Agronomy Journal 93:815-819 (2001)
© 2001 American Society of Agronomy
PRODUCTION PAPER
The Phosphorus Compositional Nutrient Diagnosis Range for Potato
Lotfi Khiaria,
Léon-Étienne Parent*,a and
Nicolas Tremblayb
a Dep. of Soil Sci. and Agri-Food Eng., Laval Univ., Paul-Comtois Building, Sainte-Foy, QC, Canada G1K 7P4
b Hortic. Res. and Dev. Cent., 430 Gouin Blvd., St-Jean-sur-Richelieu, QC, Canada J3B 3E6
* Corresponding author (leon-etienne.parent{at}sga.ulaval.ca)
Received for publication August 17, 1999.
 |
ABSTRACT
|
|---|
Leaf analysis could assist in adjusting the P fertilization of potato (Solanum tuberosum L.) to specific soil–plant systems to achieve high yield conditions. Our objective was to derive and validate Compositional Nutrient Diagnosis (CND) norms and ranges for a potato cultivar (Superior) and to compare CND to Diagnosis and Recommendation Integrated System (DRIS) and the Critical Value Approach (CVA). Survey data were obtained from 563 field observations and validated using 100 independent samples and across four P fertilizer trials. The databases included tuber yield and analyses (N, P, K, Ca, and Mg) of the upper fully expanded leaf collected at beginning of bloom. Yield cutoff between low- and high-yield subpopulations, selected from cumulative variance functions across survey data, was 34.1 Mg ha-1. The sum of squared values of CND indexes was distributed like a chi-square value. The critical chi-square value was 4.2. The critical CND P index range was between -0.80 and 0.80. Similar critical values were obtained for the validation population and fertilizer trials. The CND P index appeared symmetrical about the zero nutrient balance and was more closely related to yield compared with DRIS and CVA. The CND provides an inferential (as chi-square) and symmetrical (about zero balance) diagnosis at low cost, could provide nutrient index ranges adjusted to yield goal, and could thus be developed advantageously for specific soil–plant systems.
Abbreviations: CND, Compositional Nutrient Diagnosis • CVA, Critical Value Approach • DRIS, Diagnosis and Recommendation Integrated System
|
INTRODUCTION
|
|---|
POTATO IS A HIGH P-DEMANDING CROP, leading to P accumulation in soils and, potentially, in both surface and subsurface waters. Tissue analysis could be instrumental for targeting high yield using P rates adapted to local conditions. It would be useful for agronomists to develop nutrient norms at relatively low cost using survey databases representative of highly productive specimens growing in a given agroecosystem.
The Critical Value Approach (CVA) compares analytical results to reference nutrient ranges, disregarding nutrient interactions. However, for a tissue composition closed at 100%, i.e., the dry matter content, a change in any component concentration must affect other nutrient proportions. To avoid distorting the diagnosis, concentration values should be dual-ratioed or multiratioed (Aitchison, 1986).
The dual-ratio Diagnosis and Recommendation Integrated System (DRIS) approach diagnoses the nutrients in their order of yield limitation (Walworth and Sumner, 1987). The DRIS computes dual-ratio functions that are averaged to nutrient indexes. In the Compositional Nutrient Diagnosis (CND) approach, each nutrient is adjusted to the geometric mean of all nutrients and to a filling value, thus producing a nutrient multiratio, or row-centered log ratio (Parent et al., 1994). The CND norms are derived from survey data as mean and standard deviation of row-centered log ratios in a high-performing subpopulation. The squared values of CND nutrient indexes must sum up to a nutrient imbalanced index that is distributed like a chi-square value and could be used to define critical CND nutrient ranges. An additional step using fertilizer trials is required to ascertain the validity of nutrient ranges as row-centered log ratios computed from survey data.
