Independent observations is a key assumption for many statistical methods, including occupancy models. People are often concerned about the validity of this assumption when the presence/absence of the species is spatially correlated amongst sampling units. Such correlation is likely a reality in many applications at certain spatial scales.
In this video Darryl explains how the existence of spatial correlation does not necessarily violate the independence assumption, as the assumption is not only about the biological situation, but also about how the population is sampled.