Physiological changes throughout the ear due to age and noise - a longitudinal study

Biological and mechanical systems, whether by their overuse or their aging, will inevitably fail. Hearing provides a poignant example of this with noise-induced and age-related hearing loss. Hearing loss is not unique to humans, however, and is experienced by all animals in the face of wild and eclectic differences in ear morphology and operation. Here we exploited the high throughput and accessible tympanal ear of the desert locust, Schistocerca gregaria (mixed sex) to rigorously quantify changes in the auditory system due to noise exposure (3 kHz pure tone at 126 dB SPL) and age. We analysed tympanal dispalcements, morphology of the auditory Müller’s organ and measured activity of the auditory nerve, the transduction current and electrophysiological properties of individual auditory receptors. We found that noise mildly and transiently changes tympanal displacements, decreases both the width of the auditory nerve and the transduction current recorded from individual auditory neurons. Whereas age – but not noise - decreases the number of auditory neurons and increases their resting potential. Multiple other properties of Müller’s organ were unaffected by either age or noise including: the number of supporting cells in Müller’s organ or the nerve, membrane resistance and capacitance of the auditory neurons. The sound-evoked activity of the auditory nerve decreased as a function of age and this decrease was exacerbated by noise, with the largest difference during the middle of their life span. This ‘middle-aged deafness’ pattern of hearing loss mirrors that found for humans exposed to noise early in their life. Key point summary Age and noise lead to hearing loss. Tympanal displacements have a transient and delayed reduction after noise. The number of auditory neurons decreases with age. The width of the auditory nerve is reduced with noise exposure. Locusts repeatedly exposed to noise lost their hearing compared to silent controls, but their hearing became similar to old silent control locusts due to age-related hearing loss dominating. The electrophysiological properties of the auditory neurons and the transduction current remained unchanged for aged locusts but repeated noise exposure led to a cumulative decrease in the transduction current.

• Age and noise lead to hearing loss. 12 • Tympanal displacements have a transient and delayed reduction after noise. 13 • The number of auditory neurons decreases with age. The width of the auditory nerve is reduced with 14 noise exposure. 15 • Locusts repeatedly exposed to noise lost their hearing compared to silent controls, but their hearing 16 became similar to old silent control locusts due to age-related hearing loss dominating. 17 • The electrophysiological properties of the auditory neurons and the transduction current remained 18 unchanged for aged locusts but repeated noise exposure led to a cumulative decrease in the 19 transduction current. 20

Abstract (250 words) 21
Biological and mechanical systems, whether by their overuse or their aging, will inevitably fail. Hearing 22 provides a poignant example of this with noise-induced and age-related hearing loss. Hearing loss is not 23 unique to humans, however, and is experienced by all animals in the face of wild and eclectic differences in 24 ear morphology and operation. Here we exploited the high throughput and accessible tympanal ear of the 25 desert locust, Schistocerca gregaria (mixed sex) to rigorously quantify changes in the auditory system due 26 to noise exposure (3 kHz pure tone at 126 dB SPL) and age. We analysed tympanal dispalcements, 27 morphology of the auditory Müller's organ and measured activity of the auditory nerve, the transduction 28 current and electrophysiological properties of individual auditory receptors. We found that noise mildly and 29 transiently changes tympanal displacements, decreases both the width of the auditory nerve and the 30 transduction current recorded from individual auditory neurons. Whereas age -but not noise -decreases the 31 number of auditory neurons and increases their resting potential. Multiple other properties of Müller's organ 32 were unaffected by either age or noise including: the number of supporting cells in Müller's organ or the 33 nerve, membrane resistance and capacitance of the auditory neurons. The sound-evoked activity of the 34 auditory nerve decreased as a function of age and this decrease was exacerbated by noise, with the largest 35 difference during the middle of their life span. This 'middle-aged deafness' pattern of hearing loss mirrors that 36 found for humans exposed to noise early in their life. 37 Competing Interests: The authors declare no competing financial interests. Barnes for collecting preliminary data necessary to determine the length and sound amplitude of the 43 deafening tone that produced a temporary hearing loss. We thank Tom Austin for help with hook electrode 44 recordings. We acknowledge the help of James Blockley for help with analysis of hook electrode and laser 45 data. We acknowledge the help of the Kees Straatman and the Advanced Imaging Facility at the University 46 of Leicester. We would also like to thank Beth Fraser and Neil Rimmer for locust husbandry and Ben Cooper 47 and Brendan O'Connor for help with statistical analysis. We would like to thank the following for help with 48 blinding the treatment of the locust to the experimenter: Ben Cooper, Brendan O'Connor, Will Norton, Tom 49 Matheson, Swidbert Ott and Zoe Bailey. 50 51

