In September of 1992 I met Mike Picker from the University of Cape Town (at the meeting that produced the Biology of Xenopus book). Mike listened sympathetically to my problems with phonotaxis experiments using lab-reared animals and strongly suggested that we would have better luck with animals collected from their native habitat. Mike had been carrying out a study on hybridization between laevis and an endangered species, gilii, and a then graduate student in the lab, Ben Evans, first went to South Africa to examine hybrids using the methodology of molecular phylogenetics (Evans et al., 1996; 1998). Mike was also kind enough to play host to studies of vocal behaviors that we carried out in 1995 and 1997 and we are most grateful to him and to the University of Cape Town.

This first trip to the field was designed to resolve a paradox presented by our physiological studies. Though the male vocal system is very robust in most respects (more neurons, more muscle fibers, faster fibers etc.) it has a striking weakness: the vocal neuromuscular synapse. The connection between laryngeal motor neurons and their target muscle fibers in males is not strong enough to insure that the fiber contracts in response to an action potential in the motor nerve. Instead, with continued activity, transmission at the synapse becomes stronger (facilitates) until an action potential is achieved and the fiber contracts (Ruel et al., 1998); as more fibers contract the clicks become louder. Males use this physiological characteristic to enhance the amplitude or intensity modulation of their clicks and intensity modulation enhances the attractiveness of male songs to females. When, we wondered, did males sing these intensity modulated songs? What are they for?

Martha Tobias arrived in Cape Town in the early summer of 1995 and began prospecting for recording sites: ponds with Xenopus laevis. To catch the frogs she used bone-baited funnel traps, a method of Mike Picker's, and soon found that most ponds that she examined contained the frogs, sometimes in staggeringly large numbers. We recorded in a number of these ponds throughout this initial trip to Cape Town and were struck by how few males were calling in ponds that contained a large number of sexually active male frogs, frogs that sang spontaneously when recorded alone in a bucket. Our subsequent field trip and lab studies have suggested that males establish a vocal dominance hierarchy that restricts the number of callers (Tobias et al., 2003).

Figuring out the function of intensity-modulated songs proved very difficult because we couldn't actually see the frogs in the ponds. Most ponds were silt filled and frogs called from the layers containing decomposing leaves and plant matter. Actually observing the frogs would require a clear pond and a convenient one was at hand: the swimming pool of the house that the Tobias family had rented in Rondebosch, the University suburb of Cape Town. The plan was to have male frogs sing, spontaneously, both intensity modulated and unmodulated songs and to observe female responses to various song variants. The general approach was to put a male into a trap (one end of which was always elevated so that the frog could rise to the surface to breathe), record his songs, release a female into the pool and see what she did (the underwater pool lights were a big help here as all the observations were carried out at night). As a control we used traps containing another female. The results were that some nights the males sang intensity modulated songs and the female ended up in the male's trap. Other nights, despite the male's songs, the female ended up in the trap with the other female. The field season was drawing to a close, equipment was breaking down (Andy Bass at Cornell very kindly lent us a hydrophone) and the females were becoming unreceptive, when Martha hit on a new behavioral paradigm. The idea was that perhaps intensity-modulated trills were used in a particular phase of courtship. We might be able to observe this if we slowed courtship down enough. The male was placed in a corner of the pool (the part with the steps), barricaded with a plywood partition; the funnel trap provided an opening from the pool into the male's nuptial chamber. What we expected to happen was that the female would spend some time listening to male songs at the barrier and respond to attractive songs by swimming through the hole in the barrier, joining the male. What we had not counted on was that the soggy plywood partition was acoustically transparent making the entrance to the male's chamber impossible to find. The female, about to oviposit because we had injected her with HCG, swam rapidly up and down the length of the barrier and as she swam she began to sing, a new song, one we had never heard before. This new female song is a very rapid series of loud clicks that sounds something like machine gun fire or a Geiger counter and Martha Tobias named the song rapping.

The male's response to rapping was dramatic. Not only did he call incessantly but he also modified his advertisement call in a very characteristic way (Tobias et al., 1998) to produce an answer call. One of the characteristics of the answer call is exaggerated intensity modulation. So the answer to our question, "When do males use intensity-modulated songs?" is "When they respond to conspecific calls". We still don't know why intensity modulation is attractive; the most likely hypothesis is that intensity modulated click trains are more easily located than monotonous trains.

Figure 9 Ticking suppresses while rapping enhances male calling.

When we returned to Columbia we set out to examine rapping more carefully and carried out a series of experiments in which we used broadcasts of rapping to probe the male's response; the control stimulus was taped ticking. As we had hoped, tapes of rapping also functioned as acoustic aphrodisiacs; somewhat to our surprise, ticking tapes had the opposite effect: the males were silenced. The information carried in a single click is not sufficient to convey the message of attraction or repulsion. For example, rapping tapes made up of the clicks from ticking are as effective as rapping tapes themselves in encouraging male songs. What is important is click rate: rapid rates turn males on and slow rates are discouraging (Tobias et al., 1998). Preliminary behavioral studies suggest that males are choosy about labeling a click train as rapping but more elastic when it comes to labeling a click train as ticking (Kelley et al., 2002). In studies carried out this past summer (2003) we have pursued the labeling boundary for a male's perception of ticking and rapping and have determined that males habituate to ticking but not to rapping. This information has allowed us to design experiments in which we can examine acoustic perception and ask, for example, whether rapping and ticking are perceived categorically.

Our next trip to Cape Town (1997) focused on two issues: why don't more males sing at once and do females rap in natural ponds? We recorded from a single pond and we were able to catch the beginning of the breeding season in June and to follow the kinds of calls produced until August. We did record rapping and, in addition, a new call, chirping. From click frequencies we suspected that chirping was a male call. Again, simultaneous calling was rare in the pond. From our studies of male/female pairs, we knew that females could suppress (ticking) as well as excite (rapping) male calling (Tobias et al., 1998). We wondered whether the lack of simultaneous calling in ponds might be an effect of vocal suppression, one male suppressing each other. Perhaps males even set up a vocal dominance hierarchy that determines who gets to call, that night, that month, that season.

First, however, we had to determine the behavioral context of the different male calls and this was impossible in the ponds because we couldn't see the frogs. We also had to figure out which frog was making which song type. We couldn't tell this just by looking at the frogs because there are no externally visible movements of the body associated with calling. Bob O'Hagan solved this problem by designing the wand hydrophone, a small, relatively insensitive underwater microphone that can be attached to a rod. Using the wand hydrophone we can determine which frog is calling, even in a clasping pair. Physical constraints (a practical length for the rod, the wires leading to the amplifier) have so far limited the use of the wand hydrophone to laboratory tanks. We would very much like to be able to follow an individual frog within a natural pond as he interacts with other males and females and record his calling specifically.

The songs males sing depend on which frog they are with; some songs are produced only when paired with another male while others are produced in the presence of either sex. Female pairs don't vocalize though females do sing (ticking or rapping) when paired with males. Males vocalize as they clasp each other and one becomes vocally dominant; the subordinate male goes back to singing when the dominant male is removed. Both members of a pair call less when they are clasping but if clasping (as well as other forms of physical contact) is prevented, one singing male can silence the other (Tobias et al., 2003). Which songs are particularly important and which sensory cues are used establishing dominance are interesting, and experimentally approachable, questions.