We present a study of colony transformations during growth of Bacillus
subtilis under adverse environmental conditions. It is a continuation
of our pilot study of ''Adaptive self-organization during growth of b
acterial colonies'' (Physica A 187 (1992) 378). First we identify and
describe the transformations pathway, i.e. the excitation of the branc
hing modes from Bacillus subtilis 168 (grown under diffusion limited c
onditions) and the phase transformations between the tip-splitting pha
se (phase F) and the chiral phase (phase C) which belong to the same m
ode. This pathway shows the evolution of complexity as the bacteria ar
e exposed to adverse growth conditions. We present the morphology diag
ram of phases F and C as a function of agar concentration and pepton l
evel. As expected, the growth of phase F is ramified (fractal-like or
DLA-like) at low pepton level (about 1 g/l) and turns compact at high
pepton level (about 10 g/l). The growth of phase C is also ramified at
low pepton level and turns denser and finally compact as the pepton l
evel increases. Generally speaking, the colonies develop more complex
patterns and higher micro-level organization for more adverse environm
ents. We use the growth velocity as a response function to describe th
e growth. At low agar concentration (and low pepton level) phase C gro
ws faster than phase F, and for a high agar concentration (about 2%) p
hase F grows faster. We observe colony transformations between the two
phases (phase transformations). They are found to be consistent with
the ''fastest growing morphology'' selection principle adopted from az
oic systems. The transformations are always from the slower phase to t
he faster one. Hence, we observe F -> C transformations at low agar co
ncentrations and C -> F transformations at high agar concentrations. W
e have observed both localized and extended transformations. Usually,
the transformations are localized for more adverse growth conditions,
and extended for growth conditions close to the boundaries between mor
phologies. We have observed also transformations between different bra
nching modes, as well as transformations via virtual states. Motivated
by the contemporary knowledge about phages and plasmids, we postulate
a theoretical framework to comply with our experimental findings. We
explain our observations using these assumptions as well as our propos
al of co-mutations and auto-catalytic mutations as presented in the ab
ove mentioned pilot paper. This theoretical framework is a part of the
new evolving picture of genome cybernetics. We also discuss the conce
pt of adaptive genome changes which are based on pre-existing knowledg
e as well as the concept of genetic learning, i.e. changes (in respons
e to a new problem) which develop the potential for adaptive genome ch
anges. These concepts follow naturally if the picture of genome cybern
etics is accepted. We conclude with a discussion of the implications a
nd with further predictions (to be tested experimentally) derived from
our assumptions.