Anthropogenic habitat change is occurring rapidly, and organisms can respond through within-generation responses that improve the match between their phenotype and the novel conditions they encounter. But, plastic responses can be adaptive or maladaptive and are most likely to be adaptive only when contemporary conditions reasonably mimic something experienced historically to which a response has already evolved. Noise pollution is a ubiquitous anthropogenic stressor that accompanies expanding urbanization. We tested whether the amplitude of traffic noise influences a suite of fitness-related traits (e.g. survival, life history, reproductive investment, immunity) and whether that depends on the life stage at which the noise is experienced (juvenile or adult). Our treatments mimic the conditions experienced by animals living in urban roadside environments with variable vehicle types, but continuous movement of traffic. We used the Pacific field cricket, an acoustically communicating insect that was previously shown to experience some negative behavioral and life history responses to very loud, variable traffic noise, as a model system. 

General experimental design

We used a fully factorial design with four possible noise levels experienced during development and/or at adulthood. Each cricket had their own unique "id." We assigned juvenile crickets to one of four acoustic environments: a no traffic noise treatment or one of three chronic traffic noise treatments (50dBA, 60dBA, or 70dBA) as seen in the column "juv_treatment." At adulthood, we randomly reassigned the crickets to a noise treatment, as seen in the "adult treatment" column. The "trial_type" column represents the combination of the juvenile and adult treatments for each individual cricket (juvenile:adult). We then measured a suite of fitness-related traits, which we categorized as related to basic life-history (survival, development time, and adult size), reproductive investment (mating success, sperm viability, number of eggs hatched after 1 week of laying, and male and female investment in reproductive organs), and immunity (hemocyte counts and melanization of a foreign body).

Rearing and traffic noise exposure

To expose crickets to different noise levels, we reared the animals inside of replicate, acoustically isolated Percival incubators (model I36VLC). Because the incubators themselves generate white noise when on, we left the power off, but retained a 12:12 light:dark schedule in each using clamp lights outfitted with 60W LED bulbs attached to mechanical timers. The lighting also maintained incubator temperatures within those naturally experienced by the animals (21°C to 23°C), and we maintained 60-65% humidity by placing a 1.89L bin of water in each incubator. We measured temperature and humidity every two weeks.

Each acoustically isolated incubator housed a Bluetooth EcoXBT speaker that either played nothing (leaving background ambient noise of 37-38dBA only; hereafter “silent”) or an uncompressed .wav traffic noise track standardized to 50dBA, 60dBA, or 70dBA (when measured at 1m away). We created the traffic tracks following Gurule-Small et al. 2018 and used the same traffic noise recordings as those studies. In brief, we first spliced together ten randomly chosen 30 second clips of traffic noise originally recorded for Gurule-Small et al. 2018 to create a five-minute track. To investigate how noise levels impact plastic responses to noise, we next leveled the sound from each recording clip so that the average amplitude was consistent across the track. A sound engineer (JHG) adjusted the gain of each clip in Logic Pro X (version 10.4.8, Apple Inc., Los Altos, CA, USA) to level the sound so that each segment and the full five-minute track was at the appropriate amplitude (RMS level) for each treatment (50dBA, 60dBA, or 70dBA). Noise treatments were broadcast to crickets living inside of the incubators 24 hours a day (including during the daily peaks of communication by song), mimicking noise that would be experienced near a highway with high traffic fluidity. We measured the amplitude of our treatments using a PCE-430 sound level meter and class 1 microphone at 1m away from the speaker inside of the Percival incubators. Every two weeks we confirmed the amplitude of the traffic noise tracks.

From October 2020 to August 2021, we isolated Pacific field crickets just prior to their 2nd instar (~14 days after hatching, as early as was possible without causing mortality) from a lab-reared population that was originally collected in Hilo, Hawaii in 2017. We randomly assigned them to one of the four traffic noise treatments ("juv_treatment") and housed them in groups of 15 in 1.89L boxes inside of the incubators for the duration of their developmental exposure. Each 1.89L box contained wet cotton for water, egg carton for shelter, and ad libitum Flukers Cricket Chow for food. We did this eight times over the course of eight months, yielding eight cohorts of crickets. For each cohort ("cohort" column), we isolated 4-7 boxes ("box" column) of crickets per treatment, depending on the number of animals available in the lab colony, for a total of 3060 crickets assigned to treatments. The "cohort_box" column represents the combination of the cricket's cohort letter and box number.

