Plateau pikas were live-trapped using the using nooses anchored at burrow entrances with chopsticks (see details in the Qu et al., 2018) at the alpine meadow of Geermu, Qinghai in 10th-12th August 2019, brought back and raised in the animal laboratory at the Northwest Institute of Plateau Biology, Chinese Academy of Sciences. The trap site was about 2.0 km away from the road to ensure that the plateau pikas were not exposed to traffic noise in the natural environment.

Plateau pikas were maintained in separate polypropylene cages (47 cm × 35 cm × 20 cm) under natural light cycle condition, and the ambient temperature was kept 23 ± 1°C. Their gender was tested, and ID information were marked on the cages. The pikas were fed standard rabbit pellet (Beijing Ke Ao Food, Co., China) and pure water ad libitum. They were kept in the laboratory and adapted for 2 weeks. Then, 16 adult males and 12 adult females were used for the following experiment.

We recorded the traffic noise for playback experiments at Qinghai-Tibet highway (G109) at about 3,002 milestone. Recordings were collected 20 m away from the edge of the paved road at a height of 1 m, where plateau pikas are frequently found, on August 15^th, 2019 from 09:00 am to 12:00 am, and lasted for 3 h. A total of 1,430 vehicles were recorded during this period. We selected such date and time because they represented the normal vehicle flux during the season, according to the records of the local department of transportation. A portable sound level meter (DELIXI 2201, 0.1 dB precision) was used to measure the mean amplitude (dB) of the traffic noise, and SONY PCM–D100 recorder was utilized to obtain sounds. From the 2 h recording, a 30 min segment was selected and normalized to a mean amplitude level using Praat v 6.2 software.

The behavior and physiology of all pikas were measured at the start of the experiment in early September. Then, the pikas were randomly divided into two treatment groups (“treat” for short), i.e., “Noise” and “Control” groups, and housed in separate but adjoining rooms (18 m²) with the same conditions. In the “Noise” group room, traffic noise was broadcasted. In the “Control” group room, a silent broadcast was conducted during the whole experiment. The treatment lasted for 30 days. At the end of the experiment, the behavior and physiology of all pikas were measured once again using the same procedure.

Traffic noise was broadcasted using an amplified loudspeaker connected to a radio at the center of the “Noise” group room, with all the cages spread around the room to ensure that all the pikas heard the same noise. The noise was loop broadcasted during daytime from 07:00 am to 19:00 pm at 80 ± 2 dB sound pressure level to simulate the natural acoustic environment. Although a certain traffic noise was observed at night in G109, we did not broadcast it in the experiment because pikas are inactive and live underground overnight (Zong and Xia, 1987).

We conducted behavioral and physiological assays in early September and early October 2019, i.e., “Pro” and “Post” (“trial” for short) the traffic noise treatments, respectively. The behavioral and physiological parameters were tested in the following sequence:

We calculated the repeatability of the heart rate, exploration, RMR, and CORT as an intra-class correlation coefficient from one-way ANOVA with treat as a fixed factor using the “rptR” package (Stoffel et al., 2017). Linear mixed-effects models (LMMs) were used to test the potential effect of traffic noise on these behavioral and physiological traits. We assessed the existence of correlations between the heart rate, exploration, RMR, and CORT of each individual from different treatment and trial groups by Spearman rank correlations using the “Hmisc” package. The LMM allows the estimation of the effects of explanatory variables and their interactions as fixed effects, with ID as random effects. LMMs were fitted using the “lmer” package, and “lmerTest” package was used to obtain the summary table and p-values for LMMs via Satterthwaite’s degrees of freedom method (Kuznetsova et al., 2017).

R 4.2.1 software (R Development Core Team, 2022) was used for all statistical analyses.

The animal study was reviewed and approved by The Ethics Committee of Northwest Institute of Plateau Biology, Chinese Academy of Sciences (NWIPB-20190727). The procedure for capturing, feeding, and operating on the plateau pikas followed the guidelines of the ethical approval standard of the committee. No effect of transport and experiment was observed on pika mortality.

