The methodological and computational effort the authors undertook to prove bias is impressive, and they address a frequently mentioned concern: the circularity involved in the use of indicator values derived from species lists.

According to both Ellenberg (1974) and Landolt (1977), indicator values cannot be measured in the field, but only compared with field data. Second, Zeleny & Schaffers (2012) address circularity as a source of bias.

They find that 'to confirm the risk of this circularity in real analysis is, however, nearly impossible.
This involves a comparison of three matrices, where one of these (the plot by indicator value matrix) is a matrix product of the other two (plot by species, species by indicator values), a classical case of circularity, preventing statistical inference as proposed by Zeleny & Schaffers (2012).

There has been criticism concerning the inherent circularity and the lack of statistical calibration against measured environmental variables. While the circularity issue has been clarified and a conservative method for evaluating correlations has been devised (Zeleny & Schaffers 2012), the calibration issue has only been solved in an exemplary fashion for subsets of species that are too small to allow wider application (Ertsen et al. 1998). Wamelink et al. (2002) pointed out that mean EIVs should only be compared within broad vegetation types such as phytosociological classes (cf. Wamelink et al. 2003; Witte & von Asmuth 2003).

Заранее благодарю за ответ

в ссылке Zeleny & Schaffers 2012 говорится как раз о " circularity of reasoning"
"....Considering the use of EIVs in vegetation analysis, an often
recognized problem is the **circularity of reasoning when
attempting to explain vegetation pattern using EIVs, as
these EIVs were themselves derived from empirical experience
with vegetation..."