372 Bull. Br. arachnol. Soc. (2006) 13 (9), 372–376
The mystery of how spiders extract food without feeding have been done on groups with large chelicerae
masticating prey which masticate at least some prey (Bartels, 1930;
Zimmermann, 1934). Comstock (1948) added that, in
William G. Eberhard general, a spider ‘‘. . . sucks the juices pressed from its
Smithsonian Tropical Research Institute and prey by the mouthparts, . . .’’.
Escuela de Biología, Universidad de Costa Rica, In other spiders access to the interior of the prey is
Cuidad Universitaria, Costa Rica more complicated and, as we will show below, has been
misunderstood. The chelicerae of some groups, such as
Gilbert Barrantes and Ju-Lin Weng* filistatids, thomisids, scytodids, pholcids, some theridi-
Escuela de Biología, Universidad de Costa Rica ids, and uloborids (this list is undoubtedly incomplete),
are relatively small, and often lack teeth. Although in
Summary most of these groups the spider makes small external
Standard accounts of how spiders obtain food without holes in the prey to inject venom and to feed, the spiders
masticating their prey are probably largely wrong. Species do not masticate their prey, the exoskeleton of which
in the families Uloboridae, Thomisidae, Araneidae and remains more or less intact when the spider has finished
Theridiidae do not inject digestive fluid into the prey’s feeding (Turnbull, 1962; Kaestner, 1968; Lubin, 1986;
interior, nor do they suck fluids directly from its interior. Collatz, 1987). These spiders nevertheless gain extensive
Rather they regurgitate fluid onto the surface of the prey,
and then suck it back up from there. Philoponella vicina and access to its interior, as the discarded prey is an empty
other uloborids are extreme in this respect: they wet the husk; internal tissues of even relatively inaccessible parts
entire outer surface of the prey package simultaneously with such as the distal segments of legs are digested. Some
digestive fluid, and their mouthparts often never touch the spiders feed in both ways, masticating small, soft prey,
prey. Capillarity (along with digestion of prey membranes in but leaving large, rigid prey more or less intact (e.g. the
Philoponella) is apparently responsible both for the disper-
sion of digestive fluid into the prey, and the exit of liquids eresid Stegodyphus sp. (Y. Lubin, pers. comm.) and the
from inside the prey. araneid Allocyclosa bifurca (McCook), see below).
Descriptions from standard references imply that
spiders which leave their prey intact use changes in
Introduction pressure to feed: ‘‘. . . pump digestive juice into the
Spiders are well known to feed by regurgitating diges- body’’ of the prey; ‘‘the digested soft parts are then
tive fluid onto their prey, and then sucking up the sucked out again’’ (Kaestner, 1968: 179). Collatz (1987)
resulting nutrient-laden broth (Bertkau, 1885 in Bartels, adds that the digestive fluid is injected through small
1930; Bartels, 1930; Zimmermann, 1934; Comstock, holes that are produced by the chelicerae. Our observa-
1948; Kaestner, 1968; Collatz, 1987; Foelix, 1996). The tions, described below, of Latrodectus geometricus
regurgitated fluid, which presumably comes largely from C. L. Koch (Theridiidae), Misumenoides sp. (Thomisi-
the midgut (Kaestner, 1968) and the maxillary glands dae), Allocyclosa bifurca, and Philoponella vicina (O.
(Pickford, 1942), is very rich in proteins (about ten times Pickard-Cambridge) (Uloboridae), contradict these de-
richer than vertebrate duodenal or pancreatic juice), and scriptions, and necessitate a rethinking of how spiders
is later diluted when it enters the prey (Collatz, 1987). feed without masticating prey.
