Abstract Benthic suspension feeders like corals and sponges are important bioengineers in many marine habitats, from the shallow tropics to the depth of polar oceans. While they are generally considered opportunistic, little is known about their actual in situ diet. To tackle this limitation, fatty acid trophic markers (FATMs) have been employed to gain insights into the composition of their diet. Yet, these in situ studies have not been combined with physiological investigations to understand how physiological limitations may modulate the biochemistry of these organisms. Here, we used the cold‐water coral (CWC) Desmophyllum dianthus in its natural habitat in Comau Fjord (Northern Patagonia, Chile) as our model species to assess the trophic ecology in response to contrasting physico‐chemical conditions (variable vs. stable) and ecological drivers (food availability) at three shallow sites and one deep site. We took advantage of the expression of two distinct phenotypes with contrasting performance (growth, biomass, respiration) coinciding with the differences in sampling depth. We analysed the corals' fatty acid composition to evaluate the utility of FATM profiles to gain dietary insights and assess how performance trade‐offs potentially modulate an organism's FATM composition. We found that 20:1(n‐9) zooplankton markers dominated the deep high‐performance phenotype, while 20:5(n‐3) and 22:6(n‐3) diatom and flagellate markers, respectively, are more prominent in shallow low‐performance phenotype. Surprisingly, both energy stores and performance were higher in the deep phenotype, in spite of measured lower zooplankton availability. Essential FA concentrations were conserved across sites, likely reflecting required levels for coral functioning and survival. While the deep high‐performance phenotype met with these requirements, the low‐performance phenotype appeared to need more energy to maintain functionality in its highly variable environment, potentially causing intrinsic re‐allocations of energy and enrichment in certain essential markers (20:5(n‐3), 22:6(n‐3)). Our analysis highlights the biological and ecological insights that can be gained from FATM profiles in CWCs, but also cautions the reliability of FATM as diet tracers under limiting environmental conditions that may also be applicable to other marine organisms. Read the free Plain Language Summary for this article on the Journal blog.