Journal of Ethology

, Volume 28, Issue 3, pp 437–445

Predatory behaviour of an araneophagic assassin bug


    • Department of Brain, Behaviour and EvolutionMacquarie University
  • Phillip W. Taylor
    • Department of Brain, Behaviour and EvolutionMacquarie University

DOI: 10.1007/s10164-009-0202-8

Cite this article as:
Wignall, A.E. & Taylor, P.W. J Ethol (2010) 28: 437. doi:10.1007/s10164-009-0202-8


Assassin bugs from the genus Stenolemus (Heteroptera, Reduviidae) are predators of web-building spiders. However, despite their fascinating lifestyle, little is known about how these insects hunt and catch their dangerous prey. Here we characterise in detail the behaviour adopted by Stenolemus bituberus (Stål) during encounters with web-building spiders, this being an important step toward understanding this species’ predatory strategy. These bugs employed two distinct predatory tactics, “stalking” and “luring”. When stalking their prey, bugs slowly approached the prey spider until within striking range, severing and stretching threads of silk that were in the way. When luring their prey, bugs attracted the resident spider by plucking and stretching the silk with their legs, generating vibrations in the web. Spiders approached the luring bug and were attacked when within range. The luring tactic of S. bituberus appears to exploit the tendency of spiders to approach the source of vibrations in the web, such as might be generated by struggling prey.


PredationReduviidaeEmesinaeStenolemusLuringStalkingSensory exploitation


Spiders are best known as highly proficient and well-armed predators of insects and other terrestrial arthropods (Wise 1993). Curiously, despite the risk of injury or death, some arthropods routinely pursue spiders as prey. These predators face the challenge of bringing themselves to a position where they can launch an effective strike without being attacked by their intended prey. Whereas some araneophagic species launch comparatively crude attacks by leaping into the web directly at the resident (Jackson 1989), others rely on much more intricate routines based on stealth or deceit.

Some araneophagic arthropods stalk their spider prey, slowly bringing themselves into close proximity before striking. For example, Portia africana, P. fimbriata and P. labiata are araneophagic jumping spiders that often hunt web-building spiders by slowly stalking across the web (Jackson 1990; Harland and Jackson 2006). Some jumping spiders hunt other jumping spiders, which are a difficult and dangerous prey to catch because of their excellent visual acuity (Land 1969). Portia fimbriata uses prey-specific “cryptic stalking” that helps them to avoid detection by jumping spiders they are pursuing (Harland and Jackson 2000).

Instead of stalking, some araneophagic arthropods actively manipulate the behaviour of their prey, seemingly to reduce risk. For example, the parasitic wasp Pison morosum flies into the webs of Achaearanea veruculata, causing the spider to drop out of the web after which the wasp captures it (Laing 1988). Similar behaviour has been observed in another wasp, Sceliphron caementarium, which by flying into their webs, induces spiders from the genus Argiope to drop to the ground before capturing them (Blackledge and Pickett 2000). These predatory strategies seem to exploit the spiders’ anti-predator responses, forcing them out of their web to where they are more easily captured and less able to mount counter-attacks.

Some araneophagic arthropods use more sophisticated deceit to lure spiders within range. For example, Chalybion californicum and C. caeruleum are wasps that entangle themselves in spider webs or pluck the silk threads, luring spiders out of their retreats then attacking them (Howard 1922, cited in Coville 1976; Blackledge and Pickett 2000). There are also reports of spiders generating vibrations in other spiders’ webs that lure the resident spider in close enough to be struck (e.g., Pholcus phalangioides, jumping spiders and pirate spiders) (Jackson and Whitehouse 1986; Jackson and Brassington 1987; Jackson and Wilcox 1993; Tarsitano et al. 2000). In some instances, luring of spiders by araneophagic predators may be a straightforward form of “aggressive mimicry”, whereby a predator mimics another organism that is beneficial to its prey (Wickler 1968). Well-known examples of aggressive mimicry in other taxa include anglerfish that attract prey with a protuberance above the mouth that resembles a food item (Pietsch and Grobecker 1978), bolas spiders that attract moths using odours that closely resemble their prey’s sex pheromones (Eberhard 1977; Haynes et al. 2001), and Photuris fireflies that attract male Photinus fireflies using flash sequences that mimic the mating signals of their prey (Lloyd 1975). Similarly, vibrations generated by araneophagic predators to lure spiders may aggressively mimic the vibrations normally made by the spider’s prey (Tarsitano et al. 2000).

