Early erratic flight response of the lucerne moth to the quiet echolocation calls of distant bats

Nocturnal insects have evolved ultrasound-sensitive hearing in response to predation pressures from echolocating insectivorous bats. Flying tympanate moths take various evasive actions when they detect bat cries, including turning away, performing a steering/zigzagging flight and ceasing flight. In general, infrequent ultrasonic pulses with low sound intensities that are emitted by distant bats evoke slight turns, whereas frequent and loud ultrasonic pulses of nearby bats evoke erratic or rapid unpredictable changes in the flight path of a moth. Flight cessation, which is a freezing response that causes the moth to passively dive (drop) to the ground, is considered the ultimate last-ditch evasive behaviour against approaching bats where there is a high predation threat. Here, we found that the crambid moth Nomophila nearctica never performed passive dives in response to frequent and loud ultrasonic pulses of >60 dB sound pressure level (SPL) that simulated the attacking echolocation call sequence of the predominant sympatric insectivorous bat Eptesicus fuscus, but rather turned away or flew erratically, regardless of the temporal structure of the stimulus. Consequently, N. nearctica is likely to survive predation by bats by taking early evasive action even when it detects the echolocation calls of sympatric bats hunting other insects at a distance. Since aerially hawking bats can track and catch erratically flying moths after targeting their prey, this early escape strategy may be common among night-flying tympanate insects.


Introduction
Nocturnal insects are exposed to intense predation pressure from insectivorous echolocating bats [1,2], which has led to the evolution of ultrasound-sensitive ears and escape behaviours [3][4][5]. Many moths possess tympanal hearing organs and respond to ultrasonic pulses with diverse evasive actions [6]. In general, ultrasound with a relatively low intensity and infrequent pulse repetition rate causes flying tympanate moths to turn away from the sound source [7,8], whereas loud and frequent ultrasonic pulses evoke erratic responses, including looping,

Sampling and rearing of moths
We caught wild adult N. nearctica with an insect net in the daytime at the campus of the University of Toronto Scarborough (43˚78 0 10@N, 79˚18 0 36@W; Toronto, Ontario, Canada) from June to August 2014 and 2015. Larvae derived from gravid females were reared on an artificial diet (Silk-mate™ 2M; Nosan Corp., Yokohama, Japan) and kept at a constant temperature of 21 ± 1˚C under a 16:8 h light:dark (L:D) cycle until they were used in the experiment. Male and female moths were sexed on the day of emergence (0 days old), and housed in 30 × 30 × 30 cm nylon mesh cages with a water supply. Ethical approval was not required for this study.

Recording and analysis of bat echolocation calls
In early August 2014, we manually recorded the echolocation calls of E. fuscus, which is a dominant bat species at the site where we captured N. nearctica [28]. Recordings were made using an UltraSoundGate 416-200 (Avisoft Bioacoustics, Berlin, Germany) with a CM16/CMPA condenser microphone and Avisoft-RECORDER USGH 4.2.18 software (250-kHz sampling rate; 16-bit; .wav file format). Before making the manual recordings, we monitored bat activity using a portable bat detector (D240X; Pettersson Elektronik, Uppsala, Sweden) in heterodyne mode. We then measured the temporal characteristics of the pulse duration and inter-pulse interval (duration of silence between pulses) in the search, approach and terminal phases of the echolocation call sequences, according to the previously defined predation sequence [32]. The start and end points of a pulse were determined by discriminating sounds of >1 dB from the background noise with an oscillogram and spectrogram using BatSound 4.03 software (Pettersson Elektronik).

