How squawk frogs

When the sun sets behind the summit of El Yunque, the nightly cacophonic concert starts in the Caribbean National Park. Here in the mountainous landscape of eastern Puerto Rico, countless animals begin to croak, shout and scream, and each one tries to drown out the other. The noise level is no different than thundered subways.

One choir in particular makes students who have recently come to the research station in El Verde startle and cover their ears in horror. They are all the more astonished to learn that the originators are not even four centimeters tall frogs, which are called, as they call: Coqui (picture 1).

Strangely enough, the Antilles whistling frog with its powerful voice, also scientifically named Eleutherodactylus coqui, is sung about in many lullabies in the region. For many years it has been one of my favorite subjects to study.

When I was looking for a research project on animal communication as a young electrical engineer in 1970, I happened upon Robert R. Capranica of Cornell University in Ithaca, New York State. He made acoustic experiments with the North American bullfrog. If he played synthetic sounds to a captured male, which resembled his mating call, it barked immediately, but remained silent at the screams of 34 other frog and toad species. It was also not possible to outsmart the bullfrogs: their typical call contains parts from two frequency bands, and neither of the two components could be missing.

So I became curious how such acoustic subtleties are coordinated with the life, behavior and environment of these animals. I still find new, fascinating adaptations today.

Acoustic niches

Mainly frog calls seem to be designed to stand out against a background noise. Especially in the rainforest, where most of the frogs live, there are many noisy creatures. A single male frog (the females hardly call) not only has to be audible from the sound of other groups of animals and other frogs, but above all has to assert itself against competing members of the sexes.

One principle of the delimitation is similar to that of radio transmitters: each species occupies a certain frequency band. This can be seen impressively in the spectrogram (see box on page 92 above): The Coqui's closest relatives all have special acoustic channels, so to speak - some quack with higher frequencies, others with lower frequencies. This also seems to apply to most of the other frog species that live in the Caribbean National Park around El Yunque.

Apparently this works despite some restrictions. Big frogs have darker voices and small frogs have lighter ones. Even among the Coquis there are clear differences; for unknown reasons, the specimens in higher altitudes of Costa Rica are larger and then call a little more sonorous. In addition, the outside temperature has an effect on the cold-blooded amphibians: With cold, sluggish muscles, they take longer breaks between calls.

The species-specific frequency bands were used very effectively. In Coqui, for example, the two octave syllables have different meanings: The deep co, which sounds at an altitude of 900 meters at around 1160 Hertz, is apparently intended to mark the territory. When we played someone else's calls to one male, it promptly left out the second syllable, but continued to croak vigorously co - co - co; Nor did it matter if we electronically switched the order of the syllables in the played calls. In contrast, the high qui at frequencies around 2090 Hertz is obviously intended for females. Because only with them is there a certain large group of auditory neurons that are most sensitive to signals in this frequency range.

Furthermore, frogs croak in phases and moments that are favorable for perception. On the one hand, many species limit their vocalizations to certain times of the day; the Coqui, for example, lets his vocal organ resound from sunset to midnight. On the other hand, district neighbors practice a kind of time-division multiplex process: They use their calls primarily during the rest of the people's breaks. They succeed in doing this even though each animal has its own rhythm - larger coquis sound about every four seconds, while smaller coquis sound every two.

How they match the short gaps was something my student Randy Zelick and I wanted to examine more closely. To do this, we first confronted males with a short sinus tone every 2.5 seconds (roughly at the natural ringing frequency) and gradually increased its duration - the pause became shorter and shorter. The animals actually managed to place the majority of calls in the phase between the tones, even if it only made up 10 percent of the cycle (box on page 92 center).

Next we checked whether the frogs were adjusting to the period of the signal and how quickly they could adapt to changes. We played them randomly two notes of different length with an equally long pause of 750 milliseconds in between. Even then, the Coquis were able to use solo (box on page 92 below). As with periodic behavior patterns and functions, it can be assumed in this case that oscillations in neural circuits enable this. Obviously, with the Coqui, the excitation system that controls the vocal muscles can be switched to standby from time to time - the triggering stimulus seems to be sudden silence.

