Biology Letters


    Tree cricket males produce tonal songs, used for mate attraction and male–male interactions. Active mechanics tunes hearing to conspecific song frequency. However, tree cricket song frequency increases with temperature, presenting a problem for tuned listeners. We show that the actively amplified frequency increases with temperature, thus shifting mechanical and neuronal auditory tuning to maintain a match with conspecific song frequency. Active auditory processes are known from several taxa, but their adaptive function has rarely been demonstrated. We show that tree crickets harness active processes to ensure that auditory tuning remains matched to conspecific song frequency, despite changing environmental conditions and signal characteristics. Adaptive tuning allows tree crickets to selectively detect potential mates or rivals over large distances and is likely to bestow a strong selective advantage by reducing mate-finding effort and facilitating intermale interactions.

    1. Introduction

    Animals using acoustic signals encode considerable information about species identity and individual quality in song spectral and temporal features (reviewed in [1]). For receivers, however, separating conspecific signals from other sounds can be difficult in real-world conditions. Tree crickets signal in a crowded acoustic space with potential for masking [2]. Many insects deal with this problem through tuned auditory systems that respond selectively to frequencies representing conspecific signals [1]. However, in tree crickets, this strategy is complicated by temperature-dependent, hence variable, song frequency [3].

    Tree cricket auditory tuning is determined by frequency-specific, active amplification of tympanal vibration at low sound amplitudes [4,5]. Active amplification is a mechanical process that is achieved physiologically [6]; auditory neurons use axonemal dyneins [7,8] to generate forces [9] that preferentially amplify frequencies representing conspecific songs [4,5,10]. As a physiological rather than mechanical process determines tuning, temperature is expected to affect this process.

    Here we test, in the North American tree cricket Oecanthus nigricornis, whether the amplified frequency increases with temperature thereby allowing the auditory system to remain tuned by tracking changes in conspecific song frequency.

    2. Temperature affects self-sustained oscillations and mechanical tuning

    Self-sustained oscillations (SOs) are characteristic of active auditory mechanics and occur in tree cricket tympana [4,5]. SOs are generated by the same processes underlying active amplification, and reveal the actively amplified frequency [4,9,11]. Figure 1a shows frequency spectra of SOs measured from the anterior tympanal membrane (ATM) over temperatures ranging from 21°C to 30°C, over which SO frequency increased linearly by 112 Hz/°C (figure 1a, inset). Change in SO frequency corresponded well with the previously reported range of O. nigricornis song frequencies (3.6–4.9 kHz from 19.3°C to 29.6°C ambient temperature), and with the slope of the frequency–temperature relationship (144 Hz/°C, R2 = 0.74; data from [12]). The shallower slope that we measured may be owing to a temperature mismatch between the surface of the Peltier element that we used to control temperature and the tympanum (see the electronic supplementary material, methods).

    Figure 1.

    Figure 1. Self-sustained oscillations (SOs) and mechanical tuning of Oecanthus nigricornis ATM. (a) In an example individual, SO best frequency increases linearly with temperature. Colours in the inset correspond to the SO frequency spectra in the main plot. SO frequencies at (b) low temperature are lower than those at (c) high temperature across individuals. (d) Mechanical responses of ATM change with stimulus level and temperature as shown in an individual. Mechanical tuning is the sharpest at low stimulus amplitude and shifts to higher frequencies with increasing temperature. This shift in best frequency is consistent across individuals (e,f; stimulation level: 2.5 mPa).

    Figure 1b,c illustrates SO spectra for seven individuals (three females and four males) at low (22 ± 1°C) and high (29 ± 3°C) temperatures, respectively. SO frequencies increased with temperature despite considerable variation in both temperature ranges (table 1). In all seven individuals, best frequency was higher at higher temperatures, increasing on average by 720 ± 470 Hz.