Spiders suck up liquid from the prey by using the
muscular sucking stomach, which increases the volume
of the foregut. Ingested food is probably nearly com- Methods
pletely liquid, as thick brushes of setae in the mouth Mature female spiders collected near San José, Costa
cavity, and a second filter in the pharynx (the ‘‘palate Rica were observed feeding at room temperature under
plate’’) strain out particles as small as about 1 
m a dissecting microscope. Some behaviour of each species
(Foelix, 1996). was filmed using a digital video camera (30 frames/s)
Some spiders, such as mygalomorphs, many araneids, through the microscope. In those species with webs, the
linyphiids, agelenids and lycosids, use their large cheli- wire hoop or jar containing the web was placed under
cerae and the teeth on the distal margin of the basal the microscope. The angle of viewing varied as the
cheliceral segment to crush and masticate their prey, spider manipulated the prey, and some details were
turning it into ‘‘a pulpy mass’’ (Bertkau, 1885 in Bartels, visible in some observations but not in others. It was
1930) as they feed. Thus, the spider’s regurgitated diges- thus not possible to be sure whether certain details were
tive juices have ample access to the internal tissues of always the same. In these cases we use the phrase ‘‘in at
the prey (Kaestner, 1968; Collatz, 1987), and the spider least some cases’’.
also has direct access to the nutrient-laden fluid from
the interior of the prey. The mechanics of ingestion
seem relatively simple in these groups. The most Results
detailed studies of the behaviour and morphology of Latrodectus geometricus
*Current address: Entomology, Kansas State University, Manhatten, Two Latrodectus geometricus, which completely lacks
Kansas, USA. cheliceral teeth, were observed feeding on three rela-
Address for correspondence: William Eberhard, Biología, Univ. de tively large prey (approximately 50–100 mg muscoid
Costa Rica, Ciudad Universitaria, Costa Rica. flies) under the microscope in antero-dorsal views.
W. G. Eberhard, G. Barrantes & J.-L. Weng 373
Feeding consisted of many repeated cycles of relatively Misumenoides sp.
rapid regurgitation of clear liquid, and then slower Feeding on a relatively large prey (a honeybee),
uptake of liquid in small rapid pulses. The prey was observed under the dissecting microscope in antero-
largely intact when the spider finished, except for holes dorsal view, also involved rhythmic cycles of regurgita-
at sites where she had fed. In one individual feeding at a tion and sucking. The duration of cycles averaged
hole on the dorso-lateral surface of the anterior half of a 2–3 min/cycle during early feeding, but later became
fly’s abdomen, clear liquid periodically accumulated shorter. A clear liquid periodically appeared abruptly
rapidly during regurgitation on the anterior surface of between the distal ends of the basal cheliceral segments,
the distal portions of the basal segments of her cheli- remained there unchanged for 11
4 s (n=9), and then
cerae and the space between them, then gradually disap- slowly disappeared. The liquid made small pulsing
peared during uptake. The liquid pulsed as it movements (17 pulses in 7 s, or 2.4/s) as it disappeared.
disappeared, presumably owing to rhythmic contrac- These pulsing movements were accompanied by rhyth-
tions of the spider’s sucking stomach. In three video mic movements of similar frequency of an unidentified
records, the mean frequency of three bouts totalling 76 structure that was just visible within the spider’s cephalo-
pulses varied between 4.9 and 7.4 pulses/s. The spider thorax; presumably these were associated with pumping
also repeatedly re-grasped the prey with her cheliceral movements of the spider’s sucking stomach. The last of
fangs, usually around the time that liquid was appearing the liquid disappeared 59
3 s (n=8) after the pulses
on her chelicerae. The rhythmic appearance and disap- began, and then no liquid was visible for another
pearance of liquid on the surface of the spider indicated 89
10 s (n=8) until the moment when the next abrupt
that her mouthparts were not sealed to the surface of the wetting began. The cheliceral fangs opened slightly at
prey; the spider was thus neither forcing liquid into the the beginning of each wetting, and closed slightly as the
prey by an increase in pressure, nor drawing it directly liquid began to disappear. The opening movement of the
out of the prey by sucking. fangs might have facilitated entrance of the liquid into
Further details supported the idea that the ‘‘wet–dry’’ the prey by widening the holes they made in the prey
cycles corresponded to repeated regurgitation and inges- cuticle. As with L. geometricus, the cyclic accumulation
tion. After the spider had fed on the abdomen of the fly and disappearance of liquid on the prey’s exterior
for >10 min, the abdomen of the fly began to collapse showed that the spider’s mouthparts did not make a seal
slowly during each period of pulsing or ingestion, then with the surface of the prey.
