Final Thesis File

Understanding the Importance of Microbial Biogeography

to Australian Winemaking

Di Liu ORCID iD: https://orcid.org/0000-0003-3241-1421

Submitted in total fulfilment of the requirements of the degree of Doctor of Philosophy – Agricultural Sciences

Faculty of Veterinary and Agricultural Sciences

The University of Melbourne

August 2020

ABSTRACT

Microbes are a vital part of ecosystems and play key roles in the essential processes of the functioning. In agriculture, microbial ecology has wide reaching impacts on crop growth and quality commodity production. As a high value agricultural product, wine is a useful model for elucidating the effects of microbial ecology from the vineyard to the winery. Microbial growth and metabolism is an inherent component of wine production, influencing grapevine health and productivity, conversion of sugar to ethanol during fermentation, and the flavour, aroma and quality of finished wines. Recent advances in genetic sequencing and metagenomic approaches has extended our understanding of microbial distribution patterns and established the unique biogeography model in viticulture. While the contributions of microbial biogeography to wine metabolites and regional distinctiveness (known as terroir, a well-recognised and celebrated character in wine industry), and by which mechanisms, remain tenuous. This thesis focuses on the microbial biogeography of wine, the interplay between microbial patterns and affecting factors, and how these patterns drive wine quality and styles.

I begin by investigating the distribution patterns of bacteria and fungi at large scale, and their roles in shaping wine characteristics. Samples were collected from vineyard soil, grape must, and wine ferments across six geographically separated wine-producing regions in southern Australia (~ 400 km). Soil and grape must microbiota exhibited distinctive regional patterns, as well as wine aroma profiles. Associations among soil and wine microbiota, abiotic factors (weather and soil properties), and wine regionality were modelled, highlighting that fungal communities was the most important driver of wine aroma profiles. Source tracking wine-related fungi in the vineyard suggests that soil is a source reservoir of grape- and must-associated fungi which might be translocated via xylem sap.

I then move on to elucidate the fungal ecology within vineyards. Fungal communities were characterised over space and time that associated with the grapevine (grapes, flowers, leaves, roots, root zone soil) during the annual growth cycle (flowering, fruit set, veraison, and harvest). Fungi were significantly influenced by the grapevine habitat and plant development stage, with little influences

i from the geographic location (<5 km). The developmental stage of veraison, where grapes undergo a dramatic change in metabolism and start ripening process, saw a distinct shift in fungal communities.

A core fungal microbiota of grapevines (based on abundance-occupancy models) existed over space and time which drove the seasonal community succession. Beyond coinciding with the changing plant metabolism and physiology, strong correlations with solar radiation and water status suggests that the core microbiota changes with respect to the changing environments during plant development.

I further investigate fungal contributions to wine aroma profiles by quantifying multiple layers of fungi, combining metagenomics and population genetics. Fungal communities were characterised associated with Pinot Noir and Chardonnay grape must/juice and ferments coming from three wine estates (including 11 vineyards) in the Mornington Peninsula wine region. At this scale (< 12 km), fungal communities, yeast populations, and Saccharomyces cerevisiae populations differentiated between geographic origins (estate/vineyard), with influences from the grape variety. During spontaneous fermentation, growth and dominance of S. cerevisiae reshaped the fungal community and structured the biodiversity at strain level. Associations between fungal microbiota and wine metabolites highlights the primary role of S. cerevisiae in determining wine aroma profiles at sub- regional scale.

Overall, this thesis provides a significant body of knowledge to the microbial ecology field. Using vineyards, grapes, and wine as a model system, these findings relate microbial biogeography, environments, and quality agricultural commodity production. It provides fundamental perspectives to conserve the biodiversity and functioning for sustainable agriculture under the changing climate.

ii DECLARATION

This is to certify that:

i. the thesis comprises only my original work towards the Ph.D. except where indicated in the Preface,

ii. due acknowledgement has been made in the text to all other material used, and

iii. the thesis is fewer than 100,000 words in length, exclusive of tables, figures, bibliographies and appendices.

Di Liu August 2020

iii PREFACE

Dr Kate Howell, Prof Deli Chen, Dr Pangzhen Zhang, and Dr Qinglin Chen were supervisors of this

PhD project, and were involved in experimental design and manuscript editing. Dr Pangzhen Zhang built connections with the wine industry. Chapter 5 was performed in collaboration with Institut

National de la Recherche Agronomique (INRA), France. Dr Jean-Luc Legras performed microsatellite analysis and assisted in the interpretation of the data. Sample collection, yeast isolation, genomic

DNA extraction and sequencing, wine volatile analysis, data analysis, interpretation and writing of the manuscript were my responsibility.

The candidate’s principal supervisor, Dr Kate Howell has signed The University of Melbourne’s declaration for a thesis with publication form. For all published work in this thesis, the respective co- authors have signed The University of Melbourne’s co-author authorisation form to certify that the candidate’s contribution to the publication was greater than 50%. The manuscript intended to be submitted for publication is formatted according to the journal’s formatting requirements.

Supplemental data included in this thesis refers to supplemental files of individual published/to be submitted articles. Addendum refers to the data and discussion that was omitted from the published article.

This thesis comprises the following original manuscripts, where I was the lead author:

Review article:

• Liu, D., Zhang, P., Chen, D., and Howell, K.S. (2019). From the vineyard to the winery: how

microbial ecology drives regional distinctiveness of wine. Frontiers in Microbiology 10,

2679.

Candidate’s contribution: 80%

Work description: DL did the systematic literature review and wrote the first draft of the

manuscript. KH revised and added to the draft. All authors edited the manuscript.

iv Research articles:

• Liu D., Chen Q., Zhang P., Chen D., Howell K.S. (2020). The fungal microbiome is an

important component of vineyard ecosystems and correlates with regional distinctiveness of

wine. mSphere 5: e00534-20.

Candidate’s contribution: 80%

Work description: DL and PZ designed the experiment. DL collected samples, conducted

experiments, data analysis, data interpretation, and wrote the draft. KH and QC revised and

added to the draft. All authors edited the manuscript.

• Liu, D. and Howell, K.S.. (2020). Community succession of the grapevine fungal microbiome

in the annual growth cycle. Environmental Microbiology. Accepted Author Manuscript.

doi:10.1111/1462-2920.15172

Candidate’s contribution: 80%

Work description: Both authors designed the experiment and structured the article. DL

collected samples, conducted experiments, data analysis, data interpretation, and wrote the

draft. KH revised and added to the draft.

• Liu, D., Legras, J.-L., Zhang, P., Chen, D., and Howell, K.S. Diversity and dynamics of

microbes during spontaneous fermentation and its associations with unique aroma profiles in

wine. International Journal of Food Microbiology. Accepted Author Manuscript.

Candidate’s contribution: 70%

Work description: DL and KH designed the experiment. DL collected samples, conducted

experiments (except microsatellite analysis), data analysis, data interpretation, and wrote the

draft. JLL performed microsatellite analysis and assisted in the interpretation of the data. KH

revised and added to the draft. All authors edited the manuscript.

v There are also two co-authored papers published during my PhD, which are attached in the appendices to the thesis:

• Winters, M., Panayotides, D., Bayrak, M., Rémont, G., Gonzalez Viejo, C., Liu, D., Le, B.,

Liu, Y., Luo, J., Zhang, P., and Howell, K. (2019). Defined co‐cultures of yeast and bacteria

modify the aroma, crumb and sensory properties of bread. Journal of applied microbiology

127: 778-793.

Candidate’s contribution: Assistance in wine aroma determination method establishment, data

analysis, data interpretation, figure 2 plotting, and draft editing.

• Zhang, P., Wu, X., Needs, S., Liu, D., Fuentes, S., Howell, K., (2017). The influence of

apical and basal defoliation on the canopy structure and biochemical composition of Vitis

vinifera cv. Shiraz grapes and wine. Frontiers in chemistry 5: 48.

Candidate’s contribution: Assistance in wine aroma determination, data analysis, data

interpretation, and draft editing.

vi ACKNOWLEDGEMENTS

My PhD work was jointly sponsored by The University of Melbourne and Australia Grape and Wine

Authority (Wine Australia). The work was performed at The University of Melbourne and supported by vignerons across Victoria. I give my sincere thanks to Wine Australia and our industry partners who allowed vineyard access, enabled sampling, and provided wine samples: Stonier Wines,

Kooyong, Port Phillip Estate, Main Ridge Estate, Yabby Lake Vineyard, Barwon Ridge Wines,

Hanging Rock Winery, Curly Flat, Helen & Joey Estate, Coldstream Hills, Mount Langi Ghiran,

Best's Wines, Bass River Winery, and Fowles Wine.

I am extremely grateful to my supervisors for helping me fulfil my PhD. I would like to thank my principal supervisor Kate Howell for your wisdom, patience, understanding, inspiration, and standing by me forever. You are the great mentor to me, and I have learnt many valuable lessons that will company me throughout my career and my life. A special thanks goes to my co-supervisor Deli for bringing the idea for the project to me in the first place, being supportive and inspiring, and encouraging me to complete my wine business education. My sincere gratitude also goes to Pangzhen for your extreme enthusiasm and hard work to build collaborations with the industry, and all your suggestions throughout my project. Qinglin, thank you for training me to manipulate data and your substantial contributions to data mining in the paper. I would also like to thank A/Prof Sigfredo

Fuentes for being an excellent chair of my PhD committee.

I would like to thank my fantastic colleagues and friends throughout my PhD for all your guidance, company, and friendship. Thank you Qing Xie, Baobao Pan, Yujing Zhang, Chunhe Gu, Jiaqiang

Luo, Yit Tao Loo, Azin Moslemi, Anna Wittwer, Annli Tee, Maryam Paykary, for all the fun and relaxing break. It helps take away the confusion and fears when things go wrong and keeps me positive and persistent throughout this journey.

vii To my parents, thank you for raising me up and raising me to be an independent and honest person.

Thank you for trusting me and supporting me forever. I love you forever.

To Xi Li, thank you for being my partner, my best friend, and my secret PhD advisor. You help me be better myself and reach farther than I ever thought I could go. You are the one.

viii Declaration for a thesis with publication

PhD and MPhil students may include a primary research publication in their thesis in lieu of a chapter if: • The student contributed greater than 50% of the content in the publication and is the “primary author”, ie. the student was responsible primarily for the planning, execution and preparation of the work for publication • The student has approval to include the publication in their thesis from their Advisory Committee • It is a primary publication that reports on original research conducted by the student during their enrolment • The initial draft of the work was written by the student and any subsequent editing in response to co-authors and editors reviews was performed by the student • The publication is not subject to any obligations or contractual agreements with a third party that would constrain its inclusion in the thesis

Students must submit this form, along with Co-author authorisation forms completed by each co-author (combined as a single file), when the thesis is submitted to the Thesis Examination System: https://tes.app.unimelb.edu.au/. If you are including multiple publications in your thesis you will need to list each publication on this form. Further information on this policy is available at: gradresearch.unimelb.edu.au/preparing-my-thesis/thesis-with-publication

A. STUDENT’S DECLARATION

I declare that: • the information below is accurate • the publication(s) below meets the requirements to be included in the thesis • The advisory committee has met and agreed to the inclusion of the publication(s) in the student’s thesis • All co-authors of the publication(s) have reviewed the information below and have agreed to its veracity. Co-Author Authorisation forms for each co-author are attached.

Student’s name Student’s signature Date (dd/mm/yy)

Di Liu 07/08/2020

PRINCIPAL SUPERVISOR’S DECLARATION

Supervisor’s name Supervisor’s signature Date (dd/mm/yy)

Kate Howell 07/08/2020

B. PUBLICATION DETAILS (to be completed by the student) Click on this box and on the “+” button in the bottom right corner to enter multiple publications.

Full title From the vineyard to the winery: how microbial ecology drives regional distinctiveness of wine

Authors Di Liu, Pangzhen Zhang, Deli Chen and Kate Howell

Student’s contribution (%) 80% Volume/page numbers 10/2679

Journal or book name Frontiers in Microbiology

Status ☐Accepted and In-press ■ Published ☐ In progress Date accepted/ published 20/11/2019

The University of Melbourne ix CRICOS Provider Number: 00116K Last Updated 16 July 2020 Declaration for a thesis with publication

A. STUDENT’S DECLARATION

Student’s name Student’s signature Date (dd/mm/yy)

Di Liu 07/08/2020

PRINCIPAL SUPERVISOR’S DECLARATION

Supervisor’s name Supervisor’s signature Date (dd/mm/yy)

Kate Howell 07/08/2020

B. PUBLICATION DETAILS (to be completed by the student) Click on this box and on the “+” button in the bottom right corner to enter multiple publications.

Full title The fungal microbiome is an important component of vineyard ecosystems and correlates with regional distinctiveness of wine

Authors Di Liu, Qinglin Chen, Pangzhen Zhang, Deli Chen, and Kate Howell

Student’s contribution (%) 80% Volume/page numbers 5/e00534-20

Journal or book name mSphere

Status ■Accepted and In-press ☐ Published ☐ In progress Date accepted/ published 29/07/2020

The University of Melbourne x CRICOS Provider Number: 00116K Last Updated 16 July 2020 Declaration for a thesis with publication

A. STUDENT’S DECLARATION

Student’s name Student’s signature Date (dd/mm/yy)

Di Liu 07/08/2020

PRINCIPAL SUPERVISOR’S DECLARATION

Supervisor’s name Supervisor’s signature Date (dd/mm/yy)

Kate Howell 07/08/2020

B. PUBLICATION DETAILS (to be completed by the student) Click on this box and on the “+” button in the bottom right corner to enter multiple publications.

Full title Community succession of the grapevine fungal microbiome in the annual growth cycle

Authors Di Liu, and Kate Howell

Student’s contribution (%) 80% Volume/page numbers Not available at this stage (doi:10.1111/1462-2920.15172)

Journal or book name Environmental Microbiology

Status ☐Accepted and In-press ■ Published ☐ In progress Date accepted/ published 20/07/2020

The University of Melbourne xi CRICOS Provider Number: 00116K Last Updated 16 July 2020 Co-author authorisation form

All co-authors must complete this form. By signing below co-authors agree to the listed publication being included in the student’s thesis and that the student contributed greater than 50% of the content of the publication and is the “primary author” ie. the student was responsible primarily for the planning, execution and preparation of the work for publication.

In cases where all members of a large consortium are listed as authors of a publication, only those that actively collaborated with the student on material contained within the thesis should complete this form. This form is to be used in conjunction with the Declaration for a thesis with publication form.

Students must submit this form, along with the Declaration for thesis with publication form, when the thesis is submitted to the Thesis Examination System: https://tes.app.unimelb.edu.au/

Further information on this policy and the requirements is available at: gradresearch.unimelb.edu.au/preparing-my-thesis/thesis-with-publication

A. PUBLICATION DETAILS (to be completed by the student) From the vineyard to the winery: how microbial ecology drives regional distinctiveness Full title of wine Authors Di Liu, Pangzhen Zhang, Deli Chen and Kate Howell

Student’s contribution (%) 80%

Journal or book name Frontiers in Microbiology

Volume/page numbers 10/2679

Status ☐ Accepted and In-press ■ Published Date accepted/published

☐ In progress 20/11/2019

B. CO-AUTHOR’S DECLARATION (to be completed by the collaborator)

I authorise the inclusion of this publication in the student’s thesis and certify that: • the declaration made by the student on the Declaration for a thesis with publication form correctly reflects the extent of the student’s contribution to this work; • the student contributed greater than 50% of the content of the publication and is the “primary author” ie. the student was responsible primarily for the planning, execution and preparation of the work for publication.

Co-author’s name Co-author’s signature Date (dd/mm/yy)

Kate Howell 07/08/2020

The University of Melbourne CRICOS Provider Number: 00116K xii Last Updated 19 October 2017 Co-author authorisation form

Students must submit this form, along with the Declaration for thesis with publication form, when the thesis is submitted to the Thesis Examination System: https://tes.app.unimelb.edu.au/

Further information on this policy and the requirements is available at: gradresearch.unimelb.edu.au/preparing-my-thesis/thesis-with-publication

Student’s contribution (%) 80%

Journal or book name Frontiers in Microbiology

Volume/page numbers 10/2679

Status ☐ Accepted and In-press ■ Published Date accepted/published

☐ In progress 20/11/2019

B. CO-AUTHOR’S DECLARATION (to be completed by the collaborator)

Co-author’s name Co-author’s signature Date (dd/mm/yy)

Deli Chen 07/08/2020

The University of Melbourne CRICOS Provider Number: 00116K Last Updated 19 October 2017 xiii Co-author authorisation form

Students must submit this form, along with the Declaration for thesis with publication form, when the thesis is submitted to the Thesis Examination System: https://tes.app.unimelb.edu.au/

Further information on this policy and the requirements is available at: gradresearch.unimelb.edu.au/preparing-my-thesis/thesis-with-publication

Student’s contribution (%) 80%

Journal or book name Frontiers in Microbiology

Volume/page numbers 10/2679

Status ☐ Accepted and In-press ■ Published Date accepted/published

☐ In progress 20/11/2019

B. CO-AUTHOR’S DECLARATION (to be completed by the collaborator)

Co-author’s name Co-author’s signature Date (dd/mm/yy)

Pangzhen Zhang 07/08/2020

The University of Melbourne CRICOS Provider Number: 00116K Last Updated 19 October 2017 xiv Co-author authorisation form

Students must submit this form, along with the Declaration for thesis with publication form, when the thesis is submitted to the Thesis Examination System: https://tes.app.unimelb.edu.au/

Further information on this policy and the requirements is available at: gradresearch.unimelb.edu.au/preparing-my-thesis/thesis-with-publication

A. PUBLICATION DETAILS (to be completed by the student) The fungal microbiome is an important component of vineyard ecosystems and Full title correlates with regional distinctiveness of wine Authors Di Liu, Qinglin Chen, Pangzhen Zhang, Deli Chen and Kate Howell

Student’s contribution (%) 80%

Journal or book name mSphere

Volume/page numbers 5/e00534-20

Status ■ Accepted and In-press ☐ Published Date accepted/published

☐ In progress 29/07/2020

B. CO-AUTHOR’S DECLARATION (to be completed by the collaborator)

Co-author’s name Co-author’s signature Date (dd/mm/yy)

Kate Howell 07/08/2020

The University of Melbourne CRICOS Provider Number: 00116K Last Updated 19 October 2017 xv Co-author authorisation form

Students must submit this form, along with the Declaration for thesis with publication form, when the thesis is submitted to the Thesis Examination System: https://tes.app.unimelb.edu.au/

Further information on this policy and the requirements is available at: gradresearch.unimelb.edu.au/preparing-my-thesis/thesis-with-publication

Student’s contribution (%) 80%

Journal or book name mSphere

Volume/page numbers 5/e00534-20

Status ■ Accepted and In-press ☐ Published Date accepted/published

☐ In progress 29/07/2020

B. CO-AUTHOR’S DECLARATION (to be completed by the collaborator)

Co-author’s name Co-author’s signature Date (dd/mm/yy)

Deli Chen 07/08/2020

The University of Melbourne CRICOS Provider Number: 00116K Last Updated 19 October 2017 xvi Co-author authorisation form

Students must submit this form, along with the Declaration for thesis with publication form, when the thesis is submitted to the Thesis Examination System: https://tes.app.unimelb.edu.au/

Further information on this policy and the requirements is available at: gradresearch.unimelb.edu.au/preparing-my-thesis/thesis-with-publication

Student’s contribution (%) 80%

Journal or book name mSphere

Volume/page numbers 5/e00534-20

Status ■ Accepted and In-press ☐ Published Date accepted/published

☐ In progress 29/07/2020

B. CO-AUTHOR’S DECLARATION (to be completed by the collaborator)

Co-author’s name Co-author’s signature Date (dd/mm/yy)

Pangzhen Zhang 07/08/2020

The University of Melbourne CRICOS Provider Number: 00116K Last Updated 19 October 2017 xvii Co-author authorisation form

Students must submit this form, along with the Declaration for thesis with publication form, when the thesis is submitted to the Thesis Examination System: https://tes.app.unimelb.edu.au/

Further information on this policy and the requirements is available at: gradresearch.unimelb.edu.au/preparing-my-thesis/thesis-with-publication

Student’s contribution (%) 80%

Journal or book name mSphere

Volume/page numbers 5/e00534-20

Status ■ Accepted and In-press ☐ Published Date accepted/published

☐ In progress 29/07/2020

B. CO-AUTHOR’S DECLARATION (to be completed by the collaborator)

Co-author’s name Co-author’s signature Date (dd/mm/yy)

Qinglin Chen 07/08/2020

The University of Melbourne CRICOS Provider Number: 00116K Last Updated 19 October 2017 xviii Co-author authorisation form

Students must submit this form, along with the Declaration for thesis with publication form, when the thesis is submitted to the Thesis Examination System: https://tes.app.unimelb.edu.au/

Further information on this policy and the requirements is available at: gradresearch.unimelb.edu.au/preparing-my-thesis/thesis-with-publication

A. PUBLICATION DETAILS (to be completed by the student)

Full title Community succession of the grapevine fungal microbiome in the annual growth cycle

Authors Di Liu, and Kate Howell

Student’s contribution (%) 80%

Journal or book name Environmental Microbiology

Volume/page numbers Not available at this stage (doi:10.1111/1462-2920.15172)

Status ☐ Accepted and In-press ■ Published Date accepted/published

☐ In progress 20/07/2020

B. CO-AUTHOR’S DECLARATION (to be completed by the collaborator)

Co-author’s name Co-author’s signature Date (dd/mm/yy)

Kate Howell 07/08/2020

The University of Melbourne CRICOS Provider Number: 00116K Last Updated 19 October 2017 xix TABLE OF CONTENTS

Abstract………………………………………………………………………………………………i

Declaration…………………………………………………………………………………………..iii

Preface………………………………………………………………………………………….…….iv

Acknowledgements………………………………………………………………………...………vii

Declaration for a thesis with publication…………………………………………………….….ix

Co-author authorisation forms…………………………………………………………….….…xii

Table of contents……………………………………………………………………………...…….xx

List of Figures…………………………………………………………………………...………..xxiii

List of Tables…………………………………………………………………………..………….xxv

Chapter 1 General Introduction.…………………………………………………………………….....1

1.1 Terroir……………………………………………………………………………………...1 1.2 Microbial activities in agricultural production……………………………………………...3 1.2.1 Microbial activities in the field…………………………………………………..3 1.2.2 Fermentation…………………………………………………………………….7 1.3 Microbial biogeography……………………………………………………………………9 1.4 Objectives…………………………………………………………………………………10 1.5 Thesis overview…………………………………………………………………………..11 1.6 References………………………………………………………………………………...14

Chapter 2 From the vineyard to the winery: how microbial ecology drives regional distinctiveness of wine……………………………………………………………………………..21

2.1 Published review paper…………………………………………………………………..22 2.2 Supplementary data………………………………………………………………………35

Chapter 3 The fungal microbiome is an important component of vineyard ecosystems and correlates with regional distinctiveness of wine……………………………………….…38

3.1 Published research article……………………………………………………………….39 3.2 Supplementary data………………………………………………………………………54

xx 3.3 Addendum………………………………………………………………………………...69

Chapter 4 Community succession of the grapevine fungal microbiome in the annual growth cycle……………………………………….………………………………………………...73

4.1 Published research article……………………………………………………………….74 4.2 Supplementary data……………………………………………………………………...90

Chapter 5 Diversity and dynamics of fungi during spontaneous fermentations and association with unique aroma profiles in wine……………….…………………………….103

5.1 Introduction……………….…………………………………………………………….106 5.2 Materials and methods….……………………………………………………………….108 5.2.1 Sampling……………………………………………………………………...108 5.2.2 Wine volatile analysis………………………………………………………..109 5.2.3 DNA extraction and sequencing……………………………………………...110 5.2.4 Yeast isolation and identification…………………………………………….111 5.2.5 Microsatellite characterisation………………………………………………..111 5.2.6 Data analysis………………………………………………………………….112 5.3 Results…………………………………………………………………………………...146 5.3.1 Fungal microbiota vary by geographical origin and grape variety…………...114 5.3.2 Microbiota dynamics during spontaneous wine fermentation………………..116 5.3.3 Yeast population dynamics during spontaneous wine fermentation…………118 5.3.4 Aroma profiles are distinctive for wines of each geographical origin………..121 5.3.5 Fungal microbiota correlate to wine aroma profiles………………………….122 5.4 Discussion……………………………………………………………………………….124 5.4.1 Fungal ecology at multiple layers during spontaneous wine fermentation…..125 5.4.2 Association between fungal microbiota and composition of the wine metabolome…………………………………………………………………………127 5.5 Conclusions……………………………………………………………………………...128 5.6 References……………………………………………………………………………….130 5.7 Supplementary data……………………………………………………………………...137

Chapter 6 General discussion and conclusions………………………………………………148

6.1 Research Findings……………………………………………………………………….148 6.2 General discussion………………………………………………………………………150 6.2.1 The fungal microbiome is a component of regional distinctiveness of wine...150

xxi 6.2.2 The grapevine-associated microbiome correlates with the environment...... 151 6.2.3 The central role of Saccharomyces cerevisiae...... 152 6.3 Significance of the study...... 153 6.4 Limitations of the study...... 154 6.5 Future Directions...... 155 6.6 Conclusions...... 156 6.7 References...... 156

Appendices...... 159

Appendix I Defined co‐cultures of yeast and bacteria modify the aroma, crumb and sensory properties of bread...... 160

Appendix II The influence of apical and basal defoliation on the canopy structure and biochemical composition of Vitis vinifera cv. Shiraz grapes and wine...... 176

xxii LIST of FIGURES

Chapter 2

Figure 1 Overview of the wine-related microbiota from the vineyard to the winery

Figure 2 A scenario of wine microbial biogeography

Chapter 3

Figure 1 Wine metabolome shows regional variation across six wine-growing regions in 2017

Figure 2 Regional patterns of vineyard soil microbial community demonstrated by Bray-Curtis distance PCoA of bacterial communities (A) and fungal communities (B)

Figure 3 Microbiota exhibit regional differentiation in musts for both bacterial and fungal profiles

Figure 4 Main predictors of wine regionality

Figure 5 Direct and indirect effects of climate, soil properties, and microbial diversity (Shannon index) on wine regionality

Figure S1 Map of 15 sampling vineyards from six Pinot Noir wine-producing regions in southern

Australia, spanning 400 km (E-W)

Figure S2 Vineyard soil microbial community compositions across all samples from six grape growing regions

Figure S3 Stage of fermentation influences microbial diversities

Figure S4 Xylem sap is a potential translocation medium of yeasts from the soil to the grapevine

Chapter 4

Figure 1 Fungal communities demonstrate different structures and compositions depending on habitats in the vineyard

Figure 2 Grape developmental stage influences microbial diversity in the vineyard

Figure 3 The core microbiome in grapevine habitats and the soil over the growing season

xxiii Figure 4 The core microbiome (as OTUs) can distinguish developmental stages for plant habitats and soil. Shown are the important features (values >0.10) based on Mean Decrease Gini (MDG) of random forest models

Figure 5 Statistically significant and strong co-occurrence relationships among grapevine-associated and soil dominant fungi and weather conditions during annual growth cycle

Figure S1 Co-occurrence pattern between the soil core microbiome and weather parameters

Figure S2 Experimental design and sampling scheme

Chapter 5

Figure 1 Musts before fermentation (BF) have differential fungal communities depending geographical origins and grape varieties

Figure 2 Stage of fermentation influences microbial diversity and composition of Pinot Noir

Figure 3 Neighbour-joining tree showing the clustering of 94 S. cerevisiae strains isolated from three wine estates

Figure 4 Wine metabolites exhibit geographical variation

Figure 5 Partial least squares regression (PLSR) demonstrates microbial influence on volatile compounds of Pinot Noir and Chardonnay wines

Figure S1 Sampling map of 11 vineyards from three wine estates in Mornington Peninsula wine region, Australia.

Figure S2 Stage of fermentation influences microbial diversity and composition of Chardonnay

xxiv LIST of TABLES

Chapter 2

Table 1 Recent findings on wine microbial ecology from the vineyard to the winery.

Table S1 Studies of affect factors on vineyard and wine microbial communities

Chapter 3

Table S1 Vineyard conditions and number of soils, must and ferments, and plant samples collected in this study

Table S2 ANOVA results of wine volatile compounds among wine growing regions in 2017

Table S3 Permutational multivariate analysis of variance (PERMANOVA) using distance matrices of category effects on microbial and wine aroma diversity

Table S4 α-diversity (Shannon index) of soil and must microbial communities from six wine growing regions in 2017 and 2018

Table S5 ANOVA results of soil properties and weather conditions among wine growing regions in

2017

Chapter 4

Table 1 Experimental factors predicting α- and β-diversity of fungal communities in the vineyard

Table S1 OTUs identified as core community associated with the grapevine or within each vineyard habitat

Table S2 Classification of random forest models of the development stage

Table S3 Important features of random forest models outside the core community

Table S4 ANOVA results of weather conditions between developmental stages

Table S5 Spray programme for the studied vineyards from shoot growth until pre-veraison

Table S6 Soil properties of each sampling site at harvest

Chapter 5

xxv Table S1 Summary of microbial samples colleceted in Mornington Peninsula wine region in 2019

Table S2 Permutational multivariate analysis of variance (PERMANOVA) using distance matrices of category effects on fungal communities in spontaneous fermentations

Table S3 The occurrence and relative abundance (%) of yeast species during spontaneous fermentations of Pinot Noir and Chardonnay

Table S4 Results of FST statistic analysis

Table S5 ANOVA results of wine volatile compounds among wine estates

xxvi CHAPTER ONE

General Introduction

Microorganisms are a vital part of environmental ecosystems and play key roles in the functioning of the Earth's biosphere (Kirchman, 2018). Alongside affecting plant health, growth, and productivity in the agriculture (Müller et al., 2016), microbes have been utilised for fermentations for thousands of years to produce economically and culturally important foods including wine, beer, bread, cheese, coffee, and chocolate and thus have profound impacts on the global economy and human health (Ross et al., 2002; Fleet, 2006). Microbial ecology is a rapidly growing field, driven by recent technological advances in high throughput sequencing and metagenomics. Microbial biogeography, recently established in the agricultural sphere, has shown the potential to shape agricultural commodity and food quality (Martiny et al., 2006; Bokulich et al., 2016a). In this thesis I investigate the community patterns and ecology of wine-related microbiota and their effects in wine flavour and styles across distinct wine-growing areas in southern Australia. In doing so I aim to contribute to the increasing understanding of microbial ecology and provide perspectives to management practices to optimise food composition.

1.1 Terroir

The character and quality of agricultural products (for example wine, cheese, meat products) are formed, in part, by the place where they grow. This phenomenon centres on the idea of terroir, which claims that the relationship between the characteristics of an agricultural product (flavours, quality, style) and its geographic origin derive from an interplay of the natural environment, the humans, their agricultural practices, and their histories/traditions (Barham, 2003). The concept of terroir achieves the most elaborate expression in the case of wine. Wines made from the same grape cultivar but grown in different regions are appreciated for their distinctive characteristics (Van Leeuwen and

Seguin, 2006), which have been already a well-recognised and celebrated notion of wine identity. The

International Organisation of Vine and Wine (OIV) defines: ‘‘Vitivinicultural ‘terroir’ is a concept which refers to an area in which collective knowledge of the interactions between the identifiable

1 physical and biological environment and applied vitivinicultural practices develops, providing distinctive characteristics for the products originating from this area. ‘Terroir’ includes specific soil, topography, climate, landscape characteristics and biodiversity features’’ (OIV, 2010). The scope and breadth of its definition makes a scientific definition of terroir difficult, but researchers are attempting to find measurable definitions and quantification of scope, scale and outcomes of terroir.

Regional characteristics of wine can be identified with chemical compositions and sensory properties.

Chemical fingerprinting techniques show that wines made from particular regions are consistently different from vintage to vintage and wine compositions can be separated with statistical availability

(Giménez-Miralles et al., 1999; Pereira et al., 2005; Gonzálvez et al., 2009; Fotakis et al., 2013; Pepi and Vaccaro, 2017). Using chemometric approaches, these studies established relationships between wine composition and geographic origin.

Wine terroir has been ascribed local environmental parameters that influence grapevine growth and development, such as climate, soil, topography, and human management practices. Climate, the long- term weather patterns of the growing region, is a profound element influencing grapevine physiology and berry quality thus shaping wine quality and styles (Jackson and Lombard, 1993; Gladstones,

2011). The climate impacts viticulture and wine quality through temperature, precipitation, and solar radiation, which has been recognised at multiple levels: macroclimate at regional scale, mesoclimate at vineyard scale, and microclimate between vines and within the canopy (Van Leeuwen and Seguin,

2006).

Vineyard soil exerts a strong influence on vine growth and development thus final wine composition.

Soil types, embedding properties such as water drainage and holding capacity, aggregation (defined by texture range clay, loam and sand), and ease of root penetration, affect grapevine growth and health (White, 2015). Soil nutrient composition, in particular nitrogen, influences vine vigour, yield, and berry composition (Spayd et al., 1994; Van Leeuwen and Seguin, 2006). Multielement

2 composition of wines can be linked back to major and trace elements in the soil (Almeida and

Vasconcelos, 2003; Kment et al., 2005).

Based on the local conditions, thoughtful human-driven agricultural practices including vine management (training, pruning, trellising, canopy management), soil management (soil drainage and irrigation, fertilisers, cover crops), disease and pest control, and winemaking techniques/traditions can optimise wine quality and regional distinctiveness (Jackson and Lombard, 1993; Van Leeuwen and

Seguin, 2006; Reynolds and Heuvel, 2009; Ross et al., 2009; Capozzi et al., 2015; White, 2015;

Zhang et al., 2017).

Terroir is also shaped based on the cultural practice and tradition that have maintained over several generations (and in some areas, hundreds of years). Terroir is a spatial, ecological, and cultural concept that links the human, their histories, their social organisations, and their agricultural practices

(Bérard and Marchenay, 2006). The deeply rooted traditional knowledge and technical practices that have profound influences on development and evolution of products and flavours are also valued to terroir (Trubek, 2008).

1.2 Microbial activities in agricultural production

1.2.1 Microbial activities in the field

1.2.1.1 Plant and soil microbiota

Plants are colonised by diverse microorganisms that affect plant health and growth in a beneficial, neutral, or harmful way (Schenk et al., 2012; Müller et al., 2016). The effect that most microbes have on plants appears to be neutral, that these microbes may utilise the plant-derived organic matter as substrates for energy production, and thus may still play important roles in nutrient cycling and modifying plant environments (Schenk et al., 2012). Beneficial microbes promote plant growth and/or suppress plant diseases through a variety of mechanisms including improved nutrient uptake,

3 production of growth regulators, and biosynthesis of compounds inhibiting pathogen, such as plant- growth-promoting bacteria (PGPB) and arbuscular mycorrhizal (AM) fungi that can improve phosphorus and nitrogen acquisition (Parniske, 2008; Glick, 2012). Much rarer, plant pathogenic bacteria, fungi, oomycetes, and viruses can cause diseases on plants (Schenk et al., 2012).

The plant microbiota does not assemble randomly but represents defined phylogenetic structures

(Müller et al., 2016). Bacteria dominate these communities, while fungi, oomycetes, archaea, and viruses are also important contributors (Müller et al., 2016). With the development of next-generation sequencing (NGS), culture-independent analyses have revealed that the communities on aboveground and belowground organs exhibit a defined taxonomic structure and are consistently dominated by a few phyla including Proteobacteria, Actinobacteria, Bacteroidetes, and Firmicutes (Müller et al.,

2016). These core taxa persist across plant compartments, plant species, and multiple environments

(Bulgarelli et al., 2012; Lundberg et al., 2012; Vandenkoornhuyse et al., 2015; Zarraonaindia et al.,

2015). Identifying the core microbiota, which is expected to play pivotal roles in organising the dynamics of resident microbiomes, is the key to optimising plant-microbe interactions thus increasing resource-efficiency and stress-resistance of agroecosystems (Toju et al., 2018).

To establish the microbiota, growing plants not only provide niches to invading microbes capable of exploiting the provided resources as stochastic colonization events, but plant-microbe coevolution might lead a plant-driven selection process, resulting in active recruitment of microbiota members that provide functions to the plant host (Müller et al., 2016). Meanwhile, community assembly is a dynamic succession over time in response to plant development and environmental changes. Many factors, including source reservoir (for example soil), environmental conditions (water availability, temperature, solar radiation), plant genotype, and microbe-microbe and plant microbe interactions, influence microbial community assembly during all stages of plant growth (Chaparro et al., 2014;

Copeland et al., 2015; Edwards et al., 2015; Grady et al., 2019).

4 The plant microbiota has both direct and indirect relationships with the host, from transforming the availability of organic matter and essential nutrients in the soil (such as nitrogen-fixing rhizobacteria and arbuscular mycorrhizal fungi), modifying the tolerance to environmental stresses (such as drought, flooding, high salinity), to preventing the growth and activity of plant pathogens through competition for niches and nutrients, antibiosis, production of lytic enzymes, inhibition of pathogen- derived enzymes or toxins, and induction of plant resistance (Mastretta et al., 2006; Gilbert et al.,

2014; Müller et al., 2016).

Soil microbiota plays an important role on the agricultural products from the field to the fork (Rillig et al., 2017). On the one hand, the soil microbiome colonises the rhizosphere or the root thus influencing plant physiology and metabolism to further influence product nutrient contents and/or secondary metabolites. On the other hand, soil microbes can colonise the harvested portion of the plant internally

(endophytes) or externally (epiphytes); that is, they are directly present in the products and potentially continue to affect products in the following processing (Rillig et al., 2017). This can occur, for instance, in wine grapes. Compant et al. (2005) demonstrated that plant-growth-promoting (PGP) bacteria Burkholderia sp. strain PsJN can endophytically colonise Chardonnay plantlets and translocate onto the root surfaces, root internal tissues, and finally the internode and leaf.

Zarraonaindia et al. (2015) has shown that soil microbiome is a reservoir for grape-associated bacteria and is present in the resulting grape must, and several pathways such as dust via tillage, rain splash, winds, and insects can assist the translocation process. After harvest, soil microbiota could affect the product throughout the agrifood chain, for example spoilage and diseases which can occur during storage (Rillig et al., 2017), and construct the microbial consortia for fermented food production, for example, wine (Liu et al., 2019).

1.2.1.2 Grapevine and the associated microbiota

Grapevine and vineyard is one example to understand the interplay among soil microbiota, plant microbiota, and plant growth, and crop production (reviewed in Chapter 2). The phyllosphere

5 harbours diverse microbes including yeasts, filamentous fungi and bacteria that substantially modulate grapevine health, growth, and grape quality (Barata et al., 2012; Gilbert et al., 2014). Grape- associated microbes are transferred to the must and fermentation and have a profound influence on wine composition, flavour and quality (Zarraonaindia and Gilbert, 2015; Morrison-Whittle and

Goddard, 2018).

Along the annual growth cycle, several distinct developmental stages have been identified, the principal ones being budbreak (budburst), flowering (anthesis), fruit set (berry setting), veraison

(berry colour change), and harvest (ripeness) (Coombe, 1992; Keller, 2020). Budbreak begins the new growing cycle in spring and signals the end of dormancy. Cell division and auxin production in buds initiate small shoots emerging and growth using the carbohydrate stored in roots and wood of the vine from the last growing season (Keller, 2020). Flowering precedes the fertilisation of vine flowers and the subsequent development into berries, the fruit set process (Coombe, 1992; Keller, 2020). Green berries enlarge rapidly in the following weeks, accumulating organic acids (mainly tartrate and malate) but little sugar (Coombe, 1992; Keller, 2020). Veraison marks the beginning of ripening and is a crucial time point of grapevine metabolism and growth. Besides colour change in the berries due to anthocyanin production and accumulation, berries soften as pectin and cellulose are degraded, acidity declines, and sugar accumulates (glucose and fructose), which creates a more favourable environment for microbial colonisation (Coombe, 1992; Renouf et al., 2005). The fully ripened grapes reach the harvest stage, with sugars, acids, and secondary metabolites delivering the aroma and flavour (Coombe, 1992; Keller, 2020). Studies have attempted to characterise grapevine-associated bacteria (bulk soil, rhizosphere, plant organs) and their relevant importance on plant growth (for example PGPB) and winemaking processes (Zarraonaindia et al., 2015; Marasco et al., 2018;

Mezzasalma et al., 2018). Zarraonaindia et al. (2015) investigated the spatial and temporal dynamics of bacteria associated with grapevines of Merlot cultivars, and revealed significant differences among plant organs (root, leaf, flower, grape) and the soil, but little influences from plant development.

While fungal communities across multiple compartments and their functioning have not been

6 systematically investigated for grapevines. Grapevines and the associated microbiota deserve future research to provide perspectives to vineyard management and quality wine production.

1.2.2 Fermentation

Microorganisms have been utilised by humans for their fermentative properties to preserve and transform food products for thousands of years. Fermented foods like wine, beer, bread, cheese, yoghourt, sauerkraut, chocolate, and meat products are widely consumed throughout the world for the increasing digestibility and the desirable flavour, aroma, and texture (Ross et al., 2002; Fleet, 2006).

The aroma and flavour are one of main characteristics that define differences of quality and styles among wines produced worldwide. The aroma and flavour profiles are dependent on multiple interacting factors including but not limited to grape must composition, precursors, fermentation temperature, oxygen level, and the present microbial consortia (Mauricio et al., 1997; Bell and

Henschke, 2005; Swiegers et al., 2009; Ciani et al., 2010; Dennis et al., 2012; Bokulich et al., 2016b).

1.2.2.1 Yeast and alcoholic fermentation for wine production

Yeasts are essential for alcoholic fermentation (AF) transforming grape must to wine, converting sugar to ethanol while producing a range of metabolites which are important to the final aroma and flavour and thus wine quality. Traditionally wine is fermented spontaneously by microorganisms naturally present in the grape surface with a diversity of non-Saccharomyces yeasts present in the early stages of fermentation, which decline and are replaced by Saccharomyces yeasts, mainly S. cerevisiae (Pretorius, 2000; Goddard, 2008), and is driven by their adaptive ability to engineer the environment through fermentation. While spontaneous fermentation can be unpredictable and risks the production of off-flavours in a stuck or sluggish fermentation, it is often preferred for the complex and unique sensory profiles derived from the indigenous yeast and thus the expression of local terroir

(Pretorius, 2000; Fleet, 2003). Commercially available cultured yeasts are individual strains of S. cerevisiae, are widely used to grape must to initiate and complete fermentation. Using these commercial strains in the winemaking process significantly reduces the proportion of indigenous

7 yeasts (Ciani et al., 2006). While improving reproducibility and predictability of the quality of the wine produced, there is a common perception that these wines are attractive with generic aromas and flavours but do not possess the complexity as spontaneously fermented wines (Rainieri and Pretorius,

2000). Studies show the diverse population of fermenting yeasts (both Saccharomyces and non-

Saccharomyces) are likely to sculpt the chemical diversity and distinctive flavours (Ciani et al., 2010;

Jolly et al., 2014; Winters et al., 2019).

Yeast-derived volatile metabolites give wine its characteristic aromas during and after fermentation.

Aroma active metabolites include higher alcohols, esters, carbonyls (for example aldehydes), volatile fatty acids, and sulfur compounds and more, are derived from sugar and amino acid metabolism

(Swiegers et al., 2005). These different compound classes contribute different characteristics to the organoleptic properties of wines, such as esters imparting fresh fruity aromas and flavours and higher alcohols delivering a strong, pungent smell and taste in wine (Swiegers et al., 2005; Swiegers and

Pretorius, 2005). Within the yeast community, yeast species possess different biosynthetic pathways to produce volatile compounds, and different ability in producing extracellular enzymes, such as esterases, glycosidases, lipases, β-glucosidases, proteases and cellulases that involved in hydrolysis of structural components, thereby resulting in diverse wine composition (Swiegers and Pretorius, 2005;

Capozzi et al., 2015). Meanwhile, the interactions between yeast species affect the fermentation performance and final wine aroma and flavour, including negative interactions, such as killer characters presented by some species and competition for oxygen and nutrients between species, positive interactions (mainly between non-Saccharomyces and S. cerevisiae), and neural interactions

(Liu et al., 2017).

1.2.2.2 Bacteria and malolactic fermentation for wine production

The malolactic fermentation (MLF), conducted by lactic acid bacteria (LAB), is the conversion of stronger malic acid to softer lactic acid. Lactobacillus, Leuconostoc, Oenococcus and Pediococcus are four genera of LAB present in wine, of which Oenococcus oeni is the best adapted wine associated

8 species and used almost exclusively for MLF in red, white, and sparkling winemaking (Wibowo et al.,

1985; Henick-Kling, 1993). When needed, MLF not only provides deacidification, but can enhance the aroma and flavour profiles. The increase of lactic acid in the wine results in a softer mouth-feel, and the metabolism of other organic acids (for example citric acid) during MLF can contribute to the volatile acidity (by production of acetic acid) of the wine (Swiegers et al., 2005). Beyond the modulation of acids, LAB can produce significant amounts of 2,3-butanedione (diacetyl) to contribute a desirable ‘buttery’ or ‘butterscotch’ flavour in the wine, and metabolise volatile compounds derived from yeasts (for example acetaldehyde) to affect wine aroma and flavour (Swiegers et al., 2005).

1.3 Microbial biogeography

Microbial biogeography is the study of microbial biodiversity over time and space. It offers insights to mechanisms that generate and maintain microbial diversity, such as speciation, extinction, dispersal and species interactions (Martiny et al., 2006). Genetic methodologies, for example NGS, have revealed that past culture-dependent studies missed a large portion of microbial diversity, and have allowed recent studies to sample deeper and wider diversity than ever before (Venter et al., 2004;

Floyd et al., 2005). A growing body of evidence shows that microorganisms vary in abundance, diversity, and distribution, over multiple taxonomic and spatial scales, uncovering the role of geospatial patterns play in environmental habitats (Gibbons et al., 2013; Bahram et al., 2018), agriculture (Peiffer et al., 2013; Bokulich et al., 2014), indoor environments (Adams et al., 2013;

Kembel et al., 2014), and human health (Costello et al., 2009; Koren et al., 2012), and revealing the important links between microbial communities, local environments, and macroscopic phenomena.

Further, a decline in microbial similarity with increasing spatial distance, known as a distance-decay relationship suggests not only that composition is differentiated based on location but also that this variation correlates with spatial distance (Green et al., 2004; Soininen et al., 2007; Hanson et al.,

2012; Miura et al., 2017). Many of these studies find correlations between assemblage composition and geographical or environmental features, such as latitude (Schwalbach et al., 2004), salinity (Bahl et al., 2011), and precipitation (Tedersoo et al., 2014). Global studies demonstrate that soil fungi and

9 bacteria show niche differentiation that is associated with contrasting diversity in response to climatic factors, in particular precipitation, edaphic properties (soil pH), and spatial patterning (Tedersoo et al.,

2014; Bahram et al., 2018). Researchers have explored the applicability of existing ecological theory of larger organisms (plants and animals) to microbial ecology to provide organisation, structure, mechanistic insights (Martiny et al., 2006; Prosser et al., 2007). Hanson et al. (2012) proposes that four processes, selection, drift, dispersal and mutation, create and maintain microbial biogeographic patterns on inseparable ecological and evolutionary scales, and the interplay of these processes drive the distance-decay relationship. The relative importance of these processes seems to be related to the scale of sampling. Drift is expected of greater importance for the distance effect at large spatial scales, for example between continents or distancing tens of thousands of kilometres. While at a small scale, biogeographic patterns might be due to dispersal limitation and selection related habitat types and environments (Martiny et al., 2006; Prosser et al., 2007).

In viticulture, microbial biogeography was first reported in S. cerevisiae populations and cultivable yeasts in New Zealand (Gayevskiy and Goddard, 2012). This concept was further extended by

Bokulich et al. (2014) to demonstrate regionally structured bacterial and fungal consortia in grape musts in California. Increasing evidence supports microbial biogeography related wine production has been reported worldwide and shows influences of multiple factors, such as climate and topography, soil properties, and local anthropogenic practices, on these patterns (reviewed in Chapter 2).

Consequently, understanding microbial biogeographic patterns and affecting factors would extend our knowledge about microbial contributions to shape regional characteristics of wine.

1.4 Objectives

The main aim of this thesis is to provide a comprehensive understanding of microbial ecology and community patterns associated with grapevine and wine fermentation and investigate how these patterns contribute to wine quality and styles.

More specifically, the objectives of this thesis are to:

10 ● Based on the previous research of wine microbial biogeography conducted in other countries

and winegrowing regions of the world, investigate the distribution patterns of bacteria and

fungi of vineyard soil, grape must, and wine ferments at regional scale in southern Australia,

and in particular the role of fungi in expressing regional characteristics of wine (Chapter 3)

● Elucidate the fungal ecology in the vineyard during annual growth cycle and their

associations with plant development and the changing environment (Chapter 4)

● Investigate how fungal communities in grape must/juice, yeast populations, and S. cerevisiae

populations drive wine aroma profiles at a sub-regional scale, and the central role of S.

cerevisiae in wine microbial biogeography (Chapter 5)

1.5 Thesis overview

This thesis comprises six chapters in total. In this first chapter I have provided general background for the research performed in this thesis and outlined main objectives. The aims of subsequent chapters are detailed below.

1.5.1 Chapter 2: “From the vineyard to the winery: how microbial ecology drives regional distinctiveness of wine”

This chapter reviews the recent literature surrounding wine-related microbiome conducted using high throughput culture-independent technologies. By discussing microbial biogeography, climate, soil, and anthropogenic practices, I detail available approaches for the microbial aspect of wine terroir, and describe the knowledge gaps, which underpin the direction of this thesis.

1.5.2 Chapter 3: “The fungal microbiome is an important component of vineyard ecosystems and correlates with regional distinctiveness of wine”

This chapter addresses the first objective and investigates microbial patterns at large scale, sampling both vineyard soil and grape must/ferments from six distinct winegrowing regions in southern

Australia.

The aims of this chapter are to:

● Test the hypothesis that bacteria and fungi show distinctive community on the basis of region

● Correlate regional patterns with local environments, that are soil properties and weather

● Evaluate the contribution of biotic factors (soil and must microbiota) and abiotic factors to

wine regional characteristics

● Propose a mechanism by which fungi can translocate from the soil to the grape through the

vine

1.5.3 Chapter 4: “Community succession of the grapevine fungal microbiome in the annual growth cycle”

This chapter addresses the second objective and describes the dynamics of fungal communities associated with the grapevine (root zone soil, roots, leaves, grapes/flowers) over the annual growing season.

The aims of this chapter are to:

● Investigate the fungal community distribution patterns over space, that is different

microhabitats and geographic locations

● Investigate the fungal community succession according to plant development (flowering, fruit

set, veraison, harvest)

● Define and characterise the core microbiota associated with the grapevine

● Investigate the role of the core microbiota in seasonal community succession

1.5.4 Chapter 5: “Diversity and dynamics of microbes during spontaneous fermentation and its associations with unique aroma profiles in wine”

This chapter addresses the third objective and investigates fungal patterns at small sub-regional scale, combining culture-independent sequencing and culture experiments, and fungal contributions to wine aroma profiles.

The aims of this chapter are to:

● Test the hypothesis that fungal community, yeast populations, and S. cerevisiae populations

show distribution patterns based on the geographic location and grape variety

● Elucidate the diversity and dynamics of fungal microbiota during spontaneous fermentation

● Quantify the contributions of multiple layers of fungal microbiota to the resultant wine

1.5.5 Chapter 6: “General discussion and conclusion”

Here I summarise major findings of this thesis corresponding to the objectives and aims outlined above. I discuss how these findings will contribute to the field of microbial biogeography, microbial ecology, and winemaking practices. I also discuss the limitations of the study and end this thesis with future directions for further research.

13 1.6 References

Adams, R.I., Miletto, M., Taylor, J.W., and Bruns, T.D. (2013) Dispersal in microbes: fungi in indoor air are dominated by outdoor air and show dispersal limitation at short distances. The ISME journal 7: 1262-1273.

Almeida, C.M.R., and Vasconcelos, M.T.S. (2003) Multielement composition of wines and their precursors including provenance soil and their potentialities as fingerprints of wine origin. J Agric Food Chem 51: 4788-4798.

Bahl, J., Lau, M.C., Smith, G.J., Vijaykrishna, D., Cary, S.C., Lacap, D.C. et al. (2011) Ancient origins determine global biogeography of hot and cold desert cyanobacteria. Nature communications 2: 1-6.

Bahram, M., Hildebrand, F., Forslund, S.K., Anderson, J.L., Soudzilovskaia, N.A., Bodegom, P.M. et al. (2018) Structure and function of the global topsoil microbiome. Nature 560: 233-237.

Barata, A., Malfeito-Ferreira, M., and Loureiro, V. (2012) The microbial ecology of wine grape berries. Int J Food Microbiol 153: 243-259.

Barham, E. (2003) Translating terroir: the global challenge of French AOC labeling. Journal of rural studies 19: 127-138.

Bell, S.J., and Henschke, P.A. (2005) Implications of nitrogen nutrition for grapes, fermentation and wine. Aust J Grape Wine Res 11: 242-295.

Bérard, L., and Marchenay, P. (2006) Local products and geographical indications: taking account of local knowledge and biodiversity. International Social Science Journal 58: 109-116.

Bokulich, N.A., Thorngate, J.H., Richardson, P.M., and Mills, D.A. (2014) Microbial biogeography of wine grapes is conditioned by cultivar, vintage, and climate. Proc Natl Acad Sci USA 111: E139- E148.

Bokulich, N.A., Lewis, Z.T., Boundy-Mills, K., and Mills, D.A. (2016a) A new perspective on microbial landscapes within food production. Curr Opin Biotechnol 37: 182-189.

Bokulich, N.A., Collins, T.S., Masarweh, C., Allen, G., Heymann, H., Ebeler, S.E., and Mills, D.A. (2016b) Associations among Wine Grape Microbiome, Metabolome, and Fermentation Behavior Suggest Microbial Contribution to Regional Wine Characteristics. MBio 7: e00631-00616.

14 Bulgarelli, D., Rott, M., Schlaeppi, K., van Themaat, E.V.L., Ahmadinejad, N., Assenza, F. et al. (2012) Revealing structure and assembly cues for Arabidopsis root-inhabiting bacterial microbiota. Nature 488: 91-95.

Capozzi, V., Garofalo, C., Chiriatti, M.A., Grieco, F., and Spano, G. (2015) Microbial terroir and food innovation: the case of yeast biodiversity in wine. Microbiol Res 181: 75-83.

Chaparro, J.M., Badri, D.V., and Vivanco, J.M. (2014) Rhizosphere microbiome assemblage is affected by plant development. The ISME journal 8: 790.

Ciani, M., Beco, L., and Comitini, F. (2006) Fermentation behaviour and metabolic interactions of multistarter wine yeast fermentations. Int J Food Microbiol 108: 239-245.

Ciani, M., Comitini, F., Mannazzu, I., and Domizio, P. (2010) Controlled mixed culture fermentation: a new perspective on the use of non‐Saccharomyces yeasts in winemaking. FEMS Yeast Res 10: 123- 133.

Compant, S., Reiter, B., Sessitsch, A., Nowak, J., Clément, C., and Barka, E.A. (2005) Endophytic colonization of Vitis vinifera L. by plant growth-promoting bacterium Burkholderia sp. strain PsJN. Appl Environ Microbiol 71: 1685-1693.

Coombe, B. (1992) Research on development and ripening of the grape berry. Am J Enol Vitic 43: 101-110.

Copeland, J.K., Yuan, L., Layeghifard, M., Wang, P.W., and Guttman, D.S. (2015) Seasonal community succession of the phyllosphere microbiome. Mol Plant-Microbe Interact 28: 274-285.

Costello, E.K., Lauber, C.L., Hamady, M., Fierer, N., Gordon, J.I., and Knight, R. (2009) Bacterial community variation in human body habitats across space and time. Science 326: 1694-1697.

Dennis, E.G., Keyzers, R.A., Kalua, C.M., Maffei, S.M., Nicholson, E.L., and Boss, P.K. (2012) Grape contribution to wine aroma: Production of hexyl acetate, octyl acetate, and benzyl acetate during yeast fermentation is dependent upon precursors in the must. J Agric Food Chem 60: 2638- 2646.

Edwards, J., Johnson, C., Santos-Medellín, C., Lurie, E., Podishetty, N.K., Bhatnagar, S. et al. (2015) Structure, variation, and assembly of the root-associated microbiomes of rice. Proc Natl Acad Sci USA 112: E911-E920.

Fleet, G. (2006) The Commercial and Community Significance of Yeasts in Food and Beverage Production. In Yeasts in Food and Beverages. Querol, A., and Fleet, G. (eds). Berlin, Heidelberg: Springer Berlin Heidelberg, pp. 1-12.

15 Fleet, G.H. (2003) Yeast interactions and wine flavour. Int J Food Microbiol 86: 11-22.

Floyd, M.M., Tang, J., Kane, M., and Emerson, D. (2005) Captured diversity in a culture collection: case study of the geographic and habitat distributions of environmental isolates held at the American Type Culture Collection. Appl Environ Microbiol 71: 2813-2823.

Fotakis, C., Kokkotou, K., Zoumpoulakis, P., and Zervou, M. (2013) NMR metabolite fingerprinting in grape derived products: An overview. Food Res Int 54: 1184-1194.

Gayevskiy, V., and Goddard, M.R. (2012) Geographic delineations of yeast communities and populations associated with vines and wines in New Zealand. The ISME journal 6: 1281-1290.

Gibbons, S.M., Caporaso, J.G., Pirrung, M., Field, D., Knight, R., and Gilbert, J.A. (2013) Evidence for a persistent microbial seed bank throughout the global ocean. Proc Natl Acad Sci USA 110: 4651- 4655.

Gilbert, J.A., van der Lelie, D., and Zarraonaindia, I. (2014) Microbial terroir for wine grapes. Proc Natl Acad Sci USA 111: 5-6.

Giménez-Miralles, J.E., Salazar, D.M., and Solana, I. (1999) Regional origin assignment of red wines from Valencia (Spain) by 2H NMR and 13C IRMS stable isotope analysis of fermentative ethanol. J Agric Food Chem 47: 2645-2652.

Gladstones, J. (2011) Wine, terroir and climate change: Wakefield Press.

Glick, B.R. (2012) Plant growth-promoting bacteria: mechanisms and applications. Scientifica 2012.

Goddard, M.R. (2008) Quantifying the complexities of Saccharomyces cerevisiae's ecosystem engineering via fermentation. Ecology 89: 2077-2082.

Gonzálvez, A., Llorens, A., Cervera, M., Armenta, S., and De la Guardia, M. (2009) Elemental fingerprint of wines from the protected designation of origin Valencia. Food Chem 112: 26-34.

Grady, K.L., Sorensen, J.W., Stopnisek, N., Guittar, J., and Shade, A. (2019) Assembly and seasonality of core phyllosphere microbiota on perennial biofuel crops. Nature Communications 10: 4135.

Green, J.L., Holmes, A.J., Westoby, M., Oliver, I., Briscoe, D., Dangerfield, M. et al. (2004) Spatial scaling of microbial eukaryote diversity. Nature 432: 747-750.

Hanson, C.A., Fuhrman, J.A., Horner-Devine, M.C., and Martiny, J.B. (2012) Beyond biogeographic patterns: processes shaping the microbial landscape. Nature Reviews Microbiology 10: 497.

16 Henick-Kling, T. (1993) Malolactic fermentation. Wine microbiology and biotechnology: 289-326.

Jackson, D., and Lombard, P. (1993) Environmental and management practices affecting grape composition and wine quality-a review. Am J Enol Vitic 44: 409-430.

Jolly, N.P., Varela, C., and Pretorius, I.S. (2014) Not your ordinary yeast: non-Saccharomyces yeasts in wine production uncovered. FEMS Yeast Res 14: 215-237.

Keller, M. (2020) The science of grapevines: Academic Press.

Kembel, S.W., Meadow, J.F., O’Connor, T.K., Mhuireach, G., Northcutt, D., Kline, J. et al. (2014) Architectural design drives the biogeography of indoor bacterial communities. PLoS One 9: e87093.

Kirchman, D.L. (2018) Processes in microbial ecology. New York, USA: Oxford University Press.

Kment, P., Mihaljevič, M., Ettler, V., Šebek, O., Strnad, L., and Rohlová, L. (2005) Differentiation of Czech wines using multielement composition–A comparison with vineyard soil. Food Chem 91: 157- 165.

Koren, O., Goodrich, J.K., Cullender, T.C., Spor, A., Laitinen, K., Bäckhed, H.K. et al. (2012) Host remodeling of the gut microbiome and metabolic changes during pregnancy. Cell 150: 470-480.

Liu, D., Zhang, P., Chen, D., and Howell, K.S. (2019) From the vineyard to the winery: how microbial ecology drives regional distinctiveness of wine. Front Microbiol 10: 2679.

Liu, Y., Rousseaux, S., Tourdot-Maréchal, R., Sadoudi, M., Gougeon, R., Schmitt-Kopplin, P., and Alexandre, H. (2017) Wine microbiome: a dynamic world of microbial interactions. Crit Rev Food Sci Nutr 57: 856-873.

Lundberg, D.S., Lebeis, S.L., Paredes, S.H., Yourstone, S., Gehring, J., Malfatti, S. et al. (2012) Defining the core Arabidopsis thaliana root microbiome. Nature 488: 86.

Marasco, R., Rolli, E., Fusi, M., Michoud, G., and Daffonchio, D. (2018) Grapevine rootstocks shape underground bacterial microbiome and networking but not potential functionality. Microbiome 6: 3.

Martiny, J.B.H., Bohannan, B.J., Brown, J.H., Colwell, R.K., Fuhrman, J.A., Green, J.L. et al. (2006) Microbial biogeography: putting microorganisms on the map. Nature Reviews Microbiology 4: 102.

Mastretta, C., Barac, T., Vangronsveld, J., Newman, L., Taghavi, S., and Lelie, D.v.d. (2006) Endophytic bacteria and their potential application to improve the phytoremediation of contaminated environments. Biotechnol Genet Eng Rev 23: 175-188.

17 Mauricio, J.C., Moreno, J., Zea, L., Ortega, J.M., and Medina, M. (1997) The effects of grape must fermentation conditions on volatile alcohols and esters formed by Saccharomyces cerevisiae. J Sci Food Agric 75: 155-160.

Mezzasalma, V., Sandionigi, A., Guzzetti, L., Galimberti, A., Grando, M.S., Tardaguila, J., and Labra, M. (2018) Geographical and cultivar features differentiate grape microbiota in Northern Italy and Spain vineyards. Front Microbiol 9: 946.

Miura, T., Sánchez, R., Castañeda, L.E., Godoy, K., and Barbosa, O. (2017) Is microbial terroir related to geographic distance between vineyards? Environ Microbiol Rep 9: 742-749.

Morrison-Whittle, P., and Goddard, M.R. (2018) From vineyard to winery: a source map of microbial diversity driving wine fermentation. Environ Microbiol 20: 75-84.

Müller, D.B., Vogel, C., Bai, Y., and Vorholt, J.A. (2016) The plant microbiota: systems-level insights and perspectives. Annu Rev Genet 50: 211-234.

OIV (2010) Definition of Vitivinicultural "Terroir". The General Director of the OIV, General assembly Tbilisi (Georgia) 25: 1.

Parniske, M. (2008) Arbuscular mycorrhiza: the mother of plant root endosymbioses. Nature Reviews Microbiology 6: 763-775.

Peiffer, J.A., Spor, A., Koren, O., Jin, Z., Tringe, S.G., Dangl, J.L. et al. (2013) Diversity and heritability of the maize rhizosphere microbiome under field conditions. Proc Natl Acad Sci USA 110: 6548-6553.

Pepi, S., and Vaccaro, C. (2017) Geochemical fingerprints of “Prosecco” wine based on major and trace elements. Environ Geochem Health: 1-15.

Pereira, G.E., Gaudillere, J.-P., Van Leeuwen, C., Hilbert, G., Lavialle, O., Maucourt, M. et al. (2005) 1H NMR and chemometrics to characterize mature grape berries in four wine-growing areas in Bordeaux, France. J Agric Food Chem 53: 6382-6389.

Pretorius, I.S. (2000) Tailoring wine yeast for the new millennium: novel approaches to the ancient art of winemaking. Yeast 16: 675-729.

Prosser, J.I., Bohannan, B.J., Curtis, T.P., Ellis, R.J., Firestone, M.K., Freckleton, R.P. et al. (2007) The role of ecological theory in microbial ecology. Nature Reviews Microbiology 5: 384-392.

Rainieri, S., and Pretorius, I. (2000) Selection and improvement of wine yeasts. Ann Microbiol 50: 15- 32.

18 Renouf, V., Claisse, O., and Lonvaud‐Funel, A. (2005) Understanding the microbial ecosystem on the grape berry surface through numeration and identification of yeast and bacteria. Aust J Grape Wine Res 11: 316-327.

Reynolds, A.G., and Heuvel, J.E.V. (2009) Influence of Grapevine Training Systems on Vine Growth and Fruit Composition: A Review. Am J Enol Vitic 60: 251-268.

Rillig, M.C., Lehmann, A., Lehmann, J., Camenzind, T., and Rauh, C. (2017) Soil Biodiversity Effects from Field to Fork. Trends Plant Sci.

Ross, C.F., Weller, K.M., Blue, R.B., and Reganold, J.P. (2009) Difference testing of Merlot produced from biodynamically and organically grown wine grapes. Journal of Wine Research 20: 85- 94.

Ross, R.P., Morgan, S., and Hill, C. (2002) Preservation and fermentation: past, present and future. Int J Food Microbiol 79: 3-16.

Schenk, P.M., Carvalhais, L.C., and Kazan, K. (2012) Unraveling plant–microbe interactions: can multi-species transcriptomics help? Trends Biotechnol 30: 177-184.

Schwalbach, M.S., Hewson, I., and Fuhrman, J.A. (2004) Viral effects on bacterial community composition in marine plankton microcosms. Aquat Microb Ecol 34: 117-127.

Soininen, J., McDonald, R., and Hillebrand, H. (2007) The distance decay of similarity in ecological communities. Ecography 30: 3-12.

Spayd, S., Wample, R., Evans, R., Stevens, R., Seymour, B., and Nagel, C. (1994) Nitrogen fertilization of white Riesling grapes in Washington. Must and wine composition. Am J Enol Vitic 45: 34-42.

Swiegers, J., Bartowsky, E., Henschke, P., and Pretorius, I. (2005) Yeast and bacterial modulation of wine aroma and flavour. Aust J Grape Wine Res 11: 139-173.

Swiegers, J.H., and Pretorius, I.S. (2005) Yeast modulation of wine flavor. Adv Appl Microbiol 57: 131-175.

Swiegers, J.H., Kievit, R.L., Siebert, T., Lattey, K.A., Bramley, B.R., Francis, I.L. et al. (2009) The influence of yeast on the aroma of Sauvignon Blanc wine. Food Microbiol 26: 204-211.

Tedersoo, L., Bahram, M., Põlme, S., Kõljalg, U., Yorou, N.S., Wijesundera, R. et al. (2014) Global diversity and geography of soil fungi. Science 346: 1256688.

19 Toju, H., Peay, K.G., Yamamichi, M., Narisawa, K., Hiruma, K., Naito, K. et al. (2018) Core microbiomes for sustainable agroecosystems. Nature Plants 4: 247-257.

Trubek, A.B. (2008) The taste of place: A cultural journey into terroir: Univ of California Press.

Van Leeuwen, C., and Seguin, G. (2006) The concept of terroir in viticulture. Journal of Wine Research 17: 1-10.

Vandenkoornhuyse, P., Quaiser, A., Duhamel, M., Le Van, A., and Dufresne, A. (2015) The importance of the microbiome of the plant holobiont. New Phytol 206: 1196-1206.

Venter, J.C., Remington, K., Heidelberg, J.F., Halpern, A.L., Rusch, D., Eisen, J.A. et al. (2004) Environmental genome shotgun sequencing of the Sargasso Sea. Science 304: 66-74.

White, R.E. (2015) Understanding vineyard soils. New York, USA: Oxford University Press.

Wibowo, D., Eschenbruch, R., Davis, C., Fleet, G., and Lee, T. (1985) Occurrence and growth of lactic acid bacteria in wine: a review. Am J Enol Vitic 36: 302-313.

Winters, M., Panayotides, D., Bayrak, M., Rémont, G., Gonzalez Viejo, C., Liu, D. et al. (2019) Defined co‐cultures of yeast and bacteria modify the aroma, crumb and sensory properties of bread. J Appl Microbiol 127: 778-793.

Zarraonaindia, I., and Gilbert, J.A. (2015) Understanding grapevine-microbiome interactions: implications for viticulture industry. Microbial Cell 2: 171.

Zarraonaindia, I., Owens, S.M., Weisenhorn, P., West, K., Hampton-Marcell, J., Lax, S. et al. (2015) The soil microbiome influences grapevine-associated microbiota. MBio 6: e02527-02514.

Zhang, P., Wu, X., Needs, S., Liu, D., Fuentes, S., and Howell, K. (2017) The Influence of Apical and Basal Defoliation on the Canopy Structure and Biochemical Composition of Vitis vinifera cv. Shiraz Grapes and Wine. Frontiers in Chemistry 5: 48.

20 CHAPTER TWO

From the vineyard to the winery: how microbial ecology drives regional distinctiveness of wine

Status of chapter:

This chapter is a review article published in Frontiers in Microbiology journal. The co-authorship form is presented at the start of this thesis, after the acknowledgements.

Liu, D., Zhang, P., Chen, D., and Howell, K.S. (2019). From the vineyard to the winery: how microbial ecology drives regional distinctiveness of wine. Frontiers in Microbiology 10: 2679.

21 fmicb-10-02679 November 20, 2019 Time: 15:49 # 1

REVIEW published: 20 November 2019 doi: 10.3389/fmicb.2019.02679

From the Vineyard to the Winery: How Microbial Ecology Drives Regional Distinctiveness of Wine

Di Liu, Pangzhen Zhang, Deli Chen and Kate Howell*

School of Agriculture and Food, Faculty of Veterinary and Agricultural Sciences, The University of Melbourne, Melbourne, VIC, Australia

Wine production is a complex process from the vineyard to the winery. On this journey, microbes play a decisive role. From the environment where the vines grow, encompassing soil, topography, weather and climate through to management practices in vineyards, the microbes present can potentially change the composition of wine. Introduction of grapes into the winery and the start of winemaking processes modify microbial communities further. Recent advances in next-generation sequencing (NGS) technology have progressed our understanding of microbial communities associated with grapes and fermentations. We now have a ﬁner appreciation of microbial diversity across wine producing regions to begin to understand how diversity can contribute to wine quality and style characteristics. In this review, we highlight literature surrounding wine-related microorganisms and how these affect factors interact with and shape Edited by: microbial communities and contribute to wine quality. By discussing the geography, Giovanna Suzzi, University of Teramo, Italy climate and soil of environments and viticulture and winemaking practices, we claim Reviewed by: microbial biogeography as a new perspective to impact wine quality and regionality. Giuseppe Spano, Depending on geospatial scales, habitats, and taxa, the microbial community respond University of Foggia, Italy to local conditions. We discuss the effect of a changing climate on local conditions and Florian F. Bauer, Stellenbosch University, South Africa how this may alter microbial diversity and thus wine style. With increasing understanding *Correspondence: of microbial diversity and their effects on wine fermentation, wine production can be Kate Howell optimised with enhancing the expression of regional characteristics by understanding [email protected] and managing the microbes present.

Specialty section: Keywords: wine quality, microbial biogeography, climate, soil, bacteria, fungi This article was submitted to Food Microbiology, a section of the journal INTRODUCTION Frontiers in Microbiology Received: 16 September 2019 Wine production is a global billion-dollar industry for which regionally distinct wine Accepted: 05 November 2019 characteristics, collectively known as “terroir,” are an important collection of traits that wine Published: 20 November 2019 industry would like to define. Wines made from the same grape cultivar but grown in different Citation: regions are appreciated for their distinctive features (Van Leeuwen and Seguin, 2006). Regionality Liu D, Zhang P, Chen D and can be identified with chemical compositions and sensory properties (Pereira et al., 2005; Gonzálvez Howell K (2019) From the Vineyard to the Winery: How Microbial Ecology et al., 2009; Robinson et al., 2012), and are likely to be ascribed to local environmental parameters Drives Regional Distinctiveness that influence grapevine growth and development, such as soil types, climate, topography and of Wine. Front. Microbiol. 10:2679. human management, but the mechanisms by which these factors affect wine composition remain doi: 10.3389/fmicb.2019.02679 tenuous (Van Leeuwen and Seguin, 2006; Gladstones, 2011; Vaudour et al., 2015).

