Upload
others
View
6
Download
0
Embed Size (px)
Citation preview
Interactions between plants and antagonistic streptomycetes
A DISSERTATION SUBMITTED TO THE FACULTY OF THE GRADUATE SCHOOL
OF THE UNIVERSITY OF MINNESOTA BY
Matthew Gene Bakker
IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY
Linda L. Kinkel, Advisor
June, 2011
© Matthew Bakker 2011
i
Acknowledgements
Many people contributed to this work. I thank my advisor, Linda Kinkel, for her support
and direction, and for suggesting new perspectives and hard revisions that have
substantially improved this work. Laura McCarville, David Manning, AJ Lange, and
Lindsey Hanson provided invaluable laboratory support. The late Peter Graham provided
valuable guidance and perspective prior to his untimely death. This work has been made
possible through financial support from a variety of sources. In particular, I would like to
acknowledge the National Science Foundation's Graduate Research Fellowship Program,
the University of Minnesota's Graduate School Fellowship for incoming students and
Doctoral Dissertation Fellowship for students in their final year. Work reported in
Chapter 2 was partially funded by an award from the Land Institute. The Harvey Fellows
Program of the Mustard Seed Foundation offered me tremendous flexibility in my final
years, allowing for enriching experiences such as an extended visit to a research group in
France. Support from the Department of Plant Pathology at the University of Minnesota
for an international internship is gratefully acknowledged. I thank Eriko Takano and
Yves Dessaux for hosting me during visits to their lab groups. I thank Christine Salomon
for graciously guiding my attempts to bring research techniques from chemistry to bear
on my research questions, and for opening her laboratory to me.
Finally, thanks to Erica for encouraging me to pursue this path, and for tolerating a
strange schedule and an unsettled life, for a time.
ii
Table of Contents
Page
List of Tables iii
List of Figures v
Chapter 1 - Introduction 1
Chapter 2 - Streptomycete communities in virgin prairie soil 27
vs. never tilled, no-till annual monoculture
Chapter 3 - Accounting for sequencing errors during processing 47
of 454 pyrosequence data
Chapter 4 - Impacts of plant host and plant community richness 69
on soil Actinobacterial community structure
Chapter 5 - Do antagonistic streptomycetes play a role in 101
plant-soil feedbacks?
Bibliography 136
Appendix - Plants as modulators of antibiotic production by 163
streptomycetes
iii
List of Tables
Page
Chapter 2
Table 1 - Inhibition data for isolates belonging to OTUs with 45
at least ten members, showing differences among prairie and
monoculture treatments
Table 2 - Inhibition data for isolates belonging to OTUs with 46
at least ten members, showing differences among OTUs
Chapter 3
Table 1 - Number of sequences failing quality screening criteria 68
and total number of sequences remaining, for standard processing
pipeline and for PyroNoise processing
Chapter 4
Table 1 - ANOVA results tables showing dependence of 93
Actinobacterial density, richness and diversity upon plant community
richness and host plant species identity
Table 2 - Mean pairwise Actinobacterial community dissimilarity 94
by host plant sampled
Table 3 - Correlation coefficients for relationships between 95
Actinobacterial community characteristics and various plant community
and soil edaphic characteristics
Table 4 - Variation in soil edaphic characteristics across two levels 96
of plant community manipulation
Table S1 - Sequence yield, observed and estimated OTU richness, 97-98
OTU diversity and culturable Actinobacterial density for 60 soil samples
Table S2 - Actinobacterial OTUs that are indicative of particular 99-100
plant hosts or plant community richness treatments
iv
List of Tables, continued
Page
Chapter 5
Table 1 - ANOVA results table showing the significance of host 132
plant species, plant richness, and the interaction between host species
and plant richness on various measures of the antagonistic potential
of associated streptomycete communities
Table 2 - Pearson correlation coefficients for relationships among 133-135
conditioning plant communities, edaphic characteristics of conditioned
soil, streptomycete community antagonistic potential, and greenhouse
growth performance
Appendix
Table 1 - Plant compounds tested for impacts on antibiotic production 184
by streptomycetes, and their known roles in other interactions
Table 2 - Plants from which tissue extracts were made, and the 185-186
response of the GBL biosensor to those extracts
v
List of Figures
Page
Chapter 2
Figure 1 – Pathogen antagonism by streptomycete isolates from 41
virgin prairie meadow and never tilled no-till monoculture plots
Figure 2 – Pathogen-inhibitory phenotypes of a streptomycete isolate 42
collection from diverse prairie and from monoculture plant communities
Figure 3 – Inhibition zone sizes created by streptomycete isolates 43
against each of four target pathogens, according to the number of other
pathogens inhibited
Figure 4 – Pathogen inhibitory characteristics of streptomycete 44
operational taxonomic units
Chapter 3
Figure 1 – Changes in OTU richness and diversity as a result of 59-60
de-noising
Figure 2 – Impacts of de-noising on sample clustering 61-63
Figure 3 – Comparison of OTU richness estimates after de-noising 64-65
vs. broadening OTU cutoff thresholds
Figure 4 – Proportion of sequences that could be categorized at 66-67
different taxonomic ranks, with and without de-noising
Chapter 4
Figure 1 – Classical multidimensional scaling of pairwise dissimilarities 85
in Actinobacterial community structure
Figure 2 – Heatmaps showing significant correlations among OTUs 86-87
Figure 3 – Box and whisker plots of Actinobacterial density, richness 88
and diversity, by host species and plant community richness treatment
vi
List of Figures, continued
Page
Figure 4 – OTU richness of Actinobacterial communities associated 89
with four different host plants, grown in each of five different plant
richness treatments
Figure S1 – Rarefaction curve for Actinobacterial OTUs, for the 90
entire dataset
Figure S2 – Rarefaction curve for Actinobacterial OTUs on a per 91
sample basis
Figure S3 – A summary of sample clustering and Actinobacterial 92
community composition
Chapter 5
Figure 1 - Conditioning species and conditioning plant richness 117-118
treatment impacted streptomycete community antagonistic potential
Figure 2 - The antagonistic potential of streptomycetes associated 119
with A. gerardii was modulated by plant richness
Figure 3 - The strength of relationships among various measures 120-121
of streptomycete community antagonistic potential differed among
plant host species
Figure 4 - The strength of relationships among various measures of 122-123
streptomycete community antagonistic potential differed among
plant richness treatments
Figure 5 - Growth responses of four prairie plants varied according 124-125
to conditioning species
Figure 6 - Growth response, measured as root length, varied with 126
the conditioning plant community richness
Figure 7 - Plant community richness modulated the impacts of soil 127-128
conditioning by particular host species on subsequent growth response
vii
List of Figures, continued
Page
Figure 8 - Plant-soil feedbacks impacted the relative performance 129-130
of four prairie plants
Figure 9 - Relative root length was impacted by conditioning plant 131
richness
Appendix
Figure 1 - An example of the assay used to detect changes in 170
antibiotic production caused by the presence of plant compounds
Figure 2 - Illustrative examples of changes to patterns of 171
streptomycete antibiotic production as a result of exposure to
various plant compounds
Figure 3 - Isolates differed in sensitivity to external interference 172
with antibiotic production
Figure 4 - Indole-3-acetic acid (IAA) had differential effects on 173
antibiotic production among streptomycete isolates
Figure 5 - Intact pea seedlings were able to induce growth of the 174
GBL biosensor on a kanamycin-containing medium
Figure 6 - Extracts from lemon and lime lost their ability to elicit 175
a positive response from the GBL biosensor after alkaline treatment
Figure S1 - The presence of particular plant compounds influenced 176-183
inhibitory activity by streptomycete isolates
1
Chapter 1:
Introduction
1. The Streptomycetes
1.1. Biology and ecology of streptomycetes in soil
1.2. Streptomycetes as pathogen antagonists
1.2.1. Evidence of selection for antagonistic phenotypes
1.2.2. Importance of signaling interactions to antagonistic phenotypes
1.3. Streptomycetes as potential symbionts for diffuse mutualism with plants
2. Plants as drivers of soil microbial community characteristics
2.1. Evidence for plant-derived impacts on soil microbial communities
2.2. Mechanisms underlying plant-derived impacts on soil microbial communities
2.3. Implications of plant-derived impacts on soil microbial communities
2.3.1. Plant-soil feedbacks
3. Hypotheses and unique contributions of this dissertation
2
1. The streptomycetes
1.1. Biology and ecology of streptomycetes in soil
The work described in this thesis deals mainly with soil bacteria belonging to the genus
Streptomyces and closely related genera, together referred to as the streptomycetes. The
streptomycetes are true bacteria, belonging to the phylum Actinobacteria, class
Actinobacteria, order Actinomycetales (Anderson and Wellington, 2001). Periods of
active growth in soil appear to be discontinuous in space and time, with isolates existing
most of the time as spores (Lloyd, 1969; Mayfield et al., 1972; Williams et al., 1984).
Dissemination of spores, which are often strongly hydrophobic, can occur via arthropods
(Williams et al., 1984), or through windborne movement along with soil particles
(Llloyd, 1969; Ensign, 1978). Spore germination is sensitive to nutrient availability
(Lloyd, 1969), and leads to the formation of a filamentous substrate mycelium.
Subsequently, and perhaps triggered by nutrient limitation (Hopwood, 2006), aerial
hyphae are produced as the first step toward spore formation. The substrate mycelium is
cannibalized for energy and nutrient resources to fuel the production of aerial hyphae and
subsequent differentiation into chains of spores (Horinouchi, 2007). Morphologically, the
streptomycetes are among the most complex of bacteria.
The streptomycetes are also characterized by genetic complexity, with very large and
heavily regulated genomes. For example, the genome of Streptomyces coelicolor A3(2)
has a length of approximately nine megabases, nearly twice that of Escherichia coli
(Blattner, 1997), and thirteen percent of all genes appear to be transcriptional regulators
(Hopwood, 2006). The streptomycete genome is typically arranged in one large, linear
chromosome, along with smaller linear or circular plasmids (Kieser et al., 2000). There is
tremendous variation in metabolic or functional capacity among streptomycetes, which is
enabled by the partitioning of genes between a highly conserved core, encoding essential
functions, and chromosome arms that carry conditionally beneficial genes and appear to
undergo frequent and extensive changes (Hopwood, 2006). The metabolic potential of the
streptomycetes is legendary (Kieser et al., 2000), with strains for which genome
sequences are available showing the capacity to produce many different secondary
3
metabolites, ranging from antibiotics to siderophores and pigments (Omura et al., 2001;
Bentley et al., 2002).
These gram positive, high GC bacteria are among the dominant microbial taxa in soil
(Acosta-Martinez et al. 2008). The streptomycetes exhibit a general preference for
relatively dry habitats (Williams et al., 1972), and both acidophilic and neutrophilic taxa
exist (Williams and Mayfield, 1971). A large number of secreted proteins, proteases,
hydrolytic enzymes and biosynthetic enzymes are produced by streptomycetes (Chater et
al., 2010). Such extracellular activity is important to nutrient cycling and the degradation
of complex and recalcitrant fractions of soil organic matter (Williams et al., 1984), as
well as in interactions with other microbes sharing the same habitat (Kieser et al., 2000).
A small number of streptomycete taxa are plant pathogens, capable of causing disease on
potato, sweet potato, sugar beet, carrot, and radish (Kieser et al., 2000). Among the many
functions of streptomycetes in nature, their activity as antibiotic producers and
antagonists of plant pathogens are of greatest interest here.
1.2. Streptomycetes as pathogen antagonists
Streptomycetes have been studied for their contribution to limiting plant disease across a
wide range of pathosystems (Yuan and Crawford, 1995; Liu et al., 1995; Jones and
Samac, 1996; El-Tarabily et al., 1997; Chamberlain and Crawford, 1999; Samac and
Kinkel, 2001; Xiao et al., 2002; Samac et al., 2003; Ryan et al., 2004). Many
streptomycetes also show plant growth-promoting characteristics (Doumbou et al., 2001)
or engage in other interactions having the potential to impact plant fitness. For example,
streptomycetes have been shown to produce the plant hormone indole-3-acetic acid
(Tuomi et al., 1994), to induce plant defense responses (Lehr et al., 2007), to increase the
rate of nodulation of peas by Rhizobium (Tokala et al., 2002), to promote the formation of
mycorrhizae (Schrey et al., 2007), and to act as hyperparasites of fungal plant pathogens
(Tapio and Pohtolahdenpera, 1991).
4
Among the various mechanisms of biocontrol and plant growth promotion, antibiotic-
mediated inhibition of pathogens is of particular interest. Despite the difficulty of
observing antibiotic inhibition in situ, many lines of indirect evidence support the
importance of antibiosis to pathogen suppression in soil (Fravel, 1988; Haas and Keel,
2003; Anukool et al., 2004). Indeed, the frequency, intensity, and diversity of antibiotic
inhibitory interactions among streptomycetes have all been related to effectiveness of
disease suppression in agricultural soils (Perez et al. 2008, Wiggins and Kinkel 2005a,
Wiggins and Kinkel 2005b). However, the forces underlying the generation and
maintenance of streptomycete antibiotic phenotypes remain poorly understood. In
particular, elucidating the conditions under which selective pressures act to enhance
antibiotic production or other pathogen-antagonistic phenotypes may allow for
agricultural management that deliberately imposes such selection to enhance pathogen
antagonism and reduce plant disease.
1.2.1. Evidence of selection for antagonistic phenotypes
The astonishing capacity for antibiotic production among streptomycetes (Horinouchi
2007) is in itself strong evidence that antibiotic phenotypes respond to selection. Many
streptomycetes have dedicated genetic pathways for the production of multiple antibiotics
(Omura et al., 2001; Bentley et al., 2002), the synthesis of which is likely to be
energetically costly. Unless there were selective advantages to antibiotic production, such
genetic and energetic investment would not be sustained evolutionarily. Indeed, the
complex regulatory mechanisms that have developed to limit and optimize the expression
of antibiotic biosynthetic genes suggests a benefit to minimizing costs associated with
biosynthesis.
The concepts of synergy and contingency have been used to explain the development of
the extensive secondary metabolic potential among streptomycetes (Challis and
Hopwood, 2003). For example, coordinated production of synergistically acting
compounds may have arisen to overcome resistance, as in cases where beta-lactam
antibiotics are produced together with beta-lactamase inhibitors, together limiting the
5
effectiveness of antibiotic resistance (Challis and Hopwood, 2003). The concept of
contingency among secondary metabolites is supported by the observation of functionally
redundant capabilities, such as the production of multiple siderophores (Barona-Gomez et
al., 2006) that may compensate for loss of effectiveness through the development of
resistance or populations of cheating strains that capture the benefits of a costly
secondary metabolite without bearing the costs of production.
The antagonistic capacity of streptomycete communities varies spatially (Davelos et al.,
2004a), presumably in response to unique selective pressures. Even within taxa,
streptomycete antibiotic phenotypes differ by spatial location (Davelos-Baines et al.,
2007). Such divergence of phenotype within a taxon is evidence of local selection.
Furthermore, streptomycete antibiotic phenotypes at a population or community level can
be shifted through simple agricultural management practices. For example, the use of
green manures and cropping sequences can be used to alter the densities, relative
abundances, and inhibitory activities of streptomycete pathogen-antagonists in soil
(Wiggins and Kinkel, 2005a, 2005b).
The sensitivity of streptomycete antibiotic phenotypes to selection suggests the
possibility of enhancing pathogen-antagonistic activity through deliberately imposed
selection, such as through agricultural management practices or particular features of
crop or green manure cultivars. There have been many efforts to use agricultural
management practices to shift soil microbial community structure in ways that enhance
disease suppression (Mazzola, 2004), but to date these efforts have not met with
consistent success. Understanding the mechanisms and processes that lead to the
development of disease-suppressive soil microbial communities remains an important
priority for research. In particular, insights into the influence of various plant species and
plant community characteristics on the antagonistic phenotypes of soil streptomycetes
may have implications for agricultural management.
6
1.2.2. Importance of signaling interactions to antagonistic phenotypes
Mechanisms other than selection acting on antagonistic phenotypes or changing soil
microbial community structure may also be significant to pathogen suppression. In
particular, the ubiquity and importance of microbial signaling, mediated via chemical
messages, has become clear for a wide range of phenotypes (Shank and Kolter, 2009).
Research has revealed an astonishing degree of interdependence among members of
microbial communities, with interactions altering metabolite production (Angell et al.,
2006) and even the ability of some organisms to grow (Davis et al., 2008). For example,
exogenously produced siderophores are vital to the growth of some streptomycetes under
certain conditions (Yamanaka et al., 2005). Similarly, short peptides have been found to
act as chemical signals that trigger the growth of some bacteria (Nichols et al., 2008).
Interactions among species may even prove to be the key to unlocking metabolic
potential that has remained quiescent over decades of laboratory culture in some strains
(Omura et al., 2001). In Aspergillus, specific secondary metabolism genes were activated
by a specific interaction with Actinobacterial isolates, requiring close physical contact
(Schroeckh et al., 2009).
It is probable that continued investigation will greatly expand the repertoire of
compounds known to have signaling functions or to elicit specific responses in receiving
organisms. For example, recent work has shown that sub-inhibitory concentrations of
antibiotics can have widespread effects that resemble chemical signaling more than
inhibitory or antagonistic interactions (Goh et al., 2002; Yim et al., 2007; Aminov, 2009).
Indeed, given that all soil bacteria apparently exist in enormously complex communities,
it is reasonable that fitness benefits could be available through mechanisms for sensing
and responding to neighboring organisms. Participation in bacterial signaling interactions
by higher organisms such as plants has been demonstrated (Teplitski et al., 2000; Gao et
al., 2003). Indeed, the ability of plants to alter bacterial phenotypes and gene expression
via chemical signals is vital to the formation of nitrogen fixing and mycorrhizal
symbioses (Broughton et al., 2003; Steinkellner et al., 2007). However, the implications
7
of cross-taxa signaling interactions for plant pathogen suppression have not been
adequately explored.
For the streptomycetes in particular, it has been observed repeatedly that diffusible
compounds produced by one strain may influence antibiotic production in other strains
(Becker et al., 1997; Ueda et al., 2000). For example, one study found that approximately
25% of Streptomyces isolates produced some compound capable of inducing antibiotic
production by Streptomyces tenjimariensis (Slattery et al., 2001). Such a response to
exogenous chemical signals must occur through interactions with the regulatory pathways
that govern antibiotic production. It is not essential here to discuss all of the known
systems and pathways that regulate antibiotic production in the streptomycetes. However,
it should be noted that there exist both general regulatory systems coordinating antibiotic
production together with morphological differentiation (Horinouchi, 2007) and pathway
specific regulatory systems that can activate the biosynthesis of individual antibiotics
(Cundliffe, 2006).
Complexity is a predominant characteristic of the regulatory systems governing antibiotic
production (O’Rourke et al., 2009), with a given pathway often subject to both repressors
and activators (Cundliffe, 2006). Even regulators located within particular biosynthetic
clusters can control other pathways and have pleiotropic effects (Huang et al., 2005). In
addition to regulation at the level of gene expression, substrate and co-factor limitation
can also regulate biosynthetic rates (Huh et al., 2004). Competition for precursor
molecules among biosynthetic pathways can lead to pleiotropic effects of specific
regulation on a given pathway (Gottelt, 2010). Furthermore, nutrient status can be an
important trigger for secondary metabolism (Vilches et al., 1990; Yang et al., 2009), as in
the activation of grixazone production by phosphate depletion (Horinouchi, 2007).
Indeed, in some cases, shared regulatory systems exist for both antibiotic production and
nutritional stress response (Lian et al., 2008).
8
Despite the complexity and variety of systems regulating antibiotic production in the
streptomycetes, this thesis will emphasize diffusible signal factors capable of altering
antibiotic production through cross-taxa interactions. Several classes of such diffusible
signals have been described, including the methylenomycin furans (Corre et al., 2008;
O’Rourke et al., 2009), B-factor (Kawaguchi et al., 1988), and PI factor (Recio et al.,
2004). However, the most well-characterized and understood system involves the
gamma-butyrolactones (GBLs). These small diffusible signals, first described by
Khokhlov in 1967 (cited in Takano, 2006), control and coordinate antibiotic production,
and sometimes spore formation and morphological differentiation as well (Horinouchi
and Beppu, 2007) The GBLs have been understood to be analogous to hormones in
higher organisms (Horinouchi, 2007), responsible for triggering changes and
coordinating cellular activity across the multicellular body of an organism. Several GBL-
mediated signaling pathways have been described in detail, and evidence suggests that
this regulatory mechanism is widespread and important among streptomycetes (Hara and
Beppu, 1982; Hashimoto et al., 1992; Choi et al., 2003). The GBL receptors are repressor
proteins that bind DNA and prevent the transcription of downstream genes. Upon binding
of the GBL signal molecule, the receptor dissociates from the DNA, relieving
transcriptional repression (Takano et al., 2001, 2005).
Under laboratory conditions, GBL signals are able to diffuse between strains and
significantly influence phenotype. For example, the phenotype of a mutant GBL-deficient
strain can be restored by adjacent culture of a wildtype GBL-producing strain (Hara and
Beppu, 1982), and GBLs produced by one species are often able to alter the morphology
or antibiotic production of another species (Hashimoto et al., 1992). Across
streptomycete taxa, the specificity of GBL signaling and the potential for cross talk are
determined by a combination of structural variations to the GBL signal molecule and
relative differences in substrate affinity among GBL receptor proteins. The effects of
structural modifications on GBL signal activity have been explored for particular receptor
proteins (Nihira et al., 1988) and even single amino acid substitutions in receptor proteins
can lead to alterations in signal response, by changing DNA binding activity (Gottelt,
9
2010) or affinity for the GBL signal. In a wildtype strain of S. virginiae, the quantity of
GBL signal required to induce virginiamycin production varied by as much as 100,000
times among structural variants (Nihira et al., 1988), while with an engineered biosensor
system structural GBL variants differed up to 500 fold in activity (Hsiao et al., 2009).
To date, three major classes of GBLs have been demarcated based on differences in
chemical structure. These are typified by A-factor in S. griseus (Horinouchi, 2002;
Ohnishi et al., 2005), IM-2 in S. lavendulae (the SCBs of S. coelicolor are of this type)
(Sato et al., 1989), and the VBs of S. virginiae (Ohashi et al., 1989). Interestingly, genes
for GBL receptors appear to predate GBL synthases (Nishida et al., 2007) and may have
had earlier roles in other regulatory pathways. Thus, it may be possible that GBL receptor
homologs govern diverse aspects of streptomycete biology. Indeed, some GBL receptor
homologs have been found to bind endogenous antibiotics and elicit downstream
responses (Xu et al., 2010).
Competitive interactions are believed to be important to microbial fitness in the nutrient-
limited soil environment, although this hypothesis is difficult to test explicitly. Some
streptomycetes have been shown to achieve higher population densities when introduced
to soil alone compared to co-inoculation with another strain (Schlatter et al., 2010). A
role for antibiosis in mediating such interactions is supported by evidence of coevolution
for antibiotic production and resistance phenotypes in streptomycetes (Laskaris et al.,
2010). If antibiotics do mediate interactions among streptomycetes, interference with the
hormonal signals that regulate antibiotic production may be an important competitive
strategy.
Benefits may also be available to other organisms, such as plants, that acquire the ability
to manipulate or interact with streptomycete antibiotic regulatory pathways.
Theoretically, incentives could exist for a neighboring organism to either induce the
production of antibiotics through the synthesis of appropriate chemical signals, or to
delay or prevent antibiotic production through signal modification and degradation or by
10
producing antagonists to block proper signal reception. Insufficient attention has been
given to these possibilities among streptomycetes, although examples have been
documented in other bacterial signaling systems (Otto et al., 1999, 2001; Lyon and
Novick, 2004; Delalande et al., 2005; Uroz et al., 2007, 2008). It remains unclear, for
example, whether plants could realize a fitness benefit by inducing antibiotic production
in rhizosphere streptomycetes for protection against pathogens, or whether interference
with streptomycete antibiotic regulatory pathways by plants could be important for the
protection of symbiotic partners such as Rhizobium spp.
1.3. Streptomycetes as potential symbionts for diffuse mutualism with plants
Among the myriad of potential interacting partners for the streptomycetes, I am most
interested in relationships with plants. Several features of streptomycete biology and
ecology, outlined above, suggest that soil streptomycetes may form diffuse protective
mutualisms with plants. Potential fitness benefits exist for plants capable of appropriate
interactions with soil streptomycetes because:
- streptomycetes are ubiquitous and abundant in soil
- streptomycetes can contribute to plant growth promotion and can effectively
antagonize plant pathogens, particularly via antibiosis
- streptomycete antibiotic phenotypes are subject to selection
- streptomycete antibiotic phenotypes are susceptible to manipulation via
exogenous signaling compounds
However, is it plausible that plants could possess the capabilities necessary to engage
streptomycete populations for their own benefit?
11
2. Plants as drivers of soil microbial community characteristics
A variety of biotic and abiotic forces shape soil microbial communities, but within a
given soil type and set of climatic conditions, plants exert substantial control over the
growth conditions experienced by soil microbes. In particular, plants are capable of
altering microbial population densities, the identity and relative abundances of taxa
present, and the functional activities that are carried out by soil microbial communities,
as outlined below. Naturally, these characteristics are interrelated, although specific
relationships between microbial community structure and function are often unclear
(Fuhrman, 2009). Understanding the factors that promote and maintain microbial
diversity is an important task still before the discipline of microbial ecology. Plants may
play a significant role in this regard, through direct impacts on microbial activity and
fitness, or indirectly through changes to soil properties or microbial interactions.
2.1. Evidence for plant-derived impacts on soil microbial communities
That plant hosts significantly impact associated soil microbial communities has now been
documented repeatedly and extensively, across many locations, plant hosts, and
environmental conditions. However, some studies have found weak plant host effects
(Kielak et al., 2008), or that differences among soils are of greater effect than differences
among plant hosts (Dalmastri et al., 1999; Girvan et al., 2003; Ulrich and Becker, 2006;
Wakelin et al., 2008). Indeed, even within a soil type, mineral particles create distinct
microhabitats and select correspondingly distinct bacterial communities (Carson et al.,
2009). Nevertheless, in many cases plant host effects have been found to equal or exceed
effects due to soil type (Grayston et al., 1998; Miethling et al., 2000; Marschner et al.,
2004), and there can be significant interactions of plant host with soil type (Marschner et
al., 2001; Innes et al., 2004).
Interactions between soil microorganisms and plants occur primarily in the rhizosphere
(Barea et al., 2005; Prithiviraj et al., 2007), making this the most likely place to observe
effects of plants on associated soil microbial communities (Kowalchuk et al., 2002; Bais
et al., 2006). Indeed, the well-known rhizosphere effect, wherein bacterial populations
12
and activity are markedly higher in soil adhering to plant roots compared to bulk soil
(Starkey, 1958), is compelling evidence of the ability of plants to alter associated soil
microbial communities. However, a number of studies have found plant-driven effects
extending to microbial communities in bulk soil (Carney and Matson, 2006; Bremer et
al., 2009).
Plant hosts differ in the density of soil microbes that are supported. This has been
demonstrated for particular taxa as well for overall microbial population densities. For
example, alfalfa supported larger populations of inoculated Sinorhizobium meliloti
compared to rye (Miethling et al., 2000), and the population density of antibiotic-
producing pseudomonads in soil was found to differ among plant hosts (Bergsma-Vlami
et al., 2005a). Plant species colonizing recently deglaciated terrain were shown to have
differential effects on soil microbial biomass (Bardgett and Walker, 2004), indicating that
some plant species support more dense soil microbial communities than others.
That plant host can impact associated soil microbial community composition and
structure has been documented extensively. The richness, composition and diversity of
pathogen-antagonistic taxa were found to be plant species dependent (Berg et al., 2002),
as was specific antagonist identity (Berg et al., 2006). Plant hosts have been shown to
alter the identity of ammonia-oxidizing bacteria (Briones et al., 2002) and denitrifying
bacteria (Bremer et al., 2007) present in soil. Microbial community composition varied
according to pioneer colonizing plant species (Bardgett and Walker, 2004) and the
rhizosphere microbial communities fostered by different cultivars of canola could be
distinguished (Siciliano et al., 1998). Plant species identity was shown to impact
nematode community composition, diversity and evenness (Viketoft et al., 2005).
Interestingly, the strength of selective effect has been shown to differ among host plants;
compared to microbial communities associated with tomato, flax caused a greater shift
away from the baseline conditions of uncultivated soil (Lemanceau et al., 1995). The
ability of host plants to differentially select among soil microbes is also suggested by host
13
specificity in mycorrhizal associations, although partners in such interactions span a
continuum from specialist to generalist (Johnson et al., 2005).
Differential host plant selective effects have been documented with a wide variety of
techniques, which supports the validity and ubiquity of such findings. Significant effects
of plant host identity on soil microbial phospholipid fatty acid (PLFA) profiles have been
found, with individual fatty acid signatures showing distinct responses to plant host
(Carney and Matson, 2006). Fatty acid methyl ester (FAME) profiles were shown to
differ between bacterial strains growing on the root surface of canola and wheat
(Germida et al., 1998). Denaturing gradient gel electrophoresis (DGGE) has been used to
demonstrate that strawberry rhizosphere samples were more similar to each other than
bulk soil samples across locations (Costa et al., 2006), that soil microbial diversity
differed among host plants (Garbeva et al., 2008), and that plant-dependent shifts in the
relative abundance of bacterial populations became more pronounced over time (Smalla
et al., 2001). A similar technique, temperature gradient gel electrophoresis (TGGE), was
used to demonstrate differences in the microbial communities associated with alfalfa and
rye (Miethling et al., 2000). Terminal restriction fragment length polymorphism (T-
RFLP) analysis targeting the Actinobacteria revealed differences in soil microbial
community among plant hosts and in comparison to interspaces (Kuske et al., 2002).
Traditional, culture-based approaches have also demonstrated effects of particular plant
species on soil bacterial populations (Loranger-Merciris et al., 2006).
Studies of genetically modified plants have demonstrated that small changes in plant
genotype can result in significant impacts on associated microbial communities
(Giovanni et al., 1999). In one fascinating example, plants engineered to produce novel
carbon compounds were shown to quickly select for bacteria capable of metabolizing
those compounds, even though such capabilities were undetectable among bacteria
isolated prior to the plant-imposed selection (Oger et al., 2004). This is an indication of
the strength and rapidity with which plant-imposed selection can act on soil microbial
communities. Furthermore, invasive plants have been shown to significantly alter soil
14
microbial communities in invaded soils (Batten et al., 2006), in some cases significantly
reducing soil microbial diversity (Broz et al., 2007). Monitoring soil microbial
communities over the course of a change in plant cover can also reveal the selective
effect of host plant; at least in some cases, it appears that the soil microbial community
stabilizes after a period of adaptation to a host plant, with subsequent host switching
leading to dramatic microbial community shifts (Badri et al., 2008).
The effects of host plants have also been found to extend to many functional measures of
soil microbial activity. For example, alfalfa and rye supported microbial communities
with significantly different community level physiological profiles (Miethling et al.,
2000). The proportion of auxin-producing pseudomonads was higher for heterozygous
maize plants compared to either of the parent lines (Picard and Bosco, 2005). Rice
cultivars supported different amounts of activity by associated ammonia-oxidizing
bacteria (Briones et al., 2002), and plant identity impacted soil denitrifier activity
(Bremer et al., 2009). Even very general functional measures can be influenced by host
plant identity, as was seen for total microbial respiration (Innes et al., 2004).
Pathogen antagonism and related functional traits have been specifically studied as
variables that respond to plant host selection. The proportion, composition, richness and
diversity of pathogen-antagonistic microbes has been shown to be plant species
dependent (Berg et al., 2002, 2005; Garbeva et al., 2008). An enhanced proportion, but
reduced diversity of Verticillium antagonists was found in rhizosphere compared to bulk
soil (Berg et al., 2006). This lowered antagonist diversity in the rhizosphere is evidence
of plant-driven selection favoring some microbes over others (Berg et al., 2005). The
density of antibiotic producing pseudomonads differed by host plant (De La Fuente et al.,
2006), and even cultivars of wheat differed in their ability to support antibiotic producing
pseudomonads (Mazzola et al., 2004). From the opposite perspective, S. griseoviridis was
shown to colonize the root surface of turnip rape (Brassica rapa subspecies oleifera)
more readily than of carrot (Kortemaa et al., 1994) and the ability of various antibiotic
producing pseudomonad genotypes to colonize the rhizosphere of sugar beets was shown
15
to be variable (Bergsma-Vlami et al., 2005a). Furthermore, the amount of antibiotic
produced on a per cell basis by pseudomonads in the rhizosphere differed among plant
species (Bergsma-Vlami et al., 2005b).
While the presence of discernible effects of host plant genotype on associated soil
communities has been well established, there remains greater uncertainty about the
significance of higher-order plant community characteristics. For example, does plant
richness or diversity have implications for soil microbial community structure or
function? Both affirmative and negative conclusions have been drawn regarding the
importance of plant diversity in shaping soil microbial communities; for example, the
diversity of carbon sources that could be utilized by soil bacteria increased with plant
community diversity (Benizri and Amiaud, 2005), but plant diversity did not impact
denitrifier community composition or activity (Bremer et al., 2009). Another study found
that higher plant diversity led to higher soil bacterial diversity, which was in turn related
to the effectiveness of pathogen suppression (Garbeva et al., 2006).
Unfortunately, not all studies account for confounding factors in plant diversity
manipulations. For example, biomass production may increase with plant diversity (Zak
et al., 2003), and it is unclear in some cases whether observed effects are due to plant
diversity per se, or merely to increased amounts of plant biomass available to microbial
food webs. Nevertheless, in a study system where plant diversity manipulation did not
lead to confounding changes in biomass production, plant richness still impacted
measures of nutrient use and catabolic capacity among soil microbes (Loranger-Merciris
et al., 2006). In other studies, differences in productivity were accounted for statistically
and effects of plant richness were still observed for microbial variables such as
cellulolytic and chitinolytic capacity (Chung et al., 2007) or the catabolic activity of
culturable bacteria (Bartelt-Ryser et al., 2005).
Beyond confounding effects of productivity, there is also debate about whether diversity
may be important primarily for increasing the likelihood of the presence of particular
16
plant species having disproportionate impacts. Support for this concept, dubbed the
sampling effect (Wardle et al., 1999), does exist; bacterial catabolic activity and diversity
were found to increase with plant species number in one study, but the plant host
Trifolium repens had a substantially stronger impact than any of the other host plants
(Stephan et al., 2000). In some cases, effects of plant species identity have been found to
be more durable than effects of plant species richness (Bartelt-Ryser et al., 2005).
2.2. Mechanisms underlying plant-derived impacts on soil microbial communities
Unfortunately, few studies have gone beyond measuring plant host impact on soil
microbial community structure to investigations of actual mechanisms underlying such
observed selection by plants. There is an implicit assumption throughout much of the
literature that the chemical nature of the resources provided by plants explains differing
outcomes in terms of microbial community composition, structure or function. However,
explicit tests of this hypothesis are rare, and there is a great deal of ambiguity regarding
the relative importance of the various means of resource inputs provided to soil microbial
communities by plants.
The dominant nutrient sources available to soil microbial communities are ultimately of
plant origin: root exudates, senescent tissues, leaf litter, border cells, mucilage, and
leachates. Total nutrient inputs from plants will constrain microbial densities in soil, and
the variety of niches available to soil microbes will depend upon the chemical
composition and the spatial and temporal distribution of plant-supplied resources.
However, substantial chemical changes occur as plant biomass cycles through soil food
webs, making it unclear how persistent host plant effects may be. In particular, root
exudates are quickly assimilated and modified by soil microbes before being released
again (Dennis et al., 2010). Because portions of the soil organic matter pool can be stable
for decades or centuries (Kemmitt et al., 2008), it is possible that legacy effects of past
plant communities could persist for extended periods of time. Interactions between
contemporary inputs and legacy resources may complicate the effects of host plants on
associated microbes.
17
A number of studies have emphasized root exudates as having particular importance in
determining rhizosphere microbial community characteristics (Walker et al., 2003),
although their significance relative to other rhizodeposits has not always been clearly
demonstrated (Dennis et al., 2010). I hypothesize that while root exudation may be
critical in modulating immediate interactions within the rhizosphere, inputs such as plant
litter and senescent tissue may play a larger role in influencing bulk soil microbial
communities. Root exudation refers to the net efflux of a variety of chemicals from
actively growing plant roots (Phillips et al., 2004). Methods have been developed for the
in vitro collection of plant root exudates (Meharg and Killham, 1991; Nagahashi and
Douds, 2000), and such collected exudates can have effects similar to whole plants when
applied to soil (Badri et al., 2008; Broeckling et al., 2008). Root exudate characteristics
do indeed differ among plant hosts; grass species were found to exhibit quantitative
differences in root exudate composition (Dormaar et al., 2002), and the quantity and
chemical identity of root exudates were found to differ among sorghum accessions
(Czarnota et al., 2003).
