In order to investigate whether different food sources were treated similarly or differently by a generalist species, we followed a simple experimental procedure (A), where eggs of a single founder population were separately developed on four fruit media and one control ‘German Food’ medium for 10 days and kept on the same fruit for seven days of adult life. Both adult flies and fruit purees were processed to obtain the controlled fly and fruit datasets, with respective ions represented by colored triangles and dots. These datasets were used to perform three analyses (B, D, F) linked to three diverging expectations under our competing hypotheses of ‘metabolic generalism’ and ‘multi-host metabolic specialism’ (C, E, G). First, the comparison of the relative sizes of intersections of the complete dataset (B) should indicate whether ions are mostly shared between intersections of large sizes such as all fruits, all flies, or all samples (C, left) or whether ions are mostly confined to intersections of small size such as discrete categories or fruit-fly pairs (C, right). Second, relationships between intersections between fruits on one hand and between flies on the other (D) should indicate whether common or rare fruit ions are commonly (E, left) or rarely (E, right) found in flies. Third, while a principal component analysis (PCA) of the complete dataset should separate fruit and flies on the first dimension due to their expected dissimilarities, its second dimension should indicate whether differently-fed flies regroup into a single cluster (F, left) or whether they cluster away from each other and close to their fruit on this axis (F, right). Regarding expectations (C, E, G), all left and right panels are consistent with the ‘metabolic generalism’ and ‘multi-host metabolic specialism’ hypotheses, respectively.