A Comparison of the Vocal Repertoires
of Willow Flycatcher (Empidonax traillii)
and Alder Flycatcher (Empidonax alnorum)

D. Archibald McCallum, Ph.D.
Applied Bioacoustics
Eugene, Oregon


Skip This Introduction

Willow Flycatcher (Empidonax traillii) and Alder Flycatcher (Empidonax alnorum) are sibling species that were not distinguished as separate species by the A.O.U. (1973) until the research of Robert Carrington Stein (1958, 1963) showed that two distinguishable song-types, called "Fitzbew" and "Feebeeo," were given respectively by individuals of two distinct biological species. Peterson (1987) recounts how, before Stein's work, he and others began to appreciate the differences as their travels put them in contact with both song-forms. Lowther (1999) describes the difficulties in assigning scientific names to the song-forms, once they were recognized as distinct species. Research subsequent to Stein's has abundantly confirmed and extended his findings. Kroodsma's (1984, 2005) finding that the two species do not learn their songs, but instead develop normal songs regardless of early experience, secured the legitimacy of basing taxonomic decisions on vocalizations. The large body of research on these two species, including the endangered "Southwestern Willow Flycatcher" (E. t. extimus), is synopsized in the BNA accounts for the two species, Lowther (1999) and Sedgwick (2000). They are often referred to collectively as the "Traill's Flycatcher Complex," after one of several pre-split English names.

Despite being very difficult to distinguish visually, the species are readily distinguished by their vocalizations. Most of the main elements in their vocal repertoires are diagnostic. In this document I name and present examples of all these main elements, describe the diagnostic characteristics of each, and explain how they differ from similar sounds. I hope studying the samples and the comparisons presented here will help observers prepare for going into the field, and lead to more accurate and confident on-the-spot identifications. But, at times when, or at places where, the local status of either species is unclear, it would be a good idea to record vocalizing "Traill's Flycatchers," and compare the recorded sounds to the samples presented here. For almost all of the named sounds in the repertoires of these species, geographic and individual variation are insignificant compared to interspecific differences, and these sounds should be sufficiently representative to allow identification to species throughout the range of both species.

Willow Flycatcher is polytypic, with authorities recognizing four or five subspecies (Sedgwick 2000). Southwestern Willow Flycatcher (E. t. extimus) Fitzbew songs (see below for repertoire elements) are significantly different from those of neighboring E. t. adastus (Sedgwick 2000, 2001). Further, eastern Willow Flycatcher (E. t. traillii and E. t. campestris) songs appear to differ consistently, albeit slightly, from the two more northerly subspecies in the West, E. t. brewsteri and E. t. adastus (D. A. McCallum, unpubl. data). My goal for a future version is to document this variation, making it possible, perhaps, to identify a sample to one of three subspecies groups in the species. For now, identification to species is the limit.

Complete repertoires of main elements are presented in two formats. The Reference Page is more exhaustive and descriptive. You can follow links from the Reference Page to the Comparison Page, which presents side-by-side comparison of any two sounds in the repertoires of the two species. In both formats, the same samples are inter-leaved vertically according to acoustic similarity. The arrangement is not meant to imply functional equivalency or evolutionary homology, although comments on both function and homology are included in the text. The Traill's complex has the "roughest" sounds in the entire genus Empidonax. I use "rough" as a general descriptor of sounds that have a "saw-tooth" spectral contour, a rapid and regular oscillation of pitch over a broad frequency range. The first 13 sounds in the reference section all have "rough" parts. The final eight sounds do not. These are "smooth" or "clean" sounding.

Each sound in this document is represented by a half-wave plot, a spectrogram, and text. Half-wave plots display practically all the information displayed by a more conventional oscillogram, with twice the resolution. I use half-wave plots on the assumption that they more readily evoke time-varying loudness for those not accustomed to extracting a sense of loudness from the alternating positive and negative amplitude values of a conventional oscillogram. The wav files have been standardized to a maximum amplitude of 10 volts, which makes it possible to hear the fainter parts of each sound. In most cases the wav files presented for listening have not been high-pass filtered (which filters out the low-frequency noise, such as wind). The clip sounds more "natural" with all frequencies retained. But, I did filter the samples before preparing the spectrograms, and in some cases I have digitally replaced ambient sounds with background noise from the same recording session. I performed these operations to make the spectrograms cleaner. An unavoidable consequence is that the spectrogram does not represent all the background noise you hear on the sound clip.

Each call-type is referred to by an onomatopoeic (or, in one case, acoustically descriptive) name. Names originally intended to describe the aural qualities of a sound, though often failures in that regard, are preferable to functional names because the same sound often has more than one function (i.e., is used in more than one context). I use the name that I judge to have the greatest familiarity, regardless of its success at conveying the aural quality of the sound. In the case of Willow and Alder Flycatchers, Stein (1963) provided useful and accurate spectrographic "descriptions" (equivalent to the naming of taxa) along with onomatopoeic "names" for most call-types. They have been used so widely that it seems best to stick with them. I have revised Stein's name "Pit" to "Pip," in line with widespread practice, and have made a few other minor adjustments, detailed below. Sedgwick (2000) provided further detail on the repertoire of the Willow Flycatcher, with comments on similar sounds of the Alder Flycatcher. He changed two of Stein's names. I comment on these changes. Hereafter, WIFL = Willow Flycatcher, ALFL = Alder Flycatcher.