The sensitivity of root exudation to exogenous factors offers an explanation for
variability in host plant effects across locations, samples or experiments. For example,
root exudation has been shown to be sensitive to plant nutrient status (Johnson et al.,
1995; Shen et al., 2001), to mechanical forces (Barber and Gunn, 1974), growth substrate
(Kamilova et al., 2006a), and even to atmospheric conditions, as in a case where carbon
dioxide enrichment resulted in a decreased exudation of phenolic acids and total sugars
(Hodge et al., 1998). Root exudation can even vary within an individual plant, as in the
specialized cluster roots formed by white lupin in response to phosphate depletion;
cluster roots exude higher than normal concentrations of organic acids, with the effect of
increasing phosphorus availability. However, this change also has broader effects,
reducing bacterial density, richness, and diversity relative to other portions of the root
system (Weisskopf et al., 2005). Elephantgrass similarly undergoes specific changes in
root exudation that help to alleviate phosphorus deficiency (Shen et al., 2001).
18
Furthermore, root exudation is sensitive to the presence of rhizosphere microbes, with
various microbial metabolites significantly enhancing net efflux rate from plant roots
(Meharg and Killham, 1995; Phillips et al., 2004) or altering exudate composition
(Kamilova et al., 2006b). Thus soil microbes establishing populations in the rhizosphere
likely play a part in shaping their own selective environment by modulating plant root
exudation (De-la-Pena et al., 2008). It is possible that microbes induce patterns of root
exudation which are favorable for their own growth, or that certain keystone microbial
species alter the selective environment for the whole community by controlling root
exudation. The genetic and cellular bases for root exudation are beginning to be explored
(Badri et al., 2008, 2009), and should shed light on the mechanisms by which plant
genotypes exert differential selection on rhizosphere microbes.
The concept of plant-driven selection mediated through the chemical nature of resource
inputs invokes a deterministic model of microbial community assembly. Plants are
understood to define in important ways the shape and dimensions of niche space
available to soil microbes, and to create fitness penalties and rewards that are unequally
distributed among microbial taxa. The provision of specific chemical compounds may
offer a selective advantage to organisms with the optimal enzymatic capabilities for
accessing those substrates. Unfortunately, there is a paucity of actual data describing
relationships between the chemical identity of resources provided by plants and
corresponding changes in microbial community structure. Direct amendment of various
carbon sources to soil has demonstrated that the chemical identity of substrates for
heterotrophic microbial growth can impact microbial respiration rates and community
structure (Orwin et al., 2006). Additionally, the identity of available carbon substrates
influenced the resource use capabilities of soil streptomycetes (Schlatter et al., 2008). In
another study, bacterial isolates collected from plant root tips showed better growth in
minimal media containing a dominant root exudate component as the sole carbon source,
compared to randomly selected rhizobacteria (Kamilova et al., 2006a).
19
In some cases, observed functional differences resulting from host plant manipulation
suggest plausible connections to root exudation. For example, carbon source utilization
by soil microbes varied among host plant species, and the carbon substrates responsible
for the observed patterns matched known major components of root exudates:
carbohydrates, carboxylic acids and amino acids (Grayston et al., 1998). Similarly,
differences in soil bacterial community level physiological profiles among pioneering
plant species were driven by the utilization of particular carbon compounds (Yan et al.,
2008), although it was not demonstrated that these compounds were constituents of the
root exudates of the plants in question. In many cases, selection driven by the chemical
nature of resource inputs provided by plants has been assumed and not demonstrated
(Knee et al., 2001; Shaw et al., 2006).
Rhizosphere microbes could experience negative fitness effects in cases of plants that
produce bioactive molecules with directly inhibitory properties (Broeckling et al., 2008;
De-la-Pena et al., 2008, 2010; Badri et al., 2009). For example, root glucosinolate content
in Brassica napus was negatively correlated with root infection by Azorhizobium
caulinodans (O’Callaghan et al., 2000). Selective effects may also be indirect, resulting
from changes to the physical or chemical environment, such as modifications to water
content, soil pH (Starkey 1958) or other factors. For example, rice cultivars that
supported differing activity levels by ammonia-oxidizing bacteria were also found to
exhibit differences in oxygen availability in the root zone (Briones et al., 2002),
suggesting that the observed host plant effects were modulated through atmospheric
chemistry. Even plant traits such as rooting architecture (de Dorlodot et al., 2007) may
influence associated microbial communities through such indirect means.
Plants may also impact soil microbial communities via chemical signaling that alters gene
expression, changing microbial phenotypes and altering outcomes of microbial
competitive interactions. Outside of the well-established symbioses such as those
involving mycorrhizal fungi and nitrogen fixing bacteria, direct chemical signaling
interactions between plants and soil microbes have not received adequate research
20
attention. Chemical communication between plants and associated bacteria has been
demonstrated in both directions (Mathesius et al., 2003; Dudler and Eberl, 2006), and
direct interference of bacterial regulatory pathways by plants has been shown for gram-
negative bacteria using the N-acyl homoserine lactone quorum sensing system (Teplitski
et al., 2000). Legumes release flavonoids that alter patterns of gene expression in
rhizobia, initiating a series of complex and specific interactions that ultimately lead to the
fixation of atmospheric nitrogen inside of nodules (Oldroyd 2009). Chemical signals
produced by plants are also critical to the formation of mycorrhizae (Steinkellner et al.,
2007), with compounds in root exudates stimulating hyphal growth (Nair et al., 1991) and
branching (Nagahashi and Douds, 2000) in mycorrhizal fungi. Specific flavonoids from
plants elicit variable responses among mycorrhizal partners during pre-symbiotic growth
(Scervino et al., 2005b), providing a mechanism for the specificity of mycorrhizal
associations to be mediated through chemical signaling efficiency. Aqueous extracts of
an actinorhizal plant, Casuarina cunninghamiana, preferentially stimulated growth of
symbiotic Frankia relative to other soil bacteria (Zimpfer et al., 2004). Interestingly, a
streptomycete isolate was also stimulated by these extracts (Zimpfer et al., 2004). Other
examples of changing microbial gene expression in response to exposure to plants or
plant-derived compounds have been reported (Mark et al., 2005; Bagnarol et al., 2007;
Weir et al., 2008), and growth in the presence of plant material leads to changes in
protein expression in Streptomyces (Langlois et al., 2003). Genomic promoter regions
that are specifically activated during growth on plant roots (Ramos-Gonzalez et al., 2005)
may provide an indication of microbial genes whose regulation is susceptible to influence
by plants.
In some cases, it appears that the outcomes of competitive microbial interactions can vary
among plant hosts, as in a mixed inoculum leading to different bacterial strains
successfully colonizing the rhizosphere of various plants (De La Fuente et al., 2006).
Although the mechanisms by which competitive outcomes are altered in such cases are
unclear, one possibility is that plant signaling modulates phenotypes directly relevant to
microbial competition. For example, both indole-3-acetic acid (Matsukawa et al., 2007a)
21
and derivatives of cytokinin (Yang et al., 2006) have been found to stimulate antibiotic
production in diverse streptomycetes.
2.3. Implications of plant-derived impacts on soil microbial communities
Whether mediated through differential provision of energy- and nutrient-rich substrates,
or through more complex inhibitory and signaling interactions, in most cases the basis for
the selective effect of host plant genotype is likely to be found in particular chemical
compounds of plant origin. Understanding the mechanistic basis for plant-driven
selection and modulation of microbial competitive interactions is an important goal for
advancing plant disease control. It may be possible to deliberately select, breed or
genetically modify plants for characteristics that will lead to enhancement of beneficial
functions by associated soil microbes (Ryan et al., 2009). Indeed, a heritable basis has
been shown to underlie the ability of plants to interact with beneficial microbes (Smith et
al., 1999; Schweitzer et al., 2008) and this could be exploited to great benefit in
production agriculture. However, we are far from a predictive understanding of the
mechanisms and outcomes of plant-imposed selection and we have no detailed
understanding of the relative importance of various root exudate components in the
shaping of microbial communities, or how selective effects are modified by soil type and
initial microbial community composition and structure. There are a limited number of
examples of breeding programs that have considered rhizosphere-related traits (Wissuwa
et al., 2008), and those that have considered particular root exudation characteristics have
had a very narrow focus, such as improving nodulation (Rengel, 2002). Thus there
remains great untapped potential in using crop plants themselves as selective agents for
directing microbial activities toward agricultural benefit.
2.3.1. Plant-soil feedbacks
A coupling of the concepts that associated microbes may impact plant fitness, and that
plants are capable of exerting selection on associated microbes suggests the existence of
interactions in which plants realize fitness benefits through targetted manipulation of
associated microbes. Such interactions are likely to be dependent on the provision of
22
resources to soil microbes, suggesting a selective force that may explain the seemingly
inefficient loss of nutrients through plant roots. Plants might easily alter microbial
population densities by varying the quantity of resource inputs provided to soil microbial
communities, perhaps maintaining high microbial densities to sustain a level of
competition that will reduce pathogen viability. Microbial population density may also be
critical to maintaining selection for antagonistic phenotypes generally (Kinkel et al.,
2011). Alternatively, plants may possess mechanisms to select for pathogen-inhibitory
phenotypes among rhizosphere microbes, or to manipulate inhibitory phenotypes via
chemical signaling.
These hypotheses amount to claims for the possibility of plant-soil feedbacks having
positive outcomes for plant fitness. More significantly than simply harboring a
characteristic rhizosphere microbial community, some plant species may select for a
beneficial microbial community, which functions to increase the performance of its plant
host. Feedback effects mediated through soil microbial communities have not been
documented as extensively as the simpler case of plant-driven selection on soil
communities (Ehrenfeld et al., 2005).
Plant-soil feedbacks are believed to be important to plant community dynamics
(Reynolds et al., 2003), where the relative impact among plant species is the critical
factor (McCarthy-Neumann and Kobe, 2010a); feedbacks may be species-specific, or
may be more general, affecting all plants similarly (Casper et al., 2008). Feedbacks that
selectively improve plant performance may lead to exclusion of plant species that
experience relatively poorer performance (Batten et al., 2007). On the other hand,
specific negative feedbacks may slow competitive exclusion and work to sustain plant
diversity (Bever et al., 1997, 2010). Negative feedbacks may be more common than
positive feedbacks between plants and soil microbial communities (Mills and Bever,
1998; Hamel et al., 2005; Casper et al., 2008; McCarthy-Neumann and Kobe, 2010b),
with pathogens accumulating over time in the presence of a given host plant. Support has
been found for soil pathogens as an important factor in the spatial population dynamics of
23
dominant grassland species (Olff et al., 2000). Not all plant-soil feedbacks are mediated
through microbial activity; nutrient carry-over effects (Bartelt-Ryser et al., 2005; Casper
et al., 2008), or changes to soil physical (Ehrenfeld et al., 2005) or chemical properties
(McCarthy-Neumann and Kobe, 2010a) can also explain some plant-soil feedbacks.
Several examples of positive plant-soil feedbacks mediated through soil microbes do
exist. For example, foliar pathogen attack triggers the exudation of L-malic acid from
Arabidopsis roots, which functions to recruit beneficial Bacillus to colonize the plant root
system. By activating host plant defense responses, this change in the rhizosphere
microbial community generates a positive feedback that reduces plant disease (Rudrappa
et al., 2007). In another example, grazing caused the grass Poa pratensis to increase the
rate of carbon exudation from its roots, stimulating soil microbial biomass and activity; in
turn, this microbial activity increased available nitrogen for plant growth (Hamilton and
Frank, 2001). In the context of soilborne diseases, cucumber varieties that were resistant
or susceptible to Fusarium wilt showed distinct soil microbial community structure and
composition, suggesting that resistance may be mediated indirectly, through impacts of
cultivar-specific soil microbial activity on pathogen activity or plant defense responses
(Yao and Wu, 2010). Finally, the development of take-all-decline through continuous
wheat monoculture represents a clear example of a positive feedback between host plant
and beneficial microbes; repeated planting of wheat in the same soil leads to the
establishment of a soil microbial community that suppresses the wheat disease Take-All
(Weller et al., 2002; De La Fuente et al., 2006; Schreiner et al., 2010).
One clear weakness of many studies of plant-soil feedbacks is the failure to consider a
broader context; both soil conditioning and subsequent plant response are often carried
out with plants grown in isolation, in a single soil type. However, feedbacks are likely to
vary with soil properties and plant and microbial community characteristics. For
example, plant growth may be altered by the presence of neighboring plants, as has been
observed for morphological characteristics when plants are grown alone vs. in mixtures
24
(Bartelt-Ryser et al., 2005). It is possible that plant-driven selection may similarly vary
depending on the larger plant community context.
There are theoretical considerations that suggest that this may be the case. Because a
persistent association between two partners is a necessary condition in order for a stable
mutualism to arise (Bronstein 2009), increased plant community diversity may work
against the successful establishment of protective mutualisms between plants and
pathogen antagonists. Moreover, the impetus for engaging in such symbioses (namely,
avoidance of disease) is likely to be stronger in low diversity plant communities, since
pathogen success and plant fitness reduction due to disease are generally highest in
monocultures or low diversity plant communities (Smithson and Lenne 1996, Keesing et
al. 2006, Garrett and Mundt 1999, Mille et al. 2006). Thus, pathogen-antagonists have the
highest potential for positive impacts on plant fitness in low diversity plant communities.
Furthermore, the development and maintenance of a pathogen-inhibitory microbial
community may be energetically costly for plants, while benefits may accrue to adjacent
competing plants. In this case, a disincentive for investing in pathogen-suppressive
microbes may be experienced in higher diversity plant communities, since plants not
bearing the cost of the protective symbiosis may still reap the benefits.
Furthermore, plant-soil feedbacks may be fundamentally altered by characteristics of the
initial microbial community on which plant-driven selection can act. However, this
variable is almost always confounded with soil type, a shortcoming that has hampered
our ability to understand and predict the outcomes of plant-imposed selection on soil
microbes. The identity of microbes present at the onset will naturally constrain the
outcome of selection by plants, and final outcomes may differ even if the same microbial
taxa are present but differ in relative abundance among communities; initial root
colonizers have better success than latecomers who attempt to establish on an already
colonized root surface (Rainey, 1999), and the most abundant taxa are likely to be the
first colonizers. Thus differences in the initial relative abundances of microbial taxa could
fundamentally change the outcome of plant-imposed selection.
25
3. Hypotheses and unique contributions of this dissertation
This thesis explores the role of plants as drivers of soil microbial community structure
and functioning, with a particular emphasis on the streptomycetes as important pathogen
antagonists. I will address fundamental questions about the role of plants, in particular
plant host identity and plant community diversity, in shaping microbial communities.
Additionally, I explore the emerging possibility that plants engage in chemical signaling
with associated bacteria to modify relevant gene expression.
My research addresses the following specific hypotheses:
Ho: Plant species selectively alter the density, composition, structure and pathogen-
inhibitory activity (including intensity, diversity, and frequency of activities) of soil
streptomycete communities.
Ho: Plant community context modulates host plant species effects on associated
streptomycetes, with low plant diversity communities more effectively fostering beneficial
soil microbial associations.
Ho: Plants produce chemical signals that interact with antibiotic regulatory pathways in
streptomycetes.
This work is motivated by an ecological understanding of the significance of species
interactions in shaping emergent community functions, particularly functions of soil
microbial communities that promote plant health. An ecological perspective on plant
disease suppression will improve our understanding of plant community dynamics and
the significance of plant identity and diversity to microbial community processes.
Additionally, this research will provide fundamental insight into the regulation of
antibiotic production by streptomycetes, a topic with implications for human medicine,
plant disease management and microbial ecology. The importance of signaling to
interactions among bacteria and between bacteria and plants will be explored.
26
Chapter 2 describes an experiment testing for differences among soil streptomycete
populations associated with diverse, virgin prairie soils compared with simplified soils
supporting an annual monoculture of winter wheat.
Chapter 3 describes a detailed evaluation of the implications for biological interpretation
of changing methods for processing second generation sequence data derived from
environmental DNA.
Chapter 4 describes an experiment testing for differences among soil streptomycete
populations associated with particular host plant species, across a gradient of plant
species richness.
Chapter 5 describes a test of plant-soil feedbacks where soil conditioning depends on
two levels of plant manipulation: plant host species, by plant community richness. The
antagonistic activity of soil streptomycetes is characterized in conditioned soils, and
attempts are made to link community antagonistic potential with observed plant
performance in conditioned soils.
The Appendix describes experiments testing the hypothesis that plants produce
compounds capable of modulating streptomycete antibiotic production.
27
Chapter 2: Streptomycete communities in virgin prairie soil vs. never tilled, no-till
annual monoculture
The contents of this chapter have been published as:
M.G. Bakker, J.D. Glover, J.G. Mai, and L.L. Kinkel. 2010. Plant community effects on
the diversity and pathogen suppressive activity of soil streptomycetes. Applied Soil
Ecology 46:35-42.
28
Ecological factors that promote pathogen suppressive microbial communities remain
poorly understood. However, plants have profound impacts on the structure and
functional activities of soil microbial communities, and land-use changes which alter
plant diversity or community composition may indirectly affect the structure and function
of microbial communities. Previous research has suggested that the streptomycetes are
significant contributors to pathogen suppression in soils. We compared soil streptomycete
communities from high and low plant diversity treatments using an experimental
manipulation that altered plant diversity while controlling for soil structure and
disturbance. Specifically, we characterized an isolate collection for inhibition of plant
pathogens as a measure of functional activity, and for 16S rDNA sequence to measure
community structure. In this system, high and low diversity plant communities supported
streptomycete communities with similar diversity, phylogenetic composition, and
pathogen suppressive activity. However, inhibitory phenotypes differed among
treatments for several phylogenetic groups, indicating that local selection is leading to
divergence between streptomycetes from high and low plant diversity communities.
Although the ability to inhibit plant pathogens was common among soil streptomycetes,
pathogen-inhibitory activity differed widely among phylogenetic groups. The breadth and
intensity of pathogen inhibition by soil streptomycetes were positively related.
Introduction
There has been a long-standing interest in the manipulation of microbial communities to
enhance beneficial ecosystem services (Ducklow, 2008; Shennan, 2008). Using natural or
manipulated microbial communities to perform useful functions such as control of plant
disease holds promise for reducing environmental impacts relative to existing resource-
or chemical-intensive methods. For example, the suppression of plant pathogens by
indigenous soil microbes can enhance agricultural productivity and reduce the need for
chemical inputs such as fungicides (Emmert and Handelsman, 1999). Many attempts
have been made to emulate natural pathogen suppression through augmentative and
inoculative biocontrol. Resource manipulation has also been used in attempts to alter
microbial densities and community structure in ways that may limit pathogen activity
29
(Perez et al., 2008; Schlatter et al., 2008). To date, such attempts have had mixed results
in achieving adequate and reliable control of plant pathogens. Additional study of natural
microbial communities is needed to shed light on the factors that influence pathogen
suppression (natural biocontrol) and advance our efforts toward safe and sustainable plant
disease management.
Many soils have been characterized as possessing pathogen suppressive activities (Ghini
and Morandi, 2006; Hjort et al., 2007). Among such systems that have been well-studied,
the production of antimicrobial secondary metabolites has been identified as a significant
factor in effective pathogen suppression (Raaijmakers and Weller, 1998; Weller et al.,
2002). Enrichment of antibiotic producing bacteria in plant rhizospheres has been
demonstrated (Mazzola et al., 2004), as has the ability of plants to differentially promote
antibiotic production by associated bacteria (Bergsma-Vlami et al., 2005a; de Werra et
al., 2007; Okubara and Bonsall, 2008). However, all such studies that we are aware of
relate to 2,4-diacetylphloroglucinol-producing pseudomonads. Plant impacts on other
antibiotic producing bacteria are under explored.
The streptomycetes are ubiquitous members of soil microbial communities and are well-
known as prodigious producers of antimicrobial secondary metabolites (Bentley et al.,
2002). There is a great deal of evidence that free-living Streptomyces can protect plants
by inhibiting the causal organisms of plant disease, and members of this genus have been
studied extensively as biological control agents. For example, Streptomyces isolates have
been shown to reduce the severity of seedling diseases of alfalfa (Jones and Samac,
1996), Phytophthora root rot of soybean (Xiao et al., 2002), potato scab (Liu et al., 1995;
Ryan et al., 2004), Pythium seed and root rots (Yuan and Crawford, 1995), spring black
stem and leaf spot on alfalfa (Samac et al., 2003), pathogenic turf grass fungi
(Chamberlain and Crawford, 1999), root lesion nematodes (Samac and Kinkel, 2001),
and cavity spot disease of carrots (El-Tarabily et al., 1997). Moreover, the frequency,
intensity, and diversity of antibiotic inhibitory interactions among streptomycetes have all
been shown to be important to disease suppression in agricultural soils (Wiggins and
30
Kinkel, 2005a, 2005b; Perez et al., 2008). This suggests that strategies to enhance the
frequency, intensity, or diversity of such competitive interactions may promote disease
control.
Plant nutrient inputs into the soil microbial community, including root exudates,
senescent tissues, leaf litter, and leachates, are likely critical to mediating microbial
competitive interactions through their impacts on resource availability. In particular, total
nutrient inputs will constrain microbial densities and biomass in soil, and the variety of
niches available to soil microbes will depend upon the diversity of the resource base
available to soil food webs, in terms of chemical composition, spatial distribution, and
availability over time. A simplified plant community may be expected to provide a
concomitantly simplified suite of microbial niches as compared with a high diversity
community. Diverse plant cover may also provide opportunities for more diverse species
interactions (including plant-microbe interactions), which are vital to generating
microbial diversity (Thompson, 1999; Hansen et al., 2007). Finally, more diverse plant
communities are generally more productive than less diverse communities (Tilman et al.,
2001), suggesting greater potential resource inputs into soils, correspondingly higher
microbial population densities, and thus more frequent competitive interactions among
soil microbes. On the other hand, plants may realize the greatest benefit and be most
effective in recruiting and sustaining microbial partners for diffuse protective mutualisms
under conditions of low plant diversity. To address these uncertainties, this study
characterizes the diversity and pathogen-suppressive activity of streptomycete
communities from a diverse prairie meadow soil and a simplified agricultural
monoculture soil.
Methods
Site history:
Soil samples were collected from a study site in Ottawa County, Kansas (N’ 38.58.145,
W’ 97.28.616) that has been described in detail elsewhere (Culman et al., 2010; DuPont
et al., 2010; Glover et al., 2010). Prior to the onset of experimental manipulation, this site
31
was virgin prairie and had never been plowed. The site had been burned periodically, and
hayed annually in June or July for approximately the previous 75 years, with the hay
removed from the site. In 2004, three 20 m x 20 m research blocks were established on
the site. Two treatments (prairie meadow or no-till annual cropping) were randomly
assigned to the two halves of each block. Management of the prairie plots remained
consistent with pre-experiment practices; no agricultural inputs were applied and once-
annual removal of hay constituted the only nutrient or biomass removal. Plots assigned to
the no-till annual cropping treatment received 2 applications of glyphosate in the fall of
2004 and were subsequently planted into a rotation of soybean, sorghum, and winter
wheat from 2005 to 2007. Zero-tillage techniques were used exclusively, in order to
minimize the confounding factors of soil disturbance and degradation. However,
chemical fertilizer and herbicide have been used on the monoculture plots according to
standard agronomic practices.
Isolate collection:
In June of 2007, five soil cores were collected from random locations at least 2 m away
from the edge of each plot (2 treatments x 3 blocks). The monoculture plant community
consisted of winter wheat, which was beginning to senesce. The prairie community
consisted of a mixture of forbs, grasses, and legumes. The surface litter layer was
removed and soil was collected to a depth of 10 cm. Soil edaphic characteristics and
further site details for these plots at the time of sampling have been reported elsewhere
(DuPont et al., 2010). Cores were packed on ice and transported to the lab for refrigerated
storage until processing. For colony counts and isolations, soil samples were dried
overnight in a fume hood under 4 ply sterile cheese cloth. A 10% (w/v) soil solution in
K2HPO4/KH2PO4 buffer was shaken for 1 hour at 200 RPM, at 4 C. Samples were
serially diluted prior to plating on water agar (WA) and starch-casein agar (SCA) (Kuster
and Williams, 1964) for determination of culturable community density and selection of
isolates. Plates were incubated for 3 days at 28 C.
32
Ten isolates showing typical streptomycete morphology were collected from each soil
core (five from each of two growth media), for a total of 50 isolates per plot. Selection of
colonies for isolation was performed randomly, according to proximity to predetermined
points on a petri dish. Isolates were purified by repetitively streaking and culturing until
no contaminants were visible. Cultures were stored in 20% glycerol at -80 C.
Pathogen antagonism:
The ability of each of the 300 isolates to inhibit a set of four plant pathogens was assayed
in vitro. The plant pathogens tested were Fusarium graminearum (isolate Butte86 ADA-
11, obtained from R. Dill-Macky), Rhizoctonia solani (isolate 43, AG1, obtained from N.
Anderson), Verticillium dahliae (strain VA33A, vegetative compatibility group 4A,
obtained from N. Anderson), and Streptomyces scabies (Strain RB4, obtained from N.
Anderson). Pathogen overlays followed the method of Wiggins and Kinkel (2005b) with
minor modifications. Briefly, a dense spore suspension of each streptomycete isolate was
spotted (7 microliters per spot, 4 or 5 isolates per plate) onto 1.5% WA (18 mL/plate) and
incubated at 28 degrees C for 2 days. At this point, streptomycete isolates were not yet
differentiating to form aerial mycelium and spores, which could have complicated the
inhibition assay by dispersing the test isolate into the overlay medium. However, this
approach may not detect the full potential for antagonism, as antibiotic production is
regulated in coordination with tissue differentiation in some cases (Horinouchi and
Beppu, 1994).
A second layer of medium (14 mL) was poured over the plates for the pathogen overlays.
Plates were filled with an automatic pipetter to ensure consistent medium depth. For
Fusarium, the entire contents of a fully-colonized petri dish (oatmeal agar, OA, incubated
at room temperature for 7 days) were homogenized in a sterile Waring blender with 100
mL H2O (low speed, 2 x 5 sec, 1 x 10 sec). The resulting slurry was used to inoculate
molten potato dextrose agar (PDA, cooled to 45 C) at a rate of 20 mL inoculum per 500
mL PDA. For Rhizoctonia, liquid cultures in Czapek-Dox (CD) broth (incubated at room
temperature for 7 days) were homogenized in a sterile Waring blender (low speed, 2 x 5
33
sec, 1 x 10 sec) and added to molten CD agar (1% final concentration agar, cooled to 45
C) at a rate of 100 mL inoculum per 500 mL CD agar. For Verticillium, 10 mL of sterile
H2O were added to a sporulating culture (OA, incubated at room temperature for 7 days).
Spores were scraped loose with a sterile loop and decanted into molten PDA (cooled to
45 C) at a rate of spores from two fully-colonized plates per 500 mL PDA. For
Streptomyces, plates were covered with a layer of yeast malt-extract agar (YME; per litre:
4 g yeast extract, 10 g malt extract, 4 g glucose, 10 g Bacto agar). After solidifying, a
dense pathogen spore suspension was spread over the surface. Streptomyces overlays
were inverted and incubated at 28 C, while the other overlays were inverted and
incubated at room temperature for 2 days.
Each isolate-pathogen combination was assayed for inhibitory activity on three separate
plates. Pathogen antagonism was measured as the radius of the zone of inhibition, starting
from the edge of the inhibiting colony; average values were determined for each isolate
over the three plates. In cases where inhibition was evident but did not extend past the
edges of the inhibiting colony, a small non-zero zone size was assigned (0.01 mm).
Inhibitory activity against the four test pathogens was used to assign each streptomycete
isolate to one of 16 possible inhibitory phenotypes; each isolate was given a dichotomous
rating of ‘inhibitory’ or ‘non-inhibitory’ for each of the four test pathogens (42 = 16,
Figure 1). Four-character labels were used to denote these phenotypes, with each
character of the label corresponding to one the four test pathogens. A lower case letter
indicates no inhibition, while an uppercase letter indicates inhibition of that pathogen. For
example, isolates with the phenotype ‘FRVS’ inhibited all four of the pathogens tested,
while isolates with the phenotype ‘frvs’ did not inhibit any of the pathogens.
Community Composition and Diversity:
Streptomycete isolates were cultured in yeast dextrose (YD) broth with 0.5% glycine
(Kieser et al., 2000) on a reciprocal shaker (175 rpm, 28 C) for two to four days.
Genomic DNA was extracted using the Wizard Genomic DNA Purification Kit from
34
Promega, following the manufacturer’s directions. Primers pA and pH (Edwards et al.,
1989) were used to amplify the 16S ribosomal RNA gene by PCR, with the use of PCR
Supermix High Fidelity (Invitrogen). The following thermocycle program was used: 94 C
for 30 sec, 35 cycles of (94 C for 30 sec, 55 C for 30 sec, 70 C for 1 min 40 sec), final
extension step of 72 C for 7 min. PCR products were visualized on an agarose gel.
Products of successful PCRs were purified with the QIAquick PCR Purification Kit
(Qiagen) prior to sequencing. Sequencing was performed with the ABI PRISM 3130xl
Genetic Analyzer, using ABI BigDye version 3.1 Terminator chemistry.
Sequences were edited manually based on the chromatographs using Chromas 2
(http://www.technelysium.com.au/). Sequencing reads of fewer than 500 base pairs were
not included in the analysis. The Classifier function of the Ribosomal Database Project
(Wang et al., 2007) was used to verify the identity of each sequenced isolate. Sequences
can be found in GenBank under accession numbers EU699478-EU699737. Sequences
were aligned with Clustal W (Larkin et al., 2007) and trimmed to the same length (600
nucleotides). The resulting partial 16S ribosomal RNA gene sequence included the V2
and V3 variable regions of the 16S rDNA (Neefs et al., 1990). The V2 variable region
corresponds to the 'gamma' variable region (Stackebrandt et al., 1991) in previous studies
of streptomycete diversity (Kataoka et al., 1997; Anderson and Wellington, 2001).
Generation of a pairwise distance matrix, designation of operation taxonomic units
(OTUs) by the furthest neighbor method, and diversity analyses were performed with the
program mothur (Schloss et al., 2009).
Results
Pathogen Antagonism:
A majority of isolates showed inhibitory activity; sixty-four percent of isolates inhibited
at least one of the four plant pathogens tested. Among our isolates, the frequency of
inhibition against Fusarium, Rhizoctonia, Verticillium, and Streptomyces was 0.24, 0.40,
0.40, and 0.39, respectively. When all isolates were considered, antagonistic activity,
35
including measures of both frequency and intensity, did not differ significantly by isolate
origin (monoculture vs. prairie plant communities; Figure 1).
The inhibition assay was able to distinguish among 16 different inhibitory phenotypes.
All but 3 (‘Frvs,’ ‘FrvS,’ and ‘FRvS’) of the 16 possible phenotypes were observed. The
most frequently observed phenotypes were ‘frvs’ (36% of isolates), ‘frvS’ (13% of
isolates), and ‘FRVS’ (11% of isolates). The distribution of isolates among the
phenotypic groups differed slightly between treatments (Figure 2). A significantly higher
proportion of the monoculture isolates showed no inhibitory activity compared to the
prairie isolates (phenotype ‘frvs’; t = 2.98, p = 0.04). There was a trend toward a higher
proportion of prairie isolates inhibiting Streptomyces only compared to monoculture
isolates (phenotype ‘frvS’; t = -2.53, p = 0.06). Intensity of inhibition within phenotypes
differed only in one case; among isolates with the ‘fRVS’ phenotype, intensity of
inhibition against Streptomyces was significantly greater for prairie isolates than for
monoculture isolates (t = -3.54, p = 0.02).
Although the monoculture community included one phenotype that was not observed
among prairie isolates (‘FRvs’, Figure 2), the prairie community had modestly greater
phenotypic diversity (reciprocal Simpson diversity index (Zhou et al., 2002) of 6.13 vs.
5.48, p = 0.11). The 50 isolates from each plot divided into phenotypes as follows: prairie
plot one (PP1), 11 phenotypes, including three singletons (phenotypes represented by a
single isolate); PP2, 11 phenotypes, two singletons; PP3, 10 phenotypes, one singleton;
monoculture plot one (MP1), 11 phenotypes, three singletons; MP2, 12 phenotypes, four
singletons; MP3, 10 phenotypes, one singleton.
Streptomyces isolates that were able to inhibit a greater number of pathogens were also
better inhibitors. Inhibition of Verticillium was significantly more intense among isolates
which also inhibited two or three other pathogens, compared to those that inhibited
Verticillium alone or along with one other pathogen (Figure 3). A similar, though not
significant, trend existed for inhibition against the other pathogens (Figure 3).
36
Community Composition and Phylogenetic Diversity:
Partial 16S ribosomal RNA gene sequences were obtained for 218 isolates (118 isolates
from the prairie soil and 100 isolates from the monoculture soil) belonging to the family
Streptomycetaceae. Eight isolates were placed in the genus Kitasatospora rather than
Streptomyces; however inclusion of non-Streptomyces isolates in the collection was
expected because morphological screening for isolate selection was deliberately
permissive in order to maximize the captured diversity of culturable streptomycetes.
Streptomyces isolates were grouped into 24 operational taxonomic units (OTUs) based on
a cutoff of 2% sequence dissimilarity using the uncorrected P distance measure with gaps
considered as insertions/deletions. Comparisons of the diversity and community
composition (OTU richness and abundance) of prairie and monoculture streptomycete
communities revealed a high degree of similarity. Isolates from the prairie treatment were
included in 22 of the OTUs, while monoculture isolates were included in 20 OTUs. Four
OTUs included only prairie isolates and two OTUs included only monoculture isolates.
However, each of the OTUs that were exclusive to a single treatment contained only one
or two isolates. Various diversity indices did not differ significantly between the two
communities; for example, the reciprocal Simpson diversity index was 14.6 for the prairie
treatment and 14.1 for the monoculture treatment (p > 0.05).
Taxonomy and Pathogen Inhibition:
Among the larger OTUs (containing at least 10 isolates; n = 9 OTUs), three OTUs
showed differences in inhibitory activities between prairie and monoculture isolates
(Table 1). In OTU 14, monoculture isolates exhibited more intense inhibition against
Fusarium and Rhizoctonia than prairie isolates. In OTU 17, monoculture isolates
exhibited greater inhibitory activity against all four pathogens tested. In OTU 21,
monoculture isolates showed more intense inhibition of Fusarium, while prairie isolates
showed more frequent and intense inhibition of the pathogenic Streptomyces overlay.
37
Independent of isolate origin, phylogenetic groups (OTUs) differed in both overall
inhibitory activity and inhibition of target pathogens. In some OTUs, the majority of
isolates had very limited inhibitory activity against the pathogens tested, while other
OTUs were characterized by isolates having broad inhibitory activity against multiple
pathogens (Figure 4). Intensity of inhibition also differed significantly among OTUs
(Table 2), except against Rhizoctonia. Some OTUs, such as OTU 17, tended toward more
intense inhibition of the test pathogens, while other OTUs, such as OTU 4, tended toward
less intense inhibition. Thus, although OTU did not predict the specific inhibitory
phenotype (such as ‘FrvS’ or ‘fRVs’), the intensity and breadth of inhibitory activities did
differ among taxa.
Discussion
It is recognized that many different selective influences shape soil microbial
communities. These include microbe-microbe interactions (Marshall and Alexander,
1960), chemical and physical aspects of the soil environment (Lauber et al., 2009), and
the influence of particular host plants (Mazzola et al., 2004). We investigated host plant
community as a selective force shaping streptomycete community structure and function.
We found similar diversity, phylogenetic composition, and pathogen suppressive activity
among streptomcyete communities from high and low plant diversity treatments, three
years after the onset of experimental manipulation. However, inhibitory phenotypes
differed among treatments for three taxonomic groups, indicating that local selection is
leading to divergence between streptomycetes from high and low plant diversity
communities. In these groups, inhibitory activity against the fungal pathogens was greater
among monoculture streptomycetes, while prairie streptomycetes showed greater
inhibitory activity against the pathogenic Streptomyces overlay. Our data do not address
the abundance or activity of plant pathogens in situ, but it is tempting to speculate that
greater fungal pathogen activity in the simplified agricultural community has led to
selection for pathogen-inhibitory phenotypes among streptomycetes. It is noteworthy that
opposite trends were observed in some cases, as in the case of OTU 21; low diversity
38
plant cover enhanced inhibition of Fusarium, but high diversity plant cover enhanced
inhibition of Streptomyces.
The ability to inhibit plant pathogens in vitro was common among our collection of
streptomycete isolates. However, the suite of pathogens that each isolate inhibited varied
widely, probably as a function of the quantity and variety of toxic secondary metabolites
produced. The applicability of our in vitro assay for antibiotic production to actual
disease development has been demonstrated previously (Wiggins and Kinkel, 2005b) by
the finding that the density of antagonistic streptomycetes is negatively correlated with
the number of potato scab lesions in field plots. Approaches have been developed to
assess antagonism toward pathogens in situ or under more realistic conditions in
mesocosms (van Elsas et al., 2002). However, these approaches are better suited to
assessments of community inhibitory potential, rather than the antagonistic activity of
isolates. Isolate-based approaches are required for linking pathogen-inhibitory activity to
particular taxonomic groups.
