ZOOLOGY Zoology 109 (2006) 339–345

ARTICLE IN PRESS ZOOLOGY Zoology 109 (2006) 339–345 www.elsevier.de/zool

The effects of capture spiral composition and orb-web orientation on prey interception Brent D. Opella,Ã, Jason E. Bondb, Daniel A. Warnerc aDepartment of Biology, Virginia Tech, Blacksburg, VA 24061, USA bDepartment of Biology, East Carolina University, Greenville, NC 27858, USA cSchool of Biological Sciences, University of Sydney, Sydney, NSW 2006, Australia

Received 21 December 2005; received in revised form 15 March 2006; accepted 10 April 2006

Abstract

Cribellar prey capture threads found in primitive, horizontal orb-webs reflect more light, including ultraviolet wavelengths, than viscous threads found in more derived, vertical orb-webs. Low web visibility and vertical orientation are each thought to increase prey interception and may represent key innovations that contributed to the greater diversity of modern, araneoid orb-weaving spiders. This study compares prey interception rates of cribellate orb-webs constructed by Uloborus glomosus (Uloboridae) with viscous orb-webs constructed by Leucauge venusta (Tetragnathidae) and Micrathena gracilis (Araneidae). We placed sectors of cribellar and viscous threads side by side in frames that were oriented either horizontally or vertically. The webs of both U. glomosus and L. venusta intercepted more prey when vertically oriented. In each orientation L. venusta webs intercepted more insects than did U. glomosus. Although this is consistent with the greater visibility of cribellar threads, the more closely spaced capture spirals of L. venusta may have contributed to this difference. Micrathena gracilis webs intercepted more prey than did U. glomosus webs, although web orientation did not affect the performance of this araneoid species. The stickier and more closely spaced capture spirals of M. gracilis may have enhanced the interception rates of this species and accounted for the greater number of smaller dipterans retained in its webs. The tendency for these slow, weak flight insects to be blown into both horizontal and vertical webs may account for similar interception rates of horizontal and vertical M. gracilis webs. These observations support the enhanced prey interception of vertically oriented orb-webs, but offer only qualified support for the contributions of lower visibility viscous capture threads. r 2006 Elsevier GmbH. All rights reserved.

Keywords: Araneoidea; Cribellar capture thread; Deinopoidea orb-web evolution; Prey capture; Viscous capture thread

Introduction (Eberhard 1986), web orientation (Eberhard 1989, 1990), web invisibility or attractiveness to insects Many factors affect an orb-web’s ability to intercept (Blackledge 1998 a, b; Blackledge and Wenzel 1999; prey, including insect availability (Reichert and Cady Craig and Bernard 1990; Craig et al. 1994), the presence 1983; Wise and Barata 1983; Craig 2003), web area of a stabilimentum at the center of the web (Herberstein et al. 2000), and the visibility of a spider positioned at ÃCorresponding author. the web’s hub (Blackledge 1998a, b; Blackledge and E-mail address: [email protected] (B.D. Opell). Wenzel 1999; Craig and Ebert 1994; Zschokke 2002).

