Research Activities - Quantifying Microbial
Activity
Diversity Measures for Quantifying Microbial Activity
General problem: Is
the biodiversity of soil microbial populations important for the
maintenance of a healthy soil? Agricultural productivity depends
on the organisms that inhabit the soil. Different forms of agriculture
have very different effects on the soil’s complex ecology by impacting
on soil fertility, the cycling of inorganic compounds and the development
of soil structure.
Experimental technique: Soil
microbiologists have started using a metabolic fingerprinting technique,
available commercially as a BIOLOG kit, to build profiles of whole soil
microbial communities. A microtitre plate with 96 wells and each
containing a different carbon substrate provide information on the
capacity of the microbes present to use the range of carbon sources. Soil
microbiologists have started using a metabolic fingerprinting technique,
available commercially as a BIOLOG kit, to build profiles of whole soil
microbial communities. A microtitre plate with 96 wells and each
containing a different carbon substrate provide information on the
capacity of the microbes present to use the range of carbon sources.
Specific problem: Descriptive
data analysis techniques, particularly principal component analysis, have
been applied to investigate functional similarities or dissimilarities
across communities from different environmental conditions, using data
from BIOLOG kits. However, with microbiologists now interested in the
diversity of substrates used by microbes in the soil, a single measure
allowing comparison among microbial communities was seen as an important
quantitative tool for better understanding how microbial diversity
contributes to soil health.
Recomendations:
- Use of the Gini coefficient (Harch et al.,
1997; J. Mic. Meth. 10:33-36) as a measure of functional diversity.
Although Gini is highly correlated with other measures commonly used
in ecology (eg Shannon; r=0.967), it is preferred in this context
because of its graphical interpretation – twice the area between the
diagonal and curve, plus depicting substrate richness and evenness.
- Use of the Gini coefficient in conjunction with
hypothesis testing techniques provides scientists with different, but
complementary information to that given from descriptive data
analysis.
Other Relevant Applications: The
BIOLOG kit has also been used for assessing water quality in freshwater
and marine systems. For example, like the work done in this project we can
pose the question: "Is the biodiversity of water-borne microbial
populations important for the maintenance of a healthy river or marine
system?
Acknowledgments: This
work has been done in collaboration with Clive
Pankhurst, Clive Kirkby, Stephen Neate (CSIRO
Land & Water) and V.V.S.R. Gupta (CRC for Soil & Land
Management).
Other Areas
of Research for Analysing Microbial Activity:
Variable Selection: As is often the case when
investigating soil microbial communities, there are many more variables
measured on each sample, than is strictly necessary (95 different Carbon
substrates). This has led to the proliferation of papers on substrate
utilisation using dimension reduction techniques, such as principal
component analysis and correspondence analysis.
As criticised by Krzanowski (Applied Statistics 36
(1987):22-33), the major drawback of only using these dimension reduction
techniques is that while the dimensionality of the space may be reduced
from the original 95 substrates to say 5 principal components, all the
original 95 substrates are, in general, still needed in order to define
the new principal component variables.
This issue of reducing dimensionality is prevalent in
the literature through identification of the ‘main’ carbon substrates
contributing to community differences.
Krzanowski’s variable selection method, based on
Procrustes Analysis, will be contrasted with other methods that have been
reported in the Biolog® literature, namely: ordering F-statistics from
one-way analysis of variance and sorting eigen vectors (variable/substrate
weightings) from principal component analysis. Assessment has been based
on how well the identified subsets of Carbon substrates reproduce, as
closely as possible, the general features of the complete set of 95
substrates.
Three-mode data analaysis: Three-way statistical
methodology has been developed and applied extensively in the fields of
psychology, chemistry, agriculture and many other disciplines. We are
implementating this methodology in Genstat, specifically for ordination of
carbon substrate utilisation data from soils. The data is in the form of a
three-mode, three-way array containing soil sample by carbon substrate by
time data.
While stand-alone software is readily available for
analysing three-way data matrices (eg TUCKALS,
PARAFAC), to our knowledge mainstream statistical packages do not
incorporate procedures for using such statistical techniques. We aim is to
demonstrate it’s implementation in Genstat and how this technique can
provide a description of the main patterns in the data.
Contact: Allan Adolphson
Ph: +61-(0)2-9325-3261 Fax: +61-(0)2-9325-3200
|
To
top