- Email: [email protected]
Feeding experience affects web relocation and investment in web threads in an orb-web spider, Cyclosa argenteoalba
ANIMAL BEHAVIOUR, 1999, 57, 1251–1255 Article No. anbe.1999.1105, available online at http://www.idealibrary.com on
Feeding experience affects web relocation and investment in web threads in an orb-web spider, Cyclosa argenteoalba KENSUKE NAKATA* & ATUSHI USHIMARU†
*Department of Zoology, Faculty of Science, Kyoto University †Center for Ecological Research, Kyoto University (Received 1 September 1998; initial acceptance 22 October 1998; final acceptance 18 February 1999; MS. number: 5978R)
Orb-web spiders often relocate their webs when they assess a web site as prey poor. When a spider spins its web at a new site, it may not be able to assess the prey availability at the site accurately on the first day, owing to stochastic variation in foraging success, but it is gradually able to make an assessment. Therefore, a spider’s foraging behaviour may change according to how long it has been at its current web site. To test this possibility, we conducted a prey removal experiment, with the spider Cyclosa argenteoalba, to compare the response to prey deprivation of spiders that were on new sites with that of spiders that had been at a site for several days. Spiders in both groups had a higher relocation rate than the natural rate, but more spiders in the new-site group relocated their webs. Spiders thus seemed to use previous experience of prey capture at a web site to decide whether to relocate their web. The total length of silk thread in a web was greater on the second day at a new web site than on the first. We suggest that spiders minimize their investment in web threads until they are certain that the web site is prey rich.
because they evaluated it as prey rich. Olive (1982) suggested that spiders respond to current foraging success, and not to their average success at a site over a certain period. Janetos (1982) suggested that spiders take a day to evaluate their web sites. (2) Stochastic variation in the prey capture rate at any foraging site is so great that spiders cannot assess the availability of prey, and cannot differentiate prey-rich and prey-poor web sites (the risk sensitivity hypothesis in Caraco & Gillespie 1986). Therefore, spiders occupy rich or poor sites by chance, independent of the prey availability at a specific site. Spiders that relocate often are expected to catch the average number of prey among various sites in the habitat. If spiders remain at one site, they may catch fewer or more prey than the average, depending on the quality of the site they happen to occupy. Therefore, in a rich habitat, where the average prey availability at different web sites exceeds the threshold for successful reproduction, a spider should relocate its web frequently to avoid the risk of inadvertently staying at a poor site for a long time. On the other hand, in a poor habitat, where spiders cannot reproduce if they catch only the average number of prey in the habitat, spiders should remain at one site, because their only chance for reproduction is to occupy a relatively rich site for a long time. Gillespie & Caraco (1987) found that the web relocation rate of Tetragnatha elongata matched the prediction from the risk sensitivity
Foraging animals often face variation in their rate of prey capture. This is considered to have two components (Cartar & Abrahams 1996): temporal change in the average availability of food at a site and stochastic variation in foraging success around the average food availability. To understand a species’ foraging ecology, it is important to determine whether animals are able to assess food availability. With food availability varying both over time and between sites, animals that are able to select or remain in good locations are likely to grow better and to breed more successfully. To do this, they need to assess the availability of food at a site. However, stochastic variation in foraging success may prevent animals from making an accurate assessment. Orb-weaving spiders relocate their webs frequently. This phenomenon has been explained in two ways, depending on whether the spider is considered able to assess the availability of prey at a web site. (1) Spiders assess the availability of prey at a site promptly, and relocate their webs in response to low rates of prey capture. Spiders experimentally supplied with an abundance of prey were more tenacious than those in natural conditions (Janetos 1982; Olive 1982). The interpretation was that the spiders stayed at the site Correspondence: K. Nakata, Department of Zoology, Faculty of Science, Kyoto University, Kitashirakawa-Oiwakecho, Sakyo-ku, Kyoto, 606-8502, Japan (email: [email protected]). 0003–3472/99/061251+05 $30.00/0
1999 The Association for the Study of Animal Behaviour
1251
1999 The Association for the Study of Animal Behaviour
1252 ANIMAL BEHAVIOUR, 57, 6
hypothesis. However, Smallwood (1993) studied the same species, in the same habitat, and determined that the high relocation rate in a rich habitat was due to a high spider density, and not to risk-sensitive foraging. Our hypothesis is based on a different assumption. Stochastic variation in prey capture prevents a spider from assessing the quality of its web site immediately. When a spider occupies a new site, it initially has no information about the prey availability at that location. At the end of the first day at the new site, the spider assesses the prey availability at the site based on its current prey capture experience. This assessment is not always correct, because of stochastic variation in foraging success. For example, a spider that does not catch any prey may be occupying a rich site but may not catch any prey by chance, or it may be occupying a poor site (Vollrath 1987). The spider cannot distinguish between these two cases based on 1-day prey capture experience, but has to decide whether to leave its current web site. However, if the spider remains at the site, it is gradually