Cerveira & Jackson—Pisaurid spiders from and Sri Lanka New Zealand Journal of Zoology, 2002, Vol.Australia 29: 119–133 0301–4223/02/2902–0119 $7.00/0 © The Royal Society of New Zealand 2002

119

Prey, predatory behaviour, and anti-predator defences of Hygropoda dolomedes and Dendrolycosa sp. (Araneae: Pisauridae), web-building pisaurid spiders from Australia and Sri Lanka ANA CERVEIRA Departamento de Zoologia e Antropologia Faculdade de Ciencias Universidade de Lisboa Portugal ROBERT R. JACKSON Department of Zoology University of Canterbury Private Bag 4800 Christchurch, New Zealand email: [email protected] Abstract Pisaurids are most often described as hunting spiders, but Hygropoda dolomedes and Dendrolycosa sp. are tropical pisaurids that build sheet webs on the tops of large waxy leaves. Prey records from the field show that dipterans dominate the diets of these species (60% of 85 records for H. dolomedes and 55% of 80 records for Dendrolycosa sp.). Mosquitoes are taken especially often (32% of the records for H. dolomedes and 25% of the records for Dendrolycosa sp.). Prey-capture behaviour, feeding, and anti-predator defence are described. Methods are developed for inducing spiders to build webs on leaves and artificial leaves in the laboratory. Keywords spiders; webs; predation; anti-predator defence; mosquitoes

INTRODUCTION Many spiders use silk to build structures that function in prey detection and prey capture (Witt 1975; Shear 1994), and it is to these that the term “web” is most commonly applied when specifying broadly defined predatory strategies. Spiders are often divided informally into two groups, web builders and hunters (e.g., Foelix 1996), although Z00031; published 17 June 2002 Received 4 September 2000; accepted 19 January 2002

alternative terms for the latter group are prevalent, including “wandering spiders” and “cursorial spiders”. However, this informal classification can be misleading because even a single individual spider may sometimes be a web builder and sometimes be a hunter (Jackson 1986). Among arachnologists, it is widely appreciated that generalisation to entire families is especially problematic. Even classic “web-building families”, such as the Araneidae and Theridiidae, include many species that fail to build webs (Stowe 1986). The reverse is also common. For example, Lycosidae, Gnaphosidae, Oxyopidae, Salticidae, and Thomisidae are all typically described as “hunting-spider families”, yet there are web-builders in each of these families (Job 1974; Griswold 1983; Jackson 1985; Jarman & Jackson 1986; Jackson et al. 1995). Pisaurids are usually characterised as one of the hunting-spider families, but this is yet another family where both web-building and hunting spiders can be found (Gerhart & Kästner 1938; Blandin & Celerier 1981; Davies 1982; Carico 1985; Nentwig 1985; Heidger 1988). Here we document web structure, prey, predatory sequences, and anti-predator defences of two pisaurid species for which there have been no previous behavioural studies: Hygropoda dolomedes Simon from Australia and Dendrolycosa Doleschall sp. from Sri Lanka. In nature, the webs of H. dolomedes and Dendrolycosa sp. are always on the tops of green leaves, but inducing these spiders to build webs on leaves in captivity was initially difficult. Here we report on the procedures that succeeded. These procedures may prove applicable for laboratory rearing of other spider species that spin webs on the tops of leaves.

MATERIALS AND METHODS Hygropoda dolomedes and Dendrolycosa sp. have slender bodies and long, spiny legs (Fig. 1). Colouration of body and legs is pale green on H. dolomedes and brown on Dendrolycosa sp. Complex

120

New Zealand Journal of Zoology, 2002, Vol. 29 Fig. 1 Hygropoda dolomedes. Adult female (body length 8 mm) in typical rest posture on web. Note slender body, spiny legs, and complex abdominal markings. Web on artificial leaf.

markings (red, orange, and yellow) may be present, especially on the posterial dorsal abdomen. Adult males and females tend to be similar in size (body length c. 8 mm for H. dolomedes and c. 10 mm for Dendrolycosa sp.). Males of Dendrolycosa sp. tend to be paler than females, whereas males of H. dolomedes tend to be, on the cephalothorax and distal leg segments (tibia to tarsus), more brownish than females. Field sites were in tropical rain forest near Cairns (Queensland, Australia; 16°55′S, 145°47′E) and Galle (Sri Lanka; 6°01′N, 80°13′E). Laboratory cultures derived from spiders collected at these field

sites were established at the University of Canterbury, New Zealand, where maintenance, data analysis, and testing procedures followed those described in earlier studies (for details see Jackson & Hallas 1986). All laboratory testing was carried out between 0900 and 1700 h in a controlledenvironment laboratory (photoperiod 12L:12D, lights on 0800 h). Video tapes of prey-capture sequences and anti-predator behaviour were analysed frame by frame. No spiders were tested more than once per day. Previously used terms (Jackson & Hallas 1986) were adopted to define relative prey size: large (prey