Our objectives were to establish CND nutrient norms and critical values from a survey database for a specific potato cultivar, validate the P critical values using fertilizer trials, and compare CND with CVA and DRIS.
|
MATERIALS AND METHODS
|
|---|
We used three databases (survey, validation, and P fertilizer trial), each comprised of tuber yield and tissue analyses for N, P, K, Ca, and Mg. The survey database originated from 663 commercial potato (Superior) fields sampled in Quebec between 1988 and 1995. A validation database of 100 independent specimens was selected at random, leaving 563 specimens in the survey database. Five P trials were conducted in eastern and central Quebec during the 1993 to 1995 period. The trials comprised four to six triplicated application rates of P. Other nutrients were applied as recommended locally (CPVQ, 1996) and were assumed to be present in sufficient, nonexcessive amounts. Relative tuber yield was computed by dividing yield in the control plot by maximum yield among fertilized treatments and then multiplying by 100.
Twenty upper fully expanded leaves were collected at beginning of bloom, composited in each field or plot, and analyzed for N, P, K, Ca, and Mg. The tissues were oven-dried at 70°C and ground in a Wiley mill. Total N was determined by the micro-Kjeldahl procedure (Bremner and Mulvaney, 1982), and total P, K, Ca, and Mg by plasma emission spectroscopy (Beckman Spectra Span 6, Beckman Instruments, Fullerton, CA) after wet-digesting the tissues in a mixture of HNO3 and perchloric acid (HClO4) (Barnhisel and Bertsch, 1982). Total tuber yield was taken in middle rows of plots and in three to twelve 3-m-long rows in production fields, depending on field dimension and soil variability.
Statistical Analysis
The yield cutoff value used to divide the high- and low-yield groups was based on the cumulative functions, as outlined by Khiari et al. (2001a). Nutrient indexes were computed according to Khiari et al. (2001b) using the step-by-step CND procedure. A fifth step was added here for validating nutrient range with fertilizer trials. Nutrient norms (row-centered log ratios of concentration means and standard deviations) were computed using the Excel package (Microsoft, 1997). The theoretical CND r2 threshold was obtained by assigning the proportion of low-yield specimens at yield cutoff value as an exact probability to a chi-square cumulative probability function (Khiari et al., 2001a). The DRIS computations were performed according to Walworth and Sumner (1987). Critical CND and DRIS indexes were obtained using the Cate–Nelson procedure (Nelson and Anderson, 1977) and validated using relationships between relative tuber yield and nutrient indexes. Minimum critical values corresponded to yield percentage between 90 and 95% (Bates, 1971).
|
RESULTS AND DISCUSSION
|
|---|
Step 1. Selection of a High-Yield Subpopulation from the Survey Database (n = 563)
The cumulative variance ratio function FCi
showed a cubic relationship with tuber yield. Inflection points (-b/3a) were thus used to separate the low- and high-yield subpopulations (Table 1). The highest yield cutoff value, 34.1 Mg ha-1 tuber, was obtained with VN (Table 1). High-yield specimens accounted for 203 of the 563 specimens in the survey population. That yield cutoff value, corresponding to the minimum yield target for separating the low- and high-yield subpopulations, was lower than yield targets proposed by MacKay et al. (1987) and Parent et al. (1994). The nutrient concentrations of the high-yield subpopulation above yield cutoff were used to generate the CND norms.
Step 2. The Compositional Nutrient Diagnosis Norms Derived from the High-Yield Subpopulation
The survey population comprised 360 yield data points below and 203 above cutoff yield value of 34.1 Mg ha-1 tuber. Hence, the proportion of low-yield specimens was 360:563, or 64%. The CND r2 values were distributed as chi-square values with 6 df (r > 0.999, P < 0.001) (Fig. 1). The critical chi-square value was close to 4.2. As expected from DRIS graphs relating nutrient expressions to yield (Walworth and Sumner, 1987), higher yield targets would increase the proportion of low-yield specimens in a population and produce narrower critical nutrient ranges. For example, the proportion of low-yield specimens for a yield target of 40 Mg ha-1 tuber would be 84.2%, and the corresponding chi-square value is close to 2.7 (Fig. 2). For a yield goal of 50 Mg ha-1 tuber, the chi-square value would be 1.3. Hence, as CND r2 gets smaller and yield goal gets higher, critical nutrient index ranges must also become narrower to keep the sum of squared critical nutrient indexes equal to CND r2.
