here the use of multiple control and impact areas incorporates greater natural spatial variability, which can broaden the interpretation of results. beyond BACI can detect a greater variety of disturbance than the BACIPS design described above(underwood1991).underwood(1992)modelled press disturbance(permanent shifts of the environment to altered states) and pulse disturbance(discrete events where conditions return to pre-disturbance levels)of different sizes. an extreme pulse disturbance, which reduced population abundances by 50% and 100%, was modelled. in both cases, the disturbance as modelled produced a significant interaction between sampling times and the difference between impact and control sites. there was no interaction among times of sampling and locations before the disturbance. there was also no interaction among control location and time of sampling after the pulse disturbance was applied. underwood(1992)concluded that the changes observed in the impacted site were not matched by any change in controls. the impacted sites clearly exhibited a different pattern from that recorded at these sites before disturbance began. underwood and stewart-oaten et al. recognize that these designs are not flawless(e.g. serial correlation)and agree upon the benefits of multiple control sites.