How do we address the disconnect between genetic and morphological diversity in germplasm collections?

Th e world's food production systems must keep up with an ever-increasing demand in spite of the challenges caused by climate change and environmental degradation. To meet this challenge, plant breeders must have access to genetic diversity for crop improvement. However, genetic homogeneity in the handful of major crops that feed the world limits options for breeding progress (Swaminathan, 2009). Wild and landrace relatives of crop plants provide much needed diversity and are maintained in gene banks worldwide (McCouch, 2013). Th is germplasm resource is under-used by breeders, mainly because genotypic and phenotypic data are limited. However, using rapidly developing genomics resources, breeders are unlocking the genetic potential in gene banks and making remarkable advances (Tanksley and McCouch, 1997). Morphological traits and neutral genetic markers are commonly used to characterize germplasm collections, with the goal of improving the value of these collections to breeders (Brown, 1989). Morphological and neutral marker data provide the foundation for the assembly of a core collection, defi ned as a subset of a germplasm collection intended to represent the entire collection. Th e concept of a core collection assumes that the core has some utility for plant improvement. Th at is, the core collection is maximally diverse and is useful for breeding. Evaluating the core collection for key traits should also have some level of predictability for the rest of the collection based on the relatedness of core accessions to other accessions in the collection. It has been recognized for many years that morphological traits do not always provide a good measure of genetic values and may not accurately reveal the genetic variation in a germplasm collection (Tanksley and McCouch, 1997). Consequently, neutral molecular markers have become popular because they are thought to more accurately refl ect genetic relationships in germplasm collections (Ebana et al., 2008). But genetic relatedness measures based on neutral markers may not predict similarity in trait values or parental performance. Th e patterns of allelic variation in a species may be very diff erent for neutral markers compared with genes under selection. Based on a meta-analysis, McKay and Latta (2002) argued that allele frequencies at neutral and selected loci are not correlated because evolutionary forces act diff erently on them. A meta-analysis by Reed and Frankham (2001) showed only weak correlation between neutral molecular markers and quantitative measures of variation. Reeves, Panella, and Richards (2012) found that genetic variation at loci …