Frontiers in Microbiology | www.frontiersin.org 221 November 2019 | Volume 10 | Article 2679 fmicb-10-02679 November 20, 2019 Time: 15:49 # 2

Liu et al. Regional Traits of Wine From Microbes

From the vineyard to the winery, microorganisms play MICROBIAL BIOGEOGRAPHY: A key roles in wine production and quality. The grapevine MICROBIAL INSIGHT TO UNDERSTAND (Vitis vinifera) phyllosphere harbours diverse microbes including yeasts, filamentous fungi and bacteria that VARIATION IN WINE QUALITY substantially modulate grapevine health, growth, and grape Global studies of microbial biogeography have shown that and wine production (Figure 1; Barata et al., 2012; Gilbert distinct microbial populations are present in soil, bodies of water et al., 2014). Microbes could originate from the vineyard and ocean biomes and associated with plants (Martiny et al., soil (Zarraonaindia et al., 2015; Morrison-Whittle and 2006). Soil fungi and bacteria show global niche differentiation Goddard, 2018), air, precipitation (rainfall, hail, snow), be that is associated with contrasting diversity responses to transported by animal vectors (bees, insects, and birds) environmental filtering (for example by precipitation, soil (Goddard et al., 2010; Francesca et al., 2012; Stefanini et al., pH) and biotic interactions (Bahram et al., 2018). Culture- 2012; Lam and Howell, 2015), and be resident in nearby based microbiological methods revealed only a small portion native forests (Figures 1, 2; Knight and Goddard, 2015; of the diversity of environments (Martiny et al., 2006), and Morrison-Whittle and Goddard, 2018). Microbes that are so previous studies on the vineyard and wine microbiome grapevine-associated and are transferred to the must have do not reveal regional boundary restrained patterns (Barata a profound influence on wine composition, flavour and et al., 2012). New technologies such as NGS benefit from quality (Figure 1; Barata et al., 2012). Fermentative yeasts being culture-independent and can therefore reveal a distinct (primarily Saccharomyces cerevisiae) and lactic acid bacteria scenario of microbial biogeography. We propose that geographic (LAB, predominantly Oenococcus oeni) in the must modulate patterns favour fungi- or bacteria-driven metabolites and thus the flavour and aroma of wine (Swiegers et al., 2005). Beyond contribute to wine composition and quality. Some key wine these species, many microorganisms in the must could release microbial biogeography studies are listed in Table 1, where metabolites changing the chemical environment and/or we highlight the influence of climate, soil and anthropogenic fermentation processes and thus affect wine compositions and practices (a comparative list of these studies is given in characteristics, for example, release of inhibitory molecules Supplementary Table S1). altering Saccharomyces metabolism (Swiegers et al., 2005; Bokulich et al., 2016). Increasing evidence supports a microbial aspect to wine Microbiota-Metabolome Geographic regionality may be due, in part, to regionally structured Patterns to Elucidate Wine Regionality microbial communities, or microbial biogeography. Advanced Geographic delineations of S. cerevisiae populations and genetic-based methodologies, in particular next-generation cultivable yeasts were first reported in New Zealand vineyards, sequencing (NGS), have allowed researchers to sample microbial providing evidence for regional distribution of yeast populations diversity more deeply and widely and encouraged more (Table 1; Gayevskiy and Goddard, 2012). In the United States, comprehensive biogeographical surveys. This has marked the Bokulich et al. (2014) used NGS of 16S rRNA and internal beginning of a new era of information on the grapevine- transcribed spacer (ITS) ribosomal sequence to demonstrate associated microbiome across multiple scales (region, vineyard, regionally structured bacterial and fungal consortia in grape and vine) to elucidate wine quality and regional variation musts, with some influences from cultivar and vintage (Table 1). (Bokulich et al., 2014; Taylor et al., 2014; Burns et al., These studies posit that microbial biogeography is a contributor 2015; Morrison-Whittle and Goddard, 2018). This review to wine regionality expression. Further biogeographical aims to disentangle the role of microbial biogeography correlations between wine-related microbiota and regional on wine production, by considering how microbes interact origins have been reported, holding for newly planted or with environmental conditions and thus drive wine quality older vineyards and are representative from vineyards planted and style. around the world (Tofalo et al., 2014; Pinto et al., 2015; We commence our discussion with a review of current Wang et al., 2015; Bokulich et al., 2016; Capece et al., 2016; knowledge on microbial geographic patterns and how they Garofalo et al., 2016; Marzano et al., 2016; Portillo et al., differ from the vineyard to the winery, and then describe how 2016; El Khoury et al., 2017; Mezzasalma et al., 2017, 2018; climate, soil, and anthropogenic practices can affect microbial Singh et al., 2018; Vitulo et al., 2019). Notably, the microbial communities through the winemaking process to the finished geographic diversification in the must weakens as fermentation wine. The concept of scale is crucial to define microbial processes, due to S. cerevisiae yeasts dominance breaks the biogeography, as large-scale geographic and climatic features community diversity (Morrison-Whittle and Goddard, 2018). significantly affect microbial communities, but at smaller scales In addition, S. cerevisiae can persist perennially in a particular these differences may not be apparent. We propose that vineyard or winery within one single region, thereby enabling microbial biogeography gives a theoretical basis to wine terroir, wine style consistency between vintages (Börlin et al., 2016; which further provides information to the industry to produce Guzzon et al., 2018). distinctive and quality wines through microbial manipulation. The geographic diversification observed in microbiota has This is particularly important as wine styles are altered by been verified in wine chemical attributes (Knight et al., 2015; weather patterns and we conclude by considering how microbial Bokulich et al., 2016). Knight et al. (2015) empirically showed that biogeography and activity may respond to changing climates. regionally differentiated S. cerevisiae populations drove different

Frontiers in Microbiology | www.frontiersin.org 232 November 2019 | Volume 10 | Article 2679 fmicb-10-02679 November 20, 2019 Time: 15:49 # 3

Liu et al. Regional Traits of Wine From Microbes

FIGURE 1 | Overview of the wine-related microbiota from the vineyard to the winery. Microbiota associated with grapevine phyllosphere, especially grapes, can enter musts and constitute wine microbial consortium, in which fermentative yeasts and LAB conduct alcoholic and malolactic fermentation, respectively. The rhizosphere harbours diverse microbes that can beneﬁt plants by enabling nutrient uptake and tolerance to (a)biotic stress. Soil borne microorganisms might translocate to the phyllosphere internally (endophytes) or externally (epiphytes), thereby entering wine fermentation. Viticulture practices, for example fertilisers/compost addition, can modify soil microbiota via shifting nutrient pools or adding manure borne microorganisms (created with BioRender, https://app.biorender.com/).

metabolites in the resultant wines (Table 1; Knight et al., 2015). of kilometres (Gayevskiy and Goddard, 2012; Bokulich et al., Further, microbial patterns correlating with regional metabolite 2014; Taylor et al., 2014; Knight and Goddard, 2015) or be profiles were reported by Bokulich et al. (2016), and showed the quite a small geographic area (Pinto et al., 2015). When importance of the fermentative yeasts (for example, S. cerevisiae, comparing microbial communities in smaller geographical scales Hanseniaspora, Pichia) and LAB (Leuconostocaceae). These (at the scale of individual vineyards), the geographical patterns correlations between microbiome and metabolome can be used among populations are more evident for fungi than for bacteria to predict wine compositions and styles with their microbial (Bokulich et al., 2014; Miura et al., 2017). The leaf and grape patterns and deserve future study. For example, exploration of fungal community dissimilarities between sampling sites increase non-Saccharomyces yeasts that enhance wine aroma complexity, as geographic distance increases (Miura et al., 2017). the importance of microbial diversity in early fermentation, Non-mobile yeasts require animals (insects and birds) to and the likelihood of significant interactions between yeasts and be transferred across regions (Figure 2; Francesca et al., 2012; bacteria will held unpick mechanisms to explain the correlations Stefanini et al., 2012; Lam and Howell, 2015), animal vectors described in the studies above. are one potential biotic factor shaping the yeast and fungal community dissimilarities. When the studied scale is a small area, more factors maybe involved in the uniqueness of a site. Geospatial Scales Shape Geographic Grape varieties and clones exert marked impact on grape surface Diversification of Microbes bacteria within vineyard. For example, Bacteroidetes, Chloroflexi, The biogeographic patterns of microbes in the wine environment Acidobacteria, and Planctomycetes are clone- specific phyla are generally based on regional scales which are interpreted alongside the prevalent phylum Cyanobacteria, Proteobacteria, differently within each winegrowing country. This means that and Firmicutes (Zhang et al., 2019). This is in contrast with the notion of a “region” is not strictly defined and varies findings by Bokulich et al. (2014) who showed that the cultivar- considerably. For example, a wine region can describe an specific influence on microbial diversity is weak in larger scales. association of vineyards spanning hundreds or even thousands While Portillo et al. (2016) reported higher variability of bacteria

Frontiers in Microbiology | www.frontiersin.org 243 November 2019 | Volume 10 | Article 2679 fmicb-10-02679 November 20, 2019 Time: 15:49 # 4

Liu et al. Regional Traits of Wine From Microbes

FIGURE 2 | A scenario of wine microbial biogeography. Grapevine-associated microbiota originates from the local ecosystem encompassing soil, air, precipitation, native forests, etc. Genetic isolation is one driver of the geographic pattern that long distance decreases the gene flow that depends on physical forces and animal vectors (e.g., insects and birds). Climate is a profound environmental element shaping the microbial geographic pattern and thus affects wine quality. Macroclimate exerts influences on the regional pattern of bacteria and fungi. Mesoclimate at the vineyard scale shows weaker influences on the microbial distribution, especially for bacteria. Microclimate within the grapevine, modified by canopy management, may influence associated microbiota, this still remains to be shown (created with BioRender, https://app.biorender.com/).

between vineyards compared to intra-vineyard, Setati et al. (2012) more similar to one another than those from a larger geographic demonstrated that greater intra-vineyard variation was evident areas (Martínez et al., 2007; Miura et al., 2017), although than inter-vineyard based on fungal populations, including yeasts whether microbial patterns present a distance-dependant model (Table 1). In these cases, microclimate may play a more important is not clear. At the same time, adaptation to local environments role in structuring fungal communities (details in the section influences how microbial ecology develops and diversifies “Microclimate”), but this supposition requires further empirical (Martiny et al., 2006). For example, S. cerevisiae shows studies for confirmation. remarkable ability to adapt and thrive in human-associated food Microbial biogeography is indelibly shaped by genetic fermentative ecosystems (Legras et al., 2018). S. cerevisiae could isolation, as gene flow decreases with longer distances and be dispersed within, and successfully co-exist with other yeasts depends on vectors (Martiny et al., 2006; Kuehne et al., 2007). over a very small scale (10–200 m from the winery) (Valero Likewise, wine-related microbiota from relatively small scales are et al., 2005). How physical environments modulate wine-related

Frontiers in Microbiology | www.frontiersin.org 254 November 2019 | Volume 10 | Article 2679 fmicb-10-02679 November 20, 2019 Time: 15:49 # 5

Liu et al. Regional Traits of Wine From Microbes

TABLE 1 | Recent ﬁndings on wine microbial ecology from the vineyard to the winery.

Microorganisms Scale Habitat Methodology Major Conclusions References

Yeasts Three regions Grape juice Culture-dependent method, (i) Regional delineations were found on Gayevskiy and ITS-RFLP and D1/D2 26S yeast communities and S. cerevisiae Goddard, 2012 sequencing populations (ii) Reasonable levels of gene flow were found in S. cerevisiae populations among regions Bacteria, fungi Four regions Grape must Culture-independent method, (i) Regional origin defined grape must Bokulich et al., 16S rRNA and ITS amplicon microbial patterns, with some 2014 sequencing influences by the cultivar (ii) Weather and climate were responsible for driving biogeographical diversity (iii) Vintage exerted seasonal shifts in grape microbiota within single vineyards, especially bacteria S. cerevisiae Six regions Vineyard soil, grape Culture-dependent method, (i) Regionally genetically differentiated Knight and juice, native forest microsatellite loci amplification S. cerevisiae populations drove different Goddard, 2015; soil and fruits and genotyping wine phenotype Knight et al., 2015 (ii) Genetic similarity of S. cerevisiae populations was found between vineyards and forests within regions Fungi Within region, Grapes Culture-dependent method, (i) Intravineyard variations were greater Setati et al., 2012 three vineyards ITS-ARISA fingerprinting than intervineyard variations, possibly due to microclimate’s influences on grape microbiota (ii) The least treated vineyard (biodynamic and integrated) displayed significantly higher fungal species richness Bacteria Within region, Grape must, end Culture-independent method, (i) Bacterial community heterogeneities Portillo et al., 2016 seven vineyards malolactic ferments 16S rRNA amplicon were influenced by the cultivar and sequencing geographic orientation (ii) Intervineyard variations were greater than intravineyard variations Bacteria Within region, Soil Culture-independent method, (i) Soil bacterial communities were Burns et al., 2015; 19 vineyards 16S rRNA amplicon structured with respect to soil Burns et al., 2016 sequencing properties, location, geographic features, and management practices, e.g., conventional/organic/biodynamic systems (ii) High relative abundances of the majority of dominant taxa were found in soils with lower carbon or nitrogen contents Bacteria Within region, Soil, roots, leaves, Culture-independent method, (i) Most grapevine OTUs originated in the Zarraonaindia et al., five vineyards flowers/grapes 16S rRNA amplicon soil 2015 sequencing (ii) Soil-borne bacteria were selected by plants (iii) Microbial structure was influenced by edaphic factors, i.e., pH, C:N ration, soil carbon, etc. Fungi Six regions Vineyard soil, bark, Culture-independent method, (i) Vineyard fungi accounted for ∼40% of Morrison-Whittle juice and ferments, 26S rRNA amplicon the diversity in juice and ferments and Goddard, 2018 native forest soil sequencing and fruits (ii) The geographical diversification of must microbiome weakened during fermentation

(Continued)

Frontiers in Microbiology | www.frontiersin.org 265 November 2019 | Volume 10 | Article 2679 fmicb-10-02679 November 20, 2019 Time: 15:49 # 6

Liu et al. Regional Traits of Wine From Microbes

TABLE 1 | Continued

Microorganisms Scale Habitat Methodology Major Conclusions References

Fungi Within region, Soil, bark, grapes, Culture-independent method, Biodynamic practices signiﬁcantly Morrison-Whittle 12 vineyards juice and ferments 26S rRNA amplicon affected soil and grapevine-associated and Goddard, 2017 sequencing microbiome but not the harvest juice communities, nor on ﬁnal wine quality Fungi Within vineyard Grapes, must and Culture-independent method (i) Lower biodiversity of yeasts and Grangeteau et al., ferments for fungi with 18S rRNA fungal populations was measured 2017 amplicon sequencing, in organically- than culture-dependent method for conventionally-farmed grapes yeasts and ferments

(ii) SO2 addition favoured the domination of S. cerevisiae during fermentation Yeasts Within vineyard Must and ferments Culture-dependent method, Prefermentative cold soak modiﬁed Maturano et al., D1/D2 26S sequencing yeast dynamics in a 2015 temperature-dependent manner

Bacteria, fungi Within vineyard Must and ferments Culture-independent method, SO2 treatment altered wine microbial Bokulich et al., 16S rRNA and ITS amplicon diversity in a dose-dependent manner 2015 sequencing

microbiota are affected by climate and weather, soil, and radiation (Van Leeuwen and Seguin, 2006). In viticulture, anthropogenic practices is considered in the following sections. climatic influences are recognised at multiple geospatial scales. Macroclimate, or regional climate, is largely determined by Sampling and Methodologies latitude and altitude but also modified by moderating influences It is noteworthy that a lack of standardised sampling strategies from water such as seas or large lakes (Figure 2; Van Leeuwen, and analytical framework impedes comparisons among studies, 2010; White, 2015). The local climate of a particular vineyard hindering the global insight of microbial ecology in wine is given as the mesoclimate, which is determined primarily production. For example, the depth of soil, the volume size of by topography including altitude, aspect, and slope to impact soil and plant materials (roots, bark, leaves, flowers, grapes), as upon wine quality and style (Figure 2; Gladstones, 1992; Van well as must conditions (grapes crushed under aseptic conditions Leeuwen et al., 2004; Bramley et al., 2011). For example, vineyard or collected directly from commercial wineries), vary among orientation can affect the warmth and sunlight interception studies and generate different datasets (Supplementary Table of grapevines, and steeper slopes benefit even more from this S1) (reviewed in Morgan et al., 2017). Standardised sampling influence. At the smallest scale, microclimate, the temperature, procedures are indispensable to obtain sound data for microbial humidity and solar variations within the canopy and between biogeography studies. Metagenomic methodologies, covering vines, may be affected in part by soil conditions and leaf shading from DNA extraction, target genes for sequencing (e.g., 16S V4 and manipulated by canopy management (Figure 2; Smart and region, ITS region) to bioinformatics pipelines (e.g., QIIME), can Robinson, 1991; Van Leeuwen and Seguin, 2006). Here, we generate technical variation among individual studies [reviewed highlight the microbial presence at these different scales to in Morgan et al. (2017) and Stefanini and Cavalieri (2018)]. The disentangle the intersection of microbes and climate and how this example set by the Earth Microbiome Project1 to standardise influences wine composition and style. sampling and analysis could be a good solution to reduce the technical variation and help understand the contribution of Macroclimate biological variation, weather, climate and other general trends. Microorganisms are mainly distributed by physical forces, such as air and wind (Zhu et al., 2017). Incorporation into clouds and precipitation and into nearby ecosystems increases CLIMATE AFFECTS WINE QUALITY VIA their long-distance dispersal (Hamilton and Lenton, 1998; MICROBIOTA Figure 2). These forces shape microbial biogeographic patterns in a similar way to those of plants and animals (Hamilton and Climate, the long-term weather pattern of an area, is a profound Lenton, 1998; Zhu et al., 2017). Climate, as the most important element determining wine styles and regional characteristics and environmental factor of grapegrowing regions, exerts influences thus wine quality. Cooler climates are better suited to producing on local microbial incidence and persistence in both space and light and delicate wines, while warmer climates tend to shape time (described as the vintage effect). At the macroclimate scale, heavy and rich flavour profiles. The climate impacts viticulture grape must microbiota present regional distribution patterns and wine quality through temperature, precipitation, and solar which are significantly conditioned by local environments (e.g., temperatures and rainfall) and weakly affected by the vintage, for 1http://www.earthmicrobiome.org/ example maximum temperature and average low temperature

Frontiers in Microbiology | www.frontiersin.org 276 November 2019 | Volume 10 | Article 2679 fmicb-10-02679 November 20, 2019 Time: 15:49 # 7

Liu et al. Regional Traits of Wine From Microbes

are negatively associated with Penicillium, Pseudomonas, influence resulting wine compositions remains speculative and Enterobacteriaceae, and Leuconostocaceae (O. oeni)(Bokulich deserves further study. et al., 2014). Rainfall and humidity positively correlate with yeast Hanseniaspora and Metschnikowia, and negatively correlate with Torulaspora, Saccharomyces and Meyerozyma (Jara et al., THE INFLUENCE OF THE SOIL BORNE 2016). As many of these species (in particular Saccharomyces MICROBIOME ON WINE COMPOSITION and O. oeni) are the most abundant species present in a wine fermentation, is can be said that local climatic conditions can AND QUALITY shape wine compositions by affecting their presence as seen in The soil substrate provides the grapevine with water and these association studies. nutrients and soil type, composition and structure profoundly Mesoclimate affects vine growth and development. Soil composition can affect the composition of wine, as wine can be related to its origin At the mesoclimate level, or vineyard scale, consistency between by tracking multiple major and trace elements from the soil vintages is one dimension of wine quality that is targeted for the to the wine (Almeida and Vasconcelos, 2003; Kment et al., terroir expression (Van Leeuwen et al., 2004; Beverland, 2006). 2005). Significant correlations were attributed to the movement Vintage variations substantially influence microbial communities of elements from the soil to grapes and resultant wines (Almeida in small geospatial scales (individual vineyards) rather than and Vasconcelos, 2003). This contributes to our understanding large scales (regions), with more stable fungal patterns observed of how soil geochemistry affects wine composition but has not between vintages (Bokulich et al., 2014). Vintage can significantly been established mechanistically. An appreciation of the soil affect the biodiversity of yeast populations in the grape and microbiota to vine health is relatively recent, and we review the must (Sabate et al., 2002; Vigentini et al., 2015). As a knowledge in this area below. driver to shape mesoclimate, topographical features also exert influence on grape microbiota (Figure 2; Portillo et al., 2016). For example, Oxalobacteraceae, Haemophilus, Sphingomonas, Soil Borne Microorganisms in and Pseudomonas were identified in grapes from east-facing Association With Grapevines vineyards, while Streptococcus, Micrococcaceae, Staphylococcus, Grapevines live in biogeochemically diverse soils harbouring Enhydrobacter, and Aeromonadaceae were shown as typical taxa diverse microbiota that affect plant health and growth in of flat vineyards (Portillo et al., 2016). beneficial, commensal, or pathogenic ways (Müller et al., 2016). Soil-borne microbes can affect crop yields and metabolite Microclimate synthesis in other agricultural systems, and thus may shape It could be argued that microclimate is the environment wine colour, aroma, flavour, and quality. Serving as a reservoir most likely to affect the presence, growth and activity of for plant-associated bacteria (Zarraonaindia et al., 2015), soil microbes. Modifying the grapevine leaf area through training, borne bacteria can colonise plant organs by physical contact pruning, trellising, and defoliation alters canopy microclimate, or travel from the rhizosphere toward the phyllosphere on the in particular the solar radiation onto grapes and leaves, and to surface (epiphytes) or within plants (endophytes) (Figure 1; Chi a lesser extent, by air movement in the leaf canopy (strongly et al., 2005; Compant et al., 2005; Compant et al., 2011; Martins affecting humidity) and temperature (Figure 2; Reynolds et al., 2013; Ruiz-Pérez et al., 2016). For example, plant-growth- and Heuvel, 2009). The effect of light in the microclimate promoting (PGP) bacteria Burkholderia sp. strain PsJN can environment have been reported to affect a wide range of endophytically colonise Chardonnay plantlets and translocate aspects of berry composition, and can strongly affect the colour onto the root surfaces, root internal tissues, and finally the and flavours present in the final wine (Haselgrove et al., internode and leaf (Compant et al., 2005). Some dominant taxa in 2000; Liu et al., 2015). However, studies on the effects of the soil, such as Gammaproteobacteria (including Pseudomonas microclimate variability on microbial communities are rare. spp.) and Firmicutes (including Bacillus spp.) have been visualised Fungal communities can be discriminated within vineyards by fluorescence in situ hybridisation as endophytes inside the and this effect was ascribed to microclimate variability, but no flower ovules, and berries and seeds of Zweigelt grapevines (Vitis experimentally test has validated this correlation (Setati et al., vinifera)(Compant et al., 2011). 2012). From other studies it is clear that sunlight interception As well as bacteria, the vineyard soil is one of the natural could affect the grapevine microbiome. For example, an aquatic sources of fungi in musts (Morrison-Whittle and Goddard, yeast study suggested that Cryptococcus spp. and pigmented 2018). Notably, identical genotypes of S. cerevisiae are found yeasts Rhodotorula spp. predominate in certain environments shared between soil and fruit niches within regions (Knight as they can produce photoprotective compounds (carotenoid and Goddard, 2015). A vineyard field experiment suggested pigments and mycosporines) to adapt to pelagic sites with that S. cerevisiae yeasts can be adsorbed from the soil by roots high UV radiation (Brandão et al., 2011). These yeasts are and transported via vine to stems and surface of grapes, and also associated with grapevines (Sabate et al., 2002; Setati finally entered fermentation musts (Mandl et al., 2015). This et al., 2012), and so a similar response is possible in the fruit translocation process can enable soil borne microorganisms be a zone affect the composition of fungal communities. Whether part of grapevine-associated microbiome that influence resultant microclimate can influence the wine-related microbiome and wine quality and characteristics directly; these microbes can

Frontiers in Microbiology | www.frontiersin.org 287 November 2019 | Volume 10 | Article 2679 fmicb-10-02679 November 20, 2019 Time: 15:49 # 8

Liu et al. Regional Traits of Wine From Microbes

survive fermentation and release secondary metabolites affecting Hanseniaspora uvarum) and filamentous fungi (Aureobasidium wine profiles or interact with fermentative yeasts or bacteria pullulans, Cladosporium spp.) (Morrison-Whittle and Goddard, within the community. Of particular note for this discussion 2018). While Saccharomyces spp. and wine spoilage species are is the root associated rhizobia which has not been extensively mostly present in low numbers and in low frequencies in the studied in grapevines. This community is likely to have a soil, the most abundant genera (Amniculicola, Doratomyces, profound affect as it was observed in rice plants (Oryza sativa Endocarpon, and Tricellulortus) do not appear to be directly L.) that an ascending endophytic migration of rhizobia starting involved in wine production (Barata et al., 2012). In addition, on the rhizoplane surface, within root tissues, to reach the stem regional delineations of fungal communities were observed in base, leaf sheath, and leaves, and this process benefits rice growth musts but not in the vineyard habitats at the regional scale physiology afterward (Chi et al., 2005). (Morrison-Whittle and Goddard, 2018). Interestingly, significant differentiation of small-scale fungal communities was observed in a vineyard system (distance < 2 km), although with no Soil Properties and Interactions With vineyard demarcation between fermenting populations of S. Plants Shape Vineyard Microbiota cerevisiae (Knight et al., 2019). In-depth insight in fungal Biogeographic patterns have been observed in the microbial ecology in vineyards will inform how grapevines recruit fungal communities of vineyard soils (Table 1). Soil bacterial diversity communities during the annual plant cycle and how this affects and composition associate with the vineyard geographical the communities present on the grape. For example, incidence location, with significant influences from soil physicochemical of fermenting yeasts on grapes changes during ripening (Barata properties, such as soil texture, soil pH, temperature, moisture, et al., 2012), and can thus contribute to wine metabolite profiles. carbon and nitrogen pools (Burns et al., 2015; Zarraonaindia et al., 2015). The majority of dominant taxa such as Proteobacteria (especially Beta- and Gamma-proteobacteria), Bacteroidetes, ANTHROPOGENIC PRACTICES AFFECT Gemmatimonadetes, and Firmicutes show higher relative QUALITY WINE PRODUCTION BY abundance in soils with lower carbon or nitrogen contents, while MICROBIAL MODULATION abundance of Actinobacteria shows a negative trend (Burns et al., 2015). In addition, local climate (average annual precipitation) The natural environment of the vineyard determines grape and topography (altitude, aspect, and slope) could indirectly production and composition. Based on the local conditions, influence soil microbial communities through their impacts on thoughtful human management can optimise wine quality soil properties (Burns et al., 2015). Noticeably, fermentation- and style across vintages (White, 2015). Wine producers related bacteria in the musts are found in the vineyard soil as select vineyard sites with favourable environmental conditions dominant taxa, such as Firmicutes (encompassing LAB) and (especially climate and soil) for quality wine production, as well spoilage bacteria Acidobacteria and Proteobacteria (Burns et al., as grapevine cultivars that adapt to the local environment (Van 2015). As a reservoir of grapevine-associated bacteria, soil-borne Leeuwen and Seguin, 2006). The grapegrower manages the vine microbes can shape phyllosphere bacterial assemblages, in to achieve balance between vine vigour and yield and may use particular grapes (Zarraonaindia et al., 2015). When the grapes different training, pruning, trellising and canopy methodologies, are transferred into the winery and the microbiota persist in to harvest a healthy crop, with a desired fruit composition alcoholic fermentation, soil microbiota directly correlate with and to optimise wine quality (Jackson and Lombard, 1993; resulting wine metabolites. Reynolds and Heuvel, 2009). Pesticides (including fungicides Differentiation of vineyard soil bacterial community structure against downy mildew, powdery mildew and grey rot) and is reflected in the roots microbiome, with the relative abundance fertiliser application are common interventions in the vineyard, of several taxa occurring in a vineyard-depending manner. which are now embedded within viticulture management These taxa outcompete other bacteria for colonisation and/or are practices of conventional, organic and biodynamic systems. selected by grapevines. For example, PGP bacteria Rhizobiales, Some evidence exists to differentiate the superiority of microbial especially Bradyrhizobium spp., can benefit nitrogen fixation diversity of organic or biodynamic wines with respect to specific and antibiotic production that would promote plant growth, management practice (Ross et al., 2009; Pagliarini et al., 2013), yet disease suppression and grape composition, thereby indirectly spontaneous fermentation and related winemaking techniques influencing wine quality and characteristics (Zarraonaindia (for example cold soak, limited SO2 usage) are thought to et al., 2015). Within a vineyard, interactions between the encourage showing the indigenous microbial diversity, and thus root compartments (rhizosphere and root endosphere) and the enhance wine terroir expression (Capozzi et al., 2015). Recent rootstock can also exert a unique selective pressure enhancing studies focusing on the wine-related microbiome offer new niche differentiation of bacteria, in particular on the taxa with information to this area and provide more information to allow PGP potential, for example by producing indole acetic acid which expression of local individuality for viticultural production. affect plant hormone balances (Marasco et al., 2018). Vineyard niches (soil, bark, fruit) account for approximately Viticultural Practices 40% of fungal populations in the musts and ferments. The clear Specific human interventions including pesticides, fungicides, similarity of fungi is between musts and grapes, that communities and herbicides usage can affect microbial diversity in specific are dominated by fermenting yeasts (e.g., S. cerevisiae and habitats in vineyards (Cadežˇ et al., 2010; Fierer et al., 2012a;

Frontiers in Microbiology | www.frontiersin.org 298 November 2019 | Volume 10 | Article 2679 fmicb-10-02679 November 20, 2019 Time: 15:49 # 9

Liu et al. Regional Traits of Wine From Microbes

Perazzolli et al., 2014; Pinto et al., 2014; Chou et al., 2018). Here tends to be present only in the early stages of fermentation. we highlight the effect of viticulture practices on the vineyard Some debate exists, as Kecskeméti et al. (2016) claimed microbiology, and whether these microbes and their effects can conventional/organic/biodynamic practices in the same vineyard persist into wine production to ultimately influence wine quality. do not significantly influence grape microbial diversity. Viticulture practices can modify the belowground Inconsistency is not surprising, as local vineyard conditions, microbiome. Soil borne bacterial communities are structured specific management practices and sampling methods differ in with respect to commercial/organic/biodynamic systems, as the surveys published. mediated by shifts in soil resource pools, particularly carbon and nitrogen (Burns et al., 2016). Compost addition in organic Management of Microbes During and biodynamic vineyards increases overall bacterial diversity Winemaking and alter the community composition, and can the effect can The wine industry has developed a series of methods to promote be enhanced by tillage and cover crop management. Similarly, wine fermentations, of which the most effective methods are the soil fungal diversity and community composition of biodynamic inoculation of cultured S. cerevisiae strains and use of sulphur vineyards differ from those which are conventionally managed dioxide (SO2). Commercial fermentations are widely used to (Morrison-Whittle and Goddard, 2017; Table 1). Biodynamic reduce the risk of spoilage and unpredictable wine composition, management enhance fungal diversity in the grapevines niches and to ensure a stable wine flavour. However, inoculated (bark, fruits) but this effect did not persist into the harvested fermentations reduce the potential of microbiota to contribute to juice and fermentation (Morrison-Whittle and Goddard, 2017). regional characteristics. Spontaneous fermentations, comprising The phyllosphere microbiota can be shaped by agricultural a diversity of yeast species and S. cerevisiae strains originating management, in particular fungicide usage (Gilbert et al., from vineyard and winery, enhance wine regional expression 2014). Conventional vineyards are usually treated with several (Capozzi et al., 2015). The diversity of yeast species present agricultural chemicals, while organic/biodynamic vineyards can have profound impacts on the flavour of the resultant only recieve sulphur- and/or copper-based formulations. wine. For example, non-Saccharomyces yeasts can produce and Several studies have shown a higher microbial diversity in secrete several enzymes (esterases, β-glucosidases, proteases), to grapes present in organic and biodynamic vineyards for both synthesise volatile compounds and playing a role in varietal yeasts (Saccharomyces and non-Saccharomyces) and total fungi aroma (Romano et al., 2003). H. uvarum can positively (yeasts and filamentous fungi) (Setati et al., 2012; Martins et al., interact with S. cerevisiae to enhance fermentation (Romano 2014; Setati et al., 2015), and this could be because chemical et al., 2003). Interactions within S. cerevisiae populations treatments reduce microbial richness and diversity associated provide regional microbial signatures positively correlating with with grapevines and wine (Pinto et al., 2014; Escribano- wine aroma profiles (Knight et al., 2015). Pre-fermentative Viana et al., 2018). This effect can maintain in spontaneous cold soak, a technique widely used in red wine production fermentations from organic/biodynamic musts where higher to favour wine colour, taste and mouthfeel attributes, can yeast species richness and diversity is observed compared influence yeast population dynamics depending the temperature. to musts from conventionally managed vineyards, and this For example, a cold soak at 14 ± 1◦C can increases total is particularly true for fermentative yeasts species such as yeast populations and favour growth of H. uvarum and H. uvarum, H. vineae, H. guilliermondii, Streptomyces bacillaris, Candida zemplinina, whereas cold soak at 8 ± 1◦C favours Lachancea thermotolerans, and S. cerevisiae (Cordero-Bueso growth of S. cerevisiae (Maturano et al., 2015; Table 1). SO2 et al., 2011; Bagheri et al., 2015). Regarding grapevine-associated treatment favours the early implantation and domination of bacteria, responses to viticulture practices are weak compared S. cerevisiae and alters wine microbial diversity and fermentation to fungi, especially grape surface bacteria showing more progression in a dose-dependent manner (Bokulich et al., resilience than that of leaves (Schmid et al., 2011; Miura 2015; Grangeteau et al., 2017; Table 1). A concentration of et al., 2017). Biodynamic berries are found rich in Bacillales 25 mg/L SO2 is ideal to stabilise the microbial communities, including genera of Lysinibacillus, Bacillus, and Sporosarcin, which inhibits the growth of LAB and Gluconobacter but which are typical microbes in manure (Mezzasalma et al., not other bacteria and fungi at early fermentation, thereby 2017), but their influences on wine composition are not clear. maximising microbial diversity to benefit wine regionality Heavy use of sulphur and copper fungicides decrease the expression. Much less “manipulating the terroir,” production biodiversity of yeasts and fungi in resultant fermentations of distinctive and quality wine is achievable through microbial (Milanoviæ et al., 2013; Grangeteau et al., 2017). Particular manipulation in the winery. fungi are associated with differently managed vineyards, where Basidiomycota (especially Cryptococcus) is mainly How Will Climate Change Affect the associated with organic vineyards, while fermentative yeasts Saccharomyces, Metschnikowia, and Hanseniaspora were mainly Incidence and Activity of Microbes found in conventionally managed vineyards (Grangeteau Involved in Winemaking? et al., 2017; Table 1). The yeast-like fungus Aureobasidium Human activity has dramatically affected the global climate and pullulans dominate phyllosphere fungi in organic/biodynamic the weather patterns that grapevines experience (Olesen et al., vineyards (Schmid et al., 2011; Pancher et al., 2012; Setati 2011). Wine production is particularly sensitive to climate change et al., 2012; Martins et al., 2014; Setati et al., 2015), but as there is an inherent link between climate and wine quality

Frontiers in Microbiology | www.frontiersin.org 309 November 2019 | Volume 10 | Article 2679 fmicb-10-02679 November 20, 2019 Time: 15:49 # 10

Liu et al. Regional Traits of Wine From Microbes

and style. Matching grape varieties to a unique combination CONCLUSION AND FUTURE of climate and soil enables production of distinctive wines PERSPECTIVES worldwide (Van Leeuwen and Seguin, 2006), but this is likely to change as the temperature and water profiles of specific areas Fundamental questions about how wine quality and change. Trends in grapevine phenology associated with global distinctiveness can be derived from local environments warming are widely reported, and report earlier maturation with have been very difficult to answer. Here, we present microbial undesirable influences on grape and wine aroma and flavour biogeography as core to understanding wine regionality. As the (Palliotti et al., 2014). For example, wine grapes have been driver of wine fermentation, microorganisms inhabit and adapt ripening earlier in Australia in recent decades, driven by warming to local geography, climate, soil and anthropogenic practices. and declines in soil moisture (Webb et al., 2012). A deficit The climate shapes microbial geographic diversification at of water will affect production of colourful and flavoursome multiple scales thereby affecting wine compositions. Soil- wines rich in phenolic substances in a warmer and drier future borne microbiota shapes the grapevine-associated microbiota (Bonada et al., 2015). In the long term, climate change will and physiology and ultimately the flavour of resulting wines, affect the geographical distribution of viticulture and will require but the mechanism by which this occurs remains to be changing varieties grown to adapt to warming or providing elucidated. Human management practices modify wine-related artificial shading to reduce temperatures (Jones et al., 2005; microbiota to improve quality wine production. However, Webb et al., 2012), whereas in the short term, management our knowledge of how geographically diverse microbiota techniques may be able to mitigate negative impacts and include shape wine chemical and organoleptic characteristics is managing soil water content (irrigation), crop yield (pruning limited. In-depth insights have emerged from studies of regime), and vine response (rootstock selection, leaf removal) environment-plant-microbe interactions and will inform us (Webb et al., 2012). about how grapevines recruit their microbiome to maximise Microorganisms involved in winemaking are not discussed both nutrition and microbial diversity under local conditions. in the context of climate change. Increasing temperature and The grapevine microbiota can then be sensibly exploited in drought would strongly affect the ability to grow grapes and wine production by introducing and/or managing specific wine production, but in largely unknown ways. One area microbes in combination with optimised wine metabolites where these effects will be felt is in the soil and its microbial to sculpt fermentation consortia for quality and distinctive composition and activity. Studies outside of agriculture wine production. Anthropogenic climate change will have show that soil pH, moisture, temperature and nutrient profound consequences on wine-associated microbiota availability are the main drivers of microbial community and thus affect wine quality and style. We believe that assembly (Fierer et al., 2012b) and that soil fungi and carefully designed empirical experiments to unpick microbial bacteria occupy specific niches and respond differently to ecology and wine metabolome, surveys, and international and precipitation and soil pH, and therefore would respond interdisciplinary collaborations will be indispensable to obtain a differently to climate change on their diversity, abundance, comprehensive understanding of climate change and the future and function potential (Bahram et al., 2018). For example, of wine industry. fungal networks are more stable under drought conditions than bacteria in a grassland ecosystem (de Vries et al., 2018). Warming increases soil bacterial populations but AUTHOR CONTRIBUTIONS decreases diversity and changes the composition (Sheik et al., 2011). Warming also increases soil respiration (Luo DL wrote the first draft of the review. KH revised and et al., 2001), which can in turn reduce the abundance of added to the draft. All authors approved the final version Actinobateria as sensitive to high carbon dioxide production of the manuscript. (Sheik et al., 2011). Soil water saturation affects microbial composition by changing oxygen availability and thus microbial respiration (Carbone et al., 2011). Drought can also increase FUNDING the resistance of soil bacteria (Bouskill et al., 2013), for example Actinobateria, which is tolerant of low moisture DL acknowledges support from a Ph.D. scholarship conditions (Fierer et al., 2012b). While these studies have and funding from Wine Australia (AGW Ph1602) and been performed outside vineyards, it is likely that the same a Melbourne Research Scholarship from the University overall principals are at play. We suggest that superimposing of Melbourne. warming and drying mechanisms will affect vineyard soil microbiota abundance, diversity and functions, and thus change their capacity to support plant growth. Further studies are SUPPLEMENTARY MATERIAL needed, but it is clear that understanding wine microbial biogeography under the changing climate will help wine The Supplementary Material for this article can be found online industry to adapt to climate change and enable quality wine at: https://www.frontiersin.org/articles/10.3389/fmicb.2019. production in the future. 02679/full#supplementary-material

Frontiers in Microbiology | www.frontiersin.org 3110 November 2019 | Volume 10 | Article 2679 fmicb-10-02679 November 20, 2019 Time: 15:49 # 11

Liu et al. Regional Traits of Wine From Microbes

REFERENCES Capozzi, V., Garofalo, C., Chiriatti, M. A., Grieco, F., and Spano, G. (2015). Microbial terroir and food innovation: the case of yeast biodiversity in wine. Almeida, C. M. R., and Vasconcelos, M. T. S. (2003). Multielement composition of Microbiol. Res. 181, 75–83. doi: 10.1016/j.micres.2015.10.005 wines and their precursors including provenance soil and their potentialities as Carbone, M. S., Still, C. J., Ambrose, A. R., Dawson, T. E., Williams, A. P., Boot, fingerprints of wine origin. J. Agric. Food Chem. 51, 4788–4798. doi: 10.1021/ C. M., et al. (2011). Seasonal and episodic moisture controls on plant and jf034145b microbial contributions to soil respiration. Oecologia 167, 265–278. doi: 10. Bagheri, B., Bauer, F., and Setati, M. (2015). The diversity and dynamics of 1007/s00442-011-1975-3 indigenous yeast communities in grape must from vineyards employing Chi, F., Shen, S.-H., Cheng, H.-P., Jing, Y.-X., Yanni, Y. G., and Dazzo, F. B. (2005). different agronomic practices and their influence on wine fermentation. South Ascending migration of endophytic rhizobia, from roots to leaves, inside rice Afr. J. Enol. Vitic. 36, 243–251. plants and assessment of benefits to rice growth physiology. Appl. Environ. Bahram, M., Hildebrand, F., Forslund, S. K., Anderson, J. L., Soudzilovskaia, N. A., Microbiol. 71, 7271–7278. doi: 10.1128/aem.71.11.7271-7278.2005 Bodegom, P. M., et al. (2018). Structure and function of the global topsoil Chou, M.-Y., Heuvel, J. V., Bell, T. H., Panke-Buisse, K., and Kao-Kniffin, J. (2018). microbiome. Nature 560, 233–237. doi: 10.1038/s41586-018-0386-6 Vineyard under-vine floor management alters soil microbial composition, while Barata, A., Malfeito-Ferreira, M., and Loureiro, V. (2012). The microbial ecology the fruit microbiome shows no corresponding shifts. Sci. Rep. 8:11039. doi: of wine grape berries. Int. J. Food Microbiol. 153, 243–259. doi: 10.1016/j. 10.1038/s41598-018-29346-1 ijfoodmicro.2011.11.025 Compant, S., Mitter, B., Colli-Mull, J. G., Gangl, H., and Sessitsch, A. (2011). Beverland, M. (2006). The ‘real thing’: branding authenticity in the luxury wine Endophytes of grapevine flowers, berries, and seeds: identification of cultivable trade. J. Bus. Res. 59, 251–258. doi: 10.1016/j.jbusres.2005.04.007 bacteria, comparison with other plant parts, and visualization of niches of Bokulich, N. A., Collins, T. S., Masarweh, C., Allen, G., Heymann, H., Ebeler, colonization. Microb. Ecol. 62, 188–197. doi: 10.1007/s00248-011-9883-y S. E., et al. (2016). Associations among wine grape microbiome, metabolome, Compant, S., Reiter, B., Sessitsch, A., Nowak, J., Clément, C., and Barka, E. A. and fermentation behavior suggest microbial contribution to regional wine (2005). Endophytic colonization of Vitis vinifera L. by plant growth-promoting characteristics. mBio 7:e00631-16. doi: 10.1128/mBio.00631-16 bacterium Burkholderia sp. strain PsJN. Appl. Environ. Microbiol. 71, 1685– Bokulich, N. A., Swadener, M., Sakamoto, K., Mills, D. A., and Bisson, L. F. (2015). 1693. doi: 10.1128/aem.71.4.1685-1693.2005 Sulfur dioxide treatment alters wine microbial diversity and fermentation Cordero-Bueso, G., Arroyo, T., Serrano, A., Tello, J., Aporta, I., Vélez, M. D., et al. progression in a dose-dependent fashion. Am. J. Enol. Vitic. 66, 73–79. (2011). Influence of the farming system and vine variety on yeast communities doi: 10.5344/ajev.2014.14096 associated with grape berries. Int. J. Food Microbiol. 145, 132–139. doi: 10.1016/ Bokulich, N. A., Thorngate, J. H., Richardson, P. M., and Mills, D. A. (2014). j.ijfoodmicro.2010.11.040 Microbial biogeography of wine grapes is conditioned by cultivar, vintage, de Vries, F. T., Griffiths, R. I., Bailey, M., Craig, H., Girlanda, M., Gweon, H. S., and climate. Proc. Natl. Acad. Sci. U.S.A. 111, E139–E148. doi: 10.1073/pnas. et al. (2018). Soil bacterial networks are less stable under drought than fungal 1317377110 networks. Nat. Commun. 9:3033. doi: 10.1038/s41467-018-05516-7 Bonada, M., Jeffery, D., Petrie, P. R., Moran, M. A., and Sadras, V. O. (2015). Impact El Khoury, M., Campbell-Sills, H., Salin, F., Guichoux, E., Claisse, O., and of elevated temperature and water deficit on the chemical and sensory profiles Lucas, P. M. (2017). Biogeography of oenococcus oeni reveals distinctive but of B arossa S hiraz grapes and wines. Aust. J. Grape Wine Res. 21, 240–253. nonspecific populations in wine-producing regions. Appl. Environ. Microbiol. doi: 10.1111/ajgw.12142 83:e02322-16. doi: 10.1128/AEM.02322-16 Börlin, M., Venet, P., Claisse, O., Salin, F., Legras, J.-L., and Masneuf-Pomarede, Escribano-Viana, R., López-Alfaro, I., López, R., Santamaría, P., Gutiérrez, I. (2016). Cellar-associated Saccharomyces cerevisiae population structure A. R., and González-Arenzana, L. (2018). Impact of chemical and biological revealed high diversity and perennial persistence in Sauternes wine estates. fungicides applied to grapevine on grape biofilm, must, and wine microbial Appl. Environ. Microbiol. 82, 2909–2918. doi: 10.1128/AEM.03627-15 diversity. Front. Microbiol. 9:59. doi: 10.3389/fmicb.2018.00059 Bouskill, N. J., Lim, H. C., Borglin, S., Salve, R., Wood, T. E., Silver, W. L., et al. Fierer, N., Lauber, C. L., Ramirez, K. S., Zaneveld, J., Bradford, M. A., and Knight, (2013). Pre-exposure to drought increases the resistance of tropical forest soil R. (2012a). Comparative metagenomic, phylogenetic and physiological analyses bacterial communities to extended drought. ISME J. 7, 384–394. doi: 10.1038/ of soil microbial communities across nitrogen gradients. ISME J. 6, 1007–1017. ismej.2012.113 doi: 10.1038/ismej.2011.159 Bramley, R., Ouzman, J., and Boss, P. K. (2011). Variation in vine vigour, grape yield Fierer, N., Leff, J. W., Adams, B. J., Nielsen, U. N., Bates, S. T., Lauber, C. L., et al. and vineyard soils and topography as indicators of variation in the chemical (2012b). Cross-biome metagenomic analyses of soil microbial communities composition of grapes, wine and wine sensory attributes. Aust. J. Grape Wine and their functional attributes. Proc. Natl. Acad. Sci. 109, 21390–21395. Res. 17, 217–229. doi: 10.1111/j.1755-0238.2011.00136.x doi: 10.1073/pnas.1215210110 Brandão, L. R., Vaz, A. B. M., Rosa, C. A., Espírito Santo, L. C., Van Francesca, N., Canale, D. E., Settanni, L., and Moschetti, G. (2012). Dissemination Broock, M., Moliné, M., et al. (2011). Yeasts from an oligotrophic lake in of wine-related yeasts by migratory birds. Environ. Microbiol. Rep. 4, 105–112. Patagonia (Argentina): diversity, distribution and synthesis of photoprotective doi: 10.1111/j.1758-2229.2011.00310.x compounds and extracellular enzymes. FEMS Microbiol. Ecol. 76, 1–13. Garofalo, C., Russo, P., Beneduce, L., Massa, S., Spano, G., and Capozzi, V. (2016). doi: 10.1111/j.1574-6941.2010.01030.x Non-Saccharomyces biodiversity in wine and the ‘microbial terroir’: a survey on Burns, K. N., Bokulich, N. A., Cantu, D., Greenhut, R. F., Kluepfel, D. A., O’geen, Nero di Troia wine from the Apulian region Italy. Ann. Microbiol. 66, 143–150. A. T., et al. (2016). Vineyard soil bacterial diversity and composition revealed by doi: 10.1007/s13213-015-1090-5 16S rRNA genes: differentiation by vineyard management. Soil Biol. Biochem. Gayevskiy, V., and Goddard, M. R. (2012). Geographic delineations of yeast 103, 337–348. doi: 10.1016/j.soilbio.2016.09.007 communities and populations associated with vines and wines in New Zealand. Burns, K. N., Kluepfel, D. A., Strauss, S. L., Bokulich, N. A., Cantu, D., and ISME J. 6, 1281–1290. doi: 10.1038/ismej.2011.195 Steenwerth, K. L. (2015). Vineyard soil bacterial diversity and composition Gilbert, J. A., Van Der Lelie, D., and Zarraonaindia, I. (2014). Microbial terroir revealed by 16S rRNA genes: differentiation by geographic features. Soil Biol. for wine grapes. Proc. Natl. Acad. Sci. U.S.A. 111, 5–6. doi: 10.1073/pnas. Biochem. 91, 232–247. doi: 10.1016/j.soilbio.2015.09.002 1320471110 Cadež,ˇ N., Zupan, J., and Raspor, P. (2010). The effect of fungicides on yeast Gladstones, J. (1992). Viticulture and Environment. Broadview, SA: Winetitles. communities associated with grape berries. FEMS Yeast Res. 10, 619–630. Gladstones, J. (2011). Wine, Terroir and Climate Change. Adelaide, SA: Wakefield doi: 10.1111/j.1567-1364.2010.00635.x Press. Capece, A., Granchi, L., Guerrini, S., Mangani, S., Romaniello, R., Vincenzini, M., Goddard, M. R., Anfang, N., Tang, R., Gardner, R. C., and Jun, C. (2010). A distinct et al. (2016). Diversity of Saccharomyces cerevisiae strains isolated from two population of Saccharomyces cerevisiae in New Zealand: evidence for local Italian wine-producing regions. Front. Microbiol. 7:1018. doi: 10.3389/fmicb. dispersal by insects and human-aided global dispersal in oak barrels. Environ. 2016.01018 Microbiol. 12, 63–73. doi: 10.1111/j.1462-2920.2009.02035.x

Frontiers in Microbiology | www.frontiersin.org 3211 November 2019 | Volume 10 | Article 2679 fmicb-10-02679 November 20, 2019 Time: 15:49 # 12

Liu et al. Regional Traits of Wine From Microbes

Gonzálvez, A., Llorens, A., Cervera, M., Armenta, S., and De La Guardia, M. Martins, G., Lauga, B., Miot-Sertier, C., Mercier, A., Lonvaud, A., Soulas, M.-L., (2009). Elemental fingerprint of wines from the protected designation of origin et al. (2013). Characterization of epiphytic bacterial communities from grapes, Valencia. Food Chem. 112, 26–34. doi: 10.1016/j.foodchem.2008.05.043 leaves, bark and soil of grapevine plants grown, and their relations. PLoS One Grangeteau, C., Roullier-Gall, C., Rousseaux, S., Gougeon, R. D., Schmitt-Kopplin, 8:e73013. doi: 10.1371/journal.pone.0073013 P., Alexandre, H., et al. (2017). Wine microbiology is driven by vineyard and Martins, G., Vallance, J., Mercier, A., Albertin, W., Stamatopoulos, P., Rey, P., winery anthropogenic factors. Microb. Biotechnol. 10, 354–370. doi: 10.1111/ et al. (2014). Influence of the farming system on the epiphytic yeasts and yeast- 1751-7915.12428 like fungi colonizing grape berries during the ripening process. Int. J. Food Guzzon, R., Labagnara, T., Toffanin, A., and Villegas, T. R. (2018). Oenological Microbiol. 177, 21–28. doi: 10.1016/j.ijfoodmicro.2014.02.002 characterisation of indigenous strains of S. cerevisiae isolated in a biodynamic Martiny, J. B. H., Bohannan, B. J., Brown, J. H., Colwell, R. K., Fuhrman, J. A., winery in the Cortona DOC area. Ann. Microbiol. 68, 963–967. doi: 10.1007/ Green, J. L., et al. (2006). Microbial biogeography: putting microorganisms on s13213-018-1405-4 the map. Nat. Rev. Microbiol. 4, 102–112. doi: 10.1038/nrmicro1341 Hamilton, W. D., and Lenton, T. (1998). Spora and Gaia: how microbes fly with Marzano, M., Fosso, B., Manzari, C., Grieco, F., Intranuovo, M., Cozzi, G., et al. their clouds. Ethol. Ecol. Evol. 10, 1–16. doi: 10.1080/08927014.1998.9522867 (2016). Complexity and dynamics of the winemaking bacterial communities in Haselgrove, L., Botting, D., Van Heeswijck, R., Høj, P., Dry, P., Ford, C., et al. berries, musts, and wines from apulian grape cultivars through time and space. (2000). Canopy microclimate and berry composition: the effect of bunch PLoS One 11:e0157383. doi: 10.1371/journal.pone.0157383 exposure on the phenolic composition of Vitis vinifera L cv. Shiraz grape berries. Maturano, Y. P., Mestre, M. V., Esteve-Zarzoso, B., Nally, M. C., Lerena, M. C., Aust. J. Grape Wine Res. 6, 141–149. doi: 10.1111/j.1755-0238.2000.tb00173.x Toro, M. E., et al. (2015). Yeast population dynamics during prefermentative Jackson, D., and Lombard, P. (1993). Environmental and management practices cold soak of cabernet sauvignon and malbec wines. Int. J. Food Microbiol. 199, affecting grape composition and wine quality-a review. Am. J. Enol. Vitic. 44, 23–32. doi: 10.1016/j.ijfoodmicro.2015.01.005 409–430. Mezzasalma, V., Sandionigi, A., Bruni, I., Bruno, A., Lovicu, G., Casiraghi, M., Jara, C., Laurie, V. F., Mas, A., and Romero, J. (2016). Microbial terroir in chilean et al. (2017). Grape microbiome as a reliable and persistent signature of field valleys: diversity of non-conventional yeast. Front. Microbiol. 7:663. doi: 10. origin and environmental conditions in Cannonau wine production. PLoS One 3389/fmicb.2016.00663 12:e0184615. doi: 10.1371/journal.pone.0184615 Jones, G. V., White, M. A., Cooper, O. R., and Storchmann, K. (2005). Climate Mezzasalma, V., Sandionigi, A., Guzzetti, L., Galimberti, A., Grando, M. S., change and global wine quality. Clim. Change 73, 319–343. doi: 10.1007/ Tardaguila, J., et al. (2018). Geographical and cultivar features differentiate s10584-005-4704-2 grape microbiota in Northern Italy and Spain vineyards. Front. Microbiol. 9:946. Kecskeméti, E., Berkelmann-Löhnertz, B., and Reineke, A. (2016). Are epiphytic doi: 10.3389/fmicb.2018.00946 microbial communities in the carposphere of ripening grape clusters (Vitis Milanoviæ, V., Comitini, F., and Ciani, M. (2013). Grape berry yeast communities: vinifera L.) different between conventional, organic, and biodynamic grapes? influence of fungicide treatments. Int. J. Food Microbiol. 161, 240–246. PLoS One 11:e0160852. doi: 10.1371/journal.pone doi: 10.1016/j.ijfoodmicro.2012.12.019 Kment, P., Mihaljeviè, M., Ettler, V., Šebek, O., Strnad, L., and Rohlová, L. (2005). Miura, T., Sánchez, R., Castañeda, L. E., Godoy, K., and Barbosa, O. (2017). Is Differentiation of Czech wines using multielement composition–A comparison microbial terroir related to geographic distance between vineyards? Environ. with vineyard soil. Food Chem. 91, 157–165. doi: 10.1016/j.foodchem.2004. Microbiol. Rep. 9, 742–749. doi: 10.1111/1758-2229.12589 06.010 Morgan, H. H., Du Toit, M., and Setati, M. E. (2017). The grapevine and Knight, S., and Goddard, M. (2015). Quantifying separation and similarity in a wine microbiome: insights from high-throughput amplicon sequencing. Front. Saccharomyces cerevisiae metapopulation. ISME J. 9, 361–370. doi: 10.1038/ Microbiol. 8:820. doi: 10.3389/fmicb.2017.00820 ismej.2014.132 Morrison-Whittle, P., and Goddard, M. (2017). Fungal communities are Knight, S. J., Karon, O., and Goddard, M. R. (2019). Small scale fungal community differentially affected by conventional and biodynamic agricultural differentiation in a vineyard system. Food Microbiol. doi: 10.1016/j.fm.2019. management approaches in vineyard ecosystems. Agric. Ecosyst. Environ. 103358 246, 306–313. doi: 10.1016/j.agee.2017.05.022 Knight, S., Klaere, S., Fedrizzi, B., and Goddard, M. R. (2015). Regional microbial Morrison-Whittle, P., and Goddard, M. R. (2018). From vineyard signatures positively correlate with differential wine phenotypes: evidence for a to winery: a source map of microbial diversity driving wine microbial aspect to terroir. Sci. Rep. 5:14233. doi: 10.1038/srep14233 fermentation. Environ. Microbiol. 20, 75–84. doi: 10.1111/1462-2920. Kuehne, H. A., Murphy, H. A., Francis, C. A., and Sniegowski, P. D. (2007). 13960 Allopatric divergence, secondary contact, and genetic isolation in wild yeast Müller, D. B., Vogel, C., Bai, Y., and Vorholt, J. A. (2016). The plant microbiota: populations. Curr. Biol. 17, 407–411. doi: 10.1016/j.cub.2006.12.047 systems-level insights and perspectives. Annu. Rev. Genet. 50, 211–234. Lam, S. S., and Howell, K. S. (2015). Drosophila-associated yeast species in vineyard doi: 10.1146/annurev-genet-120215-034952 ecosystems. FEMS Microbiol. Lett. 362:fnv170. doi: 10.1093/femsle/fnv170 Olesen, J. E., Trnka, M., Kersebaum, K. C., Skjelvåg, A., Seguin, B., Peltonen-Sainio, Legras, J.-L., Galeote, V., Bigey, F., Camarasa, C., Marsit, S., Nidelet, T., et al. P., et al. (2011). Impacts and adaptation of European crop production systems (2018). Adaptation of S. cerevisiae to fermented food environments reveals to climate change. Eur. J. Agron. 34, 96–112. remarkable genome plasticity and the footprints of domestication. Mol. Biol. Pagliarini, E., Laureati, M., and Gaeta, D. (2013). Sensory descriptors, hedonic Evol. 35, 1712–1727. doi: 10.1093/molbev/msy066 perception and consumer’s attitudes to Sangiovese red wine deriving from Liu, D., Gao, Y., Li, X.-X., Li, Z., and Pan, Q.-H. (2015). Attenuated UV radiation organically and conventionally grown grapes. Front. Psychol. 4:896. doi: 10. alters volatile profile in cabernet sauvignon grapes under field conditions. 3389/fpsyg.2013.00896 Molecules 20, 16946–16969. doi: 10.3390/molecules200916946 Palliotti, A., Tombesi, S., Silvestroni, O., Lanari, V., Gatti, M., and Poni, S. (2014). Luo, Y., Wan, S., Hui, D., and Wallace, L. L. (2001). Acclimatization of soil Changes in vineyard establishment and canopy management urged by earlier respiration to warming in a tall grass prairie. Nature 413, 622–625. doi: 10. climate-related grape ripening: a review. Sci. Hortic. 178, 43–54. doi: 10.1016/j. 1038/35098065 scienta.2014.07.039 Mandl, K., Schieck, J., Silhavy-Richter, K., Alexander, P., Schneider, V., and Pancher, M., Ceol, M., Corneo, P. E., Longa, C. M. O., Yousaf, S., Pertot, I., Schmidt, H.-P. (2015). Vines take up yeasts from soil and transport them et al. (2012). Fungal endophytic communities in grapevines (Vitis vinifera through the vine to the stem and skins of grapes. Ithaka J. 349–355. L.) respond to crop management. Appl. Environ. Microbiol. 78, 4308–4317. Marasco, R., Rolli, E., Fusi, M., Michoud, G., and Daffonchio, D. (2018). Grapevine doi: 10.1128/AEM.07655-11 rootstocks shape underground bacterial microbiome and networking but not Perazzolli, M., Antonielli, L., Storari, M., Puopolo, G., Pancher, M., Giovannini, O., potential functionality. Microbiome 6:3. doi: 10.1186/s40168-017-0391-2 et al. (2014). Resilience of the natural phyllosphere microbiota of the grapevine Martínez, C., Cosgaya, P., Vásquez, C., Gac, S., and Ganga, A. (2007). High degree to chemical and biological pesticides. Appl. Environ. Microbiol. 80, 3585–3596. of correlation between molecular polymorphism and geographic origin of wine doi: 10.1128/aem.00415-14 yeast strains. J. Appl. Microbiol. 103, 2185–2195. doi: 10.1111/j.1365-2672.2007. Pereira, G. E., Gaudillere, J.-P., Van Leeuwen, C., Hilbert, G., Lavialle, O., 03493.x Maucourt, M., et al. (2005). 1H NMR and chemometrics to characterize mature

Frontiers in Microbiology | www.frontiersin.org 3312 November 2019 | Volume 10 | Article 2679 fmicb-10-02679 November 20, 2019 Time: 15:49 # 13

Liu et al. Regional Traits of Wine From Microbes

grape berries in four wine-growing areas in Bordeaux, France. J. Agric. Food Taylor, M. W., Tsai, P., Anfang, N., Ross, H. A., and Goddard, M. R. Chem. 53, 6382–6389. doi: 10.1021/jf058058q (2014). Pyrosequencing reveals regional differences in fruit-associated fungal Pinto, C., Pinho, D., Cardoso, R., Custódio, V., Fernandes, J., Sousa, S., et al. (2015). communities. Environ. Microbiol. 16, 2848–2858. doi: 10.1111/1462-2920. Wine fermentation microbiome: a landscape from different Portuguese wine 12456 appellations. Front. Microbiol. 6:905. doi: 10.3389/fmicb.2015.00905 Tofalo, R., Perpetuini, G., Fasoli, G., Schirone, M., Corsetti, A., and Suzzi, Pinto, C., Pinho, D., Sousa, S., Pinheiro, M., Egas, C., and Gomes, A. C. G. (2014). Biodiversity study of wine yeasts belonging to the “terroir” (2014). Unravelling the diversity of grapevine microbiome. PLoS One 9:e85622. of Montepulciano d’Abruzzo “Colline Teramane” revealed Saccharomyces doi: 10.1371/journal.pone.0085622 cerevisiae strains exhibiting atypical and unique 5.8 S-ITS restriction patterns. Portillo, M. D. C., Franquès, J., Araque, I., Reguant, C., and Bordons, A. (2016). Food Microbiol. 39, 7–12. doi: 10.1016/j.fm.2013.10.001 Bacterial diversity of grenache and carignan grape surface from different Valero, E., Schuller, D., Cambon, B., Casal, M., and Dequin, S. (2005). vineyards at Priorat wine region (Catalonia, Spain). Int. J. Food Microbiol. 219, Dissemination and survival of commercial wine yeast in the vineyard: a large- 56–63. doi: 10.1016/j.ijfoodmicro.2015.12.002 scale, three-years study. FEMS Yeast Res. 5, 959–969. doi: 10.1016/j.femsyr. Reynolds, A. G., and Heuvel, J. E. V. (2009). Influence of grapevine training systems 2005.04.007 on vine growth and fruit composition: a review. Am. J. Enol. Vitic. 60, 251–268. Van Leeuwen, C. (2010). Terroir: the effect of the physical environment on vine Robinson, A. L., Adams, D. O., Boss, P. K., Heymann, H., Solomon, P. S., growth, grape ripening and wine sensory attributes. Manag. Wine Qual. 1, and Trengove, R. D. (2012). Influence of geographic origin on the sensory 273–315. doi: 10.1533/9781845699284.3.273 characteristics and wine composition of Vitis vinifera cv. Cabernet Sauvignon Van Leeuwen, C., Friant, P., Chone, X., Tregoat, O., Koundouras, S., and wines from Australia. Am. J. Enol. Vitic. 63, 467–476. doi: 10.5344/ajev.2012. Dubourdieu, D. (2004). Influence of climate, soil, and cultivar on terroir. Am. J. 12023 Enol. Vitic. 55, 207–217. Romano, P., Fiore, C., Paraggio, M., Caruso, M., and Capece, A. (2003). Function Van Leeuwen, C., and Seguin, G. (2006). The concept of terroir in viticulture. of yeast species and strains in wine flavour. Int. J. Food Microbiol. 86, 169–180. J. Wine Res. 17, 1–10. doi: 10.1080/09571260600633135 doi: 10.1016/s0168-1605(03)00290-3 Vaudour, E., Costantini, E., Jones, G., and Mocali, S. (2015). An overview of the Ross, C. F., Weller, K. M., Blue, R. B., and Reganold, J. P. (2009). Difference testing recent approaches to terroir functional modelling, footprinting and zoning. Soil of merlot produced from biodynamically and organically grown wine grapes. 1, 287–312. doi: 10.5194/soil-1-287-2015 J. Wine Res. 20, 85–94. doi: 10.1080/09571260903169423 Vigentini, I., De Lorenzis, G., Fabrizio, V., Valdetara, F., Faccincani, M., Panont, Ruiz-Pérez, C. A., Restrepo, S., and Zambrano, M. M. (2016). Microbial and C. A., et al. (2015). The vintage effect overcomes the terroir effect: a three year functional diversity within the phyllosphere of Espeletia species in an Andean survey on the wine yeast biodiversity in Franciacorta and Oltrepò Pavese, two high-mountain ecosystem. Appl. Environ. Microbiol. 82, 1807–1817. doi: 10. northern Italian vine-growing areas. Microbiology 161, 362–373. doi: 10.1099/ 1128/AEM.02781-15 mic.0.000004 Sabate, J., Cano, J., Esteve-Zarzoso, B., and Guillamón, J. M. (2002). Isolation and Vitulo, N., Lemos, W. J. F. Jr., Calgaro, M., Confalone, M., Felis, G. E., identification of yeasts associated with vineyard and winery by RFLP analysis Zapparoli, G., et al. (2019). Bark and grape microbiome of Vitis of ribosomal genes and mitochondrial DNA. Microbiol. Res. 157, 267–274. vinifera: influence of geographic patterns and agronomic management doi: 10.1078/0944-5013-00163 on bacterial diversity. Front. Microbiol. 9:3203. doi: 10.3389/fmicb.2018. Schmid, F., Moser, G., Müller, H., and Berg, G. (2011). Functional and structural 03203 microbial diversity in organic and conventional viticulture: organic farming Wang, C., García-Fernández, D., Mas, A., and Esteve-Zarzoso, B. (2015). Fungal benefits natural biocontrol agents. Appl. Environ. Microbiol. 77, 2188–2191. diversity in grape must and wine fermentation assessed by massive sequencing, doi: 10.1128/AEM.02187-10 quantitative PCR and DGGE. Front. Microbiol. 6:1156. doi: 10.3389/fmicb.2015. Setati, M. E., Jacobson, D., Andong, U.-C., and Bauer, F. (2012). The vineyard 01156 yeast microbiome, a mixed model microbial map. PLoS One 7:e52609. Webb, L., Whetton, P., Bhend, J., Darbyshire, R., Briggs, P., and Barlow, E. doi: 10.1371/journal.pone.0052609 (2012). Earlier wine-grape ripening driven by climatic warming and drying Setati, M. E., Jacobson, D., and Bauer, F. F. (2015). Sequence-based and management practices. Nat. Clim. Change 2, 259–264. doi: 10.1038/ analysis of the Vitis vinifera L. cv Cabernet Sauvignon grape must nclimate1417 mycobiome in three South African vineyards employing distinct White, R. (2015). Understanding Vineyard Soils. Oxford: Oxford University Press. agronomic systems. Front. Microbiol. 6:1358. doi: 10.3389/fmicb.2015. Zarraonaindia, I., Owens, S. M., Weisenhorn, P., West, K., Hampton- 01358 Marcell, J., Lax, S., et al. (2015). The soil microbiome influences Sheik, C. S., Beasley, W. H., Elshahed, M. S., Zhou, X., Luo, Y., and Krumholz, L. R. grapevine-associated microbiota. mBio 6:e02527-14. doi: 10.1128/mBio.02 (2011). Effect of warming and drought on grassland microbial communities. 527-14 ISME J. 5, 1692–1700. doi: 10.1038/ismej.2011.32 Zhang, J., Wang, E. T., Singh, R. P., Guo, C., Shang, Y., Chen, J., et al. (2019). Singh, P., Santoni, S., This, P., and Péros, J.-P. (2018). Genotype-environment Grape berry surface bacterial microbiome: impact from the varieties and clones interaction shapes the microbial assemblage in grapevine’s phyllosphere and in the same vineyard from central China. J. Appl. Microbiol. 126, 204–214. carposphere: an NGS approach. Microorganisms 6:E96. doi: 10.1111/jam.14124 Smart, R., and Robinson, M. (1991). Sunlight Into Wine: a Handbook for Winegrape Zhu, Y.-G., Gillings, M., Simonet, P., Stekel, D., Banwart, S., and Penuelas, J. (2017). Canopy Management. Broadview, SA: Winetitles. Microbial mass movements. Science 357, 1099–1100. Stefanini, I., and Cavalieri, D. (2018). Metagenomic approaches to investigate the contribution of the vineyard environment to the quality of wine fermentation: Conflict of Interest: The authors declare that the research was conducted in the potentials and difficulties. Front. Microbiol. 9:991. doi: 10.3389/fmicb.2018. absence of any commercial or financial relationships that could be construed as a 00991 potential conflict of interest. Stefanini, I., Dapporto, L., Legras, J.-L., Calabretta, A., Di Paola, M., De Filippo, C., et al. (2012). Role of social wasps in Saccharomyces cerevisiae ecology and Copyright © 2019 Liu, Zhang, Chen and Howell. This is an open-access article evolution. Proc. Natl. Acad. Sci. U.S.A. 109, 13398–13403. doi: 10.1073/pnas. distributed under the terms of the Creative Commons Attribution License (CC BY). 1208362109 The use, distribution or reproduction in other forums is permitted, provided the Swiegers, J., Bartowsky, E., Henschke, P., and Pretorius, I. (2005). Yeast and original author(s) and the copyright owner(s) are credited and that the original bacterial modulation of wine aroma and flavour. Aust. J. Grape Wine Res. 11, publication in this journal is cited, in accordance with accepted academic practice. No 139–173. doi: 10.1111/j.1755-0238.2005.tb00285.x use, distribution or reproduction is permitted which does not comply with these terms.