Pathogen-inhibitory activity differed widely among OTUs, with taxonomic groups below
the level of genus contributing unequally to plant pathogen suppression in both prairie
and monoculture soils. This suggests that screening by taxonomy could facilitate the
identification of superior isolates for biocontrol, although the preferred taxa in which to
search for maximum antagonistic potential may vary according to the target pathogen.
Particular phylogenetic groups appear to have characteristic life history strategies, with
some groups employing broad and intense chemical inhibition against competitors, while
other groups displayed very little inhibitory activity. However, the ability to inhibit a
particular combination of pathogens or competitors (inhibitory phenotype) varied widely
among isolates within phylogenetic groups. In this regard, our research supports previous
observations about the variability of inhibitory interactions among soil microbes
(Davelos-Baines et al., 2007). The dominant inhibition phenotypes recovered among our
isolates also suggested the possibility of differing life history strategies among soil
streptomycetes; most isolates either inhibited none of the pathogens tested, inhibited all
39
four of the test pathogens, or inhibited only the pathogenic Streptomyces isolate. These
phenotypes may correspond to a non-antagonistic strategy (perhaps relying instead on an
alternative strategy such as niche differentiation), a broadly antagonistic strategy, or a
strategy of targeted competition against other streptomycetes. Future work should explore
this concept of differing life history strategies among soil streptomycetes.
Our observation that intensity of inhibition against Verticillium (with similar trends for
the other pathogens) increases with breadth of inhibitory activity indicates that isolates
with broader inhibitory capacities may be producing either superior (more effective)
antibiotics, or multiple compounds with additive or synergistic inhibitory effects on
pathogens. Screening against multiple pathogens is thus likely to be beneficial when
prospecting for pathogen antagonists for biocontrol applications because of the potential
for discovery of both the most effective inhibitory antibiotics and for maximizing the
probability for uncovering additive or synergistic antibiotic activities (Challis and
Hopwood, 2003) by isolates that can inhibit multiple plant pathogens. Although not
explored in this study, the importance of species interactions to antibiotic production
(Angell et al., 2006) and successful antagonism of pathogens (Guetsky et al., 2002) are
well known. Future work should explore the effectiveness of mixtures and consortia of
streptomycetes in limiting plant disease.
It is recognized that culture-dependent studies of microbial communities miss the
majority of microbes present, since only a small fraction of microbial cells are readily
culturable (Joseph et al., 2003). While culture-independent techniques allow for more
comprehensive sampling of microbial communities, such techniques are not yet able to
address complex microbial functions such as pathogen suppression. Isolate-based studies
continue to be necessary for providing phenotypic information on microbes, and will
illuminate the results of subsequent culture-independent studies. Furthermore, it is not
clear that resistance to cultivation is equally prevalent among microbial taxa (Nunes da
Rocha et al., 2009), and bias due to cultivation of isolates may be reduced in this study by
the emphasis on a specific group of actinobacteria that appear to be readily culturable.
40
Our work addresses the impacts of plant diversity on a narrow range of the organisms
present in soil microbial communities, and it is clear that additional work is needed to
examine the impacts of plant communities on microbial community composition,
diversity, and functional activity.
Because of the historical absence of tillage at this study site, intact soil structure and high
organic matter content may have buffered the soil microbial community from the effects
of a massive change in plant cover. A legacy effect may exist in the form of labile soil
organic matter, maintaining high resource availability for saprophytic microbial food
webs in the manipulated plots. The Streptomyces are spore-forming bacteria, and
although little information is available regarding the longevity of spores within an active
soil microbial community (Ruddick and Williams, 1972), it is possible that our collection
included isolates that had remained quiescent (not subject to selective forces) since prior
to the onset of experimental manipulation. The rate of change of community composition
would be slowed by the re-entry of community members from a dormant state.
Additionally, temporal plant diversity continues to be a feature of the monoculture
treatment, since the plots are planted in a crop rotation system. Nitrogen inputs were not
equal among the prairie and monoculture treatments. However, total soil nitrogen did not
differ significantly between treatments (DuPont et al., 2010). While chronic nitrogen
additions have been shown to change microbial community composition in some cases
(Compton et al., 2004; Nemergut et al., 2008), other studies have found no effect
(Sarathchandra et al., 2001; DeForest et al., 2004). Because our data are derived from an
isolate collection, several possible impacts of unequal nitrogen application, such as
changes in the frequency of Streptomyces within the broader soil community or changes
in total microbial biomass (Wang et al., 2008) or respiration (Bowden et al., 2004) should
not impact our results. Altered community composition within the streptomycetes was
not observed in our study.
41
Figure 1 – Pathogen antagonism by streptomycete isolates from virgin prairie meadow
and never tilled no-till monoculture plots, based on mean values by block. A) The
proportion of isolates showing inhibitory activity against each of four plant pathogens. B)
The intensity of inhibition (measured as inhibition zone size) against each of four plant
pathogens. Values shown are means and standard errors. No significant differences were
found between treatments (p > 0.1, by t-test).
0
2
4
6
8
10
Fusarium Rhizoctonia Verticillium Streptomyces
Mean
in
hib
itio
n z
on
e s
ize (
mm
)
Pathogen
Prairie
Monoculture
0.0
0.1
0.2
0.3
0.4
0.5
Fusarium Rhizoctonia Verticillium Streptomyces
Pro
po
rtio
n o
f is
ola
tes
inh
ibit
ing
Pathogen
Prairie
Monoculture
42
Figure 2 – Pathogen-inhibitory phenotypes of a streptomycete isolate collection from
diverse prairie (left) and from monoculture (right) plant communities, based on in vitro
inhibition of four test pathogens. Each letter of the phenotype label corresponds to one of
the four test pathogens. An upper case letter indicates inhibition, while a lower case letter
indicates no inhibition of that pathogen. F/f = Fusarium; R/r = Rhizoctonia; V/v =
Verticillium; S/s = Streptomyces. Differences in proportion of isolates belonging to each
phenotype were determined by t-test, assuming equality of variances.
43
Figure 3 – Inhibition zone sizes created by streptomycete isolates against each of four
target pathogens (F=Fusarium, R=Rhizoctonia, V=Verticillium, S=Streptomyces),
according to the number of other pathogens inhibited. Within each panel, box width is
proportional to the number of observations. Differences among means within each panel
were tested with a non-parametric Kruskal-Wallis test, with a multiple comparison test
performed where differences were detected. Significantly different means are indicated
by different letters. No isolate in our collection inhibited Fusarium without also
inhibiting one or more other pathogens.
44
Figure 4 – Pathogen inhibitory characteristics of streptomycete operational taxonomic
units (OTUs), showing proportion of isolates having inhibitory activity against 0, 1, 2, 3,
or 4 of the pathogens tested.
Figure 4
0.00
0.25
0.50
0.75
1.00
4 (n
=23)
6 (n
=21)
7 (n
=25)
9 (n
=10)
11 (n
=10)
14 (n
=13)
17 (n
=27)
19 (n
=14)
21 (n
=17)
OTU
Pro
po
rtio
n o
f Is
ola
tes
Inhibits 4
Inhibits 3
Inhibits 2
Inhibits 1
Inhibits 0
45
Isol
ates
OT
UTr
eatm
ent
Zon
e(m
m)
Zon
e(in
hibi
tors
on
ly; m
m)
Prop
Zon
e(m
m)
Zon
e(in
hibi
tors
on
ly; m
m)
Prop
Zon
e(m
m)
Zon
e(in
hibi
tors
on
ly; m
m)
Prop
Zon
e(m
m)
Zon
e(in
hibi
tors
on
ly; m
m)
Prop
4Pr
airie
140.
192.
630.
070.
402.
780.
141.
053.
690.
292.
024.
700.
43M
onoc
ultu
re9
0.17
0.51
0.33
0.13
0.56
0.22
1.14
2.04
0.56
2.91
4.37
0.67
6Pr
airie
120.
00.
0.00
0.74
4.47
0.17
0.12
1.42
0.08
0.32
3.83
0.08
Mon
ocul
ture
90.
00.
0.00
0.89
4.02
0.22
0.00
.0.
000.
070.
620.
117
Prai
rie11
4.17
6.55
0.64
2.51
2.76
0.91
12.1
313
.34
0.91
1.39
3.05
0.45
Mon
ocul
ture
145.
106.
490.
792.
583.
010.
8612
.55
13.5
20.
931.
844.
300.
439
Prai
rie6
1.75
10.5
20.
173.
793.
791.
003.
977.
940.
502.
563.
840.
67M
onoc
ultu
re4
0.00
.0.
002.
893.
850.
750.
873.
470.
252.
473.
290.
7511
Prai
rie5
0.46
2.32
0.20
1.35
2.25
0.60
0.24
0.59
0.40
0.00
.0.
00M
onoc
ultu
re5
0.00
.0.
001.
662.
080.
801.
032.
580.
400.
00.
0.00
14Pr
airie
83.
84**
4.39
**0.
882.
74**
2.74
**1.
009.
239.
231.
008.
58*
8.58
1.00
Mon
ocul
ture
56.
18**
6.18
**1.
003.
66**
3.66
**1.
008.
818.
811.
007.
56*
7.56
1.00
17Pr
airie
140.
00*
.0.
00*
0.00
*.
0.00
*0.
00*
.0.
00*
0.00
0.01
0.07
Mon
ocul
ture
131.
33*
5.76
0.23
*0.
39*
1.70
0.23
*1.
89*
8.20
0.23
*1.
476.
390.
2319
Prai
rie7
0.00
.0.
002.
362.
361.
000.
00.
0.00
0.54
1.27
0.43
Mon
ocul
ture
70.
00.
0.00
1.43
2.00
0.71
0.00
.0.
000.
491.
140.
4321
Prai
rie10
1.02
2.05
*0.
500.
662.
190.
304.
096.
810.
601.
78*
2.22
0.80
**M
onoc
ultu
re7
1.20
4.19
*0.
291.
214.
230.
292.
295.
340.
430.
00*
0.01
0.14
**
Tabl
e 1
- Inh
ibiti
on d
ata
for i
sola
tes b
elon
ging
to O
TUs w
ith a
t lea
st te
n m
embe
rs. Z
one
is th
e av
erag
e in
hibi
tion
zone
size
for a
ll is
olat
es. Z
one
(inhi
bito
rs o
nly)
is th
e av
erag
e in
hibi
tion
zone
size
for o
nly
thos
e is
olat
es w
hich
are
pos
itive
for i
nhib
ition
of t
he te
st p
atho
gen.
Pro
p is
the
prop
ortio
n of
isol
ates
whi
ch a
re p
ositi
ve fo
r inh
ibiti
on o
f the
test
pat
hoge
n. *
indi
cate
s tha
t the
mea
ns fo
r pra
irie
and
mon
ocul
ture
isol
ates
diff
er
at p
< 0
.10
(ttes
t, as
sum
ing
equa
lity
of v
aria
nces
). **
indi
cate
s diff
eren
ces s
igni
fican
t at p
< 0
.05.
Fusarium
Streptomyces
Verticillium
Rhizoctonia
46
OT
UIs
olat
esPr
opor
tion
of is
olat
es
inhi
bitin
g
Inhi
bitio
nzo
ne (m
m)
Prop
ortio
nof
isol
ates
in
hibi
ting
Inhi
bitio
nzo
ne (m
m)
Prop
ortio
nof
isol
ates
in
hibi
ting
Inhi
bitio
nzo
ne (m
m)
Prop
ortio
nof
isol
ates
in
hibi
ting
Inhi
bitio
nzo
ne (m
m)
423
0.17
1.0
4 b
0.17
1.67
a0.
39 2
.78
c0.
524.
54 b
621
0.00
.0.
194.
24 a
0.05
1.4
2 bc
0.10
2.23
b7
250.
72 6
.51
a0.
882.
90 a
0.92
13.4
4 a
0.44
3.73
b9
100.
1010
.52
a0.
903.
81 a
0.40
6.8
2 bc
0.70
3.60
b11
100.
10 2
.32
ab0.
702.
15 a
0.40
1.5
8 c
0.00
.14
130.
92 5
.13
ab1.
003.
09 a
1.00
9.0
6 b
1.00
8.18
a17
270.
11 5
.76
ab0.
111.
70 a
0.11
8.2
0 ab
c0.
154.
79 a
b19
140.
00 .
0.86
2.21
a0.
00 .
0.43
1.21
b21
170.
41 2
.66
b0.
293.
01 a
0.53
6.3
2 bc
0.53
1.98
b
Fusarium
Rhizoctonia
Verticillium
Streptomyces
Tabl
e 2
- Inh
ibiti
on d
ata
for i
sola
tes b
elon
ging
to O
TUs w
ith a
t lea
st te
n m
embe
rs. S
how
n ar
e m
ean
valu
es fo
r in
hibi
tion
zone
size
, con
side
ring
only
thos
e is
olat
es w
hich
succ
essf
ully
inhi
bit t
he p
atho
gen.
Diff
eren
t let
ters
indi
cate
si
gnifi
cant
diff
eren
ces b
etw
een
mea
n va
lues
with
in th
e co
lum
n (p
< 0
.05,
AN
OVA
with
Tuk
ey m
ultip
le te
st
corr
ectio
n).
47
Chapter 3: Accounting for sequencing errors during processing of 454 pyrosequence
data
The contents of this chapter have been submitted for publication as:
M.G. Bakker, Z.J. Tu, J.M. Bradeen, and L.L. Kinkel. Implications of pyrosequencing
error correction for biological data interpretation. PLoS ONE
Revisions and additional work requested by reviewers are currently in progress.
48
There has been a rapid proliferation of techniques and approaches for processing and
manipulating second generation DNA sequence data. However, there has not been
sufficient evaluation of these methods in order to detect unintended biases. In particular,
real and complex datasets should be used in detailed explorations of the implications of
various processing methods for the biological interpretation of experimental results. In
this report, we consider the PyroNoise algorithm, a recently reported strategy for 454
pyrosequencing error correction, and its implications for the biological interpretation of a
dataset composed of 60 independent soil microbial community samples. We report
subtleties in the effects of this method that should be considered in its use. Specifically,
reductions in OTU richness and abundance by the algorithm are sensitive to the structure
of the input dataset. Impacts of processing on conserved vs. variable sequence characters
were distinguished. Clustering of samples based on community-level similarity measures
was dramatically impacted by PyroNoise processing.
Introduction
Current DNA sequencing capacity offers the opportunity to study microbial communities
in unprecedented detail. However, quality standards often lag behind technical innovation
and many initial reports of microbial diversity and community composition using second
generation sequencing now appear to have been substantially overestimated (Quince et
al., 2009). New studies of large-scale sequencing of environmental DNA are expected to
follow more stringent standards for quality control and experimental design (Prosser,
2010) than were included in many initial reports based on second generation sequencing.
An expanding list of criteria has been proposed to screen out low quality reads from
pyrosequencing datasets (Huse et al., 2007; Kunin et al., 2010), but these have not proven
adequate to eliminate spurious diversity. Although high sequencing accuracy can be
achieved by removing reads that are most likely to contain errors, low error rates may
still accumulate to substantial effect in datasets with hundreds of thousands (or more)
sequence reads.
49
One approach to dealing with this problem has been to simply shed detail from a dataset
until there is a high probability that the influence of PCR or sequencing errors has been
removed. This can be seen, for example, in the use of broad criteria for delimiting
operational taxonomic units (OTUs) and in approaches that discard all of the least-
frequently occurring sequence variants (Zaura et al., 2009).
A preferable approach would be to devise means of identifying and correcting errors such
that accurate detail can be maintained in the dataset. The PyroNoise algorithm (Quince et
al., 2009) was reported as such a method for pyrosequencing error detection and
correction, and was quickly incorporated into a pipeline for analysis of high-throughput
community sequencing data (Caporaso et al., 2010). Indeed, new methods for processing
second generation sequencing data are being introduced at an astonishing rate, in step
with increasing sequencing capacity. It remains a challenge, however, to harness the
energy accompanying this rapid technical development in pursuit of meaningful
biological and ecological hypothesis testing. Toward this end, there is a clear need for
greater and more detailed exploration of the implications of processing methods for the
biological interpretation of second generation sequence data. Such evaluations will
quickly bear fruit by guiding refinements to data processing methods and ensuring
appropriate use of existing methods.
In this report, we give careful consideration to interpretive implications and unintended
biases of the PyroNoise processing method on an original dataset consisting of 60
independent soil microbial community samples. We reveal several nuances in the effects
of PyroNoise processing on data analysis and interpretation which should be considered
during the use of this procedure.
Materials and Methods
Sampling was performed at the Cedar Creek Ecosystem Science Reserve (CCESR; part
of the National Science Foundation Long-Term Ecological Research network) in July of
2009, from experimental plots that have been maintained in a long-term plant richness
50
manipulation (Tilman et al., 2001). These experimental plots were established with
defined levels of plant richness. While other colonizing species are removed from the
plots, the experimental manipulation does not control plant diversity per se, as the
relative abundance of the planted species is allowed to fluctuate as a result of natural
processes. We targeted soil under the dominant influence of each of four different plant
species (two C4 grasses: Andropogon gerardii, Schizachyrium scoparium; two legumes:
Lespedeza capitata, Lupinus perennis) by collecting soil cores from as close as possible
to the base of individual plants. Each sample consisted of four bulked soil cores, collected
to a depth of 30 cm using a 5 cm diameter soil corer and homogenized by hand. A
subsample was passed through a 2 mm screen and stored at -80C until DNA extraction.
Each plant species was sampled in five different plant richness treatments (monoculture
and assemblages of 4, 8, 16 or 32 species). There were three plot-level replicates per
host-community richness combination, except for monocultures of A. gerardii and L.
perennis, for which only two plot-level replicates were available. Two separate soil
samples were processed from one of the plots in these cases. Thus we had a total of 60
soil samples (4 plant hosts x 5 plant community diversity levels x 3 replicates).
The PowerSoil DNA kit (MO BIO; Carlsbad, CA USA) was used to extract DNA from
soil, with minor modifications to enhance recovery of DNA from target taxa (Schlatter et
al., 2010). Targeting particular taxa with optimized DNA extraction procedures is likely
to minimize the introduction of bias due to variable extraction efficiency. Extracted DNA
was passed through the QIAquick PCR Purification Kit (Qiagen; Valencia, CA USA) and
quantified with a NanoDrop ND 1000 spectrophotometer (Thermo Fisher Scientific;
Waltham, MA USA). PCRs consisted of 10 ng of template DNA in a 50 uL reaction
volume using PCR Supermix High Fidelity (Invitrogen; Carlsbad, CA USA). We used
StrepB (Rintala et al., 2001) as our forward primer, and the reverse complement of
Act283 (McVeigh et al., 1996) as our reverse primer, each at a final concentration of 200
nM. Both primers are selective for Actinobacteria and together amplify a fragment of
approximately 165 nucleotides, encompassing the V2 variable region of the 16S rRNA
gene, which is referred to as the 'gamma variable region' in some studies of
51
streptomycetes (Stackebrandt et al., 1991). Primers were modified to contain one of 30
different 10mer identifying barcodes (Parameswaran et al., 2007). PCR conditions
consisted of an initial denaturation step of 30 sec at 94 C, followed by 30 cycles of 30 sec
94 C, 30 sec 57 C, 60 sec 70 C. Products of PCRs were passed through the QIAquick
PCR Purification Kit, quantified by spectrophotometry, diluted with elution buffer to
approximately 15 ng/uL, and quantified by fluorometry (Quant-iT dsDNA HS assay kit;
Invitrogen). Thirty samples, each with a unique primer barcode, were combined in
equimolar amounts to form each of two pooled amplicon samples. Emulsion PCR and
sequencing were performed using a GS FLX emPCR amplicon kit according to the
manufacturer’s protocols (454 Life Sciences; Branford, CT USA). Each pooled sample
was run on one region of a picotitre plate on the GS FLX sequencing system (Droege and
Hill, 2008) at the University of Minnesota BioMedical Genomics Center. Resulting
sequence data have been submitted to the NCBI Sequence Read Archive as accession
SRA019985.3.
The dataset was processed through PyroNoise on a per sample basis. PyroNoise program
scripts were modified to match our forward primer and to set a minimum sequence length
of 100 nucleotides. Other settings were used at their default values (basecalling stops at
the first read with a flowgram value between 0.5 and 0.7, parameters for expectation-
maximization algorithm: sigma=1/15, c=0.05). The PyroNoise output provides a mapping
of input sequences to de-noised output sequences. We generated a single output file per
sample by duplicating each output sequence according to the number of input sequences
mapping to that output sequence.
Subsequent processing of PyroNoise output, and all processing for the standard output
comparison, was done through the program mothur (v. 1.9 and 1.10; (Schloss et al.,
2009)). Initial quality screening removed sequences with an imperfect match to the
forward primer, an unexpected length, any ambiguous bases, homopolymeric runs of
more than six nucleotides, and average sequence quality scores of less than 20 (for
standard output). After initial screening, each dataset was simplified to include only
52
unique sequences. These were aligned to the Silva reference database (Pruesse et al.,
2007) using kmer searching with a ksize of 6 to find the best template sequence and using
the Needleman-Wunsch pairwise alignment method (Needleman and Wunsch, 1970) with
a reward of +1 for a match and penalties of -1 and -2 for a mismatch and gap,
respectively. Aligned sequences were screened for chimeric sequences using the Pintail
method (Ashelford et al., 2005), as well as for sequences belonging to phyla other than
the Actinobacteria (Table 1).
The per nucleotide error rate implied by PyroNoise processing was calculated as the
average distance between the input and output sequences, where distance is defined as the
number of base differences between the two sequences divided by the length of the
shortest sequence, where terminal gaps are ignored and each internal gap contributes a
length of one. Alignments were made with ClustalW (Larkin et al., 2007) and distance
was calculated using mothur.
To determine the number of corrections made by PyroNoise at conserved vs. variable
positions, the sequences output by PyroNoise and by standard processing were combined
and aligned to a set of reference sequences derived from Actinobacterial type strains,
which was obtained from the Ribosomal Database Project Hierarchy Browser (Cole et al.,
2009). The two datasets were split apart after alignment, and the frequency of each
character state (A/C/G/T/gap) was calculated at each position in the alignment for each
dataset. The variance of the proportions of each character state was used as a measure of
conservation at each position; with this measure, perfectly conserved positions will have
the maximal variance value of 0.2, while less conserved positions will have a lower
variance. We defined conserved positions (65% of the characters in the sequence
alignment) as those for which the variance of the proportion values across the five
character states was ≥ 0.18 in either the PyroNoise or the standard processing dataset; the
remaining positions were considered to be variable.
53
We used the Classifier function of the Ribosomal Database Project (Wang et al., 2007) to
assign sequences to a taxonomic category. A bootstrapping confidence threshold of 50%
was used, which has been shown to be appropriate for classifying partial sequences
shorter than 250 nucleotides in length to the level of genus (Claesson et al., 2009).
Taxonomic assignment was performed for the ten samples having the highest PyroNoise
implied error rate, as these were considered the most likely to have been significantly
impacted by PyroNoise processing.
Dissimilarity matrices were calculated using the Vegan package for R (Oksanen et al.,
2010). For Bray-Curtis dissimilarity calculation, OTU abundances were first relativized
into proportions. Classical (metric) multidimensional scaling was performed with the
“cmdscale” function in R. Clustering patterns were determined visually.
Results and Discussion
Within our dataset, we found that net yield of quality-checked sequences was very
modestly greater with de-noising than without (Table 1; net sequence reads per sample 3
891 vs. 3 834; tpaired= -2.22, p = 0.03). Thus, in terms of net sequence yield, PyroNoise
processing offers an advantage over other methods of accounting for low quality
sequence reads. Most of the established criteria for sequence quality screening were
greatly reduced in effect after de-noising (Table 1), indicating a high degree of overlap in
the subset of sequence reads targeted by both processing methods.
Among the 242 718 sequences that were output after de-noising, 68.6% were unchanged
by the PyroNoise algorithm. Among sequences that were changed by the algorithm, the
average implied error rate was 1.51%, equal to approximately 2.5 corrections per
sequence. “Implied error rate” indicates that, for this dataset derived from unknown
organisms, we cannot know whether the changes made during de-noising correspond to
true sequencing or PCR errors. The uneven division of implied errors across sequence
reads is in keeping with previous reports that pyrosequencing errors tend to be clustered
in a subset of sequence reads (Huse et al., 2007). The overall error rate implied by
54
PyroNoise processing for our dataset was 0.47%, which is comparable to published
pyrosequencing error rate estimates (0.12% to 0.50%; (Huse et al., 2007; Quinlan et al.,
2008; Droege and Hill, 2008)). However, implied error rates across our samples ranged
from 0.17% to 2.61%. At the high end of this range, de-noising at a rate well above the
expected true error rate could mask legitimate biological variation. Indeed, by accounting
for spurious diversity without considering all sources of error, PyroNoise may be overly
aggressive in reducing sequence variation. For example, imperfections in multiple
sequence alignments can also contribute substantially to inflated estimates of sequence
diversity (Sun et al., 2009).
Technical factors may contribute to genuinely higher error rates in some samples
compared to others (Wilson, 1997). However, the wide range of implied error rates across
our samples may also be an indication of the sensitivity of PyroNoise to differences in the
structure of the input sequence set. Indeed, PyroNoise implied error rate and Shannon
diversity index were significantly negatively correlated among samples (r2=0.27, p <
0.01; data not shown), although this relationship was heavily influenced by one data point
having unusually low diversity and an unusually high implied error rate.
OTU-based Microbial Community Analysis
When characterizing complex and unknown communities by DNA sequencing, it is
common to bin sequences into OTUs based on a simple sequence similarity threshold.
Processing methods will impact the conclusions reached in such studies only if they
change patterns of sequence grouping during OTU formation. As expected, de-noising
dramatically reduced the number of OTUs formed (Table 1). However, the effects of de-
noising on OTU richness and diversity were nuanced. The reduction in observed OTU
richness as a result of de-noising varied with OTU richness (Figure 1A); a higher
proportion of OTUs were removed from samples with a higher OTU richness. Impacts on
OTU diversity were also sensitive to the structure of the input dataset; de-noising reduced
the Shannon diversity index to a greater extent in more diverse samples compared to less
diverse samples (Figure 1B). Thus de-noising limited the dispersion in OTU richness and
55
diversity among samples, with potent implications for studies comparing microbial
communities from different treatments or environments.
The positive intercepts in Figure 1 suggest that OTU richness and diversity could actually
be increased in samples with very low initial OTU diversity, although this extrapolates
beyond our data. While PyroNoise processing reduces sequence variation and will
typically lower OTU diversity in a sample, it may be possible for a reduction in sequence
variation to result in an increase in OTU diversity. The PyroNoise algorithm evaluates
individual sequences in the context of all other sequences in the sample, not in the
context of their OTU neighbors. De-noising could result in base changes that reduce the
frequency of rare sequence variants while simultaneously moving individual sequences
farther away from their nearest neighbors, creating novel divisions among OTUs.
Beyond simply describing the diversity or structure of microbial communities, many
studies aim to compare microbial communities from various treatments, locations, or
environments. One common approach is to calculate pairwise community dissimilarities
using various indices which take into account community composition and structure (e.g.,
the identity and relative abundances of OTUs present). The resulting dissimilarity matrix
can be subjected to clustering methods in order to observe patterns of similarity among
samples. With such an analysis, de-noising had a dramatic effect on clustering of samples
in our dataset (Figure 2). We examined two different indices for quantifying community
dissimilarity: the presence-absence-based Jaccard index (measuring similarity in
community composition) and the abundance-based Bray-Curtis index (measuring
similarity in community structure).
Without de-noising, our samples could not be separated into distinct clusters based on
community composition (Figure 2A), likely as a result of many uncommon OTUs found
in only one or a few samples. In contrast, after de-noising the samples could be clustered
into two clear groups (Figure 2B). For an index based on community structure, the
majority of samples fell into a single tight cluster without de-noising (Figure 2C). Two
56
additional clusters, together including 18 samples, could be distinguished. In comparison,
after de-noising, only 10 samples fell outside of the main cluster (Figure 2D). There was
relatively little overlap in the identity of samples falling outside of the main cluster in the
de-noised vs. not de-noised datasets; only four of the 10 samples falling outside the main
cluster in the de-noised dataset were similarly placed in the not de-noised dataset. Thus
de-noising switched the categorization of samples in both directions between clusters
derived from a community similarity metric. Importantly, however, de-noising led to the
same clustering patterns when either a presence-absence or an abundance-based
community similarity metric was used.
Thus PyroNoise processing has the potential to impact the conclusions drawn from
experimental data at a more fundamental level than through simple adjustments to
estimates of microbial richness and diversity. To date, the criteria for evaluating
pyrosequencing processing methods have focused almost exclusively on the ability to
successfully recreate the correct number of OTUs. Our results show that this criterion is
too simple and does not account for other effects that may accompany the choice of a
data processing method.
For our dataset, a similar number of OTUs could be formed without de-noising by simply
increasing the dissimilarity threshold for OTU formation; there was a clear
correspondence in our dataset between OTU clustering at X% after de-noising and OTU
clustering at (X+3)% without de-noising. For OTUs defined at 3% or greater dissimilarity
after de-noising, this relationship produced a similar number of sequence clusters (Figure
3A), having a similar abundance distribution between the two processing methods (two-
sample Kolmogorov-Smirnov test on the proportions of the 400 most abundant OTUs for
de-noised data at 3% dissimilarity and non-de-noised data at 6% dissimilarity; D=0.0425,
p=0.86). However, this correspondence in the number of OTUs formed with and without
de-noising differed among samples and between samples and the entire dataset (compare
panel A and panel B in Figure 3). The interpretive impacts of lumping diversity into
57
broader OTU categories are thus likely to differ from the impacts of choosing a de-
noising strategy to account for PCR and sequence error.
Classification-based Microbial Community Analysis
Taxonomic identification of sequence reads is an alternative and complementary
approach to OTU-based methods. However, not even the largest sequence databases are
comprehensive, and it remains impossible to confidently categorize many sequence reads.
Sequences that match poorly to existing databases are commonly interpreted as
representing novel diversity. However, an alternative explanation is that sequence error
contributes to novelty. If this is the case, the possibility of correcting sequence errors may
also enhance our ability to assign sequence reads derived from environmental DNA to
taxonomic categories.
In our dataset, de-noising increased the proportion of sequences that could be classified to
suborder, family, and genus (Figure 4A). Improved matching to quality-checked
databases (which are not referenced during processing) reflects favorably on the changes
made by PyroNoise and supports the idea that substantial amounts of the variation
reported as novelty may actually be erroneous. De-noising increased homology at
conserved nucleotide positions (Figure 4B; see methods for a detailed explanation),
which are likely to be the determinants of higher-level classification. At the same time,
the determinants of fine-scale differentiation among sequences remained intact, as
variable sites were handled without a consistent bias toward reduced diversity (Figure
4B). This is consistent with the idea that error correction by de-noising removes spurious
diversity from nucleotide positions that carry phylogenetic signal and can be used in
classification.
Conclusions
We give careful consideration to the implications of de-noising pyrosequence data for
biological data interpretation. Our results suggest the possibility of dramatic interpretive
implications when real, complex datasets are de-noised. PyroNoise de-noising
58
substantially reduced OTU richness and diversity, but was sensitive to features of the
input dataset when doing so. De-noising reduced the dispersion in OTU richness and
diversity across samples. Rates of de-noising differed widely among samples and were
weakly related to the input diversity within samples. Further investigations with similar
de-noising strategies should give more attention to the relative importance of actual
differences in error rate among samples versus impacts of the structure of the input
dataset on algorithm behavior. Patterns of similarity among samples were heavily
impacted by de-noising. From a sequence classification perspective, de-noising enhanced
homology at conserved nucleotide sites and increased the proportion of reads that could
be classified at taxonomic ranks as fine as the genus.
Although the rapid turnover of techniques used in microbial community analysis provides
a disincentive for researchers to invest time and energy in careful evaluation of specific
data processing methods, it is vital that such studies are undertaken. All methodologies
are likely to contain biases that may only become clear after detailed comparisons with
other methods. These biases should be revealed upfront, as methodological impacts on
the conclusions of studies can be carried into the literature and few datasets are ever
rigorously re-evaluated with updated methodologies. Furthermore, detailed analysis of
data processing methods can inform refinements to algorithms and may demonstrate the
need for entirely new data handling procedures.
59
Figure 1 – Changes in OTU richness and diversity as a result of de-noising
A) Relationship between OTU richness without de-noising and the change in OTU
richness (expressed as a percentage of the original) as a result of de-noising, by sample
(n=60). B) Relationship between OTU diversity (as calculated with the Shannon index)
without de-noising and the change in OTU diversity (expressed in Shannon diversity
index units) as a result of de-noising, by sample (n=60). OTUs were defined at 3%
dissimilarity both with and without de-noising.
A)
-80
-75
-70
-65
-60
-55
-50
-45 200 300 400 500 600 700
% c
han
ge in
OTU
ric
hn
ess
by d
e-n
ois
ing
Observed OTU richness without de-noising
r2 = 0.36
60
Figure 1, continued
B)
-‐1.5
-‐1.25
-‐1
-‐0.75
-‐0.5
-‐0.25
0 2 2.5 3 3.5 4 4.5 5 5.5
Change in OTU diversity (Shannon index units)
by de-noising
Observed OTU diversity (Shannon index) without de-noising
r2 = 0.33
61
Figure 2 – Impacts of de-noising on sample clustering
Classical multidimensional scaling plots derived from Jaccard (presence-absence based;
A, B) or Bray-Curtis (abundance based; C, D) dissimilarity matrices for data with (B, D)
and without (A, C) de-noising. In this representation, the distances between points are
approximately equal to dissimilarities. Samples are numbered consistently across panels,
using numbers 1 through 60.
62
1
2
3
4
5 6
7
8
9
10
11
12
13
14
15
16
17 18 19
20
21
22
23
24 25 26
27
28
29
30
31
32
33
34
35
36 37
38
39
40
41
42
43
44 45 46
47 48
49
50
51
52
53
54 55
56
57
58 59
60
-0.3
-0.2
-0.1
0
0.1
0.2
0.3
0.4
-0.6 -0.5 -0.4 -0.3 -0.2 -0.1 0 0.1 0.2
B - De-noised; Jaccard index
1
2
3 4
5
6 7 8
9 10
11
12
13
14 15
16 17
18
19
20
21 22 23
24
25 26
27
28
29
30
31
32
33 34
35
36
37
38
39
40
41 42
43 44
45
46
47
48
49 50
51
52
53
54
55
56 57
58
59
60
-0.3
-0.2
-0.1
0
0.1
0.2
0.3
-0.3 -0.2 -0.1 0 0.1 0.2 0.3 0.4
A - Not de-noised; Jaccard index
Figure 2, continued
63
1
2
3 4
5
6 7 8
9
10
11
12 13
14 15
16
17
18
19
20
21
22
23
24
25
26 27
28
29
30 31
32
33
34
35
36
37 38
39
40
41 42
43
44
45
46
47
48
49
50
51
52 53
54
55
56
57 58 59
60
-0.3
-0.2
-0.1
0
0.1
0.2
0.3
0.4
0.5
0.6
-0.7 -0.6 -0.5 -0.4 -0.3 -0.2 -0.1 0 0.1 0.2 0.3
C - Not de-noised; Bray-Curtis index
1
2 3
4
5 6
7 8 9
10
11
12
13
14
15
16
17
18 19 20
21
22
23
24 25
26
27
28
29
30
31
32
33
34
35
36
37 38
39
40
41
42
43 44
45
46 47
48
49
50
51
52
53 54
55
56
57
58
59
60
-0.4
-0.3
-0.2
-0.1
0
0.1
0.2
0.3
0.4
-0.7 -0.6 -0.5 -0.4 -0.3 -0.2 -0.1 0 0.1 0.2 0.3
D - De-noised; Bray-Curtis index
Figure 2, continued
64
Figure 3 – Comparison of OTU richness estimates after de-noising vs. broadening
OTU cutoff thresholds
Rarefaction curves for various OTU dissimilarity cutoffs after de-noising (shades of red)
or without de-noising (shades of green). Panel A: Curves were generated using the entire
60 sample dataset. Panel B: Curves were generated using a single, arbitrarily selected
sample from our dataset. The same legend applies to both panels.
A)
65
Figure 3, continued
B)
66
Figure 4 – Proportion of sequences that could be categorized at different taxonomic
ranks, with and without de-noising
A) For the 10 samples with the highest PyroNoise implied error rate, the proportion of
sequences that could be classified to the taxonomic rank shown. Mean and standard
errors are shown. Differences at each taxonomic rank are significant (t-test, p < 0.01). B)
Histograms showing the change resulting from de-noising in the variance of the
proportion of sequences having each character state, for each aligned nucleotide site.
