Testing Fatty Acid Based Diet Estimation Models using Diets Composed of Natural Prey Items

Briefly, We sought to test if fatty acids can be used to quantify diets of fish when the diets are composed of mixtures of prey species rather than manufactured feed ingredients. We created diets by mixing freeze dried bloodworms, daphnia, and shrimp as well as diets of each prey item by itself and fed them to juvenile lake trout. While the seven diet treatments were visible with both unconstrained (nMDS) and constrained (LDA) visualizations of lake trout fatty acid profiles at weeks 4, 8, and 12, quantitative models such as FASTAR, mixSIAR, and QFASA all returned diet estimates with errors ranging from 7 - 35% per diet item per fish. It is difficult to prescribe if the high error rates are due to malnutrition, genetics, or other un-accounted for variables in the dataset but we do note that greater accuracy did not correlate with length or mass gained by individual fish nor their lipid contents.

Excel file includes: 1)Metadata on experimental setup as well as fatty acid quantification, 2) data on the survival of fish in each treatment, 3) fatty acids expressed in milligram form, and 4) fatty acids expressed as proportions of the total fatty acids quantified.

We refer users to our manuscript for a full explanation of the methods and results of this study.
Methods from this paper are also available within the excel file of the data.

本研究旨在探究，当鱼类饲料由猎物种类的混合物而非工业制成品组成时，是否能够利用脂肪酸对饲料进行量化。为此，我们通过混合冻干血虫、挠足类及虾等饵料，以及单独每种饵料的饲料，喂养幼年湖鳟。在4周、8周和12周时，通过无约束（nMDS）和约束（LDA）可视化湖鳟脂肪酸谱，七种饲料处理均可见。然而，定量模型如FASTAR、mixSIAR和QFASA在计算饲料估计时，每种饲料每条鱼的误差范围在7%至35%之间。由于难以判断高误差率是由营养不良、遗传因素或其他未考虑的数据集变量所致，但我们注意到，更高的准确性并未与个体鱼的长度或体重增加，以及其脂质含量相关。Excel文件包括以下内容：1）关于实验设置及脂肪酸定量的元数据，2）各处理组中鱼类存活情况的数据，3）以毫克为单位的脂肪酸含量，以及4）占总脂肪酸量比例的脂肪酸含量。我们建议用户参阅我们的论文以获取本研究的详细方法和结果。此外，本文的研究方法也包含在数据集的Excel文件中。