Frontiers in Microbiology | www.frontiersin.org 3413 November 2019 | Volume 10 | Article 2679 2.2 Supplementary data

Table S1 Studies of affect factors on vineyard and wine microbial communities

Microorganisms Scale Materials Methods Factors References

Geography Bacteria, fungi Multiple regions Must Culture-independent Location, longitude, latitude, altitude Bokulich et al. 2014 Gayevskiy and Goddard, Yeasts Multiple regions Must, ferments Culture-dependent Location 2012 S. cerevisiae Multiple regions Soil, grape juice Culture-dependent Location Knight and Goddard, 2015 Bacteria Within region Soil Culture-independent Location, AVA, elevation, slope, latitude Burns et al. 2015 Bacteria, fungi Within region Must, ferments Culture-independent Location: winery-specific Stefanini et al. 2016 Yeasts Within region ferments Culture-dependent Location: vineyard-specific Tofalo et al. 2014 Bacteria, fungi Within region Grapes, leaves Culture-independent Location, spatial distance Miura et al. 2017 Location, orientation, intra- < inter- Bacteria Within region Musts, ferments samples Culture-independent Portillo et al. 2016 vineyard Yeasts Within region Grapes Culture-dependent Location, intra- > intervineyard Setati et al. 2012 Bacteria Multiple regions Grapes, bark Culture-independent Location, bark > grapes Vitulo et al. 2018 Bacteria Multiple regions Soil, grapes Culture-independent Location Mezzasalma et al. 2018 Bacteria, fungi Multiple regions Grapes, leaves Culture-independent Location, leaves > grapes Singh et al. 2018 Morrison-Whittle et al. Fungi Multiple regions Soil, grapes and ferments, bark Culture-independent Location 2018 Climate Macro-climate: temperature, net Macroclimate, precipitation, evapotranspiration, relative Bacteria, fungi mesoclimate, Must Culture-independent Bokulich et al. 2014 humidity, wind, soil termperature; Meso- microclimate & Micro- climate: vintage Culture-dependent/ Yeasts Macroclimate Grapes independent Temperature, rainfall, relatively humidity Jara et al. 2016 Macroclimate, Yeasts Air, must, wine Culture-dependent Vintage Vigentini et al. 2015 mesoclimate

35 Mesoclimate, Soil, roots, leaves, Bacteria Culture-independent Vintage: temperature Zarraonaindia et al. 2015 microclimate flowers/grapes Fungi Mesoclimate Grapes, must Culture-independent Vintage: rainfall Stefanini et al. 2017 Yeasts Mesoclimate Grapes Culture-dependent Vintage Sabate et al. 2002 Bacteria Mesoclimate Soil Culture-independent Precipitation Burns et al. 2015 Fungi Microclimate Grapes Culture-dependent Inter-vine or intra-vine climate Setati et al. 2012 Soil Soil, roots, leaves, Bacteria Within region flowers/grapes Culture-independent Soil moisture, temperature, pH, C, C:N Zarraonaindia et al. 2015 Soil type, texture, water content, clay Bacteria Within region Soil Culture-independent Burns et al. 2015 content, pH, PMN, DOC, C, N, C:N Anthropogenic Practices Vineyard management Pesticides, herbicides and fertilisers Bacteria, fungi One vineyard Leaves Culture-independent Chemical treatments Pinto et al. 2014 Culture-dependent/ Escribano-Viana et al. Bacteria, fungi One vineyard Grapes, must, wine independent Chemical and biological Fungicides 2018 Bacteria, fungi Multi vineyards Leaves Culture-independent Chemical and biological Fungicides Perazzolli et al. 2014 Bacteria, fungi One vineyard Soil, grapes Culture-independent Chemical hebicides Chou et al. 2018 Bacteria Multi vineyards Soil Culture-independent Fertilisation, cover crop, tillage Burns et al. 2016

Agricultural systems Conventional/ Biodynamic/ Integrated Fungi Multi vineyards Grapes Culture-dependent system Setati et al. 2012 Yeasts Multi vineyards Grapes, must Culture-dependent Conventional/ Organic system Cordero-Bueso et al. 2011 Conventional/ Biodynamic/ Integrated Fungi Multi vineyards Grapes Culture-independent system Setati et al. 2015 Culture-dependent/ Fungi Multiple regions Grapes independent Conventional/ Organic system Martins et al. 2014 Conventional/ Biodynamic/ Integrated Yeasts Multi vineyards Must, wine Culture-dependent system Bagheri et al. 2015 Fungi One vineyard Grapes, must, wine Culture-independent Conventional/ Organic/ Ecophyto system Grangeteau et al. 2017 Culture-dependent/ Yeasts Multi vineyards Grapes, must, wine independent Conventional/ Organic system Milanović et al. 2013

36 Culture-dependent/ Bacteria Multiple regions Grapes independent Conventional/ Organic system Martins et al. 2012 Bacteria, fungi One vineyard Grapes, leaves, shoots Culture-dependent Conventional/ Organic system Schmid et al. 2011 Bacteria, fungi Multi vineyards Grapes, must, wine Culture-independent Conventional/ Biodynamic system Mezzasalma et al. 2017 Bacteria, fungi Multi vineyards Grapes, leaves Culture-independent Conventional/ Organic system Miura et al. 2017 Conventional/ Organic/ Biodynamic Bacteria, fungi One vineyard Grapes Culture-independent system Kecskeméti et al. 2016 Culture-dependent/ Fungi Multi vineyards Shoots independent Organic/ Integrated system Pancher et al. 2012 Conventional/ Organic/ Biodynamic Bacteria Multi vineyards Soil Culture-independent system Burns et al. 2016 Soil, bark, grapes, harvested Morrison-Whittle et al. Fungi Multi vineyards juice Culture-independent Conventional/ Biodynamic system 2017 Bacteria Multi vineyards Grapes, bark Culture-independent Conventional/ Biodynamic system Vitulo et al. 2019

Winemaking practices Fungi One vineyard Grapes, must, wine Culture-independent Pressing, SO2 Grangeteau et al. 2017 Bacteria, fungi One vineyard Must, wine Culture-independent SO2 (dose-dependent) Bokulich et al. 2015 Bacteria One vineyard Must, wine Culture-independent SO2 Piao et al. 2015

37 CHAPTER THREE

The fungal microbiome is an important component of vineyard ecosystems

and correlates with regional distinctiveness of wine

Status of chapter:

This chapter has been published in mSphere journal. The co-authorship form is presented at the start of this thesis, after the acknowledgements.

Liu D., Zhang P., Chen D., Howell K.S.. (2020). The fungal microbiome is an important component of vineyard ecosystems and correlates with regional distinctiveness of wine. mSphere 5: e00534-20.

38 RESEARCH ARTICLE Applied and Environmental Science crossm

The Fungal Microbiome Is an Important Component of Vineyard Ecosystems and Correlates with Regional Distinctiveness of Wine

Di Liu,a Qinglin Chen,a Pangzhen Zhang,a Deli Chen,a Kate S. Howella aSchool of Agriculture and Food, Faculty of Veterinary and Agricultural Sciences, University of Melbourne, Parkville, Australia

ABSTRACT The flavors of fermented plant foods and beverages are formed by microorganisms, and in the case of wine, the location and environmental features of the vineyard site also imprint the wine with distinctive aromas and flavors. Microbial growth and metabolism play an integral role in wine production, by influencing grapevine health, wine fermentation, and the flavor, aroma, and quality of finished wines. The contributions by which microbial distribution patterns drive wine metabolites are unclear, and while flavor has been correlated with fungal and bacterial composition for wine, bacterial activity provides fewer sensorially active biochemical conversions than fungi in wine fermentation. Here, we collected samples across six distinct wine-growing areas in southern Australia to investigate regional distribution patterns of fungi and bacteria and the association with wine chemical composition. Results show that both soil and must microbiota distinguish wine-growing regions. We found a relationship between microbial and wine metabolic profiles under different environmental conditions, in particular measures of soil properties and weather. Fungal communities are associated with wine regional distinctiveness. We found that the soil microbiome is a source of grape- and must-associated fungi and suggest that weather and soil could influence wine characteristics via the soil fungal community. Our report describes a comprehensive scenario of wine microbial biogeography where microbial diversity responds to the surrounding environment and correlates with wine composition and regional characteristics. These findings provide perspectives for thoughtful human practices to optimize food composition through understanding fungal activity and abundance. IMPORTANCE The composition of soil has long been thought to provide wine with characteristic regional flavors. Here, we show that for vineyards in southern Australia, the soil fungal communities are of primary importance for the aromas found in wines. We propose a mechanism by which fungi can move from the soil through Citation Liu D, Chen Q, Zhang P, Chen D, Howell KS. 2020. The fungal microbiome is an the vine. important component of vineyard ecosystems and correlates with regional distinctiveness of KEYWORDS wine regionality, microbial biogeography, fungal diversity, climate, soil, wine. mSphere 5:e00534-20. https://doi.org/10 soil microbiology, yeasts .1128/mSphere.00534-20. Editor Aaron P. Mitchell, University of Georgia Copyright © 2020 Liu et al. This is an open- egional distinctiveness of wine traits, collectively known as “terroir,” can be mea- access article distributed under the terms of the Creative Commons Attribution 4.0 Rsured by chemical composition and sensory attributes (1–3), and this variation has International license. been related to the physiological responses of grapevines to local environments, such Address correspondence to Kate S. Howell, as soil properties (e.g., soil type, texture, and nutrient availability), climate (temperature, [email protected]. precipitation, and solar radiation), topography, and human-driven agricultural practices Fungi determine terroir and give flavour (4–6). Wines made from the same grape cultivar but grown in different regions are to wine. @lifeonthefly15 appreciated for their regional diversity, increasing price premiums and market demand Received 15 June 2020 Accepted 29 July 2020 (5). However, the vineyard and winery factors that drive regional wine quality traits Published 12 August 2020 remain elusive.

July/August 2020 Volume 5 Issue 4 e00534-20 39 msphere.asm.org 1 Liu et al.

Microorganisms, including yeasts, filamentous fungi, and bacteria, originate in the vineyard, are impacted upon by the built environment (winery), and play a decisive role in wine production and quality of the final wine (7–9). The fermentative conversion of grape must (or juice) into wine is a complex and dynamic process, involving numerous transformations by multiple microbial species (10). The majority of fermentations involve Saccharomyces yeasts conducting alcoholic fermentation (AF) and lactic acid bacteria (LAB) for malolactic fermentation (MLF), but many other species are present and impact the chemical composition of the resultant wine (11, 12). Recent studies propose the existence of nonrandom geographical patterns of microbiota in grapes and wines (13–19). Few studies have explored the associations between microbial communities and wine chemical composition (20, 21). Bokulich et al. (20) suggested that wine metabolites correlated with the bacterial and fungal consortia. There was a weaker correlation for fungi than bacteria with the metabolic profiles in finished Cabernet Sauvignon wines which was attributed to bacterial bioconversions during MLF. Knight et al. (21) showed empirically that regionally distinct Saccharomyces cerevisiae populations drove metabolic distinctiveness in the resultant wines, but S. cerevisiae is just one fungal species associated with winemaking. The diverse taxonomy and biochemical diversity of fungi in general are known to make important contributions to plant health and function, but their occurrence and impact beyond Saccharo- myces spp. have not been comprehensively investigated in soil, grapes, and vines. Learning whether the fungi present in vineyard ecosystems correspond to and impact upon wine production could give valuable information about how vine health and wine flavor are linked. The composition and structure of vineyard soil have long been believed to be of great importance in determining wine characteristics and flavor. Vineyard soil provides the grapevine with water and nutrients, and soil type and properties profoundly affect vine growth and development (5). Soil-borne microbiota associates with grapevines in a beneficial, commensal, or pathogenic way and determines soil quality and host growth and health. For example, soil microbes can mineralize soil organic matter and trigger plant defense mechanisms and thus influence the flavor and quality of grapes and final wines (22, 23). Alternatively, soil was previously suggested to be a potential source reservoir of grapevine-associated microbiota (15, 24) and some of soil microbes can influence fermentation and contribute to final wine characteristics (8, 24). Overall, biogeographic boundaries can constrain the vineyard soil microbiota (23, 25–28), but correlations between soil microbiota and wine attributes are weak (15). Limited but increasing evidence reveals that environmental heterogeneity conditions microbial biogeography in wine production on different spatial scales (recently reviewed by Liu et al. [29]) (13, 24, 26, 28, 30, 31). Local climatic conditions significantly correlate with microbial compositions in grape musts; for example, precipitation and temperature have been found to correlate with the abundance of filamentous fungi (for example, Cladosporium and Penicillium spp.) and of ubiquitous bacteria (for example, members of the Enterobacteriaceae family) (13), as well as of yeast populations (particularly Hanseniaspora and Metschnikowia spp.) (30). Dispersal of soil microbiota is driven by soil physicochemical properties such as soil texture, soil pH, and carbon (C) and nitrogen (N) pools (24, 26, 27), with some influences from topological characteristics (for example, orientation of the vineyard) (26, 32). Soil microbiome/bacteria may colonize grapes by physical contact (being moved by rain splashes, dust, and winds) (24) or by migration through the plant (xylem/phloem) from the rhizosphere to the phyllosphere (33). Insects help the movement and dispersal of microbes in the vineyard and winery ecosystem; for example, honeybees, social wasps, and drosophilid flies can vector yeasts among different microhabitats (34–36). Vineyard microbes enter the winery in association with grapes or must, so the effects of environmental conditions are finally reflected on microbial consortia in wine fermentation. How environmental conditions modulate microbial ecology from the vineyard to the winery and shape regional distinctiveness of wine is still largely unknown. Here, we initially tested microbial contribution to wine regional characteristics. To

July/August 2020 Volume 5 Issue 4 e00534-20 40 msphere.asm.org 2 Wine Terroir Is Associated with Environmental Fungi

FIG 1 Wine metabolome shows regional variation across six wine-growing regions in 2017. Shown are ␣-diversity (Shannon index) (A) and PCoA (B) based on Bray-Curtis dissimilarity obtained from comparing volatile proﬁles.

tackle this issue, we sampled microbial communities from the vineyard to the winery across six geographically separated wine-producing regions in southern Australia. We evaluated the volatile chemicals of wines made with Pinot Noir grapes to validate the hypothesis that these different regions have differently ﬂavored wines. Using next- generation sequencing (NGS) to proﬁle bacterial and fungal communities, we demonstrate that the soil and must microbiota exhibit distinctive regional patterns and that this correlates to the wine metabolome. Associations among soil and wine microbiome, abiotic factors (weather and soil properties), and wine regionality were modeled by random forest and structural equation modeling (SEM), highlighting the important contributions of fungal communities. We then tested a potential route of transmission of wine-related fungi from the soil to the grapes by isolating yeasts from the xylem/ phloem of grapevines to further explore the role of fungi in wine regionality. Using vineyards, grapes, and wine as a model food system, we have related the regional identity of an agricultural commodity to biotic components in the growing system to show the importance of conserving regional microbial diversity to produce distinctive foods and beverages.

RESULTS Chemical composition/aroma profiles separate wines based on geographic location. Using headspace solid-phase microextraction gas-chromatographic mass- spectrometry (HS-SPME–GC-MS), we analyzed the volatile compounds of Pinot Noir wine samples (MLF-End) to represent wine metabolite profiles coming from different growing regions and compared the results directly to the microbial communities inhabiting the musts from which these wines were fermented. In all, 88 volatile compounds were identified in these wines, containing 48 regionally differential compounds (see Table S2 in the supplemental material). Here, we used ␣- and ␤-diversity measures to further elucidate wine complexity and regionality, respectively. In wines of 2017 vintage, ␣-diversity varied with regional origins (analysis of variance [ANOVA], F ϭ 36.021, P Ͻ 0.001), with higher Shannon indices observed for the wines from regions of Mornington, Yarra Valley, and Gippsland (H ϭ 2.17 Ϯ 0.05) than for those from other regions (H ϭ 1.94 Ϯ 0.03) (Fig. 1A). Overall, wine aroma profiles displayed significant regional differentiation across both vintages based on Bray-Curtis dissimilarity (permutational multivariate analysis of variance [PERMANOVA], coefficient of determination [R2] ϭ 0.566, P Ͻ 0.001) and the clustering patterns became more distinct and the R2 values improved in comparisons of regional differences in wines of

July/August 2020 Volume 5 Issue 4 e00534-20 41 msphere.asm.org 3 Liu et al.

FIG 2 Regional patterns of vineyard soil microbial communities demonstrated by Bray-Curtis distance PCoA of bacterial communities (A) and fungal communities (B).

2017 vintage (PERMANOVA; R2 ϭ 0.703, P Ͻ 0.001) (Table S3). Principal-coordinate analysis (PCoA) showed that 74.5% of the variance was explained by the first two principal coordinates in 2017, and on PCo1 there were some wines within regions grouped together (Fig. 1B). Microbial ecology from the vineyard to the winery. To test the role of microbial diversity in regional traits of wine from the vineyard to the winery, 150 samples covering soils, musts, and fermentations were collected to analyze wine-related microbiota. A total of 11,508,480 16S rRNA and 12,403,610 internal transcribed spacer (ITS) high-quality sequences were generated from the samples, which were clustered into 13,689 bacterial and 8,373 fungal operational taxonomic units (OTUs) with a threshold of 97% pairwise identity. The dominant bacterial taxa across all soil samples were Actinobacteria, Proteobac- teria, Acidobacteria, Chloroflexi, Verrucomicrobia, Bacteroidetes, Gemmatimonadetes, Fir- micutes, Planctomycetes, and Nitrospirae (see Fig. S2A in the supplemental material). Compared with bacteria, soil fungal communities were less diverse (Table S4). Ascomy- cota was the most abundant and diverse phylum of fungi, accounting for 72% of reads, followed by Basidiomycota, Mortierellomycota, Chytridiomycota, and Olpidiomycota (Fig. S2B). The microbial diversity (␣-diversity, Shannon index) differed significantly ϭ Ͻ between regions for both bacteria and fungi (ANOVA; Fbacteria 4.645, P 0.01; ϭ Ͻ Ffungi 4.913, P 0.01). Soil microbial communities varied widely across different grape-growing regions, and significant differences were observed in both bacterial taxonomic dissimilarity and fungal taxonomic dissimilarity based on Bray-Curtis dis- 2 ϭ Ͻ tances matrices at the OTU level (PERMANOVA; R bacteria 0.318, P 0.001; 2 ϭ Ͻ R fungi 0.254, P 0.001), with clearer differences within a single vintage 2 ϭ Ͻ 2 ϭ Ͻ (PERMANOVA; R bacteria 2017 0.392, P 0.001; R fungi 2017 0.419, P 0.001) (Ta- ble S3). In 2017, soil samples from the different growing regions (except Yarra Valley and Gippsland) were able to be discriminated based on fungal communities (Fig. 2B), whereas regional separation of bacteria was weaker, with overlap of regions (Fig. 2A). Among grape musts, bacterial communities of both vintages across six winegrowing regions consisted of the ubiquitous bacteria Enterobacteriales, Rhizobiales, Burkholderiales, Rhodospirillales, Actinomycetales, Sphingomonadales, Pseudomonadales, Saprospirales, and Xanthomonadales, which do not participate in wine fermentations or spoilage (7). Members of the LAB Lactobacillales, responsible for malolactic fermentation, were present in low abundance (0.4% on average) in the must (Fig. 3A). Fungal

July/August 2020 Volume 5 Issue 4 e00534-20 42 msphere.asm.org 4 Wine Terroir Is Associated with Environmental Fungi

FIG 3 Microbiota exhibit regional differentiation in musts for both bacterial and fungal profiles. The stage of fermentation influences microbial composition of bacteria and fungi. (A) Must bacterial taxa with greater than 1.0% relative abundance at the order level, and Lactobacillales (average abundance, 0.4%). (B) Must fungal taxa with greater than 1.0% relative abundance at the genus level. (C and D) Linear discriminant analysis (LDA) effect size (LEfSe) taxonomic cladograms comparing all musts and wines categorized by fermentation stage. Significantly discriminant taxon nodes (C, bacteria; D, fungi) are colored and branch areas are shaded according to the highest-ranked stage for that taxon. For each taxon detected, the corresponding node in the taxonomic cladogram is colored according to the highest-ranked group for that taxon. If the taxon is not significantly differentially represented between sample groups, the corresponding node is colored yellow. profiles were dominated by filamentous fungi, mostly of the genera Aureobasidium, Cladosporium, Botrytis, Epicoccum, Penicillium, Alternaria, and Mycosphaerella, with notable populations of yeasts, including Saccharomyces, Hanseniaspora, and Meyerozyma, as well as the Basidiomycota genus Rhodotorula (Fig. 3B). Pinot Noir musts exhibited significant regional patterns for fungal communities across vintages 2017 and 2018 2 ϭ based on Bray-Curtis dissimilarity at the OTU level (PERMANOVA; R fungi 0.292, P Ͻ 0.001) but no significant differences for bacterial communities across both vintages 2 ϭ ϭ (PERMANOVA; R bacteria 0.108, P 0.152) (Table S3), as well as regarding community ϭ ϭ ϭ Ͻ diversity (ANOVA; Fbacteria 1.567, P 0.374; Ffungi 5.142, P 0.01) (Table S4). Within the 2017 vintage, both bacteria and fungi in the must showed distinctive compositions 2 ϭ Ͻ 2 ϭ on the basis of region (PERMANOVA; R bacteria 0.342, P 0.001; R fungi 0.565, P Ͻ 0.001) (Fig. 3A and B) (Table S3), with a more distinct trend and improved R2 coefficient values for fungi. Notably, the relative abundances of Saccharomyces yeasts between regions ranged widely from 1.3% (Macedon Range) to 65.6% (Gippsland) (Fig. 3B). As the wine fermentation proceeded, fermentative populations, including yeasts and LAB, grew and dominated, thus reshaping the community diversity (Fig. S3A and B) and composition (Fig. S3C and D). Fungal species diversity collapsed as alcoholic fermentation progressed (ANOVA; F ϭ 6.724, P Ͻ 0.01) (Fig. S3B), while the impact of the fermentation on bacterial diversity was insignificant (ANOVA; F ϭ 1.307, P ϭ 0.301), with a slight decrease at early stages and recovery at the end of fermentation (Fig. S3A). Linear discriminant analysis (LDA) effect size (LEfSe) determinations further identified differentially abundant taxa (Kruskal-Wallis rank sum test, ␣ Ͻ 0.05) associated with fermentation stages (Fig. 3). For fungal populations, Dothideomycetes (including Aureo- basidium and Cladosporium), Debaryomycetaceae (notably, yeast Meyerozyma guillier-

July/August 2020 Volume 5 Issue 4 e00534-20 43 msphere.asm.org 5 Liu et al. mondii), Penicillium corylophilum, and Filobasidium oeirense were significantly abundant in the grape must, including Leotiomycetes, Sarocladium, and Vishniacozyma victoriae in early fermentations (AF), Saccharomycetes yeasts (notably S. cerevisiae) in mid- fermentations (AF-Mid), and Tremellales (notably yeast Vishniacozyma victoriae)atthe end of fermentation (AF-End) (Fig. 3D). For bacterial communities, Acidobacteriia (spoilage), Chloroflexi, Deltaproteobacteria, Sphingobacteriia, Cytophagia, Planococcaceae, and Rhizobiaceae were observed with higher abundances in the must; Proteobacteria (including Burkholderiaceae and Tremblayales, spoilage) and Micrococcus in the AF; Burk- holderia spp. in the AF-Mid; Rhodobacterales and Pseudonocardiaceae in the AF-End; and LAB Leuconostocaceae (notably Oenococcus) in the MLF-End (Fig. 3C). Regional differences in microbial profiles were not significant in the finished wines (PERMANOVA; 2 ϭ ϭ 2 ϭ ϭ R bacteria 0.149, P 0.321; R fungi 0.109, P 0.205) (Table S3). To uncover the impact of growing season (vintage) on wine regionality and related microbiota, five vineyards in Mornington were sampled in 2017 and 2018 to perform comparisons within and between vintages. Within these five vineyards alone, both microbial communities and wine aroma showed a significant influence from vintage effects (Table S3). In large-scale comparisons of all samples, vintage only weakly impacted microbial and wine aroma profiles; in particular, an insignificant influence on 2 ϭ ϭ fungi was seen (PERMANOVA; R fungi 0.049, P 0.066) (Table S3). We used 2017 vintage data to further explore microbial biogeography and wine regionality in the following analyses. Multiple factors modify wine regionality in the vineyard. Alongside regional patterns in soil and must microbiota, environmental measures of the wine-growing regions displayed significant differences, such as in C and N levels in soil, solar radiation, and temperature and weather/climatic conditions during the growing season (October 2017–April 2018) (see Table S5 for a complete list). To disentangle the roles of microbial ecology in wine regionality, we used random forest modeling (37) to identify the biotic predictors (soil and must microbial diversity) and abiotic predictors (soil and weather parameters) structuring wine regionality and used structural equation modeling (SEM) (38) to test whether the relationship between microbial diversity and wine regionality would be able to be maintained while accounting for multiple factors simultaneously. The random forest model (R2 ϭ 0.451, P Ͻ 0.01) demonstrated that fungal diversity was a predictor for wine regionality. Not surprisingly, must fungal diversity showed higher importance on the model (increase in the mean square error [MSE]) than soil (Fig. 4). The SEM explained 93.8% of the variance found in the pattern of wine regionality (Fig. 5A). Weather correlated with wine aroma profiles directly (especially MT [mean temperature], MLT [mean low temperature], MinT [minimum temperature], and MSR [mean solar radiation]) and indirectly by powerful effects on soil and must microbial diversity, in particular, showing strong effects on soil fungal diversity (Fig. 5A). Must fungal diversity had the highest direct positive effect on wine aroma characteristics, with direct effects by soil fungal diversity (Fig. 5A). Weather, climate, and soil nutrient pools were related primarily through MSR, MLT, MinT, and MTrans (mean transpiration). Soil properties showed strong effects on soil microbial diversity and must bacterial diversity but weak effects on must fungal diversity (Fig. 5A). Must bacterial diversity had a weak effect on wine aroma profiles, as did soil bacterial diversity. Overall, must fungal diversity was the most important predictor of wine characteristics, followed by soil fungal diversity, as indicated by the standardized total effects from SEM (Fig. 5B), with effects from weather and soil properties operating both directly and indirectly (Fig. 5A). Source tracking of wine-related fungi within vineyard. As shown in the SEM, must fungal diversity was correlated with soil fungal diversity, and the former had higher effects on wine aroma profiles (Fig. 5). Given that soil is a potential source of fungi associated with wine production (14), here, we attempt to uncover the mechanism whereby soil fungi are transported from soil to the grapes. We sampled fungal communities from grapevines and soil and hypothesized that the xylem/phloem was the internal mechanism to transport microbes. A total of 2,140,820 ITS high-quality

July/August 2020 Volume 5 Issue 4 e00534-20 44 msphere.asm.org 6 Wine Terroir Is Associated with Environmental Fungi

FIG 4 Main predictors of wine regionality. The random forest mean predictor importance is shown by percentage of increase in mean square error (MSE) of climate, soil properties, and microbial diversity (Shannon index) according to wine regionality. Soil bac. div., soil bacterial diversity; Soil fung. div., soil fungal diversity; Must bac. div., must bacterial diversity; Must fung. div., must fungal diversity. Signiﬁ- cance levels: *, P Ͻ 0.05; **, P Ͻ 0.01. sequences were generated from soil and grapevine samples (grape, leaf, xylem sap, root), which were clustered into 4,050 fungal OTUs with 97% pairwise identity. Using SourceTracker (39), fungal communities in the must were matched to multiple sources from below the ground to above the ground. Results showed that grape and xylem sap were primary sources of must fungi, with 32.6% and 41.9% contributions, respectively (Fig. 6). The fungal structure of xylem sap was similar to that seen with must (Fig. S4A). Further source tracking revealed that the root and soil contributed 90.2% of fungal OTUs of xylem sap and that the latter contributed 67.9% of the fungi of grapes (Fig. 6). Notably, S. cerevisiae yeasts were found shared between microhabitats of soil, root, xylem sap, grape, and must, with the highest (1.22%) and lowest (0.038%) relative abundances in the root and soil, respectively (Fig. S4B). Could xylem vessels represent a translocation pathway of S. cerevisiae from roots to the aboveground? Chemical analysis of nutrient compositions showed that xylem sap contained nine carbohydrates (predominantly glucose, fructose, and sucrose), 15 amino acids (mainly arginine, aspar- tic acid, and glutamic acid), and six organic acids (primarily oxalic acid), which could be utilized as carbon and nitrogen sources and support yeast growth (Fig. S4E to G) (40). However, no S. cerevisiae yeasts were isolated; distinct isolates of the Basidiomycota yeasts of Cryptococcus spp. (primarily C. saitoi) and Rhodotorula sloofﬁae were found instead (Fig. S4C). The data indicating the exclusive existence of these species were validated by isolation from xylem/phloem sap coming from grapevines grown in the glasshouse (Fig. S4D).

DISCUSSION Microbial ecology can inﬂuence grapevine health and growth, fermentation, ﬂavor characteristics, and wine quality and style (13, 14, 21). We systematically investigated the microbiome from the soil to wine and found that soil and grape must microbiota exhibited regional patterns and that these patterns correlated with resulting wine metabolites. Here, we show that wine regionality is closely associated with fungal ecology, with effects from local weather, climate, and soil properties. A new mechanism to transfer fungi from the soil to grapes and must via xylem sap was investigated. A microbial component of wine terroir. Regional spatial patterns have been proposed for soil and grape must microbiota (13, 26–28, 41). The most abundant

July/August 2020 Volume 5 Issue 4 e00534-20 45 msphere.asm.org 7 Liu et al.

FIG 5 Direct and indirect effects of climate, soil properties, and microbial diversity (Shannon index) on wine regionality. (A and B) Structural equation modeling (SEM) fitted to the diversity of wine aroma profiles (A) and standardized total effects (direct plus indirect effects) derived from the model (B). Climate and soil properties represent composite variables encompassing multiple observed parameters (see Materials and Methods for the complete list of factors used to generate this model). Numbers adjacent to arrows are path coefficients and indicative of the effect size of the relationship. The width of arrows is proportional to the strength of path coefficients. R2 denotes the proportion of variance explained. (A) (0.747* MT) (0.666* MLT) (0.686* MinT) (Ϫ0.875** MSR). (B) (0.753* MinT) (0.772* MLT) (Ϫ0.683* MSR) (Ϫ0.737* MHT) (Ϫ0.843** MaxT). C, soil carbon; N, soil nitrogen; C:N, soil carbon/nitrogen ratio; MSR, mean solar radiation; MT, mean temperature; MLT, mean low temperature; MHT, mean high temperature; MinT, minimum temperature; MaxT, maximum temperature; MTrans, mean transpiration. Significance levels: *, P Ͻ 0.05; **, P Ͻ 0.01; ***, P Ͻ 0.001. bacterial phyla in the vineyard soils in our study were Actinobacteria, Proteobacteria, and Acidobacteria, which are known to be dominant and ubiquitous in vineyard soil (24, 26, 27, 42). Among fungi, we recovered 14 phyla, 30 classes, 65 orders, 125 families, and 216 genera, recording a higher diversity than reported in other wine-producing areas in the world (27, 28, 43). Glomeromycota, the phylum of arbuscular mycorrhizal fungi reported to positively affect grapevine growth, was reported as abundant in New Zealand vineyards (43). In our study, which analyzed amplicon sequences at the ITS region (rather than the D1/D2 region analyzed in the study cited in reference 43), Glomeromycota was recovered with only low frequency from Mornington and Macedon Ranges vineyards. Clearly, there are differences based on the barcoding region but geographic location may also affect distribution (44), as Coller et al. (2019), using the

FIG 6 Fungal communities in musts emerge from multiple sources in the vineyard but primarily from grapes and xylem sap. Percent composition representing the contributions of possible sources from the vineyard to the wine-related fungal community are given for must, grape, and xylem sap.

July/August 2020 Volume 5 Issue 4 e00534-20 46 msphere.asm.org 8 Wine Terroir Is Associated with Environmental Fungi same ITS1F/2 primers as in our study, retrieved Glomeromycota as a core phylum member from vineyards in Italy (27). In the must, both principal fermentation drivers (S. cerevisiae and LAB) and grapevine-associated species (such as Enterobacteriales and Aureobasidium, which may not be active in the fermentation but possibly interact with plants) were present in different abundances among regions (Fig. 3A and B). The order Lactobacillales, representing LAB, was present at an abundance of 0.4% across regions, compared to 29.7% found in California in the United States (13) and 14% in Catalonia in Spain (32). Environmental factors (such as weather and climate) and geographic features structure microbial diversity and biogeography across various habitats in the soil and plant ecosystems (13, 26, 45–47). In this study, we demonstrated that microbial biogeographic communities were distinct in both vineyard soils and grape musts in southern Australia regardless of the growing season/vintage. This aligns with previous studies on wine microbial biogeography and provides further evidence for microbial terroir (reviewed by Liu et al. [29]) (13, 15, 19, 26). Soil bacteria can be used to distinguish wine-growing regions, with impacts from soil properties (Fig. 5A), and this is supported by previous work in this field (24, 26, 41). An interesting finding was that must bacterial diversity is strongly affected by soil properties, in particular, by carbon- nitrogen (C/N) ratios. Previous work has shown that must and soil community structures are similar and that some Enterobacteriales and Actinomycetales species originate from the soil (8, 16). As C/N ratios can be manipulated by composting and cover crops (48), there is an opportunity to manipulate wine microbiota by focusing on vineyard management (29). The soil bacterial microflora is recognized as important for plant growth processes more broadly (49), but fungal diversity beyond endosymbiotic mycorrhizae has not been systematically investigated for grapevines. Here, we show that soil fungal communities are distinct between regions. Our modeling suggests that soil properties and weather strongly affect soil fungal diversity, which was in line with large-scale studies in which climatic factors (especially precipitation) and edaphic factors (especially C/N ratios) were found to be the best predictors of soil fungal richness and community composition (47, 50). Must fungal diversity was also found to be affected by weather and soil properties indirectly via soil fungi that had direct effects on wine aroma profiles (Fig. 5). Considering the limitations represented by the sampling size (15 vineyards) and the data from weather stations, numerous geophysical factors and microclimatic conditions within specific vineyards could explain microbial variations beyond the scope of our measurements. Future studies performed with further sampling within regions and more-precise weather and climate data (for example, real-time weather monitoring within the vineyard) will provide further perspectives with respect to wine microbial biogeography and the response of microbes to local environmental conditions. It is noteworthy that the drivers of microbial patterns change during wine fermentation. Microbial diversity decreases as alcoholic fermentation proceeds, with a clear loss of microorganisms, including filamentous fungi, non-Saccharomyces yeasts (for example, M. guilliermondii), spoilage bacteria (Acidobacteria and Proteobacteria), and other bacteria with unknown fermentative functions (for example, Chloroflexi)(Fig. 3C and D), and the biogeographic trend was lost by the stage of MLF-End (see Table S3 in the supplemental material). This trend was observed more distinctly in fungal communities than in bacterial communities (see Fig. S3A and B in the supplemental material). This was not unexpected as it is clear that fermentation affects fungal populations more strongly than bacterial populations, due to increasing fermentation rate, temperature, and ethanol concentration induced by S. cerevisiae growth (11, 51). In this case, fermentation conditions, such as the chemical environment and interactions and/or competition within the community (11, 52), reshape the observed microbial patterns. Despite the complex microbial ecosystem changes occurring during fermentation, we show that biogeographic patterns in the must could be reflected in the regional metabolic profiling of wine. Our modeling indicates that the indirect effects on wine aroma profiles of weather and soil properties via influencing soil and must microbial

July/August 2020 Volume 5 Issue 4 e00534-20 47 msphere.asm.org 9 Liu et al. diversity are more powerful than the direct effects (Fig. 5). In the resulting wines, the most volatile compounds were alcohols, esters, acids, and aldehydes (Table S2), many of which were likely microbial products. Some compounds, for example, monoterpenes, are derived from grapes and are modified by yeast and bacterial metabolism during fermentation (10). These modellings are potentially important to inform farming practices to structure regional microbial communities that can benefit soil quality and thus crop productivity. Fungal communities distinguish wine quality and style. In grape musts, bacterial and fungal communities exhibit different responses to site-specific and environmental effects. Bacterial regional patterns were not as distinct as fungal regional patterns and were significantly impacted by vintage (Table S3). Although they showed profound relationships with soil properties (for example, C/N ratios) and affected wine fermentation, must bacteria exhibited insignificant effects on wine aroma profiles (Fig. 5). In contrast, fungal communities displayed diverse distribution patterns at the regional scale and were weakly or insignificantly impacted by vintage in this study, aligning with results presented by Bokulich et al. (2014) (13). Soil fungal communities are less diverse than bacterial communities (Fig. S2) (44) but are of more importance to the resultant wine regionality (see Fig. 4 and 5). Must fungi, in particular, the fermentative yeasts, participate in alcoholic fermentation processes and provide aroma compounds to structure wine flavor (10). As indicated by SEM, soil fungal communities are affected by local soil properties and weather and exert impacts on must fungal communities (Fig. 5). One explanation is that grapevines filter soil microbial taxa, selecting for grape and must consortia (53, 54). Beyond fungi, plant fitness is linked strongly to the responses of soil microbial communities to environmental conditions (55). More- sensitive responses of vineyard soil fungi might improve grapevine fitness with respect to local environments, thus enhancing the expression of regional characteristics of grapes and wines. How could yeasts present in the soil be transported to the grape berry? Soil is a source reservoir of grapevine-associated microbiota (Fig. 6), an assertion that is supported by previous publications (15, 24, 41, 56). As well as transporting water and minerals absorbed by roots to the photosynthetic organs, xylem sap is also a microhabitat for microbes that can bear its nutritional environment (40). Here, we investigated xylem sap as a conduit to shape the microbiota in the grape by enrichment of the microbes recruited by roots and transported by xylem sap to the grape berries (24, 33, 49). The isolated yeasts belonged to the Cryptococcus and Rhodotorula genera, indicating that the xylem sap environment is not sterile and can potentially transport yeasts to the phyllosphere. The endophyte Burkholderia phytofirmans strain PsJN has been shown to colonize grapevine roots from the rhizosphere and spread to inflorescence tissues through the xylem (33, 57). While we were unable to find fermentative yeasts in the sap of grapevines, other yeasts (and/or spores) were present and may also be transported within the grapevine as well as making their way to the phyllosphere through other mechanisms (water splashes, insect vectors). As previous studies showed, fermentative yeasts are persistent in vineyards (58, 59, 82) and might be transported through the vine to the grapes (60). We can thus suggest fungi as a signature corresponding to consistent expression of regionality in wine production. Our study results suggest microbial contributions to wine aroma and that such contributions are related to the environment in which they are grown. Whether geographically differential microbiota can actually sculpt wine characteristics must be further empirically addressed. Fungi are implicated in the interrelationship of biotic and abiotic elements in vineyard ecosystems and could potentially be transported internally within the grapevine. Climate and soil properties profoundly structure microbial patterns from the soil to the grape must and ultimately affect the wine metabolic profile. We do not yet know how grapevines recruit their microbiome to maximize physiological development and maintain microbial diversity under local conditions. The addition of our study in Australia to support and extend investigations in other wine-growing

July/August 2020 Volume 5 Issue 4 e00534-20 48 msphere.asm.org 10 Wine Terroir Is Associated with Environmental Fungi regions worldwide contributes to a complex picture of environment-plant-microbe interactions in production of wine. Further studies focusing on empirical experiments will be indispensable to improve understanding of how agricultural production affects the ultimate ﬂavor of foods and beverages.

MATERIALS AND METHODS Sample sites and weather parameters. A total of 15 Vitis vinifera cv. Pinot Noir vineyards were selected in 2017 from among those maintained in Geelong, Mornington Peninsula (Mornington), Macedon Ranges, Yarra Valley, Grampians, and Gippsland in southern Australia, with distances between vineyards ranging from 5 km to 400 km (see Fig. S1 in the supplemental material). All these vineyards are operated under conventional management practices, and the vineyard conditions (altitude, orientation, soil conditions, cover crop) are listed in Table S1 in the supplemental material. In 2018, the sampling from the five vineyards in Mornington Peninsula (all Ͻ20 km apart) was repeated to elucidate the influence of sampling year (vintage) on microbial patterns and wine profiles. Each site’s Global Positioning System (GPS) coordinates (longitude, latitude, altitude) were utilized to extract weekly weather data from the data set provided by Australian Water Availability Project (AWAP). Variables were observed by robust topography, resolving analysis methods at a resolution of 0.05° by 0.05° (approximately 5 km by 5 km) (61). Weekly measurements for all vineyards were extracted for mean solar radiation (MSR), mean high temperature (MHT), mean low temperature (MLT), maximum temperature (MaxT), minimum temperature (MinT), mean temperature (MT), precipitation, mean relative soil moisture, mean evaporation (soil plus vegetation), and mean transpiration (MTrans) in growing seasons (October 2016/2017 to April 2017/ 2018). Collection of soil, plant, must, and ferment samples. In each vineyard, soil samples were collected from three sites covering the top, middle, and bottom of the dominant slope at harvest March to April 2017 (n ϭ 45) and 2018 (n ϭ 15) at depths of 0 to 15 cm and 30 to 50 cm from the grapevine into the interrow (three subsamples were mixed to form a composite sample at each site) (Table S1A). To further investigate fungal ecology in the vineyard, comprehensive vineyard samples (n ϭ 50) were collected from two vineyards 5 km apart in the 2018 vintage (Table S1B). These two vineyards were managed by the same winery, and the viticultural management practices were very similar; for example, grapevines were maintained under vertical shoot positioning trellising systems and the same sprays were applied at the same time of year. Five replicate Pinot Noir vines in each vineyard were selected from the top, middle, and bottom of the dominant slope, covering topological profiles of the vineyard. For each grapevine, the following five different sample types were collected at harvest in March 2018: soil (0 to 15 cm deep, root zone), roots, xylem/phloem sap, leaves, and grapes. Xylem sap (n ϭ 10) was collected from the shoots using a centrifugation method under aseptic conditions (62) (Table S1B). Details of xylem sap collection, nutrient composition analysis, and yeast isolation were provided in Text S1 in the supplemental material. Samples were stored in sterile bags, shipped on ice, and stored at Ϫ80°C until processing. Longitudinal samples were collected to study microbial communities during fermentation at the following five time points: at the must time point (destemmed, crushed grapes prior to fermentation), at early fermentation (AF, with less than 10% of the sugar fermented), at middle of fermentation (AF-Mid, with around 50% of sugar fermented), at the end of fermentation (AF-End, ϳ6°Brix, following pressing but prior to barreling), and at the end of malolactic fermentation (MLF-End, in barrels) (Table S1A). The chemical constituents of the initial musts were similar (Table S1) and were fermented in the respective wineries following similar fermentation protocols of 3 days at a cool temperature (known as cold- soaking) followed by warming the must so that fermentation could commence. Fermentations were conducted without addition of commercial yeasts and bacteria. Two fermentations from Grampians and Mornington did not complete the process and were excluded from analysis, giving wine samples from 13 vineyards in the 2017 vintage. Triplicate subsamples from tanks or barrels (from the top, middle, and bottom) were collected and mixed as composite samples. All samples (n ϭ 90) were frozen immediately after sampling in the winery, shipped on ice to the laboratory, and stored at Ϫ20°C until processing. Soil analysis. Edaphic factors were analyzed to explore the effects of soil properties on wine-related microbiota and aroma profiles. Soil pH was determined in a 2:5 soil/water suspension. Soil carbon (C), nitrogen (N), nitrate, and ammonium were analyzed by Melbourne Trace Analysis for Chemical, Earth and Environmental Sciences (TrACEES) Soil Node, at the University of Melbourne. Total C and N levels were determined using the classic Dumas method of combustion (63) and a Leco TruMac CN analyzer (Leco Corporation, MI, USA) at a furnace temperature of 1,350°C. Nitrate and ammonium were extracted with 2 M KCl and their levels determined on a segmented flow analyzer (SANϩϩ; Skalar, Breda, Netherlands) (63). Wine volatile analysis. To represent the wine aroma, volatile compounds of MLF-End samples were determined using headspace solid-phase microextraction gas-chromatographic mass-spectrometry (HS- SPME–GC-MS) (64, 65) with some modifications. Analyses were performed with an Agilent 6850 GC system and a 5973 mass detector (Agilent Technologies, Santa Clara, CA, USA) equipped with a PAL RSI 120 autosampler (CTC Analytics AG, Switzerland). Briefly, 10 ml wine was added to a 20-ml glass vial with 2 g of sodium chloride and 20 ␮l of internal standard (4-Octanol; 100 mg/liter) and then equilibrated at 35°C for 15 min. A polydimethylsiloxane/divinylbenzene (PDMS/DVB; Supelco) 65-␮m-pore-size SPME fiber was immersed in the headspace for 10 min at 35°C with agitation. The fiber was desorbed in the GC injector for 4 min at 220°C. Volatiles were separated on an Agilent J&W DB-Wax Ultra Inert capillary GC column (30 m by 0.25 mm by 0.25 ␮m) with helium carrier gas used at a flow rate of 0.7 ml/min. The column temperature program was as follows: holding at 40°C for 10 min, increasing at 3.0°C/min to

July/August 2020 Volume 5 Issue 4 e00534-20 49 msphere.asm.org 11 Liu et al.

220°C, and holding at that temperature for 10 min. The temperature of the transfer line of GC and MS was set at 240°C. The ion source temperature was 230°C. The MS was operated in positive electron ionization (EI) mode with scanning over a mass acquisition range of 35 to 350 m/z. Raw data were analyzed with Agilent ChemStation software for qualification and quantification (66). Volatile compounds (n ϭ 88) were identified in wine samples according to retention indices, reference standards, and mass spectra matching performed with the NIST11 library database. A total of 13 successive levels of standard solution in model wine solutions (12% [vol/vol] ethanol saturated with potassium hydrogen tartrate and adjusted to pH 3.5 using 40% [wt/vol] tartaric acid) were analyzed by the same protocol as was used for the wine samples to establish the calibration curves for quantification. Peak areas of volatile compounds were integrated via a target ion model. The concentrations of volatile compounds were calculated with the calibration curves and used for downstream data analysis. DNA extraction and sequencing. Genomic DNA was extracted from plant and soil samples using PowerSoil DNA isolation kits (Qiagen, CA, USA). DNA extraction from soil and xylem sap followed the kit’s protocols. Wine fermentation samples were thawed, and biomass was recovered by centrifugation at 4,000 ϫ g for 15 min, washed three times in ice-cold phosphate-buffered saline (PBS)–1% polyvinylpolypyrrolidone (PVPP), and centrifuged at 10,000 ϫ g for 10 min (12). The obtained pellets were used for DNA extraction following the kit protocol. For the grapevine samples, roots, leaves, and grapes (removed seeds and stems) were ground into powder under the protection of liquid nitrogen with 1% PVPP and DNA was isolated afterward following the kit protocol. DNA extracts were stored at Ϫ20°C until further analysis. Genomic DNA was submitted to the Australian Genome Research Facility (AGRF) for amplification and sequencing. To assess the bacterial and fungal communities, the 16S rRNA gene V3-V4 region and partial fungal internal transcribed spacer (ITS) region were amplified using universal primer pairs 341F/806R (67) and ITS1F/2 (68), respectively. The primary PCRs contained 10 ng DNA template, 2ϫ AmpliTaq Gold 360 master mix (Life Technologies, Australia), and 5 pmol of each primer. A secondary PCR was performed with TaKaRa Taq DNA polymerase (Clontech) to index the amplicons. Amplification were conducted under the following conditions: for bacteria, 95°C for 7 min, followed by 30 cycles of 94°C for 30 s, 50°C for 60 s, and 72°C for 60 s and a final extension at 72°C for 7 min; for fungi, 95°C for 7 min, followed by 35 cycles of 94°C for 30 s, 55°C for 45 s, and 72°C for 60 s and a final extension at 72°C for 7 min. PCR products were purified, quantified, and pooled at the same concentration (5 nM). The resulting amplicons were cleaned again using magnetic beads, quantified by fluorometry (Promega QuantiFluor), and normalized. The equimolar pool was cleaned a final time using magnetic beads to concentrate the pool and then measured using a D1000 high-sensitivity tape on an Agilent 2200 TapeStation. The pool was diluted to 5 nM, and molarity was confirmed again using a D1000 high- sensitivity tape. This was followed by 300-bp paired-end sequencing performed on an Illumina MiSeq system (San Diego, CA, USA). Raw sequences were processed using QIIME v1.9.2 (69). Low-quality regions (Q Ͻ 20) were trimmed from the 5= end of the sequences, and the paired ends were joined using FLASH (70). Primers were trimmed and a further round of quality control was conducted to discard full-length duplicate sequences, short (Ͻ100-nt) sequences, and sequences with ambiguous bases. Sequences were clustered followed by chimera checking using UCHIME algorithm from USEARCH v7.1.1090 (71). Operational taxonomic units (OTUs) were assigned using a UCLUST open-reference OTU-picking workflow with a threshold of 97% pairwise identity (71). Singletons or unique reads in the resultant data set were discarded; in addition, chloroplast-related and mitochondrion-related reads were removed from the OTU data set for 16S rRNA. Taxonomy was assigned to OTUs in QIIME using the Ribosomal Database Project (RDP) classifier (72) against the GreenGenes bacterial 16S rRNA database (v13.8) (73) for bacteria or the UNITE fungal ITS database (v7.2) (74) for fungi. To avoid/reduce biases generated by the use of various sequencing depths, sequence data were rarefied to the same depth per sample (the lowest sequencing depth of each batch, that is, for the soil, must and wine, and soil and plant samples) prior to downstream analysis. Data analysis. Microbial alpha-diversity was calculated using the Shannon index (H) in R (v3.5.0) with the “vegan” package (75). One-way analysis of variance (ANOVA) was used to determine whether sample classifications (e.g., region, fermentation stage) contained statistically significant differences in diversity. Principal-coordinate analysis (PCoA) was performed to evaluate the distribution patterns of wine metabolome and wine-related microbiome based on beta-diversity calculated by Bray-Curtis distance determinations performed with the “labdsv” package (76). Permutational multivariate analysis of variance (PERMANOVA) was conducted within each sample classification using distance matrices with 999 permutations to determine statistically significant differences by the use of the “adonis” function in “vegan” (75). Significant differences of wine microbiome between fermentation stages were tested based on taxonomic classification using linear discriminant analysis (LDA) effect size (LEfSe) analysis (77)(https:// huttenhower.sph.harvard.edu/galaxy/). The OTU table was filtered to include only OTUs with Ͼ0.01% relative abundance to reduce LEfSe complexity. This method applies the factorial Kruskal-Wallis rank sum test (␣ ϭ 0.05) to identify taxa with significant differential abundances between categories (using all-against-all comparisons), followed by the logarithmic LDA score (threshold ϭ 2.0) to estimate the effect size of each discriminative feature. Significant taxa were used to generate taxonomic cladograms illustrating differences between sample classes. A random forest supervised-classification model (37) was employed to identify the main predictors of wine regionality among the following variables: must and soil microbial diversity (Shannon index), soil properties, and weather. The importance of each predictor was determined by evaluating the decrease in prediction accuracy (that is, the increase in the mean square error [MSE] corresponding to comparisons

July/August 2020 Volume 5 Issue 4 e00534-20 50 msphere.asm.org 12 Wine Terroir Is Associated with Environmental Fungi

between observations and out-of-bag predictions) when the data were randomly permuted for the predictor. This analysis was conducted with 5,000 trees using the “randomForest” package in R (78). The significance of the model (P values) and the leave-one-out cross-validation R2 values were assessed using the “A3” package (ntree ϭ 5,000) (79). Structural equation modeling (SEM) (38) was used to evaluate the direct and indirect relationships among must and soil microbial diversity, soil properties, climate, and wine regionality. SEM is an a priori approach partitioning the influences of multiple drivers in a system to help characterize and comprehend complex networks of ecological interactions (80). An a priori model was established based on the known effects and relationships among these drivers of regional distribution patterns of wine aroma to manipulate the data before modeling. Weather and soil properties were used as composite variables (both random forest and SEM) to collapse the effects of multiple conceptually related variables into a single composite effect, thus aiding interpretation of model results (38). A path coefficient describes the strength and sign of the relationship between two variables (38). The good fit of the model was validated by the ␹2 test (P Ͼ 0.05), using the goodness-of-fit index (GFI Ͼ 0.90) and the root MSE of approximation (RMSEA Ͻ 0.05) (81). The standardized total effects of each factor on the wine regionality pattern were calculated by summing all direct and indirect pathways between two variables (38). All the SEM analyses were conducted using AMOS v25.0 (IBM, NY, USA). SourceTracker was used to track potential sources of wine-related fungi within the vineyards (39). SourceTracker represents a Bayesian approach that treats each give community (sink) as a mixture of communities deposited from a set of source environments and estimates the proportion of taxa in the sink community that come from possible source environments. When a sink contains a mixture of taxa that do not match any of the source environments, that portion of the community is assigned to an “unknown” source (39). In this model, we examined musts (n ϭ 2) and vineyard sources (n ϭ 50), including grapes, leaves, xylem sap, roots, and soils. The OTU tables were used as data input for modeling using the “SourceTracker” R package (https://github.com/danknights/sourcetracker). Data availability. Raw data are publicly available in the National Centre for Biotechnology Infor- mation Sequence Read Archive under BioProject accession numbers PRJNA594458 (bacterial 16S rRNA sequences) and PRJNA594469 (fungal ITS sequences).

SUPPLEMENTAL MATERIAL Supplemental material is available online only. TEXT S1, DOCX file, 0.02 MB. FIG S1, TIF file, 0.3 MB. FIG S2, TIF file, 1 MB. FIG S3, TIF file, 1.7 MB. FIG S4, TIF file, 2 MB. TABLE S1, DOCX file, 0.02 MB. TABLE S2, XLSX file, 0.02 MB. TABLE S3, XLSX file, 0.01 MB. TABLE S4, XLSX file, 0.01 MB. TABLE S5, XLSX file, 0.01 MB.

ACKNOWLEDGMENTS We give our sincere thanks to the vignerons who kindly allowed vineyard access, enabled sampling, and provided wine samples. D.L. acknowledges support from a Ph.D. scholarship and funding from Wine Australia (AGW Ph1602) and a Melbourne Research Scholarship from the University of Melbourne. We declare that the research was conducted in the absence of any commercial or ﬁnancial relationships that could be construed as a potential conﬂict of interest.

REFERENCES 1. Gonzálvez A, Llorens A, Cervera M, Armenta S, De la Guardia M. 2009. 5. Van Leeuwen C, Seguin G. 2006. The concept of terroir in viticulture. J Elemental fingerprint of wines from the protected designation of origin Wine Res 17:1–10. https://doi.org/10.1080/09571260600633135. Valencia. Food Chem 112:26–34. https://doi.org/10.1016/j.foodchem 6. Vaudour E, Costantini E, Jones G, Mocali S. 2015. An overview of the .2008.05.043. recent approaches to terroir functional modelling, footprinting and 2. Pereira GE, Gaudillere J-P, Van Leeuwen C, Hilbert G, Lavialle O, Mau- zoning. Soil 1:287–312. https://doi.org/10.5194/soil-1-287-2015. court M, Deborde C, Moing A, Rolin D. 2005. 1H NMR and chemometrics 7. Barata A, Malfeito-Ferreira M, Loureiro V. 2012. The microbial ecology of to characterize mature grape berries in four wine-growing areas in wine grape berries. Int J Food Microbiol 153:243–259. https://doi.org/10 Bordeaux, France. J Agric Food Chem 53:6382–6389. https://doi.org/10 .1016/j.ijfoodmicro.2011.11.025. .1021/jf058058q. 8. Stefanini I, Cavalieri D. 2018. Metagenomic approaches to investigate 3. Robinson AL, Adams DO, Boss PK, Heymann H, Solomon PS, Trengove the contribution of the vineyard environment to the quality of wine RD. 2012. Influence of geographic origin on the sensory characteristics fermentation: potentials and difficulties. Front Microbiol 9:991. https:// and wine composition of Vitis vinifera cv. Am J Enol Vitic 63:467–476. doi.org/10.3389/fmicb.2018.00991. https://doi.org/10.5344/ajev.2012.12023. 9. Bokulich NA, Ohta M, Richardson PM, Mills DA. 2013. Monitoring sea- 4. Gladstones J. 1992. Viticulture and environment. Winetitles, Broadview, sonal changes in winery-resident microbiota. PLoS One 8:e66437. Australia. https://doi.org/10.1371/journal.pone.0066437.

July/August 2020 Volume 5 Issue 4 e00534-20 51 msphere.asm.org 13 Liu et al.

10. Swiegers J, Bartowsky E, Henschke P, Pretorius I. 2005. Yeast and bac- terroir related to geographic distance between vineyards? Environ Mi- terial modulation of wine aroma and flavour. Aust J Grape Wine Res crobiol Rep 9:742–749. https://doi.org/10.1111/1758-2229.12589. 11:139–173. https://doi.org/10.1111/j.1755-0238.2005.tb00285.x. 32. Portillo MC, Franquès J, Araque I, Reguant C, Bordons A. 2016. Bacterial 11. Goddard MR. 2008. Quantifying the complexities of Saccharomyces diversity of Grenache and Carignan grape surface from different vine- cerevisiae’s ecosystem engineering via fermentation. Ecology 89: yards at Priorat wine region (Catalonia, Spain). Int J Food Microbiol 2077–2082. https://doi.org/10.1890/07-2060.1. 219:56–63. https://doi.org/10.1016/j.ijfoodmicro.2015.12.002. 12. Heard GM, Fleet GH. 1985. Growth of natural yeast flora during the 33. Compant S, Kaplan H, Sessitsch A, Nowak J, Ait Barka E, Clément C. 2008. fermentation of inoculated wines. Appl Environ Microbiol 50:727–728. Endophytic colonization of Vitis vinifera L. by Burkholderia phytofirmans https://doi.org/10.1128/AEM.50.3.727-728.1985. strain PsJN: from the rhizosphere to inflorescence tissues. FEMS Micro- 13. Bokulich NA, Thorngate JH, Richardson PM, Mills DA. 2014. Microbial biol Ecol 63:84–93. https://doi.org/10.1111/j.1574-6941.2007.00410.x. biogeography of wine grapes is conditioned by cultivar, vintage, and 34. Goddard MR, Anfang N, Tang R, Gardner RC, Jun C. 2010. A distinct climate. Proc Natl Acad SciUSA111:E139–E148. https://doi.org/10 population of Saccharomyces cerevisiae in New Zealand: evidence for .1073/pnas.1317377110. local dispersal by insects and human-aided global dispersal in oak 14. Gayevskiy V, Goddard MR. 2012. Geographic delineations of yeast com- barrels. Environ Microbiol 12:63–73. https://doi.org/10.1111/j.1462-2920 munities and populations associated with vines and wines in New .2009.02035.x. Zealand. ISME J 6:1281–1290. https://doi.org/10.1038/ismej.2011.195. 35. Lam SS, Howell KS. 2015. Drosophila-associated yeast species in vineyard 15. Morrison-Whittle P, Goddard MR. 2018. From vineyard to winery: a ecosystems. FEMS Microbiol Lett 362:fnv170. https://doi.org/10.1093/ source map of microbial diversity driving wine fermentation. Environ femsle/fnv170. Microbiol 20:75–84. https://doi.org/10.1111/1462-2920.13960. 36. Stefanini I, Dapporto L, Legras J-L, Calabretta A, Di Paola M, De Filippo 16. Pinto C, Pinho D, Cardoso R, Custódio V, Fernandes J, Sousa S, Pinheiro C, Viola R, Capretti P, Polsinelli M, Turillazzi S, Cavalieri D. 2012. Role of M, Egas C, Gomes AC. 2015. Wine fermentation microbiome: a landscape social wasps in Saccharomyces cerevisiae ecology and evolution. Proc from different Portuguese wine appellations. Front Microbiol 6:905. Natl Acad SciUSA109:13398–13403. https://doi.org/10.1073/pnas https://doi.org/10.3389/fmicb.2015.00905. .1208362109. 17. Portillo MC, Mas A. 2016. Analysis of microbial diversity and dynamics 37. Breiman L. 2001. Random forests. Machine Learn 45:5–32. https://doi during wine fermentation of Grenache grape variety by high- .org/10.1023/A:1010933404324. throughput barcoding sequencing. LWT Food Sci Technol 72:317–321. 38. Grace JB. 2006. Structural equation modeling and natural systems. https://doi.org/10.1016/j.lwt.2016.05.009. Cambridge University Press, Cambridge, United Kingdom. 18. Stefanini I, Albanese D, Cavazza A, Franciosi E, De Filippo C, Donati C, 39. Knights D, Kuczynski J, Charlson ES, Zaneveld J, Mozer MC, Collman RG, Cavalieri D. 2016. Dynamic changes in microbiota and mycobiota during Bushman FD, Knight R, Kelley ST. 2011. Bayesian community-wide spontaneous ‘Vino Santo Trentino’ fermentation. Microb Biotechnol culture-independent microbial source tracking. Nat Methods 8:761–763. 9:195–208. https://doi.org/10.1111/1751-7915.12337. https://doi.org/10.1038/nmeth.1650. 19. Taylor MW, Tsai P, Anfang N, Ross HA, Goddard MR. 2014. Pyrosequencing 40. Yadeta K, Thomma B. 2013. The xylem as battleground for plant hosts reveals regional differences in fruit-associated fungal communities. Environ and vascular wilt pathogens. Front Plant Sci 4:97. https://doi.org/10 Microbiol 16:2848–2858. https://doi.org/10.1111/1462-2920.12456. .3389/fpls.2013.00097. 20. Bokulich NA, Collins TS, Masarweh C, Allen G, Heymann H, Ebeler SE, 41. Mezzasalma V, Sandionigi A, Guzzetti L, Galimberti A, Grando MS, Tar- Mills DA. 2016. Associations among wine grape microbiome, metabo- daguila J, Labra M. 2018. Geographical and cultivar features differentiate lome, and fermentation behavior suggest microbial contribution to grape microbiota in Northern Italy and Spain vineyards. Front Microbiol regional wine characteristics. mBio 7:e00631-16. https://doi.org/10.1128/ 9:946. https://doi.org/10.3389/fmicb.2018.00946. mBio.00631-16. 42. Fierer N, Leff JW, Adams BJ, Nielsen UN, Bates ST, Lauber CL, Owens S, 21. Knight S, Klaere S, Fedrizzi B, Goddard MR. 2015. Regional microbial Gilbert JA, Wall DH, Caporaso JG. 2012. Cross-biome metagenomic signatures positively correlate with differential wine phenotypes: evi- analyses of soil microbial communities and their functional attributes. dence for a microbial aspect to terroir. Sci Rep 5:14233. https://doi.org/ Proc Natl Acad SciUSA109:21390–21395. https://doi.org/10.1073/pnas 10.1038/srep14233. .1215210110. 22. Müller DB, Vogel C, Bai Y, Vorholt JA. 2016. The plant microbiota: systems-level insights and perspectives. Annu Rev Genet 50:211–234. 43. Morrison-Whittle P, Goddard M. 2017. Fungal communities are differen- https://doi.org/10.1146/annurev-genet-120215-034952. tially affected by conventional and biodynamic agricultural manage- 23. Rillig MC, Lehmann A, Lehmann J, Camenzind T, Rauh C. 2018. Soil ment approaches in vineyard ecosystems. Agric Ecosyst Environ 246: biodiversity effects from field to fork. Trends Plant Sci 23:17–24. https:// 306–313. https://doi.org/10.1016/j.agee.2017.05.022. doi.org/10.1016/j.tplants.2017.10.003. 44. Öpik M, Vanatoa A, Vanatoa E, Moora M, Davison J, Kalwij J, Reier Ü, 24. Zarraonaindia I, Owens SM, Weisenhorn P, West K, Hampton-Marcell J, Zobel M. 2010. The online database MaarjAM reveals global and eco- Lax S, Bokulich NA, Mills DA, Martin G, Taghavi S, van der Lelie D, Gilbert systemic distribution patterns in arbuscular mycorrhizal fungi (Glomero- JA. 2015. The soil microbiome influences grapevine-associated microbi- mycota). New Phytol 188:223–241. https://doi.org/10.1111/j.1469-8137 ota. mBio 6:e02527-14. https://doi.org/10.1128/mBio.02527-14. .2010.03334.x. 25. Gilbert JA, van der Lelie D, Zarraonaindia I. 2014. Microbial terroir for 45. Fierer N, Jackson RB. 2006. The diversity and biogeography of soil wine grapes. Proc Natl Acad SciUSA111:5–6. https://doi.org/10.1073/ bacterial communities. Proc Natl Acad SciUSA103:626–631. https:// pnas.1320471110. doi.org/10.1073/pnas.0507535103. 26. Burns KN, Kluepfel DA, Strauss SL, Bokulich NA, Cantu D, Steenwerth KL. 46. Martiny JBH, Bohannan BJM, Brown JH, Colwell RK, Fuhrman JA, Green 2015. Vineyard soil bacterial diversity and composition revealed by 16S JL, Horner-Devine MC, Kane M, Krumins JA, Kuske CR, Morin PJ, Naeem rRNA genes: differentiation by geographic features. Soil Biol Biochem S, Ovreås L, Reysenbach A-L, Smith VH, Staley JT. 2006. Microbial 91:232–247. https://doi.org/10.1016/j.soilbio.2015.09.002. biogeography: putting microorganisms on the map. Nat Rev Microbiol 27. Coller E, Cestaro A, Zanzotti R, Bertoldi D, Pindo M, Larger S, Albanese D, 4:102–112. https://doi.org/10.1038/nrmicro1341. Mescalchin E, Donati C. 2019. Microbiome of vineyard soils is shaped by 47. Tedersoo L, Bahram M, Põlme S, Kõljalg U, Yorou NS, Wijesundera R, geography and management. Microbiome 7:140. https://doi.org/10 Villarreal Ruiz L, Vasco-Palacios AM, Thu PQ, Suija A, Smith ME, Sharp C, .1186/s40168-019-0758-7. Saluveer E, Saitta A, Rosas M, Riit T, Ratkowsky D, Pritsch K, Põldmaa K, 28. Knight SJ, Karon O, Goddard MR. 2020. Small scale fungal community Piepenbring M, Phosri C, Peterson M, Parts K, Pärtel K, Otsing E, Nouhra differentiation in a vineyard system. Food Microbiol 87:103358. https:// E, Njouonkou AL, Nilsson RH, Morgado LN, Mayor J, May TW, Majuakim doi.org/10.1016/j.fm.2019.103358. L, Lodge DJ, Lee SS, Larsson K-H, Kohout P, Hosaka K, Hiiesalu I, Henkel 29. Liu D, Zhang P, Chen D, Howell KS. 2019. From the vineyard to the TW, Harend H, Guo L-d, Greslebin A, Grelet G, Geml J, Gates G, Dunstan winery: how microbial ecology drives regional distinctiveness of wine. W, Dunk C, Drenkhan R, Dearnaley J, et al. 2014. Global diversity and Front Microbiol 10:2679. https://doi.org/10.3389/fmicb.2019.02679. geography of soil fungi. Science 346:1256688. https://doi.org/10.1126/ 30. Jara C, Laurie VF, Mas A, Romero J. 2016. Microbial terroir in Chilean science.1256688. valleys: diversity of non-conventional yeast. Front Microbiol 7:663. 48. Burns KN, Bokulich NA, Cantu D, Greenhut RF, Kluepfel DA, O’Geen AT, https://doi.org/10.3389/fmicb.2016.00663. Strauss SL, Steenwerth KL. 2016. Vineyard soil bacterial diversity and 31. Miura T, Sánchez R, Castañeda LE, Godoy K, Barbosa O. 2017. Is microbial composition revealed by 16S rRNA genes: differentiation by vineyard

July/August 2020 Volume 5 Issue 4 e00534-20 52 msphere.asm.org 14 Wine Terroir Is Associated with Environmental Fungi

management. Soil Biol Biochem 103:337–348. https://doi.org/10.1016/j done concentration in berries of Vitis vinifera L. cv. Shiraz. J Agric Food .soilbio.2016.09.007. Chem 63:4276–4283. https://doi.org/10.1021/acs.jafc.5b00590. 49. Berg G, Smalla K. 2009. Plant species and soil type cooperatively shape 66. Liu D, Gao Y, Li X-X, Li Z, Pan Q-H. 2015. Attenuated UV radiation alters the structure and function of microbial communities in the rhizosphere. volatile profile in Cabernet Sauvignon grapes under field conditions. Mol- FEMS Microbiol Ecol 68:1–13. https://doi.org/10.1111/j.1574-6941.2009 ecules 20:16946–16969. https://doi.org/10.3390/molecules200916946. .00654.x. 67. Yu Y, Lee C, Kim J, Hwang S. 2005. Group-specific primer and probe sets 50. Lauber CL, Strickland MS, Bradford MA, Fierer N. 2008. The influence of to detect methanogenic communities using quantitative real-time poly- soil properties on the structure of bacterial and fungal communities merase chain reaction. Biotechnol Bioeng 89:670–679. https://doi.org/ across land-use types. Soil Biol Biochem 40:2407–2415. https://doi.org/ 10.1002/bit.20347. 10.1016/j.soilbio.2008.05.021. 68. Gardes M, Bruns TD. 1993. ITS primers with enhanced specificity for 51. Bokulich NA, Joseph CL, Allen G, Benson AK, Mills DA. 2012. Next- basidiomycetes-application to the identification of mycorrhizae and rusts. generation sequencing reveals significant bacterial diversity of botry- Mol Ecol 2:113–118. https://doi.org/10.1111/j.1365-294x.1993.tb00005.x. tized wine. PLoS One 7:e36357. https://doi.org/10.1371/journal.pone 69. Caporaso JG, Kuczynski J, Stombaugh J, Bittinger K, Bushman FD, .0036357. Costello EK, Fierer N, Peña AG, Goodrich JK, Gordon JI, Huttley GA, Kelley 52. Liu Y, Rousseaux S, Tourdot-Maréchal R, Sadoudi M, Gougeon R, Schmitt- ST, Knights D, Koenig JE, Ley RE, Lozupone CA, McDonald D, Muegge BD, Kopplin P, Alexandre H. 2017. Wine microbiome: a dynamic world of Pirrung M, Reeder J, Sevinsky JR, Turnbaugh PJ, Walters WA, Widmann J, microbial interactions. Crit Rev Food Sci Nutr 57:856–873. https://doi Yatsunenko T, Zaneveld J, Knight R. 2010. QIIME allows analysis of .org/10.1080/10408398.2014.983591. high-throughput community sequencing data. Nat Methods 7:335–336. 53. Deveau A, Bonito G, Uehling J, Paoletti M, Becker M, Bindschedler S, https://doi.org/10.1038/nmeth.f.303. Hacquard S, Hervé V, Labbé J, Lastovetsky OA, Mieszkin S, Millet LJ, Vajna 70. Magocˇ T, Salzberg SL. 2011. FLASH: fast length adjustment of short reads B, Junier P, Bonfante P, Krom BP, Olsson S, van Elsas JD, Wick LY. 2018. to improve genome assemblies. Bioinformatics 27:2957–2963. https:// Bacterial-fungal interactions: ecology, mechanisms and challenges. doi.org/10.1093/bioinformatics/btr507. FEMS Microbiol Rev 42:335–352. https://doi.org/10.1093/femsre/fuy008. 71. Edgar RC, Haas BJ, Clemente JC, Quince C, Knight R. 2011. UCHIME 54. Panke-Buisse K, Poole AC, Goodrich JK, Ley RE, Kao-Kniffin J. 2015. improves sensitivity and speed of chimera detection. Bioinformatics Selection on soil microbiomes reveals reproducible impacts on plant 27:2194–2200. https://doi.org/10.1093/bioinformatics/btr381. function. ISME J 9:980–989. https://doi.org/10.1038/ismej.2014.196. 72. Wang Q, Garrity GM, Tiedje JM, Cole JR. 2007. Naive Bayesian classifier for rapid assignment of rRNA sequences into the new bacterial taxon- 55. Lau JA, Lennon JT. 2012. Rapid responses of soil microorganisms im- omy. Appl Environ Microbiol 73:5261–5267. https://doi.org/10.1128/AEM prove plant fitness in novel environments. Proc Natl Acad SciUSA .00062-07. 109:14058–14062. https://doi.org/10.1073/pnas.1202319109. 73. DeSantis TZ, Hugenholtz P, Larsen N, Rojas M, Brodie EL, Keller K, Huber 56. Grady KL, Sorensen JW, Stopnisek N, Guittar J, Shade A. 2019. Assembly T, Dalevi D, Hu P, Andersen GL. 2006. Greengenes, a chimera-checked and seasonality of core phyllosphere microbiota on perennial biofuel 16S rRNA gene database and workbench compatible with ARB. Appl crops. bioRxiv https://doi.org/10.1101/446369. Environ Microbiol 72:5069–5072. https://doi.org/10.1128/AEM.03006-05. 57. Compant S, Reiter B, Sessitsch A, Nowak J, Clément C, Barka EA. 2005. 74. Kõljalg U, Larsson K-H, Abarenkov K, Nilsson RH, Alexander IJ, Eberhardt Endophytic colonization of Vitis vinifera L. by plant growth-promoting U, Erland S, Høiland K, Kjøller R, Larsson E, Pennanen T, Sen R, Taylor AFS, bacterium Burkholderia sp. strain PsJN. Appl Environ Microbiol 71: Tedersoo L, Vrålstad T, Ursing BM. 2005. UNITE: a database providing 1685–1693. https://doi.org/10.1128/AEM.71.4.1685-1693.2005. Web-based methods for the molecular identification of ectomycorrhizal 58. Börlin M, Venet P, Claisse O, Salin F, Legras J-L, Masneuf-Pomarede I. fungi. New Phytol 166:1063–1068. https://doi.org/10.1111/j.1469-8137 2016. Cellar-associated Saccharomyces cerevisiae population structure .2005.01376.x. revealed high diversity and perennial persistence in Sauternes wine 75. Oksanen J, Kindt R, Legendre P, O’Hara B, Henry M, Stevens H. 2007. The estates. Appl Environ Microbiol 82:2909–2915. https://doi.org/10.1128/ vegan package. Community ecology package. http://cran.r-project.org/, AEM.03627-15. http://r-forge.r-project.org/projects/vegan/. 59. Valero E, Schuller D, Cambon B, Casal M, Dequin S. 2005. Dissemination 76. Roberts DW. 2007. labdsv: ordination and multivariate analysis for ecology. and survival of commercial wine yeast in the vineyard: a large-scale, R package version 1. https://cran.r-project.org/web/packages/labdsv/labdsv three-years study. FEMS Yeast Res 5:959–969. https://doi.org/10.1016/j .pdf. .femsyr.2005.04.007. 77. Segata N, Izard J, Waldron L, Gevers D, Miropolsky L, Garrett WS, 60. Mandl K, Schieck J, Silhavy-Richter K, Prange A, Schneider V, Schmidt Huttenhower C. 2011. Metagenomic biomarker discovery and explana- H-P. 2015. Vines take up yeasts from soil and transport them through the tion. Genome Biol 12:R60. https://doi.org/10.1186/gb-2011-12-6-r60. vine to the stem and skins of grapes. Ithaka J 2015:349–355. 78. Liaw A, Wiener M. 2002. Classification and regression by randomForest. R 61. Jones DA, Wang W, Fawcett R. 2009. High-quality spatial climate data- News 2:18–22. https://cogns.northwestern.edu/cbmg/LiawAndWiener2002 sets for Australia. Amoj 58:233–248. https://doi.org/10.22499/2.5804.003. .pdf. 62. López-Millán AF, Morales F, Abadía A, Abadía J. 2000. Effects of iron 79. Fortmann-Roe S. 2013. Accurate, adaptable, and accessible error metrics deficiency on the composition of the leaf apoplastic fluid and xylem sap for predictive models. R package version 09 2. https://CRAN.R-project in sugar beet. Implications for iron and carbon transport. Plant Physiol .org/packageϭA3. 124:873–884. https://doi.org/10.1104/pp.124.2.873. 80. Eisenhauer N, Bowker MA, Grace JB, Powell JR. 2015. From patterns to 63. Rayment GE, Lyons DJ. 2011. Soil chemical methods: Australasia, vol 3. causal understanding: structural equation modeling (SEM) in soil ecology. CSIRO Publishing, Clayton, Australia. Pedobiologia 58:65–72. https://doi.org/10.1016/j.pedobi.2015.03.002. 64. Liu D, Xing R-R, Li Z, Yang D-M, Pan Q-H. 2016. Evolution of volatile 81. Schermelleh-Engel K, Moosbrugger H, Müller H. 2003. Evaluating the fit compounds, aroma attributes, and sensory perception in bottle-aged of structural equation models: tests of significance and descriptive red wines and their correlation. Eur Food Res Technol 242:1937–1912. goodness-of-fit measures. Methods Psychol Res Online 8:23–74. https://doi.org/10.1007/s00217-016-2693-1. 82. Liu D, Howell K. 20 July 2020. Community succession of the grapevine 65. Zhang P, Barlow S, Krstic M, Herderich M, Fuentes S, Howell K. 2015. fungal microbiome in the annual growth cycle. Environ Microbiol Within-vineyard, within-vine, and within-bunch variability of the rotun- https://doi.org/10.1111/1462-2920.15172.