expanded rapidly when the spider regurgitated and Periodically the spider rotated the prey briskly with
liquid appeared on her chelicerae. In some of these cases her palps, then bit it again with her chelicerae and
it was clear that the fly’s abdomen pulsed while it performed further regurgitation–ingestion cycles. Often
collapsed at approximately the same frequency as the her chelicerae bit repeatedly at the prey, and at least
pulsing liquid. Later, movement of liquid within the prey some of these bites apparently penetrated its cuticle.
could be observed directly. Each time liquid appeared When the spider had fed at a site, a hole or holes could
between the spider’s chelicerae, there was a brief move- sometimes be seen in the cuticle. One hole near the base
ment of clear liquid within a membrane at the base of of a femur allowed direct observation of the fluid inside
the coxa of the fly’s leg III which was close to the as the spider fed more distally on a partially consumed
spider’s mouth. Liquid flowed away from the spider prey. The liquid inside rose when the spider regurgitated,
each time the spider regurgitated. The fly’s coxa was and fell when she sucked.
above the spider, so this liquid was not flowing down-
wards with gravity, but instead was moving upwards
within the prey’s body, presumably by capillarity. Later Allocyclosa bifurca
still it became clear that there was a slight delay between Five spiders were observed feeding on ten prey in
the moment when the liquid appeared on the spider’s ventral and anterior views. The chelicerae of this species
chelicerae and the fly’s abdomen inflated, and the have teeth, typical for the Araneidae, and the spiders
moment when the flow of liquid began in the base of the crushed small soft prey into a pulpy mass when feeding
fly’s leg. This delay supports the idea that the liquid on them (Fig. 1a). Large prey, such as calliphorid fly
moved into the fly by capillarity, rather than owing to about the size of the spider (20.6 mg) were, in contrast,
increased pressure, as a pressure increase would have left largely intact (Fig. 1b), and in these cases feeding
produced simultaneous or nearly simultaneous move- behaviour was similar to that described for the previous
ments in both the abdomen and the leg. species. After wrapping a calliphorid and fastening it
In sum, L. geometricus rhythmically regurgitated and near the hub, the spider grasped it with her chelicerae
ingested liquid; this liquid was not injected into the prey and repeatedly opened and closed them, apparently
under pressure, but instead flowed into the fly’s body, gnawing a hole in the lateral surface of the thorax. After
probably by capillarity. Apparently there was so much several minutes of gnawing, her chelicerae became less
liquid produced, and the fly’s partially digested abdo- active (in one case one fang was visible, and was inserted
men was so soft, that regurgitation caused the abdomen into the prey), and she began a long series of wet–dry
to inflate perceptibly, and then to collapse as the liquid cycles similar to those described above. Early in feeding
was withdrawn. As the prey’s internal tissues became on a calliphorid fly, the cycles were relatively short
digested, the liquid probably moved progressively (mean=13.7
2.2 s, n=8), but three minutes later they
deeper into the prey when the spider regurgitated. had become longer (mean=40.7=16.8 s, n=10). In at
374 How spiders extract food from prey
least some cases the spider’s endites were not pressed ning to feed. During the final burst of wrapping, clear
against the prey, but were instead up to about 1/3 the liquid appeared on the anterior surfaces of the spider’s
length of the labium away from it. This entire gap chelicerae, and her chelicerae moved actively. As soon as
between the prey and the endites (whose medial surfaces wrapping ended, the spider transferred the prey package
were separated and did not meet) was filled abruptly from her legs II and III to her palps and chelicerae, and
with clear liquid when the spider regurgitated, once immediately began to turn the package rapidly with her
again showing that there was no seal between the palps and chelicerae and simultaneously regurgitate a
spider’s mouthparts and the surface of the prey. The clear liquid which was spread over the surface of the
liquid remained more or less motionless for approxi- package as she turned it. This initial wetting behaviour
mately 10 s, and then gradually disappeared. After feed- lasted for only a minute or so with small prey, but up to
ing at one site for many minutes, the spider shifted the >20 min for very large prey. By the time initial wetting
prey, gnawed another hole, and fed. In eight calliphorid was finished (when the spider first stopped turning the
and muscid flies, feeding occurred at 2–7 sites. prey package), the entire surface of the package was wet.