Introduction: The multifaceted causes and consequences that lead to hearing loss 52
Nothing lasts forever. Left unchecked, systems inevitably fail. Two of the biggest causes are age and their 53 use. Ears are a fascinating and relevant biological example of this principle. Age-related hearing loss is 54 something that we will inevitably experience and noise-induced hearing loss is responsible for an estimated 55 third of all worldwide hearing loss (Bethesda, 1990). Audiometric measurements from workers in consistently 56 noisy environments -before the enforcement of ear protection -has gifted us a crucial quantitative and 57 longitudinal dataset to understand the interaction of noise and age on hearing loss (Burns and Robinson, 58 1970;Passchier-Vermeer, 1968, 1977. We know that threshold shifts due to noise can, with a small 59 corrective factor (Corso, 1980), be simply added to threshold shifts due to noise (Macrae, 1971; Miils, 60 1998ILR; Corso et al., 1976;Mollica, 1969). From this dataset Corso (1980) found that noise-exposed 61 individuals differed most from their silent counterparts during the middle of their life but are very similar in 62 older life. This is because age-related hearing loss dominates in later life. This simplistic mathematical 63 interaction of age and noise on hearing thresholds is in stark contrast to the complex and insidious 64 physiological deficits found in ears across animal phyla responsible for hearing loss. 65 The physiological and anatomical changes that co-occur with hearing loss present at multiple levels of the The most profound and best quantified change in the auditory system is loss of hair cells (Coleman, 1976;72 Keithley and Feldman, 1982;Bhattacharyya and Dayal, 1985;Dayal and Bhattacharyya, 1986;Li and 73 Hultcrantz, 1994;Dixon and Ward, 1971;Pinheiro et al., 1973;Clark et al., 1974;Moody et al., 1978). Loss 74 of hair cells can lead to loss of the spiral ganglion neurons onto which they synapse (Dupont et al., 1993;  Corfas, 2017). The complex correlation of multiple physiological deficits with hearing loss appears to have mistakenly led to the cause being assumed to the 79 most consistent deficit -for instance with age-related hearing loss being due to the most common stria 80 vascularis pathology (Ramadan and Schuknecht, 1989;Wu et al., 2020). In short, we lack the experiments 81 and quantitative data to track the progression of known changes in the auditory system and weigh up their 82 contribution to hearing loss 83 Adding to this complexity, it is not known which physiological deficits are caused by noise, age or a 84 combination of both. There are some breakthrough studies that have proportioned specific deficits into noise 85 or age Liberman, 2006, 2009;Fernandez et al., 2015). Kujawa and Liberman (2009) found that 86 ribbon synapses -those between hair cells and spiral ganglion neurons -sharply decrease after noise but 87 appear to be only mildly decreased with age. And Wu et al. (2020) found that most hearing loss can be 88 explained by hair cell loss not stria vascularis degeneration. 89 At present high-powered quantification of multiple physiological deficits in one model organism in response 90 to noise or age over the course of its lifespan are lacking. Further to this, age-related hearing loss and noise-91 induced hearing loss are thought to be very distinct processes (Yang et al., 2015), which could be 92 independent of each other (Corso et al., 1976). In this work we rigorously quantify the physiological changes 93 responsible for hearing loss -found across animal phyla -over the lifespan of the desert locust, Schistocerca 94 gregaria. Using a high-powered approach we measure: displacements in vivo of the tympanum, the 95 morphology and anatomy of the auditory Müller's organ, electrophysiological properties of the auditory 96 receptors and their transduction currents ex vivo and sound-induced auditory nerve responses in vivo. 97