Active times of bats and moths
To estimate the foraging times of echolocating E. fuscus, we used a D500X ultrasound detector (Pettersson Elektronik) to automatically record wild bat echolocation calls from 1 h before sunset (around 20:00) to 1 h after sunrise (7:00) from early June to late July 2014. These recordings were made at a sampling rate of 300 kHz (16-bit; .wav file format) for 5 s, including a 1-s pre-trigger buffer. In June-July 2015, we used a digital voice recorder with a voice-activated recording function (WS-822; Olympus Corp., Tokyo, Japan) that was connected via a 3.5-mm earphone jack to a heterodyne bat detector set at 30 kHz (Bat4 Bat Detector; Magenta Electronics Ltd., Tutbury, UK) to obtain 5-s-long recordings. We then analysed the time stamps of the . wav call files.
To quantify the locomotion of N. nearctica in a given day, we introduced 2-to 3-day-old virgin moths (16 males and 18 females) individually into a transparent plastic vial (90 mm high × 44 mm internal diameter), the inside walls of which had been partially sprayed with water, under the same temperature and photocycle conditions as were used during rearing. We then recorded their movement using a locomotion monitor with nine infrared beams per tube (LAM 60; Trikinetics Inc., Waltham, MA, USA) connected to DAMSystem308 software (Trikinetics Inc.) on a laptop PC. To calculate moth nocturnality (%), we first converted the individual count data (the number of passes) obtained per 10 min to binary data (0 or 1), so that any locomotion detected in a 10-min period was treated as '1' regardless of the number of passes. Since a 16L:8D photocycle was used in the experimental room, we doubled the converted values for the scotophase (8-h dark phase) and then calculated the proportion of locomotion detected during this period.

Ultrasound stimulus
To determine which pulse structures and sound levels of hunting bat echolocation calls cause evasive manoeuvres in N. nearctica, we synthesized six types of pulses with 25% rise/fall times based on echolocation calls recorded in the field using Audacity1 2.1.0 software (Audacity Team, USA; 192-kHz sampling rate; 16-bit; .wav file format). These pulses were composed of 30-kHz pure tones for simplicity. The temporal structures of the six pulse types were as follows: 1) search phase pulse with a 10-ms pulse duration (PD) and 111-ms inter-pulse interval (IPI) (8.3 pulse s −1 ); 2) early-approach phase pulse with an 8-ms PD and 88-ms IPI (10.4 pulse s −1 ); 3) middle-approach phase pulse with a 6-ms PD and 66-ms IPI (13.9 pulse s −1 ); 4) lateapproach phase pulse with a 4-ms PD and 40-ms IPI (22.7 pulse s −1 ); 5) early-terminal buzz with a 2-ms PD and 13-ms IPI (66.7 pulse s −1 ); and 6) middle-terminal buzz with a 1-ms PD and 5-ms IPI (166.7 pulse s −1 ). The late-terminal buzz, which has a <0.5-ms PD and 5-ms IPI (>181.8 pulse s −1 ), was not simulated in this study due to the limited sampling rate not allowing accurate simulation of the transient pulse as well as the lack of opportunity for a flying moth to escape an extremely close chasing bat [9,10,12,32-38].
Each ultrasound stimulus was broadcast for 2 s at 40, 50, 60, 70, 80, 90 and 100 dB SPL (re. 20 μPa) from an electrostatic speaker (ES1 speaker connected to an ED1 speaker driver; Tucker-Davis Technologies, Alachua, FL, USA) positioned 20 cm away from a tethered, resting moth (see below). The stimuli sound levels were confirmed with a 1/4-inch microphone (type 4939; grid-off, connected to a type 2670 preamplifier and a type 2690 Nexus™ conditioning amplifier with 0.02-140-kHz band-pass filter; Brüel and Kjaer, Naerum, Denmark) by referring to a known signal voltage from a sound calibrator (type 4231; 94 dB SPL at 1 kHz; Brüel and Kjaer).

