Put the frog in water: simple methods for improving individual identification as demonstrated with the agile frog

Abstract

Identification of individuals across time and space is required for investigating numerous evolutionary-ecology and conservation-related questions, and photo-based identification is commonly used for a broad taxonomic range. Systematic comparisons of available photo-matching software have been published for various taxa, and success of such software may greatly depend on image quality as well as the focal body parts. Yet, suitability assessments of images captured with different methods are missing. Here we tested the hypotheses that individual identification by colour patterns can be facilitated by taking into account the natural medium surrounding the animals and the natural body posture they tend to take. We developed new photography methods to enable individual identification by whole-body assessment using the agile frog (Rana dalmatina), a species with common yet so-far neglected melanin-based pattern on its limbs and back. We compared the reliability of different photography methods for computer-assisted identification in the HotSpotter software as well as for observers operating it. We found that photographing either hand-restrained frogs with towel-dried skin, or frogs moving freely in clean water enabled comparison of the dorsal surface of the whole body including the hind legs, and HotSpotter identified matching images at rates similar to the relatively more successful of previously published anuran studies (>92% listed with rank[≤]10). By contrast, submerging hand-restrained frogs in water significantly improved identification: images of re-captured individuals were always ranked as the most likely match. We attribute this outstanding performance to the combination of advantageous effects of in-water light refractions that improve the visibility of melanin patterns, and uniform body postures facilitating comparison across individuals. Observers in general successfully identified matching images and ruled out non-matching ones, but some mistakes were recorded when images featured freely moving frogs. The photography methods developed in this study should be easily adapted to most frog and toad species for reliable individual identification. Our study highlights that taking into account features of the natural environment of the studied species can improve individual identification by photographs. Because such methods are non-invasive and inexpensive, they should be especially beneficial for population monitoring programs of endangered species. Data availabilityData are available from the FigShare Repository [DOI will be assigned upon acceptance for publication]. Further supporting information for methods and results is provided as Supplement to this manuscript.