Minmax solutions for underdetermined isotope mixing problems: Reply to Semmens et al. (2013)

Abstract

Stable isotope tracers are used to solve ecological problems, and sophisticated modelling approaches are now available to help interpret measured isotope data. However, models such as IsoSource and SIAR can go a step too far in producing mean and median values when there are too many sources and not enough tracers. The models solve these underdetermined problems by assuming that some of the many possible or feasible solutions are more likely than others. I promote the alternative idea that all feasible solutions are equally likely, and support use of minimum and maximum ('minmax') solutions as the appropriate way to view results from underdetermined systems.