Here we examine how capture spiral composition and capture thread spacing and stickiness (Chacoń and web orientation affect the number of prey that a web Eberhard 1980). When the orientation of araneoid orb- intercepts. We focus on these features because they were webs was experimentally changed, vertical orb-webs associated with the origin of modern araneoid orb- retained prey longer than did horizontal orb-webs weaving spiders and have been considered key innova- (Eberhard 1989). Thus, vertically oriented orb-webs of tions that contribute to the diversity of these spiders araneoid spiders that incorporate viscous capture (Craig et al. 1994; Bond and Opell 1998). threads appear to represent an optimal combination of All orb-weaving spiders belong to the Orbiculariae features. However, the combined effects of these clade, a group formed of the Deinopoidea and features have not been examined. Araneoidea subclades (Coddington and Levi 1991; In this study we compared the prey interception of Griswold et al. 1998). There are only about 300 species araneoid and deinopoid orb-webs by placing large of deinopoid orb-weavers, as compared to over 4000 sectors of cribellate and viscous orb-webs side by side orb-weaving araneoid species (Griswold et al. 1998; in both horizontal and vertical orientations. Spiders Platnick 2000). Phylogenetic analyses suggest that the were not present on these web sectors and we equated origin of araneoids was associated with a transition the incidence of web damage (plus the few insects that from cribellar capture thread to wet, viscous capture remained in the web) with prey interception. Although thread (Coddington and Levi 1991; Griswold et al. 1998) our study does not attempt to compare the prey capture and with a transition from horizontal to vertical orb- or profitability of these orb-webs, it does allow us to test web orientation (Bond and Opell 1998). However, some the hypothesis that vertical orb-web orientation and araneoids, such as Leucauge venusta (Walckenaer, 1841) reduced orb-web visibility individually and collectively that is included in this study, have secondarily assumed increase prey interception. According to this hypothesis, a horizontal web orientation (Bond and Opell 1998). horizontal cribellate webs should intercept the fewest Cribellar threads of orb-weavers are formed by a pair prey and vertical viscous webs the most prey, with of supporting fibers overlain by a dense mat of horizontal viscous webs and vertical cribellate webs thousands of dry, coiled fibrils (Eberhard and Pereira intercepting intermediate numbers of prey. 1993; Opell 1999b; Peters 1986, 1992). These threads snag on the surface irregularities of insects and adhere to smooth surfaces by van der Waals and hygroscopic forces (Hawthorn and Opell 2002, 2003). In contrast, the Materials and methods supporting fibers of viscous capture threads are covered by a complex aqueous solution that coalesces into Species studied droplets that derive their stickiness from internal glycoprotein nodules (Vollrath et al. 1990; Townley et We studied the deinopoid species Uloborus glomosus al. 1991; Vollrath and Tillinghast 1991; Vollrath, 1992; (Walckenaer, 1842), family Uloboridae and the araneoid Tillinghast et al. 1993). species Leucauge venusta, family Tetragnathidae and Viscous capture threads have several apparent func- Micrathena gracilis (Walckenaer, 1805), family Aranei- tional advantages over cribellar threads: (1) they achieve dae. The first two species construct horizontal orb-webs greater stickiness relative to their volume (Opell 1997, in low vegetation in both shaded areas within forests 1998), allowing araneoid orb-weavers to construct orb- and in more exposed areas at the edges of forests. The webs that have a greater stickiness per capture area than orb-webs of M. gracilis are typically suspended 1–1.5 m those of deinopoid orb-weavers (Opell 1999a), (2) they above the floor of moist forests in orientations that vary cost less to produce (Zschokke and Vollrath 1995), and from nearly horizontal to vertical. The spectral proper- (3) their droplets are translucent and reflect less light, ties of the capture threads of these species and spacing of including ultraviolet light, than do cribellar threads their capture spirals are summarized in Table 1. (Craig et al. 1994; Zschokke 2002). As many insects Some spiders decorate the hubs of their webs with detect UV light (Briscoe and Chittka 2001) this low bands, crosses, or spirals of non-sticky threads called overall visibility combines with low UV light reflectance stabilimenta. The functions of stabilimenta and their to make araneoid orb-webs less visible to insects than effect on web visibility and prey interception have been deinopoid orb-webs (Craig and Bernard 1990; Craig et controversial (Herberstein et al. 2000). However, in the al. 1994; Zschokke 2002) and appears to increase the context of this study, we believe that, like cribellar rate of prey interception by araneoid orb-webs (Craig capture thread, the presence of stabilimenta would and Freeman 1991; Blackledge and Wenzel 1999). increase the web’s visibility and reduce the number of Web orientation also affects prey interception. Verti- insects that it intercepted. Photographs of L. venusta cally oriented artificial orb-webs captured more insects and U. glomosus webs constructed in the laboratory and and retained them for longer periods of time than did of M. gracilis webs constructed in the field (part of the horizontally oriented orb-web models having the same data base for Opell 1997, 1999a) show that L. venusta ARTICLE IN PRESS B.D. Opell et al. / Zoology 109 (2006) 339–345 341

Table 1. Capture thread and web features of the three study species (mean71 standard error, sample size)

Uloborus glomosus Leucauge venusta Micrathena gracilis

Spider mass (mg) 6.870.3, 27 20.671.7, 18 84.375.0, 21 Web orientation Horizontal Horizontal Variable Capture thread type Cribellar Viscous Viscous Capture thread stickiness (mN/mm) 15.571.1, 30 19.371.3, 25 26.471.6, 17 Relative UV reﬂectance High Low None Spectral properties UV, blue green Flat Cutoff at 410 nm Axial line diameter (mm) 0.3270.03, 17 1.0370.05, 19 3.4470.25, 19 Puff/droplet width (mm) 173.573.7, 30 10.170.4, 26 20.170.7, 20 Puffs/droplets per mm 12.770.3, 30 41.3,72.0, 26 19.970.8, 20 Spiral spacing (mm) 5.770.3, 27 2.470.2, 18 1.370.0, 21

