Research

Research Interests

Human civilization relies heavily on plants for our survival. Plants provide us with food, shelter, fiber, and the atmospheric oxygen we breathe. As human populations and resource consumption increase, we place ever-greater demands on plant communities and are now changing the climates in which they grow. Breeding plants to be efficient and inexpensive carbohydrate delivery systems has had adverse impacts on human health, while simultaneously selecting for undesirable side-effects such as inefficient use of water and nutrients by crop plants. Discovering the evolutionary, ecological and genetic basis for the wide natural variation in how plants acquire and allocate developmental, energetic and nutritional resources will be essential if we are to maintain adaptability in crop plants and natural populations alike.  This topic is the focus of research in my laboratory.

Adaptive life history traits in plants, such as growth rates and patterns, reproductive output, and resource allocation, are complex traits. Variation in complex traits is usually influenced by multiple genes (called quantitative trait loci, or QTLs), environment, and their interactions, and these traits typically show a continuous range of variation.  Variation in different complex traits is often genetically correlated, and such correlations are expected when life history trade-offs are involved.  Thus, we are interested in knowing the locations and ultimately the identities of genes underlying trait variation, the molecular and developmental pathways by which they affect suites of traits, how large their effects are, and how they interact with other genes and the environment.

Specific areas of research in my laboratory include the following:

Genetics of resource allocation in Arabidopsis lyrata:

The rock cress Arabidopsis lyrata, which is native to parts of North Carolina, is a close relative of Arabidopsis thaliana, an important model plant for genetics research and the first plant to have its complete genome sequenced. Unlike A. thaliana, A. lyrata is an outcrossing perennial that shows strong evidence of local adaptation throughout its North American and Eurasian range. The A. lyrata genome sequence has also been completed recently. This combination of features and genomic resources makes it an ideal plant for investigating the molecular genetic basis of adaptation.

 We have found that A. lyrata plants from populations in different environments, such as North Carolina and Norway (above left) show major differences in their allocation of resources to reproductive vs. vegetative processes when grown in a common environment (above right). Thus, A. lyrata provides an outstanding system for dissecting the developmental and genetic basis for variation in resource allocation patterns, which are a key aspect of local adaptation in plants but have not been studied to date in an integrated manner. In collaboration with Outi Savolainen (University of Oulu, Finland), we combined QTL mapping with structural equation modeling to characterize resource allocation trade-offs in Norway x North Carolina crosses tested in both parental environments (Remington et al. 2013).  A key finding was that QTL regions affected resource allocation largely by regulating early lateral shoot development, leading to cascading effects on later development.  North Carolina alleles increased  reproduction and concurrently reduced vegetative development in North Carolina, while reducing both survival and reproduction in Norway.  These effects appeared to involve shifts in developmental timing between the two environments, not direct trade-offs between reproduction and survival.  Thus, while genes affecting resource allocation appear to have important roles in local adaptation, this may involve complex interactions between developmental patterns and environment instead of traditional models involving direct costs of reproduction. 

 Modeling adaptive trait variation:

To support our empirical research on plant life history evolution, we are developing modeling approaches for making better inferences about the nature of gene-to-phenotype relationships in complex developmentally-related suites of traits. Multiple traits involved in plant resource allocation processes are predicted to have cause-effect relationships, which can be described in terms of trait networks. Tools to model the effects of genetic variation in such networks date back nearly a century to the pioneering research of geneticist Sewall Wright, but ironically have been little-used by geneticists until recently. Recent advances in molecular genetic and genomic techniques have facilitated the use of network approaches to evaluate causal relationships between genetic variation, gene expression, and phenotypes in the emerging field of systems biology. We have extended these approaches to evaluate how QTLs regulating trait variation at particular points in development can produce cascading effects on "downstream" traits, potentially leading to complex variation in life histories. Undergraduates participating in UNCG's NSF-funded Math-Bio Program (Robert Gove and Becca Fogel Erwin) and Quantitative Science REU Program (Will Chen, Nick Zweber and Erika Helgeson) made major contributions to developing these models (Gove et al. 2012). 

The thesis research of a previous graduate student, Jolly Shrivastava, involved development of a Bayesian model to infer the parental genotypes and effects of multiple quantitative trait loci (QTLs) from trait data in diallel mating designs.  Such models may be useful for inferring the presence and population frequency of large-effect QTLs prior to initiating expensive genetic mapping projects in previously unstudied species.

More recently, I have been interested in exploring the evolution of different patterns of genetic architecture; i.e. the extent to which variation in individual genes contributes to complex trait variation and how this variation evolves.  With the expansion of the molecular "toolkit" for these studies, it is important to distinguish the effects of individual variants from the overall effects of the alleles in which they occur (Remington 2015).  We are using genetic simulations to model the evolution of genetic architecture under different natural-selection scenarios.