The Emesinae are often observed in association with spider webs (Wygodzinsky 1966) and some assassin bugs from the genus Stenolemus have been reported to prey on web-building spiders. The few species studied to date appear to have few predatory strategies and narrow prey ranges. For example, Hodge (1984) described S. lanipes bobbing in spider webs, eliciting searching behaviour in Achaearanea tepidariorum spiderlings that were captured when they came near. Other descriptions of predatory strategies of Stenolemus assassin bugs are limited to a brief account of stalking behaviour used by S. edwardsii when hunting spiderlings of the common house spider, Badumna insignis (formerly Ixeuticus robustus) (Hickman 1969). However, we have recently reported S. bituberus (Stål) hunting spiders from five different genera in nature, employing two apparently distinct predatory tactics, “stalking” and “luring” (Wignall and Taylor 2008, 2009). This species is araneophagic, with only spiders and spider egg sacs observed being preyed on in both field and laboratory studies (Wignall and Taylor 2008). While the pursuit tendency, outcome, and predatory choices of S. bituberus have been reported elsewhere (Wignall and Taylor 2009) the behaviours used while hunting have yet to be described. We here present the first detailed description of the predatory behaviour and alternative hunting tactics of an araneophagic assassin bug, S. bituberus (hereafter “bug”).

Materials and methods

We observed the predatory behaviour of bugs hunting five web-building spider species from their natural prey range (Wignall and Taylor 2008): Achaearanea extridium (Keyserling, Theridiidae, n = 50), Achaearanea sp. (Strand, Theridiidae, n = 52), Badumna longinqua (L. Koch, Desidae, n = 54), Pholcus phalangioides (Fuesslin, Pholcidae, n = 50), and an unidentified species of Uloboridae (Thorell, n = 53). Spiders and bugs were collected between October 2005 and February 2007 from trees on the Macquarie University campus (Sydney, Australia). After use in experiments, survivors were returned to the tree that they had been collected from. Juvenile bugs cannot fly and our field observations have suggested that they are unlikely to move far, if at all, from the tree where they were collected. As we visited each location only once every 3–4 months, individual bugs were probably tested only once within an instar, although they may have been re-tested at later instars. After collection, bugs were maintained outdoors in 10 ml plastic vials, with a gauze lid for airflow and paper as substrate.

Spiders were collected 2–5 days before testing and were placed in wooden frames (200 × 200 × 30 mm) resembling those of Jackson et al. (2002) to build webs. Frames had a clear acrylic screen at the front (removed during tests) and a wooden screen at the back. All observations were conducted outdoors under shade to maintain natural light, temperature, and humidity. Interactions were staged in the morning (6–10 a.m.) or afternoon (5–7 p.m.) (Australia Eastern Summer Standard Time, AESST), times when assassin bugs have been observed hunting in the field. To stage an interaction, we removed the paper with the bug on it from the maintenance vial and placed it on the base of a frame containing a spider in its web. Behaviours observed were recorded in a notebook and timed with a stopwatch. Observations began once the bug placed a tarsus on a frame and continued until the assassin bug captured the spider, the bug abandoned the hunt (no activity for 90 min or leaving the frame), the spider caught the bug or the spider left the frame.

The terminology used to describe behaviours (e.g., phase, amplitude) follow conventions of Jackson and Hallas (1986), as does our use of the terms “usually”, “occasionally” and “rarely” to define behaviours that occur >80%, between 20 and 80%, and <20% of the time. Measurements and angles provided in the descriptions of behaviour are based on fourth instar bugs.