Two details differed from the previous species. During It was not always feasible to observe details of the
the sucking phase, the endites moved rapidly back and movements of her chelicerae; in the glimpses we ob-
forth laterally at approximately the same rate (50 in tained of her cheliceral fangs, they were grasping the silk
13.4 s, or 3.7/s) as the movement of an object (presum- shroud, not the prey. The prey was turned so rapidly
ably the palate plate) that was dimly visible within the and continuously that it seemed unlikely that her fangs
labium. These probable pumping movements within had time to penetrate the prey cuticle.
the spider may have imparted pulsing movements to the Following the initial wetting, rotation of the prey
liquid as seen in other species, but such movements were slowed considerably and became intermittent. Periodi-
not noted, probably because they were masked by the cally the spider stopped rotating the package and fed for
movements of the endites. Secondly, the liquid gradually up to >20 min at a given site. At least with relatively
drained away from the gap between the spider and the small and weak-bodied prey such as Drosophila sp., the
prey during each sucking phase. It disappeared first spider began extracting material from the prey’s interior
from the posterior surfaces of the chelicerae and even- very early during feeding: red granules of eye pigment
tually dried up completely, thus breaking the liquid from the prey began to accumulate on the spider’s
connection between the spider’s mouth and the prey. endites as little as <1 min after she first began feeding
However, the endites and the object in the labium over the surface of the eye. As the spider’s chelicerae had
continued to vibrate; these movements ceased only when apparently not pierced the prey, this material probably
the spider regurgitated again. The period without a flowed through cracks in the prey cuticle that resulted
liquid connection between the spider’s mouth and her from the collapse of the compound eye owing to the
prey lasted for about 20–30 s when the spider was compression produced by wrapping (Eberhard et al.,
regurgitating about every 60 s. 2006). During long pauses at a feeding site, the
prey was repeatedly moved slightly, possibly resulting
from movements of the spider’s chelicerae. We were
Philoponella vicina
unable to observe her chelicerae at all times, and never
Six spiders were observed feeding on ten prey. As directly observed them penetrating the prey cuticle.
described elsewhere (Eberhard et al., 2006; Weng However, the discarded body of one large muscoid fly
et al., in press; Barrantes et al., in prep.), and as typical had a hole at each of the two sites at which the spider
for uloborids in general (Lubin, 1986; Opell, 1988), P. had spent long periods of time feeding (between the eyes,
vicina wrapped the prey extensively in silk before begin- and at the tip of the abdomen). Other, smaller discarded
Fig. 1: Extreme forms of the remains of prey discarded by Allocyclosa bifurca. a A small Drosophila sp. fly that was crushed into tiny pieces; b A
large calliphorid fly that was left largely intact, apart from two holes where the spider fed (arrows). Scale lines=0.89 mm (a) and 3.2 mm (b).
W. G. Eberhard, G. Barrantes & J.-L. Weng 375
Fig. 2: SEM micrographs of prey discarded by Philoponella vicina. a Disarticulated pieces of a parasitic wasp protrude through and lie on the outer
surface of the silk shroud; b The distal end of the trochanter of a parasitic wasp, with the intersegmental membrane entirely missing; c Empty
sockets and the bases of isolated setae (arrows) from which the membranes are missing in a fly.
prey had no holes. It appeared that during feeding, as a different spot; on other occasions she rotated the prey
during wrapping, the spider’s chelicerae and mouth area immediately after ingesting.
