We report that the density profiles of globular cluster (GC) systems in a sample of 17 early-type galaxies (ETGs) show breaks at the radii where the gravitational acceleration exerted by the stars equals the galactic acceleration scale $a_0$ known from the radial acceleration relation or MOND. The match with the other characteristic radii in the galaxy is not that close. We propose possible explanations in the frameworks of the $\Lambda$CDM model and MOND. We find tentative evidence that in the $\Lambda$CDM context, GCs reveal not only the masses of the dark halos through the richness of the GC systems but also the concentrations through the break radii of the GC systems.

(abridged) Context. Gravitational fields at the outskirts of early-type galaxies (ETGs) are difficult to constrain observationally. It thus remains poorly explored how well the LCDM and MOND hypotheses agree with ETGs. Aims. This led us to gather a large sample of ETGs and examine homogeneously which dark matter halos they occupy, whether the halos follow the theoretically predicted stellar-to-halo mass relation (SHMR) and the halo mass-concentration relation (HMCR), whether ETGs obey MOND and the radial acceleration relation (RAR) observed for late-type galaxies (LTGs), and finally whether LCDM or MOND perform better in ETGs. Methods. We employed Jeans analysis of radial velocities of globular clusters (GCs). We analysed nearly all ETGs having more than about 100 archival GC radial velocity measurements. The GC systems of our 17 ETGs extend mostly over ten effective radii. A LCDM simulation of GC formation helped us to interpret the results. Results. Successful LCDM fits are found for all galaxies, but compared to the theoretical HMCR and SHMR, the best-fit halos usually have concentrations that are too low and stellar masses that are too high for their masses. This might be because of tidal stripping of the halos or because ETGs and LTGs occupy different halos. Most galaxies can be fitted by the MOND models successfully as well, but for some of the galaxies, especially those in centers of galaxy clusters, the observed GC velocity dispersions are too high. This might be a manifestation of the additional dark matter that MOND requires in galaxy clusters. Additionally, we find many signs that the GC systems were perturbed by galaxy interactions. Formal statistical criteria prefer the best-fit LCDM models over the MOND models, but this might be due to the higher flexibility of the LCDM models. The MOND approach can predict the GC velocity dispersion profiles better.

Investigating the gravitational field in the early-type galaxies (ETGs, i.e. ellipticals and lenticulars) up to large radii is observationally difficult. It is questionable how the radial acceleration (RAR) in the ETGs looks like, i.e. the relation between the dynamically inferred gravitational acceleration and the acceleration expected from the distribution of the visible matter. This relation is nearly universal for the spiral galaxies, in agreement with the MOND modified dynamics paradigm. In this contribution, we investigate a sample of 15 ETGs. We extract their full kinematic profiles out to several effective radii from their globular cluster systems and estimate their gravitational field using the Jeans equation. We parametrize the gravitational field by that produced by the stars and a Navarro-Frenk-White DM halo. We find that only 4-5 of our ETGs follow the RAR for the spiral galaxies. All these galaxies are fast rotators, have disky isophotes, appear mostly very elongated and the have bluest colors in our sample. This suggests that they might be spiral galaxies which lost their gas. Our galaxies deviating from the RAR for the spirals either disprove MOND, contain unobserved matter, or indicate a flaw in the method.