July/August 2020 Volume 5 Issue 4 e00534-20 53 msphere.asm.org 15 3.2 Supplementary data

3.2.1 Supplementary materials and methods

Xylem sap collection

Vine shoots (n = 10) were harvested from the vineyard (Table S1) under aseptic conditions and then centrifuged (1). The shoots were peeled to remove the outer bark and phloem layer, sterilised with

70% ethanol and cut to fit into 10-ml sterile centrifuge tubes that had 10 sterile glass beads at the bottom. The shoots were centrifuged at 15,000 × g for 10 min at 4°C and the xylem sap collected giving ~ 2.0 mL per sample. Additional xylem sap (n = 5) were sampled from Vitis vinifera Shiraz grapevines grown in a glasshouse at the University of Melbourne. Xylem fluid was extracted with a pressure cylinder apparatus (similar to a Scholander pressure chamber) (2). Grapevines were uprooted and the soil was carefully removed from the roots. On an aseptic bench, the main trunk was cut 5-10 cm above the roots with a sterile blade, girdled to remove phloem tissue, sterilised by immersion in

70% ethanol, and immediately inserted into a pressure cylinder. The cylinder applied 60-70 kPa pressure for two hours to extract the xylem sap (~ 4.0 mL). Xylem sap was divided into two subsamples, one used immediately to isolate living yeasts and the other flash frozen with liquid nitrogen and stored at -80°C for DNA extraction, next-generation sequencing and chemical analysis.

Chemical analysis of xylem sap composition

Carbohydrates of xylem sap were determined using enzymatic methods by Megazyme assay kits

(Megazyme, Ireland) following the manufacturer’s protocol. Amino acids (free and protein- bound) were determined using pre-column derivatisation with 6-aminoquinolyl-N-hydroxysuccinimidyl carbamate followed by separation and quantification with the ACQUITY Ultra Performance LC

(UPLC; Waters, MA, USA) system at the Australian Proteome Analysis Facility. The column was an

ACQUITY UPLC BEH C18 column (1.7 μm × 2.1 mm × 5 mm) with detection at 260 nm (UV) and a flow rate of 0.7 mL/min at 57–60°C. Identification and quantitation of the amino acids was performed against a set of prepared standards, with DL-norvaline as the internal standard (3, 4). Organic acids were determined using a Waters High-performance liquid chromatography (HPLC) (Waters, MA,

54 USA) based on Andersen et al (1989) (5) with modification. Here, 20 μL xylem sap was injected through a Synergi™ Hydro-RP LC Column (250 mm × 4.6 mm × 4 μm; Phenomenex Inc, CA, USA) at 60°C, with detection at 210 nm (UV). Mobile phases at a flow rate of 1.0 mL/min, with 20 mM potassium phosphate buffer (A, pH = 1.5) and 100% methanol (B) following a gradient programme:

(0-2.5) min, 100% A; (2.5-2.9) min, linear ramp to 30% B; (2.9-8.0) min, 30% B; (8.0-8.5) min, linear ramp to 100% A; (8.5-10) min, 100% A. Identification and quantitation of the compounds was performed using a set of prepared standards.

Isolation and identification of yeasts from xylem sap

Yeasts were isolated and identified from xylem sap to explore the potential translocation mechanism of yeasts in the vineyard. Xylem sap was serially diluted and plated (0.1 mL) onto solid yeast extract peptone dextrose (YPD) medium that was supplemented with 34 mg/mL chloramphenicol and 25 mg/mL ampicillin to inhibit bacterial growth. Plates were incubated at 28°C for 2-3 days in aerobic conditions. Single colonies with different morphological types were streaked onto Wallerstein

Nutrient (WLN) agar media to obtain pure cultures. DNA was recovered from pure colonies using the

MasterPure™ Yeast DNA Purification Kit (Epicentre, Madison, WI) following the manufacturer’s instructions. The 26S rDNA D1/D2 domain was amplified using primers NL1/4 (6) for sequencing by

Australian Genome Research Facility (AGRF). Sequences were trimmed, aligned and analysed using

BLAST at NCBI (http://blast.ncbi.nlm.nih.gov/Blast.cgi). Sequence data was uploaded to Genbank with accession numbers MN847682 - MN847692.

55 References

1. Jones DA, Wang W, Fawcett R. 2009. High-quality spatial climate data-sets for Australia. Australian Meteorological and Oceanographic Journal 58:233.

2. Schurr U. 1998. Xylem sap sampling—new approaches to an old topic. Trends in Plant Science 3:293-298.

3. Cohen SA. 2001. Amino acid analysis using precolumn derivatization with 6-aminoquinolyl-N- hydroxysuccinimidyl carbamate, p 39-47, Amino Acid Analysis Protocols. Springer.

4. Cohen SA, Michaud DP. 1993. Synthesis of a fluorescent derivatizing reagent, 6-aminoquinolyl-N- hydroxysuccinimidyl carbamate, and its application for the analysis of hydrolysate amino acids via high-performance liquid chromatography. Analytical biochemistry 211:279-287.

5. Andersen PC, Brodbeck BV, Mizell RF. 1989. Metabolism of amino acids, organic acids and sugars extracted from the xylem fluid of four host plants by adult Homalodisca coagulata. Entomologia Experimentalis et Applicata 50:149-159.

6. Kurtzman CP, Robnett CJ. 1998. Identification and phylogeny of ascomycetous yeasts from analysis of nuclear large subunit (26S) ribosomal DNA partial sequences. Antonie van Leeuwenhoek 73:331-371.

56 3.2.2 Supplementary tables and figures

Table S1 Vineyard conditions and number of soil, must and ferments, and plant samples collected in this study

(A) Soil, must and ferment samples Vintage Region Vineyard Altitude Orientation Soil Soil texture Cover Chemical Sample Fraction (V) subgroup crop constituents of harvested grapes °Brix pH Soil Must1 AF1 AF- AF- MLF- Mid1 End1 End1,2 2017 Geelong V1 82 N Sodosols Clay loam Grass 19.25 3.68 3 1 1 1 1 1 V2 92 N Sodosols Clay loam Grass 19.82 3.64 3 1 1 1 1 1 Mornington V3 101 N Sodosols Light medium Grass 19.67 3.62 3 1 1 1 1 1 clay V4 135 N Sodosols Heavy clay Grass 19.65 3.57 3 1 1 1 1 1 V5 223 N Ferrosols Clay loam Grass 19.26 3.72 3 1 1 1 1 1 V6 195 N Ferrosols Heavy clay Grass 19.15 3.56 3 1 1 1 1 1 V7 55 N Sodosols Heavy clay Grass 18.53 3.65 3 N/A N/A N/A N/A N/A Macedon V8 614 NE Chromosols Light medium Grass 18.30 3.48 3 1 1 1 1 1 Ranges clay V9 515 N Ferrosols Light medium Grass 17.90 3.46 3 1 1 1 1 1 clay Yarra V10 119 N Chromosols Heavy clay Grass 19.52 3.68 3 1 1 1 1 1 Valley V11 174 N Chromosols Heavy clay Grass 19.86 3.64 3 1 1 1 1 1 Grampians V12 344 N Sodosols Clay loam Grass 19.75 3.58 3 1 1 1 1 1 V13 356 N Sodosols Clay loam Grass 19.52 3.62 3 N/A N/A N/A N/A N/A Gippsland V14 38 NE Dermosols Clay loam Grass 18.50 3.68 3 1 1 1 1 1 V15 46 N Dermosols Clay loam Grass 18.81 3.62 3 1 1 1 1 1 2018 Mornington V3 101 N Sodosols Light medium Grass 20.32 3.43 3 1 1 1 1 1 clay V4 135 N Sodosols Heavy clay Grass 19.85 3.61 3 1 1 1 1 1

57 V5 223 N Ferrosols Clay loam Grass 20.61 3.48 3 1 1 1 1 1 V6 195 N Ferrosols Heavy clay Grass 19.70 3.65 3 1 1 1 1 1 V7 55 N Sodosols Heavy clay Grass 19.54 3.52 3 1 1 1 1 1 1 For microbial analysis: triplicates from tanks/ barrels were collected and mixed as one composite sample which was analysed by NGS 2 For wine aroma analysis: triplicates were analysed by GC-MS N/A: not applied, spontaneous fermentations did not complete and were excluded from analysis

(B) Soil and plant samples Vineyard Sample Fraction Vintage Region Vine (v) (V) Soil Root Xylem sap Leaf Grape 2018 Mornington V5 v1 1 1 1 1 1 v2 1 1 1 1 1 v3 1 1 1 1 1 v4 1 1 1 1 1 v5 1 1 1 1 1

Table S2 ANOVA results of wine volatile compounds among wine growing regions in 2017

Retenion time Volatile Compound CAS No* F p Pr (>F) (min) (5, 33) C1 3.123 Ethyl Acetate 141-78-6 5.506 0.001 ** C2 4.256 Ethyl propanoate 105-37-3 0.102 0.986 C3 5.326 Isobutyl acetate 110-19-0 5.664 8.58E-04 *** C4 5.938 Ethyl butanoate 105-54-4 6.896 2.16E-04 *** C5 7.265 Hexanal 66-25-1 0.640 0.618 C6 8.245 2-Methyl-1-propanol 78-83-1 44.540 5.42E-13 *** C7 8.734 3-Methylbutyl acetate 123-92-2 1.657 0.176 C8 9.253 Ethyl pentanoate 539-82-2 0.345 0.765 C9 9.865 1-Butanol 71-36-3 52.260 6.58E-14 *** C10 11.300 Methyl hexanoate 106-70-7 26.870 3.09E-10 *** C11 11.500 Isoamyl propanoate 105-68-0 2.015 0.078 C12 12.012 D-Limonene 5989-27-5 14.420 0.00000032 *** C13 12.457 (E)-2-Hexenal 6728-26-3 14.870 2.33E-07 *** C14 12.689 3-Methyl-1-butanol 123-51-3 1.046 0.274 C15 12.894 4-Octanone 589-63-9 2.145 0.075 C16 13.345 Ethyl hexanoate 123-66-0 10.950 0.00000463 *** C17 14.105 Styrene 100-42-5 0.370 0.865 C18 14.523 p-Cymene 99-87-6 22.128 8.91E-10 *** C19 14.923 Hexyl acetate 142-92-7 2.711 0.039 * C20 15.245 Acetoin 513-86-0 4.806 0.002 ** C21 17.312 2-Heptanol 543-49-7 8.030 0.0000669 *** C22 17.523 Ethyl heptanoate 106-30-9 4.648 0.003 ** C23 17.646 6-Methyl-5-heptene-2-one 110-93-0 1.784 0.221 C24 18.123 Ethyl lactate 97-64-3 1.085 0.389 C25 18.726 1-Hexanol 111-27-3 14.430 3.18E-07 *** C26 19.123 3-Hexanol 623-37-0 0.444 0.868 C27 19.356 Heptyl acetate 112-06-1 2.332 0.067 C28 20.463 3-Octanol 589-98-0 8.009 0.0000683 *** C29 20.912 2-Hexanol 626-93-7 0.925 0.566 C30 21.312 (E)-2-Octenal 2548-87-0 0.460 0.747 C31 21.925 Ethyl octoate 106-32-1 1.918 0.121 C32 22.702 Acetic acid 64-19-7 7.994 0.0000693 *** C33 22.813 1-Octen-3-ol 3391-86-4 16.700 6.89E-08 *** C34 22.900 Isopentyl hexanoate 2198-61-0 3.164 0.021 * C35 23.103 1-Heptanol 111-70-6 24.680 8.49E-10 *** C36 23.383 6-Methyl-5-hepten-2-ol 1569-60-4 5.329 0.001 ** C37 23.713 Octyl acetate 112-14-1 0.152 0.856 C38 24.125 Ethyl 6-heptenoate 25118-23-4 2.216 0.079 C39 24.500 2-Ethyl-1-hexanol 104-76-7 10.020 0.0000103 *** C40 25.103 Benzaldehyde 100-52-7 5.842 7.00E-04 *** C41 25.725 2-Nonanol 628-99-9 14.080 4.06E-07 ***

59 C42 26.114 Ethyl nonanoate 123-29-5 1.223 0.322 C43 26.806 Linalool 78-70-6 16.480 7.95E-08 *** C44 27.212 1-Octanol 111-87-5 7.832 0.0000816 *** C45 27.513 2-Methyl-propanoic acid 79-31-2 10.880 0.00000491 *** C46 28.012 2,3-Butanediol 513-85-9 20.140 8.87E-09 *** C47 28.426 Methyl decanoate 110-42-9 0.915 0.485 C48 28.634 Terpinen-4-ol 562-74-3 1.287 0.296 C49 29.123 4-Hydroxybutanoic acid 591-81-1 6.407 3.69E-04 *** C50 29.405 Ethyl 2-furylcarboxylate 614-99-3 1.254 0.258 C51 29.716 Benzeneacetaldehyde 122-78-1 1.887 0.126 C52 30.023 Butyric acid 107-92-6 0.105 0.891 C53 30.321 Ethyl decanoate 110-38-3 0.644 0.668 C54 30.926 3-Methylbutyl octanoate 2035-99-6 0.957 0.459 C55 31.216 1-Nonanol 143-08-8 9.500 0.0000165 *** C56 31.605 Diethyl succinate 123-25-1 3.197 0.020 * C57 32.213 Ethyl 9-decenoate 67233-91-4 0.397 0.847 C58 32.405 α-Terpineol 98-55-5 3.861 0.008 ** C59 33.012 Methionol 505-10-2 15.600 1.42E-07 *** C60 33.426 Benzyl acetate 140-11-4 2.225 0.071 C61 34.738 Methyl salicylate 119-36-8 6.335 4.00E-04 *** Naphthalene, 1,2-dihydro-1,1,6- C62 33.616 trimethyl 30364-38-6 0.145 0.899 C63 35.012 1-Decanol 112-30-1 4.640 0.003 ** C64 35.209 Citronellol 106-22-9 1.673 0.172 C65 35.529 Ethyl benzeneacetate 101-97-3 2.097 0.093 C66 36.123 Ethyl salicylate 118-61-6 5.335 0.001 ** C67 36.312 Nerol 106-25-2 4.026 0.007 ** C68 36.404 Methyl dodecanoate 111-82-0 0.011 0.956 C69 36.503 Phenethyl acetate 103-45-7 21.302 2.91E-10 *** C70 36.613 β-Damascenone 23726-93-4 26.070 4.44E-10 *** C71 37.723 Ethyl dodecanoate 106-33-2 0.121 0.987 C72 38.002 Hexanoic acid 142-62-1 3.879 0.008 ** C73 38.303 Guaiacol 90-05-1 2.391 0.061 C74 38.716 Benzyl alcohol 100-51-6 7.588 1.00E-07 *** C75 38.913 Ethyl dihydrocinnamate 2021-28-5 846.000 <2e-16 *** C76 39.812 Ethyl 3-methylbutyl succinate 28024-16-0 8.985 0.0000265 *** C77 40.023 Phenylethyl Alcohol 60-12-8 10.080 0.00000977 *** C78 40.712 β-Ionone 127-41-3 1.053 0.451 C79 44.625 Ethyl tetradecanoate 124-06-1 0.139 0.982 C80 45.217 Octanoic acid 124-07-2 3.437 0.014 * C81 46.813 Ethyl cinnamate 103-36-6 35.660 9.44E-12 *** C82 48.125 Eugenol 97-53-0 0.285 0.847 C83 48.526 4-Ethyl-phenol 123-07-9 0.156 0.457 C84 48.608 Nonanoic acid 112-05-0 1.728 0.079 C85 51.112 Ethyl hexadecanoate 628-97-7 1.061 0.401 C86 51.710 Methyl dihydrojasmonate 24851-98-7 0.945 0.243

60 C87 51.801 n-Decanoic acid 334-48-5 3.785 0.009 ** C88 52.866 2,4-bis(1,1'-dimethylethyl)phenol 96-76-4 4.387 0.004 ** *CAS Registry Number

Table S3 Permutational multivariate analysis of variance (PERMANOVA) using distance matrices of category effects on microbial and wine aroma diversity

Soil Must Wine (MLF-End) Wine Aroma

Sample Bacteria Fungi Bacteria Fungi Bacteria Fungi MLF-End Group Vintage Factor R2 p R2 p R2 p R2 p R2 p R2 p R2 p All Both Region 0.318 0.001 0.254 0.001 0.108 0.152 0.292 0.001 N/A N/A N/A N/A 0.566 0.001 All 2017 Region 0.392 0.001 0.419 0.001 0.342 0.001 0.565 0.001 0.149 0.321 0.109 0.205 0.703 0.001 All Both Vintage 0.102 0.001 0.086 0.001 0.186 0.001 0.049 0.066 N/A N/A N/A N/A 0.267 0.001 Mornington Both Vintage 0.155 0.001 0.151 0.001 0.305 0.001 0.143 0.001 N/A N/A N/A N/A 0.355 0.001 Beta-diversity calculated by the Bray–Curtis distance; Microbial diversity was tested at OTU level.

Table S4 α-diversity (Shannon index) of soil and must microbial communities from six wine growing regions in 2017 and 2018

Soil Must Vintage Region Bacteria Fungi Bacteria Fungi 2017 Geelong 5.486 ± 0.058 3.869 ± 0.181 1.838 ± 0.145 0.938 ± 0.125 Mornington 6.178 ± 0.074 4.472 ± 0.139 2.216 ± 0.242 1.840 ± 0.217 Macedon Ranges 4.317 ± 0.047 2.654 ± 0.148 2.424 ± 0.425 1.582 ± 0.246 Yarra Valley 6.201 ± 0.062 3.355 ± 0.238 1.756 ± 0.314 2.560 ± 0.123 Grampians 6.238 ± 0.120 3.615 ± 0.191 1.966 ± 0.125 1.297 ± 0.086 Gippsland 6.416 ± 0.036 4.080 ± 0.227 1.855 ± 0.245 2.200 ± 0.361 2018 Mornington 6.559 ± 0.029 4.187 ± 0.104 1.995 ± 0.357 1.694 ± 0.128

Table S5 ANOVA results of soil properties and weather conditions among wine growing regions in

2017

F (5,39) p Pr (>F)

Soil pH 2.300 0.067 C 5.891 5.37E-04 *** N 3.981 0.006 ** C: N 6.309 3.27E-04 *** Nitrate 11.640 1.56E-06 *** Ammonium 1.780 0.144 Weather Mean solar radiation 631.700 <2e-16 *** Mean high temperature 22.100 1.14E-09 *** Mean low temperature 762.600 <2e-16 *** Maximum temperature 45.580 7.09E-14 *** Minimum temperature 3823.000 <2e-16 *** Mean temperature 18.290 1.12E-08 *** Precipitation 172.300 <2e-16 *** Mean relative soil moisture 0.901 0.492 Mean evaporation 0.729 0.607 Mean transpiration 188.100 <2e-16 ***

FIG S1 Map of 15 sampling vineyards from six Pinot Noir wine-producing regions in southern Australia, spanning 400 km (E-W).

FIG S2 Vineyard soil microbial community compositions across all samples from six grape growing regions. Shown are average percentages of taxa (characterised to the phylum level) across sites in each region: (A) dominant bacterial phyla with greater than 1.0% relative abundance; (B) fungal phyla.

FIG S3 Stage of fermentation influences microbial diversities. Bacterial (A) and fungal (B) α-diversity (Shannon index) changes during wine fermentation. Bray-Curtis distance PCoA of bacterial communities (C) and fungal communities (D) according to the fermentation stage.

FIG S4 Xylem sap is a potential translocation medium of yeasts from the soil to the grapevine. (A) Fungal taxa in the genus level of xylem sap from the vineyard. (B) Relative abundances of S. cerevisiae in the soil, root, xylem sap, grape, must. Cultured yeast species from the xylem sap from vineyard (C) and glasshouse (D). Nutrient compositions of xylem sap: carbohydrates (E), total amino acids (protein and free) (F), and organic acids (G).

68 3.3 Addendum

This part data based on co-occurrence network was omitted from the published paper.

Microbial patterns correlate to wine aroma profiling

The co-occurrence/interaction patterns between wine metabolome and microbiota in the soil and must were explored in network analysis using Gephi (v0.9.2) (Bastian et al., 2009). Only top 100 OTUs

(both bacteria and fungi) based on the relative abundance were used to construct the network. A correlation matrix was calculated with all possible pair-wise Spearman's rank correlations between volatile compounds and selected OTUs. Correlations with a Spearman correlation coefficient ρ ≥ 0.8 and a p < 0.01 (false discovery rate [FDR]-corrected p value) were considered statistically robust and displayed in the networks (Junker and Schreiber, 2008).

Network analysis was used to explore connections between regional microbial and wine metabolic patterns. 57 volatile compounds and 246 OTUs were screened out based on strong correlation coefficients (Spearman correlation coefficient, ρ ≥ 0.8; p < 0.01), forming the co-occurrence patterns.

These variables showed a sophisticated internal structure, consisting of 303 nodes and 610 edges

(average degree = 4.026), with mostly positive correlations (90.66%) (Figure 1A). Both soil (average degree = 3.333; Figure 1B) and must microbiota (average degree = 3.791; Figure 4C) presented high connectivity with wine metabolites (Figure 1B, C). Soil fungal OTUs (69 of 5610 OTUs) were more densely connected with volatile compounds than bacteria (42 of 12117 OTUs) (Figure 1). This indicated that soil microbial communities, especially fungi, might play a role in wine properties. The most densely connected volatile compounds were assigned to groups of monoterpenes (C18, p- cymene; C58, α-terpineol), phenylpropanoids (C40, benzaldehyde; C69, phenethyl acetate), and acetoin (C20) (Figure 1). Correspondingly, high-connectivity nodes with these compounds were

OTUs of bacterial taxa Enterobacteriales, Rhizobiales, Burkholderiales and Xanthomonadales, and fungal taxa Penicillium, Rhodotorula and Neocatenulostroma from soil and must (Figure 1B, C). The

Figure 1 Statistically significant and strong co-occurrence relationships between wine volatile compounds and bacterial and fungal OTUs in vineyard soils and musts across six wine growing regions. Network plots for wine aroma with soil and must microbiota (A), and simplified network plots for wine aroma with soil microbiota (B) and with must microbiota (C), separately. Circle nodes represent assigned volatile compounds, bacterial and fungal OTUs, with different colours. Direct connections between compounds and OTUs indicate strong correlations (Spearman correlation coefficient, ρ ≥ 0.8; p < 0.01). The colour of edges describes the positive correlation (pink) or the negative correlation (green). The size of nodes is proportioned to the interconnected degree

70 group of esters, which are yeast-driven products (Bokulich et al., 2013), were highly associated with fungal OTUs (in particular Hypocreales, Alternaria, Cryptococcus and Mortierella) from the vineyard soil, for example, ethyl hexanoate (C16) and ethyl dihydrocinnamate (C75) (Figure 1B). Fatty acids

(including C49, 4-hydroxybutanoic acid; C80, octanoic acid) were mostly negatively connected with fungi (including Acremonium, Penicillium) from the must (Figure 1C).

Co-occurrence networks demonstrate that the soil and must microbiota correlate closely to the wine metabolome, highlighting potential contributions from fungal taxa. These interactions also indicate potential modulations of soil and must microbiota on wine chemistry, in particular those taxa that are numerically dominant early in the fermentation. For example, bacteria Enterobacteriales positively correlated with some grape-derived volatile compounds (monoterpenes and phenylpropanoids), thus potentially enhancing varietal aroma in wine (Swiegers et al., 2005). Enterobacteriales have been also previously related to wine fermentation rate by Bokulich et al. (2016). Fungal genera abundance (for example Penicillium) negatively correlated with the medium-chain fatty acid octanoic acid, which can contribute to mushroom flavour and also alter Saccharomyces metabolism (Bisson, 1999). These microorganisms correlate with wine metabolites and/or can change chemical environments and fermentation behaviours and finally exert substantial effects on wine metabolite profiles (Swiegers et al., 2005; Bokulich et al., 2016).

71 References

Bastian, M., Heymann, S., and Jacomy, M. (2009) Gephi: an open source software for exploring and manipulating networks. In Third international AAAI conference on weblogs and social media.

Bisson, L.F. (1999) Stuck and sluggish fermentations. Am J Enol Vitic 50: 107-119.

Bokulich, N.A., Ohta, M., Richardson, P.M., and Mills, D.A. (2013) Monitoring seasonal changes in winery-resident microbiota. PLoS One 8: e66437.

Bokulich, N.A., Collins, T.S., Masarweh, C., Allen, G., Heymann, H., Ebeler, S.E., and Mills, D.A. (2016) Associations among Wine Grape Microbiome, Metabolome, and Fermentation Behavior Suggest Microbial Contribution to Regional Wine Characteristics. MBio 7: e00631-00616.

Junker, B.H., and Schreiber, F. (2008) Analysis of biological networks: Wiley Online Library.

Swiegers, J., Bartowsky, E., Henschke, P., and Pretorius, I. (2005) Yeast and bacterial modulation of wine aroma and flavour. Aust J Grape Wine Res 11: 139-173.

72 CHAPTER FOUR

Community succession of the grapevine fungal microbiome in the

annual growth cycle

Status of chapter:

This chapter has been published in Environmental Microbiology journal. The co-authorship form is presented at the start of this thesis, after the acknowledgements.

Liu, D. and Howell, K.S.. (2020). Community succession of the grapevine fungal microbiome in the annual growth cycle. Environmental Microbiology. Accepted Author Manuscript. doi:10.1111/1462-

2920.15172

73 Environmental Microbiology (2020) 00(00), 00–00 doi:10.1111/1462-2920.15172

Community succession of the grapevine fungal microbiome in the annual growth cycle

Di Liu and Kate Howell * modulate vine health, growth, and productivity (Gilbert School of Agriculture and Food, Faculty of Veterinary et al., 2014; Müller et al., 2016). Microbes originate from and Agricultural Sciences, University of Melbourne, the surrounding environment (including soil, air, precipita- Parkville, 3010, Australia. tion, animal vectors, native forests) and colonize both exterior (epiphytes) and interior (endophytes) (Goddard et al., 2010; Stefanini et al., 2012; Lam and Howell, 2015; Summary Zarraonaindia et al., 2015; Morrison-Whittle and Microbial ecology and activity in wine production Goddard, 2018). This microbiota can be transferred to influences grapevine health and productivity, conver- the grape must (or juice) and thus have a profound influ- sion of sugar to ethanol during fermentation, wine ence on wine composition, flavour, aroma, and quality aroma, wine quality and distinctiveness. Fungi in the (Ciani et al., 2010; Barata et al., 2012; Morrison-Whittle vineyard ecosystem are not well described. Here, we and Goddard, 2018). Saccharomyces yeasts (primarily characterized the spatial and temporal dynamics of Saccharomyces cerevisiae) and lactic acid bacteria (pre- fungal communities associated with the grapevine dominantly Oenococcus oeni) are primary drivers of wine (grapes, flowers, leaves, and roots) and soils over an fermentation and to form wine flavour and aroma annual growth cycle in two vineyards to investigate (Swiegers et al., 2005). Beyond these species, our the influences of grape habitat, plant developmental knowledge about the microbial ecology associated with stage (flowering, fruit set, veraison, and harvest), grapevines and its role in the vineyard ecosystem is vineyards, and climatic conditions. Fungi were limited. influenced by both the grapevine habitat and plant Recent studies posit the existence of geographically development stage. The core microbiome was priori- structured microbiota that contribute to the expression of fi tized over space and time, and the identi ed core regional distinctiveness of agricultural products, specifi- members drove seasonal community succession. cally, known as ‘terroir’ in viticulture [reviewed by Liu The developmental stage of veraison, where the et al. (2019)]. Significant biogeographic patterns are gen- grapes undergo a dramatic change in metabolism erally observed at scales of hundreds to thousands of and start accumulating sugar, coincided with a dis- kilometres (Gayevskiy and Goddard, 2012; Bokulich tinct shift in fungal communities. Co-occurrence net- et al., 2014; Taylor et al., 2014; Knight and Goddard, works showed strong correlations between the plant 2015) and influenced by multiple factors, such as climate microbiome, the soil microbiome, and weather indi- and topography, soil properties, and even local anthropo- ces. Our study describes the complex ecological genic practices (Bokulich et al., 2014; Pinto et al., 2014; dynamics that occur in microbial assemblages over a Burns et al., 2015; Zarraonaindia et al., 2015; Jara growing season and highlight succession of the core et al., 2016; Portillo et al., 2016; Morrison-Whittle and community in vineyards. Goddard, 2017). Microbial spatial pattern analyses have mainly been conducted at large scales and focused on Introduction studying grapes and must, the communities present at the start of fermentation, and the soil, but the nature and The grapevine (Vitis vinifera) is naturally colonized by the importance of microbial ecology within vineyards complex and diverse microorganisms, such as bacteria, towards defining terroir, is largely unknown. filamentous fungi, and yeasts (Barata et al., 2012; Metagenomics tools have been actively used to map Stefanini and Cavalieri, 2018), which substantially the plant microbiome, in particular the phyllosphere and the rhizosphere (Turner et al., 2013; Müller et al., 2016). In viticulture, many studies have been dedicated to Received 15 May, 2020; revised 16 July, 2020; accepted 17 July, 2020. *For correspondence. E-mail [email protected]; Tel: characterizing grapevine-associated bacteria and their +61 3 90353119. relevant importance on plant growth and winemaking

74 2 D. Liu and K. Howell processes (Zarraonaindia et al., 2015; Marasco knowledge of the core microbiome associated with et al., 2018). Zarraonaindia et al. (2015) investigated the grapevine across different soil types and environmental spatial and temporal dynamics of the bacterial microbiota conditions to allow optimization of the growth of this associated with grapevine organs of Merlot cultivars in important economic crop. New York state, and revealed significant differences among grapevine habitats and the soil, although the latter serving as a source reservoir for the former. The endo- Results phytic fungi are diverse and structured in rhizo- Fungal communities vary by grapevine habitat within compartments and the vascular system (Deyett and vineyards Rolshausen, 2020). Fungal communities across multiple compartments have not been systematically investigated. A total of 4 786 543 ITS high-quality sequences were As described in other plant systems, microbiome compo- generated from all samples (n = 160), which were clus- sition is not temporally stable but change in response to tered into 2286 fungal OTUs with a threshold of 97% plant development (Mougel et al., 2006; Jumpponen and pairwise identity. No OTUs were found to be common Jones, 2010; Chaparro et al., 2014; Copeland et al., across all the samples. Among all the OTUs identified in 2015; Grady et al., 2019) and can fundamentally affect the present study, 62.90% were detected in more than the health and productivity of plants. Core microbiomes one grapevine habitat (root zone soil, root, leaf, flower are commonly defined based on abundance-occupancy and grape), of which 54 OTUs were shared across sam- distributions describing microbes, which are highly abun- ple types (Fig. 1A). The soil contained the most OTUs dant and ubiquitous occurring the habitat/plant species and of these, 37.67% of OTUs found in soils were also over time and are hypothesised to reflect underlying func- associated with the plant, with 32.75% of OTUs shared tional relationships with the host (Lundberg et al., 2012; between soils and roots which was the highest portion of Shade and Stopnisek, 2019). Identifying members and OTUs shared between soil and plant organs (5.86% characteristics of core microbiomes, which are expected between soils and leaves, 11.58% between soils and to play pivotal roles in organizing the dynamics of flowers, 7.95% between soils and grapes). Each plant resident microbiomes, will be the key to optimizing inter- organ shared the majority of fungal taxa with other actions between plants and indigenous microorganisms organs and the soil. Some OTUs were unique for specific and increasing resource-efficiency and stress-resistance organs, with low occurrence frequencies. Seventy-nine of agroecosystems (Toju et al., 2018). In grapevines, OTUs found in grapes were observed specific for this however, this remains to be shown. organ (Fig. 1A) with a low occurrence frequency ≤ 10%, In the present study, we sampled microbial communi- accounting for 28.11% of total grape-associated OTUs; ties associated with Pinot Noir grapevines from two vine- likewise, the organ-specific OTU portion was 9.11% for yards within the same grape growing region to provide flowers (frequency ≤ 30%), 8.24% for leaves (fre- insights into fungal ecology in the vineyard. Using internal quency ≤ 2.5%), and 8.53% for roots (frequency ≤ 7.5%). transcribed spacer (ITS) amplicon sequencing to charac- Ascomycota was the most abundant phylum in the vine- terize fungal communities, we disentangled the influ- yard habitats comprising 77.69% of all sequences, ences of geographic location (between two vineyards), followed by Basidiomycota and Mortierellomycota in the grapevine habitat (soil, roots, leaves, flowers and plant, and Mortierellomycota and Basidiomycota in the grapes), grapevine developmental stage (flowering, fruit soil respectively (data not shown). The major fungal genera set, veraison, harvest), and climatic conditions on the fun- were Aureobasidium, Cladosporium, Epicoccum, Mor- gal diversity, composition, and structure from root zone tierella, Cryptococcus, Debaryomyces, Saccharomyces, soil, roots, leaves, and grapes or flowers. Specifically, we Mycosphaerella, Lophiostoma, Alternaria,andPenicillium, aimed to prioritize the core microbiome over space and with relative abundance of more than 1.0% across all sam- time and further confirm the role of this core community ples (Fig. 1B). Some major taxa associated with plants, such on seasonal community succession based on Random as Aureobasidium, Debaryomyces, Saccharomyces and Forest supervised learning models. The influence of the Mycosphaerella, were present in soil samples at very low changing environment was elucidated with network anal- abundances (<0.05%). The genus Mortierella had the ysis. We show that fungal communities differ according highest relative abundance (26.00%) in the soil and to grapevine habitats and developmental stages, with lit- showed lower abundances in the plant (Fig. 1B). tle geographic influences at this scale (5 km). The core Grapevine-associated fungi varied with vine habitat microbiome correlates closely with solar radiation and (root zone soil, root, leaf, flower and grape) and exerted water status during the annual growth cycle and drives significant differences in both community diversity seasonal community succession. Combining our study (α-diversity, Shannon index; ANOVA, F(4,155) = 5.206, with studies in other grape growing areas will expand our P < 0.001) and dissimilarity (β-diversity, Bray–Curtis

75 The core fungal microbiome of grapevines 3

A Root C 65

Soil 428 4 0 1254 73 7 0.4 31 13 14 24 20 28 Leaf 7 12 5 54 51 3 3 6 5 0.0 13

9 PCo2 (16.9%) 1 3 1 8 Flower 25 3 Grape 79 Leaf −0.4 Root Grape Soil Proportion of OTUs

0.00 0.25 0.50 0.75 1.00 Detect in more than one habitat −0.5 0.0 0.5 Detect in only one habitat PCo1 (30.5%)

B D 100 Ascomycota a: Cladosporium_ramotenellum Grape Dothideomycetes b: Cladosporium Leaf Dothideales c: Cladosporiaceae Root d: Mycosphaerella_tassiana Soil e: Mycosphaerella f: Mycosphaerellaceae g: Aureobasidium_pullulans 75 Genus Agaric h: Aureobasidium Aureobasidium i: Dothioraceae omycetes Cladosporium j: Pyrenochaetopsis_leptospora Epicoccum k: Pyrenochaetopsis Mortierella l: Cucurbitariaceae Cryptococcus m: Didymella_pedeiae 50 Debaryomyces n: Exophiala_equina o: Penicillium_onobense Saccharomyces p: Mortierella_amoeboidea

Relative Abundance (%) Mycosphaerella q: Mortierellaceae Lophiostoma r: Saccharomyces_cerevisiae Alternaria s: Saccharomyces

Penicillium Mortierellales

Xylariales t: Saccharomycetaceae Mortierellomycota 25 Others u: Debaryomyces_hansenii v: Debaryomyces

w: Saccharomycetales_fam_Incertae_sedis Mortierellomycotina_cls_Incertae_sedis x: Fusarium_denticulatum

y: Amphisphaeriaceae

Saccharomycetales z: Cryptococcus_saitoi Saccharomycetes

Soil Root Leaf Flower Grape

Fig 1. Fungal communities demonstrate different structures and compositions depending on habitats in the vineyard. A. Fungal OTUs shared among soil and grapevine habitats (roots, leaves, flowers, grapes) in the vineyard. B. Microbial community composition characterized to the genus level (relative abundance >1.0% shown). C. PCoA among plant organs and soil samples at harvest stage based on Bray-Curtis distances. D. Linear discriminant analysis effect size taxonomic cladogram identifying significantly discriminant taxa associated with habitats at harvest. distance; PERMANOVA, R2 = 0.271, P < 0.001) indepen- (including Pyrenochaetopsis), Didymella pedeiae, Exophiala dent of phenological development stage (Table 1). This trend equina and Penicillium onobense were observed with wasmoredistinctwithanimprovedcoefficient of determina- higher abundances in the soil. For grapevine habitats, Sac- tion (R2) within a certain stage (Table 1). Geographic location charomycetaceae (notably fermentative yeast S. cerevisiae), only weakly impacted microbial diversity and composition, Cryptococcus saitoi (Basidiomycota yeasts), Dothioraceae except for fungal diversity associated with roots, which was (including Aureobasidium)andMycosphaerellaceae (includ- significantly impacted by vineyard (ANOVA, F(1,38) =5.878, ing Mycosphaerella)weresignificantly more abundant in P < 0.01) (Table 1). Principal coordinate analysis (PCoA) roots than in other habitats, whereas Saccharomycetes showed the habitat-specific patterns (95% confidence inter- (including Debaryomyces)andFusarium denticulatum were val) of fungal communities, with 47.4% of total variance more abundant in leaves, and Ascomycota (notably explained by the first two principal coordinate (PC) axes using Cladosporium) in grapes. the harvest stage as an example (Fig. 1C). Linear discriminant analysis (LDA) effect size (LEfSe) analysis further confirmed that these patterns related to significant associations Temporal fungal community dynamics at grapevine (Kruskal–Wallis sum-rank test, α < 0.05) between fungal taxa phenological stages and sample types (Fig. 1D). At harvest, Mortierellomycota (notably Mortierella amoeboidea), Agaricomycetes, Significant variances in the fungal diversity were recorded Xylariales (including Amphisphaeriaceae), Cucurbitariaceae at each grapevine habitat over time (Table 1; Fig. 2A).

76 4 D. Liu and K. Howell

Table 1. Experimental factors predicting α- and β-diversity of fungal communities in the vineyard.

α-diversity (Shannon)a β-diversity (Bray–Curtis)b Factor Sample group Df F P Rb P

Habitat All stages 4, 155 5.206 1.05E-02 0.271 0.001 Flowering 3, 36 18.109 4.00E-11 0.366 0.001 Fruit set 3, 36 38.109 4.00E-11 0.385 0.001 Veraison 3, 36 5.668 2.93E-03 0.466 0.001 Harvest 3, 36 22.848 2.27E-08 0.350 0.001 Stage Soil 3, 36 3.858 0.018 0.229 0.001 Root 3, 37 1.390 0.026 0.155 0.008 Leaf 3, 38 3.396 0.001 0.120 0.014 Grape 2, 27 6.392 2.23E-02 0.426 0.001 Vineyard Soil 1, 38 3.969 0.055 0.034 0.203 Root 1, 38 5.878 0.008 0.052 0.073 Leaf 1, 38 0.018 0.894 0.017 0.806 Flower 1, 8 0.988 0.353 0.513 0.757 Grape 1, 28 0.788 0.382 0.025 0.551

Bold values indicate statistically signiﬁcant results, P < 0.05. aOne-way ANOVA. bPermutational multivariate analysis of variance (PERMANOVA) using distance matrices.

AB 3.5 Flower Grape Leaf Root Fruit set 3.0 Soil Veraison 0.25 Harvest

2.5 Shannon

2.0 PCo2 (10.8%) 0.00

1.5

−0.25 1.0

Flowering Fruit set Veraison Harvest −0.25 0.00 0.25 0.50 PCo1 (54.9%)

Fig 2. Grape developmental stage inﬂuences microbial diversity in the vineyard. A. α-diversity (Shannon index) changes during the growing season (bars indicate SE). B. Bray–Curtis distance PCoA of grape fungal communities according to the developmental stage. Shown are grapes sampled from fruit set, veraison, and harvest stage. Flowering was excluded as ﬂowers were sampled and counted as another grapevine habitat.

Soil fungi showed higher diversity than that of grapevine veraison. Interestingly, veraison was found to be a key organs (except flowers) throughout the growing season, stage of diversity changes in grapevine-associated fungi with a rapid increase before fruit set, and then a decrease microbiota (Fig. 2A). There were significant differences in by veraison, and finally an increase at harvest. The trend fungal community composition over time regardless of was also observed in root fungal diversity. After fruit set, grapevine habitat (Table 1). We observed the most distinct the diversity of grape fungi significantly increased pattern in grapes with the highest R2 coefficients 2 (ANOVA, F(2,27) = 6.392, P < 0.001) over the development (PERMANOVA, R = 0.426, P < 0.001), followed by the and ripening process, reaching a maximum at harvest. For soil (PERMANOVA, R2 =0.229,P < 0.001). As an example, leaves, the fungal communities tended to be less complex PCoA showed a strong separation of grape samples, espe- and diverse as grapevine development progressed, cially samples at harvest from earlier stages on PC1, which although an increased Shannon index was observed at explained 54.9% of the variance (Fig. 2B).

77 The core fungal microbiome of grapevines 5 The core microbiome fluctuates according to plant a key stage where the core community was distinct from development other stages (Fig. 3C–F). Some taxa changed in a similar manner in plant organs. For instance, the most dominant Here, we prioritized the core microbiome over space and genera, Cladosporium, accumulated in relative abun- time to further investigate fungal community succession dance, between post-veraison grapes and leaves, while in the vineyard. We first identified the most dominant Aureobasidium and Cryptococcus declined in those OTUs based on the abundance-occupancy distributions post-veraison organs (Fig. 3C–E). Fermentative yeasts (see Experimental procedures for details) to include in S. cerevisiae and Debaryomyces hansenii persisted in the core (see Supporting Information Table S1 for a com- the grapevine throughout the growth cycle but decreased plete list) for the grapevine and the soil respectively. in relative abundances of post-veraison grapevine habitats About 1.51% of OTUs (15 OTUs) constituted a core that (Supporting Information Table S1). However, S. cerevisiae accounted for 74.70 ± 2.70% of the sequences across all increased in the post-veraison root microbiome. Some plant samples (Fig. 3A; Supporting Information Table S1A). species like Epicoccum sp. and Alternaria sp. showed On average, the core community was highly abundant in changes in developing grapes and became much more each grapevine habitat, with a maximum recorded in abundant at harvest, but these genera only showed mod- grapes (90.55 ± 2.27% in relative abundance) (Fig. 3B; erate changes in leaves and roots (Fig. 3C–E; Supporting Supporting Information Table S1A). A wide variety of fungi Information Table S1A2–A4). It is noteworthy that a few was observed within those core members and included fer- core members only appeared or appeared in high occu- mentative yeasts (Saccharomyces, Debaryomyces), yeast- pancy in specific developmental stages. The best example like fungi (Aureobasidium, Cryptococcus, Vishniacozyma), is OTUs assigned to Alternaria infectoria and Alternaria filamentous fungi (Cladosporium, Alternaria, Penicillium, rosae where taxa only were detected in grape samples at Fusarium), and other genera (Mycosphaerella, Didymella, veraison and accumulated afterwards (Supporting Ramularia, Epicoccum) (Supporting Information Table S1A). Information Table S1A2). For the soil habitat, the most Within grapevine habitats, some species outside the grape- dominant genus Mortierella decreased in the relative vine core dominated and developed over time and also abundance during the plant development (especially after defined the core specific to this organ (Supporting Information veraison), accompanied by an increase in ascomycetous Table S1A1–A4). Specifically, three core species from genera (Fig. 3F). Within ascomycetous genera, those taxa Mortierella genus and Microascaceae family were as part shared between soil and plants accumulated as the of core community of flowers (Supporting Information growth cycle processed, including Epicoccum, Alternaria, Table S1A1), with three from Filobasidium and Alternaria Fusarium, Cladosporium and Penicillium,withnotably genera for grapes (Supporting Information Table S1A2), great changes at veraison, as well as soil yeasts and three from Hypocreales order, Thermoascaceae fam- E. equina and Solicoccozyma phenolica, and some soil- ily and Lophiostoma genus for roots (Supporting Informa- specificfungi,suchasTrichoglossum, Podospora and tion Table S1A4). For leaves, only eight OTUs within the Humicola (Fig. 3F; Supporting Information Table S1B). grapevine core dominated the community, while other core OTUs showed lower occupancy and/or abundance during the growth cycle (Supporting Information Table S1A3). For Core microbiome drives seasonal community soil, 1.74% of OTUs (35 OTUs) were the core members succession accounting for 63.40 ± 3.54% of the reads across all sam- To further confirm the robustness of these observations ples, which presented a high abundance at each stage across the growth cycle, Random Forest supervised (Supporting Information Table S1B). The soil core taxa learning models (Breiman, 2001) were employed to clas- belonged to Ascomycota, Mortierellomycota (Mortierella), sify samples and identify which taxa explain the strongest Basidiomycota (e.g., Clavaria and Auricularia) and variations during grapevine developmental stages. The Chytridiomycota (Rhizophlyctis), including some filamen- fungal microbiota had high discriminative power to distin- tous fungi like Podospora, Humicola, Ilyonectria and guish samples coming from various stages, especially for Sarocladium, and yeasts including Exophiala and basidio- grapes and soils (class errors, 0.100 and 0.125 on aver- mycetous Solicoccozyma. age, respectively; Supporting Information Table S2). Tracking core OTUs within individual habitats revealed Compared with all other stages, veraison had the lowest distinct temporal dynamics and succession associated predictive error (class error = 0.150 on average), for with plant development (Fig. 3C–F). All the core genera example, all soil samples at veraison were correctly iden- in the alluvial diagrams presented significantly different tified (class error = 0.000), suggesting that veraison was relative abundances during the annual growth cycle the most distinct stage of the development. [ANOVA, false discovery rate (FDR)-corrected P value of All the core OTUs were important features of these <0.05]. Overall, in each habitat, veraison appeared to be models for grapevine habitats (Fig. 4). Filamentous fungi

78 6 D. Liu and K. Howell A B

Flower 1. 5 1 % 7. 1 2 % Grape

98.49% 74.70% Leaf

Root 15 core OTUs Others 0 25 50 75 100 C D Relative Abundance (% all reads) 100 Grape 100 Leaf

80 80 Genus Genus Aureobasidium Aureobasidium Cladosporium Cladosporium Epicoccum Cryptococcus 60 Cryptococcus 60 Saccharomyces Alternaria Debaryomyces Mycosphaerella Mycosphaerella Debaryomyces Epicoccum Didymella Saccharomyces 40 Vishniacozyma 40 Relative Abundance (%) Abundance Relative Filobasidium (%) Abundance Relative Ramularia Fusarium

20 20

0 0

Fruit set Veraison Harvest Flowering Fruit set Veraison Harvest EF 100 Root 60 Soil

Genus Genus 80 Aureobasidium Mortierella Cryptococcus Trichoglossum Cladosporium Epicoccum Mycosphaerella Podospora Lophiostoma 40 Humicola Epicoccum Ilyonectria 60 Debaryomyces Rhizophlyctis Ramularia Archaeorhizomyces Alternaria Clavaria Didymella Exophiala Vishniacozyma Penicillium Saccharomyces Pyrenochaetopsis

40 Penicillium (%) Abundance Relative Alternaria Relative Abundance (%) Abundance Relative Fusarium Chaetomium 20 Solicoccozyma Auricularia Fusarium Sarocladium 20 Cladosporium Gibberella

0 0

Flowering Fruit set Veraison Harvest Flowering Fruit set Veraison Harvest

Fig 3. The core microbiome in grapevine habitats and the soil over the growing season. A. Percentage (left) and relative abundance (right) of OTUs representing the core versus the remaining microbiome associated with the grapevine (ﬂowers, grapes, leaves, roots). B. Relative abundance (mean ± SE) of dominant OTUs across plant habitats. C–F. Relative abundance changes of dominant fungal genera of grapes (C), leaves (D), roots (E), and soil (F) from the beginning to the end of growing season.

79 The core fungal microbiome of grapevines 7 (e.g., Cladosporium ramotenellum and Fusarium distinguished based on the associated core microbiome denticulatum), fermentative yeasts S. cerevisiae and and defined predictive features for dynamic fungal ecol- D. hansenii, and some yeast-like fungi (Aureobasidium ogy in the vineyard. pullulans and Cryptococcus saitoi) within the core (Supporting Information Table S1) were top features in the classification models for vine habitats. Several non- Co-occurrence patterns between the core microbiome core taxa were also important for the models (see and vineyard weather Supporting Information Table S3 for a complete list). For instance, yeast Hanseniaspora uvarum (OTU_33) which During the annual growth cycle, all weather parameters was present in low occupancy (≤50%), was an important displayed significant differences between the phenologi- feature distinguishing developmental stage, with cal stages (Supporting Information Table S4). Stage- higher abundance at veraison in grapes or harvest in dependent patterns in grapevine-associated microbiota leaves and roots; also, yeasts Torulaspora delbrueckii (Fig. 3; Supporting Information Table S1, S3) also sug- (OTU_121) were important for both grapes and leaves, gest that longitudinal environmental conditions are likely and Rhodotorula babjevae (OTU_40) for grapes. In addi- responsible for structuring the microbial communities. tion, some soil core species were important for random Network analysis was conducted to explore the co- forest models of plant organs, including Mortierella spp., occurrence patterns between plant core microbiome and Sarocladium bactrocephalum, E. equina, which were environmental matrices including abiotic (weather) and present only at specific stages (Supporting Information biotic factor (soil microbiota), based on the strong correla- Table S2B). A portion of soil core taxa explained the tion coefficients (Spearman correlation coefficient, highest variations of some phenological stages, for ρ ≥ 0.8; P < 0.01) (Fig. 5). Overall, the plant microbiota instance, Epicoccum nigrum (OTU_15) and Pyr- presented high connectivity with soil microbiota and enochaetopsis leptospora (OTU_62) were the most weather conditions. The more densely connected module important predictors (Fig. 4; Supporting Information was observed in grape microbiota (average Table S1). Non-core taxa were mainly detected at early degree = 3.610; Fig. 5A) than leaf (average stages (flowering and fruit set; e.g., Archaeorhizomyces degree = 2.381; Fig. 5B) and root (average sp., Solicoccozyma terricola) or later stages (e.g., Peni- degree = 2.316; Fig. 5C). In addition, grape microbiota cillium multicolor, Aspergillus ardalensis), and contrib- was significantly correlated with each weather parameter. uted to the models. When we used the core subset to Across habitats, the most closely connected fungi were build the random forest model, a higher classification A. pullulans (OTU_1), C. ramotenellum (OTU_2), accuracy (class error = 0.075 on average) was achieved, C. saitoi (OTU_3), and F. denticulatum (OTU_2019). with improved discrimination for fruit set and harvest Correspondingly, high-connectivity nodes with these (0.200 and 0.000, respectively). Taken together, these OTUs were soil species Penicillium sp., Auricularia sp., results confirmed that plant development stages can be Humicola sp., Mortierella sp., and Ilyonectria sp., and

Grape Leaf Root Soil OTU_2 OTU_9 OTU_6 OTU_15 OTU_8 OTU_1808 OTU_2 OTU_62 OTU_5 OTU_2 OTU_3 OTU_1093 OTU_1 OTU_9 OTU_584 OTU_15 OTU_3 OTU_26 OTU_21 OTU_74 OTU_6 OTU_5 OTU_2019 OTU_14 OTU_5 OTU_1 OTU_16 OTU_501 OTU_3 OTU_2019 OTU_2019 OTU_146 OTU_1093 OTU_22 OTU_151 OTU_64 OTU_21 OTU_7 OTU_8 OTU_54 OTU_15 OTU_17 OTU_88 OTU_9 OTU_48 OTU_15 OTU_1 OTU_4 OTU_1093 OTU_1093 OTU_26 OTU_2078 OTU_4 OTU_49 OTU_33 OTU_24 OTU_8 OTU_519 OTU_6 OTU_20 OTU_43 OTU_40 OTU_38 OTU_47 OTU_179 OTU_20 OTU_2201 OTU_16 OTU_35 OTU_37 OTU_23 OTU_51 OTU_2201 OTU_37 OTU_16 OTU_122 OTU_38 OTU_763 OTU_121 OTU_47 OTU_38 OTU_47 Core OTUs OTU_323 Core OTUs OTU_33 Core OTUs OTU_223 Core OTUs OTU_4 Others OTU_35 Others OTU_29 Others OTU_392 Others OTU_121 OTU_20 OTU_1808 OTU_94 OTU_323 OTU_33 OTU_2201 OTU_128 OTU_26 OTU_1808 OTU_341 0.00 0.20 0.40 0.60 0.00 0.25 0.50 0.75 1.00 0.00 0.20 0.40 0.60 0.00 0.10 0.20 0.30 Mean Decrease Gini Mean Decrease Gini Mean Decrease Gini Mean Decrease Gini

Fig 4. The core microbiome (as OTUs) can distinguish developmental stages for plant habitats and soil. Shown are the important features (values >0.10) based on Mean Decrease Gini (MDG) of random forest models. Core members were coloured in pink (grape), green (leaf), blue (root), and navy (soil). Identiﬁed OTU species are given in the Supporting Information Table S1.

80 8 D. Liu and K. Howell

g_OTU_16 A g_OTU_1 BC g_OTU_15 g_OTU_2 s_OTU_30 g_OTU_2019 s_OTU_88 g_OTU_26 s_OTU_10 s_OTU_89 g_OTU_64 s_OTU_88 l_OTU_2 s_OTU_23 s_OTU_49 l_OTU_9 g_OTU_5 s_OTU_17 r_OTU_8 g_OTU_3 l_OTU_26 s_OTU_1093 s_OTU_89 s_OTU_72 g_OTU_20 r_OTU_1 s_OTU_66 s_OTU_72 g_OTU_6 s_OTU_1093 l_OTU_2019

s_OTU_66 s_OTU_46 r_OTU_26 s_OTU_21 g_OTU_74 l_OTU_1 s_OTU_185 r_OTU_3 s_OTU_19 s_OTU_30 g_OTU_1093 s_OTU_2 l_OTU_3 s_OTU_17 s_OTU_129 r_OTU_15 g_OTU_9 s_OTU_35 l_OTU_6 s_OTU_2 s_OTU_46 g_OTU_8 s_OTU_19 r_OTU_22 s_OTU_185 Env8 s_OTU_88 Env1 s_OTU_23 r_OTU_2 s_OTU_150 Env1 s_OTU_17 s_OTU_37 s_OTU_91 Env6 s_OTU_89 Env6 s_OTU_105 r_OTU_2019 s_OTU_105 Env5 s_OTU_91 s_OTU_179 s_OTU_56 Env8 s_OTU_35 s_OTU_56 s_OTU_14 Env7 s_OTU_23 Env2 Env3 s_OTU_179 Env4 Env8 s_OTU_51 Grape OTUs Leaf OTUs Root OTUs Soil OTUs Environmental factors

Fig 5. Statistically significant and strong co-occurrence relationships among grapevine-associated and soil dominant fungi and weather conditions during annual growth cycle. Network plots for grape (A), leaf (B), and root (C) core microbiota with weather matrices and soil core microbiota. Circle nodes represent core OTUs and environmental variables, with different colours. Direct connections between nodes indicate strong correlations (Spearman correlation coefficient, ρ ≥ 0.8; P < 0.01). The colour of edges describes the positive correlation (pink) or the negative correlation (grey). The size of nodes is proportioned to the interconnected degree. The width of edges is proportioned to the correlation coefficients. Env1, solar radiation; Env2, mean high temperature; Env3, mean low temperature; Env4, mean temperature; Env5, precipitation; Env6, evaporation; Env7, transpiration; Env8, relative soil moisture. changing solar radiation (Env1) and vine water status functioning of the vineyard ecosystem and opens the related parameters (Env6, evaporation; Env8, relative soil potential to use specific fungi to deliberately manipulate moisture). Likewise, soil taxa were also strongly corre- the microbiome to affect grapevine health, growth, and lated with weather parameters (in particular Env1, 6, 8) grape and wine production. (Supporting Information Fig. S1). In addition, habitat- specific correlations were also found. For example, fi grape-speci c core taxon A. infectoria (OTU_74) and Community succession of the grapevine-associated fi root-speci c core taxon Lophiostoma sp. (OTU_22) were microbiota over space and time densely connected nodes for them respectively. The below ground root microbiota was significantly correlated Our data show that fungal communities are not stochasti- with relative soil moisture. cally assembled but have a distinct community with habitat patterns according to the development stage of grapevines (Fig. 1). Fungal communities varying by sites has previously been shown in vineyard systems at small Discussion scales (Coller et al., 2019; Knight et al., 2019). In our Microbiomes are dynamic over space and time. Despite study, grapevine habitats and plant development within the overwhelming diversity of microbial communities, rel- the sampled vineyards outweighed biogeographic trends atively few microbial taxa dominate the ecosystems at this scale (5 km), which were only significant for root locally or globally (Delgado-Baquerizo et al., 2018; Egidi fungal diversity, suggesting that fungal assemblages et al., 2019; Grady et al., 2019; Shade and Stopnisek, have extensive local heterogeneity associated with their 2019). Describing the core microbiome is essential for host plant (Table 1). Grapevines maintain distinct fungal unravelling the stable, consistent components across microbial communities in a habitat-specific manner, but complex microbial assemblages. The definition of a core also share many taxa. Fungal diversity is higher below microbiome provides the scope to define a reproducible the ground (root zone soil, roots) than the aboveground and conservative perspective for the dynamic microhabitats (flowers, leaves, and grapes), and this is in line biome and allows for systematic discovery of persistent with other studies on grapevines (Zarraonaindia et al., taxa, especially for microbiota associated with perennial 2015; Morrison-Whittle and Goddard, 2017; Deyett and plants (Shade and Stopnisek, 2019). Here, our study Rolshausen, 2020). Soil has been suggested as the narrows down the immense number of measurable fungal source and reservoir of plant microbiota (Zarraonaindia taxa to the core microbiome to advance understanding of et al., 2015; Morrison-Whittle and Goddard, 2018; Grady grapevine-associated fungal microbiota. This is an impor- et al., 2019); yet many core taxa associated with grape- tant step to further investigate contributions of fungi to vine organs (e.g., Cladosporium, Aureobasidium, and

81 The core fungal microbiome of grapevines 9 Saccharomyces) are in low abundances in our soil sam- influence the composition of the rhizosphere microbiome ples (Fig. 1B and D). This differentiation indicates that (Berg and Smalla, 2009; Chaparro et al., 2014). During there are ecological filters operated by the plant to favour the growing season, vineyards were treated regularly with some species while excluding others (Chesson, 2000; sulfur- and copper-based fungicides. The influences of Leibold and McPeek, 2006). The core taxa found here fungicides on microbial diversity and composition associ- persist across habitats in differing abundance with some ated grapevine have been widely discussed (Setati species only existing to a specific plant organ. We et al., 2012; Martins et al., 2014; Morrison-Whittle and observed a decrease in fungal diversity between the root Goddard, 2017). Whilst this is non-negligible, the fact of zone soil and the roots but an enrichment of saprophytic microbial assemblage and community succession fungi (Mycosphaerella and Lophiostoma, core taxa in according to plant development is of significance. The roots; Fig. 3E) and dilution of potentially pathogenic fungi dominance of yeast-like fungus A. pullulans in the vine- (Ilyonectria, core taxa in soil; Fig. 3F). These fungi have yard has previously been attributed to its resistance to been previously reported colonizing the rhizocompartments fungicides, that the ability to detoxify copper compete of grapevines (Martínez-Diz et al., 2019; Deyett and against other fungi (Comitini and Ciani, 2008; Schmid Rolshausen, 2020). Arbuscular mycorrhizal fungi including et al., 2011). Martins et al. (2014) demonstrated that Glomeromycota: Rhizophagus, Funneliformis, and growth-stage factor was preponderant in shaping fungal Acaulospora, which positively affect grapevine growth in a diversity and richness in grape surface throughout the mutualistic manner (White, 2015), were recovered with low ripening process, followed by fungicide treatments prevalence and abundance in root zone soil and root (encapsulated in the farming system). samples in our study (data not shown). This could be due Our study shows that a core microbiome of grapevines in part to the choice of barcoding region used here, rather exists over space and time in the vineyard ecosystem. Core than the large subunit rRNA, which is the preferred bio- microorganisms are expected to play important roles in marker for these fungi (Lekberg et al., 2012). However, organizing assembly of plant-associated microbiomes geographic location may also influence distribution (Kivlin rather than promoting plant growth directly (Toju et al., 2011; Hazard et al., 2013), as found in vineyards by et al., 2018). A core microorganism can be a mutualist, a Coller et al. (2019) who retrieved Glomeromycota as a core commensalist or even an antagonist when it is the sole inoc- phylum in Italy, using the same ITS1F/2 primers as in our ulum in a plant (Lundberg et al., 2012; Toju et al., 2018). In study. Interestingly, the genus Mortierella (Mortierello- our study, the core species Mycosphaerella tassiana, Cla- mycota phylum), an important component of phosphorus dosporium ramontellum, Didymella negriana are normally cycling in the rhizosphere (Osorio and Habte, 2001), were considered to be grapevine pathogens, whereas we observed as core members associated with both soil and observed no symptoms of plant diseases in the vines. We flowers (Fig. 1B; Supporting Information Table S1). The cannot easily explain these results but suggest that the presence of this genus in flowers is unexpected and may induction of disease by these species may be dependent point to a role of flowers in forming and establishing the not just on presence, but the community-scale influences or grapevine microbiome and their functionality. other plant factors. We observed an increase in the relative The grapevine-associated microbiota is also affected abundance of the soil core taxa that were shared with plants by the plant development stage. The leaf fungal diversity over the growing cycle, and this lends further support to the decreased during the growth cycle (Fig. 2A), as previ- influence from the roots and plant development on the soil ously described for grapevine and other plant systems microbiome (Fig. 3F). Aureobasidium are the most wide- (Pinto et al., 2014; Copeland et al., 2015). This may be spread fungi in the vineyard and persist throughout the partly attributed to experiencing greater climatic stress growing season and decline at later stages across plant (UV radiation, temperature, and humidity) in the organs (Fig. 3), supporting previous studies in vineyard eco- phyllosphere and thus only eight core OTUs persist over systems (Renouf et al., 2005; Martins et al., 2014; Pinto time due to environmental fluctuations. Microbial diversity et al., 2014). Using high-throughput sequencing, our data increased in the grapes from fruit set to harvest stage reveal the persistence of S. cerevisiae across habitats and (Fig. 2A) and may be explained by sugar accumulation developmental stages as core taxa, which has previously and exudation as ripening proceeds. It is well docu- been detected in mature grapes based on enrichment mented that accumulation of sugars favours fungal colo- culture-dependent methods (Fleet et al., 2002; Mannazzu nization (Renouf et al., 2005; Martins et al., 2014). et al., 2002; Renouf et al., 2005) and at low frequencies in Similar trends between soil and root fungal diversity were leaves (Pinto et al., 2014). We did not observe an expected as expected, suggesting that root microbiomes are partly increase in the relative abundance of S. cerevisiae after derived from the rhizosphere (Compant et al., 2011; veraison in grape and leaf samples (Fig. 3C and D), but Lundberg et al., 2012; Deyett and Rolshausen, 2020), there could be increased competition with other species, for and in turn, that root morphology and root exudation can example Cladosporium as sugar becomes more available.

82 10 D. Liu and K. Howell Meanwhile, some non-core OTUs displayed temporal suc- colonization in grapes (Coombe, 1992; Renouf cession, which were screened by a supervised learning et al., 2005). Mougel et al. (2006) recorded for Medicago approach (Fig. 4). These predictors associated with grape- truncatula that the rhizosphere microbial communities dif- vine displayed seasonal volatility, for instance, Filobasidium fered according to the plant development stage, with the stepposum, E. equina,andClonostachys rosea were not most significant shifts observed during the transition from detected at veraison yet reappeared at harvest in roots, as vegetative to reproductive stages. Mounting evidence did M. amoeboidea for leaves, and S. bactrocephalum for suggest regulation by root exudates as one mechanism grapes (Fig. 4; Supporting Information Table S3). Cyclic (Broeckling et al., 2008; Chaparro et al., 2013; Chaparro patterns have been reported in the human gut microbiome et al., 2014; Zhalnina et al., 2018). Plants secrete blends and indicate that some dynamic lineages of microbes of compounds and specific phytochemicals (sugars, decreased in prevalence and abundance in modernized sugar alcohols, phenolics, and amino acids) in the populations (Smits et al., 2017). Conditionally rare taxa root exudates that are differentially produced at distinct (here, specific to the habitat and the development stage) developmental stages acting as signalling molecules to can disproportionately contribute to temporal changes in help orchestrate rhizosphere microbiome assemblage microbial communities, for example, acting as a reservoir (Chaparro et al., 2014). Grapevines may select a subset that can rapidly respond to environmental changes, thus of microbes at different stages of development presum- playing roles in ecosystem resilience and stability (Shade ably for specific functions that the core microbiome or the et al., 2014; Lynch and Neufeld, 2015). In this study, those rare taxa express and thus improving plant fitness in the non-core OTUs were not considered as the main driver of vineyard ecosystem. Future experiments targeting rhizo- seasonal community succession due to lack of persistence sphere microbiome and chemical environments are and clear relationships with grapevines and wine fermenta- essential to elucidate the exact mechanism. tion, but their dynamics provide a more complete view of Alongside the influences of changing metabolism microbial diversity and also indicate some adaptation pro- and physiology during plant development, grapevine- cess for the plant or plant–microbe interactions (Shade and associated microbiome is also subject to dynamic envi- Gilbert, 2015). Further research will determine whether rare ronmental conditions where plants grow. Soil core taxa and related volatility have an ecological or functional microbiome closely correlated with the abiotic matrices role in the vineyard. (weather parameters; Supporting Information Fig. S1), and it is in turn the biotic factor for the plant microbiome. Different plant organs respond to changes in their envi- Grapevine-associated microbiome in response to plant ronment in different manners, via different and/or habitat- development and the changing environment specific core taxa that closely correlated with different The microbiome succession is associated with plant environmental matrices (Fig. 5). Some important features development. Studies have shown that phyllosphere distinguishing stages of soil microbiome identified by ran- (Copeland et al., 2015) and rhizospheric fungal and dom forest models are also highly connected with plant bacterial communities of a wide range of plants, such as core microbiome, including Mortierella sp., Penicillium Arabidopsis, Medicago, wheat, maize, pea, and sugar sp., and Ilyonectria sp.. Water status (evaporation and beet, change according to plant developmental stages relative soil moisture) and solar radiation had a strong (Baudoin et al., 2002; Mougel et al., 2006; Houlden correlation with the microbiome. For instance, higher et al., 2008; Micallef et al., 2009). In our study, veraison solar radiation and temperature, decreased precipitation, is the stage where the fungi microbiome of the grapevine increased evaporation and transpiration, and lower rela- changes very distinctly. Significant changes in the fungal tive soil moisture were observed in the period from fruit diversity and most distinct community structures of both set to veraison in both vineyards (data not shown), and the core (Fig. 3) and the whole (Supporting Information this water stress event coincided with dramatic changes Table S2) datasets were observed at veraison. In particu- in the microbiome during the period discussed above. lar, this observation is clearer in root zone soil and grape Meanwhile, drought triggers substantial plant responses, microbiome, in which many non-core OTUs were only from physiological changes including root-shoot signal- detected before or after veraison (Supporting Information ling (e.g. abscisic acid), the decreased photosynthesis Table S3). In grapevine phenology, veraison marks the and chlorophyll contents, disturbed plant water relations beginning of ripening and is a crucial time point of grape- and osmolyte accumulation, to metabolic perturbations vine metabolism and growth. Besides colour change in like nitrogen metabolism and secondary metabolite syn- the berries due to anthocyanin production and accumula- thesis (Anjum et al., 2011; Hochberg et al., 2013), which tion, berries soften as pectin and cellulose are degraded, may also affect grapevine-associated fungal communi- acidity declines, and sugar accumulates. All these factors ties. In viticulture, moderate water stress during ripening create a more favourable environment for microbial is recognized as beneficial for grape quality, resulting in

83 The core fungal microbiome of grapevines 11 lower yields and a concentration of berry metabolites at two sites in one region in Australia, our research high- (Van Leeuwen and Seguin, 2006). Our work may provide lights the need to further investigate core microbial mem- a microbial insight to understand additional mechanisms bers for functions and interactions with the grapevine to of how water status can affect quality wine production. understand the interactive functionalities that occur Global studies show that climate (especially precipita- between the host crop and their microbiome. Increasing tion) (Tedersoo et al., 2014; Cavicchioli et al., 2019), our understanding of fungal ecology in the vineyard and followed by edaphic factors (such as soil pH, aridity) their responses to the changing environment over the (Maestre et al., 2015; Bahram et al., 2018), drive soil fun- course of the growing season would help improve man- gal community assembly and hence multifunctionality in agement techniques to maintain healthy vines, produce ecosystems (Bahram et al., 2018; Cavicchioli et al., quality grapes and allow expression of wine regionality. 2019; Li et al., 2019). Our recent study revealed that wine-related fungal communities structured and distinguished vineyard ecosystems by impacting the flavour and quality of wine, and weather (in particular solar radia- Experimental procedures tion and temperature) strongly affected soil and must fun- Sample sites gal communities and the resultant wines across six winegrowing regions in southern Australia (Liu et al., Grapevine samples were collected at two vineyards in 2020). Vine water availability can condition the grape and Mornington Peninsula wine region during the 2018 vin- must microbiota at regional scales, for example, precipi- tage (Supporting Information Fig. S2; grapevine images tation and humidity correlate with the presence of fila- were created with BioRender, https://app.biorender.com/). mentous fungi (e.g. Penicillium) (Bokulich et al., 2014) Vineyard A (1.4 ha, altitude 55 m) and B (1.7 ha, altitude and yeasts (such as Saccharomyces, Hanseniaspora, 195 m) are 5 km apart, planted with Vitis vinifera cv. Pinot and Metschnikowia) (Jara et al., 2016). In the present Noir (clone MV6). Both vineyards are commercially oper- study, microbial heterogeneity and seasonal dynamics ated by a single winery under very similar viticultural across habitats highlight that water availability may influ- practices. Grape marc was spread at 2 t per hectare for ence the microbiota present within vineyards. Grapevine both vineyards in August 2017. Grapevines were under phenology is shifting with warming and drying induced vertical shoot positioning trellising systems and were by climate change (Webb et al., 2012; Cameron treated with fungicides of sulfur (Microthio Disperse), et al., 2020), and our work suggests that the grapevine- copper hydroxide (Kocide), metalaxyl + copper hydroxide associated microbiota will change as well, thus poten- (Ridomil), and myclobutanil (Mycloss) for the protection tially affect the distinctiveness of regional wines (Liu of powdery mildew and downy mildew, from shoot growth et al., 2020). Microbial assemblages responding to cli- (October 2017) until pre-veraison (January 2018) (see mate change may also affect plant physiology, as Supporting Information Table S5). In each vineyard, five reported for rhizosphere microorganisms which influ- replicate vines were selected from the top, middle, and ence plant phenology and alter the flowering time bottom of the dominant slope covering topological (Wagner et al., 2014; Panke-Buisse et al., 2015; Lu profiles of the vineyard, with pairwise distances ranging et al., 2018). Interactions between grapevines and asso- from 15 m to 135 m (Supporting Information Fig. S2A). ciated fungi deserve future study and would provide fur- Soil types in both vineyards are clay loam (Isbell, 2016). ther perspectives on vineyard establishment and Soil chemical analyses (pH, carbon, nitrogen) did not management practices such as managing soil moisture show significant differences between vineyards at and crop yield (Webb et al., 2012; Palliotti et al., 2014). harvest (Supporting Information Table S6). GPS coordi- As agricultural practices adapt to a changing climate, nates were recorded for each site and utilized to extract understanding the core microbiome, transient taxa weekly weather data provided by Australian Water Avail- and specific interactions between plant and microbe ability Project (AWAP) (Jones et al., 2009) from two become more important to maintaining plant productivity weather stations. Variables were observed by robust (Cavicchioli et al., 2019). topography resolving analysis methods at a resolution of 0.05 × 0.05 (approximately 5 km × 5km).Weekly measurements were extracted for precipitation (m), Conclusions − solar radiation (MJ m 2), total evaporation (m) (soil + Our study describes the core microbiome of grapevines vegetation), transpiration (m), mean high temperature and illustrates the seasonal community succession driven (C), mean low temperature (C), mean temperature by the core microbiome. Plants recruit their microorgan- (C), and mean relative soil moisture for each develop- isms based on habitats at different stages of development stage during the growing season (October 2017 to ment yet conserve core fungal species. While undertaken April 2018).