Locust husbandry 99
Desert locusts (Schistocerca gregaria) were reared in crowded conditions (phase gregaria) on a 12:12 h 100 light:dark cycle at 36:25°C. Locusts were fed on a combination of fresh wheat and bran ab libitum. We used 101 nearly equal numbers of female and male Schistocerca gregaria from the Leicester 'Locust Labs' laboratory 102 stock. The cuticle of locusts continues to sclerotize up until they change (yellow for males, grey for females) 103 colour at ~ 10 days post their last moult. And recordings from single auditory neurons are not possible until 104 this time. Therefore we used locusts between 10 and 34 days post imaginal moult. In all experiments day 1 105 corresponds to locusts 10 days post their last moult. We used new locusts for each set of experiments. We 106 are willing and open to share our locust strains with other research groups. 107

Survival assay and life span in locusts 108
We counted the deaths of locusts over the 24 days ( Fig. 1) whilst subjecting the locusts to identical conditions 109 as the experimental locusts. Approximately 50% of locusts were dead at 12 days post first noise exposure or 110 mock noise exposure (for control group) and there was no stark difference between noise-exposed and 111 control locusts. After this time the rate of locust mortality decreased drastically so that very few locusts died 112 over the next 12 days until we stopped measuring. The wings of all locusts (control and noise-exposed) were cut off at their base to increase noise exposure of 118 the conditioning tone to their tympanal ears, which are otherwise covered by their wings. Between ten and 119 twenty locusts, for both the noise-exposed group and the control group, were placed in a cylindrical wire 120 mesh cage (8 cm diameter, 11 cm height). Both cages were placed directly under a speaker (Visaton FR 10 121 HM 4 OHM, RS Components Ltd). For the noise-exposed group only, the speaker was driven by a function 122 generator (Thurlby Thandar Instruments TG550, RS Components Ltd) and a sound amplifier (Monacor PA-123 702, Insight Direct Ltd) to produce a 3 kHz tone at 126 dB SPL, measured at the top of the cage where 124 locusts tended to accumulate. Throughout the paper we refer to noise that the locusts are exposed to as this 125 3 kHz 126 dB SPL pure tone. This tone was played continuously for 12 hours overnight (21:00-09:00) for the 126 noise-exposed group during their natural darkness period. The control group was housed in an identical cage 127 with a silent speaker for 12 hours. All recordings were performed within a twelve hour window during the day.

Biomechanical measurements of the tympanum with laser Doppler vibrometry 138
For in vivo measurements of the tympanum, locusts were mounted in natural dorso-ventral orientation 139 following removal of wings and hind legs, and fixed to a copper platform using blue tac (Bostik, Colombes, 140  Displacements were measured as the average peak-to-peak displacement. 154

In vivo hook electrode recordings from auditory nerve six 155
Locusts were secured ventral side up in plasticine. A section of the second and third ventral thoracic segment 156 was cut with a fine razor blade and removed with fine forceps. Tracheal air sacks were removed to expose 157 nerve six and the metathoracic ganglia. Hook electrodes constructed from silver wire 18 μm diameter 158 (AG549311, Advent Research Materials Ltd) were hooked under the nerve and the nerve was lifted out of 159 the haemolymph. A mixture of 70% Vaseline and 30% paraffin oil was applied through a syringe to coat the 160 auditory nerve to stop it drying out. Signals were amplified 1,000 times by a differential amplifier (Neurolog 161 System) then filtered with a 500 Hz high pass filter and a 50 kHz low pass filter. This amplified and filtered 162 data was sampled at 25 kHz by Spike2 (version 8) software running on Windows (version 10). To quantify 163 the compound spiking activity of the auditory nerve we used Matlab (Version R2020a, Mathworks Inc.) and 164 rectified the nerve signal and integrated the area underneath. We computed this for the 0.5 s of sound-evoked 165 neural activity and for 60 s background nerve activity before the tones and the background activity between 166 the tones. To compute the σ ratio we divided the sound-evoked response by the background neural activity. 167 N.B. the locust treatment was blinded to the experimenter until all data was collected and analysed. 168

