Application-specific data aggregation can play a significant role in energy-efficient operation of wireless sensor networks. Existing aggregation techniques rely heavily on the routing protocol to build shortest paths to route node measurements to the base station and are limited in the types of supported queries. We propose an aggregation scheme that utilizes the inherent information gradients present in the network. The query is directed to the source of information, resulting in better load sharing in the network. We support a variety of queries ranging from simple maximum, minimum or average of the readings of sensor nodes to more complex quantile queries such as k highest values or k^th highest value through a generic query algorithm. The query algorithm shifts the computation to the querying agent, thus eliminating any in-network aggregation.