View larger version (18K):
[in this window]
[in a new window]
|
Fig. 1. Comparison between the chi-square cumulative function and the Compositional Nutrient Diagnosis (CND) r2 distribution function to obtain the theoretical threshold CND r2 value (4.2) at yield cutoff in the survey population (n = 563).
|
|
Step 3. Validation of the Threshold Nutrient Imbalance Index
The CND nutrient norms are the means and standard deviations of row-centered log ratios from the high-yield subpopulation (Table 2). The DRIS norms are the mean and coefficient of variation of dual ratios (Table 3). In the validation population, yield decreased with increasing CND r2 or DRIS nutrient imbalance index (Fig. 3). The same pattern was shown by Walworth and Sumner (1987) between the sum of DRIS indexes, irrespective of sign and yield of sugarcane (Saccharum officinarum L.). The yield–CND r2 relationship was iterated following the Cate–Nelson ANOVA procedure for separating the low- and high-yield subpopulations. The critical CND r2 value was found to be 4.3, which corresponded to a yield cutoff value of 34.7 Mg ha-1, similar to the results obtained at Step 2. The critical DRIS nutrient imbalance index corresponded to a yield cutoff of 37.8 Mg ha-1. However, using the Cate–Nelson partitioning procedure, DRIS was less effective (R2 = 0.13) than CND (R2 = 0.34) at separating the high- from the low-yield subpopulations.
View this table:
[in this window]
[in a new window]
|
Table 2. The Compositional Nutrient Diagnosis (CND) norms for the high-yield potato (‘Superior’) subpopulation (>34.1 Mg ha-1), assuming a low-yield subpopulation proportion of 64%.
|
|
View this table:
[in this window]
[in a new window]
|
Table 3. Diagnosis and Recommendation Integrated System (DRIS) norms for the high-yield potato (‘Superior’) subpopulation (>34.1 Mg ha-1).
|
|
View larger version (27K):
[in this window]
[in a new window]
|
Fig. 3. Relationship between the Compositional Nutrient Diagnosis (CND) nutrient imbalance indexes (CND r2) or the Diagnosis and Recommendation Integrated System (DRIS) nutrient imbalance index (DRIS NII) and tuber yield in the validation population (n = 100) as indicated by the variance method of the Cate–Nelson procedure (R2 of the partition between two classes).
|
|
Step 4. Nutrient Sufficiency Ranges
The squared values of CND nutrient indexes (IX) were related to tuber yield for partitioning the population according to the Cate–Nelson procedure. Critical values were found to be 0.67 for CND I2N, 0.64 for CND I2P, 0.45 for CND I2K, 0.85 for CND I2Ca, 0.83 for CND I2Mg, and 0.75 for CND I2R5 (Table 4). The sum of squared values of critical CND indexes was thus 4.2, the theoretical chi-square value or critical CND r2 obtained at Steps 2 and 3 (Table 4), as follows:
 | (1) |
View this table:
[in this window]
[in a new window]
|
Table 4. Critical squared Compositional Nutrient Diagnosis (CND) indexes in the validation samples using Cate–Nelson Partitioning procedure for potato (‘Superior’).
|
|
Indeed, the squared individual nutrient indexes must sum to the critical CND r2 as a control procedure (Khiari et al., 2001a). The critical CND nutrient index ranges were as follows: -0.82 to +0.82 for CND IN, -0.80 to +0.80 for CND Ip, -0.67 to +0.67 for CND IK, -0.92 to +0.92 for CND ICa, -0.91 to +0.91 for CND IMg, and -0.87 to +0.87 for CND IR5. Those sufficiency ranges for P were ascertained at Step 5 using the P fertilizer trials.
