Spectrographic analyses of songs

Using SonaGrams to characterize calling songs
During the first 40 years of T. J. Walker's studies of the songs of crickets and katydids (1955-1994), his principal means of analyzing these songs were two models of a device known as the Kay Sona-Graph. By a combination of vacuum-tube electronics and complex machinery, Sona-Graphs produced "SonaGrams" (=audiospectrograms) by using a fine, precisely controlled electrical arc to write on the light-gray surface of special SonaGram paper.

One of more SonaGrams were made of 1- or 2-second samples of thousands of tape recorded songs. The rate and temporal pattern of the cycles of wing movements that produced the recorded songs were then determined by counts and manual measurements along the horizontal axis of the SonaGram. Measurements along the vertical axis were used to quantify carrier frequency. These counts and measurements were time-consuming and limited to brief samples of the songs. Furthermore, monitoring for changes in the electronic and mechanical workings of the SonaGraph required time-consuming calibration procedures.

Digital technology now makes determining the physical features of recordings of calling songs more accurate and much faster than in the Sona-Graph era.

Graphed-data files
For many North American species of crickets and katydids, T. J. Walker and those working with him used SonaGrams to analyze calling songs from diverse geographic localities and at a range of temperatures. To make the results of these analyses easily accessible, the data will be entered into Excel worksheets and graphed to show the rate of wing cycling as a function of temperature. Access to the resulting "graphed-data files" (nnngd.xls) will be via links in SINA species accounts. In the initial stage of this project, completed in October 2003, graphed-data files were posted for 11 species of Nemobiinae (ground crickets), 11 species of Conocephalus (lesser meadow katydids), and 8 species of Neoconocephalus (common coneheads).

Interpretation of graphed-data files
Caged vs. free. Accurately measuring the ambient temperature of a calling caged cricket or katydid is easy. On the other hand, when an individual is calling while free in the outdoors, the temperature of its immediate surroundings is usually difficult and often impossible to determine. For example, a katydid calling 20 meters away in the top of a pine tree is unlikely to be experiencing the same temperature as a thermometer at a parabolic reflector that is facilitating the recording. Furthermore, the calling katydid may be warmed or cooled in the short term by breezes of varying temperatures. Similarly, a field cricket calling from the interface of soil and grass is likely, because of the thermal inertia of soil and the insulating effect of grass, to be at a temperature different from that of a thermometer a few inches above. Thus in weighing the significance of a record that creates an "outlier" on a graph of wingstroke rate vs. temperature, one should consider whether the recording was of a caged or free individual. The circumstance for any record can almost always be correctly judged by comparing the CollDate and TapeDate fields. It they are the same, the recording is of a free individual.

Another caveat relative to recordings of free individuals is that in most cases the identification is based only on the calling song itself. For some groups of species, the identification is made solely on the wingstroke rate. Thus an outlier could be the result of a misidentification (or what would have been an outlier may be concealed by a misidentification).

Same or different individual. Records that are known to be of the same individual calling on different occasions are identified as being different "takes" (a, b, c, etc.) of the same individual. Individuals are specified by a SpeciesNo (nnn) followed by an IndivNo (1, 2, 3, etc.).

More information about a record. Additional information about a record is contained in tblRecordedSongs in the GrylTett.mdb database. The best way to go from a record in a graphed-data file to a record in tblRecordedSongs is to use the combination of Year and CutNo.

Abbreviations. Abbreviations that appear in graphed-data files are explained in tblAbbreviations in the GrylTett.mdb database. Two of likely interest are OSU and FASTAX. OSU records are from recordings made at Ohio State University during 1955-1957. The original tapes are in the Borror Laboratory of Bioacoustics (https://blb.osu.edu/). Records that have FASTAX in the Condition field are at unknown temperatures because the ambient temperature could not be accurately measured as the caller was photographed under the bright light used with an ultra-high-speed FASTAX camera.

Digital analyses
As stated above, digital analyses of recorded songs should be more accurate than those made from SonaGrams. Those wishing to re-analyze the songs in the graphed-data files should be advised that most of the recordings in the Walker Tape Library (WTL) will be digitally archived by the Macauley Library of Natural Sounds (MLNS) (http://birds.cornell.edu/LNS/). Within about three years, all WTL recordings should be available in digital format from MLNS. In addition, a set of CD-ROMs of the digitized songs and many of the specimens that made the songs will be preserved in the Florida State Collection of Arthropods (http://www.fsca-dpi.org/). The MLNS status of WTL recordings can be determined by referring to the MLNS and LNSCatNo fields of tblRecordedSongs of the GrylTett.mdb database.