Colony-specific trace-element fingerprints in Atlantic puffin Fratercula arctica faeces across Iceland and associations with diet and breeding performance.

Trace elements include essential micronutrients and potentially toxic contaminants and can vary spatially in marine food webs. We evaluated whether fresh Atlantic puffin Fratercula arctica droppings resolve colony-scale multi-element variation across Iceland and explored associations with chick-provisioning diet and breeding performance. During 21-28 July 2018, 154 fresh droppings, each collected immediately after a directly observed defecation from a different puffin, were obtained from nine colonies and analysed for As, Co, Cr, Mo, Ni, Pb and Se. Among-colony differences were assessed with Kruskal-Wallis tests and Dunn comparisons, multivariate structure with principal component analysis (PCA), and colony discrimination with repeated stratified random-forest cross-validation, naive baselines and a full-pipeline label-permutation test. Colony-level diet and breeding analyses (n = 9) were treated as exploratory. All seven elements differed among colonies, with the largest effect sizes for Mo and Cr. PCA PC1 and PC2 explained 44.6% and 17.9% of variance, respectively. Random-forest discrimination was moderate but clearly above null expectation (balanced accuracy = 0.449 ± 0.022, label-permutation null mean = 0.113, 99 permutations, p = 0.010). The strongest element-performance association was Se with burrow occupancy (ρ = 0.883, exact p = 0.0031), but no association remained significant after false-discovery-rate correction across 28 tests (Se-BO qBH = 0.086). Diet composition differed markedly among colonies but showed weak multivariate concordance with elemental profiles. Fresh droppings therefore contained detectable, internally cross-validated colony-associated multi-element structure within the 2018 breeding season. Multi-year and external validation are required before these profiles can be used for monitoring or source attribution.