Influence of Lab Adapted Natural Diet, Genotype, and Microbiota On

INFLUENCE OF LAB ADAPTED NATURAL DIET, GENOTYPE, AND MICROBIOTA ON

DROSOPHILA MELANOGASTER LARVAE

ANDREI BOMBIN

LAURA K. REED, COMMITTEE CHAIR JULIE B. OLSON JOHN H. YODER STANISLAVA CHTARBANOVA-RUDLOFF CASEY D. MORROW

A DISSERTATION

Submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy in the Department of Biological Sciences in the Graduate School of The University of Alabama

TUSCALOOSA, ALABAMA

2020

ABSTRACT

Obesity is an increasing pandemic and is caused by multiple factors including genotype, psychological stress, and gut microbiota. Our project investigated the effects produced by microbiota community, acquired from the environment and horizontal transfer, on traits related to obesity. The study applied a novel approach of raising Drosophila melanogaster, from ten wild-derived genetic lines on naturally fermented peaches, preserving genuine microbial conditions. Larvae raised on the natural and standard lab diets were significantly different in every tested phenotype. Frozen peach food provided nutritional conditions similar to the natural ones and preserved key microbial taxa necessary for survival and development. On the peach diet, the presence of parental microbiota increased the weight and development rate. Larvae raised on each tested diet formed microbial communities distinct from each other. In addition, we evaluated the change in microbial communities and larvae phenotypes due to the high fat and high sugar diet modifications. We observed that presence of symbiotic microbiota often mitigated the effect that harmful dietary modifications produced on larvae and was crucial for

Drosophila survival on high sugar peach diets. Although genotype of the host was the most influential factor shaping the microbiota community, several dominant microbial taxa were consistently associated with nutritional modifications across lab and peach diets. The effect that individual microbial taxa produced on the host varied significantly with changing environmental and genetic conditions.

DEDICATION

This dissertation is dedicated to my wife Shun Yan, who tirelessly supported me during the PhD journey and was my inspiration, during all these years.

iii

LIST OF ABBREVIATIONS AND SYMBOLS

16S rRNA 16S ribosomal ribonucleic acid gene

CAFE Capillary feeder

DGRP Drosophila Genetic Reference Panel

DSPR Drosophila Synthetic Population Resource

D Diet

G Genotype

NS Non-sterilized

P Colorimetric assay plate

PCR Polymerase chain reaction

PHF Peach high fat diet

PHFA Peach high fat autoclaved diet

PHS Peach high sugar diet

PHSA6 Peach high sugar autoclaved 6% diet

PHSA11 Peach high sugar autoclaved 11% diet

PR Peach regular diet

R Regular diet

RHF Regular high fat diet

RHS Regular high sugar diet

Ro Round

RT Radioactive tracer

S Sterilized

T Treatment

WGS Whole Genome Sequencing

ACKNOWLEDGMENTS

I greatly appreciate the help of my academic advisor Laura K. Reed. Laura has not only guided me through academic and scientific work but has been very understanding and attentive mentor, with whom I could always share my sorrows and happiness. During the exhausting PhD journey, I avoided a need for professional psychological help largely due to Laura. I would like to express my appreciation to my committee members: John Yoder, who always served me as an example, with his research enthusiasm and tireless experimenting work, and who’s influence shaped my understanding of what does it mean to be a scientist; Julie Olson, who’s expertise in microbiology and microbial databases has navigated me through the microbiological part of my work, and who always encouraged me to think broader; Stanislava Chtarbanova-Rudloff, who has provided her guidance and research materials for dissections, to whom, I always have been able to come to talk about research and any other issue and who has been very sympathetic and encouraging, which meant a lot to me; Casey Morrow, who’s help with sequencing, troubleshooting, and obtaining funding has been crucial for generating our dataset, and who was always very supportive. I would like to thank Juan Lopez-Bautista, who provided his guidance for my whole academic career, from whom, I first learned what does it mean to do a research, and who was not only my teacher but also my friend.

I would like to express my gratitude to my brother Sergei Bombin, who helped me with bioinformatical analysis, persevered all challenges of coming across the ocean with me, and a friendly competition with whom has kept me going through all these years. I would like to thank my students and researchers who put their time and effort in this work, without whom generating the sample size of more than 130,000 would be absolutely impossible, and who’s example fueled my ambitions: Owen Cunneely, Kira Ecikman, Abigail Ruesy, Rachael Cowan, Mengting Su, and Abigail Myers. As well as previous members of the team Caroline Hart, Ryan O'Rourke, and

Kara MacIntyre. I appreciate the help of my colleagues: Logan Griffin, who significantly improved the sterilization method that I applied, and who became my friend; as well as Vishal

Oza, Kelsey Lowman, Clare Scott, and YounJi Nam. I would like to thank the University of

Alabama, College of Arts and Sciences, and the Department of Biological Sciences that provided me with financial resources without which, I would not be able to afford to finish this degree.

I would like to thank my family; my mother Natalia Stroeva, without whom, I would not start my new life in the United States and most likely would not even become a scientist; my grandfather Vladilen Nichaev whose wisdom has guided me through my life, and from whom I learned most of my life skills; my grandmother Nadejda Nichaeva, and my sister Veronica

Bombina for her cheerful example. I would like to thank my wife Shun Yan for her love and to whom this work is dedicated.

vii

CONTENTS

ABSTRACT ...... ii

DEDICATION ...... iii

LIST OF ABBREVIATIONS AND SYMBOLS ...... iv

ACKNOWLEDGMENTS ...... vi

LIST OF TABLES ...... x

LIST OF FIGURES ...... xiii

CHAPTER ONE ...... 1

INTRODUCTION ...... 1

GOALS AND HYPOTHESES ...... 13

CHAPTER TWO ...... 16

INTRODUCTION ...... 16

MATERIALS AND METHODS ...... 19

RESULTS ...... 28

DISCUSSION ...... 55

CHAPTER THREE ...... 66

INTRODUCTION ...... 66

MATERIALS AND METHODS ...... 71

viii

RESULTS ...... 79

DISCUSSION ...... 148

CHAPTER FOUR ...... 164

SYNTHESIS ...... 164

CONCLUSIONS ...... 178

FINAL NOTES ...... 179

REFERENCES ...... 180

LIST OF TABLES

Table 2.1A: The difference in larvae phenotypes based on the diet ...... 30

Table 2.1B: The difference in fresh diets’ nutritional values ...... 30

Table 2.1C: The difference in fresh diets’ nutritional values ...... 30

Table 2.2: The influence of parental microbiota on larvae phenotypes ...... 33

Table 2.3: The contribution of diet, genotype, treatment, and their interactive effects on formation of larvae life history traits and metabolic phenotypes ...... 34

Table 2.4: Influence of a diet on alpha diversity measurements of larval bacterial community .. 36

Table 2.5: Influence of treatment on alpha diversity measurements of larval bacterial community ...... 43

Table 3.1: Comparison of phenotypes of the larvae raised on high fat and high sugar diets ...... 82

Table 3.2: The difference in phenotypes between the larvae with (NS) and without (S) parental microbiota ...... 87

Table 3.3: The influence of diet, genotype, treatment, and their interactive effect on phenotype of the larvae raised on high fat diets ...... 91

Table 3.4A: The influence of diet, genotype, treatment, and their interactive effect on phenotype of the larvae raised on high sugar diets ...... 91

Table 3.4B: The influence of diet, genotype, and their interactive effect on phenotypes of the larvae raised on high sugar diets ...... 91

Table 3.5: The influence of diet modification on larval phenotypes ...... 96

Table 3.6: The influence of diet on alpha diversity measurements of larval bacterial communities ...... 101

Table 3.7: Influence of a diet on bacterial community composition measured with Bray-Curtis distances ...... 103

Table 3.8: Influence of a diet on phylogenetic diversity of larval bacterial community measured with Weighed Unifrac distances ...... 107

Table 3.9: Influence of treatment on alpha diversity measurements of larval bacterial communities ...... 110

Table 3.10: Influence of treatment on bacterial community composition measured with Bray- Curtis distances ...... 110

Table 3.11: Influence of treatment on phylogenetic diversity of larval bacterial community measured with Weighted Unifrac distances ...... 112

Table 3.12: Influence of nutritional modifications on alpha diversity measurements of larval bacterial communities ...... 114

Table 3.13: Influence of nutritional modifications on bacterial community composition measured with Bray-Curtis distances ...... 115

Table 3.14: Influence of nutritional modifications on phylogenetic diversity of larval bacterial community measured with Weighted Unifrac distances ...... 118

xii

LIST OF FIGURES

Figure 2.1: The presence of dietary and/or maternally inherited bacteria substantially impacts metabolic and life history phenotypes in flies ...... 31

Figure 2.2: Larvae raised on peach and regular lab diets form distinct bacterial communities ... 36

Figure 2.3: Bacterial communities of the larvae raised on all of the diets could be differentiated by the abundance of dominant bacteria phyla, classes, and orders ...... 38

Figure 2.4: Bacterial communities of the larvae raised on all of the diets could be differentiated by the abundance of dominant bacteria families and genera ...... 39

Figure 2.5: Bacterial communities of the larvae raised on all of the diets could be differentiated by the abundance of all identified bacteria phyla, classes, and orders ...... 41

Figure 2.6: Bacterial communities of the larvae raised on all of the diets could be differentiated by the abundance of all identified bacteria families, genera, and ZOTUs ...... 42

Figure 2.7: Inheriting parental bacteria influences the formation of larvae symbiotic bacterial community with the effect being unequal among the diets ...... 44

Figure 2.8: Microbial communities of the larvae raised with or without parental bacteria could be differentiated by the abundance of all bacterial phyla, classes, and orders ...... 47

Figure 2.9: Microbial communities of the larvae raised with or without parental microbiota could be differentiated by the abundance of all bacteria families, genera, and ZOTUs ...... 48

xiii

Figure 2.10: Microbial communities of the larvae raised with or without parental microbiota could be differentiated by the abundance of dominant bacteria phyla, classes, and orders ...... 50

Figure 2.11: Microbial communities of the larvae raised with or without parental microbiota could be differentiated by the abundance of dominant bacteria families and genera ...... 50

Figure 2.12: The influence of symbiotic microbial taxa on larvae metabolic phenotypes vary with the diets ...... 53

Figure 2.13: The effect that bacterial taxa produce on larvae life history traits and metabolic phenotypes vary with the larvae diet ...... 54

Figure 2.14: The interaction between microbial genera varies between diets and treatments ..... 60

Figure 2.15: The effect of the whole bacterial communities on larvae metabolic phenotypes and life history traits varies with the diet...... 65

Figure 3.1: The presence of environmental and parental microbiota is beneficial for larvae raised on high fat diets ...... 83

Figure 3.2: The presence of environmental and parental microbiota is beneficial for larvae raised on high sugar diets ...... 84

Figure 3.3: The response of non-sterilized larvae to high fat nutritional modification varies between dietary types ...... 97

Figure 3.4: The response of sterilized larvae to high fat nutritional modification varies between dietary types ...... 98

xiv

Figure 3.5: The response of non-sterilized larvae to high sugar nutritional modification varies between dietary types ...... 99

Figure 3.6: The response of sterilized larvae to high sugar nutritional modification varies between dietary types ...... 100

Figure 3.7: Larvae raised on peach and lab diets form distinct microbial communities even with high fat and high sugar nutritional modifications ...... 104

Figure 3.8: Larvae raised on peach and lab diets form phylogenetically distinct microbial communities even with high fat and high sugar nutritional modifications ...... 107

Figure 3.9: Parental microbiota significantly influences the formation of bacterial community of the larvae, only on the diets subjected to HS modification ...... 111

Figure 3.10: The response of microbial community to dietary modifications varied with the origin of a diet and presence of parental microbiota ...... 116

Figure 3.16: Nutritional modifications cause shifts in bacterial composition and are associated with changes in abundance of the dominant bacteria phyla ...... 132

Figure 3.17: Nutritional modifications cause shifts in bacterial composition and are associated with changes in abundance of the dominant bacteria classes ...... 134

Figure 3.18: Nutritional modifications cause shifts in bacterial composition and are associated with changes in abundance of the dominant bacteria orders ...... 136

Figure 3.19: Nutritional modifications cause shifts in bacterial composition and are associated with changes in abundance of the dominant bacteria families ...... 138

Figure 3.20: Nutritional modifications cause shifts in bacterial community composition and are associated with changes in abundance of the dominant bacteria genera ...... 140

Figure 3.21: On high fat diets, the influence of bacterial taxa on larval phenotype varies with diet, treatment, and their interactive effect ...... 144

Figure 3.22: On high sugar diets, the influence of bacterial taxa on larval phenotype varies with diet, treatment, and their interactive effect ...... 145

xvi

Figure 3.23: The influence of bacteria abundance on larval phenotypes, varies with diet, genotype, and their interactive effect ...... 147

xvii

CHAPTER ONE

INTRODUCTION

The hologenome theory of evolution

In the 20th century the theory of modern synthesis was primarily nucleocentric, such that speciation was largely explained by Mendelian genetics and micro-evolutionary principles

(Bordenstein and Theis 2015). At that time, studies of host-microbial interaction were mostly limited to bacterial pathogenicity (Inglis 2007, Allen-Vercoe 2013). The major change in a view of the bacterial role in a host-microbial interaction happened with a development of the holobiont theory in 1994 by Richard Jefferson (Jefferson 1994). The theory states that a host and its commensal microbiota possess a metagenome that expresses a synergistic phenotype, which is subjected to evolutionary forces as one complex organism (Rosenberg et al. 2010, Zilber‐

Rosenberg and Rosenberg 2008). A phenotype of this unit could be varied by genome modifications of the host, as well as its commensal bacteria. Metagenomic changes induced by bacteria have more potential for genetic variability and could arise by altering dominant species of bacteria, as well as acquisition of new strains of microorganisms from an environment

(Bordenstein and Theis 2015). Impressively, the hologenome concept brings back several of

Lamarck’s principles of the evolutionary theory (Bordenstein and Theis 2015, Rosenberg,

Sharon, and Zilber‐Rosenberg 2009). Thus, the symbiotic microbiome or its parts may be acquired from an environment during the host’s life period and has a potential to be inherited

(Bordenstein and Theis 2015).

Symbiotic microorganisms play a crucial role in a host’s development, fitness, behavior, speciation, and other processes that had been thought to be independent from microbial activities

(Archie and Theis 2011, Stilling et al. 2014, Brucker and Bordenstein 2012). Interestingly, the molecular interaction between a host and its commensal microbiota resembles two types of interplay: intergenomic and genotype-by-environment interactions (Bordenstein and Theis

2015). In addition, a host organism possesses means to differentiate between commensal and pathogenic microorganisms (Shanbhag et al. 2017).

Currently, in the times when community DNA could be sequenced right from the environmental samples, also known as metagenomics, the complexity of the interaction between a host and its symbiotic microbiota is better understood and appreciated, allowing us to see that evolutionary forces influence a holobiont as one entity (Brucker and Bordenstein 2012, Gordon et al. 2013, Bordenstein and Theis 2015, Meyer et al. 2008). To investigate the role of microbiota in host’s health and development, one of the first steps is to determine the composition of microbiota which can be accomplished through marker gene profiling (Jandhyala et al. 2015).

This technique primarily consists of two steps: 16S rRNA sequencing and bioinformatics analyses of the resulted sequences for microbial taxonomic groups determination (Jandhyala et al. 2015). This method allows for collection of large amount of data, relatively inexpensively

(Meyer et al. 2008). The taxonomic data can be further used for inferring the metagenome which essentially enables predicting the functional abilities of microbiome, based on the sequencing of one gene (Langille et al. 2013). Works on organism interactions with their microbiota expanded the boundaries of microbiology research and showed that symbiotic bacteria could enhance host adaptive capabilities. (Shin et al. 2011, Ceja-Navarro et al. 2015). In fact, a bacterial genome often complements the host’s genome and could be solely responsible for the host's adaptation to

2 food toxins as well as contribute to a normal development and survival of the host (Ceja-Navarro et al. 2015).

Obesity and the link between gut microbiota and metabolic phenotype

Obesity is a worldwide epidemic that continues to grow and contributes to the development of various diseases, including but not limited to: type two diabetes mellitus, stroke, asthma, arthritis, coronary heart disease, arterial hypertension, all components of metabolic syndrome (Seganfredo et al. 2017, Wahba and Mak 2007, Thompson et al. 2007). Obesity does not only induce health risks but is also an important social factor. Stereotypically, the major cause of obesity is considered to be overeating and lack of exercise. Obese individuals are often stigmatized as lazy and unsuccessful, making them vulnerable to discrimination (Puhl and Heuer

2010). Thus, obesity may lead to physical health risks, moral suffering, and poor mental health of the patients. Contrary to the stereotypic view, the development of obesity and metabolic syndrome can be caused by genotype, epigenetic factors, sleep deprivation, malfunction of endocrine system, psychological stress, and gut microbiota, in addition to an excess of calorie intake and lack of sufficient physical activity (Seganfredo et al. 2017, Han and Lean 2016)

Therefore, an efficient treatment of obesity is complicated by the complexity of its origin

(Thompson et al. 2007, Kaur 2014).

Gut microbiota is one of the most important factors shaping metabolic phenotypes and, as a consequence, is a key element in the development of metabolic and autoimmune diseases, as well as cancer and asthma (Read and Holmes 2017, Leitão-Gonçalves et al. 2017). Alterations in gut microbiota biodiversity and community structure are correlated with the development of the obese phenotype (Turnbaugh et al. 2006, Flint, Duncan, and Louis 2017). Transfer of the microbiota from an obese to a lean, axenic (microbiota free) individual significantly increases

3 weight gain and adiposity, compared to axenic mice colonized with microbiota from a lean organism (Ridaura et al. 2013, Tilg and Moschen 2016, Turnbaugh et al. 2006). These results suggest that obesity can be transferred from one individual to another; therefore, exhibiting some characteristics of an infectious disease (Tilg and Moschen 2016). In addition, early life alterations of the microbiota, via antibiotic treatment or infant birth through Caesarean section, can increase a risk of developing obesity during the childhood (Tilg and Moschen 2016).

Different members of microbial community do not equally contribute to the host’s phenotype. Although a negative correlation between microbial diversity and metabolic syndrome was reported, recent studies suggested that presence of the specific microbial groups and the genetic constitution of the microbiome may well be a more reliable predictor of metabolic phenotype than overall diversity (Ma et al. 2017, Galenza et al. 2016, Flint, Duncan, and Louis

2017). In addition, the presence of particular microbial groups, such as Lactobacillus, is associated with high species richness of the microbial metagenome and therefore could serve as an initial predictor for a total gene count (Tilg and Moschen 2016).

Treatment of obesity is complicated and even with the most efficient treatments regain of weight is common and can reach up to 75% of the initial weight (Seganfredo et al. 2017). The methods suggested to combat obesity are restrictive diet/low calorie intake, exercise, weight reduction drugs, and bariatric surgery (Seganfredo et al. 2017, Kaur 2014). Interestingly, most of these treatments are associated with changes in gut microbiota composition (Seganfredo et al.

2017, Furet et al. 2010, Everard et al. 2013). Restriction diets and diets with high protein content reduce abundance and biodiversity of the symbiotic microbial community (Seganfredo et al.

2017). Bariatric surgery, one of the most efficient treatments for weight loss, has been shown to result in an abundance shift of the dominant microbiota strains (Zhang et al. 2009, Furet et al.

2010). The amount of the exercise is also positively correlated with the prevalence of

Akkermansia muciniphila, and treatment of obese mice with A. muciniphila improves insulin sensitivity, reduces fat mass and adipose tissue inflammation, and can partially reverse the obese phenotype induced by a high fat diet (Everard et al. 2013, Patterson et al. 2016). Thus, studying the composition of the gut microbiota and its influence on host metabolic phenotype could provide new options that combined with other therapies, such as exercise, would increase the success rate of the obesity and metabolic syndrome treatments.

One of the mechanisms through which symbiotic microbiota influence host metabolism is activating signal transduction pathways. Activation of the free fatty acids G-protein-coupled receptors inhibits the insulin signaling in adipose tissue and may induce lean phenotype, independent of calorie intake, through suppression of fat accumulation (Kimura et al. 2013, Tilg and Moschen 2016). In fact, microbiota derived short chain fatty acids may be the primary agonists for such receptors as Gpr43, the overexpression of which is negatively associated with adiposity (Tilg and Moschen 2016). Microbiota can also regulate host gene expression epigenetically via histone modifications (Davison et al. 2017).

Another mechanism used by gut microbiota to influence the host’s metabolism is by changing the quality of the food. Nutrition energy values of a particular diet can be modified via microbiome-based nutrient transformation, alteration of nutrient absorption, and production of additional digestive enzymes which produce more energy available to the host (Leulier et al.

