Datasets used for this research.

The forensic application of stable oxygen isotope data from human tissues depends on naturally occurring isotopic variation in drinking water across geographic areas. One factor which complicates interpretation of forensic data is local variability: if a wide range of drinking water values is in a small geographic region it may be difficult to identify or rule out that region as a location of origin. We examine data from community collection programs documenting tap water isotope variation within 30 cities\developed areas throughout the United States. Isotopic variation within individual developed areas ranged widely, from essentially nil to greater than 9‰ (δ18O interdecile range). Many (14/30) of the study areas exhibited multi-modal isotope distributions, even in cases where the isotopic range was very small (e.g., <2.0‰), suggesting that the use of multiple, isotopically different sources was common. Most variation was attributed to differences in the source-region altitude or degree of evaporation for different water sources, and we observe limited variation in areas where contrasts in these variables are small. Variation tended to be largest in cities across the western USA. We explored correlations between the magnitude of local isotopic variation and climatic, environmental, and socioeconomic variables. We found that higher levels of variation occurred at sites where local water resources were likely to be more isotopically heterogeneous and in lower supply, consistent with the development of supply systems in these areas that access diverse and isotopically heterogeneous water resources. We also found that variation was positively correlated with larger developed areas, suggesting that pressures related to city growth may affect the degree to which infrastructure tapping diverse water resources is developed. Collectively, our results suggest that the influence of local tap water isotope heterogeneity on the precision of forensic geolocation is systematic and somewhat predictable, information to be factored into future applications.