Phenotypic plasticity is the ability of an organism or genotype to express an alternative morphology, phys iology or behaviour in response to environmental stimuli (Schlichting 1986). Plasticity is a trait in itself, subjected to natural selection and evolutionary change (Bradshaw 1965) Foraging strategies of gen- eralist predators am чагу from n single generalized strategy to a repertoire of very specialized strategies. Plasticity in foraging strategies should be favoured if specialized strategics are levs efficient in capturing the different prey types than generalized ones. Orb weaving spiders are usually considered generalist predators but there is no еч ideiKe that they can alter their webs in response to different types of prey Although within-spccics variation in web design has been well documented, it has been attributed to fac tors other than prey capture, such as restricted space for the web. wind, rain or starvation (Craig 1989: sev eral examples cited in F.berhard 1990 and Hcnschcl & Lubin 1990).
How should web design vary with prey type? To maximize the probability that prey will be captured when a limited amount of silk is produced, a web should have the largest area that does not compromise the efficiency of contact between sveb and insect; this is achieved when the mesh size equals the jirey sic. For the same amount of silk, a smaller web with finer mesh may be required to stop and retain a heavy or fast-flying prey without it breaking through the web (F.bcrtiard 1990) (the words stopping’ and ‘retention’ arc used as defined in the review by F.bcrh«rd 1990 meaning the absorption of the prey energy and the adhesion of the prey until the spider arrises, respec tively». This is because the kinetic energy of the prey is a positive function of its mass and velocity. Thus, a spider’s capture success may depend on the size (area or standard length), weight and Hying speed of llie prcv. Evidence from comparaiive studies suggests that some nocturnal spiders that capture heavy, fast flying prey have webs of relatively narrow mesh (Fbcrhard 1986) but in general, web design is not considered to be fmc-luncd to specific prey types (F.berhard 1990). A Brazilian colonial spider Parawixia bistrkihi Rengger 1936 (Arancidac) has been observed spin ning two distinct types of webs which differed in size and architecture. The question is, can these differ ences be attributed to sanation in prey type avail ability? Parawixia bistrxata is a colonial orb-weaving spider commonly found in cerrado (savanna) vegeta tion in South America.
Its life cycle is univoltinc and dcs-clopnirnt within the colony and the population as a whole is synchronous (Gorgonio 1978: Gobbi 1979; Sandoval 1987; Fowler & Gobbi 1988). Spidcrlings from lbe second to fifth instars typically feed during sunset; after the fifth instar spiders feed only at night, For both activity periods, all member» of each colony spin their individual orb webs simultaneously, term ing a large net of linked orb-wchs. At the end of the feeding period, each spider ingests its own web and retreats to a communal aggregation until the next day (Gobbi 1979; Sandovul 1987). Unusual webs were occasionally spun during the day: their occurrence was temporally correlated with local termite swarms. These ‘daytime’ webs were observed every Septem ber. during a 3-year study of the foraging and social behaviour of P. bistriata Here, die temporal and mor phological characteristics of webs and prey observed at sunset and daytime are compared.
Materials and methods
This study was conducted over 5ha of undisturbed habitat (cerrado vegetation) of P. bistriaia in Itira pina. Siio Paulo, Brazil. The occurrence of each web type of II colonics in .September 1984 and five colonies in September 1985 was counted over several days. These web types were easily distinguished (sec Fig. I). To quantify differences between web types, detailed measurements of web dimensions were taken from randomly selected webs of oik colony in 1986 because, in this colony, all individuals spun sunset webs and most of them also spun daytime webs the following day. Ibis indicates that both web types were spun by the same individuals within a short time period, litis procedure was necessary a» individuals were nor marked because previous attempts had shown that marked individuals may not spin webs the next day. Furthermore, if a different number of spi ders spun w ebs on different days, the cause of change in web design could be attributed to variation in space available for web building. From the webs of this colony web and hub diameters were measured, and the spiral and rndii numbers of 10 randomly chosen webs spun at sunset were counted (see Table 2 for definition of web characteristics.
