Water shrews detect movement, shape, and smell to find prey underwater

Abstract

American water shrews (Sorex palustris) are aggressive predators that feed on a variety of terrestrial and aquatic prey. They often forage at night, diving into streams and ponds in search of food. We investigated how shrews locate submerged prey using high-speed videography, infrared lighting, and stimuli designed to mimic prey. Shrews attacked brief water movements, indicating motion is an important cue used to detect active or escaping prey. They also bit, retrieved, and attempted to eat model fish made of silicone in preference to other silicone objects showing that tactile cues are important in the absence of movement. In addition, water shrews preferentially sniffed model prey fish and crickets underwater by exhaling and reinhaling air through the nostrils, suggesting olfaction plays an important role in aquatic foraging. The possibility of echolocation, sonar, or electroreception was investigated by testing for ultrasonic and audible calls above and below water and by presenting electric fields to foraging shrews. We found no evidence for these abilities. We conclude that water shrews detect motion, shape, and smell to find prey underwater. The short latency of attacks to water movements suggests shrews may use a flush-pursuit strategy to capture some prey.

Fig. 1.

Water shrews in various stages of pursing and capturing prey. Shrews attacked fish and crayfish and often lunged with mouth open during pursuits. (Copyright for this figure is held by K.C.C.)

Fig. 2.

Evidence for water shrew sensory abilities. (A) Latency to capturing fish (time from entering water to grasping fish) was not different for three shrews under lighted or infrared conditions. Bars are SEM. Shrews can thus efficiently capture prey without eyesight. (B) Water shrew face under the SEM, illustrating the prominent sensory vibrissae. The eyes are small and in this case not visible behind the vibrissae.

Fig. 3.

Responses to brief water movements. (A) A frame from the high-speed video showing a shrew in midattack to a 75-ms water pulse (arrow) emanating from the outlet. (B) Schematic of the chamber used to test water shrews. Responses to water movements were scored for the four quadrants as indicated (0° corresponds to the active outlet). (C) Histogram showing the average number of attacks (open-mouthed lunges) in each quadrant for 20 trials for each of four shrews. Bars are standard error of the mean. See SI Movie 2 for behavior.

Fig. 4.

Paradigm for investigating responses to model fish. (A) Three silicone rectangles and three silicone cylinders were placed in the chamber (items 1–6) along with a silicone model fish (item F). (B) A single frame from the video showing a water shrew biting the model silicone fish (which was retrieved). (C) Histogram showing the average number of times the model fish was bitten (see SI Movie 3) compared with the other six items for the first four trials of the four shrews tested (after which three of four shrews stopped responding). (D) Histogram showing the average number of responses (responses included retrieving, biting, or lunging at the fish with open mouth) to the moving model fish for the first four trials of the three shrews that responded (see SI Movie 5). Bars are SEM. Contrast was enhanced in B.

Fig. 5.

Underwater sniffing of four different objects. (A) Shrews attended particularly to the model cricket, which was extensively explored compared with other objects and repeatedly sniffed by each shrew as illustrated (see SI Movie 6). (B) The average number of sniffs made to different objects. Shrews occasionally sniffed the silicone cylinders and rectangles, usually the model fish, and the model cricket multiple times. Bars are SEM. Contrast was enhanced in A.

Fig. 6.

Latency to attack. To provide a detailed measure of reaction time, behavior was videotaped at 1 ms per frame as a shrew responded to a water movement (selected frames shown above). The water pulse came from an opening (arrow) covered with a flexible rubber tab, so stimulus onset could be visualized. The stimulus began at plate 2 (time 0), as indicated by the smallest initial movement of the rubber cover, and reached a maximum at 17 ms (plate 4). The first movement of the shrew toward the stimulus occurred by 20 ms (plate 5), and by 50 ms (plate 8), the shrew's open mouth was at the point of stimulation caused by the water (which was deflected leftward by the rubber tab).