frequently failed to contact the prey directly, but only Examination of prey that had been discarded after
contacted the shroud covering the prey. feeding often showed that portions of their bodies, such
While feeding, the spider cyclically (approximately as their legs, were disarticulated (Fig. 2a). This raised the
once every 30 s) regurgitated clear fluid abruptly, and possibility that we had missed direct manipulation of the
then more slowly ingested it. As in the previous species, prey by the spider. Closer examination of the ends of
it was clear that the spider’s mouthparts did not make a disarticulated segments of legs showed, however, that the
seal with the surface of the prey. As the liquid was intersegmental membranes had disappeared completely
regurgitated, it flowed away from the spider’s mouth in (Fig. 2b). In addition, free setae, which lacked their basal
an expanding circle; this ‘‘wave’’ was visible at distances membranes, were often scattered nearby, sometimes
from her mouth up to approximately the length of her adhering to the inner surface of the shroud (Fig. 2c). The
sternum. Following regurgitation, the area on the sur- loss of membranes suggests that prey became disarticu-
face of the prey where the liquid was perceptibly deeper lated as a result of the spider having digested those
began to shrink gradually in small, barely perceptible membranes. Hard-bodied prey, such as ants and small
pulses of about 5/s (the exact frequency was difficult to beetles, had no obvious breaks in their cuticle when
measure in video recordings). In one case the spider’s they were discarded, but were nevertheless also partially
abdomen and legs IV pulsed with this same rhythm. At disarticulated and more or less empty inside.
later stages, it was possible to observe movements of the
liquid directly by following the movement of small
Discussion
bubbles that formed under the shroud and inside por-
tions of the prey such as inside leg segments: the liquid Extended cycles of rhythmic regurgitation and suck-
moved gradually towards the spider as she sucked, and ing occurred in all four species. This may be an ancient
abruptly away from her when she regurgitated. Coordi- and generalised method of feeding in arachnids, as
nation between regurgitation, ingestion, and rotation of similar cycles have been described in a pseudoscorpion
the prey varied: sometimes the spider regurgitated just by Schlottke (1933, in Pickford, 1942), and in the
before rotating the prey and then sucked up liquid from agelenid Tegenaria domestica (Clerck) (Bartels, 1930).
376 How spiders extract food from prey
In none of the four species studied here did the Our observations indicate that fluid flowed into and
spider’s mouthparts form a tight seal against the surface out of the prey’s interior by capillarity, rather than as a
of the prey during feeding. Digestive liquid welled into a result of pressure changes, as suggested by previous
pool between and around the distal portions of the authors. Access to the prey’s interior was probably
spider’s basal cheliceral segments each time she regurgi- provided by holes made when the spider injected venom
tated. This pool spread over the entire prey in the or, in P. vicina, by breaks in the cuticle resulting from
uloborid. The digestive liquid was thus free on the compression from wrapping and by the ability of the
surface of the prey, and was not pumped directly into its digestive fluid to digest prey membranes (Eberhard
interior. The slow subsidence and disappearance of this et al., 2006; Weng et al., in press). The low surface
pool while the spider sucked also indicates that the tension of the digestive fluid of P. vicina compared with
spiders did not make a seal with the prey’s surface to tap water (Weng et al., in press) probably facilitates
suck liquid directly from its interior. Instead, the spiders movement of fluid into and out of the prey; the surface
sucked only from the pool of liquid on the surface of the tension of the digestive fluid of other species has not to
prey. As noted in the introduction, these observations our knowledge been measured.
contradict standard accounts of feeding by spiders, and Given these indications of capillary flow, and the poor
necessitate rethinking the mechanism by which spiders design of the relatively thin cheliceral fangs of spiders
extract nutrients from their prey. for pressing more than a small fraction of the liquid
A spider’s problems of getting digestive fluid into and from a prey, we doubt that Comstock (1948) was correct
then out of prey can be appreciated intuitively by in his description of spiders feeding by pressing out fluid
imagining the difficulty a human would have in extract- when prey is masticated. Simple sucking on such prey is
ing nutrients, spider-style, from a soft drink bottle probably enough to extract most of its liquid contents.
