Many ecological studies use Two-Term Local Quadrat Variance Analysis (TTLQV) and its derivatives for spatial pattern analysis. Currently, rules for determining variance peak significance are arbitrary. Variance peaks found at block size 1 and at > 50 % of the transect length are the only peaks whose use is explicitly prohibited. Although the use of variance peaks found at block sizes > 10 % of the transect length have also been warned against, many researchers interpret them regardless. We show in this paper that variance peaks derived from TTLQV are subject to additional ‘rules of thumb’. Through the use of randomization and permutation analyses on real and simulated data of species abundance in contiguous plots along a single transect, we show that variance peaks found at block sizes 1, 2 and 3 occur frequently by chance and thus likely do not indicate biologically meaningful patterns. The use of multiple replicate transects decreases the probability of Type II error.
Campbell, Jonathan E., Franklin, Scott B., Gibson, David J. and Newman, Jonathan A. "Permutation of Two-Term Local Quadrat Variance Analysis: General Concepts for Interpretation of Peaks." (Feb 1998).