2017, Turnbaugh et al. 2006, Huang and Douglas 2015, Davison et al. 2017). The microbiome of obese mice is enriched in glycoside hydrolases that increase host energy extracting capabilities

(Turnbaugh et al. 2006). Presence of Acetobacter tropicalis may reduce food sugar content up to

75% (Huang and Douglas 2015). Gut microbiota can also influence the host’s brain functions

5 connected with metabolic phenotype such as bulk food intake and nutrients preference (Leitão-

Gonçalves et al. 2017). Therefore, the microbiome plays an important role in a host’s response to a diet and can serve as a predictor of the individual’s metabolic response to dietary treatments

(Read and Holmes 2017).

Formation and persistence of the gut microbiota community

The digestive system microbiota community exhibits relative stability in its structure over time; however, it can be modified by host’s nutrition, stress levels, antibiotic treatments, and inter-microbial interactions (Mackie, Sghir, and Gaskins 1999, Tefit et al. 2017, Flint, Duncan, and Louis 2017). A newborn organism or the one that has been treated with antibiotics provides a mostly uninhabited environment for microorganisms. A new habitat supplies a variety of unoccupied ecological niches and the order in which species would colonize an environment would influence a final community structure (Li et al. 2004, Janos 1980). Successful introduction and succession of many organisms depend on the presence of the commensal taxa (Janos 1980,

Cortines and Valcarcel 2009). As in the case with most ecological habitats, the sequence of colonization of the host with bacterial species is an important determinant in community development (Li et al. 2004).

Commensal pioneer microbiota species provide necessary conditions for subsequent colonization and often prevent pathogenic bacteria from populating the host (Li et al. 2004,

Matamoros et al. 2013). In dental microbial communities, actinomyces and streptococci provide biofilms as surfaces for subsequent colonies to attach (Li et al. 2004). In addition, several species such as Streptococcus mitis and Streptococcus sanguinis are capable of producing hydrogen peroxide, which may assist in preventing pathogenic bacteria growth to the level at which they can cause periodontitis in the host (Li et al. 2004). In infants’ digestive systems,

Enterobacteriaceae drastically reduce oxygen concentration during early colonization, which is necessary for anaerobic bacteria's subsequent establishment (Matamoros et al. 2013). In addition, gut microbiota can actively respond upon the introduction of pathogenic species (Haag et al.

2012). For example, in infant mice upon the infection with Campylobacter jejuni, the microbiota composition shifts toward elevated concentration of Escherichia coli thus preventing the fixation of the pathogen (Haag et al. 2012). In addition, reintroduced commensal bacteria can outcompete undesirable strains such as antibiotic resistant Enterococcus, opening a possibility for the development of new treatments (Ubeda et al. 2013).

A Drosophila model for symbiotic microbiota research

The Drosophila intestinal tract has a similar structure and physiology as a vertebrate gut

(Ma et al. 2017). The microbiota in vertebrate models is estimated to contain around 1000 symbiotic taxa of the digestive system that are unculturable (Tilg and Moschen 2016). In contrast, the D. melanogaster gut microbiota consists of about 20 dominant strains of the culturable microorganisms that provide the relative simplicity necessary for advancing this field

(Leulier et al. 2017, Tefit et al. 2017). In contrast to other model organisms with simple gut microbiota, such as Caenorhabditis elegans, symbiotic microbiota of D. melanogaster has a more similar structure to vertebrates (Leulier et al. 2017). For example, the effect of several microbial strains, such as Lactobacillus, on their hosts is conserved across Drosophila and mammals (Leulier et al. 2017). In lab, D. melanogaster has a short generation time (14 days) which allows for a quick survey of the developing metabolic phenotype. In addition, the

Drosophila Genetic Reference Panel and The Drosophila Synthetic Population Resource offer a variety of diverse genotypes, with sequenced parental genomes, that allow for testing the microbiota effects across various genetic backgrounds and provide potential for studying genetic

7 interaction between host and its symbionts, even mapping the specific genetic loci responsible for the interactions (King, Macdonald, and Long 2012, Mackay et al. 2012).

D. melanogaster is a valuable model organism for studying human related metabolic diseases. Drosophila possesses organs that regulate metabolic homeostasis and are analogous

(and often homologous) to human organs (Leulier et al. 2017, Baker and Thummel 2007,

Bharucha 2009). A Drosophila digestive tract is compartmentalized and consists of foregut, midgut, and hindgut (Bharucha 2009, Lemaitre and Miguel-Aliaga 2013). Genes coding for distinct enzyme groups are differentially expressed through the digestive tract; suggesting that nutrients are processed sequentially through the gut (Lemaitre and Miguel-Aliaga 2013).

Drosophila’s fat body stores lipids and glycogen, as well as metabolizes nutrients and toxins, therefore performing functions similar with vertebrate’s adipose tissue and liver (Baker and

Thummel 2007, Bharucha 2009). In addition, the Drosophila brain has several areas responsible for metabolic regulation and food preference (Bharucha 2009, Leitão-Gonçalves et al. 2017). In

D. melanogaster, hemolymph sugar concentrations are detected by the ring gland and regulated by insulin production from glia cells (Strigini and Leulier 2016). Furthermore, Drosophila can develop obese and diabetic phenotypes (Bharucha 2009).

The influence of genotype and environment on the Drosophila Phenotype

D. melanogaster metabolism has been extensively studied and was shown to be influenced by genotype, diet, exercise, and symbiotic microbiota (Reed et al. 2010, Jumbo-

Lucioni et al. 2010, Musselman et al. 2011, Mendez et al. 2016, Newell and Douglas 2014).

Many genetic mechanisms that are responsible for regulation of the metabolic homeostasis are conserved between Drosophila and vertebrate models (Jumbo-Lucioni et al. 2010). For example,

Jumbo-Lucioni et. al (2010) showed that D. melanogaster body weight, glycogen, glycerol,

8 triacylglycerol, and metabolic rate are all significantly influenced by genotype and could be mapped to specific loci in the genome. In addition, they reported that loci that were correlated with the difference in metabolic responses were enriched in genes coding for hydrolase and alpha-glucosidase enzymes. Interestingly, the enzymes with those same enzymatic functions are also encoded in symbiotic bacterial genomes that were shown to increase energy extracting capabilities of the host (Turnbaugh et al. 2006). These findings provide food for thought about the deep evolutionary conservation of the catabolic mechanisms and commensal interaction within a holobiont. In these results, one can see an analogy with evolutionary effects of gene duplication where the function of the ancestral gene can be shared between paralogs within one genome (Force et al. 1999).

Diet and genotype influence metabolic phenotypes and longevity of Drosophila. Raising

D. melanogaster larvae on high sugar diet (86.4% of total calories available from carbohydrates) resulted in a smaller body size, delayed development, elevated glucose, trehalose, and reduced glycogen concentrations of the larvae (Musselman et al. 2011). In addition, those larvae developed a reduced insulin sensitivity and had higher triglyceride concentrations, overall exhibiting a diabetic phenotype (Musselman et al. 2011). A high sugar diet also significantly decreased the pupal weight (Reed et al. 2010). Obesity, induced by high sugar diet, increased with aging of the flies (Skorupa et al. 2008). A high fat diet can also produce an obese phenotype in Drosophila, which resulted in higher triglyceride concentrations and disturbs insulin homeostasis (Birse et al. 2010). In addition, a diet rich in fats led to similar effects in body mass and development rate as a high sugars diet and also produced transgenerational effects

(Musselman et al. 2011, Dew-Budd, Jarnigan, and Reed 2016). An obese phenotype is harmful for Drosophila fitness. Higher fat mass storage was shown to reduce longevity of flies and

9 resulted in lower fecundity (Moghadam et al. 2015). On the opposite side of the spectrum, diets restricted in sugars and/or yeast consumption were shown to increase life span (Mair, Piper, and

Partridge 2005).

However, diet alone could not be the only predictor of the metabolic phenotype. The phenotypic response to a diet modification often varies with the genotype (Reed et al. 2010,

Reed et al. 2014). In fact, diet-by-genotype (DxG) interaction may explain more variance than diet alone in the metabolic response of such traits as triglyceride and carbohydrates concentrations (Reed et al. 2010). In addition, recent findings showed that genotype-by-diet interactions significantly influences metabolomic profiles; hence, laying the foundation for explaining the mechanism through which DxG influences metabolic traits (Reed et al. 2014,

Williams et al. 2015).

The influence of the symbiotic microbiota on Drosophila phenotype and fitness

Symbiotic microbiota play an important role in D. melanogaster development and metabolic phenotype. Axenic flies have longer development time, lower weight, protein, and glycogen content but higher free glucose and triglyceride concentrations (Newell and Douglas

2014, Dobson et al. 2015, Huang and Douglas 2015). Axenic Drosophila, exhibit phenotype similar with flies raised with harmful nutritional modifications, such as high sugar and high fat diets. Although axenic Drosophila consume less food, their energy storage index stays significantly higher than that of the conventional flies (Wong, Dobson, and Douglas 2014). In addition, presence of the commensal microbiota allows Drosophila to maximize lifespan and reproductive output (Leitão-Gonçalves et al. 2017). Dobson et. al (2015) showed that the abundance of specific bacterial taxa was associated with a change in a metabolic phenotype:

Acetobacter, Gluconobacter and Komagataibacter were negatively correlated with a fly’s energy

10 storage index. Lactobacillus was positively associated with triglyceride concentration and

Achromobacter and Xanthomonadaceae were positively correlated with glycogen concentrations

(Chaston et al. 2016). In addition, Drosophila microbiota varies across genetic backgrounds which makes it possible to establish associations between host’s genes and microbiota dependent metabolic responses as well as particular symbiotic species (Chaston et al. 2016, Dobson et al.

2015, Early et al. 2017).

Symbiotic microbiota allow Drosophila to overcome nutritional limitations of their diet.

Shin et. al (2011) showed that axenic larvae raised on a casamino acid diet experienced 90% body size reduction and were not able to survive to form pupae. However, the presence of only one bacterial species, Acetobacter pomorum, could restore survival and the normal rate of larval development via induction of the hosts’ insulin-like growth factor signaling (Shin et al. 2011).

Leitão-Gonçalves et. al (2017) demonstrated that axenic flies express a strong preference for yeast-rich food due to the demand for essential amino acids. Acetobacter promorum and serval

Lactobacillus species are able to suppress the yeast appetite in Drosophila and shift preference for nutrition toward high-sugar concentration diets (Leitão-Gonçalves et al. 2017). The change in nutritional preference may be explained by competition between the host and its symbiotic microbiota for available sugars and through production of essential amino acids by the microbial community (Leitão-Gonçalves et al. 2017). This explanation finds further support in the work of

Huang et. al (2015) which showed that the presence of A. tropicalis in the food can reduce glucose content of the diet up to 75%, which results in lower triglyceride and glucose concentrations of its Drosophila melanogaster host. In addition, symbiotic microbiota may supply its host with short chain fatty acids and vitamins (Ma et al. 2017, Wong, Dobson, and

Douglas 2014).

The formation and persistence of the Drosophila gut microbiota community

Microbial composition of lab dwelling Drosophila primarily consists of

Acetobacteraceae, Enterococcaceae, and Lactobacillaceae families (Chandler et al. 2011, Early et al. 2017). Within these families, the influence of Acetobacter tropicalis, Enterococcus faecalis, Lactobacillus brevis, and L. plantarum on host metabolic phenotypes have been studied more than others (Early et al. 2017, Shin et al. 2011, Huang and Douglas 2015). However, the microbiota of wild fly populations are more diverse and differ in the abundance of the dominant species (Chandler et al. 2011). In the lab, Drosophila raised on fruits still exhibit a more complex and diverse community of symbiotic microbes compared to conventionally raised flies (Vacchini et al. 2017). Lab food preservatives, especially methylparaben sodium salt (moldex), largely contribute to the difference between natural and laboratory associated microbial communities

(Tefit et al. 2017).

Gut microbial composition of D. melanogaster can be easily manipulated. During colonization of a fruit, Drosophila inoculate the substrate with their microbiota (Morais et al.

1995). In addition, when females deposit embryos on a substrate, parental microbiota resides on the chorion of the egg (Ryu et al. 2008). The initial microbiota is gained by the first instar larvae, through consumption of the chorion (Ryu et al. 2008). Later, during feeding, larvae acquire additional microorganisms from the environment, representatives of which remain in the larval and pupal intestine until eclosion of the adult fly (Ridley et al. 2012). Removing the chorion and its microbiota then placing the egg on sterile food allows us to obtain axenic larvae.

Drosophila’s commensal microbial composition can also be influenced by nutritional conditions.

A diet with a high glucose concentration increases the biodiversity of microbiota (Galenza et al.

2016).

D. melanogaster is a frugivorous species which feeds and reproduces on decaying fruits

(Strickberger 1962, Sturtevant 1916). Therefore, to maximize the number of microbial taxa that a

Drosophila is exposed to, it is necessary to raise flies on naturally rotten fruits. However, previous research suggested that during the decay process the chemical composition of the fruit and its microbial community structure change, which makes it problematic to determine at what time point microbial community would be optimal for fly development (Hoffmann 1985, da

Veiga Moreira et al. 2013). Several studies evaluated the influence of the food rotting stages on

Drosophila behavior and fitness. Nunney (1990) showed that pierced oranges that were left outdoors for four days produced the largest number of flies. However, D. melanogaster females prefer to oviposit on apples and oranges that have been rotting for nine days (Hoffmann 1985).

In addition, among mycophagous Drosophila, adults and larvae were most attracted to mushrooms that had been decaying for four days (Kimura 1980). Results of the previous research did not provide a unanimous opinion about the optimum state of decaying material to raise Drosophila and suggested that it may vary, depending on the substrate type.

GOALS AND HYPOTHESES

The first goal of my doctoral studies was to develop methodology for establishing a natural microbiota of the fly gut in the lab environment and evaluate its influence on metabolic phenotypes in several genetic backgrounds. I hypothesized that flies raised on naturally rotten peaches would exhibit a distinct microbial community from larvae raised on lab diets. Previous research identified that wild populations of Drosophila exhibit a higher diversity of

Acetobacteraceae and Enterococcaceae with increased abundances of Gluconobacter,

Acetobacter malorum, A. pomorum, and Enterococcus respectively, as well as more representatives of rare taxa, such as Clostridiales (Chandler et al. 2011). Therefore, we expect

13 that flies that raised on the natural diets will have a similar microbial profile. In addition,

Drosophila raised on natural microbial communities would have lower triglyceride and glucose concentrations due to the previously demonstrated inverse relationship between the diversity of the gut microbiome and these phenotypes. Acetobacter is well known for reducing available dietary sugars; therefore, we hypothesize that due to higher abundances of Acetobacter in natural diets the sugar concentrations in food and flies will be lower, as well as utilized more quickly by the holobiont after consumption than in the sterilized diet which will at least partially explain the difference in observed phenotypes. My second goal was to tudy the difference in metabolic response due to the supplementation of nutrients (additional fats and sugars) using the standard lab diet and the natural food/microbiota composition. I hypothesized that, overall, the high fat and high sugar food would cause an obese phenotype with higher triglyceride and glucose concentrations, as well as a significant difference in weight. In addition, the negative effects of high fat and high sugar diets would be reduced in flies eating the peach-based diet, due to the higher microbial diversity. We expected that flies raised on naturally rotten peaches would contain representatives of Clostridiales order, of which members such as Clostridium are known for their lipolytic activity. Third, I would evaluate how succession of parental pioneer microbiota species influenced metabolic phenotypes. I hypothesized that larvae that were cleared from their parental microbiota (via embryo sterilization with the removal of the chorion) would acquire different bacterial community composition than those that received the parental microbiome due to pioneer species effects on the construction of the ecosystem of the gut, and as a consequence will develop different metabolic phenotypes. Finally, I wanted to characterize the interactive effects between diet, genotype, parental, and environmental microbiota on metabolic phenotype. I hypothesized that interactive effects would explain a large portion of

14 variation in the Drosophila metabolic response. We expected to see that higher abundances of

Acetobacterecae and Clostridiales on the natural diets would lead to more rapid utilization of dietary sugars and lipids, thus producing a leaner phenotype on the peach-based diets compared with lab diets. However, we also expected that the effect produced by individual microbial taxa would vary with diet, genotype, and presence of parental microbiota.

I present the major findings of my research focused on these four goals in the following two dissertation chapters.

CHAPTER TWO

INFLUENCE OF LAB ADAPTED NATURAL DIET AND MICROBIOTA ON LIFE HISTORY AND METABOLIC PHENOTYPE OF DROSOPHILA MELANOGASTER

INTRODUCTION

The holobiont theory states that a host and its commensal microbiota possess a metagenome that expresses a synergistic phenotype, which is subjected to evolutionary forces as one complex organism (Rosenberg et al. 2010). A phenotype of this unit could be varied by genome modifications of the host, as well as its commensal bacteria. Metagenomic changes induced by bacteria have more potential for genetic variability and could arise by altering dominant species of bacteria, as well as acquisition of new strains of microorganisms from an environment

(Bordenstein and Theis 2015).

Initial parental microbiota is deposited on the chorion of Drosophila embryo (Ryu et al.

2008). In addition, fruit colonized by Drosophila is inoculated with flies’ symbiotic microbiota, simultaneously (Morais et al. 1995). Thus first instar larvae obtain their earliest symbiotic microbial community by consuming the chorion and further modifies this community through feeding on the inoculated fruit (Ryu et al. 2008, Ridley et al. 2012). Wong et. al (2015) showed that parental microbiota transferred with the chorion of the egg could modify the microbial community composition in a food substrate and in the offspring. In addition, the transfer of

16 axenic Drosophila on food substrate can change the food microbial community to resemble the symbiotic microbiota composition that would develop in the host (Wong et al. 2015).

One of the benefits that Drosophila microbiota provides to the host is the ability to overcome nutritional limitations of the food. Axenic larvae raised on the casamino acids diet, lacking any protein supply experienced dramatic reduction in size and were not able to enter a pupation developmental stage (Shin et al. 2011). The adverse effects of a low protein diet could be mostly rescued by the presence of symbiotic microbiota that can induce host’s insulin-like growth factor signaling (Shin et al. 2011). Symbiotic microbiota is also likely to be able to supply host with essential amino acids, thus reducing its appetite for protein rich diet to sugar rich diet (Leitão-Gonçalves et al. 2017). Natural fly food is poor in proteins therefore, the presence of symbiotic microbiota is greatly beneficial for Drosophila development in the native environment (Pais et al. 2018, Bing et al. 2018).

The microbial community of flies raised on lab diets is dominated by Lactobacillaceae,

Acetobacteraceae, and Enterococcaceae. However, the microbiota of wild fly populations may differ in the diversity and the abundance of dominant species. However, previous research indicated that Drosophila in natural population and raised on fruits in the lab possessed a more complex microbiota composition (Chandler et al. 2011, Vacchini et al. 2017). Lab food preservatives, especially methylparaben sodium salt (moldex), significantly contribute to the difference in gut microbiota structure between lab and wild flies (Tefit et al. 2017). In addition, microbial taxa associated with natural Drosophila populations may exhibit different qualities from lab adapted microbial taxa, such as more efficient and stable colonization of the host (Pais et al. 2018) Therefore, studying the evolutionary relationship of Drosophila and its microbiota, as well as the symbiont’s influence on fly’s metabolic phenotype only on standard lab

17 microbiota, may be insufficient to understand the natural relationship and co-evolution of the fly and its microbiota.

With this work we wanted to address a series of specific questions and hypotheses:

1) How does a natural diet with a naturally occurring and/or maternally inherited communities of microbes influence the life history and metabolic phenotypes of flies relative to a standard lab diet? Is there genetic variation in the phenotypic response to nutritional change and dietary and parental microbiota availability?

1.1 We hypothesized that larvae raised on a natural diet will exhibit different life history traits and metabolic phenotypes, comparing with larvae raised on a standard lab diet.

1.2 We hypothesized that the presence of a maternally transmitted microbiota will significantly impact larval phenotypes, and that this impact may vary across dietary treatments.

1.3 Given prior findings on the roles of genetic variation on metabolic phenotypes, we hypothesized that there is genetic variation in phenotype that interacts with the dietary conditions and the availability of maternally transmitted microbiota.

2) Will the symbiotic microbiota community of the larvae raised on the natural diet be different from the lab food raised larvae? Will the presence of maternally inherited microbiota influence the formation of microbial communities? Is the microbiota community variable across host genotypes?

2.1 We hypothesized that the gut bacterial community composition and diversity will vary substantially across both dietary and parental microbiota conditions.

2.2 We hypothesized that the maternally transmitted microbiota will have “founder effects” in the formation of the larval gut microbiome.

2.3 We hypothesized that the composition of the microbial community will exhibit variation with host genotype.

3) Will specific microbial taxa and/or microbiota communities as a whole influence the larvae phenotypes differently across diets and genotypes?

3.1 We hypothesized that some microbial taxa will have consistent correlations with host phenotype across diets and treatments while others will have a diet or treatment specific relationship.

MATERIALS AND METHODS

Food preparation

On August 28th, 2017, we put approximately 200 peaches outdoors and allowed them to decay for six days. On September 3rd 2017, the fruits were collected, manually ground, and stored in freezers at -20 °C. Peach food (PR) preparation protocol was the following: we allowed approximately one liter of the peach food to thaw, homogenized it with an immersion blender, and distributed it into vials, with approximately 10 ml of food per vial. In order to prepare autoclaved peach food (PA), vials containing peach food were autoclaved for 25 min at 121 °C.

