Meta-analysis as a tool to improve robustness and reliability of yield
response models to nitrogen fertilization
Valkama, E., Turtola, E.
MTT Agrifood Research Finland, Plant Production/ Soil and Plant
Nutrition, FI-31600, Jokioinen, Finland
Fertilizer applications should be based on relevant yield response
models and be economically justified. Yield responses to nitrogen (N)
fertilizer usually vary among years, sites, crops and varieties that
lead to numerous response curves, uncertainty in the models’
development, and difficulties in obtaining the economically optimal N
rates (Nopt). To overcome these challenges the technique of
meta-analysis was applied to improve the robustness and the reliability
of the yield response models.
Meta-analysis is the statistical analysis of a large collection of
results for the purpose of integrating the findings. In a
meta-analysis, the result of each independent experiment is usually
expressed as an index of effect (effect estimate). To produce a summary
of the findings, these effect estimates are then combined across
studies using a weighting procedure. Sub-groupings of experiments are
often examined separately to determine whether experimental results
differ among the categorical groups (e.g. decades, soil types, soil pH,
cultivation zones and yield without N). When the statistically
significant difference is found, the yield response models should be
fitted separately for each categorical group.
We gathered 60 Finnish field experiments from the years 1947–2004 in
order to develop the yield response models of cereals to nitrogen (N)
fertilization. The Mitscherlich-type exponential models fitted the data
better than the quadratic models. Among the studied categorical groups,
yield without N fertilization (the “low”, “medium” or “high” group) was
a key factor determining both the magnitude of yield response and
models’ parameters. Taking this into account, from tens of independent
experiments it allowed for obtaining only one response curve and one
value of average Nopt for a certain cereal, yield group and the
fertilizer-to-cereal price ratio (for Nopt).
Valkama_Malliseminaari_Helsinki_ 02 02 2012.pdf