Vibrational playback by means of airborne stimuli

As one of the most common forms of communication in animals (Hill, 2008), substrate-borne vibrational signaling is attracting attention from researchers interested in the ecology and evolution of behavior. Animals use a variety of substrates to transmit vibrational signals, and how these signals are transmitted depends on the properties of the substrate (Čokl and Virant-Doberlet, 2003; Cocroft and Rodríguez, 2005; Elias et al., 2010). For example, plants that are used as vibrational substrates vary in their transmission properties among species and individuals (Henry and Wells, 2004; Cocroft et al., 2006; Čokl et al., 2006; McNett et al., 2006; McNett and Cocroft, 2008; Čokl et al., 2009). While these attributes impose interesting challenges for animals to communicate effectively with one another, they pose equal challenges to researchers interested in studying vibrational communication.

Currently, vibrational playback methods are such that the playback device (e.g. shaker, actuator) must be in direct contact with the substrate (Cocroft and Rodríguez, 2005). While these contact-based methods are effective, they have the limitation that only one or a few substrates can be stimulated at any one time. If one wants to isolate individuals so that they experience only the treatment without addition of social interactions, the ability to stimulate several substrates simultaneously would be helpful. Additionally, by contacting the substrate, current vibrational playback methods can alter the transmission features of the substrate (Cocroft and Rodríguez, 2005).

One possible alternative to contact-based methods is to take advantage of the fact that airborne signals are picked up by some substrates, such as the tissues of plants and leaf litter (Saxena and Kumar, 1980; Hill, 2008). Because airborne signals maintain the same timing and pitch as vibrational signals (Cocroft and Rodríguez, 2005), signals may be effectively and adequately imparted from a loudspeaker onto a substrate. However, the pickup of airborne signals by the substrate introduces a new level of potential variation on top of the variation due to transmission along the substrate. Here, we present a methodological framework to determine if it is appropriate to use airborne induction of vibrational stimuli. We assess this method for broad-band and pure-tone stimuli, representing a range of signals used by vibrationally signaling animals and the types of stimuli used in vibrational playback experiments.

We validate the use of airborne stimuli for vibrational playback in four steps. First, we ask if we can compensate for the differential attenuation of frequencies that occurs due to pickup by a plant substrate and transmission of that stimulus along the plant. We determine if the standard compensation procedure for contact-based playback of stimuli containing a range of frequencies (Cocroft, 1996; Cocroft, 2005) also works for airborne vibrational playback. Second, we ask if the stimulus compensation for one plant individual is adequate for use with multiple individuals of the same species and describe the degree of variation introduced by using this compensated stimulus. Third, we look at how well pure-tone stimuli are picked up by the plants by looking at the waveform's shape, the variation in amplitude due to varying leaf number, and we describe the degree of variation introduced within a plant and across plants by repositioning each plant to the loudspeaker. Fourth, we look at how vibrationally signaling animals interact with the airborne stimuli picked up by the plants and compare these interactions to established vibrational playback techniques.

We address both broad-band and pure-tone stimuli in an ecologically relevant case study. We used a member of the Enchenopa binotata Say species complex of treehoppers (Hemiptera: Membracidae) and its host plant, Viburnum lentago L. (Caprifoliaceae). These plant-feeding insects communicate with plant-borne vibrational signals (Cocroft et al., 2008; Cocroft et al., 2010). Our findings indicate that airborne signals can be used as vibrational playbacks for Enchenopa but that validation must be verified for each substrate and species in a case-by-case manner.

We used potted exemplars of Viburnum lentago plants. Plants were moved to a greenhouse in February 2011 and 2012 to promote leaf growth and remained there until they were used. We selected 11 plants for the following experiments, choosing similarly sized individuals (~30 cm high with 6–8 leaves each). Experiments were performed in an anechoic chamber held at a temperature of 24±0.5°C during February and March 2011 and 2012.

