A global estimator of C and N isotope baselines for fresh waters

Abstract Baselines are the pebbles in the shoes of isotope ecologists. The extreme variability of the isotope composition of resources at the base of food webs governs the spatial differences of consumers' isotope composition, so that isotope‐inferred trophic properties can be compared across ecosystems only after correction for baseline effects. However, acquiring comparable and reproducible isotopic baselines in different lakes and rivers has been so challenging that many isotope datasets lack baseline values. Global estimates of C and N isotopic baselines would considerably expand the scope of large‐scale isotope analyses in ecology. Cross‐referencing the global freshwater isotope database ISOFRESH (>800 sites across five continents) with a set of environmental attributes describing hydrology, physiography, climate, land use, soils and anthropogenic influences, we built data‐driven models that predict C and N stable isotope compositions of benthic and pelagic/open‐water baseline organisms for lakes and rivers with an error within 11%–13% of the overall range of values observed worldwide. We then applied the models globally to predict spatial patterns in isotope baselines. We showed simulated baselines accurately preserve patterns in across‐site variability for most of the common isotope‐derived trophic metrics computed for freshwater fish at the population and community levels. We conclude with guidance on the best use of such predictive baseline models, highlighting their usefulness for filling in gaps in meta‐ecological analyses that test regional or global drivers of food web structure, but caution against substituting them for measured values in local‐scale studies.