Digital phenotyping of individual feed intake in Atlantic salmon (Salmo salar) with the X-ray method and image analysis

Abstract

The primary barrier to research into feed efficiency of Atlantic salmon (Salmo salar) is the lack of a reliable method to assess individual feed intake in large cohorts of fish over a growth period. A method with potential is the X-ray method, which images radio-opaque markers (beads) in feed consumed by fish. However, the time taken to count the markers in the digestive tract of fish is extremely onerous and the method has previously been shown to have low repeatability. Furthermore, the method has not been assessed and optimized for Atlantic salmon. Firstly, we made use of image analysis to count beads within the digital radiographs, which was highly correlated to manual counting by human observers (R2 = 0.99). Remarkably, image counting was up to 6 times faster than human counting when the number of beads per fish was high (> 300 per fish). We investigated the potential effect of different sources of error on the feed mass to bead count calibration equation and found the effects of X-ray exposure setting and the position of pellets on the X-ray plate to be negligible on the bead counts of both human and image analysis. We tested different feeding periods with the time of the X-ray images to minimize the loss of beads through defecation. We found that fish should be X-rayed within 6.5–11 hours of first being offered feed for a 6-hour feeding period which fed the entire daily ration. Lastly, we assessed the repeatability of feed intake over a 70-day growth period from 80–300 grams and found feed intake to be significantly and moderately repeatable (r = 0.45 ± 0.11), indicating developments in the method result in a consistent ranking of individual fish based on feed intake from three repeated measurements. The X-ray method combined with image analysis greatly reduces counting time without compromising accuracy, achieves promising repeatability, and is feasible in Atlantic salmon parr.