We provided clean housing and fresh food and water twice per week. To ensure that animals remained unmated prior to mating trials (see below), we isolated them individually in 0.47L deli cups when they approached eclosion (during the final instar before adulthood) with water, shelter and food (ad libitum Kaytee rabbit chow). We checked the deli cups twice weekly for eclosion. At eclosion we randomly reassigned individuals to one of the four noise treatments ("adult_treatment"). Throughout the experiment, we changed the stacking order of the boxes within incubators once a week (to account for the possibility that crickets would experience different loudness of noise depending on their location within the incubator) and rotated treatments among incubators every two weeks to avoid incubator effects.

Assay Overview

To measure a comprehensive set of fitness-related traits following developmental and/or adult exposure to different noise levels, we used the following general workflow: (1) at eclosion we determined development time, survival to eclosion, and measured pronotum width; (2) two weeks later we assessed mating success in controlled mating trials, allowed females to lay eggs for one week, and measured male sperm viability; (3) following the sperm viability assay or week of egg laying, we moved the crickets into the immune assays; (4) finally, when all other assays were complete, we dissected out and weighed the reproductive organs of all males and females and determined survival from eclosion to dissection. 

Life History Trait Assays

When we assigned newly eclosed adults to their adult noise treatments, we measured the width of their pronotum ("pronotum" column) to the nearest mm using digital calipers (n = 454), recorded their sex ("sex" column), and recorded the date on which they eclosed. To obtain a metric of development time ("days_to_eclosion" column), we subtracted the date at which they were assigned to treatments ("hatch_date" column) from the date of eclosion ("ecolsion_upper" column). Thus, development time here is from treatment assignment (0-14 days after hatching) to adulthood (assay: development time, n = 518). Survival to eclosion is the percent of crickets that lived to eclosion of the original 3060 crickets assigned to treatments (assay: juvenile survival, n = 3060), whereas the survival to dissection describes the percent of survivors from eclosion to the end of the experiment (assay: adult survival, n = 518). 

Reproductive Investment Assays

Mating Trials

To determine whether experience with noise impacts the likelihood of individuals mating, we conducted standardized no-choice mating trials. We allowed each individual to mate only once. Whenever possible, we paired males and females from the same adult treatment for mating, but because the sexes eclosed at different rates this was not always possible. In these cases, we paired the focal cricket with a mate pulled randomly from the large, freely breeding laboratory colony (51% of trials; whether the mate was pulled from the breeding colony or the experimental group is included as a covariate in the model). The column "pop_pair" indicates whether this stimulus cricket came from the same experimental treatment or the laboratory colony, "pair_virgin" indicates whether this stimulus cricket was a virgin, and "pronotum_pair" indicates the pronotum size of the stimulus male. Each mating trial consisted of placing the male and female together inside of a deli cup (9cm diameter) at ambient room temperature under dim light for up to two hours each day for up to four days (or until a single successful mating took place). We checked deli cups every 15 minutes for successful mating as evidenced by spermatophore transfer. After the crickets mated, or four days of unsuccessful mating trials, we again isolated the males and females in separate deli cups. We recorded the number of days that they failed in the "num_fails" column, whether they mated in the "mated" column, and the date of successful mating in the "date_mating" column. Using the date of the mating and the cricket's ecolsion date, we calculated the "age_mating." Not all individuals successfully mated in the time allotted, so we were able to assess mating success (mated or not mated in four opportunities) (assay: mated, n = 445).

Number of Hatchlings

Next, allowed all females who mated successfully to lay eggs for one week. After one week, we recorded the date as "end_egg_lay." We then placed each female’s cotton pad in a labeled Tupperware container, providing food and shelter when hatchlings appeared. Not all females that mated had offspring hatch, allowing us to assess hatching success (zero live hatchlings or more than zero live hatchlings) (assay: hatching success, n = 178). Five weeks after placing the cotton pad in the Tupperware, we counted the number of offspring ("num_f2" column, n = 138). We recorded the date we counted the number of offspring in the "date_f2_counted" column and calculated the number of days these offspring had to hatch in the "days_F2_hatched" column.

