**6. Conclusions**

This study investigated the replicability of PLS regression methods, including PLS with waveband selection, for predicting nutrient content in plant and grain material across multiple environments. Using *Eragrostis tef* (tef) as a target crop, this study compared PLS model fits and selected wavebands across two environments in the U.S. and Ethiopia to determine the extent to which the methods and finding are replicable. Three main findings emerge from this study:

First, the model fits and wavebands selected as important for nutrient prediction were not replicable across the two study sites in the US and Ethiopia. Eleven of the 12 comparisons had statistically different model fits across 1000 bootstrapped iterations, and the Jaccard index for similarity indicated very low similarities in the wavebands selected.

Combining samples from both environments improved model fits, suggesting that increasing within-sample variation may improve the predictability of PLS models across study areas, though caution is reserved if great disparities in sample values are great. Our recommendation is to build a more open and transparent culture of data sharing within the remote sensing community that will permit data sharing in order to advance modeling capabilities and promote development of more generalizable predictive models.

Results using PLS regression with hyperspectral data from non-milled grains were generally positive, and wavebands for protein prediction generally agreed with other studies. While more research is needed to determine whether these consistencies are true positives or are affected by other factors, this study contributes to the gap in the literature related to non-milled grains.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2072-4292/12/18/2867/s1, Figure S1: Bootstrapping results for plant and grain analyses with the PLS-Full model, Figure S2: Bootstrapping results for plant and grain analyses with the PLS-Wave model, Table S1: Selected wavebands from the partial least square regression (PLS) using waveband selection (PLS-Wave) for protein in plant and grain for the combined United State and Ethiopia samples (USET).

**Author Contributions:** Conceptualization: K.C.F., A.E.F.; Methods: K.C.F., A.E.F., S.A.; Fieldwork/Data Collection: K.C.F., A.E.F., S.A.; Data Analysis: K.C.F., A.E.F.; Writing/Editing: K.C.F., A.E.F., S.A. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Acknowledgments:** The authors would like to thank the many that supported this research including the Ethiopian Biodiversity Institute for hosting and supporting the research/field work, Oklahoma State University's Soil, Water, and Forage Analytical Laboratory and the Ethiopian Public Health Institute of Addis Ababa for providing the required nutrient analyses, Kensuke Kawamura for sharing his partial least square regression with waveband selection code, and the Fulbright U.S. Student Scholars Program for providing fiscal support. Special thanks also to Dejene Dida, Tariku Geda, Thiago Souza, Nathalia Gratchet, Andy Han, and Emily Ellis for their support and aid in field and laboratory experiments. Moreover, we would like to thank the editors for their time and consideration of the manuscript. USDA is an Equal Opportunity Provider and Employer.

**Conflicts of Interest:** The authors declare no conflict of interest.
