Flexible Algorithms for Visualizing Geophylogenies

Abstract. Geophylogenies enrich the leaves (species) of a phylogenetic tree with geographic locations (sites) that are biologically meaningful, such as sightings, habitats, or fossil finds. To extract geographic patterns from this data, geophylogenies are customarily visualized using a map to indicate the locations, a tree drawing for the phylogenetic tree, and some sort of linking mechanism, such as labels or leaders, which matches leaves and locations. So far, such geophylogeny visualizations are mostly created without sophisticated algorithmic assistance. An exception is recent work by Klawitter et al. (2023) which focusses mainly on point sites as geographic locations, tree drawings with fixed leaf positions on a line, and linking via leaders. In this paper we significantly extend their work by adding additional flexibility to all components of the visualization pipeline: we support not only point, but also region sites, the locations of leaves are adaptive to the data and can lie on either a line or a circle, and the locations are visually linked to leaves without the need for explicit connections via leaders. We implemented our algorithms and evaluated them experimentally. Our results show that the added flexibility indeed results in visualizations of higher quality for datasets of up to medium size; for large datasets the leaves of the phylogenetic tree tend to be so crowded around the map that the difference between adaptive and fixed leaf positions becomes negligible. However, the possibility of placing leaves on a circle and the linking without leaders still improve the readability of our visualizations when compared Klawitter et al. (2023).