Turbulence phenomena is present in a broad variety of flows: from geo- and astrophysical flows to industrial flows. One of its main characteristics is the efficiency to transport quantities that are dragged by the fluid, and the study of this phenomenon is relevant to accurately model spray dynamics in the atmosphere, cloud formation, or corrosion processes generated by droplet impacts in industrial systems. All of these examples involve inertial particles (i.e., particles denser than the fluid) interacting with and dragged by the turbulent flow. In this case, a phenomenon called preferential concentration has also been reported: inertial particles have the tendency to distribute inhomogeneously in space, forming clusters and depleted regions. Two possible mechanisms have been proposed to explain it: When the Stokes number is lower than unity, centrifugal forces are believed to be dominant, resulting in depletion of regions of high vorticity. For Stokes number larger than unity, the sweep-stick mechanism results in accumulation of particles in points with low Lagrangian acceleration. However, these arguments neglect the effect of gravity acting in the particles, a force that is relevant in many geophysical processes such as in droplet suspension and precipitation. For this case, it has been suggested that particles may stick preferentially in points where the Lagrangian acceleration equals the gravitational one, but few numerical studies have verified this prediction. In this work we focus on the study the dynamics of inertial particles under the effect of gravity in isotropic and homogeneous turbulence, using direct numerical simulations (DNS) in 512³ grids. We integrate 10⁶ particles with different Stokes numbers and under different gravitational accelerations. We consider conditions in which particles are suspended by the turbulence, and cases in which particles fall and reach a terminal velocity. Statistical studies are made of the fluid Lagrangian accelerations and vorticity at the points in which the particles are, as well as of the particles velocity and acceleration. Finally, using Voronoi diagrams we characterize the clustering level and statistical properties of the clusters, and explore the correlation between cluster volume, Lagrangian acceleration, and vorticity.