2017MNRAS.472.1346G

It has been suggested that the internal dynamics of dwarf spheroidal galaxies (dSphs) can be used to test whether or not ultralight axions with m_a ∼ 10^–22 eV are a preferred dark matter candidate. However, comparisons to theoretical predictions tend to be inconclusive for the simple reason that while most cosmological models consider only dark matter, one observes only baryons. Here, we use realistic kinematic mock data catalogues of Milky Way (MW) dSph's to show that the 'mass-anisotropy degeneracy' in the Jeans equations leads to biased bounds on the axion mass in galaxies with unknown dark matter halo profiles. In galaxies with multiple chemodynamical components, this bias can be partly removed by modelling the mass enclosed within each subpopulation. However, analysis of the mock data reveals that the least-biased constraints on the axion mass result from fitting the luminosity-averaged velocity dispersion of the individual chemodynamical components directly. Applying our analysis to two dSph's with reported stellar subcomponents, Fornax and Sculptor, and assuming that the halo profile has not been acted on by baryons, yields core radii r_c > 1.5 and 1.2 kpc, respectively, and m_a < 0.4 x 10^–22 eV at 97.5 per cent confidence. These bounds are in tension with the number of observed satellites derived from simple (but conservative) estimates of the subhalo mass function in MW-like galaxies. We discuss how baryonic feedback might affect our results, and the impact of such a small axion mass on the growth of structures in the Universe.