Host shifts in biological weed control: Real problems, semantic difficulties or poor science?

Many biologists perceive organisms as constantly evolving and therefore consider the host plant ranges of biological control agents as labile. Host plant ranges are thus likely to undergo adaptive change should environmental conditions change, for example following successful biological control. As a consequence, the introduction of biological control agents against weeds is considered by many to be an inherently unsafe practice with non-target plants at risk of attack. However, despite the introduction of over 600 insect species from one geographic region to another for biological weed control during this century, there are relatively few documented cases of changes in host plant range. Purported instances are discussed in relation to behavioural and genetic concepts. It is concluded that apparent additions to the host range can, in all of the cases examined, be explained in terms of established behavioural concepts of preadaptation, threshold change resulting from host deprivation, and effects of experience (learning). The inappropriateness of the often-used term 'host shift' to describe these cases is demonstrated, and it is concluded that evidence from biological weed control contradicts some aspects of ecological and evolutionary theory.