Sperm Viability

Following successful mating trials, we isolated males for sperm viability analysis the following day. If males did not mate in the four days (n = 18/243, 7.4%) or if we were not able to conduct the assay the exact day after mating, we standardized time since spermatophore production by manually removing their spermatophore (externally with no injury to the cricket) 24 hours before their sperm viability assay. We used a THERMOFisher LIVE/DEAD sperm viability kit to stain live and dead sperm following the protocol outlined in Garcia-Gonzalez et al. 2005. (assay: sperm viability, n = 167). Briefly, we removed a spermatophore non-invasively from the male cricket, placed it on a glass slide in Beadle saline, and cut it open with dissection scissors to evacuate the sperm. We pipetted 5µL of the sperm and Beadle saline mixture to a clean part of the glass slide and gently mixed it with a sterilized pin. Then, we stained the sperm with SYBR-14 and propidium iodide (each addition followed by a 10-minute incubation period in the dark) and photographed the sperm with a Leica M165FC scope outfitted with an EC3 camera on a computer running LAS X imaging software. The GFP3 (blue) and DSR (green) fluorescent filters allowed us to take pictures of live (green) and dead (red) sperm, respectively, from the same view window on each sample. We recorded the date on which we performed this assay ("date_sv") and used the cricket's eclosion date to calculate the "age_sv."

We then crowdsourced the live and dead sperm counting using the world’s largest community science platform Zooniverse ([61]; https://www.zooniverse.org/projects/marywestwood/the-cricket-wing). We divided each sperm image into 36 smaller images and uploaded the resulting 12,304 images to our Zooniverse project. We developed a detailed protocol and training tutorial that taught volunteers to identify sperm cells and click on each cell in an image that either showed live (green) or dead (red) sperm. Volunteers were not aware of the crickets’ noise treatment, and we had 784 volunteers count sperm cells. Clicks were automatically counted and deposited in a spreadsheet accessible to the researchers. Each image was counted 6 times. We also scanned for errors in all data caused by such issues as volunteers submitting highly inaccurate or empty counts, and removed these from the dataset. Next, we calculated the mean number of sperm in each image and identified outlier images (those below the 1st quartile or above the 3rd quartile); we did not find that certain individual volunteers regularly submitted counts that were identified as outliers. Thus, the “wisdom of the crowd” offset any one-off outlier counts. After removing erroneous counts, each image was counted an average of 5.2 +/- 0.98 times. Finally, to calculate sperm viability, we summed the counts for each male cricket across the 36 images and divided the number of live sperm by the number of dead sperm. Numbers of live sperm are found in the "live" column, number of dead sperm in the "dead.y" column, and sperm viabilty in the "sperm_viability" column.

Reproductive Organs

We dissected male and female reproductive organs after the completion of all other assays. We recorded the date we dissected them ("date_weigh_w") and used this date and the eclosion date to calculate "age_dissection." We froze crickets at -20°C for 15 minutes to euthanize them, then dissected them under a dissecting scope (Wild Heerbrugg M3Z) to extract and weigh reproductive organs. We recorded the initials of the dissector in the "dissector_initials" column. For males we removed the testes, spermatophore mold, and accessory glands (assays: testes, n = 158, spermatophore mold, n = 149, accessory gland, n = 157). For females we removed the ovaries (assay: ovaries, n = 196). We weighed each organ type separately on a VWR-64B scale immediately after dissection. Females were 33.3 +/- 9.4 days post eclosion and males were 34.8 +/- 13.8 days post eclosion at the time of dissection (age is included as a covariate in the reproductive organ models; see below). Female ovary weight is recorded in the "ovary_w" column, male testes in the "testes_w" column, spermatophore mold in the "smold_w" column, and accessory glands in the "ag_w" column. All weights were measured in grams.