Numerical values of thread features are from Opell (1998, 1999a, unpublished data) and Opell and Bond (2001). Spectral features of thread are from Craig et al. (1994). and M. gracilis cut out the hub of their webs and that grass and fringed on its circumference by a 1–3 m wide neither produces a stabilimentum. In contrast, 14 of the weedy area. At this site, we placed web samples 4–5 m 30 U. glomosus webs in these earlier studies had linear inside the forest edge. stabilimenta. Although our experimental design did not address the presence of stabilimenta in U. glomosus Collecting web sectors webs, these features would further increase the visibility of these already more highly visible orb-webs. Adult female L. venusta and U. glomosus were collected from local populations and placed individually Estimating web visibility in 25 37 cm plastic boxes. Wooden dowel rods attached to the perimeter of each box provided Table 1 compares the three species’ thread and web attachment surfaces for the framework lines of webs. features. U. glomosus produces wider capture threads The width of these boxes exceeded the diameters of the that reflect more UV light, suggesting that these webs webs of these two species that we observed in the field are more visible to insects. Threads produced by L. (Opell 1999a). Consequently, we were able to obtain venusta and M. gracilis have narrower droplets and normal, undamaged webs. We collected a large portion reflect less UV light, indicating that the webs of these of the web’s center that included the hub on the 4 mm two species are less visible to insects. However, it is not wide rim of a 173 mm diameter ring. This wooden ring clear which of these two webs is more visible. The was painted flat black to reduce its visibility and one of threads of L. venusta have narrower but more closely its rims was covered by double-sided tape to securely spaced droplets and appear to reflect slightly more UV hold the orb-web sector. A small brush was used to light than those of M. gracilis. The capture spirals of M. chase a spider from its web before we collected its web gracilis are more closely spaced than those of L. venusta, on the ring. We used the same procedures to collect which probably increases web visibility. sectors of orb-webs constructed by adult female M. gracilis in the field. Study sites U. glomosus females mature before most L. venusta and all M. gracilis. Therefore, we collected many of the We deployed web samples at two sites located on the U. glomosus webs that were used in this study prior to edge of the Virginia Tech campus (Blacksburg, Mon- the field trials and stored them in a dust-free, low tgomery County, VA), a field site and a forest site. Both humidity laboratory environment. U. glomosus webs sites were away from human traffic and were chosen to were stored for 1–30 days before being deployed. As the provide different light environments. The field site was width of the uloborid cribellar thread decreased by only on the western edge of a narrow band of forest that 4% after 14 months of storage (Opell 1993), we believe bordered a continuously flowing stream. Here we placed that the properties of these webs were not affected by web samples in an overgrown field 2–3 m from the forest this short storage period. In contrast, the viscous edge. The forest canopy did not extend over the web capture threads of L. venusta and M. gracilis webs do samples. We cleared a 1 m wide area on each side of the change with age as their viscous droplets dry and a deployed web samples to reduce the likelihood of insects meaningful study of these webs required us to use newly jumping into the webs from surrounding vegetation. The spun webs. Therefore, each morning before deploying forest site was a 0.4 ha, roughly circular deciduous forest web samples in the field, we collected L. venusta webs remnant that was surrounded on all sides by mowed that had been produced in the laboratory during the ARTICLE IN PRESS 342 B.D. Opell et al. / Zoology 109 (2006) 339–345 previous night or M. gracilis webs that had been number of prey intercepts during the first two intervals, collected from the field soon after they were constructed that is, during the 08:00–14:00 h period. in mid to late morning of the previous day and stored Comparison of U. glomosus and M. gracilis web overnight in an environmental chamber at 80% relative performance was similar to that described above. humidity to maintain viscous droplet volume. However, frames containing these paired webs were deployed only in the field site on 12, 15, 22, 28, 30 August 1998 and 1 September 1998. As in the previous Deploying web sectors comparison, extensive web damage limited the useful daily observation period to 08:00–14:00 h. Web damage We secured deinopoid and araneoid web rings side by of 50% or greater also compromised the paired side in a rectangular frame that was also painted flat experimental design for both sites on 22 August and black to reduce its visibility (Fig. 1) and attached these for one of the sites on 28 August, forcing us to exclude frames to wooden supports. One frame was attached these observations. horizontally so that the plane of its webs was 90 cm above the ground. The other frame was attached vertically, so that the centers of the orb-webs that it Scoring insect interception contained were 90 cm above the ground. A distance of 1 m separated these two frames. Another set of We counted each newly detected area of web damage, horizontal and vertical frames was deployed 4 m away regardless of its size, and each newly ensnared insect as from the first set, providing a total of eight web samples one incident of prey interception. We did not attempt to for each day’s observation. correlate the size of the damaged area with insect size, One series of observations compared the performance because insects belonging to different taxa struggle of U. glomosus and L. venusta webs. We placed web differently when caught in webs (Nentwig 1982). frames in the field site on 3, 8, and 9 July and in the Additionally, when an insect struggles free of a forest site on 11, 15, and 16 July 1998. We deployed the horizontal web it simply falls out, but when it struggles frames at 08:00 h and examined each web at 3-h from a vertical web it is likely to encounter additional intervals, at which time we sketched the damaged areas capture threads and generate further damage before resulting from insect impact and struggle. If an insect exiting the web. We divided the total number of remained in the web we also noted its order, and interceptions per web by the 6 h of the observation position. Ensnared insects (in most cases, small dipter- period to determine prey interceptions per hour. ans) were not removed from the web. At the end of the As the diameter of our web collecting rings was second observation period none of the webs had smaller than that of the capture area of the three species sustained damage that exceeded 50% of their capture webs, capture threads extended to the edge of the rings. area. However, at the end of the third period many webs However, capture threads were absent from the web’s had sustained damage to 50% or more of their area, hub and surrounding free zone. Thus, differences in the making it difficult to evaluate prey interceptions and to sizes of these central regions had the potential to affect compare their performance with their paired web. the sensitivity of our assay: webs with smaller central Although we observed only one instance of damage regions will tend to register more prey strikes than webs from a falling leaf, excluding webs with 50% or more with larger central regions. To account for this, we damage reduced the likelihood of these events affecting photographed webs of each species and from these our observations. Therefore, we analyzed only the photographs measured the diameters of the webs’ central regions. From this, we determined the mean capture area of each species’ web. For L. venusta this was 1311 mm2 (N ¼ 15, SD ¼ 318), for M. gracilis 2080 mm2 (N ¼ 26, SD ¼ 236), and for U. glomosus 1012 mm2 (N ¼ 17, SD ¼ 551). The area encompassed by the collecting ring was 23,506 mm2. Therefore, for L. venusta the average area containing capture threads was 22,199 mm2, for M. gracilis 21,426 mm2, and for U. glomosus 22,494 mm2. As the webs of U. glomosus had the greatest capture area, we multiplied the prey interception values of L. venusta and M. gracilis by 1.013 and 1.050 (the values of U. glomosus capture area Fig. 1. A frame containing rings that hold orb-web sectors divided by their respective capture areas) to adjust for collected from U. glomosus and either L. venusta or M. gracilis their smaller capture areas and resulting lower potential webs. to register damage from insect interception. ARTICLE IN PRESS B.D. Opell et al. / Zoology 109 (2006) 339–345 343