84 12 D. Liu and K. Howell Sample collection pool was cleaned a final time using magnetic beads to concentrate the pool and then measured using a High- In each grapevine, four different grapevine habitats were Sensitivity D1000 Tape on an Agilent 2200 TapeStation. studied: root zone soil, roots, leaves, and grapes or The pool was diluted to 5 nM and molarity was confirmed flowers (season dependent) (Supporting Information again using a High-Sensitivity D1000 Tape. This was Fig. S2B). Samples were aseptically collected from the followed by 300 bp paired-end sequencing on an Illumina same vines at four developmental stages [as defined by MiSeq (San Diego, CA, USA). Coombe (1992)] at flowering (November 2017), fruit-set Raw sequences were processed using QIIME v1.9.2 (young berries enlarging, December 2017), veraison (Caporaso et al., 2010). Low-quality regions (Q < 20) (berry colour change, January to February 2018), and were trimmed from the 50 end of the sequences, and the harvest (berries ripe, March 2018) from both vineyards. paired ends were joined using FLASH (Magocˇ and Root zone soil was taken from the surrounding roots at Salzberg, 2011). Primers were trimmed and a further the depth of 0–15 cm. Three sub-samples (the sun- round of quality control was conducted to discard full- exposed side, the shade side, and the middle) were length duplicate sequences, short sequences (<100 nt), mixed to form a composite sample. Healthy leaves of and sequences with ambiguous bases. Sequences were similar sizes, with no discernible evidence of disease clustered followed by chimera checking using UCHIME were collected. Undamaged grapes were chosen from algorithm from USEARCH v7.1.1090 (Edgar et al., 2011). different grape bunches. Samples were immediately Operational taxonomic units (OTUs) were assigned using stored in sterile bags and transported to the laboratory on UCLUST open-reference OTU-picking workflow with a ice. Roots were collected and rinsed and cleaned of all threshold of 97% pairwise identity (Edgar, 2010). Single- visible soil using distilled water and flame-sterilized tons or unique reads in the resultant data set were dis- utensils back in the laboratory. Plant materials were flash carded. Taxonomy was assigned to OTUs in QIIME frozen in liquid nitrogen and all the samples were stored using the UNITE fungal ITS database (v7.2) (Kõljalg at −80C until processing. A total of 160 vineyard et al., 2005) and BLAST classifier. To avoid/reduce samples were collected and processed. biases generated by varying sequencing depth, sequences were rarefied to 10 000 per sample (the lowest sequencing depth) prior to downstream analysis. Raw DNA extraction and sequencing sequencing reads have been deposited at the National Genomic DNA was extracted from soil and botanical Centre for Biotechnology Information Sequence Read samples using PowerSoil™ DNA Isolation kits (QIAgen, Archive under the bioproject PRJNA626608. CA, USA). DNA isolation of soil (0.25 g) samples followed the kit protocol. For plant parts, roots, leaves, grapes Data analysis (removed seeds and stems) or flowers, were ground into powder in liquid nitrogen with 1% polyvinylpolypyrrolidone Fungal OTUs shared among habitats in the vineyard and then DNA isolated following the kit protocol. Thus, were illustrated by the ‘VennDiagram’ package (Chen endophytes and epiphytes were extracted, sequenced, and Boutros, 2011) in R (v3.5.0). Microbial alpha-diversity and analysed simultaneously from all plant organs. DNA was calculated using the Shannon index with the ‘vegan’ extracts were stored at −20C until further analysis. package (Oksanen et al., 2007). One-way analysis of Genomic DNA was submitted to Australian Genome variance (ANOVA) was used to determine whether sam- Research Facility (AGRF) for amplification and sequenc- ple classifications (e.g., habitat, developmental stage, ing. To assess the fungal communities, partial fungal vineyard) contained statistically significant differences in internal transcribed spacer (ITS) region was amplified the diversity. Principal coordinate analysis (PCoA) was using the universal primer pairs ITS1F/2 (Gardes and performed to evaluate the distribution patterns of Bruns, 1993). The primary PCR reactions contained grapevine-associated microbiome based on beta- 10 ng DNA template, 2× AmpliTaq Gold® 360 Master diversity calculated by the Bray–Curtis distance with the Mix (Life Technologies, Australia), 5 pmol of each primer. ‘labdsv’ package (Roberts, 2016). Permutational multivar- A secondary PCR to index the amplicons was performed iate analysis of variance (PERMANOVA) using distance with TaKaRa Taq DNA Polymerase (Clontech). Amplifi- matrices with 999 permutations was conducted within cation was conducted under the following conditions: each sample category to determine the statistically signif- 95C for 7 min, followed by 35 cycles of 94C for 30 s, icant differences with ‘adonis” function in ‘vegan’ 55C for 45 s, 72C for 60 s and a final extension at 72C (Oksanen et al., 2007). Significant taxonomic differences for 7 min. The resulting amplicons were cleaned of fungi between vineyard habitats were tested using again using magnetic beads, quantified by fluorometry linear discriminant analysis (LDA) effect size (LEfSe) (Promega Quantifluor) and normalized. The equimolar analysis (Segata et al., 2011) (https://huttenhower.sph.

85 The core fungal microbiome of grapevines 13 harvard.edu/galaxy/). The OTU table was filtered to Acknowledgements include only OTUs >0.1% relative abundance to reduce The authors give their sincere thanks to the vignerons in – LEfSe complexity. The factorial Kruskal Wallis sum-rank Mornington Peninsula who kindly allowed vineyard access test (α = 0.05) was applied to identify taxa with significant and enabled sampling. D.L. acknowledges support from a differential abundances between classes (all-against-all Ph.D. scholarship and funding from Wine Australia (AGW comparisons), followed by the logarithmic LDA score Ph1602) and a Melbourne Research Scholarship from the (threshold = 2.0) to estimate the effect size of each dis- University of Melbourne. Dr. Qinling Chen, Prof. Deli Chen criminative feature. Significant taxa were used to gener- and Dr. Pangzhen Zhang are acknowledged for helpful dis- cussions and support in the course of this work. ate taxonomic cladograms illustrating differences between sample categories. Thecoremicrobiomewasdefined based on abundance- References occupancy distributions that include highly abundant and ubiquitous taxa (Delgado-Baquerizo et al., 2018; Shade Anjum, S.A., Xie, X.-y., Wang, L.-c., Saleem, M.F., Man, C., and Stopnisek, 2019). For soil core microbiome, OTUs and Lei, W. (2011) Morphological, physiological and bio- were filtered out with top 1% mean relative abundance chemical responses of plants to drought stress. African J Agric Res 6: 2026–2032. (MRA) across all samples (Galand et al., 2009) and an Bahram, M., Hildebrand, F., Forslund, S.K., Anderson, J.L., occurrence frequency found in more than 50% of samples Soudzilovskaia, N.A., Bodegom, P.M., et al. (2018) Struc- or 100% occupancy at any time point (Delgado-Baquerizo ture and function of the global topsoil microbiome. Nature et al., 2018; Grady et al., 2019). For the grapevine, OTUs 560: 233–237. with top 10% MRA across all samples (roots, leaves, Barata, A., Malfeito-Ferreira, M., and Loureiro, V. (2012) The grapes, flowers) and 50% occupancy of all samples or microbial ecology of wine grape berries. Int J Food – 100% occupancy of one habitat samples were defined as Microbiol 153: 243 259. Baudoin, E., Benizri, E., and Guckert, A. (2002) Impact of the core members (Delgado-Baquerizo et al., 2018; Grady growth stage on the bacterial community structure along fl et al., 2019). For each plant organ (except owers), the maize roots, as determined by metabolic and genetic fin- core OTUs were those top 10% abundant across all sam- gerprinting. Appl Soil Ecol 19: 135–145. ples with 50% occupancy of all samples or 100% occu- Berg, G., and Smalla, K. (2009) Plant species and soil type pancy at any developmental stage (Delgado-Baquerizo cooperatively shape the structure and function of microbial et al., 2018; Grady et al., 2019). Dynamics and successions communities in the rhizosphere. FEMS Microbiol Ecol – of the core microbiome were illustrated by alluvial diagrams 68:1 13. Bokulich, N.A., Thorngate, J.H., Richardson, P.M., and using the ‘ggalluvial’ package (Brunson, 2018) in ‘ggplot2’. Mills, D.A. (2014) Microbial biogeography of wine grapes fi Random forest supervised classi cation models is conditioned by cultivar, vintage, and climate. Proc Natl (Breiman, 2001) were used to evaluate the discriminative Acad Sci 111: E139–E148. power of fungal microbiota (all OTUs) to distinguish Breiman, L. (2001) Random forests. Mach Learn 45:5–32. developmental stages and the robustness of the group- Broeckling, C.D., Broz, A.K., Bergelson, J., Manter, D.K., ings themselves. The importance of each predictor is and Vivanco, J.M. (2008) Root exudates regulate soil fun- measured by the mean decrease in Gini coefficient gal community composition and diversity. Appl Environ Microbiol 74: 738–744. (MDG) (i.e. how each variable contributes to the homoge- Brunson, J.C. (2018) Ggalluvial: Alluvial diagrams in ggplot2. neity of the nodes and leaves in the resulting random for- R package version 0.9.1. https://CRAN.R-project.org/ est, and variables resulting in nodes with higher purity package=ggalluvial have a higher MDG). Models were constructed with Burns, K.N., Kluepfel, D.A., Strauss, S.L., Bokulich, N.A., 10 000 trees, with OTUs as predictors and developmen- Cantu, D., and Steenwerth, K.L. (2015) Vineyard soil bac- tal stages as class labels using the ‘randomForest’ pack- terial diversity and composition revealed by 16S rRNA age (Liaw and Wiener, 2002). genes: differentiation by geographic features. Soil Biol Biochem 91: 232–247. The co-occurrence/interaction patterns between the Cameron, W., Petrie, P.R., Barlow, E.W.R., Patrick, C.J., core microbiome associated with grapevine habitats Howell, K., and Fuentes, S. (2020) Advancement of grape (grapes, leaves, roots) and soil, and weather conditions maturity: comparison between contrasting cultivars and during the growing season was explored using network regions. Austral J Grape Wine Res 26:53–67. analysis with Cytoscape (Shannon et al., 2003). Correla- Caporaso, J.G., Kuczynski, J., Stombaugh, J., Bittinger, K., tion matrices were calculated with all possible pair-wise Bushman, F.D., Costello, E.K., et al. (2010) QIIME allows Spearman’s rank correlations between selected OTUs of analysis of high-throughput community sequencing data. Nat Methods 7: 335–336. plant and soil, or OTUs and weather indexes. Correla- Cavicchioli, R., Ripple, W.J., Timmis, K.N., Azam, F., fi ρ ≥ tions with a Spearman correlation coef cient 0.8 and Bakken, L.R., Baylis, M., et al. (2019) Scientists’ warning a P < 0.01 were considered statistically robust and dis- to humanity: microorganisms and climate change. Nat played in the networks (Junker and Schreiber, 2008). Rev Microbiol 17: 569–586.

86 14 D. Liu and K. Howell

Chaparro, J.M., Badri, D.V., and Vivanco, J.M. (2014) Rhizo- identification of mycorrhizae and rusts. Mol Ecol 2: sphere microbiome assemblage is affected by plant devel- 113–118. opment. ISME J 8: 790–803. Gayevskiy, V., and Goddard, M.R. (2012) Geographic delin- Chaparro, J.M., Badri, D.V., Bakker, M.G., Sugiyama, A., eations of yeast communities and populations associated Manter, D.K., and Vivanco, J.M. (2013) Root exudation of with vines and wines in New Zealand. ISME J 6: phytochemicals in Arabidopsis follows specific patterns 1281–1290. that are developmentally programmed and correlate with Gilbert, J.A., van der Lelie, D., and Zarraonaindia, I. (2014) soil microbial functions. PLoS One 8: e55731. Microbial terroir for wine grapes. Proc Natl Acad Sci Chen, H., and Boutros, P.C. (2011) VennDiagram: a pack- 111:5–6. age for the generation of highly-customizable Venn and Goddard, M.R., Anfang, N., Tang, R., Gardner, R.C., and Euler diagrams in R. BMC Bioinformatics 12: 35. Jun, C. (2010) A distinct population of Saccharomyces Chesson, P. (2000) Mechanisms of maintenance of species cerevisiae in New Zealand: evidence for local dispersal by diversity. Ann Rev Ecol Syst 31: 343–366. insects and human-aided global dispersal in oak barrels. Ciani, M., Comitini, F., Mannazzu, I., and Domizio, P. (2010) Environ Microbiol 12:63–73. Controlled mixed culture fermentation: a new perspective Grady, K.L., Sorensen, J.W., Stopnisek, N., Guittar, J., and on the use of non-saccharomyces yeasts in winemaking. Shade, A. (2019) Assembly and seasonality of core FEMS Yeast Res 10: 123–133. phyllosphere microbiota on perennial biofuel crops. Nature Coller, E., Cestaro, A., Zanzotti, R., Bertoldi, D., Pindo, M., Communications 10: 4135. Larger, S., et al. (2019) Microbiome of vineyard soils is Hazard, C., Gosling, P., Van Der Gast, C.J., Mitchell, D.T., shaped by geography and management. Microbiome Doohan, F.M., and Bending, G.D. (2013) The role of local 7: 140. environment and geographical distance in determining Comitini, F., and Ciani, M. (2008) Influence of fungicide treat- community composition of arbuscular mycorrhizal fungi at ments on the occurrence of yeast flora associated with the landscape scale. ISME J 7: 498–508. wine grapes. Ann Microbiol 58: 489–493. Hochberg, U., Degu, A., Toubiana, D., Gendler, T., Compant, S., Mitter, B., Colli-Mull, J.G., Gangl, H., and Nikoloski, Z., Rachmilevitch, S., and Fait, A. (2013) Metab- Sessitsch, A. (2011) Endophytes of grapevine flowers, olite profiling and network analysis reveal coordinated berries, and seeds: identification of cultivable bacteria, changes in grapevine water stress response. BMC Plant comparison with other plant parts, and visualization of Biol 13: 184. niches of colonization. Microb Ecol 62: 188–197. Houlden, A., Timms-Wilson, T.M., Day, M.J., and Bailey, M. Coombe, B. (1992) Research on development and ripening J. (2008) Influence of plant developmental stage on micro- of the grape berry. Am J Enol Vitic 43: 101–110. bial community structure and activity in the rhizosphere of Copeland, J.K., Yuan, L., Layeghifard, M., Wang, P.W., and three field crops. FEMS Microbiol Ecol 65: 193–201. Guttman, D.S. (2015) Seasonal community succession of Isbell, R. (2016) The Australian soil classification, 2nd edn. the phyllosphere microbiome. Mol Plant Microbe Interact Clayton South, Australia: CSIRO publishing. 28: 274–285. Jara, C., Laurie, V.F., Mas, A., and Romero, J. (2016) Micro- Delgado-Baquerizo, M., Oliverio, A.M., Brewer, T.E., bial Terroir in Chilean valleys: diversity of non- Benavent-González, A., Eldridge, D.J., Bardgett, R.D., conventional yeast. Front Microbiol 7: 663. et al. (2018) A global atlas of the dominant bacteria found Jones, D.A., Wang, W., and Fawcett, R. (2009) High-quality in soil. Science 359: 320–325. spatial climate data-sets for Australia. Australian Meteoro- Deyett, E., and Rolshausen, P.E. (2020) Endophytic micro- logical and Oceanographic Journal 58: 233. bial assemblage in grapevine. FEMS Microbiol Ecol 96: Jumpponen, A., and Jones, K. (2010) Seasonally dynamic fiaa053. fungal communities in the Quercus macrocarpa Edgar, R.C. (2010) Search and clustering orders of magni- phyllosphere differ between urban and nonurban environ- tude faster than BLAST. Bioinformatics 26: 2460–2461. ments. New Phytol 186: 496–513. Edgar, R.C., Haas, B.J., Clemente, J.C., Quince, C., and Junker, B.H., and Schreiber, F. (2008) Analysis of Biological Knight, R. (2011) UCHIME improves sensitivity and speed Networks, Hoboken, USA: Wiley-Interscience. of chimera detection. Bioinformatics 27: 2194–2200. Kivlin, S.N., Hawkes, C.V., and Treseder, K.K. (2011) Global Egidi, E., Delgado-Baquerizo, M., Plett, J.M., Wang, J., diversity and distribution of arbuscular mycorrhizal fungi. Eldridge, D.J., Bardgett, R.D., et al. (2019) A few Soil Biol Biochem 43: 2294–2303. Ascomycota taxa dominate soil fungal communities world- Knight, S., and Goddard, M. (2015) Quantifying separation wide. Nat Commun 10:1–9. and similarity in a Saccharomyces cerevisiae Fleet, G., Prakitchaiwattana, C., Beh, A., and Heard, G. metapopulation. ISME J 9: 361–370. (2002) The yeast ecology of wine grapes. In Biodiversity Knight, S.J., Karon, O., and Goddard, M.R. (2019) Small and Biotechnology of Wine Yeasts, Vol. 95. Kerala, India: scale fungal community differentiation in a vineyard sys- Research Signpost, pp. 1–17. tem. Food Microbiol 87: 103358. Galand, P.E., Casamayor, E.O., Kirchman, D.L., and Kõljalg, U., Larsson, K.H., Abarenkov, K., Nilsson, R.H., Lovejoy, C. (2009) Ecology of the rare microbial biosphere Alexander, I.J., Eberhardt, U., et al. (2005) UNITE: a data- of the Arctic Ocean. Proc Natl Acad Sci 106: 22427–22432. base providing web-based methods for the molecular Gardes, M., and Bruns, T.D. (1993) ITS primers with identification of ectomycorrhizal fungi. New Phytol 166: enhanced specificity for basidiomycetes-application to the 1063–1068.

87 The core fungal microbiome of grapevines 15

Lam, S.S., and Howell, K.S. (2015) Drosophila-associated progression of bacterial community succession in the Ara- yeast species in vineyard ecosystems. FEMS Microbiol bidopsis rhizosphere. Plant Signal Behav 4: 777–780. Lett 362: fnv170. Morrison-Whittle, P., and Goddard, M. (2017) Fungal com- Leibold, M.A., and McPeek, M.A. (2006) Coexistence of the munities are differentially affected by conventional and niche and neutral perspectives in community ecology. biodynamic agricultural management approaches in vine- Ecology 87: 1399–1410. yard ecosystems. Agric Ecosyst Environ 246: 306–313. Lekberg, Y., Schnoor, T., Kjøller, R., Gibbons, S.M., Morrison-Whittle, P., and Goddard, M.R. (2018) From vine- Hansen, L.H., Al-Soud, W.A., et al. (2012) yard to winery: a source map of microbial diversity driving 454-sequencing reveals stochastic local reassembly and wine fermentation. Environ Microbiol 20:75–84. high disturbance tolerance within arbuscular mycorrhizal Mougel, C., Offre, P., Ranjard, L., Corberand, T., fungal communities. J Ecol 100: 151–160. Gamalero, E., Robin, C., and Lemanceau, P. (2006) Li, J., Delgado-Baquerizo, M., Wang, J.-T., Hu, H.-W., Dynamic of the genetic structure of bacterial and fungal Cai, Z.-J., Zhu, Y.-N., and Singh, B.K. (2019) Fungal rich- communities at different developmental stages of Medi- ness contributes to multifunctionality in boreal forest soil. cago truncatula Gaertn. cv. Jemalong line J5. New Phytol Soil Biol Biochem 136: 107526. 170: 165–175. Liaw, A., and Wiener, M. (2002) Classification and regres- Müller, D.B., Vogel, C., Bai, Y., and Vorholt, J.A. (2016) The sion by randomForest. R news 2:18–22. plant microbiota: systems-level insights and perspectives. Liu, D., Zhang, P., Chen, D., and Howell, K.S. (2019) From Annu Rev Genet 50: 211–234. the vineyard to the winery: how microbial ecology drives Oksanen, J., Kindt, R., Legendre, P., O’Hara, B., regional distinctiveness of wine. Front Microbiol 10: 2679. Stevens, M.H.H., Oksanen, M.J., and Suggests, M. (2007) Lu, T., Ke, M., Lavoie, M., Jin, Y., Fan, X., Zhang, Z., et al. The vegan package. Community ecology package 10: (2018) Rhizosphere microorganisms can influence the 631–637. timing of plant flowering. Microbiome 6: 231. Osorio, N., and Habte, M. (2001) Synergistic influence of an Liu, D., Chen, Q., Zhang, P., Chen, D., & Howell, K.S. arbuscular mycorrhizal fungus and a P solubilizing fungus (2020). The fungal microbiome is an important component on growth and P uptake of Leucaena leucocephala in an of vineyard ecosystems and correlates with regional dis- Oxisol. Arid Land Res Manag 15: 263–274. tinctiveness of wine. mSphere, 5: e00534-20. Palliotti, A., Tombesi, S., Silvestroni, O., Lanari, V., Gatti, M., Lundberg, D.S., Lebeis, S.L., Paredes, S.H., Yourstone, S., and Poni, S. (2014) Changes in vineyard establishment Gehring, J., Malfatti, S., et al. (2012) Defining the core and canopy management urged by earlier climate-related Arabidopsis thaliana root microbiome. Nature 488: grape ripening: a review. Sci Hortic 178:43–54. 86–90. Panke-Buisse, K., Poole, A.C., Goodrich, J.K., Ley, R.E., Lynch, M.D., and Neufeld, J.D. (2015) Ecology and explora- and Kao-Kniffin, J. (2015) Selection on soil microbiomes tion of the rare biosphere. Nat Rev Microbiol 13: 217–229. reveals reproducible impacts on plant function. ISME J 9: Maestre, F.T., Delgado-Baquerizo, M., Jeffries, T.C., 980–989. Eldridge, D.J., Ochoa, V., Gozalo, B., et al. (2015) Pinto, C., Pinho, D., Sousa, S., Pinheiro, M., Egas, C., and Increasing aridity reduces soil microbial diversity and Gomes, A.C. (2014) Unravelling the diversity of grapevine abundance in global drylands. Proc Natl Acad Sci 112: microbiome. PLoS One 9: e85622. 15684–15689. Portillo, M.D.C., Franquès, J., Araque, I., Reguant, C., and Magocˇ, T., and Salzberg, S.L. (2011) FLASH: fast length Bordons, A. (2016) Bacterial diversity of Grenache and adjustment of short reads to improve genome assemblies. Carignan grape surface from different vineyards at Priorat Bioinformatics 27: 2957–2963. wine region (Catalonia, Spain). Int J Food Microbiol 219: Mannazzu, I., Clementi, F., and Ciani, M. (2002) Strategies 56–63. and criteria for the isolation and selection of autochtho- Renouf, V., Claisse, O., and Lonvaud-Funel, A. (2005) nous starters. In Biodiversity and Biotechnology of Wine Understanding the microbial ecosystem on the grape Yeasts, pp. 19–34. Kerala, India: Research Signpost. berry surface through numeration and identification of Marasco, R., Rolli, E., Fusi, M., Michoud, G., and yeast and bacteria. Austral J Grape Wine Res 11: Daffonchio, D. (2018) Grapevine rootstocks shape under- 316–327. ground bacterial microbiome and networking but not Roberts, D.W. (2016) labdsv: Ordination and multivariate potential functionality. Microbiome 6:3. analysis for ecology. R package version 1. 8-0. https:// Martínez-Diz, M.d.P., Andrés-Sodupe, M., Bujanda, R., CRAN.R-project.org/package=labdsv. Díaz-Losada, E., Eichmeier, A., and Gramaje, D. (2019) Segata, N., Izard, J., Waldron, L., Gevers, D., Miropolsky, L., Soil-plant compartments affect fungal microbiome diversity Garrett, W.S., and Huttenhower, C. (2011) Metagenomic and composition in grapevine. Fungal Ecology 41: biomarker discovery and explanation. Genome Biol 234–244. 12: R60. Martins, G., Vallance, J., Mercier, A., Albertin, W., Schmid, F., Moser, G., Müller, H., and Berg, G. (2011) Func- Stamatopoulos, P., Rey, P., et al. (2014) Influence of the tional and structural microbial diversity in organic and con- farming system on the epiphytic yeasts and yeast-like ventional viticulture: organic farming benefits natural fungi colonizing grape berries during the ripening process. biocontrol agents. Appl Environ Microbiol 77: 2188–2191. Int J Food Microbiol 177:21–28. Setati, M.E., Jacobson, D., Andong, U.-C., and Bauer, F. Micallef, S.A., Channer, S., Shiaris, M.P., and Colón- (2012) The vineyard yeast microbiome, a mixed model Carmona, A. (2009) Plant age and genotype impact the microbial map. PLoS One 7: e52609.

88 16 D. Liu and K. Howell

Shade, A., and Gilbert, J.A. (2015) Temporal patterns of rar- White, R.E. (2015) Understanding vineyard soils, 2nd edn. ity provide a more complete view of microbial diversity. New York, USA: Oxford University Press. Trends Microbiol 23: 335–340. Zarraonaindia, I., Owens, S.M., Weisenhorn, P., West, K., Shade, A., and Stopnisek, N. (2019) Abundance-occupancy Hampton-Marcell, J., Lax, S., et al. (2015) The soil micro- distributions to prioritize plant core microbiome member- biome influences grapevine-associated microbiota. MBio ship. Curr Opin Microbiol 49:50–58. 6: e02527–e02514. Shade, A., Jones, S.E., Caporaso, J.G., Handelsman, J., Zhalnina, K., Louie, K.B., Hao, Z., Mansoori, N., da Knight, R., Fierer, N., and Gilbert, J.A. (2014) Condition- Rocha, U.N., Shi, S., et al. (2018) Dynamic root exudate ally rare taxa disproportionately contribute to temporal chemistry and microbial substrate preferences drive pat- changes in microbial diversity. MBio 5: e01371–e01314. terns in rhizosphere microbial community assembly. Nat Shannon, P., Markiel, A., Ozier, O., Baliga, N.S., Wang, J.T., Microbiol 3: 470–480. Ramage, D., et al. (2003) Cytoscape: a software environment for integrated models of biomolecular interaction networks. Genome Res 13: 2498–2504. Supporting Information Smits, S.A., Leach, J., Sonnenburg, E.D., Gonzalez, C.G., Additional Supporting Information may be found in the online Lichtman, J.S., Reid, G., et al. (2017) Seasonal cycling in version of this article at the publisher’s web-site: the gut microbiome of the Hadza hunter-gatherers of Tan- zania. Science 357: 802–806. Fig. S1. Co-occurrence pattern between the soil core micro- Stefanini, I., and Cavalieri, D. (2018) Metagenomic biome and weather parameters. Circle nodes represent soil approaches to investigate the contribution of the vineyard core OTUs and environmental variables, with different col- environment to the quality of wine fermentation: potentials ours. Direct connections between nodes indicate strong cor- and difficulties. Front Microbiol 9: 991. relations (Spearman correlation coefficient, ρ ≥ 0.8; Stefanini, I., Dapporto, L., Legras, J.-L., Calabretta, A., Di p < 0.01). The colour of edges describes the positive correla- Paola, M., De Filippo, C., et al. (2012) Role of social tion (pink) or the negative correlation (grey). The size of wasps in Saccharomyces cerevisiae ecology and evolu- nodes is proportioned to the interconnected degree. The tion. Proc Natl Acad Sci 109: 13398–13403. width of edges is proportioned to the correlation coefficients. Swiegers, J., Bartowsky, E., Henschke, P., and Pretorius, I. Abbreviations: Env1, solar radiation; Env2, mean high tem- (2005) Yeast and bacterial modulation of wine aroma and perature; Env3, mean low temperature; Env4, mean temper- flavour. Austral J Grape Wine Res 11: 139–173. ature; Env5, precipitation; Env6, evaporation; Env7, Taylor, M.W., Tsai, P., Anfang, N., Ross, H.A., and transpiration; Env8, relative soil moisture. Goddard, M.R. (2014) Pyrosequencing reveals regional Fig. S2. Experimental design and sampling scheme. differences in fruit-associated fungal communities. Environ (A) Sampling map of two vineyards in Mornington Peninsula – Microbiol 16: 2848 2858. wine region, Australia. Five Vitis vinifera Pinot Noir vines Tedersoo, L., Bahram, M., Põlme, S., Kõljalg, U., Yorou, N. were selected in each vineyard. (B) Soils, roots, leaves, and S., Wijesundera, R., et al. (2014) Global diversity and grapes/ flowers were collected in each grapevine throughout geography of soil fungi. Science 346: 1256688. the annual growth cycle: flowering, fruit set, veraison (berry Toju, H., Peay, K.G., Yamamichi, M., Narisawa, K., colour change), and harvest. Calendar dates of sample col- Hiruma, K., Naito, K., et al. (2018) Core microbiomes for lection are indicated for each developmental stage. Grape- – sustainable agroecosystems. Nat Plants 4: 247 257. vines were created with BioRender: https://app.biorender. Turner, T.R., James, E.K., and Poole, P.S. (2013) The plant com/. microbiome. Genome Biol 14: 209. Table S1. OTUs identified as core community associated Van Leeuwen, C., and Seguin, G. (2006) The concept of with the grapevine or within each vineyard habitat terroir in viticulture. J Wine Res 17:1–10. Table S2. Classification of random forest models of the Wagner, M.R., Lundberg, D.S., Coleman-Derr, D., Tringe, S. development stage G., Dangl, J.L., and Mitchell-Olds, T. (2014) Natural soil microbes alter flowering phenology and the intensity of Table S3. Important features of random forest models out- selection on flowering time in a wild Arabidopsis relative. side the core community Ecol Lett 17: 717–726. Table S4. ANOVA results of weather conditions between Webb, L., Whetton, P., Bhend, J., Darbyshire, R., Briggs, P., developmental stages and Barlow, E. (2012) Earlier wine-grape ripening driven Table S5. Spray programme for the studied vineyards from by climatic warming and drying and management prac- shoot growth until pre-veraison tices. Nat Clim Change 2: 259–264. Table S6. Soil properties of each sampling site at harvest

89 4.2 Supplementary data

Table S1 OTUs identified as core community associated with the grapevine or within each vineyard habitat The first column shows the OTU id and the second column shows the lowest taxonomy level the OTU could be classified to. The third to the last column show the mean relative abundance (outside the parentheses) and occupancy of samples (inside the parentheses) across plant organs(A) or within each vineyard habitat(A1 - A4, grapevine; B soil). Core OTUs are coloured for each plant organ, of which bold ones are core OTUs that are specific to this organ. For example, core OTUs of grapes are those filled with red, of which OTU_20, OTU_74, OTU_64 are grape-specific core OTUs.

(A) Grapevine ID Taxonomy Overall Flower Grape Leaf Root k__Fungi; p__Ascomycota; c__Dothideomycetes; o__Dothideales; f__Dothioraceae; g__Aureobasidium; OTU_1 s__Aureobasidium_pullulans 24.37% (98.33%) 25.00% (100.00%) 28.29% (100.00%) 19.63% (95.00%) 25.80% (100.00%) k__Fungi; p__Ascomycota; c__Dothideomycetes; o__Capnodiales; f__Cladosporiaceae; g__Cladosporium; OTU_2 s__Cladosporium_ramotenellum 12.75% (98.33%) 11.59% (100.00%) 19.00% (100.00%) 12.06% (95.00%) 8.92% (100.00%) k__Fungi; p__Basidiomycota; c__Tremellomycetes; o__Tremellales; f__Tremellaceae; g__Cryptococcus; OTU_3 s__Cryptococcus_saitoi 8.61% (92.50%) 2.31% (100.00%) 7.75% (96.67%) 10.15% (82.50%) 9.38% (97.50%) k__Fungi; p__Ascomycota; c__Dothideomycetes; o__Capnodiales; f__Mycosphaerellaceae; g__Mycosphaerella; OTU_6 s__Mycosphaerella_tassiana 5.51% (90.83%) 1.94% (100.00%) 4.40% (96.67%) 8.10% (82.50%) 4.76% (92.50%) k__Fungi; p__Ascomycota; c__Saccharomycetes; o__Saccharomycetales; OTU_5 f__Saccharomycetales_fam_Incertae_sedis; g__Debaryomyces; s__Debaryomyces_hansenii 5.34% (91.67%) 4.30% (100.00%) 4.09% (93.33%) 9.40% (87.50%) 2.60% (92.50%) k__Fungi; p__Ascomycota; c__Saccharomycetes; o__Saccharomycetales; f__Saccharomycetaceae; OTU_9 g__Saccharomyces; s__Saccharomyces_cerevisiae 4.55% (94.17%) 8.96% (100.00%) 2.29% (100.00%) 9.08% (92.50%) 0.74% (90.00%) k__Fungi; p__Ascomycota; c__Dothideomycetes; o__Pleosporales; f__Pleosporaceae; g__Epicoccum; OTU_1093 s__Epicoccum_nigrum 3.28% (76.67%) 5.95% (90.00%) 6.06% (83.33%) 1.77% (55.00%) 1.91% (90.00%) k__Fungi; p__Ascomycota; c__Dothideomycetes; o__Pleosporales; f__Pleosporaceae; g__Epicoccum; OTU_15 s__Epicoccum_nigrum 2.67% (73.33%) 3.06% (90.00%) 6.35% (86.67%) 0.91% (47.50%) 1.44% (85.00%) k__Fungi; p__Ascomycota; c__Dothideomycetes; o__Pleosporales; f__Pleosporaceae; g__Alternaria; OTU_8 s__Alternaria_betae-kenyensis 1.84% (59.17%) 0.47% (70.00%) 4.47% (73.33%) 0.43% (20.00%) 1.54% (85.00%) k__Fungi; p__Ascomycota; c__Dothideomycetes; o__Capnodiales; f__Cladosporiaceae; g__Cladosporium; OTU_1808 s__Cladosporium_delicatulum 1.63% (51.67%) 0.50% (90.00%) 0.29% (53.33%) 4.13% (52.50%) 0.51% (40.00%) k__Fungi; p__Ascomycota; c__Dothideomycetes; o__Pleosporales; f__Pleosporales_fam_Incertae_sedis; OTU_16 g__Didymella; s__Didymella_negriana 1.24% (59.17%) 0.46% (70.00%) 3.10% (76.67%) 0.20% (25.00%) 1.03% (77.50%) k__Fungi; p__Ascomycota; c__Dothideomycetes; o__Capnodiales; f__Mycosphaerellaceae; g__Ramularia; OTU_26 s__Ramularia_pratensis 1.20% (53.33%) 0.19% (50.00%) 0.83% (73.33%) 0.57% (25.00%) 2.34% (67.50%) k__Fungi; p__Basidiomycota; c__Tremellomycetes; o__Tremellales; f__Bulleribasidiaceae; g__Vishniacozyma; OTU_38 s__Vishniacozyma_carnescens 0.90% (53.33%) 0.35% (60.00%) 1.44% (73.33%) 0.75% (30.00%) 0.76% (60.00%) k__Fungi; p__Ascomycota; c__Eurotiomycetes; o__Eurotiales; f__Trichocomaceae; g__Penicillium; OTU_47 s__Penicillium_halotolerans 0.65% (52.50%) 2.55% (90.00%) 0.13% (56.67%) 0.75% (32.50%) 0.45% (60.00%) k__Fungi; p__Ascomycota; c__Sordariomycetes; o__Hypocreales; f__Nectriaceae; g__Fusarium; OTU_2019 s__Fusarium_denticulatum 0.18% (90.83%) 0.42% (100.00%) 0.20% (96.67%) 0.08% (82.50%) 0.22% (92.50%) Core community (Total relative abundance) 74.70 ± 2.70% 68.04 ± 10.44% 88.70 ± 2.27% 78.01 ± 4.53% 62.41 ± 5.40%

(A1) Flower

ID Taxonomy Overall (Flowering) k__Fungi; p__Ascomycota; c__Dothideomycetes; o__Dothideales; f__Dothioraceae; g__Aureobasidium; OTU_1 s__Aureobasidium_pullulans 25.00% (100.00%) k__Fungi; p__Ascomycota; c__Dothideomycetes; o__Capnodiales; f__Cladosporiaceae; g__Cladosporium; OTU_2 s__Cladosporium_ramotenellum 11.59% (100.00%)

90 k__Fungi; p__Basidiomycota; c__Tremellomycetes; o__Tremellales; f__Tremellaceae; g__Cryptococcus; OTU_3 s__Cryptococcus_saitoi 2.31% (100.00%) k__Fungi; p__Ascomycota; c__Saccharomycetes; o__Saccharomycetales; OTU_5 f__Saccharomycetales_fam_Incertae_sedis; g__Debaryomyces; s__Debaryomyces_hansenii 4.30% (100.00%) k__Fungi; p__Ascomycota; c__Dothideomycetes; o__Capnodiales; f__Mycosphaerellaceae; g__Mycosphaerella; OTU_6 s__Mycosphaerella_tassiana 1.94% (100.00%) k__Fungi; p__Ascomycota; c__Dothideomycetes; o__Pleosporales; f__Pleosporaceae; g__Alternaria; OTU_8 s__Alternaria_betae-kenyensis 0.47% (70.00%) k__Fungi; p__Ascomycota; c__Saccharomycetes; o__Saccharomycetales; f__Saccharomycetaceae; OTU_9 g__Saccharomyces; s__Saccharomyces_cerevisiae 8.96% (100.00%) k__Fungi; p__Ascomycota; c__Dothideomycetes; o__Pleosporales; f__Pleosporaceae; g__Epicoccum; OTU_15 s__Epicoccum_nigrum 3.06% (90.00%) k__Fungi; p__Ascomycota; c__Dothideomycetes; o__Pleosporales; f__Pleosporales_fam_Incertae_sedis; OTU_16 g__Didymella; s__Didymella_negriana 0.46% (70.00%) k__Fungi; p__Ascomycota; c__Dothideomycetes; o__Capnodiales; f__Mycosphaerellaceae; g__Ramularia; OTU_26 s__Ramularia_pratensis 0.19% (50.00%) k__Fungi; p__Basidiomycota; c__Tremellomycetes; o__Tremellales; f__Bulleribasidiaceae; g__Vishniacozyma; OTU_38 s__Vishniacozyma_carnescens 0.35% (60.00%) k__Fungi; p__Ascomycota; c__Eurotiomycetes; o__Eurotiales; f__Trichocomaceae; g__Penicillium; OTU_47 s__Penicillium_halotolerans 2.55% (90.00%) k__Fungi; p__Ascomycota; c__Dothideomycetes; o__Capnodiales; f__Cladosporiaceae; g__Cladosporium; OTU_1808 s__Cladosporium_delicatulum 0.50% (90.00%) k__Fungi; p__Ascomycota; c__Dothideomycetes; o__Pleosporales; f__Pleosporaceae; g__Epicoccum; OTU_1093 s__Epicoccum_nigrum 5.95% (90.00%) k__Fungi; p__Ascomycota; c__Sordariomycetes; o__Hypocreales; f__Nectriaceae; g__Fusarium; OTU_2019 s__Fusarium_denticulatum 0.42% (100.00%) k__Fungi; p__Mortierellomycota; c__Mortierellomycotina_cls_Incertae_sedis; o__Mortierellales; OTU_4 f__Mortierellaceae; g__Mortierella; s__Mortierella_amoeboidea 1.00% (70.00%) k__Fungi; p__Mortierellomycota; c__Mortierellomycotina_cls_Incertae_sedis; o__Mortierellales; OTU_21 f__Mortierellaceae; g__Mortierella; s__Mortierella_elongata 0.88% (60.00%) OTU_113 k__Fungi; p__Ascomycota; c__Sordariomycetes; o__Microascales; f__Microascaceae 0.43% (60.00%) Core community (Total relative abundance) 69.82 ± 10.44%

(A2) Grape ID Taxonomy Overall Fruit set Veraison Harvest k__Fungi; p__Ascomycota; c__Dothideomycetes; o__Dothideales; f__Dothioraceae; g__Aureobasidium; OTU_1 s__Aureobasidium_pullulans 28.29% (100.00%) 41.15% (100.00%) 29.62% (100.00%) 14.09% (100.00%) k__Fungi; p__Ascomycota; c__Dothideomycetes; o__Capnodiales; f__Cladosporiaceae; g__Cladosporium; OTU_2 s__Cladosporium_ramotenellum 19.00% (100.00%) 11.07% (100.00%) 13.20% (100.00%) 32.74% (100.00%) k__Fungi; p__Basidiomycota; c__Tremellomycetes; o__Tremellales; f__Tremellaceae; g__Cryptococcus; OTU_3 s__Cryptococcus_saitoi 7.75% (96.67%) 12.78% (100.00%) 10.03% (100.00%) 0.43% (90.00%) k__Fungi; p__Ascomycota; c__Dothideomycetes; o__Capnodiales; f__Mycosphaerellaceae; g__Mycosphaerella; OTU_6 s__Mycosphaerella_tassiana 4.40% (96.67%) 6.95% (100.00%) 5.09% (90.00%) 1.17% (100.00%) k__Fungi; p__Ascomycota; c__Saccharomycetes; o__Saccharomycetales; OTU_5 f__Saccharomycetales_fam_Incertae_sedis; g__Debaryomyces; s__Debaryomyces_hansenii 4.09% (93.33%) 2.75% (80.00%) 9.25% (100.00%) 0.27% (100.00%) k__Fungi; p__Ascomycota; c__Saccharomycetes; o__Saccharomycetales; f__Saccharomycetaceae; OTU_9 g__Saccharomyces; s__Saccharomyces_cerevisiae 2.29% (100.00%) 3.43% (100.00%) 3.14% (100.00%) 0.29% (100.00%) k__Fungi; p__Ascomycota; c__Dothideomycetes; o__Pleosporales; f__Pleosporaceae; g__Epicoccum; OTU_1093 s__Epicoccum_nigrum 6.06% (83.33%) 1.90% (60.00%) 3.00% (90.00%) 13.27% (100.00%) k__Fungi; p__Ascomycota; c__Dothideomycetes; o__Pleosporales; f__Pleosporaceae; g__Epicoccum; OTU_15 s__Epicoccum_nigrum 6.35% (86.67%) 1.03% (70.00%) 3.80% (90.00%) 14.23% (100.00%) k__Fungi; p__Ascomycota; c__Dothideomycetes; o__Pleosporales; f__Pleosporaceae; g__Alternaria; OTU_8 s__Alternaria_betae-kenyensis 4.47% (73.33%) 2.20% (40.00%) 1.65% (80.00%) 9.55% (100.00%)

91 k__Fungi; p__Ascomycota; c__Dothideomycetes; o__Capnodiales; f__Cladosporiaceae; g__Cladosporium; OTU_1808 s__Cladosporium_delicatulum 0.29% (53.33%) 0.46% (50.00%) 0.39% (50.00%) 0.01% (60.00%) k__Fungi; p__Ascomycota; c__Dothideomycetes; o__Pleosporales; f__Pleosporales_fam_Incertae_sedis; OTU_16 g__Didymella; s__Didymella_negriana 3.10% (76.67%) 0.90% (40.00%) 3.46% (90.00%) 4.95% (100.00%) k__Fungi; p__Ascomycota; c__Dothideomycetes; o__Capnodiales; f__Mycosphaerellaceae; g__Ramularia; OTU_26 s__Ramularia_pratensis 0.83% (73.33%) 0.81% (50.00%) 1.33% (70.00%) 0.36% (100.00%) k__Fungi; p__Basidiomycota; c__Tremellomycetes; o__Tremellales; f__Bulleribasidiaceae; g__Vishniacozyma; OTU_38 s__Vishniacozyma_carnescens 1.44% (73.33%) 1.14% (60.00%) 1.35% (60.00%) 1.84% (100.00%) k__Fungi; p__Ascomycota; c__Eurotiomycetes; o__Eurotiales; f__Trichocomaceae; g__Penicillium; OTU_47 s__Penicillium_halotolerans 0.13% (56.67%) 0.27% (40.00%) 0.10% (50.00%) 0.03% (80.00%) k__Fungi; p__Ascomycota; c__Sordariomycetes; o__Hypocreales; f__Nectriaceae; g__Fusarium; OTU_2019 s__Fusarium_denticulatum 0.20% (96.67%) 0.46% (100.00%) 0.14% (100.00%) 0.01% (90.00%) k__Fungi; p__Basidiomycota; c__Tremellomycetes; o__Filobasidiales; f__Filobasidiaceae; g__Filobasidium; OTU_20 s__Filobasidium_stepposum 0.86% (76.67%) 0.90% (70.00%) 0.50% (60.00%) 1.19% (100.00%) k__Fungi; p__Ascomycota; c__Dothideomycetes; o__Pleosporales; f__Pleosporaceae; g__Alternaria; OTU_74 s__Alternaria_infectoria 0.36% (46.67%) 0.00% (0.00%) 0.41% (40.00%) 0.66% (100.00%) k__Fungi; p__Ascomycota; c__Dothideomycetes; o__Pleosporales; f__Pleosporaceae; g__Alternaria; OTU_64 s__Alternaria_rosae 0.25% (40.00%) 0.00% (0.00%) 0.25% (20.00%) 0.52% (100.00%) Core community (Total relative abundance) 90.55 ± 2.27% 87.92 ± 5.61% 87.30 ± 4.00% 96.41 ± 1.67%

(A3) Leaf ID Taxonomy Overall Flowering Fruit set Veraison Harvest k__Fungi; p__Ascomycota; c__Dothideomycetes; o__Dothideales; f__Dothioraceae; g__Aureobasidium; OTU_1 s__Aureobasidium_pullulans 19.63% (95.00%) 24.88% (100.00%) 16.42% (100.00%) 24.09% (100.00%) 12.98% (80.00%) k__Fungi; p__Ascomycota; c__Dothideomycetes; o__Capnodiales; f__Cladosporiaceae; g__Cladosporium; OTU_2 s__Cladosporium_ramotenellum 12.06% (95.00%) 12.68% (100.00%) 17.98% (100.00%) 3.72% (90.00%) 14.12% (90.00%) k__Fungi; p__Basidiomycota; c__Tremellomycetes; o__Tremellales; f__Tremellaceae; g__Cryptococcus; OTU_3 s__Cryptococcus_saitoi 10.15% (82.50%) 6.71% (80.00%) 6.02% (90.00%) 18.31% (100.00%) 9.12% (60.00%) k__Fungi; p__Ascomycota; c__Dothideomycetes; o__Capnodiales; f__Mycosphaerellaceae; g__Mycosphaerella; OTU_6 s__Mycosphaerella_tassiana 8.10% (82.50%) 7.97% (90.00%) 15.77% (100.00%) 6.45% (70.00%) 1.54% (70.00%) k__Fungi; p__Ascomycota; c__Saccharomycetes; o__Saccharomycetales; OTU_5 f__Saccharomycetales_fam_Incertae_sedis; g__Debaryomyces; s__Debaryomyces_hansenii 9.40% (87.50%) 7.90% (90.00%) 11.31% (80.00%) 11.05% (90.00%) 6.93% (90.00%) k__Fungi; p__Ascomycota; c__Saccharomycetes; o__Saccharomycetales; f__Saccharomycetaceae; OTU_9 g__Saccharomyces; s__Saccharomyces_cerevisiae 9.08% (92.50%) 8.67% (100.00%) 9.27% (100.00%) 14.39% (100.00%) 3.38% (70.00%) k__Fungi; p__Ascomycota; c__Dothideomycetes; o__Pleosporales; f__Pleosporaceae; g__Epicoccum; OTU_1093 s__Epicoccum_nigrum 1.77% (55.00%) 2.35% (70.00%) 1.08% (40.00%) 1.89% (70.00%) 1.85% (40.00%) k__Fungi; p__Ascomycota; c__Dothideomycetes; o__Pleosporales; f__Pleosporaceae; g__Epicoccum; OTU_15 s__Epicoccum_nigrum 0.91% (47.50%) 2.04% (70.00%) < 0.01% (30.00%) 0.09% (50.00%) 1.71% (40.00%) k__Fungi; p__Ascomycota; c__Dothideomycetes; o__Pleosporales; f__Pleosporaceae; g__Alternaria; OTU_8 s__Alternaria_betae-kenyensis 0.43% (20.00%) 1.66% (50.00%) 0.14% (10.00%) 0.00% (0.00%) < 0.01% (20.00%) k__Fungi; p__Ascomycota; c__Dothideomycetes; o__Capnodiales; f__Cladosporiaceae; g__Cladosporium; OTU_1808 s__Cladosporium_delicatulum 4.13% (52.50%) 0.93% (40.00%) 1.80% (40.00%) 3.22% (90.00%) 10.95% (40.00%) k__Fungi; p__Ascomycota; c__Dothideomycetes; o__Pleosporales; f__Pleosporales_fam_Incertae_sedis; OTU_16 g__Didymella; s__Didymella_negriana 0.20% (25.00%) 0.35% (40.00%) 0.37% (20.00%) 0.06% (30.00%) < 0.01% (10.00%) k__Fungi; p__Ascomycota; c__Dothideomycetes; o__Capnodiales; f__Mycosphaerellaceae; g__Ramularia; OTU_26 s__Ramularia_pratensis 0.57% (25.00%) 1.55% (50.00%) 0.10% (20.00%) 0.67% (10.00%) < 0.01% (20.00%) k__Fungi; p__Basidiomycota; c__Tremellomycetes; o__Tremellales; f__Bulleribasidiaceae; g__Vishniacozyma; OTU_38 s__Vishniacozyma_carnescens 0.75% (30.00%) 0.96% (60.00%) 1.21% (20.00%) 0.77% (30.00%) < 0.01% (10.00%) k__Fungi; p__Ascomycota; c__Eurotiomycetes; o__Eurotiales; f__Trichocomaceae; g__Penicillium; OTU_47 s__Penicillium_halotolerans 0.75% (32.50%) 0.75% (50.00%) 1.00% (20.00%) 0.81% (40.00%) 0.41% (20.00%) k__Fungi; p__Ascomycota; c__Sordariomycetes; o__Hypocreales; f__Nectriaceae; g__Fusarium; OTU_2019 s__Fusarium_denticulatum 0.08% (82.50%) 0.05% (80.00%) 0.05% (80.00%) 0.06% (90.00%) 0.15% (80.00%) Core community (Total relative abundance) 74.32 ± 4.41% 72.08 ± 7.92% 79.65 ± 7.75% 83.12 ± 6.19% 60.87 ± 12.48%

92 (A4) Root ID Taxonomy Overall Flowering Fruit set Veraison Harvest k__Fungi; p__Ascomycota; c__Dothideomycetes; o__Dothideales; f__Dothioraceae; g__Aureobasidium; OTU_1 s__Aureobasidium_pullulans 25.80% (100.00%) 25.28% (100.00%) 16.10% (100.00%) 36.62% (100.00%) 26.30% (100.00%) k__Fungi; p__Ascomycota; c__Dothideomycetes; o__Capnodiales; f__Cladosporiaceae; g__Cladosporium; OTU_2 s__Cladosporium_ramotenellum 8.92% (100.00%) 9.19% (100.00%) 4.40% (100.00%) 11.87% (100.00%) 10.54% (100.00%) k__Fungi; p__Basidiomycota; c__Tremellomycetes; o__Tremellales; f__Tremellaceae; g__Cryptococcus; OTU_3 s__Cryptococcus_saitoi 9.38% (97.50%) 9.65% (90.00%) 3.81% (100.00%) 12.92% (100.00%) 11.51% (100.00%) k__Fungi; p__Ascomycota; c__Dothideomycetes; o__Capnodiales; f__Mycosphaerellaceae; g__Mycosphaerella; OTU_6 s__Mycosphaerella_tassiana 4.76% (92.50%) 4.58% (100.00%) 2.44% (90.00%) 5.27% (90.00%) 6.79% (90.00%) k__Fungi; p__Ascomycota; c__Saccharomycetes; o__Saccharomycetales; OTU_5 f__Saccharomycetales_fam_Incertae_sedis; g__Debaryomyces; s__Debaryomyces_hansenii 2.60% (92.50%) 4.12% (100.00%) 0.50% (90.00%) 3.19% (90.00%) 2.65% (90.00%) k__Fungi; p__Ascomycota; c__Saccharomycetes; o__Saccharomycetales; f__Saccharomycetaceae; OTU_9 g__Saccharomyces; s__Saccharomyces_cerevisiae 0.74% (90.00%) 0.61% (90.00%) 0.33% (80.00%) 0.78% (100.00%) 1.23% (90.00%) k__Fungi; p__Ascomycota; c__Dothideomycetes; o__Pleosporales; f__Pleosporaceae; g__Epicoccum; OTU_1093 s__Epicoccum_nigrum 1.91% (90.00%) 2.98% (100.00%) 1.08% (90.00%) 1.58% (80.00%) 1.99% (90.00%) k__Fungi; p__Ascomycota; c__Dothideomycetes; o__Pleosporales; f__Pleosporaceae; g__Epicoccum; OTU_15 s__Epicoccum_nigrum 1.44% (85.00%) 2.03% (90.00%) 0.45% (100.00%) 1.56% (70.00%) 1.74% (80.00%) k__Fungi; p__Ascomycota; c__Dothideomycetes; o__Pleosporales; f__Pleosporaceae; g__Alternaria; OTU_8 s__Alternaria_betae-kenyensis 1.54% (85.00%) 1.08% (90.00%) 1.28% (90.00%) 1.86% (80.00%) 1.98% (80.00%) k__Fungi; p__Ascomycota; c__Dothideomycetes; o__Capnodiales; f__Cladosporiaceae; g__Cladosporium; OTU_1808 s__Cladosporium_delicatulum 0.51% (40.00%) 1.09% (40.00%) 0.08% (20.00%) 0.59% (60.00%) 0.27% (40.00%) k__Fungi; p__Ascomycota; c__Dothideomycetes; o__Pleosporales; f__Pleosporales_fam_Incertae_sedis; OTU_16 g__Didymella; s__Didymella_negriana 1.03% (77.50%) 0.93% (90.00%) 0.65% (70.00%) 1.81% (80.00%) 0.79% (70.00%) k__Fungi; p__Ascomycota; c__Dothideomycetes; o__Capnodiales; f__Mycosphaerellaceae; g__Ramularia; OTU_26 s__Ramularia_pratensis 2.34% (67.50%) 2.48% (80.00%) 5.19% (80.00%) 0.41% (50.00%) 1.09% (60.00%) k__Fungi; p__Basidiomycota; c__Tremellomycetes; o__Tremellales; f__Bulleribasidiaceae; g__Vishniacozyma; OTU_38 s__Vishniacozyma_carnescens 0.76% (60.00%) 1.66% (70.00%) 0.51% (60.00%) 0.46% (70.00%) 0.40% (40.00%) k__Fungi; p__Ascomycota; c__Eurotiomycetes; o__Eurotiales; f__Trichocomaceae; g__Penicillium; OTU_47 s__Penicillium_halotolerans 0.45% (60.00%) 0.29% (60.00%) 0.51% (70.00%) 0.39% (50.00%) 0.63% (60.00%) k__Fungi; p__Ascomycota; c__Sordariomycetes; o__Hypocreales; f__Nectriaceae; g__Fusarium; OTU_2019 s__Fusarium_denticulatum 0.22% (92.50%) 0.15% (100.00%) 0.13% (70.00%) 0.25% (100.00%) 0.34% (100.00%) OTU_7 k__Fungi; p__Ascomycota; c__Eurotiomycetes; o__Eurotiales; f__Thermoascaceae 5.15% (62.50%) 0.24% (50.00%) 17.75% (90.00%) 2.04% (90.00%) 0.26% (20.00%) OTU_22 k__Fungi; p__Ascomycota; c__Dothideomycetes; o__Pleosporales; f__Lophiostomataceae; g__Lophiostoma 4.53% (77.50%) 6.14% (80.00%) 11.20% (100.00%) 0.23% (80.00%) 0.12% (50.00%) OTU_43 k__Fungi; p__Ascomycota; c__Sordariomycetes; o__Hypocreales 2.29% (55.00%) 1.33% (40.00%) 2.38% (90.00%) 0.46% (40.00%) 4.80% (50.00%) Core community (Total relative abundance) 73.87 ± 3.81% 72.74 ± 8.42% 68.72 ± 6.65% 81.67 ± 8.03% 73.14 ± 7.95%

(B) Soil ID Taxonomy Overall Flowering Fruit set Veraison Harvest k__Fungi; p__Mortierellomycota; c__Mortierellomycotina_cls_Incertae_sedis; o__Mortierellales; OTU_4 f__Mortie+C112:C146_Mortierella; s__Mortierella_amoeboidea 13.19% (100.00%) 14.30% (100.00%) 12.82% (100.00%) 17.12% (100.00%) 8.88% (100.00%) k__Fungi; p__Mortierellomycota; c__Mortierellomycotina_cls_Incertae_sedis; o__Mortierellales; OTU_21 f__Mortierellaceae; g__Mortierella; s__Mortierella_elongata 9.58% (100.00%) 26.09% (100.00%) 9.71% (100.00%) 1.13% (100.00%) 0.56% (100.00%) OTU_10 k__Fungi; p__Ascomycota; c__Geoglossomycetes; o__Geoglossales; f__Geoglossaceae; g__Trichoglossum 8.22% (67.50%) 0.02% (20.00%) 0.29% (50.00%) 23.25% (100.00%) 10.03% (100.00%) k__Fungi; p__Ascomycota; c__Dothideomycetes; o__Pleosporales; f__Pleosporaceae; g__Epicoccum; OTU_15 s__Epicoccum_nigrum 4.70% (95.00%) 0.46% (100.00%) 1.79% (100.00%) 0.08% (80.00%) 15.73% (100.00%) OTU_18 k__Fungi; p__Ascomycota; c__Dothideomycetes; o__Pleosporales 2.80% (87.50%) 1.33% (100.00%) 2.65% (100.00%) 7.49% (70.00%) 0.17% (80.00%) k__Fungi; p__Ascomycota; c__Dothideomycetes; o__Pleosporales; f__Pleosporaceae; g__Epicoccum; OTU_1093 s__Epicoccum_nigrum 2.56% (100.00%) 3.93% (100.00%) 3.52% (100.00%) 0.22% (100.00%) 2.44% (100.00%) OTU_17 k__Fungi; p__Basidiomycota; c__Agaricomycetes; o__Auriculariales 2.18% (65.00%) 0.01% (40.00%) < 0.01% (20.00%) 6.54% (100.00%) 2.38% (100.00%) OTU_14 k__Fungi; p__Ascomycota; c__Sordariomycetes; o__Sordariales; f__Lasiosphaeriaceae; g__Podospora 1.79% (67.50%) < 0.01% (10.00%) 0.09% (100.00%) 0.06% (60.00%) 6.68% (100.00%)

93 k__Fungi; p__Ascomycota; c__Sordariomycetes; o__Sordariales; f__Chaetomiaceae; g__Humicola; OTU_19 s__Humicola_nigrescens 1.67% (100.00%) 5.60% (100.00%) 0.53% (100.00%) 0.23% (100.00%) 0.05% (100.00%) k__Fungi; p__Ascomycota; c__Sordariomycetes; o__Hypocreales; f__Hypocreales_fam_Incertae_sedis; OTU_88 g__Ilyonectria; s__Ilyonectria_macrodidyma 1.37% (100.00%) 0.28% (100.00%) 0.87% (100.00%) 1.16% (100.00%) 3.10% (100.00%) k__Fungi; p__Chytridiomycota; c__Rhizophlyctiomycetes; o__Rhizophlyctidales; f__Rhizophlyctidaceae; OTU_28 g__Rhizophlyctis; s__Rhizophlyctis_rosea 1.10% (85.00%) 0.04% (50.00%) 4.33% (100.00%) 0.11% (90.00%) 0.16% (100.00%) OTU_66 k__Fungi; p__Basidiomycota; c__Agaricomycetes; o__Auriculariales 1.07% (62.50%) 3.80% (100.00%) 0.28% (100.00%) < 0.01% (20.00%) < 0.01% (30.00%) k__Fungi; p__Ascomycota; c__Archaeorhizomycetes; o__Archaeorhizomycetales; f__Archaeorhizomycetaceae; OTU_105 g__Archaeorhizomyces 1.05% (65.00%) 1.39% (90.00%) 2.80% (100.00%) 0.08% (40.00%) < 0.01% (30.00%) OTU_30 k__Fungi; p__Basidiomycota; c__Agaricomycetes; o__Agaricales; f__Clavariaceae; g__Clavaria 0.92% (85.00%) 0.54% (100.00%) 2.88% (100.00%) 0.25% (80.00%) 0.14% (60.00%) k__Fungi; p__Ascomycota; c__Eurotiomycetes; o__Chaetothyriales; f__Herpotrichiellaceae; g__Exophiala; OTU_29 s__Exophiala_equina 0.79% (100.00%) 0.65% (100.00%) 0.86% (100.00%) 1.04% (100.00%) 0.65% (100.00%) OTU_72 k__Fungi; p__Ascomycota; c__Sordariomycetes; o__Sordariales; f__Chaetomiaceae 0.78% (100.00%) 0.90% (100.00%) 1.25% (100.00%) 0.70% (100.00%) 0.33% (100.00%) k__Fungi; p__Ascomycota; c__Dothideomycetes; o__Pleosporales; f__Cucurbitariaceae; g__Pyrenochaetopsis; OTU_62 s__Pyrenochaetopsis_leptospora 0.77% (100.00%) 0.09% (100.00%) 2.34% (100.00%) 0.19% (100.00%) 0.58% (100.00%) k__Fungi; p__Ascomycota; c__Dothideomycetes; o__Pleosporales; f__Pleosporaceae; g__Alternaria; OTU_8 s__Alternaria_betae-kenyensis 0.77% (100.00%) 0.11% (100.00%) 0.98% (100.00%) 0.28% (100.00%) 1.68% (100.00%) k__Fungi; p__Basidiomycota; c__Tremellomycetes; o__Filobasidiales; f__Piskurozymaceae; g__Solicoccozyma; OTU_46 s__Solicoccozyma_phenolica 0.68% (100.00%) 0.27% (100.00%) 0.27% (100.00%) 1.01% (100.00%) 1.17% (100.00%) OTU_51 k__Fungi; p__Ascomycota; c__Sordariomycetes; o__Xylariales; f__Amphisphaeriaceae 0.62% (100.00%) 0.08% (100.00%) 0.10% (100.00%) 1.54% (100.00%) 0.79% (100.00%) OTU_35 k__Fungi; p__Ascomycota; c__Leotiomycetes; o__Helotiales 0.60% (97.50%) 1.83% (100.00%) 0.14% (100.00%) 0.33% (90.00%) 0.04% (100.00%) OTU_70 k__Fungi; p__Ascomycota; c__Sordariomycetes; o__Sordariales; f__Chaetomiaceae; g__Chaetomium 0.59% (100.00%) 1.34% (100.00%) 0.40% (100.00%) 0.34% (100.00%) 0.25% (100.00%) OTU_23 k__Fungi; p__Ascomycota; c__Leotiomycetes; o__Helotiales; f__Helotiaceae 0.58% (55.00%) 1.12% (90.00%) 1.18% (100.00%) < 0.01% (30.00%) 0.00% (0.00%) OTU_43 k__Fungi; p__Ascomycota; c__Sordariomycetes; o__Hypocreales 0.54% (97.50%) 0.59% (90.00%) 0.62% (100.00%) 0.70% (100.00%) 0.26% (100.00%) k__Fungi; p__Basidiomycota; c__Agaricomycetes; o__Auriculariales; f__Auriculariales_fam_Incertae_sedis; OTU_89 g__Auricularia 0.53% (47.50%) < 0.01% (10.00%) < 0.01% (20.00%) 0.05% (60.00%) 1.98% (100.00%) OTU_150 k__Fungi; p__Ascomycota; c__Sordariomycetes; o__Hypocreales; f__Nectriaceae; g__Fusarium 0.46% (100.00%) 0.15% (100.00%) 0.82% (100.00%) 0.12% (100.00%) 0.75% (100.00%) OTU_56 k__Fungi; p__Ascomycota; c__Sordariomycetes; o__Hypocreales; f__Nectriaceae 0.44% (95.00%) 0.33% (90.00%) 1.10% (100.00%) 0.09% (90.00%) 0.26% (100.00%) OTU_2041 k__Fungi; p__Ascomycota; c__Sordariomycetes; o__Sordariales; f__Chaetomiaceae 0.41% (100.00%) 0.38% (100.00%) 0.47% (100.00%) 0.46% (100.00%) 0.33% (100.00%) OTU_185 k__Fungi; p__Ascomycota; c__Leotiomycetes; o__Helotiales 0.39% (87.50%) 0.24% (60.00%) 0.04% (90.00%) 0.45% (100.00%) 0.80% (100.00%) k__Fungi; p__Ascomycota; c__Sordariomycetes; o__Hypocreales; f__Hypocreales_fam_Incertae_sedis; OTU_37 g__Sarocladium; s__Sarocladium_bactrocephalum 0.38% (72.50%) 0.10% (60.00%) 1.30% (100.00%) < 0.01% (30.00%) 0.18% (100.00%) k__Fungi; p__Ascomycota; c__Eurotiomycetes; o__Eurotiales; f__Trichocomaceae; g__Penicillium; OTU_91 s__Penicillium_nodositatum 0.38% (87.50%) 0.04% (90.00%) 0.02% (60.00%) 0.70% (100.00%) 0.76% (100.00%) k__Fungi; p__Ascomycota; c__Dothideomycetes; o__Capnodiales; f__Cladosporiaceae; g__Cladosporium; OTU_2 s__Cladosporium_ramotenellum 0.36% (100.00%) 0.41% (100.00%) 0.43% (100.00%) 0.12% (100.00%) 0.46% (100.00%) k__Fungi; p__Ascomycota; c__Eurotiomycetes; o__Eurotiales; f__Trichocomaceae; g__Penicillium; OTU_49 s__Penicillium_paczoskii 0.31% (90.00%) 0.01% (80.00%) 0.01% (80.00%) 0.08% (100.00%) 1.08% (100.00%) k__Fungi; p__Ascomycota; c__Sordariomycetes; o__Hypocreales; f__Nectriaceae; g__Gibberella; OTU_179 s__Gibberella_zeae 0.25% (100.00%) 0.06% (100.00%) 0.14% (100.00%) 0.38% (100.00%) 0.43% (100.00%) OTU_129 k__Fungi; p__Ascomycota; c__Sordariomycetes; o__Sordariales; f__Chaetomiaceae; g__Chaetomium 0.17% (100.00%) 0.26% (100.00%) 0.21% (100.00%) 0.11% (100.00%) 0.10% (100.00%) Core community (Total relative abundance) 63.40 ± 3.54% 66.77 ± 4.43% 55.21 ± 4.09% 66.39 ± 5.10% 63.21 ± 3.60%

Table S2 Classification of random forest models of the development stage

Flowering Fruit set Veraison Harvest Class Error Grape Fruit set N.A. 9 1 0 0.100 Veraison N.A. 1 9 0 0.100 Harvest N.A. 0 1 9 0.100 Leaf Flowering 6 2 1 1 0.400 Fruit set 1 6 0 3 0.400 Veraison 0 2 7 1 0.300 Harvest 0 1 3 6 0.400 Root Flowering 8 2 0 0 0.200 Fruit set 3 7 0 0 0.400 Veraison 0 0 9 1 0.200 Harvest 0 0 3 7 0.300 Soil Flowering 9 1 0 0 0.100 Fruit set 3 7 0 0 0.300 Veraison 0 0 10 0 0.000 Harvest 0 0 1 9 0.100 N.A.: Not applied

Table S3 Important features of random forest models outside the core community The same as Table S1.