Observations with values greater than zero indicate increased sequence homology across
the dataset at that position as a result of de-noising. Left panel, conserved sites. Right
panel, variable sites. The X-axis scale differs between panels.
A)
0
0.2
0.4
0.6
0.8
1
Order Suborder Family Genus
Prop
ortio
n of
sequ
ence
s ass
igne
d
Taxonomic Rank
Not de-noised
De-noised
67
Figure 4, continued
B)
68
Table 1 - Number of sequence reads failing quality screening criteria and total number of
sequence reads remaining (bold, italics), for standard processing pipeline and for
PyroNoise processing. The sum of reads failing each criterion in the initial screening is
greater than the number of reads dropped because some reads failed on multiple criteria.
PyroNoise processing includes a test of matching to the 5' primer, and does not make use
of quality scores. OTUs were defined based on a 3% sequence dissimilarity threshold,
using the furthest neighbor method.
Screening Criteria Standard Output
PyroNoise Output
Initial 409 997 242 718
< 100% match to 5' primer 31 948 N/A
Unexpected sequence length 153 915 525
Ambiguous bases present 36 113 14
Homopolymers > 6 bases 32 073 0
Avg Quality Score < 20 32 174 N/A
Remainder after 1st stage screening 254 654 242 188
Uniques 28 746 4 522
Insufficient match to be aligned 14 3
Flagged as chimeras 24 551 8 707
Phyla other than target 29 28
Net sequence read yield 230 060 233 450
Sequence reads per sample (+/- SE) 3 834 +/- 98 3 891 +/- 96
Unique sequence reads 26 452 4 326
OTUs 3 858 1 329
OTUs per sample (+/- SE) 433 +/- 9.4 143 +/- 2.4
69
Chapter 4: Impacts of plant host and plant community richness on soil Actinobacterial
community structure
70
Introduction
Empowered by new tools that allow for deeper sampling, much current research in
microbial ecology aims to describe the structure of microbial communities in diverse
environments. In particular, culture-independent approaches using bulk community DNA
have revealed an astonishing diversity of microbes in environments such as soil,
seawater, and the digestive tracts of higher animals (Acosta-Martinez et al., 2008; Gilbert
et al., 2009; Qin et al., 2010). Elucidating the forces that structure and maintain microbial
diversity is one of the central tasks of the discipline of microbial ecology. While the
importance of abiotic environmental characteristics, biotic interactions, and stochastic
events (eg. immigration, order of colonization) have been highlighted (Palacios et al.,
2008; Chase, 2010; Caruso et al., 2011), our understanding of the determinants of
microbial community structure and composition remains limited. We address these gaps
in knowledge by characterizing Actinobacterial community composition from dozens of
samples over a field scale and closely probing the specific impacts of plant host across a
gradient of plant species richness.
For soil microbial communities, factors such as pH (Fierer and Jackson, 2006), parent
material (Ulrich and Becker, 2006), and plant community or host plant genotype (Innes et
al., 2004; Marschner et al., 2004; Garbeva et al., 2008) have been shown to be important
determinants of community structure. However, there are likely to be interactions among
these and other factors. We focus on host plant effects, and test the hypothesis that plant
community context modulates the impacts of a given host plant species on associated soil
microbial communities. Specifically, we compare Actinobacterial community structure
from soils associated with particular host plant species when grown in monoculture or in
plant communities of increasing richness.
Impacts of changing plant diversity on microbial community composition have been
documented previously (Carney and Matson, 2006). However, it has been difficult to
distinguish between effects due to diversity per se and effects due to the increasing
likelihood of the presence of particular plant species having a strong effect (Wardle,
71
1999). One way around this dilemma is to investigate the impacts of particular plant
species across a gradient of plant richness; if host species have consistent impacts on soil
microbial communities regardless of the richness or diversity of the surrounding plant
community, then the impacts of plant diversity may be limited to the additive effects of
individual host species. On the other hand, if a changing plant community context alters
the impact of individual plant hosts, then plant diversity may be an important variable in
its own right for understanding soil microbial community structure and dynamics.
Theoretically, plants may experience incentives for altered interactions with soil
microbial partners when growing in isolation versus in more diverse communities. In
particular, if the development of a beneficial microflora requires costly inputs, there may
be selection against such investment where neighboring plants could share the benefits
without incurring costs (Strassmann et al., 2000; West et al., 2002). There are multiple
mechanisms, both direct and indirect, by which plants may exert selection on associated
soil microbes. For example, the provision of specific chemical compounds may offer a
selective advantage to organisms with the optimal enzymatic capabilities for accessing
those substrates, while bioactive molecules in root exudates may directly inhibit
particular microbial taxa (Broeckling et al., 2008; Badri et al., 2009; De-la-Pena et al.,
2010). Plant compounds may act as signals that trigger changes in microbial gene
expression (Mark et al., 2005; Weir et al., 2008) and alter outcomes of competitive
interactions among microbes. Significantly, the results of such plant-driven selection may
feedback in ways that impact plant fitness. Although there is a body of research that
investigates these plant-microbe feedbacks (Olff et al., 2000; Reynolds et al., 2003;
Casper et al., 2008; McCarthy-Neumann and Kobe, 2010b), insufficient attention has
been given to the possibility that feedback dynamics for individual plant hosts may differ
as a function of the broader plant community context.
For the censusing of complex soil microbial communities, depth of coverage remains
shallow even where tens or hundreds of thousands of DNA sequences are sampled, and
many datasets are unable to address the biogeography of lower-order microbial taxa.
72
Indeed, we have extremely limited information on dispersal abilities and biogeography or
degree of endemism for the majority of microbial taxa, and this places a major constraint
on the debate over the relative importance of deterministic vs. stochastic forces in
shaping microbial communities. To address this gap in knowledge, this work also offers
general insights into the structure and variability among soil Actinobacterial
communities. The Actinobacteria are among the dominant members of soil bacterial
communities and are well known for their role in organic matter decomposition and for
the production of diverse antibiotics and other secondary metabolites (Genilloud et al.,
2011).
Here, we combine taxonomically-selective PCR with massively parallel sequencing to
investigate Actinobacterial community composition and biogeography over a field scale,
with manipulation of plant cover occuring across two factors: plant host species and plant
community richness. This work provides fundamental insight into the structure and
variability among soil Actinobacterial communities. We test plant species identity and
plant richness as separate and interacting factors shaping associated soil microbial
communities.
Methods
Sampling was performed at the Cedar Creek Ecosystem Science Reserve (CCESR; part
of the National Science Foundation Long-Term Ecological Research network) in July of
2009, from experimental plots that have been maintained in a long-term plant richness
manipulation (Tilman et al., 2001). These experimental plots were established in 1994
with defined levels of plant richness. For plant richness treatments of up to 16 species,
plant species were drawn from a pool of 16 core native prairie plant species. For 32
species-plots, additional plant species were included beyond the pool of 16 species used
for the lower richness treatments (Tilman et al., 1997). While other colonizing species are
removed from the plots, the experimental manipulation does not control plant diversity
per se, as the relative abundance of the planted species is allowed to fluctuate as a result
of natural processes. We targeted soil under the dominant influence of each of four
73
different plant species (two C4 grasses: Andropogon gerardii, Schizachyrium scoparium;
two legumes: Lespedeza capitata, Lupinus perennis) by collecting soil cores from as
close as possible to the base of individual plants. Each sample consisted of four bulked
soil cores, collected to a depth of 30 cm using a 5 cm diameter soil corer and
homogenized by hand. A subsample was passed through a 2 mm screen and stored at -
80C until DNA extraction. Each plant species was sampled in five different plant richness
treatments (monoculture and assemblages of 4, 8, 16 or 32 species). There were three
plot-level replicates per host-community richness combination, except for monocultures
of A. gerardii and L. perennis, for which only two plot-level replicates exist. Two
separate soil samples were processed from one of the plots in these cases. Thus we had a
total of 60 soil samples (4 plant hosts x 5 plant community diversity levels x 3 replicates).
Culturable streptomycete densities were determined by dilution plating onto water agar
plates and then covering with 5 mL of cooled, molten starch-casein agar. This method
allows filamentous Actinobacteria to grow up through the overlay medium, while
suppressing the growth of many other bacteria (Wiggins and Kinkel, 2005b).
The PowerSoil DNA kit (MO BIO; Carlsbad, CA USA) was used to extract DNA from
soil, with minor modifications to enhance recovery of DNA from target taxa (Schlatter et
al., 2010). Targeting particular taxa with optimized DNA extraction procedures is likely
to minimize the introduction of bias due to variable extraction efficiency. Extracted DNA
was passed through the QIAquick PCR Purification Kit (Qiagen; Valencia, CA USA) and
quantified with a NanoDrop ND 1000 spectrophotometer (Thermo Fisher Scientific;
Waltham, MA USA). PCRs consisted of 10 ng of template DNA in a 50 uL reaction
volume using PCR Supermix High Fidelity (Invitrogen; Carlsbad, CA USA). We used
StrepB (Rintala et al., 2001) as our forward primer, and the reverse complement of
Act283 (McVeigh et al., 1996) as our reverse primer, each at a final concentration of 200
nM. Both primers are selective for Actinobacteria and together amplify a fragment of
approximately 165 nucleotides, encompassing the V2 variable region of the 16S rRNA
gene, which is referred to as the 'gamma variable region' in some studies of
74
streptomycetes (Stackebrandt et al., 1991). Primers were modified to contain one of 30
different 10mer identifying barcodes (Parameswaran et al., 2007). PCR conditions
consisted of an initial denaturation step of 30 sec at 94 C, followed by 30 cycles of 30 sec
94 C, 30 sec 57 C, 60 sec 70 C. Products of PCRs were passed through the QIAquick
PCR Purification Kit, quantified by spectrophotometry, diluted with elution buffer to
approximately 15 ng/uL, and quantified by fluorometry (Quant-iT dsDNA HS assay kit;
Invitrogen). Thirty samples, each with a unique primer barcode, were combined in
equimolar amounts to form each of two pooled amplicon samples. Emulsion PCR and
sequencing were performed using a GS FLX emPCR amplicon kit according to the
manufacturer’s protocols (454 Life Sciences; Branford, CT USA). Each pooled sample
was run on one region of a picotitre plate on the GS FLX sequencing system (Droege and
Hill, 2008) at the University of Minnesota BioMedical Genomics Center. Resulting
sequence data have been submitted to the NCBI Sequence Read Archive as accession
SRA019985.3.
The PyroNoise algorithm (Quince et al., 2009) was used to de-noise the sequence data,
with the raw flowgram signals as the input to the algorithm. The dataset was processed
through PyroNoise on a per sample basis. PyroNoise program scripts were modified to
match our forward primer and to set a minimum sequence length of 100 nucleotides.
Other settings were used at their default values (basecalling stopped at the first read with
a flowgram value between 0.5 and 0.7, parameters for expectation-maximization
algorithm: sigma=1/15, c=0.05). The PyroNoise output provides a mapping of input
sequences to de-noised output sequences. We generated a single output file per sample by
duplicating each output sequence according to the number of input sequences mapping to
that output sequence, in order to retain information on the relative abundance of sequence
variants.
Subsequent processing was done with the program mothur (Schloss et al., 2009) using
versions 1.9 and 1.10. Initial quality screening removed sequences containing ambiguous
bases or of unexpected length (shorter than 175 or longer than 215 nucleotides).
75
Sequences were aligned to the Silva reference database (Pruesse et al., 2007) using kmer
searching with a ksize of 6 to find the best template sequence and using the Needleman-
Wunsch pairwise alignment method (Needleman and Wunsch, 1970) with a reward of +1
for a match and penalties of -1 and -2 for a mismatch and gap, respectively. Aligned
sequences were screened for chimeric sequences using the Pintail method (Ashelford et
al., 2005), as well as for sequences belonging to phyla other than the Actinobacteria.
Sequences passing these quality criteria were clustered into operational taxonomic units
(OTUs) based on sequence dissimilarity, using the furthest neighbor method. All analyses
with the exception of rarefaction curves presented in Figure S1 were performed using a
clustering criterion of 3% dissimilarity.
Statistical analyses were performed in R. Diversity indices, OTU richness estimates and
pairwise community similarities were calculated in mothur and with the Vegan package
in R (Oksanen et al., 2010). The function “envfit” in the Vegan package was used to fit
plant and soil characteristics onto an ordination derived from pairwise Actinobacteiral
community similarities. Enrichment of particular OTUs by experimental treatments was
tested with indicator species analysis, using the LabDSV package in R (Roberts, 2010).
The Tukey method was used for post hoc multiple comparisons following analysis of
variance procedures. Patterns of OTU co-occurrence were observed by testing for
correlations in the relative abundances of OTUs across samples (Ravel et al., 2010). The
significance of correlation coefficients was adjusted with the FDR (false discovery rate)
method for multiple test correction. The OTUs with the fifty highest cumulative
correlation scores (defined as the sum of the absolute values of the statistically significant
correlation coefficients of each OTU to every other OTU) were used to generate a
heatmap. Clustering of the heatmap was based on either similarity of correlation profile,
or on genetic distance using the F84 distance measure and one representative sequence
variant per OTU.
Soil edaphic characteristics were measured at the University of Minnesota soil-testing
lab, using standard procedures. Plant diversity, percent cover and above- and
76
belowground biomass data were accessed through CCESR Long Term Ecological
Research network database (http://www.cedarcreek.umn.edu/research/data/).
Results
Approximately 250,000 partial Actinobacterial 16S rDNA sequence reads were collected
from 60 independent soil samples (average of 3,900 sequence reads per sample; range
1,200 to 5,500; Table S1). The selectivity of our PCR amplification successfully limited
most sequence reads to DNA belonging to members of the class Actinobacteria. Eighteen
different genera were detected within this class, but 84% of all sequence reads belonged
to a single genus, the Streptomyces (data not shown). Because it is difficult to accurately
assign short 16S rDNA sequence reads to finer taxonomic divisions than the genus, we
based our analysis on operational taxonomic units (OTUs) defined on the basis of
sequence similarity (cutoff of 3% dissimilarity).
Across all samples, sequences clustered into 1329 OTUs and rarefaction analysis
suggested that further sampling would have continued to reveal additional diversity
(Figure S1). On a per sample basis, however, our level of sequencing depth was sufficient
to detect most of the diversity present; although the per-sample rarefaction curve had not
yet plateaued, the rate of detection of additional taxa had declined substantially (Figure
S2). Furthermore, OTU richness and diversity were not significantly correlated with the
number of sequence reads per sample (data not shown), indicating good depth of
coverage. Average observed richness of OTUs was 143 per sample (range 88 to 193),
with an average Chao richness estimate of 186 OTUs per sample (range 103 to 259). The
Shannon diversity index ranged from 2.1.6 to 4.20, with an average value of 3.71.
Culturable streptomycete density averaged 1.6 x 106 colony forming units / g (range 4.9 x
105 to 2.9 x 106; Table S1) and was significantly correlated with OTU richness (r2 = 0.26,
p < 0.001) and diversity (r2 = 0.13, p = 0.03).
Similarity in Actinobacterial community structure among samples was summarized with
an ordination based on pairwise dissimilarity values (inverse of the Bray-Curtis index).
77
Independent of experimental treatment (host species or plant community richness),
samples fell into two distinct clusters (Figure 1; dashed circles). Community composition
differed dramatically among samples belonging to these clusters (Figure S3). For
simplicity, we will refer to the larger cluster (n = 50) as the ‘dominant community state,’
and to the smaller cluster (n = 10) as the ‘minority community state.’ Samples
demonstrating these two community states did not differ significantly (p > 0.05) in terms
of the number of sequence reads sampled, Actinobacterial density or diversity, or any
measured soil edaphic properties (data not shown). Neither was there a clear relationship
between community state and host plant or plant richness treatment (Figure S3).
However, samples having the minority community state had significantly higher
Actinobacterial richness compared to samples having the dominant community state
(average of 153 vs. 141 observed OTUs; t = -2.45; p = 0.02).
With our sampling scheme, different host species were sometimes sampled in the same
experimental plot. On average, samples from the same plot had more similar community
structure than samples from different plots (Bray-Curtis = 0.43 vs. 0.31 respectively; t = -
2.86, p = 0.008). However, in several cases samples from within the same experimental
plot demonstrated two community types and were highly dissimilar from each other
(Figure 1; points connected by lines). Considering only those samples having the
dominant community state, samples from the same plot were also significantly more
similar than samples from different plots (Bray-Curtis = 0.55 vs. 0.38 respectively; t = -
5.51, p < 0.001)
Co-occurrence of Actinobacterial OTUs among samples
Analysis of correlation among OTU relative abundances revealed non-random patterns of
co-occurrence for some OTUs. Correlation analyses considered only OTUs present in
samples having the dominant community state, due to low overlap of OTUs among
community states. As a whole, genetic distance among OTUs was not consistently related
to degree of correlation in relative abundance (Mantel test with 1000 permutations; r =
0.011, p = 0.19). However, in specific cases, relationships could be observed between
78
genetic distance and relative abundance across samples; for example, OTUs 371, 389 and
734 were most similar to each other and were positively correlated in abundance (Figure
2A, highlighted with intersecting black lines), while the relative abundance of OTU 673
was negatively correlated with that of its three most similar OTUs (Figure 2A; white
lines). Patterns in the co-occurrence of OTUs were more evident when genetic similarity
was ignored; for example, OTUs 205, 578, 588, and 1139 tended to be abundant in the
same samples (Figure 2B, white lines). On the other hand, OTUs 151, 332 and 993 all
tended to be rare when other particular OTUs were abundant (Figure 2B, black lines).
Variation in rhizosphere Actinobacterial communities associated with plant species
identity:
Actinobacterial density did not vary with host species, although Actinobacterial richness
did (Table 1); Andropogon gerardii supported significantly lower Actinobacterial
richness than S. scoparium (Figure 3). Diversity can be measured as a property of
individual samples (alpha diversity), or as a measurement of the variation across samples
(beta diversity). Average alpha diversity on a per sample basis did not differ significantly
among host species (Table 1). However, beta diversity was highest for L. perennis (Table
2) among the host plant species tested. Indeed, L. perennis-associated communities were
as different from each other as they were from communities associated with other host
species (Table 2).
Indicator species analysis revealed that particular OTUs were significantly enriched and
preferentially associated with each of the four host species (Table S2). Six OTUs were
significant indicators of A. gerardii samples, while six, five, and two OTUs were
significant indicators of Actinobacterial communities from L. capitata, L. perennis, and
S. scoparium, respectively.
Variation in rhizosphere Actinobacterial communities associated with plant community
richness:
Actinobacterial richness and diversity varied by plant community richness treatment,
79
while Actinobacterial density did not (Table 1). Monoculture Actinobacterial richness
was significantly lower than for richer plant community treatments, and monoculture
Actinobacterial diversity was significantly lower than in 8-species plots (Figure 3).
Significant indicator OTUs were found for each plant richness treatment (Table S2).
Although ANOVA did not reveal a significant interaction between host plant species and
plant community diversity treatment (Table 1), effects of plant community richness on
Actinobacterial richness varied among individual host plant species. Both A. gerardii and
L. capitata supported significantly lower Actinobacterial richness when grown in
monoculture compared to more diverse plant communities (Figure 4). In contrast, there
was no effect of plant community richness treatment on Actinobacterial richness for L.
perennis or S. scoparium (Figure 4). Thus, plant community richness treatment changed
host plant species impact on associated soil microbial communities for some plant species
but not for others.
Plant community and soil edaphic characteristics as correlates of Actinobacterial
community characteristics:
Several measures of plant diversity and productivity were significantly correlated with
Actinobacterial culturable density and OTU richness (Table 3): more diverse plant
communities and more productive plant communities had denser and richer
Actinobacterial communities. Notably, none of the available plant community
characteristics were significantly correlated with Actinobacterial diversity. Furthermore,
percent plant cover showed patterns that were coherent with patterns of similarity in
Actinobacterial community structure; plant cover data could be fit onto an ordination
based on Actinobacterial community structure (Figure 1).
Measures of soil carbon (C), nitrogen (N), potassium (K), organic matter (OM), and pH
could also be correlated with Actinobacterial density and richness (Table 3). Soil pH was
the only variable that was correlated with Actinobacterial diversity. Locations with higher
pH had greater Actinobacterial density, richness and diversity. Patterns of soil pH and
80
soil K concentrations could also be fit to patterns of Actinobacterial community similarity
(Figure 1).
There were also significant correlations among many soil and plant community
characteristics (data not shown). Furthermore, experimental treatments sometimes
differed in soil characteristics. Samples from L. perennis had higher pH than A. gerardii
samples, and higher K than S. scoparium samples (Table 4). Samples from plant
communities with a richness of 16 species had higher soil C, N, OM, and K than samples
from monocultures (Table 4). Thus plant host and plant community richness treatment
altered the soil physical and chemical environment.
Discussion
This work investigated the structure of soil Actinobacterial communities in soils
associated with different plant host species and across a gradient of plant community
richness. Our results reinforce previous findings that host plant species have differential
effects on associated soil microbial communities (Miethling et al., 2000; Mazzola et al.,
2004; Funnell-Harris et al., 2010). Beyond this, however, we found that plant community
context can change host plant impact on associated soil microbial communities in some
cases. This finding may offer a partial explanation for observed variability in soil
microbial community responses to host plants, and suggests that methodologies for
investigations of feedback between plants and soil microorganisms should be re-
considered. In particular, using monocultures to either condition soil or to test effects of
soil conditioning is likely to offer an incomplete perspective since surrounding plant
community richness can alter the impacts of host species on soil microbial communities.
Plant host species and plant richness treatments significantly altered Actinobacterial
richness more frequently than Actinobacterial diversity. Furthermore, Actinobacterial
richness was related to many soil and plant community characteristics, while
Actinobacterial diversity showed a relationship only with soil pH. This suggests that
management practices or changing plant cover are more likely to have impacts on soil
81
microbial richness than diversity. This is in agreement with other studies that have shown
soil microbial diversity to be remarkably resilient to change (Hirsch et al., 2009).
The increase in Actinobacterial richness and diversity with plant species richness
treatment may be the result of several different mechanisms. Plant identity and diversity
are related to overall plant productivity and to soil edaphic characteristics. This suggests a
mechanism whereby plants exert selection on soil microbial populations by modifying
resource availability and the chemical environment in soil. In our results, Actinobacterial
density and richness increased with measures of plant productivity, including percent
cover and above ground biomass. This is consistent with a mechanism of plants
impacting soil microbial communities through the simple quantity of resource inputs
provided. Plant richness and productivity are confounded in the plots from which we
sampled (Zak et al., 2003); more species-rich communities have consistently greater
above- and below-ground productivities. On the other hand, more diverse chemical inputs
accompanying a rise in plant species richness may provide a greater array of nutrients,
and thereby a greater number of niches for soil microbes. A greater number of available
niches may allow for the coexistence of more microbial taxa.
Beyond direct effects of the quantity or diversity of the resource base provided by plants
to soil microbial food webs, indirect impacts on associated microbes are also probable.
Plant species and plant richness treatments differentially modified soil chemical
properties such as pH and N, C, and K concentration. Such changes to the soil chemical
environment may have exerted physiological stresses distinct from nutritive effects.
Individual plant hosts may impact soil properties in ways that are distinct from plant
community level impacts. In our system, plant host species altered soil pH while plant
community richness treatment did not. Soil pH is an important correlate of soil bacterial
diversity and richness at continental geographic scales (Fierer and Jackson, 2006) and is a
significant environmental constraint for some Actinobacteria specifically (Williams et al.,
1971). Our data are consistent with these findings, even over the narrow range of pH
82
values (5.5 to 6.5) represented in our soil samples.
There remain multiple hypotheses that may explain how manipulating plant species
richness could lead to changes in Actinobacterial richness and diversity. In particular,
additional work is needed to clarify the relative importance of resource quantity versus
diversity in the context of plant impacts on soil microbial communities. Similarly, future
studies should specifically explore the relative importance of selective effects exerted
directly by plants (as through the provision of specific chemical resources that differ in
accessibility to various microbial taxa) versus ways in which variable resource
availability might alter the outcomes of microbial interactions (as through general
resource provision altering the overall density of microbial populations and changing
microbial dynamics through density-dependent mechanisms).
Among the observed host species effects, the high beta diversity of Actinobacterial
communities associated with L. perennis is of particular interest. The selective forces
exerted by L. perennis may be quite different than those of the other plant species tested.
Given that L. perennis is ruderal and adapted to highly disturbed habitats (Pavlovic and
Grundel, 2009), it is possible that its life history strategy is relatively independent of
interactions with Actinobacteria; a weak selective effect by L. perennis could allow
underlying spatial differences in Actinobacterial community structure to persist. In
contrast, stronger selection by the other plant species could contribute to homogenization
of Actinobacterial communities associated with different individuals of the same plant
species. Apart from comparisons with L. perennis, similarity in Actinobacterial
community structure was not higher among samples from the same host plant species
compared to samples from different host plant species. Thus, selective pressures
experienced by soil microbes in association with A. gerardii, L. capitata and S.
scoparium may be broadly similar. However, the presence of unique aspects to the
selective environment of each host species is suggested by the significant enrichment of a
small number of particular Actinobacterial taxa (OTUs) in samples from each host plant
species.
83
Despite relationships with plant and soil characteristics, the most striking differences
between Actinobacterial communities were not clearly connected to these variables. We
repeatedly recovered examples of two distinct community types, but these community
types were not consistently associated with plant host, plant community richness
treatment, or soil properties. In aquatic systems, priority effects, or the order in which
species colonize a habitat, have been suggested to lead to the formation of distinct
community states (Chase, 2010). However, there was no coherent spatial pattern to the
distinct community states in our study, with distinct states observed in the same
experimental plot. This lack of spatial pattern argues against a definitive role for
immigration or dispersal events in defining these distinct community types.
Although efforts were made to homogenize soil samples, it may be that distinct
microenvironments were represented in different samples. Alternatively, the development
of distinct community states may be driven by species interactions within microbial
communities, potentially including top-down controls (eg. phage activity, predation) or
competitive or antagonistic interactions. Culturable streptomycetes from samples having
the minority community state showed a trend toward more pronounced antagonistic
phenotypes relative to streptomycetes from samples having the dominant community
state (data not shown). The fact that Actinobacterial richness differed among community
states suggests that the selective effects of microbial interactions may also differ.
Finally, this work sheds light on the structure and variability among soil Actinobacterial
communities. Our results suggest that a given soil sample is likely to contain several
hundred Actinobacterial taxa, while the number of taxa present over the field scale is in
the thousands. Some biogeographic patterns were observed, with samples from the same
experimental plot (located within a few meters of each other) showing higher
Actinobacterial community similarity than samples from different plots (located tens of
meters apart). The higher similarity among samples collected in close proximity but from
different host species suggests an important role for immigration and dispersal in
84
structuring Actinobacterial communities. We were also able to observe patterns of co-
occurrence among taxa, possibly revealing shared habitat preferences or the effects of
competitive interactions. Interestingly, patterns of both positive and negative correlation
were observed among Actinobacterial OTUs. This suggests that multiple forces are at
play to influence the dynamics of microbial taxa individually and as a result of
interactions. For example, positive correlations in taxon abundance across samples could
suggest either shared habitat preferences or a mutualistic symbioses or syntrophy
(Freilich et al., 2010). Similarly, negative correlations in taxon abundance across samples
could suggest either disparate habitat preferences or a competitive or antagonistic
dynamic (Fuhrman and Steele, 2008; Fuhrman, 2009). Although theoretical predictions
suggest that more closely related taxa should compete more strongly with each other
(Tonsor, 1989), we observed instances of genetically similar taxa sharing similar patterns
of abundance. This suggests that, at least in some cases, closely related taxa may coexist
without competitive exclusion.
85
Figure 1 – Classical multidimensional scaling of pairwise dissimilarities in
Actinobacterial community structure (inverse of Bray-Curtis index). In this
representation, the distances between points are approximately equal to dissimilarities.
Lines connecting points indicate different samples taken from within the same
experimental plot. Unconnected points are from plots that were sampled only once.
Vector length and direction indicate the strength and orientation of fitted relationships
between plant community or soil edaphic characteristics and the arrangement of samples
in ordination space.
Soil pH
Soil K Percent Plant Cover
Figure 1
86
Figure 2 – Heatmaps showing significant correlations among OTUs (Pearson's
correlation, with FDR multiple test correction, p < 0.05). A) OTUs are clustered
according to genetic distance. B) OTUs are clustered according to similarity in
correlation profile. Particular comparisons are highlighted by intersecting white or black
lines (see text).
A)
87
Figure 2, continued B)
88
Figure 3 – Box and whisker plots of Actinobacterial density, richness, and diversity, by
host species (left side; Ag = A. gerardii; Lc = L. capitata; Lp = L. perennis; Ss = S.
scoparium) or plant community richness treatment (right side; Div01 = monoculture;
Div04 = assemblage of 4 plant species; Div08 = 8 spp.; Div16 = 16 spp; Div 32 = 32
spp). Different letters indicate means which differ significantly; ns = no significant
differences.
89
Figure 4 – OTU richness of Actinobacterial communities associated with four different
host plants, grown in each of five different plant richness treatments (indicated along the
x axis). Symbols show individual data points, while horizontal bars display mean values.
For each host plant, different letters indicate that means differ significantly among
treatments; ns = no significant differences.
Figure 4
50
100
150
200
250
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
OTU
ric
hn
ess
(C
hao
est
imate
)
1 s
pp
4 s
pp
8 s
pp
16 s
pp
32 s
pp
A B B B AB A AB B A A ns ns 1 s
pp
4 s
pp
8 s
pp
16 s
pp
32 s
pp
1 s
pp
4 s
pp
8 s
pp
16 s
pp
32 s
pp
1 s
pp
4 s
pp
8 s
pp
16 s
pp
32 s
pp
A. gerardii L. capitata L. perennis S. scoparium
90
Figure S1 – Rarefaction curve for Actinobacterial OTUs, considering the entire dataset
as a whole.
0
500
1000
1500
2000
2500
3000
OTU
s de
tect
ed
Sequence reads (thousands)
0.01
0.03
0.05
0.10
OTU clustering criterion
(sequence dissimilarity)
Figure S1
91
Figure S2 – Rarefaction curve for Actinobacterial OTUs on a per-sample basis. Shown is
the mean across samples, flanked by the standard error, of the estimated number of OTUs
observed at each sampling intensity.
0
20
40
60
80
100
120
140
160
1 500 1000 1500 2000 2500 3000 3500 4000 4500
OTU
s de
tect
ed
Sequence reads
Figure S2
92
Figure S3 – A summary of sample clustering and Actinobacterial community
composition. The dendrogram shows clustering based on Bray-Curtis dissimilarity. The
left column of colored boxes indicates which host plant was sampled. The right column
of colored boxes indicates the plant community richness treatment from which the sample
was taken. Vertical bars to the right of the colored boxes indicate the presence and
relative abundance (increase from grey through black to red) of each OTU in each
sample.
93
Table 1 - ANOVA results tables showing dependence of Actinobacterial density (A),
richness (B) and diversity (C) upon host plant species identity and plant richness.
A) Culturable density (log transformed):
Degrees of
Freedom
Sum of Squares
Mean of Squares
F-value p-value
Host plant species 3 0.087 0.029 0.73 0.540 Plant richness 4 0.257 0.064 1.62 0.190 Interaction 12 0.151 0.013 0.32 0.982 Residuals 38 1.510 0.040
B) OTU richness (Chao estimate):
Degrees of
Freedom
Sum of Squares
Mean of Squares
F-value p-value
Host plant species 3 5438.9 1813.0 2.92 0.045 ** Plant richness 4 15852.7 3963.2 6.39 < 0.001 *** Interaction 12 8282.1 690.2 1.11 0.377
Residuals 40 24806.7 620.2
C) OTU diversity (Shannon index):
Degrees of
Freedom
Sum of Squares
Mean of Squares
F-value p-value
Host plant species 3 0.470 0.157 1.78 0.167 Plant richness 4 0.775 0.194 2.19 0.087 *
Interaction 12 0.574 0.048 0.54 0.874 Residuals 40 3.531 0.088
94
Table 2 - Mean pairwise Actinobacterial community dissimilarity (inverse of Bray-Curtis
index), by host plant sampled. Significant differences are indicated down columns, by
different letters; ns = no significant differences. Values along the diagonal (bold) are a
measure of beta diversity for each host plant.
Host 1 A. gerardii L. capitata L. perennis S. scoparium Host 2 (p<0.001) (p=0.003) (ns) (p<0.001)
A. gerardii 0.641 a 0.670 a 0.714 0.663 a L. capitata 0.670 a 0.677 ab 0.717 0.673 a L. perennis 0.714 b 0.717 b 0.744 0.726 b S. scoparium 0.663 a 0.673 ab 0.726 0.682 ab
95
Table 3 - Correlation coefficients (p-values) for relationships between Actinobacterial
community characteristics and various plant community and soil edaphic characteristics.
Culturable density
Observed richness
Estimated richness (Chao)
Diversity (Shannon
index) Belowground plant biomass
ns
ns 0.34 (0.06)
ns
Plant diversity by cover 0.32 (0.05) 0.36 (0.02) 0.42 (0.01)
ns Total plant cover
ns 0.35 (0.03) 0.42 (0.01)
ns
Plant diversity by clip strip 0.44 (0.01) 0.35 (0.06) 0.42 (0.02)
ns Aboveground plant biomass 0.32 (0.09) 0.46 (0.01) 0.47 (0.01)
ns
Soil pH 0.36 (0.03) 0.46 (<0.01)
ns 0.34 (0.03) Soil K 0.30 (0.07) 0.30 (0.07) 0.36 (0.02)
ns
Soil organic matter 0.34 (0.04) 0.29 (0.09) 0.30 (0.08)
ns Total soil N 0.32 (0.06) 0.29 (0.09)
ns
ns
Total soil C
ns
ns 0.28 (0.10)
ns
96
Table 4 - Variation in soil edaphic characteristics across two levels of plant
community manipulation. Different letters indicate significant differences between mean
values within a given comparison (p < 0.05, ANOVA with Tukey multiple test
correction); ns = no significant differences.
pH Potassium
(ppm) Organic
matter (%) Nitrogen
(%) Carbon
(%) A. gerardii 5.95 a 42.1 ab 1.11 ns 0.037 ns 0.562 ns L. capitata 6.03 ab 44.9 ab 1.05
0.035
0.511
L. perennis 6.17 b 52.4 b 1.09
0.042
0.545 S. scoparium 6.10 ab 38.1 a 1.03
0.034
0.486
Monocultures 6.01 ns 34.1 a 0.88 a 0.028 a 0.421 a 4 plant spp 6.03
42.0 ac 1.06 ab 0.034 ab 0.515 ab
8 plant spp 6.16
41.6 ac 1.06 ab 0.035 ab 0.510 ab 16 plant spp 6.05
57.4 b 1.31 b 0.052 b 0.678 b
32 plant spp 6.08
46.8 bc 1.05 ab 0.035 ab 0.506 ab
97
Table S1 - Sequence yield, observed and estimated OTU richness, OTU diversity and
culturable Actinobacterial density for 60 soil samples. Sample labels indicate the host
plant targetted (digits 1-2; Ag = Andropogon gerardii, Lc = Lespedeza capitata, Lp =
Lupinus perennis, Ss = Schizachyrium scoparium), the plot number (digits 3-5), the
within plot replicate (digit 6), and the plant species richness in that plot (digits 7-8).
