Using Bayesion mixing models on mean-centered δ¹³C values of essential amino acids (eAA) extracted from Drosophila melanogaster flies and ontogenetic environments reared in a semi-natural experimental set-up, basal resource eAA contribution to fly tissue is estimated. Experiments were set up in the laboratory facilities of University of Bremen, Germany in October 2021; amino acid extraction and δ¹³C measurement were executed at the University of Göttingen, Germany and the Kompetenzzentrum Stabile Isotope, Göttingen, Germany. Samples were collected to deduce fly eAA origin-bacterial, fungal or plant derived- acquired during fly development under five different symbiont-environment conditions. Samples of ontogenetic environment reflecting resources available prior to and after fly development are included. In an experimental set-up single initially axenic first-instar larvae were introduced to semi-natural conditions using 1 mL of fruit puree inoculated with 50 µL of fly fecal matter suspension collected from 30 female flies maintained in microcosms (see Riedel and Rohlfs (in preparation) for set-up of microcosms and Riedel et al. (preprint, https://doi.org/10.1101/2025.10.14.682149) for fecal matter inoculation); or 1 mL of artificial lab diet. After development to adult flies and eclosion from the pupal case, flies were collected and dried. The experimental units in which flies had developed in, were frozen along samples of fresh substrate (not inoculated with fecal matter suspension). Fly and respective samples of ontogenetic environments post eclosion were pooled using a randomized stratified approach to achieve a minimal collective fly weight of 2 g per sample. For five treatment groups five samples were pooled (nFly = 25) and the respective ontogenetic environments were pooled accordingly (nontoenv = 25). Substrate samples were dehydrated first in a drying oven then in a lyophilizer and then homogenized alongside fly samples. Amino acids were extracted as described in Larsen et al. (2016) including three external standards and nor-leucine as internal standard. Extracted amino acid mixes were then separated and analyzed using mass spectrometry following (Pollierer et al., 2020; also described in Riedel et al., preprint, https://doi.org/10.1101/2025.10.14.682149). The triplicate measurements were aligned to the external standard and extracted measurements were inspected visually and manually corrected if necessary. Mean δ¹³C isotope values were corrected for carbon added during the derivatization of samples following O'Brien et al. (2002). Then implementing Bayesian mixing models from the MixSIAR package (Stock et al., 2018) in an R environment, δ¹³C isotope values of eAA (biotracers), isoleucine, leucine, methionine, phenylalanine, threonine, and tyrosine were centered to the mean, both of samples (mixture) as well as training data (source: Larsen et al., 2009; 2016; Pollierer et al., 2020) We used sample identifiers and treatment combination as factors and run the model at "short" length. After model convergence, the summary output was extracted, and metadata appended.

Funded by the University of Bremen in cooperation with the AG Scheu at the University of Göttingen (Georg-August Universität)