Published in Scientific American
Reprinted with Permission
Copyright © 1995 by Scientific American, Inc. All rights reserved
Article will be removed from posting on September 16, 1998
by Peter M. Narins
When the sun sets behind the peak of El Yunque in Puerto Rico, the hillside comes alive with sound. But not exactly with song. Calls of myriad forest creatures build up to a cacophony as loud as a subway train passing six meters away. One particular bellow is apt to make my uninitiated students jump and clap their hands to their ears, shouting, "What is that?" More often than not, it is the Eleutherodactylus coqui, or coqui, a 36-millimeter long amphibian that is the star amazingly enough, of local lullabies--as well as of my ongoing studies on frog signals.
In 1970, as a young electrical engineer looking for a research project in communications, I chanced on Robert R. Capranica of Cornell University describing an experiment with the North American bullfrog. Electronically synthesizing its mating call, he broadcast it to a captive male, which instantly emitted a similar bellow. But when Capranica presented the croaks of 34 other species of frogs and toads, not once did the bullfrog respond. Capranica also found that both frequency bands present in the call were needed to elicit a reply.
I began to wonder how the subtleties of these sounds related to the environments for which the frogs are adapted. Two decades later I am still studying problem. For the most part, the calls seem to be solutions to the challenge of signal processing in the presence of noise. Most frogs live in rain forests, which also support numerous other signaling creatures. The small number of sounds a male frog produces has to be easily distinguished from those of other individuals and species in the acoustically complex environment. (Female frogs, on the other hand, rarely call.)
One amphibian strategy, that of spectral separation, is also utilized by radio stations in a large city, which each transmit within a separate segment of the of the electromagnetic spectrum. In a rain forest, different amphibian "users" of the acoustic space may likewise occupy private channels. Nearly every species of frog in the Caribbean National Forest, where El Yunque is located, seems to call within a band of frequencies not shared by any other group. Such a separation persists even though the frequencies are constrained by other factors. Large frogs have deep voices--that is, they call at low frequencies; small frogs emit high chirps. Because coquis at higher elevations of El Yunque are-for unknown reasons-larger, their voices are deeper. And cold frogs repeat their calls at slower rates, because the muscles controlling the sounds slow down.
The frequencies claimed by a species are used with much efficiency. The coqui has a simple, two-note, co- qui call. The co note, at about 1,160 hertz (for a male at an altitude of 900 meters), appears to indicate territorial boundaries to other males: on presenting a signaling male with the calls of another, we found that the first dropped its qui note and called exclusively co. Nor did the first male care if the call was electronically changed to qui-co. The qui note, on the other hand, entices females. Neurophysiological studies show that a large population of auditory neurons found exclusively in female coquis are maximally sensitive around 2,090 hertz, in the frequency range of the qui.
In addition to spectrum sharing, frogs resort to time-sharing. Some restrict their calls to particular times of day. The coqui, for example, regales the island of Puerto Rico from sunset to midnight, repeating the call every few seconds. Bigger coquis call once in four seconds; smaller coquis chirp every two seconds. Astonishingly, these sounds can be timed to fit in the brief moments when the nearest neighbors are taking a break.
Randy Zelick, a former graduate student, and I tried to evaluate how precisely a frog can place its calls to avoid temporal overlap with another loud sound. We presented coqui males with bursts of pure tones, repeated every 2.5 seconds, to coincide with the natural calling rate of that population. Gradually we increased the duration of these interfering tones. Even when the broadcast occupied 90 percent of its normal calling period, a frog was able to initiate its vocalizations in the narrow window of (relative) silence.
To test if the males were predicting the occurrence of the gap by observing its periodicity, we presented them with randomly occurring tones of two different durations, separated by 750-millisecond gaps. The amphibians were still able to avoid the loud sounds. In these creatures, as in all animals, periodic functions are believed to be driven by oscillations in a neuronal circuit. The coqui's extraordinary feat implies that the periodic circuitry driving the vocal muscles could be reset cycle by cycle, being triggered each time by the onset of the gap.