Regular Drosophila lab food (R) was cooked according to the protocol described in previous works (Dew-Budd, Jarnigan, and Reed 2016, Mendez et al. 2016).

To ensure that the autoclaved peach food did not contain any live microorganisms, we used 1g of autoclaved and non-autoclaved peach materials and diluted them in 9 ml of sterile

Phosphate-buffered saline solution (PBS) (Leitão-Gonçalves et al. 2017, Carvalho, Kapahi, and

Benzer 2005). Then, we performed a serial dilution in PBS to get food dilutions (Leboffe and

Pierce 2012). We mixed 1 ml of each dilution with standard methods agar (Criterion) via the pour plate method (Leboffe and Pierce 2012). The agar was prepared according to the manufacturer's directions. Samples were incubated at 35 °C for 48 hours and visually evaluated for the presence of microbial colonies (Maturin, Peeler, and Drug Administration 2001). The independent variable for the diet component will be referred to as D.

Drosophila stocks and husbandry

We used 10 naturally derived genetic lines created by the DGRP2 project: 142, 153, 440,

748, 787, 801, 802, 805, 861, and 882 (Mackay et al. 2012, Huang et al. 2014). Stocks were maintained at constant temperature, humidity and light/dark cycle on a molasses-based lab diet as described in previous works (Reed et al. 2014, Dew-Budd, Jarnigan, and Reed 2016, Mendez et al. 2016). The independent variable for the genetic component will be referred to as G.

Drosophila embryos sterilization

In order to remove parental microbiota, we sterilized ~12-hour old embryos with subsequent two-minute washes in 2.5% active hypochlorite solution, 70% ethanol solution, and sterilized distilled water (Leitão-Gonçalves et al. 2017, Carvalho, Kapahi, and Benzer 2005).

After sterilization, embryos were placed on the apple agar plates, and incubated until the first instar stage under fly-rearing conditions described above (Ashburner 1989). The non-sterilized control embryos (NS) were allowed to develop for ~24 hours (until the 1st instar larvae stage) on

20 the apple agar plates, on which they had been deposited. In order to demonstrate that sterilized embryos did not possess parental microbiota, 20 sterilized 1st instar larvae were collected and grinded in 200 l of the sterile PBS using a mechanical homogenizer. The resulting mixtures were plated on nutrient and standard method agars (Criterion). The agars were prepared according to the manufacturer's directions. Plates were incubated for 48 hours at 35 °C and visually evaluated for the presence of microbial colonies (Maturin, Peeler, and Drug

Administration 2001). NS larvae were used as the positive control according to the same procedure. The independent variable for the sterilization treatment component will be referred to as T.

Larvae rearing and collection

In three separate time periods (~ 30 days apart), 50 sterilized and non-sterilized larvae, of each genetic line were put in at least three vials of PA, PR and R food, each. The independent variable for the time period (round) component will be referred to as Ro. Larvae were allowed to develop until the late third instar wandering stage (when they stopped moving but before the pupation started) and then collected in micro centrifuge tubes with sterile Ringer’s solution

(Ashburner 1989). Each vial was checked for the presence of larvae, at the right developmental stage, four times per day at 9 am, 11 am, 2:30pm, and 5 pm for 18 days after larval colonization.

Larvae were inspected for the presence of any damage (damaged ones were removed), cleaned with at least two washes in a sterile Ringer’s solution, and stored in the Ringer’s solution at -20

°C in Eppendorf tubes with 10 larvae per tube.

Measuring Experimental Phenotypes

Survival. The total number of third instar larvae collected per vial was divided by the number of first instar larvae used to seed the vial (50) to evaluate the proportion of larvae that survived till the late 3rd instar stage. Developmental rate: The developmental rate, in days, was calculated for each larva individually, from the day it was put in the food vial to the collection date. We then calculated median developmental time per vial and used it in our statistical analysis (Ridley et al. 2012). Weight: In order to measure the dry weight, larvae were taken from the -20 °C freezer, allowed to reach room temperature and placed in a VWR standard oven at 37

°C overnight. After drying, each larva was weighed individually using Mettler Toledo XS 105 microbalance. Weights were recorded with LabX direct software v. 2.2. Triglyceride: With the exception of four samples (due to low survival of larvae of certain D/G/T/Ro combinations), we homogenized 10 larvae per sample to determine total triglyceride concentration using the Sigma

Triglyceride Determination Kit (Clark and Keith 1988, De Luca et al. 2005, Reed et al. 2010,

Dew-Budd, Jarnigan, and Reed 2016). Results were adjusted to represent the average triglyceride concentration per mg of dry larval weight. Protein: Protein concentrations were quantified using the Bradford’s method with 10 homogenized larvae per sample (with the exception of 3% of the samples in which we used one to nine larvae, due to especially low survival rates of the specific groups) (Bradford 1976, Dew-Budd, Jarnigan, and Reed 2016). Protein concentrations were averaged to represent the protein concentration per mg of dry larvae weight. Glucose: For most of the samples, combined trehalose and glucose concentrations were quantified via homogenization of 10 larvae (with the exception of 5% of the samples in which we used four to nine larvae) with subsequent overnight incubation in 1 μg/mL trehalase solution and further application of the Sigma Glucose Determination Kit (Rulifson, Kim, and Nusse 2002, Reed et al.

2014, Dew-Budd, Jarnigan, and Reed 2016). Glucose concentrations were averaged and adjusted to represent the amount of glucose per mg of dry weight.

To assess the triglyceride, protein and glucose concentrations in the diets, we used freshly unfrozen food and unfrozen food that was incubated in the food vials for seven days at 25 °C, in the same incubator as the experimental fly stocks, where larvae development was taking place.

For the food assays, we used the same procedure as for the larvae but with ~12.35 mg of the food sample. For the glucose assay, instead of incubating samples in trehalase, we incubated them in

100μl of 1 mg/μl solution of invertase (to convert sucrose into glucose) (Ward's Natural Science) overnight at 37 °C. For analysis, the results were adjusted to represent the amount of the measured compound per mg of the sample.

DNA extraction and sequencing

DNA was extracted from 10 larvae with the Qiagen blood and tissue DNA extraction kit according to the standard protocol, with overnight incubation of the samples in proteinase K at

56 °C. DNA extractions were used for sequencing the V4 region of the 16S ribosomal RNA subunit (16S rRNA), which was performed in the Microbiome Core Facility of The University of

Alabama in Birmingham, AL according to the previously published method on the Illumina

MiSeq platform (Kumar et al. 2014).

Trimmomatic-0.36 (Bolger, Lohse, and Usadel 2014) was used to process demultiplexed

DNA sequences. Standard Illumina-specific barcode sequences, all sequences with less than 36 bases, and leading and trailing low-quality bases were all removed using the default settings of the Trimmomatic-0.36 program. The USEARCH-fastq_mergepairs tool was used to combine forward and reverse readings. All reads with an expected error greater than 1 were removed, as

23 well as chimeric reads and singletons. Combined readings without a merging pair were filtered using fastq_filter command. The -unoise3 tool was used to cluster readings into zero-radius OTU

(ZOTU) 100% identity. OTUs were then designated with the lowest taxonomic rank using the

UCULT algorithm implemented in QUIME 1.9.1 (Caporaso, Kuczynski, et al. 2010, Edgar

2010) along with SILVA reference database version 132 (Quast et al. 2012). Using SILVA v.

132 database, PyNAST (Caporaso, Bittinger, et al. 2010) with default options was used for sequence alignment. The phylogenetic trees of ZOTUs and OTUs were assembled using the default options of QUIIME 1.9.1 with the FastTree program (Price et al. 2009). Alpha and beta diversities were rarefied with QUIIME 1.91 -single_rarefaction.py using the -- subsample_multinomial option in order to subsample the replacements. Rarefaction for all samples was performed to the depth of 4,500 reads. This was the lowest possible number of reads between samples.

Statistical Analysis

Data transformation. Normality tests, data transformations and statistical models were done with JMP Pro 14.0. Phenotype measurements were tested for normality with the Shapiro-

Wilk test and an outlier box plot. Only larvae survival (ST. 1) showed normal distribution.

Therefore, all other phenotypic measurements data were transformed. We performed a cube root transformation on the data for development rate and glucose by weight, a square root transformation on data for weight and protein by weight, and a log transformation on data for triglyceride by weight concentrations. The bacterial abundance was log(x+1) transformed for all parametric analyses.

Bacterial Diversity. Alpha and beta diversities were computed in QIIME v. 1.9.1. To estimate alpha diversity, we used Shannon, Simpson, and PD Whole Tree metrices. As all of the alpha diversity indices were not normally distributed, we performed a pairwise comparison of them with a Wilcoxon rank-sum test in R v. 3.5.1 with “matrixTests” package v. 0.17 (Bruno et al. 2019). Beta diversity was estimated with Bray-Curtis and Weighted Unifrac distances. The similarity between each sample’s beta diversity distance was evaluated via hierarchical clustering, applying a ward method for distance calculation, and visualized with a constellation plot in JMP v. 14.0.

Statistical modeling: In order to evaluate the contribution of each variable and their interactive effect to each phenotypic development, we used standard least squares model with model effects to include Diet (D), Genotype (G), Sterilization Treatment (T) and their specific interactive effect: D*G (diet-by-genotype), D*T (diet-by-treatment), G*T (genotype-by- treatment), D*G*T (diet-by-genotype-by-treatment). In order to verify that the built model fits the data, a Lack of Fit test was performed. If the time period of the experiment (Ro) and/or the variance between the colorimetric assay runs (triglyceride, protein, and glucose) (P) produced a significant effect, these variables were included in the model’s effects, unless their addition caused the model to fail the lack of fit test. Thus, the models for larvae survival, development time, and weight were the following:

푦 = 훽 + 훽퐷 + 훽퐺 + 훽푇 + 훽(퐷 ∗ 퐺) + 훽(퐷 ∗ 푇) + 훽(퐺 ∗ 푇)

+ 훽(퐷 ∗ 퐺 ∗ 푇) + 휀

For triglyceride by weight:

푦 = 훽 + 훽퐷 + 훽퐺 + 훽푇 + 훽(퐷 ∗ 퐺) + 훽(퐷 ∗ 푇) + 훽(퐺 ∗ 푇)

+ 훽(퐷 ∗ 퐺 ∗ 푇) + 훽 푃 + 휀

And for protein and glucose by weight:

푦 = 훽 + 훽퐷 + 훽퐺 + 훽푇 + 훽(퐷 ∗ 퐺) + 훽(퐷 ∗ 푇) + 훽(퐺 ∗ 푇)

+ 훽(퐷 ∗ 퐺 ∗ 푇) + 훽 푃 + 훽푅 + 휀

Where 푦 is the response, 훽 values are constants, and 휀 is a random error term.

All models for interactive effects of diet, genotype, and treatment were done with all 10 genetic lines, with the exception of glucose which was done without 861 due to the low survival rate of this genetic line.

In order to verify that the built model fits the data, a Lack of Fit test was performed. Only the models with non-significant lack of fit p-value were kept. To assess the pairwise difference between diets (R vs PR and PR vs PA) and treatments (NS vs S), we used the least square model with one main explanatory variable of interest (Diet or Treatment). We also included time period and assay plate variance as additional explanatory variables if they produced a significant effect.

The pairwise comparisons were performed with Post-Hoc Student’s t-test.

Bacterial abundance. In order to evaluate if the diet and treatment could serve as categorical predictors for classification of the larvae bacterial samples, we performed discriminant analysis at phylum, class, order, family, and genus taxonomic levels, as well as at the level of individual ZOTUs. The results were visualized with a canonical plot in JMP v. 14.0

(JMP manual). For the 10 most abundant representatives of each taxonomic level, we applied the linear covariance method for the discriminant analysis which allowed us to visualize the

26 covariates in the form of vector rays. Using this method allowed us to represent which of them drove the separation of the clusters (JMP manual). The length of the ray is correlated with the strength of the impact that it produced on the samples to be separated in the vector direction, on a canonical plot. When we ran the analysis with all identified taxa, we applied a wide linear method for the discriminant analysis. To compare the abundance of bacterial taxa between the diets and treatments, we performed a Wilcoxon test as described above. The p values were adjusted for the false discovery rate with Benjamini-Hochberg correction and added to data tables as FDR p. The threshold for the value of FDR p that should be considered significant could be subjective and vary from 0.25 to 0.05 among microbiology studies (Wu et al. 2011,

Bruce-Keller et al. 2015). To evaluate the interactive effect of the variables on the abundance of each identified bacterial taxa, we used the three-way linear interaction model.

푦 = 훽 + 훽퐷 + 훽퐺 + 훽푇 + 훽(퐷 ∗ 퐺) + 훽(퐷 ∗ 푇) + 훽(퐺 ∗ 푇)

+ 훽(퐷 ∗ 퐺 ∗ 푇) + 훽푅 + 휀

In order to identify the correlations between larvae phenotypes and bacterial abundances, we found the mean phenotype for each combination of diet, genetic line, treatment, and round.

Spearman’s rank correlation between the microbial abundances and tested phenotypes was calculated with Hmisc v. 4.3-0 in R v. 3.5.1, with the adjustment of p values for FDR p as described above. We also tested the possible interactive effect of each identified bacterial taxa and one of the independent variables on the formation of the tested phenotypes according to the formula:

푦 = 훽 + 훽푥 + 훽푥 + 훽(푥 ∗ 푥)+ 훽푅 + 휀

where, 푥 was the abundance of the microbial taxa, 푥was one of the independent variables (D, G, or T) and R was time component. Development, weight, triglyceride, protein, and glucose were normalized with log, square root, log, log, and cube root, transformations respectively.

RESULTS

1) There are substantial differences in the life history and metabolic phenotypes for larvae raised on a natural peach diet with a naturally occurring and/or maternally inherited community of bacteria relative to a standard lab diet

1.1 Larvae raised on a natural diet exhibited different life history traits and metabolic phenotypes from larvae raised on a standard lab diet. Survival. The proportion of larvae that survived on the lab diet was significantly higher than the larval survival on a natural peach diet regardless of sterilization (Fig. 2.1A, Table 2.1A & S2.1.1). The undisturbed bacterial community of the peach diet produced a significant positive effect on larvae survival, when compared with the autoclaved peach diet (Fig. 2.1A, Table 2.1A & S2.1.1). Development rate. Within both controlled and sterilized treatments, larvae developed faster on the regular lab diet compared to the natural diet and faster on the regular peach food compared to the autoclaved peach food (Fig. 2.1B, Table

2.1A & S2.1.1). Weight. Third instar larvae raised on the lab food were significantly heavier than those that were raised on the natural diet (Fig. 2.1C, Table 2.1A & S2.1.1). Among the peach diets, larvae consuming the autoclaved diet were significantly lighter (Fig. 2.1C, Table

2.1A & S2.1.1), suggesting that under natural nutritional conditions, mmicroorganisms in the food substrate facilitate growth and weight gain of the larvae. Triglyceride. Although fresh and incubated R food had higher triglyceride concentrations than the PR food (Tables 2.1B, C), larvae raised on the PR diet had significantly higher triglyceride concentrations by weight than

28 those that were raised on a lab diet, independent of sterilization treatment (Fig. 2.1D, Table 2.1A

& S2.1.1). Incubated autoclaved peach food had significantly higher triglyceride content compared with regular peach food (Table 2.1C) and produced larvae with higher triglyceride concentration by weight (Fig. 2.1D, Table 2.1A & S2.1.1). Protein. Independent of the treatment, larvae raised on the regular lab food had higher protein by weight concentrations compared to larvae raised on peach food (Fig. 2.1E, Table 2.1A & S2.1.1). Larvae raised on the

PA diet had significantly higher protein by weight levels compared to PR raised larvae, but only in the absence of parental bacterial community (Fig. 2.1E, Table 2.1A & S2.1.1). Evaluating the protein quantity in fresh food itself, we found significantly higher protein concentration in the regular lab diet compared to the peach diet (Table 2.1B). Between fresh peach diets there was no significant difference in protein (Table 2.1B). However, after incubation, autoclaved peach food had significantly less protein than regular peach food (Table 2.1C). Glucose. Larvae raised on a lab food diet had significantly higher glucose concentrations than larvae raised on the peach food diet (Fig. 2.1F, Table 2.1A & S2.1.1). Larvae that consumed PR food had lower glucose by weight concentrations compared with larvae raised on the autoclaved version, which was consistent with our 1.1 hypothesis (Fig. 2.1F, Table 2.1A & S2.1.1). Interestingly, the fresh peach food itself had a significantly higher glucose concentration than the lab food (Table 2.1B).

However, after incubating the peach food, the concentration of glucose was lower in PR than in both R and PA diets (Table 2.1C), suggesting a strong impact from the live bacterial community.

Treatment Survival Development Weight Triglyceride Protein Glucose NS R> PR ** R< PR *** R> PR *** R< PR *** R> PR ** R> PR *** S R> PR ** R< PR *** R> PR *** R< PR *** R> PR *** R> PR *** NS PR> PA *** PR< PA *** PR> PA *** PR< PA *** PR< PA PR< PA ** S PR> PA *** PR< PA *** PR> PA *** PR< PA *** PR< PA *** PR< PA ***

Table 2.1A: The difference in larvae phenotypes based on the diet. Comparison of larvae life history traits and metabolic phenotypes between larvae, from all 10 genetic lines, raised on regular lab diet (R), peach diet (PR), and autoclaved peach diet (PA). NS stands for non- sterilized larvae, S stands for sterilized larvae. Asterisks indicate the significance of comparisons p< 0.001 ***, p< 0.01 **, and p< 0.05 *.

Triglyceride R > PR R > PA PR > PA fresh food p =0.0344 p <0.0001 p= 0.0003 Protein R > PR R > PA PR < PA fresh food p <0.0001 p= 0.0003 p= 0.347 Glucose R < PR R < PA PR > PA fresh food p <0.0001 p= 0.588 p <0.0001

Table 2.1B: The difference in fresh diets’ nutritional values. Comparison of triglyceride, protein, and glucose concentrations in fresh regular diet (R), peach diet (PR), and autoclaved peach diet (PA).

Triglyceride R > PR R > PA PR > PA incubated food p =0.0344 p <0.0001 p= 0.0003 Protein R > PR R > PA PR < PA incubated food p <0.0001 p= 0.0003 p= 0.347 Glucose R < PR R < PA PR > PA incubated food p <0.0001 p= 0.588 p <0.0001

Table 2.1C: The difference in fresh diets’ nutritional values. Comparison of triglyceride, protein, and glucose concentrations in diets incubated without larvae, regular diet (R), peach diet

(PR), and autoclaved peach diet (PA).

(PR)

(PA)

(R)

Figure 2.1: The presence of dietary and/or maternally inherited bacteria substantially impacts metabolic and life history phenotypes in flies. A) mean survival until the late 3rd instar stage is increased on lab diet and with the presence of bacteria on peach diets B) mean development time is decreased on a lab diet and with the presence of dietary and/or maternal bacteria on peach diets; C) mean larval weight is higher on a lab diet and in the presence of bacteria on peach diets; D) Triglyceride concentrations by weight are reduced on a lab diet and in the presence of environmental bacteria on peach diets; E) Protein by weight is lower on a peach diet, in the presence of bacteria; F) Glucose by weight is lowest on a peach diet in the presence of bacteria. Error bars indicated one standard error of the mean. S stands for sterilized larvae and

NS stands for non-sterilized larvae.

1.2 We observed that the presence of a maternally transmitted bacteria significantly impacted larvae phenotypes, and that impact varied across dietary treatments. Survival. The presence of the parental bacteria significantly enhanced overall survival of larvae on the autoclaved peach diet but did not produce a significant effect on larval survival on the lab diet or the non-autoclaved peach diet (Fig. 2.1A, Table 2.2, S2.1.2). This suggested that the availability of bacterial taxa was necessary for a successful transition of the larvae through the instar stages under natural nutritional conditions, and that these bacterial species could be acquired from the food substrate if available and/or inherited maternally. Development rate. Presence of parental bacteria on the peach diet reduced the number of days necessary for larvae to reach the third instar stage on the autoclaved peach diet but not on the regular lab diet (Fig. 2.1B, Table 2.2,

S2.1.2), suggesting that under natural nutritional conditions, maternal microbes might influence the larval developmental rate independent of bacteria acquired from the food substrate. Weight.

Maternally inherited bacteria produced a significant positive effect on larval weight on all of the tested diets (Fig. 2.1C, Table 2.2, S2.1.2). This indicated the universality of their influence on larval growth across food substrates. Triglyceride. Parental bacteria did not influence the triglyceride concentrations significantly on any diet (Fig. 2.1D, Table 2.2, S2.1.2). Protein.

