Abstract
Introduction Quantitative fatty acid signature analysis (QFASA) can provide species level diet estimates integrated over weeks to months, which are valuable for assessing health, ecological roles, and disturbance vulnerability. However, the approach has seen limited use in cetaceans. Calibration coefficients (CCs) have mainly been derived from non-cetacean mammals, best-fit QFASA model parameters are undefined, and the temporal integration of blubber fatty acids (FAs) remains poorly resolved. Method We used bottlenose dolphins ( Tursiops truncatus , n = 3, hereafter "dolphin") under professional care with known, varied diets to develop and evaluate species- and blubber layer-specific CCs and explore model performance under different parameter combinations. For each dolphin, we calculated CCs for the inner and outer blubber, compared these to published non-dolphin CCs, and evaluated QFASA-estimated diets across different FA sets, distance measures, CC sources (dolphin and non-dolphin), and FA integration periods. Model performance was assessed using prey distinctiveness, the percentage of predator FAs that fell outside prey ranges [predator-beyond-prey (PBP) values], and weighted error between estimates and the dolphins’ known diets. Results Dolphin CCs differed between the inner and outer blubber and from non-dolphin CCs for many FAs. Dolphin-specific, layer-matched CCs produced lower-error estimates and identified key prey species more accurately than non-dolphin CCs. Inner and outer blubber estimates were consistent with prey consumption integrated over weeks to months, supporting QFASA’s long-term nature. However, model performance was sensitive to FA set, distance measure, CC source, and dolphin diet complexity. In some parameter combinations, the augmented FA contributed a large portion of the model signal, reducing interpretability. This highlights the need for cautious parameter selection. Discussion These results provide the first layer-specific CCs for bottlenose dolphins and illustrate the utility and limitations of QFASA for cetacean diet estimation. We recommend that investigators use species- and layer-specific CCs where possible and consider prey distinctiveness, PBP values, and the augmented FA’s contribution when selecting model parameters. We also caution against over-interpreting best-fit parameter sets and diet estimates derived from small calibration datasets. To yield the most complete understanding of free-ranging cetacean diet, QFASA is best applied as one of several complementary methods rather than as a standalone approach.