Dissection of Müller's Organ and isolation of Group-III auditory neurons. 169
Whole cell patch clamp recordings per performed on group-III auditory neurons because they form the 170 majority of auditory neurons of Müller's organ (~46 out of ~80) (Jacobs et al., 1999), they are the most 171 sensitive auditory neurons of Müller's organ (Römer, 1976) and are broadly tuned to the 3 kHz we used for were low-pass filtered at 2.9 kHz with a four-pole Bessel filter. Compensation of the offset potential were 206 performed using the "automatic mode" of the EPC10 amplifier and the capacitive current was compensated 207 manually. The calculated liquid junction potential between the intracellular and extracellular solutions was 208 also compensated (15.6 mV; calculated with Patcher's-PowerTools plug-in from 209 www3.mpibpc.mpg.de/groups/neher/index.php?page=software). Series resistance was compensated at 210 77% with a time constant of 100 μs. 211

Staining and confocal microscopy 212
Locusts were secured ventral side up in plasticine. A square section of the second and third ventral thoracic 213 segment was cut with a fine razor blade and removed with fine forceps and set aside. Tracheal air sacks 214 were removed to expose nerve six and the metathoracic ganglia. The thoracic cavity was filled with locust 215 extracellular saline and auditory nerve six was pinched and broken at the top of the nerve, close to the 216 metathoracic ganglion with fine forceps and the cut end of nerve six placed on the thoracic cuticle outside 217 the thoracic cavity at the posterior end. A well was formed around the nerve end with petroleum jelly 218 (Vaseline, Boots) using a syringe (1.1mm diameter). The well was filled with Neurobiotin (5% m/v, SP-1120, 219 Vector Laboratories) dissolved in distilled water before a lid was fashioned with more petroleum to seal the 220 well. The square of thoracic cuticle was replaced onto the locust thoracic cavity and sealed back in place with 221 petroleum jelly, to prevent desiccation. Locusts were incubated overnight at 4°C to allow Neurobiotin to 222 diffuse along the nerve and fill the auditory neurons of Müller's organ. Following overnight incubation, whole 223 ears were excised from the first abdominal segment, as described above. Whole locust ears were fixed in 224 Fluorescence images (pixel size 0.31 μm 2 , with a total of 65 z stacks) were captured with a confocal 235 microscope (FV1000 CLSM, Olympus) equipped with Plan-UPlanSApo 20x (0.75 numerical aperture) lens. 236 Fluorescence emission of Dylight 488 was collected through a 505-530 nm bandpass filter and 237 fluorescence emission of DAPI was collected through a 485 nm low pass filter. Confocal images were 238 adjusted for contrast and brightness, overlaid and stacked in ImageJ (version 1.51, National Institutes of 239 Health). 240

Morphological analysis and cell counting of Müller's organ 241
To count the auditory neurons, a maximum intensity z-projection was produced for each z-stack image in 242 ImageJ, displaying the auditory neurons as a 2D image. Neurons were manually counted using the Cell 243 Counter plugin (Kurt de Vos, University of Sheffield, https://imagej.nih.gov/ij/plugins/cell-counter.html). 244 Contrast/ brightness parameters were adjusted to ensure all neurons were counted. 245 Imaris software was used for automated cell counts of the DAPI stained nuclei present in Müller's organ and 246 nerve. Automated cell counting was performed on 3D images using the "Spots" tool, with consistent 247 parameters (15 μm spot size) used across all Müller's organs. The "Surfaces" tool was used to ensure only 248 cells within the Müller's organ were included in the automated counting. The width of the auditory nerve was 249 also measured in Imaris using the MeasurementPro tool. 250