We next decided to find out if a coqui would call during the "window" if, instead of turning off the interfering sound, we simply played one that was less loud during that period. Zelick and I presented 23 males on El Yunque with a 1.5-second control tone with a pitch similar to that of the male being studied. The sound was immediately followed by a one-second "test" tone. This sequence was repeated every 2.5 seconds, the same rate as its calls. We increased the loudness of this test tone until the animal no longer called preferentially in the window, perhaps because at this point both tones sounded equally loud to it. But even when the test tone was within four to six decibels of the control tone--a difference barely discernible to humans against the background noise of the forest--59 percent of the males were able to place their calls in the window. Such a small shift in intensity was apparently sufficient for resetting their neural oscillator.
Despite such sensitive adaptations, neatly partitioning frequency and time among all local frog species is not always possible. So various other strategies have evolved. Males of many species produce a periodic, stereotyped call that increases redundancy, so that the caller can be identified and located even if some of its croaks are drowned out. Moreover, the eardrum and other parts of the auditory receptors of an amphibian are tuned to the tone and characteristic period of that species' call, allowing for sharp, selective hearing.
By far the most obvious adaptation is loudness. In forested areas one male coqui occurs in every 10 square meters, so it is subjected to intense pressure to drown out its neighbors for the benefit of a distant female. A coqui sitting half a meter away calls at between 90 and 95 dB SPL, close to the human threshold of pain. (SPL, for sound pressure level, refers to the lowest pressure difference audible to the human ear at 1,000 hertz, Po = 0.0002 dyne per square centimeter. Sound pressures P are described in decibels with respect to this standard, as dB = 20 log P/Po. A jackhammer, for example, produces noise that is 100 dB SPL.) As a result, the male is exposed to potentially damaging levels of sound from its own call for 11 months a year.
Wondering how such a small creature protects itself from its own racket, I decided to measure how loud it sounds at the ear. In the early evening, as a coqui chirped in the vegetation near E1 Verde, I carefully extended a probe microphone to within 13 to 35 millimeters of its eardrum. Peak levels of its co and qui notes at such distances were an ear-splitting 114 and 120 dB SPL, respectively.
Although these external pressures are very high, the amount by which the eardrum moves--and thereby stimulates or possibly overstimulates the sensitive inner ear--also depends on the pressure inside the eardrum. Male frogs vocalize by squeezing their lungs with their nostrils and mouth shut. Air is forced over the vocal cords and into a closed system of chambers that includes the mouth cavity. A thin-walled sac at the base of the mouth then blows up like a balloon, radiating the call from the vocal cords into the environment.
To learn how much the eardrum actually moves as a coqui calls, I collected 10 of the loudest males from the trees around the field station in E1 Verde and flew with them to the University of Konstanz in Germany. I had arranged to collaborate with neuroethologists Günther Ehret and Jürgen Tautz, using their laser Doppler vibrometer (LDV), which produces a low-powered helium-neon laser beam. When the beam is aimed at a male frog's eardrum, some of the light scattered by the membrane reenters the laser. If the eardrum is stationary, the wavelength of the returned light is the same as that of the outgoing laser beam. But if the wavelengths differ, the LDV calculates the velocity with which the eardrum moves and from that, the amount it moves--to within a billionth of a meter.
We set up 10 individual aquariums for the displaced males, complete with tropical plants, high temperature and humidity to simulate their home. Much to our chagrin, not one of the frogs called at all for the three weeks I was there. We concentrated on measuring their eardrum displacement in response to the playback of a coqui's call, adjusted so that the levels of the co and the qui notes at the eardrum were 66 and 73 dB SPL, as they would be for a neighbor's call in the wild.
Our experiments serendipitously offered a very interesting observation. The experimental protocol involved a double-blind format in which I aimed the laser at the eardrum and presented the call, 130 times, while Ehret and Tautz monitored the eardrum response. Then we ran a control trial, in which I aimed the laser at a point on the skull while the call was rebroadcast, to ensure that the entire frog was not vibrating in response to the sound. (We had earlier discovered that shining the laser beam on the frog in the darkened laboratory would make it completely immobile.) We subtracted the control spectra from the experimental spectra to get the net motion of the eardrum.
Late one night my hand slipped during a control run, and I inadvertently aimed the laser to one side of the frog. To our astonishment, the LDV revealed that the skin overlying the lungs was clearly vibrating in response to the sound. We immediately proceeded to "map" the frog's body. It turned out that a small region of the lateral body wall responded to sound, and it was only slightly less sensitive than the eardrum.