    Table 1.Mechanical and neuronal tuning in different temperature regimes. Data are given as mean±s.d. (n). Asterisks indicate statistically significant differences.

    low temperature (22±1°C) high temperature (29±3°C) paired t-test
    SO frequency (kHz) 3.7 ± 0.3 (7) 4.4 ± 0.4 (7) *one-tailed, p = 0.003
    mechanical best frequency (kHz) 3.8 ± 0.4 (8) 4.4 ± 0.4 (8) *one-tailed, p = 2×10−4
    neuronal best frequency (kHz) 3.3 ± 0.2 (8) 4.3 ± 0.21 (8) *one-tailed, p = 3.67×10−5
    Q factor of mechanical response 7.7 ± 3.6 (8) 5.6 ± 1.8 (8) two-tailed, p = 0.08

    ATM motion showed the stimulus amplitude-dependent tuning well known from actively amplified auditory systems (figure 1d) [4,5]. At low stimulus amplitudes, a peak emerged in the ATM frequency response (figure 1d). Figure 1e,f shows frequency response functions at low and high temperatures for 2.5 mPa (42 dB SPL) stimuli (four individuals of each sex). As with the SOs, peak frequency was higher at higher temperatures in each individual. On average, best frequencies of sound-driven responses increased by 632 ± 270 Hz. The sharpness of the peak (Q factor) was not significantly different between the two temperature ranges (table 1).

    3. Temperature affects neural tuning

    We recorded extracellularly from the prothoracic ganglion, the first site of central neuronal processing of auditory information in crickets (figure 2a,b). Recordings were made in the hemiganglion contralateral to the side of stimulation in eight O. nigricornis (four of each sex). Four animals had only one ear, because of injury either during development or dissection. In field crickets, the only prothoracic neuron readily recorded contralateral to the site of afferent input is ON1 (see the electronic supplementary material, methods), whose frequency tuning is representative of the auditory system [13]. Although prothoracic auditory neurons of Oecanthus have not yet been characterized, given the recording site, and the occurrence of an ON1-like neuron in several ensiferans [1315], it is likely that these recordings were from a neuron homologous to ON1.

    Figure 2.

    Figure 2. Sound-evoked responses of O. nigricornis putative ON1 interneuron at different temperatures. Example responses of one individual; 60 dB SPL stimuli at different frequencies for (a) low and (b) high temperatures. Coloured boxes indicate stimulus duration. (c) Tuning of an individual, expressed as sensitivity (neural response per unit sound pressure) varies with stimulus amplitude and temperature. Increases in best frequency with temperature are consistent across animals (d,e). (f) Thresholds are lower for on-peak frequencies than for off-peak frequencies (n = 8). Error lines indicate 95% prediction intervals for logistic fits. Threshold, saturation and dynamic range (mean ± s.d.) are indicated; variation is estimated by jack-knife resampling. Colours represent different individuals; males are indicated by circles and females by squares.

    To allow comparison with mechanics, neural responses are represented per unit sound pressure (figure 2c–e). As observed in the mechanics, both neuronal tuning and sensitivity were amplitude-dependent. Tuning to conspecific frequency was sharpest near threshold and broadened as amplitude increased (figure 2c). Note, however, that amplitude-dependent tuning at the neural level does not require active auditory mechanics. Variation among receptors in sensitivity and tuning, and response saturation at high stimulus level, can account for responses that are more frequency-selective at low than at high stimulus levels. Indeed, we observed this in ON1 responses in the field cricket Gryllus assimilis, a species in which there is no sign of active tympanal mechanics (electronic supplementary material, figure S1).

    Crucially, however, neural tuning in O. nigricornis shifted up with temperature, mirroring the mechanics (figure 2d,e and table 1). The average per-individual shift in best frequency was 938 ± 320 Hz. No such tuning shift occurred in G. assimilis (electronic supplementary material, figure S1), reinforcing the conclusion that the shift in frequency tuning is implemented by active mechanisms.

    4. The benefit of temperature-dependent tuning

    Without active tuning, how much would a temperature mismatch reduce male active acoustic space? To estimate this, we compared the neuronal response at the best frequency at the temperature at which the recording was made (on-peak) and best frequency in the other temperature range (off-peak). Thresholds were determined from logistic fits to neuronal responses (figure 2f).