Now there is always background noise in nature. What would happen if a somewhat quieter tone sounds during the signal pause? To do this, we observed 23 males in the field, to whom we in turn played a 1.5-second loud control tone of the ringing frequency and a weaker one-second test tone in a 2.5-second cycle. Gradually we increased its volume to determine what difference the frogs could no longer perceive.

They proved to be unexpectedly sensitive: even if the control and test tones differed by only four to six decibels, 16 of the 23 males began to croak after the louder signal was interrupted. Humans would hardly perceive such intensity gradations with the background noise prevailing in this environment.

In spite of all such behavioral adaptations, when many frog species are resident, individual calls can be drowned out in the general noise. Many of them respond to this with redundancy. Tireless sounding is not only typical for advertising frogs: If an animal lets out its voice rhythmically and stereotypically for minutes, it hits a quieter phase in which it attracts attention. In addition, as is the case with some chirping insects for frogs, for example, the transmitter and receiver are matched to one another: the hearing system primarily processes the characteristic frequency and time patterns of calls of its own kind (Fig. 2 above).

Built-in silencer

The most striking thing about the frog calls remains the volume (Fig. 2 below). In the rainforest near our research station there is a male Coqui about every ten square meters, which has to assert itself against competitors. The sound pressure level of the calls is 90 to 95 decibels half a meter away, more than that of a jackhammer seven meters away.

One early evening I succeeded in pushing a microphone up to about 20 millimeters in front of the large superficial eardrum of a calling district owner without worrying him. The sound pressure level there was 114 decibels for the co and 120 for the qui (it was very close to the pain threshold for humans). How can a little amphibian endure eleven months a year without hearing damage?

The vibration intensity of the eardrum - and thus the stimulation of the sensory hairs in the inner ear - does not only depend on the external pressure; the counter pressure from the inside plays a role. When croaking, frogs squeeze their lungs while keeping their mouths and noses closed. The air is pressed past the vocal cords into a closed system of chambers and into the oral cavity, the membranous throat pouch inflates like a balloon to form a sound bladder and transmits the vibrations to the outside - but they also penetrate the auditory canals (Fig. 4).

The neuroethologists Günther Ehret and Jürgen Tautz had developed a sophisticated system at the University of Konstanz with which the movements of the eardrum can be measured. I went there with ten of the strongest screamers that I had freshly caught at the El Verde field station. The device is a laser Doppler vibrometer whose low-energy helium-neon laser beam can be directed directly onto the eardrum. If the membrane does not move, the wavelength of the reflected light remains unchanged; if it oscillates, the wavelength changes, and the device uses this to determine the deflection to within a millionth of a millimeter (see Figure 2 above).

We set up all the males their own terrarium with tropical plants and kept them at their usual temperature and humidity. But none of them uttered a single sound in the three weeks that I was able to stay in Germany. We had no choice but to use tape recordings, which we played back in such a way that they had a sound pressure level on the eardrum that corresponded to the calls from neighbors: 66 decibels the co and 73 the qui.

The experiments were designed as a double-blind study: my colleagues used to watch the movements of the eardrum while I aimed the laser beam. For each test series we played 130 calls, and then I directed the light to a point on the skull as a countercheck. In this way we made sure that we were not only detecting vibrations of the whole body in the strong sound field. (In addition, the animal had to sit completely still; but conveniently, as we had previously found out, the frogs froze as soon as the laser beam hit them in the dark; Figure 4, left.) The net movement of the eardrum was easy to calculate from test and control measurements by us subtracted the two spectra from each other.

We made one of the most revealing discoveries when my hand slipped during an inspection. The jet accidentally hit the side of the animal's body at the level of the lungs. Amazingly, the measuring device now also showed a reaction: the skin vibrated clearly below the frog call from the tape.

Immediately we measured the entire surface of the body. Only a small spot on the flanks was sensitive - but it was hardly less responsive than the eardrum itself, almost like a second pair of ears (picture 4 above).

Next, together with Barbara Schmitz from the University of Konstanz, we determined the pressure fluctuations in the oral cavity when a small loudspeaker placed directly on the skin set the area at the level of the lungs vibrating. Because the eardrum then also vibrated, there must be a continuous air connection from the lungs to this point. Such a cavity system can be used to explain two things: how frogs locate sound sources and how they protect themselves from their own noise.

