Proc. NatL Acad. Sci. USA Vol. 80, pp.

3382-3385, June 1983 Evolution

Spider leg autotomy induced by prey venom injection: An adaptive response to "pain"?*
(chemical defense/chemoreception/coevolution/pharmacology)

THOMAS EISNER AND SCOTT CAMAZINE

Section of Neurobiology and Behavior, Cornell University, Ithaca, New York 14853

Contributed by Thomas Eisner, February 22, 1983 ABSTRACT Field observations showed orb-weaving spiders (Argiope spp.) to undergo leg autotomy if they are stung in a leg by venomous insect prey (Phymatafasciata}. The response occurs within seconds, before the venom can take lethal action by spread to the body of the spiders. Autotomy is induced also by honeybee venom and wasp venom, as well as by several venom components (serotonin, histamine, phospholipase A2, melittin) known to be responsible for the pain characteristically elicited by venom injection in humans. The sensing mechanism by which spiders detect injected harmful chemicals such as venoms therefore may be fundamentally similar to the one in humans that is coupled with the perception of pain.

Bee stings hurt. So do wasp stings, scorpion stings, the bites of centipedes, and the venom injections of many other animals, including snakes. To inflict pain is not necessarily to the advantage of an animal that uses its venom strictly for incapacitation of prey. In fact, it may be to its disadvantage because pain may induce increased struggling on the part of the prey. But venoms are also used' defensively, and it is in that context that they may derive their effectiveness largely, if not exclusively, from their pain-inducing qualities. It is principally because venoms are painful that they can function in defense. Pain, in the sense of a consciously perceived experience, remains a subjective notion applicable to humans but untestable with animals. But when defined operationally as a physiological phenomenon induced in an animal by stimuli that are painful to us and resulting in a protective stimulus-avoidance response in that animal, pain is amenable to testing with nonhuman subjects. We here report that' certain chemical venom components that are painful to humans are also responsible for eliciting a natural self-preserving response in spiders: leg autotomy. Leg autotomy had previously been noted to occur in spiders when a leg was pulled or injured. It occurs consistently at the level of the coxa-trochanter joint near the base' of the leg. A special mechanism provides for minimization of bleeding at the site of leg detachment, and spiders can withstand the loss of several legs (1, 2). We found autotomy to occur also when a spider is stung in the leg while attempting to capture venomous
insect prey. Here we present evidence that, under such circumstances, the response is adaptive to spiders because it prevents sytemic spread of the poisons and that it is triggered by action of the venom itself rather than by the mechanics of cuticle perforation or fluid injection. Moreover, we show that certain well-known pain-inducing compounds found in venoms are capable of inducing the autotomy response.
The publication costs of this article were defrayed in part by page charge payment. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. 1734 solely to indicate this fact.

EXPERIMENTAL PROCEDURES AND OBSERVATIONS Our study was prompted by field observations made while studying prey-capture behavior in orb-weaving spiders of the genus Argiope. A small ambush bug, Phymatafasciata, had just been seen to fly into a web of Argiope aurantia, when the spider pounced upon it in typical fashion, in anticipation of wrapping it in silk and killing it. Suddenly the behavior came to a halt. The phymatid had grasped the tip of one of the spider's legs with its prehensile forelegs, and it was holding on while seemingly attempting to puncture the leg with its proboscis. The spider remained motionless at first, but then abruptly autotomized the seized appendage and retreated to the hub of the web. The relinquished leg remained in the hold of the phymatid (Fig. 1 D-F). Tests with captured phymatids that were flipped individually from vials into natural webs of Argiope (A. aurantia and A. trifasciata) showed autotomy to follow almost invariably when a spider's leg was grasped by a phymatid. But leg seizure occurred only rarely. Of 43 phymatids that were offered, 36 failed to gain a hold on the spider and were quickly spun in and killed. Six of the remaining seven phymatids seized the spider's leg and induced autotomy. The seventh also seized a leg, but autotomy did not follow. The encounter had an unexpected outcome. The spider seemed to attempt to pull away from the phymatid as soon as it had wrapped the insect, but it was unable to do so and became stationary and, within seconds, entirely motionless. For a full 20 min, during which the phymatid never released its hold and the spider remained still, the pair was kept under observation. When at the end of this period the spider was prodded, it failed to respond and was found to be dead (Fig.