able to assess the quality of the site. Therefore, a spider’s response to its current prey capture experience will depend on whether it has had previous experience at the current web site. To test this, we conducted a prey removal experiment, using the orb-web spider, Cyclosa argenteoalba, to compare the web relocation rates of spiders that were occupying a new web site with spiders that had been at a site for several days. Our hypothesis leads to two predictions. (1) Prey removal causes a high rate of web relocation in both groups of spiders. (2) Spiders at an unfamiliar site (a newly occupied site) are more likely to relocate their webs than spiders at a familiar site (where they have remained for several days). In the experiment, the spiders in the first group experienced only experimentally induced foraging failure, while the spiders in the second group had presumably caught some prey at the site. Therefore, spiders in the first group should be more likely to interpret current foraging failure as an indication of a poor-quality site. Our assumption raises another question. Does a spider change its investment in web threads in the course of becoming acquainted with its web site? Spiders adjust their web structure according to their energetic state (Sherman 1994), prey capture rate (Higgins & Buskirk 1992) and prey type (Sandoval 1994). This evidence suggests that the cost of constructing a silk web is high, and that spiders must conserve the cost. Since a spider has to construct a web before capturing any prey, it risks wasting its investment if it is constructed in a poor location. To minimize this risk, the spider should minimize its investment until it is sure that the web site is prey rich. To test this hypothesis, we compared the total length of web threads on the first 2 days at a new site. METHODS Cyclosa argenteoalba is a multivoltine orb spider, which occurs in East Asia. It typically inhabits open, sunny places, such as the forest edge or bamboo forests. The adult females are 5–6 mm in size. The spider usually weaves its web at dawn and consumes it at night. It
decorates its web with debris, but unlike other Cyclosa spiders which camouflage themselves by sitting on the debris, it hangs its debris away from the centre hub and does not transfer it to a new web during web relocation. The study area was the southern part of the Botanical Garden of the Faculty of Science, Kyoto University, Japan. The area was approximately 1040 m. We surveyed the study area daily from May until October, to observe the natural pattern of web relocation. When we found a new adult or subadult female in the area, we removed it from its web and marked it on the femur with an individual dot pattern, using enamel paint. Then, we carefully returned the spider to its web. The majority of spiders quickly returned to the centre of the hub, the normal place where spiders wait for prey, and sat there motionless. A maximum of 17 marked spiders were in the study area at one time, although marked spiders disappeared from early July to early August. For each marked spider, we recorded the location of the web and whether it was relocated. Of 111 spiders marked, 23 could not be found the day after marking; therefore the observations are for 88 spiders. To minimize the effect of marking, we waited 2 days before observing each marked spider. When a spider was not found in the study area or adjacent areas for 10 days, it was considered to be dead. Starting in September, we also observed individual web morphology in relation to web relocation. We recorded the diameters of the innermost and outermost catching spirals, the number of radii, and the number of catching and hub spirals. From these data, we calculated the total thread length of the web, according to formulas in Sherman (1994). To compare these parameters on the first and second days at a new site, we used the Wilcoxon signed-ranks test. To investigate whether prior information affected the decision to relocate the web, we conducted a prey removal experiment. At dawn, we checked whether individual marked spiders had relocated their webs. All spiders that had relocated their webs and an equal number of spiders that had not relocated were selected for manipulation. With tweezers, we kept removing prey, during the day, from the webs of the spiders in both groups. Care was taken not to damage the web or disturb the spiders’ behaviour. When we found spiders already wrapping or feeding on prey, we forcibly removed the prey from the spiders’ legs or jaws. In such cases, we observed that the spiders rapidly jerked the web with their legs and turned in circles. This ‘turn-and-jerk’ behaviour (Suter 1978) is thought to indicate that the spiders were looking for lost prey. The spiders soon stopped this behaviour and sat still. We began prey removal at 0700 hours, about 90 min after sunrise, and terminated it at sunset. The majority of webs had no holes at 0700 hours, suggesting that no prey had been caught that day. Since we inspected each web every 3 min, the prey removal decreased the foraging success of each spider to virtually zero. The next day, we checked to see whether each spider had relocated in response to the manipulation. Since we could not obtain an adequate sample size for statistical analysis in 1 day, we conducted the same experiment five times from mid-August to late
NAKATA & USHIMARU: SPIDER WEB RELOCATION 1253
40 Observed Expected
35 30 Frequency
September, and manipulated 23 relocating and 23 nonrelocating spiders. We did not use the same spider more than once, irrespective of whether it was relocating or nonrelocating. The web relocation rate did not change over the course of the experimental period: 21–31 August: 19.8%; 1–10 September: 19.6%; 11–20 September: 21.2%; 21–30 September: 17.6% (G test: G3 =0.501, NS). We assumed that the day of the experiment had negligible effect on the results, and so we pooled the data. We used a chi-square test to analyse the difference in response between relocating and nonrelocating spiders.
