2005. The Journal of Arachnology 33:1±6

BEHAVIOR OF WEB-INVADING ARGENTATUS () IN APPENSA (ARANEIDAE) HOST WEBS IN

Alexander M. Kerr1: Marine Laboratory, University of Guam, Mangilao GU 96923 USA

ABSTRACT. Most Argyrodes live in the webs of other spiders, stealing food from the host, scavenging small prey from the web or killing and eating the host. I observed the behavior of A. argentatus from Guam, where it is a frequent inhabitant of the large orb webs of Argiope appensa. I examined the pro- portion of time spent in different activities, whether behavior differed between the sexes and if population density of Argyrodes on a host web affects Argyrodes behavior. Argyrodes spent 55% of the time hanging immobile and inverted in the support strands at the webs' margin. This was signi®cantly more time than that spent in stationary activity, forward movement at the web's margin, feeding, foraging on the sticky spiral or in aggressive interaction. Females foraged signi®cantly more often than did males, though the sexes spent about the same amount of time feeding and in other activities. Females also engaged in more bouts of feeding and 21% of these bouts were at prey bundles prepared by the host. In contrast, males invariably foraged for small insects unnoticed by the host. Keywords: Kleptoparasite, Mariana Islands, Micronesia, Araneae

Most members of the large, cosmopolitan 1880; Exline & Levi 1962; photo in Koh genus of Argyrodes Simon 1864 live in the 2000). General observations of this species webs of other spiders. They feed on small in- have been made on host webs of Argiope ar- sects that have gone unnoticed by the host gentata (Fabricius 1775) in Panama (Robin- (Whitehouse 1986), prey stolen from the host son & Olazarri 1971) and Nephila maculata (Robinson & Olazarri 1971), the host itself (Fabricius 1793) from (Robinson (Trail 1980; Tanaka 1984; Larcher & Wise & Robinson 1973). on 1985) or host-web silk (Shinkai 1988; Tso & the island of Guam in western Micronesia is Severinghaus 1998). Argyrodes may also cap- a frequent inhabitant of the large orb webs of ture prey themselves using an abandoned host several species. Kerr and Quenga (2003) re- web (Larcher & Wise 1985) or use their own port on population variation in different host small web (Whitehouse 1986). species and habitats for Guamanian Argyro- Despite a growing literature on the ecology des, including A. argentatus. The most com- of this interesting genus (e.g., Henaut mon orb-weaving spider on Guam hosting A. 2000; Miyashita 2001, 2002) and the prospect argentatus in their webs is Argiope appensa of powerful comparative phylogenetic ap- (Walckenaer 1841) (Araneidae) (25 mm), proaches (Agnarsson 2002; Whitehouse et al. which occurs from and across 2002), the behavior of most species is still the tropical western Paci®c to (Levi poorly known. One little studied species is Ar- 1983). It builds a nearly vertical planar orb gyrodes argentatus O.P. Cambridge 1880, a web with sticky spiral strands, occasionally small (adult female body length c. 5 mm) spi- with cruciate or diagonal strips of white silk der with a tall, conical, silvered abdomen that near the center (Kerr 1993). In this paper, I is reported from Madagascar eastward through record further aspects of the behavior of Ar- southeast Asia to South America (Cambridge gyrodes argentatus from Guam. Speci®cally, I asked: (1) What is the proportion of time 1 Current address: Department of Ecology, Evolu- spent in different activities? (2) How does be- tion and Marine Biology, University of California, havior differ between the sexes? (3) Does Santa Barbara CA 93601 USA. E-mail: alexander. population density on a host web affect Ar- [email protected] gyrodes behavior?

