Bulk tissue versus amino acid carbon stable isotopes to reveal the diet and basal resource use of an invasive rodent

Stable isotope analysis is widely used to reconstruct food webs,
characterize trophic relationships, and estimate dietary composition.
However, bulk tissue stable isotope analysis (BSIA) can yield diffuse or
imprecise estimates when dietary sources are isotopically similar. We
compared carbon stable isotope analysis of amino acids (δ13C-EAA values)
with BSIA (δ13C and δ15N values) to estimate diet composition and basal
resource use of non-native, invasive house mice (Mus musculus) on Sand
Island, part of a remote atoll in the north Pacific Ocean. We aimed to
assess whether δ13C-EAA values could improve source differentiation and
yield more precise diet estimates and to evaluate whether δ13C-EAA
fingerprints could provide insight into the origin and use of basal
resources by mice. We applied a Bayesian mixing model using δ13C-EAA
values from a subset of 10 mice and associated diet items, with sample
selection guided by existing BSIA and next-generation sequencing datasets.
Diet estimates from the δ13C-EAA model closely resembled those from the
BSIA-based model (n = 90): arthropods dominated the mouse diet, with
smaller contributions from seabirds and plants. Moreover, estimates from
the δ13C-EAA model were more consistent and precise than those from a BSIA
model using the same 10 individuals, underscoring the value of informed
sample selection. Although the use of δ13C-EAA values did not
substantially improve separation among isotopically similar sources, it
reinforced our previous findings. We also used δ13C-EAA fingerprinting to
reconstruct basal resource use. Mice obtained most of their essential
amino acids from aquatic basal sources—likely through direct consumption
of seabird tissues or indirect consumption of scavenging
arthropods—whereas terrestrial resources contributed far less. These
findings highlight the role of marine-derived nutrients in this food web
and suggest that mice may reroute energy and nutrients in the Sand Island
system by exploiting marine subsidies. Collectively, our results suggest
that δ13C-EAA values are most effective when paired with pre-existing
ecological knowledge, especially for omnivores in complex food webs.
Applied judiciously, this method may strengthen ecological inferences and
provide additional insight into impacts of invasive species—particularly
in vulnerable island ecosystems.