Cerveira & Jackson—Pisaurid spiders from Australia and Sri Lanka

121

Fig. 2 Taxonomic categories of prey on which spiders were found feeding in the field (for more detailed taxonomic information, see Appendices 1 and 2) by Hygropoda dolomedes (N = 85) and Dendrolycosa sp. (N = 80).

about 50% larger in estimated body volume than the spider); medium (prey and spider about equal in size); small (prey about half the size of the spider); very small (prey about one tenth size of the spider). “Larger” is a collective term for medium and large prey. “Smaller” means small and very small. The expressions “usually” or “often”, “sometimes” or “occasionally”, and “rarely” or “infrequently” are used, respectively, for frequencies of occurrence of 80% or more, 20–80%, and 20% or less. The spider’s legs are specified as pairs I-IV (anterior to posterior). Hygropoda dolomedes and Dendrolycosa sp. were similar in most respects. Whenever there is no need to designate a particular species, the simpler term “spider” was used when referring collectively to these two species. PREY RECORDS FROM THE FIELD Prey records were based entirely on spiders that were feeding when observed. Spiders were always in their webs when observed feeding. Prey remains were rarely present in webs, as H. dolomedes and Dendrolycosa sp. discard prey remains after feeding. Whenever we found a spider feeding, we recorded prey size and collected the prey for identification (taken to the lowest taxonomic level feasible). Working with a poorly known fauna, and with prey that were often in poor condition when collected, species determination was often impossible. Although records were not kept of spiders that were not feeding when seen, a rough estimate is that fewer than 5% of the spiders seen were feeding.

Prey records for the two species were similar (Appendices 1 and 2). Using pooled data, more than 95% of the records were insects, the remainder being spiders (Fig. 2). Dipterans were by far the most common prey (60% for H. dolomedes, 55% for Dendrolycosa sp.). Mosquitoes accounted for an especially large percentage of the total number of prey recorded for each species (32% for H. dolomedes and 25% Dendrolycosa sp.). Hemipterans were also well represented in the prey records (13% for H. dolomedes and 19% for Dendrolycosa sp.), but the majority were “homopterans” (nine out 11 for H. dolomedes and 14 out of 15 for Dendrolycosa sp.). Lepidoptera accounted for 9% of the prey records for H. dolomedes and 3% for Dendrolycosa sp. Except for two caterpillars taken by H. dolomedes, all lepidopterans were moths. Ephemeroptera (mayflies) accounted for 4% of the prey records for H. dolomedes and 6% for Dendrolycosa sp. Winged termites were taken by both species (4% of records for H. dolomedes and 7% for Dendrolycosa sp.). Additional prey were taken by only one species: three psocids (H. dolomedes), two winged ants (Dendrolycosa sp.), two mantises (H. dolomedes), one cockroach (Dendrolycosa sp.) and one mecopteran (Dendrolycosa sp.). Four insects fed on by Dendrolycosa sp. could not be identified to order. Prey size ranged from very small to large for both species. Smaller prey dominated the records: more than 70% of H. dolomedes and more than 90% of Dendrolycosa sp. (Fig. 3).

122

New Zealand Journal of Zoology, 2002, Vol. 29 Fig. 3 Size categories (see text) of prey on which spiders were found feeding in the field (for more details, see Appendices 1 and 2) by Hygropoda dolomedes (N = 85) and Dendrolycosa sp. (N = 80).

Webs in nature The web was a horizontal sheet of non-sticky silk lines on the top of a green leaf. Leaves were usually waxy and more or less ellipsoid in shape (5–10 mm wide and 10–30 mm long). Juveniles and adults of both sexes built similar webs. More than one web was never found on a single leaf, and usually each web almost entirely covered the surface of the leaf on which it was situated. Single webs of Dendrolycosa sp. sometimes extended across two leaves, but only rarely across three or more leaves. Each web of H. dolomedes was always on a single leaf. Usually the leaf was positioned horizontally and had edges that curled upwards slightly (i.e., the top surface of the leaf was somewhat concave). The web was suspended over this concave surface, the margins of the leaf providing anchorage points. The spider sometimes sat on a small silk pit in the silk sheet. The pit was similar in size to the spider and located more or less in the centre of the web. The bottom of the pit was often attached by silk to the upper surface of the leaf. The pit and surrounding web region tended to be more densely woven than the rest of the web. Procedures for inducing web building in the laboratory “Natural leaves” were green leaves of Pseudopanax spp. collected locally in New Zealand. These leaves, like the leaves on which these pisaurids built their webs in nature, were waxy, slightly concave (uppersurface bent up slightly around sides), more or less ellipsoid in shape and about 120 mm long and

