# Diet and trophic structure of fishes in the Barents Sea: between empty and full stomachs – large individual variability follows a common pattern

## TL;DR

Fish in the Barents Sea show large individual variation in stomach contents, but their feeding patterns follow a common statistical pattern.

## Contribution

A common log-normal-like statistical pattern in fish stomach contents is identified across multiple species.

## Key findings

- Stomach content data for six fish species showed a consistent log-normal-like distribution pattern.
- High stomach contents were rare and not plateaued, suggesting infrequent large feeding events.
- Mean and median stomach content values were low, indicating generally low feeding rates among fish.

## Abstract

More than 27,000 stomachs from 70 species of fish were collected from the Barents Sea in 2015. Quantitative stomach content expressed relative to the body weight of the predator fish (g g−1 as %) varied by four to five orders of magnitude for six species with the largest sample size (Atlantic cod Gadus morhua, haddock Melanogrammus aeglefinus, Greenland halibut Reinhardtius hippoglossoides, long rough dab Hippoglossoides platessoides, polar cod Boreogadus saida, and Atlantic capelin Mallotus villosus). The quantitative stomach contents of individual fish followed a common and strict statistical relationship for predator species or groups of species (by families), and for prey categories across predator species. The common pattern was log‐normal‐like and was modelled with good fit by different types of right‐skewed distributions, that is, variants of the Box–Cox, generalized inverse Gaussian, inverse gamma, or gamma distributions. The long tail in the high end reflects high variation with no clear sign of a plateau, as could be expected from the concept of a “full stomach”. This is interpreted to reflect that high stomach contents are rare events that are sampled at low frequencies. The maximum recorded stomach content varied from 1% to 34% of body weight for 55 species of fish, being positively correlated (R
2 = 0.45) with sample size. About a third of the stomachs were empty, and the low tail of the log‐normal‐like distribution represents the transition to empty stomachs. The amount of food in the stomachs was overall low compared to maximum values, with mean and median of 2.0% and 1.1%, respectively, for the 17,873 stomachs containing food. Supported by bioenergetic considerations, this suggests relatively low feeding rates of the various fish predators but sufficient to meet their energy demands.