Distinct pigmentation patterns formed by black spots or stripes are common in a range of species throughout the animal kingdom – from insects to mammals, birds, and fish. This project is about the characteristic black spots on the otherwise silvery skin of salmon. These spots consist of the black pigment melanin, the same molecule that provides mammals such as tigers and cheetahs with their characteristic pattern. In salmon the number and size of melanin-based skin spots vary considerably between individuals, and the pattern reveals a surprising amount of information about each animal. Fishes that are sensitive to changes in the environment, stress, and infections as a rule have small and few spots. These individuals will also typically be socially subordinate, and show higher levels of the “stress hormone” cortisol. Also in other animals there are clear links between pigment patterns and social position. The dark mane of dominant male lions is but one example. Notably, in salmon the spot pattern is a heritable individual characteristic, like a fingerprint. It is tempting to say that since the fish don’t have internet, they carry their social media profile on the outside of their body. Strictly, however, we do not yet know if fishes check each other’s spots and adjust their behaviour based on the visual appearance of others. Another question is whether the pigment fades in states of stress and disease. Or in other words, whether day-to-day changes in fitness are considered public information or kept private among fishes. To this task, the project will develop machine vision and AI-algorithms which can both recognise individually consistent spot patterns and adjudge their signal strength in terms of the level of contrast vs fading as a response to changing environmental and physiological conditions. Conjointly, this constitutes a new level of collaboration between fundamental behavioural ecology, applied biomedical studies, and present-day information technology.
Animal pigmentation has spurred some of the most controversial and productive fields in biology, including sexual selection, social signalling, sympatric speciation, crypsis and mimicry. Pigmentation patterns of live animals are however notoriously difficult to describe in mathematical terms that can be used for further statistical analysis. It is thus becoming increasingly important to develop means of quantifying the various properties of visual signals in a manner independent of human perception. Here we suggest a project to integrate state of the art machine vision and AI-based biometric sensing to reliably capture and describe a hitherto unrecognized but highly conspicous visual signal: Distinct melanin-based skin pigmentation patterns seen in salmonid fishes, the information content of which has remained openly hidden to human observers. Using individually recognizable patterns in high numbers of fish undergoing stressfull rearing procedures in aquaculture, we ask the question whether stress induced modification of melanin-based pigmentation patterns represent an external visual representation of internal state of the organism. The proposed project will investigate this possible ramification while elucidating the underlying physiological and molecular-genetic mechanisms linking individual variation in stress coping ability to pigmentation patterns. Gene-phenotype interactions are herein revealed by combining a systems biology approach at the genomic and transcriptomic level with forefront research in machine vision and deep learning algorithms. In this, we will achieve integration between forefront biological research and some of the most effective contemporary technological innovation environments in Norway.