Evaluating the role of parental bacteria, we observed that sterilized larvae had higher protein by weight concentrations but only on the PA diet (Fig. 2.1E, Table 2.2, S2.1.2). This suggested that the core bacteria involved in a natural metabolic phenotype formation might be inherited or acquired from the environment. Glucose. The parental bacteria reduced the glucose by weight concentrations only on PA food (Fig. 2.1F, Table 2.2, S2.1.2), indicating that both parental and environmental microbial taxa might be sufficient to reduce glucose concentrations in larvae.

Genetic line Diet Survival Development Weight Triglyceride Protein Glucose All PA S< NS *** S> NS *** S< NS *** S> NS S> NS ** S> NS ** All PR S< NS S> NS *** S< NS *** S> Ns S< NS S> NS All R S< NS S> NS S< NS S> NS S> NS S> NS

Table 2.2: The influence of parental microbiota on larvae phenotypes. Comparison of larvae life history traits and metabolic phenotypes between larvae, from all 10 genetic lines, raised on regular lab diet (R), peach diet (PR), and autoclaved peach diet (PA). NS stands for non- sterilized larvae, S stands for sterilized larvae. Asterisks indicate the significance of comparisons p< 0.001 ***, p< 0.01 **, and p< 0.05 *

1.3 Evaluating the contribution of tested independent variables on larvae phenotypes, we observed a genetic variation in most of the tested life history traits and phenotypes that interacted with the dietary conditions and the availability of maternally transmitted microbiota. Survival.

All of the independent variables included in the model produced a significant effect on larval survival until the late third instar stage. Of of the tested variables, diet was the strongest predictor of survival, followed by the interactive effect of the diet by treatment and genetic line (Table 2.3,

S2.1.3). Development. The development rate of the larvae was significantly influenced by diet, genotype, and treatment (Table 2.3, S2.1.3). Among the specific interaction of these variables, only D*T and G*T produced a significant effect on development (Table 2.3, S2.1.3). Diet was the key factor that influenced the time necessary for the larvae to reach the late third instar stage and explained almost half of all variance followed by the genetic line (Table 2.3, S2.1.3). The combination of the rest of the variables was responsible only for 8.42 % of variation in developmental time (Table 2.3, S2.1.3).

Weight. All of the independent variables, with the exception of D*T, produced a significant effect on dry larval weight with diet being the best predictor, followed by genotype, and D*G interaction (Table 2.3, S2.1.3). Triglyceride. Once again, diet explained the largest portion of variance across all independent variables. The interactive effect of D*G was a better predictor of triglyceride concentrations than the genotype (Table 2.3, S2.1.3). Protein. In contrast with other measured phenotypes, the variance explained by the model was predominantly evenly distributed across the independent variables, with genotype having the highest predicting power followed by D*G interactive effect (Table 2.3, S2.1.3). Glucose. Diet was the strongest predictor of larvae glucose concentrations (Table 2.3, S2.1.3). Other variables that produced a significant effect on glucose concentrations were treatment, D*G, and D*T

(Table 2.3, S2.1.3).

Independent variable Survival Development Weight Triglyceride Protein Glucose Diet VE=28.4% *** VE= 47.5% *** VE= 31.4% *** VE= 41.4% *** VE= 4.24% *** VE= 17.8% *** Genetic line VE=5.13% *** VE=4.98% *** VE=7.44% *** VE=2.36% ** VE=5.71% *** VE=3.15% Treatment VE=4.69% *** VE=2.41% *** VE=0.74% *** VE= 0.36% VE=0.40% * VE= 2.61% *** Diet*Genetic line VE=3.13% *** VE=0.96% VE=2.88% *** VE=5.03% *** VE=5.09% *** VE=5.52% Diet*Treatment VE=8.02% *** VE=1.41% *** VE=0.10% VE=0.07% VE=0.45% VE= 3.55% *** Genetic line*Treatment VE=0.98% * VE=2.82% *** VE=1.06% *** VE= 1.81% * VE=3.53% *** VE=0.72% Diet*Treatment*Genetic line VE= 2.37% *** VE=0.82% VE=2.75% *** VE=4.81% *** VE= 4.92% *** VE= 2.37%

Table 2.3: The contribution of diet, genotype, treatment, and their interactive effects on formation of larvae life history traits and metabolic phenotypes. VE stands for variance explained, by each independent variable. Asterisks indicate the significance of comparisons p<

0.001 ***, p< 0.01 **, and p< 0.05 *

2) The symbiotic bacteria community composition of the larvae raised on the natural diet was different from the lab food raised larvae and was influenced by maternally inherited bacteria and the host’s genotype

2.1 The gut bacterial community composition and diversity varied substantially across dietary and treatment conditions. Alpha diversity. We characterized a total of 6,763 unique

ZOTUs across the whole dataset with the number of ZOTUs per sample ranging from 55 to

886. The total number of reads per sample ranged from 4,685 to 908,308. Of the ZOTUs that could be assigned a taxonomic classification, we identified 134 classes, 218 families, and 394 genera represented across the samples. The response of alpha diversity to changing diets varied with the larval sterilization treatment. For NS larvae, we found that the Shannon index of larvae raised on PR and R diets was significantly higher than those raised on the PA diet (Table 2.4).

All other comparisons were not significant. However, if the embryos were subjected to sterilization, the microbial species richness of larvae raised on the PA diet was significantly higher than larvae raised on the PR diet (Table 2.4). In addition, there was no significant difference in bacterial species richness between larvae raised on regular or peach regular diets

(Table 2.4). We observed the exact same pattern for the PD whole tree index. Larvae raised on any diet were not significantly different in Shannon’s index (Table 2.4).

Treatment Richness Shannon Simpson PD Whole Tree NS R = PR R = PR R = PR R = PR S R = PR R = PR R = PR R = PR NS R = PA R > PA * R > PA * R = PA S R = PA R = PA R = PA R = PA NS PR = PA PR > PA *PR > PA * PR = PA S PR < PA * PR = PA PR = PA PR < PA *

Table 2.4: Influence of a diet on alpha diversity measurements of larval bacterial community

Beta diversity. The hierarchical clustering of the Bray-Curtis distances indicated that the most distant bacterial communities were formed between larvae raised on the R and PR diet (Fig.

2.2A). This pattern held true for both sterilized and non-sterilized larvae (Fig. 2.2A). Clustering

Weighted Unifrac distances suggested that PR and R diets may produce symbiotic bacterial communities that were phylogenetically distant from each other, especially if the parental microbiota had been removed (Fig. 2.2B).

(PR) (PA)

(R)

Figure 2.2: Larvae raised on peach and regular lab diets form distinct bacterial communities. Constellation plot based on hierarchal clustering of larvae bacterial community beta diversity distances A) Bray-Curtis distances indicate that symbiotic bacteria of the larvae raised on a peach diet is different from larvae raised on a regular diet. Symbiotic bacterial communities of PA raised larvae is more similar with R raised larvae than with PR raised larvae,

36 especially if larvae were not sterilized B) Weighted Unifrac distances indicate that based on phylogenetic beta diversity distances PR and R diets form distinct communities. Samples from sterilized larvae are marked with a blue color and samples from non-sterilized larvae are marked with the red color. Samples raised on regular lab diet are marked with circled, on a peach diet with crosses and on peach autoclaved diet with triangles.

Taxa composition. Applying discriminant analysis on the ten most abundant bacteria at each taxonomic level revealed which organisms were largely responsible for the differentiation of the bacterial composition on the canonical plot, based on diet. Thus, PR food is largely defined by the abundance of Cyanobacteria at the phylum level (Fig. 2.3A, B),

Epsilonproteobacteria at the class level (Fig. 2.3C, D), Streptophyta at the order level (Fig. 2.3E,

F), Leuconostocaceae sequences at the family level (Fig. 2.4A, B), and Leuconostoc at the genera level (Fig. 2.4C, D). In turn, the lab diet was defined by Firmicutes, Bacilli,

Lactobacillales, Lactobacillaceae, and Lactobacillus respectively (Fig. 2.3-4). Interestingly, when we considered only the 10 most abundant organisms at each taxonomic level, we did not see a full separation between R and PA diets unless the larvae were sterilized (Fig. 2.3-4). If parental microbiota were removed, the differentiation of the PA diet was led by Actinobacteria at the phylum level, Actinobacteria and Alphaproteobacteria at the class level, Clostridiales and

Rickettsiales at the order level, Lachnospiraceae and Rickettsiaceae at the family level, and

Bacteroides and Wolbachia at the genus level (Fig. 2.3-4). Including all identified bacterial groups in the discriminant analysis revealed that diet was a good predictor of bacterial taxa composition at phylum, class, order, family, genus, and even individual ZOTU levels (Fig. 2.5-

6).

Figure 2.3: Bacterial communities of the larvae raised on all of the diets could be differentiated by the abundance of dominant bacteria phyla, classes, and orders.

Discriminant analysis of symbiotic bacterial communities based on taxa relative abundances. A)

10 most abundant microbial phyla found in non-sterilized larvae B) 10 most abundant bacterial phyla found in sterilized larvae C) 10 most abundant bacterial classes found in non-sterilized larvae D) 10 most abundant bacterial classes found in sterilized larvae E) 10 most abundant bacterial orders found in non-sterilized larvae F) 10 most abundant bacterial orders found in sterilized larvae. Diet serves as a good predictor for differentiation of assosiated bacterial communities. The length of the vector is correlated with the strength of the impact that it produced for the samples to be separated, in the vector direction, on the canonical plot.

Figure 2.4: Bacterial communities of the larvae raised on all of the diets could be differentiated by the abundance of dominant bacteria families and genera. Discriminant

39 analysis of symbiotic bacterial community based on taxa relative abundances in larvae. A) 10 most abundant bacterial families found in non-sterilized larvae B) 10 most abundant bacterial families found in sterilized larvae C) 10 most abundant bacterial genera found in non-sterilized larvae D) 10 most abundant bacterial genera found in sterilized larvae.

Figure 2.5: Bacterial communities of the larvae raised on all of the diets could be differentiated by the abundance of all identified bacteria phyla, classes, and orders.

Discriminant analysis of symbiotic bacterial community based on taxa relative abundances in larvae. A) All identified bacterial phyla found in non-sterilized larvae B) All identified bacterial phyla found in sterilized larvae C) All identified bacterial classes found in non-sterilized larvae

D) All identified bacterial classes found in sterilized larvae E) All identified bacterial orders found in non-sterilized larvae F) All identified bacterial orders found in sterilized larvae. Diet serves as a good predictor for bacterial communities differentiation.

Figure 2.6: Bacterial communities of the larvae raised on all of the diets could be differentiated by the abundance of all identified bacteria families, genera, and ZOTUs.

Discriminant analysis of symbiotic bacterial community based on taxa relative abundances in larvae. A) All identified bacterial families found in non-sterilized larvae B) All identified bacterial families found in sterilized larvae C) All identified bacterial genera found in non- sterilized larvae D) All identified bacterial genera found in sterilized larvae E) All identified bacterial ZOTUs found in non-sterilized larvae F) All identified bacterial ZOTUs found in sterilized larvae. Diet serves as a good predictor for bacterial communities differentiation.

Overall, for non-sterilized larvae the abundance of eight phyla, 12 classes, 20 orders, 27 families, 40 genera, and 141 ZOTUs were significantly different between PR and R food (S2.2.1-

6). Comparing PR and PA food, we found that the abundance of four phyla, six classes, nine orders, 16 families, 20 genera, and 76 ZOTUs were significantly different (S2.2.1-6). Lastly, we observed the significant difference for the abundance of four phyla, three classes, five orders and families, three genera, and 27 ZOTUs between R and PA food (S2.2.1-6). This indicated the minimal difference between microbial communities of these diets to be consistent with the discriminant analysis.

In larvae lacking parental microbiota, we observed that the abundance of six phyla, 10 classes, 18 orders, 34 families, 37 genera, and 114 ZOTUs were significantly different between lab and peach diets (S2.2.1-6). Comparing PR and PA diets, we found a significant difference in abundance of seven phyla, 14 classes, 32 orders, 48 families, 67 genera, and 200 ZOTUs (S2.2.1-

6). R and PA diets were significantly different in the abundance of five phyla, six classes, nine orders, 13 families, 18 genera, and 87 ZOTUs (S2.2.1-6).

2.2 The maternally transmitted microbiota influenced the composition of the larvae’s symbiotic bacterial communities. Alpha diversity. S larvae had higher values for species richness, Shannon, and PD whole tree indexes on the PA diet (S2.2). NS larvae had a significantly higher Simpson index value on the PR diet (S2.2). All other comparisons were not significantly different.

Diet Richness Shannon Simpson PD Whole Tree PA S > NS * S > NS * S = NS S > NS * PR S = NS S = NS S < NS * S = NS R S = NS S = NS S = NS S = NS

Table 2.5: Influence of treatment on alpha diversity measurements of larval bacterial community

Beta diversity. When comparing the difference between beta diversity metrics in NS and

S treatments for each diet, we observed a distinctive clustering, based on the treatment of samples that were raised on PA food for Bray-Curtis Distance (Fig. 2.7A) and Weighted Unifrac distance (Fig. 2.7D). For the samples that were raised on R food, we observed the clustering for

Weighted Unifrac distance only (Fig 2.7F).

Figure 2.7: Inheriting parental bacteria influences the formation of larvae symbiotic bacterial community with the effect being unequal among the diets. A-C) Bray-Curtis distances between NS and S larvae raised on PA, PR, and R diets, respectively. NS and S communities are clearly separated on A) PA and C) R diets, and B) PR food is less distinct. D-F)

Weighted Unifrac distances. Larvae raised on A) PA and C) R diets form more distinct communities than larvae raised on B) PR diet.

Taxa composition. The discriminant analysis indicated that the status of inheritance of the parental bacteria could serve as a good predictor for differentiation of the bacterial community as indicated with the canonical plot on all taxonomic levels (phylum, order, class, family, genus, and ZOTU) (Fig. 2.8-9). Among the 10 most abundant phyla that defined the differentiation of the NS community were Firmicutes and Bacterioidetes on the PA diet (Fig.

2.10A), Firmicutes on the PR diet (Fig. 2.10B), and Proteobacteria, Tenericutes, and

Flusobacteria on the R diet (Fig. 2.10C). Phyla that were influential for differentiation of the S community were Actinobacteria, Fusobacteria and Cyanobacteria on the PA diet (Fig. 2.10A),

Actinobacteria, Tenericutes, and Flusobacteria on the PR food (Fig. 2.10B), and Planctomycets and Bacterioides on the R diet (Fig. 2.10C). Considering bacterial classes, the NS treatment was strongly defined by Bacilli on the PA (Fig. 2.10D), and the PR diets (Fig. 2.10E), and

Alphaproteobacteria and Bacteroida on the R food diet (Fig. 2.10F). The sterilization treatment was mostly separated due to Actinobacteria on the PA diet, Alpha and Beta proteobacteria on the

PR diet, and Gammaproteobacteria on the R diet (Fig. 2.10D-E). On R food, microbial communities from the sterilized and non-sterilized larvae were not fully separated on the canonical plot (Fig. 2.10F). At the order level, the NS community was defined by Lactobcillales on the PA and the PR diets, and Actinomycetales and Rhodospirillales on the R food (Fig.

2.10G-I). Sterilized larvae were associated with abundances of Streptophyta on the PA food, as well as Burkholderiales and Rhodospirillales on the PR diet (Fig. 2.10G-I). On the R diet, four out of ten tested orders were strongly associated with the S treatment (Fig. 2.10I). At the family level, the NS larvae were correlated with Lactobacillaceae on the PA and the PR diets and

Acetobacteraceae on the R food diet (Fig. 2.11A-C). Sterilized larvae were defined by the abundance of Nocardiaceae on the PA food and Leuconostocaceae and Caulobactereceae on the

R diet (Fig. 2.11A-C). At the genera level, NS was primarily separated by Lactobacillus on the

PA diet and Acetobacter and Agrobacterium on the R food (Fig. 2.11D-F). On the PR food diet,

95% confidence ellipses almost overlapped, indicating that sterilization status might not be the decisive predictor for abundance of the 10 most common genera (Fig. 2.11B). The S treatment was primarily defined by the abundance of Leuconostoc and Gluconobacter on the PA diet and

Lactobacillus and Leuconostoc on the R diet (Fig. 2.11A, C).

Multiple microbial groups were significantly different in their distribution between NS and S treatments across the diets on all taxonomic levels. On regular food, there existed an abundance of four phyla (S2.3.1), four classes (S2.3.2), five orders (S2.3.3), six families

(S2.3.4), nine genera (S2.3.5), and 41 ZOTUs (S2.3.6). On the PR diet we saw a significant difference in the abundance of one phylum (S2.3.1), two classes (S2.3.2), six orders (S2.3.3), nine families (S2.3.4), 11 genera (S2.3.5), and 61 ZOTUs (S2.3.6). The highest number of significantly different taxa was observed on the PA diet with seven phyla (S2.3.1), 12 classes

(S2.3.2), 21 orders (S2.3.3), 34 families (S2.3.4), 43 genera (S2.3.5), and 148 ZOTUs (S2.3.6).

Figure 2.8: Microbial communities of the larvae raised with or without parental bacteria could be differentiated by the abundance of all bacterial phyla, classes, and orders.

Discriminant analysis of symbiotic bacterial community based on taxa relative abundances in larvae from non-sterilized (red) and sterilized (blue) treatments. The analysis includes all identified phyla A) on a PA diet B) on a PR diet, and C) on a R diet, all identified classes D) on a

PA diet E) on a PR diet, and F) on a R diet, all identified orders G) on a PA diet H) on a PR diet,

47 and I) on a R diet. On all diets, treatment serves as a good predictor for a differentiation of the symbiotic bacterial communities.

Figure 2.9: Microbial communities of the larvae raised with or without parental microbiota could be differentiated by the abundance of all bacteria families, genera, and ZOTUs.

Discriminant analysis of symbiotic bacterial community based on taxa relative abundances in larvae from non-sterilized (red) and sterilized (blue) treatments. The analysis includes all identified families A) on a PA diet B) on a PR diet, and C) on a R diet, all identified genera D) on a PA diet E) on a PR diet, and F) on a R diet, all identified ZOTUs G) on a PA diet H) on a

PR diet, and I) on a R diet. On all diets, treatment serves as a good predictor for a differentiation of the symbiotic bacterial communities.

Figure 2.10: Microbial communities of the larvae raised with or without parental microbiota could be differentiated by the abundance of dominant bacteria phyla, classes, and orders. Discriminant analysis of symbiotic bacterial community based on taxa relative abundances in larvae from non-sterilized (red) and sterilized (blue) treatments. The analysis includes 10 dominant phyla A) on a PA diet B) on a PR diet, and C) on a R diet, 10 dominant classes D) on a PA diet E) on a PR diet, and F) on a R diet, 10 dominant orders G) on a PA diet

H) on a PR diet, and I) on a R diet. Although on all diets, treatment serves as a good predictor for a differentiation of the symbiotic bacterial communities, it is especially influential on PA food.

Figure 2.11: Microbial communities of the larvae raised with or without parental microbiota could be differentiated by the abundance of dominant bacteria families and

50 genera. Discriminant analysis of symbiotic bacterial community based on taxa relative abundances in larvae from non-sterilized (red) and sterilized (blue) treatments. The analysis includes 10 dominant families A) on a PA diet B) on a PR diet, and C) on a R diet and 10 dominant genera D) on a PA diet E) on a PR diet, and F) on a R diet. Although on all diets, treatment serves as a good predictor for a differentiation of the symbiotic bacterial communities, it is especially influential on PA food.

2.3 The composition of the microbial community exhibited variation with host genotype, which further exhibited a significant interactive effect with diet and treatment. We also tested the influence of genotype and other variables’ interactive effect on the abundance of bacteria. At the phyla level, 14 were significantly influenced by genotype, three by D*G interaction, five by

G*T, five by D*T, and six by D*G*T (S2.4.1). Abundances of 30 classes were significantly influenced by genotype, eight by D*G, G*T, and D*T, and 10 by D*G*T interaction (S2.4.2).

Among the orders, an abundance of 46 was significantly influenced by genotype, 13 by D*G, 15 by G*T, 16 by D*T, and 15 by D*G*T (S2.4.3). The abundance of 72 families was significantly influenced by genotype, 20 by D*G, 15 by G*T, 18 by D*T, and 23 by D*G*T (S2.4.4). Lastly, genotype significantly influenced 94 genera, D*G influenced 44, G*T influenced 46, D*T influenced 30, and D*G*T influenced 45 genera (S2.4.5).

3.1 We identified microbial taxa that exhibited correlations with host phenotype across diets and treatments, with many that had a diet, treatment or genotype specific relationship.