(A) Grape ID Taxonomy Overall Fruit set Veraison Harvest k__Fungi; p__Ascomycota; c__Dothideomycetes; o__Capnodiales; f__Capnodiales_fam_Incertae_sedis; OTU_146 g__Rachicladosporium; s__Rachicladosporium_cboliae 0.04% (30.00%) 0.00% (0.00%) 0.00% (0.00%) 0.12% (90.00%) k__Fungi; p__Ascomycota; c__Dothideomycetes; o__Pleosporales; f__Pleosporaceae; g__Stemphylium; OTU_54 s__Stemphylium_herbarum 0.51% (46.67%) 0.00% (0.00%) 0.70% (40.00%) 0.83% (100.00%) k__Fungi; p__Ascomycota; c__Saccharomycetes; o__Saccharomycetales; f__Saccharomycodaceae; g__Hanseniaspora; OTU_33 s__Hanseniaspora_uvarum 0.06% (36.67%) 0.04% (20.00%) 0.11% (50.00%) 0.02% (40.00%) k__Fungi; p__Basidiomycota; c__Microbotryomycetes; o__Sporidiobolales; f__Sporidiobolales_fam_Incertae_sedis; OTU_40 g__Rhodotorula; s__Rhodotorula_babjevae 0.44% (30.00%) 0.67% (10.00%) 0.23% (10.00%) 0.41% (70.00%) k__Fungi; p__Ascomycota; c__Sordariomycetes; o__Hypocreales; f__Hypocreales_fam_Incertae_sedis; g__Sarocladium; OTU_37 s__Sarocladium_bactrocephalum 0.06% (23.33%) 0.12% (10.00%) 0.00% (0.00%) 0.06% (60.00%) OTU_2201 k__Fungi; p__Ascomycota; c__Sordariomycetes; o__Hypocreales; f__Nectriaceae; g__Fusarium; s__Fusarium_denticulatum 0.05% (46.67%) 0.01% (70.00%) 0.12% (70.00%) 0.00% (0.00%) k__Fungi; p__Ascomycota; c__Saccharomycetes; o__Saccharomycetales; f__Saccharomycetaceae; g__Torulaspora; OTU_121 s__Torulaspora_delbrueckii 0.08% (36.67%) 0.00% (0.00%) 0.23% (50.00%) 0.01% (60.00%) k__Fungi; p__Mortierellomycota; c__Mortierellomycotina_cls_Incertae_sedis; o__Mortierellales; f__Mortierellaceae; OTU_4 g__Mortierella; s__Mortierella_amoeboidea 0.52% (26.67%) 0.83% (30.00%) 0.72% (40.00%) < 0.01% (10.00%) k__Fungi; p__Ascomycota; c__Dothideomycetes; o__Pleosporales; f__Pleosporales_fam_Incertae_sedis; g__Didymella; OTU_323 s__Didymella_negriana 0.20% (30.00%) 0.00% (0.00%) 0.24% (50.00%) 0.36% (40.00%) k__Fungi; p__Ascomycota; c__Dothideomycetes; o__Capnodiales; f__Cladosporiaceae; g__Cladosporium; OTU_128 s__Cladosporium_sphaerospermum 0.11% (30.00%) 0.00% (0.00%) 0.33% (40.00%) 0.01% (50.00%)

(B) Leaf ID Taxonomy Overall Flowering Fruit set Veraison Harvest k__Fungi; p__Mortierellomycota; c__Mortierellomycotina_cls_Incertae_sedis; o__Mortierellales; f__Mortierellaceae; OTU_21 g__Mortierella; s__Mortierella_elongata 0.04% (20.00%) 0.15% (10.00%) < 0.01% (10.00%) < 0.01% (10.00%) 0.01% (50.00%) k__Fungi; p__Ascomycota; c__Dothideomycetes; o__Pleosporales; f__Pleosporales_fam_Incertae_sedis; g__Boeremia; OTU_48 s__Boeremia_noackiana 2.74% (17.50%) 0.14% (20.00%) 0.00% (0.00%) < 0.01% (10.00%) 11.10% (40.00%) k__Fungi; p__Mortierellomycota; c__Mortierellomycotina_cls_Incertae_sedis; o__Mortierellales; f__Mortierellaceae; OTU_4 g__Mortierella; s__Mortierella_amoeboidea 0.26% (27.50%) 0.91% (30.00%) 0.00% (0.00%) 0.12% (30.00%) 0.02% (50.00%) k__Fungi; p__Ascomycota; c__Dothideomycetes; o__Capnodiales; f__Cladosporiaceae; g__Cladosporium; OTU_2078 s__Cladosporium_delicatulum 0.01% (30.00%) < 0.01% (20.00%) < 0.01% (30.00%) < 0.01% (50.00%) 0.01% (20.00%) OTU_24 k__Fungi; p__Ascomycota; c__Eurotiomycetes; o__Eurotiales; f__Trichocomaceae; g__Aspergillus; s__Aspergillus_conicus 1.96% (15.00%) 0.11% (30.00%) 6.77% (10.00%) 0.38% (20.00%) 0.00% (0.00%) k__Fungi; p__Basidiomycota; c__Tremellomycetes; o__Filobasidiales; f__Filobasidiaceae; g__Filobasidium; OTU_20 s__Filobasidium_stepposum 1.42% (25.00%) 1.64% (30.00%) 0.25% (20.00%) 1.19% (40.00%) 2.58% (10.00%) < 0.01% OTU_2201 k__Fungi; p__Ascomycota; c__Sordariomycetes; o__Hypocreales; f__Nectriaceae; g__Fusarium; s__Fusarium_denticulatum (35.00%) < 0.01% (20.00%) < 0.01% (30.00%) < 0.01% (40.00%) < 0.01% (50.00%) < 0.01% OTU_23 k__Fungi; p__Ascomycota; c__Leotiomycetes; o__Helotiales; f__Helotiaceae (10.00%) 0.00% (0.00%) 0.00% (0.00%) 0.00% (0.00%) 0.01% (40.00%)

96 k__Fungi; p__Ascomycota; c__Dothideomycetes; o__Pleosporales; f__Pleosporales_fam_Incertae_sedis; g__Didymella; < 0.01% OTU_323 s__Didymella_negriana (5.00%) 0.00% (0.00%) < 0.01% (20.00%) 0.00% (0.00%) 0.00% (0.00%) OTU_35 k__Fungi; p__Ascomycota; c__Leotiomycetes; o__Helotiales 0.65% (5.00%) 2.68% (20.00%) 0.00% (0.00%) 0.00% (0.00%) 0.00% (0.00%) k__Fungi; p__Ascomycota; c__Saccharomycetes; o__Saccharomycetales; f__Saccharomycetaceae; g__Torulaspora; OTU_121 s__Torulaspora_delbrueckii 0.21% (17.50%) 0.78% (20.00%) 0.02% (30.00%) 0.07% (20.00%) 0.00% (0.00%) k__Fungi; p__Ascomycota; c__Saccharomycetes; o__Saccharomycetales; f__Saccharomycodaceae; g__Hanseniaspora; OTU_33 s__Hanseniaspora_uvarum 2.15% (25.00%) 0.69% (20.00%) 1.25% (30.00%) 1.33% (20.00%) 5.29% (30.00%)

(C) Root ID Taxonomy Overall Flowering Fruit set Veraison Harvest OTU_17 k__Fungi; p__Basidiomycota; c__Agaricomycetes; o__Auriculariales 0.80% (22.50%) 2.70% (40.00%) 0.41% (50.00%) 0.00% (0.00%) 0.00% (0.00%) k__Fungi; p__Mortierellomycota; c__Mortierellomycotina_cls_Incertae_sedis; o__Mortierellales; f__Mortierellaceae; OTU_4 g__Mortierella; s__Mortierella_amoeboidea 0.67% (45.00%) 0.00% (0.00%) 0.55% (90.00%) 0.91% (40.00%) 1.25% (50.00%) OTU_51 k__Fungi; p__Ascomycota; c__Sordariomycetes; o__Xylariales; f__Amphisphaeriaceae 0.09% (22.50%) 0.00% (0.00%) 0.08% (40.00%) 0.00% (0.00%) 0.29% (50.00%) k__Fungi; p__Ascomycota; c__Sordariomycetes; o__Hypocreales; f__Bionectriaceae; g__Clonostachys; OTU_122 s__Clonostachys_rosea 0.34% (32.50%) 0.04% (30.00%) 0.27% (50.00%) 0.00% (0.00%) 1.01% (50.00%) k__Fungi; p__Ascomycota; c__Saccharomycetes; o__Saccharomycetales; f__Saccharomycodaceae; g__Hanseniaspora; OTU_33 s__Hanseniaspora_uvarum 0.03% (40.00%) 0.02% (40.00%) 0.02% (50.00%) < 0.01% (30.00%) 0.08% (40.00%) k__Fungi; p__Ascomycota; c__Eurotiomycetes; o__Chaetothyriales; f__Herpotrichiellaceae; g__Exophiala; OTU_29 s__Exophiala_equina 0.25% (32.50%) 0.30% (20.00%) 0.42% (80.00%) 0.00% (0.00%) 0.23% (30.00%) k__Fungi; p__Basidiomycota; c__Tremellomycetes; o__Filobasidiales; f__Filobasidiaceae; g__Filobasidium; OTU_20 s__Filobasidium_stepposum 0.18% (20.00%) 0.57% (50.00%) < 0.01% (10.00%) 0.00% (0.00%) 0.14% (20.00%) OTU_2201 k__Fungi; p__Ascomycota; c__Sordariomycetes; o__Hypocreales; f__Nectriaceae; g__Fusarium; s__Fusarium_denticulatum 0.01% (52.50%) 0.01% (60.00%) < 0.01% (40.00%) < 0.01% (30.00%) 0.01% (80.00%)

(D) Soil ID Taxonomy Overall Flowering Fruit set Veraison Harvest k__Fungi; p__Ascomycota; c__Eurotiomycetes; o__Eurotiales; f__Trichocomaceae; g__Penicillium; OTU_584 s__Penicillium_multicolor 0.01% (32.50%) 0.00% (0.00%) 0.00% (0.00%) < 0.01% (20.00%) 0.02% (100.00%) k__Fungi; p__Ascomycota; c__Dothideomycetes; o__Pleosporales; f__Pleosporales_fam_Incertae_sedis; g__Didymella; OTU_501 s__Didymella_pedeiae 0.14% (75.00%) 0.03% (30.00%) 0.03% (70.00%) 0.12% (100.00%) 0.36% (100.00%) OTU_151 k__Fungi; p__Ascomycota; c__Leotiomycetes; o__Helotiales 0.11% (42.50%) 0.42% (100.00%) < 0.01% (60.00%) 0.00% (0.00%) 0.00% (0.00%) k__Fungi; p__Ascomycota; c__Archaeorhizomycetes; o__Archaeorhizomycetales; f__Archaeorhizomycetaceae; OTU_519 g__Archaeorhizomyces 0.03% (57.50%) 0.06% (100.00%) 0.07% (100.00%) < 0.01% (20.00%) 0.00% (0.00%) k__Fungi; p__Mortierellomycota; c__Mortierellomycotina_cls_Incertae_sedis; o__Mortierellales; f__Mortierellaceae; OTU_763 g__Mortierella; s__Mortierella_elongata 0.13% (60.00%) 0.41% (100.00%) 0.09% (90.00%) 0.01% (30.00%) < 0.01% (10.00%) k__Fungi; p__Basidiomycota; c__Tremellomycetes; o__Filobasidiales; f__Piskurozymaceae; g__Solicoccozyma; OTU_223 s__Solicoccozyma_terricola 0.04% (65.00%) 0.07% (100.00%) 0.09% (100.00%) < 0.01% (20.00%) < 0.01% (30.00%) OTU_392 k__Fungi; p__Ascomycota; c__Lecanoromycetes; o__Lecanorales; f__Cladoniaceae; g__Cladonia; s__Cladonia_cervicornis 0.01% (45.00%) 0.02% (70.00%) 0.03% (100.00%) 0.00% (0.00%) 0.00% (0.00%) k__Fungi; p__Ascomycota; c__Eurotiomycetes; o__Eurotiales; f__Trichocomaceae; g__Aspergillus; OTU_94 s__Aspergillus_ardalensis 0.01% (60.00%) < 0.01% (20.00%) < 0.01% (30.00%) 0.02% (80.00%) 0.03% (100.00%) OTU_341 k__Fungi; p__Ascomycota; c__Sordariomycetes; o__Sordariales; f__Lasiosphaeriaceae; g__Podospora 0.05% (50.00%) 0.00% (0.00%) 0.00% (0.00%) 0.08% (80.00%) 0.12% (100.00%)

Table S4 ANOVA results of weather conditions between developmental stages

Factor Weather parameter Df F p Pr (>F) Env1 Solar radiation 3 98.42 3.33E-04 *** Env2 Mean high temperature 3 49.88 0.002 ** Env3 Mean low temperature 3 165.4 1.20E-04 *** Env4 Mean temperature 3 29.72 0.003 ** Env5 Precipitation 3 87.13 4.24E-04 *** Env6 Evaporation 3 279.6 4.22E-05 *** Env7 Transpiration 3 635.5 8.21E-06 *** Env8 Relative soil moisture 3 118.2 2.32E-04 ***

Table S5 Spray programme for the studied vineyards from shoot growth until pre-veraison

Microthio Kocide Ridomil Mycross Spray Vineyard Date Dispersea (kg/ha)b (kg/ha)c (mL/ha)d (kg/ha)

A 10/24/2017 1 3.0 0.2 N.A. N.A. B 10/24/2017 A 10/31/2017 2 4.0 0.6 N.A. N.A. B 10/31/2017 A 11/15/2017 3 4.1 1.1 2.3 N.A. B 11/08/2017 A 11/30/2017 4 5.4 3.6 N.A. N.A. B 11/24/2017 A 12/21/2017 5 5.0 2.0 N.A. 160.0 B 12/15/2017 A 01/05/2018 6 5.0 2.0 2.3 160.0 B 12/30/2017 A 01/17/2019 7 4.2 1.2 N.A. N.A. B 01/17/2018 a Microthio Disperse: 800g/kg sulphur bKocide: 53.8 % w/w copper hydroxide c Ridomil: 5% w/w metalaxyl-m + 60% w/w copper hydroxide d Mycross: 200 g/L myclobutanil

Table S6 Soil properties of each sampling site at harvest

Site Soil type pH Total C (%) Total N (%) C: N Vineyard A V1 Clay loam 6.29 4.94 0.33 14.83 V2 Clay loam 6.74 4.84 0.32 15.17 V3 Clay loam 6.69 4.73 0.32 14.73 V4 Clay loam 6.48 5.93 0.38 15.40 V5 Clay loam 6.18 4.91 0.33 14.93 Vineyard B V6 Clay loam 6.56 4.84 0.32 15.26 V7 Clay loam 6.11 3.67 0.23 16.19 V8 Clay loam 6.28 4.72 0.29 16.36 V9 Clay loam 6.26 4.52 0.32 14.05 V10 Clay loam 6.46 3.58 0.25 14.39

100

Fig. S1 Co-occurrence pattern between the soil core microbiome and weather parameters. Circle nodes represent soil core OTUs and environmental variables, with different colours. Direct connections between nodes indicate strong correlations (Spearman correlation coefficient, ρ ≥ 0.8; p < 0.01). The colour of edges describes the positive correlation (pink) or the negative correlation (grey). The size of nodes is proportioned to the interconnected degree. The width of edges is proportioned to the correlation coefficients. Abbreviations: Env1, solar radiation; Env2, mean high temperature; Env3, mean low temperature; Env4, mean temperature; Env5, precipitation; Env6, evaporation; Env7, transpiration; Env8, relative soil moisture.

101

Fig. S2 Experimental design and sampling scheme. (A) Sampling map of two vineyards in Mornington Peninsula wine region, Australia. Five Vitis vinifera Pinot Noir vines were selected in each vineyard. (B) Soils, roots, leaves, and grapes/ flowers were collected in each grapevine throughout the annual growth cycle: flowering, fruit set, veraison (berry colour change), and harvest. Calendar dates of sample collection are indicated for each developmental stage. Grapevines were created with BioRender: https://app.biorender.com/.

102 CHAPTER FIVE

Diversity and dynamics of fungi during spontaneous fermentations and

association with unique aroma profiles in wine

Status of chapter:

This chapter has been accepted for publication in International Journal of Food Microbiology journal.

I am the lead author in collaboration with Dr Jean-Luc Legras in INRA, France.

Liu D., Legras J.-L., Zhang P., Chen D., Howell K.S.. The fungal microbiome is an important component of vineyard ecosystems and correlates with regional distinctiveness of wine. International

Journal of Food Microbiology journal. Accepted Author Manuscript.

103 Diversity and dynamics of fungi during spontaneous fermentations and association with unique aroma profiles in wine

Di Liu a, Jean-Luc Legras b, Pangzhen Zhang a, Deli Chen a, Kate Howell a * a School of Agriculture and Food, Faculty of Veterinary and Agricultural Sciences, University of Melbourne Parkville 3010 Australia b SPO, Université Montpellier INRA, Montpellier SupAgro, F-34060 Montpellier, France

*Corresponding author. [email protected]

104 Abstract

Microbial activity is an integral part of an agricultural ecosystem and influences the quality of agricultural commodities. Microbial ecology influences grapevine health and crop production, conversion of sugar to ethanol during fermentation, thus wine aroma and flavour. There are regionally differentiated microbial patterns in grapevines and must but how microbial patterns contribute to wine regional distinctiveness (terroir) at small scale (<100 km) is not well defined. Here we characterise fungal communities, yeast populations, and Saccharomyces cerevisiae populations during spontaneous fermentation using metagenomics and population genetics to investigate microbial distribution and fungal contributions to the resultant wine. We found differentiation of fungi, yeasts, and S. cerevisiae between geographic origins (estate/vineyard), with influences from the grape variety. Growth and dominance of S. cerevisiae during fermentation reshaped the fungal community and showed geographic structure at the strain level. Associations between fungal microbiota diversity and wine chemicals suggest that S. cerevisiae plays a primary role in determining wine aroma profiles at a sub-regional scale. The geographic distribution at scales of less than 12 km supports that differential microbial communities, including the dominant fermentative yeast S. cerevisiae can be distinct in a local setting. These findings provide further evidence for microbial contributions to wine terroir, and perspectives for sustainable agricultural practices to maintain microbial diversity and optimise fermentation function to craft beverage quality.

Keywords: microbial biogeography, fungal microbiota, Saccharomyces cerevisiae, wine, spontaneous fermentation

105 5.1 Introduction

Wine grapes (Vitis vinifera) are an economically and culturally important agricultural commodity for which microbial activity plays key roles in grape and wine production and quality (Barata et al., 2012;

Swiegers et al., 2005). The grapevine harbours complex and diverse microbiota, such as bacteria, filamentous fungi, and yeasts (Barata et al., 2012; Liu and Howell, 2020; Stefanini and Cavalieri,

2018), which substantially modulate vine health, growth, and crop productivity (Berg et al., 2014;

Gilbert et al., 2014; Müller et al., 2016). Grapevine-associated microbiota can be transferred to the grape must/juice and have an influence on wine composition, aroma, flavour, and quality (Barata et al., 2012; Ciani et al., 2010; Morrison-Whittle and Goddard, 2018). Wine fermentation is a complex and multispecies process, involving numerous transformations by fungi and bacteria to sculpt chemical and sensory properties of the resulting wines (Swiegers et al., 2005; Verginer et al., 2010).

While these consortia all contribute to wine flavour formation, the fermentation process is principally driven by diverse populations of Saccharomyces cerevisiae (Fleet, 2003; Goddard, 2008; Howell et al., 2006).

Microbial biogeography contributes to regional distinctiveness of agricultural products, known as

“terroir” in viticulture [reviewed by Liu et al. (2019)]. Biogeographical patterns in the microbiota associated with the grape and must have been demonstrated for both fungi and bacteria at a regional scale (Bokulich et al., 2014; Mezzasalma et al., 2018; Pinto et al., 2015; Taylor et al., 2014), which are conditioned by multiple factors, such as cultivar, climate and vintage, topography, and soil properties (Bokulich et al., 2014; Liu et al., 2019; Miura et al., 2017; Portillo et al., 2016;

Zarraonaindia et al., 2015). Bokulich et al. (2016) showed that the bacterial and fungal consortia correlated with metabolites in finished wines, highlighting the importance of fermentative yeasts (for example, S. cerevisiae, Hanseniaspora uvarum, Pichia guilliermondii) and lactic acid bacteria

(Leuconostocaceae) on the abundance of regional aromatic signatures. Our previous research revealed that wine-related fungal communities structured and distinguished vineyard ecosystems by impacting the flavour and quality of wine, and weather with a contribution by soil properties affected soil and

106 must fungal communities and thus the composition of wines across six winegrowing regions in southern Australia (Liu et al., 2020). However, whether grape-associated microbiota exhibit distinct patterns of distribution at smaller geographic scales (for example individual vineyards) and their associations with wine aroma profiles are not well understood.

Geographic differentiation of S. cerevisiae populations is evident at global (Legras et al., 2007; Liti et al., 2009) and regional scales (Gayevskiy and Goddard, 2012; Knight and Goddard, 2015), revealing a picture of distinctive populations at large scales more than ~ 100 km. Drumonde-Neves et al. (2018) showed higher genetic divergence among S. cerevisiae populations between rather than within islands/regions (~ 1.5 – 260 km scale) and suggested a prevailing role of geography over ecology

(grape varieties and agricultural cultivation) in shaping diversification, as previously reported

(Goddard et al., 2010). At small scales, several studies characterised significant genetic differences between S. cerevisiae populations residing in different vineyards within the same region [Börlin et al.

(2016); <10 km] and different sites within a vineyard [Schuller and Casal (2007); 10 – 400 m]. Knight et al. (2015) experimentally demonstrated that regional strains of S. cerevisiae produce distinct wine chemical compositions, suggesting a prominent route by which regional S. cerevisiae shape wine terroir. While few studies have investigated how S. cerevisiae differentiation can affect wine aroma, flavour, and characteristics, and none have considered S. cerevisiae, fermentative yeasts, and the global fungal communities simultaneously to quantify their contributions to the resultant wine.

To investigate these questions, we sampled microbial communities associated with Pinot Noir and

Chardonnay grape must and juice from three wine estates with 8 - 12 km pairwise distances to include grapes from 11 vineyards in the Mornington Peninsula wine region of Victoria, Australia. Using culture-independent sequencing to characterise fungal communities, we disentangled the influences of geographic origin (estate/vineyard), grape variety, and (spontaneous) fermentation stage on the diversity, structure, and composition of the fungal communities. Yeast populations were isolated during spontaneous wine fermentation and taxonomically identified, and the S. cerevisiae populations differentiated using microsatellite analysis. To identify the volatiles that differentiated the wine estates

107 we used headspace solid-phase microextraction gas-chromatographic mass-spectrometric (HS-SPME–

GC-MS) for metabolite profiling of the resultant wines. Associations between fungal communities and wine metabolites were elucidated with partial least squares regression (PLSR) and structural equation model (SEM). We demonstrate that the grape/wine microbiota and metabolites are geographically distinct, identify multiple layers of fungal microbiota that correlate with wine aroma profiles, and demonstrate that distinctive S. cerevisiae exert the most powerful influences on wine quality and style at small geographic scales.

5.2 Materials and methods

5.2.1 Sampling

Five Vitis vinifera cv. Pinot Noir and six Vitis vinifera cv. Chardonnay vineyards from three wine estates (designated A, B, and C) in the Mornington Peninsula region were selected to conduct this study in 2019 (Supplementary Fig. S1). The distance between wine estates A and B, A and C, B and C is 8 km, 12 km, and 10 km, respectively. Within these estates, vineyards are within a 5 km radius of one another. All vineyards were commercially managed using similar viticultural practices, for example, grapevines were under vertical shoot positioning trellising systems and were applied with the same sprays. Chemical constituents of harvested grapes (°Brix, pH, total acidity) were similar and listed in Supplementary Table S1. Fermentations sampled in this study were conducted without addition of commercial yeasts following similar fermentation protocols. Tanks were cleaned and decontaminated before filling grapes. For Pinot Noir, crushed grapes were held for three days at a cool temperature (known as cold-soaking) and followed by warming so fermentation could commence. Fermentation samples were collected at three time points in duplicate: before fermentation

(BF; destemmed and crushed grape musts of Pinot Noir, Chardonnay juice following clarification), at the middle of fermentation (MF, around 50% of sugar fermented), and at the end of fermentation (EF, before pressing, 6-7 °Brix) (Supplementary Table S1). Samples (n = 66) were shipped on ice to the laboratory. Each sample was divided into two subsamples, one was used immediately to isolate yeasts

108 and the other was stored at -20°C for DNA extraction, next-generation sequencing and wine volatile analysis.

5.2.2 Wine volatile analysis

Volatile compounds of EF samples were determined using headspace solid-phase microextraction gas-chromatographic mass-spectrometric (HS-SPME–GC-MS) method (Liu et al., 2016; Zhang et al.,

2015) with some modifications. Analyses were conducted with Agilent 6850 GC system and a 5973 mass detector (Agilent Technologies, Santa Clara, CA, USA), equipped with a PAL RSI 120 autosampler (CTC Analytics AG, Switzerland). In brief, 10 mL wine sample was added to a 20 ml glass vial containing 2 g sodium chloride and 20 μL internal standard (4-Octanol, 100 mg/L), and then equilibrated at 35 °C for 15 min. A polydimethylsiloxane/divinylbenzene (PDMS/DVB, Supelco) 65

μm SPME fibre was immersed in the headspace for 10 min at 35°C with agitation, and followed by desorbing in the GC injector for 4 min at 220 °C. Volatile compounds were separated on an Agilent

J&W DB-Wax Ultra Inert capillary GC column (30 m × 0.25 mm × 0.25 μm), with helium carrier gas at a flow rate of 0.7 mL/min. The column temperature program was as follows: holding 40 °C for 10 min, increasing at 3.0 °C/min to 220 °C and holding at this temperature for 10 min. The temperature of the transfer line of GC and MS was set at 240 °C. The ion source temperature was 230 °C. The MS was operated in positive electron ionization (EI) mode with scanning over a mass acquisition range of

35 to 350 m/z. Raw data were processed with Agilent ChemStation Software for qualification and quantification. Volatile compounds (n = 79) were identified in wine samples according to retention indices referencing standards and mass spectra matching with NIST11 library. 13 successive levels of standards in model wine solutions (12% v/v ethanol saturated with potassium hydrogen tartrate and adjusted to pH 3.5 using 40% w/v tartaric acid) were analysed by the same protocol as wine samples to establish the calibration curves for quantification. Peak areas of compounds were integrated via target ions model. The concentrations of compounds were calculated with the calibration curves and used for downstream data analysis.

109 5.2.3 DNA extraction and sequencing

Samples were thawed, and biomass was recovered by centrifugation at 4,000 × g for 15 min, washed three times in ice-cold phosphate buffered saline (PBS) with 1% polyvinylpolypyrrolidone (PVPP) and centrifuged at 10,000 × g for 10 min (Bokulich et al., 2014). The obtained pellets were used for

DNA extraction using PowerSoil™ DNA Isolation kits (QIAgen, CA, USA). DNA extracts were stored at -20 °C until further analysis.

Genomic DNA was submitted to Australian Genome Research Facility (AGRF) for amplicon sequencing. To analyse the fungal communities, partial fungal internal transcribed spacer (ITS) region was amplified using the universal primer pairs ITS1F/2 (Gardes and Bruns, 1993). The primary PCR reactions contained 10 ng DNA template, 2× AmpliTaq Gold® 360 Master Mix (Life Technologies,

Australia), 5 pmol of each primer. A secondary PCR to index the amplicons was performed with

TaKaRa Taq DNA Polymerase (Clontech). Amplification was conducted as follows: 95 °C for 7 min, followed by 35 cycles of 94 °C for 30 s, 55 °C for 45 s, 72 °C for 60 s, and a final extension at 72 °C for 7 min. The resulting amplicons were cleaned again using magnetic beads, quantified by fluorometry (Promega Quantifluor), and normalised. The equimolar pool was cleaned a final time using magnetic beads to concentrate the pool and measured using a High-Sensitivity D1000 Tape on an Agilent 2200 TapeStation. The pool was diluted to 5nM and molarity was confirmed again with a

High-Sensitivity D1000 Tape. This was followed by 300 bp paired-end sequencing on an Illumina

MiSeq (San Diego, CA, USA).

Raw sequences were processed using QIIME v1.9.2 (Caporaso et al., 2010). Low quality regions (Q <

20) were trimmed from the 5′ end of the sequences, and the paired ends were joined using FLASH

(Magoč and Salzberg, 2011). Primers were trimmed and a further round of quality control was conducted to discard full length duplicate sequences, short sequences (< 100 nt), and sequences with ambiguous bases. Sequences were clustered followed by chimera checking using UCHIME algorithm from USEARCH v7.1.1090 (Edgar et al., 2011). Operational taxonomic units (OTUs) were assigned

110 using UCLUST open-reference OTU-picking workflow with a threshold of 97% pairwise identity

(Edgar, 2010). Singletons or unique reads in the resultant data set were discarded. Taxonomy was assigned to OTUs in QIIME using the UNITE fungal ITS database (v7.2) (Kõljalg et al., 2005). To avoid/reduce biases generated by varying sequencing depth, sequences were rarefied to 10,000 per sample prior to downstream analysis. Raw sequencing reads are publicly available in the National

Centre for Biotechnology Information (NCBI) Sequence Read Archive under the bioproject

PRJNA646865.

5.2.4 Yeast isolation and identification

Yeast populations were evaluated during spontaneous fermentations (BF, MF, EF). Aliquots of 0.1 ml from serially diluted samples (from 10–2 to 10–5) were spread on Wallerstein Laboratory Nutrient

(WLN) agar plates (Oxoid, Australia) that were supplemented with 34 mg/mL chloramphenicol and

25 mg/mL ampicillin to inhibit bacterial growth. After 5 days of incubation at 28°C, colonies were counted (30-300 colonies) and colony morphology was recorded to differentiate yeast species according to Pallmann et al. (2001) and Romancino et al. (2008). Characteristic yeast colonies were isolated by subculturing on fresh WLN plates. DNA was extracted from pure colonies using the

MasterPure™ Yeast DNA Purification Kit (Epicentre, Madison, WI) following the manufacturer’s instructions. The 26S rRNA D1/D2 domain was amplified using primers NL1/4 (Kurtzman and

Robnett, 1998) for sequencing by AGRF. Species identity was determined using BLAST hosted by the NCBI (http://blast.ncbi.nlm.nih.gov/Blast.cgi), considering an identity threshold of at least 98%.

The colonies identified as S. cerevisiae were analysed by microsatellites as described in the following paper. Sequence data was uploaded to Genbank with accession numbers MT821080-MT821098.

5.2.5 Microsatellite characterisation

A set of 94 S. cerevisiae strains, listed in Supplementary Table S1, was characterised for allelic variation at 12 microsatellites as previously described (Legras et al., 2007). Briefly, two multiplex assays of primers corresponding to loci C5, C3, C8, C11, SCYOR267c and C9, YKL172w, ScAAT1,

C4, SCAAT5, C6, YPL009c, were amplified using the QIAGEN Multiplex PCR Kit according to the

111 manufacturer’s instructions. PCRs were run in a final volume of 12.5 μL containing 10 – 250 ng yeast

DNA, and used the following program: initial denaturation at 95°C for 15 min, followed by 34 cycles of 94°C for 30 s, 57°C for 2 min, 72°C for 1 min, and a final extension at 60°C for 30 min. PCR products were sized for 12 microsatellite loci on an ABI 310 DNA sequencer (Applied Biosystems) using the size standards HD400ROX. Allele distribution into classes was carried out using Genious software v9.1.6 (Biomatters) and the corresponding alleles classes were described in Legras et al.

2007.

5.2.6 Data analysis

Alpha diversities of fungal communities, yeast populations, and wine volatile compounds were calculated using the Shannon index with the “vegan” package (Oksanen et al., 2007). One-way analysis of variance (ANOVA) was used to determine whether sample classifications (e.g., fermentation stage, wine estate) contained statistically significant differences in the alpha-diversity.

Principal coordinate analysis (PCoA) was performed to evaluate the distribution patterns of fungal communities, yeast populations, and wine aroma based on beta-diversity calculated by the Bray–

Curtis distance with the “labdsv” package (Roberts, 2007). Permutational multivariate analysis of variance (PERMANOVA) using distance matrices with 999 permutations was conducted within each sample class to determine the statistically significant differences with “adonis” function in “vegan”

(Oksanen et al., 2007). Significant taxonomic differences of fungi in the BF must between sample categories (wine estate, grape variety) were tested using linear discriminant analysis (LDA) effect size

(LEfSe) analysis (Segata et al., 2011) (https://huttenhower.sph.harvard.edu/galaxy/). The OTU table was filtered to include only OTUs > 0.01% relative abundance to reduce LEfSe complexity. The factorial Kruskal–Wallis sum-rank test (α = 0.05) was applied to identify taxa with significant differential abundances between groups (all-against-all comparisons), followed by the logarithmic

LDA score (threshold = 2.0) to estimate the effect size of each discriminative feature. Significant taxa were used to generate taxonomic cladograms illustrating differences between sample categories.

Dynamics and successions of the dominant species (relative abundance > 1.00%) demonstrating

112 significant differences among stages (ANOVA; FDR-corrected p value < 0.05) were illustrated by alluvial diagrams using the “ggalluvial” package (Brunson, 2018) in “ggplot2”.

Multilocus genotypes were calculated with the “poppr” package (Kamvar et al., 2014). The Bruvo’s distance (Bruvo et al., 2004) was calculated between each strain by the “ape” (Paradis et al., 2004) and “poppr” packages. The tree was obtained from the distance matrices with “poppr”, and drawn using MEGA X (Kumar et al., 2018). The tree was rooted by the midpoint method. An alternative method of genetic clustering- discriminant analysis of principal components (DAPC) was also applied to infer the population structure with “adegenet” package (Jombart et al., 2010). Analysis of molecular variance (AMOVA) was conducted to determine the degree of differentiation of variations between designated partitions (estates, vineyards, varieties) with “poppr” (Excoffier et al., 1992). The genetic distance FST (Reynolds et al., 1983) between estates were calculated by the “hierfstat” package

(Goudet, 2005). PCoA based on the Bruvo’s distance was conducted to evaluate the geographic distribution pattern of S. cerevisiae populations with “vegan” (Oksanen et al., 2007).

Partial least squares regression (PLSR) analysis with cross-validation was used to model associations between normalised mean values for fungal taxa (species with relative abundance > 0.10%) and wine volatile compounds using the “pls” package (Wehrens and Mevik, 2007). The structural equation model (SEM) (Grace, 2006) was used to evaluate the direct and indirect relationships between must fungal communities and yeast populations, S. cerevisiae populations, and resulting wine aroma (the first axis values of PCoA analysis). SEM is an a priori approach partitioning the influence of multiple drivers in a system to help characterise and comprehend complex networks of ecological interactions

(Eisenhauer et al., 2015). An a priori model was established based on the known effects and relationships among drivers of distribution patterns of wine aroma to manipulate the data before modelling (including must aroma, °Brix, pH; data not shown). A path coefficient described the strength and sign of the relationship between two variables (Grace, 2006). The good fit of the model was validated by the χ2-test (p > 0.05), using the goodness of fit index (GFI > 0.90) and the root MSE of approximation (RMSEA < 0.05) (Schermelleh-Engel et al., 2003). Standardised total effects of

113 each factor on the wine aroma distribution pattern were calculated by summing all direct and indirect pathways between two variables (Grace, 2006). All these analyses were conducted using AMOS v25.0 (AMOS IBM, NY, USA).

5.3 Results

5.3.1 Fungal microbiota vary by geographical origin and grape variety

To elucidate the influences of geographic locations, grape variety, and fermentation process on the wine microbiota, 66 duplicate samples covering three wine estates, Pinot noir and Chardonnay, from the beginning, middle and end of fermentation were collected to analyse fungal communities. A total of 1,566,576 ITS high-quality sequences were generated from all samples, which were clustered into

277 fungal OTUs with a threshold of 97% pairwise identity. Ascomycota was the most abundant phylum in the grape must/juice before fermentation (BF) comprising 92.00% of all sequences, followed by Basidiomycota (7.98%) and Mortierellomycota (< 0.01%) (Data not shown). Fungal profiles were dominated by filamentous fungi, mostly of the genera Cladosporium, Aureobasidium, and Epicoccum, with notable populations of fermentative yeasts including Saccharomyces and

Hanseniaspora, as well as the basidiomycetous yeasts Vishniacozyma, Rhodotorula, and

Sporobolomyces (Fig. 1A).

Individual vineyards and wine estates were distinguished based on the microbial community present in the grape must/juice before fermentation (Fig. 1B; Supplementary Table S2). Permutational multivariate analysis of variance (PERMANOVA) based on the Bray–Curtis distance confirmed that fungal composition was significantly different between estates (R2 = 0.310, p < 0.001) and at least two vineyards (R2 = 0.573, p < 0.001) regardless of grape variety. Pinot Noir demonstrated stronger

2 2 geographical differentiation (PERMANOVA, R Estate = 0.683, p < 0.001; R Vineyard = 0.779, p < 0.001)

2 2 than Chardonnay (PERMANOVA, R Estate = 0.470, p < 0.001; R Vineyard = 0.566, p < 0.001). Grape variety weakly but significantly impacted fungal composition (PERMANOVA, R2 = 0.103, p =

0.021). This impact was more distinct with an improved coefficient of determination (R2) within a certain wine estate (Fig. 1B; Supplementary Table S2). Principal coordinate analysis (PCoA) showed

114

Fig. 1 Musts before fermentation (BF) have differential fungal communities depending geographical origins and grape varieties. (A) Microbial community composition characterised to the genus level (11 genera, relative abundance > 1.00% shown); (B) Principal coordinate analysis (PCoA) based on Bray-Curtis distances among wine estates across both varieties; (C) Linear discriminant analysis (LDA) effect size (LEfSe) taxonomic cladogram identifying significantly discriminant (Kruskal–Wallis sum-rank test α < 0.05; LDA score > 2.00) taxa associated with wine estates; (D) LEfSe taxonomic cladogram identifying significantly discriminant taxa associated with grape varieties.

115 the geographic patterns (95% confidence interval based on estates) of fungal communities across both varieties, with 65.0% of total variance explained by the first two principal coordinate (PC) axes (Fig.

1B). Linear discriminant analysis (LDA) effect size (LEfSe) analysis further confirmed that these patterns related to significant associations (Kruskal–Wallis sum-rank test, α < 0.05) between fungal taxa and wine estates (Fig. 1C), and grape varieties (Fig. 1D), respectively. Microbotryomycetes, notably Rhodotorula babjevae and Sporobolomyces roseus, were observed with higher abundances in the grape must/juice from wine estate A, with Saccharomycetaceae (notably fermentative yeast S. cerevisiae), Tremellomycetes (including Filobasidium spp.), Cladosporium grevilleae,

Mycosphaerellaceae, Stemphylium, Didymella from wine estate B, and Capnodiales (notably

Cladosporium spp.), Pleosporales (including Didymellaceae, Alternaria, Epicoccum,

Sporormiaceae), Exobasidiomycetes, Leotiomycetes (including Botrytis), Sordariomycetes (including

Cystofilobasidiales, Trichosphaeriales, Diaporthales) from wine estate C. For varietal influences,

Phaeomoniellales including Phaeomoniellaceae and Entylomataceae, Dothiora, Epicoccum brasiliense, Sporobolomyces phaffii, Blastobotrys, and Ilyonectria were significantly abundant in

Chardonnay juice, while Botryosphaeriales, Capnodiales (in particular Cladosporium ramotenellum),

Dothideales (in particular Aureobasidium pullulans), Eurotiales (including Penicillium),

Saccharomycetes (in particular Hanseniaspora spp.), Microbotryomycetes_ord_Incertae_sedis,

Boeremia, Diaporthaceae, Sarocladium, and Amphisphaeriaceae in Pinot Noir musts.

5.3.2 Microbiota dynamics during spontaneous wine fermentation

Significant decreases in the fungal diversity (ANOVA, p < 0.001) were recorded between fermentation stages regardless of grape variety (Fig. 2A, Supplementary Fig. S2A). PCoA showed that ferments were grouped according to their fermentative stage, where PC1 explained 66.0% and

57.3% of the total variance for Pinot Noir (Fig. 2B; PERMANOVA, R2 = 0.648, p < 0.001) and

Chardonnay (Supplementary Fig. S2B; PERMANOVA, R2 = 0.608, p < 0.001), respectively. BF samples were clearly separated from MF and EF, while the latter groups were partially overlapped.

116

Fig. 2 Stage of fermentation influences microbial diversity and composition of Pinot Noir. (A) α-diversity (Shannon index) significantly decreases (p < 0.001***) during the fermentation; (B) Bray-Curtis distance PCoA of fungal communities according to the fermentation stage (BF, before fermentation; MF, at the middle of fermentation; EF, at the end of fermentation); (C) Relative abundance changes of major fungal species (9 species; relative abundance > 1.00%) that significantly differed by fermentation stage (ANOVA; FDR-corrected p value < 0.05).

117 Within grape varieties, tracking major species across vineyards and estates (relative abundance >

1.00%) revealed fungal dynamics and succession during fermentation. All these species shown in the alluvial diagrams presented significantly different relative abundances as the fermentation progressed

(ANOVA; FDR-corrected p value < 0.05) (Fig. 2C, Supplementary Fig. S2C). BF musts/juice displayed diverse and variable collections of fungi, of which A. pullulans and C. ramotenellum were the most abundant species for Pinot Noir and Chardonnay, respectively. In the beginning of Pinot

Noir fermentation, filamentous fungi and non-Saccharomyces yeasts dominated 89.9% of the community, with 5.45% relative abundance for S. cerevisiae. After fermentation started, S. cerevisiae grew and gradually dominated the community, occupying 66.5% of the MF community and 80.9% of the EF community, while other species underwent drastic decreases in relative abundances during the fermentation (Fig. 2C). Likewise, this succession pattern was also observed in Chardonnay ferments, except that few taxa (A. pullulans, R. babjevae, Hanseniaspora lachancei) showed slight recoveries in relative abundances between MF and EF (Supplementary Fig. S2C). Correspondingly, geographical differences in fungal communities were not significant based on estates in both MF (PERMANOVA,

R2 = 0.162, p = 0.104) and EF (PERMANOVA, R2 = 0.145, p = 0.162) ferments, although fungi differentiated vineyard origin of some vineyards in both stages (p < 0.001) (Supplementary Table S2).

5.3.3 Yeast population dynamics during spontaneous wine fermentation

To estimate yeast population dynamics during spontaneous fermentations, we isolated colonies based on morphology on WLN medium and identified yeasts using the taxonomically distinctive 26S rRNA

D1/D2 region. A total of 359 yeast isolates were obtained (Supplementary Table S1), corresponding to 14 species: Hanseniaspora uvarum, Hanseniaspora opuntiae, Torulaspora delbrueckii,

Metschnikowia andauensis, Meyerozyma guilliermondii, Candida africana, Candida intermedia,

Candida oleophila, Candida ishiwadae, Rhodotorula mucilaginosa, Pichia membranifaciens,

Wickerhamomyces anomalus, Saccharomyces bayanus, and S. cerevisiae (Supplementary Table S3).

All of these species had also been identified with ITS amplicon sequencing (give figure/table number). S. cerevisiae and the non-Saccharomyces species of H. uvarum, H. opuntiae, T. delbrueckii,

M. andauensis dominated the isolates, whereas other species appeared sporadically. At the beginning

118 of fermentation, the grape must/juice harboured high species diversity (Shannon index) of yeasts, with non-Saccharomyces yeasts dominating the populations (Supplementary Table S3). The amount and distribution of yeast species differed among vineyards, estates, and varieties (Supplementary Table

S3), with significant geographical differences observed in population compositions (PERMANOVA

2 2 based on Bray–Curtis distance, R Estate = 0.287, p = 0.019; R Vineyard = 0.426, p = 0.002). As fermentation proceeded, the viable population of yeast in the must increased from initial values of

104–106 to 107–108 CFU/mL in the middle fermentation (MF) and declined coinciding with decreased species diversity during fermentation (Supplementary Table S3). Non-Saccharomyces H. uvarum and

H. opuntiae were isolated throughout some fermentations, with T. delbrueckii isolated at the beginning and middle fermentation points. In spite of low initial abundance, S. cerevisiae dominated the ferments at middle and end fermentation points and occupied 100% of some yeast populations from wine estate B (Supplementary Table S3).

From the isolated yeasts, 94 S. cerevisiae strains were further analysed through 12 microsatellite loci, resulting in 80 multilocus genotypes, with 14 strains showing genotypes identical to others in this study. The 12 microsatellite loci recorded from 4 to 23 different alleles per locus, of which C5 and

SCAAT1 displayed the highest diversity with 24 and 21 alleles, respectively. Two strains were shared between vineyards and varieties, while other identical strains were isolated from within a single vineyard and are likely the same strain. The neighbour-joining tree built from the Bruvo’s distance showed clustering linked to the geographical origin of strains (Fig. 3A). Some branches clustered isolates from one wine estate with very close genetic relationships as group Ⅰ (A), group Ⅱ (B), and group Ⅲ (C). Some strains from estate A and B gathered in clusters, with only few strains from estate

C (group Ⅳ, Ⅴ). Group Ⅵ was composed of clusters originating from all three estates. The relationships among S. cerevisiae strains inferred using DAPC were consistent with the genetic tree

(Fig. 3B). The total amount of genetic variation explained by the first 35 principal components was

93.8%, of which 24.12% was conserved by the first two axes. Most populations were clearly separated into estate A, B, and C, with overlaps observed among groups (Fig. 3B). Strains V1.2.1,

V9.1.2, and V11.2.4 among three estates were coinciding with group VI in the phylogenetic tree, as

119

Fig. 3 Saccharomyces cerevisiae populations show geographic clustering. (A) Neighbour-joining tree showing the clustering of 94 S. cerevisiae strains isolated from three wine estates. The tree was constructed from the Bruvo’s distance between strains based on the polymorphism at 12 loci and is rooted according to the midpoint method. Branches are coloured according to the wine estate from which strains have been isolated. (B) Scatterplot from discriminant analysis of principal components (DAPC) of the first two principal components discriminating S. cerevisiae populations by estates. Points/diamonds represent individual observations, and lines represent population memberships. Inertia ellipses represent an analog of a 67% confidence interval based on a bivariate normal distribution. Number of principal components (n = 35) at which the maximal reassignment of samples occurred are depicted as black lines the PCA graph on the topright corner, with subsequent components in grey line.

120 well as strains V4.3.4 and V7.2.6 between estate A and B clusters with group IV (Fig.3). AMOVA confirmed significant influences of geographic origins (estate/vineyard, p = 0.001) and the grape variety (p = 0.023) on the S. cerevisiae population structure based on, in particular based on estates

(Supplementary Table S4). Within estates, the difference between vineyards or varieties was not significant (Supplementary Table S4). Higher differentiation between was observed between wine estates A and C (12 km; FST = 0.143, p < 0.001), B and C (10 km; FST = 0.133, p < 0.001), than between A and B (8 km; FST = 0.076, p < 0.001) (Supplementary Table S4).

5.3.4 Aroma profiles are distinctive for wines of each geographical origin

Using GC-MS, we analysed the volatile compounds of Pinot Noir and Chardonnay wine samples (at the end of fermentation) in triplicate to represent wine metabolite profiles. In all, 79 volatile compounds were identified in these wines. Pinot Noir wines contained 37 geographically differential compounds based on wine estates, and Chardonnay wines contained 46 (Supplementary Table S5).

Within grape varieties, wine complexity (as determined by Shannon index) was not significantly different amongst wine estates (ANOVA; FPinot Noir (2, 12) = 2.393, p = 0.161; FChardonnay (2, 15) = 2.313, p =

0.284). Wine aroma profiles are clearly separated with PCoA based on Bray–Curtis dissimilarity according to estates, where PC1 explained 65.8% and 62.7% of the total variance for Pinot Noir wines

(PERMANOVA, R2 = 0.722, p = 0.011) and Chardonnay wines (PERMANOVA, R2 = 0.773, p =

0.002), respectively (Fig. 4).

Fig. 4 Wine metabolites exhibit geographical variation. PCoA based on Bray-Curtis dissimilarity obtained from comparing volatile profiles of Pinot Noir and Chardonnay wines.

121 5.3.5 Fungal microbiota correlate to wine aroma profiles

To elucidate the relationship between geographically differential fungal microbiota and wine metabolites, partial least squares regression (PLSR) was used to model covariance between fungal species and volatile compounds. PSLR projections were made with dominant species in the grape must (relative abundance > 0.01% across samples; Pinot Noir, 20 species; Chardonnay, 23 species) and volatile compounds in the resulting wines. Compounds shown in the plots explained > 20% of the variance in the first two components (Fig. 5), of which many were the same compounds identified by

ANOVA (Supplementary Table S5, S6). PLSR showed highly covariable relationships between fermentative yeasts and volatile compounds. In Pinot Noir wines, some interesting correlations were observed; for example S. cerevisiae and T. delbrueckii correlated strongly with several esters (C33,

Ethyl octoate; C26, Ethyl lactate; C55, 3-Methylbutyl octanoate; C38, Ethyl 6-heptenoate); M. guilliermondii with monoterpenes (C43, linalool; C48, terpinen-4-ol; C68, nerol; C59, α-terpineol), β- damascenone (C70), nonanals (C41, 2-nonanol; C56, 1-nonanol) and the related ester (C42, ethyl nonanoate); and H. uvarum and H. lachancei with alcohols (C46, 2,3-butanediol; C63, 1-decanol) and esters (C62, methyl salicylate; C76, ethyl tetradecanoate). In Chardonnay wines, S. cerevisiae was associated with some esters (for example, C33; C55; C16, ethyl hexanoate; C61, benzyl acetate; C20, hexyl acetate), T. delbrueckii and M. guilliermondii with some alcohols (C63; C44, 1-octanol) and esters (C26; C38; C42; C76), and H. uvarum and H. lachancei with C46, C62, and fatty acids (C45, 2- methyl-propanoic acid; C72, hexanoic acid; C77, octanoic acid).

To disentangle the role of microbial communities on wine metabolites, we used structural equation modelling (SEM) (Grace, 2006) to testify fungal community compositions (the first axis of PCoA) at multiple levels simultaneously: fungal communities, yeasts (S. cerevisiae and non-Saccharomyces yeasts), and S. cerevisiae populations. The SEM explained 84.9% of the variance found in the geographical pattern of wine aroma (Fig. 6). Fungal communities in the must drove wine aroma profiles directly (path coefficient = 0.286***) and indirectly by effects on yeasts and S. cerevisiae populations, in particular strong influences on yeast populations (path coefficient = 0.562). S. cerevisiae populations had the highest direct positive effects on the resulting wine aroma

122

Fig. 5 Partial least squares regression (PLSR) demonstrates microbial influence on volatile compounds of Pinot Noir and Chardonnay wines.

123 characteristics, while yeast populations had the lowest but significant effects (path coefficient = -

0.142**) (Fig. 6A). Overall, S. cerevisiae populations were the most important driver of wine characteristics (Fig. 6B).

Fig. 6 Direct and indirect effects of fungal community composition (the first axis of PCoA) at multiple levels on wine aroma profiles. Structural equation model (SEM) fitted to wine aroma composition (A) and standardised total effects (direct plus indirect effects) derived from the model

(B).

5.4 Discussion

There is mounting evidence for geographical differentiation of wine-related microbial communities at regional scales (Bokulich et al., 2014; Gayevskiy and Goddard, 2012; Jara et al., 2016; Pinto et al.,

2015; Taylor et al., 2014). Our previous work revealed that different wine-producing regions in southern Australia possess distinct, distinguishable microbial patterns (especially fungal microbiota) at the scale of 400 km, correlated with local weather conditions and soil properties (Liu et al., 2020).

In the current study, within a single sub-region spanning 12 km, we demonstrate geographical differentiation of grape must/juice fungal communities, yeast populations, and S. cerevisiae populations, with influences from the grape variety.

124 5.4.1 Fungal ecology at multiple layers during spontaneous wine fermentation

In the freshly crushed grape must/juice, fungal communities were highly diverse and characterised by ubiquitous genera such as Aureobasidium, Cladosporium, Saccharomyces, and Rhodotorula, deriving from the vineyard ecosystem. Geographical origin had a greater impact on the fungal community

(Supplementary Table S2) and yeast populations (results 3.2) than grape variety, and this is in line with other studies in grapevine-associated microbiota (Bokulich et al., 2014; Mezzasalma et al., 2018;

Wang et al., 2015). In particular, S. cerevisiae yeasts was one of geographical features (Fig. 1C).

Given the small scale of the vineyards in this study (< 12 km), macroclimate does not differentiate the sites, and so local conditions appear to modulate communities. Geographical features (for example vineyard orientation), microclimate, and soil properties (for example nutrient availability) could explain some variation among vineyard sites, but these measures were beyond the scope of this study.

Within a single estate, grape variety played a significant role in shaping fungal communities (Fig. 1B;

Supplementary Table S2), suggesting a genetic component to plant-microbial interactions (Bokulich et al., 2014; Mezzasalma et al., 2018). Cultivar variation in the microhabitat and environmental stress responses may explain how grapevines recruit their associated microbiota, including both the normal microbiota and cultivar-specific susceptibilities to disease pressures (Fung et al., 2008). Our results show that differential taxa between varieties are ubiquitous fungi in agricultural ecosystems, and some of these taxa are considered to be grapevine pathogens, for example C. ramotenellum and

Mycosphaerella tassiana (Bensch et al., 2012), were more abundant in Pinot Noir musts (Fig. 1D).

Geographical signatures diminished during spontaneous wine fermentation as growth of fermentative yeasts reshaped the community diversity and composition, in particular S.cerevisiae (Fig. 2, S2;

Supplementary Table S2, S3). Regardless of grape variety, fungal and yeast species diversity collapsed as alcoholic fermentation progressed, with a loss of the environmental fungi (Fig. 2A, S2A;

Supplementary Table S3), indicating evolution through selection associated with wine fermentation.

Non-Saccharomyces yeasts start the fermentation process (especially genera Hanseniaspora,

Candida, Pichia and Metschnikowia), but they are quickly replaced by S.cerevisiae that lead fermentation until the end (Capozzi et al., 2015; Fleet, 2003). The inability of non-Saccharomyces

125 yeasts to sustain their presence in ferments has been attributed to their oxidative and weakly fermentative metabolism, their sensitivity to increasing fermentation rate, ethanol and heat production induced by S.cerevisiae growth, and being less tolerant to low oxygen availability (Bokulich et al.,

2016; Combina et al., 2005; Goddard, 2008). Thus, fermentation conditions, such as the chemical environment and interactions within the community drive the microbial pattern into a population dominated by S. cerevisiae (Fig. 2C, S2C; Supplementary Table S3) (Liu et al., 2017).

Spontaneous wine fermentation is characterised not only by significant intraspecific biodiversity

(Cocolin et al., 2000), but also by high genetic polymorphism in the S.cerevisiae population present during the fermentation. Biogeography of S.cerevisiae has been previously reported from regional

(Gayevskiy and Goddard, 2012; Knight and Goddard, 2015) and global (Legras et al., 2007; Liti et al.,

2009) scales to small scales between vineyards within a single region (Börlin et al., 2016; Schuller and Casal, 2007), while Knight et al. (2019) demonstrated no vineyard demarcation between S. cerevisiae that could be ascribed to gene flow. Here, we show that distinct geographical differentiation between wine estates, with increased genetic divergence with distance (Schuller and

Casal, 2007). A certain degree of mixed strains from various vineyard sites indicates gene flow among populations at small scales. Within the wine estates, the vineyards are within a 5 km radius from one another, and thus insect vectors like honeybees, wasps, and fruit flies, as well as birds, may homogenise the yeast populations (Francesca et al., 2012; Goddard et al., 2010; Lam and Howell,

2015; Stefanini et al., 2012). Shared staff and agricultural implements may also facilitate the movement and exchanges of S.cerevisiae populations between vineyards managed by the same estate

(Goddard et al., 2010). Harvested grapes from various vineyards within the estate were processed at the same winery, where equipment surfaces may harbour large populations of S.cerevisiae and other yeasts under normal cleaning conditions, acting as potential reservoir for wine microbiota during spontaneous fermentation (Bokulich et al., 2013; Ciani et al., 2004; Sabate et al., 2002). All these factors contribute to the estate-specific pattern of S.cerevisiae distribution.

126 5.4.2 Association between fungal microbiota and composition of the wine metabolome

Previous research suggested that fungal microbiota structured and distinguished vineyard ecosystems impacting the aroma and quality of wine at the regional scale (Liu et al., 2020). Here we show that the geographical diversification observed in fungal communities, yeasts, and S.cerevisiae could translate to aromatic differences in wines within a single region. Wine aroma profiles highly associated with fermentative yeasts, in particular, S.cerevisiae displayed the most important effects on wine characteristics at this scale (Fig. 5, 6). Other fungal species, although non-fermentative (or not known to be associated) with wine fermentation, could produce some sensory-active compounds associated with wine aroma formation (Verginer et al., 2010), or substantially modulate vine health, growth, and fruit quality in the vineyard (Berg et al., 2014; Gilbert et al., 2014). The most widespread fungi A. pullulans are known to have an antagonistic effect on mould development, like Botrytis cinerea, causing grey rot, and Aspergillus spp., producing ochratoxin (Barata et al., 2012). Further, host- microbe interactions could promote plant resistance to environmental/abiotic stress thus benefiting crop production (Berg, 2009; Lugtenberg and Kamilova, 2009).

In the resultant wines, most volatile compounds were esters, alcohols, acids and aldehydes, many of which are fermentation-derived products. Some compounds were grape-derived, for example monoterpenes, and potentially modified by microbial metabolism during fermentation (Swiegers et al., 2005). Non-Saccharomyces yeasts dominating the initial spontaneous fermentation can contribute to the overall wine aroma profiles by producing flavour-active compounds, which depends on yeast species and strains (Capozzi et al., 2015; Jolly et al., 2014). Here, our work shows the presence of T. delbrueckii, M. guilliermondii, and Hanseniaspora spp. correlated with some volatile compounds, such as higher alcohols, ethyl esters, and acids (Fig. 5), thus potentially affecting wine characteristics.

M. guilliermondii, which is known to produce β-glucosidases to release bound terpenoids (Silva et al.,

2005), associated with monoterpenes including linalool, terpinen-4-ol, nerol, and α-terpineol, thus enhancing varietal aroma in wine and local terroir expression. Beyond fermentative contributions, the presence of non-Saccharomyces yeasts and the species diversity indirectly affect wine ecosystem function by altering the ecological dominance of S. cerevisiae through antagonistic interactions in

127 early succession (Bagheri et al., 2017; Boynton and Greig, 2016). S. cerevisiae eventually dominate the fermentation, where they are naturally initially rare, by modifying the environment through fermentation, and drive wine ecosystem function (Boynton and Greig, 2016; Goddard, 2008). S. cerevisiae show genetic diversity at strain level and it is well documented that different genotypes produce variable amounts of volatile compounds as fermentation by-products, with desirable or undesirable impacts on wine aroma and flavour (Capece et al., 2012; Howell et al., 2004; Pretorius,

2000; Romano et al., 2003). Knight et al. (2015)found that S. cerevisiae genotypes and wine phenotypes were correlated with geographic dispersion, and regional populations produced distinct aroma profiles in Sauvignon Blanc wine fermentation. Our results further suggest that within a single region, wine aroma profiles were affected by the geographical origin and genomics of S. cerevisiae natural strains. As the keystone driver of the wine ecosystem, S. cerevisiae was associated with many ethyl esters and acetate esters, exerting the most powerful influences of multiple layers of fungal microbiota on wine quality and style (Fig. 5, 6). The negative effect of yeast populations on wine aroma characteristics is likely due to the inhibiting of diverse non-Saccharomyces yeasts on the ecological dominance of S. cerevisiae (Bagheri et al., 2017; Boynton and Greig, 2016; Goddard,

2008) (Supplementary Table S3). A causative relationship between the geographically differentiated fungi, yeasts, and S. cerevisiae should be established to show the impact on wine compounds for regional differentiation based on sensory characteristics.

5.5 Conclusions

Our study describes the diversity of fungal communities during spontaneous wine fermentation in carefully selected vineyards comprising two cultivars. As predicted, we observed ecological dominance Saccharomyces spp., but showed that geographical diversification is evident in the initial fungal community composition and the strain level diversity of S. cerevisiae. Fungal species correlated with wine volatile compounds, of which S. cerevisiae is likely the primary driver of wine aroma and characteristics within the sub-region (less than 12 km). A better understanding of how multiple layers of fungal microbiota vary at different scales, and the effects of these communities on

128 agricultural ecosystems, provides perspectives for sustainable management practices maintaining the biodiversity and functioning thus optimising plant food and beverage production.

Declaration of competing interest We declare that this research was conducted in the absence of any commercial or financial relationships that could be constructed as a potential conflict of interest.

Acknowledgements We give our sincere thanks to the vignerons who kindly enabled sampling and provided wine samples. DL acknowledges support from a Ph.D. scholarship and funding from Wine Australia (AGW

Ph1602) and a Melbourne Research Scholarship from the University of Melbourne. We would also like to acknowledge Master student Haoran Liu at the University of Melbourne for assistance in sample collection and yeast culture experiments.

129 5.6 References

Bagheri, B., Bauer, F.F., Setati, M.E., 2017. The impact of Saccharomyces cerevisiae on a wine yeast consortium in natural and inoculated fermentations. Front. Microbiol. 8, 1988.

Barata, A., Malfeito-Ferreira, M., Loureiro, V., 2012. The microbial ecology of wine grape berries. Int. J. Food Microbiol. 153, 243-259.

Bensch, K., Braun, U., Groenewald, J.Z., Crous, P.W., 2012. The genus cladosporium. Stud. Mycol. 72, 1-401.

Berg, G., 2009. Plant–microbe interactions promoting plant growth and health: perspectives for controlled use of microorganisms in agriculture. Appl. Microbiol. Biotechnol. 84, 11-18.

Berg, G., Grube, M., Schloter, M., Smalla, K., 2014. Unraveling the plant microbiome: looking back and future perspectives. Front. Microbiol. 5, 148.

Bokulich, N.A., Collins, T.S., Masarweh, C., Allen, G., Heymann, H., Ebeler, S.E., Mills, D.A., 2016. Associations among Wine Grape Microbiome, Metabolome, and Fermentation Behavior Suggest Microbial Contribution to Regional Wine Characteristics. MBio 7, e00631-00616.

Bokulich, N.A., Ohta, M., Richardson, P.M., Mills, D.A., 2013. Monitoring seasonal changes in winery-resident microbiota. PLoS One 8, e66437.

Bokulich, N.A., Thorngate, J.H., Richardson, P.M., Mills, D.A., 2014. Microbial biogeography of wine grapes is conditioned by cultivar, vintage, and climate. Proc. Natl. Acad. Sci. USA 111, E139- E148.

Börlin, M., Venet, P., Claisse, O., Salin, F., Legras, J.-L., Masneuf-Pomarede, I., 2016. Cellar- associated Saccharomyces cerevisiae population structure revealed high diversity and perennial persistence in Sauternes wine estates. Appl. Environ. Microbiol., 03627-03615.

Boynton, P.J., Greig, D., 2016. Species richness influences wine ecosystem function through a dominant species. Fungal Ecol. 22, 61-72.

Brunson, J.C., 2018. ggalluvial: Alluvial diagrams in'ggplot2'. R package version 0.9.1.

Bruvo, R., Michiels, N.K., D’SOUZA, T.G., Schulenburg, H., 2004. A simple method for the calculation of microsatellite genotype distances irrespective of ploidy level. Mol. Ecol. 13, 2101- 2106.

130 Capece, A., Romaniello, R., Siesto, G., Romano, P., 2012. Diversity of Saccharomyces cerevisiae yeasts associated to spontaneously fermenting grapes from an Italian “heroic vine-growing area”. Food Microbiol. 31, 159-166.

Caporaso, J.G., Kuczynski, J., Stombaugh, J., Bittinger, K., Bushman, F.D., Costello, E.K., Fierer, N., Pena, A.G., Goodrich, J.K., Gordon, J.I., 2010. QIIME allows analysis of high-throughput community sequencing data. Nat. Methods 7, 335.

Capozzi, V., Garofalo, C., Chiriatti, M.A., Grieco, F., Spano, G., 2015. Microbial terroir and food innovation: the case of yeast biodiversity in wine. Microbiol. Res. 181, 75-83.

Ciani, M., Comitini, F., Mannazzu, I., Domizio, P., 2010. Controlled mixed culture fermentation: a new perspective on the use of non‐Saccharomyces yeasts in winemaking. FEMS Yeast Res. 10, 123- 133.

Ciani, M., Mannazzu, I., Marinangeli, P., Clementi, F., Martini, A., 2004. Contribution of winery- resident Saccharomyces cerevisiae strains to spontaneous grape must fermentation. Antonie Van Leeuwenhoek 85, 159-164.

Cocolin, L., Bisson, L., Mills, D., 2000. Direct profiling of the yeast dynamics in wine fermentations. FEMS Microbiol. Lett. 189, 81-87.

Combina, M., Elía, A., Mercado, L., Catania, C., Ganga, A., Martinez, C., 2005. Dynamics of indigenous yeast populations during spontaneous fermentation of wines from Mendoza, Argentina. Int. J. Food Microbiol. 99, 237-243.

Drumonde-Neves, J., Franco-Duarte, R., Vieira, E., Mendes, I., Lima, T., Schuller, D., Pais, C., 2018. Differentiation of Saccharomyces cerevisiae populations from vineyards of the Azores Archipelago: Geography vs ecology. Food Microbiol. 74, 151-162.

Edgar, R.C., 2010. Search and clustering orders of magnitude faster than BLAST. Bioinformatics 26, 2460-2461.

Edgar, R.C., Haas, B.J., Clemente, J.C., Quince, C., Knight, R., 2011. UCHIME improves sensitivity and speed of chimera detection. Bioinformatics 27, 2194-2200.

Eisenhauer, N., Bowker, M.A., Grace, J.B., Powell, J.R., 2015. From patterns to causal understanding: structural equation modeling (SEM) in soil ecology. Pedobiologia 58, 65-72.

Excoffier, L., Smouse, P.E., Quattro, J.M., 1992. Analysis of molecular variance inferred from metric distances among DNA haplotypes: application to human mitochondrial DNA restriction data. Genetics 131, 479-491.

131 Fleet, G.H., 2003. Yeast interactions and wine flavour. Int. J. Food Microbiol. 86, 11-22.

Francesca, N., Canale, D.E., Settanni, L., Moschetti, G., 2012. Dissemination of wine‐related yeasts by migratory birds. Environ. Microbiol. Rep. 4, 105-112.

Fung, R.W.M., Gonzalo, M., Fekete, C., Kovacs, L.G., He, Y., Marsh, E., McIntyre, L.M., Schachtman, D.P., Qiu, W., 2008. Powdery Mildew Induces Defense-Oriented Reprogramming of the Transcriptome in a Susceptible But Not in a Resistant Grapevine. Plant Physiol. 146, 236-249.

Gayevskiy, V., Goddard, M.R., 2012. Geographic delineations of yeast communities and populations associated with vines and wines in New Zealand. The ISME journal 6, 1281-1290.

Gilbert, J.A., van der Lelie, D., Zarraonaindia, I., 2014. Microbial terroir for wine grapes. Proc. Natl. Acad. Sci. USA 111, 5-6.

Goddard, M.R., 2008. Quantifying the complexities of Saccharomyces cerevisiae's ecosystem engineering via fermentation. Ecology 89, 2077-2082.

Goddard, M.R., Anfang, N., Tang, R., Gardner, R.C., Jun, C., 2010. A distinct population of Saccharomyces cerevisiae in New Zealand: evidence for local dispersal by insects and human‐aided global dispersal in oak barrels. Environ. Microbiol. 12, 63-73.

Goudet, J., 2005. Hierfstat, a package for R to compute and test hierarchical F‐statistics. Mol. Ecol. Notes 5, 184-186.

Grace, J.B., 2006. Structural equation modeling and natural systems. Cambridge University Press.

Howell, K.S., Cozzolino, D., Bartowsky, E.J., Fleet, G.H., Henschke, P.A., 2006. Metabolic profiling as a tool for revealing Saccharomyces interactions during wine fermentation. FEMS Yeast Res. 6, 91- 101.

Howell, K.S., Swiegers, J.H., Elsey, G.M., Siebert, T.E., Bartowsky, E.J., Fleet, G.H., Pretorius, I.S., de, M.A., 2004. Variation in 4-mercapto-4-methyl-pentan-2-one release by Saccharomyces cerevisiae commercial wine strains. FEMS Microbiol. Lett. 240, 125-129.

Jara, C., Laurie, V.F., Mas, A., Romero, J., 2016. Microbial Terroir in Chilean Valleys: Diversity of Non-conventional Yeast. Front. Microbiol. 7, 663.

Jolly, N.P., Varela, C., Pretorius, I.S., 2014. Not your ordinary yeast: non-Saccharomyces yeasts in wine production uncovered. FEMS Yeast Res. 14, 215-237.

132 Jombart, T., Devillard, S., Balloux, F., 2010. Discriminant analysis of principal components: a new method for the analysis of genetically structured populations. BMC Genet. 11, 94.

Kamvar, Z.N., Tabima, J.F., Grünwald, N.J., 2014. Poppr: an R package for genetic analysis of populations with clonal, partially clonal, and/or sexual reproduction. PeerJ 2, e281.

Knight, S., Goddard, M., 2015. Quantifying separation and similarity in a Saccharomyces cerevisiae metapopulation. The ISME journal 9, 361-370.

Knight, S., Klaere, S., Fedrizzi, B., Goddard, M.R., 2015. Regional microbial signatures positively correlate with differential wine phenotypes: evidence for a microbial aspect to terroir. Sci. Rep. 5, 14233.

Knight, S.J., Karon, O., Goddard, M.R., 2019. Small scale fungal community differentiation in a vineyard system. Food Microbiol. 87, 103358.

Kõljalg, U., Larsson, K.H., Abarenkov, K., Nilsson, R.H., Alexander, I.J., Eberhardt, U., Erland, S., Høiland, K., Kjøller, R., Larsson, E., 2005. UNITE: a database providing web‐based methods for the molecular identification of ectomycorrhizal fungi. New Phytol. 166, 1063-1068.

Kumar, S., Stecher, G., Li, M., Knyaz, C., Tamura, K., 2018. MEGA X: molecular evolutionary genetics analysis across computing platforms. Mol. Biol. Evol. 35, 1547-1549.

Kurtzman, C.P., Robnett, C.J., 1998. Identification and phylogeny of ascomycetous yeasts from analysis of nuclear large subunit (26S) ribosomal DNA partial sequences. Antonie Van Leeuwenhoek 73, 331-371.

Lam, S.S., Howell, K.S., 2015. Drosophila-associated yeast species in vineyard ecosystems. FEMS Microbiol. Lett. 362, fnv170.

Legras, J.-L., Merdinoglu, D., Cornuet, J.-M., Karst, F., 2007. Bread, beer and wine: Saccharomyces cerevisiae diversity reflects human history. Mol. Ecol. 16, 2091-2102.

Liti, G., Carter, D.M., Moses, A.M., Warringer, J., Parts, L., James, S.A., Davey, R.P., Roberts, I.N., Burt, A., Koufopanou, V., Tsai, I.J., Bergman, C.M., Bensasson, D., O’Kelly, M.J.T., van Oudenaarden, A., Barton, D.B.H., Bailes, E., Nguyen, A.N., Jones, M., Quail, M.A., Goodhead, I., Sims, S., Smith, F., Blomberg, A., Durbin, R., Louis, E.J., 2009. Population genomics of domestic and wild yeasts. Nature 458, 337.

Liu, D., Chen, Q., Zhang, P., Chen, D., Howell, K.S., 2020. The Fungal Microbiome Is an Important Component of Vineyard Ecosystems and Correlates with Regional Distinctiveness of Wine. mSphere 5, e00534-00520.

133 Liu, D., Howell, K., 2020. Community succession of the grapevine fungal microbiome in the annual growth cycle. Environ. Microbiol., doi:10.1111/1462-2920.15172.

Liu, D., Xing, R.-R., Li, Z., Yang, D.-M., Pan, Q.-H., 2016. Evolution of volatile compounds, aroma attributes, and sensory perception in bottle-aged red wines and their correlation. Eur. Food Res. Technol. 242, 1937-1948.

Liu, D., Zhang, P., Chen, D., Howell, K.S., 2019. From the vineyard to the winery: how microbial ecology drives regional distinctiveness of wine. Front. Microbiol. 10, 2679.

Liu, Y., Rousseaux, S., Tourdot-Maréchal, R., Sadoudi, M., Gougeon, R., Schmitt-Kopplin, P., Alexandre, H., 2017. Wine microbiome: a dynamic world of microbial interactions. Crit. Rev. Food Sci. Nutr. 57, 856-873.

Lugtenberg, B., Kamilova, F., 2009. Plant-growth-promoting rhizobacteria. Annu. Rev. Microbiol. 63, 541-556.

Magoč, T., Salzberg, S.L., 2011. FLASH: fast length adjustment of short reads to improve genome assemblies. Bioinformatics 27, 2957-2963.

Mezzasalma, V., Sandionigi, A., Guzzetti, L., Galimberti, A., Grando, M.S., Tardaguila, J., Labra, M., 2018. Geographical and cultivar features differentiate grape microbiota in Northern Italy and Spain vineyards. Front. Microbiol. 9, 946.

Miura, T., Sánchez, R., Castañeda, L.E., Godoy, K., Barbosa, O., 2017. Is microbial terroir related to geographic distance between vineyards? Environ. Microbiol. Rep. 9, 742-749.

Morrison-Whittle, P., Goddard, M.R., 2018. From vineyard to winery: a source map of microbial diversity driving wine fermentation. Environ. Microbiol. 20, 75-84.

Müller, D.B., Vogel, C., Bai, Y., Vorholt, J.A., 2016. The plant microbiota: systems-level insights and perspectives. Annu Rev Genet 50, 211-234.

Oksanen, J., Kindt, R., Legendre, P., O’Hara, B., Stevens, M.H.H., Oksanen, M.J., Suggests, M., 2007. The vegan package. Community ecology package 10, 631-637.

Pallmann, C.L., Brown, J.A., Olineka, T.L., Cocolin, L., Mills, D.A., Bisson, L.F., 2001. Use of WL medium to profile native flora fermentations. Am. J. Enol. Vitic. 52, 198-203.

Paradis, E., Claude, J., Strimmer, K., 2004. APE: analyses of phylogenetics and evolution in R language. Bioinformatics 20, 289-290.

134 Pinto, C., Pinho, D., Cardoso, R., Custódio, V., Fernandes, J., Sousa, S., Pinheiro, M., Egas, C., Gomes, A.C., 2015. Wine fermentation microbiome: a landscape from different Portuguese wine appellations. Front. Microbiol. 6, 905.

Portillo, M.d.C., Franquès, J., Araque, I., Reguant, C., Bordons, A., 2016. Bacterial diversity of Grenache and Carignan grape surface from different vineyards at Priorat wine region (Catalonia, Spain). Int. J. Food Microbiol. 219, 56-63.

Pretorius, I.S., 2000. Tailoring wine yeast for the new millennium: novel approaches to the ancient art of winemaking. Yeast 16, 675-729.

Reynolds, J., Weir, B.S., Cockerham, C.C., 1983. Estimation of the coancestry coefficient: basis for a short-term genetic distance. Genetics 105, 767-779.

Roberts, D.W., 2007. labdsv: Ordination and multivariate analysis for ecology. R package version 1.

Romancino, D., Di Maio, S., Muriella, R., Oliva, D., 2008. Analysis of non‐Saccharomyces yeast populations isolated from grape musts from Sicily (Italy). J. Appl. Microbiol. 105, 2248-2254.

Romano, P., Fiore, C., Paraggio, M., Caruso, M., Capece, A., 2003. Function of yeast species and strains in wine flavour. Int. J. Food Microbiol. 86, 169-180.

Sabate, J., Cano, J., Esteve-Zarzoso, B., Guillamón, J.M., 2002. Isolation and identification of yeasts associated with vineyard and winery by RFLP analysis of ribosomal genes and mitochondrial DNA. Microbiol. Res. 157, 267-274.

Schermelleh-Engel, K., Moosbrugger, H., Müller, H., 2003. Evaluating the fit of structural equation models: Tests of significance and descriptive goodness-of-fit measures. Methods Psychol. Res. Online 8, 23-74.

Schuller, D., Casal, M., 2007. The genetic structure of fermentative vineyard-associated Saccharomyces cerevisiae populations revealed by microsatellite analysis. Antonie Van Leeuwenhoek 91, 137-150.

Segata, N., Izard, J., Waldron, L., Gevers, D., Miropolsky, L., Garrett, W.S., Huttenhower, C., 2011. Metagenomic biomarker discovery and explanation. Genome Biol 12, R60.

Silva, E.G.d., Borges, M.d.F., Medina, C., Piccoli, R.H., Schwan, R.F., 2005. Pectinolytic enzymes secreted by yeasts from tropical fruits. FEMS Yeast Res. 5, 859-865.

135 Stefanini, I., Cavalieri, D., 2018. Metagenomic Approaches to Investigate the Contribution of the Vineyard Environment to the Quality of Wine Fermentation: Potentials and Difficulties. Front. Microbiol. 9, 991.

