Abstract: Populations of many species of birds are declining at an alarmingly rapid rate, and we must assess these changes quickly and carefully if we are to avoid extinctions and detect further deterioration in the birds' environments. Yet, despite large-scale censussing over the past several decades, estimation of birds' populations remain problematic. The basic method of estimation is inexpensive but can be very inaccurate and is prone to several kinds of error. Recently, this method has been supplemented by a much more detailed but labor-intensive procedure that has its own set of biases. Both methods are useful and will continue to be used, but there is a need for a procedure that is accurate and inexpensive.

Recordings of birds' songs offers such a procedure. Recording birds' songs can be used to extract the "signatures" of individual singers. Identification of signatures uses temporal shifts in patterns of sound frequencies, and the necessary analyses can be done on personal computers. Dr. Smith will describe a program in which he and his colleagues have begun to investigate the specificity, repeatability, limits, and feasibility of this technique. The technique that they have developed is an efficient, teachable procedure and will enable the field work (and, perhaps, the initial analyses) to be done by a network of "citizen scientists".