Neural representation of bat predation risk and evasive flight in moths: A modelling approach

Highlights

• How can prey animals evaluate predator threat based on perceivable sensory cues?

• We combine empirical data and modelling in a multi-species bat-moth-community.

• Across 14 bat species, echolocation call frequency predicts bat predation threat.

• The neural audiograms of 12 moth species exploit this link for adaptive escape.

• Even simple sensory systems can trigger appropriate actions for multiple predators.

Abstract

Most animals are at risk from multiple predators and can vary anti-predator behaviour based on the level of threat posed by each predator. Animals use sensory systems to detect predator cues, but the relationship between the tuning of sensory systems and the sensory cues related to predator threat are not well-studied at the community level. Noctuid moths have ultrasound-sensitive ears to detect the echolocation calls of predatory bats. Here, combining empirical data and mathematical modelling, we show that moth hearing is adapted to provide information about the threat posed by different sympatric bat species. First, we found that multiple characteristics related to the threat posed by bats to moths correlate with bat echolocation call frequency. Second, the frequency tuning of the most sensitive auditory receptor in noctuid moth ears provides information allowing moths to escape detection by all sympatric bats with similar safety margin distances. Third, the least sensitive auditory receptor usually responds to bat echolocation calls at a similar distance across all moth species for a given bat species. If this neuron triggers last-ditch evasive flight, it suggests that there is an ideal reaction distance for each bat species, regardless of moth size. This study shows that even a very simple sensory system can adapt to deliver information suitable for triggering appropriate defensive reactions to each predator in a multiple predator community.

Introduction

Sensory system adaptations for detecting and responding to predator cues are well-known in animals, and the specific neural activity required to generate anti-predator behaviour has been documented for many species (e.g. arctiid moths: Ratcliffe et al., 2009; crickets: Nolen and Hoy 1984; locusts: Fotowat and Gabbiani, 2007; cockroaches: Ritzmann et al., 1980; crayfish: Edwards et al., 1999; fish: Eaton et al., 2001). Most animals co-occur with multiple predator species, each of which provides different cues and poses a different level of threat (Smolka et al., 2011; Falk et al., 2015). Animals often demonstrate anti-predator behaviours that are proportional to the risk posed by the detected predator (e.g. moths: Roeder, 1974, Ratcliffe et al., 2011; crabs: Smolka et al., 2011; frogs: Fraker, 2008; fish: Helfman, 1989; birds: Templeton et al., 2005), but many aspects of predator threat cannot be directly encoded by sensory systems, such as the speed and manoeuvrability of a predator that is stationary at the time it is detected. Behavioural studies suggest that some animals might overcome this limitation by assessing a single detectable trait that is comparable across predators, such as size, to estimate predator threat (Templeton et al., 2005). Few studies, however, have assessed whether sensory systems are specifically tuned to cues that correlate with the predator threat posed by different predators in the community.

Eared moths and echolocating bats are ideal study animals for addressing this question. Echolocating bats are significant nocturnal predators of moths (reviewed in Fullard, 1998). Bats produce ultrasonic echolocation calls for orientation and prey detection, and ultrasound-sensitive ears evolved in many moth families with the primary function of detecting bat echolocation calls (Fullard, 1998; ter Hofstede and Ratcliffe, 2016). With only 1–4 auditory receptor neurons depending on the moth family, these ears are the simplest ears in nature (Yack, 2004). Moths in the family Noctuidae have two auditory receptor cells, called A1 and A2. The A2 cell is approximately 20 dB less sensitive than the A1 cell (Fig. S1). Noctuid moths also have a two-staged anti-bat response; they show directional flight away from quiet ultrasonic pulses, typical of a distant bat, and last-ditch flight manoeuvres in response to loud ultrasonic pulses, typical of a close bat (Roeder, 1962, 1964; Agee, 1969). Directional flight is likely triggered at intensities just above A1 threshold (Roeder, 1964, 1967) and at much greater distances than those at which the bat can detect the moth's echo (Roeder, 1998; Surlykke et al., 1999; Goerlitz et al., 2010), meaning that moths can initiate directional flight to avoid being detected by the bat. Directional flight, however, should no longer be effective once the bat has detected the moth (Corcoran and Conner, 2016) due to differences in flight speeds between bats (4–9 m/s: Table S1) and moths (0.5–6 m/s; Riley et al., 1992; Vickers and Baker, 1997; Luo et al., 2002; Chapman et al., 2008; Corcoran and Conner, 2016). Instead, moths require more drastic erratic flight or other secondary defence strategies (e.g., Ratcliffe and Fullard, 2005, Corcoran and Conner, 2012) to avoid being captured by the attacking bat. Due to the A2 cell's higher threshold, meaning it is only activated by high amplitude sounds, Roeder (1974) hypothesized that A2 cell activity might trigger these last-ditch flight maneuvers. This hypothesis, however, has never been empirically tested in noctuid moths.

The frequency tuning curves of moth A-cells vary depending on the sympatric bat community (reviewed in Fullard, 1998; ter Hofstede et al., 2013), suggesting a strong functional link between frequency tuning of moth ears and the echolocation call frequency of sympatric bats. Although the A1 cell is more sensitive than the A2 cell, both cells have the same shaped tuning curve (Fig. S1), meaning that moths cannot discriminate between different bat species based on frequency (ter Hofstede et al., 2013). Larger moths have lower A1 cell thresholds (i.e. are more sensitive) than smaller moths, presumably to compensate for the greater distances at which bats can detect the louder echo reflected from their larger surface area (Surlykke et al., 1999; ter Hofstede et al., 2013). Despite differences in tuning and sensitivity between moth species, the general shape of the auditory tuning curve is similar across most moth species, with greatest sensitivity to lower ultrasonic frequencies and decreasing sensitivity at higher ultrasonic frequencies (Fig. S1; Fullard, 1998; ter Hofstede et al., 2013; ter Hofstede and Ratcliffe, 2016). The relatively consistent shape of the moth auditory tuning curve across species suggests that it is either constrained by morphological or physiological factors or that it is a functional adaptation, i.e. a matched filter (Wehner, 1987; Römer, 2016; von der Emde and Warrant, 2016).

Here we use both empirical data and mathematical modelling to assess whether the shape of the moth auditory tuning curve is adapted such that moths detect different bat species at times when they pose a similar level of threat. First, we tested for relationships between bat call frequency, which varies across bat species in a community, and four variables that influence the detectability of calls by the moth and echoes by the bat (call level, duration and repetition rate) or the time required for the bat to intercept the moth (bat flight speed). Strong relationships between these variables would allow for the evolution of moth ear frequency tuning that reflects the threat posed by different bat species. Second, we incorporated data on bat echolocation call parameters, bat and moth hearing thresholds and estimated detection distances between bats and moths into multiple models of moth escape behaviour to develop testable hypotheses about the adaptive value of moth ear tuning for evading bat predators at a distance or at close range.