60 mm wide. “Artificial leaves” were made by cutting green cardboard (c. 0.5 mm thick) into ellipses comparable in size and shape to the natural leaves. We bent the cardboard up at the sides, giving the artificial leaf a slightly concave shape, more or less comparable to that of natural leaves. A leaf (natural or artificial) was placed concave side up in a clear plastic petri dish (diameter 140 mm). A damp cotton roll was placed inside the petri dish, positioned to one side. Spiders usually built webs on these leaves within 1–2 days. Webs on natural and artificial leaves were similar to each other and to webs in nature. Webs usually covered the entire upper surface of the leaf (natural or artificial), with lines attached to the margins holding the sheet taut (Fig. 4). Spiders sometimes extended the web out and used the sides of the petri dish as additional anchorage points, but the bulk of the silk was always over the leaf rather than the plastic of the petri dish. Webs could be readily removed intact simply by picking up the leaf because lines fastened to the edge of the petri dish broke away without distorting the web. Predatory sequences For qualitative information, prey-capture sequences were observed in the field by dropping prey onto webs or by pushing prey onto webs with a brush. Other webs were taken intact from the field (with the spider in the web) into the laboratory to observe predation. Predatory sequences were also investigated using webs built in the laboratory by first removing a leaf (natural or artificial) from a petri

Cerveira & Jackson—Pisaurid spiders from Australia and Sri Lanka

123

Fig. 4 Hygropoda dolomedes (adult female) walking on web on artificial leaf (cardboard). Web is a sheet of silk across concave upper-leaf surface. Note taut lines in front of spider. Body close to silk and palps on silk.

dish and suspending it by a clip from a stand, then dropping or pushing prey with a brush onto the web. Predatory sequences observed by all of these methods were comparable. More than 500 sequences were observed for each species.

ELEMENTS OF PREDATORY BEHAVIOUR The following (listed alphabetically) are described: apply silk to prey (9), bite and stab (4), discard prey remains (12), feeding (10), lunging attack on prey (6), manipulate with palps (11), pull on prey (7), raised-body posture (2), rest posture (1), retractedlegs posture (3), rotate over prey (8), simple attack on prey (5). 1. Rest posture. Spiders in this posture (Fig. 1) held their bodies close and parallel to the web (ventral surface of body no more than 2–3 mm above silk). Legs were almost fully extended at each joint. Legs I and II were close together and oriented forward. Legs III and IV were oriented rearward.

Palps were held anterior to the chelicerae, angling down and usually in contact with the web. The posture of walking spiders (Fig. 4) deviated only slightly from the resting posture (body a few millimetres higher; palps remaining on or near the silk). 2. Raised-body posture. This posture (Fig. 5) resembled the rest posture except that spiders stood with bodies held higher (ventral surface of body 5–6 mm above the silk). 3. Retracted-legs posture. In this posture, the spider’s body was sometimes elevated above the silk as in the raised-body posture. Other times the spider’s body was no more elevated than when in the rest posture. However, leg positioning was always very different from that in either the rest posture or the raised-body posture. Femur-patella joints of legs I and II, and sometimes III, were highly flexed. The spider held the tarsi of these legs just in front of its chelicerae. Legs IV were stretched rearward, with almost all joints fully extended.

124

New Zealand Journal of Zoology, 2002, Vol. 29

Fig. 5 Hygropoda dolomedes (adult female), in raised-body posture while shifting feeding position on a house fly. Biting fly’s abdomen, manipulating fly with palps, and taking hold of fly’s abdomen with chelicerae.

