2013A&A...553A..94L

A number of large spectroscopic surveys of stars in the Milky Way are under way or are being planned. In this context it is important to discuss the extent to which elemental abundances can be used as discriminators between different (known and unknown) stellar populations in the Milky Way. We aim to establish the requirements in terms of precision in elemental abundances, as derived from spectroscopic surveys of the Milky Way's stellar populations, in order to detect interesting substructures in elemental abundance space. We used Monte Carlo simulations to examine under which conditions substructures in elemental abundance space can realistically be detected. We present a simple relation between the minimum number of stars needed to detect a given substructure and the precision of the measurements. The results are in agreement with recent small- and large-scale studies, with high and low precision, respectively. Large-number statistics cannot fully compensate for low precision in the abundance measurements. Each survey should carefully evaluate what the main science drivers are for the survey and ensure that the chosen observational strategy will result in the precision necessary to answer the questions posed.