Across the NS larvae, we found four significant interactions on the R food, 14 on PR diet, and 14 on PA diet at the phylum level (S2.5.1). For the S larvae at the same taxonomic level, we found eight significant interactions on the R diet, five on the PR and two on PA food (S2.5.1). At the class taxonomic level, nine significant interactions were found on R food, 35 on PR, and 24 on

PA diet (S2.5.2). Considering S larvae, there were 23 significant correlations on the R diet, 11 on

PR, and four on the PA diet (S2.5.2). For NS larvae at the order level, we found 23 significant correlations on the R diet, 57 on PR, and 46 on PA diet (S2.5.3). For S larvae we found 29 significant correlations on R food, 22 on PR, and nine on PA diet (S2.5.3). At the family level, we found 34 significant correlations on R food, 88 on PR, and 67 on PA diet for NS larvae

(S2.5.4). For S larvae, we observed 51 significant interactions on R food, 29 on PR, and 14 on

PA diets (S2.5.4). Across the genera, we found 40 significant interactions on R, 105 on PR, and

64 on PA diets, for NS larvae (Fig. 2.12-13, S2.5.5). Considering S larvae, we found 76 significant interactions between tested taxa and phenotypes on R, 46 on PR, and 33 on PA diets

(Fig. 2.12-13, S2.5.5). At the level of individual ZOTUs, for NS larvae, we found 226 significant interactions on R, 283 on PR, and 225 on PA diets (S2.5.6). For S larvae the number of significant interactions between ZOTUs abundances and larvae phenotypes were as follow 313 on R, 164 on PR, and 230 on PA diets (S2.5.6).

We evaluated the interactive effect of the abundance of microbial taxa and diet, genotype, and treatment on forming the tested phenotypes. D*A produced a significant effect in 11 cases at the phylum level (S2.6.01), in 26 at the class level (S2.6.02), in 52 at the order level (S2.6.03), in

72 at the family level (S2.6.04), and in 95 cases at the genus level (S2.6.05). We found a significant G*A interaction in eight cases at the phylum level (S2.6.06), in 10 at the class level

(S2.6.07), in 39 at the order level (S2.6.08), in 57 at the family level (S2.6.09), and in 105 cases at the genus level (S2.6.10). T*A produced a significant effect in 13 cases at the phylum level

(S2.6.11), in 27 at class level (S2.6.12), in 36 at the order level (S2.6.13), in 60 at the family level (S2.6.14), and in 87 cases at the genus level (S2.6.15).

(PA)

(PR)

(R)

(PA)

(PR)

(R)

(PA)

(PR)

(R)

Figure 2.12: The influence of symbiotic microbial taxa on larvae metabolic phenotypes vary with the diets. Spearman’s rank correlation coefficients between the abundance of 10 dominant bacterial genera and larvae phenotypes, on each diet. The color of the bars corresponds to the level of significance for each correlation. A) Weight of NS larvae B) Weight of S larvae

C) Triglyceride levels of NS larvae D) Triglyceride levels of S larvae E) Glucose levels of NS larvae F) Glucose levels of S larvae. The effect that symbiotic bacteria produce on all metabolic phenotypes may change with the diet.

(PA)

(PR)

(R)

(PA)

(PR)

(R)

(PA)

(PR)

(R)

Figure 2.13: The effect that bacterial taxa produce on larvae life history traits and metabolic phenotypes vary with the larvae diet. Spearman’s rank correlation coefficients between the abundance of 10 dominant symbiotic bacterial genera and larvae phenotypes on each diet. The color of the bars corresponds to the level of significance for each correlation. A)

Total proportion of NS larvae reaching the 3rd instar stage; B) Total proportion of S larvae reaching the 3rd instar stage; C) Median number of days to reach pre-pupation stage for NS larvae; D) Median number of days to reach pre-pupation stage for S larvae; E) Protein concentrations of NS larvae; F) Protein concentrations of S larvae. Symbiotic bacteria may influence life history traits and protein concentrations of the larvae differently depending on a diet.

DISCUSSION

1.1: Overall, it appeared that frozen peach food was capable of providing nutritional conditions similar to the natural ones and can preserve key microbial taxa necessary for survival and development of Drosophila larvae

The reduction in survival, increase in development time and increase in triglyceride concentrations, as well as decreased weight and protein concentrations of the larvae raised on the natural food compared with the larvae raised on the R lab food resembles the phenotype generated by a reduced protein diet. These findings correlate with our evaluation of the protein concentrations in different diets (Klepsatel, Procházka, and Gáliková 2018, Bing et al. 2018,

Skorupa et al. 2008, Sang 1956). In addition, the adaptation of Drosophila to the lab environment was connected to increased weight and reduced stress tolerance (Sgro and Partridge

2000, Hoffmann et al. 2001, Russell, Kurtz, and Reviewers-July 2012). Therefore, nutritional and pathogenic stresses associated with the natural food conditions could further contribute to the decrease in survival and development rate of larvae raised on the PR food compared to the standard diet (Staubach et al. 2013, Bing et al. 2018, Pais et al. 2018, Sang 1956).

The pattern regarding glucose concentration was more interesting. Freshly unfrozen peach food had a higher glucose concentration than regular lab food, but larvae raised on the PR diet had the lowest concentration compared to larvae raised on any other diet. This pattern was likely caused by the activity of naturally acquired microbes since it was shown that the presence of several microbial taxa that naturally associate with Drosophila, such as Acetobacter, is correlated with decreased sugars in fly food and Drosophila itself (Huang and Douglas 2015,

Dobson et al. 2015). In addition, incubation of the PR food, even without the larvae, led to a drastic reduction of glucose concentrations compared to the R and PA diets.

Furthermore, the difference between all phenotypes (with the exception of glucose) increased if the peach food diet was autoclaved and even more (with the exception of triglyceride) if the parental microbiota were not transferred to the autoclaved diet. This suggested that symbiotic bacteria drove the phenotypic change between PR and PA raised larvae. These findings are consistent with previous studies which showed that the presence of naturally associated bacteria was advantageous for Drosophila melanogaster and Drosophila suzukii on fresh fruit diets, which are also poor in protein content (Pais et al. 2018, Bing et al. 2018). In fact, larvae raised on a PA diet closely resembled the phenotype of axenic larvae and axenic larvae raised under low protein nutritional conditions. Examples of this similar phenotype include lower survival and body size/weight (Shin et al. 2011, Wong et al. 2015, Dobson et al.

2015), longer development time (Shin et al. 2011, Newell and Douglas 2014), elevated glucose

(Huang and Douglas 2015) and triglyceride concentrations (Newell and Douglas 2014, Dobson et al. 2015).

1.2 Maternally deposited microbes produced positive effects on larvae that were raised on the peach diets

Interestingly, the presence of parental microbiota did not produce a significant effect on any of the tested phenotypes, when larvae were raised on the lab diet. Contrarily, on the peach diet, the presence of parental microbota increased the weight and development rate even if the original peach microbiota were still present. These findings are consistent with the reports of beneficial effects, of the maternally deposited microbiota, for larvae on a fruit diet. These results also indicate the importance of considering an organism’s natural environmental conditions when addressing the questions about symbiotic relationships and evolutionary patterns (Pais et al. 2018, Bing et al. 2018).

1.3 Genotype was one of the key factors that influenced larvae phenotypes

It is important to note that although the described patterns were observed for the total experimental population of larvae, the genetic component still played a significant role in generating all but the glucose phenotype. In addition, consistent with previous research, we observed that D*G interaction played a significant role in forming metabolic phenotype as well as contributed to the survival of the organism (Reed et al. 2010, Reed et al. 2014). Furthermore, most of the tested phenotypes were significantly correlated with G*T and even D*G*T, indicating the importance of considering multiple factors to understand the development of complex traits.

2.1 Bacteria of the larvae raised on PR food exhibit a distinct community structure

Multiple studies were performed to evaluate the gut microbiota composition of lab and wild populations of Drosophila (Chandler et al. 2011, Adair et al. 2018, Douglas 2018, Pais et al.

2018, Wong, Chaston, and Douglas 2013). Although most of them consistently report the prevalence of different members of Alphaproteobacteria, Bacilli, or Gammaproteobacteria in lab and wild populations, the relative abundance of the taxa, especially at lower taxonomic levels, often varies between studies (Douglas 2018). In our work, larvae raised on the PR food diet formed a distinct community clearly separated from the larvae raised on the R food diet, as displayed on the canonical plot. We observed a higher prevalence of Gluconobacter and

Leuconostoc and lower abundance of Lactobacillus in larvae raised on the PR diet compared with the R food diet (Pais et al. 2018, Staubach et al. 2013, Corby-Harris et al. 2007, Chandler et al. 2011), which is consistent with previous findings performed on natural populations of

Drosophila.

However, it is difficult to judge how well the microbial community of our experimental larvae represent the microbial community of wild population, as a variety of factors could influence gut microbiota composition in flies which certainly complicates comparisons between studies (Douglas 2018, Staubach et al. 2013, Jehrke et al. 2018). As such, it was shown that the gut microbial composition of lab reared flies may vary with diet (and even among the standard diets with the major carbohydrate source), genetic line, development stage temperature, and etc.

(Jehrke et al. 2018, Douglas 2018, Wong, Ng, and Douglas 2011, Moghadam et al. 2018). The wild populations of gut microbiota in Drosophila were shown to vary with collection location and diet (Adair et al. 2018, Wong, Chaston, and Douglas 2013, Staubach et al. 2013, Martinez-

Porchas et al. 2017). In other insects and wild populations of vertebrates, gut microbiota was shown to change even with seasonality (Behar, Jurkevitch, and Yuval 2008, Ferguson et al.

2018, Tong and Zhang 2019, Maurice et al. 2015).

In addition, the relationship between Drosophila gut microbiota during the developmental and adult stages is a subject of controversy between a few studies that compared those relationships (Jehrke et al. 2018, Wong, Ng, and Douglas 2011, Vacchini et al. 2017).

Furthermore, to the best of our knowledge, the gut bacteria of the larvae from the natural populations was not assessed at all. This is likely due to the complexity of identifying

Drosophila species during the larval stage. Therefore, we hope to provide the methodology for the possibility of exploring the effects of a natural diet, and the microbial community associated with it, in a controlled lab environment. This setting provides the opportunity to work not only with adult flies but also with larvae.

2.2-2.3 Community structure of symbiotic bacteria were correlated with diet, treatment, host genotype and their specific interactive effects

The development of symbiotic microbiota populations was shown to be correlated with the available nutrients present in the diet, the host’s genotype and parental microbiota left on the chorion of the egg (Douglas 2018, Jehrke et al. 2018, Wong et al. 2015). Complementary to the results reported by Jehkre (2018), we also observed that the genotype of the host may influence the abundance of bacterial taxa more than the diet. Wong (2015) reported that the bacterial population deposited on the Drosophila embryo may shift the symbiotic microbiota population of the offspring, even in the presence of bacteria that previously colonized the food substrate. We observed similar results in most cases.

However, among the 10 most abundant genera on the PR diet, the full separation of the S and NS larvae microbial community compositions was not present on the canonical plot indicating the possibility of a difference in the response of the lab and the natural microbial population to the presence of Drosophila parental bacteria. This differentiation was not likely caused by the nutrition composition of the food since the PA separation, represented on the canonical plot, between S and NS treatments was obvious in all cases. In addition, for beta diversity distances, the abundance of individual microbial taxa, as well as the correlations between the abundances of microbial taxa within the microbial community, represented patterns found in PA food that resembled the ones in the R food raised larvae, if parental microbiota were not removed (Fig. 2.2-6, 2.14). Overall, consistently with previous studies, our findings indicated the dependency of relative bacterial abundances on all of the tested variables and demonstrated the interactive effect between these variables (Wong et al. 2015, Jehrke et al. 2018, Douglas

2018).

Figure 2.14: The interaction between microbial genera varies between diets and treatments. The network plot of spearman rank correlation between abundances of 10 dominant bacterial genera A) from NS larvae on a PA diet , B) from NS larvae on a PR diet , C) from NS larvae on a R diet D) from S larvae on a PA diet, E) from S larvae on a PR diet, F) from S larvae on a R diet. Microbial correlations of non-sterilized larvae on PA diet are more similar with R diet than with PR raised larvae. Red lines indicate negative correlation and blue lines indicate positive correlation. The density of the color is positively correlated with the strength of the correlation. The correlations less than 0.5 were filtered out.

3.1 The influence of individual microbial taxa as well as the influence of the whole microbial community on the host may vary with the diet and other environmental and genetic conditions

Genotype and gut microbiota composition are among the major factors that control the development of obesity traits (Parks et al. 2013). Changes in some key microbiota populations are associated with the rapid expansion in the prevalence of metabolic syndrome (Zhang et al.

2010). Alterations in the gut microbiota community can modulate insulin secretion and sensitivity, thus contributing to diabetes susceptibility (Kreznar et al. 2017). Moreover, previous research indicates that genetic variation considerably influences the gut microbiota composition

(Zhang et al. 2010, Kreznar et al. 2017, Jehrke et al. 2018). However, most of the studies mentioned above have used less than ten genotypes to study the correlation between gut microbiota and the pathogenesis of obesity in mice. The challenges of using a mouse model involve relatively high expenses for husbandry and logistics (Rosenthal and Brown 2007, Berger et al. 2005, Martin et al. 2016).

Drosophila melanogaster is an exceptional model to study the effect of genotype on the phenotype formation, due to the variety of established tools such as Drosophila Genetic

Reference Panel and The Drosophila Synthetic Population Resource. These resources offer a variety of diverse genotypes, with sequenced parental genomes, that allow for testing the microbiota effects across various genetic backgrounds and provides potential for studying genetic interaction between host and its symbionts, and even mapping the specific genetic loci responsible for the interactions (Mackay et al. 2012, King, Macdonald, and Long 2012, Chaston et al. 2016). The phenotypic response to a diet modification often varies with the genotype (Reed et al. 2010, Reed et al. 2014). In fact, diet-by-genotype (DxG) interaction may explain more variance than diet alone in the metabolic response of such traits, such as triglyceride and carbohydrate concentrations (Reed et al. 2010). In addition, recent findings showed that genotype-by-diet interactions significantly influences metabolomic profiles; hence, laying the

61 foundation for explaining the mechanism through which DxG influences metabolic traits (Reed et al. 2014, Williams et al. 2015). Our results are consistent with previous research, in that phenotypic response varied significantly between genetic lines (Reed et al. 2014, Williams et al.

2015). Genotype had a significant effect on survival, development rate, and triglyceride concentrations, and was the second-best predictor of weight and the best predictor of protein concentrations.

Obesity and type two diabetes are associated with elevated weight, high blood glucose concentrations, and excess accumulation of adipose tissue (Martyn, Kaneki, and Yasuhara 2008,

Akter et al. 2017). Consistent with recent studies linking Lactobacillus and Coprococcus to obesity in humans, our results show that these genera are positively associated with glucose concentrations (Million et al. 2012, Ignacio et al. 2016, Murugesan et al. 2015, Armougom et al.

2009). In addition, previous research has shown an overall decrease in the abundance of

Firmicutes in obese humans (Schwiertz et al. 2010). Similarly, we observed that the total abundance of Firmicutes is negatively associated with triglyceride concentrations. It should be noted that the correlation between metabolic phenotype and particular microbial taxa could vary between studies (Ley et al. 2006, Furet et al. 2010, Turnbaugh et al. 2006).

Consistently with previous Drosophila research, we observed that the abundance of microbial taxa was correlated with measured phenotypes. Acetobacteraceae was negatively correlated with larval glucose concentrations (Chaston et al. 2016, Douglas 2018, Chaston,

Newell, and Douglas 2014). Additionally, Acetobacter increased development time while

Lactobacillus and Firmicutes decreased it (Chaston, Newell, and Douglas 2014, Chaston et al.

2016, Newell and Douglas 2014, Storelli et al. 2011). Previous work showed that Acetobacter species reduced triglyceride concentrations while most Lactobacillus species had no effect

(Chaston et al. 2016, Newell and Douglas 2014). In contrast, our data shows that Acetobacter did not significantly affect triglyceride concentrations, and Lactobacillus showed a negative correlation. Consistent with Newell and Douglas (2014), we found that L. brevis and L. plantarum had no significant effect on protein concentrations, but in addition, our results indicated that Acetobacter abundance was negatively correlated with protein concentrations.

Consistent with Jehrke (2018), we observed that most of the correlations between the tested phenotype and abundance of bacteria were relatively weak. Weaker correlations observed with large sample sizes in microbiome research, while significant, failed to hold up to the use of stricter FDR values or other conservative p adjustment methods (Jehrke et al. 2018, Wu et al.

2011). Expanding the analysis of the bacterial species abundance for each phenotype beyond the most dominant species, while providing a more complete overview of the correlation between tested phenotypes and microbial abundance, also raises FDR values as a result of increasing sample sizes (Wu et al. 2011). Previous microbiome studies have dealt with high FDR values by accepting higher thresholds, so as to not miss possible correlations (Wu et al. 2011). Since the level of FDR that should be tolerated is poorly defined and often widely variable compared to accepted p-values, its value is often seen as arguable (Pawitan et al. 2005). Considering the large sample sizes used in our analysis, using a low FDR value may obscure important correlations between the tested phenotypes and their abundance of microbiota (Wu et al. 2011).

Some of the inconsistencies between our work and previous studies on the correlation of the abundance of microbial taxa and measured metabolic phenotype perhaps may be addressed to the interactive effect between the variables included in the experiments. Several studies showed that the contribution of the symbiotic microbiota to the host may be observed only in a diet dependent manner. For example, Shin et al. (2011) showed that axenic Drosophila larvae would

63 not be able to develop on a protein poor diet without activation of the insulin signaling pathway by its symbiotic microbes. Wong et al. (2014) found diet-dependent differences in microbiota produced effects, including reduction of vitamins requirements on a low-yeast diet and suppression of lipid and carbohydrate storage on a high-sugar diet. Bing et. al (2018) found that symbiotic microbiota of D. suzukii are critical for providing proteins for development of flies raised on fresh fruit, but that these microbial proteins are not essential for development of flies raised on a nutrient sufficient diet.

In our study, we also observed that the effect bacterial abundance produced at the level of individual taxa on larvae phenotype varied with diet. In a few cases, even the direction of correlation between the abundance of microbial taxa and tested phenotype was opposite on different diets. In addition, using PCA, we observed that correlational effects microbial abundance (as an example at the family level) produced on measured metabolic (Fig. 2.15A) and fitness phenotypes (development rate and survival) (Fig. 2.15B) varied between the diets. The correlation coefficients for the influence of all microbial taxa on measured metabolic phenotypes clustered together for the PR diet but not for other diets (Fig. 2.15A). The correlation coefficients between bacteria abundances and fitness phenotypes clustered for all but PA diets (Fig 2.15B).

Perhaps taking into account all of the described findings, in future studies that will aim to understand the mechanisms responsible for formation of metabolic phenotypes, we will see a consistent control for the gut bacterial composition in a similar fashion, as now we can see it for genetic lines and nutritional composition of the food.

Figure 2.15: The effect of the whole bacterial communities on larvae metabolic phenotypes and life history traits varies with the diet. A) Principal component analysis of correlation coefficients between bacterial families and metabolic phenotype of the sterilized larvae from 10 genetic lines. R diet is marked with circles, PR diet is marked with crosses, and PA diet is marked with triangles. Larvae glucose concentration is marked with red, protein is marked with green, triglyceride is marked with blue, and weight is marked with brown colors. B) Principal component analysis of correlation coefficients between bacterial families and life history of the sterilized larvae from 10 genetic lines. R diet is marked with circles, PR diet is marked with crosses, and PA diet is marked with triangles. Larvae development rate is marked with red and survival is marked with blue.

CHAPTER THREE

THE EFFECT OF HIGH FAT AND HIGH SUGAR DIETARY MODIFICATIONS ON DROSOPHILA MELANOGASTER METABOLIC PHENOTYPE AND SYMBIOTIC MICROBIOTA COMMUNITY

INTRODUCTION

The pandemic of obesity contributes to the increase in multiple health conditions including coronary heart disease, asthma, arthritis, hypertension, and type two diabetes mellitus among many others (Seganfredo et al. 2017, Wahba and Mak 2007, Thompson et al. 2007). However, the negative effects of obesity are not limited to health-related problems. Obesity is also accompanied by social stigmatization that makes affected individuals vulnerable to discrimination, further social injustices and poor mental health (Puhl and Heuer 2010). The problem of obesity is further complicated by the diversity of causal factors such as dietary habits, genotype, epigenetic regulation, psychological stress, sleep deprivation, and gut microbiota composition (Seganfredo et al. 2017, Han and Lean 2016). These issues make the development of an efficient treatment a complex problem requiring a multivariate research approach

(Thompson et al. 2007, Kaur 2014).

The complexity of obesity often means that even with the most efficient treatments, weight regain is common and often substantial (Seganfredo et al. 2017). Common treatments that are often used to combat obesity, such as restriction diets, high protein diets, bariatric surgery,

66 and others are often correlated with changes in gut microbiota composition (Zhang et al. 2009,

Furet et al. 2010, Everard et al. 2013, Seganfredo et al. 2017). Dietary administration of symbiotic microbial taxa such as Akkermansia muciniphila was sufficient to reverse some of the phenotypes associated with obesity (Everard et al. 2013, Patterson et al. 2016). Thus, future research on the effects of gut microbiota may assist the development of more efficient treatments, to combat obesity.

Symbiotic microbiota plays a major role in regulation of normal functioning of the metabolic and immune systems. As a result, disturbance of the normal microbial flora may lead to autoimmune and metabolic disorders (Read and Holmes 2017, Leitão-Gonçalves et al. 2017).