4. Bite and stab. Spiders attacked prey by stabbing or biting. Biting: the spider placed chelicerae around a prey and held on. Stabbing: with fangs extended, the spider lunged at a prey, punctured it with extended fangs, then immediately moved back (did not hold on). Sometimes spiders combined the two attack modes by biting immediately after stabbing. 5. Simple attack on prey. A spider suddenly and rapidly rushed toward the prey, moved over it, bit it, and held on (Fig. 6). 6. Lunging attack on prey. From 10–20 mm away, a spider with fangs extended made a sudden dash toward a prey. Legs I were also extended, angling c. 45° up from the substrate (more or less parallel and forward, with most joints fully extended). Instead of biting, the spider usually stabbed the prey upon contact, keeping legs I elevated until after backing away. 7. Pull on prey. First the spider placed legs I and II, and occasionally III, on the prey. Pulling was achieved by rapidly and simultaneously extending

legs IV fully, lowering the body and flexing legs I–III. The prey was thereby moved under the spider’s body. Palps reached over then flexed against the prey once they made contact. After pulling on the prey, the spider was in the retractedlegs posture, having either stabbed or bitten the prey. 8. Rotate over prey. In the raised-body posture, while using its chelicerae to hold onto its prey, a spider stepped and pivoted about, frequently pausing to lower its abdomen and fasten silk lines to the web. The spider eventually lowered its abdomen, released the prey and assumed the raised-body posture again, but left the prey on the web, then turned in circles. Its chelicerae remained positioned more or less over the prey. Initially the spider’s fangs were extended. While rotating, the spider’s palps tended to stay in contact with prey. Spiders usually completed two or three full turns (either clockwise or anticlockwise, no obvious rules regarding the direction adopted), then paused 5–20 s after

Cerveira & Jackson—Pisaurid spiders from Australia and Sri Lanka

125

Fig. 6 Hygropoda dolomedes (adult female) biting house fly.

having released the prey. With larger prey, there might be as many as 10 full turns before the spider paused. Pauses lasted from a few seconds to several minutes, after which spiders usually began feeding, although on rare occasions they rotated again. 9. Apply silk to prey. While rotating over prey, spiders intermittently lowered their abdomens and fastened silk lines to the web or the prey. Fastening lines to the prey was achieved by temporarily moving forward, bringing the spinnerets over the prey. Spinnerets were lowered and silk was fastened to the web or prey, after which the spider backed up and repositioned its chelicerae over the prey. All the while, the spider continued to rotate over the prey. Silk attachment tended to happen 5–10 times per circuit around the prey, with each circuit taking 2–5 s. 10. Feeding. Although spiders were in the raisedbody posture when they began feeding, they usually adopted the rest posture soon afterwards (Fig. 7). This meant that the prey was normally

lying on the web while the spider fed. Spiders held onto the prey with their chelicerae and fed uninterrupted for 1–6 h. Palps usually rested on top of large prey. Otherwise, the spider’s palps were in the usual rest posture for most of the feeding bout. 11. Manipulate with palps. Using their palps, spiders intermittently manipulated the prey by relaxing the grip of the chelicerae around the prey and simultaneously moving the palps against the prey. With larger prey, spiders adopted the raisedbody posture while repositioning (Fig. 5). After repositioning, the spider again tightly pressed its chelicerae against the prey. Manipulating was especially frequent with larger prey. 12. Discard prey remains. Prey remains, held by the spider’s chelicerae, were carried to the edge of the leaf. The spider leaned out and released its grip, letting the remains drop away from the leaf. If unsuccessful (Fig. 8), the spider gripped the prey remains again, leaned out further and dropped the prey over the edge.

126

New Zealand Journal of Zoology, 2002, Vol. 29

Fig. 7 Hygropoda dolomedes (adult female) feeding on house fly. In rest posture. Fly lies on web.

Typical predatory sequences Time elapsing between initial reaction by the spider and prey capture was usually 0.5–15 s. Predatory sequences typically began with spiders in the rest posture in the centre of the web (on the pit, if present). Very small prey usually elicited rapid, simple attacks, often in a single burst covering up to 100 mm to reach the prey in less than 1 s, followed by immediately returning to the centre of the web with the prey. Simple attacks were also made against larger prey, but with the spiders usually failing to hold on. Although hard to discern, spiders appeared usually to stab, instead of bite, larger prey. After attacking larger prey, spiders typically moved partly or all the way off the prey. Subsequent behaviour of the spider depended on prey activity. As long as the prey remained quiescent, the spider tended to remain quiescent, but prey activity brought the spider running back. Often prey were active only intermittently, with spiders approaching in successive short runs of only a few millimetres, pausing in the web each time prey stopped moving. After pausing with prey only a few millimetres away,

spiders tended to make lunging attacks if the prey became active after a short pause. However, should the prey remain quiescent for several minutes, spiders usually returned to the centre of the web even if they had been only a few millimetres from the prey at the time. Additional prey occasionally landed on webs while spiders were still feeding. When this happened, spiders sometimes initiated predatory sequences towards a new prey while carrying a previously captured prey in their chelicerae (Fig. 9). Just before contacting the new prey (typically when 2–3 mm away), the spider dropped the previously-captured prey on the web. A few seconds after capturing the second prey, the spider usually picked up the first prey and now held both prey items in its chelicerae. However, when the first prey was very small, spiders sometimes captured a second prey while still gripping the first. A spider might capture as many as five very small prey in succession, holding all of them by the chelicerae in one bundle while feeding. Spiders usually applied silk to bundles resulting from multiple captures, but only rarely to single prey