25 20 15 10 5
RESULTS
0
1
2
5 3 4 6 Duration of residence (days)
7
>– 8
Figure 1. Frequency distribution of the number of days spiders were resident at a web site. The expected distribution is geometric with P=0.27.
6000
Length on second day (cm)
Spiders were observed for 10.3611.62 (XSD) days (N=88), for a total of 913 spider-days of observation. Webs were relocated 162 times. The other 751 times, the spiders remained at the same site that they had occupied the day before. When we divide the number of web relocations (162) by the total number of observation days (913), we obtain a relocation rate of 17.7%. Of the 162 relocations, 123 occurred after the spider had stayed at the same site for 2 consecutive days. In other words, web relocation of a spider from a familiar site was observed 123 times. The natural relocation rate at a familiar site was 14.1% (123 relocations versus 751 stays). The other 39 relocations occurred from an unfamiliar site. That is, they occurred on the day immediately after a previous relocation. The natural relocation rate at an unfamiliar site (31.7%) was calculated by dividing the number of consecutive relocations (39) by the number of the relocations from familiar sites (123). The averageSD number of days in residence was 3.683.85. If web relocation occurs with a constant probability, p, the days in residence (x) are expected to have a geometric distribution, with a probability function of p(x)=pqx1, where q=1p. Since the average of a geometric distribution is 1/p, p is assumed to be 0.27, the reciprocal of 3.68; however, the observed distribution is significantly different (÷26 =13.61, P<0.05; Fig. 1). The observed frequency was higher than expected for both the shortest (1 day) and longest (d8 days) number of days in residence. The total thread length on the first day (XSD=3128.51244.8 cm) was significantly shorter than that on the second day (3710.31131.8 cm; Wilcoxon signed-ranks test: Z= 2.373, N=16, P=0.0176; Fig. 2). Of the 16 webs observed, the total thread length increased in 12 webs. In the prey removal experiment, spiders at an unfamiliar site relocated more often than spiders at a familiar site (÷21 =4.293, P<0.05; Fig. 3). However, the relocation rate at a familiar site was significantly higher than the natural rate of 14.1% (÷21 =4.841, P<0.05). The relocation rate at an unfamiliar site was also significantly higher than the natural rate of 31.7% (÷21 =7.126, P<0.01).The average number of removed preySD was 1.681.45 in unfamiliar sites and 2.321.70 in familiar sites (range 0–5 in both sites; Mann–Whitney test: Z= 1.124, N1 =N2 =23, P>0.2).
4000
2000
0
2000
4000
6000
Length on first day (cm) Figure 2. Change in total length of web thread between the first and second days spiders were resident at a new web site. Each square represents one web. Squares above the diagonal are webs in which the total thread length increased.
DISCUSSION The results of the prey removal experiment support both of our predictions. In response to prey removal, spiders at both the unfamiliar and familiar sites relocated their webs at a higher rate than the naturally occurring rate (prediction 1). In addition, spiders that had not relocated the previous day were more likely to remain at their current web site (prediction 2), although there was no obvious difference in the expected prey capture rate (equivalent to the number of prey removed). This is consistent with field observations, which found that web relocation occurred in a nonrandom manner over time. The higher rate of 1-day residence periods indicates that spiders relocated
1254 ANIMAL BEHAVIOUR, 57, 6
unknown site, to reduce the risk of investing too much in a site with poor return. These results seem to show that spiders change their energetic investment in a web as they become acquainted with the site. Although the web silk is consumed each night and recycled, spiders are known to adjust their investment in the web in response to energy input and output (e.g. Sherman 1994), suggesting its importance in their energy budget. The change in the investment in web threads, together with the use of prey capture experience in decision making for web relocation, should be interpreted as adaptation by spiders to environmental uncertainty.
