Bianca Moebius-Clune, Harold van Es, and Jeff Melkonian, Department of Crop and Soil Sciences, Cornell University
For over fifteen years, the corn stalk nitrate test (CSNT; Binford et al., 1992; Blackmer & Mallarino, 1996) has been promoted as a tool to determine whether a corn crop received deficient, adequate, or excessive nitrogen (N) amounts during a growing season. In recent years, the test has been strongly promoted as part of the adaptive N management approach, and its adoption has increased geographically beyond Iowa where it was initially developed. Little attention has been given to whether the test is sufficiently precise for field-level N management.
The Test
The basic concept of the CSNT is that the nitrate-N concentration of the lower corn stalk at the end of the season is indicative of whether sufficient N was applied to the corn crop, as plants suffering from N deficiency remove more N from the lower stalk than those with adequate or excess N supply. Universities and grower associations generally suggest the following interpretations of the test:
- Low (less than 250 ppm nitrate-N, in some states 450 or 750): high probability that the crop was N deficient.
- Optimal (generally between 250 and 2000 ppm nitrate-N, in some states also including a “marginal” range when below 750): high probability that yields were not limited by N, and no apparent excess.
- Excessive (>2000 ppm nitrate-N): high probability that N uptake exceeded plant requirement and that N was applied at excessive rates.
The CSNT is promoted as a tool that provides a post–mortem evaluation, but concerns have emerged about its utility to growers. All reported data on the CSNT in journal articles and fact sheets show that yield adequacy is often observed with CSNT values in the “low” range, which raises doubts about whether the test is a powerful indicator of N deficiency. Indeed, a recent Iowa report based on a large data set of N rate trials (Sawyer, 2010) indicated that 15% of CSNT values in the “low” range were false positives, while of cases with field-verified N deficits, 30% of CSNT results were false negatives. A recent Maryland study involving 10 experiments (Forrestal et al., 2012) found about a third of “low” CSNT values to be false positives for deficiencies.
Arguably, the primary value of the CSNT is related to determining excessive N rates, because N deficiencies can also be determined from leaf yellowing during the growing season. Recent New York research reports suggest that fields with high excessive N applications may still show low or optimum CSNT values (What’s Cropping Up?, Vol.21 No.3) and that site differences affect CSNT values more than excess or deficient fertilizer rates (Katsvairo et al., 2003). The above-mentioned Iowa report (Sawyer, 2010) indicated that 33% of cases with field-verified excess N applications were not identified through the test, i.e., one third were false negatives for excess N. Moreover, the Maryland study (Forrestal et al., 2012) found as much as half the CSNT results to be false negatives for excess N. These results challenge the notion that the CSNT is an effective tool for adaptive nitrogen management in corn production.
Methods
As part of an Adapt-N beta testing effort (http://adapt-n.cals.cornell.edu/), we conducted 35 replicated strip trials on commercial and research farms throughout New York (17 trials) and Iowa (18 trials) in 2011. They involved two rates of N (a conventional “Grower” rate and an “Adapt-N” rate), which resulted in field-scale strips with N rate differences ranging from 15 to 140 lbs/ac. Trials had 3 to 8 replications for each treatment (except for 2 of the trials, NY8 and NY9, with only single strip yield measures, but replicated CSNT values). Trials were distributed across both states under a wide range of weather conditions, and involved grain and silage corn, with and without manure application, and rotations of corn after corn, corn after soybean, and corn after a clover cover crop (Table 1). New York yield results were reported in a recent What’s Cropping Up? article (Vol.22 No.2).
To allow for comparison across all trials, silage yield values were converted to grain equivalents (8.14 bu grain per ton silage, calculated by using a harvest index of 0.55). The yield results from a majority of the trials showed unambiguous over-fertilization associated with the higher N rate (same yields for both rates). In these cases, the amount of “effective excess N applied” was set to the N rate difference between treatments (Table 1). In some cases the low rate provided insufficient N (reduced yields), and the optimum N level appeared to be between the high and low rates. In these cases, the amount of excess N applied was estimated by subtracting a conservative 1.25 lb N from the N rate difference between the treatments per bushel of yield lost due to the lower rate.
Fifteen corn stalk sections, sampled from each replicate strip, were dried, ground, and analyzed for nitrate content, according to published protocols. Means for each treatment are presented in Table 1. The utility of the CSNT was then assessed by evaluating the relationship between N rates, test values, and yield losses, and determining whether it accurately diagnosed field-demonstrated deficient or excessive N levels.
Results
Figure 1 shows the relationship between yield loss and CSNT results (the critical 250 and 2000 ppm levels are indicated on the graph). This pattern is similar to those in the original publications, but our data also indicate that:
- While yield losses are strongly associated with low CSNT values, the reverse does not hold: Low stalk nitrate levels do not necessarily imply yield losses.
- Adequate N rates (no yield losses) can result in a wide range of CSNT values. i.e. the power of the test to detect adequate or excess N rates is limited because low CSNT values may be observed when yield losses did not occur.
- Conversely, high CSNT values correctly imply a high probability of excess N rates.
Figure 1. Yield losses (bu/ac) and CSNT values (ppm) from the lower N rate treatments in all trials, and for the higher N rate treatments in those trials where excess was unambiguous (implying no yield gain with further added N).
In most trials, but not all, CSNT values for the upper N rates were higher than for the lower ones, indicating that the test shows some sensitivity to N levels (Table 1). However, in only 8 out of 35 trials (6 of them from Iowa) did the CSNT for the upper rate fall into a higher category than the CSNT for the lower rate.