Sample Sequence
yield Observed richness
Estimated richness (Chao)
Diversity (Shannon
index) Culturable
density Ag005a01 4014 133 154 3.70 4.88E+05 Ag045a04 3884 132 177 3.66 1.11E+06 Ag070a04 3498 132 178 3.55 1.16E+06 Ag078a32 4224 153 185 3.84 1.68E+06 Ag105a32 4064 151 208 3.91 1.29E+06 Ag109a01 3353 88 103 2.87 1.47E+06 Ag109b01 4119 103 119 3.07 . Ag146a08 2871 135 200 3.48 1.18E+06 Ag220a16 3875 129 200 3.68 6.42E+05 Ag229a04 4908 132 147 3.70 8.67E+05 Ag273a16 3917 135 197 3.61 1.22E+06 Ag283a08 3080 112 139 3.75 1.38E+06 Ag292a08 3099 157 192 3.98 2.74E+06 Ag295a32 4898 128 169 3.48 2.13E+06 Ag299a16 3578 146 198 3.80 1.98E+06 Lc002a01 3700 126 138 3.73 7.47E+05 Lc029a01 3857 153 185 3.82 1.18E+06 Lc057a08 3868 158 208 3.87 2.76E+06 Lc064a32 3221 141 195 3.85 1.42E+06 Lc078a32 4059 146 177 3.95 2.11E+06 Lc082a16 2468 126 220 3.45 1.51E+06 Lc094a01 3582 146 179 3.76 2.38E+06 Lc110a04 4752 132 158 3.52 9.46E+05 Lc146a08 3549 126 178 3.50 6.76E+05 Lc201a04 4301 141 168 3.69 2.91E+06 Lc206a08 4635 157 185 3.90 1.81E+06 Lc249a32 4642 157 206 3.91 1.40E+06 Lc253a16 4020 158 223 3.91 2.22E+06 Lc273a16 4756 169 247 3.63 2.12E+06 Lc286a04 3503 141 171 3.80 1.80E+06
98
Table S1, continued
Sample Sequence
yield Observed richness
Estimated richness (Chao)
Diversity (Shannon
index) Culturable
density Lp012a08 3045 154 210 3.81 1.72E+06 Lp070a04 3576 131 147 3.67 1.02E+06 Lp082a16 2978 142 195 3.54 9.53E+05 Lp083a01 3911 121 144 2.16 5.58E+05 Lp093a04 3824 169 224 3.91 2.70E+06 Lp110a04 4467 142 171 3.67 1.68E+06 Lp202a16 4670 178 221 3.65 2.80E+06 Lp206a08 4545 141 197 3.68 2.14E+06 Lp220a16 4902 136 171 3.66 8.48E+05 Lp249a32 4575 169 204 3.90 1.95E+06 Lp262a32 3634 146 181 3.88 1.44E+06 Lp265a01 1217 150 209 4.14 1.71E+06 Lp265b01 5514 150 191 3.60 . Lp292a08 4934 162 195 3.95 1.74E+06 Lp295a32 5063 127 162 3.46 1.35E+06 Ss012a08 4273 193 259 4.20 1.89E+06 Ss031a01 3015 103 138 3.63 1.02E+06 Ss057a08 2668 164 211 4.04 1.54E+06 Ss064a32 3781 145 182 3.73 1.82E+06 Ss070a04 3679 141 174 3.84 1.17E+06 Ss105a32 3247 146 183 3.94 1.90E+06 Ss135a01 3641 130 152 3.76 8.80E+05 Ss202a16 4094 143 174 3.58 1.80E+06 Ss229a04 3884 122 196 3.56 7.53E+05 Ss253a16 3287 164 201 4.06 2.53E+06 Ss262a32 4841 174 235 3.90 9.46E+05 Ss280a01 3775 161 229 3.88 1.36E+06 Ss283a08 4785 142 219 3.76 1.81E+06 Ss286a04 3455 148 181 3.86 2.09E+06 Ss299a16 3875 134 224 3.71 8.45E+05 Average 3891 143 186 3.71 1.56E+06 Minimum 1217 88 103 2.16 4.88E+05 Maximum 5514 193 259 4.20 2.91E+06
99
Table S2 - Actinobacterial OTUs that are indicative of particular plant hosts or plant
community richness treatments. Indicator value has a maximum of 1. Fold enrichment
describes the proportional abundance of a given OTU for the indicated group compared
to all other samples. Incidence is the proportion of samples in which a given OTU was
observed.
Grouping OTU Indicator value
p-value
Fold enrichment
Incidence: indicated group
Incidence: other
A. gerardii 205 0.43 0.002 2.3 1.00 0.89 297 0.46 0.001 7.4 0.67 0.27 492 0.37 0.03 2.2 0.87 0.64 887 0.29 0.02 4.3 0.47 0.11 900 0.26 0.04 3.1 0.53 0.18 1008 0.40 0.005 2.1 1.00 0.71 L. capitata 301 0.27 0.04 4.1 0.47 0.22 389 0.41 0.003 4.1 0.67 0.29 421 0.32 0.008 9.7 0.40 0.11 645 0.47 0.004 3.1 0.93 0.56 849 0.34 0.04 6.2 0.53 0.36 1125 0.29 0.05 4.5 0.47 0.22 L. perennis 30 0.33 0.004 17.7 0.40 0.04 175 0.44 0.02 41.2 0.47 0.16 310 0.30 0.05 10.3 0.40 0.18 358 0.18 0.04 5.6 0.27 0.07 892 0.28 0.03 4.0 0.47 0.20 S. scoparium 296 0.26 0.04 6.5 0.40 0.16 1037 0.44 0.02 16.6 0.53 0.22
100
Table S2, continued
Grouping OTU Indicator value p-value Fold
enrichment
Incidence: indicated group
Incidence: other
Monocultures 253 0.39 0.02 56.5 0.42 0.23 280 0.44 0.003 5.3 0.83 0.60 310 0.49 0.001 21.5 0.58 0.15 889 0.32 0.05 8.0 0.50 0.23 958 0.35 0.02 2.3 1.00 0.88 1262 0.41 0.02 21.5 0.50 0.17 4 plant spp 747 0.28 0.03 3.3 0.67 0.27 929 0.30 0.01 32.9 0.33 0.06 8 plant spp 70 0.32 0.03 2.5 0.83 0.63 739 0.34 0.01 7.0 0.58 0.17 1037 0.40 0.04 20.0 0.50 0.25 1097 0.26 0.05 11.4 0.33 0.08 1108 0.36 0.009 3.4 0.83 0.38 16 plant spp 51 0.28 0.03 16.7 0.33 0.10 108 0.33 0.02 8.7 0.50 0.23 266 0.37 0.05 3.0 0.83 0.77 298 0.41 0.03 15.8 0.50 0.25 330 0.34 0.02 4.2 0.67 0.38 344 0.34 0.01 8.3 0.50 0.19 355 0.26 0.03 19.3 0.33 0.06 518 0.33 0.02 5.8 0.50 0.17 717 0.50 0.0004 482.0 0.50 0.02 799 0.43 0.01 7.8 0.67 0.29 892 0.25 0.05 4.5 0.50 0.21 993 0.35 0.03 4.7 0.67 0.48 1206 0.43 0.03 30.4 0.50 0.19 32 plant spp 50 0.52 0.001 13.9 0.67 0.17 77 0.31 0.02 3.9 0.58 0.27 381 0.33 0.007 5.5 0.58 0.15 422 0.25 0.03 13.4 0.33 0.08 863 0.31 0.03 10.4 0.42 0.13 955 0.33 0.04 3.1 0.75 0.63 1044 0.41 0.009 213.0 0.42 0.08
101
Chapter 5: Do antagonistic streptomycetes play a role in plant-soil feedbacks?
102
Introduction
Non-pathogenic soil microbes can limit plant disease (Ezziyyani et al., 2007; Hiltunen et
al., 2009) and positively impact plant performance through a variety of mechanisms,
including antibiotic-mediated antagonism of pathogens (Haas and Keel, 2003; Anukool et
al., 2004), enhanced nutrient availability (Yehuda et al., 2000; Verbruggen and Kiers,
2010), production of plant hormones (Ortiz-Castro et al., 2011), or activation of innate
plant defense responses (Compant et al., 2005; Lehr et al., 2008). At the same time, plant
genotype-specific impacts on associated soil microbial communities have been well
documented (Bardgett and Walker, 2004; Viketoft et al., 2005; Carney and Matson,
2006). Taken together, these two concepts suggest the potential for positive plant-soil
feedbacks, with plant hosts driving changes in associated soil microbial communities that
in turn enhance plant performance.
Positive plant-soil feedbacks occur when the changes to the soil environment imposed by
a particular plant species enhance the performance of conspecific individuals. In contrast,
negative plant-soil feedbacks occur when changes in the soil environment imposed by a
particular plant species reduce the performance of that species (Kulmatiski et al., 2008).
Changes to the soil environment brought about by the presence of one plant species may
also have implications for the subsequent growth of other species (Callaway et al., 2008).
We refer to this as the impacts of a conditioning species on the growth performance of a
response species.
Plant-soil feedbacks have been studied extensively in ecology because of their potential
impacts on plant community dynamics (Klironomos, 2002; Eppinga et al., 2006;
Petermann et al., 2008). Negative feedbacks may prevent dominant species from
excluding other species (Bever et al., 1997, 2010) by reducing plant performance when
the same soil is occupied over time. On the other hand, positive feedbacks may play a
role in facilitating invasion by exotic species (Inderjit and Van der Putten, 2010) as
invaders condition soil in ways that make it more suitable for their own growth.
However, plant-soil feedbacks also have implications for agricultural systems, where the
103
same plants are often grown repeatedly or in a short rotation for extended periods of time.
Better understanding of the forces that attenuate negative or promote positive feedbacks
may suggest methods of managing agroecosystems to limit plant disease and improve
plant health.
Many plant-soil feedbacks appear to be mediated through soil microbial communities
(Olff et al., 2000; McCarthy-Neumann and Kobe, 2010b), although mechanisms
involving soil chemistry or nutrient levels may play a larger role in some cases
(Ehrenfeld et al., 2005; Casper et al., 2008; McCarthy-Neumann and Kobe, 2010a). There
is an unfortunate tendency in studies of plant-soil feedbacks to treat soil microbial
communities as a black box. For example, soil sterilization is typically used to
demonstrate the importance of microbial players in the strength or direction of observed
feedbacks (Klironomos, 2002; McCarthy-Neumann and Kobe, 2010b), but this approach
offers no insight into the identity of organisms or the mechanisms that may underlie
microbial impacts on plant performance. At the same time, studies that characterize plant
host impacts on soil microbial communities in detail have rarely provided complementary
data on feedback effects on plant performance, limiting the applicability of these studies
to agriculture or plant ecology.
To better understand the development and implications of plant-soil feedbacks for plant
productivity, the involvement of specific microbial taxa and particular microbial
functions that may feedback to impact plant fitness need to be considered. Toward this
end, we address the antagonistic activity of streptomycetes in soils conditioned by four
different prairie plant species. Streptomycetes have been widely studied for their
contribution to limiting plant disease across a wide range of pathosystems (Liu et al.,
1995; Jones and Samac, 1996; Samac and Kinkel, 2001; Xiao et al., 2002). Furthermore,
measures of in vitro antagonistic activity have been shown to relate well to plant disease
levels in field settings (Wiggins and Kinkel, 2005a, 2005b)
104
Plant growth occurs in a community context, and it is possible that a changing context for
growth could alter plant-driven impacts on soil microbes that feedback to influence plant
performance. As aerial plant morphology may be altered by the presence of neighboring
plants (Bartelt-Ryser et al., 2005), so it is also possible that root architecture and root
exudation may vary depending on the specific context in which a plant is grown. This
suggests the possibility that plant-specific impacts on associated soil microbial
communities and resulting plant-soil feedbacks may exhibit unique dynamics as a
function of community context.
Plant community diversity may be an important factor in the development of plant-soil
feedbacks in which pathogen antagonists limit plant disease. Because a persistent
association between two partners is a necessary condition in order for a stable mutualism
to arise (Bronstein 2009), increased plant community diversity may work against the
successful establishment of protective mutualisms between plants and pathogen
antagonists. Moreover, the negative impacts of plant disease are likely to be stronger in
low diversity plant communities (Smithson and Lenne 1996, Keesing et al. 2006, Garrett
and Mundt 1999, Mille et al. 2006). Thus, pathogen-antagonists have the highest
potential for positive impacts on plant fitness in low diversity plant communities.
Furthermore, the development and maintenance of a pathogen-inhibitory microbial
community may be energetically costly for plants, while benefits may accrue to adjacent
competing plants. In this case, a disincentive for investing in pathogen-suppressive
microbes may be experienced in higher diversity plant communities, since plants not
bearing the cost of the protective symbiosis may still reap the benefits.
Here we test for connections between plant community impacts on associated microbes
and subsequent feedbacks on plant growth performance. Conditioning soil with plant host
identity and plant richness as distinct factors, we couple measurements of subsequent
greenhouse growth performance with assays for the antagonistic potential of
streptomycete communities.
105
Methods
Sampling was performed at the Cedar Creek Ecosystem Science Reserve (CCESR; part
of the National Science Foundation Long-Term Ecological Research network), from
experimental plots that have been maintained in a long-term plant richness manipulation
(Tilman et al., 2001). These experimental plots were established in 1994 with defined
levels of plant richness. For plant richness treatments of up to 16 species, plant species
were drawn from a pool of 16 core native prairie plant species. For 32 species-plots,
additional plant species were included beyond the pool of 16 species used for the lower
richness treatments (Tilman et al., 1997). While other colonizing species are removed
from the plots, the experimental manipulation does not control plant diversity per se, as
the relative abundance of the planted species is allowed to fluctuate as a result of natural
processes and species are not replaced if they undergo local extinction (Tilman et al.,
2001).
We targeted soil under the dominant influence of each of four different plant species,
including two C4 grasses: Andropogon gerardii, Schizachyrium scoparium and two
legumes: Lespedeza capitata, Lupinus perennis, each growing in five different plant
richness treatments (monoculture and assemblages of 4, 8, 16 or 32 species). Soil cores
were collected from as close as possible to the base of individual plants. Each sample
consisted of four bulked soil cores, collected to a depth of 30 cm using a 5 cm diameter
soil corer and homogenized by hand. There were three plot-level replicates per plant host-
community richness combination, except for monocultures of A. gerardii and L. perennis,
for which only two plot-level replicates were available. Soil samples were transported to
the laboratory on ice and were at 4 C until processing.
Streptomycete antagonistic potential
Approximately 5 g sub-samples of soil were dried overnight under sterile cheesecloth
before being thoroughly dispersed in a 10-fold volume of water on a reciprocal shaker
(175 rpm, 60 min, 4 C). Soil dilutions were spread onto water agar plates and then
covered with 5 mL of cooled, molten starch-casein agar (SCA). This method suppresses
106
the growth of many unicellular bacteria, while allowing filamentous streptomycetes to
grow up through the SCA (Wiggins and Kinkel, 2005b). After five days of incubation at
28 C, streptomycete colonies were enumerated based on morphology. Streptomycete
inhibitory activities were assessed using a modified Herr's assay (Wiggins and Kinkel,
2005a, 2005b). Briefly, soil dilution plates that had been incubated for five days were
covered with a thin layer of medium and then spreading an overlay isolate as a dense
spore suspension (approximately 1.5 x 107 colony-forming units [CFU]/plate). Three
different overlay isolates were used: one plant pathogen (Streptomyces scabies 87) and
two nonpathogenic Streptomyces isolates (Streptomyces sp. 4-21 and Streptomyces sp.
1324-2). Multiple overlay isolates were used in order to limit possible bias due to
inherent differences in antibiotic resistance and susceptibility among overlay isolates.
Antagonistic activity was quantified as the density (CFU/g) and proportion of
streptomycete colonies that produced a clear zone of inhibition against the overlay
isolate. The radius of each inhibition zone was used as a measure of the intensity of
inhibitory activity. Values of inhibitory intensity and antagonist density and frequency
were averaged across three soil dilution plates for each overlay strain from each soil
sample.
Feedbacks of soil conditioning on plant performance
In a complementary test for plant-soil feedbacks, the same soil samples were used for a
greenhouse test of plant growth performance. Specifically, each of the four target plant
species was grown in soil conditioned by all combinations of host species and plant
community richness, and response variables related to plant performance were assessed.
Seeds of each plant species were surface-disinfested by vortexing for 90 s in 15% H2O2
and four seeds were planted into each conetainer (approximately 165 mL volume).
Osmocote slow release fertilizer (15-9-12) and a top layer of sand were added to each
conetainer at planting. Seedlings were thinned to one per container as they emerged. Five
replicate conetainers were planted per treatment (conditioning soil treatment X response
plant species), but due to low germination some treatments ended up with fewer than five
107
replicates. Conetainers were watered lightly twice a day until seedling emergence, at
which time watering was performed on an as-needed basis.
At the end of a 12-week growth period, all plants were harvested for evaluation of growth
performance. Above- and belowground tissues were harvested separately for drying and
biomass determination. Soil was washed from roots with water. The length of the root
system was measured for each plant, roots were visually inspected for symptoms of
disease, and root nodules were counted for the legumes L. capitata and L. perennis. Mean
values across replicate containers were carried forward for analysis so that statistical
replicates corresponded to soil samples collected from the field experiment.
Statistical analyses
Statistical analyses were performed in R (R Development Core Team, 2011).
Streptomycete density measures were log-transformed and inhibition zone sizes were
square root-transformed to improve assumptions of normality. Measures of antagonistic
activity were averaged across overlay strains, except where indicated otherwise.
Differences in antagonistic activity and plant growth performance as a result of soil
conditioning treatment were tested by two-way ANOVA, with conditioning host species
and conditioning plant richness as the two factors. Where significant effects were evident,
Tukey's HSD was used for multiple test correction in contrasts among treatments.
Correlation analyses included associated data on streptomycete community richness and
diversity (as described in Chapter 4) and on plant cover, diversity and productivity in the
experimental plots [obtained through the CCESR Long Term Ecological Research
network database (http://www.cedarcreek.umn.edu/research/data/)]. The significance of
correlation coefficients was adjusted with the FDR (false discovery rate) method for
multiple test correction.
Results
Plant impacts on streptomycete antagonistic potential
108
Streptomycete densities and antagonistic activities varied among communities. Across all
experimental treatments, culturable streptomycete densities ranged from 4.9 x 105 to 2.9
x 106 CFU per gram of soil, with a mean of 1.6 x 106 CFU/g. Inhibitor densities ranged
from 1.5 x 105 to 8.1 x 105 CFU/g, with a mean of 2.9 x 105 CFU/g. Inhibitor frequency
ranged from 10 to 54% of colonies, with a mean of 28% of colonies showing in vitro
inhibitory activity. Intensity of inhibition, measured as the average radius of each
inhibition zone, ranged by a factor of 4.5 among samples (1.4 to 6.1 mm), with a mean
value of 2.9 mm.
Several measures of streptomycete community antagonistic potential were impacted by
the host plant or plant community (Table 1). The density of antagonistic streptomycetes
differed among plant hosts; both L. capitata and S. scoparium supported significantly
higher antagonist densities compared to A. gerardii (Figure 1A). The density of
antagonistic streptomycetes also differed with plant community richness; 8-species plots
supported significantly higher antagonist densities than 16-species plots (Figure 1B) The
proportion of streptomycete isolates showing inhibitory activity also differed among plant
community richness treatments. Plant monocultures supported significantly higher
proportions of antagonistic streptomycetes than 16-species assemblages (Figure 1C).
Finally, the intensity of antagonistic activity among inhibitory streptomycetes was also
impacted by plant community richness; monocultures and 4-species assemblages
supported more strongly antagonistic streptomycetes compared to assemblages of sixteen
plant species (Figure 1D).
Importantly, the impacts of particular host plant species on the antagonistic potential of
streptomycete communities were dependent on the surrounding plant community
richness. For example, among soil streptomycete communities associated with A.
gerardii, the proportion of isolates showing inhibitory activity was highest when A.
gerardii was grown as a monoculture (Figure 2A), while the density of inhibitory isolates
was highest when A. gerardii was grown in eight species assemblages (Figure 2B). Thus
109
plant community context influenced host species effects on associated soil microbial
communities.
The strength of certain relationships among the various measures of antagonistic potential
differed among plant species and plant richness treatments. The correlation between
streptomycete density and antagonist density was positive for all plant species, but was
not significant for A. gerardii (Figure 3A; for A. gerardii, r2 = 0.14, p = 0.19; for L.
capitata, r2 = 0.24, p = 0.06; for L. perennis, r2 = 0.51, p = 0.004; for S. scoparium, r2 =
0.59, p = 0.001). There were no significant correlations for any host species between
antagonist density and antagonist frequency (Figure 3B). Antagonist density and the
intensity of inhibition were significantly positively correlated only for L. capitata (Figure
3C; r2 = 0.44, p = 0.007). Antagonist frequency and intensity of inhibition were also
weakly positively correlated only for L. capitata (Figure 3D; r2 = 0.21, p = 0.09).
Grouping the data according to plant richness treatment, streptomycete density and
antagonist density were significantly positively correlated for all but the most diverse
plant communities (Figure 4A; for monocultures, r2 = 0.85, p = 0.001; for 4 species plots,
r2 = 0.68, p = 0.006; for 8 species plots, r2 = 0.48, p = 0.04; for 16 species plots, r2 = 0.43,
p = 0.06). In contrast, antagonist density and antagonist frequency were significantly
positively correlated only in the most diverse plant communities (Figure 4B; r2 = 0.58, p
= 0.02). There were no significant correlations between intensity of inhibition and either
antagonist density or frequency within individual plant richness treatments (Figure 4C,
D).
Feedbacks of soil conditioning on plant performance
Growth performance among the four plant species varied among soil conditioning
treatments. We found evidence of differential plant-soil feedback dynamics as a function
of conditioning plant species, conditioning plant community richness, and response plant
species. Little disease was evident at the time of harvest, as revealed by visual inspection
110
of washed roots (data not shown), indicating that feedback effects may occur even in the
absence of significant disease.
Several of the response hosts showed differential growth depending on the conditioning
host species. For example, S. scoparium produced significantly more aboveground
biomass in soils conditioned by the legume L. perennis compared to soils conditioned by
either of the two grasses (Figure 5A). In contrast, L. capitata produced significantly more
belowground biomass in soils conditioned by A. gerardii than in conspecific soils (Figure
5B). Biomass allocation in A. gerardii was also altered by soil conditioning, with a
significantly higher ratio of above- to belowground biomass in soils conditioned by L.
capitata than in soils conditioned by L. perennis (Figure 5C).
Plant community richness during soil conditioning also impacted subsequent plant
growth, although to a lesser extent than conditioning host species. Significant differences
in growth among plant richness treatments were found only for root length (Figure 6). In
particular, root length for both L. perennis and S. scoparium was greatest in soil
conditioned by the richest plant communities and there was a similar, though not
significant, trend for A. gerardii (Figure 6).
Furthermore, plant community richness modulated the impacts of soil conditioning by
particular plant species on subsequent plant performance. For example, L. capitata
produced greater aboveground biomass in soils conditioned by A. gerardii grown in
monoculture compared to soils conditioned by A. gerardii grown in more diverse plant
communities (Figure 7A). As another example, total biomass production by L. capitata
showed an interaction between conditioning species and conditioning plant richness (2-
way ANOVA, pinteraction = 0.07); growth response to soil conditioning by A. gerardii and
L. capitata was reduced in high plant community richness, while the opposite was true
for conditioning by L. perennis and S. scoparium (Figure 7B). These results suggest that
host-specific feedback dynamics may vary as a function of surrounding plant richness.
111
In addition to direct impacts on individual hosts, plant-soil feedbacks also altered the
relative performance of the response plants. In most cases, these shifts in relative
performance were consistent with negative conspecific feedbacks. For example, L.
perennis produced the lowest amount of aboveground biomass relative to S. scoparium
when grown in conspecific soil, and S. scoparium similarly produced the lowest amount
of aboveground biomass relative to L. perennis when grown in conspecific soil or soil
conditioned by the other grass species, A. gerardii (Figure 8A). L. capitata produced the
lowest amount of belowground biomass relative to S. scoparium when grown in
conspecific soil (Figure 8B). Similarly, L. capitata produced the shortest root length
relative to S. scoparium, and S. scoparium produced the shortest root length relative to L.
capitata, in conspecific soils (Figure 8C).
Plant richness treatments during soil conditioning had little impact on relative plant
performance compared to the effects of conditioning species. For example, relative
biomass production was not impacted by manipulation of plant richness (data not shown).
While relative root length was impacted in several cases by the plant richness treatment
during soil conditioning, the differences in each case were connected to L. perennis
producing shorter roots in soil conditioned by eight-species plant mixtures (Figure 9).
Correlations between antagonistic streptomycetes and plant-soil feedbacks
We used correlation analyses to test for relationships among soil properties,
streptomycete antagonistic potential and plant growth performance in conditioned soil.
Impacts of plant species and plant community richness on soil edaphic characteristics
were explored earlier (see Table 4, Chapter 4). Specifically, soil collected from L.
perennis had significantly elevated pH relative to A. gerardii, and significantly higher
potassium relative to S. scoparium. Soil collected from sixteen species plots had
significantly higher organic matter, nitrogen and carbon than soil from monocultures.
Additionally, soil potassium levels were greater in the most diverse plant communities
than in plant communities of lower richness (see Table 4, Chapter 4).
112
In general, measures of streptomycete antagonistic potential were negatively correlated
with measures of productivity or fertility, while streptomycete densities were positively
correlated with many of the same variables. Streptomycete density was significantly
positively correlated with plant diversity, soil pH, and soil organic matter, potassium and
nitrogen concentrations (Table 2A). Antagonist density was significantly positively
correlated with soil pH (Table 2A). Antagonist frequency was significantly negatively
correlated with conditioning plant diversity and productivity (belowground biomass,
percent cover, aboveground biomass) and with soil organic matter, potassium, nitrogen
and carbon levels (Table 2A). Intensity of inhibition was similarly negatively correlated
with conditioning plant productivity (belowground biomass, percent cover, aboveground
biomass) and with soil organic matter, potassium, nitrogen and carbon levels, though not
with conditioning plant diversity (Table 2A).
Plant growth performance showed remarkably little relationship with edaphic
characteristics of conditioned soil (Table 2B, C), perhaps because fertilization in the
greenhouse reduced differences in nutrient status among conditioned soils. Root length of
A. gerardii showed the strongest relationships with parameters of conditioned soil, with
greater root length in conditioned soils of lower pH and of higher soil organic matter,
higher soil nitrogen and soil carbon content, or higher plant productivity during
conditioning (Table 2C). Belowground biomass of S. scoparium was also positively
correlated with potassium concentrations of conditioned soils (Table 2B).
Measures of streptomycete antagonistic potential were not consistently related to plant
growth performance. However, streptomycete density showed a positive relationship with
S. scoparium belowground biomass production and root length, and the proportion of
inhibitory streptomycetes was negatively related to S. scoparium belowground biomass
(Table 2A). Streptomycete diversity was also positively related to L. capitata root length
(Table 2C).
113
Discussion
Although many studies have addressed the impacts of plant host species on associated
microbial communities and subsequent plant performance, the impacts of plant diversity
have received far less attention, and only very rarely have the two factors been
deliberately brought together. In this work, both plant species and plant community
richness impacted the antagonistic potential of associated streptomycete communities.
Furthermore, in some cases the impacts of a given host species on associated microbes
were modified by the surrounding plant community richness.
There are two explanations that are commonly invoked to explain impacts of plant
diversity on a variety of ecosystem properties and functions, including impacts on
associated microbial communities. The first highlights the common confounding of plant
diversity and productivity (Zak et al., 2003). It is possible that microbial community
changes resulting from plant diversity manipulations may be due simply to changes in the
amount of plant biomass available to microbial food webs. It should be noted, however,
that not all study systems confound diversity and productivity (eg. Loranger-Merciris et
al., 2006), and statistical approaches can be used to account for productivity differences
among treatments (Bartelt-Ryser et al., 2005; Chung et al., 2007). The second common
explanation, dubbed the sampling effect (Wardle et al., 1999), suggests that diversity may
be important primarily for increasing the likelihood of the presence of particular plant
species having disproportionate impact. However, we suggest an alternative mechanism
by which plant diversity may impact associated microbial communities: surrounding
plant diversity may modify the impacts of a particular host species on the soil microbial
community. This work demonstrates that the impacts of a given host plant species on the
antagonistic potential of associated streptomycetes varies with surrounding plant
richness.
For agricultural application, it will be important to elucidate the mechanistic basis for
plant-driven effects on streptomycete antagonistic potential. Strong relationships between
soil edaphic characteristics and measures of streptomycete antagonistic potential suggest
114
that plant-driven effects may be partially mediated through changes to the chemical
environment in soil. In this regard, it is interesting that antagonist frequency and the
intensity of inhibition were inversely related to measures of soil fertility including
organic matter, potassium, carbon and nitrogen content. If these soil carbon and nutrient
levels are indicative of resource availability for microbes, resource competition among
microbes may be less in soils with higher levels of organic matter, potassium, carbon, and
nitrogen. High resource availability may reduce selective pressure for antagonistic
phenotypes, lowering community level antagonistic potential. This hypothesis should be
explored further with direct manipulative experiments.
Relationships among the various measures of antagonistic potential may provide insights
into the forces that generate and maintain antagonistic phenotypes among soil
streptomycetes (Kinkel et al., 2011). We found that the strength of relationships among
measures of antagonistic potential differed among host plant species and plant richness
treatments. In particular, we observed a steady decline with increasing plant richness in
the strength of the relationship between streptomycete density and antagonist density. In
communities with the highest plant richness, increases in streptomycete density were not
accompanied by corresponding increases in antagonist density. One possible explanation
is that resource diversity for microbes increases with plant species richness. A more
diverse resource base may support niche differentiation as an alternative evolutionary
trajectory to direct resource competition, reducing selection for antagonistic capacity
(Kinkel et al., 2011).
This work also sheds light on the feedback dynamics of four important prairie plant
species. Changes in the relative performance of these species as a result of soil
conditioning provide a mechanism for plant-soil feedbacks to impact plant community
dynamics. Examples of facilitation were observed, where soil conditioned by one species
enhanced the subsequent growth of another species. Relatively poorer performance of
plants in their own conditioned soil suggests that negative conspecific feedbacks operate
in these species, and may contribute to the maintenance of plant diversity (Bever et al.,
115
2010). Negative conspecific feedbacks were observed despite the absence of visible
disease symptoms. This may suggest an important role for chemical mechanisms or
nutrient effects. Alternately, negative conspecific feedbacks in the absence of visible
disease may be explained through changes to communities of plant-growth promoting
microbes or the development of asymptomatic infections.
A legacy effect of the prior plant community, mediated through the soil, had an impact on
subsequent relative performance among these species. By extension, competitive
dynamics among plant species may differ depending on the history of plant species and
plant community richness at a given site. Notably, plant community richness was shown
to modulate host-specific feedback effects. This possibility has rarely been considered in
the ecological literature. A recent meta-analysis of plant-soil feedback studies found that
most have used monocultures (Kulmatiski et al., 2008). Incorporating the impacts of
plant richness on plant host effects on soil microbial communities has profound
implications for our understanding of plant-soil feedbacks. One focal point for studies of
plant-soil feedback has been understanding plant invasion (Klironomos, 2002). Invasion
by an exotic species is often accompanied by a reduction in plant richness, suggesting
that plant-soil feedbacks may be variable over the course of an invasion. The relevance of
broader plant community characteristics has been best appreciated in studies that
approach plant-soil feedbacks to understand successional changes in plant communities.
For example, feedback effects have been shown to differ with plant community
successional stage (Kardol et al., 2006). However, different plant species were studied
among successional stages (Kardol et al., 2006). In contrast, we suggest that the feedback
effects experienced by a given plant species may differ depending on surrounding plant
richness, and potentially due to other community-level characteristics as well.
This work investigated relationships between plant growth performance and
streptomycete community antagonistic potential. There were broad shared patterns of
relationship with soil edaphic characteristics for both streptomycete antagonistic activity
and plant growth performance, suggesting the possibility that both antagonistic potential
116
and plant growth performance respond to soil fertility. However, no observable disease
symptoms were present at the time of plant harvest. It remains possible that antagonistic
streptomycetes could be involved in plant-soil feedbacks under experimental conditions
of greater pathogen activity.
117
Figure 1 - Across plant richness treatments, conditioning plant species impacted the
density of antagonistic streptomycetes in soil (A). Across conditioning plant species,
plant richness influenced the density of antagonistic streptomycetes (B), the frequency of
antagonistic streptomycetes (C) and the intensity of antagonistic activity among
inhibitory streptomycetes (D). Means and standard errors are shown; different letters
indicate significant differences among means (p < 0.05, ANOVA with Tukey multiple
test correction).
A)
B)
5
5.2
5.4
5.6
5.8
6
A. gerardii L. capitata L. perennis S. scoparium
Log antagonist CFU/g
Conditioning plant
a b b ab
5.0
5.2
5.4
5.6
5.8
6.0
1 spp 4 spp 8 spp 16 spp 32 spp
Log antagonist CFU/g
Conditioning plant richness
ab ab a b ab
118
Figure 1, continued
C)
D)
0.0
0.1
0.2
0.3
0.4
0.5
1 spp 4 spp 8 spp 16 spp 32 spp Proportion inhibitory colonies
Conditioning plant richness
a ab ab b ab
0.0
0.5
1.0
1.5
2.0
2.5
1 spp 4 spp 8 spp 16 spp 32 spp
Inhibition zone size
(mm, sqrt transformed)
Conditioning plant richness
a a ab b ab
119
Figure 2 - The antagonistic potential of streptomycetes associated with A. gerardii was
modulated by plant richness. A) The frequency of antagonistic streptomycetes. B) The
density of antagonistic streptomycetes. Means and standard errors are shown; different
letters indicate significant differences among means (p < 0.1, ANOVA with Tukey
multiple test correction).
A)
B)
0
0.1
0.2
0.3
0.4
0.5
0.6
A. gerardii 1 spp
A. gerardii 4 spp
A. gerardii 8 spp
A. gerardii 16 spp
A. gerardii 32 spp
Proportion inhibitory colonies
Conditioning treatment
a
b
ab ab ab
5
5.2
5.4
5.6
5.8
6
A. gerardii 1 spp
A. gerardii 4 spp
A. gerardii 8 spp
A. gerardii 16 spp
A. gerardii 32 spp
Log antagonist CFU/g
Conditioning treatment
a
b ab
ab
b
120
Figure 3 - The strength of relationships among various measures of streptomycete
community antagonistic potential differed among plant host species. Trend lines are
shown for significant relationships only (Pearson correlation with FDR multiple test
correction, p < 0.1). A) Streptomycete density by antagonist density. B) Antagonist
density by antagonist frequency. C) Antagonist density by intensity of inhibition. D)
Antagonist frequency by strength of inhibition.
A)
B)
5.00
5.25
5.50
5.75
6.00
5.50 5.75 6.00 6.25 6.50
Log antagonist CFU/g
Log streptomycete CFU/g
A. gerardii
L. capitata
L. perennis
S. scoparium
L. capitata
L. perennis
S. scoparium
Conditioning species
0.0
0.2
0.4
0.6
5.00 5.25 5.50 5.75 6.00
Proportion inhibitory colonies
Log antagonist CFU/g
A. gerardii
L. capitata
L. perennis
S. scoparium
Conditioning species
121
Figure 3, continued
C)
D)
1
2
3
4
5
6
5.00 5.25 5.50 5.75 6.00
Inhibition zone size (mm)
Log antagonist CFU/g
A. gerardii
L. capitata
L. perennis
S. scoparium
L. capitata
Conditioning species
1
2
3
4
5
6
0.0 0.2 0.4 0.6
Inhibition zone size (mm)
Proportion inhibitory colonies
A. gerardii
L. capitata
L. perennis
S. scoparium
L. capitata
Conditioning species
122
Figure 4 - The strength of relationships among various measures of streptomycete
community antagonistic potential differed among plant richness treatments. Trend lines
are shown for significant relationships only (Pearson correlation with FDR multiple test
correction, p < 0.1). A) Streptomycete density by antagonist density. B) Antagonist
density by antagonist frequency. C) Antagonist density by intensity of inhibition. D)
Antagonist frequency by strength of inhibition.
A)
B)
5.00
5.25
5.50
5.75
6.00
5.50 5.75 6.00 6.25 6.50
Log antagonist CFU/g
Log streptomycete CFU/g
1 spp
4 spp
8 spp
16 spp
32 spp
1 spp
4 spp
8 spp
16 spp
Conditioning plant
richness
0.0
0.2
0.4
0.6
5.00 5.25 5.50 5.75 6.00
Proportion inhibitory colonies
Log antagonist CFU/g
1 spp
4 spp
8 spp
16 spp
32 spp
32 spp
Conditioning plant
richness
123
Figure 4, continued
C)
D)
1
2
3
4
5
6
5.00 5.25 5.50 5.75 6.00
Inhibition zone size (mm)
Log antagonist CFU/g
1 spp
4 spp
8 spp
16 spp
32 spp
Conditioning plant
richness
1
2
3
4
5
6
0.0 0.2 0.4 0.6
Inhibition zone sie (mm)
Proportion inhibitory colonies
1 spp
4 spp
8 spp
16 spp
32 spp
Conditioning plant
richness
124
Figure 5 - Growth responses of four prairie plants varied according to conditioning
species (across conditioning plant richness treatments). A) Aboveground biomass. B)
Belowground biomass. C) Ratio of above- to belowground biomass. Means and standard
errors are shown; different letters indicate significant differences among means (p < 0.1,
ANOVA with Tukey multiple test correction). ns = no significant differences. Ag = A.
gerardii; Lc = L. capitata; Lp = L. perennis; Ss = S. scoparium.
A)
B)
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
A. gerardii L. capitata L. perennis S. scoparium
Aboveground biomass (g)
Response plant
Ag
Lc
Lp
Ss
a
b
a
ns ns ns
ab
Conditioning species
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
A. gerardii L. capitata L. perennis S. scoparium
Below
ground biomass (g)
Response plant
Ag
Lc
Lp
Ss
a
ns
ns
ns
ab ab b
Conditioning species
125
Figure 5, continued
C)
0.0
0.5
1.0
1.5
2.0
2.5
3.0
A. gerardii L. capitata L. perennis S. scoparium
Ratio of above- to
below
ground biomass
Response plant
Ag
Lc
Lp
Ss
a
ns
ns
ns
ab ab b
Conditioning species
126
Figure 6 - Growth response, measured as root length, varied with the conditioning plant
community richness (across conditioning species). Means and standard errors are shown;
within each comparison, different letters indicate significant differences among means (p
< 0.1, ANOVA with Tukey multiple test correction). ns = no significant differences.
