GenGIS is an open-source software package that allows users to visualize and interactively explore georeferenced genetic datasets in order to investigate the influence of geography and the environment on spatial and temporal patterns of biodiversity. Specific attention has been given to the visualization of 2D and 3D tree structures within their associated geographic context along with statistical tests for evaluating the degree to which the topology of a tree correlates with a geographic or ecological gradient. GenGIS can be applied to a wide range of datasets and has been used to explore the geographic distribution of HIV-1 subtypes in Africa, the early spatiotemporal dynamics of the recent H1N1 pandemic, and the influence of geography on eukaryotic phylogenies such as the Ensatina salamanders of the Pacific coast.

Here we investigate 19 metagenomic samples of marine microbial communities collected as part of Global Ocean Sampling expedition. These samples cover a wide latitudinal gradient extending from Halifax, Nova Scotia to the Panama Canal (~9°N to 45°N) and a range of habitat types (i.e., a freshwater sample, a human impacted embayment, three estuarine samples, and 14 open ocean samples). Using 16S ribosomal DNA genes from these samples as indication of their phylogenetic composition, qualitative and quantitative beta diversity is calculated using the UniFrac diversity measure. Correlations between microbial community composition and geographic proximity or habitat type are assessed using visualizations and analytical tests provided by GenGIS.

Results/Conclusions

Our analysis suggests environmental conditions, particularly salinity, strongly influences microbial community composition. The two low salinity estuarine samples (3 to 10 ppt) from the Delaware and Chesapeake Bays are similar in both species richness and evenness. The freshwater sample taken from Lake Gatun, Panama (0.1 ppt) has similar richness to these estuarine samples, but appears distinct from all samples when species abundance is considered. Furthermore, our analysis reveals that the Bay of Fundy estuarine sample (25 to 31 ppt) appears similar in composition to the open ocean samples from the North Atlantic (~30 ppt) which may be explained by their similar salinity levels.

A naïve linear regression of these samples provides statistical evidence for a latitudinal gradient of species richness. Using GenGIS we show that a visual analysis of this dataset combined with a new tree-based statistical measure provides strong evidence that the 9 open ocean Northern Atlantic samples are compositionally distinct from the 5 open ocean Caribbean samples. However, within these distinct open ocean habitats there is little evidence of latitudinal structure.