N Deficient Cases: An evaluation of the power of the CSNT to detect N deficiencies is presented in Table 1 and Figure 2. Of all CSNT values in the “low” range (25 instances), 60% were measured when N rates were in fact known to be adequate or even excessive (i.e., more than half were false positives of deficiency; Fig 2a). For only six of these trials, yield reductions were statistically significant and the CSNT correctly supported the results (highlighted in green in Table 1). Of the 11 trials where significant yield losses were measured with the lower N rate (and deficiencies occurred), the CSNT identified six (54%) correctly in the “low” range (Fig 2b), while CSNT results for the remaining 46% of trials were false negatives for deficiency.
N Excess Cases: Instances with excessive CSNT values (>2000 ppm) were in fact known to have excess N or there was no evidence to the contrary (Table 1, Fig 2c). Therefore, the test was 100% accurate when showing excessive CSNT values, similar to Sawyer’s (2010) results. However, the opposite was not the case. We found that in only 11 of 35 cases (31%, 24% of those in NY) where unambiguous surplus N applications occurred, the CSNT correctly identified excess N levels (CSNT>2000 ppm, Fig. 2d). Conversely, for 24 of these 35 cases (69% overall; 76% for those in NY) the CSNT erroneously diagnosed non-excessive levels (i.e. more than two-thirds were false negatives). Many of these could be considered serious misdiagnoses (highlighted in red in Table 1, excess of 30 lb N or more). This includes Trial NY4 where at least 140 lbs N/ac excess were applied, Trial NY18 with an excess of at least 106 lb N/ac, and Trial NY27 and NY28 where at least 75 lbs N/ac excess were applied. In the latter case, the CSNT values suggested deficiency when in fact N was applied in considerable excess.
Conclusions
We used 35 strip trials to make an assessment of the utility of the CSNT for corn nitrogen management on a field-by-field basis. We conclude from this year’s data and other published work that the test has limited ability to support management decisions. The primary question is whether the test can effectively detect excessive N applications. The answer appears to be “no.” Over two-thirds of the cases with substantially over-fertilized crops (up to 140 lbs/ac) did not show CSNT values in the excessive range (>2000 ppm), i.e. a majority of those cases were false negatives. Since the test’s primary need is related to determining excessive N rates, it appears to perform weakly in serving its main purpose. A second issue is whether the CSNT precisely determines N deficiencies. In this case the problem is with high rates of false positives, i.e., low CSNT values while N rates were in fact adequate or even excessive.
An additional concern is that end-of-season evaluations of the current growing season have limited value for the predictability of N needs in future growing seasons. Research has demonstrated (summarized by van Es et al., 2007) that weather conditions during the early growing season greatly affect N losses and are a critical factor in determining optimum N rates. This implies that an interpretation of CSNT values requires an evaluation of the complex growing conditions of the past season, and that test results from one growing season have limited value for predicting N needs for the next year when the weather may be very different.
Overall, we conclude from previous research reports and our own 35 strip trials that the CSNT is not an effective tool for use in field-specific adaptive N management, especially in the Northeast. We suggest that users of the test recognize its inherent weaknesses, and we recommend caution with the adoption of the CSNT for field-level adaptive N management.
Acknowledgements
This work was supported by grants from the New York Farm Viability Institute and the USDA-NRCS Conservation Innovation Grants Program. We are grateful for the cooperation in field activities from Bob Schindelbeck, Keith Severson, Kevin Ganoe, Sandra Menasha, and Anita Deming of Cornell Cooperative Extension, from Dave DeGolyer, Dave Shearing and other staff at the Western NY Crop Management Association, from Eric Bever and Mike Contessa at Champlain Valley Agronomics, and from Shannon Gomes, Hal Tucker, Michael McNeil and Frank Moore of MGT Envirotec in Iowa. We also are thankful for the cooperation of the many farmers who implemented these trials.
References
Binford, G.D., A.M. Blackmer, and B.G. Meese. 1992. Optimal concentrations of nitrate in cornstalks at maturity. Agron. J. 84:881–887.
Blackmer, A.M. and A.P. Mallarino. 1996. Cornstalk Testing to Evaluate Nitrogen Management (PM-1584). Iowa State Univ. Extension. Available on the Web at: http://www.extension.iastate.edu/Publications/PM1584.pdf. [URL verified 2/14/12].
Forrestal, P.J., R.J. Kratochvil, and J.J. Meisinger. 2012. Late-Season Corn Measurements to Assess Soil Residual Nitrate and Nitrogen Management. Agron. J. 104:148–157 (2012)
Katsvairo, T.W., W. J. Cox, and H. M. van Es. 2003. Spatial Growth and Nitrogen Uptake Variability of Corn at Two Nitrogen Levels. Agron. J. 95:1000–1011
Sawyer, John. 2010. Corn Stalk Nitrate Interpretation. Integrated Crop Management News. Iowa State University Extension and Outreach. http://www.extension.iastate.edu/CropNews/2010/0914sawyer.htm [URL verified 3/11/12].
van Es, H.M., B.D. Kay, J.J. Melkonian, and J.M. Sogbedji. 2007. Nitrogen Management Under Maize in Humid Regions: Case for a Dynamic Approach. In: T. Bruulsema (ed.) Managing Crop Nutrition for Weather. Intern. Plant Nutrition Institute Publ. pp. 6-13. http://adapt-n.cals.cornell.edu/pubs/pdfs/vanEs_2007_Managing N for Weather_Ch2.pdf . [URL verified 2/22/12].
