Analytical error in stable isotope ecology

Abstract

The increasing popularity of stable isotope analysis (SIA) as an ecological research tool and the ease of automated analysis have created a knowledge gap between ecologists using SIA and the operators of isotope ratio mass spectrometry (IRMS) equipment. This has led to deterioration in the understanding of IRMS methodology and its proper dissemination in the ecological literature. Of 330 ecological research papers surveyed, 63 (19%) failed to report any form of analytical error associated with IRMS. Of the 267 papers that reported analytical error, there was considerable variation both in the terminology and approach used to quantify and describe error. Internal laboratory standards were often used to determine the analytical error associated with IRMS, so chosen because they are homogenous and have isotopic signatures that do not vary over time. We argue that true ecological samples collected in the field are complex bulk mixtures and often fail to adhere to these two criteria. Hence the analytical error associated with samples is potentially greater than that of standards. A set of standard data run over time with a precision typically reported in the ecological literature (1 standard deviation: 1SD = 0.26 per thousand) was simulated to determine the likelihood of spurious treatment effects depending on timing of analysis. There was a 90% likelihood of detecting a significant difference in the stable nitrogen ratio of a single sample (homogenized bovine liver) run in two time periods when n > 30. Minor protocol adjustments, including the submission of blind replicates by researchers, random assignment of sample repeats within a run by analytical labs, and reporting 1SD of a single sample analyzed both within and between runs, will only serve to strengthen the interpretation of true ecological processes by both researchers and reviewers.