BALZARINI MONICA GRACIELA
Congresos y reuniones científicas
Título:
Dimension reduction techniques for assessing spatial patterns of genetic variation in native forests.
Autor/es:
TEICH, I.; BRUNO, C.; PEÑA MALAVERA, A.; BALZARINI, M.
Lugar:
La Falda, Córdoba.
Reunión:
Congreso; Cuarta Escuela de Matemática y Biología; 2010
Resumen:
As many ecological and evolutionary factors that influence genetic variation are mediated by space, the joint analysis of genetic and spatial information can lead to a better understanding of the underlying processes. Geographic patterns of intraspecific genetic variability are central to many fields of evolutionary biology, molecular ecology, and conservation biology. Multivariate analyses are useful for extracting meaningful information from georreferenced genomic data. Particularly, dimension reduction techniques (DRT) aim at summarizing multilocus-multiallele molecular marker data into a few uncorrelated synthetic variables that are chosen so as to reflect most of the variability in data, as defined by an optimized criterion. Such methods can be applied on allelic frequencies calculated for each marker as well as on distance metrics derived from the multivariate profile of binary molecular data to explore genetic variability among individuals or populations. Spatial information is commonly used a posteriori of applying DRT and for graphical display purposes. DRT are useful descriptive tools to visualize, quantify and test spatial structure but they are not properly designed to investigate spatial patterns. For instance, ordinary ordination methods may reveal spatial patterns wherever they are obvious, but they are not constrained to do so. When spatial information is used a priori, as a component of the adjusted model or the optimized criterion, it allows investigating spatial structures other than the most evident, by focusing on the part of the variability which is spatially structured. For instance, the Spatial Principal Components Analysis (sPCA) relies on a modification of PCA such that not only the variance of the synthetic variables, but also their spatial autocorrelation, is optimized. In this work we describe a number of analytical DRT which are used for ordination of multivariate profiles of genetic data in reduced spaces, give an illustration of its applications in studies of spatial distribution of genetic variability within native tree species of Argentina, and discuss about the type of biological information that can be extracted in studies at different geographic scales.