0
4
8
12
16
20
A. gerardii L. capitata L. perennis S. scoparium
Root length (cm
)
Response plant
1 spp
4 spp
8 spp
16 spp
32 spp
a b a
b ab ab
ns
ns a a
ab ab
Conditioning plant
richness
127
Figure 7 - Plant community richness modulated the impacts of soil conditioning by
particular host species on subsequent growth response. A) Aboveground biomass
production of four prairie plants in soil conditioned by A. gerardii growing in
assemblages of increasing plant richness. B) Total biomass production by L. capitata
when grown in soils conditioned by various plant species in either low (monoculture or 4
species assemblages) or high diversity (assemblages of 16 or 32 species) plant
communities. Ag = A. gerardii. Means and standard errors are shown; different letters
indicate significant differences among means (p < 0.1, ANOVA with Tukey multiple test
correction). ns = no significant differences.
A)
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
A. gerardii L. capitata L. perennis S. scoparium
Aboveground biomass (g)
Response plant
Ag; 1 spp
Ag; 4 spp
Ag; 8 spp
Ag; 16 spp
Ag; 32 spp
bc a c b abc
ns
ns
ns Conditioning treatment
128
Figure 7, continued
B)
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
A. gerardii L. capitata L. perennis S. scoparium
Total L. capitata biomass (g)
Conditioning species
Low diversity
High diversity
129
Figure 8 - Plant-soil feedbacks impacted the relative performance of four prairie plants.
For each measure of performance, the vertical axis displays the ratio of the value for the
first species relative to the value for the second species. A) Aboveground biomass. B)
Belowground biomass. C) Root length. Ag = A. gerardii; Lc = L. capitata; Lp = L.
perennis; Ss = S. scoparium. Means and standard errors are shown; within each
comparison, different letters indicate significant differences among means (p < 0.1,
ANOVA with Tukey multiple test correction). Cases in which there was no significant
effect of conditioning host are not displayed.
A)
B)
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
Ag:Ss Lp:Ss Ss:Lp
Ratio of aboveground
biomasses
Ag
Lc
Lp
Ss
a b
a
a b a
ab
ab
ab
ab
ab b
Conditioning species
0.0
0.5
1.0
1.5
2.0
2.5
3.0
Lc:Ss Ss:Lc
Ratio of below
ground
biomasses Ag
Lc
Lp
Ss
a b
a b
ab ab
ab ab Conditioning species
130
Figure 8, continued
C)
0.0
0.4
0.8
1.2
1.6
2.0
Lc:Ss Lp:Lc Ss:Lc
Ratio of root lengths
Ag
Lc
Lp
Ss
a b
a b
ab
ab ab
ab ab
a b ab
Conditioning species
131
Figure 9 - Relative root length was impacted by conditioning plant richness (across
conditioning species). The vertical axis displays the ratio of the value for the first species
relative to the value for the second species. Ag = A. gerardii; Lc = L. capitata; Lp = L.
perennis; Ss = S. scoparium. Means and standard errors are shown; within each
comparison, different letters indicate significant differences among means (p < 0.1,
ANOVA with Tukey multiple test correction). Cases in which there was no significant
effect of conditioning host are not displayed.
0.0
0.5
1.0
1.5
2.0
Ag:Lp Lc:Lp Lp:Ag Lp:Lc Lp:Ss Ss:Lp
Ratio of root lengths
1 spp
4 spp
8 spp
16 spp
32 spp
a a
a b
b
b a
a a ab
ab ab
ab ab
ab ab
ab ab
a
b
ab
b
b a
a b
ab ab
b
a
Conditioning plant richness
132
Table 1 - ANOVA results table showing the significance of host plant species, plant
richness, and the interaction between host species and plant richness on various measures
of the antagonistic potential of associated streptomycete communities. * indicates
significant effects.
Streptomycete density
(log CFU/g)
Antagonist density
(log CFU/g)
Proportion inhibitory colonies
Inhibition zone size (mm; sqrt transformed)
Df F p-value F p-value F p-value F p-value Host species 3 0.66 0.58 3.43 0.026 * 0.94 0.43 0.45 0.72 Plant richness 4 1.68 0.17 2.95 0.032 * 3.82 0.011 * 2.75 0.042 * Host species : Plant richness 12 0.32 0.98 0.96 0.50 1.42 0.20 0.74 0.70
Residuals 38
133
Table 2 - Pearson correlation coefficients (p-values) for relationships among
conditioning plant communities, edaphic characteristics of conditioned soil,
streptomycete community antagonistic potential, and greenhouse growth performance.
Highlighted are: A) streptomycete antagonistic potential, B) belowground biomass of
response plants, C) root length of response plants.
A)
(ns) (ns) -0.45 (0.02) -0.37 (0.06)(ns) (ns) -0.54 (<0.01) -0.35 (0.04)
0.44 (0.02) (ns) -0.39 (0.04) (ns)(ns) (ns) -0.50 (<0.01) -0.46 (0.01)(ns) (ns) -0.44 (<0.01) -0.37 (0.03)
0.30 (0.1) (ns) -0.45 (<0.01) -0.33 (0.06)0.33 (0.06) (ns) -0.50 (<0.01) -0.39 (0.02)0.33 (0.06) 0.41 (0.01) (ns) (ns)0.33 (0.06) (ns) -0.65 (<0.01) -0.49 (<0.01)
. 0.57 (<0.01) -0.62 (<0.01) (ns)0.57 (<0.01) . (ns) (ns)
-0.62 (<0.01) (ns) . 0.32 (0.06)(ns) (ns) 0.32 (0.06) .
0.37 (0.03) (ns) -0.30 (0.09) -0.37 (0.02)0.42 (0.01) 0.34 (0.05) (ns) (ns)
A. gerardii (ns) (ns) (ns) (ns)L. capitata (ns) (ns) (ns) (ns)L. perennis (ns) (ns) (ns) (ns)S. scoparium (ns) (ns) (ns) (ns)A. gerardii (ns) (ns) (ns) (ns)L. capitata (ns) (ns) (ns) (ns)L. perennis (ns) (ns) (ns) (ns)S. scoparium 0.33 (0.06) (ns) -0.38 (0.02) (ns)A. gerardii (ns) (ns) (ns) (ns)L. capitata (ns) (ns) (ns) (ns)L. perennis (ns) (ns) (ns) (ns)S. scoparium 0.39 (0.02) (ns) (ns) (ns)
Streptomycetedensity
(log CFU/g)
Antagonistdensity
(log CFU/g)
Inhibitionzone (mm; sqrt transformed)
Antagonist frequency
(proportion of isolates)
Below-ground
biomass (g)
Root length (cm)
Streptomycete density (log CFU/g)Antagonist density (log CFU/g)
Inhibition zone (mm; sqrt transformed)Antagonist frequency
Soilstreptomycete communities
Greenhousegrowth response
Soil carbon (%)
Belowground biomass (g/m2, 2006)Total plant cover (%, 2007)Plant diversity (Shannon index, 2008)Aboveground biomass (g/m2, 2009)
Streptomycete richness (Chao estimator)Streptomycete diversity (Shannon index)
Soil pHSoil organic matter (%)
Soil potassium (ppm)
Characteristics of plant
community
Characteristicsof conditioned
soil
Soil nitrogen (%)
Above-ground
biomass (g)
134
(ns) (ns) (ns) (ns)(ns) (ns) (ns) (ns)(ns) (ns) (ns) (ns)(ns) (ns) (ns) (ns)(ns) (ns) (ns) (ns)(ns) (ns) (ns) (ns)(ns) (ns) (ns) (ns)(ns) (ns) (ns) (ns)(ns) (ns) (ns) 0.31 (0.09)(ns) (ns) (ns) 0.33 (0.06)(ns) (ns) (ns) (ns)(ns) (ns) (ns) -0.38 (0.02)(ns) (ns) (ns) (ns)(ns) (ns) (ns) (ns)(ns) (ns) (ns) (ns)
A. gerardii 0.49 (<0.01) (ns) (ns) (ns)L. capitata (ns) (ns) (ns) (ns)L. perennis (ns) (ns) 0.34 (0.05) (ns)S. scoparium (ns) 0.32 (0.06) (ns) 0.83 (<0.01)A. gerardii . 0.48 (<0.01) (ns) (ns)L. capitata 0.48 (<0.01) . (ns) (ns)L. perennis (ns) (ns) . (ns)S. scoparium (ns) (ns) (ns) .A. gerardii (ns) (ns) (ns) (ns)L. capitata 0.31 (0.07) (ns) (ns) (ns)L. perennis (ns) (ns) (ns) (ns)S. scoparium (ns) (ns) (ns) (ns)
Greenhouse belowground biomass (g)
A. gerardii L. capitata L. perennis S. scoparium
Characteristics of plant
community
Characteristicsof conditioned
soil
Soilstreptomycete communities
Greenhousegrowth response
Total plant cover (%, 2007)Plant diversity (Shannon index, 2008)Aboveground biomass (g/m2, 2009)
Belowground biomass (g/m2, 2006)
Streptomycete density (log CFU/g)Antagonist density (log CFU/g)
Inhibition zone (mm; sqrt transformed)
Soil carbon (%)
Above-ground
biomass (g)
Below-ground
biomass (g)
Root length (cm)
Streptomycete richness (Chao estimator)Streptomycete diversity (Shannon index)
Antagonist frequency
Soil pHSoil organic matter (%)
Soil potassium (ppm)
Soil nitrogen (%)
(ns) (ns) -0.45 (0.02) -0.37 (0.06)(ns) (ns) -0.54 (<0.01) -0.35 (0.04)
0.44 (0.02) (ns) -0.39 (0.04) (ns)(ns) (ns) -0.50 (<0.01) -0.46 (0.01)(ns) (ns) -0.44 (<0.01) -0.37 (0.03)
0.30 (0.1) (ns) -0.45 (<0.01) -0.33 (0.06)0.33 (0.06) (ns) -0.50 (<0.01) -0.39 (0.02)0.33 (0.06) 0.41 (0.01) (ns) (ns)0.33 (0.06) (ns) -0.65 (<0.01) -0.49 (<0.01)
. 0.57 (<0.01) -0.62 (<0.01) (ns)0.57 (<0.01) . (ns) (ns)
-0.62 (<0.01) (ns) . 0.32 (0.06)(ns) (ns) 0.32 (0.06) .
0.37 (0.03) (ns) -0.30 (0.09) -0.37 (0.02)0.42 (0.01) 0.34 (0.05) (ns) (ns)
A. gerardii (ns) (ns) (ns) (ns)L. capitata (ns) (ns) (ns) (ns)L. perennis (ns) (ns) (ns) (ns)S. scoparium (ns) (ns) (ns) (ns)A. gerardii (ns) (ns) (ns) (ns)L. capitata (ns) (ns) (ns) (ns)L. perennis (ns) (ns) (ns) (ns)S. scoparium 0.33 (0.06) (ns) -0.38 (0.02) (ns)A. gerardii (ns) (ns) (ns) (ns)L. capitata (ns) (ns) (ns) (ns)L. perennis (ns) (ns) (ns) (ns)S. scoparium 0.39 (0.02) (ns) (ns) (ns)
Streptomycetedensity
(log CFU/g)
Antagonistdensity
(log CFU/g)
Inhibitionzone (mm; sqrt transformed)
Antagonist frequency
(proportion of isolates)
Below-ground
biomass (g)
Root length (cm)
Streptomycete density (log CFU/g)Antagonist density (log CFU/g)
Inhibition zone (mm; sqrt transformed)Antagonist frequency
Soilstreptomycete communities
Greenhousegrowth response
Soil carbon (%)
Belowground biomass (g/m2, 2006)Total plant cover (%, 2007)Plant diversity (Shannon index, 2008)Aboveground biomass (g/m2, 2009)
Streptomycete richness (Chao estimator)Streptomycete diversity (Shannon index)
Soil pHSoil organic matter (%)
Soil potassium (ppm)
Characteristics of plant
community
Characteristicsof conditioned
soil
Soil nitrogen (%)
Above-ground
biomass (g)
Table 2, continued
B)
135
Table 2, continued
C)
(ns) (ns) (ns) (ns)(ns) (ns) (ns) (ns)(ns) (ns) (ns) (ns)
0.46 (0.01) (ns) (ns) (ns)0.33 (0.06) (ns) (ns) (ns)0.3 (0.09) (ns) (ns) (ns)
0.40 (0.02) (ns) (ns) (ns)-0.33 (0.06) (ns) (ns) (ns)
(ns) (ns) (ns) (ns)(ns) (ns) (ns) 0.39 (0.02)(ns) (ns) (ns) (ns)(ns) (ns) (ns) (ns)(ns) (ns) (ns) (ns)(ns) (ns) (ns) (ns)(ns) 0.33 (0.06) (ns) (ns)
A. gerardii -0.41 (0.01) (ns) (ns) -0.32 (0.06)L. capitata (ns) (ns) (ns) (ns)L. perennis (ns) (ns) (ns) (ns)S. scoparium -0.35 (0.04) (ns) -0.41 (0.01) (ns)A. gerardii (ns) 0.31 (0.07) (ns) (ns)L. capitata (ns) (ns) (ns) (ns)L. perennis (ns) (ns) (ns) (ns)S. scoparium (ns) (ns) (ns) (ns)A. gerardii . (ns) 0.42 (0.01) 0.47 (<0.01)L. capitata (ns) . (ns) (ns)L. perennis 0.42 (0.01) (ns) . 0.43 (0.01)S. scoparium 0.47 (<0.01) (ns) 0.43 (0.01) .
L. perennis S. scoparium
Greenhouse root length (cm)
A. gerardii L. capitata
Below-ground
biomass (g)
Belowground biomass (g/m2, 2006)Total plant cover (%, 2007)Plant diversity (Shannon index, 2008)Aboveground biomass (g/m2, 2009)
Characteristics of plant
community
Soilstreptomycete communities
Greenhousegrowth response
Soil carbon (%)Characteristicsof conditioned
soil
Root length (cm)
Streptomycete richness (Chao estimator)Streptomycete diversity (Shannon index)
Soil pHSoil organic matter (%)
Soil potassium (ppm)
Soil nitrogen (%)
Streptomycete density (log CFU/g)Antagonist density (log CFU/g)
Inhibition zone (mm; sqrt transformed)Antagonist frequency
Above-ground
biomass (g)
136
Bibliography
Acosta-Martinez, V., S. Dowd, Y. Sun, and V. Allen. 2008. Tag-encoded pyrosequencing analysis of bacterial diversity in a single soil type as affected by management and land use. Soil Biol Biochem. 40(11): 2762-2770.
Aminov, R.I. 2009. The role of antibiotics and antibiotic resistance in nature. Environ Microbiol. 11(12): 2970-2988.
Anderson, A.S., and E.M.H. Wellington. 2001. The taxonomy of Streptomyces and related genera. Int J Sys Evol Microbiol. 51: 797-814.
Angell, S., B.J. Bench, H. Williams, and C.M.H. Watanabe. 2006. Pyocyanin isolated from a marine microbial population: Synergistic production between two distinct bacterial species and mode of action. Chem Biol. 13: 1349-1359.
Anukool, U., W.H. Gaze, and E.M.H. Wellington. 2004. In situ monitoring of streptothricin production by Streptomyces rochei F20 in soil and rhizosphere. Appl Environ Microbiol. 70(9): 5222-5228.
Ashelford, K.E., N.A. Chuzhanova, J.C. Fry, A.J. Jones, and A.J. Weightman. 2005. At least 1 in 20 16S rRNA sequence records currently held in public repositories is estimated to contain substantial anomalies. Appl Environ Microbiol. 71(12): 7724-7736.
Badri, D.V., V.M. Loyola-Vargas, C.D. Broeckling, C. De-la-Pena, M. Jasinski, D. Santelia, E. Martinoia, L.W. Sumner, L.M. Banta, F. Stermitz, and J.M. Vivanco. 2008. Altered profile of secondary metabolites in the root exudates of Arabidopsis ATP-binding cassette transporter mutants. Plant Phys. 146(2): 762-771.
Badri, D.V., N. Quintana, E.G. El Kassis, H.K. Kim, Y.H. Choi, A. Sugiyama, R. Verpoorte, E. Martinoia, D.K. Manter, and J.M. Vivanco. 2009. An ABC transporter mutation alters root exudation of phytochemicals that provoke an overhaul of natural soil microbiota. Plant Phys. 151(4): 2006-2017.
Bagnarol, E., J. Popovici, N. Alloisio, J. Marechal, P. Pujic, P. Normand, and M.P. Fernandez. 2007. Differential Frankia protein patterns induced by phenolic extracts from Myricaceae seeds. Physiologia Plantarum. 130(3): 380-390.
Bais, H.P., T.L. Weir, L.G. Perry, S. Gilroy, and J.M. Vivanco. 2006. The role of root exudates in rhizosphere interactions with plants and other organisms. Annu Rev Plant Biol. 57(1): 233-266.
137
Barber, D.A., and K.B. Gunn. 1974. The effect of mechanical forces on the exudation of organic substances by the roots of cereal plants grown under sterile conditions. New Phytol. 73: 39-45.
Bardgett, R.D., and L.R. Walker. 2004. Impact of coloniser plant species on the development of decomposer microbial communities following deglaciation. Soil Biol Biochem. 36(3): 555-559.
Barea, J.M., M.J. Pozo, R. Azcon, and C. Azcon-Aguilar. 2005. Microbial co-operation in the rhizosphere. J Exp Botany. 56(417): 1761-1778.
Barona-Gomez, F., S. Lautru, F.X. Francou, P. Leblond, J.L. Pernodet, and G.L. Challis. 2006. Multiple biosynthetic and uptake systems mediate siderophore-dependent iron acquisition in Streptomyces coelicolor A3(2) and Streptomyces ambofaciens ATCC 23877. Microbiology. 152: 3355-3366.
Bartelt-Ryser, J., J. Joshi, B. Schmid, H. Brandl, and T. Balser. 2005. Soil feedbacks of plant diversity on soil microbial communities and subsequent plant growth. Perspective Plant Ecol Evol Sys. 7(1): 27-49.
Batten, K.M., K.M. Scow, K.F. Davies, and S.P. Harrison. 2006. Two invasive plants alter soil microbial community composition in serpentine grasslands. Biol Invasion. 8(2): 217-230.
Batten, K.M., K.M. Scow, and E.K. Espeland. 2007. Soil microbial community associated with an invasive grass differentially impacts native plant performance. Microb Ecol. 55(2): 220-228.
Becker, D.M., L.L. Kinkel, and J.L. Schottel. 1997. Evidence for interspecies communication and its potential role in pathogen suppression in a naturally occurring disease suppressive soil. Can J Microbiol. 43(10): 985-990.
Benizri, E., and B. Amiaud. 2005. Relationship between plants and soil microbial communities in fertilized grasslands. Soil Biol Biochem. 37(11): 2055-2064.
Bentley, S.D., K.F. Chater, A.M. Cerdeno-Tarraga, G.L. Challis, N.R. Thomson, K.D. James, D.E. Harris, M.A. Quail, H. Kieser, D. Harper, A. Bateman, S. Brown, G. Chandra, C.W. Chen, M. Collins, A. Cronin, A. Fraser, A. Goble, J. Hidalgo, T. Hornsby, S. Howarth, C.H. Huang, T. Kieser, L. Larke, L. Murphy, K. Oliver, S. O’Neil, E. Rabbinowitsch, M.A. Rajandream, K. Rutherford, S. Rutter, K. Seeger, D. Saunders, S. Sharp, R. Squares, S. Squares, K. Taylor, T. Warren, A. Wietzorrek, J. Woodward, B.G. Barrell, J. Parkhill, and D.A. Hopwood. 2002. Complete genome sequence of the model actinomycete Streptomyces coelicolor A3(2). Nature. 417(6885): 141-147.
138
Bergsma-Vlami, M., M.E. Prins, and J.M. Raaijmakers. 2005a. Influence of plant species on population dynamics, genotypic diversity and antibiotic production in the rhizosphere by indigenous Pseudomonas spp. FEMS Microbiol Ecol. 52(1): 59-69.
Bergsma-Vlami, M., M.E. Prins, M. Staats, and J.M. Raaijmakers. 2005b. Assessment of genotypic diversity of antibiotic-producing Pseudomonas species in the rhizosphere by denaturing gradient gel electrophoresis. Appl Environ Microbiol. 1(2): 993-1003.
Berg, G., K. Opelt, C. Zachow, J. Lottmann, M. Gotz, R. Costa, and K. Smalla. 2006. The rhizosphere effect on bacteria antagonistic towards the pathogenic fungus Verticillium differs depending on plant species and site. FEMS Microbiol Ecol. 56(2): 250-261.
Berg, G., N. Roskot, A. Steidle, L. Eberl, A. Zock, and K. Smalla. 2002. Plant-dependent genotypic and phenotypic diversity of antagonistic rhizobacteria isolated from different Verticillium host plants. Appl Environ Microbiol. 68(7): 3328-3338.
Berg, G., C. Zachow, J. Lottmann, M. Gotz, R. Costa, and K. Smalla. 2005. Impact of plant species and site on rhizosphere-associated fungi antagonistic to Verticillium dahliae Kleb. Appl Environ Microbiol. 71(8): 4203-4213.
Besserer, A., V. Puech-Pages, P. Kiefer, V. Gomez-Roldan, A. Jauneau, S. Roy, J.C. Portais, C. Roux, G. Becard, and N. Sejalon-Delmas. 2006. Strigolactones stimulate arbuscular mycorrhizal fungi by activating mitochondria. PLoS Biology. 4(7): 1239-1247.
Bever, J.D., I.A. Dickie, E. Facelli, J.M. Facelli, J. Klironomos, M. Moora, M.C. Rillig, W.D. Stock, M. Tibbett, and M. Zobel. 2010. Rooting theories of plant community ecology in microbial interactions. Trend Ecol Evol. 25(8): 468-478.
Bever, J., K. Westover, and J. Antonovics. 1997. Incorporating the soil community into plant population dynamics: The utility of the feedback approach. J Ecol. 85: 561-573.
Blattner, F.R., et al. 1997. The complete genome sequence of Escherichia coli K-12. Science. 277(5331): 1453-1462.
Bowden, R.D., E. Davidson, K. Savage, C. Arabia, and P. Steudler. 2004. Chronic nitrogen additions reduce total soil respiration and microbial respiration in temperate forest soils at the Harvard Forest. Forest Ecol Manage. 196(1): 43-56.
Bremer, C., G. Braker, D. Matthies, C. Beierkuhnlein, and R. Conrad. 2009. Plant presence and species combination, but not diversity, influence denitrifier activity
139
and the composition of nirK-type denitrifier communities in grassland soil. FEMS Microbiol Ecol. 70(3): 377-387.
Bremer, C., G. Braker, D. Matthies, A. Reuter, C. Engels, and R. Conrad. 2007. Impact of plant functional group, plant species, and sampling time on the composition of nirK-type denitrifier communities in soil. Appl Environ Microbiol. 73(21): 6876-6884.
Briones, A.M., S. Okabe, Y. Umemiya, N.B. Ramsing, W. Reichardt, and H. Okuyama. 2002. Influence of different cultivars on populations of ammonia-oxidizing bacteria in the root environment of rice. Appl Environ Microbiol. 68(6): 3067-3075.
Broeckling, C.D., A.K. Broz, J. Bergelson, D.K. Manter, and J.M. Vivanco. 2008. Root exudates regulate soil fungal community composition and diversty. Appl Environ Microbiol. 74(3): 738-744.
Broughton, W.J., F. Zhang, X. Perret, and C. Staehelin. 2003. Signals exchanged between legumes and Rhizobium: Agricultural uses and perspectives. Plant Soil. 252(1): 129-137.
Broz, A.K., D.K. Manter, and J.M. Vivanco. 2007. Soil fungal abundance and diversity: Another victim of the invasive plant Centaurea maculosa. ISME J. 1(8): 763-765.
Callaway, R.M., D. Cipollini, K. Barto, G.C. Thelen, S.G. Hallett, D. Prati, K. Stinson, and J. Klironomos. 2008. Novel weapons: Invasive plant suppresses fungal mutualists in American but not in its native Europe. Ecology. 89(4): 1043-1055.
Caporaso, J.G., J. Kuczynski, J. Stombaugh, K. Bittinger, F.D. Bushman, E.K. Costello, N. Fierer, A.G. Peña, J.K. Goodrich, J.I. Gordon, G.A. Huttley, S.T. Kelley, D. Knights, J.E. Koenig, R.E. Ley, C.A. Lozupone, D. McDonald, B.D. Muegge, M. Pirrung, J. Reeder, J.R. Sevinsky, P.J. Turnbaugh, W.A. Walters, J. Widmann, T. Yatsunenko, J. Zaneveld, and R. Knight. 2010. QIIME allows analysis of high-throughput community sequencing data. Nat Meth. 7(5): 335-336.
Carney, K.M., and P.A. Matson. 2006. The influence of tropical plant diversity and composition on soil microbial communities. Microb Ecol. 52(2): 226-238.
Carson, J.K., L. Campbell, D. Rooney, N. Clipson, and D.B. Gleeson. 2009. Minerals in soil select distinct bacterial communities in their microhabitats. FEMS Microbiol Ecol. 67(3): 381-388.
Caruso, T., Y. Chan, D.C. Lacap, M.C.Y. Lau, C.P. McKay, and S.B. Pointing. 2011. Stochastic and deterministic processes interact in the assembly of desert microbial communities on a global scale. ISME J. E-pub ahead of print. DOI: 10.1038/ismej.2011.21.
140
Casper, B.B., S.P. Bentivenga, B. Ji, J.H. Doherty, H.M. Edenborn, and D.J. Gustafson. 2008. Plant-soil feedback: Testing the generality with the same grasses in serpentine and prairie soils. Ecology. 89(8): 2154-2164.
Challis, G.L., and D.A. Hopwood. 2003. Synergy and contingency as driving forces for the evolution of multiple secondary metabolite production by Streptomyces species. PNAS. 100: 14555-14561.
Chamberlain, K., and D.L. Crawford. 1999. In vitro and in vivo antagonism of pathogenic turf grass fungi by Streptomyces hygroscopicus strains YCED9 and WYE53. J Indust Microbiol Biotech. 23(1): 641-646.
Chase, J.M. 2010. Stochastic community assembly causes higher biodiversity in more productive environments. Science. 328(5984): 1388-1391.
Chater, K.F., S. Biró, K.J. Lee, T. Palmer, and H. Schrempf. 2010. The complex extracellular biology of Streptomyces. FEMS Microbiol Rev. 34(2): 171-198.
Choi, S.U., C.K. Lee, Y.I. Hwang, H. Kinosita, and T. Nihira. 2003. Gamma-butyrolactone autoregulators and receptor proteins in non-Streptomyces actinomycetes producing commercially important secondary metabolites. Arch Microbiol. 180(4): 303-307.
Chung, H., D.R. Zak, P.B. Reich, and D.S. Ellsworth. 2007. Plant species richness, elevated CO2, and atmospheric nitrogen deposition alter soil microbial community composition and function. Global Change Biol. 13(5): 980-989.
Claesson, M.J., O. O’Sullivan, Q. Wang, J. Nikkilä, J.R. Marchesi, H. Smidt, W.M. de Vos, R.P. Ross, and P.W. O’Toole. 2009. Comparative analysis of pyrosequencing and a phylogenetic microarray for exploring microbial community structures in the human distal intestine. PLoS ONE. 4(8): e6669.
Cole, J.R., Q. Wang, E. Cardenas, J. Fish, B. Chai, R.J. Farris, A.S. Kulam-Syed-Mohideen, D.M. McGarrell, T. Marsh, G.M. Garrity, and J.M. Tiedje. 2009. The Ribosomal Database Project: Improved alignments and new tools for rRNA analysis. Nuc Acid Res. 37: D141-D145.
Compant, S., B. Duffy, J. Nowak, C. Clement, and E.A. Barka. 2005. Use of plant growth-promoting bacteria for biocontrol of plant diseases: Principles, mechanisms of action, and future prospects. Appl Environ Microbiol. 71(9): 4951-4959.
Compton, J.E., L.S. Watrud, L.A. Porteous, and S. DeGrood. 2004. Response of soil microbial biomass and community composition to chronic nitrogen additions at Harvard Forest. Forest Ecol Manage. 196(1): 143-158.
141
Corre, C., L. Song, S. O’Rourke, K.F. Chater, and G.L. Challis. 2008. 2-alkyl-4-hydroxymethylfuran-3-carboxylic acids, antibiotic production inducers discovered by Streptomyces coelicolor genome mining. PNAS. 105(45): 17510-17515.
Costa, R., M. Gotz, N. Mrotzek, J. Lottmann, G. Berg, and K. Smalla. 2006. Effects of site and plant species on rhizosphere community structure as revealed by molecular analysis of microbial guilds. FEMS Microbiol Ecol. 56(2): 236-249.
Culman, S.W., S.T. DuPont, J.D. Glover, D.H. Buckley, G.W. Fick, H. Ferris, and T.E. Crews. 2010. Long-term impacts of high-input annual cropping and unfertilized perennial grass production on soil properties and belowground food webs in Kansas, USA. Ag Ecosys Environ. 137(1-2): 13-24.
Cundliffe, E. 2006. Antibiotic production by actinomycetes: The Janus faces of regulation. J Indust Microbiol Biotech. 33(7): 500-506.
Czarnota, M.A., A.M. Rimando, and L.A. Weston. 2003. Evaluation of root exudates of seven sorghum accessions. J Chem Ecol. 29(9): 2073-2083.
Dalmastri, C., L. Chiarini, C. Cantale, A. Bevivino, and S. Tabacchioni. 1999. Soil type and maize cultivar affect the genetic diversity of maize root-associated Burkholderia cepacia populations. Microb Ecol. 38(3): 273-284.
Davelos-Baines, A.L., K. Xiao, and L.L. Kinkel. 2007. Lack of correspondence between genetic and phenotypic groups amongst soil-borne streptomycetes. FEMS Microbiol Ecol. 59(3): 564-575.
Davelos, A.L., L.L. Kinkel, and D.A. Samac. 2004a. Spatial variation in frequency and intensity of antibiotic interactions among streptomycetes from prairie soil. Appl Environ Microbiol. 70(2): 1051-1058.
Davelos, A.L., K. Xiao, J.M. Flor, and L.L. Kinkel. 2004b. Genetic and phenotypic traits of streptomycetes used to characterize antibiotic activities of field-collected microbes. Can J Microbiol. 50(2): 79-89.
Davis, M.J., S.N. Mondal, H. Chen, M.E. Rogers, and R.H. BrIansky. 2008. Co-cultivation of Candidatus Liberibacter asiaticus with Actinobacteria from citrus with Huanglongbing. Plant Disease. 92(11): 1547-1550.
De-la-Pena, C., Z. Lei, B. Watson, D. Badri, M. Brandao, M. Silva-Filho, L. Sumner, and J. Vivanco. 2010. Root secretion of defense-related proteins is development-dependent and correlated with flowering time. J Biol Chem. 285(40): 30645-30665.
142
De-la-Pena, C., Z. Lei, B.S. Watson, L.W. Sumner, and J.M. Vivanco. 2008. Root-microbe communication through protein secretion. J Biol Chem. 283(37): 25247-25255.
DeForest, J.L., D.R. Zak, K.S. Pregitzer, and A.J. Burton. 2004. Atmospheric nitrate deposition, microbial community composition, and enzyme activity in northern hardwood forests. Soil Sci Soc Am J. 68(1): 132-138.
Delalande, L., D. Faure, A. Raffoux, S. Uroz, C. D’Angelo-Picard, M. Elasri, A. Carlier, R. Berruyer, A. Petit, P. Williams, and Y. Dessaux. 2005. N-hexanoyl-L-homoserine lactone, a mediator of bacterial quorum-sensing regulation, exhibits plant-dependent stability and may be inactivated by germinating Lotus corniculatus seedlings. FEMS Microbiol Ecol. 52(1): 13-20.
Dennis, P.G., A.J. Miller, and P.R. Hirsch. 2010. Are root exudates more important than other sources of rhizodeposits in structuring rhizosphere bacterial communities? FEMS Microbiol Ecol. 72(3): 313-327.
de Dorlodot, S., B. Forster, L. Pages, A. Price, R. Tuberosa, and X. Draye. 2007. Root system architecture: Opportunities and constraints for genetic improvement of crops. Trend Plant Sci. 12(10): 474-481.
Dormaar, J.F., B.C. Tovell, and W.D. Willms. 2002. Fingerprint composition of seedling root exudates of selected grasses. J Range Manage. 55(4): 420-423.
Doumbou, C.L., M.K. Hamby Salove, D.L. Crawford, and C. Beaulieu. 2001. Actinomycetes, promising tools to control plant diseases and to promote plant growth. Phytoprotection. 82(3): 85-102.
Droege, M., and B. Hill. 2008. The Genome Sequencer FLXTM System - Longer reads, more applications, straight forward bioinformatics and more complete data sets. J Biotech. 136(1-2): 3-10.
Ducklow, H. 2008. Microbial services: Challenges for microbial ecologists in a changing world. Aquat Microb Ecol. 53(1): 13-19.
Dudler, R., and L. Eberl. 2006. Interactions between bacteria and eukaryotes via small molecules. Curr Op Biotech. 17(3): 268-273.
DuPont, S.T., S.W. Culman, H. Ferris, D.H. Buckley, and J.D. Glover. 2010. No-tillage conversion of harvested perennial grassland to annual cropland reduces root biomass, decreases active carbon stocks, and impacts soil biota. Ag Ecosys Environ. 137(1-2): 25-32.
143
Edwards, U., T. Rogall, H. Blocker, M. Emde, and E.C. Bottger. 1989. Isolation and direct complete nucleotide determination of entire genes - Characterization of a gene coding for 16S-ribosomal RNA. Nuc Acid Res. 17(19): 7843-7853.
Ehrenfeld, J.G., B. Ravit, and K. Elgersma. 2005. Feedback in the plant-soil system. Annu Rev Environ Res. 30: 75-115.
El-Tarabily, K.A., G.E.S.J. Hardy, K. Sivasithamparam, A.M. Hussein, and D.I. Kurtboke. 1997. The potential for the biological control of cavity-spot disease of carrots, caused by Pythium coloratum, by streptomycete and non-streptomycete actinomycetes. New Phytol. 137(3): 495-507.
van Elsas, J.D., P. Garbeva, and J. Salles. 2002. Effects of agronomical measures on the microbial diversity of soils as related to the suppression of soil-borne plant pathogens. Biodegradation. 13(1): 29-40.
Emmert, E.A.B., and J. Handelsman. 1999. Biocontrol of plant disease: A (Gram-) positive perspective. FEMS Microbiol Letter. 171(1): 1-9.
Ensign, J.C. 1978. Formation, properties, and germination of Actinomycete spores. Annu Rev Microbiol. 32: 185-219.
Eppinga, M.B., M. Rietkerk, S.C. Dekker, P.C. De Ruiter, W.H. Van der Putten, and W.H. Van der Putten. 2006. Accumulation of local pathogens: A new hypothesis to explain exotic plant invasions. Oikos. 114(1): 168-176.
Ezziyyani, M., M.E. Requena, C. Egea-Gilabert, and M.E. Candela. 2007. Biological control of Phytophthora root rot of pepper using Trichoderma harzianum and Streptomyces rochei in combination. J Phytopath. 155(6): 342-349.
Fierer, N., and R.B. Jackson. 2006. The diversity and biogeography of soil bacterial communities. PNAS. 103(3): 626-631.
Fravel, D.R. 1988. Role of antibiosis in the biocontrol of plant diseases. Annu Rev Phytopathol. 26: 75-91.
Freilich, S., A. Kreimer, I. Meilijson, U. Gophna, R. Sharan, and E. Ruppin. 2010. The large-scale organization of the bacterial network of ecological co-occurrence interactions. Nuc Acid Res. 38(12): 3857-3868.
Fuhrman, J., and J. Steele. 2008. Community structure of marine bacterioplankton: Patterns, networks, and relationships to function. Aquat Microb Ecol. 53: 69-81.
Fuhrman, J.A. 2009. Microbial community structure and its functional implications. Nature. 459(7244): 193-199.
144
Funnell-Harris, D.L., J.F. Pedersen, and S.E. Sattler. 2010. Soil and root populations of fluorescent Pseudomonas spp. associated with seedlings and field-grown plants are affected by sorghum genotype. Plant Soil. 335(1-2): 439-455.
Gao, M.S., M. Teplitski, J.B. Robinson, and W.D. Bauer. 2003. Production of substances by Medicago truncatula that affect bacterial quorum sensing. Mol Plant-Microbe Interact. 16(9): 827-834.
