Underwood (1994)

Underwood, A. J. 1994. On beyond BACI: sampling designs that might reliably detect environmental disturbances. Ecological Applications 4(1):3-15.

 Much sampling to detect and quantify human environmental disturbances is flawed by a lack of appropriate replication. BACI (Before-After-Control-Impact) designs have only a single control location, and any conclusions from them are illogical. Asymmetrical designs using one putatively impacted and several control locations can reliably detect a variety of environmental impacts, including those that do not affect long-run mean abundances, but do alter temporal variance. When abundances of populations in different locations show temporal interaction, the asymmetrical designs are also extendable to sample at hierarchical spatial and temporal scales.
     The power of tests using asymmetrica designs is great for non-interactive sets of abundances, but greatest for pulse (short-term) responses to disturbances, large alterations of temporal variance, or combinations of sustained, press responses in mean abundance coupled with altered temporal heterogeneity. Power in temporally interactive sets of data is generally poor.
     Alternatives to pre-disturbance sampling, including generalized assessment of spatial and temporal variances and experimental impacts, may provide better guidance for detection of human disturbances.