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