Accelerating Plant Breeding: Past, Present and Future

Since the start of modern plant breeding a century ago, there has been focused effort on how to optimize selections and maximize the rate of gain.  Mid-century, during the heyday of quantitative genetics, advancement of the theory behind expected gain from selection was empirically explored and implemented to accelerate breeding.  New statistical approaches to account for field variation lead to higher selection accuracy.  During this time, the mechanization of breeding programs drastically increased population size and testing environments leading to more accurate assessments of breeding values and higher intensity of selection.  With the strong foundation of robust field trials, the advent of inexpensive molecular markers found fertile ground to push the speed and population sizes further.  At the turn of this century, implementation of molecular breeding is currently enabling selection at a speed not possible with phenotypic selection.  Spanning the scope from allele enrichment to whole-genome prediction, the breeding methodologies now possible through direct selection on genotypes gives new scope to breeding.  Gazing into the future it is likely that we will see phenotyping once again coming front and center.  Though, in this new century, phenotyping being automated and high-throughput at the intersection of classical breeding, physiology, engineering, machine vision, artificial intelligence and high-performance computing.  Though genomics tools are revolutionizing breeding, in the end, the prediction models are only as good as the phenotypes that are fed in.