According to the variance proportions explained by between-individual variations, the behavioral and physiological traits of plateau pikas were repeatable (Table 1). The repeatability of heart rate and RMR showed similar patterns (R_adj = 0.333 [0.099, 0.707], R_adj = 0.333 [0.120, 0.698], respectively), followed by CORT (R_adj = 0.321 [0.025, 0.667]) and exploration (R_adj = 0.303 [0.068, 0.673]), which suggest that all behavioral and physiological traits measured were medium repeatable over time. At the start of the experiment, all traits were not significantly correlated with each other in the two groups. However, at the end of the experiment, exploration and CORT were significantly positively correlated, and RMR was significantly negatively correlated with exploration or CORT in the Noise group (p < 0.001; Figure 1A). However, all traits were not significantly correlated with each other in the Control group (Figure 1B).

Before the experiments, the heart rates of plateau pikas were in the range of 140.70–413.82 beats per min, with an average of 342.22 ± 2.13 beats per min. At the end of the experiment, the average exploration of the Noise and Control groups changed to 383.60 ± 4.24 and 372.93 ± 2.67 beats per min, respectively. Treatment, trial, and their interaction and gender had no significant effects on heart rate (Figure 2A; Table 2).

Before the start of experiments, the exploration ranged from 0.50 to 6.09, with an average of 4.44 ± 0.08. By contrast, at the end of the experiment, the average exploration of Noise and Control groups changed to 42.93 ± 0.76 and 4.13 ± 0.06, respectively. The treatment and its interaction with the trial had significant effects on exploration (Figure 2B; Table 2).

Before the experiments, the RMR of plateau pikas ranged from 100.83 ml O₂/[g·h] to 348.01 ml O₂/[g·h], with the mean of 244.07 ± 2.26 ml O₂/[g·h]. At the end of the experiment, the mean RMR of the Noise and Control groups changed to 205.83 ± 4.40 and 228.74 ± 3.17 ml O₂/[g·h], respectively. Although the trial had no significant effects on RMR, treatment, gender, and the interaction between treatment and trial had significant effects on RMR (Figure 2C; Table 2).

Similarly, before the start of the experiment, the CORT of plateau pikas ranged from 5.612 ng/mg to 11.799 ng/mg, with a mean of 8.32 ± 0.06 ng/mg. At the end of the experiment, the mean CORT of the Noise and Control groups changed to 9.88 ± 0.06 and 7.76 ± 0.09 ng/mg, respectively. The treatment and its interaction with trial had significant effects on CORT (Figure 2D; Table 2).

When wild animals face repeated stimulus, their behavior may be modified, a process called habituation (Rankin et al., 2009). Heart rate and RMR tests involve the capture and handling of subjects, which may elicit significant stress responses (Careau et al., 2008). In this study, although neither behavioral (i.e., exploration) nor physiological traits (heart rate, RMR, and CORT) were singly affected by trial, exploration, RMR, and CORT were interactively affected by trial and traffic noise, suggesting that plateau pikas habituated to the experimental trial, which may weaken the impacts of traffic noise, as shown by the decreases or homogenizations in the exploration, RMR, and CORT of the Control group (Figure 2). In this scenario, we still detected the significant effects of traffic noise on exploration, RMR, and CORT, suggesting the crucial effects of traffic noise on pikas.

Numerous studies have reported the high repeatability estimate for animal behavior over time and/or context environment, which is termed personality (Réale et al., 2010). When animals face external challenges or disturbances, they may reflect behavioral variations, which can be explained through individual consistency and plasticity (Clermont et al., 2019; Couchoux et al., 2021). Using meta-analysis, Bell et al. (2009) observed a mean repeatability of 0.37; our previous study showed that the repeatability of exploration and CORT ranged from 0.34 to 0.63 in the natural population of plateau pikas (Qu et al., 2018). In the present study, the repeatability of exploration, heart rate, RMR, and CORT was around 0.32, indicating that plateau pikas showed lower behavioral and physiological consistency, but higher plasticity under traffic noise stress compared with the natural environment. Plasticity benefits animals to alter their behavior and physiology to better cope with variable environments. For example, urban birds show behavioral plasticity of the acoustic phenotype to increase their survival and reproduction rate during the urbanization process following artificial noise (Slabbekoorn, 2013). A number of insect species show behavioral plasticity in foraging, reproduction, habitat choice, and modified social behavior to cope with artificial rapid environmental changes (van Baaren and Candolin, 2018). This study demonstrated that rodents can change their behavior and physiology to mitigate the potential adverse effects of traffic noise.

