In this work we explore the application of data mining techniques to the problem of acoustic recognition of bird species. Most bird song analysis tools produce a large amount of spectral and temporal attributes from the acoustic signal. The identification of distinctive features has become critical in resource constrained applications such as habitat monitoring by sensor networks. Reducing computational requirements makes affordable to run a classifier on devices with power consumption constraints, such as nodes in a sensor network. Experimental results demonstrate that considerable dimensionality reduction can be achieved without significant loss in classification efficiency.