Cerveira & Jackson—Pisaurid spiders from Australia and Sri Lanka

127

Fig. 8 Hygropoda dolomedes (adult female) returning to remains of house fly at edge of artificial leaf after unsuccessful attempt to discard it.

items. After silk application, prey bundles sometimes remained attached to the web while the spider fed, but smaller bundles were usually pulled free during feeding. Prey sometimes moved away before contact, and spiders usually ran quickly after these prey, sometimes overtaking and capturing them before they left the web. However, spiders never left their webs while chasing prey. After an unsuccessful chase, spiders usually returned to the centre of the web where they resumed the rest posture. When tested with medium, small, and very small dipterans and homopterans, the most common groups from the field prey records, spiders usually reacted with a simple or a lunging attack, subduing the prey within 0.5–3 s. Large dipterans and homopterans were rarely captured, although they were usually approached, with the spider usually running away immediately after contact. Regardless of size, moths (the third most common group from the field records) provoked simple attacks, but were rarely captured because they usually flew away before the spider reached them.

Small and very small salticid spiders were often captured in sequences comparable to how the pisaurids captured dipterans and homopterans. Pisaurids also rushed toward larger salticids, but the salticid usually fled from the web before contact. On the rare occasions when a pisaurid managed to contact a medium or large salticid, a fight ensued (i.e., the two spiders, often venter to venter and with fangs extended, grappled at each other with their legs), after which the salticid fled from the web and the pisaurid either returned to the centre of the web or moved under the leaf (see below). Ant workers and beetles were common on leaves in the habitats of H. dolomedes and Dendrolycosa sp., but there were no records of feeding on these arthropods. In laboratory tests using small and medium coccinellid beetles and medium ant workers, attack-and-release sequences were common: spiders made a simple or lunging attack, but backed a few millimetres away and became quiescent immediately after touching the beetle or ant. Spiders sometimes approached a beetle once more when movement was resumed, but ran away

128

New Zealand Journal of Zoology, 2002, Vol. 29

Fig. 9 Hygropoda dolomedes (adult female) approaching second house fly while feeding on a previously-captured house fly. On artificial leaf.

almost immediately. After this, the spider left the beetle alone, allowing it to leave the web. Ants were rarely approached more than once. A beetle or an ant worker might walk across a web and come close to a spider, touch it or even walk under it. When this happened, spiders always moved a few millimetres away and often adopted the raisedbody posture. Anti-predator behaviour Predator simulation was achieved by: (1) lightly touching a spider’s legs or body, or its web, with a brush (once or 4–10 times in rapid succession); (2) forcefully striking a spider’s legs or body, or its web, with a brush (once or 4–10 times in rapid succession); (3) enticing a very large salticid spider (more than 2 times the resident spider’s size) onto a spider web. The salticids used were juveniles and adult females of Mopsus mormon, a species that is sympatric with H. dolomedes, is found on leaves, and sometimes preys on H. dolomedes (Jackson 1983).

Spiders usually reacted to being lightly touched once on their legs, body, or web by stepping a few millimetres away. Infrequently, there was no reaction whatsoever. When forcefully struck once on their legs, body, or web, spiders sometimes moved 10 mm or further away. After being struck once, spiders sometimes adopted the retracted-legs posture for a few seconds. Spiders never left their webs completely after being touched or struck once with a brush. When light touching or forceful striking was applied 4–10 times in rapid succession, spiders usually moved off the web. However, leaving the web was usually only partial. Spiders typically flipped around the edge of the leaf then came to rest close to the edge, with one or more legs flexed around the leaf edge so that at least one leg’s tarsus stayed in contact with the web (Fig. 10). Spiders moved completely away from the web only after very persistent striking. Even then the spider usually remained on the underside of the leaf. Using a brush, it was very difficult to persuade the spider to move off the leaf.

Cerveira & Jackson—Pisaurid spiders from Australia and Sri Lanka

129

Fig. 10 Hygropoda dolomedes (adult female) underneath artificial leaf with left legs II and III flexed around edge of leaf and in contact with web.