Elements of general behaviour

Rest position

When at rest, bugs held a characteristic posture, with the body parallel to and <1 mm from the substrate (Fig. 1). The mid and hind-legs were positioned with the tarsi approximately equidistant from the body and flexed ≤60° at the femur–tibia joint. The forelegs were held anterior to the head with the tibiae flexed so that they were held against the femora and perpendicular to the coxae. The antennae were flexed with segment 1 held against segment 2, segment 3 at 140°–160° angle to segment 2 and segment 4 parallel to segments 1 and 2.
Fig. 1

Assassin bug in the rest position


Walking by bugs in or on the periphery of spider webs was generally slow (<5 mm/s) and characterised by frequent pauses lasting up to 5 min accompanied by antennal waving, antennal tapping and grooming (see below). Body movements tended to be slow and deliberate and bugs often tapped the substrate up to 5 times with the tarsus before placing the leg and moving forward. Taps were slow with irregular intervals of up to 3 s. Each step was accompanied by irregular up-and-down bouncing of the body, generated by flexing and extending the mid and hind-legs. Faster locomotion was accompanied by faster and often higher amplitude bouncing. Bouncing during locomotion was variable in amplitude (1–4 mm) and speed (0.5–4/s).

When outside webs, bugs tended to walk much faster than when inside webs (10–30 mm/s). Pauses rarely lasted more than 5 s. Each tarsus was usually placed in contact with the substrate with few (<3) or no preceding taps. As when inside webs, bugs usually bounced while walking outside webs.

Groom antennae

Bugs groomed their antennae by rubbing the fore-tibiae together in an up-and-down motion, then placing an antenna between the fore-tibiae and slowly running the tibiae down its length. Bugs usually began at the proximal end of segment 4 and ran the tibiae down to the distal end. Occasionally, while running the tibiae down the antenna, the bug would either start again before reaching the distal end or repeat rubbing the fore-tibiae together before starting again at segment 2.

Elements of predatory behaviour (in webs)

Antennal wave

When antennal waving, bugs moved the antennae up and down either in phase with each other or very slightly out of phase (Fig. 2). Antennal segments 1 were oriented upward, with the distal segments stretched anterior to the body and at circa 90° angle to segment 1. Antennal segments 2–4 were waved up and down 10°–90° (equivalent to a 1–10 mm amplitude movement of antennae segment 4) from the segment 1-segment 2 joint. Antennal waving was often irregular, with each complete wave taking 1–4 s to complete.
Fig. 2

Assassin bug in the middle of an antennal wave

While antennal waving, the bug often slowly turned to one side. The body was usually oriented at an angle to the substrate (30°–60°) so that the head was much higher than the distal end of the abdomen. Antennal waving was usually performed before commencing or in the early stages of hunting, although it was also occasionally also observed during later stages. Bugs tended to perform antennal waving more often when hunting uloborids than other spider species.

Antennal tap

To antennal tap, the distal ends of antennal segments 4 were brought into contact or near contact (<1 mm) with a substrate or prey. The antennae moved either in phase or slightly out of phase. When antennal tapping, movement of antennae segments 4 was of smaller amplitude (0.1–5 mm) than when antennal waving. While antennal tapping, the body was usually motionless and parallel to the substrate. Antennal tapping of the substrate was particularly common when hunting uloborids.

Leg probe

Leg probing was usually performed by the forelegs, but was also occasionally performed by the mid or hind-legs when the bug appeared to be searching for a foothold in a web. When performed by the forelegs, the tibia–femur joints flexed up and down by as much as 90° while the coxae–trochanter and trochanter–femur joints also flexed, producing a circular 5–10 mm motion of the tarsi.

Leg probing by the mid and hind-legs also produced a circular motion of the tarsi. Usually only one mid or hind-leg probed at a time and this was also often associated with locomotion in spider webs. Because of the greater length of the appendages, a larger amplitude movement (up to 15 mm) was observed than in the forelegs. Most movement was generated from the femur–tibia joint, although some flexion was also observed in the coxa–trochanter and trochanter–femur joints.

Severing with forelegs

Bugs were observed severing the silk threads of spider webs with their forelegs. The assassin bug first grasped the silk with its tarsi (and pretarsi) and then made low amplitude (<5 mm) up and down movements, with most of the movement being generated at the tibia–tarsus joint. Often the tarsi moved more than once, usually in alternating phase, before the silk was severed.

Severing with proboscis

Occasionally a bug severed silk threads using its proboscis. This was seen especially when a bug was cutting itself free of silk thrown on it by a resident spider. To sever silk threads using its proboscis, the bug extended its proboscis above the silk by simultaneously flexing each segment circa 45°. Then segment 3 was brought down with flexion at each segment of the proboscis contributing to the angle of movement. Often, more than one movement of the proboscis was required to sever the silk thread. Whilst severing with the proboscis, the tarsus of one or both forelegs held the silk thread taunt.