1G). Phymatids can inflict a noticeable "sting" with their proboscis. They do this whenever given the opportunity (as when they are held between the fingers) to bring the tip 'of the proboscis to bear against the skin. The sting mechanism is essentially similar to what it is in the well-known relatives of phymatids called assassin bugs (family Reduviidae). Four needlesharp stylets, tightly apposed to form a slender bundle, effect the puncture, which occurs with simultaneous injection of venomous saliva (3). Prey insects, including powerful species such as bumblebees, are quickly immobilized by phymatid venom (Fig. 1 A and B). In humans, the sting induces an instantaneous sharply localized pain followed by a mild itch. Laboratory observations showed that it is the sting, rather than the mere mechanical grasp of the spider's leg by the phymatid, that induces the autotomy. Spiders (A. trifasciata) were
*

This is report no. 74 of the series "Defense Mechanisms of Arthropods." Report no. 73 is Nowicki, S. & Eisner, T., Psyche, in press.

3382

Evolution: Eisner and Camazine

Proc. NatL Acad. Sci. USA 80 (1983)

3383

FIG. 1. (A-C) Phymata fasciata as predator and prey (arrows point to dorsum of thorax of phymatid): (A) feeding on syrphid fly; (B) feeding on a hesperid butterfly which has been caught by its slender proboscis; (C) being eaten by an araneid spider. (D-F) Sequence of events in an attack of A. aurantia upon P. fasciata: (D) as spider inspects phymatid, one of its legs is caught by the latter; (E) stung in the leg, the spider autotomizes that leg and retreats to the hub; (F) moments later the spider preens in its normal upside-down stance at the hub, while the detached leg remains in the hold of the phymatid. (G) Comparable encounter, of fatal outcome to Argiope; the phymatid, already spun in, has grasped and stung a leg of the spider, but the latter failed to autotomize and died as a result. The spider is here shown hanging limply by its seized appendage shortly after dying. (H) Close-up view of phymatid stinging an Argiope leg (photograph of dead specimen, killed by immersion in liquid Freon at - 190'C while it was stinging, and subsequently lyophilized while frozen). Bars: D, 10 mm; H, 1 mm.

fastened ventral side down on a flat surface with wax droplets applied to the venter of the "abdomen" (technically the spider's opisthosoma) and to the tips of their outstretched legs. Individual phymatids attached by their backs to tethers were then brought into contact with an articular joint (the femur/patella or patella/tibia joint) of a spider's leg. They reached out with

the forelegs, grasped the leg, and proceeded to probe the joint membrane with the proboscis. Autotomy never occurred during such preliminary probings, but it followed quickly (within 5 see in 10 of 10 cases) once the phymatid began pressing its proboscis against a given site of the membrane, a moment that is coincident with the onset of stinging. Stylet penetration at

