distinction between alpha and beta diversity is thus not used
hereafter.
It is important to take into account the sensitivity
thresholds related to the techniques used to characterize
microbial diversity. Techniques related to PCR can in theory
provide a signal, e.g., a band detectable on a gel, from a
targeted sequence in the reaction volume if it yields at least
one billion fragments after 30 amplification cycles. However, several factors diminish the amplification yield below
the theoretical level. The first is the resistance of the polymerase used. The latter is a thermostable enzyme but it must
undergo cycles of heating at the limits of its resistance, and
part of the molecules is thus lost during each cycle. The
second is the presence of polymerase inhibitors: like any
enzyme, Taq polymerase functions optimally in the environment proposed by the supplier, but factors such as the presence of soil humic acids (Watson and Blackwell 2000) or
damaged nucleotides (Jenkins et al. 2000) are known to
decrease the efficiency of amplification. Finally, although
there are thousands of bacterial species per gram of soil
(Torsvik et al. 1990) or other complex environments, the
literature does not indicate the presence of more than a few
tens of taxa in most collections of amplicons. Indeed, in a
given reaction mixture, some DNA sequences with low GC are
likely to be more rapidly amplified and will thus eventually
monopolize the Taq polymerase. Overall, it is estimated that
below 1 % of initial abundance, a taxon cannot be detected
even if the experience of intensive sequencing for complex
environments has never been attempted convincingly. This is
the objective of ongoing projects (Vogel et al. 2009).
Beyond this intrinsic limit, a wide range of methods exist
today to characterize microbial diversity, each method
apprehending a different aspect of diversity.
8.4.1 Approaches Targeting Phenotype
Diversity
8.4.1.1 Cultural Approaches
For a long time, diversity studies were performed using
cultivation techniques, which implies the isolation and cultivation of microorganisms. Counting techniques, such as the
most probable number (MPN), or the determination of usage
patterns of growth substrates such as done with BIOLOG
(Bossio and Scow 1998), have been developed to quantify
microorganisms and estimate their metabolic profiles.
Although these methods have provided the fundamental
basis for the study of microbial communities in their environment (Garland and Mills 1991), they have some
limitations, the main ones being:
(i) Representativeness: Media and culture conditions
impose constraints such that only a small fraction of
microorganisms from a given environment can be
cultured. According to some estimates, the cultivable
diversity represents only 0.1–1 % of the actual bacterial
diversity (Amann et al. 1995).
(ii) Identification: These methods require to further characterize the bacteria by cultivating them to better know
their taxonomic affiliation.
In particular, it is recognized that the general use of rich
media can strongly bias the type of bacterial strains able to
grow as compared to the taxa most common in natural, often
oligotrophic environments. The use of a range of selective
media may allow better access to a range of microbial
diversity, but the multiplication of media may prove to be
time-consuming and tedious. The development of new selective media and of automatic approaches using robots should
partly alleviate these limitations and allow the isolation and
characterization of new bacterial species in the future, which
would help quantifying the actual level of microbial diversity in situ.
8.4.1.2 Approaches Targeting the Diversity
of Metabolites
Lipids
Different lipid classes can be targeted to assess microbial
diversity. The most widely used are:
(i) The phospholipid fatty acids (PLFA) characterized by
chemical analysis and/or molecular isotopy (Frostegard
Box 8.1 (continued)
strains (e.g., number of fungal propagules/g soil), replicate measurements are required for each test parameter
for each different soil treatment. To compare the frequency of highly aggressive strains for different
treatments, several options are available. For each sample replicate of each treatment, the proportion of strains
of a given level of aggressiveness is calculated and the
differences can be evaluated with parametric tests. An
alternative option would be to characterize a random
sample of strains for 2 or 3 levels of aggressiveness (low
vs. strong, or no, intermediate, high) and then compare
the bi- or trinomial frequencies obtained with nonparametric tests. It would be necessary to estimate the effect
of sample size on the power of statistical tests for these
two options in order to assess which analysis strategy is
the most efficient in terms of time and labor involved.
For a basic statistical approach such as an analysis of
variance and a χ
2
, such estimates are easily obtained
with statistical tools.