Single pluck

To single pluck, the bug used one fore-tarsus to pull and then release the silk. The other fore-tarsus may or may not be in contact with the silk. High amplitude plucks (>3 mm) were generated mainly by flexing the femur–tibia joint, whereas low amplitude plucks (<3 mm) were generated mainly by flexing the tibia–tarsus joint.

Bugs hunting A. extridium and Achaearanea sp. usually plucked at an amplitude of 1–3 mm and 1–2 plucks/s, although the amplitude and speed usually increased (amplitude 3–4 mm, speed 2–3 plucks/s) when the wind was blowing. Bugs hunting B. longinqua, P. phalangioides and the uloborid usually plucked at a higher amplitude of up to 5 mm, with up to 4 plucks/s. The amplitude and speed of plucks when hunting B. longinqua, P. phalangioides and the uloborids also usually increased when the wind was blowing (amplitude 4–5 mm, speed 3–4 plucks/s).

Double pluck

When double plucking, the way the bug moved its forelegs was similar to when single plucking except that both fore-tarsi, rather than only one fore-tarsus, were placed on the silk thread. When double plucking, leg movement was usually generated by flexing the tibia–tarsus joint and was usually slightly out of phase.

Bugs that double plucked when hunting A. extridium and Achaearanea sp. used low-amplitude (0.1–1 mm) movements of the tarsus and generated up to 2 plucks/s. The highest rates of double plucking were observed in the presence of wind. Bugs hunting other spider species double plucked using 0.5–1 mm movements and circa 3 plucks/s, with faster double plucks observed in the presence of wind for all spider species.

Rotary pluck

When rotary plucking, bugs plucked the silk thread with large amplitude movements of the femur–tibia joints of the forelegs (up to 90° flexion of the joint) with some flexion also at the coxa–trochanter and trochanter–femur joints (up to 45°). Often, the forelegs moved slightly out of phase with each other. A bout of rotary plucking most often comprised 2 plucks, one with each foreleg.


To stretch, the bug used one fore-tarsus to hold the silk thread, whilst the other fore-tarsus was placed very close to it (<1 mm) on the silk (Fig. 3). The fore-tarsus then very slowly (up to 0.3 mm/s) stretched the thread up to 10 mm either laterally to the other tarsus (which held onto the silk) or orthogonally. Bugs appeared to stretch silk threads from B. longinqua webs slower (circa 0.1 mm/s) than threads from any of the other spider webs.
Fig. 3

Assassin bug stretching a thread of silk. The right foreleg is holding the thread stationary while the alternate foreleg is stretching the silk orthogonally

Most of the movement for the stretch was generated from the femur–tibia, coxa–trochanter, and trochanter–femur joints. The tarsus holding the silk thread also occasionally stretched the silk a little (up to 5 mm). Occasionally, the foreleg performing the majority of the silk stretching released the thread and then stretched again from the starting position. The silk thread being stretched occasionally broke.

Pre-strike behaviour

Pre-strike behaviour was usually performed once the bug was positioned with its head and pronotum above the spider’s body (Fig. 4). The bug slowly extended its forelegs at the femur–tibia joint to circa 90° angle and extended its proboscis in a manner similar to when severing silk threads. The bug often antennal tapped the spider during this stage.
Fig. 4

Assassin bug in the “pre-strike” posture


To strike, the bug rapidly (<0.1 s) brought the anterior of its body downward on to the dorsal surface of the spider by flexing the femur–tibia and the trochanter–femur joint of the mid and hind-legs. The proboscis pierced the body of the spider and the forelegs flexed at the femur–tibia joint to grasp the prey. Occasionally, the mid and hind-legs flexed to propel the body of the bug slightly forward to bring it into position above the cephalothorax of the spider if it was not already positioned there. While striking, assassin bugs quickly (<0.1 s) moved their antennae up and behind their head (antenna segment 1 moved back making 150° angle from the top of the head and antenna segment 2 moved back parallel to segment 1). Strikes were performed head-on, side-on or from behind the spider. Prey usually became inactive 5–10 s after being struck by the bug and all strikes observed (n = 15) were successful.