Next we measured the pressure fluctuations inside the mouth cavity as a small speaker pressed to the lateral body wall applied sound. With Barbara Schmitz, also at the University of Konstanz, we showed that these sounds caused the eardrum to vibrate. These studies indicated an unbroken air link from the lung to the eardrum. Such an internal pathway suggests not only how frogs locate the source of a sound but also how they protect themselves from their own calls.
Our own ears and those of other mammals and birds are pressure receivers: the eardrum is stimulated by sound coming from the outside, while the inside remains at constant pressure. A receptor of this type cannot distinguish between directions. (At rather high frequencies, however, the waves have lengths that are small enough to be blocked by the head, allowing for some directionality.) The origin of a sound is usually sensed by analyzing both the time of arrival and the intensity differences between the two ears.
In a pressure-gradient receiver ear, such as that of many insects, the sound reaches both sides of the eardrum, which reacts according to the pressure difference across it. This type of ear is inherently directional, because the pressure (or phase) that sound presents at either side depends on how much farther it has to travel to get to one side than to the other. The path lengths depend, in turn, on the angle at which the sound is incident. Comparison between two ears can enhance directionality.
Frogs have an intermediate arrangement. Their ears are asymmetrical pressure-gradient receivers: sound impinges on both sides of the eardrum, but with unequal pressures, the difference between which is measured. This receiver is also directional, but more complex than the insect ear. At the internal surface, sound arrives by at least two routes--via the opposite ear and through the acoustic pathway from the lungs.
The air routes suggest a novel way in which frogs might protect themselves from overstimulation. When a male calls, the high air pressure in its mouth cavity is communicated to the eardrum, which bulges out. Because that membrane is pulled tight, the response to sound is dampened. In addition, sounds generated by the vocal cords impinge both on the inner surface of the eardrums and on the outer surface, after being radiated from the vocal sac. If the sounds arrive nearly in phase--so that periods of high pressure coincide on both sides--the eardrum will not move much in response.
To test this theory by observing the eardrum motion of calling frogs, we had to return to E1 Verde. Pamela Lopez, my graduate student at the time, and I used a portable LDV for the delicate measurements. Because the target has to be extremely stable, we were restricted to studying only those individuals calling from solid objects, such as large tree trunks or houses. More problematic was that the coqui call most vigorously in humidity approaching 100 percent, during or just after rain. The LDV does not operate in rain, because the laser beam must be uninterrupted on its way to and from the target.
That summer was one of the driest on record for Puerto Rico. To induce the frogs to call, we found that spraying water on the test animal for a few minutes worked very well. The other problem we faced was that the laser required a 110-volt power source. Fortunately, local residents were most cooperative when asked if we could plug in our extension cord to make laser measurements on a frog's eardrum.
Our results confirmed that the eardrum does in fact vibrate in response to a male's own call, but with a very small vibration amplitude. We are now investigating the phases at which sound arrives on the inner and outer sides of the eardrum to test the theory further.
A few frogs, unable to compete with the coqui and others in loudness, shrillness or persistence, exploit entirely different media. The white-lipped frog Leptodactylus albilabris, a ground-dwelling, nocturnal creature found along marshes, ditches and mountain streams throughout much of Puerto Rico, employs this tack. Males call from within clumps of dense grass, under fallen vegetation or from shallow depressions or burrows in the muddy soil. (Females of the species are camouflaged and silent.)
The night I first heard a white-lipped frog, I carefully approached it, but he instantly ceased calling. After several futile attempts, I managed to capture some calling males and bring them back to California. In the laboratory of Edwin R. Lewis at the University of California at Berkeley, we investigated the frog's extraordinary sensitivity to my distant footfalls. Transmitted through the ground, the steps must have made the frog shake. But how did the creature detect such delicate vibrations?
We found a population of nerve fibers originating in the inner ear of this frog that responded vigorously when it was vibrated at frequencies between 20 and 160 hertz. The most sensitive fibers responded to peak accelerations of about 0.000001 g (where g = 32 feet per second squared, the acceleration caused by gravity), making them 100 times more sensitive than mammalian inner-ear organs.