8 Biodiversity and Microbial Ecosystems Functioning
267
hereafter.
It is important to take into account the sensitivity
thresholds related to the techniques used to characterize
microbial diversity. Techniques related to PCR can in theory
provide a signal, e.g., a band detectable on a gel, from a
targeted sequence in the reaction volume if it yields at least
one billion fragments after 30 amplification cycles. However, several factors diminish the amplification yield below
the theoretical level. The first is the resistance of the polymerase used. The latter is a thermostable enzyme but it must
undergo cycles of heating at the limits of its resistance, and
part of the molecules is thus lost during each cycle. The
second is the presence of polymerase inhibitors: like any
enzyme, Taq polymerase functions optimally in the environment proposed by the supplier, but factors such as the presence of soil humic acids (Watson and Blackwell 2000) or
damaged nucleotides (Jenkins et al. 2000) are known to
decrease the efficiency of amplification. Finally, although
there are thousands of bacterial species per gram of soil
(Torsvik et al. 1990) or other complex environments, the
literature does not indicate the presence of more than a few
tens of taxa in most collections of amplicons. Indeed, in a
given reaction mixture, some DNA sequences with low GC are
likely to be more rapidly amplified and will thus eventually
monopolize the Taq polymerase. Overall, it is estimated that
below 1 % of initial abundance, a taxon cannot be detected
even if the experience of intensive sequencing for complex
environments has never been attempted convincingly. This is
the objective of ongoing projects (Vogel et al. 2009).
Beyond this intrinsic limit, a wide range of methods exist
today to characterize microbial diversity, each method
apprehending a different aspect of diversity.
8.4.1 Approaches Targeting Phenotype
Diversity
8.4.1.1 Cultural Approaches
For a long time, diversity studies were performed using
cultivation techniques, which implies the isolation and cultivation of microorganisms. Counting techniques, such as the
most probable number (MPN), or the determination of usage
patterns of growth substrates such as done with BIOLOG
(Bossio and Scow 1998), have been developed to quantify
microorganisms and estimate their metabolic profiles.
Although these methods have provided the fundamental
basis for the study of microbial communities in their environment (Garland and Mills 1991), they have some
limitations, the main ones being:
(i) Representativeness: Media and culture conditions
impose constraints such that only a small fraction of
microorganisms from a given environment can be
cultured. According to some estimates, the cultivable
diversity represents only 0.1–1 % of the actual bacterial
diversity (Amann et al. 1995).
(ii) Identification: These methods require to further characterize the bacteria by cultivating them to better know
their taxonomic affiliation.
In particular, it is recognized that the general use of rich
media can strongly bias the type of bacterial strains able to
grow as compared to the taxa most common in natural, often
oligotrophic environments. The use of a range of selective
media may allow better access to a range of microbial
diversity, but the multiplication of media may prove to be
time-consuming and tedious. The development of new selective media and of automatic approaches using robots should
partly alleviate these limitations and allow the isolation and
characterization of new bacterial species in the future, which
would help quantifying the actual level of microbial diversity in situ.
8.4.1.2 Approaches Targeting the Diversity
of Metabolites
Lipids
Different lipid classes can be targeted to assess microbial
diversity. The most widely used are:
(i) The phospholipid fatty acids (PLFA) characterized by
chemical analysis and/or molecular isotopy (Frostegard
Box 8.1 (continued)
strains (e.g., number of fungal propagules/g soil), replicate measurements are required for each test parameter
for each different soil treatment. To compare the frequency of highly aggressive strains for different
treatments, several options are available. For each sample replicate of each treatment, the proportion of strains
of a given level of aggressiveness is calculated and the
differences can be evaluated with parametric tests. An
alternative option would be to characterize a random
sample of strains for 2 or 3 levels of aggressiveness (low
vs. strong, or no, intermediate, high) and then compare
the bi- or trinomial frequencies obtained with nonparametric tests. It would be necessary to estimate the effect
of sample size on the power of statistical tests for these
two options in order to assess which analysis strategy is
the most efficient in terms of time and labor involved.
For a basic statistical approach such as an analysis of
variance and a χ
2
, such estimates are easily obtained
with statistical tools.
8 Biodiversity and Microbial Ecosystems Functioning
267