Moth avoidance behaviour
We examined the frequency and mode of avoidance manoeuvres exhibited by N. nearctica in response to simulated echolocation pulses of E. fuscus using 2-to 3-day-old virgin moths of both sexes (49 males and 47 females). Individual moths had a fine nylon fishing line tied around their cervix and were suspended from a metal rod attached to the side wall of the experimental room, which stabilized their flight in a restricted space. These tethered moths generally started to fly when a small piece of paper that served as a platform was removed from beneath their legs, with those moths that did not exhibit flight being removed from the experiments. Immediately after stable flight had been confirmed, the ultrasound stimuli were presented to each tethered flying moth in a random order, with different stimuli being broadcast to the same moth at ca. 20-min intervals to avoid sensory behavioural fatigue. Moth behaviour was then recorded with a digital high-definition (HD) camcorder in infrared recording mode (HDR-PJ760V; Sony, Tokyo, Japan) connected to a heterodyne bat detector (Magenta Electronics Ltd.) that was set at 30 kHz to monitor the ultrasound stimuli presented. Recordings were made for at least 3 s from >1 s before the broadcast of the 2-s ultrasound stimulus to the end of the broadcast. All experiments were carried out in the scotophase under a dim red light and behavioural responses observed during the stimulus presentation were analysed blindly.
The incidence of avoidance responses was analysed with likelihood-ratio (LR) tests in cumulative link mixed models (CLMM) using the 'ordinal' package in R 3.2.4. In these models, three conventional flying insect escape response behaviours to bat echolocation [1,2,7] were included as an ordered response variable ('no response' = '0', 'turning-away response' = '1' and 'erratic response' = '2'; see Results for definitions), and moth ID was included as a random effect. The six pulse types, two moth sexes and seven sound levels were incorporated into the statistical models as fixed effects. The deviance between the full model and each model that lacked one of the explanatory variables outlined above was analysed. To test the effect of pulse type on the behavioural response, multiple comparisons were made for all combinations of pulse type and each P-value was then adjusted after controlling for the false discovery rate [39]. The effects of pulse type, sex and sound level on the overall pooled evasive actions and each individual avoidance response category were analysed with LR tests in generalized linear mixed models (GLMM) with a binomial error distribution using the 'lme4' package. The highest sound level that caused evasive responses in the greatest proportion of moths (100 dB SPL; see Results) was then compared with the other sound levels and the P-value was adjusted as above.

Nocturnal activities of bats and moths
To eliminate the possibility that N. nearctica is temporally isolated from E. fuscus, we examined the overlap in active times between the bat predator and moth prey. We obtained 681 audio files of echolocation calls of E. fuscus in the field in 2014 and 890 in 2015. We pooled the time stamps of these 1571 files and generated a histogram of the number of files in 10-min time bins (Fig 2A; S1 Data). Most of the files contained echolocation calls were recorded between 20 min after sunset (10 min before dusk) and 210 min after sunset (180 min after dusk, 330 min before sunrise or 300 min before dawn) at the study site in June and July, with peak activity (42.6% of the call files) occurring 50-80 min after sunset (20-50 min after dusk).

Evasive manoeuvres
We tested that ultrasonic pulses of >60 dB SPL induced frequent evasive actions regardless of the attacking stage in the echolocation sequence. We categorised the behavioural reactions of flying N. nearctica to bat call simulations into three types: no response, turning-away response, which was defined as directional flights with courses away from the sound source, and erratic response, which was defined as flights with unpredictable trajectories, including looping-flight, zigzagging-flight and spiral-flight (S1 Video) [7,9,32,33,38]. Flight cessation (passive dive) was never observed in this species, which contrasts with noctuid, erebid, pyralid and geometrid moths [7,9,10,38,[40][41][42][43][44]. No significant differences were observed in the mode and incidence of evasive actions among the six pulse types that were broadcast at 40-100 dB SPL (LR test in CLMM; LR statistic < 8.00, adjusted P > 0.070) (Fig 3; S1 Data). There were also no differences between the sexes in the evasive manoeuvres performed (LR statistic = 0.15, P = 0.70). However, sound levels did significantly affect the evasive manoeuvres (LR statistic = 54.90, P < 0.0001).