Vibration sensitivity of frogs and toads is known to reside in the sacculus of the inner ear [see "The Hair Cells of the Inner Ear," by A. J. Hudspeth; SCIENTIFIC AMERICAN, January 1983]. In the white-lipped frog, this organ consists of a sac filled with a slurry of dense calcium carbonate crystals, resting on 600 sensory hair cells. When the animal shakes, the upper surface of the hair cells and the roots of the hairs move back and forth; the massive sac at the tips of the hairs remains stationary because of inertia. As a result, the hairs bend, modulating the normal discharge rate of the nerve fibers.
But a seismometer as sensitive as that of the white-lipped frog must have some function other than detecting the presence of researchers. To investigate this possibility, Lewis and I used a geophone. This device consists of a coil held by a spring within a strong magnetic field; vertical movements of the earth generate a voltage in the coil. Connecting amplifiers and headphones and placing the geophone near a calling white-lipped frog, we were quite pleased to find that it recorded a heavy thump simultaneously with the call.
After a rain, the male white-lipped frog buries its rear end in muddy soil, leaving its head and forelimbs exposed. When the frog croaks, its vocal pouch expands explosively, striking the ground. The impact generates a Rayleigh wave of vertical vibrations that travels along the ground's surface at roughly 100 meters per second (and with a peak acceleration of 0.002 g at a distance of one meter). The energy in this wave contains frequencies in the same range in which we had found the nerve fibers to be most sensitive.
To test if the thumps were being detected by the neighboring white-lipped frogs, Lewis and his colleagues constructed a "thumper" from the solenoid of an electric typewriter. We triggered the device with a tape recording of a male's call, thus simulating the animal's thump rate and pattern. Even though we insulated the thumper so that the airborne sounds it made could not be heard, males within three meters of our artificial frog consistently entrained their calls to its, producing a chorus.
We do not yet know if the males respond differently to the acoustic and seismic components of a neighbor's call. Although vibrations are usually associated with bulk movements of the body, and sound with airborne pressure changes impinging on the ears, the perception of these stimuli appears to be intimately related. Recent studies of leopard frogs in our laboratory and elsewhere have revealed two significant populations of nerve fibers. One set originates in the amphibian papillia, a low-frequency sound sensor, and the other set in the sacculus, with its vibration-sensitive hair cells. But both bundies of fibers respond to sound as well as to vibrations. The main difference between sound and vibration in this case could well be the route these stimuli take to reach the sensors. Such pathways have not been fully elucidated.
In addition to the white-lipped frog, three other species of vertebrates--the blind mole rat of Israel (Spalax), the Cape mole rat of South Africa (Georychus) and the bannertail kangaroo rat of the U.S. Southwest (Dipodomys)--have now been shown to communicate seismically. Recently Albert S. Feng of the University of Illinois, Jakob Christensen-Dalsgaard of Odense University and I discovered that one species of Malaysian tree frog (Polypedates leucomys tax) exhibits a particularly unusual behavior. During courtship, females living in dense mats of floating vegetation perch on a reed or blade of grass and tap their rear toes rhythmically. This activity occurs in the dark, persists for several minutes and is only occasionally accompanied by calls. Males on neighboring reeds quickly locate and mate with the tapping female.
The taps seem to serve as vibrational signals that indicate the female∆s presence. This behavior is remarkable: in most tree frog species, stationary males call, and females localize them. Also, this is the first time a terrestrial vertebrate has been found to transmit seismic signals through a substrate other than earth. Probably, the different environments in which frogs live offer several other media that are likewise being exploited. I expect to spend many more years deciphering these diverse signals.
PETER M. NARINS has studied communication in frogs, golden moles, krill, forest birds and Cape mole rats across all seven continents. After earning a master's degree in electrical engineering in 1966 from Cornell University, he went to Chile as a Peace Corps volunteer for three years. Returning to Cornell, he completed a doctorate in neurobiology and behavior in 1976. Two years later he joined the faculty of the University of California, Los Angeles, where he is a professor of physiological science. In his free time he enjoys amateur radio, playing guitar and watching birds.
Return to 1995 Science Writers Award Winners Return to ASA Press Room