These measure ments were repeated the next day when the spiders built daytime webs. From these data, die mean mesh size was calculated, defined here as the mean distance between two spiral turns. To calculate mesh size, the web radii minus the hub radii was divided by the number of spiral turns, The total length of silk used per web was calculated by adding the length of all spirals and radii. The size of insects captured was observed and recorded in five colonics (three colonics in 1984 and two colonies in 19851 at sunset, when one of the web types was spun daily. The length and width of the prey trapped in an area of 1 m2 dial comprised several webs, during a period of 30 min was measured All the prey captured in daytime webs of three colonics in 1984 were recorded immediately after termites hеd swan nod. Available prey (flying insects) were sampled with standard sticky traps. Three 40 x 40cm clear plastic sheets coated with polybutanc (an odourless, trans- parent substance i were attached to a pole at heights of 0 5 m. 15 m and 2-5 m. This was the height range where the spiders normally spun their webs. Sticky trap» were about 5 m away from the colony from which web dimensions and observations of prey cap- ture were recorded All the insects were collected from the traps after six I h intervals to estimate diurnal variation in flight activity of prey (from 6.Ш to 20 00 h). This procedure was necessary because preliminary sampling had indicated that the availability of different prey types changed rapidly throughout the day. Trapping was repeated on 4 separate days: 3 days in 1984 (27 September, termite swarms present; 24 and 25 October, termites not present) and I day in 1985 (28 September, termites not prescru). To evaluate the effectiveness of tlic two web types in trapping prey of different sizes the size of prey cap- tured by the webs was compared with the size of those prey caught by sticky traps using a Kolmogorov- Smimov test. Only die samples and obsersations that were taken simultaneously were used for these com- parisons
Results
MIL OF FORAGING ACTIVITY OF SPIDERS AND THEIR PREY Of the II colonics whose sumct webs were counted (90-800 individuals, mean ±SL> = 290012150). Only one colony contained some individuals that did not construct webs. Thus. 92-5% of all spiders observed in September 1984 and 1985 during sunset constructed sunset webs In these colonies, sunset web building began between 16.00 and I7.00h . Sunset occurred at approximately 18 00h The time of foraging activity at sunset coincided with the lime that small prey were most abundant The periodicity of flight activity of these prey was very predictable and varied little between days . In comrast. daytime webs were spun at variable times, always around the peak of termite swarms; usually within 30 mm before or after the termite swarm began (Table II. The variation in time of web-spin ning activity was not owing to variation in age or m/с of spiders as different colonies of P. bistiiaia have synchronous development (Sandoval 1987) All ter mite swarms observed occurred during rain Unlike the small (lies, termites had variable periodicity of flight activity. Nonetheless, the spiders were able to track their availability over time The two types of webs were never observed to occur simultaneously and daytime webs were ingested before the sunset webs were spun. Termites were never observed dur ing sunset and daytime webs were never observed in the absence of termite swarms.
WEB DESIGNS
Daytime and sunset webs differed significantly in a number of key dimensions in tlic colony studied and could easily be distinguished by eye. The mesh size of daytime webs was. on average, three times larger than sunset webs (Fig. I. Table 2). Mesh enlargement resulted from both an 82% increase in web diameter and a 57% decrease in the number of spiral turns. Tlvcic was no overlap in these dimensions for the two web types The amount of silk did not differ between the two web types (Table 2). Because the number of radii and spirals were smaller and the area was larger in daytime webs than in sunset webs, then, all else being equal, daytime webs were probably weaker for stopping prey. Although web measurements were taken from a single colony, the dimorphism in mesh si/c and diameter was observed m 10 other colonies (Table 11. Ibis dimorphism was not a result of varia tion in sie, age or individual variation in behaviour within colonics because the development within colonies was extremely synchronous (Sandoval 1987) and all individuals spun their webs at both activity periods when webs were measured Furthermore, daytime webs wen; unique to day activity and sunset webs were unique to sunset activity. ‘Thus, it is clear that web dimorphism ill P. bixtriata is the result of plasticity in the web building behaviour of each indi vidual.
