The Chondrichthyan Tree of Life Project: Taking Stock of the World's Sharks and Rays

Methods

Summary

Sampling: Accurately estimating the tree of life depends on sampling as many species as possible. We've been collecting tissue samples for genetic analysis for 20+ years, and have amassed samples of more than 800 species, representing all families and genera of chondrichthyans. This includes described and newly described species, as well as undescribed species. Any species that are missing from our collection are rare animals that are difficult to sample.

Phylogenetic analyses: each gene in the genome can have a different history (for example, due to differences in mutation and/or recombination rates or as a result of hybridisation, selection etc). The challenge when inferring the tree of life is to sample a sufficient number of different genes to make sense of these differences and reveal the true species tree. Technological advances in recent years mean it is now easier than ever to collect large amounts of sequence data. We have developed a new gene capture method that allows us to target more than 1000 different genes simultaneously. We designed probes for these 1000+ genes based on the available whole genome sequence of the elephant shark. We hybridise these probes to the genomic DNA of our target species, then wash away other regions of the genome that we are not interested in. This leaves us with a target DNA library for each chondrichthyan species that is enriched to contain only the 1000+ genes of interest, which we then sequence on a high throughput sequencing instrument. This data is then subjected to a suite of model optimised phylogenetic analyses. We work closely with taxonomists throughout this process such that we are constantly considering the molecular data in conjunction with the set of relationships that are suggested by classical taxonomy and morphometric data. This allows us to address any regions of ambiguity or conflicting topology based on multidisciplinary data.

CT scanning: Our CT scans can be thought of as comprising thousands of adjacent 2-D x-ray slices arranged from head to tail - for each specimen. We convert the sequential 2-D information into 3-D volumes through a process that is referred to as "segmentation". Segmentation involves recognizing corresponding components of the skeleton in each sequential CT slice. The process requires an understanding of the anatomy. We use computer software to help us, but most of the "decision making" involved in the process is done by a human (usually one of our very hard working graduate students). The process is time consuming. On average it takes us about 30 hours to segment a single specimen before we have a 3 dimensional representation of the skeleton that we can load on our website for everybody to explore.