Our experiments involved creating broad-band and pure-tone stimuli and broadcasting them from a loudspeaker to impart them onto plants. We created and controlled all stimuli using custom-written programs in MATLAB 7.11.0 (m-files courtesy of R. B. Cocroft). We used a Dell Precision T3500 computer (Dell Computer, Round Rock, TX, USA) connected to an Edirol UA-25 USB audio controller (Roland Corp., Los Angeles, CA, USA) connected to a Rokit 8 powered monitor speaker (KRK Systems, Deerfield Beach, FL, USA) to deliver all stimuli. We suspended the speaker from the ceiling of the anechoic chamber such that the speaker cone was parallel to the ground of the chamber (Fig. 1). We placed plants 42 cm away from the center of the cone and isolated them from ground vibrations by placing them on shock-absorbing sorbothane (Edmund Scientifics, Tonawanda, NY, USA). We recorded stimuli picked up by the plants by focusing the beam of a laser vibrometer (Polytec PDV 100; Polytec, Auburn, MA, USA) onto a small (~2 mm) piece of reflective tape placed about halfway up the main plant stem (~14–18 cm high). We sent the vibrational signals from the laser to an HP Pavilion dm4 computer (Hewlett-Packard, Palo Alto, CA, USA) through another Edirol UA-25 USB audio controller and recorded them (44.1 kHz sampling rate) using the program AUDACITY 1.3.13 (http://audacity.sourceforge.net). While our use of a single laser vibrometer underestimates maximum signal amplitude variation in the plant stem (McNett et al., 2006), our purpose was to sample variation as would be most often detected in playback experiments.

We asked if we can compensate for each plant, adjusting for its pickup of the airborne stimulus and filtering during stimulus transmission. The goal of the compensation routine is to minimize the difference in dB at any one frequency between the broadcast signal and the signal recorded from the stem (Cocroft, 1996; Cocroft, 2005). We broadcast white noise ranging from 50-2000 Hz through a loudspeaker and recorded it with a laser vibrometer. We obtained a transfer function (fft size=2048) of the recorded and input signals, which described the relative amplitude difference between the signals. This transfer function was used to create a digital filter that we used to create a compensated stimulus (MATLAB m-files courtesy of R. B. Cocroft). We then played back the compensated airborne stimulus and recorded it with the laser vibrometer.

We evaluated the effectiveness of the compensation by looking at the transfer function of the new compensated stimulus and laser recording. We considered compensation successful if the frequency-response of the plant stem was, on average, within ±3 dB across the sampled range, which is the standard criterion in acoustic studies (e.g. Gerhardt et al., 2007).

We randomly selected one of the 11 plants and generated a compensated stimulus as above. We broadcast the compensated airborne stimulus and recorded it with the laser vibrometer from each of the other 10 plants. This experiment thus describes how well compensation for one plant applies to other plants. We tested each plant once a day for three consecutive days, randomizing plant order each day.

We analyzed plant variation in two ways. First, we looked at variation due to repositioning of plants by assessing variation within plants over the course of three days of testing. For each day tested, we determined the maximum amplitude in dB for each plant across the frequency range and set that value to 0 dB, adjusting all other amplitudes accordingly. We performed pairwise comparisons in amplitude between the three days at each measured frequency from 50 to 2000 Hz for each plant and then calculated the mean relative amplitude for each plant from those comparisons. We then used plant means to determine the overall mean difference in relative amplitude between the broadcast airborne stimulus and the laser recording for each measured frequency. We focused on variation in the 50–500 Hz range as this encompasses the range of signals used by members of the E. binotata species complex (Cocroft et al., 2008; Cocroft et al., 2010). In addition, we noted the maximum amplitude difference within any one plant across the three days of testing for each frequency sampled.

Second, we looked at variation between plants. As before, we adjusted amplitude values for each plant (maximum set to 0 dB) and calculated the mean response across days for each plant. These means were then used to calculate all pairwise differences, resulting in 45 total comparisons for the 10 plants, again noting the maximum difference.

We synthesized five pure-tone stimuli that mimicked signals of males in the E. binotata species complex. These stimuli have previously been used to study the communication system of this complex (e.g. Rodríguez et al., 2006; Fowler-Finn and Rodríguez, 2012). Stimuli ranged in frequency from 85 to 285 Hz, which covers 100 Hz above and below the mean signal frequency for males of the E. binotata ‘V. lentago’ species (185 Hz). We randomly selected one of the 11 plants and placed it in the anechoic chamber as previously described. We then calibrated each of the five stimuli to an amplitude of 0.1 mm s⁻¹ by using an oscilloscope (Kikusui 5020, Yokohama, Japan). We then broadcast the calibrated stimuli for this plant to each of the remaining 10 plants. Randomizing order, we broadcast each stimulus and recorded it from the plant stem and, in addition, estimated the amplitude of the wave from the oscilloscope each time. We repeated this process two more times so each plant was tested three times for each stimulus.

We also looked at how pickup by plants varies with the numbers of leaves. Using the same five synthetic pure-tone stimuli, we calibrated each stimulus to an amplitude of 0.1 mm s⁻¹ and recorded the signal from the plant stem. We then removed half of the leaves and broadcast that same calibrated stimulus, again recording from the stem. We finally removed all the leaves and repeated this process one more time. In total, 10 plants were tested for each stimulus, and each stimulus was calibrated to the plant tested.