The CND nutrient ranges appeared to be crop specific because the potato nutrient index ranges were different than those derived for sweet corn (Khiari et al., 2001b). Adopting a nutrient range diagnostic approach, the more distant from the critical range a nutrient would be, the more limiting the nutrient. This approach contrasts with the DRIS nutrient index interpretation using the relative importance of a given nutrient, i.e., the numerical order of nutrient indexes (Walworth and Sumner, 1987). Our results support the view that nutrient ranges should be defined for CND, as was also suggested for DRIS (Savoy and Robinson, 1990).
Step 5. Validation of Critical Phosphorus Ranges Using Fertilizer Trials
While selecting critical nutrient ranges with fertilizer trials, one assumes that all nutrients are in sufficient amounts except the ones being varied. A close examination of the results of the five P fertilizer trials indicated that one trial should be discarded. Yield potential at most sites appeared to be largely controlled by a limiting nutrient (Mg) not being varied although supplemented at recommended rate. However, crop response to the varying nutrient was still obtained as far as yield potential was not attained. In one P trial, the nutrient imbalance indexes due to large Mg deficiency produced CND r2 values exceeding an average of 12 and no significant crop response to P fertilizers. The CND r2 values at the other four sites did not exceed an average of 11 and produced significant crop response to P fertilizers. Those four sites were thus retained to validate the CND IP range computed at Step 4.
Range of Compositional Nutrient Diagnosis Phosphorus Index
Across the four P fertilizer trials, tuber yield was related to CND IP by a highly significant (P < 0.01) cubic effect (Fig. 4). At 90% maximum yield, the lower limit of the critical CND IP range was found to be –0.84, and the computed upper limit was 0.79. That range derived independently from fertilizer trials was close to the critical CND IP range between -0.80 and 0.80 obtained at Step 4 from population data (Table 4). The -0.8 to 0.8 critical range was about four times the standard deviation of 0.191 (Table 2). The maximum yield was obtained near CND IP = 0, thus validating the symmetrical property of critical CND ranges. The DRIS P index was not symmetrical about zero, as also reported by Sumner (1979). As shown in Fig. 4, the yield–index relationship was closer with CND (R2 = 0.83) compared with DRIS (R2 = 0.72). The P concentration was poorly related to yield (R2 = 0.23).
View larger version (24K):
[in this window]
[in a new window]
|
Fig. 4. Relationship between the Compositional Nutrient Diagnosis (CND) or Diagnosis and Recommendation Integrated System (DRIS) P indexes or P concentration and tuber yield across four P fertilizer trials.
|
|
At two sites where yield potential >40 Mg ha-1 and treatment means >95% of maximum yield (three treatments only), the CND IP were as follows: -0.04 (first site) and -0.64 and +0.59 (second site). Should CND r2 be 2.7 for yield goal of 40 Mg ha-1 (Step 2 and Fig. 2), one would expect the following solution to Eq. [1]:
 | (2) |
and
 | (3) |
As a result, the IP critical range would be between -0.64 and +0.64, as found above. In contrast with DRIS, the additivity of squared CND indexes to CND r2 could thus be a useful concept to adjust nutrient index ranges to yield goal. However, this concept should be further tested by varying the five nutrients in fertilizer trials across a large number of sites showing increasing yield potentials because only P was tested here with a very limited number of results.
|
CONCLUSION
|
|---|
A method is described to generate CND norms from survey databases and validate them with fertilizer trials. A nutrient imbalance index computed as the sum of squared values of CND nutrient indexes was distributed like a chi-square value. Using that generic model, CND norms could be elaborated from limited crop databases. The symmetrical CND IP range computed from a survey database was validated by independent P fertilizer trials. The DRIS critical P index range did not appear symmetrical. The DRIS or CND P indexes were significantly related to yield, and nutrient P concentrations were related to yield to a small extent. The CND proved superior to DRIS for diagnosing P. Conceptually, CND appeared superior to DRIS because the sum of squared values of CND indexes is distributed like a chi-square value. The CND nutrient ranges could thus be derived inferentially from survey data and adjusted to yield goal. Consequently, CND norms and nutrient ranges could be easily updated using small databases for improving fertilizer management in specific potato agroecosystems.
|
ACKNOWLEDGMENTS
|
|---|
We thank the Conseil de Recherches en Pêche et Agroalimentaire du Québec, RBF Technologies, La Coopération Fédérée de Québec, and the Natural Sciences and Engineering Research Council of Canada for financial support and Catherine Tremblay for her kind professional support. The contribution of participating potato producers is gratefully acknowledged.
|
REFERENCES
|
|---|
- Aitchison, J. 1986. Statistical analysis of compositional data. Chapman and Hall, New York, NY.
