All Xenopus species that have been examined express distinctive and species-typical vocal behaviors (Kobel et al., 1996). Calls are produced and responded to underwater. The most common call is the mating or advertisement call of the male given when alone, when with other males or when paired with females. Calls consist of trains of clicks, brief and noisy bursts of sound produced when the sound producing elements within the larynx move. The sound frequencies that make up these clicks differ somewhat by species and by sex. The most distinctive distinguishing feature of click trains, however, is their temporal pattern and intensity modulation (changes in click amplitude within a call). Click trains (trills) vary in length and in temporal pattern. The inter-trill interval between calls and the interclick interval within a call are also distinguishing features. Since the temporal patterns of the calls reflect a pattern generated within the central nervous system (Yamaguchi and Kelley, 2000), the way in which vocal patterns are produced by the brain must vary across species.
A note on technique: The figure above (modified from South African Frogs by Passmore and Carruthers) depicts sound spectrograms from three species of Xenopus: laevis, muelleri and gilii. The sounds were recorded using a hydrophone (an underwater microphone) and analyzed with a sonagraph. The result is a sound spectrogram, a representation of the presence of energy at different sound frequencies (0.5 to 3.0 kilohertz or 500 to 3000 cycles per second) as a function of time. As you can see, each click (each black trace) contains a range of sound frequencies. That range differs somewhat across species as does the temporal pattern, interclick interval and duration of of the click train or trill. Sounds can also be portrayed as oscillographs that depict sound intensity as a function of time. Oscillographs are particularly useful for comparing the timing of activity in parts of the nervous system to the timing of the sounds that are produced by neural activity.
"Presumably, early anuran vocalizations functioned primarily for the attraction of mates" (Duellman and Trueb, 1994, p. 107). If we accept this presumption, we can regard the divergence of advertisement calls in different species to be the result of a variety of selective pressures (see below). The dominant hypothesis for differences in advertisement calls of extant (present day) species is a species isolation mechanism (premating isolation) that operates via female choice. In a wide variety of species females chose the calls of conspecifics (same species) over those of heterospecifics (different species) and mate preferentially with the former. The argument is strengthened by the observation that production of inviable or infertile hybrids results in a severe loss of reproductive output; the pressure for females to chose male conspecifics is particularly acute since the number of eggs that a female can produce in her lifetime is much more limited than the number of sperm that a male can produce. Given that species isolation could be of current importance (especially when populations are sympatric or share a locale), how did differences in male songs come about in the first place? Since the temporal patterns that distinguish species-specific songs in Xenopus are produced by central nervous system circuitry, we can seek to determine how changes in pattern generation by the brain affect reproductive isolation and how isolation and changes in gene expression affect neural activity. The question is particularly interesting in Xenopus because of the hybrid hypothesis for speciation (and polyploidization). Interspecific hybrids are viable and they vocalize although they are usually infertile (some hybrid females can produce viable offspring when backcrossed to parental males; Kobel, 1996).
Another issue is the divergence of call types within a species and their functions in social communication. While the advertisement call is the most common anuran vocalization, most species also express a release call given in response to amplexus (clasping) by a male. Both sexes can make release calls and it is possible that the release call rather than the advertisement call is ancestral. Advertisement calls can play multiple roles (see Wells, 1977) both in attracting females and in establishing male calling territories. Another common call type is the encounter call, produced when males come into close proximity. In addition to release calls, females in some frog species produce a "reciprocation" call in response to male calling (Heinzmann,1970; Dixon, 1957). The six call types we have identified in X. laevis seem to represent one of the more complex anuran vocal repertoires. We do not know whether this complexity is specific to laevis or common to all Xenopodinae.
How did the calls of the various Xenopus species come to differ (within a species and across species) and what is the relation between this divergence and the divergence of species? Attempts to use communication signals as characters in a phylogeny (map or history of species divergence) have had a mixed record of success (Duellman, 1970; Nelson, 1973). The very strong influence of environmental conditions on signal propagation can result in remarkable similarity of calls (convergence) from very different phyletic groups. For example, several species produce calls while floating at the surface after torrential rains. Despite marked differences in phyletic origins, the calls of a North American pelobatid, a South American leptodactylid and an Australian myobactrachid are structurally very similar (cited Duellman and Trueb, 1994, p. 106), On the other hand, acoustic competition between different species can drive divergence that results in reduced interference; the combination of species present at a particularly favorable breeding site (or environmental conditions that influence sound transmission) can influence call types without regard to the phylogeny of the species under study.