    At both temperatures, thresholds were much lower for the amplified frequency (figure 2f, low temperature: p = 1.4 × 10−4; high temperature: p = 4 × 10−4; one-tailed paired t-tests). Based on the shift in threshold, with fixed tuning to the low-temperature frequency, song produced at the best frequency for high temperatures would have to be 18.5 ± 1.8 dB SPL louder to be detected by a low-temperature ear. Considering attenuation owing to spherical spreading alone, active amplification increases detection distance by a factor of 8.6 ± 1.7 and active area by a factor of 76.0 ± 30.9. Similarly, song at best auditory frequency for low temperature would have to be 11.4 ± 1.8 dB SPL louder to be heard by a warm ear. This conservatively corresponds to an increase in detection distance by a factor of 3.8 ± 0.8 and active area by a factor of 14.8 ± 6.3.

    The main effect of active amplification, thus, was to lower auditory thresholds at the temperature-appropriate frequency, thereby increasing the distance over which a male signalling at that frequency could be heard.

    It is important to note that the threshold achieved by the actively amplified O. nigricornis auditory system, at about 45 dB SPL, is similar to ON1 thresholds in field crickets [13,16,17]. Indeed, even at the very highest observed amplification, O. nigricornis tympana rarely achieve mechanical sensitivity higher than G. assimilis (figure 1; electronic supplementary material, figure S1, [5]). It has been argued that an actively amplified auditory system produces a high level of spontaneous noise near criticality, and so cannot lower the minimal detectable sound level [18]. Thus, we find that the primary function of active amplification in tree crickets is not to lower the absolute auditory threshold below that of other crickets but rather to modulate auditory tuning to match conspecific song frequency even as it changes with temperature.

    Active amplification has been invoked to explain the remarkable sensitivity, dynamic range and frequency resolution of many auditory systems [19]. However, few studies demonstrate the contribution of active amplification to these processes, especially in ecologically relevant settings. Given that some features, such as fine frequency tuning [20] and high dynamic range [21], are achievable in systems that are, to the best of our knowledge, purely passive, the advantage of active amplification remains to be demonstrated.

    Perhaps the most compelling demonstrations of functional roles for active amplification come from fruit flies and mosquitoes. In fly courtship, active amplification contributes to species isolation by matching antennal tuning to conspecific song frequency [10]. However, passive resonance can also achieve matched filtering (electronic supplementary material, figure S1a,b, [22]), and the specific advantage of active tuning is unclear. In mosquitoes, mate recognition is believed to be mediated by convergence of higher harmonics of male and female flight tones [23]. In the genus Culex, mismatched harmonics are thought to be detected using low-frequency difference tones generated by active mechanics [24]. However, in Aedes aegypti mosquitoes, the frequencies of the higher harmonics can be detected directly by the auditory organ [25], again questioning the specific advantage of the active process.

    We show that an important selective advantage of active amplification is its flexibility and responsiveness to environmental conditions. While sender–receiver match can be achieved through resonance for a fixed frequency signal [22], resonant systems can only remain sensitive to a variable frequency signal by being broadly tuned, thus compromising selectivity. Active systems can modify the amplified frequency and bypass this trade-off.

    In insects, active amplification depends on the action of ATP-dependent molecular motors [8,26]. Models suggest that the amplified frequency, and hence tuning may be determined by the ATPase rate of these motors [7,27]. Indeed, the temperature dependence of SO frequency has been observed in two poikilothermic organisms: frogs [28] and mosquitoes [7]. Here we show that tree crickets have both temperature-dependent SOs and mechanical and neural tuning, and that they exploit this property to maintain auditory sensitivity to song even as it changes with temperature, without loss of selectivity.


    This work complies with ethical guidelines in Canada.

    Data accessibility

    The dataset supporting this article has been uploaded to the Dryad repository:

    Authors' contributions

    N.M. made mechanical measurements and analysed data; G.P. made neuronal measurements and analysed data; N.M., G.P. and A.M. designed the study and wrote the manuscript. All authors agree to be held accountable for the content therein and approve the final version of the manuscript.

    Competing interests

    The authors have no competing interests.


    This research was funded by NSERC Discovery grants to G.P. and A.M.


    Published by the Royal Society. All rights reserved.