Visualizing morphometric differences between geographic populations

Anja C. Schunke, Max Hermann, and Reinhard Klein
Poster presentation at 9th Int. Congress of Vertebrate Morphology in Punta del Este, Uruguay, July 2010
 

Abstract

Morphometric analyses of variation as a method to distinguish between differing geographic populations is a widespread method that works particularly well for patchy occurrences. However, when investigated species have a seemingly continuous distribution area, the analysis of their geographic variation is confronted with some difficulties. The majority of morphometric methods require a pre-grouping either for data presentation or for the calculations (e.g. discriminant function analysis, canonical variates analysis), which is generally based on independent information, e.g. differences in coloration, or hypotheses of geographic barriers like rivers or mountains. This a priori grouping is necessarily more or less arbitrary and can easily lead to false negative results, when differing groups are (partially) lumped together and/or separated. Here we present an interactive visualization system which enables a better supported grouping based on the morphometric data.

At the heart of our system are iconic information visualization techniques which are capable to display all attributes of a specimen at once in a graphical manner. The idea is that overlaying these icons on a geographic map makes differences between single specimens or whole populations apparent in their spatial context. To facilitate rapid direct quantification of attributes as well as qualitative comparison of attribute differences between different regions, our system provides two icon types which exhibit adequate perceptual qualities for those tasks. The underlying attribute set can be changed interactively in our system. Furthermore a spatial grouping can be applied which eases the investigation of large datasets by visualizing group representatives instead of individuals.

Images

Download Paper

Download Paper

Bibtex

@MISC{hermann-2010-starvis-poster,
        author = {Schunke, Anja C. and Hermann, Max and Klein, Reinhard},
         title = {Visualizing morphometric differences between geographic populations},
          year = {2010},
         month = jul,
  howpublished = {Poster presentation at 9th Int. Congress of Vertebrate Morphology in Punta del Este, Uruguay},
      abstract = {Morphometric analyses of variation as a method to distinguish between differing geographic
                  populations is a widespread method that works particularly well for patchy occurrences. However,
                  when investigated species have a seemingly continuous distribution area, the analysis of their
                  geographic variation is confronted with some difficulties. The majority of morphometric methods
                  require a pre-grouping either for data presentation or for the calculations (e.g. discriminant
                  function analysis, canonical variates analysis), which is generally based on independent
                  information, e.g. differences in coloration, or hypotheses of geographic barriers like rivers or
                  mountains. This a priori grouping is necessarily more or less arbitrary and can easily lead to false
                  negative results, when differing groups are (partially) lumped together and/or separated. Here we
                  present an interactive visualization system which enables a better supported grouping based on the
                  morphometric data. 
                  
                  At the heart of our system are iconic information visualization techniques which are capable to
                  display all attributes of a specimen at once in a graphical manner. The idea is that overlaying
                  these icons on a geographic map makes differences between single specimens or whole populations
                  apparent in their spatial context. To facilitate rapid direct quantification of attributes as well
                  as qualitative comparison of attribute differences between different regions, our system provides
                  two icon types which exhibit adequate perceptual qualities for those tasks. The underlying attribute
                  set can be changed interactively in our system. Furthermore a spatial grouping can be applied which
                  eases the investigation of large datasets by visualizing group representatives instead of
                  individuals.}
}