2006. The Journal of Arachnology 34:456–466
THE PREY ATTACK BEHAVIOR OF ACHAEARANEA
TESSELATA (ARANEAE, THERIDIIDAE)
Gilbert Barrantes and Ju-Lin Weng: Escuela de Biologı́a, Ciudad Universitaria
Rodrigo Facio, Universidad de Costa Rica, San Pedro, San José, Costa Rica.
E-mail: gbarrantes@biologia.ucr.ac.cr
ABSTRACT. The attack behavior of the cobweb spider Achaearanea tesselata (Keyserling 1884) is
roughly separated into three sequential steps: descend from the suspended retreat, pass through the sheet
threads, and wrap the prey from underneath the sheet. The position and speed as the spider descended
varied apparently according to prey type. In the fastest descent, A. tesselata fell free upside down, with
all legs free and stretched upward. Two other relatively slow types of descent occurred when spiders
approached the sheet head down or climbing down on a mesh thread. The behavior used to pass between
the sheet lines showed little variation. It occurred at high speed with the legs folded dorsally; when the
legs were in this position the spider offered a very small area of impact, apparently permitting the femora
to penetrate or open a space between the lines of the sheet. The spider then opened the femora of the legs
to create enough space for the cephalothorax, and seizing the sheet from underneath with legs I, II, and
III, the spider pulled the abdomen and hind legs through the sheet. Then the spider rushed to the prey,
flung viscid lines at the prey, and wrapped it. Attacks occurred in as little as 0.11 s after the spider began
its descent.
The design of the webs of A. tesselata transmits information regarding the location of the prey trapped
on the sheet to reach the resting spider inside the retreat. The first response of the spider in her retreat
was to turn to face the prey; the spider then climbed down along mesh threads following a nearly straight
line to the prey.
Keywords: Cobweb spider, prey capture, orientation, web use, tangle web, silk
The design of both a spider’s web and its sheet, with a dense mesh above and a few
attack behavior are likely complementary and support lines below (Fig. 1, Eberhard 1972).
together affect prey capture success (Eberhard A curled leaf, tiny twigs, small dried flowers,
1990). Webs generally trap and partially im- or other debris serve as the spider’s retreat in
mobilize prey (Chacón & Eberhard 1980; nearly all webs. The spider rests upside down
Nentwig 1982), but usually prey escape unless within the retreat, which is suspended near the
the spider further immobilizes them (Eberhard middle of the upper mesh. The web of A. tes-
1990). Therefore, prey capture success is par- selata works as a knock down trap for flying
tially determined by the speed with which the and jumping insects. Insects that strike the up-
spider reaches the prey, which in turn is prob- per mesh fall down onto the sheet where they
ably largely dependent on the information the are attacked by spiders. Jörger and Eberhard
spider receives regarding prey location in the (pers. comm.) suggested that lines in the mesh
web (Biere & Uetz 1981). The speed and of A. tesselata could also offer information to
complexity of attack behavior may be at least spiders regarding prey location in the web.
partially related to the web design (Eberhard The design of this web is apparently shared
1990), since some webs may impose some re- by two other species in the genus: A. dispar-
striction on the spider’s movements. ata Denis 1965 (Darchen 1968) and A. japon-
Achaearanea tesselata (Keyserling 1884) is ica (Bosenberg & Strand 1906) ( Theridium
relatively common in bushes in urban areas japonicum). However, A. tesselata seems to be
and highly disturbed vegetation throughout its unique in resting above the sheet and passing
distribution in Central and South America through it to attack prey (Eberhard 1972).
(Levi 1959; Eberhard 1972). The web of this According to Eberhard (1972), the first re-
spider consists of a finely-spun horizontal sponse of A. tesselata to prey in its web was
456
BARRANTES & WENG—ATTACK BEHAVIOR OF A. TESSELATA 457
Figure 1.—Web of A. tesselata: notice the upper mesh (solid arrow) and the sheet (dotted arrow).
to descend from the retreat. Passing rapidly (Sony DCR-VX 1000) that recorded 30
through the sheet without causing any appar- frames/s.
ent damage, the spider rushed directly to the Detailed descriptions of attack behavior are
prey wrapping and/or biting it. Considering based on video analyses. Drawings are based
the high density of threads in the sheet (Eber- on digital-video images imported into a com-
hard 1972; Jörger & Eberhard pers. comm.), puter using the program iMovie, version 2.0.
