How Robust Are Genomic Offset Predictions to Methodological Choices? Insights From Perennial Ryegrass

Area of Science:

• Population Genetics
• Climate Change Adaptation
• Bioinformatics

Background:

• Genomic offsets quantify genetic mismatch with environmental change, crucial for assessing climate maladaptation.
• Understanding methodological sensitivity in genomic offset prediction is vital for reliable climate risk assessment.

Purpose of the Study:

• To compare Gradient Forest (GF) and Canonical Correlation Analysis (CANCOR) for outlier detection in genomic offset studies.
• To assess the sensitivity of genomic offset predictions to sampling design and identify robust methods for climate adaptation research.

Main Methods:

• Utilized 457 perennial ryegrass populations with 189,968 SNPs and 75 climatic variables.
• Compared GF (non-linear, non-parametric) and CANCOR (linear, parametric) for outlier detection.
• Experimentally validated genomic offsets against 105 phenotypic traits in common gardens.

Main Results:

• Both GF and CANCOR identified significant outlier loci associated with environmental gradients (2113 for GF, 653 for CANCOR).
• Genomic offset predictions were spatially congruent across both outlier detection methods.
• GF demonstrated lower sensitivity to sample size and geographic biases compared to CANCOR.

Conclusions:

• Non-linear, non-parametric methods like GF may offer more robust genomic offset predictions due to reduced sensitivity to sampling biases.
• Findings provide practical guidance for designing genomic offset studies in agricultural and natural systems.
• Phenotypic traits correlated with genomic offsets can serve as valuable fitness proxies for climate risk assessment.