Text includes recording data, the significance of the call-type for identification of the two species, comments on the spectrographic distinguishing features of the call-type, similar sounds, and common combinations including the focal sound. The generalizations I offer are based on spectrographic and aural study of representative samples from throughout the breeding range of the two species, and include my own recordings, the Borror Laboratory of Bioacoustics online catalogue, commercially-distributed samplers, and private recordings graciously provided by recordists.

All sounds presented here are copyrighted, and may not be published in any way (including on the WorldWide Web) or distributed for commercial gain, without the express written consent of the copyright holder. The Copyright holder is indicated in the figure legend of the sound. The text and figures of this document are copyrighted (2007, 2008) by Dougald Archibald McCallum, and may not be published in any way (including on the WorldWide Web) or distributed for commercial gain, without the express written consent of Dougald Archibald McCallum.

Technical Data: For digitization, audio tapes were played back on an Onkyo TA-RW400 stereo deck, shown through spectrographic analysis of simulated frequency tones to have very little harmonic distortion. Audio on video digital- 8 tapes, although recorded in digital format, was played back on a Sony DCR- TRV320 or DCR-TRV340 camcorder, i.e., converted to analogue. Analogue output, from the line out jack in the case of audiotape, and from the headphone jack in the case of videotape, was digitized at a sample rate of 50,000 points per second with a National Instruments DAQCard 6062E analogue-to-digital acquisition card deployed in a Toshiba 2595CDT computer, controlled by NIDisk TM software developed and licensed by Engineering Design, Berkeley, CA. Spectrograms were produced with SIGNAL TM sound analysis software, also from Engineering Design.

All spectrograms presented here were produced with 256-point FFTs (Fast Fourier Transforms), which yields complementary (Beecher 1988) frequency and time resolution of 195.3 Hz and 5.1 msec. Granularity of spectrograms was reduced by using 2000 FFTs per sound segment. Additionally, as with all digital spectrograms, those produced by SIGNAL are smoothed with linear interpolation. All spectrograms were output as bitmap files and transformed to jpeg format with Sony Picture Gear 4.1 Lite software. The bitmap files used for this paper were 600 x 400 pixels in size, resulting in approximately 26 Hz per pixel and 1.3 msec per pixel. The granularity of the graphic therefore does not obscure the level of detail intrinsic to the digital spectrogram.


1. Get started by clicking one of the links below. On your first time through, start with the Reference Page to see how the sounds are organized. For side-by-side comparison (requires a screen 1200 pixels wide for full display), either click the link below, or follow the link associated with each sound sample in the Reference Page.

2. NAVIGATION. Once your have entered the Reference Page, get started by either scrolling down until you reach the first spectrogram, or click any of the links in the Table of Contents.

3. Click anywhere in the spectrogram to hear the sound. You can replay the same sound by clicking the "Play" arrow on your sound player, or by clicking again in the spectrogram.

The sounds will play in the default player on your system. In Windows, the sound player is specified in the File Types tab of Folder Options. Most sound players will take up a large part of your display; but, you can reset display options within the sound player so it will not monopolize screen space. Good luck, the methods for adjusting the sound player displays vary among applications. I would be interested in feedback on technical problems and solutions you come up with. I can't provide technical support for each user personally, but have compiled some hints on sound player problems, which you may consult and follow at your own risk.

4. You can open a separate, free-floating window for each sound by clicking the link "Open Spectrogram in New Window." You will have to close these windows manually, by clicking the "X" in the upper righthand corner.

5. You can play the sound at half-speed by clicking the "Half Speed" link. Remember that halving the speed of the sound also halves its frequency (pitch). The main purpose of the half-speed sounds is to help you resolve the details of the natural sound. After listening to a half-speed sound several times, go back to the full-speed sound and check whether you hear more detail. You also have the option of listening to tenth-speed sounds by clicking the "Tenth Speed" link. These sounds are so low that they sound grouse-like. They have an eerie quality, but at this speed it is possible to resolve the short note at the end of ALFL Feebeeo, and to distinguish the two short notes at the beginning of P2 of WIFL Fitzbew from the one short note at the beginning of P2 of WIFL Fizzbew. At this speed you can also appreciate the difference between a trill ("Fizz" of WIFL Fizzbew) and a buzz ("Fee" of ALFL Feebeeo). Bear in mind that these small birds probably hear much more detail than we do in their rapidly-uttered trills and buzzes. In general, despite major exceptions, the smaller the singer, the higher the pitch of the sound, and also the faster the modulation (speed of trills and buzzes). These correlate with other size-scaled physiological rates. If you've listened to a speeded-up humpback whale song you were probably impressed with how much it sounds like birdsong. More than likely, most animals perceive about the same level of detail in their communication signals.

If you would like to document technical problems with this site, please email me.

Reference Page    Comparison Page