The variation in gut microbiota composition and diversity has been shown to be correlated with changes of metabolic phenotype and obesity (Turnbaugh et al. 2006, Flint, Duncan, and Louis

2017). In fact, even a transfer of microbiota from an obese organism to an axenic one can cause the development of the obese phenotype in the recipient (Ridaura et al. 2013, Tilg and Moschen

2016, Turnbaugh et al. 2006); which suggests that in certain cases obesity may be transferred as an infectious disease.

Drosophila melanogaster was shown to be an excellent model for studying symbiotic microbiota and its interactive effect with other conditions on the development of metabolic phenotypes (Dobson et al. 2015, Chaston et al. 2016, Jehrke et al. 2018, Wong, Dobson, and

Douglas 2014). In D. melanogaster, microbiota influence several life history traits and metabolic phenotypes such as survival until pupation, development time, weight, protein, triglyceride, and glucose concentrations (Newell and Douglas 2014, Dobson et al. 2015, Huang and Douglas

2015). Interestingly, the impact of harmful diets, such as high fat and high sugar, forms a phenotype similar to flies that have a removal of symbiotic microbiota. For example, axenic flies

67 often exhibit elevated glucose and triglyceride concentrations (Wong, Dobson, and Douglas

2014, Henry, Tarapacki, and Colinet 2020). In addition, certain microbial taxa from Acetobacter and Lactobacillus genera may reduce host appetite for essential amino acids and increase the appetite for sugar consumption; thereby potentially providing the host with the essential amino acids and competing for available sugars (Leitão-Gonçalves et al. 2017).

Several studies demonstrated the differences in the symbiotic bacterial community of wild flies and lab raised flies (Chandler et al. 2011, Vacchini et al. 2017, Tefit et al. 2017). In our previous work, we also observed that larvae raised on the natural peach diet and standard lab food exhibit a substantial difference in the microbiota composition. It was also demonstrated that in Drosophila and mouse models, the effect of symbiotic microbiota on host phenotype may vary with diet, genotype, and their interactive effects (Wong, Dobson, and Douglas 2014, Kreznar et al. 2017, Henry, Tarapacki, and Colinet 2020, Zhang et al. 2009, Org et al. 2015).

Our work aims to answer several specific questions:

1. Will larvae raised on high fat (HF) and high sugar (HS) diets exhibit a variation in a

phenotypic response? Will parental microbiota influence the response of the larvae raised

on modified diets equally among lab and peach-based diets? Will the genotype and

specific interactive effects between independent variables influence larvae phenotypes?

Will the response to diet modification be consistent across all the diets (peach-based or

lab-based)?

1.1. We hypothesized that a unique interaction between environmental bacteria and

nutritional conditions of the diets will result in different phenotypes of the larvae

raised on HF or HS diets, depending on the basis of the original diet (peach-based or

lab-based diets).

1.2. Based on the previous research and our own work, we hypothesized that presence of

parental microbiota will influence larval phenotypes more on autoclaved diets than on

any other diet types and especially on HS diets.

1.3.We hypothesized that genotype and diet-genotype and/or diet-bacterial interactive

effects will influence larvae phenotypes, but the strength of the effect will vary with

diet modifications.

1.4. We hypothesized that larvae raised on the peach-based diets will respond differently

to diet modification (N vs modified) than larvae raised on the lab-based diets, and that

on the natural diets, presence of environmental and parental microbiota will make

larvae less responsive to diet modifications.

2. Will bacterial communities of the larvae raised on the natural and lab HF and HS

modified diets be different, in diversity and composition? Will parental microbiota

remain one of the key factors in shaping bacterial communities if the diets are modified?

Will microbial communities of the peach diet change more with diet modifications? Will

any of the bacterial taxa exhibit a consistent shift in abundance based on the diet origin

and diet modification? Will we observe a variation in the microbial interactions and

influence of the environmental variables interactive effect on bacterial community based

on the diet modification?

2.1. We hypothesized that independent of the diet modifications, bacterial communities

of the peach diet will still be differentiated from the bacterial communities of the lab-

based diet.

2.2. We hypothesized that parental microbiota will influence the formation of bacterial

community of the larvae, especially on the HS diets.

2.3. We hypothesized that the microbial communities of autoclaved diets will exhibit

higher variation due to modification of the diets, since no established bacterial

community would be present before the modification.

2.4.We hypothesized that several dominant bacterial taxa will be consistently associated

with base diet (aka peach base or lab base) and certain modification of the food (HS

or HF).

2.5.We hypothesized that based on the diet modification, we will observe a variation in

the effects of environmental and genetic variables and their interactive effects on the

bacterial communities, as well as variation in the microbial interactions.

3. Will the influence of microbial taxa on larval phenotype vary within HF and HS diets?

Will we observe a variation in the influence of genetic, diet, treatment, and bacterial

interactive effects on larvae phenotypes based on nutritional modification and original

bases of the diet?

3.1.We hypothesized that microbial taxa will be correlated with larval phenotype;

however, the effect will vary with diet modifications.

3.2. We hypothesized that we will observe specific interactive effects between microbial

abundances and independent variables and that these effects will vary between HF

and HS diets.

MATERIALS AND METHODS

Diet Preparation. Mixtures of decayed peaches were prepared and stored using the procedures outlined in the second chapter. Peach food (PR) and standard lab food (R) diets were prepared according to the protocols described previously (Dew-Budd, Jarnigan, and Reed 2016, Mendez et al. 2016, Watanabe and Riddle 2017) and in the second chapter of this work. In addition, high sugar and high fat food variants were prepared for both natural peach food and regular

Drosophila lab food diets. All diet types were distributed evenly in the same manner, with approximately 10 mL per vial.

Peach high-fat food (PHF) preparation protocol was the following: we allowed one liter of the peach food to thaw before supplementing it with coconut oil in an amount equal to 3% of the total mixture weight, then warmed the mixture to 28 °C (to melt the coconut oil), and completely homogenized it with an immersion blender before distributing it into vials. Regular high-fat food (RHF) preparation protocol was the following: we supplemented one liter of

Drosophila lab food with 30g of coconut oil (3%) before the food solidified during cooking, thoroughly mixed it and distributed it into vials. Peach high-sugar food (PHS) preparation protocol was the following: we allowed one liter of the peach food to thaw before supplementing it with sucrose in an amount equal to 11% by weight, then completely homogenized it with an immersion blender before distributing it into vials. In order to prepare autoclaved versions of peach food (PHFA, PHSA11, and PHSA6), vials containing the corresponding food were

71 autoclaved for 25 min at 121 °C. Since our first results indicated that most of the tested genotypes could not survive on PHSA11 if the embryos went through sterilization treatment,

PHSA6 diet was introduced in the second and third rounds of the experiment. PHSA6 diet was prepared as described above for PHSA11 but with only 6% extra sucrose added by weight.

Regular high-sugar food (RHS) preparation protocol was the following: we supplemented one liter of the Drosophila lab food described above with 90 g of sucrose before the mixture solidified during cooking, then it was completely homogenized by thoroughly mixing it before distributing it into vials. The independent variable for the diet component will be referred to as

Drosophila stocks and husbandry. For the high-sugar portion of the study, the following five naturally derived genetic lines were sourced from the DGRP2 project: 153, 748,

787, 802, and 805 (Mackay et al. 2012, Huang et al. 2014). For the high-fat portion of the study, the following 10 naturally derived genetic lines were sourced from the DGRP2 project: 142, 153,

440, 748, 787, 801, 802, 805, 861, and 882 (Mackay et al. 2012, Huang et al. 2014). These stocks were maintained using the procedures as described in previous works, maintaining them at a constant temperature, humidity, and light/dark cycle (Dew-Budd, Jarnigan, and Reed 2016,

Mendez et al. 2016). The independent variable for the genetic component will be referred to as

Drosophila embryos sterilization. In order to test for the effects of parental microbiota, sterilized and non-sterilized embryos were prepared as described in the second chapter. The independent variable for the sterilization treatment component will be referred to as T.

Larval rearing and collection. Due to the nature of rearing and collecting and the large number of larvae required, this portion of the study was carried out in separate high-fat and high- sugar portions. Each portion of the study followed the same pattern. Across three separate time periods (~25-30 days apart), 50 sterilized or non-sterilized larvae of each genetic line were added to all diet types. For the high-fat component, there were at least three vials each of PHF, PHFA,

RHF, PA, PR and R food. For the high-sugar component, there were at least three vials each of

PHS, PHSA (from the second round, PHSA6% and PHSA11% to help mitigate low survival on food containing 11% sucrose), RHS, PA, PR, and R food. For the HS study, only NS larvae were used for PR and R diets as controls. The independent variable for the time period (round) component will be referred to as Ro. Larvae were allowed to develop until the late 3rd instar stage and were subsequently collected, sorted, and stored following the same protocols outlined in the second chapter of this work. The total time spent collecting larvae for the high-fat and high-sugar portions of the experiment was 98 and 106 days, respectively.

Measuring Experimental Phenotypes. The measured experimental phenotypes of survival, developmental rate, and weight were measured using the same procedures detailed in the second chapter. Triglyceride concentrations were quantified, by homogenizing 10 larvae per sample and measuring the total triglyceride concentration using the Sigma triglyceride

Determination Kit (Clark and Keith 1988, De Luca et al. 2005, Reed et al. 2010). Results were adjusted to represent the average triglyceride concentration per mg of dry larval weight. Protein concentrations were quantified using the Bradford’s method with 10 homogenized larvae per sample (with the exception of 3% of HF and 1% of HS samples, in which we used one to nine larvae due to especially low survival rates of the specific groups) (Bradford 1976, Dew-Budd,

Jarnigan, and Reed 2016). Protein values were averaged to represent the protein concentration

73 per mg of dry larvae weight. Glucose values, using combined trehalose and glucose concentrations for most samples, were quantified via homogenization of 10 larvae (with the exception of several HF and HS samples in which we used one to nine larvae) with subsequent overnight incubation in 1 μg/mL trehalase solution and further application of the Sigma Glucose

Determination Kit as described in the second chapter. Glucose concentrations were averaged and adjusted to represent the amount of glucose per mg of dry larval weight.

DNA extraction and sequencing. Five genetic lines 153, 748, 787, 802, and 805 raised on HF and HS diets were used for DNA sequencing. DNA extraction, sequencing, and processing were carried out using the methods described in detail in the second chapter, with the following modifications. SILVA 132 Qiime release database was used as the reference database for the taxonomic assignment. Alignments were filtered by QIIME’s filter_alignment.py script.

Reads cumulative sum scaling (CSS) normalization for alpha and beta diversities were performed through QIIME 1.91 with metagenomeSeq 1.26.1(Paulson et al. 2013). The phylogenetic trees of ZOTUs were assembled using the default options of QIIME 1.9.1 with the

FastTree program (Price et al. 2009).

Statistical Analysis

Data transformation. Normality tests, data transformations, and statistical models for larval phenotypes were done with JMP Pro 15.0. Phenotype measurements were tested for normality with the Shapiro-Wilk test and an outlier box plot. All phenotypic measurements data except survival were transformed (S3.1).

Larval phenotype statistical modeling: The contribution of each variable and their interactive effect on phenotype development was evaluated using a standard least squares model

74 with model effects to include Diet (D), Genotype (G), Sterilization Treatment (T) and their specific interactive effect: D*G (diet-by-genotype), D*T (diet-by-treatment), G*T (genotype-by- treatment), D*G*T (diet-by-genotype-by-treatment). If the time period of the experiment (R) and/or the variance between the colorimetric assay runs (triglyceride, protein, and glucose) (P) produced a significant effect, these variables were included in the model’s effects unless adding them would make the model not pass a lack of fit test. Data transformations for all of the models as well as the list of included variables are in S3.1.

The general three-way interaction model for evaluation of the influence of environmental and genetic factors on larvae phenotype:

푦 = 훽 + 훽퐷 + 훽퐺 + 훽푇 + 훽(퐷 ∗ 퐺) + 훽(퐷 ∗ 푇) + 훽(퐺 ∗ 푇)

+ 훽(퐷 ∗ 퐺 ∗ 푇) + 휀 with the model for the square root transformation of glucose by weight including the additional significant variable P (… + 훽푃) and the model for the cube root transformation of protein by weight including the additional significant variables Ro and P (… + 훽푃 + 훽푅).

The general two-way interaction model for evaluation of the influence of independent variables and their interactive effects on larvae phenotype (applied when the deleterious effects of the diet modification and sterilization would not provide sufficient samples to permit evaluation of the three-way interaction):

푦 = 훽 + 훽퐷 + 훽퐺 + 훽푇 + 훽(퐷 ∗ 퐺) + 훽(퐷 ∗ 푇) + 훽(퐺 ∗ 푇) + 휀 with models for the square root transformation of glucose by weight and the cube root transformation of triglyceride by weight including the additional significant variable P (…

+ 훽푃) and the model for the cube root transformation of protein by weight including the additional significant variables R and P (… + 훽푃 + 훽푅).

The general one variable comparison model for comparisons of HF and HS diets with a single treatment across all measured phenotypes is the following (applied for evaluation of the independent variable effect on larval phenotypes and for the post-hoc test. S and NS treatments were always separated):

푦 = 훽 + 훽푥 + 휀

For all models above, 푦 is the response, 훽 values are constants, and 휀 is a random error term.

Verification that the built model fit the data and post-hoc pairwise comparisons were carried out as described in the second chapter of this work.

All the statistical tests that were performed in R were run with version 3.6.1 (R core team 2019).

Microbial Diversity. Alpha diversity indices (species richness, Shannon and Simpson diversity indices) were calculated using R package vegan v.2.5-6 (Oksanen et al. 2009). In order to evaluate the effect of diet and treatment on alpha diversity indices, as well as pairwise difference in alpha diversities between groups of samples, we applied a linear analysis of variance model using the aov function (base R) and Fisher’s LSD post-hoc test using DescTools v.0.99.36 (Signorell et al. 2020, Team and DC 2019). Weighted Unifrac distances were calculated by beta_diversity_through_plots.py script with R 3.6.1 and Vegan v2.4-2 package

(Oksanen et al. 2009).

Sample groups were compared with permutational analysis of variance using the anosim function in vegan v.2.5-6 with 999 permutations (Oksanen et al. 2009). Bray-Curtis distances were calculated and compared with the anosim function as described above.

Microbial abundance. The difference in abundance of microbial taxa was evaluated with the Wilcoxon test as described in the second chapter of this work. In order to evaluate if the diets could serve as categorical predictors for classification of the larvae microbial samples as well as to which of the microbial taxa drive the differentiation of the diets, linear discriminant analysis was performed with the lda function in MASS v.7.3-51.4 (Venables and Ripley 2002) at the phylum, class, order, family, and genus taxonomic levels for the 10 most abundant representatives of the taxa at each taxonomic level. The first and second linear discriminants were visualized with a ggplot2 v.3.2.1 (Wickham 2016).

The correlation between microbial abundances on each diet was evaluated with the

Spearman rank correlation test as described in the second chapter of this work. The only exception was the analysis of sterilized larvae raised on PHSA6 due to the limitation of available samples (4). The correlations in this group were evaluated with linear regression model using the lm function (base R) and glance function from broom v.0.7.0 (Robinson 2014). Due to the technical limitations at the ZOTU level, we used a linear regression model using JMP v.15.

We evaluated the influence of diet, genotype, treatment and their interactive effect on the abundance of microbial taxa with linear regression model using lm and anova functions (base

R). A square root transformation was applied to normalize the microbial abundances.

In order to evaluate the influence of microbial taxa on larvae phenotypes, we randomly assigned each phenotypic measurement within each larvae group (based on diet, treatment, and

77 genetic line) to one of the three groups and found an average per group.. The correlation coefficients and p values were calculated with a Spearman rank correlation test as described in the second chapter of this work. Since we were only able to obtain four sequencing samples for S larvae raised on the PHSA6 diet (an insufficient sample size for the test described above), we evaluated the correlations with generalized linear quasi-Poisson models with the glm and anova functions (base R). The quasi-Poisson model was selected as it fits over dispersed data better than a linear regression model (Ver Hoef and Boveng 2007).

푌 ~ Poi(μ, 휃)

푣푎푟(푌) = 푣(휇) = 휃휇

where the mean μi for the ith observation is allowed to vary as a function of the phenotypic measurements and μi > 0 (Ver Hoef and Boveng 2007).

The model can then be generated using

μ = expβ + β푥, + ⋯+ β푥,

where βp is the estimated regression coefficient for the pth term and each x is a component of a larger set of p explanatory values for each phenotypic measurement (Ver Hoef and Boveng

2007).

In order to further measure the interactive effects between diet, genotype, and treatment

(separately) and abundance of each microbial taxa on larvae phenotypes, we used a quasi-

Poisson model with the R functions described above.

RESULTS

1.1 Larval life history traits and metabolic phenotype varies with a diet

Survival. The number of larvae that survived on a regular high fat diet was significantly higher than the larval survival on a peach high fat diet (NS: p< 0.0001, S: p< 0.0001) (Fig. 3.1A, Table

3.1). However, more larvae survived on the PHF diet comparing with the PHFA diet (NS: p<

0.0001, S: p< 0.0001) (Fig. 3.1A, Table 3.1). Independent from the presence of parental microbiota, larvae raised on a RHS diet had higher survival rates than those raised on a PHS diet

(NS: p<0.0001, S: p< 0.0001) (Fig. 3.2A, Table 3.1). The number of larvae that survived on a

PHS diet was significantly higher than on the PHSA11 diet, whether parental microbes were present or not (NS: p = 0.0001, S: p< 0.0001) (Fig. 3.2A, Table 3.1). More larvae survived on a

PHS diet compared to a PHSA6 diet, but only when parental microbes were removed (S: p<0.0001) (Fig. 3.2A, Table 3.1). Between the autoclaved peach diets, the survival of the larvae was higher on a 6% diet compared to 11% diet but only if parental microbiota was present (NS: p< 0.0001) (Fig. 3.2A, Table 3.1).

Development rate. The development rate was slower among larvae raised on PHF food compared to RHF (NS: p< 0.0001, S: p< 0.0001) (Fig. 3.1B, Table 3.1). However, it was faster than on PHFA diet (NS: p< 0.0001, S: p< 0.0001) (Fig. 3.1B, Table 3.1). For both sterilized and non-sterilized treatments, larvae developed faster on RHS compared to a PHS diet (NS: p<

0.0001, S: p< 0.0001) (Fig. 3.2B, Table 3.1). In the presence of parental microbiota, larvae developed faster when raised on the peach high sugar diet compared to those raised on either

PHSA6 or PHSA11 (both p < 0.0001) (Fig. 3.2B, Table 3.1). Also, only in the presence of

79 parental microbiota, larvae took longer to develop on a PHSA11 diet when compared to larvae raised on the PHSA6 diet (p < 0.0001) (Fig. 3.2B, Table 3.1).

Weight. The weight of RHF diet raised larvae was significantly higher than PHF raised larvae (NS: p< 0.0001, S: p< 0.0001) (Fig. 3.1C, Table 3.1). However, compared to larvae raised on PHFA, larvae raised on PHF food were significantly heavier (NS: p< 0.0001, S: p< 0.0001)

(Fig. 3.1C, Table 3.1). Independent of treatment, larvae raised on a RHS diet were significantly heavier than those raised on a PHS diet (NS: p <0.0001, S: p=0.0138) (Fig. 3.2C, Table 3.1).

Larvae raised on a PHS diet weighed significantly more than larvae raised on a PHSA11 diet only when parental microbes were present (NS: p=0.0197) (Fig. 3.2C, Table 3.1). There was no significant difference in larval weight between a PHS diet and a PHSA6 diet or between the two

PHSA diets for either treatment (Fig. 3.2C, Table 3.1).

Triglyceride. Larvae raised on a RHF diet had significantly lower triglyceride concentrations compared to larvae raised on a PHF diet (NS: p< 0.0001, S: p< 0.0001) (Fig.

3.1D, Table 3.1). Independent of the treatment, larvae raised on the PHFA diet had higher triglyceride concentrations by weight compared to larvae raised on PHF diet (NS: p= 0.0061, S: p< 0.0001) (Fig. 3.1D, Table 3.1). Larvae raised on a peach high sugar diet had significantly higher triglyceride concentrations than those that were raised on a regular high sugar diet but only when parental microbes were present (NS: p<0.0001) (Fig. 3.2D, Table 3.1). If parental microbiota were present, larvae raised on a PHSA11 diet had significantly higher triglyceride concentrations compared to those raised on a PHS diet (NS: p< 0.0001) (Fig. 3.2D, Table 3.1).

However, independent of the treatment, larvae raised on PHSA6 diet were not different in triglyceride concentrations comparing with larvae raised on PHS diet (Fig. 3.2D, Table 3.1).

When parental microbes were present, larvae raised on the PHSA11 diet contained higher

80 triglyceride concentrations than those on the corresponding 6% sugar diet (NS: p=0.0016), but there was no significant difference when the larvae were sterilized (Fig. 3.2D, Table 3.1).

