca and 4.5-7.7 pH. Comparisons between
Morgan and modified Morgan P analyses provided a close relationship: Morgan P (ppm) =
0.90*modified Morgan P (ppm). Mehlich-III P results from Brookside and A&L were
virtually identical. Spectrum consistently measured a slightly higher (6%) P level.
Regression analyses between modified Morgan or Morgan and Mehlich-III
extractable P (averages of Brookside and A&L) showed results similar to those reported by
Klausner and Reid (1996); a very poor correlation (identified by a low r
2
value) existed
when analyses were compared across all soil types and chemical characteristics. However,
including pH, extractable Al, and Ca in the analysis resulted in greatly improved
predictions:
Morgan STP =
1.617 + 0.5574*M3P – 0.001809*M3Ca – 12.97*pH +
0.05799*M3Al – 0.00002743*M3Al
2
+ 1.2794*pH
2
+
0.00004445*M3P*M3Ca - 0.0009237*M3P*M3Al +
0.00000038*M3P*M3Al
2
(r
2
=0.88) [Model 1]
In this equation all data are in ppm. Morgan STP is Morgan extractable soil test P, M3P is
Mehlich-III extractable P, M3Al is Mehlich-III extractable Al, M3Ca is Mehlich-III
extractable Ca, and pH is the soil pH in water (1:1). An r
2
value of 1 indicates a perfect
correlation (and thus prediction). For field data, an r
2
of 0.75 or higher is generally
considered good.
Because most soil testing laboratories presently do not include Al in their standard
packages, we developed a second equation without Al (all data in ppm):
Morgan STP =
-55.53 + 1.366*M3P – 0.001284*M3Ca + 21.78*pH +
0.00005626*M3P*M3Ca – 0.5244*M3P*pH – 2.028*pH
2
+
0.0490*M3P*pH
2
(r
2
=0.82) [Model 2]
Figure 1 shows measured versus predicted values for both models. Model [1] predicted
86% within 5 ppm (10 lbs/acre) of the measured value. The predictions for model [2] (i.e.
no Al included) were slightly less accurate: 79% of the samples were predicted with a
maximum deviation of 5 ppm (Figure 2). Deviations between measured and predicted
values did not correlate with measured STP (i.e. deviations occurred throughout the range
of measured soil test values).
5
All equations assume soil test values in ppm. To convert lbs/acre to ppm, divide by 2. To convert ppm to
lbs/acre, multiply by 2.
Implications for Recommendations
Although a deviation of 10 lbs P/acre (5 ppm) in soil test P may seem large, such a
deviation will not necessarily result in different P fertilizer recommendations. The “Cornell
Guide” recommends a P application of 20 (± 5) lbs P
2
O
5
/acre for corn grown on soils
testing high for available P (9-39 lbs P/acre Morgan soil test P). No P addition is
recommended for optimal economic yield when the STP is very high (≥40 lbs P/acre or 20
ppm P) while for soils with Morgan P levels less than 9 lbs P/acre, the recommendation is
(65-[5*STP]) ± 25%. In this calculation, STP is Morgan soil test P in lbs/acre.
Recommendations are given as ranges because the relationship between soil test results
and yield response is not perfect. The goal is to ensure that the true value for P application
falls within the ± 25% range 90-95% of the time.
We used the ACS 2000/2001 dataset to see how often we derive P
recommendations for corn (using pH, Mehlich-III P and Ca and model [2]) that are not
within the acceptable range. This comparison showed that for almost 60% of the 10,331
samples, recommendations based on predicted Morgan and those based on measured
Morgan soil test P were identical. An additional 30% of the predictions generated 5-10 lbs
P
2
O
5
/acre difference in recommendation and a total of 8% deviated 15-20 lbs P
2
O
5
. Almost
95% of the time, recommendations derived using pH, Mehlich-III P and Ca data fell within
the +/- 25% range for recommendations. A slight improvement can be expected if
Mehlich-III Al is included and model [1] is used for the conversion.