18 16
Relocate Stay
Number of spiders
14 12 10 8 6 4 2 0
Acknowledgments Unfamiliar site
Familiar site
Figure 3. The difference in response to prey removal between spiders at an unfamiliar site (spiders had just relocated their webs to the new site) and spiders at a familiar site (spiders had been at the site for several days).
more often on the day after web relocation. These results suggest that a spider’s relocation of its web site is based on its assessment of the site’s quality, and it uses current and previous prey capture experience at the site to make its decision. If spiders were unable to assess the prey availability of a site because of stochastic variation, as the risk sensitivity hypothesis assumes (Caraco & Gillespie 1986), then neither group of spiders would respond to prey removal and their relocation rate would not be different from the natural relocation rate. If spiders were able to assess the prey availability promptly, and respond only to current prey capture experience as Janetos (1982) and Olive (1982) suggested, then prediction 1 would be met, but prediction 2 would not. Neither the risk sensitivity hypothesis nor the prompt assessment hypothesis can therefore explain the results of our experiment. Many studies have revealed that spiders leave their web site when they capture few prey (Turnbull 1964; Janetos 1982; Olive 1982; Rypstra 1983; Vollrath 1985). Vollrath & Houston (1986) found that Nephila clavipes randomly relocated its web with time; however, their observation was conducted under a constant ‘no prey’ condition. The absence of stochastic variation in foraging success may explain why their result differs from ours. Vollrath & Houston (1986) also found that previous experience did not affect the decision to leave a site, and suggested that spiders had no long-term memory. Similar results have been found in patch selection in the crab spider Misumena vatia (Morse 1993), and the wolf spider Schizocosa ocreata (Persons & Uetz 1996). Our results, however, suggest that spiders do use previous prey capture experience at their current sites to decide whether to leave. There is also some evidence that previous experience affects behaviour other than web relocation (Se´brier & Krafft 1993; Whitehouse 1997). We also observed an increase in the total length of the web thread on the second day at a new site. This suggests that spiders initially restrict their investment at an
We thank T. Watanabe and S. Yabuta and members of the Laboratory of Animal Ecology, Kyoto University for their discussions and comments. We also thank the Botanical Garden, Faculty of Science, Kyoto University, for substantial help. References Caraco, T. & Gillespie, R. G. 1986. Risk-sensitivity: foraging mode in an ambush predator. Ecology, 67, 1180–1185. Cartar, R. V. & Abrahams, M. V. 1996. Risk-sensitive foraging in a patch departure context: a test with worker bumble bees. American Zoologist, 36, 447–458. Gillespie, R. G. & Caraco, T. 1987. Risk-sensitive foraging strategies of two spider population. Ecology, 68, 887–899. Higgins, L. E. & Buskirk, R. E. 1992. A trap-building predator exhibits different tactics for different aspects of foraging behaviour. Animal Behaviour, 44, 485–499. Janetos, A. C. 1982. Foraging tactics of two guilds of web-spinning spiders. Behavioral Ecology and Sociobiology, 10, 19–27. Morse, D. H. 1993. Choosing hunting site with little information: patch-choice responses of crab spiders to distant cues. Behavioral Ecology, 4, 61–65. Olive, C. W. 1982. Behavioral response of a sit-and-wait predator to spatial variation in foraging gain. Ecology, 63, 912–920. Persons, M. H. & Uetz, G. W. 1996. The influence of sensory information on parch residence time in wolf spiders (Araneae: Lycosidae). Animal Behaviour, 51, 1285–1293. Rypstra, A. L. 1983. The importance of food and space in limiting web-spider densities; a test using field enclosures. Oecologia, 59, 312–316. Sandoval, C. P. 1994. Plasticity in web design in the spider Parawixia bistriata: a response to variable prey type. Functional Ecology, 8, 701–707. Se´brier, M. A. & Krafft, B. 1993. Influence of prior experience on prey consumption behaviour in the spider Zygiella x-notata. Ethology, Ecology and Evolution, 5, 541–547. Sherman, P. M. 1994. The orb-web: an energetic and behavioural estimator of a spider’s dynamic foraging and reproductive strategies. Animal Behaviour, 48, 19–34. Smallwood, P. D. 1993. Web-site tenure in the long-jawed spider: is it risk-sensitive foraging, or conspecific interaction? Ecology, 74, 1826–1835. Suter, R. 1978. Cyclosa turbinata (Araneae, Araneidae): prey discrimination via web-borne vibrations. Behavioral Ecology and Sociobiology, 3, 283–296. Turnbull, A. L. 1964. The search for prey by a web-building spider Achaearanea tepidariorum (C. L. Koch) (Araneae, Theridiidae). Canadian Entomologist, 96, 568–579.
NAKATA & USHIMARU: SPIDER WEB RELOCATION 1255
Vollrath, F. 1985. Web spider’s dilemma: a risky move or site dependent growth. Oecologia, 68, 69–72. Vollrath, F. 1987. Growth, foraging and reproductive success. In: Ecophysiology of Spiders (Ed. by W. Nentwig), pp. 357–370. Berlin: Springer-Verlag.
Vollrath, F. & Houston, A. 1986. Previous experience and site tenacity in the orb spider Nephila (Araneae, Aranidae). Oecologia, 70, 305–308. Whitehouse, M. E. A. 1997. Experience influences male–male contests in the spider Argyrodes antipodiana (Theridiidae: Araneae). Animal Behaviour, 53, 913–923.