Protein. Larvae raised on a RHF diet had significantly higher protein by weight concentrations, compared to larvae raised on a peach high fat diet, but only in the absence of parental microbiota (S: p=0.01) (Fig. 3.1E, Table 3.1). Sterilized larvae consuming the autoclaved peach high fat diet had significantly higher protein concentrations than the ones consuming the regular high fat peach diet (S: p< 0.0001) (Fig. 3.1E, Table 3.1). Only when parental microbes were removed, larvae raised on a RHS diet had higher concentrations of protein by weight than larvae raised on a PHS diet (S: p=0.0010) (Fig. 3.2E, Table 3.1).

Independent from treatment, there was no significant difference in protein concentration for larvae raised on a PHS diet compared to a PHSA11 diet (Fig. 3.2E, Table 3.1). Non-sterilized larvae had higher protein concentrations when raised on a PHS diet compared to a PHSA6 diet

(Fig. 3.2E, Table 3.1). In contrast, sterilized larvae raised on a PHS diet had lower protein concentrations than those raised on a PHSA6 diet (NS: p=0.0365, S: p=0.0338) (Fig. 3.2E, Table

3.1). Non-sterilized larvae raised on a PHSA11 diet contained a higher protein concentration than larvae on the PHSA6 diet (NS: p=0.0159), but there was no significant difference for the sterilized treatment (Fig. 3.2E, Table 3.1).

Glucose. Larvae raised on the RHF diet had significantly higher glucose by weight concentrations compared to larvae raised on the PHF diet for both treatments (NS: p< 0.0001, S: p< 0.0001) (Fig. 3.1F, Table 3.1). Independent of the treatment, larvae raised on the PHFA diet had higher glucose concentrations by weight compared to larvae raised on PHF diet (NS: p<0.0001, S: p< 0.0001) (Fig. 3.1F, Table 3.1). Larvae raised on a PHS diet had higher glucose concentrations than those raised on a RHS diet regardless of treatment (NS: p=0.0063, S:

81 p=0.0469) (Fig. 3.2F, Table 3.1). Sterilized larvae raised on a PHS diet had a significantly higher glucose concentration by weight than larvae on a PHSA11 diet (S: p=0.0152) (Fig. 3.2F, Table

3.1), but there was no significant difference in glucose concentrations when parental microbes were present (Fig. 3.2F, Table 3.1). There was no significant difference in the glucose concentration by weight of larvae raised on a PHS diet compared to a PHSA6 diet or between the two autoclaved peach high sugar diets regardless of whether parental microbes were present or absent (Fig. 3.2F, Table 3.1).

Treatment Survival Development Weight Triglyceride Protein Glucose NS PHF > PHFA *** PHF < PHFA *** PHF > PHFA *** PHF < PHFA ** PHF < PHFA PHF < PHFA *** S PHF > PHFA *** PHF < PHFA *** PHF > PHFA *** PHF < PHFA *** PHF < PHFA *** PHF < PHFA *** NS RHF > PHF *** RHF < PHF *** RHF > PHF *** RHF < PHF *** RHF > PHF RHF > PHF *** S RHF > PHF *** RHF < PHF *** RHF > PHF *** RHF < PHF *** RHF > PHF * RHF > PHF *** NS PHS < RHS *** PHS > RHS *** PHS < RHS *** PHS > RHS *** PHS > RHS PHS > RHS ** S PHS < RHS *** PHS > RHS *** PHS < RHS * PHS > RHS PHS < RHS ** PHS > RHS * NS PHS > PHSA11 ** PHS < PHSA11 *** PHS > PHSA11 * PHS < PHSA11 *** PHS < PHSA11 PHS < PHSA11 S PHS > PHSA11*** PHS < PHSA11 PHS > PHSA11 PHS < PHSA11 PHS < PHSA11 PHS > PHSA11 * NS PHS < PHSA6 PHS < PHSA6 *** PHS > PHSA6 PHS < PHSA6 PHS > PHSA6 * PHS < PHSA6 S PHS > PHSA6 *** PHS < PHSA6 PHS > PHSA6 PHS < PHSA6 PHS < PHSA6 * PHS > PHSA6 NS PHSA11 < PHSA6 *** PHSA11 > PHSA6 *** PHSA11 < PHSA6 PHSA11 > PHSA6 ** PHSA11 > PHSA6 * PHSA11 < PHSA6 S PHSA11 < PHSA6 PHSA11 > PHSA6 PHSA11 > PHSA6 PHSA11 > PHSA6 PHSA11 < PHSA6 PHSA11 > PHSA6

Table 3.1: Comparison of phenotypes of the larvae raised on high fat and high sugar diets.

Asterisks indicate the significance of comparisons p< 0.001 ***, p< 0.01 **, and p< 0.05 *.

Figure 3.1: The presence of environmental and parental microbiota is beneficial for larvae raised on high fat diets. A) mean survival until the late 3rd instar stage is increased on lab diet and with presences of bacteria on peach diets; B) median development time is decreased on a lab diet and with the presence of dietary and/or maternal microbiota on peach diets; C) mean larval weight is higher on a lab diet and in the presence of microbiota on peach diets; D) Triglyceride concentrations by weight are reduced on a lab diet and in the presence of environmental microbiota on peach diets; E) Protein by weight is lower on a peach diet, in the presence of microbiota; F) Glucose by weight is lowest in larvae raised on a peach diet in the presence of microbes. Error bars indicated one standard error of the mean. Larvae raised on high fat

83 modified diets largely maintain the same relations between peach and lab based food as larvae raised.

Figure 3.2: The presence of environmental and parental microbiota is beneficial for larvae raised on high sugar diets. A) mean survival until the late 3rd instar stage is increased on lab diet and with presences of microbiota on peach diets; B) median development time is decreased on a lab diet and with the presence of dietary and/or maternal microbiota on peach diets; C) mean larval weight is higher on a lab diet and in the presence of microbiota; D) Triglyceride concentrations by weight are reduced on a lab diet in the presence of parental microbiota; E)

Protein by weight is lower on RHS and PHSA6 diets, in the presence of microbiota; F) Glucose by weight phenotype exhibits much variation between diets and treatments. Error bars indicated

84 one standard error of the mean. Larvae raised on high sugar modified diets exhibit much variation in the phenotypic response between diets and treatments.

1.2 Parental microbiota influences larval life history traits and metabolic phenotypes

Survival. The presence of parental microbiota did not affect survival of the larvae raised on RHF diet but significantly increased the survival of larvae on PHF (p= 0.0138) and PHFA (p<

0.0001) diets (Fig. 3.1A, Table 3.2). On the RHS diet there was no significant difference in larval survival between sterilized and non-sterilized treatments (Fig. 3.2A, Table 3.2). For each peach high sugar diets (PHS, PHSA6, and PHSA11), larvae that retained parental microbiota survived in significantly higher numbers (PHS: p<0.0001, PHS: p=0.0008, PHSA11: p<0.0001) (Fig.

3.2A, Table 3.2).

Development rate. There was no significant difference between sterilized and non- sterilized treatments on the RHF diet (Fig. 3.1B, Table 3.2). However, the presence of parental microbiota significantly reduced the development time of larvae raised on the PHF (p< 0.0001) and PHFA (p< 0.0001) diets (Fig. 3.1B, Table 3.2). Presence of parental microbiota increased the development rate on RHS and PHS diets (both p<0.0001), while no significant effects on other

HS diets were observed (Fig. 3.2B, Table 3.2). Interestingly, on the peach HS diets the presence of parental microbiota played a more noticeable role in the reduction of larval development time when compared to environmental microbiota (Fig. 3.2B, Table 3.2).

Weight. The absence or presence of parental microbiota did not have a significant effect on the weight of the larvae raised on RHF and PHF diets, but sterilized larvae raised on the

PHFA diet weighed significantly less than non-sterilized larvae on the same diet (p< 0.0001)

(Fig. 3.1C, Table 3.2). The presence of parental microbes increased larval weight on a RHS diet

(p=0.0497) and decreased the weight of the larvae on a PHSA11 diet (p=0.0092) (Fig. 3.2C,

Table 3.2). Treatment did not influence the weight of larvae raised on other diets.

Triglyceride. The absence of parental microbiota increased the triglyceride concentration only for larvae raised on a PHFA diet (p=0.0005) and produced no significant effect on other HF diets (Fig. 3.1D, Table 3.2). The removal of parental microbes significantly increased the concentration of triglyceride by weight for larvae raised on a RHS diet (p<0.0001), but did not have a significant effect on the triglyceride concentrations of the larvae raised on any of the other

HS diets (Fig. 3.2D, Table 3.2).

Protein. Parental microbes had no effect on the larvae raised on RHF or PHF diets (Fig.

2.1E, Table 2.2). The absence of parental microbiota significantly increased protein concentrations in larvae raised on the PHFA diet (p= 0.0408) (Fig. 3.1E, Table 3.2). The removal of parental microbes increased the concentration of protein in the larvae raised on the PHSA6 diet and RHS diets (both p<0.0001) (Fig. 3.2E, Table 3.2). The sterilization treatment did not produce a significant effect on protein concentration for the larvae raised on either the PHSA11 diet or the PHS diet (Fig. 3.2E, Table 3.2).

Glucose. Parental microbiota did not produce a significant effect on the glucose concentrations of the larvae raised on any of the high fat diets (Fig. 3.1F, Table 3.2). The presence of parental microbiota positively affected the larvae glucose concentrations on PHSA6

(p= 0.0428) diet but did not produce any effect on all other high sugar diets (Fig. 3.2F, Table

3.2).

Diet Survival Development Weight Triglyceride Protein Glucose PHF S < NS * S > NS *** S < NS S > NS S < NS S < NS PHFA S < NS *** S > NS *** S < NS *** S > NS ** S > NS * S > NS RHF S < NS S > NS S < NS S < NS S > NS S < NS PHSA6 S < NS *** S > NS S > NS S > NS S > NS *** S < NS * PHSA11 S < NS *** S < NS S > NS ** S > NS S > NS S < NS PHS S < NS ** S > NS *** S > NS S < NS S > NS S < NS RHS S < NS S > NS *** S < NS * S > NS *** S > NS *** S < NS

Table 3.2: The difference in phenotypes between the larvae with (NS) and without (S) parental microbiota. Asterisks indicate the significance of comparisons p< 0.001 ***, p< 0.01

**, and p< 0.05 *.

1.3 Diet, genotype, treatment and their interactive effect significantly influence most of the larval phenotypes

Survival. On the high fat diets, all of the independent variables significantly influenced the survival of the larvae (diet, genetic line, treatment, diet-by-genotype (D*G), diet-by- treatment (D*T) p<0.0001; genotype-by-treatment (G*T) and diet-by-genotype-by-treatment

(D*G*T) p=0.0002) (Table 3.3). Diet was the most influential variable and explained 46.90% of the variance (VE) in larvae survival. Following the diet, genotype was the second most influential factor (VE=4.74%) (Table 3.3). Although the variables’ interactive effects produced a significant effect on the survival, they explained less variance in the data than independent variables alone, D*T (VE=2.91%), D*G (VE=2.66%), treatment (VE=2.56%), D*T*G

(VE=1.53%), and G*T (VE=1.04%) (Table 3.3). On the high sugar diets, all variables except for

G*T were significant in influencing survival (genetic line p<0.0001, diet p<0.0001, treatment p<0.0001, G*D p<0.0001, D*T p<0.0001, G*D*T p=0.0063). Diet was the most influential variable (VE=26.11%) followed by treatment (VE=19.08%), and D*T (VE=13.35%) (Table

3.4A).

Development rate. All of the variables and their interactive effects except for D*T*G significantly influenced the development rate of larvae on the high fat diets (diet, genetic line, treatment, G*T, and D*G p<0.0001; D*T p=0.001) (Table 3.3). Diet produced the biggest effect on the development rate (VE=30.80%), followed by the genetic line (VE=5.58%), D*G

(VE=4.75%), G*T (VE=4.39%), treatment (VE=2.49%), and D*T (VE=1.04%) (Table 3.3). On the high sugar diet, genetic line (p<0.0001), diet (p<0.0001), treatment (p<0.0001), G*D

(p=0.0225), and D*T (p=0.0404) had a significant effect on development rate of the third instar larvae (Table 3.4A). Diet (VE=44.12%), treatment (VE=8.83%), and genetic line (VE=6.41%) were the most influential variables (Table 3.4A).

Weight. On the high fat diets, all of the independent variables and their interactive effects significantly affected the weight of the third instar larvae (diet, genetic line, treatment, D*G,

G*T, D*T*G p<0.0001; D*T p=0.0015) (Table 3.3). The diet explained the biggest portion of the variance (VE=19.44%), with genetic line contributing the second highest (VE=8.54%) followed by D*T*G interactive effect (VE=5.29%) (Table 3.3). Due to the low survival rates of the sterilized larvae raised on high sugar autoclaved diets, we used statistical models that both included and excluded treatment as a variable. These were assessed to account for those diets that produced insufficient sample size of third instar larvae, primarily PHSA11 and PHSA6.

Taking into account all of the variables on high sugar diets, genetic line (p<0.0001), diet

(p<0.0001), G*D (p<0.0001), D*T (p=0.0413), and G*D*T (p=0.035) significantly affected the weight of the larvae (Table 3.4A). Diet explained most of the variation in weight (VE=7.75%) followed by G*D (VE=4.95%), genetic line (VE=4.86%), and G*D*T (VE=1.66%) (Table

3.4A). Excluding the treatment but increasing the number of genetic lines used in the model produced the following results: genetic line (p<0.0001), diet (p<0.0001), and G*D (p<0.0001)

(Table 2.4B). Of these variables, diet had the greatest variance explained (VE=18.51%), followed by genetic line (8.23%) and the G*D interaction (6.28%) (Table 3.4B).

Triglyceride. All of the tested variables and their interactive effects significantly influenced triglyceride concentrations in larvae on HF diets (diet, genetic line, D*G, and G*T p<0.0001; D*T*G p=0.0002; treatment p=0.0006; D*T p=0.031) (Table 3.3). Diet explained the largest portion of the variance in triglyceride concentrations (VE=41.70%) which was followed by G*T (VE=9.19%), genetic line (VE=7.62%), D*T*G (VE=7.37%), D*G (VE=3.15%), D*T

(VE=2.77%), and treatment (VE=1.799%) (Table 3.3). For the model that included all of the independent variables, for the HS diets, only the G*D*T and G*T variables did not produce a significant effect on larvae triglyceride concentrations (diet p<0.0001, treatment p<0.0001, D*T p<0.0001, genetic line p=0.0045, G*D p=0.0134) (Table 3.4A). Comparing with other variables, diet explained biggest portion of the variance (VE=13.89%), followed by treatment (VE=9.09%),

D*T (VE=8.21%), genetic line (VE=5.80%), and G*D (VE=4.798%) (Table 3.4A). In the model that excluded T variable, all four variables produced a significant effect on larvae triglyceride concentrations (diet p<0.0001, plate p<0.0001, G*D p=0.0074, genetic line p=0.0142) (Table

3.4B). Of these variables, diet still explained the highest proportion of the variance

(VE=44.18%) (Table 3.4B).

Protein. Similar to triglyceride, all of the independent variables demonstrated a significant effect on the protein concentrations of third instar larvae on the high fat diets (all p<0.0001) (Table 3.3). This phenotype was the only one tested that did not have the largest variance explained by diet. Instead, G*T produced the largest effect (VE=13.95%), followed by

D*T*G (VE=12.79%), genetic line (VE=12.28%), diet (VE=10.78%), D*T (VE=7.27%), D*G

(VE=6.10%), and finally, treatment (VE=3.81%) (Table 3.3). On the HS diets, in the model that

89 included all of the independent variables, the significant variables were treatment (p<0.0001),

D*T (p=0.0008), and genetic line (p=0.0242) (Table 3.4A). Treatment explained most of the variance (VE=4.97%), with other variables contributing even smaller values: D*T (VE=2.88%), and genetic line (VE=2.87%) (Table 3.4A). Excluding the treatment from the model produced the following results: genetic line (p=0.0277), and G*D (p=0.0182), with G*D explaining the largest portion of the variance (VE=6.48%) (Table 3.4B).

Glucose. Only three of the independent variables demonstrated a significant effect on the glucose concentrations of the third instar larvae on the high fat diets. Diet explained most of the variance in glucose concentration (p<0.001, VE=43.58%) (Table 3.3). Genetic line (p= 0.0074) and D*G (p= 0.0198) were the other significant independent variables (genetic line VE=5.75% and G*T VE=4.83%) (Table 3.3). On high sugar diets, only the diet produced a significant effect on larvae the glucose concentrations (p=0.0002) and was the most impactful out of all experimental variables (VE=7.16%), according to the model with all tested variables (Table

3.4A). Similar results were observed if treatment was removed from the model; only diet produced a significant effect on the glucose concentrations of the larvae (p=0.0001) and was still the most influential experimental variable (VE=9.72%) (Table 3.4B).

Independent variable Survival Development Weight Triglyceride Protein Glucose Diet VE=46.9% *** VE=30.8% *** VE=19.4% *** VE=41.7% *** VE=10.8% *** VE=43.6% *** Genetic line VE=4.74% *** VE=5.58% *** VE=8.54% *** VE=7.61% *** VE=12.3% *** VE=5.75% ** Treatment VE=2.56% *** VE=2.49% *** VE=.608% *** VE=1.799% ** VE=3.81% *** VE=0.00281% Diet*Genetic line VE=2.66% *** VE=4.75% *** VE=3.45% *** VE=3.15% *** VE=6.099% *** VE=0.195% Diet*Treatment VE=2.91% *** VE=1.04% ** VE=.424% ** VE=2.77% * VE=7.27% *** VE=2.95% Genetic line*Treatment VE=1.04% ** VE=4.39% *** VE=2.50% *** VE=9.19% *** VE=13.95% *** VE=4.83% * Diet*Treatment*Genetic line VE=1.53% ** VE=2.14% VE=5.29% *** VE=7.37% ** VE=12.8% *** VE=1.32%

Table 3.3: The influence of diet, genotype, treatment, and their interactive effect on phenotype of the larvae raised on high fat diets. Asterisks indicate the significance of comparisons p< 0.001 ***, p< 0.01 **, and p< 0.05 *.

Independent variable Survival Development Weight Triglyceride Protein Glucose Genetic line VE=1.51% *** VE=6.41% *** VE=4.86% *** VE=5.80% ** VE=2.87% * VE=1.83% Diet VE=26.11% *** VE=44.12% *** VE=7.75% *** VE=13.89% *** VE=0.448% VE=7.16% ** Treatment VE=19.08% *** VE=8.83% *** VE=0.236% VE=9.09% *** VE=4.97% *** VE=1.89% Genetic line*Diet VE=1.41% *** VE=1.33% * VE=4.95% *** VE=4.798% * VE=0.8697% VE=2.87% Genetic line*Treatment VE=0.135% VE=0.978% VE=0.829% VE=1.93% VE=1.91% VE=1.49% Diet*Treatment VE=13.35% *** VE=0.488% * VE=0.778% * VE=8.21% *** VE=2.88% ** VE=0.0667% Genetic line*Diet*Treatment VE=0.683% ** VE=0.256% VE=1.66% * VE=2.01% VE=1.39% VE=3.56%

Table 3.4A: The influence of diet, genotype, treatment, and their interactive effect on phenotype of the larvae raised on high sugar diets. Asterisks indicate the significance of comparisons p< 0.001 ***, p< 0.01 **, and p< 0.05 *.

Independent variable Weight Triglyceride Protein Glucose Genetic line VE=8.23% *** VE=2.25% * VE=1.98% * VE=0.58% Diet VE=18.51% *** VE=44.18% *** VE=1.53% VE=9.72% ** Genetic line*Diet VE=6.28% *** VE=7.15% ** VE=6.48% * VE=6.01%

Table 3.4B: The influence of diet, genotype, and their interactive effect on phenotypes of the larvae raised on high sugar diets. Asterisks indicate the significance of comparisons p<

0.001 ***, p< 0.01 **, and p< 0.05 *.

1.4 Larval response to nutritional modification varies between lab-based and peach-based diets

Survival. There was no difference in survival between standard lab diet and lab high fat diet raised larvae, for either treatment (Fig.3.3A, 3.4A, Table 3.5). Compared to larvae raised on a high fat peach diet, larvae raised on the regular peach diet had higher survival rates (NS: p=

0.0055, S: p< 0.0001) (Fig.3.3A, 3.4A, Table 3.5). The number of larvae that survived on PA

91 food was significantly higher than larval survival on the PHFA diet (NS: p< 0.0001, S: p<

0.0001) (Fig.3.3A, 3.4A, Table 3.5). Our results suggested that HF modification of the food negatively influenced the survival of the larvae on the natural diets. However, both environmental and parental bacteria could increase the survival of the larvae independently as well as have beneficial additive effects. These findings suggested that symbiotic bacteria could enhance a host’s ability to adapt to nutritional stress.

