The Encyclopedia Brittanica (LINK) description of an ecosystem begins with a topic on energy flow. A model that takes energy into account in a food web was developed in the1970’s at the University of Wisconsin and is called the Wisconsin Biometrics Model (LINK). In 2000, a study was done on the stellar’s sea lion in Alaska. (LINK) The paper says that they were testing the “Junk Food Hypothesis”. This hypothesis “generally posits that when high-energy, high-lipid prey (e.g., herring, capelin, or other fatty fish) become less available, predators switch to lower-quality alternatives (lower energy density, less nutritious “junk food”), leading to nutritional stress, reduced reproduction, poorer body condition, or population declines—even if overall prey abundance seems sufficient.” (LINK) (LINK)
In the 1980’s, NOAA developed the Ecopath with Ecosin (EwE) (LINK) model which is used for modeling a food web and is based on biomass, not energy. This model can be modified to use energy, and in another paper, (LINK), back-to-back studies were done comparing biomass-based study with an energy-based study. This study found that the biomass model underestimated the impact of the high energy content herring by 20%-110%. The study also found that warm-blooded predators (birds and mammals) were much more affected than fish, which is what one would expect since warm-blooded animals burn calories just to stay warm. This gives greater importance to the lack of menhaden for ospreys in Chesapeake Bay.
There are low energy fish, like flounder, which do well with a low energy content forage, but active fish like striped bass, tuna and cod need high energy content forage. For examples of bioenergetic studies, see this (LINK).