368 
Psyche 
[Vol. 94 
to move. Some undetected environmental and/or physiological fac- 
tors must contribute to the decision to relocate. 
The two categories of residence times used in comparison of feed- 
ing rates were based on the average residence time of 6.7 days. 
Spiders staying at web sites for longer periods than six days had 
significantly higher feeding rates than those staying at web sites for 
five days or less. Olive (1982) has similarly shown that Argiope spp. 
moved more often when current feeding levels were decreased, and 
Vollrath (1985) observed that Nephila clavipes in poor environ- 
ments changed sites more often than spiders in rich environments. 
Martyniuk (1983) found that filmy-dome spiders (Linyphiidae) 
which were moved to previously abandoned web sites of low prey 
availability and were provided with supplemental prey, remained at 
those sites. 
Janetos (1982) has designated sheet-web weavers as “sit-and-wait” 
predators and orb weavers as “active” foragers. He assumed that 
orb webs could be put nearly anywhere, resulting in high variance in 
payoffs. Therefore, he predicted that orb weavers would do better 
by moving frequently among sites to find potential “hot spots.” The 
assumption that orb webs can be put anywhere may not be correct. 
Several studies have shown that M. gracilis requires specific structu- 
ral characteristics for web construction (Biere 1977; Hartsock 1983; 
Hodge 1985). Neither does M. gracilis follow Janetos’s (1982) pre- 
diction of frequent movement among sites. Studies of M. gracilis 
and M. schreibersi (Perty) (Shelly 1984; Hodge 1985) have shown 
that these spiders often remain at web sites for 15 days or more. This 
contests the generalization that orb weavers are “active” foragers. 
If spider web site relocations occur at random, then the expected 
distribution of intervals between relocations will follow a negative 
exponential distribution, which is a Poisson process (Bailey 1964; 
Ross 1970). If so, then the stimuli that induce a spider to leave a web 
site occur with a constant small probability in any short period of 
time. Such stimuli could be physical disturbance of the web due to 
climatic factors, or predation attempts (Janetos 1982). Comparison 
of the actual distribution of residence times with that expected from 
the random movement hypothesis indicates that M. gracilis does not 
move at random. A previous study of macrohabitat selection by M. 
gracilis found that spiders moved with a random pattern until they 
encountered a macrohabitat where environmental factors are ame- 
liorated (Hodge 1988). This macrohabitat was a deciduous forest 