Immunity Assays

We assessed cricket immunity using hemocyte counts and melanization. We conducted the hemocyte counts after sperm viability (males) or one week of egg laying (females) (average age: 25 +/- 3.8 days post eclosion (females) and 23.9 +/- 5.4 days post eclosion (males)). We counted the numbers of two different types of hemocytes, plasmatocytes ("total_count_plasmatocytes" column) and granulocytes ("" column), following Triggs et al. 2012. We placed the crickets at 5°C for two minutes to anesthetize them, and then poked the cricket’s pronotum using a sterilized pin and pipetted 2µL of the hemolymph that emerged into 4µL of an anticoagulant buffer Smilanich et al. 2018. We then pipetted 5µL of the mixture onto a Weber Scientific Hemocytometer and placed it under a Keyence VXH Digital Microscope with a Keyence VH-Z100UR/W/T lens. We imaged the 5x5 grid of the hemocytometer at 400X and employed the Zooniverse platform again to count the number of each hemocyte type (on average 10 counts per image; assay: granulocytes, n= 213; plasmatocytes, n = 212). We recorded this date as the "date_hem" and used the ecolsion date to calculate the "age_immunity." We recorded the total number of hemocytes in the "total_hems" column.

Immediately following the hemocyte assay, we inserted a 3mm long piece of nylon monofilament fishing line with a knot tied at the end into the previously created hole in the cricket’s pronotum. We left the filament in the crickets for 24 hours then removed and imaged the filaments under a Keyence VHX Digital Microscope (Keyence Corporation, Itasca, IL USA) scope with a Keyence VH-Z20R/W/T lens at 50x magnification, recording this date as "date_fil_out". We used the GNU Image Manipulation Program (v 2.10) to measure the amount of melanization on the filaments (assay: melanization, n = 359). The level of melanization was calculated in the "mean_white" column.

Survival Measurements

To calculate the proportion of individuals that survived to eclosion and dissection, we created a separate data sheet titled “20231027_F1_SurvivalDataset.” All column names with the same name as those in “20231027_F1_CompleteDataset” contain the same type of data. However, each box of crickets started with 15 individuals, but many died before reaching adulthood and therefore never received an ID. The “individual” column provides a temporary individual ID for each of these crickets. The “cohort_start_date” column contains the date the crickets were placed in the treatments, and the “dead_date” contains the day the individual cricket died. These two dates were used to calculate “days_until_death.”

Date Cleaning

Following data collection, we cleaned our dataset by fixing any errors, such as typos, and making sure there was only one row for each individual ID.

Statistical Analyses

We performed all statistical analyses in R (v2022.12.0+353). First, we visualized histograms and q-q plots and determined the best distribution fit for each variable (see accompanying code and data); all variables fit the assumptions of a normal distribution. We then ran linear mixed-effect models to test whether the continuous fitness traits we measured depended on juvenile and/or adult noise treatment and generalized linear mixed-effect models for the four binomial traits (juvenile survival, mated or not, successful hatching or not, and adult survival). The basic model structure included juvenile treatment (silent, 50dBA, 60dBA, or 70dBA), adult treatment (silent, 50dBA, 60dBA, or 70dBA), and their interaction as main effects; the interaction allowed us to determine whether plasticity is stage dependent. For pronotum, development time, and juvenile survival, juvenile noise treatment (silent, 50dBA, 60dBA, or 70dBA) was the only main effect because the crickets had not experienced the adult noise treatment at the time of the assays. All models included the cohort blocking variable as a random effect. We Bonferroni corrected all p-values to account for the number of models run. 

We included several appropriate covariates, and these differed across models. For pronotum, development time, and adult survival, we included sex as a covariate. The adult survival model also included mating success. The binomial model addressing whether crickets mated or not also included sex as a covariate as well as whether the cricket they were paired with was from the experimental treatment or from the breeding colony. Both the binomial model assessing mated females’ hatching success and the linear model investigating the number of hatchlings included pronotum size and age at the end of the egg-laying period as covariates. We included two covariates in the model assessing sperm viability, pronotum size, and age at the time of the assay. The model structures for all reproductive organs included pronotum size and age at dissection as covariates. In the reproductive organ models, we also included whether the cricket successfully mated or not as a covariate, as mating may generate differences in reproductive organ mass. For the ovary model, we also included the number of hatchlings as a covariate. Finally, for both hemocyte models (number of granulocytes, number of plasmatocytes) and the melanization model, the covariates were pronotum size, mating success, and age at the time of the assay.