Post-strike behaviour

Bugs usually began feeding immediately after striking. Feeding lasted longer than 1 h and the forelegs were used to hold the spider at all times. Usually, the bug pierced the spider’s body with its proboscis at several sites during feeding. There was no obvious pattern to where they pierced the spider’s body. Whilst feeding, bugs usually bounced slowly up and down (circa 1 cycle/s) with irregular, low amplitude (1–2 mm) movements of the body.

Sequences of predatory behaviour

Three main phases were discernible in predatory sequences. Phase 1 begins with the bug antennal waving and walking along the frame to the edge of the spider web, which was taken as the start of predatory behaviour (Fig. 5). Often, during Phase 1 the bug paused to groom for periods lasting up to 7 min. Once the bug reached the edge of the spider web, walking was often much slower with frequent pauses to groom, antennal wave and antennal tap the silk threads. The beginning of Phase 2 was characterised by silk manipulation behaviours (Fig. 5). During this phase, bugs hunting P. phalangioides usually began a bout of rotary plucking, behaviour that occasionally elicited whirling from the resident spider (an anti-predator behaviour consisting of a circular movement of the body, Jackson et al. 1990). Bugs usually increased their rate of approach during bouts of whirling. Bugs hunting Achaeranea sp., A. tepidariorum, and the uloborid began single or double plucking a silk thread when at the edge of the web, while bugs hunting B. longinqua most often began by stretching the silk thread.
Fig. 5

Idealised sequences of typical hunts. Dashed lines indicate rare sequences. Stalking sequences are represented by assassin bug approaching the spider prior to antennal tapping, luring sequences by the spider approaching the assassin bug. Probability of entering each phase of predatory behaviour from the previous phase is written next to the phase headings

Once bugs entered Phase 2, there were discernable differences related to the spider species. Bugs hunting B. longinqua usually used stretching and severing with forelegs more often than any of the plucking behaviours, and tended to stay within 3 cm of the web periphery. However, bugs hunting P. phalangioides often very quickly moved away from the edge of the web toward the resident spider, using rotary plucking, stretching, and severing with forelegs to move silk threads from their path. Bugs also often walked into the web when hunting uloborids, often antennal waving and occasionally plucking, stretching, or severing silk threads.

Once in Phase 2, bugs hunting Achaearanea sp. and A. tepidariorum often used all of the types of plucking, stretching, and severing behaviours, aside from severing with the proboscis which was rarely observed. Bugs often single, double, and rotary plucked the silk threads in bouts lasting up to 20 min before stretching and severing with their forelegs to clear silk threads from their path. Bugs would often alternate bouts of plucking, stretching and severing with slowly walking forward into the web (usually no more than 1 cm at a time), antennal tapping and antennal waving while walking, before beginning another bout of plucking, stretching, or severing.

Two distinct predatory tactics were observed at the beginning of Phase 3, “stalking” and “luring” (Fig. 5). Stalking bugs approached the spider very slowly (usually <1 mm/s) until within striking range, with frequent pauses of up to 20 min to sever silk threads and occasionally to stretch silk. Spiders rarely responded to stalking bugs, apparently failing to detect these enemies in their web. Bugs stalked spiders along the frame and into the web. Whereas stalking bugs approached the spider, luring bugs were approached by the spider. Luring bugs stretched, single plucked, double plucked, and rotary plucked the threads in bouts lasting up to 20 min. These silk manipulations produced vibrations in the web to which the spiders usually responded by first orienting toward the bug, and then pausing for up to 10 min. During pauses, the bug usually continued manipulating the silk. The spider then usually approached the bug. Several approach and pause sequences were commonly observed before the spider reached the luring bug. Usually the spider ceased approaching once within 5 mm of the bug’s head. Only rarely (n = 9) were spiders observed to attack bugs. All attacks were observed when hunts were in Phase 2.

Regardless of whether the luring or the stalking tactic was adopted, the bug usually antennal tapped the spider in the pre-strike behaviour, usually beginning close to a leg then slowly moving closer toward the body. Bouts of antennal tapping lasted from 5 s to 3 min during which spiders rarely responded. In all observations, bugs that antennal tapped the spider succeeded in capturing the spider.


Stenolemus bituberus bugs use two distinct predatory tactics when hunting web-building spiders, stalking and luring. Systematic observations from this study confirm our previous opportunistic observations of predatory behaviour in the field (Wignall and Taylor 2008). Although both predatory tactics include manipulation of the silk threads by the bugs, the progress of the hunt is very different. When stalking, bugs sever and stretch silk threads between themselves and the spider, most often appearing to do so without alerting the prey spider. When luring, silk manipulations produce vibrations to which the spider responded by approaching the bug.

