Premium
Increasing Precision in Agronomic Field Trials Using Latin Square Designs
Author(s) -
Jones Marcus,
Woodward Richard,
Stoller Jerry
Publication year - 2015
Publication title -
agronomy journal
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.752
H-Index - 131
eISSN - 1435-0645
pISSN - 0002-1962
DOI - 10.2134/agronj14.0232
Subject(s) - sorghum , randomized block design , agronomy , field trial , field experiment , blocking (statistics) , field (mathematics) , spatial variability , crop , mathematics , mean squared error , statistics , biology , pure mathematics
Spatial variation from soil and related factors often affects the outcome of agronomic field experiments. The randomized complete block (RCB) is the most prevalent design despite inefficiencies that can result in inflated error terms. Experimental designs such as the Latin square (LS) allow for bidirectional blocking and offer the potential to account for spatial variability better. The objectives of this research were to investigate the occurrence of two‐way gradients in agronomic field trials and compare the estimated relative efficiency (ERE) of a LS to a RCB. Thirty LS trials were evaluated in 10 states during 2013 across the midwestern United States investigating crop yields of corn ( Zea mays L.), soybean [ Glycine max (L.) Merr.], and sorghum [ Sorghum bicolor (L.) Moench]. The results show that 47% of the trials exhibited a two‐way gradient, indicating this characteristic is widespread across a large geographic region. Overall, the ERE was increased in 70% of the trials by using the LS design. A lower ERE occurred in 7% of the trials conducted using a LS. Multiple gradients appear common in agronomic field plot trials and enough variation existed between the two blocking directions to justify the use of a LS design. Our data indicate the LS offers a low risk, high reward option of experimental design for controlling spatial heterogeneity and increasing precision. When possible, the LS design should be used in field experiments where the trial area appears uniform and gradients to block against are not obvious.