Feeding experience affects web relocation and investment in web threads in an orb-web spider, Cyclosa argenteoalba KENSUKE NAKATA* & ATUSHI USHIMARU†
*Department of Zoology, Faculty of Science, Kyoto University †Center for Ecological Research, Kyoto University (Received 1 September 1998; initial acceptance 22 October 1998; final acceptance 18 February 1999; MS. number: 5978R)
Orb-web spiders often relocate their webs when they assess a web site as prey poor. When a spider spins its web at a new site, it may not be able to assess the prey availability at the site accurately on the first day, owing to stochastic variation in foraging success, but it is gradually able to make an assessment. Therefore, a spider’s foraging behaviour may change according to how long it has been at its current web site. To test this possibility, we conducted a prey removal experiment, with the spider Cyclosa argenteoalba, to compare the response to prey deprivation of spiders that were on new sites with that of spiders that had been at a site for several days. Spiders in both groups had a higher relocation rate than the natural rate, but more spiders in the new-site group relocated their webs. Spiders thus seemed to use previous experience of prey capture at a web site to decide whether to relocate their web. The total length of silk thread in a web was greater on the second day at a new web site than on the first. We suggest that spiders minimize their investment in web threads until they are certain that the web site is prey rich.
because they evaluated it as prey rich. Olive (1982) suggested that spiders respond to current foraging success, and not to their average success at a site over a certain period. Janetos (1982) suggested that spiders take a day to evaluate their web sites. (2) Stochastic variation in the prey capture rate at any foraging site is so great that spiders cannot assess the availability of prey, and cannot differentiate prey-rich and prey-poor web sites (the risk sensitivity hypothesis in Caraco & Gillespie 1986). Therefore, spiders occupy rich or poor sites by chance, independent of the prey availability at a specific site. Spiders that relocate often are expected to catch the average number of prey among various sites in the habitat. If spiders remain at one site, they may catch fewer or more prey than the average, depending on the quality of the site they happen to occupy. Therefore, in a rich habitat, where the average prey availability at different web sites exceeds the threshold for successful reproduction, a spider should relocate its web frequently to avoid the risk of inadvertently staying at a poor site for a long time. On the other hand, in a poor habitat, where spiders cannot reproduce if they catch only the average number of prey in the habitat, spiders should remain at one site, because their only chance for reproduction is to occupy a relatively rich site for a long time. Gillespie & Caraco (1987) found that the web relocation rate of Tetragnatha elongata matched the prediction from the risk sensitivity
Foraging animals often face variation in their rate of prey capture. This is considered to have two components (Cartar & Abrahams 1996): temporal change in the average availability of food at a site and stochastic variation in foraging success around the average food availability. To understand a species’ foraging ecology, it is important to determine whether animals are able to assess food availability. With food availability varying both over time and between sites, animals that are able to select or remain in good locations are likely to grow better and to breed more successfully. To do this, they need to assess the availability of food at a site. However, stochastic variation in foraging success may prevent animals from making an accurate assessment. Orb-weaving spiders relocate their webs frequently. This phenomenon has been explained in two ways, depending on whether the spider is considered able to assess the availability of prey at a web site. (1) Spiders assess the availability of prey at a site promptly, and relocate their webs in response to low rates of prey capture. Spiders experimentally supplied with an abundance of prey were more tenacious than those in natural conditions (Janetos 1982; Olive 1982). The interpretation was that the spiders stayed at the site Correspondence: K. Nakata, Department of Zoology, Faculty of Science, Kyoto University, Kitashirakawa-Oiwakecho, Sakyo-ku, Kyoto, 606-8502, Japan (email: [email protected]). 0003–3472/99/061251+05 $30.00/0
1999 The Association for the Study of Animal Behaviour
1251
1999 The Association for the Study of Animal Behaviour
1252 ANIMAL BEHAVIOUR, 57, 6
hypothesis. However, Smallwood (1993) studied the same species, in the same habitat, and determined that the high relocation rate in a rich habitat was due to a high spider density, and not to risk-sensitive foraging. Our hypothesis is based on a different assumption. Stochastic variation in prey capture prevents a spider from assessing the quality of its web site immediately. When a spider occupies a new site, it initially has no information about the prey availability at that location. At the end of the first day at the new site, the spider assesses the prey availability at the site based on its current prey capture experience. This assessment is not always correct, because of stochastic variation in foraging success. For example, a spider that does not catch any prey may be occupying a rich site but may not catch any prey by chance, or it may be occupying a poor site (Vollrath 1987). The spider cannot distinguish between these two cases based on 1-day prey capture experience, but has to decide whether to leave its current web site. However, if the spider remains at the site, it is gradually able to assess the quality of the site. Therefore, a spider’s response to its current prey capture experience will depend on