it is puzzling how the spider passes so rapidly Different portions of the spider were not al-
through the sheet and how this structure suf- ways in focus, hence sample sizes for different
fers no apparent damage. In this paper we de- analyses differed. Descriptions are all based
scribe in detail how A. tesselata descends on samples of at least 10 individuals, sample
from the retreat and passes through the sheet sizes lower than 10 are indicated in the text.
and discuss the advantages and possible evo- Density of sheet threads.—We estimated
lution of this behavior. Furthermore, we test the density of threads in sheets of five webs
the hypothesis that the upper mesh lines give of adult females by counting the number of
information regarding prey position on the threads crossing the diameter of the 1.82 mm
sheet to the spider. field of view of a sample on a glass micro-
METHODS scope slide under a compound microscope.
Field observations.—Observations on at- Threads were counted from 10 to 15 fields of
tack behavior of A. tesselata were made in the view in each of the five sheets and then the
field, from March to September 2004, within average number of threads per millimeter was
the campus of the Universidad de Costa Rica, calculated. Samples from the sheets were col-
San Pedro, San José Province, Costa Rica lected on slides framed with strips of double-
(954N, 8403W; elevation 1200 m). Attack sided adhesive tape (1.5 mm thick  2.5 mm
behavior of A. tesselata occurs so rapidly that wide). Slides were carefully lifted from below
the spider’s descent and details of its move- the web into the threads of the sheet; the
ments as it passes through the sheet cannot be threads extending beyond the slide were cut
distinguished with the naked eye. Thus, in ad- so that only threads adhering to the tape were
dition to field observations, we video recorded collected. This method allows observation of
the complete attack sequence of 37 adult fe- threads without modifying their original ar-
males of A. tesselata in the field; each indi- rangement. Additionally, on each of these
vidual was recorded from one to four times, slides, we randomly selected 10 sheet lines of
giving a total of 52 sequences. Video record- approximately 10 mm long. Along each se-
ings were made with a digital video camera lected line, thread connections and number of
458 THE JOURNAL OF ARACHNOLOGY
threads crossing over, but not attached to the dom, hence the deviation from the prey po-
line were counted in 10 fields of view (0.45 sition could vary from 0 (no deviation from
mm each). The width and length of the ceph- prey position) to 180 (maximum deviation
alothorax and abdomen were also measured from prey position). We used a Monte Carlo
on 30 adult females of A. tesselata for com- Analysis (Statistica package, version 5.0) to
parison with thread density of sheets (14 from produce 500 random samples of possible de-
the collection of the Museo de Zoologı́a, viation angles. The mean of departure angles
Universidad de Costa Rica, and 16 used in a of A. tesselata was compared with a distri-
prey location experiment). Voucher specimens bution of random means using a one sample
of spiders have been deposited in the Museo Student’s t-test. We complemented the infor-
de Zoologı́a, Universidad de Costa Rica. mation obtained in the laboratory with field
Experiments on orientation and descend- observations on spiderlings and adult female
ing speed.—Sixteen mature females with re- orientation.
treats and egg sacs were each placed on a We also measured the time a spider spent
three-dimensional wire structure with six ex- in descending from the retreat to the sheet and
tensions projected downward, forming a hexa- the attack time from the retreat to the moment
gon, hanging from a thin fishing line at 2 m the attack of the prey began. Time was esti-
above the floor. The fishing line makes it dif- mated from the video records of different spi-
ficult for the spiders to escape upward as they ders using a frame (30 frames/s) as a time-
cannot climb it. They frequently descend, but reference unit. Times were compared among
usually at about a meter they turn back if they descent and prey types using one way ANO-
have not encountered an object below. The re- VAs. Velocity of descent was calculated for
treats in the webs of these spiders were small 15 webs in which a house fly or a blow fly of
enough to allow observations of the move- similar size was dropped in the center of the
ments of the spider inside. After one or two sheet. The descent time was estimated in all
nights spiders had spun a complete web with cases from the video records and the distance
the sheet approximately following the hexag- from the retreat to the sheet was measured to
onal shape of the three-dimensional structure. calculate the velocity. We also calculated the
To determine the orientation of the spiders free fall velocity by dropping two recently
during the attack behavior, we dropped Dro- dead mature females. These spiders were fro-
sophila sp. flies on the sheets of the finished zen until dead, and then allowed to thaw. We
webs and video recorded the spider’s move- next dropped and recorded the falling time for
ments in the retreat and its orientation as it each spider from a platform placed at 5.5 cm
approached the prey. To video record spider’s above a landing surface; we repeated this
movements the camera was fixed to a platform three times for each spider.