Garbeva, P., J.D. van Elsas, and J.A. van Veen. 2008. Rhizosphere microbial community and its response to plant species and soil history. Plant Soil. 302(1-2): 19-32.
Garbeva, P., J. Postma, J.A. van Veen, and J.D. van Elsas. 2006. Effect of above-ground plant species on soil microbial community structure and its impact on suppression of Rhizoctonia solani AG3. Environ Microbiol. 8(2): 233-246.
Genilloud, O., I. González, O. Salazar, J. Martín, J.R. Tormo, and F. Vicente. 2011. Current approaches to exploit actinomycetes as a source of novel natural products. J Ind Microbiol Biotechnol. 38(3): 375-389.
Germida, J.J., S.D. Siciliano, J.R. de Freitas, and A.M. Seib. 1998. Diversity of root-associated bacteria associated with field-grown canola (Brassica napus L.) and wheat (Triticum aestivum L.). FEMS Microbiol Ecol. 26(1): 43-50.
Ghini, R., and M.A.B. Morandi. 2006. Biotic and abiotic factors associated with soil suppressiveness to Rhizoctonia solani. Scientia Agricola. 63(2): 153-160.
Gilbert, J.A., D. Field, P. Swift, L. Newbold, A. Oliver, T. Smyth, P.J. Somerfield, S. Huse, and I. Joint. 2009. The seasonal structure of microbial communities in the Western English Channel. Environ Microbiol. 11(12): 3132-3139.
Giovanni, G.D.D., L.S. Watrud, R.J. Seidler, and F. Widmer. 1999. Comparison of parental and transgenic alfalfa rhizosphere bacterial communities using Biolog GN metabolic fingerprinting and enterobacterial repetitive intergenic consensus sequence-PCR (ERIC-PCR). Microb Ecol. 37(2): 129-139.
Girvan, M.S., J. Bullimore, J.N. Pretty, A.M. Osborn, and A.S. Ball. 2003. Soil type is the primary determinant of the composition of the total and active bacterial communities in arable soils. Appl Environ Microbiol. 69(3): 1800-1809.
Glover, J.D., S.W. Culman, S.T. DuPont, W. Broussard, L. Young, M.E. Mangan, J.G. Mai, T.E. Crews, L.R. DeHaan, and D.H. Buckley. 2010. Harvested perennial grasslands provide ecological benchmarks for agricultural sustainability. Ag Ecosys Environ. 137(1-2): 3-12.
145
Goh, E.B., G. Yim, W. Tsui, J. McClure, M.G. Surette, and J. Davies. 2002. Transcriptional modulation of bacterial gene expression by subinhibitory concentrations of antibiotics. PNAS. 99(26): 17025-17030.
Gottelt, M. 2010. Regulation of antibiotic production in Streptomyces coelicolor: A bacterial hormone receptor variant and the awakening of a cryptic antibiotic biosynthesis gene cluster. PhD thesis. Rijksuniversiteit Groningen.
Grayston, S.J., S.Q. Wang, C.D. Campbell, and A.C. Edwards. 1998. Selective influence of plant species on microbial diversity in the rhizosphere. Soil Biol Biochem. 30(3): 369-378.
Guetsky, R., D. Shtienberg, Y. Elad, E. Fischer, and A. Dinoor. 2002. Improving biological control by combining biocontrol agents each with several mechanisms of disease suppression. Phytopathology. 92(9): 976-985.
Haas, D., and C. Keel. 2003. Regulation of antibiotic production in root-colonizing Pseudomonas spp. and relevance for biological control of plant disease. Annu Rev Phytopathol. 41: 117-153.
Hamel, C., V. Vujanovic, R. Jeannotte, A. Nakano-Hylander, and M. St-Arnaud. 2005. Negative feedback on a perennial crop: Fusarium crown and root rot of asparagus is related to changes in soil microbial community structure. Plant Soil. 268(1): 75-87.
Hamilton, E.W., and D.A. Frank. 2001. Can plants stimulate soil microbes and their own nutrient supply? Evidence from a grazing tolerant grass. Ecology. 82(9): 2397-2402.
Hansen, S.K., P.B. Rainey, J.A.J. Haagensen, and S. Molin. 2007. Evolution of species interactions in a biofilm community. Nature. 445(7127): 533-536.
Hara, O., and T. Beppu. 1982. Mutants blocked in streptomycin production in Streptomyces griseus - The role of A-factor. J Antibiot. 35(3): 349-358.
Hashimoto, K., T. Nihira, and Y. Yamada. 1992. Distribution of virginiae butanolides and IM-2 in the genus Streptomyces. J Ferment Bioeng. 73(1): 61-65.
Hiltunen, L.H., T. Ojanpera, H. Kortemaa, E. Richter, M.J. Lehtonen, and J.P.T. Valkonen. 2009. Interactions and biocontrol of pathogenic Streptomyces strains co-occurring in potato scab lesions. J Appl Microbiol. 106(1): 199-212.
Hirsch, P.R., L.M. Gilliam, S.P. Sohi, J.K. Williams, I.M. Clark, and P.J. Murray. 2009. Starving the soil of plant inputs for 50 years reduces abundance but not diversity of soil bacterial communities. Soil Biol Biochem. 41(9): 2021-2024.
146
Hjort, K., A. Lembke, A. Speksnijder, K. Smalla, and J.K. Jansson. 2007. Community structure of actively growing bacterial populations in plant pathogen suppressive soil. Microb Ecol. 53(3): 399-413.
Hodge, A., E. Paterson, S.J. Grayston, C.D. Campbell, B.G. Ord, and K. Killham. 1998. Characterisation and microbial utilisation of exudate material from the rhizosphere of Lolium perenne grown under CO2 enrichment. Soil Biol Biochem. 30(8-9): 1033-1043.
Hopwood, D.A. 2006. Soil to genomics: The Streptomyces chromosome. Annu Rev Genet. 40: 1-23.
Horinouchi, S. 2002. A microbial hormone, A-factor, as a master switch for morphological differentiation and secondary metabolism in Streptomyces griseus. Frontier Biosci. 7: D2045-D2057.
Horinouchi, S. 2007. Mining and polishing of the treasure trove in the bacterial genus Streptomyces. Biosci Biotech Biochem. 71(2): 283-299.
Horinouchi, S., and T. Beppu. 1994. A-factor as a microbial hormone that controls cellular differentiation and secondary metabolism in Streptomyces griseus. Mol Microbiol. 12(6): 859-864.
Horinouchi, S., and T. Beppu. 2007. Hormonal control by A-factor of morphological development and secondary metabolism in Streptomyces. Proc J Acad B - Phys Biol Sci. 83(9-10): 277-295.
Hsiao, N.-H., S. Nakayama, M.E. Merlo, M. de Vries, R. Bunet, S. Kitani, T. Nihira, and E. Takano. 2009. Analysis of two additional signaling molecules in Streptomyces coelicolor and the development of a butyrolactone-specific reporter system. Chem Biol. 16(9): 951-960.
Huang, J., J. Shi, V. Molle, B. Sohlberg, D. Weaver, M.J. Bibb, N. Karoonuthaisiri, C.J. Lih, C.M. Kao, M.J. Buttner, and S.N. Cohen. 2005. Cross-regulation among disparate antibiotic biosynthetic pathways of Streptomyces coelicolor. Mol Microbiol. 58(5): 1276-1287.
Huh, J.H., D.J. Kim, X.Q. Zhao, M. Li, Y.Y. Jo, T.M. Yoon, S.K. Shin, J.H. Yong, Y.W. Ryu, Y.Y. Yang, and J.W. Suh. 2004. Widespread activation of antibiotic blosynthesis by S-adenosylmethionine in streptomycetes. FEMS Microbiol Letter. 238(2): 439-447.
Huse, S.M., J.A. Huber, H.G. Morrison, M.L. Sogin, and D.M. Welch. 2007. Accuracy and quality of massively parallel DNA pyrosequencing. Genome Biol. 8(7): R143.
147
Inderjit, and W.H. Van der Putten. 2010. Impacts of soil microbial communities on exotic plant invasions. Trend Ecol Evol. 25(9): 512-519.
Innes, L., P.J. Hobbs, and R.D. Bardgett. 2004. The impacts of individual plant species on rhizosphere microbial communities in soils of different fertility. Biol Fert Soil. 40(1): 7-13.
Johnson, D., M. Ijdo, D.R. Genney, I.C. Anderson, and I.J. Alexander. 2005. How do plants regulate the function, community structure, and diversity of mycorrhizal fungi? J Exp Bot. 56(417): 1751-1760.
Johnson, J.F., D.L. Allan, C.P. Vance, and G. Weiblen. 1995. Effects of phosphorus stress on carbon metabolism and root exudation in Lupinus albus. Plant Phys. 108(2): 64-64.
Jones, C.R., and D.A. Samac. 1996. Biological control of fungi causing alfalfa seedling damping-off with a disease-suppressive strain of Streptomyces. Biol Cont. 7(2): 196-204.
Joseph, S.J., P. Hugenholtz, P. Sangwan, C.A. Osborne, and P.H. Janssen. 2003. Laboratory cultivation of widespread and previously uncultured soil bacteria. Appl Environ Microbiol. 9(12): 7210-7215.
Kamilova, F., L.V. Kravchenko, A.I. Shaposhnikov, T. Azarova, N. Makarova, and B. Lugtenberg. 2006a. Organic acids, sugars, and L-tryptophane in exudates of vegetables growing on stonewool and their effects on activities of rhizosphere bacteria. Mol Plant-Microbe Interact. 19(3): 250-256.
Kamilova, F., L.V. Kravchenko, A.I. Shaposhnikov, N. Makarova, and B. Lugtenberg. 2006b. Effects of the tomato pathogen Fusarium oxysporum f. sp radicis-lycopersici and of the biocontrol bacterium Pseudomonas fluorescens WCS365 on the composition of organic acids and sugars in tomato root exudate. Mol Plant-Microbe Interact. 19(10): 1121-1126.
Kardol, P., T.M. Bezemer, and W.H. van der Putten. 2006. Temporal variation in plant–soil feedback controls succession. Ecol Letter. 9: 1080-1088.
Kataoka, M., K. Ueda, T. Kudo, T. Seki, and T. Yoshida. 1997. Application of the variable region in 16S rDNA to create an index for rapid species identification in the genus Streptomyces. FEMS Microbiol Letter. 151(2): 249-255.
Kawaguchi, T., M. Azuma, S. Horinouchi, and T. Beppu. 1988. Effect of B-factor and its analogs on rifamycin biosynthesis in Nocardia sp. J Antibiot. 41(3): 360-365.
Kemmitt, S.J., C.V. Lanyon, I.S. Waite, Q. Wen, T.M. Addiscott, N.R.A. Bird, A.G. O’Donnell, and P.C. Brookes. 2008. Mineralization of native soil organic matter
148
is not regulated by the size, activity or composition of the soil microbial biomass - A new perspective. Soil Biol Biochem. 40: 61-73.
Kielak, A., A.S. Pijl, J.A. van Veen, and G.A. Kowalchuk. 2008. Differences in vegetation composition and plant species identity lead to only minor changes in soil-borne microbial communities in a former arable field. FEMS Microbiol Ecol. 63(3): 372-382.
Kieser, T., M.B. J, M.J. Buttner, K.F. Chater, and D.A. Hopwood. 2000. Practical Streptomyces Genetics. The John Innes Foundation, Norwich.
Kinkel, L., M. Bakker, and D. Schlatter. 2011. A coevolutionary framework for managing disease-suppressive soils. Annu Rev Phytopathol. 49: in press.
Klironomos, J.N. 2002. Feedback with soil biota contributes to plant rarity and invasiveness in communities. Nature. 417(6884): 67-70.
Knee, E.M., F.C. Gong, M.S. Gao, M. Teplitski, A.R. Jones, A. Foxworthy, A.J. Mort, and W.D. Bauer. 2001. Root mucilage from pea and its utilization by rhizosphere bacteria as a sole carbon source. Mol Plant-Microbe Interact. 14(6): 775-784.
Kortemaa, H., H. Rita, K. Haahtela, and A. Smolander. 1994. Root-colonization ability of antagonistic Streptomyces griseoviridis. Plant Soil. 163(1): 77-83.
Kowalchuk, G.A., D.S. Buma, W. de Boer, P.G.L. Klinkhamer, and J.A. van Veen. 2002. Effects of above-ground plant species composition and diversity on the diversity of soil-borne microorganisms. Ant van Leeuw Int J Gen Mol Microbiol. 81(1-4): 509-520.
Kulmatiski, A., K.H. Beard, J.R. Stevens, and S.M. Cobbold. 2008. Plant-soil feedbacks: A meta-analytical review. Ecology Letters. 11(9): 980-992.
Kunin, V., A. Engelbrektson, H. Ochman, and P. Hugenholtz. 2010. Wrinkles in the rare biosphere: Pyrosequencing errors can lead to artificial inflation of diversity estimates. Environ Microbiol. 12(1): 118-123.
Kuske, C.R., L.O. Ticknor, M.E. Miller, J.M. Dunbar, J.A. Davis, S.M. Barns, and J. Belnap. 2002. Comparison of soil bacterial communities in rhizospheres of three plant species and the interspaces in an arid grassland. Appl Environ Microbiol. 68(4): 1854-1863.
Kuster, E., and S.T. Williams. 1964. Selection of media for isolation of Streptomyces. Nature. 202: 928-929.
149
De La Fuente, L., B.B. Landa, and D.M. Weller. 2006. Host crop affects rhizosphere colonization and competitiveness of 2,4-diacetylphloroglucinol-producing Pseudomonas fluorescens. Phytopathology. 96(7): 751-762.
Langlois, P., S. Bourassa, G.G. Poirier, and C. Beaulieu. 2003. Identification of Streptomyces coelicolor proteins that are differentially expressed in the presence of plant material. Appl Environ Microbiol. 69(4): 1884-1889.
Larkin, M.A., G. Blackshields, N.P. Brown, R. Chenna, P.A. McGettigan, H. McWilliam, F. Valentin, I.M. Wallace, A. Wilm, R. Lopez, J.D. Thompson, T.J. Gibson, and D.G. Higgins. 2007. Clustal W and Clustal X version 2.0. Bioinformatics. 23(21): 2947-2948.
Laskaris, P., S. Tolba, L. Calvo-Bado, and L. Wellington. 2010. Coevolution of antibiotic production and counter-resistance in soil bacteria. Environ Microbiol. 12(3): 783-796.
Lauber, C.L., M. Hamady, R. Knight, and N. Fierer. 2009. Pyrosequencing-based assessment of soil pH as a predictor of soil bacterial community structure at the continental scale. Appl Environ Microbiol. 75(15): 5111-5120.
Lehr, N.A., S.D. Schrey, R. Bauer, R. Hampp, and M.T. Tarkka. 2007. Suppression of plant defence response by a mycorrhiza helper bacterium. New Phytol. 174(4): 892-903.
Lehr, N.A., S.D. Schrey, R. Hampp, and M.T. Tarkka. 2008. Root inoculation with a forest soil streptomycete leads to locally and systemically increased resistance against phytopathogens in Norway spruce. New Phytologist. 177(4): 965-976.
Lemanceau, P., T. Corberand, L. Gardan, X. Latour, G. Laguerre, J.M. Boeufgras, and C. Alabouvette. 1995. Effect of two plant species, flax (Linum usitatissinum L) and tomato (Lycopersicon esculentum Mill), on the diversity of soilborne populations of fluorescent pseudomonads. Appl Environ Microbiol. 61(3): 1004-1012.
Lian, W., K.P. Jayapal, S. Charaniya, S. Mehra, F. Glod, Y.-S. Kyung, D.H. Sherman, and W.-S. Hu. 2008. Genome-wide transcriptome analysis reveals that a pleiotropic antibiotic regulator, AfsS, modulates nutritional stress response in Streptomyces coelicolor A3(2). BMC Genom. 9: 56.
Liu, D., N.A. Anderson, and L.L. Kinkel. 1995. Biological control of potato scab in the field with antagonistic Streptomyces scabies. Phytopathology. 85(7): 827-831.
Llloyd, A.B. 1969. Dispersal of streptomycetes in air. J Gen Microbiol. 57: 35-40.
Lloyd, A.B. 1969. Behavior of streptomycetes in soil. J Gen Microbiol. 56: 165-170.
150
Loranger-Merciris, G., L. Barthes, A. Gastine, and P. Leadley. 2006. Rapid effects of plant species diversity and identity on soil microbial communities in experimental grassland ecosystems. Soil Biol Biochem. 38(8): 2336-2343.
Lyon, G.J., and R.P. Novick. 2004. Peptide signaling in Staphylococcus aureus and other Gram-positive bacteria. Peptides. 25(9): 1389-1403.
Mark, G.L., J.M. Dow, P.D. Kiely, H. Higgins, J. Haynes, C. Baysse, A. Abbas, T. Foley, A. Franks, J. Morrissey, and F. O’Gara. 2005. Transcriptome profiling of bacterial responses to root exudates identifies genes involved in microbe-plant interactions. PNAS. 102(48): 17454-17459.
Marschner, P., D. Crowley, and C.H. Yang. 2004. Development of specific rhizosphere bacterial communities in relation to plant species, nutrition and soil type. Plant Soil. 261(1-2): 199-208.
Marschner, P., C.H. Yang, R. Lieberei, and D.E. Crowley. 2001. Soil and plant specific effects on bacterial community composition in the rhizosphere. Soil Biol Biochem. 33(11): 1437-1445.
Marshall, K., and M. Alexander. 1960. Competition between soil bacteria and Fusarium. Plant Soil. 12: 143-153.
Mathesius, U., S. Mulders, M.S. Gao, M. Teplitski, G. Caetano-Anolles, B.G. Rolfe, and W.D. Bauer. 2003. Extensive and specific responses of a eukaryote to bacterial quorum-sensing signals. PNAS. 100(3): 1444-1449.
Matsukawa, E., Y. Nakagawa, Y. Iimura, and M. Hayakawa. 2007a. A new enrichment method for the selective isolation of streptomycetes from the root surfaces of herbaceous plants. Actinomycetologica. 21: 66-69.
Matsukawa, E., Y. Nakagawa, Y. Iimura, and M. Hayakawa. 2007b. Stimulatory effect of indole-3-acetic acid on aerial mycelium formation and antibiotic production in Streptomyces spp. Actinomycetologica. 21(1): 32-39.
Mayfield, C.I., S.T. Williams, S.M. Ruddick, and H.L. Hatfield. 1972. Studies on the ecology of actinomycetes in soil: IV. Observations on the form and growth of streptomycetes in soil. Soil Biol Biochem. 4: 79-91.
Mazzola, M. 2004. Assessment and management of soil microbial community structure for disease suppression. Annu Rev Phytopathol. 42: 35-59.
Mazzola, M., D.L. Funnell, and J.M. Raaijmakers. 2004. Wheat cultivar-specific selection of 2,4-diacetylphloroglucinol-producing fluorescent Pseudomonas species from resident soil populations. Microb Ecol. 48(3): 338-348.
151
McCarthy-Neumann, S., and R.K. Kobe. 2010a. Conspecific and heterospecific plant-soil feedbacks influence survivorship and growth of temperate tree seedlings. J Ecol. 98(2): 408-418.
McCarthy-Neumann, S., and R.K. Kobe. 2010b. Conspecific plant-soil feedbacks reduce survivorship and growth of tropical tree seedlings. J Ecol. 98(2): 396-407.
McVeigh, H.P., J. Munro, and T.M. Embley. 1996. Molecular evidence for the presence of novel actinomycete lineages in a temperate forest soil. J Indust Microbiol. 17(3-4): 197-204.
Meharg, A.A., and K. Killham. 1991. A novel method of quantifying root exudation in the presence of soil microflora. Plant Soil. 133(1): 111-116.
Meharg, A.A., and K. Killham. 1995. Loss of exudates from the roots of perennial ryegrass inoculated with a range of microorganisms. Plant Soil. 170(2): 345-349.
Miethling, R., G. Wieland, H. Backhaus, and C.C. Tebbe. 2000. Variation of microbial rhizosphere communities in response to crop species, soil origin, and inoculation with Sinorhizobium meliloti L33. Microb Ecol. 40(1): 43-56.
Mills, K.E., and J.D. Bever. 1998. Maintenance of diversity within plant communities: Soil pathogens as agents of negative feedback. Ecology. 79(5): 1595-1601.
Nagahashi, G., and D.D. Douds. 2000. Partial separation of root exudate components and their effects upon the growth of germinated spores of AM fungi. Mycol Res. 104: 1453-1464.
Nair, M.G., G.R. Safir, and J.O. Siqueira. 1991. Isolation and identification of vesicular-arbuscular mycorrhiza-stimulatory compounds from clover (Trifolium repens) roots. Appl Environ Microbiol. 57(2): 434-439.
Needleman, S., and C. Wunsch. 1970. A general method applicable to the search for similarities in the amino acid sequence of two proteins. J Mol Biol. 48(3): 443-453.
Neefs, J.M., Y. Vandepeer, L. Hendriks, and R. Dewachter. 1990. Compilation of small ribosomal-subunit RNA sequences. Nuc Acid Res. 18: 2237-2317.
Nemergut, D.R., A.R. Townsend, S.R. Sattin, K.R. Freeman, N. Fierer, J.C. Neff, W.D. Bowman, C.W. Schadt, M.N. Weintraub, and S.K. Schmidt. 2008. The effects of chronic nitrogen fertilization on alpine tundra soil microbial communities: Implications for carbon and nitrogen cycling. Environ Microbiol. 10(11): 3093-3105.
152
Nichols, D., K. Lewis, J. Orjala, S. Mo, R. Ortenberg, P. O’Connor, C. Zhao, P. Vouros, T. Kaeberlein, and S.S. Epstein. 2008. Short peptide induces an “uncultivable” microorganism to grow in vitro. Appl Environ Microbiol. 74(15): 4889-4897.
Nihira, T., Y. Shimizu, H.S. Kim, and Y. Yamada. 1988. Structure-activity relationships of virginiae butanolide-C, an inducer of virginiamycin production in Streptomyces virginiae. J Antibiot. 41(12): 1828-1837.
Nishida, H., Y. Ohnishi, T. Beppu, and S. Horinouchi. 2007. Evolution of gamma-butyrolactone synthases and receptors in Streptomyces. Environ Microbiol. 9(8): 1986-1994.
Nunes da Rocha, U., L. Van Overbeek, and J.D. Van Elsas. 2009. Exploration of hitherto-uncultured bacteria from the rhizosphere. FEMS Microbiol Ecol. 69(3): 313-328.
O’Callaghan, K.J., P.J. Stone, X.J. Hu, D.W. Griffiths, M.R. Davey, and E.C. Cocking. 2000. Effects of glucosinolates and flavonoids on colonization of the roots of Brassica napus by Azorhizobium caulinodans ORS571. Appl Environ Microbiol. 66(5): 2185-2191.
O’Rourke, S., A. Wietzorrek, K. Fowler, C. Corre, G.L. Challis, and K.F. Chater. 2009. Extracellular signalling, translational control, two repressors and an activator all contribute to the regulation of methylenomycin production in Streptomyces coelicolor. Mol Microbiol. 71(3): 763-778.
Oger, P., H. Mansouri, X. Nesme, and Y. Dessaux. 2004. Engineering root exudation of lotus toward the production of two novel carbon compounds leads to the selection of distinct microbial populations in the rhizosphere. Microb Ecol. 47(1): 96-103.
Ohashi, H., Y.H. Zheng, T. Nihira, and Y. Yamada. 1989. Distribution of virginiae butanolides in antibiotic-producing actinomycetes, and identification of the inducing factor from Streptomyces antibioticus as virginiae butanolide-A. J Antibiot. 42(7): 1191-1195.
Ohnishi, Y., H. Yamazaki, J. Kato, A. Tomono, and S. Horinouchi. 2005. AdpA, a central transcriptional regulator in the A-factor regulatory cascade that leads to morphological development and secondary metabolism in Streptomyces griseus. Biosci Biotech Biochem. 69(3): 431-439.
Oksanen, J., F.G. Blanchet, R. Kindt, P. Legendre, R.B. O’Hara, G.L. Simpson, P. Solymos, M.H.H. Stevens, and H. Wagner. 2010. Vegan: Community ecology package. R package version 1.17-4.
153
Okubara, P., and R. Bonsall. 2008. Accumulation of Pseudomonas-derived 2,4-diacetylphloroglucinol on wheat seedling roots is influenced by host cultivar. Biol Cont. 46(3): 322-331.
Olff, H., B. Hoorens, R.G.M. de Goede, W.H. van der Putten, and J.M. Gleichman. 2000. Small-scale shifting mosaics of two dominant grassland species: The possible role of soil-borne pathogens. Oecologia. 125(1): 45-54.
Omura, S., H. Ikeda, J. Ishikawa, A. Hanamoto, C. Takahashi, M. Shinose, Y. Takahashi, H. Horikawa, H. Nakazawa, T. Osonoe, H. Kikuchi, T. Shiba, Y. Sakaki, and M. Hattori. 2001. Genome sequence of an industrial microorganism Streptomyces avermitilis: Deducing the ability of producing secondary metabolites. PNAS. 98(21): 12215-12220.
Ortiz-Castro, R., C. Diaz-Perez, M. Martinez-Trujillo, R.E. del Rio, J. Campos-Garcia, and J. Lopez-Bucio. 2011. Transkingdom signaling based on bacterial cyclodipeptides with auxin activity in plants. PNAS. 108(17): 7253-7258.
Orwin, K.H., D.A. Wardle, and L.G. Greenfield. 2006. Ecological consequences of carbon substrate identity and diversity in a laboratory study. Ecology. 87(3): 580-593.
Otto, M., H. Echner, W. Voelter, and F. Gotz. 2001. Pheromone cross-inhibition between Staphylococcus aureus and Staphylococcus epidermidis. Infect Immun. 69(3): 1957-1960.
Otto, M., R. Sussmuth, C. Vuong, G. Jung, and F. Gotz. 1999. Inhibition of virulence factor expression in Staphylococcus aureus by the Staphylococcus epidermidis agr pheromone and derivatives. FEBS Letter. 450(3): 257-262.
Palacios, C., E. Zettler, R. Amils, and L. Amaral-Zettler. 2008. Contrasting microbial community assembly hypotheses: A reconciling tale from the Río Tinto. PLoS ONE. 3(12): e3853.
Parameswaran, P., R. Jalili, L. Tao, S. Shokralla, B. Gharizadeh, M. Ronaghi, and A.Z. Fire. 2007. A pyrosequencing-tailored nucleotide barcode design unveils opportunities for large-scale sample multiplexing. Nuc Acid Res. 35(19): e130.
Pavlovic, N.B., and R. Grundel. 2009. Reintroduction of wild lupine Lupinus perennis depends on variation in canopy, vegetation, and litter cover. Rest Ecol. 17(6): 807-817.
Perez, C., R. Dill-Macky, and L.L. Kinkel. 2008. Management of soil microbial communities to enhance populations of Fusarium graminearum antagonists in soil. Plant Soil. 302(1-2): 53-69.
154
Petermann, J.S., A.J.F. Fergus, L.A. Turnbull, and B. Schmid. 2008. Janzen-Connell effects are widespread and strong enough to maintain diversity in grasslands. Ecology. 89(9): 2399-2406.
Phillips, D.A., T.C. Fox, M.D. King, T.V. Bhuvaneswari, and L.R. Teuber. 2004. Microbial products trigger amino acid exudation from plant roots. Plant Phys. 136(1): 2887-2894.
Picard, C., and M. Bosco. 2005. Maize heterosis affects the structure and dynamics of indigenous rhizospheric auxins-producing Pseudomonas populations. FEMS Microbiol Ecol. 53(3): 349-357.
Prithiviraj, B., M. Paschke, and J. Vivanco. 2007. Root communication: The role of root exudates. Encyclo Plant Crop Sci. 1: 1-4.
Prosser, J.I. 2010. Replicate or lie. Environ Microbiol. 12(7): 1806-1810.
Pruesse, E., C. Quast, K. Knittel, B.M. Fuchs, W. Ludwig, J. Peplies, and F.O. Glockner. 2007. SILVA: A comprehensive online resource for quality checked and aligned ribosomal RNA sequence data compatible with ARB. Nuc Acid Res. 35(21): 7188-7196.
Qin, J., R. Li, J. Raes, M. Arumugam, K.S. Burgdorf, et al. 2010. A human gut microbial gene catalogue established by metagenomic sequencing. Nature. 464(7285): 59-65.
Quince, C., A. Lanzén, T.P. Curtis, R.J. Davenport, N. Hall, I.M. Head, L.F. Read, and W.T. Sloan. 2009. Accurate determination of microbial diversity from 454 pyrosequencing data. Nat Meth. 6(9): 639-641.
Quinlan, A.R., D.A. Stewart, M.P. Strömberg, and G.T. Marth. 2008. Pyrobayes: An improved base caller for SNP discovery in pyrosequences. Nat Meth. 5(2): 179-181.
R Development Core Team. 2011. R: A language and environment for statistical computing. R Foundation for Statistical Computing. Vienna, Austria.
Raaijmakers, J.M., and D.M. Weller. 1998. Natural plant protection by 2,4-diacetylphloroglucinol-producing Pseudomonas spp. in take-all decline soils. Mol Plant-Microbe Interact. 11(2): 144-152.
Rainey, P.B. 1999. Adaptation of Pseudomonas fluorescens to the plant rhizosphere. Environ Microbiol. 1(3): 243-257.
Ramos-Gonzalez, M.I., M.J. Campos, and J.L. Ramos. 2005. Analysis of Pseudomonas putida KT2440 gene expression in the maize rhizosphere: In vitro expression
155
technology capture and identification of root-activated promoters. J Bacteriol. 187(12): 4033-4041.
Ravel, J., P. Gajer, Z. Abdo, G.M. Schneider, S.S.K. Koenig, S.L. McCulle, S. Karlebach, R. Gorle, J. Russell, C.O. Tacket, R.M. Brotman, C.C. Davis, K. Ault, L. Peralta, and L.J. Forney. 2010. Colloquium Paper: Vaginal microbiome of reproductive-age women. PNAS. 108(S1): 4680-4687.
Recio, E., A. Colinas, A. Rumbero, J.F. Aparicio, and J.F. Martin. 2004. PI factor, a novel type quorum-sensing inducer elicits pimaricin production in Streptomyces natalensis. J Biol Chem. 279(40): 41586-41593.
Rengel, Z. 2002. Breeding for better symbiosis. Plant Soil. 245: 147-162.
Reynolds, H.L., A. Packer, J.D. Bever, and K. Clay. 2003. Grassroots ecology: Plant-microbe-soil interactions as drivers of plant community structure and dynamics. Ecology. 84(9): 2281-2291.
Rintala, H., A. Nevalainen, E. Ronka, and M. Suutari. 2001. PCR primers targeting the 16S rRNA gene for the specific detection of streptomycetes. Mol Cell Probes. 15(6): 337-347.
Roberts, D.W. 2010. Labdsv: Ordination and multivariate analysis for ecology. R package.
Ruddick, S.M., and S.T. Williams. 1972. Studies on the ecology of actinomycetes in soil: V. Some factors influencing the dispersal and adsorption of spores in soil. Soil Biol Biochem. 4: 93-103.
Rudrappa, T., J. Bonsall, J. Gallagher, D. Seliskar, and H. Bais. 2007. Root-secreted allelochemical in the noxious weed Phragmites australis deploys a reactive oxygen species response and microtubule assembly disruption to execute rhizotoxicity. J Chem Ecol. 33(10): 1898-1918.
Ryan, A.D., L.L. Kinkel, and J.L. Schottel. 2004. Effect of pathogen isolate, potato cultivar, and antagonist strain on potato scab severity and biological control. Biocont Sci Tech. 14(3): 301-311.
Ryan, P.R., Y. Dessaux, L.S. Thomashow, and D.M. Weller. 2009. Rhizosphere engineering and management for sustainable agriculture. Plant Soil. 321(1-2): 363-383.
Samac, D.A., and L.L. Kinkel. 2001. Suppression of the root-lesion nematode (Pratylenchus penetrans) in alfalfa (Medicago sativa) by Streptomyces spp. Plant Soil. 235(1): 35-44.
156
Samac, D.A., A.M. Willert, M.J. McBride, and L.L. Kinkel. 2003. Effect of antibiotic-producing Streptomyces on nodulation and leaf spot in alfalfa. Appl Soil Ecol. 22(1): 55-66.
Sarathchandra, S.U., A. Ghani, G.W. Yeates, G. Burch, and N.R. Cox. 2001. Effect of nitrogen and phosphate fertilisers on microbial and nematode diversity in pasture soils. Soil Biol Biochem. 33(7-8): 953-964.
Sato, K., T. Nihira, S. Sakuda, M. Yanagimoto, and Y. Yamada. 1989. Isolation and structure of a new butyrolactone autoregulator from Streptomyces sp. Fri-5. J Ferment Bioeng. 68(3): 170-173.
Scervino, J.M., M.A. Ponce, R. Erra-Bassells, H. Vierheilig, J.A. Ocampo, and A. Godeas. 2005a. Arbuscular mycorrhizal colonization of tomato by Gigaspora and Glomus species in the presence of root flavonoids. J Plant Phys. 162(6): 625-633.
Scervino, J.M., M.A. Ponce, R. Erra-Bassells, H. Vierheilig, J.A. Ocampo, and A. Godeas. 2005b. Flavonoids exhibit fungal species and genus specific effects on the presymbiotic growth of Gigaspora and Glomus. Mycol Res. 109: 789-794.
Schlatter, D.C., A. Fubuh, K. Xiao, D. Hernandez, S. Hobbie, and L.L. Kinkel. 2008. Influence of carbon source amendments on population density, resource use, and antibiotic phenotypes of soilborne Streptomyces. Phytopathology. 98(6): S140-S141.
Schlatter, D.C., D.A. Samac, M. Tesfaye, and L.L. Kinkel. 2010. Rapid and specific method for evaluating Streptomyces competitive dynamics in complex soil communities. Appl Environ Microbiol. 76(6): 2009-2012.
Schloss, P.D., S.L. Westcott, T. Ryabin, J.R. Hall, M. Hartmann, E.B. Hollister, R.A. Lesniewski, B.B. Oakley, D.H. Parks, C.J. Robinson, J.W. Sahl, B. Stres, G.G. Thallinger, D.J.V. Horn, and C.F. Weber. 2009. Introducing mothur: Open-source, platform-independent, community-supported software for describing and comparing microbial communities. Appl Environ Microbiol. 75(23): 7537-7541.
Schreiner, K., A. Hagn, M. Kyselkova, Y. Moenne-Loccoz, G. Welzl, J.C. Munch, and M. Schloter. 2010. Comparison of barley succession and Take-All disease as environmental factors shaping the rhizobacterial community during Take-All Decline. Appl Environ Microbiol. 76(14): 4703-4712.
Schrey, S.D., V. Salo, M. Raudaskoski, R. Hampp, U. Nehls, and M.T. Tarkka. 2007. Interaction with mycorrhiza helper bacterium Streptomyces sp AcH 505 modifies organisation of actin cytoskeleton in the ectomycorrhizal fungus Amanita muscaria (fly agaric). Curr Genet. 52(2): 77-85.
157
Schroeckh, V., K. Scherlach, H.-W. Nutzmann, E. Shelest, W. Schmidt-Heck, J. Schuemann, K. Martin, C. Hertweck, and A.A. Brakhage. 2009. Intimate bacterial-fungal interaction triggers biosynthesis of archetypal polyketides in Aspergillus nidulans. PNAS. 106(34): 14558-14563.
Schweitzer, J.A., J.K. Bailey, D.G. Fischer, C.J. LeRoy, E.V. Lonsdorf, T.G. Whitham, and S.C. Hart. 2008. Plant-soil-microorganism interactions: Heritable relationship between plant genotype and associated soil microorganisms. Ecology. 89(3): 773-781.
Shank, E.A., and R. Kolter. 2009. New developments in microbial interspecies signaling. Curr Op Microbiol. 12(2): 205-214.
Shaw, L.J., P. Morris, and J.E. Hooker. 2006. Perception and modification of plant flavonoid signals by rhizosphere microorganisms. Environ Microbiol. 8(11): 1867-1880.
Shennan, C. 2008. Biotic interactions, ecological knowledge and agriculture. Phil Trans R Soc B - Biol Sci. 363(1492): 717-739.
Shen, H., X.C. Wang, W.M. Shi, Z.H. Cao, and X.L. Yan. 2001. Isolation and identification of specific root exudates in elephantgrass in response to mobilization of iron- and aluminum-phosphates. J Plant Nut. 24(7): 1117-1130.
Siciliano, S.D., C.M. Theoret, J.R. de Freitas, P.J. Hucl, and J.J. Germida. 1998. Differences in the microbial communities associated with the roots of different cultivars of canola and wheat. Can J Microbiol. 44(9): 844-851.