Bugs have characteristic locomotion, with irregular up-and-down bouncing whilst walking and substrate tapping with the tarsus prior to placing it down. Similar “choppy” locomotion has been observed in the web-invading jumping spider P. fimbriata (Jackson and Hallas 1986). Because they lack the regular patterns that characterise locomotion in most animals, these unusual stepping behaviours in araneophagic web-invaders may diminish the likelihood that spiders will identify resulting vibrations as coming from an invader. By generating additional cyclical vibrations in the web, bouncing may further break up the pattern of vibrations generated by stepping, or generate noise (a “smokescreen”) against which the vibrations of stepping are less likely to be detected.

Increases in the speed of stretching and severing of silk threads in the presence of wind by stalking bugs suggest that this bug exploits these opportunities to approach undetected while the spider’s sensory systems are overwhelmed by environmental disturbance. Araneophagic jumping spiders have been reported to exploit wind and other disturbances to mask their predatory approaches toward web residents (Wilcox et al. 1996; Cerveira et al. 2003). We also observed that bugs approached spiders more when P. phalangioides was whirling. Perhaps bugs exploit the pholcids’ own anti-predator response, and environmental disturbances, to mask their approach.

When stalking, bugs seem to rely on stealth, as indicated by their seemingly cryptic style of locomotion and exploitation of environmental disturbance to approach. On the other hand, luring bugs appear to deliberately generate vibrations that reveal their presence and location to the prey spider. Interestingly, just as stalking bugs appear to adjust their behaviour to advance while remaining cryptic during periods of environmental disturbance, there is evidence that luring bugs adjust their behaviour to remain detectable. Specifically, luring bugs increased the speed and amplitude of single plucking, double plucking and rotary plucking during episodes of wind disturbance and this may serve to increase the spider’s ability to detect the resulting vibrations over the background noise. Changes in signal structure to maintain efficacy in the presence of environmental noise have been reported for both visual (Ord et al. 2007; Peters et al. 2007) and acoustic signals (Lengagne et al. 1999; Brumm et al. 2004), and our observations suggest that bugs may show similar flexibility in their use of vibratory signals.

One of the most interesting behaviours that we observed in hunting bugs is the apparently counter-intuitive antennal tapping of prey during pre-strike behaviour. Similar tapping of prey has been observed in spitting spiders (Li et al. 1999), jumping spiders (Jackson and Willey 1994) and digger wasps (Anton and Gnatzy 1998). While there have been no studies directly investigating the function of prey tapping, the presence of similar behaviours across such taxonomically diverse predators suggest that this behaviour serves a common function. By tapping their prey, predators may gather tactile and contact chemical information about the identity or orientation of the prey before striking. Alternatively, tapping prey may habituate prey to physical contact and thereby reduce their responsiveness or aggressiveness toward later stimuli. Bugs must be very close to their prey to strike and risk alerting the spider to their presence. If alerted, spiders might run away or, worse, attack. By habituating the spider to external stimuli, the bug may reduce the likelihood that the spider will attack during the final approach and attack.

Spiders responded to single, double and rotary plucking and stretching of the silk by luring bugs in a manner that was extremely similar to how they respond to prey struggling in the web. These observations raise the question of whether S. bituberus aggressively mimic the specific cues produced by insect prey and/or mates in spider webs, as has been suggested for other web-invading araneophagic arthropods (Jackson and Whitehouse 1986; Blackledge and Pickett 2000; Tarsitano et al. 2000) or instead exploit a more general tendency in spiders to approach the source of vibrations in their webs (Jackson and Pollard 1997).


We thank Chris Evans and Robert Jackson for helpful comments throughout the study. We also thank Marie Herberstein and Aaron Harmer for comments on the manuscript. This study was supported by a grant from the Australian Research Council. AEW was supported by a RAACE scholarship from Macquarie University.

Supplementary material

View video

Supplementary material 1 (MOV 5529 kb)

View video

Supplementary material 2 (MOV 5704 kb)

View video

Supplementary material 3 (MOV 3550 kb)

Copyright information

© Japan Ethological Society and Springer 2010