The parasitoid fraction is defined for a given set of arthropods as th
e number of parasitoid individuals divided by the number of all arthro
pod individuals. The hymenopteran parasitoid fraction was computed for
arthropods caught in sticky traps on islands at two sites in the Baha
mas. At Staniel, twenty islands were cross-classified by area (large,
small) and distance (near, far). At Abaco, twenty-one islands were cro
ss-classified by area (large, small) and lizard treatment (lizards occ
urring naturally and unmanipulated, lizards not occurring naturally bu
t experimentally introduced, lizards not occurring naturally and unman
ipulated). The parasitoid fraction was larger for large islands and la
rger for near islands wherever this could be tested. Range of variatio
n was especially great at Staniel, from 0.13 (small-far) to 0.48 (larg
e-near). Absolute number of parasitoids, the numerator of the parasito
id fraction, was larger for large islands at Abaco but not at Staniel;
however, it was somewhat larger for near islands at Staniel. At Abaco
, parasitoid fractions and absolute numbers were highest overall for i
slands having lizards naturally and lowest for islands not having liza
rds naturally; lizard-introduction islands were intermediate. A relati
ve-attenuation hypothesis, in which parasitoids and other arthropods b
oth diminish with decreasing immigration or increasing extinction but
the former does so relatively more strongly, is partly supported by th
e Staniel distance and Abaco area data. Food-web-interaction hypothese
s may explain the other results: a differential reduction of other art
hropods compared to parasitoids on large islands fits the Staniel area
data, and a negative effect of lizards on parasitoid predators fits t
he Abaco treatment data. However, a tendency for islands favorable for
lizards to be favorable for parasitoids seems at least an equally pla
usible explanation for the latter result.