with the lens at 10 cm above the retreat and
aligned perpendicular to its top. Before each RESULTS
attack, we randomly selected one of the six Orientation in retreat and mesh.—The
sections of the approximately hexagonal sheet first response of A. tesselata in the retreat to
on which the prey was to be dropped. Each a Drosophila dropped onto the sheet was to
spider was recorded only once. We applied a turn to orient facing ‘‘toward’’ the same sector
binomial analysis to evaluate if the attack ap- of the sheet. The orientation movements of
proach of the spider was random with respect spiders in the retreat occurred in all 16 trials
to the six segments of the sheet. More precise conducted in the laboratory: four times prey
measurements of the attack orientation of the was in front of the spider, six times it was
spider were obtained from video records; spe- lateral to her, and six times behind her. While
cifically we obtained the difference in degrees descending to Drosophila prey, spiders
of the approaching direction of the spider to climbed down along the mesh lines, following
the prey and prey position in the sheet. We a nearly straight line. In 15 out of 16 times
compared the mean differences between ap- the spider moved to the 60 section of the
proaching direction and prey position using a sheet on which a prey was dropped (P 
random distribution of means. This distribu- 0.0001, Binomial test). The average angle de-
tion was constructed under the assumption viation of spiders to prey location at arrival to
that the attack direction of spiders was ran- the sheet was 7.0 (8.2). This small deviation
BARRANTES & WENG—ATTACK BEHAVIOR OF A. TESSELATA 459
the mesh. The longitudinal plane of the spider
formed an angle of 30 (SD  6, n  3),
relative to the sheet, with the cephalothorax
directed downward, and dorsum facing the
sheet (Fig. 2). Given the position of the legs,
it is likely that A. tesselata hangs from at least
four or five different threads while in the re-
treat.
Descending behavior.—The position of
the body and legs when A. tesselata came
Figure 2.—Resting position of A. tesselata in a down from the retreat, after a prey falls on the
retreat; retreat was not drawn to show the position sheet, varied between individuals and within
of the legs (dash lines indicate the possible position the same spider depending on prey type, and
of threads from which the spider hangs). possibly on hunger and experience of the spi-
der. Relatively large flies (e.g., house flies) re-
is significantly smaller than a random mean leased the fastest reaction from the spider, pro-
(87.6  48.51; t  9.87, df  15, P  voking a very fast descent to the sheet (Table
0.0001). The lines on which A. tesselata de- 1). Treehoppers, of about the size of a house-
scended probably intersected several other fly, evoked a relatively slow descent; similar
threads, since the upper mesh in the web is reactions were elicited by Drosophila. The
relatively dense and lines are interconnected, variation fits relatively well into three cate-
but the spiders continued apparently selecting gories: free fall, head-down, and walking on
the lines running more directly toward the mesh threads.
prey. We could not, however, see the lines in Free fall: The spider fell rapidly while up-
most cases. On three occasions we could see side down (dorsum toward the sheet) with all
that the spider paused briefly at an intersection legs apparently directed upward. Falls were so
of lines; she briefly jerked the threads with her rapid (52.3 cm/s, SD  13.8, range  30.3–
legs I, and then moved on the line that ran 80.0; Table 1) that the spider was generally a
most nearly toward the prey. blur in the video (Figs. 3–6). The spider fre-
In the field we video recorded the orienta- quently bounced when her body first struck
tion behavior of three fourth or fifth instar ju- the sheet, but sometimes passed through on
veniles on their mother’s web. The spiderlings first contact. When this type of descent oc-
were on threads about three centimeters to the curred, hind legs apparently did not contact
side and two centimeters below the retreat the dragline. The falling speed of dead spiders
when a Drosophila was dropped on the sheet measured in the laboratory was similar to the
5 cm away on the opposite sector of the sheet. speed calculated for spiders in the field (54.1
The spiderlings did not descend directly to the cm/s, SD  5.6). However, field data may be
sheet, but walked directly toward the prey affected by wind currents, which reduce the
through the mesh, passing under the retreat. speed, and by different weights of spiders. In
Position in the retreat.—The spider hung addition, the spider was apparently capable of
inside the retreat (if present; Fig. 2), above the directing her descent toward the prey position
sheet, where several threads converged from as she fell. The Figures 3–6, obtained from a
Table 1.—Average time (standard deviation in parentheses) spent by Achaearanea tesselata descending
from the retreat to the sheet web and attacking prey. Time of attack includes from the moment the spider
initiates its descent to the moment the attack begins. Information is separated by descent type; units in
seconds.