whether it has had previous experience at the current web site. To test this, we conducted a prey removal experiment, using the orb-web spider, Cyclosa argenteoalba, to compare the web relocation rates of spiders that were occupying a new web site with spiders that had been at a site for several days. Our hypothesis leads to two predictions. (1) Prey removal causes a high rate of web relocation in both groups of spiders. (2) Spiders at an unfamiliar site (a newly occupied site) are more likely to relocate their webs than spiders at a familiar site (where they have remained for several days). In the experiment, the spiders in the first group experienced only experimentally induced foraging failure, while the spiders in the second group had presumably caught some prey at the site. Therefore, spiders in the first group should be more likely to interpret current foraging failure as an indication of a poor-quality site. Our assumption raises another question. Does a spider change its investment in web threads in the course of becoming acquainted with its web site? Spiders adjust their web structure according to their energetic state (Sherman 1994), prey capture rate (Higgins & Buskirk 1992) and prey type (Sandoval 1994). This evidence suggests that the cost of constructing a silk web is high, and that spiders must conserve the cost. Since a spider has to construct a web before capturing any prey, it risks wasting its investment if it is constructed in a poor location. To minimize this risk, the spider should minimize its investment until it is sure that the web site is prey rich. To test this hypothesis, we compared the total length of web threads on the first 2 days at a new site. METHODS Cyclosa argenteoalba is a multivoltine orb spider, which occurs in East Asia. It typically inhabits open, sunny places, such as the forest edge or bamboo forests. The adult females are 5–6 mm in size. The spider usually weaves its web at dawn and consumes it at night. It
decorates its web with debris, but unlike other Cyclosa spiders which camouflage themselves by sitting on the debris, it hangs its debris away from the centre hub and does not transfer it to a new web during web relocation. The study area was the southern part of the Botanical Garden of the Faculty of Science, Kyoto University, Japan. The area was approximately 1040 m. We surveyed the study area daily from May until October, to observe the natural pattern of web relocation. When we found a new adult or subadult female in the area, we removed it from its web and marked it on the femur with an individual dot pattern, using enamel paint. Then, we carefully returned the spider to its web. The majority of spiders quickly returned to the centre of the hub, the normal place where spiders wait for prey, and sat there motionless. A maximum of 17 marked spiders were in the study area at one time, although marked spiders disappeared from early July to early August. For each marked spider, we recorded the location of the web and whether it was relocated. Of 111 spiders marked, 23 could not be found the day after marking; therefore the observations are for 88 spiders. To minimize the effect of marking, we waited 2 days before observing each marked spider. When a spider was not found in the study area or adjacent areas for 10 days, it was considered to be dead. Starting in September, we also observed individual web morphology in relation to web relocation. We recorded the diameters of the innermost and outermost catching spirals, the number of radii, and the number of catching and hub spirals. From these data, we calculated the total thread length of the web, according to formulas in Sherman (1994). To compare these parameters on the first and second days at a new site, we used the Wilcoxon signed-ranks test. To investigate whether prior information affected the decision to relocate the web, we conducted a prey removal experiment. At dawn, we checked whether individual marked spiders had relocated their webs. All spiders that had relocated their webs and an equal number of spiders that had not relocated were selected for manipulation. With tweezers, we kept removing prey, during the day, from the webs of the spiders in both groups. Care was taken not to damage the web or disturb the spiders’ behaviour. When we found spiders already wrapping or feeding on prey, we forcibly removed the prey from the spiders’ legs or jaws. In such cases, we observed that the spiders rapidly jerked the web with their legs and turned in circles. This ‘turn-and-jerk’ behaviour (Suter 1978) is thought to indicate that the spiders were looking for lost prey. The spiders soon stopped this behaviour and sat still. We began prey removal at 0700 hours, about 90 min after sunrise, and terminated it at sunset. The majority of webs had no holes at 0700 hours, suggesting that no prey had been caught that day. Since we inspected each web every 3 min, the prey removal decreased the foraging success of each spider to virtually zero. The next day, we checked to see whether each spider had relocated in response to the manipulation. Since we could not obtain an adequate sample size for statistical analysis in 1 day, we conducted the same experiment five times from mid-August to late
NAKATA & USHIMARU: SPIDER WEB RELOCATION 1253
40 Observed Expected
35 30 Frequency
September, and manipulated 23 relocating and 23 nonrelocating spiders. We did not use the same spider more than once, irrespective of whether it was relocating or nonrelocating. The web relocation rate did not change over the course of the experimental period: 21–31 August: 19.8%; 1–10 September: 19.6%; 11–20 September: 21.2%; 21–30 September: 17.6% (G test: G3 =0.501, NS). We assumed that the day of the experiment had negligible effect on the results, and so we pooled the data. We used a chi-square test to analyse the difference in response between relocating and nonrelocating spiders.