Experimental design and statistical analysis 251
We designed all experiments to have a power above 95%, which gives a false positive rate (probability of 252 finding an effect that is not there) of 5%. In order to calculate the power we used the raw data, and the  locust (noise-exposed or control) was blinded to the experimenter; lone working conditions, due to Covid 281 restrictions, made complete blinding impossible. All data either remained blinded or was recoded to be 282 completely blind when analysing the data to avoid unconscious bias. 283 To test for differences and interactions between control, noise-exposed and aged locusts we used either a 284 differences. Curves where fitted to the data using Matlab (version R2018a) for hook electrode recordings or 290 the drm package in R for patch-clamp recordings (Ritz, 2016). The drm package was also used to compute 291 t and p values when comparing control and noise-exposed four part Log-Linear models. F statistics of the 292 Log-Linear model fits were computed by excluding treatment (noise-exposed or control) as a factor. Higher 293

F statistics donate a stronger effect of treatment. 294
In order to compare responses between noise-exposed and control locusts across SPLs we adopted an 295 approach first implemented in pharmacology research. In our work the "dose-response curves" are equivalent 296 to SPL-auditory response curves. This allowed us to maximise the information contained in each dataset and 297 to quantitatively compare model parameters such as: Hill coefficient (steepness of slope), maximal asymptote 298 (maximum σ ratio), and inflexion point (σ ratio at the steepest part of the slope). We did this using the drm 299 function of drc package (Version 3.1-1, Ritz and Strebig, 2015). We fitted four-part Log-Linear models with 300 auditory nerve responses (σ ratio) as the dependent variable with treatment (control or noise-exposed) and 301 SPL as the independent variables. This analysis was done for each day and the t and p value reported for 302 each model parameter: Hill coefficient, maximal asymptote and inflexion point on each graph in Figures 2  303   and 3. 304 To test whether the factor of treatment (noise-exposed or control) significantly affected auditory nerve 305 response we compared the above model to a model in which treatment was omitted as an independent 306 variable, using the anova function (Ritz et al., 2015). This gave an F statistic labelled on each graph in Figures  307 2 and 3. The p value for this analysis remained below 0.001 for all graphs. We used identical analysis for 308 whole-cell patch-clamp data where transduction current was the dependant variable. 309 Analysed datasets are freely available to download: https://www2.le.ac.uk/departments/npb/people/bw120. contactless laser Doppler vibrometry. Tone-evoked displacements rose above the noise floor above sound 318 amplitudes of 60 dB SPL ( Fig. 2A-C) and higher SPLs remained linear on a log-log plot. Thus, tone-evoked 319 tympanal displacements can be explained by a simple power law. Directly after, and at 48 hours after noise 320 exposure, there was no difference in tone-evoked tympanal displacements for control and noise exposed 321 locusts (LMEM with SPL as a random effect; t(297)=-0.152, p=0.879; t(209)=0.069, p=0.945). The tympanal 322 displacements of the noise-exposed locusts 24 hours after noise exposure were higher (t(298)=-3.05, 323 p=0.0025).