The resident spider’s reaction to Mopsus mormon was to move away, all the while staying on the web, or to go around the edge of the leaf, all the while keeping one or more leg tarsi on the web. Resident spiders only rarely moved completely away from their webs, and never left the leaves, in tests with M. mormon.

DISCUSSION Besides H. dolomedes and Dendrolycosa sp., there are other pisaurids that build prey-capture webs. Architis nitidopilosa from Central America may be the best known of these. It builds a large 3dimensional web in the vegetation, consisting of a short horizontal funnel which widens at both openings to form two extensive sheets. The outer sides of the sheets bend upward. These webs are typically found between two forked stems, in the umbels or panicles of grasses and herbs, or under large leaves (Nentwig 1985). Details of web shape

depend on the local configuration of attachment possibilities in the vegetation. Juveniles of Pisaura mirabilis from Europe and Pisaurina mira from North America build webs (Lenler-Eriksen 1969; Carico 1985) that resemble those constructed by A. nitidopilosa. Webs built by P. mirabilis juveniles are used for prey capture, but there are no reports of this for P. mira (Carico 1985). Inola amicabilis, I. cracentis, and I. subtilis are Australian pisaurids that build horizontal sheet webs, usually placed one below the other in tiers against tree trunks or rock faces (Davies 1982). Details concerning web structure and web use are not available. Web building is also known in Euprosthenops, Euprosthenopsis, Vuattouxia, and Tetragnophthalma (Gerhart & Kästner 1937; Blandin & Celerier 1981), all of which are African pisaurids, and for Eurychoera and Polyboea from Singapore (Koh 1989). Few details are available for the webs of these genera, but Euprosthenops proximus builds its webs under especially unusual circumstances. Its

130 hammock-shaped web is found in abandoned mammal burrows positioned over the webs of an agelenid spider, Agelena ocellata (Heidger 1988). Although use of prey-capture webs may not be routine in the Pisauridae, a special kind of web, not known to function as a prey-capture device, is well known in two pisaurid genera, Dolomedes and Pisaura. Species in these genera capture prey as hunters, but females spin large “nursery webs” in which they place their eggs shortly before hatching (Bristowe 1941; Carico 1973; D’Andrea 1987). These webs are presumed to serve primarily as shelters for newly-hatched juveniles, because they play no apparent role in prey capture (Bristowe 1941; Forster & Forster 1999), but this is only a tentative conclusion. There have been no detailed studies of how these spiders use their nursery webs. All prey-capture webs known for pisaurids might be described as sheet webs, but the webs of H. dolomedes and Dendrolycosa sp. appear to be exceptionally simple: horizontal sheets are usually positioned on the tops of leaves, without tunnels or retreats. The affinity of H. dolomedes and Dendrolycosa sp. for their webs appears to be extreme, as these spiders rarely leave the web when tested with predator simulations and with actual predators. When they do leave their webs, it is only partially, as the spider usually keeps a leg on the silk even after going under the leaf. Hygropoda dolomedes and Dendrolycosa sp. rest on leaf tops, seemingly exposed to potential predators. However, the spider and its web both tend to be inconspicuous. The spider tends go unnoticed because of its camouflaging markings, and in nature the web resembles little more than a sheen of light reflected off a leaf top. These leaf-top webs may also go unnoticed by many unsuspecting dipterans and other prey that make the mistake of landing where H. dolomedes or Dendrolycosa sp. have positioned their webs. Spider webs are often portrayed as “traps” or “snares”, but these terms appear to have limited applicability to the webs of H. dolomedes and Dendrolycosa sp. Most prey appear to be detained little if at all by the non-sticky silk on these simple sheets. It is probably more appropriate to envisage the webs of H. dolomedes and Dendrolycosa sp. as arenas in which prey capture takes place, and as critical components of the spider’s sensory system (see Witt 1975).

New Zealand Journal of Zoology, 2002, Vol. 29 Movement appears to be a critical prey-attack cue for H. dolomedes and Dendrolycosa sp. From a distance, prey activity elicited accurate and rapid dashes to the prey. Even at close quarters, prey were rarely attacked unless moving, except when spiders were already touching a quiescent prey. Almost all of the prey taken in the field were fast-moving arthropods against which these pisaurids’ rapid movement-provoked attacks are appropriate. Caterpillars, which were rare in the prey records, are perhaps the primary exception. Beetles, although common on leaf tops in tropical rain forests, were absent from the prey records. A significant factor may be that H. dolomedes and Dendrolycosa sp. can not readily pierce the hard cuticle of beetles with their fangs. Ant workers were also noticeably absent from the prey records. Often equipped with poison, stings, and strong mandibles, ants at close quarters can be a formidable challenge for a spider. In laboratory trials, beetles and ants were never captured, but they usually elicited an initial attack followed by release. For these pisaurids, decisions about whether a moving arthropod on a web is acceptable or not as prey appear to be made largely after contact. Dipterans, especially mosquitoes, dominated the prey records for H. dolomedes and Dendrolycosa sp. This has potential applied significance, as mosquitoes are notorious as vectors of malaria, yellow and dengue fever, elephantiasis, and other tropical diseases. Research is needed on the influence of web-building pisaurids on mosquito populations.