25 20 15 10 5
RESULTS
0
1
2
5 3 4 6 Duration of residence (days)
7
>– 8
Figure 1. Frequency distribution of the number of days spiders were resident at a web site. The expected distribution is geometric with P=0.27.
6000
Length on second day (cm)
Spiders were observed for 10.3611.62 (XSD) days (N=88), for a total of 913 spider-days of observation. Webs were relocated 162 times. The other 751 times, the spiders remained at the same site that they had occupied the day before. When we divide the number of web relocations (162) by the total number of observation days (913), we obtain a relocation rate of 17.7%. Of the 162 relocations, 123 occurred after the spider had stayed at the same site for 2 consecutive days. In other words, web relocation of a spider from a familiar site was observed 123 times. The natural relocation rate at a familiar site was 14.1% (123 relocations versus 751 stays). The other 39 relocations occurred from an unfamiliar site. That is, they occurred on the day immediately after a previous relocation. The natural relocation rate at an unfamiliar site (31.7%) was calculated by dividing the number of consecutive relocations (39) by the number of the relocations from familiar sites (123). The averageSD number of days in residence was 3.683.85. If web relocation occurs with a constant probability, p, the days in residence (x) are expected to have a geometric distribution, with a probability function of p(x)=pqx1, where q=1p. Since the average of a geometric distribution is 1/p, p is assumed to be 0.27, the reciprocal of 3.68; however, the observed distribution is significantly different (÷26 =13.61, P<0.05; Fig. 1). The observed frequency was higher than expected for both the shortest (1 day) and longest (d8 days) number of days in residence. The total thread length on the first day (XSD=3128.51244.8 cm) was significantly shorter than that on the second day (3710.31131.8 cm; Wilcoxon signed-ranks test: Z= 2.373, N=16, P=0.0176; Fig. 2). Of the 16 webs observed, the total thread length increased in 12 webs. In the prey removal experiment, spiders at an unfamiliar site relocated more often than spiders at a familiar site (÷21 =4.293, P<0.05; Fig. 3). However, the relocation rate at a familiar site was significantly higher than the natural rate of 14.1% (÷21 =4.841, P<0.05). The relocation rate at an unfamiliar site was also significantly higher than the natural rate of 31.7% (÷21 =7.126, P<0.01).The average number of removed preySD was 1.681.45 in unfamiliar sites and 2.321.70 in familiar sites (range 0–5 in both sites; Mann–Whitney test: Z= 1.124, N1 =N2 =23, P>0.2).
4000
2000
0
2000
4000
6000
Length on first day (cm) Figure 2. Change in total length of web thread between the first and second days spiders were resident at a new web site. Each square represents one web. Squares above the diagonal are webs in which the total thread length increased.
DISCUSSION The results of the prey removal experiment support both of our predictions. In response to prey removal, spiders at both the unfamiliar and familiar sites relocated their webs at a higher rate than the naturally occurring rate (prediction 1). In addition, spiders that had not relocated the previous day were more likely to remain at their current web site (prediction 2), although there was no obvious difference in the expected prey capture rate (equivalent to the number of prey removed). This is consistent with field observations, which found that web relocation occurred in a nonrandom manner over time. The higher rate of 1-day residence periods indicates that spiders relocated
1254 ANIMAL BEHAVIOUR, 57, 6
unknown site, to reduce the risk of investing too much in a site with poor return. These results seem to show that spiders change their energetic investment in a web as they become acquainted with the site. Although the web silk is consumed each night and recycled, spiders are known to adjust their investment in the web in response to energy input and output (e.g. Sherman 1994), suggesting its importance in their energy budget. The change in the investment in web threads, together with the use of prey capture experience in decision making for web relocation, should be interpreted as adaptation by spiders to environmental uncertainty.
18 16
Relocate Stay
Number of spiders
14 12 10 8 6 4 2 0
Acknowledgments Unfamiliar site
Familiar site
Figure 3. The difference in response to prey removal between spiders at an unfamiliar site (spiders had just relocated their webs to the new site) and spiders at a familiar site (spiders had been at the site for several days).