In vivo electrophysiology 333
Alix Blockley quantified the in vivo performance of the auditory neurons to produce sound-evoked spikes in 334 response to repeated noise exposure over an extended 24 day period. She recorded spiking activity directly 335 from the auditory nerve (nerve 6) using hook electrodes in response to 3 kHz tones produced by a speaker 336 (Fig. 3A, B). A 3 kHz tone matched the frequency of our noise exposure and the frequency to which the 337 majority of auditory neurons (Group III neurons) in Müller's organ are broadly tuned. These recordings left the bilateral ears, in the first abdominal segment, intact (Fig. 3A). She exposed locusts to a 126 dB SPL 3 339 kHz tone overnight for 12 hours every three days and recorded nerve activity at days: 1,2,3,4,5,6,12,18,24 340 ( Fig. 3 C). 341 The tone-evoked activity of the auditory nerve were composed of biphasic spikes that summated together at 342 tone onset and for loud SPLs. Therefore, to quantify the tone-evoked activity of the auditory nerve she 343 rectified the tone-evoked signals, integrated the area underneath and averaged the three tone presentations 344 at each SPL. She then computed the σ ratio by dividing the average area during the 0.5 s tone by the average 345 area over 0.5s during 60 seconds of background activity before tone presentations began (Fig. 3Bii). This 346 normalises for differences in background spiking activity and noise of the recording. To maximise the 347 information from the recordings we fitted four part Log-Linear equations to model the relationship between 348 transduction current and SPL (Fig. 3Di-Dix). We compared three specific parameters of four part Log-Linear 349 equations between the nerve response of control and noise exposed locusts: Hill coefficient, maximum   N.B. days 1 and 4 where directly after noise-exposure and days 2 and 5 24hrs after noise 372 exposure. Recordings on days 3, 6, 12, 18, 24 where all 48 hrs after noise-exposure and directly before the 373 next noise exposure). Cii Plot of the maximal σ ratio for control and noise exposed locusts at each recording 374 day with positive standard deviation plotted. Di Plot of the σ ratio (nerve response) as a function of SPL of a 375 3 kHz tone for control and noise exposed locusts within 12 hours after cessation of noise exposure. 376 Responses for individual locusts are plotted as thin grey or thin red lines for control and noise-exposed 377 respectively. Dots are the average for each treatment at each SPL, shaded area is the positive standard 378 deviation at each SPL and thick lines are four-parameter Log-Linear fits for each treatment. The extent of the 379 difference between each group and a statistical test of a difference are displayed as t and p values next to 380 three icons donating (from top to bottom) Hill coefficient, maximum asymptote and the inflexion point (p values 381 below 0.05 are in bold). The F statistic compares the fit of the model when the treatment of the group (noise-382 exposed or control) was omitted. Higher F statistics donate a larger effect of treatment. Dii and Diii plot σ 383 ratio against SPL for noise-exposed and control locusts 24 hrs and 48 hrs after cessation of noise exposure 384 (or silent speaker for control). Div is directly after a second noise exposure and Dv and Dvi 24hrs and 48hrs 385 after the second noise exposure. Dvii, Dviii, and Dix are 48 hours after the 4 th , 6 th and 8 th noise exposure 386 respectively. 387

Morphology and number of auditory neurons and supporting cells in Müllers organ 388
Daisy Ogle backfilled the auditory nerves of control locusts and noise-exposed locusts in vivo with of auditory neurons in noise-exposed and control locusts' ears over 24 days. F Number of cells in Müller's 416 organ in noise-exposed and control locusts over 24 days. 417 Plot of average maximum transduction current for noise-exposed and control locusts for each day. Di Plot of 422 the transduction current as a function of SPL of a 3 kHz tone for control and noise exposed locusts within 12 423 hours after cessation of noise exposure. Transduction currents for individual locusts are plotted as thin grey 424 or thin red lines for control and noise-exposed respectively. Dots are the average transduction current for 425 each treatment at each SPL and shaded region is positive standard deviation for each treatment and SPL. The F statistic compares the fit of the model when the treatment of the group (noise-exposed or control) was 430 omitted. Higher F statistics donate a larger effect of treatment. Dii and Diii plot transduction current against 431 SPL for noise-exposed and control locusts 24 hrs and 48 hrs cessation of noise exposure (or silent speaker 432 for control). Div, Dv, Dvi and Dvii are 48 hours after the 2 nd , 4 th , 6 th and 8 th noise exposure respectively. E 433 Plots the average discrete depolarisation amplitude for each auditory neuron for auditory neurons from noise-434 exposed and control locusts. The extent of the difference and the statistical test for any difference are plotted 435 as t and p values respectively. F, G, H Plots the membrane resistance, membrane capacitance and 436 membrane potential for each auditory neuron from noise-exposed and control locusts. Linear models were 437 used to test for differences for electrophysiological properties of auditory neurons from control and noise-438 exposed locusts, with their t and p values displayed on the figures. 439