Slattery, M., I. Rajbhandari, and K. Wesson. 2001. Competition-mediated antibiotic induction in the marine bacterium Streptomyces tenjimariensis. Microb Ecol. 41(2): 90-96.
Smalla, K., G. Wieland, A. Buchner, A. Zock, J. Parzy, S. Kaiser, N. Roskot, H. Heuer, and G. Berg. 2001. Bulk and rhizosphere soil bacterial communities studied by denaturing gradient gel electrophoresis: Plant-dependent enrichment and seasonal shifts revealed. Appl Environ Microbiol. 67(10): 4742-4751.
Smith, K.P., J. Handelsman, and R.M. Goodman. 1999. Genetic basis in plants for interactions with disease-suppressive bacteria. PNAS. 96(9): 4786-4790.
Stackebrandt, E., D. Witt, C. Kemmerling, R. Kroppenstedt, and W. Liesack. 1991. Designation of streptomycete 16S and 23S ribosomal-RNA-based target regions for oligonucleotide probes. Appl Environ Microbiol. 57(5): 1468-1477.
Starkey, R.L. 1958. Interrelationships between microorganisms and plant roots in the rhizosphere. Bacteriol Rev. 22(3): 154-172.
158
Steinkellner, S., V. Lendzemo, I. Langer, P. Schweiger, T. Khaosaad, J.-P. Toussaint, and H. Vierheilig. 2007. Flavonoids and strigolactones in root exudates as signals in symbiotic and pathogenic plant-fungus interactions. Molecules. 12(7): 1290-1306.
Stephan, A., A.H. Meyer, and B. Schmid. 2000. Plant diversity affects culturable soil bacteria in experimental grassland communities. J Ecol. 88(6): 988-998.
Strassmann, J., Y. Zhu, and D. Queller. 2000. Altruism and social cheating in the social amoeba Dictyostelium discoideum. Nature. 408: 965-967.
Sun, Y., Y. Cai, L. Liu, F. Yu, M.L. Farrell, W. McKendree, and W. Farmerie. 2009. ESPRIT: Estimating species richness using large collections of 16S rRNA pyrosequences. Nuc Acid Res. 37(10): e76-e76.
Takano, E. 2006. Gamma-butyrolactones: Streptomyces signalling molecules regulating antibiotic production and differentiation. Curr Op Microbiol. 9(3): 287-294.
Takano, E., R. Chakraburtty, T. Nihira, Y. Yamada, and M.J. Bibb. 2001. A complex role for the gamma-butyrolactone SCB1 in regulating antibiotic production in Streptomyces coelicolor A3(2). Mol Microbiol. 41(5): 1015-1028.
Takano, E., H. Kinoshita, V. Mersinias, G. Bucca, G. Hotchkiss, T. Nihira, C.P. Smith, M. Bibb, W. Wohlleben, and K. Chater. 2005. A bacterial hormone (the SCB1) directly controls the expression of a pathway-specific regulatory gene in the cryptic type I polyketide biosynthetic gene cluster of Streptomyces coelicolor. Mol Microbiol. 56(2): 465-479.
Tapio, E., and A. Pohtolahdenpera. 1991. Scanning electron microscopy of hyphal interaction between Streptomyces griseoviridis and some plant pathogenic fungi. J Agric Sci Finland. 63(5): 435-441.
Teplitski, M., J.B. Robinson, and W.D. Bauer. 2000. Plants secrete substances that mimic bacterial N-acyl homoserine lactone signal activities and affect population density-dependent behaviors in associated bacteria. Mol Plant-Microbe Interact. 13(6): 637-648.
Thompson, J.N. 1999. The evolution of species interactions. Science. 284(5423): 2116-2118.
Tilman, D., J. Knops, D. Wedlin, P. Reich, M. Ritchie, and E. Siemann. 1997. The Influence of Functional Diversity and Composition on Ecosystem Processes. Science. 277(5330): 1300-1302.
Tilman, D., P.B. Reich, J. Knops, D. Wedin, T. Mielke, and C. Lehman. 2001. Diversity and productivity in a long-term grassland experiment. Science. 294(5543): 843-845.
159
Tokala, R.K., J.L. Strap, C.M. Jung, D.L. Crawford, M.H. Salove, L.A. Deobald, J.F. Bailey, and M.J. Morra. 2002. Novel plant-microbe rhizosphere interaction involving Streptomyces lydicus WYEC108 and the pea plant (Pisum sativum). Appl Environ Microbiol. 68(5): 2161-2171.
Tonsor, S. 1989. Relatedness and intraspecific competition in Plantago lanceolata. Am Nat. 134(6): 897-906.
Tuomi, T., S. Laakso, and H. Rosenqvist. 1994. Indole-3-acetic acid (IAA) production by a biofungicide Streptomyces griseoviridis strain. Ann Bot Fennici. 31(1): 59-63.
Ueda, K., S. Kawai, H. Ogawa, A. Kiyama, T. Kubota, H. Kawanobe, and T. Beppu. 2000. Wide distribution of interspecific stimulatory events on antibiotic production and sporulation among Streptomyces species. J Antibiot. 53(9): 979-982.
Ulrich, A., and R. Becker. 2006. Soil parent material is a key determinant of the bacterial community structure in arable soils. FEMS Microbiol Ecol. 56(3): 430-443.
Uroz, S., P.M. Oger, E. Chapelle, M.-T. Adeline, D. Faure, and Y. Dessaux. 2008. A Rhodococcus qsdA-encoded enzyme defines a novel class of large-spectrum quorum-quenching lactonases. Appl Environ Microbiol. 74(5): 1357-1366.
Uroz, S., P. Oger, S.R. Chhabra, M. Camara, P. Williams, and Y. Dessaux. 2007. N-acyl homoserine lactones are degraded via an amidolytic activity in Comamonas sp strain D1. Arch Microbiol. 187(3): 249-256.
Verbruggen, E., and E.T. Kiers. 2010. Evolutionary ecology of mycorrhizal functional diversity in agricultural systems. Evol Appl. 3(5-6): 547-560.
Viketoft, M., C. Palmborg, B. Sohlenius, K. Huss-Danell, and J. Bengtsson. 2005. Plant species effects on soil nematode communities in experimental grasslands. Appl Soil Ecol. 30(2): 90-103.
Vilches, C., C. Mendez, C. Hardisson, and J.A. Salas. 1990. Biosynthesis of oleandomycin by Streptomyces antibioticus - Influence of nutritional conditions and development of resistance. J Gen Microbiol. 136: 1447-1454.
Wakelin, S.A., L.M. Macdonald, S.L. Rogers, A.L. Gregg, T.P. Bolger, and J.A. Baldock. 2008. Habitat selective factors influencing the structural composition and functional capacity of microbial communities in agricultural soils. Soil Biol Biochem. 40(3): 803-813.
Walker, T.S., H.P. Bais, E. Grotewold, and J.M. Vivanco. 2003. Root exudation and rhizosphere biology. Plant Phys. 132(1): 44-51.
160
Wang, Q., G.M. Garrity, J.M. Tiedje, and J.R. Cole. 2007. Naive Bayesian classifier for rapid assignment of rRNA sequences into the new bacterial taxonomy. Appl Environ Microbiol. 73(16): 5261-5267.
Wang, Z.-P., X.-G. Han, and L.-H. Li. 2008. Effects of grassland conversion to croplands on soil organic carbon in the temperate Inner Mongolia. J Environ Manage. 86(3): 529-534.
Wardle, D.A., K.I. Bonner, G.M. Barker, G.W. Yeates, K.S. Nicholson, R.D. Bardgett, R.N. Watson, and A. Ghani. 1999. Plant removals in perennial grassland: Vegetation dynamics, decomposers, soil biodiversity, and ecosystem properties. Ecol Monograph. 69(4): 535-568.
Wardle, D. 1999. Is “sampling effect” a problem for experiments investigating biodiversity-ecosystem function relationships? Oikos. 87(2): 403-407.
Weir, T.L., V.J. Stull, D. Badri, L.A. Trunck, H.P. Schweizer, and J. Vivanco. 2008. Global gene expression profiles suggest an important role for nutrient acquisition in early pathogenesis in a plant model of Pseudomonas aeruginosa infection. Appl Environ Microbiol. 74(18): 5784-5791.
Weisskopf, L., N. Fromin, N. Tomasi, M. Aragno, and E. Martinoia. 2005. Secretion activity of white lupin’s cluster roots influences bacterial abundance, function and community structure. Plant Soil. 268(1-2): 181-194.
Weller, D.M., J.M. Raaijmakers, B.B.M. Gardener, and L.S. Thomashow. 2002. Microbial populations responsible for specific soil suppressiveness to plant pathogens. Annu Rev Phytopathol. 40: 309-348.
de Werra, P., E. Baehler, A. Huser, C. Keel, and M. Maurhofer. 2007. Detection of plant-modulated alterations in antifungal gene expression in Pseudomonas fluorescens CHA0 on roots by flow cytometry. Appl Environ Microbiol. 74(5): 1339-1349.
West, S., E. Kiers, I. Pen, and R. Denison. 2002. Sanctions and mutualism stability: When should less beneficial mutualists be tolerated? J Evolution Biol. 15(5): 830-837.
Wiggins, B.E., and L.L. Kinkel. 2005a. Green manures and crop sequences influence alfalfa root rot and pathogen inhibitory activity among soil-borne streptomycetes. Plant Soil. 268: 271-283.
Wiggins, B.E., and L.L. Kinkel. 2005b. Green manures and crop sequences influence potato diseases and pathogen inhibitory activity of indigenous streptomycetes. Phytopathology. 95(2): 178-185.
161
Williams, S.T., and C.I. Mayfield. 1971. Studies on the ecology of actinomycetes in soil: III. The behaviour of neutrophilic streptomycetes in acid soil. Soil Biol Biochem. 3: 197-208.
Williams, S.T., F.L. Davies, C.I. Mayfield, and M.R. Khan. 1971. Studies on the ecology of actinomycetes in soil: II. The pH requirements of streptomycetes from two acid soils. Soil Biol Biochem. 3: 187-195.
Williams, S.T., S. Lanning, and E.M.H. Wellington. 1984. Ecology of actinomycetes. p. 481-528. In Goodfellow, M., Mordarski, M., Williams, S.T. (eds.), The Biology of Actinomycetes. Academic Press, New York.
Williams, S.T., M. Shameemullah, E.T. Watson, and C.I. Mayfield. 1972. Studies on the ecology of actinomycetes in soil: VI. The influence of moisture tension on growth and survival. Soil Biol Biochem. 4: 215-225.
Wilson, I. 1997. Inhibition and facilitation of nucleic acid amplification. Appl Environ Microbiol. 63(10): 3741-3751.
Wissuwa, M., M. Mazzola, and C. Picard. 2008. Novel approaches in plant breeding for rhizosphere-related traits. Plant Soil. 321(1-2): 409-430.
Xiao, K., L.L. Kinkel, and D.A. Samac. 2002. Biological control of Phytophthora root rots on alfalfa and soybean with Streptomyces. Biol Cont. 23(3): 285-295.
Xu, G., J. Wang, L. Wang, X. Tian, H. Yang, K. Fan, K. Yang, and H. Tan. 2010. “Pseudo”-butyrolactone receptors respond to antibiotic signals to coordinate antibiotic biosynthesis. J Biol Chem. 285(35): 27440-27448.
Yamanaka, K., H. Oikawa, H.O. Ogawa, K. Hosono, F. Shinmachi, H. Takano, S. Sakuda, T. Beppu, and K. Ueda. 2005. Desferrioxamine E produced by Streptomyces griseus stimulates growth and development of Streptomyces tanashiensis. Microbiol-SGM. 151: 2899-2905.
Yang, Y.Y., X.Q. Zhao, Y.Y. Jin, J.H. Huh, J.H. Cheng, D. Singh, H.J. Kwon, and J.W. Suh. 2006. Novel function of cytokinin: A signaling molecule for promotion of antibiotic production in streptomycetes. J Microbiol Biotech. 16(6): 896-900.
Yang, Y.-H., E. Song, E.-J. Kim, K. Lee, W.-S. Kim, S.-S. Park, J.-S. Hahn, and B.-G. Kim. 2009. NdgR, an IclR-like regulator involved in amino-acid-dependent growth, quorum sensing, and antibiotic production in Streptomyces coelicolor. Appl Microbiol Biotechnol. 82(3): 501-511.
Yan, W., R.R.E. Artz, and D. Johnson. 2008. Species-specific effects of plants colonising cutover peatlands on patterns of carbon source utilisation by soil microorganisms. Soil Biol Biochem. 40(2): 544-549.
162
Yao, H., and F. Wu. 2010. Soil microbial community structure in cucumber rhizosphere of different resistance cultivars to Fusarium Wilt. FEMS Microbiol Ecol. 72(3): 456-463.
Yehuda, Z., M. Shenker, Y. Hadar, and Y. Chen. 2000. Remedy of chlorosis induced by iron deficiency in plants with the fungal siderophore rhizoferrin. J Plant Nutr. 23(11): 1991-2006.
Yim, G., H.M.H. Wang, and J. Davies. 2007. Antibiotics as signalling molecules. Phil Trans R Soc B - Biol Sci. 362(1483): 1195-1200.
Yuan, W.M., and D.L. Crawford. 1995. Characterization of Streptomyces lydicus WYEC108 as a potential biocontrol agent against fungal root and seed rots. Appl Environ Microbiol. 61(8): 3119-3128.
Zak, D.R., W.E. Holmes, D.C. White, A.D. Peacock, and D. Tilman. 2003. Plant diversity, soil microbial communities, and ecosystem function: Are there any links? Ecology. 84(8): 2042-2050.
Zaura, E., B.J. Keijser, S.M. Huse, and W. Crielaard. 2009. Defining the healthy “core microbiome” of oral microbial communities. BMC Microbiol. 9(1): 259.
Zhou, J., D. Treves, L. Wu, T. Marsh, R. O’Neill, A. Palumbo, and J. Tiedje. 2002. Spatial and resource factors influencing high microbial diversity in soil. Appl Environ Microbiol. 68(1): 326-334.
Zimpfer, J.F., J.M. Igual, B. McCarty, C. Smyth, and J.O. Dawson. 2004. Casuarina cunninghamiana tissue extracts stimulate the growth of Frankia and differentially alter the growth of other soil microorganisms. J Chem Ecol. 30(2): 439-452.
163
Appendix: Plants as modulators of antibiotic production by streptomycetes
164
Background
The ability of plants to impact microbial community structure is well documented in the
literature (Bardgett and Walker, 2004; Viketoft et al., 2005; Carney and Matson, 2006).
This phenomenon has been attributed to plant-driven selective effects including
differential resource provision (Knee et al., 2001; Shaw et al., 2006; Yan et al., 2008) or
direct inhibitory effects of plant compounds (Broeckling et al., 2008; Badri et al., 2009).
However, plants may also influence associated microbes through other mechanisms such
as chemical signaling that modifies microbial gene expression (Teplitski et al., 2000; Gao
et al., 2003). In particular, we investigated the possibility that plants may produce
chemical compounds capable of modifying antibiotic production among associated
streptomycetes.
We screened a set of compounds produced by plants for effects on antibiotic production
by a collection of streptomycete isolates. These compounds, including flavonoids and
sesquiterpene lactones, are involved in other plant-microbe interactions and are known to
be present in biologically relevant concentrations in the rhizosphere (Scervino et al.,
2005a, 2005b; Besserer et al., 2006). We also tested the plant hormone indole-3-acetic
acid (IAA) because it has been reported previously as enhancing antibiotic production
among streptomycetes (Matsukawa et al., 2007b). As a complementary approach, we also
screened plant extracts for the ability to interact with a specific regulatory mechanism
that governs antibiotic production among streptomycetes. The gamma-butyrolactones
(GBLs) are hormonal signals that regulate antibiotic production among streptomycetes
through specific signal-receptor binding interactions (Takano, 2006). We used a
bioreporter system (Hsiao et al., 2009) to screen a collection of plant extracts for the
presence of GBLs, or other compounds having the potential to interact with GBL receptor
proteins.
Methods
Screening for impacts of plant compounds on streptomycete antibiotic production
165
A collection of 24 antibiotic-producing streptomycete isolates was used to screen plant
compounds for their impacts on antibiotic production. This screening panel included 20
isolates collected from a Kansas prairie soil (see Chapter 2), and four isolates originally
collected from a Minnesota prairie soil (Davelos et al., 2004b). These isolates were
chosen because of a body of background knowledge gained from their use in prior
studies, and for the diversity of inhibitory activities represented across the collection.
Supplemented minimal medium solid (SMMS) was used because streptomycete isolates
tend to produce antibiotics well on this medium (Kieser et al., 2000), it is pH buffered,
and it can be produced consistently across batches.
We tested seven different plant compounds, each at three different concentrations. This
collection consisted of compounds that have been implicated in plant-microbe
interactions in the rhizosphere, and included four flavonoids, two sesquiterpene lactones,
and one plant hormone (Table 1). Plant compounds were dissolved in an appropriate
solvent and added to cooled, molten SMMS to create a final concentration of 20 nM, 1
uM or 50 uM. An equal volume of solvent was added to produce each final
concentration, and solvent-only controls were performed separately. Amended media was
poured into 20 x 20 cm bioassay dishes (250 mL/dish) and streptomycete spore
suspensions were spotted onto the plate. Spore suspensions were adjusted by plate count
to a standard concentration of 1 x 108 CFU/mL, and 4 uL was applied per spot. For each
experimental treatment, three replicate colonies per isolate were spotted across two
bioassay dishes (36 colonies per plate). Plates were incubated at 30 C in a single layer
and in an upright orientation.
Antibiotic production was assayed after two, three, and four days of incubation, using a
sensitive overlay strain, Bacillus sp. 22U-2. An overnight liquid culture of B. 22U-2 in
nutrient broth was adjusted to an optical density at 600 nm of 0.475. This concentration-
adjusted culture was amended at a rate of 10% to cooled, molten soft nutrient agar
(nutrient broth plus 0.5% final agar concentration). Streptomycete colonies in the
bioassay dishes were covered with the seeded soft nutrient agar, using 175 mL/dish.
166
Overlaid plates were inverted and incubated at 30 C in a single layer for 24 hours. The
assay is illustrated in Figure 1. Antibiotic production was measured as the radius of the
resulting zones of inhibition, measured from the edge of each streptomycete colony. Two
measurements of inhibition zone size were taken at right angles and averaged. For each
isolate, data were expressed over time as mean values among replicate colonies. Data
were examined for changes in inhibition zone sizes, the timing of inhibition, and the
duration of active inhibition in the presence vs. absence of plant compounds.
Screening for GBLs in plant extracts
Screening for GBLs in plant extracts was performed with the use of a biosensor designed
for the detection of these molecules (Hsiao et al., 2009). The biosensor works by an
inducible antibiotic resistance mechanism, such that growth on a kanamycin-containing
medium only occurs when an appropriate signal binds the GBL receptor in the biosensor
(Hsiao et al., 2009). This biosensor is most sensitive to the GBLs produced by S.
coelicolor: SCB1, SCB2, and SCB3. Other GBL structural variants can be detected at
sufficiently high concentrations, but up to 500 times more compound may be required for
some structural variants relative to the cognate GBLs (Hsiao et al., 2009).
Initial screening was performed with intact seedlings germinated from surface-sterilized
seeds. For surface sterilization, seeds were washed in 70% ethanol for 2 minutes, drained,
washed for 30 min in 0.5% NaClO, and triple-rinsed with sterile water. Seeds were
transferred to nutrient agar to reveal remaining contamination. Germinated seeds showing
no evidence of contamination were transferred to soft plant growth medium (half strength
MS + half strength Gamborg's B5 vitamins [Sigma], 0.75% agar) and incubated under
florescent lights on the lab bench. When roots had grown, a layer of cooled, molten
nutrient agar containing kanamycin (5 ug/mL) was poured to submerge roots. Spores of
the GBL biosensor were applied to the kanamycin-containing medium adjacent to and
away from live plant roots.
167
Because the presence of live plants opened the assay to undesirable potential effects, such
as translocation or modification of kanamycin by a living plant, we generated extracts
from plant tissues for testing with the GBL biosensor. A wide variety of plant tissues
were purchased at a local grocery store. Liquid was collected from plant tissues by
processing through a kitchen juicer (Hamilton Beach). Remaining solids were removed
by sequential filtration and centrifugation (filtered through 6-ply cheesecloth, centrifuged
at 4000 RPM for 5 min, supernatant collected through 6-ply cheesecloth, centrifuged at
4000 RPM for 20 min). Fifty milliliters of the resulting liquid were mixed with an equal
volume of ethyl acetate. Aqueous and organic fractions were separated by centrifugation
(4000 RPM, 5 min) and then collected with a separatory funnel. The organic fraction was
concentrated to dryness with a rotovap, re-suspended in 1 mL methanol and filtered
through a 0.2 micron nylon filter into a clean tube. Extracts were stored dry at -20 C.
Testing of plant extracts with the GBL biosensor followed the recommended procedure
for the biosensor strain (Hsiao et al., 2009). Briefly, nutrient agar was prepared with
kanamycin at a concentration of 5 ug/mL. Spores of the biosensor strain were spread
across each plate and lids were vented to allow the surface of the medium to dry. Plant
extracts were re-suspended in 75 uL of methanol, and 2 uL was spotted onto the center of
each plate. Plates were incubated at 30 C for three days. Growth of the biosensor in a
halo around the spotted extract indicated that the extract contained a compound capable
of binding the GBL receptor protein (for illustration, see Figure 6, discussed below).
Lack of growth by the biosensor indicated a negative result. Positive results were
confirmed in a second assay using plant tissues obtained from a different source than the
first sample.
Among positive plant extracts, a crucial question was whether or not the positive activity
was due to the specific GBL whose cognate receptor has been engineered into the
biosensor, or whether the receptor could bind a different plant compound. Because the
GBL contains a lactone ring that is known to be sensitive to hydrolysis under alkaline
conditions (Takano, 2006), the stability of plant extract activity under elevated pH
168
provided an initial test to begin addressing this question; continued ability to activate the
biosensor after alkaline treatment would indicate that the interaction with the GBL
receptor involved a different chemical compound than the known cognate signal. A
volume of extract sufficient to activate the GBL biosensor was transferred into a small
tube. Extract pH was determined by spotting a small volume onto pH paper. Alkaline
treatment was achieved through the addition of 10M NaOH until pH reached
approximately 13. To avoid direct effects of high pH on the biosensor, extracts were
returned to a moderate pH by the addition of 10% HCl. The entire contents of the tube
were applied to the biosensor plate after pH adjustment.
Results
Screening for impacts of plant compounds on streptomycete antibiotic production
Theoretically, several aspects of antibiotic production could be susceptible to
modification by chemical signals from neighboring organisms. These include the quantity
of antibiotic produced, the timing of the onset of production, and the duration of time
over which active antibiotics are present. For clarity, illustrative examples are pulled for
presentation here (for the full dataset, see Figure S1). In some cases, the presence of
particular plant compounds increased the intensity of antibiotic inhibition. This was not
always consistent across time points (Figure 2A), although occasionally a uniform and
temporally consistent increase was observed (Figure 2B). In other cases, the presence of
particular plant compounds reduced the intensity of antibiotic inhibition, either at all
concentrations tested (Figure 2C) or in a dose-dependent manner (Figure 2D).
Accelerated antibiotic production in response to plant compounds led to detectable
inhibition at an earlier time point (Figure 2E). In contrast, delayed antibiotic production
was revealed as an increase in the duration of time required to reach full inhibitory
activity (Figure 2F). For some isolates in our collection, antibiotic activity appeared to be
transitory. Certain plant compounds were able to extend the longevity of antibiotic
activity (Figure 2G). Finally, in some cases no discernible impacts of the presence of
plant compounds on antibiotic production were observed (Figure 2H).
169
Streptomycete isolates showed differences in sensitivity toward external interference with
antibiotic production. Some isolates were relatively insensitive, responding with minor
variations in inhibition zone size to even the highest concentrations of plant compounds
tested (Figure 3A). In contrast, other isolates showed a much more variable response to
the imposed treatments (Figure 3B). As this implies, responses to a given plant
compound differed among isolates. For example, while the flavonoid chrysin reduced
antibiotic production by isolate 2-12 at all concentrations tested (Figure 2C), it had no
effect on antibiotic production by isolate TLI040 at the lowest dose, and accelerated
antibiotic production at higher doses (Figure 2E). Similarly, IAA increased antibiotic
production by isolate TLI224 (Figure 4A), decreased antibiotic production by isolate
TLI175 (Figure 4B), and had no impact on antibiotic production by isolate TLI030
(Figure 4C).
Screening for streptomycete GBLs in plant extracts
Pilot screening using intact seedlings of various species germinated from surface-
sterilized seeds suggested that some plant species produced compounds capable of
interacting the GBL receptor protein in the biosensor strain (Figure 5). These preliminary
results were followed up by tests of the ability of plant tissue extracts to activate the
biosensor strain. Approximately 13% of extracts yielded a positive growth response by
the GBL biosensor (Table 2). Positive results included extracts from several Allium spp.,
varieties of citrus and potato, and one variety of apple (Table 2). Both potato and onion
samples collected from three independent sources consistently gave a positive response
from the GBL biosensor (data not shown). Extracts from a broad range of other plant
tissues did not produce a positive response from the GBL biosensor (Table 2). Extracts of
lemon and lime were tested for pH-dependent stability, and were found to lose activity
after alkaline treatment (Figure 6). This is consistent with the known susceptibility of
GBLs to alkaline hydrolysis.
170
Figure 1 - An example of the assay used to detect changes in antibiotic production caused by the presence of plant compounds. See methods for details.
171
Figure 2 - Illustrative examples of changes to patterns of streptomycete antibiotic production as a result of exposure to various plant compounds. Treatments are indicated by the combination of line color and dash pattern, according to the legend in each panel.
A) B)
G) H)
E) F)
C) D)
Isolate: TLI254
Isolate: TLI098
Isolate: 93
Isolate: TLI092
Isolate: TLI040
Isolate: 2-12
Isolate: TLI030
Isolate: TLI254
172
Figure 3 - Isolates differed in sensitivity to external interference with antibiotic production. Shown are the responses of isolate 93 (A) and isolate TLI175 (B) to the imposed treatments. Treatments are indicated by the combination of line color and dash pattern, according to the legend in each panel. A)
B)
!"
#"
$"
%&"
%'"
&!"
&" (" #"
Inhi
bitio
n zo
ne s
ize
(mm
)
Days of incubation prior to overlay
Isolate: 93
!"
#"
$"
%&"
%'"
&!"
&" (" #"
Inhi
bitio
n zo
ne s
ize
(mm
)
Days of incubation prior to overlay
Isolate: TLI175
173
Figure 4 - Indole-3-acetic acid (IAA) has differential effects on antibiotic production among streptomycete isolates. Shown are the responses of isolate TLI224 (A), isolate TLI175 (B), and isolate TLI030 (C) to the addition of IAA at various concentrations to the culture medium. Treatments are indicated by the combination of line color and dash pattern, according to the legend in each panel. A)
B)
C)
!"
#"
$!"
$#"
%!"
%#"
%" &" '"
Inhi
bitio
n zo
ne s
ize
(mm
)
Days of incubation prior to overlay
()*+,)-."/+01"
233"%!"*4"
233"$"54"
233"#!"54"
Isolate: TLI224
!"
#"
$!"
$#"
%!"
%#"
%" &" '"
Inhi
bitio
n zo
ne s
ize
(mm
)
Days of incubation prior to overlay
()*+,)-."/+01"
233"%!"*4"
233"$"54"
233"#!"54"
Isolate: TLI175
!"
#"
$!"
$#"
%!"
%#"
%" &" '"
Inhi
bitio
n zo
ne s
ize
(mm
)
Days of incubation prior to overlay
()*+,)-."/+01"
233"%!"*4"
233"$"54"
233"#!"54"
Isolate: TLI030
174
Figure 5 - Intact pea seedlings were able to induce growth of the GBL biosensor on a
kanamycin-containing medium. An equal volume of biosensor spore suspension was
applied to each position indicated with an arrow. Note the amount of biosensor growth
near the plant (red arrow) compared to a location distant from the plant (blue arrow).
175
Figure 6 - Extracts from lemon and lime lost their ability to elicit a positive response
from the GBL biosensor after alkaline treatment. A positive growth response is evident in
the top panels, as the halo of growth seen surrounding extracts that have been spotted
onto the center of each plate.
Alkaline-treated
Initial extract
Lemon Lime
176
Figure S1 - The presence of particular plant compounds influences inhibitory activity by streptomycete isolates. Treatments are indicated by the combination of line color and dash pattern, according to the legend in each panel.
!"
#"
$"
%&"
%'"
&!"
&" (" #"
Inhi
bitio
n zo
ne s
ize
(mm
)
Days of incubation prior to overlay
Isolate: 2-12
!"
#"
$"
%&"
%'"
&!"
&" (" #"
Inhi
bitio
n zo
ne s
ize
(mm
)
Days of incubation prior to overlay
Isolate: 4-16
!"
#"
$"
%&"
%'"
&!"
&" (" #"
Inhi
bitio
n zo
ne s
ize
(mm
)
Days of incubation prior to overlay
Isolate: 4-20
177
Figure S1, continued
!"
#"
$"
%&"
%'"
&!"
&" (" #"
Inhi
bitio
n zo
ne s
ize
(mm
)
Days of incubation prior to overlay
Isolate: 93
!"
#"
$"
%&"
%'"
&!"
&" (" #"
Inhi
bitio
n zo
ne s
ize
(mm
)
Days of incubation prior to overlay
Isolate: TLI030
!"
#"
$"
%&"
%'"
&!"
&" (" #"
Inhi
bitio
n zo
ne s
ize
(mm
)
Days of incubation prior to overlay
Isolate: TLI040
178
Figure S1, continued
!"
#"
$"
%&"
%'"
&!"
&" (" #"
Inhi
bitio
n zo
ne s
ize
(mm
)
Days of incubation prior to overlay
Isolate: TLI060
!"
#"
$"
%&"
%'"
&!"
&" (" #"
Inhi
bitio
n zo
ne s
ize
(mm
)
Days of incubation prior to overlay
Isolate: TLI078
!"
#"
$"
%&"
%'"
&!"
&" (" #"
Inhi
bitio
n zo
ne s
ize
(mm
)
Days of incubation prior to overlay
Isolate: TLI092
179
Figure S1, continued
!"
#"
$"
%&"
%'"
&!"
&" (" #"
Inhi
bitio
n zo
ne s
ize
(mm
)
Days of incubation prior to overlay
Isolate: TLI098
!"
#"
$"
%&"
%'"
&!"
&" (" #"
Inhi
bitio
n zo
ne s
ize
(mm
)
Days of incubation prior to overlay
Isolate: TLI122
!"
#"
$"
%&"
%'"
&!"
&" (" #"
Inhi
bitio
n zo
ne s
ize
(mm
)
Days of incubation prior to overlay
Isolate: TLI129
180
Figure S1, continued
!"
#"
$"
%&"
%'"
&!"
&" (" #"
Inhi
bitio
n zo
ne s
ize
(mm
)
Days of incubation prior to overlay
Isolate: TLI145
!"
#"
$"
%&"
%'"
&!"
&" (" #"
Inhi
bitio
n zo
ne s
ize
(mm
)
Days of incubation prior to overlay
Isolate: TLI148
!"
#"
$"
%&"
%'"
&!"
&" (" #"
Inhi
bitio
n zo
ne s
ize
(mm
)
Days of incubation prior to overlay
Isolate: TLI152
181
Figure S1, continued
!"
#"
$"
%&"
%'"
&!"
&" (" #"
Inhi
bitio
n zo
ne s
ize
(mm
)
Days of incubation prior to overlay
Isolate: TLI175
!"
#"
$"
%&"
%'"
&!"
&" (" #"
Inhi
bitio
n zo
ne s
ize
(mm
)
Days of incubation prior to overlay
Isolate: TLI202
!"
#"
$"
%&"
%'"
&!"
&" (" #"
Inhi
bitio
n zo
ne s
ize
(mm
)
Days of incubation prior to overlay
Isolate: TLI224
182
Figure S1, continued
!"
#"
$"
%&"
%'"
&!"
&" (" #"
Inhi
bitio
n zo
ne s
ize
(mm
)
Days of incubation prior to overlay
Isolate: TLI254
!"
#"
$"
%&"
%'"
&!"
&" (" #"
Inhi
bitio
n!"#$%
!&'"%!()
)*!
Days of incubation prior to overlay
Isolate: TLI259
!"
#"
$"
%&"
%'"
&!"
&" (" #"
Inhi
bitio
n zo
ne s
ize
(mm
)
Days of incubation prior to overlay
Isolate: TLI265
183
Figure S1, continued
!"
#"
$"
%&"
%'"
&!"
&" (" #"
Inhi
bitio
n zo
ne s
ize
(mm
)
Days of incubation prior to overlay
Isolate: TLI267
!"
#"
$"
%&"
%'"
&!"
&" (" #"
Inhi
bitio
n zo
ne s
ize
(mm
)
Days of incubation prior to overlay
Isolate: TLI278
!"
#"
$"
%&"
%'"
&!"
&" (" #"
Inhi
bitio
n zo
ne s
ize
(mm
)
Days of incubation prior to overlay
Isolate: TLI296
184
Table 1 - Plant compounds tested in this study for impacts on antibiotic production by
streptomycetes, and their known roles in other interactions.
Class Compound Known roles Solvent Reference
Flavonoid Chrysin Nodulation, Mycorrhizae formation DMSO Scervino 2005
Morin
Nodulation, Mycorrhizae formation EtOH Scervino 2005
Quercetin
Nodulation, Mycorrhizae formation EtOH Tsai 1991
Rutin
Nodulation, Mycorrhizae formation DMSO Scervino 2005
Sesquiterpene lactone Artemisinin
Mycorrhizae formation Infection by parasitic plants MeOH Besserer 2006
Parthenolide
Mycorrhizae formation Infection by parasitic plants Acetone Besserer 2006
Plant hormone
Indole-3-acetic acid
Plant hormone Produced by some streptomycetes EtOH
Matsukawa 2007
185
Table 2 - Plants from which tissue extracts were made, and the response of the GBL
biosensor to those extracts. Positive results are highlighted in grey.
Scientific name Common name Response by GBL
biosensor Actinidia chinensis Kiwi - Allium ampeloprasum var. porum Leek + Allium cepa Onion, Red + Allium cepa Onion, Yellow + Allium cepa var. aggregatum Shallot - Allium sativum Garlic + Ananas comosus Pineapple - Apium graveolens Celery root - Armoracia rusticana Horseradish - Asparugus officinalis Asparagus - Beta vulgaris Beet - Brassica napobrassica Rutabaga - Brassica oleracea var. botrytis Cauliflower - Brassica oleracea var. gongylodes Kohlrabi - Brassica oleracea var. italica Broccoli - Brassica rapa var. rapa Turnip - Calopogonium caeruleum Jicama - Capsicum annuum Pepper, green bell - Capsicum annuum Pepper, jalapeno - Capsicum annuum Pepper, orange bell - Capsicum annuum Pepper, poblano - Capsicum annuum Pepper, red bell - Capsicum annuum Pepper, yellow bell - Citrullus sp. Watermelon - Citrus aurantifolia Lime + Citrus limon Lemon + Citrus maxima Pummelo - Citrus x paradisi Grapefruit, red - Citrus x sinensis Orange, navel - Cucumis melo Honeydew - Cucumis melo var. cantalupensis Cantaloupe - Cucumis sativus Cucumber -
186
Table 2, continued
Scientific name Common name Response by GBL
biosensor Cucurbita moschata Squash, butternut - Cucurbita pepo Squash, acorn - Cucurbita pepo (green) Squash, green straight neck - Cucurbita pepo (yellow) Squash, yellow straight neck - Daucus carota Carrot - Fragaria x ananassa Strawberry - Ipomoea batatas Sweet potato - Malus domestica Apple, Golden Delicious - Malus domestica Apple, Granny Smith + Malus domestica Apple, Gala - Mangifera indica Mango - Phaseolus vulgaris Bean, green - Physallis philadelphica Tomatillo - Pimpinella anisum Fennel - Prunus armeniaca Apricot - Prunus avium Cherries, Bing - Prunus persica Nectarine - Prunus persica Peach - Punica granatum Pomegranate - Pyrus communis Pear, green d'Anjou - Pyrus communis Pear, red d'Anjou - Raphanus sativus Radish - Raphanus sativus var. longipinnatus Radish, Daikon - Rubus fruticosus Blackberry - Sechium edule Squash, Chayote - Solanum lycopersicum Tomato, Bushel Boy - Solanum lycopersicum Tomato, Roma - Solanum melongena Eggplant - Solanum tuberosum Potato, Red + Solanum tuberosum Potato, Russet + Solanum tuberosum Potato, Yukon Gold - Vitis sp. Grape, red - Vitis sp. Grape, green - Yucca sp. Yucca root - Zingiber officinale Ginger -
187