more often on the day after web relocation. These results suggest that a spider’s relocation of its web site is based on its assessment of the site’s quality, and it uses current and previous prey capture experience at the site to make its decision. If spiders were unable to assess the prey availability of a site because of stochastic variation, as the risk sensitivity hypothesis assumes (Caraco & Gillespie 1986), then neither group of spiders would respond to prey removal and their relocation rate would not be different from the natural relocation rate. If spiders were able to assess the prey availability promptly, and respond only to current prey capture experience as Janetos (1982) and Olive (1982) suggested, then prediction 1 would be met, but prediction 2 would not. Neither the risk sensitivity hypothesis nor the prompt assessment hypothesis can therefore explain the results of our experiment. Many studies have revealed that spiders leave their web site when they capture few prey (Turnbull 1964; Janetos 1982; Olive 1982; Rypstra 1983; Vollrath 1985). Vollrath & Houston (1986) found that Nephila clavipes randomly relocated its web with time; however, their observation was conducted under a constant ‘no prey’ condition. The absence of stochastic variation in foraging success may explain why their result differs from ours. Vollrath & Houston (1986) also found that previous experience did not affect the decision to leave a site, and suggested that spiders had no long-term memory. Similar results have been found in patch selection in the crab spider Misumena vatia (Morse 1993), and the wolf spider Schizocosa ocreata (Persons & Uetz 1996). Our results, however, suggest that spiders do use previous prey capture experience at their current sites to decide whether to leave. There is also some evidence that previous experience affects behaviour other than web relocation (Se´brier & Krafft 1993; Whitehouse 1997). We also observed an increase in the total length of the web thread on the second day at a new site. This suggests that spiders initially restrict their investment at an
We thank T. Watanabe and S. Yabuta and members of the Laboratory of Animal Ecology, Kyoto University for their discussions and comments. We also thank the Botanical Garden, Faculty of Science, Kyoto University, for substantial help. References Caraco, T. & Gillespie, R. G. 1986. Risk-sensitivity: foraging mode in an ambush predator. Ecology, 67, 1180–1185. Cartar, R. V. & Abrahams, M. V. 1996. Risk-sensitive foraging in a patch departure context: a test with worker bumble bees. American Zoologist, 36, 447–458. Gillespie, R. G. & Caraco, T. 1987. Risk-sensitive foraging strategies of two spider population. Ecology, 68, 887–899. Higgins, L. E. & Buskirk, R. E. 1992. A trap-building predator exhibits different tactics for different aspects of foraging behaviour. Animal Behaviour, 44, 485–499. Janetos, A. C. 1982. Foraging tactics of two guilds of web-spinning spiders. Behavioral Ecology and Sociobiology, 10, 19–27. Morse, D. H. 1993. Choosing hunting site with little information: patch-choice responses of crab spiders to distant cues. Behavioral Ecology, 4, 61–65. Olive, C. W. 1982. Behavioral response of a sit-and-wait predator to spatial variation in foraging gain. Ecology, 63, 912–920. Persons, M. H. & Uetz, G. W. 1996. The influence of sensory information on parch residence time in wolf spiders (Araneae: Lycosidae). Animal Behaviour, 51, 1285–1293. Rypstra, A. L. 1983. The importance of food and space in limiting web-spider densities; a test using field enclosures. Oecologia, 59, 312–316. Sandoval, C. P. 1994. Plasticity in web design in the spider Parawixia bistriata: a response to variable prey type. Functional Ecology, 8, 701–707. Se´brier, M. A. & Krafft, B. 1993. Influence of prior experience on prey consumption behaviour in the spider Zygiella x-notata. Ethology, Ecology and Evolution, 5, 541–547. Sherman, P. M. 1994. The orb-web: an energetic and behavioural estimator of a spider’s dynamic foraging and reproductive strategies. Animal Behaviour, 48, 19–34. Smallwood, P. D. 1993. Web-site tenure in the long-jawed spider: is it risk-sensitive foraging, or conspecific interaction? Ecology, 74, 1826–1835. Suter, R. 1978. Cyclosa turbinata (Araneae, Araneidae): prey discrimination via web-borne vibrations. Behavioral Ecology and Sociobiology, 3, 283–296. Turnbull, A. L. 1964. The search for prey by a web-building spider Achaearanea tepidariorum (C. L. Koch) (Araneae, Theridiidae). Canadian Entomologist, 96, 568–579.
NAKATA & USHIMARU: SPIDER WEB RELOCATION 1255
Vollrath, F. 1985. Web spider’s dilemma: a risky move or site dependent growth. Oecologia, 68, 69–72. Vollrath, F. 1987. Growth, foraging and reproductive success. In: Ecophysiology of Spiders (Ed. by W. Nentwig), pp. 357–370. Berlin: Springer-Verlag.
Vollrath, F. & Houston, A. 1986. Previous experience and site tenacity in the orb spider Nephila (Araneae, Aranidae). Oecologia, 70, 305–308. Whitehouse, M. E. A. 1997. Experience influences male–male contests in the spider Argyrodes antipodiana (Theridiidae: Araneae). Animal Behaviour, 53, 913–923.