Electrical properties of auditory neurons and transduction current dependence on sound pressure level 440
We measured the electrical properties and tone-elicited currents of Group III neurons in Müller's organ 441 through whole-cell patch-clamp recordings from individual neurons from excised ears (Fig. 5A). We isolated 442 and optimised the transduction current at the distal ciliated end of the auditory neuron using pharmacology, 443 voltage protocols and the optimal sound stimulus ( Fig. 5B) (detailed in methods). We recorded directly after 444 the first 12 hour noise exposure at day 1 then 24 hours and 48 hours afterwards on day 2 and 3. Recordings 445 at day 6, 12, 18, 24 were all made 48 hours after the last 12 hour noise exposure (Fig. 5C). The transduction 446 current increased with an increased sound amplitude and followed a log-linear model (Fig. 5D). Directly after 447 noise-exposure the maximum transduction current is reduced (Fig. 4Di, t=(17)-3.64, p=0.0018, Cohen's effect 448 size d (at 90 dB SPL)= -1.03) and reduces further 24 hours after noise exposure (Fig. 5Dii, t=(17)-8.08,  449 p<0.00001 Cohen's effect size d (at 100 dB SPL)= -1.28). Only 48 hours after the first noise exposure is the 450 maximum transduction current comparable with non-noise-exposed controls. Interestingly the Hill coefficient, 451 which mathematically derives the interaction between sound pressure level and transduction current, does 452 not change directly after noise-exposure ( Fig. 5Di t=(17)1.66, p=0.113), but changed after 24 hours (Fig.5Dii  453 t=3.76, p=0.0014 and remains different 48 hours later (Fig. 5Diii, t=(17)3.87, p=0.0011). After two noise 454 exposures and a 48 hour recovery (Day 6) both groups are still fairly similar (Fig. 5Div). Upon further noise-455 exposure there is further decline in the maximum transduction current of the noise-exposed group (Fig. 5C, 456 Dv-Dvii Cohen's effect size d (at 110 dB SPL)= -1.24) with the F statistic at its highest (492) when comparing 457 control and noise-exposed locusts at 24 days whereas the control locusts show no age-related decline (Fig.  458 5C, Dv-Dvii). 459 We recorded smaller discrete depolarisations in the auditory neurons of noise-exposed locust auditory 460 Repeated or prolonged noise exposure causes hearing loss which, for humans, is most different from non-470 noise-exposed individuals during the middle of their life (Passchier-Vermeer, 1974). As humans age, age-471 related hearing loss dominates and noise-exposure plays less of a role in determining hearing loss (Corso, 472 1980;Echt et al., 1998). The interaction of age and noise on hearing loss is only known for the human auditory 473 system thanks to extensive longitudinal audiometric measurements including from industrial workers before 474 ear protection (Passchier-vermeer, 1968;Corso, 1963). Here, we repeatedly exposed large numbers of 475 locusts to noise and quantified their ability to hear over their life by measuring the response of their auditory 476 nerve to sound. We also measured components of the auditory system known to correlate with hearing loss: 477 decrease in auditory receptors, auditory nerve thinning, decrease of electrochemical gradients and other 478 components crucial to the function of the auditory system: supporting cells, auditory receptor 479 electrophysiological properties and transduction. We found striking parallels with the pattern of human 480 hearing loss and rigorously quantify -for the first time in any auditory system -how noise and age affect 481 specific components in the ear. 482 Before sound can impinge on the delicate inner ear sensory structures it must be captured mechanically 483 through a tympanum (for sound pressure changes) or antennae (for particle velocity changes). Although We measured compound spiking potentials from the auditory nerve in response to a 3 kHz pure tone at SPLs 496 from 20-110 dB SPL. We found a decrease in maximal nerve response directly after noise exposure (Fig.  497 3Di, Div). This is similar to a decrease in the first wave of auditory brainstem responses (ABRs) and . Noise-exposed locusts recovered their maximal nerve response within 24 hours 500 after first noise exposure but it then decreased again at 48 hours later (Fig. Di-Diii), a 48 hour pattern repeated 501 after the second noise-exposure on day 4 (Div-Dvi). By comparison mice (with an accelerated hearing loss 502 phenotype) recover quickest in the first three days before their recovery slows with no sign of a reversal 503 apparent within the days measured (straight after, 1, 3, 7 and 14 days after noise exposure) (Chuang et a., Liberman, 2009). Although the reversal of recovery found for locusts appears paradoxical, no study has 507 measured recovery twice within 48 hours to uncover short term changes. This pattern of recovery mirrors 508 some traumatic brain injuries where initial recovery can be followed by worsening of symptoms. The 509 physiological basis of such non-linear recovery is due to different aetiologies presenting over distinct time 510 courses. In the case of TBI initial electrophysiological dysfunction is followed by structural damage then toxic-511 metabolic and vascular effects (Giza and Hovda, 2001). We suggest that noise exposure has a complex 512 recovery process where an initial promising recovery is followed shortly after by deterioration and is the result 513 of multiple processes, triggered by noise exposure, acting over distinct time courses. We do not yet know if 514 this reversal of recovery after noise exposure is an exception compared to the ears of other animals exposed 515 to noise. 