Collective control of distributed energy resources (DER)-such as photovoltaic (PV) inverters or battery storage-have the potential to provide regulation services to the bulk electric grid. While optimal power flow techniques may be used to coordinate DER for this purpose, these approaches typically rely on accurate network models and a large number of system measurements. In this paper, we consider an approach that alleviates these modeling and measurement requirements. Here, we consider a two-dimensional adaptive control scheme known as extremum seeking, or ES, to perform optimization without knowledge of a model of the distribution network. We apply this scheme to enable simultaneous feeder head active power and voltage magnitude reference tracking, as well as feeder voltage regulation. From the perspective of the transmission grid, this approach essentially transforms the distribution feeder into a controllable (P,V) bus. Simulation results confirm the ability of the approach to track substation real power and voltage reference signals while maintaining distribution system voltages within acceptable tolerances.