Descent type
Free fall (n  21) Head down (n  10) Walking (n  15)
Descent 0.11 (0.02) 1.36 (0.65) 4.08 (3.46)
Attack 1.54 (1.42) 4.64 (5.41) 5.86 (5.45)
460 THE JOURNAL OF ARACHNOLOGY
Figures 3–6.—Sequence of events during the free fall descent of A. tesselata. 3. The spider inside the
retreat. 4. The spider beginning to fall (blur indicated by the arrow). 5. Note that the spider is only a blur
due to the descent speed (arrow). 6. Point where spider struck the sheet (arrow). Note the angle of falling
direction of the spider toward the left of the picture where the prey contacted the sheet (indicated by a
star).
video record taken in the lab to avoid wind ative to the sheet plane. While walking, the
currents, shows how the spider oriented her spider moved forward and alternately extend-
fall toward the prey position (left-inferior cor- ed her anterior legs, while one leg IV held the
ner of the picture). From video records, it was dragline. This type of descent was much more
not possible to obtain information on how frequent when prey fell near the edge of the
such orientation was attained by falling spi- sheet.
ders. Crossing through the sheet threads.—
Head down: The spider advanced more The spider passed her body through the sheet
slowly toward the sheet maintaining a more threads with a series of extraordinarily rapid
or less head-down position (Table 1, Figs. 7– movements of legs and body that enabled her
8). The spider advanced in short jerks: it first to ease her relatively large body (Table 2)
extended the anterior and second pair of legs through the small spaces between the sheet
forward and, waving them, possibly grabbed lines (Table 3). The movements of the spider
some mesh threads and moved downward a were so rapid that it was necessary to use 32
short distance. This sequence was repeated video records of attack behavior to assemble
several times until the spider reached the the complete sequence of movements as the
sheet. In this descent the spider moved almost spider passed through the sheet. In all cases,
perpendicularly to the sheet through the mesh, independently of how she arrived at the sheet,
possibly using its lines as ‘‘the steps of a lad- the spider passed this structure dorsal side
der’’. During this type of descent the spider first. Hence, when the spider descended either
alternated the leg IV that held the dragline. head down or walking along mesh threads,
Walking on mesh threads: A. tesselata ap- she repositioned her body before passing
proached the sheet by walking directly toward through the sheet. In these cases, when the
the prey along one or a few mesh threads. In spider was nearly touching the sheet, she ex-
this descent the spider’s body was oblique rel- tended her first legs and grabbed a sheet line
BARRANTES & WENG—ATTACK BEHAVIOR OF A. TESSELATA 461
Figure 9.—Ventral view of A. tesselata pulling
sheet lines with her first legs. Note most of the other
legs directed backwards and grabbing mesh threads
to possibly counteract the forward tension exerted
by pulling sheet threads.
dragged forward, indicating that the tension
was counteracted by tension on mesh threads
held by the other legs, which were directed
backward, except one hind leg that was in
contact with the dragline (Fig. 9). From this
position, A. tesselata released all but the first
legs at once, falling in a very fast movement
(ca. 0.03 s), dorsum against the sheet (Figs.
10–11).
Independently of how the spider descended
from the retreat, as she passed through the
sheet, her legs were tightly folded against the
body, forming a compact structure that likely
Figures 7–8.—Position and movements of the
legs of A. tesselata in a head-down descent. 7. Ex-
tending legs I–II and waving them to grab some Table 2.—Length and width of abdomen (Ab)
mesh threads. 8. Movement of legs as the spider and cephalothorax (Cph) in millimeters for 30 ma-
descended a short distance. ture females of Achaearanea tesselata.
Ab Ab Cph Cph
length width length width
with each one (Fig. 9), and then pulled these
threads toward her body by folding the first Average 3.1 2.3 1.8 1.4
legs. The sheet threads she had grabbed were S. D. 0.39 0.23 0.10 0.07
bent toward the spider’s body, but she was not Range 2.6–3.7 2.0–2.6 1.7–2.0 1.3–1.5
462 THE JOURNAL OF ARACHNOLOGY
Figures 10–17.—Sequence of movements and position of legs of A. tesselata passing through the sheet.