ACKNOWLEDGMENTS This work was supported by grants from the Marsden Fund of the Royal Society of New Zealand (UOC512), the National Geographic Society (2330-81) and the United States National Science Foundation (BNS 8617078). Valuable assistance was provided by the Department of Wildlife Conservation in Sri Lanka. Special thanks are due to Roy Bulner (deceased) and Major General M. Madawela in Sri Lanka for assistance with the research. G. B. Edwards provided invaluable taxonomic assistance. Tracey Robinson and Helen Spinks are gratefully acknowledged for their help with preparation of the manuscript. Voucher specimens have been lodged in the Florida Collection of Arthropods (Division of Plant Industry, Gainesville, Florida, United States).

Cerveira & Jackson—Pisaurid spiders from Australia and Sri Lanka REFERENCES Blandin, P.; Celerier, M. L. 1981: Les araignées des savanes de Lamo (Cote d’Ivoire). Ecole Normale Superieure, Publications du Laboratoire de Zoologie 21: 1–586. Bristowe, W. S. 1941: The comity of spiders. Vol. 2. London, Ray Society. Carico, J. E. 1973: The Neartic species of the genus Dolomedes (Araneae: Pisauridae). Bulletin of Museum of Comparative Zoology, Harvard University 144: 435–488. Carico, J. E. 1985: Description and significance of the juvenile web of Pisaura mira. Bulletin of the British Arachnological Society 6: 295–296. D’Andrea, M. 1987: Social behaviour in spiders (Arachnida, Araneae). Monitore Zoologico Italiano. Monografia: Italian Journal of Zoology. Monograph 3: 1–156. Davies, V. T. 1982: Inola nov. gen., a web-building pisaurid (Araneae: Pisauridae) from northern Australia with descriptions of three species. Memoirs of the Queensland Museum 20: 479–487. Foelix, R. F. 1996. Biology of spiders. Cambridge, Massachusetts, Harvard University Press. Forster, R. R.; Forster, L. M. 1999: Spiders of New Zealand and their worldwide kin. Dunedin, New Zealand, University of Otago Press. Gerhart, U.; Kästner, A. 1938: Araneae. In: Kükenthal, W. G.; Krumbach, T. ed. Handbuch der Zoologie. Vol. 3. Berlin, de Gruyter. Pp. 394–656. Griswold, C. E. 1983: Tapinillis longipes (Taizanowski), a web-building lynx spider from the American tropics (Araneae: Oxyopidae). Journal of Natural History 17: 979–985. Heidger, C. 1988: Ecology of spiders inhabiting abandoned mammal burrows in South African savanna. Oecologia (Heidelberg) 76: 303–306. Jackson, R. R. 1983: The biology of Mopsus mormon, a jumping spider (Araneae: Salticidae) from Queensland: intraspecific interactions. Australian Journal of Zoology 31: 39–53. Jackson, R. R. 1985: A web-building jumping spider. Scientific American 253: 102–115. Jackson, R. R. 1986: Web building, predatory versatility, and the evolution of the Salticidae. In: W. A. Shear. ed. Spiders: webs, behavior, and evolution. Stanford, California, Stanford University Press. Pp. 232–268.