Statistical analysis interactions of these transformed values (F ¼ 4.80, P ¼ 0.0056) showed that the site variable did not We used the S.A.S. statistical package (S.A.S. contribute to this model (P ¼ 0.7005), but that the Institute Inc., Cary, NC) to analyze data. We examined species and web orientation variables did (P ¼ 0.0106 the Shapiro-Wilk-W-Statistics of mean values to deter- and 0.0107, respectively). Therefore, we combined data mine if they were normally distributed (P40.05). If they from the field and forest sites in the subsequent analysis. were not, we transformed data as described under each Prey interception values were normally distributed for section. each of the four resulting species-web orientation categories. A two factor ANOVA with interaction (F ¼ 3.84, P ¼ 0.0159) showed that species and web Results orientation each explained differences in prey interception values (Fig. 2). Species and web orientation each contributed to this model (P ¼ 0.0280 and 0.0167, Total prey intercepted respectively), although the interaction of species and orientation did not (P ¼ 0.6833). Our comparisons of the performance of the three species’ webs described in the following paragraphs are based on 812 incidents of prey interception. Webs Uloborus glomosus and Micrathena gracilis retained a total of 164 insects during the 9 h that they were in the field. This total includes webs that were later Prey interception values for each of the four treat- excluded from the analysis because they suffered more ments (Fig. 3) were normally distributed except for the than 50% damage or as a result of compromised paired horizontal webs of U. glomosus, which did not become design. These retained insects included 150 small normal when log transformed. A two factor ANOVA with interaction was not significant (F ¼ 2.33, dipterans (body length p3 mm), three larger dipterans, two hemipterans, six hymenopterans, one coleopteran P ¼ 0.0925). Although species contributed to this model larva, and two coleopteran adults. The taxonomic (P ¼ 0.0374), neither orientation nor the interaction of diversity of these insects was low and only small species and orientation did (P ¼ 0.3328 and 0.2590, dipterans were present in sufficient numbers to permit respectively). an analysis: U. glomosus horizontal 9, vertical 12; L. venusta horizontal 6, vertical 14; M. gracilis horizontal 65, vertical 44). An ANOVA showed no effect of web Discussion orientation (F ¼ 0.02, P ¼ 0.8859) on dipteran number, but spider species affected dipteran number (F ¼ 15.24, Our results support the hypothesized complementary P ¼ 0.0268). A Ryan-Einot-Gabriel-Welsch multiple benefits of viscous capture threads and vertical orb-web range test (a ¼ 0.05) ranked the number of dipterans orientation. However, as this is the first attempt at a in M. gracilis webs above those in webs of the other two side-by-side comparison of the performance of webs, species. Both prey interception and prey retention our preliminary study bears repeating with other species. contribute to the presence of dipterans in webs. There- fore, it is possible that the greater stickiness of M. 2.5 gracilis capture threads (Table 1) is a key factor in explaining the higher number of dipterans retained in 2.0 the webs of this species. It is also possible that the later deployment of the M. gracilis webs may have exposed them to higher densities of dipterans. These factors and 1.5 the higher contribution of retained dipterans to the prey interception values of M. gracilis suggest caution in 1.0 comparing the performance of M. gracilis webs with webs of the other two species. Interceptions per Hour 0.5