(Lines represent the approximate position of the sheet; threads grabbed by spider’s legs in 10 and 18 were
clearly observed in the video records. All legs were not always drawn when video images were not focused
or angle did not allow a clear view). 10. The spider falling on the sheet after releasing its grasp on mesh
threads. 11. The spider contacting the platform as she falls, note how legs began to fold on the cephalo-
thorax. 12. The platform is deformed by the impact of the spider, the legs of the spider are tightly folded
BARRANTES & WENG—ATTACK BEHAVIOR OF A. TESSELATA 463
Table 3.—Average of attachment points and un- (Figs. 12–13). Then the spider grabbed some
attached lines crossing over along 10 randomly se- sheet threads from underneath with legs I, II,
lected lines in five different sheets; counts were ob- and III and pulled herself downward, pulling
tained in 10 fields of view (0.45 mm each) per and freeing first the legs IV (Fig. 16), which
sheet. Mean density of sheet lines per mm calcu-
lated as number of lines across 10 to 15 fields of grabbed some lines as soon as they passed
view (1.82 mm each) for five different sheets. through the sheet. The spider continued pull-
ing, by stretching the legs ventrally, and
Attachment Unattached dragged the abdomen through the sheet and
points lines Density then hung with some legs free (Fig. 17) before
Average 1.15 1.60 1.96 rushing toward the prey. In a few occasions,
S. D. 0.61 0.80 0.59 the spider showed some difficulty dragging
Range 0.58–2.16 1.08–3.00 0.52–3.85 the abdomen through the sheet so that she had
to struggle to free her abdomen from the lines.
The entire process of passing through the
sheet took from 0.03 to 0.20 s (mean  0.10,
facilitated penetration between the dense wo- SD  0.06, n  12).
ven threads of the sheet (Fig. 12; Table 3). Capturing and transporting prey.—As
Femora (and possibly coxae and trochanters)
soon as the spider was hanging from the sheet
of legs I, II, and III of both sides were directed
dorsally over the cephalothorax, nearly touch- lines, she rushed directly to the prey that rest-
ing the anterior surface of the abdomen. The ed on the sheet. With the prey at reach, the
femur-patella joints touched or nearly touched spider began the attack by flinging lines with
each other and the more distal segments were large viscid globs up onto it with her fourth
pointed ventrally with tarsus and metatarsus tarsi (Griswold et al. 1998). Immediately after
(possibly for legs I and II) bent over the ceph- the first attack, A. tesselata started wrapping
alothorax (Fig. 12). The femora of legs IV the prey, pausing frequently to either bite it or
were approximately parallel to the others, but clean the tarsi of legs IV, whereas the other
the more distal segments of these legs were legs grabbed sheet lines from below. When a
directed backward and pressed against the large prey was fully wrapped, the spider rap-
sides of the abdomen (Fig. 12); one leg IV idly broke some sheet threads, attached a
held the dragline. dragline by pressing the spinnerets to the prey
The spider struck the sheet with the folded and pulled it up to the retreat with a fourth
legs resting against the cephalothorax first, leg, climbing up along a mesh line. The drag-
while her abdomen was bent upward at an an- line was frequently attached half way to the
gle of 30 (SD  5, n  3; Figs. 11–12). As retreat, and the spider climbed down by the
the spider contacted the sheet, femur-patella same line, completely releasing the prey from
joints of possibly all legs were apparently the sheet threads, wrapping the prey with a
pushed through a space between the sheet few more lines, and attaching another line,
threads (Fig. 13). Then, the femora of all legs pulled up the prey closer to the retreat. This
began to stretch out laterally, widening the procedure was repeated until the prey was at
space between threads of the sheet to permit about 1 cm from the retreat. If the prey was
the cephalothorax to pass through (Fig. 14– small, the spider directly carried it up nearby
15). During some of these movements the dis- the retreat with one hind leg. Occasionally, a
tal segments of legs IV were maintained ap- prey escaped before the spider reached it (n
parently immobile and pressed against the  6). In such cases, spiders returned to the
abdomen, which was still above the sheet retreat passing back upward through the sheet.
←
on the cephalothorax, except distal segments of legs IV that are pressed against the sides of the abdomen.
13. Joints femur-patella-tibia of all legs passing through the sheet. 14. Femora begin to stretch out, opening
enough space to seize the cephalothorax through the sheet. 15. Femora widely stretched. 16. Most legs
grabbing sheet threads from underneath, the abdomen half way passing the sheet. 17. The spider hanging
from the underside of the sheet.