The ear of mammals and birds is biophysically a closed pressure sensor (Fig. 3 left). With a single receiver of this type, the direction of sound cannot be determined; both ears have to work together, with differences in arrival time and intensity of the respective signal being offset. (The function of the auricles in mammals and the skull, which attenuates high frequencies, are negligible in this consideration.)

In contrast, the hearing organs of many insects detect pressure gradients. The sound hits the eardrum from both sides and varies in strength depending on the direction of origin. This pressure difference is registered, more precisely the phase difference between the two oscillations, which results from the fact that the sound takes longer to one side than to the other. The direction can thus be localized with just one hearing organ (Fig. 3 center).

With the frogs, both are now combined. Your ears are, so to speak, asymmetrical pressure gradient sensors: with them, too, the sound hits the eardrum on both sides, albeit with a very uneven pressure amplitude, and this difference results in the vibration pattern of the eardrum. The direction of the sound source is determined in a very complex way, because at least two sound paths meet on the inside of the eardrum on each side: from the other ear and from the lungs (Fig. 3 right).

This cavity system could also explain how the sensitive structures of the ear are protected from over-stimulation during their own vocalization. When a male calls, we suspected that the high air pressure in the oral cavity would have to communicate to the eardrum, which would then be allowed to bulge outwards, that is, to stretch it. The sound coming from outside should now deflect it less strongly: its effect would be attenuated. In addition, the tones generated on the vocal cords would not only be emitted from the sound bladder to the outside and then picked up by the eardrum, but would also be passed on in the organism to the ear. If both wave packets arrived almost in phase, i.e. if the periods of high pressure coincided, the eardrum would hardly be deflected.

Whether this idea was correct could only be checked when the frog males croaked, for which we could not bring them in captivity. My doctoral student Pamela Lopez and I therefore had to carry out the next experiments in the field station in El Verde. We used a portable laser Doppler vibrometer for this.

Because the animals were not allowed to sway with a branch or leaf during the highly sensitive measurements, we were only able to examine males who, for example, had chosen a tree trunk or a house roof to be their territory. It was more problematic that the Coqui screams most vigorously at almost 100 percent humidity, i.e. in the rain or shortly afterwards in the wet leaves; then our device does not work, because neither the laser beam nor the reflected light should hit water droplets.

Now it was a particularly dry summer in Puerto Rico at that time. Eventually we decided to give the frogs artificial irrigation for a few minutes - this stimulated them to croak, and we were able to prove that a coquis's eardrum actually vibrates very weakly when it shouts unbearably loud for human ears. We are currently trying to find out the phase difference with which the sounds arrive on both sides of the eardrum.

The frogs seismometer

A few species that do not have such a piercing, high-pitched voice as the Coqui make themselves heard in their own way. They include the nocturnal white-lipped whistling frog (Leptodactylus albilabris), which occurs in large parts of Puerto Rico in the marshes of wetlands and on watercourses. The vocal males crouch in dense tufts of grass, fallen leaves, shallow depressions or self-dug holes; the females are well camouflaged and remain silent.

When I first heard the call and tried to sneak up on me, the animals immediately fell silent. It was only with extreme caution that I was able to catch some males (picture 5 left). I studied with Edwin R. Lewis of the University of California at Berkeley how the frogs perceive slight ground vibrations from a distance.

As is known, sensory cells with a tuft of fine, sensitive hairs in the saccule of the inner ear are responsible for the vibration sense of frogs and toads (compare "The hair cells of the inner ear" by A. J. Hudspeth, Spectrum of Science, March 1983, page 108); the white-lipped whistle frog has 600. On them rests an accumulation of calcium carbonate crystals, which remain sluggishly when the sensory hairs vibrate when the animal is shaken, so that they are deflected. This stimulus changes the normal pulse rate of the hair cells, which are otherwise constantly firing.

We discovered a group of nerve fibers that come from the inner ear and respond very violently when we subject an animal to deep vibrations between 20 and 160 Hertz, which are transmitted in the ground. Some reacted to accelerations of around one millionth of the acceleration of gravity, so they are a hundred times as sensitive as those of mammals.