Uloborus glomosus and Leucauge venusta 0.0 Uloborus Leucauge Uloborus Leucauge Prey interception values for the eight treatments (U. Horizontal Orientation Vertical Orientation glomosus horizontal webs, L. venusta horizontal webs, Fig. 2. Mean prey interception values for U. glomosus and L. U. glomosus vertical webs, and L. venusta vertical webs venusta orb-web sectors placed in horizontal and vertical for each of the two study sites) were natural log orientations. The sample size in each treatment was 12; error transformed to make them normal. An ANOVA with bars represent 71 standard deviation. ARTICLE IN PRESS 344 B.D. Opell et al. / Zoology 109 (2006) 339–345

2.5 The same study that associates a transition to vertical orb-web orientation with the origin of the Araneoidea (Bond and Opell 1998), also documents that some 2.0 araneoids, such as L. venusta, have reacquired a horizontal orb-web orientation. Thus, there must be 1.5 some combination of factors, such as microhabitat, spider size, and the strength and architecture of the web that favors horizontal orientation, despite the advan- 1.0 tages of vertical orientation suggested by this study and other studies. It may be that smaller species which Interceptions per Hour 0.5 construct horizontal orb-webs can more easily ﬁnd physical spaces in which to place their webs or that horizontal orb-webs are less taut and oscillate in air 0.0 currents, thereby intercepting weakly ﬂying insects Uloborus Micrathena Uloborus Micrathena (Craig 1987a, b). These factors could explain why orb- Horizontal Orientation Vertical Orientation weaving uloborids have persisted and why a few Fig. 3. Mean prey interception values for U. glomosus and M. araneoids have secondarily acquired vertical web gracilis orb-web sectors placed in horizontal and vertical orientation. orientations. The sample size in each treatment was 11; error bars represent 71 standard deviation.

