Data and code from: Severity of charcoal rot disease in soybean genotypes inoculated with Macrophomina phaseolina isolates differs among growth environments

This dataset includes all the raw data and all the R statistical software code that we used to analyze the data and produce all the outputs that are in the figures, tables, and text of the associated manuscript:

Mengistu, A., Q. D. Read, C. R. Little, H. M. Kelly, P. M. Henry, and N. Bellaloui. 2025. Severity of charcoal rot disease in soybean genotypes inoculated with Macrophomina phaseolina isolates differs among growth environments. Plant Disease. DOI: 10.1094/PDIS-10-24-2230-RE.

The data included here come from a series of tests designed to evaluate methods for identifying soybean genotypes that are resistant or susceptible to charcoal rot, a widespread and economically significant disease. Four independent experiments were performed to determine the variability in disease severity by soybean genotype and by isolated variant of the charcoal rot fungus: two field tests, a greenhouse test, and a growth chamber test. The tests differed in the number of genotypes and isolates used, as well as the method of inoculation. The accuracy of identifying resistant and susceptible genotypes varied by study, and the same isolate tested across different studies often had highly variable disease severity. Our results indicate that the non-field methods are not reliable ways to identify sources of charcoal rot resistance in soybean.

The models fit in the R script archived here are Bayesian general linear mixed models with AUDPC (area under the disease progress curve) as the response variable. One-dimensional clustering is used to divide the genotypes into resistant and susceptible based on their model-predicted AUDPC values, and this result is compared with the preexisting resistance classification. Posterior distributions of the marginal means for different combinations of genotype, isolate, and other covariates are estimated and compared. Code to reproduce the tables and figures of the manuscript is also included.

The following files are included:

• README.pdf: Full description, with column metadata for the data spreadsheets and text description of each R script
• data2023-04-18.xlsx: Excel sheet with data from three of the four trials
• cleaned_data.RData: all data in analysis-ready format; generates a set of data frames when imported into an R environment
• Modified Cut-Tip Inoculation on DT974290 and LS980358 on first 32 isolates.xlsx: Excel spreadsheet with data from the fourth trial
• data_cleaning.R: Script required to format data from .xlsx files into analysis-ready format (running this script is not necessary to reproduce the analysis; instead you may begin with the following script importing the cleaned .RData object)
• AUDPC_fits.R: Script containing code for all model fitting, model predictions and comparisons, and figure and table generation