Dear everyone especially Jan, Paul and Bjarke
Perhaps I can add another two cents to the discussion on quantitisation in
biofilm research. I agree almost completely with what has been said however
I think you have all missed one extremely important element in biofilm
research, indeed in all microbial ecological research. This is of course
the part that chance plays in colonising a surface. Remember Jacques Monod
'Hazarde et Necessité'.
If you were to take 100 identical clean surfaces and expose to a group of
ten different organisms, some of which interacted positively with other
members, some which were inhibitory; incubated each system under constant
environmental conditions how reproducible would the final biofilm become??
I suggest not very.
Many things will influence the outcome of this experiment. Just for a start
the growth phase of the inoculum. An exponential or random population might
attach differently depending on whether it had just hatched or was about to
procreate. This might dictate if 2 cells attached next one another, which
would start to grow first. As our cellular automaton model showed, even
identical cells will behave differently depending on the substrate
distribution. As soon as one cell starts to grow it will deplete a zone
around it reducing the ability of a neighbouring cell to grow.
Much more important, and a well known ecological principle, is the size of
the landscape. Thus the smaller the landscape the lower the diversity of
interacting communities. So what is the critical landscape size which would
allow all our 10 organisms to grow and survive? What questions could we ask
as we gradually increased the surface area for attachment? Well we could
decide how many and which survive, the ratio of each species one to another
or one to the whole population. We could monitor who lives with who and the
size of the clusters. We could determine 'free' area with nothing attached
and growing. We could measure pore size if present, spacing between stalks
etc. etc.
Now we have a possible experiment. Obviously if we gather data for an
infinite landscape we will obtain an average constant ratio of each species
present and all our measurements mentioned above will asymptote to such
values. The question becomes, how big a landscape is needed for this to
occur?
So at least we can determine some quantitative data this way.
On pattern formation, I think Paul could be right in thinking that fractal
analysis could give some clues. I was thinking about all this in my car on
the way into work. Take an acre of empty field and let a random set of
seeds grow without touching the field. We will see colonies of plants great
humps of brambles, young trees and so on. If we had a hundred such fields
we would see the same 'pattern' though the distribution of plants would be
entirely different. Take an oak tree or more. Each tree is quite different
yet is *obviously* still an oak tree. So we recognise it by a range of
fundamental attributes, branching pattern, bark texture and leaf shape to
name a few.
So in the biofilm we need to identify the pattern forming attributes. I
guess presence of clusters, presence of pores, stalks and mushrooms etc.
Such *significant* attributes could be put into a neural network and used
as a recognition system.
On interactions life is obviously hard when it comes to a complex
community. Much can be learned from the newer techniques, fluorescent
probes etc. reporter genes. gfp etc., though I still believe that
individual interactions ought to be determined in simpler laboratory
models. This would give good quantitative data, rate constants, affinity
constants, yields, inhibition constants, chemical signals, the effects of
environmental physico-chemical characteristics on rates etc. A simple,
model I would recommend is to put interacting partners into a
one-dimensional gel-stabilised system so that the environmental chemistry
can be investigated easily by for example taking gel cores, slicing them
and measuring things. I realise that you young things will regard some of
these techniques as rather quaint......! Other possibilities include
chemostats, though the relevance of chemostat data to attached
heterogeneous populations is open to question. Of course the transfer of
any quantitative data derived in this way to the in vivo system is open to
even more questions but it may be the best thing we have.
Just to finish, I want reiterate that we *must develop* good statistical
tools to cope with the random chance behaviour of microbial communities.
But we also need a more general understanding that these stochastic effects
are important. Clearly we also need more and more sophisticated
investigative techniques *including* simplified models plus mathematics to
answer some of the questions Jan, Paul and Bjarke raised!
Its been a privilegeŠŠŠŠŠ
Julian
PS That was about three cents worth!
PPS Thanks Bob for starting this hare.
**********************************************************
Professor Julian Wimpenny
School of Pure and Applied Biology
Cardiff University
Cardiff CF1 3TL
Wales, UK
tel: +44 (0) 1222 874974
fax +44 (0) 1222 874305
email: WimpennyJ at cf.ac.uk
**********************************************************
-------------------------------------------------------------------
To reply to the group as well as to the originator, make sure that
the address biofilms at net.bio.net is included in the "To:" field.
See the BIOFILMS homepage at http://www.im.dtu.dk/biofilms for info
on how to (un)subscribe and post to the Biofilms newsgroup.
