Observational Evidence of Velocity Anisotropy Assembly Bias in Galaxy Clusters
Cosmological simulations show that halo clustering depends on secondary internal properties beyond mass, an effect known as assembly bias. In particular, numerical models have shown that internal velocity anisotropy ($β$) is the halo property most tightly linked to secondary halo bias: at fixed mass, haloes with isotropic velocity profiles are more strongly clustered, whereas haloes in which radial motions dominate are more weakly clustered. Despite this robust theoretical prediction, direct observational detection of velocity anisotropy assembly bias has remained elusive. Here we present observational evidence for this effect using a sample of massive galaxy clusters from the eROSITA and DESI surveys. We measure the internal orbital anisotropy $β$ for 91 clusters via Jeans dynamical modelling with MAMPOSSt, considering three anisotropy profile models to ensure robustness. We divided the sample into two subpopulations: isotropic and radially anisotropic, verifying that they show no significant differences in halo mass. To quantify their large-scale clustering, we measured the projected cross-correlation function with We report evidence of velocity anisotropy assembly bias: clusters with isotropic internal kinematics exhibit stronger clustering than radially dominated ones, a segregation that is unlikely to be attributable to residual differences in cluster mass. This trend shows varying significance across three anisotropy profiles, from a mild tendency to a highly pronounced signal. Furthermore, the assembly bias remains distinct when restricting the sample to clusters consistently classified across all three models.