In both horizontal and vertical orientation, viscous orb- Acknowledgements webs of L. venusta intercepted more prey than did the cribellate orb-webs of U. glomosus. Regardless of their Kelly Rausch helped tabulate data. This material is capture thread composition, vertical webs intercepted based on work supported by National Science Founda- more prey than did horizontal webs. This comparison tion Grant IBN-9417803. appears to suggest that the benefits of producing viscous capture threads and of constructing vertical orb-webs are roughly equal and that these benefits are additive. However, as the spiral spacing of U. glomosus webs is References greater than that of L. venusta webs, we cannot rule out the possibility that this difference favored the prey Blackledge, T.A., 1998a. Signal conflict in spider webs driven interception of L. venusta. Capture thread composition by predators and prey. Proc. Roy. Soc. Lond. B 265, and web orientation may influence prey interception in 1991–1996. different ways. The sun’s angle may also alter a web’s Blackledge, T.A., 1998b. Stabilimentum variation and fora- visibility. Vertical and horizontal webs may intercept ging success in Argiope aurantia and Argiope trifasciata different guilds of insects or insects with different flight (Araneae, Araneidae). J. Zool. 246, 21–27. characteristics (Zschokke and Vollrath 2000). However, Blackledge, T.A., Wenzel, J.W., 1999. Do stabilimenta in orb in each orientation, the lower web visibility would webs attract prey or defend spiders? Behav. Ecol. 10, 372–376. increase the incidence of prey interception. Bond, J.E., Opell, B.D., 1998. Testing adaptive radiation and The webs of M. gracilis outperform those of U. key innovation hypotheses in spiders. Evolution 52, glomosus in both horizontal and vertical orientation. 403–414. Although this is consistent with the greater visibility of Briscoe, A.D., Chittka, L., 2001. The evolution of color vision the cribellar capture threads, we cannot rule out the in insects. Annu. Rev. Entomol. 46, 471–510. possibility that the more closely spaced capture spirals Chacoń, P., Eberhard, W.G., 1980. Factors affecting numbers of M. gracilis intercepted more insects and that the and kinds of prey caught in artificial spider webs with stickier capture threads of this species retained more considerations of how orb-webs trap prey. Bull. Br. insects, particularly small dipterans. U. glomosus webs Arachnol. Soc. 5, 29–38. performed as they had in the previous comparison, with Coddington, J.A., Levi, H.W., 1991. Systematics and evolu- vertical webs exhibiting interception rates twice those of tion of spiders (Araneae). Annu. Rev. Ecol. Syst. 22, 565–592. horizontal webs. However, the interception rates of Craig, C.L., 1987a. The significance of spider size to the vertical and horizontal M. gracilis webs did not differ. diversification of spider-web architectures and spider As dipterans accounted for a large number of the prey reproductive modes. Am. Nat. 129, 47–68. intercepted by these webs, it may be that wind currents Craig, C.L., 1987b. The ecological and evolutionary inter- had a greater effect on these small, slow flying insects dependence between web architecture and web silk spun by and blew them into webs regardless of their orientations. orb web weaving spiders. Biol. J. Linn. Soc. 30, 135–162. ARTICLE IN PRESS B.D. Opell et al. / Zoology 109 (2006) 339–345 345