3384

Evolution: Eisner and Camazine

such moments could be seen through the translucent joint membrane by observation with a stereomicroscope. Further testing with immobilized spiders showed that mere mechanical perforation of a leg was no substitute for the sting. A single puncture was made, at the femur/patella joint or at the midsection of the femur or tibia, in one leg of each of eight spiders (four A. aurantia and four A. trifasciata) by using a flamesterilized steel pin. Autotomy followed in one case only. There was some bleeding from the wounds, but the spiders showed no long-range ill-effects. That failure to autotomize following a phymatid sting may lead to lethal systemic envenomization in Argiope was already apparent from the single field encounter of fatal outcome. Argiope are unable to withstand phymatid stings to the body. Eight spiders (three A. aurantia and five A. trifasciata) that were caused to be stung on the dorsal surface of the abdomen by tethered phymatids all eventually died. Kept under observation in test arenas immediately after the sting, they showed locomotory impairment within seconds and total immobility and lack of responsiveness within minutes. Quantitative data on the autotomy-inducing capacity of venoms and venom components was obtained by injecting fixed volumes (1.0 /.l) of test substance into individual legs of A. trifasciata. Injection was effected into the femur/patella or patella/tibia joint of a leg with a glass micropipette operated by a micrometer-advanced microsyringe. The injection took a fraction of a second. The delay to autotomy was timed to the nearest second from the moment of injection, by using a foot-operated stopwatch. Up to four legs per spider were injected, each never more than once; at least 1 min transpired between consecutive leg injections. All autotomies occurred within 19 sec (x SD, 4.5 + 3.7; n = 101 legs); failure to autotomize within 30 sec was scored as no response (experience had shown that autotomy is not usually delayed beyond that period). Because hemipteran venoms are poorly characterized chemically and unavailable commercially, we used honeybee and wasp
100-

(Vespa maculata) venom for our tests, as well as known major components of hymenopteran venoms. The venom components were injected in physiological saline (4), at concentrations approximating those in natural honeybee or wasp venoms (5, 6); the venoms, available in lyophilized form, were brought to natural concentrations by dilution with distilled water. Saline and hypertonic NaCl (7%) were injected as controls. As is clear from the results (Fig. 2), the venoms themselves as well as four of the venom components (serotonin, histamine, phospholipase A2, and melittin) were active, inducing autotomy in upward of 47% of cases. Other venom components were inactive (acetylcholine, bradykinin, hyaluronidase, adrenaline, and dopamine). The controls were inactive or minimally active.
DISCUSSION Three of the four active components-serotonin, phospholipase A2, and histamine-are known to be pain-inducing in humans (7-9). Whatever the neural basis of their detection in spiders-whether it be comparable to the pain-coupled venomsensing mechanism in humans or not-it is clear that spiders are highly sensitive to these substances and that they respond defensively to them in a manner that prevents their systemic spread. All these compounds are of relatively widespread occurrence in animal (and even plant) venoms (10-12), a reflection perhaps of their broad potential effectiveness vis a vis enemies. Arthropods in particular may need to cope with vertebrate and arthropod predators alike, and it is interesting that serotonin, phospholipases, and histamine are present in many of their venoms (10). The fourth active component, the polypeptide melittin, is also pain-inducing in humans but it is of restricted distribution and known from honeybee venom only (13). Of the inactive venom components tested, three-hyaluronidase, dopamine, and adrenaline-are reportedly painless, or at least not potently pain-inducing (8, 9), in humans. Although these compounds contribute to the overall pharma-

20

75
E 0
0

27

25

F7 I//

24

50-

19

7-

0-0

lp:

O0

4-

,g 00
9

0-

0o

00

'I'

CO

0 0z p .'

vrTo

Com
bA
Co

Q)

Hlo :e. *oo1 0\0p~

d@ ~r

IV,

FIG. 2. Incidence of leg autotomy in A. trifasciata in response to venom or venom component injected. Numbers above bars or in parenthesis after substance give sample size (= number of legs injected). Percentages give concentration (wt/vol) in physiological saline; honeybee and wasp venom concentrations were 12% and 7.5%, respectively (weight lyophilized venom/volume distilled water).

Proc. Natl. Acad. Sci. USA 80 (1983)

15

X10
'
"~
.

,.~

wXMX"

tp

+.

.,C,

lb

e.'

Evolution: Eisner and Camazine cological effect of venoms and may even (as in the case of hyaluronidase) (8) act to potentiate the activity of other venom components, they are apparently not pain-inducing as such. But two components, acetylcholine and bradykinin, which proved ineffective vis a vis Argiope (at least at the concentrations tested), are definitely pain-inducing in humans (8, 9). Therefore, there may be no absolute concordance in the chemical sensitivities of the venom-detecting systems of humans and Argiope. Given the phyletic gap between arachnids and vertebrates, this should come as no surprise. What is remarkable is that there should even be substantial overlap in the pharmacological spectrum of what induces pain in us and what acts as if it were painful to
Argiope.

