Diet alteration may lead to nutrient limitations even in the absence of food limitation, and this may affect physiological functions, including immunity. Nutrient limitations may also affect the maintenance of body mass and key life‐history events that may affect immune function. Yet, variation in immune function is largely attributed to energetic trade‐offs rather than specific nutrient constraints. To test the effect of diet on life‐history traits, we tested how diet composition affects innate immune function, body mass and moult separately and in combination with each other, and then used path analyses to generate hypotheses about the mechanistic connections between immunity and body mass under different diet compositions. We performed a balanced parallel and crossover design experiment with omnivorous common bulbuls Pycnonotus barbatus in out‐door aviaries in Nigeria. We fed 40 wild‐caught bulbuls ad libitum on fruits or invertebrates for 24 weeks, switching half of each group between treatments after 12 weeks. We assessed innate immune indices (haptoglobin, nitric oxide and ovotransferrin concentrations, and haemagglutination and haemolysis titres), body mass and primary moult, fortnightly. We simplified immune indices into three principal components (PCs), but we explored mechanistic connections between diet, body mass and each immune index separately. Fruit‐fed bulbuls had higher body mass, earlier moult and showed higher values for two of the three immune PCs compared to invertebrate‐fed bulbuls. These effects were reversed when we switched bulbuls between treatments after 12 weeks. Exploring the correlations between immune function, body mass and moult, showed that an increase in immune function was associated with a decrease in body mass and delayed moult in invertebrate‐fed bulbuls, while fruit‐fed bulbuls maintained body mass despite variation in immune function. Path analyses indicated that diet composition was most likely to affect body mass and immune indices directly and independently from each other. Only haptoglobin concentration was indirectly linked to diet composition via body mass. We demonstrated a causal effect of diet composition on innate immune function, body mass and moult: bulbuls were in a better condition when fed on fruits than invertebrates, confirming that innate immunity is nutrient specific. Our results are unique because they show a reversible effect of diet composition on wild adult birds whose immune systems are presumably fully developed and adapted to wild conditions—demonstrating a short‐term consequence of diet alteration on life‐history traits.

Ecological research is often hampered by the inability to quantify animal diets. Diet composition can be tracked through DNA metabarcoding of fecal samples, but whether (complex) diets can be quantitatively determined with metabarcoding is still debated and needs validation using free-living animals. This study validates that DNA metabarcoding of feces can retrieve actual ingested taxa, and most importantly, that read numbers retrieved from sequencing can also be used to quantify the relative biomass of dietary taxa. Validation was done with the hole-nesting insectivorous Pied Flycatcher whose diet was quantified using camera footage. Size-adjusted counts of food items delivered to nestlings were used as a proxy for provided biomass of prey orders and families, and subsequently, nestling feces were assessed through DNA metabarcoding. To explore potential effects of digestion, gizzard and lower intestine samples of freshly collected birds were subjected to DNA metabarcoding. For metabarcoding with Cytochrome Oxidase subunit I (COI), we modified published invertebrate COI primers LCO1490 and HCO1777, which reduced host reads to 0.03%, and amplified Arachnida DNA without significant changing the recovery of other arthropod taxa. DNA metabarcoding retrieved all commonly camera-recorded taxa.Overall, and in each replicate year (N = 3), the relative scaled biomass of prey taxa and COI read numbers correlated at R = .85 (95CI:0.68-0.94) at order level and at R = .75 (CI:0.67-0.82) at family level. Similarity in arthropod community composition between gizzard and intestines suggested limited digestive bias. This DNA metabarcoding validation demonstrates that quantitative analyses of arthropod diet is possible. We discuss the ecological applications for insectivorous birds.

Ecological research is often hampered by the inability to quantify animal diets. Large-scale changes in arthropod diversity, abundance and phenologies urge the need to understand the consequences for trophic interactions. Diet composition of insectivorous predators can be tracked through DNA metabarcoding of faecal samples, but to validate the quantitative accuracy of metabarcoding, validation using free-living animals for which their diet can be approximated, is needed.This validation study assesses the use of DNA metabarcoding in quantifying diets of an insectivorous songbird. Using camera footage, we documented food items delivered to nestling Pied Flycatchers Ficedula hypoleuca, and subsequently sequenced the Cytochrome Oxidase subunit I (COI) in their faeces. Our special interest was to retrieve the relative contribution of arthropod taxa with a PCR-based protocol.Assessment of taxonomic coverage of the invertebrate COI primers LCO1490 and HCO1777, previously applied in insectivorous songbirds, demonstrated that COI barcodes were predominantly assigned to arthropod taxa, however, substantial amounts of reads (2–60%) were assigned to flycatchers. Modified primers reduced vertebrate reads to 0.03%, and yielded more spider DNA without significant changing the recovery of other arthropod taxa.To explore digestive biases, contents of stomachs and lower intestines were compared in eight adult male flycatchers, victims of competitors for nest boxes. Similarity in arthropod community composition between stomach and intestines confirmed limited digestive bias.With nest box cameras in 2013, 2015 and 2016, size-adjusted counts of prey items fed to nestlings were recorded, to approximate provided biomass of arthropod orders and families which allowed comparison with abundance of COI barcode reads in nestling faeces. The relative abundances of size-adjusted prey counts and COI reads correlated at R = 0.85 (CI:0.68-0.94) at order level and at R=0.75 (CI:0.67-0.82) at family level. Each common order and all common taxa within Lepidoptera, Diptera and Coleoptera were retrieved in similar proportions, while the taxonomic resolution was higher in barcodes than in camera records.This DNA metabarcoding validation demonstrates that quantitative arthropod diet analyses is possible in songbirds. We discuss the ecological application of our approach in approximating the arthropod diets in insectivorous birds.

Ecological research is often hampered by the inability to quantify animal diets. Diet composition can be tracked through DNA metabarcoding of faecal samples, but whether (complex) diets can be quantitatively determined with metabarcoding is still debated and needs validation using free-living animals. This study validates that DNA metabarcoding of faeces can retrieve actual ingested taxa, and most importantly, that read numbers retrieved from sequencing can also be used to quantify relative abundances of dietary taxa. Validation was done with the hole-nesting insectivorous Pied Flycatcher whose diet was quantified using camera footage. Size-adjusted counts of food items delivered to nestlings were used to approximate provided biomass of prey orders and families and subsequently nestling faeces were assessed through DNA metabarcoding. To explore potential effects of digestion, stomach and lower intestine samples of freshly collected birds were subjected to DNA metabarcoding. For metabarcoding with Cytochrome Oxidase subunit I (COI), we modified published invertebrate COI primers LCO1490 and HCO1777, which reduced host reads to 0.03%, and amplified Arachnida DNA without significant changing the recovery of other arthropod taxa. DNA metabarcoding retrieved all commonly camera-recorded taxa. Overall, and in each replicate year (N = 3), the relative abundances of size-adjusted prey counts and COI read numbers correlated at R=0.85 (CI:0.68-0.94) at order level and at R=0.75 (CI:0.67-0.82) at family level. Similarity in arthropod community composition between stomach and intestines suggested limited digestive bias. This DNA metabarcoding validation demonstrates that quantitative analyses of arthropod diet is possible. We discuss the ecological applications for insectivorous birds.