When recently diverged species occupy very similar habitats the form of the communication signal can be an informative phylogenetic character. Hopkins and his colleagues, for example, have shown that the presence of an electric organ discharge prepulse (produced by a separate electric organ) distinguishes a sub-group of the Rhamphichtyoid clade of electric fish from South America (Hopkins, 1999).
Species and sub-species differences in anuran vocalizations are believed to have a genetic basis. The primary evidence in support of this hypothesis is that hybrid calls are intermediate between the parental species (but see Gehrhardt, 1974). The calls themselves should reflect changes in gene expression driven by mutation, by geographic isolation and by selection. Natural selection, for example, should act strongly on the localizability of acoustic signals and sexual selection should contribute to competition-driven variation in male calls.
So what are the genes that influence calls? One approach to this question is to mutagenize frogs and screen for call variants; I do not know of an example of this approach - which has been applied to the courtship behavior of genetically tractable organisms such as Drosophila (Eberl et al.,1997; Megighi et al., 2001) - in frogs. Another is to take closely related species and attempt to correlate changes in gene expression (using, for example, DNA microarrays) with differences in vocal attributes. Both of these approaches are "black box" in the sense that they do not require any a priori knowledge of how vocalizations are actually produced by the organisms. The advantage of a back box approach is that results are not influenced by preexisting notions of which genes should be important. These approaches may be the only way in which we can surprise ourselves, come up with novel insights into mechanism. The disadvantage is that, given the complexity of assembling sensory, neural and muscular into a functional system that produces songs, we might expect that a very large number of genes would be involved and that the contributions of some of these would be trivial or uninformative. For example, deaf male frogs would not produce the answer call. Loss of function androgen receptor mutants would not produce advertisement calls because all would be phenotypically female. While this latter mutant would be very useful in terms of understanding sexual differentiation, it would be less informative in terms of the evolution of call structure or the mechanism of call production itself.
The second approach is the candidate gene approach. This approach relies on prior knowledge of how the character under study develops or functions, on how specific proteins contribute to development and adult function and how those proteins are coded for at the genomic level. For example, we know that the CNS is responsible for producing the different temporal patterns of Xenopus songs and that these patterns differ in males and females (Yamaguchi and Kelley, 2000). The complement of ion channels expressed by vocal motor neurons also differs dramatically in the sexes (Yamaguchi et al., 2000). An example of the candidate gene approach would be to interfere, within vocal motor neurons, with an ion channel that is sexually differentiated in its expression and determine whether the interference produces changes in vocal patterns.
Experimental tools for both of these approaches have been developed for Xenopus because of its importance for molecular studies of development. Transgenic Xenopus can be produced using a method developed by Kris Kroll and Enrique Amaya (1996) that involves incubating sperm nuclei with DNA (typically tagged with sequences coding for green fluorescent protein or GFP). Effective integration into the germ line produces transgenic lines. Genes can be expressed in the wrong place or at the wrong time using appropriate promoters. Gene function can be knocked down via competition with constructs coding for non-functional variants of the protein of interest. This latter approach works best when the protein of interest must associate (dimerize) with a partner to be effective (since the result phenocopies certain genetic manipulations it is called a dominant negative approach). Transgenic approaches have been used extensively in laevis and, in theory, should be applicable to any Xenopus. Other methods (i.e. transfection with lentiviruses; Lois et al., 2002) could, in theory, also be used to generate transgenic Xenopus.
Mutational analysis is another story because it requires backcrosses and is thus highly dependent on generation time. In laevis it can take 2 years for a female to reach reproductive maturity. This characteristic, together with the pseudotetraploid nature of the genome, mitigates against analysis using mutagenesis in laevis. X. tropicalis on the other hand has a much shorter generation time (as short as 3 months) and is diploid. These two characteristics have led to the adoption of this species by frog developmental biologists (the tropicalis project) and to plans for sequencing its genome, generating ESTs, microarray chips( http://www.xenbase.org/). The project also includes mutational analyses (chemical mutagenesis and insertional mutagenesis using transposons, Amaya et al., 2002).
Given our interest in evolution of vocal communication we have also begun investigating tropicalis vocal behaviors in the context of a molecular phylogetic analysis of the Xenopodinae (Tobias et al., 2002). From our point of view the two main disadvantages of tropicalis are that it is a more fragile species and that its vocal repertoire has not been examined in detail. Using our past work on laevis (see above) as a template, we are gathering data on calls and on vocal physiology. We have established a breeding colony and an effective, low cost husbandry paradigm and have collected a related population from Gabon in Equatorial Africa for molecular and vocal comparisons.