Heart rate under handling indicates the individual physiological index of coping style and reflects the response of the animal autonomic nervous system to predation risk or fearful stimuli rather than disturbance stress (Koolhaas et al., 2010). Our previous studies demonstrated that in the natural population, the heart rate and personality were correlated in plateau pikas, where the predation risk is high (Qu et al., 2018; Wei et al., 2020), as pikas are the primary prey for almost all predators on the Qinghai-Tibetan Plateau (Harris et al., 2014). In this study, the heart rate was consistent at the start and end of the experiment, indicating that plateau pikas maintained a stable response to human handling, regardless of whether they were exposed to traffic noise exposure; thus, artificial traffic noise may have different affecting mechanisms for wildlife compared with predation risk, and animals show variable coping styles to predation and artificial challenges (Francis and Barber, 2013).

As an external stressor, noise can hyperactivate the sympathetic autonomic nervous system and activate the hypothalamic–pituitary–adrenal axis, which increase the organism CORT and finally alter metabolism (Schmidt et al., 2013). Two weeks of noise enhanced the blood glucose and CORT levels and varied the glucose metabolism and gut microbiota composition in rats (Cui et al., 2016). Noise exposure also increased the corticosterone level of male treefrog but decreased their immune function (Troïanowski et al., 2017). Our results showed that at the end of the experiment, the CORT of plateau pikas in the Noise group was significantly higher than that at the start, whereas no significant variations in the CORT pre and post experiment were detected in the Control group. Thus, traffic noise caused evident stress to the pikas, and the significant interactions between treatment and trial on CORT suggested that pikas might have accustomed to the experimental environment over 1 month, which relieved the potential stress from an unfamiliar environment. Exploration reflects an animal’s response to space, and it is associated with anxiety; anxious animals consistently show heightened exploration (Karl et al., 2006). In this study, the increased CORT and exploration suggest that persistent traffic noise is an interference pressure resource to pikas, and it affects their hormone sensitivity and anxiety, finally influencing their body conditions.

Metabolism is a proximate mechanism underlying behavioral and physiological variations (Careau et al., 2008). According to the Pace-of-life Syndrome (POLS) hypothesis, animal life history, behavior, and physiology are correlated, and their coupling is used to cope with external challenges (Réale et al., 2010; Boulton et al., 2015; Houslay et al., 2017). In this study, at the start of the experiment in both groups and at the end of experiment in the Control group, no significant correlations were detected among traits, whereas significant correlations between exploration, RMR, and CORT appeared in the Noise group at the end of the experiment, suggesting that traffic noise promoted the correlations among traits, that is, the occurrence of coping style. Although the presence of personality traits has been found in different external stresses, such as predation risk (Dingemanse et al., 2007, 2009) and artificial harvest (Biro and Post, 2008), we emphasized for the first time a significant relationship among behavior, RMR, and CORT induced by traffic noise.

Two distinct conceptual models explain the relationships between animal metabolism and behavior (Careau et al., 2008). The performance model refers to the situation in which a high metabolic rate requires more active behavioral traits. Thus, exploration, aggressiveness and boldness should positively correlate with energy metabolism. A positive correlation was detected between boldness and metabolism in wild-caught orb-weaving spider (Larinioides cornutus; Shearer and Pruitt, 2014) and fall field crickets (Gryllus pennsylvanicus; Careau et al., 2019). By contrast, the allocation model argues that with individuals having to allocate a fixed amount of energy to maintain their normal metabolism, limited energy is expended in the behaviors associated with activity, exploration, etc. The active mosquitofish (Gambusia holbrooki) has a low RMR, which suggests that high activity rates require individuals to allocate less energy toward maintenance (Biro et al., 2020). In this study, the decrease of RMR, as well as negative relationship between RMR and exploration in the Noise group at the end of experiment, provided evidence for the allocation model of energy management. Under ad libitum food conditions in the laboratory, traffic noise caused trade-offs between RMR and exploration. The relationships between behavior and metabolism are not always stable (Mathot et al., 2019). For instance, the risk-taking behavior (i.e., boldness) and basal metabolic rate of the great tit (Parus major) changed from positive to negative in two consecutive years (Mathot et al., 2015). Under artificial light stimulus, the common toad (Bufo bufo) allocates less energy on activities but more energy on RMR (Touzot et al., 2019). Similarly, given that hormone production is a high-energy-consuming process, the negative relationships between CORT and RMR occurred under noise exposure conditions, which is consistent with the result of Guenther et al. (2014), who observed negative correlations between CORT and RMR in juvenile and mature cavies (Cavia aperea).