Astronomy and population ecology seemingly have little in common. But they meet in a paper by Zaven Arzoumanian and colleagues, who have developed a way of recognizing individual spot patterns of whale sharks by adapting an algorithm used for comparing star patterns in images of the night sky (J. Appl. Ecol. doi:10.1111/j.1365-2664.2005.01117.x).

Many animals have beautiful spot patterns that differ subtly between individuals of the same species. Photographs of these markings allow ecologists to identify individuals in population studies, but sifting through them is laborious and often unreliable when done by eye. The aim of the authors — an astronomer, a computer scientist and a marine biologist — was to automate the process.

The whale shark (Rhincodon typus, pictured) is the world's largest fish, reaching lengths of more than 14 metres, and bears patterns of white spots on its dark dorsal surface. The leap forwards made by Arzoumanian et al. was to view these patterns as celestial constellations. They adapted an algorithm for identifying characteristic stellar patterns from the geometric properties of triangles made by the lines formed between all possible combinations of three-spot coordinates within a defined region of space. In the same way, by concentrating on a highly variable region of spots above the pectoral fin of whale sharks, the spatial relationships between spots formed the basis for identifying a shark's unique spot pattern.


The modified algorithm was used to calculate similarity scores by comparing the geometry of each spot triangle in a shark photograph with all spot-triangle combinations from a second image. This procedure was repeated between all image-pair combinations to locate high-scoring matches. The researchers then tested the technique by scanning a web-based database of around 1,500 whale shark photographs, and achieved a 92% success rate in matching previously identified pairs of images of individual sharks, as well as resolving new matches not detected by eye. The approach should improve our understanding of whale shark population dynamics, about which little is known. More information is needed urgently, as increasing exploitation of whale sharks means their populations are now vulnerable.

Many animal species have star-quality skin patterns, and the study shows how individual animals can be ‘tagged’ at low cost for long periods, and without physical interference. Linking the pattern-matching algorithm to a web-based photo-identification database, so that it performs like an Internet search engine, will allow researchers around the world to compare new images with vast numbers of virtually tagged animals.