516 The long-term effects of repeated noise exposure result in a decreased maximum nerve response and higher 517 SPL at the inflexion point (indicative of increased hearing thresholds). But age-related hearing loss leads to 518 narrowing of the differences between control and noise-exposed locusts at older ages. This follows human 519 hearing loss -the only other organism with a high enough powered comparative longitudinal dataset (Corso, 520 1980) and suggests this pattern of degeneration is fundamental across diverse biological auditory systems. exposure, used here to deafen locusts, is at similar sound levels (to the above studies) but resulted in little 524 effect on the loss of locust auditory receptors. Instead auditory receptor number decreased steadily due to 525 age with a mild compounding effect of noise. Mild (100 dB SPL) noise exposure in mice appears to do little 526 to compound a strongly age-related auditory receptor decline (Fernandez et al., 2015). If this is similar in 527 humans then (we cautiously conclude) noise-induced hearing loss may have little to do with loss of hair cells 528 consequence of noise exposure. The number of non-sensory cell types has not been counted in any auditory 532 system and we found no decrease in non-neuronal cell types, either in the nerve or Müller's organ, in the 533 locusts either as a function of noise or age. 534 In individual auditory neurons of Müller's organ, membrane resistance and cell capacitance were unaffected 535 either by noise or age. There was a trend for older auditory neurons (irrespective of treatment) to be more 536 depolarised ~3 mV, although the magnitude of the change is small (t=-2.382, p=0.0177). Whilst there are no 537 changes in the membrane resistance and membrane potential of hair cells of mice, their capacitance 538 decreased as a function of age (Jeng et al., 2020). Properties of the transduction current in mice were 539 unaffected by age which matches that found here for auditory neurons of locusts. This reveals remarkable 540 homeostatic mechanisms at play to maintain auditory receptor function. Repeated exposure to noise, 541 however, results in a steady decrease in the electrochemical driving force for the transduction current. In 542 mammals the electrochemical driving force for the scala media is established by specialised ion-pumping 543 cells in the stria vascularis. Dysfunction of the stria vascularis has been recently popularised as responsible 544 for the majority of age-related hearing loss (Ramadan and Schuknecht, 1989;Vaden et al., 2017). More 545 recent interpretations (Wu et al., 2020) suggest that although the correlation of stria vascularis atrophy in 546 patients with hearing loss is high (Schuknecht, 1974;Schuknecht, 1993) it does little to predict the extent of 547 hearing loss. 548 Despite the clear anatomical differences in auditory systems of insects and vertebrates they share equivalent 549 components: transduction on ciliated endings, electrochemical gradient across the auditory receptors and 550 spiking axons that carry auditory information to the central nervous system. In addition, they share 551 developmentally interchangeable transcription factors for ear development (Wang et al., 2002) and evolved 552 from the same ancestral ciliated sensory appendage (Nowotny and Warren, 2020). As such, insects are a 553 valuable resource to derive general principles of how auditory systems are affected by age and noise. This 554 is keenly evidenced by the similarities in the pattern, causes and consequences of hearing loss which stand 555 in stark contrast to what can be "misleading" anatomical differences. 556 557

Author contributions 558
Charlie Woodrow and Alix Blockley collected and analysed the Doppler laser data (Fig. 2). Alix Blockley and 559 analysed the hook electrode data (Fig 3.). Daisy Ogle collected and analysed all morphological data on 560 Müller's organs and composed Figure 4. Alix together with Ben Warren conceived the experimental workflow 561 of the paper. BW conceived the idea for the paper, collected and analysed the patch-clamp 562 electrophysiological data and composed Figure 5, wrote the paper and performed all statistical analyses. All 563 authors helped refine figures, write the paper and interpret data. 564