Phylogenetic analysis is a powerful methodology with broad explanatory power derived from its ability to make predictions about evolutionary process. However, the process of arriving at a classification involves decisions about how to reconcile multiple tree solutions, how to visualize data changes on phylogenetic tree topologies, how to represent conflicts among solutions based on different data sets and finally, methods for presenting these results in publishable form. The field of scientific visualization specializes in the development of visual display techniques to reveal (1) patterns and structure within a data set, and (2) patterns and structure between data sets. The phylogenetic tree, or cladogram, is an example of type (1) visualization because it reveals the relationships among taxa within a single data set. Type (2) visualizations instead focus upon the simultaneous presentation of multiple data sets or multiple representations of the same data set. Within this proposal we outline a series of highly visual, data driven methods for data inspection, simultaneous multiple tree visualization, evaluation of assumptions about the analysis (e.g., choice of root taxon), presentation techniques for hierarchical data and methods for linking, through the World Wide Web (WWW), phylogenetic trees to the data that generated the result(s). Few or none of these techniques are in current use in systematics, in fact, they represent very recent research within computer science. We -believe that the richness and complexity of phylogenetic analysis invites new methods for visualizing the analytical process. MacClade Maddison92 , a Macintosh program for visualizing data change on a single tree, enhanced our understanding of the how character changes occurred on single tree topologies. We will expand that type of analysis to include far richer visual metaphors for viewing the da ta matrix to phylogeny pathway. The research we are proposing is exploratory and by design we will investigate methods that may not prove fruitful for all data sets. However, the applicability of these methods for other forms of hierarchical data makes the likelihood high that some or even most of the visualization tools will assist researchers in all areas of comparative biology. In addressing the visualization problem, the assumptions--a 2D, static medium, Euclidean drawing space, separate figures, data and explanations and homogeneous levels of detail in single illustrations--implicit in current tree visualization techniques must be re-examined. We envision that as electronic journals become ever more prominent in all-areas of science the need will exist for display techniques optimized to on-line viewing. This research should provide significant insight into means for displaying classifications and systematic analyses within this new medium. Although a large number of the specific deliverables for this research will be "backend" programs for creating the visualizations .and database architectures for storing trees and data, we will provide access to the entire suite of technologies through WWW browsers (by mid-1996, browsers for all major workstation and PC platforms should incorporate the capability for viewing these results). Users will be able to input and store trees and NEXUS (a standard file format for phylogenetic data) data matrices for subsequent analysis, all within their browser.
