Sound garden: How snakes respond to airborne and groundborne sounds

Abstract

Evidence suggests that snakes can hear, but how snakes naturally respond to sound is still unclear. We conducted 304 controlled experiment trials on 19 snakes across five genera in a sound-proof room (4.9 x 4.9 m) at 27ºC, observing the effects of three sounds on individual snake behavior, compared to controls. We quantified eight snake behaviors (body movement, body freezing, head-flicks, tongue-flicks, hissing, periscoping, head fixation, lower jaw drop) in response to three sounds, which were filtered pink-noise within the following frequency ranges: 0-150Hz (sound 1, which produced ground vibrations, as measured by an accelerometer), 150-300Hz (sound 2, which did not produced ground vibrations), 300-450Hz (sound 3, which did not produced ground vibrations). All snake responses were strongly genus dependent. Only one genus (Aspidites, Woma Pythons) significantly increased their probability of movement in response to sound, but three other genera (Acanthophis (Death Adders), Oxyuranus (Taipans), and Pseudonaja (Brown Snakes)) were more likely to move away from sound, signaling potential avoidance behavior. Taipans significantly increased their likelihood of displaying defensive and cautious behaviors in response to sound, but three of the five genera exhibited significantly different types of behaviors in sound trials compared to the control. Our results highlight potential heritable behavioral responses of snakes to sound, clustered within genera. Our study illustrates the behavioral variability among different snake genera, and across sound frequencies, which contributes to our limited understanding of hearing and behavior in snakes.

Fig 1. Diagram of sound-proof room for controlled trials.

The experimental set-up is shown on the left side, and the data collection area and sound system station are shown on the right side. Image not made to scale. See text for dimensions.

Fig 2. Genus-level behavioral comparisons of snakes in control trials.

Genus groupings of snakes, snake count per species (n = 2–6) and presence of three dominant behaviors (tongue flicks, defensive/cautious behaviors and >20cm movement) (from n = 4 trials per snake) as a proportion of control trials (n = 4 per snake). Defensive/cautious behaviors included cautious exploration, fixation, freezing, hisses, head jerks, jaw drops and periscoping (Table 1). Snake images are not to relative scale. See S1 Table in S1 File for the length of all included snakes. Image credits to CNZ: Acanthophis, Aspidites, Hoplocephalus; to CJH: Oxyuranus, Pseudonaja.

Fig 3. Defensive/cautious behavior responses of 19 snakes to sound compared to controls.

(A) Mean probability of exhibiting defensive or cautious behavior in control trials for three snake genera with sufficient response (Fig 2). Behaviors included cautious exploration, fixation, freezing, hisses, head jerks, jaw drops and periscoping from n = 4 trials per snake. Lines represent 95% credible intervals. (B) Change in probability of exhibiting defensive or cautious behaviors from the control to each of three sound treatments. Values above zero indicate increase in probability, and vice versa. Faded points had credible intervals crossing zero, suggesting no change in probability from control (n = 4 per snake). (C) Non-metric multidimensional scaling of defensive/cautious behavior composition for each genus-sound treatment combination. Points with similar behavior composition are clustered on the plot, and are sized based on the total count of behaviors. Points in the same direction as arrows exhibited more of those behaviors, with arrow length proportional to strength of influence. Differences between control (“C”) and treatment points (“S1”, “S2” and “S3”) suggest the type of exhibited defensive behaviors changed in response to sound treatments. Symbols for the control and S1 are omitted for Acanthophis due to no behaviors being exhibited in those trials. Numbers are omitted from Acanthophis and Hoplocephalus symbols due to space constraints.

Fig 4. Snake movement in response to sound compared to controls.

(A) Histogram of snake movement by genus (colored groupings) and sound treatment (C, S1-S3) (n = 4 per treatment per snake). Histogram bars are grouped in units of 10cm, except above 100cm (right of dashed line) where they are grouped in units of 50cm. Alternating colors on bars represent blocks of up to 20% of trials. Grey shading indicates the cut-off used to distinguish 20cm threshold for binary movement response used in probability model. (B) Mean probability of >20cm snake movement in control (n = 4 per snake). (C) Change in probability of movement from the control to each of three sound treatments. Values above zero (dashed line) indicate increase in probability, and vice versa. (D) Mean probability of movement away from speaker for each genus in control trials, using only snakes with >20cm of movement. (E) Change in probability of movement away from speaker compared to control means for each of three sound treatments, as per (B). Lines in (B-E) represent 95% credible intervals from Bayesian hierarchical models. Points in C and E with 95% credible intervals that do not cross zero are highlighted as significant effects.