Stefanini, I., Dapporto, L., Legras, J.-L., Calabretta, A., Di Paola, M., De Filippo, C., Viola, R., Capretti, P., Polsinelli, M., Turillazzi, S., 2012. Role of social wasps in Saccharomyces cerevisiae ecology and evolution. Proc. Natl. Acad. Sci. USA 109, 13398-13403.

Swiegers, J., Bartowsky, E., Henschke, P., Pretorius, I., 2005. Yeast and bacterial modulation of wine aroma and flavour. Aust. J. Grape Wine Res. 11, 139-173.

Taylor, M.W., Tsai, P., Anfang, N., Ross, H.A., Goddard, M.R., 2014. Pyrosequencing reveals regional differences in fruit‐associated fungal communities. Environ. Microbiol. 16, 2848-2858.

Verginer, M., Leitner, E., Berg, G., 2010. Production of volatile metabolites by grape-associated microorganisms. J. Agric. Food. Chem. 58, 8344-8350.

Wang, C., García-Fernández, D., Mas, A., Esteve-Zarzoso, B., 2015. Fungal diversity in grape must and wine fermentation assessed by massive sequencing, quantitative PCR and DGGE. Front. Microbiol. 6, 1156.

Wehrens, R., Mevik, B.-H., 2007. The pls package: principal component and partial least squares regression in R.

Zarraonaindia, I., Owens, S.M., Weisenhorn, P., West, K., Hampton-Marcell, J., Lax, S., Bokulich, N.A., Mills, D.A., Martin, G., Taghavi, S., 2015. The soil microbiome influences grapevine- associated microbiota. MBio 6, e02527-02514.

Zhang, P., Barlow, S., Krstic, M., Herderich, M., Fuentes, S., Howell, K., 2015. Within-vineyard, within-vine, and within-bunch variability of the rotundone concentration in berries of Vitis vinifera L. cv. Shiraz. J. Agric. Food. Chem. 63, 4276-4283.

136 5.7 Supplementary data

Table S1 Summary of microbial samples colleceted in Mornington Peninsula wine region in 2019 Chemical constituents of Sample Fraction crushed grape musts (BF) Number of S.c.strains Wine Grape Vineyard yeast analysed by Estate variety Total °Brix pH acidity BF MF EF isolates microsatellites (g/L) V1 Pinot Noir 19.75 3.62 5.55 2 2 2 29 9 V2 Pinot Noir 19.75 3.57 5.85 2 2 2 27 8 A V3 Chardonnay 18.45 3.15 5.85 2 2 2 41 9 V4 Chardonnay 17.80 3.24 5.75 2 2 2 34 7 V5 Pinot Noir 19.55 3.73 5.00 2 2 2 40 9 V6 Pinot Noir 18.55 3.66 5.20 2 2 2 32 8 B V7 Chardonnay 18.10 3.34 5.65 2 2 2 29 9 V8 Chardonnay 17.35 3.14 5.45 2 2 2 37 8 V9 Chardonnay 17.85 3.10 5.80 2 2 2 32 9 V10 Pinot Noir 19.10 3.54 5.65 2 2 2 23 8 C V11 Chardonnay 17.50 3.23 5.00 2 2 2 35 10 Total 66 359 94 Note: Samples were collected before fermentation (BF), at the middle of fermentation (MF), and at the end of fermentation (EF).

137

Table S2 Permutational multivariate analysis of variance (PERMANOVA) using distance matrices of category effects on fungal communities in spontaneous fermentations Factor Sample Group R2 p Wine Estate BF 0.310 0.001 BF_Pinot Noir 0.683 0.001 BF_Chardonnay 0.470 0.001 MF 0.162 0.104 EF 0.145 0.162 Vineyard BF 0.573 0.001 BF_Pinot Noir 0.779 0.001 BF_Chardonnay 0.566 0.001 MF 0.305 0.001 EF 0.257 0.001 Variety BF 0.103 0.021 BF_Estate A 0.742 0.001 BF_Estate B 0.264 0.001 Stage Pinot Noir 0.648 0.001 Chardonnay 0.608 0.001 Bold values indicate statistically significant results, p < 0.05

138

Table S3 The occurrence and relative abundance (%) of yeast species during spontaneous fermentations of Pinot Noir and Chardonnay (A) Wine Estate A Yeast V1 V2 V3 V4 BF MF EF BF MF EF BF MF EF BF MF EF Population size (CFU/mL) 5E+06 9E+07 3E+07 4E+04 3E+07 1E+07 1E+06 1E+07 3E+05 2E+04 2E+07 7E+05 α-diversity (Shannon index) 0.93 0.25 0.07 1.31 0.36 0.08 1.02 0.55 0.35 1.12 0.65 0.40 Saccharomyces cerevisiae 5.40 94.44 98.73 22.22 91.25 98.45 1.54 81.75 91.18 9.76 73.06 86.11 Hanseniaspora uvarum 30.25 3.26 11.50 25.26 Hanseniaspora opuntiae 61.35 2.30 1.27 35.78 3.75 1.55 56.28 15.87 5.88 25.11 13.89 Torulaspora delbrueckii 30.50 5.00 3.08 2.38 5.85 1.83 Metschnikowia andauensis 2.80 1.54 23.90 Meyerozyma guilliermondii Candida africana 1.54 Candida intermedia 0.20 Candida oleophila 0.98 Candida ishiwadae 0.49 Rhodotorula mucilaginosa Pichia membranifaciens 2.94 Wickerhamomyces anomalus 59.02 Saccharomyces bayanus 0.77

139

(B) Wine Estate B Yeast V5 V6 V7 V8 V9 BF MF EF BF MF EF BF MF EF BF MF EF BF MF EF Population size (CFU/mL) 1E+06 9E+07 8E+07 3E+06 1E+08 1E+07 1E+06 2E+07 7E+06 1E+06 2E+07 4E+06 2E+06 3E+07 1E+07 α-diversity (Shannon index) 1.37 0.21 0.00 0.80 0.43 0.00 1.33 0.70 0.22 1.29 0.33 0.00 0.59 0.00 0.00 Saccharomyces cerevisiae 33.88 94.74 100.00 0.33 84.62 100.00 35.85 77.06 94.29 36.07 92.12 100.00 27.85 100.00 100.00 Hanseniaspora uvarum 30.26 42.10 20.75 10.12 5.71 29.53 3.94 72.15 Hanseniaspora opuntiae 28.42 55.60 29.25 25.26 3.94 Torulaspora delbrueckii 3.30 5.26 1.67 15.38 14.15 12.82 9.13 Metschnikowia andauensis 0.83 0.33 Meyerozyma guilliermondii 2.48 Candida africana Candida intermedia Candida oleophila Candida ishiwadae Rhodotorula mucilaginosa 0.83 Pichia membranifaciens Wickerhamomyces anomalus Saccharomyces bayanus

140

(C) Wine Estate C Yeast V10 V11 BF MF EF BF MF EF Population size (CFU/mL) 5E+06 1E+08 3E+06 2E+05 1E+07 1E+06 α-diversity (Shannon index) 1.21 0.59 0.24 1.03 0.19 0.08 Saccharomyces cerevisiae 6.00 81.82 93.33 15.50 95.45 98.37 Hanseniaspora uvarum 52.00 12.38 6.67 12.70 4.55 1.63 Hanseniaspora opuntiae 30.00 5.80 7.50 Torulaspora delbrueckii 64.30 Metschnikowia andauensis 6.00 Meyerozyma guilliermondii Candida africana Candida intermedia Candida oleophila 6.00 Candida ishiwadae Rhodotorula mucilaginosa Pichia membranifaciens Wickerhamomyces anomalus Saccharomyces bayanus

141

Table S4 Results of FST statistic analysis

(A) FST statistic analysis results FST p Estate 0.024 0.001 Vineyard 0.011 0.001 Variety 0.0005 0.403

(B) Pairwise FST statistic values between wine estates

FST or p value for estate* Estate A B C A 0.001 0.001 B 0.076 0.001 C 0.143 0.133

*FST values are lightface, and p values are boldface.

142

Table S5 ANOVA results of wine volatile compounds among wine estates

Retention CAS Pinot Noir Chardonnay Volatile Compound time (min) Number F(2, 12) p Pr (>F) F(2, 15) p Pr (>F) C1 3.185 Ethyl Acetate 141-78-6 3.870 0.050 1.053 0.373 C2 4.256 Ethyl propanoate 105-37-3 0.104 0.956 2.101 0.087 C3 5.509 Isobutyl acetate 110-19-0 10.065 0.003 ** 1.688 0.218 C4 6.150 Ethyl butanoate 105-54-4 2.856 0.097 3.848 0.045 * C5 7.215 Butyl acetate 123-86-4 2.856 0.097 3.849 0.045 * C6 7.285 Hexanal 66-25-1 0.620 0.608 1.243 0.320 C7 8.111 2-Methyl-1-propanol 78-83-1 6.536 0.012 * 33.415 2.98E-06 *** C8 9.056 3-Methylbutyl acetate 123-92-2 52.534 1.16E-06 *** 31.232 4.49E-06 *** C9 9.353 Ethyl pentanoate 539-82-2 2.025 0.079 0.950 0.482 C10 10.073 1-Butanol 71-36-3 7.999 0.006 ** 2.927 0.085 C11 11.027 Pentyl acetate 628-63-7 127.180 8.36E-09 *** 110.020 1.09E-09 *** C12 11.500 Isoamyl propanoate 105-68-0 1.567 0.174 1795.700 1.39E-18 *** C13 12.031 D-Limonene 5989-27-5 0.354 0.746 1.214 0.218 C14 12.709 3-Methyl-1-butanol 123-51-3 16.694 3.42E-04 *** 15.405 0.000 *** C15 13.012 4-Octanone 589-63-9 1.903 0.192 2928.600 3.59E-20 *** C16 13.606 Ethyl hexanoate 123-66-0 10.640 0.002 ** 28.799 7.31E-06 *** 4,6-Dimethyl-2- C17 13.873 19549-80-5 heptanone 110.380 1.88E-08 *** 1.246 0.287 C18 14.239 Styrene 100-42-5 14.496 6.29E-04 *** 3.281 0.066 C19 14.502 1-Pentanol 71-41-0 3.693 0.056 1.147 0.344 C20 15.259 Hexyl acetate 142-92-7 30.880 1.85E-05 *** 250.160 3.02E-12 *** C21 15.545 Acetoin 513-86-0 10.403 0.002 ** 976.810 1.30E-16 *** C22 15.658 2-Octanone 111-13-7 0.327 0.825 1.782 0.116 3-Methyl-2- C23 16.839 2758-18-1 cyclopentenone 1.534 0.255 1.431 0.270 C24 17.629 2-Heptanol 543-49-7 1.843 0.200 32.930 3.26E-06 ***

143 C25 17.876 Ethyl heptanoate 106-30-9 48.978 1.69E-06 *** 3.259 0.067 C26 18.337 Ethyl lactate 97-64-3 5.047 0.026 * 8.939 0.003 ** C27 19.006 1-Hexanol 111-27-3 30.600 1.94E-05 *** 7.825 0.005 ** C28 19.514 3-Hexanol 623-37-0 1.026 0.204 0.123 0.871 C29 20.152 2-Nonanone 821-55-6 2.252 0.057 32.979 3.23E-06 *** C30 21.239 Hexadecanal 629-80-1 1.485 0.212 3.356 0.062 C31 21.588 2-Hexanol 626-93-7 1.058 0.398 3.566 0.054 C32 22.011 2-Octanol 123-96-6 0.424 0.828 0.641 0.667 C33 22.371 Ethyl octoate 106-32-1 5.731 0.018 * 8.124 0.004 ** C34 23.087 Acetic acid 64-19-7 28.634 2.70E-05 *** 0.013 0.987 C35 23.154 1-Octen-3-ol 3391-86-4 54.304 9.70E-07 *** 0.957 0.429 C36 23.243 Isopentyl hexanoate 2198-61-0 2.227 0.150 12.294 0.001 *** C37 23.372 1-Heptanol 111-70-6 34.123 1.12E-05 *** 3.517 0.056 C38 24.354 Ethyl 6-heptenoate 25118-23-4 0.540 0.596 9.442 0.002 ** C39 24.793 2-Ethyl-1-hexanol 104-76-7 3.718 0.055 8.173 0.004 ** C40 25.422 Benzaldehyde 100-52-7 1.967 0.182 3.460 0.058 C41 26.060 2-Nonanol 628-99-9 5.391 0.021 * 92.613 3.62E-09 *** C42 26.441 Ethyl nonanoate 123-29-5 5.413 0.021 * 10.247 0.002 ** C43 27.123 Linalool 78-70-6 2.602 0.115 4.739 0.025 * C44 27.578 1-Octanol 111-87-5 1.175 0.342 144.360 1.59E-10 *** C45 27.559 2-Methyl-propanoic acid 79-31-2 6.248 0.014 * 45.028 4.57E-07 *** C46 28.391 2,3-Butanediol 513-85-9 23.574 6.97E-05 *** 1.967 0.174 C47 28.763 Methyl decanoate 110-42-9 1.411 0.282 15.713 0.000 *** C48 28.801 Terpinen-4-ol 562-74-3 6.665 0.011 * 28.847 7.24E-06 *** C49 29.472 4-Hydroxybutanoic acid 591-81-1 15.991 0.000 *** 33.456 2.96E-06 *** C50 29.444 Methyl benzoate 93-58-3 2.223 0.057 40.738 8.66E-07 *** C51 29.572 Butyrolactone 96-48-0 15.803 0.000 *** 0.389 0.838 3-(Methylthio)propyl C52 29.950 16630-55-0 acetate 0.927 0.556 2.014 0.056 C53 30.161 Benzeneacetaldehyde 122-78-1 0.497 0.774 0.146 0.709

144 C54 30.536 Ethyl decanoate 110-38-3 3.140 0.080 20.358 5.32E-05 *** C55 31.250 3-Methylbutyl octanoate 2035-99-6 0.663 0.533 17.847 0.000 *** C56 31.555 1-Nonanol 143-08-8 3.213 0.076 3.382 0.061 C57 31.971 Diethyl succinate 123-25-1 6.815 0.011 * 74.015 1.69E-08 *** C58 32.469 Ethyl 9-decenoate 67233-91-4 7.148 0.009 ** 2.198 0.145 C59 32.800 α-Terpineol 98-55-5 1.988 0.180 7.695 0.005 ** C60 33.463 Methionol 505-10-2 183.370 1.01E-09 *** 30.938 4.76E-06 *** C61 33.677 Benzyl acetate 140-11-4 3.132 0.080 374.690 1.57E-13 *** C62 35.082 Methyl salicylate 119-36-8 9.172 0.004 ** 1.256 0.313 C63 35.374 1-Decanol 112-30-1 3.606 0.059 28.647 7.54E-06 *** C64 35.472 Citronellol 106-22-9 20.382 0.000 *** 4.555 0.028 * Hexyl 4- C65 35.866 6259-76-3 hydroxybenzoate 22.915 7.98E-05 *** 53.971 0.000 *** C66 35.082 Ethyl salicylate 118-61-6 3.370 0.069 1.607 0.117 C67 36.533 Methyl dodecanoate 111-82-0 1.981 0.101 4.987 0.022 * C68 36.529 Nerol 106-25-2 6.015 0.016 * 1.112 0.355 C69 36.817 Phenethyl acetate 103-45-7 101.610 3.01E-08 *** 63.767 4.64E-08 *** C70 36.945 β-Damascenone 23726-93-4 3.402 0.068 0.324 0.728 C71 38.044 Ethyl dodecanoate 106-33-2 4.936 0.027 * 1.647 0.226 C72 38.271 Hexanoic acid 142-62-1 0.699 0.516 93.784 3.32E-09 *** C73 38.313 Guaiacol 90-05-1 0.125 0.586 2.072 0.079 C74 39.108 Benzyl alcohol 100-51-6 3.051 0.085 12.150 0.001 *** C75 40.263 Phenylethyl Alcohol 60-12-8 31.516 1.67E-05 *** 7.104 0.007 ** C76 44.951 Ethyl tetradecanoate 124-06-1 31.130 1.78E-05 *** 5.039 0.021 * C77 45.449 Octanoic acid 124-07-2 0.387 0.687 106.360 1.38E-09 *** C78 51.289 Ethyl hexadecanoate 628-97-7 4.261 0.039983 * 9.232 0.002 ** C79 52.056 Methyl dihydrojasmonate 24851-98-7 0.199 0.822 170.660 4.81E-11 ***

145

Fig. S1 Sampling map of 11 vineyards from three wine estates in Mornington Peninsula wine region, Australia.

146

Fig. S2 Stage of fermentation influences microbial diversity and composition of Chardonnay. (A) α-diversity (Shannon index) significantly decreases (p < 0.001***) during the fermentation; (B) Bray-Curtis distance PCoA of fungal communities according to the fermentation stage (BF, before fermentation; MF, at the middle of fermentation; EF, at the end of fermentation); (C) Relative abundance changes of major fungal species (12 species; relative abundance > 1.00%) that significantly differed by fermentation stage (ANOVA; FDR-corrected p value < 0.05).

147 CHAPTER SIX

General Discussion and Conclusions

Microbial biogeography is a contributor to regional distinctiveness of agricultural products, which is commonly known as terroir in viticulture. I evaluated the microbial contributions at multiple scales to wine characteristics from the soil, plant, grapes, must and ferments in wine regions across Victoria,

Australia. My thesis describes a comprehensive scenario of wine microbial biogeography where vineyard ecosystems are structured and distinguished by fungal communities, with influences from local environments, impacting the flavour and quality of wine. I present the first study of the core fungal microbiota for grapevines. In the vineyard, fungal ecology is dependent on the grapevine habitat and plant developmental stage displaying complex dynamics and succession over the annual growth cycle, which is driven by a core microbiota. Microbial biogeographic patterns decay during fermentation. The yeast Saccharomyces cerevisiae is regionally distinctive and play a central role in shaping wine characteristics within a single region. These findings add a significant body of knowledge to the microbial ecology field and provide perspectives to restore biodiversity and functioning for sustainable agriculture and for management practices to optimise agricultural commodity production.

6.1 Research Findings

In Chapter 2, I reviewed recent studies of microbiome associated with grapes and wine production based on next-generation sequencing and proposed microbial biogeography as a new perspective to investigate wine regional distinctiveness (terroir). Depending on geospatial scales, microhabitats, and bacterial/fungal taxa, microbial communities are responsive to the local environment, in particular soil and climatic conditions. While the contributions by which microbial distribution patterns drive wine metabolites remain tenuous, I emphasise the necessity of integrative microbiome and metabolome analysis, as well as uncovering microbial ecology (soil and plant) in the vineyard ecosystem to explore microbial diversity and dynamics during wine fermentation. This chapter shows that microbial biogeography at all scales and its impact on wine distinctiveness is far from resolved.

148

Microbial biogeography and its associations with wine distinctiveness at a regional scale (~ 400 km) were reported in Chapter 3. Soil and grape must microbiota (both bacteria and fungi), as well as wine aroma profiles distinguished winegrowing regions in southern Australia, regardless of the growing season/ vintage. Microbial communities and the surrounding environments structured wine characteristics, explaining 93.8% of the variance of aroma profiles. Must fungal diversity was the most important predictor of wine regionality, with additional effects from soil fungal diversity, soil properties, and weather. This paper demonstrated that xylem sap was a possible translocation pathway of fungi from the soil to the grape and must, where yeasts could be transported to the phyllosphere through other mechanisms (water splashes, insect vectors), thereby introducing effects of soil fungi to the resultant wine.

Fungal ecology within vineyards was comprehensively investigated during the annual growth cycle and reported in Chapter 4. Rather than stochastically assembled, I found that grapevine-associated fungal communities varied according to habitat (root zone soil, root, leaf, flower and grape) and plant developmental stage (flowering, fruit set, veraison, harvest). I described the first core microbiota of grapevines that existed over space and time to drive seasonal community succession in the vineyard ecosystem. Grapevine-associated microbiome succession coincided with the changing plant metabolism and physiology, and this study showed the importance of the veraison stage, and the impact of a changing environment in particular solar radiation and vine water status.

Chapter 5 revealed fungal microbiota during spontaneous fermentation and their contributions to the resultant wine at a sub-regional scale (~ 12 km). Fungal communities, yeast populations, and

Saccharomyces cerevisiae populations differentiated between geographic origins, with influences from the grape variety. Growth and dominance of S. cerevisiae reshaped the fungal community and shifted biodiversity from the species level to the strain level. These multiple layers of fungal microbiota associated with wine aroma characteristics, in which distinctive S. cerevisiae exerted the most powerful effects at small geographic scales.

149

My thesis examines theoretical considerations of microbial biogeography in an applied agricultural ecosystem. My findings on microbial contributions from larger scales (~ 400 km) to smaller scales (~

12 km), the core and rare taxa involved in plant production and translocations of microbes in the environment provide fundamental knowledge that will assist with efficient production of quality agricultural commodities and conserving regional biodiversity for sustainable agriculture.

6.2 General discussion

My thesis investigated microbial biogeography at multiple scales to add comprehensive perspectives to wine terroir. At the larger regional scale, I systematically investigated the microbiota from the soil to wine and found that soil and grape must microbiota exhibited regional patterns and that these patterns correlated with resulting wine metabolites. In particular, wine regionality is closely associated with fungal ecology with effects from local weather, climate and soil properties. The nature and importance of fungal ecology towards defining terroir was further elucidated at smaller sub- regional scales. Within vineyards, grapevine habitats and plant development outweighed biogeographic trends (at ~ 5 km distances), suggesting that fungal assemblages have extensive local heterogeneity associated with their host plant. This local heterogeneity was driven by a core microbiota which correlated with the changing climatic conditions. Within wineries, fungal communities, including the dominant fermentative yeast S. cerevisiae were geographically distinct in the local setting (~ 12 km), and associated with wine aroma profiles, in which S. cerevisiae exerted the most powerful influences on wine quality and style. A better understanding of fungal microbiota and their influences at multiple scales will help understand the scale of terroir, which is one of the great unanswered questions.

6.2.1 The fungal microbiome is a component of regional distinctiveness of wine

There is mounting evidence for geographical differentiation of wine-related microbiota at regional scales worldwide (Gayevskiy and Goddard, 2012; Bokulich et al., 2014; Taylor et al., 2014; Pinto et al., 2015; Jara et al., 2016), while the central question of the mechanisms that link microbial

150 biogeography with wine regional distinctiveness have not been answered. This thesis reveals the fungal microbiota as critical in translating microbial abundance and distribution to distinctive aroma profiles. At regional scales, grape must fungal communities demonstrated more distinct distribution patterns than bacteria and were weakly affected by the vintage, aligning with results presented by

Bokulich et al. (2014) (Chapter 3). At sub-regional scales, geographical differentiation was also evident for must fungi (Chapter 5). This distinct and stable pattern was of the most importance structuring wine regional characteristics when modelling multiple biotic (soil and must microbiota) and abiotic (soil and weather) factors in the vineyard ecosystem simultaneously (Chapter 3). The role of fungi in shaping wine terroir can be further understood from two scenarios. In the vineyard ecosystem, the grapevine-associated fungal microbiota assembled according to plant development indicating far-reaching plant-microbe interactions (Chapter 4). Grapevines could recruit microbes for specific functions at different stages (Chaparro et al., 2014), and in turn, microbiota could also affect plant physiology and metabolism, plant fitness under a/biotic stress, and fruit quality (Lau and

Lennon, 2012; Gilbert et al., 2014; Müller et al., 2016).

In a previous study conducted in Californian winegrowing regions in the northern hemisphere, metabolite profiles in the red wine were found more closely associated with bacterial profiles

(Bokulich et al., 2016). While, in the winemaking process, fungi, as the extended community resources of fermentative yeasts, conduct alcoholic fermentation and produce volatile compounds, thus providing more sensorially active biochemical conversions than bacteria (Swiegers et al., 2005).

As discussed in Chapter 3 and 5, fungal species highly associated with fermentation-derived esters, alcohols, and acids, as well as grape-derived compounds, thus reinforcing the significance of microbial biogeography on wine terroir.

6.2.2 The grapevine-associated microbiota correlates with the environment

Climate has been recognised as the most profound abiotic influence on vine growth, development, and grape quality, thus determining wine composition and regional characteristics (Van Leeuwen and

Seguin, 2006; Gladstones, 2011). This thesis suggests that microbial communities are responsive to

151 local environments over space and time, providing further evidence for environment-plant-microbe interactions and the mechanism of microbial terroir. In terms of space, various macroclimate and soil properties at regional scale showed strong effects on the soil and must microbial diversity thus the resultant wines, in particular solar radiation and soil C: N on soil fungal diversity (Chapter 3). This is consistent with studies on large and global scales that climate and soil nutrient availability drive soil fungal community assembly and enhance multifunctionality in ecosystems (Lauber et al., 2008;

Tedersoo et al., 2014; Bahram et al., 2018). In terms of time, vineyard fungal diversity and composition changed coinciding with the changing weather during the growing season at sub-regional scale. Solar radiation and vine water status highly correlated with the core fungi of the soil and plants, which drove the seasonal community succession (Chapter 4). The effect of water availability has also echoed on the presence of yeasts and fungi in other winegrowing regions (Bokulich et al., 2014; Jara et al., 2016). Increasing understanding of the modification of grapevine-associated microbiota by environmental factors will inform us about how plants recruit their microbiome to maximise both physiology and microbial diversity under different conditions, and provide perspectives to viticultural practices, such as water status, soil biodiversity, and canopy management to positively shape the microbial consortium in attempt to enhance wine quality and characteristics. Further, these findings suggest that climate change threatens not only the geographic range and water availability for viticulture, grapevine physiology and phenology, but also the microbial ecology in the vineyard and winemaking. Predicting the microbial community under the changing environments and modifying them through agricultural approaches will help the wine industry to adapt to climate change, maintain and improve wine quality and styles in the future.

6.2.3 The central role of Saccharomyces cerevisiae

S. cerevisiae plays an important role in the fungal ecology of the vineyard and the winery (Chapter 3,

4, 5). Using high-throughput sequencing, S. cerevisiae was revealed as a core species across grapevine organs and persistent throughout the growth cycle in the vineyard, rather than specific in mature grapes based on enrichment culture-dependent methods (Fleet et al., 2002; Renouf et al.,

152 2005). Interestingly, the relative abundance of S. cerevisiae changed according to plant development and was higher in the roots than the grape microbiota at harvest (Chapter 4). Xylem sap was a potential conduit to transport yeast and/ or spores from roots to grapes within the grapevine as well as making their way to the phyllosphere through external movements (insect vectors, rain splashes)

(Chapter 3). In the grape must before fermentation, S. cerevisiae was one of signatures of geographical distribution patterns among winegrowing regions at larger scales (Chapter 3) and between wine estates at smaller sub-regional scales (Chapter 5). As fermentation proceeded, S. cerevisiae grew and eventually dominated the fermentation (Chapter 3, 5) by modifying the chemical environment through fermentation, and drove wine ecosystem function (Goddard et al., 2010;

Boynton and Greig, 2016). High genetic polymorphism of S. cerevisiae defines microbial biogeography at the strain level for the wine ecosystem, at a regional scale (Gayevskiy and Goddard,

2012; Knight and Goddard, 2015) to the sub-regional scale (Chapter 5). Regional S. cerevisiae populations have been experimentally verified to produce distinct wine aroma profiles (Knight et al.,

2015). As the principal driver of wine fermentation, this study revealed that S. cerevisiae exerted more powerful effects than fungal microbiota and yeast populations on the resultant wines at small scales (< 12 km). Understanding of the role of S. cerevisiae in microbial biogeography and its contributions to wine composition will be a key step towards manipulating and optimising the microbiome thus enhancing the expression of regional characteristics and presents a considerable advance in our understanding of microbial ecology in the wine and grape ecosystem.

6.3 Significance of the study

Wine is a high value agricultural product and the fundamental role of microbes in the development of flavour and distinctiveness is very important to define and distinguish the provenance of wine. This study describes a comprehensive scenario of microbial biogeography where the fungal microbiome

(including yeasts) sits in the centre, responds to the surrounding environment, and correlates with wine composition and characteristics. The addition of these findings in Australia to support and extend investigations in other winegrowing regions worldwide supports a complex picture of environment-plant-microbe interactions in production of wine. With extended understanding of

153 microbial contributions to wine composition and characteristics, wine production can be better manipulated by thoughtful human practices to introduce and/or manage specific microbes to optimise wine flavour and enhance the expression of regional styles.

This study described the core fungal microbiota for grapevines which drives the complex ecological dynamics occurring in microbial assemblages over a growing season. Defining the core microbiome is an essential step to provide a consistent and reproducible scope to unravel the dynamic microbiome and further investigate their interactions with the plant and the functionality in the vineyard ecosystem. Understanding the seasonal patterns of these core taxa is promising for microbiome management to improve the growth of this important economic crop, yields, and quality wine production.

These exploratory findings pose a shift in our knowledge of agricultural and food systems beyond grape and wine production, wherein microbial activities play important roles in product quality and characteristics. Elaboration of the interrelationship between growing areas, climate, soil, microbial patterns, host-microbiome interactions, and quality outcomes may raise further incentive to restore the biodiversity and functioning for sustainable agriculture, improving the quality, consumer acceptance, and economic value of agricultural commodities.

6.4 Limitations of the study

This study suggests microbial contributions to wine aroma and that this contribution is related with local environments where they are grown. These findings represent evidence that was based on the methodology exploring correlations and predictions, in particular algorithms to describe interactions such as networks, random forest predictions and structural equation modeling. Whether these regionally differential microbiotas actively respond to the surrounding environment and actively modulate wine chemosensory qualities must be empirically determined to provide causative evidence

154 and fully address microbial biogeography as a determining feature of wine terroir, to which the thesis presented here provides a basis and direction to explore these relationships on multiple scales.

Fungi are indicated assembled according to plant development in the vineyard and could potentially be transported internally within the grapevine. The soil microbiome, as a source reservoir of grape associated microbiome, closely correlated with resultant wine composition. We do not yet know how grapevines recruit their microbiome (in particular for endophytes) to maximise nutrition and maintain microbial diversity under local conditions. Combining this study with studies in other grape growing areas will expand our knowledge of the core microbiome associated with the grapevine across different soil types and environmental conditions. Beyond the core microbiome, rare taxa that may disproportionately contribute to changes in microbial communities thus ecosystem resilience and stability (Shade et al., 2014; Lynch and Neufeld, 2015) deserve further research to provide a more complete view of microbial assemblages and plant-microbe interactions.

While undertaken in six winegrowing regions in southern Australia, this thesis highlights the need to further model microbiota-metabolite associations in other winegrowing regions at larger scales to characterise core microbial members and provide compelling evidence for this research area and perspectives for agricultural practices.

6.5 Future Directions

Future studies are necessary to establish causative relationships between microbial consortia, wine metabolome, and sensory profiles. More work should be done to elucidate the roles of many environmental fungi/ bacteria that dominate grapes and early fermentations, and their contributions to flavour production. Microbial interactions in the regional community may also influence the metabolites and metatranscriptomics would be a useful tool to investigate the functional behaviours of active populations during wine fermentation. Sensory studies are necessary to confirm whether microbial contributions and wine compositions can extend to consumer perception in wine traits.

155

For fungal ecology in the vineyard, next steps would be to investigate the functionality of core microbiome and their direct interactions with the plant, including investigations of the interactions between core members, between core and rare members, and with the host. This exploration will lay the foundation for technological approaches to achieve sustainable agriculture.

Anthropogenic climate change will have profound consequences on grapevine-associated microbiota thus affecting wine quality and style. Modeling the effects of climate change at large scales, combined with advances in microclimate arising from precision viticulture (for example remote sensing), will expand our knowledge on environment-plant-microbe interactions. This will help the wine industry and agricultural production in general to adapt to climate change.

6.6 Conclusions

The research presented in this thesis addresses both theoretically and commercially important concepts and adds a significant body of knowledge to the fields. It provides a significant step forward in our understanding of the contributions of microbial communities and patterns on wine regional identity and agricultural commodities in general. This thesis reveals the nature and importance of fungal ecology in the vineyard towards shaping grape and wine quality. Better understanding of environments, plants, and microbial ecology will provide more comprehensive and fundamental perspectives to restore the biodiversity and functioning for sustainable agriculture under the changing climate.

6.7 References

Bahram, M., Hildebrand, F., Forslund, S.K., Anderson, J.L., Soudzilovskaia, N.A., Bodegom, P.M. et al. (2018) Structure and function of the global topsoil microbiome. Nature 560: 233-237.

156 Bokulich, N.A., Collins, T.S., Masarweh, C., Allen, G., Heymann, H., Ebeler, S.E., and Mills, D.A. (2016) Associations among Wine Grape Microbiome, Metabolome, and Fermentation Behavior Suggest Microbial Contribution to Regional Wine Characteristics. MBio 7: e00631-00616.

Boynton, P.J., and Greig, D. (2016) Species richness influences wine ecosystem function through a dominant species. Fungal Ecol 22: 61-72.

Fleet, G., Prakitchaiwattana, C., Beh, A., and Heard, G. (2002) The yeast ecology of wine grapes. Biodiversity and biotechnology of wine yeasts Research Signpost, Kerala, India 95: 1-17.

Gayevskiy, V., and Goddard, M.R. (2012) Geographic delineations of yeast communities and populations associated with vines and wines in New Zealand. The ISME journal 6: 1281-1290.

Gilbert, J.A., van der Lelie, D., and Zarraonaindia, I. (2014) Microbial terroir for wine grapes. Proc Natl Acad Sci USA 111: 5-6.

Gladstones, J. (2011) Wine, terroir and climate change: Wakefield Press.

Goddard, M.R., Anfang, N., Tang, R., Gardner, R.C., and Jun, C. (2010) A distinct population of Saccharomyces cerevisiae in New Zealand: evidence for local dispersal by insects and human‐aided global dispersal in oak barrels. Environ Microbiol 12: 63-73.

Jara, C., Laurie, V.F., Mas, A., and Romero, J. (2016) Microbial Terroir in Chilean Valleys: Diversity of Non-conventional Yeast. Front Microbiol 7: 663.

Knight, S., and Goddard, M. (2015) Quantifying separation and similarity in a Saccharomyces cerevisiae metapopulation. The ISME journal 9: 361-370.

Knight, S., Klaere, S., Fedrizzi, B., and Goddard, M.R. (2015) Regional microbial signatures positively correlate with differential wine phenotypes: evidence for a microbial aspect to terroir. Sci Rep 5: 14233.

Lau, J.A., and Lennon, J.T. (2012) Rapid responses of soil microorganisms improve plant fitness in novel environments. Proc Natl Acad Sci USA 109: 14058-14062.

Lauber, C.L., Strickland, M.S., Bradford, M.A., and Fierer, N. (2008) The influence of soil properties on the structure of bacterial and fungal communities across land-use types. Soil Biol Biochem 40: 2407-2415.

Lynch, M.D., and Neufeld, J.D. (2015) Ecology and exploration of the rare biosphere. Nature Reviews Microbiology 13: 217-229.

157 Müller, D.B., Vogel, C., Bai, Y., and Vorholt, J.A. (2016) The plant microbiota: systems-level insights and perspectives. Annu Rev Genet 50: 211-234.

Pinto, C., Pinho, D., Cardoso, R., Custódio, V., Fernandes, J., Sousa, S. et al. (2015) Wine fermentation microbiome: a landscape from different Portuguese wine appellations. Front Microbiol 6: 905.

Renouf, V., Claisse, O., and Lonvaud‐Funel, A. (2005) Understanding the microbial ecosystem on the grape berry surface through numeration and identification of yeast and bacteria. Aust J Grape Wine Res 11: 316-327.

Shade, A., Jones, S.E., Caporaso, J.G., Handelsman, J., Knight, R., Fierer, N., and Gilbert, J.A. (2014) Conditionally rare taxa disproportionately contribute to temporal changes in microbial diversity. MBio 5: e01371-01314.

Swiegers, J., Bartowsky, E., Henschke, P., and Pretorius, I. (2005) Yeast and bacterial modulation of wine aroma and flavour. Aust J Grape Wine Res 11: 139-173.

Taylor, M.W., Tsai, P., Anfang, N., Ross, H.A., and Goddard, M.R. (2014) Pyrosequencing reveals regional differences in fruit‐associated fungal communities. Environ Microbiol 16: 2848-2858.

Tedersoo, L., Bahram, M., Põlme, S., Kõljalg, U., Yorou, N.S., Wijesundera, R. et al. (2014) Global diversity and geography of soil fungi. Science 346: 1256688.

Van Leeuwen, C., and Seguin, G. (2006) The concept of terroir in viticulture. Journal of Wine Research 17: 1-10.

158 APPENDICES

Co-authored papers published during my PhD:

Appendix I

Winters, M., Panayotides, D., Bayrak, M., Rémont, G., Gonzalez Viejo, C., Liu, D., Le, B., Liu, Y.,

Luo, J., Zhang, P., and Howell, K. (2019). Defined co‐cultures of yeast and bacteria modify the aroma, crumb and sensory properties of bread. Journal of applied microbiology 127: 778-793.

Appendix II

Zhang, P., Wu, X., Needs, S., Liu, D., Fuentes, S., Howell, K., (2017). The influence of apical and basal defoliation on the canopy structure and biochemical composition of Vitis vinifera cv. Shiraz grapes and wine. Frontiers in chemistry 5: 48.

159 Journal of Applied Microbiology ISSN 1364-5072

ORIGINAL ARTICLE Deﬁned co-cultures of yeast and bacteria modify the aroma, crumb and sensory properties of bread

M. Winters1, D. Panayotides1, M. Bayrak1,G.Remont 1,2, C.G. Viejo1, D. Liu1,B.Le1, Y. Liu1, J. Luo1, P. Zhang1 and K. Howell1

1 School of Agriculture and Food, Faculty of Veterinary and Agricultural Sciences, University of Melbourne, Parkville, Vic, Australia 2 AgroParisTech, Paris, France

Keywords Abstract lactic acid bacteria, leavening, Sourdough, starter, wheat. Aims: Yeast and bacterial communities inhabit a sourdough starter to make artisanal bread. This study shows whether the interactions of micro-organisms Correspondence derived from Australian sourdough starters provide some of the positive Kate Howell, School of Agriculture and Food, flavour, and aroma properties to bread by using defined sourdough cultures as Faculty of Veterinary and Agricultural the sole leaven in bread production. Sciences, University of Melbourne, Parkville, Methods and Results: An investigation of Australian sourdough starters found Vic. 3010, Australia. E-mail: [email protected] that they contained Saccharomyces cerevisiae and Kazachstania exigua yeasts. When these yeasts were inoculated alone to ferment wheat flour in an extended 2019/2131: received 14 November 2018, fermentation, the bread had a heterogeneous crumb structure, a deeper colour revised 27 March 2019 and accepted 4 June and a distinctive chemical aroma profile than those made with commercial 2019 baker’s yeast. When bread was made combining these yeasts individually and in combinations with lactic acid bacteria also isolated from these sourdough doi:10.1111/jam.14349 starters, including Lactobacillus plantarum, L. brevis, L. rossiae, L. casei, the bread aroma profiles and crumb structure were more distinctive, with compounds associated with sour aromas produced, and preferred by sensory panels. Conclusions: The use of defined mixed cultures as the leaven in bread making, by exploiting the microbial diversity of artisanal Australian starters, can produce bread with distinctive and attractive aromas. Significance and Impact of the Study: Understanding and identifying the community ecosystems found in sourdough cultures and using them as the sole leaven in bread production provide novel insights into microbial interactions and how they affect food quality by removing the effects of commercial yeast strains.

2014), chocolate (Ho et al. 2014) and bread (Aslankoohi Introduction et al. 2016). Yeasts are an essential part of the production of food Sourdough is a mixture of ﬂour and water fermented by and beverages (Legras et al. 2007; Sicard and Legras lactic acid bacteria (LAB) and yeasts and is used as a leaven 2011; Steensels and Verstrepen 2014). Yeast fermenta- in artisanal breads. The mixture of micro-organisms in the tions convert the raw materials into a wide range of sourdough culture produce CO2 by fermenting starch and compounds, including ethanol and carbon dioxide sugars found in the ﬂour to produce an aerated bread pro-

(CO2), glycerol, acids and beyond (Fleet 2007). Beyond duct (Arendt et al. 2007; De Vuyst et al. 2009). This pro- the dominance of Saccharomyces cerevisiae in the process usually involves extended fermentation time duction in food, the use of non-Saccharomyces species is compared to bread made with industrial methods that use increasingly considered an important way to improve a deﬁned species of S. cerevisiae (Gobbetti and G€anzle the ﬂavour, aroma and functionality of a wide variety of 2012). The fermentation outcomes can be variable but, in foods (Carrau et al. 2015) including wine (Jolly et al. most cases, result in an attractive loaf which is increasingly

160 favoured by consumers (Corsetti et al. 2000; Clarke et al. the presence of LAB (Maloney and Foy 2003). Bringing 2002; Gobbetti et al. 2005; De Vuyst and Vancanneyt the benefits of microbial diversity and extended fermenta- 2007). The microbes present in sourdough starters have tion of sourdough fermentation to conventional bread been catalogued and reviewed (de Vuyst et al. 2014; Ganzle making can thus be used to make bread a healthier food. and Ripari 2016; van Kerrebroeck et al. 2017). Most LAB Understanding the consequences of individual yeast and isolated from sourdough cultures belong to the Lactobacil- bacterial strains and their interactions within cocultures lus genus (Vogel et al. 1999; Gul et al. 2005; Galle and during fermentation is essential for this to occur. Arendt 2014). Some studies have characterized more than The origins and composition of yeasts and bacteria in 60 LAB species from mature sourdoughs and more than a sourdough starter reflect the grain, and flour used, and 30 Lactobacillus species alone have been isolated from tra- the bakery environment (Vrancken et al. 2010; Ercolini ditional sourdoughs (Minervini et al. 2015; Dertli et al. et al. 2013; Minervini et al. 2015; Rizzello et al. 2015; 2016). Some species such as Lactobacillus sanfranciscensis Alfonzo et al. 2017). A recent report of the yeast and bac- and Lactobacillus panis have only been isolated from sour- teria found on insect frass in a grain silo and their trans- dough breads, while others are commonly found in other fer into the flour provides an insight into the transfer of environments and foods (Catzeddu 2011). The range of these organisms into the sourdough (Boiocchi et al. yeast species in sourdough microbiota is less diverse than 2017). These reports paint an important picture to the that of LAB. While more than 20 species of yeast have been influence of the immediate environment on the establish- reported to be found in sourdough, only six to seven spe- ment of the sourdough ecosystem, and recent compre- cies are commonly found (De Vuyst et al. 2009; Lattanzi hensive surveys have uncovered previously undescribed et al. 2013). These species include Kazachstania exigua yeasts and bacteria in sourdoughs produced in France (previously known as Saccharomyces exigua), Pichia kudri- (Jacques et al. 2016; Lhomme et al. 2016). The impor- avzevii and Torulaspora delbrueckii, as well as S. cerevisiae tance of place is clearly necessary for the introduction and Kazachstania humilis (formerly Candida humilis) and establishment of the microbes present in sourdough, which are the two most common species (Minervini et al. but as yet the surveys of yeasts and bacteria have all taken 2015). Yeast and LAB coexist in stable relationships within place in the northern hemisphere. sourdough cultures and the nature of their interactions This paper describes the microbial diversity of yeasts depends on the metabolic pathways of each micro-organ- and bacteria in sourdoughs originating from Australia. ism (Corsetti et al. 2000; De Vuyst and Neysens 2005). Furthermore, it outlines the fermentation characteristics A number of quality features are found to be enhanced of wheat doughs inoculated with single yeasts and defined when sourdough is used as a leaven for bread making. mixed cultures with selected LAB. This study aims to This includes nutritional benefits, increased shelf life, observe the effects of sourdough-derived micro-organisms protection from spoilage, increased proteolysis and in bread making by using defined cocultures to produce improved aroma, crumb and flavour profiles. Some leavens and comparing these breads with those made nutritional and health benefits associated with sourdough with commercial yeast strains. Our methodology system- are increased protein digestibility, total phenols, antioxi- atically understands these interactions through analysis of dant activity and nutrient bioavailability (Rizzello et al. aroma, crumb colour and structure, and sensory panel 2010; De Vuyst et al. 2016). Shelf life extension is feedback it is suggested that the combination of LAB and achieved through the synthesis of both antibacterial and yeast results in breads with increased complexity and antifungal compounds (Gobbetti et al. 2005; Rizzello attractive aromas and flavours. Our paper demonstrates et al. 2011) and by delaying staling and firmness (Corsetti that the consequences of yeast and bacterial diversity in et al. 2000; Katina et al. 2006; Arendt et al. 2007; Moroni traditional artisan bread production fundamentally et al. 2009). Sourdough LAB also produce proteolytic changes the final bread to distinguish it from bread made enzymes and increase the activity of these enzymes via with a commercial strain of S. cerevisiae. Understanding acidification which causes increased gluten breakdown these interactions will enable food producers to rationally (Arendt et al. 2007; Rizzello et al. 2007). Research has apply fermentation protocols to make bread with defined also found that bread made with sourdough LAB has an flavour, aroma and structural characteristics. increased number of aroma compounds (Petel et al. 2017) and produce a wider range of aromatic acids (Han- Materials and methods sen and Schieberle 2005). Additionally, sourdough leavens improves the crumb structure and loaf volume of bread Isolation of yeasts and bacteria from sourdough starters due to acidification (Barber et al. 1992; Gobbetti and G€anzle 2012). In terms of flavour, some yeast species Eleven sourdough starters were donated to this project have been found to produce more flavour metabolites in from various bakeries within Victoria, Australia. The

161 original names, identifiers and origins of each are listed (DNA) was recovered from pure colonies using the Mas- in Table 1. For each starter, 1Á0 g was diluted with 0Á1% terPureTM Yeast DNA purification kit (Epicentre, Madi- peptone water and serially diluted and plated onto de son, WI) following the manufacturer’s instructions. MAN Rogosa and Sharpe, yeast extract peptone dextrose Primers ITS1 and ITS4 were used to amplify the ITS and (YPD) and Wallerstein Nutrient (WLN) agar media using 5Á8S rRNA DNA regions using Kapa HiFi HotStart the spread plate method. Sourdough starters SNO, PHI, ReadyMix PCR Kit (Kapa Biosystems, Wilmington, MA) TRS, TVR, RAW and RAG required supplementation as described previously (Lam and Howell 2015). The 26S with antibiotic compounds to control fungal overgrowth rDNA region was amplified using primers NL1 and NL4 À by adding 50 mg ml 1 cycloheximide to WLN media. To as described by Kurtzman and Robnett (1998)for control yeast growth without bacteria, YPD media was sequencing by AGRF (Melbourne, Australia). Returned À supplemented with 34 mg ml 1 chloramphenicol and sequences were trimmed, aligned and the consensus À1 25 mg ml ampicillin. All plates were incubated at 27°C sequences checked for similarity using the BLAST algorithm for 2–3 days in aerobic conditions and single colonies and the nucleotide databases (NCBI) using GENEIOUS were taken and replated onto fresh media to obtain pure R.10.2.2 (BioMatters, Auckland, New Zealand). colonies, which were stored in 60% glycerol at À70°C. Gram-staining, catalase and oxidase tests broadly char- Gas production of small-scale fermentations acterized LAB isolates by defining their biochemical properties (De Vuyst and Vancanneyt 2007; Mattarelli et al. Gas production of the different cultures was measured in 2014). For yeasts, plate morphology was observed and small dough fermentations with an ANKOM Gas Produc- recorded, as well as cell morphology under a light micro- tion system (ANKOM Technology, Macedon, NY). scope. Dough containing 25 g of wheat flour (Laucke T55; Taxonomic assignations of bacteria were carried out Laucke, Vic., Bridgewater, Australia) and 50 g of distilled with the 16S rRNA gene using a commercial service pro- water was inoculated with a cellular suspension contain- vided by the Australian Genome Research Facility ing 106 CFU per ml (0Á153 g) of yeast and 107 CFU per (AGRF). For yeast identification, deoxyribonucleic acid ml (0Á042 g) of bacteria according to composition given

Table 1 Origin of leavens and the yeast and bacteria identiﬁed in each sourdough

Leaven Origin Code Species

POR The Bakehouse, Portland VIC Y01 Kazachstania bulderi B02 Lactobacillus plantarum SNO Snow Gum Road, Beaufort VIC SNOSY1 Kazachstania humilis SNOSB1 Lactobacillus plantarum SNOSB2 Acetobacter cerevisiae RAG Ragland Road, Beaufort VIC RAGSY1 Saccharomyces cerevisiae RAGSB1 Lactobacillus casei TRS Tivoli Road Bakery, South Yarra VIC TRSSY3 Saccharomyces cerevisiae TRSSB1 Acetobacter cerevisiae TVR Tivoli Road Bakery, South Yarra VIC TVRSY1 Pichia membranifaciens TVRSB1 Lactobacillus hammesii TAM Natimuk VIC TAMSY1 Kazachstania humilis TAMSB2 Kocuria rhizophila TAMSB5 Lactobacillus plantarum RAW Natimuk VIC RAWSY2 Saccharomyces cerevisiae RAWSB1 Lactobacillus brevis RAWSB2 Lactobacillus plantarum PHI Phillippa’s Bakery, Richmond VIC PHISY1 Kazachstania humilis PHISB1 Lactobacillus paralimentarius ALA Ox the Baker, Mornington Peninsula VIC AlaMT1 Saccharomyces cerevisiae AlaT2 Lactobacillus rossiae AlaMT1 Lactobacillus plantarum CO1 T2 Lactobacillus brevis AlaT5 Acetobacter malorum BLBN Ox the Baker, Mornington Peninsula VIC BLBNDT1 Saccharomyces cerevisiae BLBNDT2 Lactobacillus casei

162 Table 2 Composition of the mixed cultures used in cultured experiments in this paper

Abbreviation Species present Description Source

BY Saccharomyces cerevisiae Baker’s yeast (commercial S. cerevisiae) Anchor yeast (South Africa) Y1 Y01 Kazachstania bulderi This paper Lbp B02 Lactobacillus plantarum This paper Lbp2 AlaMT1 Lactobacillus plantarum This paper Lbr AlaT2 Lactobacillus rossiae This paper Lbc BLBNDT2 Lactobacillus casei This paper Lbb CO1 T2 Lactobacillus brevis This paper Am AlaT5 Acetobacter malorum This paper M1 K. bulderi Y01 Lb. rossiae AlaT2 M2 K. bulderi Y01 This is the reconstructed ecology of the POR starter culture in Table 1. Lb. plantarum B02 M3 K. bulderi Y01 Lb. casei BLBNDT2 M4 K. bulderi Y01 POR starter culture with Lb. rossiae Lb. rossiae AlaT2 Lb. plantarum B02 M5 K. bulderi Y01 Lb. rossiae AlaT2 Lb. casei BLBNDT2 M6 K. bulderi Y01 POR starter culture with Lb. casei Lb. plantarum B02 Lb. casei BLBNDT2 M7 S. cerevisiae AlaMT1 This is the reconstructed ecology of the Ala starter culture in Table 1. Lb. plantarum AlaMT1 M8 S. cerevisiae AlaMT1 Lb. brevis CO1 T2 M9 S. cerevisiae AlaMT1 Lb. rossiae AlaT2 M10 S. cerevisiae AlaMT1 Lb. casei BLBNDT2 M11 S. cerevisiae AlaMT1 A. malorum AlaT5 in Table 2. The dough was then placed into the sealed was divided into three pieces of 300 g, shaped and put bottles and in a water bath maintained at 30°C for 24 h. into loaf tins (14Á6 9 7Á62 cm). The loaves were proofed Measurements of pressure were taken every 0Á25 min. for 2 h in a commercial oven (Convotherm 4 EasyDial 10.10; Moffat, Vic., Australia) at 30°C (Combi-steam setting: fan 2, HumidityPro 2). The loaves were baked for Bread making 20 min at 200°C, cooled for 90 min and then sliced trans- Wheat flour (Bilby Type 55; Laucke), tap water and cook- versely into 2-cm-thick slices. The baker’s yeast bread used ing salt (Black & Gold, NSW, Australia) were used to pro- the same protocol as the sourdough, but the second proof duce the leavens and duplicates of bread dough. The was shortened to 1 h. inoculum amounts used for the experimental treatments were 106 cells per gram sourdough of yeast and 107 cells Analysis of aroma of baked bread by HS-SPME-GC-MS per gram sourdough for LAB (Katina et al. 2006). The leaven was prepared by mixing 50 g flour and 50 g water and The aroma profile of bread was analysed using headspace fermented for 20 h at room temperature (approx. 25°C). solid phase micro-extraction gas chromatography mass This leaven was mixed with 325 g of water and 500 g of spectrometry (HS-SPME-GC-MS), the method was flour in a BM2500 Compact Bakehouse (Sunbeam, NSW, adapted from Aslankoohi et al. (2016). Briefly, after the Australia) for 5 min. After 30 min, salt (10 g) and water baked bread had cooled, 2 g sample was taken from the (25 ml) were added and fermented for 3 h. The dough middle of the loaf. Three samples were taken from each

163 loaf to give a total of 18 samples for each treatment. An binary image were also filled using an algorithm based on internal standard comprising 20 ll 4-octanol (Sigma- morphological reconstruction (Soille 2004). Cell detection Aldrich, Castle Hill, NSW, Australia, internal standard, from the binary image was based on an algorithm that 0Á01%) was added and agitated for 15 min at 40°C before defines cells to be any 8-connected region of pixels with being sampled by a fused silica SPME fibre (poly- value 1 (MathWorks 2016). The crumb grain features were dimethylsiloxane/divinylbenzene; Supelco Co, Bellefonte, extracted from the resulting binary image and included PA) for 30 min. The compounds were desorbed for the void fraction (VF, cell area/total area9100), mean cell 10 min at the injection port at 250°C in splitless mode. area (MCA, mm2), small to large cell ratio (SLR) and cell Agilent Technologies 6850 Series II gas chromatograph area range (R, mm2). connected to an Agilent Technologies 5973 Network mass spectrometer coupled with PAL multipurpose autosam- Sensory evaluation pling system was used to achieve the sample analysis (Agilent Technologies, Santa Clara, CA). Data were col- An untrained sensory consumer panel was conducted. A lected in Agilent G1701EA MSD software. The gas chro- total of 40 participants aged between 18 and 25 (32Á5% matograph was fitted with Agilent J&W DB-Wax Ultra males, 67Á5% females) attended two sensory panels held Inert GC column (30 m, 0Á25 mm i.d. and 0Á25 µm). on different days; the bread treatments were baked the day The carrier gas was helium with a flow rate of prior to the panels and the loaves were cut into 1-cm-thick À 1Á4 ml min 1. The temperature was initially held at 36°C slices on the day. Each participant was presented with À for 10 min and then rose to 220°C at a rate of 6°C min 1, eight slices of bread, given a three-digit random code. Par- where it was held for a further 5 min. The mass spectrom- ticipants were asked to score the treatments based on eter was operated in scan mode (35–600 amu). MS trans- aroma and flavour. Questions related to the intensity of fer line, ion source and quadrupole temperature were set aroma (not intense to very intense), the complexity of the at 250, 230 and 150°C respectively. C7-C30 alkane stan- aroma (simple to complex), colour (light to dark), unifor- dard was analysed to establish retention index. Identifica- mity (irregular to uniform), porosity (dense to porous), tion and quantification of volatile compounds were overall appearance (dislike to like), texture (dislike to like) achieved by comparing retention index and MS spectrum and overall preference (dislike to like) using a 7-point with NIST spectral library database ver. 2011 (NIST, MD). semi-structured scale, in which the extremes were The internal standard was used to semi-quantify volatile described (Granato et al. 2010; Hayes et al. 2013). compounds based on peak area ratio. Approval for this panel was given by the Faculty of Veteri- nary and Agricultural Sciences Human Research Ethics Committee (HREC 1545786.2). Crumb analysis Three images of loaf cross sections were taken in a light Statistics box (Photosimile 50; Ortery, Irvine, CA) using an iPhone SE (Apple Inc., Cupertino, CA) using the locked manual Statistical tests were conducted by MINITAB 17 (Minitab functions for focus and exposure from iOS 10. A total of Pty Ltd, Sydney, Australia) and METABOANALYST 3Á0 18 images were taken from the 6 loaves produced in (McGill University, Montreal, Canada). Minitab was used duplicated bakes for each treatment. Images were cropped to perform analysis of variance (ANOVA) on gas produc- to the largest rectangle possible containing only crumb. tion, crumb grain features and sensory results. Where an For the crumb colour analysis these cropped images were ANOVA found a significant difference (P < 0Á05), Tukey’s â analysed using an algorithm in Matlab R2018b (Math- post-hoc analysis was conducted to determine which treat- works Inc., Matick, MA) to output L*a*b colour values ments were significantly different from each other. for each image using a modified code from Gonzalez MetaboAnalyst was used conduct statistical similarity and Viejo et al. (2016). graphical tests on the GC data set. The principal compo- For the crumb structure analysis, these images were nent analysis was run on the computer vision analysis input into a customized code developed in Matlab R2018b (CVA) and aroma data using MATLAB R2018b. which converts the image to grayscale and then segments it using a locally adaptive threshold method using a sensi- Results tivity factor of 0Á68. This image was inverted such that a value of 0 (black) meant cell wall material and a value of 1 Microbial composition of the sourdough starters (white) meant cells. Small objects were cleaned out by an opening operation (erosion and dilation) using a square The identification of yeasts and bacteria found in the 393 pixel structuring element. Any holes in the resultant sourdough starters revealed a diversity of species. All

164 isolated colonies were examined morphologically, with yeasts fermentative capacity alone (Y1). Acidity (titratable À some simple biochemical tests and cellular morphology, acidity g 100 g 1 of dough) was measured and found to with the results given in Table 1. Confirmation of the be slightly increased when bacteria were included in the species identity was performed by DNA sequencing, cocultures (data not shown). which targeted the 16S rRNA gene for bacteria and the D1/D2 region of the 26S rDNA gene for yeasts. All DNA- Small-scale bread making level identifications confirmed the taxonomy at genera level suggested by colony and cell morphologies (data not Different combinations of yeast and bacteria were selected shown). For the bacteria, percent identity for hits by a for small-scale bread making. Since few small-scale bread- BLAST database search ranged from 90 to 100%, and query making experiments have considered the biochemical cover percentage was relatively high for all yeast BLAST activity of K. bulderi, this yeast was combined with vari- hits, with a range of 95–99%. All starters contained at ous bacteria to unique combinations (Table 2). Optimiz- least one isolated yeast species, and between one and five ing small-scale bread-making produced 150 g loaves for bacteria partners. Four different yeast species were iso- analysis, as detailed below. lated from the sourdough starters, these were S. cerevisiae, two from the Kazachstania genus and one from the Pichia Aroma analysis genus. Furthermore, a total of six Lactobacillus species were isolated, two Acetobacter species and one Kocuria. Fifteen volatile compounds were identified from the HS- SPME-GC-MS analysis, including a mix of alcohols, aldehydes and ketones. Ten of these compounds were present Gas production of small-scale fermentations in all the bread treatments (Table S1) with distinctive Gas production of different strain combinations (Table 2) aromas (Table S2). The pairing of K. bulderi and L. plan- were investigated through mini-fermentations in an tarum (M2) had the highest ratios for most volatiles. ANKOM gas production system over 12 h. Comparison When L. rossiae was paired with L. plantarum and K. bul- of total gas production at 12 h of fermentation (Fig. 1) deri (M4), it lowered the amounts of volatiles, suggesting found that cultures containing S. cerevisiae (M7–M11, possible competition or a pathway divergence with this BY) produced significantly greater amounts of gas than combination. Lactobacillus casei did not perform well cultures containing K. bulderi (M1–M6, Y1). Maximum with K. bulderi (M3) in terms of aroma as it had lower gas production was observed in the culture containing L. ratios of most compounds compared to the other sour- casei and S. cerevisiae (M10). Lactobacillus casei and L. doughs containing one bacterium. However, the combi- plantarum were the only two bacteria that produced an nation of L. rossiae and L. casei with K. bulderi (M5) increase in gas produced when combined with S. cere- produced the most volatile compounds with two LAB visiae (M7, M10) compared to the yeast alone (BY). which indicates a beneficial relationship between these When K. bulderi was combined with bacteria, the mixed two bacteria. cultures either reduced gas production (M2, M4, M6) or Based on the similarity of relative amounts of the vola- were not significantly different (M1, M3, M5) to the tiles in each treatment, the volatile and bread treatments

Figure 1 Gas production of small-scale fermentations using the ANKOM gas production system with standard error in the mean shown as error bars.

165 were clustered and analysed (Fig. 2). There were two dis- paired bacteria (M4, M5, M6). No significant differences tinct groups in the treatments; one contained BY and Y1 were observed between the different single bacteria leav- and the other containing the sourdough treatments. From ens (M1, M2, M3). Loaves made with yeast alone had the cluster heatmap, a clear separation was distinguished more negative a* values than the loaves made with sour- between the yeast-leavened and sourdough breads. These dough leavens. There was no significant difference in b* compounds (heptanal, 1,2,4,5-tetramethylbenzene and 2- value between the yeast species or between the sourdough pentylfuran) had higher ratios in the yeast-leavened treatments and the controls. Results of the four crumb breads, followed by the sourdoughs with one bacterium structure features analysed through CVA are shown in and the lowest amounts in the sourdoughs with two bac- Fig. 3. The VF and MCA values are analogous to crumb teria (Fig. 2). The addition of bacteria tended to lower porosity. The control loaves (BY, Y1) were not found to the amounts volatiles compared with the control Y1. be significantly different in terms of VF which suggests that the species of yeast does not affect this parameter. Only sourdough treatments containing L. rossiae (M1, Computer vision analysis of crumb structure M4, M5) had a significantly greater VF than the control The structure and colour of the crumb were assessed treatments. The MCA of the sourdough treatments was using computer vision analysis (CVA) of sliced bread found to be significantly greater than the controls. The using a custom code developed on MATLAB R2018b. The greatest MCA value was found in the L. plantarum treat- crumb colour results are summarized in Fig. 3. L* values ment (M2), suggesting this bacterium has the greatest suggest that both control treatments (BY and Y1) are sig- ability to produce larger cell sizes. The addition of other nificantly lighter than the sourdough leavens containing LAB species seemed to reduce the MCA since both mixed 2 1 0 –1 –2 6 7 9 4 1 14 2 11 13 5 8 class 10 3 12

M4 Figure 2 Aroma compounds in mixed culture bread. Treatments included BY ( ), M1 ( ), M2 ( ), M3 ( ), M4 ( ), M5 ( ), M6 ( ) and M2 Y1 ( ). The volatile compounds identiﬁed for each bread treatment were clustered based on similarity using Euclidean distance and ratios have been auto-scaled. Volatile Y1 compounds are listed by the number on the right hand side. 1: Ethanol, 2: Hexanal, 3: Heptanal, 4: 2-pentylfuran, 5: 1-pentanol, 6: BY Acetoin, 7: 2-heptenal, 8: 1-hexanol, 9: 3- ethyl-2-methyl-1,3-hexadiene, 10: 1,2,4,5- tetramethylbenzene, 11: 1-octen-3-ol, 12: 1- heptanol, 13: Benzaldehyde, 14: 2- phenylethanol.

166 Figure 3 Bread crumb colour and structure analysis of the small-scale loaves in the mixed culture fermentations by digital imaging. (a) Means of L*, a* and b* values with the standard error in the mean shown as error bars. (b) Means of void fraction, mean cell area (MCA), small to large cell ratio (SLR) and cell size range (R) with the standard error in the mean shown as error bars. treatments containing L. plantarum (M4, M6) had signifi- Figure 4 shows the score plot for the first two principal cantly lower MCA values than the single bacteria treat- components grouped according to the leaven treatment. ment (M2). The SLR and R measurements are analogous The clustering of the two yeast controls towards the bot- to the cell size uniformity within the crumb. The control tom left corner of the plot suggest that the use of bacteria loaves (BY, Y1) displayed the highest SLR values, mean- in sourdough cultures does alter the crumb and aroma of ing they possess few large cells and are thus the most uni- bread produced, particularly in terms of lightness (L) and form. All the sourdough treatments had significantly SLR. The first principal component (PC1) was found to lower SLR values than the Y1 control treatment. Overall, explain 37Á61% of the variance. The largest coefficients in loaves made with L. plantarum had the lowest SLR and PC1 corresponded to benzaldehyde, 1-octen-3-ol and 1- were thus the least uniform. Furthermore, most of the pentanol. The second principal component (PC2) was sourdough treatments (M1, M2, M3, M4, M6) had a sig- found to explain 22Á05% of the variance. The largest nificantly lower R value compared to both controls (BY, coefficients in PC2 corresponded to a*, 1,2,3,4-tetram- Y1). ethylbenzene and 2-pentylfuran.

Effect of fermentation mode on the aroma, crumb Sensory evaluation structure and colour of bread A consumer panel was performed on the bread leavened A principal component analysis of the aroma, crumb with the different sourdough treatments. Figure 5 gives structure features and colour obtained from CVA. the mean scores for each treatment based on the

167 Figure 4 Principal component analysis of the bread crumb structure and colour from CVA (void fraction, mean cell area, SLR, L*, a* and b*) and aroma compounds from GC-MS analysis (ethanol, hexanal, heptanal, 2-pentylfuran, 1-pentanol, acetoin, 2-heptenal, 1-hexanol, 3-ethyl-2- methyl-1,3-hexadiene, 1,2,4,5-tetramethylbenzene, 1-octen-3-ol, 1-heptanol, benzaldehyde and 2-phenylethanol). Treatments included BY ( ), M1 ( ), M2 ( ), M3 ( ), M4 ( ), M5 ( ), M6 ( ) and Y1 ( ). questions from the questionnaire. No significant differ- indicates the use of bacteria in sourdough cultures to lea- ences were found for the questions relating to the overall ven bread influences consumer scoring of crumb unifor- crumb appearance, texture, colour, overall aroma or mity and porosity. Comparing individual bacterial overall preference but significant differences were found species, L. rossiae (M1) scored significantly higher in unifor sour aroma intensity, fruity aroma intensity, aroma formity than L. plantarum (M2). In terms of porosity, complexity, uniformity and porosity. The three sour- only treatments containing L. plantarum (M2, M4, M6) dough treatments containing L. plantarum (M6, M2, M4) scored significantly higher compared to the controls (BY, were scored higher than the control treatments for sour Y1). aroma intensity while loaves leavened with K. bulderi alone (Y1) were scored the lowest. For intensity of fruity Discussion aroma, the loaves made with S. cerevisiae (BY) scored the highest while the sourdough containing L. rossiae (M1) Our study isolated and identified yeasts and bacteria pre- scored the lowest. Treatments that scored highest for sent in the bread making starter of Australian sourdough aroma complexity were the S. cerevisiae control (BY) and producers. By producing bread with these yeasts and bac- the sourdough treatments containing L. plantarum (M2, teria, in isolation or combination, the impact on bread is M4, M6). Overall the aroma sensory results suggest that apparent. Inclusion of bacteria with yeasts to ferment L. plantarum produces loaves with a perceptible sour wheat dough affects the aroma, crust colour, crumb aroma and high complexity while the control containing structure and sensory attributes of the final bread. S. cerevisiae generally scored higher for positive quality More than 70 species of LAB and 25 species of yeast features and appeared to be preferred by consumers. The have been identified in wheat sourdoughs (Moroni et al. two control treatments containing yeast alone (BY, Y1) 2009; Nionelli and Rizzello 2016). Minervini et al. (2015, were not significantly different in terms of uniformity or 2016) found that the two most dominant yeast species in porosity but some sourdough treatments were. This sourdough were S. cerevisiae and K. humilis. This is

168 Figure 5 Sensory analysis of baked loaves. Bar charts of mean sensory panel scores of different bread treatments with error bars represented by the standard error. (a) Colour (P-value 0Á139), (b) uniformity (P- value = 0Á00), (c) porosity (P-value = 0Á000), (d) overall visual (P-value = 0Á165), (e) texture (P-value = 0Á733), (f) sour/vinegar aroma (P- value = 0Á001), (g) fruity aroma (P- value = 0Á029), (h) aroma complexity (P- value = 0Á0Á003), (i) overall aroma (P- value = 0Á359) and (j) overall liking (P- value = 0Á977). consistent with the findings of this study as yeasts of 2007; Galle and Arendt 2014). Three of the identified genus Saccharomyces or Kazachstania were found in 9 of bacterial species are commonly found in sourdough. the 10 starters investigated. While Kazachstania yeast spe- These include Lactobacillus rossiae, Lactobacillus plan- cies have often been described in sourdoughs, Vrancken tarum and Lactobacillus brevis (Vogel et al. 1999; Corsetti et al. (2010) suggested that it is not as commonly found and Settanni 2007; Scheirlinck et al. 2008). Less com- as Saccharomyces species. The starters in this study were monly identified were the two Acetobacter species, which sourced from artisan bakers who restrict the yeast in the have been found to be the predominant bacteria in sour- bakery to naturally derived leaven, and do not generally dough (Scheirlinck et al. 2008; Li et al. 2017). Lactobacil- use commercial preparations of S. cerevisiae. Most bacte- lus casei was identified in two of the starters. While this ria identified from the sourdough starters were LAB of bacterium is regularly used in the dairy industry, espe- the Lactobacillus genus which is consistent with previous cially for milk and cheese fermentation (Reale et al. research (Vogel et al. 1999; De Vuyst and Vancanneyt 2016), it is not commonly found in the sourdough

169 environment. Notably, L. sanfranscisensis was not isolated casei may activate relevant pathways (Zotta et al. 2017) in this study. It is a common partner in sourdough fer- which could explain the presence of acetoin in only cer- mentations around the world (de Vuyst and Neysens tain sourdough treatments. Also, Petel et al. (2017) 2005). It may not be a common partner in the Australian reported that 2-heptenal was only produced by L. plan- context, but further comprehensive surveys shed light tarum, but the current study suggested that L. casei may onto its role in bread making in Australia. also be capable in producing this compound. Further The sourdough yeast species K. bulderi has previously research is required to profile bread volatiles on a more been found to have a lower fermentative capacity than detailed level, to investigate the odour activity value in commercial S. cerevisiae strains (Gobbetti et al. 1994). the bread matrix, and therefore better illustrate the This is consistent with our findings as K. bulderi pro- importance of microbial species on volatile production. duced about half the amount of gas compared to S. cere- Colour analysis from baked bread crumb suggested visiae during the 12-h model fermentation (Fig. 1). that interactions between LAB and yeast decrease light- Strains of S. cerevisiae are highly specialized in the rapid ness, increase redness and have no effect on yellowness. production of CO2 and ethanol from sugars (Maloney This is consistent with previous research which found and Foy 2003). Cocultures of S. cerevisiae and certain that the colour of loaves fermented with LAB were signif- species of LAB increase yeast CO2 production compared icantly different from the controls and were not influ- to S. cerevisiae alone. Lactobacillus plantarum, particularly, enced by the LAB strains used (Settanni et al. 2013). has been found to improve fermentation and gas produc- Research conducted by Bartkiene et al. (2017), Settanni ing ability of S. cerevisiae (Martinez-Anaya et al. 1990; et al. (2013) and Rizzello et al. (2012) found that the Gobbetti et al. 1995; Gobbetti 1998). Similarly, in this measure of redness (a*) for the crumb of sourdough study L. plantarum increased the gas producing ability of loaves increased and yellowness was lower in sourdough S. cerevisiae, but not K. bulderi. The increase in the fer- while Rizzello et al. (2012) found them to be higher in mentative capacity of S. cerevisiae in the presence of LAB control loaves. This difference in colour may be due to a may be due to LAB proteases generating peptides to sup- change in the concentration of Maillard reaction prod- port the growth of yeast cells (Maloney and Foy 2003). ucts. Since sourdough fermentation results in increased Alternatively, the decreased leavening capacity observed proteolysis (Thiele et al. 2002; Gobbetti et al. 2005; in certain bacteria–yeast combinations may be associated Arendt et al. 2007) and carbohydrate metabolism to with nutrient depletion and inhibition of yeast growth by organic acids (Ganzle et al. 2007) this method is likely to acid production (Valmorri et al. 2008). alter the amount of amino acids and sugars available for Most aroma compounds are formed during the baking Maillard reactions. of bread (Gobbetti 1998; Corsetti and Settanni 2007), Porosity was measured through VF and MCA using however, sourdough volatiles are also created during fer- CVA. Both of these measurements suggested that sour- mentation (Petel et al. 2017). While 14 major volatile dough bread made with a mixture of yeast and LAB was compounds were identified here, previous research on more porous than loaves made with yeast only. Particu- sourdough and other wheat breads have found more larly, L. rossiae produced loaves with a greater VF, even volatiles (Quılez et al. 2006; Paucean et al. 2013; Petel when paired with other bacteria. Rizzello et al. (2012) et al. 2017). The time of fermentation here was a rela- found that the VF was significantly higher for sourdough tively short 5–6 h (Katina et al. 2006) compared to gen- breads made with L. sanfranciscensis and L. plantarum eral sourdough fermentation times of up to 6 days compared to bread made with baker’s yeast alone but (G€anzle 2014). The aroma analysis clearly distinguished Settanni et al. (2013) found that the VF of sourdough the yeast vs sourdough leavened breads. Generally, the containing L. plantarum was not significantly different mixed cultures containing both bacteria and yeast tended from the control which is more consistent with our to be lower in the amounts of some volatiles produced. experimental results. In the same study, other LAB spe- This could be explained by the interaction of yeast and cies were found to produce significant differences in VF. LAB, which could affect the use of carbohydrates by LAB The MCA result suggested that sourdough leavens have during sourdough fermentation (Paucean et al. 2013), the ability to increase the mean cell area of crumb cells, and therefore production of metabolites. The highest rel- and hence the perceived porosity, compared to leavens ative production for most volatiles was found for the containing yeast alone. Settanni et al. (2013) also found yeast and LAB combination of K. bulderi and L. plan- that the MCA of sourdough made with L. plantarum was tarum, which suggests a potential beneficial interaction significantly lower but only when made with sterile flour. between the two microbes. The production of acetoin The reasons for a lower MCA may be due to a crumb may be manipulated by the genetic potential of certain with more gas cells or smaller gas cell size, with the coa- bacteria and it has been shown that L. plantarum and L. lescence of close cells leading to the formation of larger

170 gas cells (Jahromi et al. 2014). Crumb uniformity was are valuable as a sensory panel is able to distinguish our measured through the small to large cell ratio (SLR) and breads (Lassoued et al. 2008; Mogol & Gokmen, 2014). cell size range (R). CVA results of cell size range suggests Specific metabolic interactions between yeasts and bac- that the sourdough had a greater cell size range and were teria involved in sourdough bread production are likely thus less uniform than the control loaves leavened with to result in the formation of desirable breads. As shown yeast alone. SLR results also suggested that the addition of in our study, the fermentation rate is affected by inclu- bacteria produces less uniform loaves with a greater num- sion of bacteria in the fermentation, but the metabolic ber of large cells compared to loaves leavened with yeast interactions of these microbes are likely to dramatically alone. In contrast to our results, Clarke et al. (2002) found affect the aroma compounds produced. Cross-feeding of sourdough style loaves were more uniform than control essential compounds is likely to play an important part loaves as they had a greater SLR. It may be that the choice in metabolic output. For example, maltose utilization is of bacteria can significantly affect all these measures in likely to be a major factor in enhancing growth of both yeast and bacterial fermentations in bread making. yeast and bacteria in a coculture. Here, maltose-positive L. sanfranciscensis can release glucose for maltose-negative K. humilis to utilize (recently reviewed by Carbonetto Sensory evaluation et al. 2018). Additionally, release of amino acids by yeasts The sensory data revealed some significant differences can maintain and enhance growth of LAB (see, for exam- regarding sour and fruity aroma intensities and aroma ple, Ponomarova et al. 2017). We do not know if these complexities between treatments, but there were no dif- factors are at play to maintain the coculture in our sys- ferences in the overall liking of the aroma nor general tem, but further studies will allow us to understand these preference. Sourdoughs containing L. plantarum scored metabolic interactions. However, it is clear that metabolic highest for sour aroma which suggests that it can pro- interactions between micro-organisms in wine can signifi- duce compounds that contribute to sourness being cantly impact the production of fermentation aroma detected by panellists. Paterson and Piggott (2006) found (Howell et al., 2006) and this was shown in this study. that bread made with sourdough containing L. plantarum We suggest that rational inclusion of LAB with chosen was preferred in terms of taste and aroma compared to biochemical attributes to enhance and complement the the yeast-leavened loaf. Additionally, yeast-leavened bread growth of yeasts will enable the baking industry to pro- samples had the lowest scores for sour aroma. This is as duce distinctive breads. expected since LAB are responsible for acid production The research carried out in this paper investigated the which are usually linked with sour odour and taste due diversity of microbes present in sourdough cultures, to their pungent nature (Makhoul et al. 2015). Fruity which is the first of its kind in Australia. Using deliberate aroma is often associated with volatile esters and these additions of these yeasts and bacteria as the main leaven have been reported in higher amounts when nonconven- in bread production, we have directly observed how these tional yeast species are used (Aslankoohi et al. 2016), micro-organisms affected the gas production, aroma and such as the Kazachstania yeast. Surprisingly, the samples crumb features of bread. Our results show that combin- made with baker’s yeast scored the highest for fruity aro- ing yeast and bacteria can alter the composition, struc- mas. It is possible that the untrained sensory panel used ture, quality and perception by consumers of the final in this study were not able to detect subtle differences in loaf. Depending on consumer preferences and desires, aroma. In terms of crumb colour and structure, sensory these findings may thus be used by bread producers to results either showed no significant differences between alter their products to suit a particular market. This the sourdough treatments and the controls, or that the method of bread making reflects the production of tradi- sourdough was scored less uniform and more porous for tional artisan bread. The use of S. cerevisiae yeast may some treatments. In general, previous research has found hide the subtleties of the more interesting and less notice- that sourdough loaves have improved sensory properties able aspects of the sourdough ecosystem. The mechanism compared to doughs fermented with either yeast or LAB by which micro-organisms interact within sourdough cul- alone (Edeghor et al. 2016). Breads made with LAB have tures affect aroma, flavour, structure and fermentation been found to score higher for texture, crumb colour and can be used to create bread with distinctive features. (Rizzello et al. 2012; Saeed et al. 2016; Hadaegh et al. 2017), overall acceptability (Saeed et al. 2016; Hadaegh et al. 2017), appearance and taste (Hadaegh et al. 2017) Acknowledgements compared to bread made with baker’s yeast alone. Our results confirm these findings, and show that the chemi- The bakers of Victoria who generously gave their sour- cal and computer vision measures applied in this paper dough starters to our project are gratefully acknowledged.