Craig, C.L., 2003. Spider Webs and Silk: Tracing Evolution Opell, B.D., 1997. The material cost and stickiness of capture from Molecules to Genes to Phenotypes. Oxford University threads and the evolution of orb-weaving spiders. Biol. J. Press, New York. Linn. Soc. 62, 443–458. Craig, C.L., Bernard, G.D., 1990. Insect attraction to Opell, B.D., 1998. Economics of spider orb-webs: the benefits ultraviolet-reflecting spider webs and web decorations. of producing adhesive capture thread and of recycling silk. Ecology 71, 616–624. Funct. Ecol. 12, 613–624. Craig, C.L., Ebert, K., 1994. Color and pattern in predator–- Opell, B.D., 1999a. Redesigning spider webs: stickiness, prey interactions: the bright body colors and patterns of capture area, and the evolution of modern orb-webs. Evol. tropical orb-weaving spiders attract flower-seeking prey. Ecol. Res. 1, 503–516. Funct. Ecol. 8, 616–620. Opell, B.D., 1999b. Changes in spinning anatomy and thread Craig, C.L., Freeman, C.R., 1991. Effects of predator visibility stickiness associated with the origin of orb-weaving spiders. on prey encounter. A case study on aerial web weaving Biol. J. Linn. Soc. 68, 593–612. spiders. Behav. Ecol. Sociobiol. 29, 249–254. Opell, B.D., Bond, J.E., 2001. Changes in the mechanical Craig, C.L., Bernard, G.D., Coddington, J.A., 1994. Evolu- properties of capture threads and the evolution of modern tionary shifts in the spectra properties of spider silks. orb-weaving spiders. Evol. Ecol. Res. 3, 567–581. Evolution 48, 287–296. Peters, H.M., 1986. Fine structure and function of capture Eberhard, W.G., 1986. Effects of orb-web geometry on prey threads. In: Nentwig, W. (Ed.), Ecophysiology of Spiders. interception and retention. In: Shear, W.A. (Ed.), Spiders: Springer, New York, pp. 187–202. Webs, Behavior, and Evolution. Stanford University Press, Peters, H.M., 1992. On the spinning apparatus and structure Stanford, pp. 70–100. of the capture threads of Deinopis subrufus (Araneae, Eberhard, W.G., 1989. Effects of orb-web orientation and Deinopidae). Zoomorphology 112, 27–37. spider size on prey retention. Bull. Br. Arachnol. Soc. 8, Platnick, N.I., 2000. Estimating spider numbers. Am. Ara- 45–48. chnol. 61, 8–9. Eberhard, W.G., 1990. Function and phylogeny of spider Reichert, S.E., Cady, A.B., 1983. Patterns of resource use and webs. Annu. Rev. Ecol. Syst. 21, 341–372. tests for competitive release in a spider community. Ecology Eberhard, W.G., Pereira, F., 1993. Ultrastructure of 64, 899–913. cribellate silk of nine species in eight families and possible Tillinghast, E.K., Townley, M.A., Wight, T.N., Uhlenbruck, taxonomic implications (Araneae: Amaurobiidae, G., Janssen, E., 1993. The adhesive glycoprotein of the orb Deinopidae, Desidae, Dictynidae, Filistatidae, Hypochili- web of Argiope aurantia (Araneae, Araneidae). Mater. Res. dae, Stiphidiidae, Tengellidae). J. Arachnol. 21, Soc. Symp. Proc. 292, 9–23. 161–174. Townley, M.A., Bernstein, D.T., Gallanger, K.S., Tillinghast, Griswold, C.E., Coddington, J.A., Hormiga, G., Scharff, N., E.K., 1991. Comparative study of orb-web hydroscopicity 1998. Phylogeny of the orb-web building spiders (Araneae, and adhesive spiral composition in three areneid spiders. J. Orbiculariae: Deinopoidea, Araneoidea). Zool. J. Linn. Exp. Zool. 259, 154–165. Soc. 123, 1–99. Vollrath, F., 1992. Spider webs and silks. Sci. Am. 266, 70–76. Hawthorn, A.C., Opell, B.D., 2002. Evolution of adhesive Vollrath, F., Tillinghast, E.K., 1991. Glycoprotein glue beneath a mechanisms in spider cribellar prey capture threads: spider web’s aqueous coat. Naturwissenschaften 78, 557–559. evidence for Van der Waals and hygroscopic forces. Biol. Vollrath, F., Fairbrother, W.J., Williams, R.J.P., Tillinghast, J. Linn. Soc. 77, 1–8. E.K., Bernstein, D.T., Gallagher, K.S., Townley, M.A., Hawthorn, A.C., Opell, B.D., 2003. Van der Waals and 1990. Compounds in the droplets of the orb spider’s viscid hygroscopic forces of adhesion generated by spider capture spiral. Nature 345, 526–528. threads. J. Exp. Biol. 206, 3905–3911. Wise, D.H., Barata, J.L., 1983. Prey of two syntopic spiders Herberstein, M.E., Craig, C.L., Coddington, J.A., Elgar, with different web structures. J. Arachnol. 11, 271–281. M.A., 2000. The functional significance of silk decorations Zschokke, S., 2002. Ultraviolet reflectance of spiders and their of orb-web spiders: a critical review of the empirical webs. J. Arachnol. 30, 246–254. evidence. Biol. Rev. 75, 649–669. Zschokke, S., Vollrath, F., 1995. Unfreezing the behaviour of Nentwig, W., 1982. Why do only certain insects escape from a two orb spiders. Physiol. Behav. 58, 1167–1173. spider’s web? Oecologia 53, 412–417. Zschokke, S., Vollrath, F., 2000. Planarity and size of orb- Opell, B.D., 1993. What forces are responsible for the webs built by Araneus diadematus (Araneae: Araneidae) stickiness of spider cribellar threads? J. Exp. Zool. 265, under natural and experimental conditions. Ekolo´gia 19 469–476. (Suppl. 3), 307–318.