464 THE JOURNAL OF ARACHNOLOGY
The process was slow and clumsy. The spider Griswold et al. 1998; Benjamin et al. 2002;
introduced different legs into different spaces Benjamin & Zschokke 2004), but in other
between the sheet lines, making it difficult to families (e.g., Lycosidae, Pisauridae) the spi-
pass through. Usually she only succeeded af- der runs on top of the sheet to reach the prey
ter several attempts, and cut and broke some (Eberhard 1990). Theridiidae webs with aerial
lines before getting free. sheets are present in some Anelosimus, Tidar-
Time spent during the descent and at- ren and Achaearanea, e.g., A. disparata, A.
tack.—The time the spider spent descending tesselata, and A. japonica (Darchen 1968;
from the retreat to the sheet was determined Eberhard 1972; Shinkai & Takano 1984).
by how the spider approached the sheet. The However, the attack behavior associated with
average time that the spider spent in a free fall the presence of a sheet in the web is very
descent was significantly lower than the time different between genera and species. Some
she spent in a head down or walking descent social species of Anelosimus attack prey from
(F2,43 17.48, P  0.0001; Table 1). Accord- below the platform (Levi 1972), similar to
ingly, the mean time from the moment the spi- some Tidarren (pers. obs.), other Anelosimus
der began her descent until the moment the that build sheets with knockdown lines, attack
attack (attack time) of the prey began was sig- insects entangled either in trap lines or in the
nificantly lower when she fell free from the sheet (Avilés & Salazar 1999; Vakanas &
retreat (F2,43 5.29, P  0.01). However, dif- Krafft 2001). In Achaearanea, the spiders of
ferences in attack time are primarily related to the social species A. disparata descend di-
the lower time of free fall descents, since the rectly from the retreat to the prey on the sheet
time the spider lasted from the moment she without passing through this structure to ac-
contacted the sheet to start attacking the prey cess prey (Darchen 1968; Darchen & Ledoux
did not differ among spiders using different 1978) contrary to A. tesselata; no information
descent types (F(2,38) 1.74, P  0.19). is available for A. japonica.
The type of prey apparently determined The unusual web design probably has likely
how the spider descended from the retreat to shaped, at least partially, the evolution of the
the sheet. For example, free fall was more fre- complex attack behavior of A. tesselata. Many
quent when prey were blow flies or house Achaearanea species construct three-dimen-
flies, walking along lines when prey were dro- sional gum-foot webs, specialized for walking
sophilids, and head down descents were more prey and/or working as knock down traps
frequent when prey were treehoppers (	2 (Gertsch 1949); this type of web is ancestral
42.4, df  4, P  0.0001). in Theridiidae (Levi & Levi 1962; Benjamin
Prey entangled in mesh lines.—Occasion- & Zschokke 2003; Agnarsson 2004). Similar-
ally a prey was entangled in mesh lines at ly, the presence of a retreat suspended in the
about the same level of the retreat (n  4). upper mesh is common in species of Achaear-
When this occurred the spider descended, anea, and is also a feature present in other
passed through the sheet, and shook it from genera, such as Tidarren and Theridion (Bris-
underneath, apparently trying to determine the towe 1958; Agnarsson 2003). Falling upside
prey’s location. After a few seconds the spider down from the retreat also occurs as an escape
climbed through the sheet into the mesh. The response in Argyrodes (Whitehouse 1986), A.
movements as she climbed in the mesh were lunata (Clerck 1757) ( T. lunatum) (Nielsen
clumsy and her orientation toward the prey 1932), A. tepidariorum (C. L. Koch 1841) and
imprecise, compared with the rapid and pre- in Tidarren sp. (G. Barrantes unpubl. data)
cise movements when locating prey on the when spiders are disturbed, suggesting that
sheet. this behavior is also widespread in theridiids.
Therefore, the construction of a sheet in
DISCUSSION Achaearanea seems to be newly evolved.
Horizontal aerial sheets with mesh above Thus, A. tesselata, retaining the theridiid re-
and/or below have independently evolved in a treat trait, presents a novel behavior that al-
wide variety of separated spider groups (Shear lows her to pass through the sheet and access
1986; Eberhard 1990). In some families (e.g., prey from underneath; this novel behavior
Linyphiidae) the spiders run upside down on probably derived from a former escape behav-
the lower surface of the sheet (Bristowe 1958; ior.