171 In particular, Kim Hol from the Bakehouse in Portland, Corsetti, A. and Settanni, L. (2007) Lactobacilli in sourdough Michael James from Tivoli Road Bakery in South Yarra, fermentation. Food Res Int 40, 539–558. Cameron Russell from Sparrow’s in Beaufort, Phillippa Corsetti, A., Gobbetti, M., Marco, Bd, Balestrieri, F., Paoletti, Grogan from Phillippa’s in Collingwood and Stephen F., Russi, L. and Rossi, J. (2000) Combined effect of Noonan from Ox the Baker in the Mornington Peninsula sourdough lactic acid bacteria and additives on bread 48 – gave their starters and knowledge very generously. Dr firmness and staling. J Agric Food Chem , 3044 3051. Damir Torrico assisted in planning the sensory trials and De Vuyst, L. and Neysens, P. (2005) The sourdough his expertise is gratefully acknowledged. microflora: biodiversity and metabolic interactions. Trends Food Sci Technol 16,43–56. Conflict of Interest De Vuyst, L. and Vancanneyt, M. (2007) Biodiversity and identification of sourdough lactic acid bacteria. Food The authors declare no conflict of interest. Microbiol 24, 120–127. De Vuyst, L., Vrancken, G., Ravyts, F., Rimaux, T. and Weckx, S. (2009) Biodiversity, ecological determinants, and References metabolic exploitation of sourdough microbiota. Food 26 – Alfonzo, A., Miceli, C., Nasca, A., Franciosi, E., Ventimiglia, Microbiol , 666 675. G., Di Gerlando, R., Tuohy, K., Francesca, N. et al. (2017) De Vuyst, L., Van Kerrebroeck, S., Harth, H., Huys, G., Monitoring of wheat lactic acid bacteria from the field Daniel, H.-M. and Weckx, S. (2014) Microbial ecology of until the first step of dough fermentation. Food Microbiol sourdough fermentations: diverse or uniform? Food 37 – 62, 256–269. Microbiol ,11 29. Arendt, E.K., Ryan, L.A.M. and Bello, F.D. (2007) Impact of De Vuyst, L., Harth, H., Van Kerrebroeck, S. and Leroy, F. sourdough on the texture of bread. Food Microbiol 24, (2016) Yeast diversity of sourdoughs and associated 165–174. metabolic properties and functionalities. Int J Food 239 – Aslankoohi, E., Herrera-Malaver, B., Naser Rezaei, M., Microbiol ,26 34. ß Steensels, J., Courtin, C.M. and Verstrepen, K.J. (2016) Dertli, E., Mercan, E., Arıcı, M., Yılmaz, M.T. and Sagdıc,O. Non-conventional yeast strains increase the aroma (2016) Characterisation of lactic acid bacteria from complexity of bread. PLoS ONE 11,1–18. Turkish sourdough and determination of their Barber, B., Ortola, C., Barber, S. and Fernandez, F. (1992) exopolysaccharide (EPS) production characteristics. LWT 71 – Storage of packaged white bread. III. Effects of sour dough Food Sci Technol , 116 124. and addition of acids on bread characteristics. Z Lebensm Edeghor, U., Lennox, J., Agbo, B.E. and Aminadokiari, D. Unters Forch 194, 442–449. (2016) Bread fermentation using synergistic activity Bartkiene, E., Bartkevics, V., Pugajeva, I., Krungleviciute, V., between lactic acid bacteria (Lactobacillus bulgaricus) and 26 Mayrhofer, S. and Domig, K. (2017) The contribution of baker’s yeast (Saccharomyces cerevisae). J Food Sci , – P. acidilactici, L. plantarum, and L. curvatus starters and L- 46 53. (+)-lactic acid to the acrylamide content and quality Ercolini, D., Pontonio, E., De Filippis, F., Minervini, F., La parameters of mixed rye-Wheat bread. LWT 80,43–50. Storia, A., Gobbetti, M. and Di Cagno, R. (2013) Boiocchi, F., Porcellato, D., Limonta, L., Picozzi, C., Vigentini, Microbial ecology dynamics during rye and wheat 79 I., Locatelli, D.P. and Foschino, R. (2017) Insect frass in sourdough preparation. Appl Environ Microbiol , – stored cereal products as a potential source of 7827 7836. Lactobacillus sanfranciscensis for sourdough ecosystem. J Fleet, G.H. (2007) Yeasts in foods and beverages: impact on 18 Appl Microbiol 123, 944–955. product quality and safety. Curr Opin Biotechnol , – Carbonetto, B., Ramsayer, J., Nidelet, T., Legrand, J. and 170 175. Sicard, D. (2018) Bakery yeasts, a new model for studies Galle, S. and Arendt, E.K. (2014) Exopolysaccharides from 54 in ecology and evolution. Yeast 35, 591–603. sourdough lactic acid bacteria. Crit Rev Food Sci Nutr , – Carrau, F., Gaggero, C. and Aguilar, P.S. (2015) Yeast diversity 891 901. and native vigor for flavor phenotypes. Trends Biotechnol Ganzle, M. and Ripari, V. (2016) Composition and function 33, 148–154. of sourdough microbiota: from ecological theory to bread 239 – Catzeddu, P. (2011) Sourdough breads. In Flour and Breads quality. Int J Food Microbiol ,19 25. and Their Fortification in Health and Disease Prevention ed. Ganzle, M.G., Vermeulen, N. and Vogel, R.F. (2007) Preedy, V. R. and Watson, R. R. pp. 37–46. London: Carbohydrate, peptide and lipid metabolism of 24 Elsevier. lactic acid bacteria in sourdough. Food Microbiol , – Clarke, C.I., Schober, T.J. and Arendt, E.K. (2002) Effect of 128 138. € single strain and traditional mixed strain starter cultures Ganzle, M.G. (2014) Enzymatic and bacterial on rheological properties of wheat dough and on bread conversions during sourdough fermentation. Food 37 – quality. Cereal Chem 79, 640–647. Microbiol ,2 10.

172 Gobbetti, M. (1998) The sourdough microflora: interactions of Kazachstania humilis fa comb. nov. and Candida lactic acid bacteria and yeasts. Trends Food Sci Technol 9, pseudohumilis to Kazachstania pseudohumilis fa comb. nov. 267–274. Int J Syst Evol Microbiol 66, 5192–5200. Gobbetti, M. and G€anzle, M. (2012) Handbook on Sourdough Jahromi, S., Karimi, M., Yazdi, F. and Mortazavi, S. (2014) Biotechnology. New York: Springer Science & Business Response surface optimisation of Barbari bread-making Media. process variables: interrelationship of texture, image and Gobbetti, M., Corsetti, A. and Rossi, J. (1994) The sourdough organoleptic characteristics; using image analysis for microflora. Interactions between lactic acid bacteria and quality and shelf life prediction. J Food Process Preserv 38, yeasts: metabolism of carbohydrates. Appl Microbiol 1608–1621. Biotechnol 41, 456–460. Jolly, N.P., Varela, C. and Pretorius, I.S. (2014) Not your Gobbetti, M., Corsetti, A. and Rossi, J. (1995) Interaction ordinary yeast: non-Saccharomyces yeasts in between lactic acid bacteria and yeasts in sour-dough wine production uncovered. FEMS Yeast Res 14, using a rheofermentometer. World J Microbiol Biotechnol 215–237. 11, 625–630. Katina, K., Heinio,€ R.L., Autio, K. and Poutanen, K. (2006) Gobbetti, M., De Angelis, M., Corsetti, A. and Di Cagno, R. Optimization of sourdough process for improved sensory (2005) Biochemistry and physiology of sourdough lactic profile and texture of wheat bread. LWT Food Sci Technol acid bacteria. Trends Food Sci Technol 16,57–69. 39, 1189–1202. Gonzalez Viejo, C., Fuentes, S., Li, G., Collmann, R., Conde, van Kerrebroeck, S., Maes, D. and de Vuyst, L. (2017) B. and Torrico, D. (2016) Development of a robotic Sourdoughs as a function of their species diversity and pourer constructed with ubiquitous materials, open process conditions, a meta-analysis. Trends Food Sci hardware and sensors to assess beer foam quality using Technol 68, 152–159. computer vision and pattern recognition algorithms: Kurtzman, C.P. and Robnett, C.J. (1998) Identification and RoboBEER. Food Res Int 89, 504–513. phylogeny of ascomycetous yeasts from analysis of nuclear Granato, D., De Castro, I.A., Ellendersen, L.S.N. and Masson, large subunit (26S) ribosomal DNA partial sequences. M.L. (2010) Physical stability assessment and sensory Antonie Van Leeuwenhoek 73, 331–371. optimization of a dairy-free emulsion using response Lam, S.S. and Howell, K.S. (2015) Drosophila-associated yeast surface methodology. J Food Sci 75, S149–S155. species in vineyard ecosystems. FEMS Microbiol Lett 362, Gul, H., Ozcelik, S., Sagdic, O. and Certel, M. (2005) 20. Sourdough bread production with lactobacilli and S. Lassoued, N., Delarue, J., Launay, B. and Michon, C. (2008) cerevisiae isolated from sourdoughs. Process Biochem 40, Baked product texture: correlations between instrumental 691–697. and sensory characterization using Flash Profile. J Cereal Hadaegh, H., Seyyedain Ardabili, S., Tajabadi Ebrahimi, M., Sci 48, 133–143. Chamani, M. and Azizi Nezhad, R. (2017) The impact of Lattanzi, A., Minervini, F., Di Cagno, R., Diviccaro, A., different lactic acid bacteria sourdoughs on the quality Antonielli, L., Cardinali, G., Cappelle, S., De Angelis, M. characteristics of toast bread. J Food Qual 2017,1–11. et al. (2013) The lactic acid bacteria and yeast microbiota Hansen, A. and Schieberle, P. (2005) Generation of aroma of eighteen sourdoughs used for the manufacture of compounds during sourdough fermentation: applied and traditional Italian sweet leavened baked goods. Intl J Food fundamental aspects. Trends Food Sci Technol 16,85–94. Microbiol 163,71–79. Hayes, J.E., Allen, A.L. and Bennett, S.M. (2013) Direct Legras, J., Merdinoglu, D., Cornuet, J. and Karst, F. comparison of the generalized visual analog scale (gVAS) (2007) Bread, beer and wine: Saccharomyces and general labeled magnitude scale (gLMS). Food Qual cerevisiae diversity reflects human history. Mol Ecol 16, Prefer 28,36–44. 2091–2102. Ho, V.T.T., Zhao, J. and Fleet, G. (2014) Yeasts are essential Lhomme, E., Urien, C., Legrand, J., Dousset, X., Onno, B. and for cocoa bean fermentation. Int J Food Microbiol 174,72– Sicard, D. (2016) Sourdough microbial community 87. dynamics: an analysis during French organic bread-making Howell, K.S., Cozzolino, D., Bartowsky, E.J., Fleet, G.H. and processes. Food Microbiol 53,41–50. Henschke, P.A. (2006) Metabolic profiling as a tool for Li, H., Li, Z., Qu, J. and Wang, J. (2017) Bacterial diversity in revealing Saccharomyces interactions during wine traditional Jiaozi and sourdough revealed by high- fermentation. FEMS Yeast Res 6,91–101. throughput sequencing of 16S rRNA amplicons. LWT Jacques, N., Sarilar, V., Urien, C., Lopes, M.R., Morais, C.G., Food Sci Technol 81, 319–325. Uetanabaro, A.P.T., Tinsley, C.R., Rosa, C.A. et al. (2016) Makhoul, S., Romano, A., Capozzi, V., Spano, G., Aprea, E., Three novel ascomycetous yeast species of the Cappellin, L., Benozzi, E., Scampicchio, M. et al. (2015) Kazachstania clade, Kazachstania saulgeensis sp. nov., Volatile compound production during the bread-making Kazachstaniaserrabonitensis sp. nov. and Kazachstania process: effect of flour, yeast and their interaction. Food australis sp. nov. Reassignment of Candida humilis to Bioproc Tech 8, 1925–1937.

173 Maloney, D. and Foy, J. (2003) Yeast fermentations Chapter 3. of respirative cultures of Lactobacillus casei on model In Handbook of Dough Fermentations Marcel Dekker ed. sourdough fermentation. LWT Food Sci Technol 73, 622– Kulp, K. and Lorenz, K. New York: Marcel Dekker. 629. Martinez-Anaya, M., Pitarch, B., Bayarri, P. and Benedito de Rizzello, C.G., De Angelis, M., Di Cagno, R., Camarca, A., Barber, C. (1990) Microflora of the sourdoughs of wheat Silano, M., Losito, I., De Vincenzi, M., De Bari, M.D. flour bread. X. Interactions between yeasts and lactic acid et al. (2007) Highly efficient gluten degradation by bacteria in wheat doughs and their effects on bread lactobacilli and fungal proteases during food processing: quality. Cereal Chem 67,85–91. new perspectives for celiac disease. Appl Environ Microbiol MathWorks Inc. (2016) Image Processing Toolbox: User’s 73, 4499–4507. Guide. Natick: The MathWorks Inc. Rizzello, C., Nionelli, L., Coda, R., De Angelis, M. and Mattarelli, P., Biavati, B., Hammes, W. and Holzapfel, W.H. Gobbetti, M. (2010) Effect of sourdough fermentation on (2014) Guidelines for characterizing LAB, bifidobacteria stabilisation, and chemical and nutritional characteristics and related genera for taxonomic purposes. In Lactic Acid of wheat germ. Food Chem 119, 1079–1089. Bacteria: Biodiversity and Taxonomy ed. Holzapfel, W. H. Rizzello, C.G., Cassone, A., Coda, R. and Gobbetti, M. (2011) and Wood, B. J. B. pp. 583–592. Chichester, UK: John Antifungal activity of sourdough fermented wheat germ Wiley and Sons. used as an ingredient for bread making. Food Chem 127, Minervini, F., Lattanzi, A., De Angelis, M., Celano, G. and 952–959. Gobbetti, M. (2015) House microbiotas as sources of Rizzello, C.G., Coda, R., Mazzacane, F., Minervini, D. and lactic acid bacteria and yeasts in traditional Italian Gobbetti, M. (2012) Micronized by-products from sourdoughs. Food Microbiol. 52,66–76. debranned durum wheat and sourdough fermentation Minervini, F., Celano, G., Lattanzi, A., De Angelis, M. and enhanced the nutritional, textural and sensory features of Gobbetti, M. (2016) Added ingredients affect the bread. Food Res Int 46, 304–313. microbiota and biochemical characteristics of durum Rizzello, C.G., Cavoski, I., Turk, J., Ercolini, D., Nionelli, L., wheat type-I sourdough. Food Microbiol 60, 112–123. Pontonio, E., De Angelis, M., De Filippis, F. et al. (2015) Mogol, B. and Gokmen, V. (2014) Computer vision-based The organic cultivation of Triticum turgidum spp. durum analysis of foods: a non-destructive colour measurement reflects on the axis flour, sourdough fermentation and tool to monitor quality and safety. J Sci Food Agric 94, bread. Appl Environ Microbiol 81, 3192–3204. 1259–1263. Saeed, M., Yasmin, I., Khan, W.A., Pasha, I., Sharif, M.K. and Moroni,A.V.,DalBello,F.andArendt,E.K.(2009)Sourdough Shabbir, M.A. (2016) Utilization of indigenously isolated in gluten-free bread-making: an ancient technology to solve single strain starter cultures for the production of a novel issue? Food Microbiol 26,676–684. sourdough bread. Int J Food Allied Sci 2,8–14. Nionelli, L. and Rizzello, C.G. (2016) Sourdough-based Scheirlinck, I., Van der Meulen, R., Van Schoor, A., biotechnologies for the production of gluten-free foods. Vancanneyt, M., De Vuyst, L., Vandamme, P. and Huys, Foods 5, 65. G. (2008) Taxonomic structure and stability of the Paterson, A. and Piggott, J.R. (2006) Flavour in sourdough bacterial community in Belgian sourdough ecosystems as breads: a review. Trends Food Sci Technol 17, 557–566. assessed by culture and population fingerprinting. Appl Paucean, A., Vodnar, D.-C., Socaci, S.-A. and Socaciu, C. Environ Microbiol 74, 2414–2423. (2013) Carbohydrate metabolic conversions to lactic acid Settanni, L., Ventimiglia, G., Alfonzo, A., Corona, O., Miceli, and volatile derivatives, as influenced by Lactobacillus A. and Moschetti, G. (2013) An integrated technological plantarum ATCC 8014 and Lactobacillus casei ATCC 393 approach to the selection of lactic acid bacteria of flour efficiency during in vitro and sourdough fermentation. origin for sourdough production. Food Res Int 54, 1569– Eur Food Res Technol 237, 679–689. 1578. Petel, C., Onno, B. and Prost, C. (2017) Sourdough volatile Sicard, D. and Legras, J.-L. (2011) Bread, beer and wine: yeast compounds and their contribution to bread: a review. domestication in the Saccharomyces sensu stricto complex. Trends Food Sci Technol 59, 105–123. C R Biol 334, 229–236. Ponomarova, O., Gabrielli, N., Sevin, D.C., Mulleder,€ M., Soille, P. (2004) Segmentation. In Morphological Image Zirngibl, K., Bulyha, K., Andrejev, S., Kafkia, E. et al. Analysis: Principles and Applications (2nd edn) ed. Soille, (2017) Yeast creates a niche for symbiotic lactic acid P. pp. 229–254. New York: Springer. bacteria through nitrogen overflow. Cell Syst 5, 345–357. Steensels, J. and Verstrepen, K.J. (2014) Taming wild yeast: Quılez, J., Ruiz, J. and Romero, M. (2006) Relationships potential of conventional and nonconventional yeasts in between sensory flavor evaluation and volatile and industrial fermentations. Annu Rev Microbio 68,61–80. nonvolatile compounds in commercial wheat bread type Thiele, C., Gaenzle, M.G. and Vogel, R.F. (2002) Contribution baguette. J Food Sci 71, 423–427. of sourdough lactobacilli, yeast, and cereal enzymes to the Reale, A., Di Renzo, T., Zotta, T., Preziuso, M., Boscaino, F., generation of amino acids in dough relevant for bread Ianniello, R., Storti, L.V., Tremonte, P. et al. (2016) Effect flavor. Cereal Chem 79,45–51.

174 Valmorri, S., Mortensen, H.D., Jespersen, L., Corsetti, A., response and potential applications in the food industry. J Gardini, F., Suzzi, G. and Arneborg, N. (2008) Variations Appl Microbiol 122, 857–869. of internal pH in typical Italian sourdough yeasts during co-fermentation with lactobacilli. LWT Food Sci Technol Supporting Information 41, 1610–1615. Vogel, R.F., Knorr, R., Muller,€ M.R., Steudel, U., G€anzle, M.G. Additional Supporting Information may be found in the and Ehrmann, M.A. (1999) Non-dairy lactic online version of this article: fermentations: the cereal world. Antonie Van Leeuwenhoek 76 – , 403 411. Table S1. The presence or absence of identiﬁed volatile Vrancken, G., De Vuyst, L., Van der Meulen, R., Huys, G., compounds of different bread treatments from GC-MS Vandamme, P. and Daniel, H.-M. (2010) Yeast species analysis. composition differs between artisan bakery and spontaneous Table S2. Odour description of the volatile aroma laboratory sourdoughs. FEMS Yeast Res 10,471–481. compounds identiﬁed in bread making from GC-MS Zotta, T., Parente, E. and Ricciardi, A. (2017) Aerobic analysis. metabolism in the genus Lactobacillus: impact on stress

175 ORIGINAL RESEARCH published: 07 July 2017 doi: 10.3389/fchem.2017.00048

The Inﬂuence of Apical and Basal Defoliation on the Canopy Structure and Biochemical Composition of Vitis vinifera cv. Shiraz Grapes and Wine

Pangzhen Zhang †, Xiwen Wu †, Sonja Needs, Di Liu, Sigfredo Fuentes and Kate Howell *

School of Agriculture and Food, University of Melbourne, Parkville, VIC, Australia

Defoliation is a commonly used viticultural technique to balance the ratio between grapevine vegetation and fruit. Defoliation is conducted around the fruit zone to reduce the leaf photosynthetic area, and to increase sunlight exposure of grape bunches. Apical leaf removal is not commonly practiced, and therefore its influence on canopy structure and resultant wine aroma is not well-studied. This study quantified the influences Edited by: Andrea Buettner, of apical and basal defoliation on canopy structure parameters using canopy cover Fraunhofer Institute for Process photography and computer vision algorithms. The influence of canopy structure changes Engineering and Packaging (FHG), Germany on the chemical compositions of grapes and wines was investigated over two vintages Reviewed by: (2010–2011 and 2015–2016) in Yarra Valley, Australia. The Shiraz grapevines were Marcin Szymanski,´ subjected to five different treatments: no leaf removal (Ctrl); basal (TB) and apical (TD) Poznan University of Medical leaf removal at veraison and intermediate ripeness, respectively. Basal leaf removal Sciences, Poland Michael Rychlik, significantly reduced the leaf area index and foliage cover and increased canopy porosity, Technische Universität München, while apical leaf removal had limited influences on canopy parameters. However, the Germany latter tended to result in lower alcohol level in the finished wine. Statistically significant *Correspondence: Kate Howell increases in pH and decreases in TA was observed in shaded grapes, while no significant [email protected] changes in the color profile and volatile compounds of the resultant wine were found. †These authors have contributed These results suggest that apical leaf removal is an effective method to reduce wine equally to this work. alcohol concentration with minimal influences on wine composition.

Specialty section: Keywords: canopy management, defoliation, canopy structure, image analysis, shiraz, wine, aroma proﬁle This article was submitted to Food Chemistry, a section of the journal INTRODUCTION Frontiers in Chemistry

Received: 07 April 2017 Climate change has already had many impacts on the global wine industry, some of them that could Accepted: 21 June 2017 be considered beneﬁcial or detrimental depending on the growing region (Jones and Webb, 2010; Published: 07 July 2017 Mira De Orduna, 2010; Mozell and Thach, 2014). The phenological development of grapevines is Citation: altered mainly due to the changes in global temperature, carbon dioxide concentration [CO2], and Zhang P, Wu X, Needs S, Liu D, solar radiation, leading to compression of phenological stages, accelerated grape maturation and Fuentes S and Howell K (2017) The earlier harvest dates (Mira De Orduna, 2010). One of the major consequences to grape chemistry is Inﬂuence of Apical and Basal the elevated berry sugar level, which is mainly caused by excessive evaporative loss of water, rather Defoliation on the Canopy Structure and Biochemical Composition of Vitis than increased photosynthesis (Keller, 2010; Mozell and Thach, 2014). Elevated wine alcohol levels vinifera cv. Shiraz Grapes and Wine. are reported in major existing wine regions of the world (Jones, 2007; Van Leeuwen and Darriet, Front. Chem. 5:48. 2016). Increasing grape sugar and consequently wine alcohol levels is deemed undesirable for the doi: 10.3389/fchem.2017.00048 wine industry, which can be related to a range of problems, such as: stuck alcoholic fermentation,

Frontiers in Chemistry | www.frontiersin.org 1 July 2017 | Volume 5 | Article 48 176 Zhang et al. Defoliation and the Effect on Grape and Wine Composition increased levels of acetic acid, higher energy requirements MATERIALS AND METHODS for cooling, generation of unwanted by-products, alteration of wine sensory profile, potential influences on human health and Vineyard Site reducing competitiveness due to alcohol associated taxation This study was conducted on a commercial Shiraz (Vitis vinifera (Erasmus et al., 2003; Coulter et al., 2008; Mira De Orduna, 2010; cv.) vineyard located at Coldstream, Victoria, Australia (latitude − Ashenfelter and Storchmann, 2016). Thus, techniques controlling 37.72, longitude 145.41, elevation 83 m) in two contrasting sugar concentration in grapes and wine alcohol levels with growing seasons (2010–2011 and 2015–2016) The weather data minimal influences on wine biochemical profile are favored by was obtained from the closest Bureau of Meteorology weather ∼ the wine industry in general. station ( 1.5 km; BOM weather station No. 086383). The mean Four major approaches may be used to reduce alcohol levels in January temperature (MJT) and total rainfall from budburst (October) to harvest for the two studied growing seasons were wine: (i) microbiological interventions; (ii) genetic engineering ◦ ◦ of fermenting yeast; (iii) physical/enzymatic treatment of grapes 20.4 C; 706.8 mm and 20.7 C; 204 mm, respectively. Grapevines and wine and (iv) agronomical management in the vineyard. were trained using a vertical shooting positioning trellis system. The microbiological approach selects yeast strains with lower All vine rows were orientated from north to south. A total of ethanol yield during the alcoholic fermentation. Milanovic 25 panels of vines were established in each row with 4 vines et al. (2012) observed reduced ethanol production with the in each panel. Vine spacing was 2.8 m between rows and 1.8 m co-fermentation of immobilized Strarmerella bombicola and between vines. Canopy management practices were restricted to Saccharomyces cerevisiae. Genetic engineering techniques could the treatments from this trial. Other agronomical management control the genes expression of S. cerevisiae, and therefore practices were applied with normal commercial standards. No convert sugar into other by-product and create new yeast strains specific pest and disease pressure was observed during both (Ozturk and Anli, 2014). experimental seasons. Physical or enzymatic removal of ethanol can also effectively reduce alcohol level in wine. For example, membrane-based Experimental Design and Berry systems are now commonly used in new world wine countries, Measurements however, may have some adverse impacts on the aroma of wine Vitis vinifera L. cv. Shiraz grapevines included in this study were (Diban et al., 2008). On the other hand, adding the enzyme subjected to five treatments: (i) no defoliation treatment (Ctrl); glucose oxidase (GOX) during fermentation from the fungus (ii) 5–7 basal leaves surrounding bunches removed from each Aspergillus niger catalyzes the conversion of glucose into gluconic shoot at Veraison (0–50%) (TB-v) or at mid ripeness (TB-m); (iii) acid and hydrogen peroxide, thus leading to lower alcohol in the seven fully expanded apical leaves removed from each shoot at resultant wine (Biyela et al., 2009; Ozturk and Anli, 2014), but Veraison (0–50%) (TD-v) or mid ripeness (TD-m) in the 2015– this intervention is controlled in many winegrowing regions. 2016 growing season, and only Ctrl and TA-v treatment in the Grapevine canopy management has been commonly used to 2011–2012 growing season. Five replicates were conducted for control sugar accumulation during grape maturation to achieve each treatment with one panel of grapevines as one individual lower alcohol wines. The rate of sugar accumulation in berries replicate in the 2015–2016 growing season, while 12 replicates is largely dependent on the ratio of leaf area to fruit weight were conducted in 2011–2012, following a completely random (LA/FW), and lower sugar accumulation in grape berries can block design. be achieved by reducing leaf area (Ozturk and Anli, 2014). The Grape samples (200 g per replicate) were collected in zip- influences of leaf trimming on grape biochemical composition lock plastic bags before treatment at baseline, then fortnightly has been researched on many grapevine cultivars, while limited since treatment until commercial harvest on 5th April 2011 and research has been conducted on V. vinifera cv. Shiraz, which is 2nd March 2016. For each sampling date, fruit was transferred one of the major cultivars planted in Australia (Kozina et al., to the laboratory of the winery for immediately analysis in ◦ 2008; Lanari et al., 2013; Baiano et al., 2015; Feng et al., 2015; the same day (2011–2012). Samples were stored at −20 C in Osrecakˇ et al., 2016). In addition, defoliation studies tend to the winery laboratory, and later on transferred the University focus on bunch zone leaf removal, which not only modifies of Melbourne in Styrofoam boxes on dry ice, then stored at ◦ the LA/FW ratio, but also increases the direct solar exposure −20 C until analysis (2015–2016). In the 2011–2012 growing and alter bunch zone microclimate (Spayd et al., 2002). The season, grape samples from individual replicates were blended influence of apical leaf trimming on the canopy structure and together and subjected to chemical analysis. Total soluble solid ◦ biochemical composition of resultant grapes and wines has not (TSS, Brix), pH, titratable acidity (TA), total anthocyanins, and been well-studied. total phenolics were analyzed using a refractometer, pH meter, The objective of this research was to quantify the impacts of alkaline titration, and spectrophotometer, respectively, following apical and basal leaf removal at different grapevine phenological the published protocol (Iland, 2004). stages on the canopy structure of grapevines. Furthermore, this research investigates the consequences of leaf removal on the Leaf Area Index (LAI) and Canopy Structure biochemical composition of grape berries and wine in V. vinifera Measurements cv. Shiraz. The findings of this project provide vineyard managers All the experimental replicates were subjected to canopy an alternative canopy management method to manipulate grape structure measurements at mid-ripeness after defoliation sugar accumulation and to reduce wine alcohol levels with treatments in the 2015–2016 growing season following the minimal influences on the wine volatile profile. canopy cover photography method (Fuentes et al., 2008, 2014).

Frontiers in Chemistry | www.frontiersin.org 1772 July 2017 | Volume 5 | Article 48 Zhang et al. Defoliation and the Effect on Grape and Wine Composition

For this, an iPhone 5s (Apple lnc. Cupertino, CA, USA) was used BOC Australia, North Ryde, NSW, Australia), and the flow rate to acquire digital images of grapevine canopy using the back was 2.0 ml/min in constant flow mode. The GC inlet was fitted camera. Upward-looking images were obtained from under the with a borosilicate glass SPME inlet liner (Agilent, 6.3 mm o.d., grapevine canopies at 0◦ zenith angle and around 20 cm above 78.5 mm length) held at 220◦C. ground. Three images were obtained from each treated panel The SPME fiber was desorbed in the pulsed splitless mode around the middle grapevine. Images were labeled and saved in and the splitter at 50:1, was opened after 30 s. The fiber was fine JPEG format. The images collected were then analyzed using allowed to bake in the inlet for 10 min. The oven was started computer vision algorithms through a customized code written at 40◦C, held for 4 min, then increased to 220◦C at 5◦C, and in Matlab. v2015b (Mathworks Inc., Matick, MA, USA; Fuentes held for 20 min. The MS source, quadruple and transfer line et al., 2008, 2014). was held at 230◦C, 106◦C, and 250◦C, respectively. The MS was operated in positive EI mode at 70 eV with scanning over a Microvinification mass acquisition range of 35–350 m/z. The standards solution Additional 2 kg of grape bunches were collected from each of 62 commonly found compounds in wine were prepared at replicate at harvest and cooled down to 4◦C on dry ice to 13 scales dilutions in model wine solution (13% alcohol, titrate be transferred to the laboratory. Small scale fermentation was buffer, pH 3.2), and analyzed using the GC as wine samples to performed following the protocol established by the Irymple generate standard calibration curves for volatiles quantification. Research Centre of the Department of Economics, Development, Wine volatiles were identified by comparing the mass spectra Jobs, Transport, and Resources of the Victorian Government and retention indices with the NIST library in ChemStation (DEDJTR) as described earlier (Kilmister et al., 2014). Briefly, and NIST Chemistry Webbook database and standard solutions harvested grapes were destemmed, crushed in sanitized 1.5–2 L obtained. All compounds were quantified based on GC peak containers, pH of the musts were adjusted to 3.4–3.5 with tartaric ratio of individual compounds and internal standard, and the solution (10%), then diammonium phosphate (10%) at 1 ml/L calibration curves generated from the standard solutions. The of juice was added plus commercial yeast at 0.2 g/L (EC1118). concentration of ethyl non-anoate (Quality control), blank SPME Primary fermentation in microvinification were done at 18◦C in runs and blank internals standards were checked regularly. a temperature controlled room for 7 days until Baume readings reached 1–2 for pressing using a ratchet style press. Wines were Statistical Analysis transferred into clean 500 ml glass bottles, with addition of The grapevine canopy structure parameters, grape and wine potassium metabisulfite (10%) at 0.5 ml/L and copper sulfate biochemical test results and wine volatiles of different treatment (0.4%) at 1 ml/L, and placed in at 16◦C controlled cool room for groups were subjected to one-way analysis of variance (ANOVA) < 14 days. After, wines were racked into new bottles filled with CO2, at p 0.05 significance level (CoStat, version 6.4, CoHort and placed in a 1–2◦C cool room for stabilization. Wines were software, Monterey, USA). then bottled into clean 375 ml bottles and placed in 14◦C cool room until analysis. RESULTS AND DISCUSSION Preparation of Samples and Headspace Solid-Phase Leaf Area Index Analysis Microextraction and Gas Chromatography Mass The impacts of defoliation on the canopy structure of the Spectrometry (HS-SPME-GC-MS) Analysis of Wine grapevine was quantified using leaf area index (LAI), foliage Volatiles projective cover (Ff), crown cover (Fc) and porosity (φ). Wine samples were prepared for wine volatile analysis based Statistically significant lower LAI, Ff and higher φ were observed on the protocol proposed by Siebert et al. (2005) with some in basal defoliation groups (TB-v and TB-m), compared to modifications. For this, 1 ml of wine samples were diluted with the non-defoliated group (Ctrl; Figure 1). This study clearly 9 ml of milli Q water into HS-SPME vial (Agilent Technologies, quantifies the impacts of basal defoliation, with around 33% 20 ml) with addition of 2 g of sodium chloride and 200 µL of reduction in LAI, 27% reduction in Ff and 23% increase in φ 4-octanol (Internal standard; 10 mg/L) and ethyl non-anoate (Figure 1). However, no significant differences were observed (quality control; 10 ml/L). The vial and its contents were shaken between Ctrl and apical defoliation groups (TA-v and TA-m) in at 220 rpm and heated to 35◦C. A polydimethylsiloxane (PDMS, the same parameters. No significant differences were observed Agilent) 100 µm fiber was exposed to the headspace and agitated amongst all experimental groups in Fc in the 2015–2016 growing for 10 min. season. An Agilent Technologies 6850 series II (GC; Agilent LAI and Ff explains the area of leave tissue and plant foliage Technologies, Santa Clara, CA) equipped with an Agilent PAL per unit ground surface, respectively, while φ explains the light 120 multipurpose auto-sampler and coupled with an Agilent penetration rate through the canopy (Fuentes et al., 2008, 2014). 5,873 mass selective detector (MSD) was used for volatile Basal defoliation decreases the leaf area in the central part of the assessments. The instruments were controlled using Agilent canopy and increase light penetration, and therefore explains the G1701EA MSC ChemStation software in conjunction with decrease in LAI and Ff, and increase in φ. Minimal influences Agilent PAL Sampler Software Control B.01.04 for ChemStation. were observed in apical treatment groups, this was likely due The GC was fitted with a J&W DB-wax column (30 m × 0.25 to the VSP trellis system of the grapevine, where apical leaves mm, 0.25 µm film df) with helium as carrier gas (ultrahigh purity, mainly stay on top of the canopy. Therefore, apical defoliation

Frontiers in Chemistry | www.frontiersin.org 1783 July 2017 | Volume 5 | Article 48 Zhang et al. Defoliation and the Effect on Grape and Wine Composition

FIGURE 1 | Influence of leaf trimming on canopy structure parameters measured at mid-ripeness in the 2015–2016 growing season: (A) leaf area index (LAI); (B) Foliage projective cover (Ff); (C) Crown cover (Fc); (D) Porosity (φ)(p < 0.05), no unites for each parameter. One-way ANOVA were conducted to compare different parameters at p < 0.05, and a, b were used to indicate statistically significant differences. Boxplot shows median value and standard deviation for each treatment groups. Control group (Ctrl), basal leaves removal at veraison (TB-v) at mid ripeness (TB-m), apical leaves removal at veraison (TD-v) at mid ripeness (TD-m). has minimal impacts on grapevine canopy shading and canopy groups for finished wine. No significant changes were observed structure parameters. Fc represented area of the canopy per unit amongst experimental groups in the total anthocyanins and ground surface (Fuentes et al., 2008, 2014), and defoliation within phenolics of grape berries at harvest in the 2015–2016 growing the canopy does not influence the outer size of the canopy and season, which is further confirmed in the wine profile where no therefore no significant differences were observed. differences were observed in wine color density, total phenolics, red pigments, color hue, and degree of red pigment coloration. Grape and Wine Composition Analysis However, reduction in grape total anthocyanins, wine color Due to large variations amongst replicates, no statistically density and total red pigments were observed in TD-v treatment significant differences could be established among treatment group compared to Ctrl in the 2010–2011 growing season. groups in berry weigh and brix at the four sampling points The impacts of defoliation on the composition of harvested (Table 1). However, there is a trend that apical defoliation tends grape has had varying reports and most of the research has to lead to slightly lower TSS (◦Brix) at harvest in both seasons focused on basal defoliation. Baiano et al. (2015) reported (2010–2011: Ctrl 22.00, TD-v 21.20; 2015–2016: Ctrl 23.2 ± increased berry TSS and decreased berry TA, anthocyanins 0.45, TD-v 22.9 ± 0.74, TD-m 22.5 ± 0.47). This trend can and phenolics in basal defoliation treatments at veraison and be confirmed by observing the alcohol strength (%v/v) of the intermediate ripeness in Nero di Troia grapes. This is consistent finished wine (2010–2011: Ctrl 12.40, TD-v 11.72; 2015–2016: to another study in Sauvignon Blanc and Riesling showing that Ctrl 13.17 ± 0.69, TD-v 12.94 ± 0.43, TD-m 13.06 ± 0.17). basal leaf removal increase TSS and decrease TA and pH (Kozina Significant increases in the pH of harvested grape samples were et al., 2008). However, Di Profio et al. (2011) observed slightly observed in basal defoliation treatment at veraison compared to increase of minimal changes in TSS, lower TA and higher pH Ctrl (Table 1). Significantly decrease in TA was observed in all in basal defoliation treatments at veraison in Merlot, Cabernet treatments in the 2015–2016 growing season and a similar trend Franc and Cabernet Sauvignon. The same study also observed was observed in the 2011–2012 growing season. The pH and inconsistent results in berry anthocyanins and phenolics over TA was adjusted during winemaking process, and therefore pH three different vintages. A recent research reported that basal and TA of the wine were not compared amongst experimental leaf removal at veraison did not significantly modify grape TSS,

Frontiers in Chemistry | www.frontiersin.org 1794 July 2017 | Volume 5 | Article 48 Zhang et al. Defoliation and the Effect on Grape and Wine Composition

TABLE 1 | Summary of grape and wine chemical parameters.

Sampling time 2010–2011 vintagea 2015–2016 vintageb

Ctrl TD-vc Ctrl TD-v TB-v TD-m TB-m

GRAPE PARAMETERS Berry weight (100 berries) Veraison 161.6 148.0 81.0 ± 4.9 79.1 ± 8.0 86.0 ± 24.3 n/ad n/a 2 weeks Post Veraison 185.0 190.2 121.4 ± 13.4 129.28 ± 10.0 131.3 ± 12.7 126.6 ± 11.1 132.9 ± 6.7 4 weeks Post Veraison 171.2 173.8 126.1 ± 9.5 128.8 ± 17.3 134.4 ± 16.1 123.8 ± 12.7 116.5 ± 20.5 Harvest 159.8 164.4 107.4 ± 3.4 110.1 ± 9.2 119.5 ± 9.3 110.4 ± 8.2 113.8 ± 7.9 ◦Brix Veraison 14.2 13.9 11.0 ± 0.3 11.4 ± 0.4 11.5 ± 0.5 n/a n/a 2 weeks Post Veraison 18.3 17.3 16.4 ± 1.4 15.7 ± 0.7 15.7 ± 0.5 15.5 ± 0.4 15.3 ± 0.3 4 weeks Post Veraison 20.8 20.2 20.7 ± 0.5 20.7 ± 0.4 20.4 ± 1.0 20.6 ± 0.6 20.6 ± 0.8 Harvest 22.0 21.2 23.2 ± 0.5 22.9 ± 0.7 23.2 ± 1.0 22.5 ± 0.5 22.6 ± 0.5 pH Veraison 2.84 2.81 2.71 ± 0.04 2.71 ± 0.01 2.71 ± 0.06 n/a n/a 2 weeks Post Veraison 3.06 3.02 3.19 ± 0.08 3.17 ± 0.05 3.11 ± 0.09 3.12 ± 0.05 3.11 ± 0.03 4 weeks Post Veraison 3.20 3.16 3.49 ± 0.02 3.5 ± 0.09 3.48 ± 0.1 3.51 ± 0.14 3.42 ± 0.04 Harvest 3.36 3.39 3.79 ± 0.06b 3.94 ± 0.23ab 4.14 ± 0.22a 4.06 ± 0.1ab 4.01 ± 0.07ab TA Veraison 14.21 14.33 6.16 ± 0.35 6.38 ± 0.35 6.84 ± 0.35 n/a n/a 2 weeks Post Veraison 9.30 9.50 6.55 ± 0.85 6.42 ± 0.63 6.78 ± 0.56 7 ± 0.59 7.05 ± 0.08 4 weeks Post Veraison 7.70 7.75 4.49 ± 0.08 4.48 ± 0.26 4.48 ± 0.29 4.43 ± 0.37 4.84 ± 0.37 Harvest 7.25 6.60 3.84 ± 0.17a 3.06 ± 0.6b 2.6 ± 0.22b 2.59 ± 0.31b 2.85 ± 0.25b Total anthocyanins (mg/g) Harvest 1.68 1.28 0.7 ± 0.11 0.67 ± 0.06 0.62 ± 0.06 0.67 ± 0.08 0.64 ± 0.08 Total phenolics (a.u./g) Harvest n/a n/a 0.35 ± 0.04 0.32 ± 0.02 0.32 ± 0.03 0.34 ± 0.02 0.31 ± 0.04 WINE PARAMETERS Total phenolics (a.u.) Finished Wine n/a n/a 34.78 ± 5.91 35.27 ± 5.51 37.42 ± 5.49 34.78 ± 3.03 36.61 ± 3.63 Wine color density Finished Wine 11.04 9.01 5.56 ± 0.8 5.61 ± 0.57 5.5 ± 0.22 5.47 ± 0.5 5.88 ± 0.6 Total red pigments (a.u.) Finished Wine 16.78 13.06 11.24 ± 3.15 11.22 ± 2.26 12.33 ± 1.9 11.77 ± 1.36 12.31 ± 2.06 Wine color hue Finished Wine 0.75 0.73 0.61 ± 0.04 0.62 ± 0.04 0.6 ± 0.04 0.58 ± 0.03 0.6 ± 0.03 Degree of red pigment Finished Wine 37.67 40.00 31.86 ± 6.11 31.44 ± 3.51 28.21 ± 3.29 29.5 ± 2.56 30.11 ± 2.16 coloration (%) Alcohol (% v/v) Finished Wine 12.40 11.72 13.27 ± 0.69 12.94 ± 0.43 13.28 ± 0.54 13.06 ± 0.17 13.28 ± 0.24 aReplicates if Grape samples in each group in the 2010–2011 vintage were mixed in to one sample and analyzed for the biochemical composition. bOne-way ANOVA conduced for the samples in the 2015–2016 vintage at p < 0.05, a, b were used to indicate statistically signiﬁcant differences. cControl group (Ctrl), basal leaves removal at veraison (TB-v) at mid ripeness (TB-m), apical leaves removal at veraison (TD-v) at mid ripeness (TD-m). dn/a, not available.

TA, anthocyanins and phenolics content in Merlot and Teran, not in Sangiovese. Another study in Sangiovese also observed but did slightly increase TSS and decrease TA in Plavac mali lower berry TSS and wine alcohol level in treatment with in one of the 2-year study (Osrecakˇ et al., 2016). Similarly, defoliation in medium top part of the canopy, but minimal another research in Merlot with basal leaf removal a pre-veraison change in anthocyanins and phenolics, which is consistent to showed no significantly differences in TSS, pH and TA, and current results (Palliotti et al., 2013). increased anthocyanins in one season but decreased in another (King et al., 2012). Minimal changes in grape TSS, pH, and HS-SPME-GC-MS Analysis of Wine TA of basal defoliation treatment at pre-veraison and veraison Volatiles were also observed in Pinot and Grenache (Tardáguila et al., The volatile compounds in the wine samples of both vintage were 2008; Feng et al., 2015). Tardáguila et al. (2008) also reported analyzed using an HS-SPME-GC-MS method, which analyses minimal changes to total phenolics concent, while inconsistent 62 commonly found wine aroma compounds, and here only results were observed over the 3 year trial in Pinot (Feng et al., 17 volatile compounds were found of high concentrations in 2015). Very few reports have studied the influences of apical the sample wine (Table 2). The volatile compounds identified leaf removal on grape content. Lanari et al. (2013) removed can classified as: fatty acid derived esters, other esters, the apical part the canopy using mechanical leaf stripper at alcohol derived acetate, and miscellaneous compounds. No intermediate ripeness in Sangiovese and Montepulciano, and statistically significant differences were observed amongst observed lower TSS in harvested grapes, which is consistent with different treatment groups in the concentration of these volatile the results presented here. The same study also observed lower compounds, and large differences were observed between two anthocyanins and polyphenols in harvested Montepulciano, but vintages. Despite this, apical defoliaton resulted in much lower

Frontiers in Chemistry | www.frontiersin.org 1805 July 2017 | Volume 5 | Article 48 Zhang et al. Defoliation and the Effect on Grape and Wine Composition 28 0.2 5 191 2.2 24 0.2 18 196 5 96 193 8 17 9 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± n/a n/a 34 183 0.01 1.2 2 12 178 222 1.1 5.5 12 343 0.3 2.9 7 78 365 638 5 22 180 775 171 690 10 39 22 41 13 32 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± n/a n/a f 23 171 0.1 1.1 2 14 85 178 0.7 5.9 18 345 0.3 2.8 20 80 145 567 10 29 195 810 204 719 4 34 4 36 10 44 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± n/a n/a 2015–2016 vintage 29 184 0.2 1.0 3 15 131 187 1.3 4.5 19 344 0.9 2.6 22 80 236 385 9 27 148 731 227 539 8 42 14 33 12 37 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± n/a n/a 41 182 0.3 1.1 2 14 72 188 0.7 5.3 15 347 0.4 2.9 31 85 485 525 19 27 117 778 358 636 14 37 8 38 6 37 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± n/a n/a Ctrl TD-v TB-v TD-m TB-m 15 96 33 42 34 42 1.2 2.6 180 158 339 593 719 774 d,e g 42 21 98 73 91 57 n/a n/a n/a n/a 160 181 800 212 149 399 2010–2011 vintage 49 n/a n/a 105 490 300 Ctrl TD-v 4,358 2,901 5.3 c Waxy, fruity, sweet waxy, Fruity, sweet Green, fermented, nutty n/a Green, fruity, oily Whiskey, malt, burnt 187 sweet, Fruity, green 1,234 Odor Fruity, sweet, green 371 Fruity, juicy, pineapple 215 Fruity, green, apple Creamy, fruity, vanilla n/a 0.05, a, b were used to indicate statistically signiﬁcant differences. < g/L) Apple, fruity, fresh g/L) Sweet, apple, pineapple 87 g/L) n/a g/L) Fruity, sweet, banana 172 g/L) Fruity, apple, ylang g/L) g/L) g/L) g/L) g/L) g/L) g/L) Whiskey, fusel, ethereal 71 0.05). < Compound name b LRI-Act a 880 841 Ethyl acetate ( µ 1,638 1,638 Ethyl decanoate ( µ 1,439 1,435 Ethyl octanoate (mg/L) 1,461 1,461 1-Heptanol ( µ 1,913 1,910 Phenylethyl Alcohol (mg/L) Foral, rose, dried rose 22 1,355 1,358 1-Hexanol (mg/L) 1,208 1,210 Isopentanol (mg/L) 1,233 1,231 Ethyl hexanoate (mg/L) 1,066 1,066 Ethyl 3-methylbutanoate ( µ 1,272 1,270 Hexyl acetate ( µ 1,126 1,111 3-Methylbutyl acetate ( µ 1,339 1,345 Butyl lactate ( µ 1,680 1,677 Diethyl butanedioate ( µ 1,399 1,426 Ethyl 2-methyloctanoate ( µ LRI-Lit Summary of HS-SPME-GC-MS results ( p h Replicates if Grape samples in each group in the 2010–2011 vintage were mixed in to one sample and analyzed for the biochemical composition. Actual linear retention index calculated based on alkane standards. Internal standard 4-octanol (Peak 12) and quality control ethyl non-anoate (Peak 16) not shown in the table. Control group (Ctrl), basal leaves removal at veraison (TB-v) at mid ripeness (TB-m), apical leaves removal at veraison (TD-v) at mid ripeness (TD-m). Linear retention index obtained from NIST chemistry webbook. Odor of compounds obtained from Flavornet, The Good Scents CompanyConcentration and The of Pherobase. ethyl acetate, butyl lactate, diethyl butanedioate, isobutanol and ethyl 2-methyloctanoate were expressed as ug/L in 4-octanol correspondent, while ethyl hexanoate, ethyl octanoate, isopentanol, 1-hexanol and One-way ANOVA conduced for the samples in the 2015–2016 vintage at p ALCOHOL DERIVED ACETATE STRAIGHT-CHAIN ESTERS OTHER ESTERS ALCOHOLS 17 14 15 19 11 7 8 TABLE 2 | Peak number a b c d e phenylethyl alcohol were expressedf in mg/L. 1 2 1,0343 1,016 Ethyl butanoate ( µ 1,050 1,0525 Ethyl 2-methylbutanoate ( µ 1,094 1,085 Isobutanol ( µ g h 4 9 6 10 18 13

Frontiers in Chemistry | www.frontiersin.org 1816 July 2017 | Volume 5 | Article 48 Zhang et al. Defoliation and the Effect on Grape and Wine Composition concentration of ethyl acetate, ethyl hexanoate, ethyl octanoate, 2015a, 2016c), it is essential to consider the location and ethyl decanoate, 3-methylbutyl acetate, and 1-hexanol in aspects of the vineyard when conducting defoliation treatments. 2010–2011 vintage, but fewer differences were observed between Additionally, canopy defoliation is associated with other factors, Ctrl and TD-v in the 2015–2016 vintage. Higher concentration such as solar exposure, which is further associated with bunch of diethyl butanedioate, ethyl 2-methylbutanoate, and ethyl 3- zone microclimate (Zhang et al., 2015b). Thus, these factors have methylbutanoate were also observed in the apical defoliation the potential to influence the defoliation outcome and therefore group at veraison in the 2010–2011 vintage. However, in the should be considered separately as suggested previously (Spayd 2015–2016 vintage the concentration of diethyl butanedioate in et al., 2002). More importantly, terpenoids in grape berries TD-v was lower than that of Ctrl, while ethyl 2-methylbutanoate are actively produced in different phenological stages (Zhang and ethyl 3-methylbutanoate were not detected. All defoliation et al., 2016a,b), and therefore apical defoliation at different treatments had slightly higher average concentration of 1- phenological stages should also be investigated. heptanol and slightly lower average concentration of hexyl acetate, diethyl butanedioate, isobutanol, isopentanol, and ethyl CONCLUSIONS 2-methylbutanoate, compared to non-defoliated group in the 2015–2016 vintage. Defoliation is a commonly used technique to manipulate grape The influences of defoliation on wine volatiles profile are and wine composition, and the present study compared both associated with five major factors, the first being different basal defoliation and the less studied apical defoliation and its cultivars and clones having different responses to defoliation influences on grape and wine composition. Compared to basal treatment. Significant reductions in the concentration of most defoliation, apical defoliation has less influence on the grapevine wine volatile esters were observed in basal defoliated Sauvignon canopy and therefore less potential effect on the bunch zone Blanc and Riesling, however Sauvignon Blanc is more sensitive microclimate. Apical defoliation near veraison resulted in slightly to defoliation treatment and has much higher percentage of increased pH and decreased TA in harvested grape berries, reduction (Kozina et al., 2008). Significant differences in wine and could lower the alcoholic strength in the resultant wine. volatile were reported amongst defoliation treatment groups of Reduction of wine anthocyanins and color profile in apical two different Sauvignon blanc clone (Šuklje et al., 2016).The treatment groups were observed in one season, but not the timing of defoliation may have dramatic influence on grape other. Minimal changes were observed in major wine volatile composition, as shown by Šuklje et al. (2014) who showed compounds amongst non-defoliation and defoliation treatments. that the influence of basal defoliation on Sauvignon Blanc was These results demonstrate that apical defoliation is a novel and similar as that of Kozina et al. (2008) (treatment at veraison), effective way to moderate wine alcohol with minimal influenced but when performed at an earlier berry development stage on wine aromatic properties, and therefore may be a technique (treatment at pepper core size), observed higher concentration to mitigate the influences of global warming on increasing wine of volatile esters in Sauvignon Blanc. Similarly, another study alcohol level. Further research on apical defoliation is needed in Tempranillo reported much higher total acetates and lower to study the influence on other important aromatic compounds. wine alcohols in defoliation group at pre-bloom compared to The timing and location of apical defoliation and its interaction fruit-set (Vilanova et al., 2012). Defoliation by mechanical or with other viticulture and climate factors need to be investigated manual methods are another factor may affect volatiles profile to determine the optimal apical defoliation technique suitable for of the resultant wine. Dramatic reduction of wine volatile commercial vineyards. esters were only observed in manual defoliation treatment, but not in mechanical treatment groups in Tempranillo (Vilanova AUTHOR CONTRIBUTIONS et al., 2012). Further harvest time can also alter the result of defoliation. Significantly higher concentrations of wine esters PZ and KH conceived and designed the study, analyzed the were observed in the defoliation treatment group at first data, and wrote the paper. SF and DL contributed analytical harvest, but no significant differences when a second harvest methods and finalized the data. XW, SN, and PZ performed the was conducted 12 days later (Verzera et al., 2016). Finally, experiments. PZ, KH, SF, SN, and XW participated in the design the location of the defoliation as given in the current study and critical revision. All authors read and approved the final is another cause of variation in wine volatiles. This study is manuscript. the first to study the effect of apical defoliation, but together with two studies shows that apical leaf removal modifies grape ACKNOWLEDGMENTS berry composition differently compared to basal leaf removal and therefore influences wine volatiles differently (Lanari et al., 2013; This work was sponsored by the Rathbone Wine Group and Palliotti et al., 2013). The University of Melbourne. The work was performed at the The influences of apical defoliation on grape volatiles such as University of Melbourne and Yering Station Winery. Funding terpenes, have yet to be elucidated. This is especially true when was received from the Faculty of Veterinary and Agricultural terpenoid compounds, such as rotundone, are associated with Sciences, University of Melbourne. The authors thank Mr. wine quality (Herderich et al., 2013, 2015). Since terpenes may Andy Clark, Mr. Willy Lunn, and the late Mr. Nathan Scarlett be produced differently within individual bunches, individual (Rathbone Wine Group) for their support and assistance with the grapevines, and individual vineyard blocks (Zhang et al., 2013, field trials.

182 Frontiers in Chemistry | www.frontiersin.org 7 July 2017 | Volume 5 | Article 48 Zhang et al. Defoliation and the Effect on Grape and Wine Composition

REFERENCES ‘Sangiovese’ and ‘Montepulciano’ grapevines. ISHS Acta Hoticult. 978, 327–334. doi: 10.17660/ActaHortic.2013.978.38 Ashenfelter, O., and Storchmann, K. (2016). Climate change and wine: a review Milanovic, V., Ciani, M., Oro, L., and Comitini, F. (2012). Starmerella of the economic implications. J. Wine Econ. 11, 105–138. doi: 10.1017/jwe. bombicola influences the metabolism of Saccharomyces cerevisiae at 2016.5 pyruvate decarboxylase and alcohol dehydrogenase level during mixed Baiano, A., De Gianni, A., Previtali, M. A., Del Nobile, M. A., Novello, V., wine fermentation. Microb. Cell Fact. 11:18. doi: 10.1186/1475-2859-11-18 and De Palma, L. (2015). Effects of defoliation on quality attributes of Nero Mira De Orduna, R. (2010). Climate change associated effects on grape di Troia (Vitis vinifera L.) grape and wine. Food Res. Int. 75, 260–269. and wine quality and production. Food Res. Int. 43, 1844–1855. doi: 10.1016/j.foodres.2015.06.007 doi: 10.1016/j.foodres.2010.05.001 Biyela, B., Du Toit, W., Divol, B., Malherbe, D., and Van Rensburg, P. Mozell, M. R., and Thach, L. (2014). The impact of climate change on the (2009). The production of reduced-alcohol wines using Gluzyme MonoR global wine industry: challenges & solutions. Wine Econ. Policy 3, 81–89. 10.000 BG-treated grape juice. South Afr. J. Enol. Vitic. 30, 124–132. doi: 10.1016/j.wep.2014.08.001 doi: 10.21548/30-2-1432 Osrecak,ˇ M., Karoglan, M., and Kozina, B. (2016). Influence of leaf removal and Coulter, A., Henschke, P., Simos, C., and Pretorius, I. (2008). When the heat is on, reflective mulch on phenolic composition and antioxidant activity of Merlot, yeast fermentation runs out of puff. Aust. N.Z. Wine Ind. J. 23:26. Teran and Plavac mali wines (Vitis vinifera L.). Sci. Hortic. 209, 261–269. Di Profio, F., Reynolds, A., and Kasimos, A. (2011). Canopy management doi: 10.1016/j.scienta.2016.07.005 and enzyme impacts on Merlot, Cabernet Franc, and Cabernet Sauvignon. Ozturk, B., and Anli, E. (2014). Different techniques for reducing alcohol levels in I. Yield and berry composition. Am. J. Enol. Viticult. 62, 139–151. wine: a review? BIO Web Confer. 3:02012. doi: 10.1051/bioconf/20140302012 doi: 10.5344/ajev.2010.10024 Palliotti, A., Panara, F., Silvestroni, O., Lanari, V., Sabbatini, P., Howell, G. S., et al. Diban, N., Athes, V., Bes, M., and Souchon, I. (2008). Ethanol and aroma (2013). Influence of mechanical postveraison leaf removal apical to the cluster compounds transfer study for partial dealcoholization of wine using membrane zone on delay of fruit ripening in Sangiovese (Vitis vinifera L.) grapevines. Aust. contactor. J. Memb. Sci. 311, 136–146. doi: 10.1016/j.memsci.2007.12.004 J. Grape Wine Res. 19, 369–377. doi: 10.1111/ajgw.12033 Erasmus, D., Van Der Merwe, G., and Van Vuuren, H. (2003). Genolewide Siebert, T. E., Smyth, H. E., Capone, D. L., Neuwohner, C., Pardon, K. H., expression analyses: metabolic adaptation of Saccharomyces cerevisiae to Skouroumounis, G. K., et al. (2005). Stable isotope dilution analysis of high sugar stress. FEMS Yeast Res. 3, 375–399. doi: 10.1016/S1567-1356(02) wine fermentation products by HS-SPME-GC-MS. Anal. Bioanal. Chem. 381, 00203-9 937–947. doi: 10.1007/s00216-004-2992-4 Feng, H., Yuan, F., Skinkis, P. A., and Qian, M. C. (2015). Influence of cluster Spayd, S. E., Tarara, J. M., Mee, D. L., and Ferguson, J. C. (2002). Separation of zone leaf removal on Pinot noir grape chemical and volatile composition. Food sunlight and temperature effects on the composition of Vitis vinifera cv. Merlot Chem. 173, 414–423. doi: 10.1016/j.foodchem.2014.09.149 Berries. Am. J. Enol. Vitic. 53, 171–182. Fuentes, S., Palmer, A., Taylor, D., Zeppel, M., Whitely, R., and Eamus, D. Šuklje, K., Antalick, G., Buica, A., Langlois, J., Coetzee, Z. A., Gouot, J., et al. (2016). (2008). An automated procedure for estimating the leaf area index (LAI) Clonal differences and impact of defoliation on Sauvignon blanc (Vitis vinifera of woodland ecosystems using digital imagery, MATLAB programming and L.) wines: a chemical and sensory investigation. J. Sci. Food Agric. 96, 915–926. its application to an examination of the relationship between remotely doi: 10.1002/jsfa.7165 sensed and field measurements of LAI. Funct. Plant Biol. 2008, 1070–1079. Šuklje, K., Antalick, G., Coetzee, Z., Schmidtke, L. M., Baša Cesnik,ˇ H., doi: 10.1071/FP08045 Brandt, J., et al. (2014). Effect of leaf removal and ultraviolet radiation on Fuentes, S., Poblete-Echeverría, C., Ortega-Farias, S., Tyerman, S., and De Bei, R. the composition and sensory perception of Vitis vinifera L. cv. Sauvignon (2014). Automated estimation of leaf area index from grapevine canopies using Blanc wine. Aust. J. Grape Wine Res. 20, 223–233. doi: 10.1111/ajgw. cover photography, video and computational analysis methods. Aust. J. Grape 12083 Wine Res. 20, 465–473. doi: 10.1111/ajgw.12098 Tardáguila, J., Diago, M. P., De Toda, F. M., Poni, S., and Vilanova, M. (2008). Herderich, M. J., Siebert, T. E., Parker, M., Capone, D. L., Mayr, C., Zhang, P., et al. Effects of timing of leaf removal on yield, berry maturity, wine composition (2013). Synthesis of the ongoing works on Rotundone, an aromatic compound and sensory properties of cv. Grenache grown under non irrigated conditions. responsible for the peppery notes in wines. Internet J. Enol. Vitic. 6/1, 1–6. OENO One 42, 221–229. doi: 10.20870/oeno-one.2008.42.4.810 Herderich, M., Barter, S., Black, C. A., Bramley, R., Capone, D., Dry, P., et al. Van Leeuwen, C., and Darriet, P. (2016). The impact of climate (2015). “Terroir effects on grape and wine aroma compounds,” in Advances change on viticulture and wine quality. J. Wine Econ. 11, 150–167. in Wine Research, eds S. B. Ebeler, G. Sacks, S. Vidal, and P. Winterhalter doi: 10.1017/jwe.2015.21 (American Chemical Society), 131–146. Verzera, A., Tripodi, G., Dima, G., Condurso, C., Scacco, A., Cincotta, F., et al. Iland, P. (2004). Chemical Analysis of Grapes and Wine: Techniques and Concepts. (2016). Leaf removal and wine composition of Vitis vinifera L. cv. Nero Campbelltown, SA: Patrick Iland Wine Promotions. d’Avola: the volatile aroma constituents. J. Sci. Food Agric. 96, 150–159. Jones, G. V. (2007). Climate change: observations, projections, and general doi: 10.1002/jsfa.7075 implications for viticulture and wine production. Econ. Dep. Work. Paper 7, Vilanova, M., Diago, M. P., Genisheva, Z., Oliveira, J., and Tardaguila, J. (2012). 15. doi: 10.1787/177754708811 Early leaf removal impact on volatile composition of Tempranillo wines. J. Sci. Jones, G. V., and Webb, L. B. (2010). Climate change, viticulture, Food Agricult. 92, 935–942. doi: 10.1002/jsfa.4673 and wine: challenges and opportunities. J. Wine Res. 21, 103–106. Zhang, P., Barlow, S., Krstic, M., Herderich, M., Fuentes, S., and Howell, K. doi: 10.1080/09571264.2010.530091 (2015a). Within-Vineyard, Within-Vine, and Within-Bunch Variability of the Keller, M. (2010). Managing grapevines to optimise fruit development in a Rotundone Concentration in Berries of Vitis vinifera L. cv. Shiraz. J. Agricult. challenging environment: a climate change primer for viticulturists. Aust. J. Food Chem. 63, 4276–4283. doi: 10.1021/acs.jafc.5b00590 Grape Wine Res. 16, 56–59. doi: 10.1111/j.1755-0238.2009.00077.x Zhang, P., Fuentes, S., Siebert, T., Krstic, M., Herderich, M., Barlow, E. W. R., Kilmister, R. L., Mazza, M., Baker, N. K., Faulkner, P., and Downey, M. O. (2014). A et al. (2016a). Comparison data of common and abundant terpenes at different role for anthocyanin in determining wine tannin concentration in Shiraz. Food grape development stages in Shiraz wine grapes. Data Brief 8, 1127–1136. Chem. 152, 475–482. doi: 10.1016/j.foodchem.2013.12.007 doi: 10.1016/j.dib.2016.07.010 King, P. D., McClellan, D. J., and Smart, R. E. (2012). Effect of severity of leaf and Zhang, P., Fuentes, S., Siebert, T., Krstic, M., Herderich, M., Barlow, E. W. R., crop removal on grape and wine composition of Merlot vines in Hawke’s Bay et al. (2016b). Terpene evolution during the development of Vitis vinifera L. vineyards. Am. J. Enol. Viticult. 2012:12020. doi: 10.5344/ajev.2012.12020 cv. Shiraz grapes. Food Chem. 204, 463–474. doi: 10.1016/j.foodchem.2016. Kozina, B., Karoglan, M., Herjavec, S., Jeromel, A., and Orlic, S. (2008). Influence 02.125 of basal leaf removal on the chemical composition of Sauvignon Blanc and Zhang, P., Fuentes, S., Wang, Y., Deng, R., Krstic, M., Herderich, M., et al. (2016c). Riesling wines. J. Food Agric. Environ. 6, 28–33. Distribution of rotundone and possible translocation of related compounds Lanari, V., Lattanzi, T., Borghesi, L., Silvestroni, O., and Palliotti, A. amongst grapevine tissues in Vitis vinifera L. cv. Shiraz. Front. Plant Sci. 7:859. (2013). Post-veraison mechanical leaf removal delays berry ripening on doi: 10.3389/fpls.2016.00859

Frontiers in Chemistry | www.frontiersin.org 1838 July 2017 | Volume 5 | Article 48 Zhang et al. Defoliation and the Effect on Grape and Wine Composition

Zhang, P., Howell, K., Krstic, M., Herderich, M., Barlow, E. W. R., and Fuentes, S. Conflict of Interest Statement: The authors declare that the research was (2015b). Environmental factors and seasonality affect the concentration of conducted in the absence of any commercial or financial relationships that could rotundone in Vitis vinifera L. cv. Shiraz Wine. PLoS ONE 10:e0133137. be construed as a potential conflict of interest. doi: 10.1371/journal.pone.0133137 Zhang, P., Scarlett, N., Sheehan, D., Barlow, S., Krstic, M., Herderich, M., Copyright © 2017 Zhang, Wu, Needs, Liu, Fuentes and Howell. This is an open-access et al. (2013). “Intra-bunch variability of rotundone concentration in Vitis article distributed under the terms of the Creative Commons Attribution License (CC vinifera cv. Shiraz wine grapes at harvest,” in Australian Wine Industry BY). The use, distribution or reproduction in other forums is permitted, provided the Technical Conference, eds. E. M. C. Robinson, P. W. Godden, and D. L. original author(s) or licensor are credited and that the original publication in this Johnson (Adelaide, SA: Australian Wine Industry Technical Conference journal is cited, in accordance with accepted academic practice. No use, distribution Inc.), 158. or reproduction is permitted which does not comply with these terms.

Frontiers in Chemistry | www.frontiersin.org 1849 July 2017 | Volume 5 | Article 48

Minerva Access is the Institutional Repository of The University of Melbourne

Author/s: Liu, Di

Title: Understanding the Importance of Microbial Biogeography to Australian Winemaking

Date: 2020

Persistent Link: http://hdl.handle.net/11343/251818

File Description: Final thesis file

Terms and Conditions: Terms and Conditions: Copyright in works deposited in Minerva Access is retained by the copyright owner. The work may not be altered without permission from the copyright owner. Readers may only download, print and save electronic copies of whole works for their own personal non-commercial use. Any use that exceeds these limits requires permission from the copyright owner. Attribution is essential when quoting or paraphrasing from these works.