1 2 THE JOURNAL OF ARACHNOLOGY

METHODS that activity divided by the total number of Surveys.ÐThe surveys were performed in observations on that spider. Intersexual differ- Mangilao and Ukudu (Dededo), Guam, 6±31 ences in time spent in a behavior were ex- amined with one-way anovas or Kruskal-Wal- August 1989 in native forest, beach strand or lis procedures. I examined the relationship disturbed vegetation. Guam (13Њ N, 145Њ E; between the density of Argyrodes argentatus 540 km2) is a volcanic and tectonically up- and behavior using simple linear regressions. lifted limestone arc island in the western Pa- Recently hatched spiderlings were sometimes ci®c Ocean. Voucher specimens of Argyrodes extremely numerous on a web, but such ag- argentatus (adult males and females) and its gregations dispersed quickly. To prevent them host Argiope appensa (adult females) are de- from in¯ating estimates of population density posited at the Department of Zoology, South- on webs, spiderlings were excluded from the ern Illinois University at Carbondale and the analyses. To meet assumptions of parametric University of Guam Herbarium. Censusing of procedures, densities of adult Argyrodes were haphazardly encountered webs was conducted square-root or log transformed and propor- during periods of no rain between 0900±2100 tions were arc-sine-square root transformed h, since a preliminary survey (A. Kerr pers. when necessary. Statistical outliers were de- obs.) suggested that the spiders were most ac- tected using Dixon's tests (Sokal & Rohlf tive during this time. A dim red light was used 1981). Tests of association with categorical during nocturnal observations to avoid dis- variables were done with a G test with Wil- turbing the spiders or attracting insects to the liams' correction for small sample size. Then host web. The density of Argyrodes was mea- homogeneity of variances was checked with sured as number of spiders per unit area of Bartlett's tests and normality con®rmed with host web. The area of Argiope host webs, as Rankit plots (Sokal & Rohlf 1981). Otherwise de®ned by the outermost spiral strands, was analogous nonparametric procedures were computed as an ellipse based on horizontal used. and vertical web diameters. To determine the behavior of Argyrodes, I recorded for 20 Ar- RESULTS giope appensa host webs the activities of each General observations.ÐArgyrodes argen- Argyrodes found, after which the web was no tatus would sometimes glean small insects longer used. In these webs, a total of forty- from the host web. These were sometimes eat- eight Argyrodes argentatus (23 females, 21 en where found or wrapped and taken to the males and 4 juveniles) were each observed re- margin of the host web and eaten there. Ar- peatedly for ca. 15 s at 5-min intervals (sensu gyrodes would also remove and eat spiral Vollrath 1976) over a period of 1.5±4 h per catching silk from a host web, sometimes re- web for a total of 1,286 separate observations moving large sections. The host Argiope ap- in 107.2 spider hours. These same host webs pensa would sometimes leave wrapped small and spiders, and their recorded behaviors, prey at the capture site. These prey bundles were used in all analyses. Twelve webs (60%) were sometimes removed by Argyrodes ar- were from beach-strand vegetation, six (30%) gentatus, who would cut them from the web were from disturbed vegetation and two and hoist them to the host web's barrier (10%) occurred in native forest. I categorized strands. At other times, A. argentatus would behavior as quiescence (immobile and invert- feed together on prey bundles held in the ed), foraging (moving forward on prey-catch- host's mouth as the host rested at the web's ing spiral while rotating leg pair I sensu hub. I also observed one possible instance of Whitehouse 1986), forward movement on oth- on the host by A. argentatus. In this er silks at the web's margin, feeding (contact case, several A. argentatus were feeding at the between mouthparts and prey), stationary ac- web hub on a dead adult female Argiope ap- tivity (grooming, modifying silk, mating), or pensa host that had been alive the previous as agonism-avoidance (rushing towards or re- day. treating from other spiders). Proportion of time in activities.ÐPropor- Statistical analyses.ÐThe proportion of tional time differed signi®cantly between ac- time a spider spent performing an activity was tivities via a Kruskal-Wallis procedure (H ϭ computed as the number of observations for 121.62, P Ͻ 0.001). The 48 Argyrodes (fe- KERRÐARGYRODES BEHAVIOR 3

