This first-attempt study used constructed bacterial consortia containing Escherichia coli DH5 alpha, (a weak decolorizer) and its UV-irradiated mutants (E. coli UVT1 and UV68; strong decolorizers) via equilateral triangle diagram and mixture experimental design to assess color removal during species evolution. The results showed that although strain DH5 alpha was not an effective decolorizer, its presence might still played a significant role in affecting optimal color removal capabilities of mixed consortia (e.g., E. coli DH5 alpha, UVT1 and UV68) for two model azo dyes-, namely, reactive red 22 (RR22) and reactive black 5 (RB5). Contour analysis of ternary systems also clearly showed that decolorization of RR22 and RB5 by DH5 alpha-containing active mixed consortia was more effective than mono-cultures of the stronger decolorizer alone (e.g., UVT1). The optimal composition of the mixed consortium (UV68, UVT1, DH5 alpha) achieving the highest specific decolorization rate was (13%:58%:29%) and (0%:74%:26%) for decolorization of RR22 and RB5, respectively, with initial total cell density fixed at OD600 = 3.5 +/- 0.28. (c) 2006 Elsevier B.V. All rights reserved.