1975] 
Tolbert — Avoidance Behavior in Orb Weavers 
45 
At times no response was observed even with repeated presentation 
of the model. 
Differences in Component Use 
Any insect or arachnid that approaches an orb web falls into one 
of four categories: (i) potential prey, (2) potential predator or 
parasite, (3) potential mate and (4) “neutral” (unpalatable prey, 
too large or strong an animal for the spider to subdue, etc.). An 
orb weaver that responds incorrectly may fail to obtain sufficient 
food, may be killed or may fail to mate. While this study does not 
identify the specific cues involved in these decisions, one would not 
expect all spiders of the same species or the same individual at dif- 
ferent times to respond identically to a standard stimulus. The vari- 
ous response components are analyzed below under a variety of con- 
ditions to delineate their use. 
Dorsal vs. ventral model presentation 
A. aurantia and A. trifasciata were tested separately for differ- 
ences in response components elicited by dorsal vs. ventral presenta- 
tions of the simulated predator. A chi-square test (Conover, 1971) 
was employed in the analyses and unless otherwise stated this is the 
test used throughout the paper. The response of A. aurantia differed 
significantly (n = 94, x 2 = 41.23, p<o.OOi) with dorsal and 
ventral presentations (fig. 4). Stilting and web flexing followed by 
stilting accounted for over 50% of the above variability. They 
occurred with substantially greater frequency than expected when 
the model was presented ventrally. Switching sides of the web fol- 
lowed by web flexing occurred more often than expected on dorsal 
presentations. A. trifasciata also exhibited a significant difference 
(n = 1 8 1 , x 2 = 102.84, p < 0.001) in response components. Stilt- 
ing again proved to be primarily a response to ventral presentations 
of the model. Web flexing, moving away from the hub and rebuff 
occurred much more often than expected when spiders were touched 
dorsally (figs. 4 and 5). It should be mentioned that the majority 
of responses (>95%) occurred after the model touched either the 
web or the spider with less than 5% occurring on the actual ap- 
proach. Due to the small number of responses to the actual approach 
of the simulated predator these data are included, without distinction, 
in the various statistical tests. Also, the term “presentation” refers 
to both approach and touch responses. 