131

Jackson, R. R.; Hallas, S. E. A. 1986: Comparative biology of Portia africana, P. albimana, P. fimbriata, P. labiata, and P. schultzi, araneophagic webbuilding jumping spiders (Araneae: Salticidae): utilisation of silk, predatory versatility, and intraspecific interactions. New Zealand Journal of Zoology 13: 423–489. Jackson, R. R.; Taylor, P. W.; McGill, A. S.; Pollard, S. D. 1995: The web and prey-capture behaviour of Diaea sp., a crab spider (Thomisidae) from New Zealand. Records of the Western Australian Museum, Supplement No. 52: 33–37. Jarman, E. A. R.; Jackson, R. R. 1986: The biology of Taieria erebus (Araneae, Gnaphosidae), an araneophagic spider from New Zealand: silk utilisation and predatory versatility. New Zealand Journal of Zoology 13: 521–541. Job, W. 1974: Beiträge für Biologie der fangnetzbouenden Wolfspinne Aulonia albimana (Walckenaer 1805) (Arachnida, Araneae, Lycosidae, Hippasinae). Zoologische Jahrbucher Abteilung für Systematik, Ökologie und Geographie der Tier 101: 560–608. Koh, J. 1989: A quide to common Singapore spiders. Singapore, Singapore Science Centre. Lenler-Eriksen 1969: The hunting web of the young spider Pisaurina mirabilis. Journal of Zoology London 157: 391–398. Li, D.; Jackson, R. R.; Cutler, B. 1996: Prey-capture techniques and prey preferences of Habrocestum pulex (Hentz), an ant-eating jumping spider (Araneae, Salticidae) from North America. Journal of Zoology London 240: 551–562. Nentwig, W. 1985: Architis nitidopilosa, a neotropical pisaurid with a permanent catching web (Araneae, Pisauridae). Bulletin of the British Arachnological Society 6: 297–303. Shear, W. A. 1994: Untangling the evolution of the web. American Scientist 82: 256–266. Stowe, M. K. 1986: Prey specialisation in the Araneidae. In: Shear, W. A. ed. Spiders: webs, behavior, and evolution. Stanford University Press, Stanford, California. Pp. 101–131. Witt, P. N. 1975: The web as a means of communication. Biosciences Communication 1: 7–23.

132

New Zealand Journal of Zoology, 2002, Vol. 29

Appendix 1 Prey records for Hygropoda sp. in nature. Number of records given in parentheses. For each record, H. dolomedes sp. in act of feeding when found. A. Araneae 1. Anyphaenidae (cursorial spider): medium (1) 2. Salticidae (jumping spider): Cyteae alburna small (1), Tau ala lepidus small (1), medium (1) B. Diptera 1. Chironomidae (midge): very small (6) 2. Culicidae (mosquito): very small (22), small (5) 3. Diopsidae (stalked-eye fly): small (2), medium (1) 4. Dolichopodidae (fly): small (2), medium (1) 5. Micropezidae (fly): medium (2) 6. Tabanidae (horse fly): large (1) 7. Tephritidae (fruit fly): medium (1) 8. Unknown: small (6), medium (2) C. Ephemeroptera 1. Unknown mayfly: small (3) D. Hemiptera 1. Cicadellidae (homopteran): small (2) 2. Dictypharidae (homopteran): large (1) 3. Flatidae (homopteran): medium (1) 4. Lygaeidae (heteropteran): medium (1) 5. Ricaniidae (homopteran): medium (1) 6. Tingidae (homopteran): small (1) 7. Unknown homopteran: very small (2), small (2) E. Isoptera 1. Alate termite: small (1), medium (2) F. Lepidoptera 1. Yponomeutidae (moth): small (2) 2. Unknown caterpillar: small (1), medium (1) 3. Unknown moth: medium (2), large (2) G. Mantodea 1. Unknown mantis: medium (2) H. Psocoptera 1. Unknown: very small (3)

Cerveira & Jackson—Pisaurid spiders from Australia and Sri Lanka

133

Appendix 2 Prey records for Dendrolycosa sp. in nature. Number of records given in parentheses. For each record, Dendrolycosa sp. in act of feeding when found. A. Araneae 1. 2. B. Blattodea 1. C. Diptera 1. 2. 3. 4. D. Ephemeroptera 1. E. Hemiptera 1. 2. 3. 4. 5. F. Hymenoptera 1. G. Isoptera 1. H. Lepidoptera 1. I. Mecoptera 1. J. Unknown insect 1.

Salticidae (jumping spider): small (1) Unknown hunting spider: small (1) Unknown cockroach: small (1) Chironomidae (midge): very small (2) Culicidae (mosquito): very small (16), small (4) Dolichopodidae (fly): small (2) Unknown: very small (15), small (3), medium (2) Unknown mayfly: very small (3), small (1) Cicadellidae (homopteran): very small (2), small (2) Flatidae (homopteran): small (1) Miridae (heteropteran): medium (1) Ricaniidae (homopteran): small (1) Unknown homopteran: very small (5), small (2), medium (1) Formicidae (alate ant): small (2) Alate termite: small (4), medium (2) Unknown moth: small (1), medium (1) Unknown: medium (1) very small (4)

134

New Zealand Journal of Zoology, 2002, Vol. 29