Proc. Natl. Acad. Sci. USA 80 (1983)

3385

The five inactive substances that failed to induce autotomy invariably proved nonlethal to the spiders. But death did not always follow in spiders that retained their leg after injection of active samples. Although honeybee venom and phospholipase A2 were always lethal to nonautotomizing spiders (three and six spiders respectively), histamine, serotonin, and melittin were consistently not (eight, four, and five spiders). And, somewhat surprisingly, neither was wasp venom (three spiders). The chemosensory system that "forewarns" spiders against systemic toxins is thus not attuned strictly to molecules that are lethal. Our tests with phymatids offered no evidence that these insects might secure their release from predators by defensive use of their venoms. They remained trapped in the Argiope webs whether they induced leg autotomy or not, and all were eventually eaten. But tests that we did with a jumping spider, Phidippus audax, a most likely natural enemy of phymatids, gave different results. Four individual Phymata were offered to an equal number of female Phidippus, held in small arenas in which the spiders had been maintained for months on a diet of live insects. Three of the phymatids were promptly killed by the spiders and eaten. The fourth was also pounced upon, but it grasped a leg of the spider, stung it, and caused the appendage to be autotomized. The spider backed away, leaving the phymatid unencumbered and potentially free to make an immediate escape. Leg autotomy occurs also in arthropods other than spiders. If seized by a leg, katydids often relinquish the appendage, as

do many grasshoppers, crickets, and cockroaches. Tests that we did with unidentified katydids showed them to autotomize readily in response to Phymata stings (n = 5 katydids), indicating that they too have the means for quick-sensing of potentially hazardous toxins. And so do spiders other than Argiope. Each of seven unidentified spiders of three families (Salticidae, Thomisidae, and Agelenidae) autotomized in response to phymatid stings. But one spider-the common house spider Achaearanea tepidariorum (family Theridiidae)-proved exceptional in that it never autotomized (n = 11 spiders) despite its evident susceptibility to envenomization (7 of 11 died). Whether this spider lacks the appropriate venom-sensing mechanism or is physically unable to autotomize and whether its vulnerability to venoms is correlated with a life style that provides for minimization of exposure to venoms remains unknown.
The initial field observations and laboratory tests were made at the Huyck Preserve, Rensselaerville, NY, in collaboration with Maria Eisner, who also helped with the photography. This study was supported byGrant AI-02908 from the National Institutes of Health and by a generous stipend from Cornell University.
1. 2. 3. 4.

558. 5. O'Connor, R. & Peck, M. L. (1978) in Arthropod Venoms, ed. Bettini, S. (Springer, Berlin), pp. 613-659. 6. Edery, H., Ishay, J., Gitter, S. & Joshua, H. (1978) in Arthropod Venoms, ed. Bettini, S. (Springer, Berlin), pp. 691-771. 7. Schmidt, J. 0. (1982) Annu. Rev. Entomol 27, 339-368. 8. Chahl, L. A. & Kirk, E. J. (1975) Pain 1, 3-49. 9. Keele, C. A. & Armstrong, D. (1964) Substances Producing Pain
10. Bettini, S. (1978) Arthropod Venoms (Springer, Berlin). 11. Habermehl, G. (1976) Gift-Tiere und ihre Waffen (Springer, Ber12. Thurston, E. L. & Lersten, N. R. (1969) Bot. Rev. 35, 393-412. 13. Habermann, E. (1971) in Venomous Animals and Their Venoms, Venomous Invertebrates, eds. Bucherl, W. & Buckley, E. E. (Academic, New York), Vol. 3, pp. 61-93.

Weber, H. (1930) Biologie der Hemipteren (Springer, Berlin). Griffiths, J. T. & Tauber, O. E. (1943) J. Gen. Physiol 26, 541-

Parry, D. A. (1957) Q. J Microsc. Sci. 98, 331-340. Wood, F. D. (1926)J. Morphol Physiol 42, 143-195.

and Itch (Edward Arnold, London).

lin)..