Discussion
In the present study, we found that flying N. nearctica exhibited both turning-away and erratic flight in response to fairly quiet (60 dB SPL) bat-like ultrasound stimuli, regardless of the temporal structures of the echolocation pulses. This sound level corresponds to echolocation calls emitted by E. fuscus at more than 20 m away [28,45,46], from which distance the observed evasive manoeuvres could allow the moths to escape to a safety zone before the bat approaches, making them crucial for avoiding bat predation. The auditory cells of N. nearctica can respond to search phase pulses of E. fuscus at >48 dB SPL, approach phase pulses at >51 dB SPL and the terminal buzz at >53 dB SPL [28], implying that any observed evasive actions in response to stimuli of up to 40 dB SPL were simply response-like behaviours. Although we focused on the defensive behaviour of a single moth prey (N. nearctica) against a single bat predator (E. fuscus) in one location, both neural [28] and behavioural data (this study) support our observation that moths exhibit early escape strategies in response to >60 dB SPL ultrasound stimuli. The response-inducing sound levels and pulse structures indicate that N. nearctica escapes bat predators both while other flying insects are being chased (and subsequently captured) by the bats and while the bats are approaching and representing a real predation threat [7,33]. The emission of warning or jamming signals by other moths species, such as tiger moths, helps them to avoid being captured after their accurate localisation by a bat [10,14]. However, sound generation has not been found in N. nearctica.
Diving and dropping to the ground is likely to be the ultimate last-ditch response of moths to a predation threat, being performed shortly before they would be captured by the bat [33]. The lucerne moth takes early escape actions to >60 dB SPL ultrasounds regardless of pulse structures PLOS ONE | https://doi.org/10.1371/journal.pone.0202679 August 20, 2018 However, we did not observe passive diving in N. nearctica and flight cessation has never been observed in this moth or hedylid moths (nocturnal butterflies) [47]. Similarly, tethered Asian corn borer moths, Ostrinia furnacalis (which are in the same family as N. nearctica), do not exhibit a substantial number of passive dives but evasive action is evoked by 62-dB ultrasonic pulses in ca. 80% of the individuals [48]. Therefore, since O. furnacalis and N. nearctica have similar auditory thresholds and frequency-tuning curves [28,49], it seems likely that early avoidance behaviours are a common counter-measure in crambid moths against predacious bats. By contrast, noctuid (two auditory cells in individual ear) [7,9], erebid (two cells) [38], notodontid (one cell) [50], geometrid (four cells) [7] and pyralid (four cells) [42,44] moths show flight cessation behaviour upon the detection of approach-phase echolocation pulses of bats. Therefore, it appears that flight cessation does not depend on the number of auditory cells, since crambid moths have four cells in each ear, like pyralid and geometrid moths [4,6,11].
Intense ultrasound evokes flight cessation and erratic behavioural responses in noctuid moths [9]. However, in N. nearctica, relatively quiet ultrasonic pulses (60 dB SPL) induced turning-away and erratic flight, whereas even intense ultrasound stimuli did not evoke flight cessation (Figs 3 and 4). It is risky to cease flight in proximity to an approaching bat because bats can track and often catch diving moths [10]. The behavioural data obtained in this study suggest that one of the optimal evasion tactic for flying N. nearctica is to fly away early to avoid being locked onto and chased by a bat. Although Krauel and the colleagues [31] state that N. nearctica can fly high for the autumn migration, we frequently observed this moth flying close to the ground or grasses in summer, suggesting that the observed erratic response allows lowflying moths to escape from bats [20,21,51].
Nomophila nearctica is abundant in North America, where it is commonly seen and easily collected in summer [29], and the active time of N. nearctica overlaps with that of E. fuscus (Fig  2). However, DNA fragments of N. nearctica have not been detected from the faeces of E. fuscus and M. lucifugus [52,53], suggesting that wild bats rarely consume N. nearctica in Canada. Thus, since noctuid moth species that are captured by bats do not show erratic flight in response to relatively quiet ultrasound stimuli, it appears that N. nearctica can successfully escape insectivorous bats through a combination of early turning away from cruising bats and erratic flights against hunting bats at a distance.

S1 Data. All raw data for Figs 1-4.
(XLSX) S1 Video. Examples of the three behavioural responses (no response, turning away and erratic flight) of Nomophila nearctica to an early-approach phase pulse broadcast at 90 dB sound pressure level (SPL).