We looked at variation due to repositioning the plants by assessing variation within and between plants across the three rounds of testing as in step 2 above. The oscilloscope output was converted to velocity (mm s⁻¹), which was then translated to dB (re. 0.1 mm s⁻¹) for the pairwise comparisons.

We used the 185 Hz synthesized pure-tone stimulus in playbacks to males. We placed an individual male on a plant and allowed him to settle for one minute. We then broadcast the airborne stimulus as a three-call bout, allowing 30 s afterwards to assess if the male was sexually active. This process was repeated a maximum of five times or until the male responded. Males that responded to the stimulus were then broadcast an extended 12-call bout to see if they would engage in chorusing behavior in which their calls alternated with the broadcast stimulus.

For females, we synthesized seven pure-tone stimuli ranging from 145 to 225 Hz (just extending beyond the natural range of male advertisement calls in this species). Following previously established methods (Fowler-Finn and Rodríguez, 2012), we presented a random sequence of seven stimuli to a female, with each stimulus presented as a three-call bout. We tested female preferences through playbacks using a piezo-electric stack (contact-based method) and by broadcasting airborne stimuli through the aforementioned setup, the assay of preference being the production of duetting signals by females (Rodríguez et al., 2004; Rodríguez et al., 2006; Fowler-Finn and Rodríguez, 2012). We then generated preference functions using the responses of all females from each playback technique by constructing cubic spline regressions (glms40 cubic spline program; http://www.zoology.ubc.ca/~schluter/wordpress/software) (Schluter, 1988).

We were able to compensate for the pickup and transmission of each plant. The mean difference between the compensated broadcast airborne stimulus and laser recording for any frequency was less than 2 dB and had a maximum difference of 5.4 dB (Fig. 2).

On the whole, average variation within and between plants was similar across the range tested, with a mean difference of 3.2 dB and 4.3 dB for within and between plants, respectively, but there was considerable variability at certain frequencies (Fig. 3A,B). Over the 50–500 Hz range, responses were adequately flat across days, with a mean difference of 2.28 dB and 2.88 dB for within and between plants, respectively (Fig. 3C,D).

The structure of pure-tone signals was minimally altered (paired t-test comparing frequency of playback stimulus and laser recording, all P>0.17; Fig. 4). These signals appear to be predominantly picked up by the leaves, as removal of all the leaves results in a substantial decrease in signal amplitude (Fig. 4C). Signal pickup and transmission was consistent despite repositioning the plant relative to the loudspeaker, with a mean difference of 1.75 dB and 1.74 dB within and between plants, respectively, for the five signals tested (Fig. 5).

Eleven of 14 male Enchenopa treehoppers called within 30 s of being presented with the airborne stimulus, with six calling after one bout, three more after the second, and the remaining two after the third bout was presented. Of those that called, two males engaged in chorusing behavior (Fig. 6A).

Females duetted with the airborne stimuli and differentially responded according to their frequency (Fig. 6B). Female preference functions overlapped entirely between playback methods (Fig. 7).

We tested whether airborne signals could be adequately imparted onto a substrate as vibrational playbacks. First, we demonstrate that it is possible to compensate effectively for the pickup and transmission properties of a plant stem across a broad frequency range. Second, we describe the variation within and between plants across days, highlighting that in the range used by a group of signaling animals, responses are fairly equal for both measures. Third, we demonstrate that a broadcast pure-tone stimulus is minimally altered by the plant and that these signals are picked up best by the leaves. In addition, repositioning of a plant to the loudspeaker only slightly changes the intensity of a pure-tone stimulus within and between plants across trials. Finally, we show that both male and female Enchenopa treehoppers respond similarly to airborne and vibrational stimuli in playback experiments in spite of the multiple inputs into the plant. However, it is unclear how signal localization may be affected, a point that will require further study.

For a single plant, the recorded signals from airborne playbacks were similar to vibrational playbacks. How well this works across plants depends on the frequency range considered. Some frequency ranges are relatively flat with repeatable responses, such as a subset of the range addressed here. The low and similar variation in dB within and between plants shows that individuals exposed to complex signals in this range will be subject to the same stimulus variability. However, the type of stimulus used is important. Broad-band signals may be more problematic than pure-tone signals because of the frequency range they span and the likely increased variability in transmission as the range increases. We stress that this technique needs to be tested for each substrate, frequency range and stimulus type. This technique may be applicable to other types of substrate-borne signals, such as leaf litter, presenting further opportunities to expand our understanding of vibrational communication.

We would like to thank T. Schuck for help with plant rearing. We also thank K. D. Fowler-Finn and D. Kim for helpful feedback on a previous version of this manuscript.