- Barnhisel, R., and P.M. Bertsch. 1982. Digestion with perchloric-nitric acids. p. 279–280. In A.L. Page (ed.) Methods of soil analysis. Part 2. SSSA Book Ser. 5. SSSA, Madison, WI.
- Bates, T.E. 1971. Factors affecting critical nutrient concentrations in plant and their evaluation: A review. Soil Sci. 112:116–130.
- Bremner, J.M., and C.S. Mulvaney 1982. Nitrogen—total. p. 595–624. In A.L. Page et al. (ed.) Methods of soil analysis. Part 2. SSSA Book Ser. 5. SSSA, Madison, WI.
- [CPVQ] Conseil des Productions Végétales du Québec. 1996. Grilles de référence en fertilisation (in French). 2nd ed. Agdex 540. CPVQ, Canada.
- Khiari, L., L.E. Parent, and N. Tremblay. 2001a. Selecting the high-yield subpopulation for diagnosing nutrient imbalance in crops. Agron. J. 93:802–808 (this issue).[Abstract/Free Full Text]
- Khiari, L., L.E. Parent, and N. Tremblay. 2001b. Critical compositional nutrient indexes for sweet corn at early growth stage. Agron. J. 93:809–814 (this issue).[Abstract/Free Full Text]
- MacKay, D.C., J.M. Carefoot, and T. Entz. 1987. Evaluation of the DRIS procedure for assessing the nutritional status of potato (Solanum tuberosum L.). Commun. Soil Sci. Plant Anal. 18:1331–1353.
- Microsoft. 1997. Microsoft Excel 97. Incline Village, NV.
- Nelson, L.A., and R.L. Anderson. 1977. Partitioning of soil test–crop response probability. p. 19–38. In M. Stelly (ed.) Soil testing: Correlating and interpreting the analytical results. ASA Spec. Publ. 29. ASA, Madison, WI.
- Parent, L.E., A.N. Cambouris, and A. Muhawenimana. 1994. Multivariate diagnosis of nutrient imbalance in potato crops. Soil Sci. Soc. Am. J. 58:1432–1438.[Abstract/Free Full Text]
- Savoy, H.J., and D.L. Robinson. 1990. Norm range size effects in calculating Diagnosis and Recommendation Integrated System indexes. Agron. J. 82:592–596.[Abstract/Free Full Text]
- Sumner, M.E. 1979. Interpretation of foliar analyses for diagnostic purposes. Agron. J. 71:343–348.[Abstract/Free Full Text]
- Walworth, J.L., and M.E. Sumner. 1987. The Diagnosis and Recommendation Integrated System (DRIS). Adv. Soil Sci. 6:149–188.
This article has been cited by other articles:
|
|
|
|
 
M-C. Belanger, A. A. Viau, G. Samson, and M. Chamberland
Determination of a Multivariate Indicator of Nitrogen Imbalance (MINI) in Potato Using Reflectance and Fluorescence Spectroscopy
Agron. J.,
October 19, 2005;
97(6):
1515 - 1523.
[Abstract]
[Full Text]
[PDF]
|
|
|
|
|
|
|
L. Khiari, L.-E. Parent, and N. Tremblay
Critical Compositional Nutrient Indexes for Sweet Corn at Early Growth Stage
Agron. J.,
July 1, 2001;
93(4):
809 - 814.
[Abstract]
[Full Text]
[PDF]
|
|
|
TOP
ABSTRACT
INTRODUCTION