BARRANTES & WENG—ATTACK BEHAVIOR OF A. TESSELATA 465
Some elements of the sequence of the at- protective barrier between the spider and the
tack behavior were fairly variable such as the prey.
descent to approach the prey, while the posi- The tangled mesh of the web of A. tesselata
tion of legs and body when the spider passed transmits precise information on prey location
through the sheet showed very little variation. to the spider in her retreat. Spiderlings are also
The speed and movements of this spider when capable of perceiving prey location through
the prey is approached seemed to be deter- information transmitted by mesh lines. This is
mined primarily by the size and type of prey, probably due to the convergence of most mesh
though position of prey on the sheet and hun- lines that connect with the sheet below on five
ger may also determine attack speed (Riechert or six more or less horizontal threads attached
& Luczak 1982). Highly rewarding harmless to the mouth of the retreat where the spider
prey with high probability of escaping (e.g., hangs (Jörger & Eberhard pers. comm.). This
houseflies) released the fastest reaction: free structure may channel vibrations to the central
fall descent (Table 1). When the prey was a point where the spider rests. Probably infor-
treehopper, the approach was slow. The strong mation is yielded by vibrations or changes in
kicks from the hind legs of treehoppers can tension produced by the prey, as occurs in
probably cause serious injuries to spiders. The Latrodectus (Lamoral 1968). This information
harmless drosophilids were also approached allowed the spider to orient her attack before
slowly, possibly as a consequence of the little initiating her descent. In contrast, information
reward these prey represent to mature females of prey position was not efficiently transmit-
of A. tesselata. This indicates that A. tesselata ted to the spider when prey fell and entangled
is either capable of sensing prey type and how on mesh lines, as the spider’s orientation was
dangerous the prey could be, and of modify- imprecise. The reason for this imprecision, in
ing the capture sequence accordingly (Riech- terms of thread connection, is not clear. Thus
ert & Luczak 1982); or it may attack faster the design of the mesh in A. tesselata may
prey sending strong signals (e.g., high buzz- increase efficiency of capture success of prey
ing) as they may have greater reward. trapped on the sheet.
The spiders’ consistent positions while The diversity on web designs and frequent
passing through the sheet are probably a re- convergences in the Theridiidae (Benjamin &
sponse to the high density of threads in this Zschokke 2003; Agnarsson 2004) offer an op-
structure relative to the size of the spider. The portunity to study the interrelation of web de-
position of the body and legs when striking sign and attack behavior. For example a com-
the sheet offered the least contact area, and parative study of Achaearanea species with
the high speed at the moment of impact (even similar web design, could provide a better un-
after a slow descent) may help to insert femur- derstanding of the evolution of the attack be-
patella joints between two threads and open havior in the genus, and the function of the
enough space for the spider to pass through. upper mesh in transmitting information on
The absence of perceptible damage caused to prey location.
the sheet during an attack, reported by Eber-
hard (1972), is largely determined by the nu- ACKNOWLEDGMENTS
merous unattached lines that form part of this We thank W.G. Eberhard, I. Agnarsson, G.
structure (Table 3). These lines are probably Stratton, and two anonymous reviewers for
easily separated when spiders strike the sheet. their valuable comments on the manuscript.
The consistent orientation of the body, posi- We are indebted to W.G. Eberhard for provid-
tions of the legs, and behavior as the spider ing references and the picture of an A. tesse-
passed down through the sheet largely con- lata web, but particularly for his intellectual
trasted with the inconsistent uncoordinated stimulus throughout this study. We owe the
movements of the spider going upward design of slides to collect web samples to the
through the sheet after an unsuccessful attack. generosity of C. Grismado.
Intense selection on prey capture success may
have reduced variation in downward move- LITERATURE CITED
ments to increase the rapidity of attack (Eber- Agnarsson, I. 2003. The phylogenetic placement
hard 2000). An obvious advantage of attack- and circumscription of the genus Synotaxus (Ar-
ing prey from underneath is to interpose a aneae: Synotaxidae), a new species from Guya-
466 THE JOURNAL OF ARACHNOLOGY
na, and notes on theridioid phylogeny. Inverte- Wendilgarda spiders (Araneae: Theridiosomati-
brate Systematics 17:719–734. dae). Ethology Ecology and Evolution 12:223–
Agnarsson, I. 2004. Morphological phylogeny of 235.
cobweb spiders and their relatives (Araneae, Ar- Griswold, C.E., J.A. Coddington, G. Hormiga & N.
aneoidea, Theridiidae). Zoological Journal of the Scharff. 1998. Phylogeny of the orb-web build-
Linnaean Society 141:447–626. ing spiders (Araneae, Orbiculariae: Deinopoidea,
Avilés, L. & P. Salazar. 1999. Notes on the social Araneoidea). Zoological Journal of the Linnean
structure, life cycle, and behavior of Anelosimus Society 123:1–99.