Figure 1.ÐTime budget of adult male and female Argyrodes argentatus. Males denoted by striped bars. Asterisk indicates a signi®cant difference via a one-way anova. males, males and juveniles combined) in the Ϯ 0.010; H ϭ 1.57, P ϭ 0.210), or agonism 20 host webs spent signi®cantly more time (females: 0.011 Ϯ 0.010; males: 0.005 Ϯ hanging immobile and inverted in the support 0.005; H ϭ 0.89, P ϭ 0.344). Of the 50 bouts strands at the webs' margin than in stationary of feeding (continuous feeding through mul- activity, forward movement at the web's mar- tiple observations), 38 were by females and gin, foraging, feeding or agonism (minimum 12 were by males. Females more often fed on pairwise G ϭ 11.02, P Ͻ 0.001 adjusted for large prey bundles prepared by the host (21% multiple comparisons). Other activities on av- of feeding events) than did males, who were erage occupied no more than 9% of the time. never observed feeding on prey bundles, but Instances of agonism were least frequent, oc- rather fed on small prey (tiny dipterans and curring 0.7% of the time. homopterans) caught on the spiral strands, but Sex related differences.ÐWhen sexes unnoticed by the host. This intersexual differ- were considered separately, there was one no- ences in type of prey used was not signi®cant ϭ ϭ table signi®cant difference in behavior. Fe- (Gadj 3.01; P 0.0829). males foraged signi®cantly more often than Effects of density.ÐThe density of A. ar- did males (females: mean proportion Ϯ 1SE gentatus on host webs varied by about an or- ϭ Ϯ Ϯ 0.172 0.039; males: 0.019 0.010; F1,42 der of magnitude, from a minimum of 0.027/ ϭ 19.776, P Ͻ 0.001) (Fig. 1). Other behav- 100 cm2 to 0.26/100 cm2 or 1±8 individuals iors were not signi®cantly different (P Ͼ 0.05) per web. The proportion of time that adult A. between the sexes: quiescence (females: 0.506 argentatus spent feeding on prey was weakly Ϯ 0.066; males: 0.625 Ϯ 0.065; H ϭ 0.798, but signi®cantly and inversely related with the P ϭ 0.372), stationary activities (females: density of this species on Argiope webs (n ϭ 0.079 Ϯ 0.014; males: 0.113 Ϯ 0.026; H ϭ 44; P ϭ 0.0462; r2 ϭ 0.142; y ϭϪ0.776 x ϩ 2.19, P ϭ 0.138), movement on support 0.430) (Fig. 2). I observed only six instances strands (females: 0.138 Ϯ 0.039; males: 0.160 of agonism between Argyrodes individuals or 4 THE JOURNAL OF ARACHNOLOGY

