rized in Fig. 13.1 (for a more detailed
description, see van Lenteren et al., 2003).Direct effects of released organism on other
organisms in ecosystemDirect effects of releases may be extinction
or reduction in numbers of native non-tar-
get species. Certain categories of insects,
such as hymenopteran and dipteran para-
sitoids, are extremely specific in host use
and attack only one or a few species within
an insect genus. Biocontrol scientists have
therefore concluded that the introduction of
these specific agents would not endanger
the native fauna, contrary to the introduc-
tion of generalist vertebrate predators (e.g.
Harris, 1990). Polyphagous insect predators
may, however, attack non-target organisms
and be harmful. Howarth (1985, 1991) pre-sents circumstantial evidence that natural-
enemy introductions may have led to a
reduction of non-target species, but others
(e.g. Funasaki et al., 1988) believe these
charges to be unjustified.
It is not an easy task to show that, in an
ecosystem with a rare plant-eating insect
species that is attacked by many different
natural-enemy species, the introduction of a
new natural enemy further reduces the den-
sity of the herbivore or merely replaces
some other mortality factor. The extinction
of pest or non-target organisms as a result
of biological control is, however, extremely
unlikely. Pests have seldom, if ever, been
exterminated in the more than 100 years of
insect biological control. Rather, a low pop-
ulation level of both pest and natural
enemy has developed, as in natural ecosys-
tems. In nature, it is the rule, rather than the
exception, to find extremely low densities of198 J.C. van Lenteren et al.Step 1
Petri-dish non-choice black-box test
Are non-target species attacked?Step 2
Petri-dish non-choice behavioural test
Are non-target species attacked?
Step 3
Petri-dish choice test
Are non-target species attacked?Step 4
Large-cage choice test
Are non-target species attacked?Step 5
Field test (if possible, see text)
Are non-target species attacked?YesYes, at constant fixed rate of attackYes, easy attack or switch
in host rangeYes, easy attack or switch
in host range
High hazard ratingYes
High hazard ratingNoNo, or only at end of observationNo, or low attack rate and
no shift in host preferenceNo, or low attack rate and
no shift in host preferenceNoLow hazard ratingLow hazard ratingLow hazard ratingLow hazard ratingLow hazard ratingFig. 13.1.Sequential host-specificity testing scheme.