(representing the hard cuticle of the prey) that was open
at one end (the wound produced by the spider’s bite) and Acknowledgements
full of liquid (the prey’s blood and internal tissues). One We thank Yael Lubin and Brent Opell for useful ideas
could only add digestive liquid at the mouth of the and discussions regarding spiders, Arno Lise for
bottle, and only small quantities of liquid at that. identifying the thomisid, Roberto Cordero and Kyle
Forcing additional liquid into an unyielding bottle that Harms for information regarding capillarity, and the
is already full is not feasible; and in any case it would not Smithsonian Tropical Research Institute and the Uni-
be possible to create pressure to force it in, if one’s versidad de Costa Rica for financial support.
mouth failed to make a tight seal with the bottle (as is
the case for spiders). This analogy is somewhat over-
References
drawn, because at least some prey are not entirely rigid
(see the description of the abdomen of a prey of Latro- BARRANTES, G., EBERHARD, W. G. & WENG, J.-L. in prep.:
dectus expanding and collapsing), but it illustrates the Prey wrapping and feeding by Philoponella vicina (Araneae,Uloboridae).
basic problem. BARTELS, M. 1930: Ueber den Fressmechanismus und den che-
It might seem that at least part of the problem of getting mischen Sinn einiger Netzspinnen. Revue suisse Zool. 37: 1–41.
digestive fluid into the prey could be solved by first creat- COLLATZ, K.-G. 1987: Structure and function of the digestive tract.
ing several holes in the prey’s exterior and then sucking In W. Nentwig (ed.), Ecophysiology of spiders: 229–239. New
out some blood, so as to create an empty space within the York: Springer-Verlag.COMSTOCK, J. H. 1948: The spider book. Revised and edited by
prey, into which the digestive fluid could be added. But W. J. Gertsch. Comstock Pub. Assoc., Ithaca, NY.
feeding did not begin with repeated bites at different sites EBERHARD, W. G., BARRANTES, G. & WENG, J.-L. 2006: Tie
in L. geometricus, A. bifurca or P. vicina (we did not see its them up tight: Philoponella vicina spiders mangle and kill their
initiation in the thomisid); and at least in the uloborid, prey by wrapping. Naturwissenschaften.
spiders routinely regurgitated onto the external surface of FOELIX, R. 1996: Biology of spiders. Cambridge, MA: HarvardUniversity Press.
their prey before either perforating it or sucking. KAESTNER, A. 1968: Invertebrate zoology Vol II. Translated and
The soft drink bottle analogy also illustrates a spider’s adapted by H. W. Levi & L. R. Levi. New York: John Wiley &
problems with the second process, sucking up nutrients Sons.
from the prey. Unless one made an additional hole in the LUBIN, Y. D. 1986: Web building and prey capture in the UIobori-
rigid walls of the bottle, one could not suck out fluid dae. In W. A. Shear (ed.), Spiders: webs, behavior, and evolution:132–171. Stanford, CA: Stanford University Press.
from inside, as sucking liquid from a closed-ended OPELL, B. D. 1988: Prey handling and food extraction by the
container is not feasible unless the walls of the bottle triangle-web spider Hyptiotes cavatus (Uloboridae). J. Arach-
collapse. In addition, when one’s mouth does not make nol. 16: 272–274.
a tight seal with the surface of the bottle (as in spiders), PICKFORD, G. E. 1942: Studies on the digestive enzymes of spiders.
negative pressure cannot be produced inside the bottle Trans. Conn. Acad. Arts Sci. 35: 33–72.TURNBULL, A. L. 1962: Quantitative studies of the food of Linyphia
by sucking; thus sucking could only remove the fluid triangularis Clerck (Araneae: Linyphiidae). Can. Ent. 94: 1233–
that had accumulated on the outside of the bottle 1249.
around the opening. Finally, in contrast with the behav- WENG, J. L., BARRANTES, G. & EBERHARD, W. G. in press:
iour of a human who is attempting to extract liquid from Feeding by Philoponella vicina (Araneae, Uloboridae). Can. J.
a soft drink bottle, spiders did not tip their prey upside Zool.ZIMMERMANN, E. W. 1934: Untersuchungen über den Bau des
down to allow the liquid to run out; some spiders even Mundhöhlendaches der Gewebespinnen. Revue suisse Zool. 41:
fed exclusively from the upper half of the prey! 149–176.