Evolution of advertisement calls in African clawed frogs

Abstract

For most frogs, advertisement calls are essential for reproductive success, conveying information on species identity, male quality, sexual state and location. While the evolutionary divergence of call characters has been examined in a number of species, the relative impacts of genetic drift or natural and sexual selection remain unclear. Insights into the evolutionary trajectory of vocal signals can be gained by examining how advertisement calls vary in a phylogenetic context. Evolution by genetic drift would be supported if more closely related species express more similar songs. Conversely, a poor correlation between evolutionary history and song expression would suggest evolution shaped by natural or sexual selection. Here, we measure seven song characters in 20 described and two undescribed species of African clawed frogs (genera Xenopus and Silurana) and four populations of X. laevis. We identify three call types - click, burst and trill - that can be distinguished by click number, call rate and intensity modulation. A fourth type is biphasic, consisting of two of the above. Call types vary in complexity from the simplest, a click, to the most complex, a biphasic call. Maximum parsimony analysis of variation in call type suggests that the ancestral type was of intermediate complexity. Each call type evolved independently more than once and call type is typically not shared by closely related species. These results indicate that call type is homoplasious and has low phylogenetic signal. We conclude that the evolution of call type is not due to genetic drift, but is under selective pressure.

Keywords: Silurana; Xenopus; call types; phylogenetic signal; vocal behaviour.

Figure 1

The units of advertisement calling in African clawed frogs. This example is taken from X. clivii. (A) Click: the smallest unit of sound. Calibration bar = 10 ms. (B) SVU: a repeating pattern of clicks, equivalent to one advertisement call. Here, one SVU is composed of 3 clicks, increasing in amplitude throughout the SVU. The click shown in A is the middle click in B. An asterisk (*) indicates each click and was used to determine click number. Line labeled ICI indicates the time interval from the beginning of one click to the beginning of the next. DF1 and DF2 are indicated on the spectrogram (B, right). Amplitude vs. time (ms) is shown on the left and frequency vs time is shown on the right. Calibration bar = 10 ms. (C) Bout: a bout contains a number of SVUs occurring at regular intervals. Here, a bout contains 3 SVUs; the middle SVU is that shown in B. Line labeled iSVUi indicates the time interval from the beginning of one SVU to the beginning of the next. Amplitude vs time (s). Calibration bar = 100 ms.

Figure 2

Advertisement calls and molecular phylogeny of African clawed frogs. Only species recorded from in this study are shown. A waveform of one SVU from each species is indicated to the right of the species name (or location, for the four populations of X. laevis). Numbers in parentheses refer to chromosome number (for Silurana 20 = diploid, 40 = tetraploid; for Xenopus 36 = tetraploid, 72 = octoploid and 108 = dodecaploid). Calibration bar = 50 ms.

Figure 3

Advertisement calls are categorized into three call types — click (left), burst (middle) and trill (right). (A) Spectrogram (top) and waveform (bottom) are shown for one SVU of each call type. Calibration bars: click and burst = 50 ms, trill = 200 ms. (B) A bout of calling from which the SVU in A was taken. Calibration bars: click = 200 ms, burst = 250 ms, trill = 1 s.

Figure 4

Silurana and Xenopus species are distinguished by spectral profile. Silurana species (left; one SVU from S. tropicalis) have one low frequency dominant peak (approx. 500 Hz), while Xenopus species (right; one SVU from X. andrei) have two higher frequency dominant peaks (approx. 2.0 and approx. 2.7 kHz). Top traces indicate spectrograms (frequency vs. time); bottom traces indicate waveforms (amplitude vs. time). Calibration bars: 500 ms (left), 100 ms (right).

Figure 5

Single phase calls have one and biphasic calls have two temporal patterns. A representative single phase call from X. gilli (left) and a representative biphasic call from X. laevis South Africa (right). Top traces are spectrograms (frequency vs. time); bottom traces are waveforms (amplitude vs. time). Calibration bars: 50 ms (left), 100 ms (right).

Figure 6

The temporal pattern of biphasic calls varies by species. (A) Representative SVUs from three biphasic callers. (Left) X. itombwensis calls consist of a rapid trill followed by a slower burst (unlike most bursts, there is no intensity modulation for this slow burst). (Middle) X. laevis Congo consists of a burst (4 clicks) followed by a single click after a longer pause. (Right) X. laevis Malawi consists of a rapid trill followed by a slower trill. Calibration bar = 100 ms. (B) Portion of a bout from X. laevis Malawi illustrating the characteristic regular pattern of SVUs, here alternating fast and slow trills. Calibration bar = 500 ms. (C) SVU of S. epitropicalis. The most common pattern of calling for this species is that shown here, a long rapid trill followed by a bout of bursts. Calibration bar = 500 ms. (D) A bout of calling from S. epitropicalis illustrating the irregular pattern of short to long trills. Calibration bar = 2.5 s.

Figure 7

Call types mapped onto a molecular phylogenetic tree for African clawed frogs. All known species, including museum samples and one extant species not recorded (gray lines) are shown. Predicted ancestral species with unknown same ploidy descendants are indicated by crosses. Call types are colour-coded as indicated. Each call type has evolved at least twice. Maximum parsimony analyses suggest that the ancestral call type was a burst, a call of intermediate complexity. This figure is published in colour in the online edition of this journal, which can be accessed via http://www.brill.nl/beh