Impact of tidal environment on galaxy clustering in GAMA

We constrain models of the galaxy distribution in the cosmic web using data from the Galaxy and Mass Assembly (GAMA) survey. We model the redshift-space behaviour of the 2-point correlation function (2pcf) and the recently proposed Voronoi volume function (VVF) -- which includes information beyond 2-point statistics. We extend the standard halo model using extra satellite degrees of freedom and two assembly bias parameters, $\alpha_{\rm cen}$ and $\alpha_{\rm sat}$, which respectively correlate the occupation numbers of central and satellite galaxies with their host halo's tidal environment. We measure $\alpha_{\rm sat}=1.44^{+0.25}_{-0.43}$ and $\alpha_{\rm cen}=-0.79^{+0.29}_{-0.11}$ using a combination of 2pcf and VVF measurements, representing a detection of assembly bias at the 3.3$\sigma$ (2.4$\sigma$) significance level for satellite (central) galaxies. This result remains robust to possible anisotropies in the halo-centric distribution of satellites as well as technicalities of estimating the data covariance. We show that the growth rate ($f\sigma_8$) deduced using models with assembly bias is about 7\% (i.e. $1.5\sigma$) lower than if assembly bias is ignored. When projected onto the $\Omega_m$-$\sigma_8$ plane, the model constraints without assembly bias overlap with Planck expectations, while allowing assembly bias introduces significant tension with Planck, preferring either a lower $\Omega_m$ or a lower $\sigma_8$. Finally, we find that the all-galaxy weak lensing signal is unaffected by assembly bias, but the central and satellite sub-populations individually show significantly different signals in the presence of assembly bias. Our results illustrate the importance of accurately modelling galaxy formation for cosmological inference from future surveys.