Conclusions
Cornell’s fertilizer recommendations are based on soil tests obtained using the
Morgan extraction solution. Thus, the most accurate recommendations are obtained using
the Morgan solution for soil testing. However, the results of this study have shown that
recommendations can be derived with modified Morgan as well as with Mehlich-III P
input data if the soil pH and Mehlich-III Ca are known. The predictions can be improved
by using an equation that includes Mehlich-III Al.
Conversions from other extractants (e.g. P Bray, Olsen) to Morgan P values may or
may not correlate as well as the Mehlich-III to Morgan conversions in this study. Separate
studies are needed to address conversions for other extractants. Separate studies are also
needed if laboratory procedures are changed.
The P conversion models will be programmed into Cropware (Cornell’s nutrient
management software) that will be released in May 2001 and used to determine the NY P
index for fields that have Mehlich-III soil test data. In a future article in “What’s Cropping
Up?” we will discuss Morgan equivalents for Mehlich-III K, Ca, and Mg.
Acknowledgments
We owe thanks to Francoise Vermeylen for her help with the statistical analyses,
Ray Bryant for his assistance in determining the sampling matrix, and Stu Klausner for his
review of an earlier draft of this article. Thanks to Scott Anderson (Spectrum Analytic
Laboratories Inc.), Paul Chu (A&L Laboratories Inc.) and Mark Flock (Brookside
Laboratories Inc.) for collaborating on this project and donating services. We thank Agway
Inc., ConsulAgr Inc., Cooks Consulting Services and the Miner Institute for their
involvement in field sampling and the Cornell Nutrient Analyses Laboratory staff for their
help in processing the samples. This project was funded by a grant from the Natural
Resources Conservation Service, and NY State’s Departments of Agriculture & Markets
and Environmental Conservation.
References
1. Pote, D.H., T.C. Daniel, D.J. Nichols, A.N. Sharpley, P.A. Moore, Jr., D.M. Miller,
and D.R. Edwards. 1996. Relationships between phosphorus levels in three Ultisols
and phosphorus concentrations in runoff. Journal of Environmental Quality 28:170-
175.
2. Klausner, S. and W.S. Reid. 1996. Comparing soil test results between laboratories.
In: Cornell Cooperative Extension. What’s Cropping Up? 6(3):2-4.
3. Magdoff, F.R., C. Hryshko, W.E. Jokela, R.P. Durieux, and Y. Bu. 1999. Comparison
of phosphorus soil test extractants for plant availability and environmental assessment.
Soil Science Society of America Journal 63:999-1006.
4. McIntosh, J.L. 1969. Bray and Morgan soil test extractants modified for testing acid
soils from different parent materials. Agronomy Journal 61:259-265.
Figures Not Available
Figure 1: Measured versus predicted Morgan extractable P for 235 New York soils.
Predicted values were obtained using a model that included Mehlich-III P, Ca, Al and pH
as inputs (model [1]) and a model that included Mehlich-III P, Ca and pH only (model [2]).
See text for the models.
Figure 2: Cumulative percentage of samples as a function of the difference in predicted and
measured Morgan soil test P for 235 samples from New York.
Table 1: A total of 235 soils from 8 major agricultural areas in New York State were
sampled to derive Morgan to Mehlich-III conversion equations. (#) = number of locations
sampled per soil type. The soil type of four samples remained unidentified.
Northern Tier Till
High lime
Southern Tier and
Catskill Till – Acid
Valley and Lake Plain,
lacustrine/marine
Outwash
A. Well drained A. Well drained A. Well drained A. High pH
Honeoye (21) Lordstown (6) Collamer (7) Arkport (6)
Ontario (21) Mardin (21) Hudson (3) Howard (3)
Madrid (6) Schroon (1) Hamlin (3) Braceville (1)
Hogansburg (6) Bath (7)
B. Poorly drained B. Poorly drained B. Poorly drained B. Low pH
Lima (2) Volusia (30) Rhinebeck (18) Chenango (29)
Appleton (7) Fremont (7) Munuscong (1) Colonie (1)
Ovid (11) Malone (2) Niagara (7)
Angola (3) Madalin (1)