Figure 2.ÐProportion of time spent feeding by Figure 3.ÐThe rate of agonism and avoidance of adult Argyrodes argentatus versus its population hosts and other Argyrodes versus adult Argyrodes density on host webs of Argiope appensa. argentatus population density on host webs of Ar- giope appensa. between Argyrodes and their host. In three of these encounters, the aggression resulted in did males (38 versus 12, respectively). Hence one of the pair of Argyrodes leaving the web. feeding in females is spent in more, but short- Pooled intra-Argyrodes and Argyrodes-host er feedings. This difference might occur if agonism was not signi®cantly correlated with some of the feeding events by females were Argyrodes density (r2 ϭ 0.005; n ϭ 44; P Ͼ shorter because of higher food quality. Fe- 0.05) (Fig. 3). Other behaviors were also not males spent 21% of feeding events at large signi®cantly correlated with the density of prey bundles, while males were never ob- adult A. argentatus: quiescence (r2 ϭ 0.009), served doing so. One possibility that might stationary activities (r2 ϭ 0.051), movement account for these differences between the sex- on support strands (r2 ϭ 0.051), or foraging es is if females possess a larger energy and (r2 ϭϪ0.021). nutrient budget, such as needed for egg de- velopment and egg-sac construction (Toft DISCUSSION 1999). Of the Argyrodes species that take food in Food availability can affect a spider's the webs of larger spiders, sometimes foraging growth rate, adult size, fecundity (Miyashita is for small prey unnoticed by the host 1990, 1991) and web-site tenacity (Caraco & (Whitehouse 1986; Pasquet et al. 1997), while Gillespie 1986; but see Smallwood 1993), as at other times Argyrodes concentrates on shar- well as in¯uence the degree of intra- and in- ing large prey captured by the host (Vollrath terspeci®c competition (Spiller 1984). Argy- 1979; Whitehouse 1997). There may also be rodes antipodianus are known to aggressively species-speci®c differences in the use of these compete for food (Whitehouse 1997), sug- strategies (Tso & Severinghaus 2000). Inter- gesting that food is a limiting resource. In this sexual differences in foraging mode have also study, the time adult Argyrodes argentatus been noted. Cangialosi (1990) found that male spent feeding was inversely proportional to Argyrodes ululans O.P. Cambridge 1880 in their density on host webs (Fig. 2). One pos- webs of Anelosimus eximius (Keyserling sible explanation for this pattern is that as Ar- 1884) spend more time foraging than do fe- gyrodes density increases, so does competi- males. This is apparently because females tion for food because of increased wait until alerted by the hosts to steal freshly intraspeci®c agonism among Argyrodes and captured prey, while males more often search aggression by hosts which limits access to for previously caught prey for which it pre- food. However, aggressive interactions were sumably spends more time in locating. Con- very few and uncorrelated with Argyrodes versely, in this study, female A. argentatus density (Fig. 3). Another possibility is that spent more time foraging than males. I also time spent feeding is positively correlated did not ®nd an intersexual difference in pro- with an unmeasured variable, such as food portional feeding times (Fig. 1) (contra Can- quality, which itself negatively correlates with gialosi 1990). However, females engaged in Argyrodes density. Webs may be densely pop- over three times as many bouts of feeding as ulated with Argyrodes because of better food KERRÐARGYRODES BEHAVIOR 5 that requires less time to consume. For ex- Cangialosi, K.R. 1990. Life cycle and behavior of ample, feeding bouts (de®ned as a putatively the kleptoparasitic spider, Argyrodes ululans (Ar- continuous term of feeding through consecu- aneae, Theridiidae). Journal of Arachnology 18: tive observations made ®ve minutes apart) are 347±358. shorter when feeding with the host on large Caraco, T.W. & R.G. Gillspie. 1986. Risk sensitiv- predigested prey bundles. This has also been ity: foraging mode in an ambush predator. Ecol- ogy 67:1180±1185. observed by Whitehouse (1997) in another Exline, H. & H.W. Levi. 1962. American spiders of species, A. antipodianus O.P. Cambridge, the genus Argyrodes (Araneae, Theridiidae). Bul- 1880. letin of the Museum of Comparative Zoology There is growing interest in the evolution 127:75±202. of behavior in Argyrodes (Whitehouse et al. Henaut, Y. 2000. Host selection by a kleptoparasitic 2002). Several researchers are generating a spider. Journal of Natural History 34:747±753. phylogeny of the genus to use, for example, Kerr, A.M. 1993. Low frequency of stabilimenta in in ancestral state analyses of the correlated orb webs of Argiope appensa (Araneae: Aranei- evolution of kleptoparasitism and araneopha- dae) from Guam: an indirect effect of an intro- gy. The success of this promising approach duced avian predator? Paci®c Science 47:328± will depend not only on the quality of the phy- 337. logenetic estimates, but also in large measure Kerr, A.M. and A.S. Quenga. 2003. Population var- iation of web-invading spiders (Theridiidae: Ar- on natural history information from a behav- gyrodes spp.) on host webs in Guam, Mariana iorally diverse suite of Argyrodes. Robinson Islands, Micronesia. Journal of Natural History. and Olazarri (1971) listed several observa- In press. tions on the behavior of A. argentatus in Ar- Koh, J.K.H. 2000. A Guide to Common Singapore giope argentatus host webs from Panama. Spiders. Singapore Science Centre, Singapore. Most of the behaviors of Panamanian Argy- Larcher, S.F. & D.H. Wise. 1985. Experimental rodes appeared to parallel those of popula- studies of the interactions between a web-invad- tions in Argiope appensa host webs from ing spider and two host species. Journal of Ar- Guam in the western Paci®c. Guamanian pop- achnology 13:43±59. ulations also gleaned insects from the host Levi, H.W. 1983. The orb-weaver genera Argiope, web, stole host food bundles, fed with the host Gea, and Neogea from the Paci®c region (Ara- and appeared to occasionally attack, kill and neae: Araneidae: Argiopinae). Bulletin of the feed on the host. However, I did not ®nd, as Museum of Comparative Zoology 150:247±338. Miyasita, T. 1990. Decreased reproductive rates of reported for Panamanian populations, that Ar- the spider, Nephila clavipes, inhabiting small gyrodes removed host-wrapped prey bundles woodlands in urban areas. Ecological Research from the host web. Despite this possible dif- 5:341±351. ference, there is apparently little intraspeci®c Miyasita, T. 1991. Direct evidence of food limita- variation in these behaviors within the nearly tion for growth rate and body size in the spider pantropical species A. argentatus. Nephila clavipes. Acta Arachnologica 40:17±21. Miyasita, T. 2001. Competition for a limited space ACKNOWLEDGMENTS in kleptoparasitic Argyrodes spiders revealed by I thank J. Beatty for advice, for providing ®eld experiments. Population Ecology 43:97± literature and for identifying the spiders. I also 103 thank A. Rypstra, C. Craig, and the students Miyasita, T. 2002. Population dynamics of two spe- of Craig's behavioral ecology seminar for cies of kleptoparasitic spiders under different comments, as well as I. Schreiner for trans- host availabilities. Journal of Arachnology 30: lating Vollrath (1976). 31±38. Pasquet, A, R. Leborge & T. Canterral. 1997. Op- LITERATURE CITED portunistic egg feeding in the kleptoparasitic spi- Agnarsson, I. 2002. Sharing a webÐon the relation der Argyrodes gibbosus. Ethology 103:160±170. of sociality and kleptoparasitism in theridiid spi- Robinson, M.H. & J. Olazarri. 1971. Units of be- ders (Theridiidae, Araneae). Journal of Arach- havior and complex sequences in the predatory nology 30:181±188 behavior of (Fabricius) (Ara- Cambridge, O.P. 1880. On some new and little- neae: Araneidae). Smithsonian Contributions to known spiders of the genus Argyrodes. Proceed- Zoology 65:1±36. ings of the Zoological Society, London 1880: Robinson, M.H. and B. Robinson. 1973. Ecology 320±342. and behavior of the giant wood spider Nephila 6 THE JOURNAL OF ARACHNOLOGY

