Aflatoxin contamination in post-harvest maize threatens food safety and security; therefore, there is a value in finding ways to reduce mycotoxin contamination in maize. Our study focused on using a pre-calibrated multi-spectral sorting machine based on visual characteristics previously associated with mycotoxin contamination of maize kernels to reduce aflatoxin contamination in maize. Maize samples (n=45 each) were collected from Ghana and Kenya. A pre-existing linear discriminant analysis (LDA) classification algorithm was used to reject kernels with high visual risk features. 5g subsamples of the sorted samples were used for a one-step lateral flow immunochromatographic assay and ELISA for Ghana and Kenya samples, respectively. After sorting, aflatoxin mean difference between the accept and the reject stream for Ghana was -0.28 log(ng/g), paired t-test showed significant difference (p<2.2e-16) between the accept and reject levels for each individual sample . Kenya sample’s mean difference was -0.28 log(ng/g) (p=0.08 by paired t test ). The Ghana and Kenya samples had an average mass rejection of 37.9% (range 15.9-10.7), 7.17% (range 1.10-25.2%) respectively. This shows that performing multi-spectral sorting under less controlled field conditions still retained efficacy to reduce mycotoxin in maize, despite having a lower effect than when aflatoxin sorting performed under highly controlled laboratory settings.
Data analysis was performed in R 4.4.1 and RStudio Desktop with the following packages:
- BayesianReasoning
- car
- datasets
- data.table
- gapminder
- ggplot2
- ggpubr
- grid
- gridExtra
- multcompView
- patchwork
- plotly
- reshape2
- ragg
- rstatix
- tidyverse
Data and code are located in "data and analysis".
R analysis can be run by opening the "data and analysis.Rproj" file, followed by "Julie_Ghana_Kenya_revised.Rmd" and running the required code chunks.
You can view the list of authors in the AUTHORS file.
Corresponding author: Matthew J. Stasiewicz
103 Agricultural Bioprocess Lab
1302 W. Pennsylvania
Urbana, IL, 1361801
USA
+1-217-265-0963
[email protected]
Publication pending.
This project's code is licensed under the GNU General Public License v3.0 and dataset is licensed the Creative Commons Attribution Share Alike 4.0 International license. Please see the LICENSE.code and LICENSE.dataset files for details.
This work was supported by the University of Illinois at Urbana‐Champaign College of Agriculture, Consumer and Environmental Sciences (ACES) Office of International Programs seed grant to Matthew J Stasiewicz, and Global Food Security Internship Program to Julie Hwang. As well as an ACES James Scholar Honors Program travel grant to Julie Hwang. As well as a University of Illinois I-MMAS program and Tecnológico de Monterrey partnership research internship to Mauricio Canales. As well as a USDA Research Capacity Fund (Hatch) project Food and Nutrition Systems for Safety, Security and Sustainability [ILLU-698-930] to Matthew J Stasiewicz.