TITLE:
Foliar Nutrient Balance Standards for Maize (Zea mays L.) at High-Yield Level
AUTHORS:
Viviane Cristina Modesto, Serge-Étienne Parent, William Natale, Léon Etienne Parent
KEYWORDS:
Compositional Data Analysis; Critical Range; Diagnosis and Recommendation Integrated System; DRIS; Ionomics; Nutrient Balance; Nutrient Interactions
JOURNAL NAME:
American Journal of Plant Sciences,
Vol.5 No.4,
February
26,
2014
ABSTRACT: Maize is one of the most nutrient demanding staple
crops. Tissue nutrient diagnosis of maize is currently conducted using critical
nutrient concentration or dual ratio ranges, but such diagnoses are
pathological as biased by data redundancy, sub-compositional incoherence and
non-normal distribution. The use of orthogonal balances, a compositional data
analysis technique, avoids such biases. Our objective was to develop foliar
nutrient balance standards for maize. We collected 758 grain yields (15.5%
moisture content) and foliar samples at silk stage in maize fields of southern Quebec,
Canada, and analyzed ten
nutrients in tissues (N, P, K, Ca, Mg, B, Cu, Zn, Mn, Fe). Nutrients were
arranged into ad hoc balances and
computed as isometric log ratios (ilr).
An optimized binary classification performed by a customized receiver operating
characteristic procedure showed that a critical Mahalanobis distance of 4.21
separated balanced from imbalanced specimens about yield cut-off of 11.83 Mg
grain·ha-1 with test performance of 86%. Quebec maize balance standards differed from
published standards computed from DRIS norms collected in other agroecosystems.
The Redfield N/P ratio in maize leaves was found to be the least variable
balance across regions of the world. The DRIS dual ratios and raw concentration
values were found to be geometrically inadequate for conducting diagnosis. The
unbiased nutrient balance diagnosis combined
the critical Mahalanobis distance and a mobile representation of nutrient
balances with ilr means of
true negative (TN) specimens centered at fulcrums and back-transformed ilr values of TN specimens into raw
concentrations loading the buckets below. Nutrients can be appreciated as
relative shortage, adequacy or excess in the concentration domain following
statistical analysis and diagnosis in the unbiased balance domain.