Do constrained immigration rates and high β diversity explain contrasting productivity–diversity patterns measured at different scales?