The plant-to-plant variability, which is caused by non-genetic factors, is called micro-environmental variation and can be evaluated and controlled for with experimental designs. We are a leading group working on different aspects of field experimental design to optimize genotypic comparisons. One of the key aspects of agricultural research is discerning noise from signal that can be accomplished with proper experimental designs and analysis models. However, in the last few decades, the need to evaluate increasingly large genomic populations and the increasing availability of geographical-information systems and computational capabilities created the belief that a substitution of experimental designs for spatial modeling would better account for micro-environmental variability. Our group used the latest advancement in geographical-information systems, high throughput computing and classic experimental design theory to prove that spatial modeling cannot substitute for proper experimental designs (Borges et al., 2019). Furthermore, spatially minded experimental designs such as alpha-designs perform better than traditional experimental designs like randomized complete blocks. This provided a key tool for researchers to decide on experimental design strategies. We then extended the analysis to control both micro and macro environmental variability (i.e. genotype by environment interaction) simultaneously and proposed a new experimental design strategy called Mega-Environmental Design that can be used to design optimal experiments in the presence of genotype by environment interaction improving genotypic mean estimation, reducing standard error, and increasing the expected response to selection (González- Barrios et al., 2019).

We have also optimized phenotyping resources for traits expensive to measure (Lado et al., 2018). One of the major accomplishments in this area includes the development of improved methodological approaches for studying complex traits. Our group has also worked on other strategies to improve mean comparisons: optimizing phenotyping experiments (Quero et al., 2013; Rosas et al., 2016), using soil parameters for spatial modeling (González et al., 2012; González-Barrios et al., 2015a), approaches to evaluate genetic components of time-scale parameters (González-Barrios et al., 2015b) and spatio-temporal models (González-Barrios et al., in review). All these phenotyping approaches produce better signal to noise ratio improving plant evaluations. The experimental design optimization for large genomic studies which is a recent line of work involves among others a $0.5 million federally funded NIFA grant in which Dr. Gutierrez is the PI.

Future work includes extending the spatial models of Borges et al. (2019) for an invited paper (Hoefler et al., in prep), the generalization to large population sizes and genotype-by-environment structure of the Mega-Environmental Design paper (Berro et al., in prep.) proposed in our renewal of the Hatch grant, and the large scale implementation of these strategies as part of a pending NSF grant.