Statistical Analysis of Results from Irrigation Experiments Using the Line‐Source Sprinkler System

TLDR

The line‑source sprinkler system is widely used to study how treatment variables and irrigation influence yield and other traits, offering a space‑efficient design that visually displays irrigation effects. Statistical tests for treatment effects and treatment‑by‑irrigation interactions are feasible when treatments are replicated and randomized in strips perpendicular to the sprinkler line, but no test exists for irrigation level effects because irrigation amounts are applied systematically without randomization. Because irrigation effects are typically large and obvious, significance testing is unnecessary, though care is needed when interpreting small differences between adjacent irrigation levels and when estimating yield‑irrigation regressions due to systematic bias.

Abstract

Abstract The line‐source irrigation system has been used in many studies to determine the influence of several treatment variables and irrigation on yield and other traits. It has the advantage of using a minimum of space and shows visually the irrigation effects for some traits. Valid statistical tests for treatment effects and the treatment by irrigation interaction are possible if the treatments are replicated and randomized in strips at right angles to the sprinkler line. No statistical test is available for the effect of irrigation level on yield because irrigation amount is always applied systematically with no randomization. Since the irrigation effects are usually large and obvious there is no need to assign a probability level to their significance. However, caution should be exercised in the interpretation of small differences between adjacent irrigation levels and for potential bias in the estimates of the regressions of yield on irrigation level brought about by the systematic arrangement.