eximius. Journal of Arachnology 27:497–502. Lamoral, B.H. 1968. On the nest and web structure
Benjamin, S.P. & S. Zschokke. 2002. Untangling of Latrodectus in South Africa, and some obser-
the tangle-web: web construction behavior of the vations on body colouration of L. geometricus
comb-footed spider Steatoda triangulosa and (Araneae:Theridiidae). Annals of the Natal Mu-
comments on the phylogenetic implications (Ar- seum. 20:1–14.
aneae: Theridiidae). Journal of Insect Behavior Levi, H.W. 1959. The spider genera Achaearanea,
15:791–809. Theridion and Sphyrotinus from Mexico, Central
Benjamin, S.P. & S. Zschokke. 2004. Homology, America and West Indies (Araneae, Theridiidae).
behaviour and spider webs: web construction be- Bulletin of the Museum of Comparative Zoology
haviour of Linyphia hortensis and L. triangularis 121:57–163.
(Araneae Linyphiidae) and its evolutionary sig- Levi, H.W. 1972. Taxonomic-nomeclatorial notes
nificance. Journal of Evolutionary Biology 17: on misplaced theridiid spiders (Araneae: Theri-
120–130. diidae) with observations on Anelosimus. Trans-
Benjamin, S.P., M. Düggelin & S. Zschokke. 2002. actions of the American Microcopical Society
Fine structure of sheet-webs of Linyphia trian- 91:533–538.
gularis (Clerck) and Microlinyphia pusilla (Sun- Levi, H.W. & L.R. Levi. 1962. The genera of the
devall), with remarks on the presence of viscid spider family Theridiidae. Bulletin of the Muse-
silk. Acta Zoologica 83:49–59. um of Comparative Zoology 127:1–71.
Biere, M. & G. Uetz. 1981. Web orientation in the Nentwig, W. 1983. The non-filter function of orb-
spider Micrathena gracilis (Araneae: Araneidae). webs in spiders. Oecologia 58:418–420.
Ecology 62:336–344. Nielsen, E. 1932. The Biology of Spiders with Es-
Bristowe, W.S. 1958. The World of Spiders. Will- pecial Reference to the Danish Fauna. Volumes
mer Brothers and Haram Ltd., London. 308 pp. 1 & 2. Levin & Munksgaard, Copenhagen, Den-
Chacón, P. & W.G. Eberhard. 1980. Factors affect- mark. 723 pp.
ing numbers and kinds of prey caught in artificial Riechert, S.E. & J. Luczak. 1982. Spider foraging:
spider webs, with considerations of how orb behavioral responses to prey. Pp. 353–386. In
webs trap prey. Bulletin of the British Arachnol- Spider Communication: Mechanisms and Eco-
logical Significance (P.N. Witt & J.S. Rovner,
ogical Society 5:29–38.
eds.). Princeton University Press, New Jersey.
Darchen, R. 1968. Ethologı́e d’Achaearanea dis-
Shear, W.A. 1986. The evolution of web-building
parata Denis, Aranea, Theridiidae, araignée so- behavior in spiders: a third generation of hypoth-
ciale du Gabon. Extrait de la Revue Biologia Ga- eses. Pp. 364–400. In Spiders—Webs, Behavior,
bonica. 4:5–25. and Evolution (W.A. Shear, ed.). Stanford Uni-
Darchen, R. 1978. Achaearanea disparata, araignée versity Press, Stanford, California.
sociale du Gabon, synonyme ou espece jumelle Vakanas, G. & B. Krafft. 2001. Coordination of be-
d A. tessellata, solitaire. Revue Arachnologique havioral sequences between individuals during
1:121–132. prey capture in a social spider, Anelosimus exi-
Eberhard, W.G. 1972. Observations on the biology mius. Journal of Insect Behavior 14:777–798.
of Achaearanea tesselata (Araneae: Theridiidae). Whitehouse, M.E.A. 1986. The foraging behaviours
Psyche 79:209–212. of Argyrodes antipodiana (Theridiidae), a klep-
Eberhard, W.G. 1990. Function and phylogeny of toparasitic spider from New Zealand. New Zea-
spider webs. Annual Review of Ecology and land Journal of Zoology 13:151–168.
Systematics 21:341–372.
Eberhard, W.G. 2000. Breaking the mold: behav- Manuscript received 17 August 2005, revised 12
ioral variation and evolutionary innovation in July 2006.