maculata (Fabricius) in New Guinea. Smithson- Vollrath, F. 1976. Konkurrenzvermeidung bei tro- ian Contributions to Zoology 149:1±76. pischen kleptoparasitischen Haubennetzspinnen Shinkai, A. 1988. A note on the web silk theft by der Gattung Argyrodes (Arachnida: Araneae: Argyrodes cylindratus (Thorell) (Araneae: Ther- Theridiidae). Entomologica Germanica 3:104± idiidae). Acta Arachnologica 38:116±119. 108. Smallwood, P.D. 1993. Web-site tenure in the long- Vollrath, F. 1979. Behaviour of the kleptoparasitic jawed spider: is it risk-sensitive foraging, or con- spider Argyrodes elevatus (Araneae, Theridi- speci®c interactions? Ecology 74:1826±1835. idae). Behaviour 27:515±521. Sokal, R.R. & F.J. Rohlf. 1981. Biometry. Freeman Whitehouse, M.A. 1986. The foraging behaviours and Company, San Francisco. of Argyrodes antipodiana (Theridiidae), a klep- Spiller, D.A. 1984. Competition between two spider toparasitic spider from New Zealand. New Zea- species: experimental ®eld study. Ecology 65: land Journal of Zoology 13:151±168. 909±919. Whitehouse, M.A. 1993. Group structure and time Tanaka, K. 1984. Rate of predation by a kleptopar- budgets of Argyrodes antipodiana (Araneae, asitic spider, Argyrodes ®ssifrons, upon a large Theridiidae), a kleptoparasitic spider from New host spider, Agelena limbata. Journal of Arach- Zealand. New Zealand Journal of Zoology 20: nology 12:363±367. 201±206. Whitehouse, M.E.A. 1997. The bene®ts of stealing Toft, S. 1999. Prey choice and spider ®tness. Jour- from a predator: Foraging rates, predation risk, nal of Arachnology 27:301±307. and intraspeci®c aggression in the kleptoparasitic Trail, D.S. 1980. Predation by Argyrodes (Theridi- spider Argyrodes antipodiana. Behavioral Ecol- idae) on solitary and communal spiders. Psyche ogy 8:663±667. 87:349±355. Whitehouse, M.A., I. Agnarsson, T. Miyashita, D. Tso, I-M. & L.L. Severinghaus. 1998. Silk stealing Smith, K. Cangialosi, T. Masumoto, D. Li and Y. by Argyrodes lanyuensis (Araneae: Theiriidae): Henaut. 2002. Argyrodes: phylogeny, sociality A unique form of kleptoparasitism. Animal Be- and interspeci®c interactions±a report on the Ar- haviour 56:219±225. gyrodes symposium, Badplaas 2001. Journal of Tso, I-M. & L.L. Severinghaus. 2000. Argyrodes Arachnology 30:238±245. ®ssifrons inhabiting webs of Cyrtophora hosts: Prey size distribution and population character- Manuscript received 3 June 2002, revised 27 May istics. Zoological Studies 39:236±242. 2003.