It seemed improbable to us that such a fine sensorium should only perceive the steps of an approaching enemy, and we suspected that it also serves the communication between conspecifics. To test this, we placed a geophone (or seismophone), which measures ground vibrations very accurately, right next to a frog. In fact, it registered a powerful bump with every croak.

The males of the white-lipped whistling frog dig themselves into the muddy ground after rain; only the head and forelegs are still peeking out. When they croak, the sound bubble expands explosively and hits the floor. The vibration propagates at about 100 meters per second as a surface wave (as a Rayleigh wave); after one meter, the peak acceleration is still two thousandths of the acceleration due to gravity, and the vibration frequencies are in the same range as the greatest sensitivity of the inner ear sensory cells.

Lewis and his colleagues have now constructed a device that causes the ground to shake like a frog, but does not generate any sound that can be heard through the air. They used the magnetic coil of an electric typewriter and synchronized the trigger mechanism with tape recordings of frog calls so that the throbbing corresponded exactly to the natural one. Within a radius of three meters, all the males reacted as precisely as if the device was conducting a choir.

We do not yet know whether the frogs can distinguish the acoustic and seismic components of the signal from neighbors. But the perceptions of vibrations affecting the body and those of changes in pressure in the ears seem to be closely linked.

Recently, we and other researchers identified two important groups of nerve fibers in leopard frogs. One arises from a special structure in the inner ear of amphibians - the papilla amphibiorum - and responds to lower-frequency sounds (when hearing, it seems to play the main role as a sensory terminal for amphibians). The other, which we already mentioned with the white-lipped whistle frog, comes from the sacculus and is responsible for vibrational stimuli. A distinction is apparently more functional, because both groups also react to the other signal. The only thing that is essentially different is the way in which the stimuli reach the sensory cells. These pathways are not yet fully understood.

Seismic communication, as in amphibians, has now also been detected in blind mice (genus Spalax) living in south-eastern Europe and in the eastern Mediterranean, in Bleßmullen (Georychus) in South Africa and in kangaroo rats (Dipodomys) in the southwest of the United States.

Interesting in this context is the behavior of a Malay tree frog, Polypedates leucomystax, as Albert S. Feng from the University of Illinois in Urbana, Jakob Christensen-Dalsgaard from the University of Odense (Denmark) and I recently discovered: the females come out of the Floating plant mats on the water, in which they live, climb on a stalk and drum rhythmically against it with the toes of their hind feet - for minutes, with the occasional croak. Males then quickly find each other to mate (Fig. 5 right).

This communication by means of vibration is not only remarkable because of the inversion of the situation (in tree frogs it is usually the males that have the enticing part). The Polypedates females are also the first known terrestrial vertebrates that do not use the ground for seismic signals. Frogs in particular inhabit so many different habitats that many surprises are still to be expected.


- Are there frogs with 4 ears? Laser measurements show new sound input in the tree frog, Eleutherodactylus coqui. By J. Tautz, G. Ehret and P. Narins in: Negotiations of the German Zoological Society, Volume 81, 1988, page 303.

- Accessory Pathway for Sound Transfer in a Neotropical Frog. By P. M. Narins, G. Ehret and J. Tautz in: Proceedings of the National Academy of Sciences, Volume 85, Issue 5, pages 1508 to 1512, March 1988.

- Seismic Communication in Anuran Amphibians. By P. M. Narins in: Bioscience, Volume 40, Issue 4, pages 268 to 274, April 1990.

- Biological Constraints on Anuran Acoustic Communication: Auditory Capabilities of Naturally Behaving Animals. By P. M. Narins in: Evolutionary Biology of Hearing. Edited by D. B. Webster, R. R. Fay, and A. N. Popper. Springer, Heidelberg 1992.

- Reduction of Tympanic Membrane Displacement during Vocalization of the Arboreal Frog, Eleutherodactylus coqui. By P. M. Narins in: Journal of the Acoustical Society of America, Volume 91, Issue 6, pages 3551-3557, June 1992.

- Comparative Aspects of Interactive Communication. By P. M. Narins in: Active Hearing. Published by Å. Flock, D. Ottosen and M. Ulfendahl. Elsevier Science, 1995.

From: Spektrum der Wissenschaft 11/1995, page 90
© Spektrum der Wissenschaft Verlagsgesellschaft mbH

This article is contained in Spectrum of Science 11/1995