There was no significant difference in survival between larvae raised on the regular and regular high sugar diets for either treatment (Fig.3.5A, 3.6A, Table 3.5). More larvae survived on the PR diet compared to the PHS diet, but only for the sterilized treatment (S: p=0.0179)

(Fig.3.5A, 3.6A, Table 3.5). When compared to the PA diet, less larvae survived on both the

PHSA11 (S: p< 0.0001) and PHSA6 (S: p<0.0001) diets, but only for the sterilized treatment

(Fig.3.6A, Table 3.5). Surprisingly, more larvae survived on the PHSA6 diet compared to the PA diet if parental microbiota were present (NS: p= 0.0063) (3.5A, Table 3.5). Our results suggested that the presence of environmental and parental microbiota could improve larvae survival on HS peach diets. The presence of parental microbiota was sufficient to rescue the negative effect that increased sugar amounts produced on the survival of larvae.

Development. HF modification of the regular lab and peach diets did not produce a significant effect on developmental rate (Fig.3.3B, 3.4B, Table 3.5). With both control and sterilized treatments, larvae developed faster on PHFA food compared to the PA diet (NS: p<

0.0001, S: p< 0.0001) (Fig.3.3B, 3.4B, Table 3.5). Addition of fat increased the development rate of larvae raised on the autoclaved peach diet only, suggesting that environmental microbiota could reduce the effect that HF produced on larva development.

Similar to results seen for survival, there was no significant difference in development rate between larvae raised on the regular versus the regular high sugar diets (Fig.3.5B, 3.6B,

Table 3.5). For both treatments, larvae raised on the PR diet developed faster than those raised on the PHS diet (NS: p<0.0001, S: p<0.0001) (Fig.3.5B, 3.6B, Table 3.5). Non-sterilized larvae raised on the PA diet developed faster than those raised on the PHSA11 diet (NA: p<0.0001)

(Fig.3.5B, Table 3.5). In contrast, sterilized larvae developed faster on the PHSA11 diet than the

PA diet (S: p= 0.0043) (Fig. 3.6B, Table 3.5). Compared to the PHSA6 diet, flies raised on the

PA diet took longer to develop, but only for the sterilized treatment (S: p= 0.0002) (Fig.3.5B,

3.6B, Table 3.5).

Weight. Compared to larvae raised on a regular high fat diet, larvae raised on a regular diet were significantly heavier (NS: p< 0.0001, S: p< 0.0001) (Fig.3.3C, 3.4C, Table 3.5). We found that only non-sterilized larvae consuming the regular peach diet weighed significantly more than the ones consuming a high fat peach diet (Fig.3.3C, 3.4C, Table 3.5). Our findings suggested that on high fat diets parental and environmental microbiota might produce interactive effects on larvae weight.

For both treatments, larvae raised on the regular diet weighed more than those on the regular high sugar diet (NS: p<0.0001, S: p<0.0001) (Fig.3.5C, 3.6C, Table 3.5). There was no significant difference in weight between flies raised on the PR diet and the PHS diet for the sterilized treatment, but larvae raised on the peach regular diet weighed more if parental microbiota were inherited (NS: p<0.0001) (Fig.3.5C, 3.6C, Table 3.5). Sterilized larvae raised on both PHSA11 (S: p<0.0001) and PHSA6 (S: p<0.0001) diets weighed more compared to larvae raised on the PA diet, but there was no significant difference for the non-sterilized larvae

(Fig.3.5C, 3.6C, Table 3.5).

Triglyceride. On the standard lab diet, only sterilized larvae experienced a significant increase in triglyceride concentrations produced by high fat modification of the food (S: p=

0.0449) (Fig.3.3D, 3.4D, Table 3.5). Addition of extra fats did not produce a significant effect on larvae raised on the peach food for either treatment (Fig.3.3D, 3.4D, Table 3.5). However, sterilized larvae raised on a PHFA diet had significantly higher triglyceride concentrations than those that were fed a PA diet (S: p= 0.0015) (Fig. 3.4D, Table 3.5). Our findings suggested that on the natural diet, both parental and environmental microbiota may be sufficient to reduce larvae triglyceride concentrations and even eliminate the elevated triglyceride concentrations caused by a HF modification of the food.

Larvae raised on the regular food had lower triglyceride concentrations than those on the regular high sugar food, but only when parental microbes were removed (S: p<0.0001)

(Fig.3.5D, 3.6D, Table 3.5). In contrast, flies raised on the PHS diet had higher triglyceride concentrations by weight than those on the PR, but only in the presence of parental microbes

(NS: p= 0.0107) (Fig.3.5D, 3.6D, Table 3.5). There was no significant difference in triglyceride levels between flies raised on the PA diet and the PHSA11 diet or the PHSA6 diet for either treatment (Fig.3.5D, 3.6D, Table 3.5).

Protein. Independent of the treatment, larvae raised on the regular diet had higher protein concentrations by weight compared to larvae raised on the regular high fat diet (NS: p=

0.0433, S: p= 0.0248) (Fig.3.3E, 3.4E, Table 3.5). In contrast, there was no significant difference between peach regular and peach high fat diets (Fig.3.3E, 3.4E, Table 3.5). Larvae raised on a

PHFA diet had significantly higher protein by weight concentrations, compared to larvae raised on a PA diet, but only in the absence of parental microbiota (S: p= 0.0027) (Fig.3.3E, 3.4E,

Table 3.5). Our findings suggested that on the peach diets both parental and environmental

94 microbiota were sufficient to reduce elevated protein concentrations produced by HF modification. Interestingly, additional fats produced the opposite effect on protein concentrations of the larvae raised on the standard lab diet and the presence of parental microbiota did not produce any effect in this case.

Larvae raised on the regular diet had a higher concentration of protein by weight compared to the regular high sugar diet but only when parental microbiota were present (NS: p=0.0003) (Fig.3.5E, 3.6E, Table 3.5). Independent of treatment, there was no significant difference in protein concentrations between larvae raised on the PR diet compared to the PHS diet (Fig.3.5E, 3.6E, Table 3.5). Larvae raised on the PA diet had a higher protein concentration by weight than those raised on PHSA11 for both treatments (NS: p= 0.0106, S: p= 0.0290)

(Fig.3.5E, 3.6E, Table 3.5). In the presence of parental microbes, larvae on the PA diet had a higher protein concentration compared to larvae raised on a PHSA6 diet (NS: p< 0.0001) but no significant difference was observed if embryos were sterilized (Fig.3.5E, 3.6E, Table 3.5).

Glucose. Compared to larvae raised on a regular lab diet, larvae raised on a RHF diet had significantly higher glucose concentrations but only in the presence of parental microbiota (NS: p= 0.0337) (Fig.3.3F, 3.4F, Table 3.5). In contrast, sterilized larvae raised on a PHF diet had significantly lower glucose concentrations than those raised on a PR diet (S: p= 0.0456)

(Fig.3.3F, 3.4F, Table 3.5). Similarly, only sterilized larvae raised on a PHFA diet had significantly lower glucose concentrations than larvae raised on the PA diet (S: p= 0.0059)

(Fig.3.3F, 3.4F, Table 3.5). Once again, larvae raised on natural and lab diets exhibited exactly opposite responses to the addition of extra fat.

Independent of treatment, glucose concentrations of larvae raised on R and RHS diets were not significantly different (Fig.3.5F, 3.6F, Table 3.5). Flies raised on the PHS diet had higher glucose concentrations than the PR diet only when parental microbes were removed (S: p<0.0001) (Fig.3.5F, 3.6F, Table 3.5). Only non-sterilized larvae raised on the PA diet had lower glucose concentrations than those on PHSA11 (NS: p= 0.0023) (Fig.3.5F, 3.6F, Table 3.5). In the presence of parental microbiota, PHSA6 raised larvae also had higher glucose concentrations than larvae raised on the PA diet (NS: p<0.0001) (Fig.3.5F, Table 3.5).

Treatment Survival Development Weight Triglyceride Protein Glucose NS R > RHF R < RHF R > RHF *** R < RHF R > RHF * R < RHF * S R > RHF R < RHF R > RHF *** R < RHF * R > RHF * R < RHF NS PR > PHF ** PR < PHF PR > PHF * PR < PHF PR > PHF PR > PHF S PR > PHF *** PR < PHF PR > PHF PR < PHF PR > PHF PR > PHF * NS R < RHS R > RHS R > RHS *** R > RHS R > RHS ** R > RHS S R < RHS R < RHS R > RHS *** R < RHS *** R > RHS R > RHS NS PR > PHS PR < PHS *** PR > PHS *** PR < PHS * PR < PHS PR < PHS S PR > PHS * PR < PHS *** PR < PHS PR < PHS PR > PHS PR < PHS *** NS PA > PHSA PA < PHSA *** PA < PHSA PA < PHSA PA > PHSA * PA < PHSA ** S PA > PHSA *** PA > PHSA ** PA < PHSA *** PA < PHSA PA > PHSA * PA > PHSA NS PA < PHSA6 ** PA > PHSA6 PA < PHSA6 PA > PHSA6 PA > PHSA6 *** PA < PHSA6 *** S PA > PHSA6 *** PA > PHSA6 ** PA < PHSA6 *** PA > PHSA6 PA > PHSA6 PA < PHSA6

Table 3.5: The influence of diet modification on larval phenotypes. Asterisks indicate the significance of comparisons p< 0.001 ***, p< 0.01 **, and p< 0.05 *.

Figure 3.3: The response of non-sterilized larvae to high fat nutritional modification varies between dietary types. A) HF modification caused a decrease in mean survival until the 3rd instar stage on peach based diets; B) HF modification caused a decrease in median development time only on peach autoclaved diet; C) HF modification caused a decrease in larval weights on lab and peach diet in the presence of environmental microbes; D) HF modification did not cause an increase in triglyceride concentrations; E) HF modification caused a decrease in larval protein concentrations only on the lab diet; F) HF modification caused a decrease in larval glucose concentrations only on the lab diet.

Figure 3.4: The response of sterilized larvae to high fat nutritional modification varies between dietary types. A) HF modification caused a decrease in mean survival until the 3rd instar stage on peach based diets; B) HF modification caused a decrease in median development time only on peach autoclaved diet; C) HF modification caused a decrease in larval weights only on a lab diet; D) HF modification caused an increase in triglyceride concentrations only on lab and autoclaved peach diet; E) The response in larval protein concentration to HF modification varied between diet types; F) HF modification caused a decrease in larval glucose concentrations only on peach diets.

Figure 3.5: The response of non-sterilized larvae to high sugar nutritional modification varies between dietary types. A) HS modification largely did not influence larval survival; B)

HS modification caused an increase in median development time only on peach diets; C) HS modification caused a decrease in larval weights on lab and peach diets but not on an autoclaved peach diet; D) HS modification caused an increase in triglyceride concentrations only on a peach diet in the presence of environmental microbiota; E) HS modification did not cause a decrease in protein concentrations only on a peach die if environmental microbiota were present; F) HS modification caused an increase in larval glucose concentrations only on autoclaved peach diet.

Figure 3.6: The response of sterilized larvae to high sugar nutritional modification varies between dietary types. A) HS modification decreased larval survival on both peach diets; B) In response to HS modification, the development time varied between diet types ; C) In response to

HS modification, larval weight varied between diet types; D) HS modification caused an increase in triglyceride concentrations only on a lab diet; E) HS modification caused a decrease in protein concentrations only on the autoclaved peach diet; F) HS modification caused an increase in larval glucose concentrations only on a peach diet in a presence of environmental microbiota.

2.1 Diet produced a significant effect on bacterial community composition but had few impacts on alpha diversity

Alpha diversity. Surprisingly, we did not observe any significant difference in alpha diversity indices between non-sterilized and sterilized treatments for larvae raised on a HF diet.

100

For larvae raised on HS diets, only the species richness of NS larvae raised on PHS diet was significantly higher than that of larvae raised on RHS diet (p= 0.0439) (Table 3.6). All other comparisons were not significant.

Treatment Shannon index Simpson index Richness NS PHFA > PHF PHFA > PHF PHFA > PHF NS RHF > PHF RHF > PHF RHF > PHF NS RHF > PHFA RHF > PHFA RHF < PHFA S PHFA > PHF PHFA < PHF PHFA > PHF S RHF > PHF RHF > PHF RHF < PHF S RHF < PHFA RHF > PHFA RHF < PHFA NS PHSA11 < PHS PHSA11 < PHS PHSA11 < PHS NS PHSA6 < PHS* PHSA6 < PHS PHSA6 < PHS NS RHS < PHS RHS < PHS RHS < PHS* NS PHSA6 < PHSA11 PHSA6 < PHSA11 PHSA6 < PHSA11 NS RHS < PHSA11 RHS > PHSA11 RHS < PHSA11 NS RHS > PHSA6 RHS > PHSA6 RHS > PHSA6 S PHSA11 > PHS PHSA11 > PHS PHSA11 > PHS S PHSA6 < PHS PHSA6 < PHS PHSA6 < PHS S RHS < PHS RHS < PHS RHS < PHS S PHSA6 < PHSA11 PHSA6 < PHSA11 PHSA6 < PHSA11 S RHS < PHSA11 RHS < PHSA11 RHS < PHSA11 S RHS > PHSA6 RHS > PHSA6 RHS > PHSA6

Table 3.6: The influence of diet on alpha diversity measurements of larval bacterial communities. Asterisks indicate the significance of comparisons p< 0.05 *.

Bray-Curtis Distances. Bray-Curtis distances displayed a significant separation between microbial communities associated with larvae raised on RHF and PHF diets for both sterilized and non-sterilized treatments (NS: p = 0.001; S: p = 0.001) (Fig.3.7A, B, Table 3.7). However, there was no separation between microbial communities in larvae raised on RHF and PHFA diets for either treatment (Fig.3.7A, B, Table 3.7). There was a distinct separation between microbial

101 communities of larvae raised on PHF and PHFA diets for both sterilized and non-sterilized treatments (NS: p = 0.004; S: p = 0.003) (Fig.3.7A, B, Table 3.7). These results indicate that environmental microbes associated with a natural peach diet have a significant impact on larval microbial communities regardless of the presence or absence of parental microbiota.

Microbes associated with larvae raised on RHS and PHS diets showed a distinct separation in microbial communities for both sterilized and non-sterilized treatments (NS: p =

0.002; S: p = 0.001) (Fig.3.7C, D, Table 3.7). For both treatments, larvae raised on a RHS diet did not show a significant separation of microbial communities when compared to a PHSA6 diet.

In addition, S larvae raised on a RHS diet did not show any significant difference in Bray-Curtis distances when compared to a PHSA11 diet, though it should be noted that NS larvae exhibited a potential marginal difference (p=0.05) (Fig.3.7C, D, Table 3.7). Larvae raised on a PHS diet showed a significant separation of microbial communities compared to a PHSA6 diet for both treatments (NS: p = 0.016; S = 0.009) (Fig.3.7C, D, Table 3.7). When larvae raised on a PHS diet were compared to those on a PHSA11 diet, NS larvae displayed a significant difference in microbial community composition (p = 0.028) (Fig.3.7C, Table 3.7). When larvae raised on both

PHSA diets were compared to each other, no significant difference in Bray-Curtis distances were observed (Fig.3.7C, D, Table 3.7). As indicated by the results, environmental microbes present in the high sugar variants of peach food played a key role in the formation of distinct microbial communities.

102

p value Comparison Type 0.001 PHFNS vs RHFNS HF 0.353 PHFANS vs RHFNS HF 0.004 PHFNS vs PHFANS HF 0.001 PHFS vs RHFS HF 0.118 PHFAS vs RHFS HF 0.003 PHFS vs PHFAS HF 0.002 PHSNS vs RHSNS HS 0.05 PHSA11NS vs RHSNS HS 0.171 PHSA6NS vs RHSNS HS 0.028 PHSNS vs PHSA11NS HS 0.016 PHSNS vs PHSA6NS HS 0.541 PHSA11NS vs PHSA6NS HS 0.001 PHSS vs RHSS HS 0.526 RHSS vs PHSA11S HS 0.514 PHSA6S vs RHSS HS 0.799 PHSS vs PHSA11S HS 0.009 PHSS vs PHSA6S HS 0.8 PHSA6S vs PHSA11S HS

Table 3.7: Influence of a diet on bacterial community composition measured with Bray-

Curtis distances. Bray-Curtis distances are calculated based on all ZOTUs identified in larval samples.

103

Figure 3.7: Larvae raised on peach and lab diets form distinct microbial communities even with high fat and high sugar nutritional modifications. A) Non-sterilized larvae raised on HF diets; B) Sterilized larvae raised on HF diets; C) Non-sterilized larvae raised on HS diets; D)

Sterilized larvae raised on HS diets. For this figure, Bray-Curtis distances were calculated based on the abundance of 100 dominant ZOTUs and visualized with a multidimensional scaling plot.

104

We further evaluated the differences between larvae raised on HF and HS diets using

Bray-Curtis distances calculated for each taxonomic level. For larvae raised in the presence of parental microbiota (NS), we observed the following results. At the phylum level, no comparisons were significant. At the class level, only larvae raised on a PHF diet were significantly different from those raised on an RHF diet (p=0.011) (S3.2). At the order level, larvae raised on a PHF diet were significantly different from those raised on both RHF (p=0.01) and PHFA (p=0.047) diets, while larvae raised on a PHS diet were significantly different from those raised on a RHS diet (p=0.035) (S3.2). At the family level, larvae raised on a PHF diet were significantly different from those raised on both RHF (p=0.007) and PHFA (p=0.026) diets, while larvae raised on a PHS diet were significantly different from those raised on both RHS

(p=0.037) and PHSA6 (p=0.036) diets (S3.2). At the genera level, larvae raised on a PHF diet were significantly different from those raised on both RHF (p=0.001) and PHFA (p=0.011) diets, while larvae raised on a PHS diet were significantly different from those raised on both RHS

(p=0.038) and PHSA6 (p=0.037) diets (S3.2).

For larvae raised in the absence of parental microbiota (S), we observed the following results. At the phylum level, larvae raised on a PHF diet were significantly different from those raised on a PHFA diet (p=0.019) (S3.2). At the class level, larvae raised on a PHF diet were significantly different from those raised on a PHFA diet (p=0.009), with no other comparisons being significant (S3.2). At the order level, larvae raised on a PHF diet were significantly different from those raised on both RHF (p=0.048) and PHFA (p=0.011) diets, while larvae raised on a PHS diet were significantly different from those raised on a RHS diet (p=0.012)

(S3.2). At the family level, larvae raised on a PHF diet were significantly different from those raised on both RHF (p=0.014) and PHFA (p=0.004) diets, while larvae raised on a PHS diet were

105 significantly different from those raised on an RHS diet (p=0.011) (S3.2). At the genera level, larvae raised on a PHF diet were significantly different from those raised on both RHF (p=0.015) and PHFA (p=0.002) diets, while larvae raised on a PHS diet were significantly different from those raised on both RHS (p=0.011) and PHSA6 (p=0.023) diets (S3.2).

Phylogenetic diversity with Weighed Unifrac Distances. We also evaluated if the samples raised on different diets are significantly different in the phylogenetic diversity of their gut bacteria. We observed that samples raised on PHF diets were significantly different from those raised on both RHF (NS: p=0.001, S: p=0.002) and PHFA (NS: p=0.004, S: p=0.001) diets

(Table 3.8, Fig. 3.8A, B). No other HF diet comparison was significantly different (Table 3.8,

Fig. 3.8A, B). Interestingly, if the parental microbiota was not removed, all HS diets were not significantly different from each other (Table 3.8, Fig. 3.8C, D). For Sterilized larvae, those raised on PHS and RHS diets were significantly different (p=0.01) (Table 3.8, Fig. 3.8C, D). p value Comparison Type 0.001 RHFNS vs PHFNS HF 0.27 RHFNS vs PHFANS HF 0.004 PHFNS vs PHFANS HF 0.002 RHFS vs PHFS HF 0.196 RHFS vs PHFAS HF 0.001 PHFS vs PHFAS HF 0.149 PHSNS vs RHSNS HS 0.243 PHSA11NS vs RHSNS HS 0.381 PHSA6NS vs RHSNS HS 0.162 PHSNS vs PHSA11NS HS 0.077 PHSNS vs PHSA6NS HS 0.372 PHSA11NS vs PHSA6NS HS 0.01 PHSS vs RHSS HS 0.636 RHSS vs PHSA11S HS 0.362 RHSS vs PHSA6S HS 1 PHSS vs PHSA11S HS 0.119 PHSS vs PHSA6S HS 1 PHSA6S vs PHSA11S HS

106

Table 3.8: Influence of a diet on phylogenetic diversity of larval bacterial community measured with Weighed Unifrac distances. Weighed Unifrac distances are calculated based on all ZOTUs identified in larval samples.

Figure 3.8: Larvae raised on peach and lab diets form phylogenetically distinct microbial communities even with high fat and high sugar nutritional modifications. A) Non-sterilized

107 larvae raised on HF diets; B) Sterilized larvae raised on HF diets; C) Non-sterilized larvae raised on HS diets; D) Sterilized larvae raised on HS diets. Weighed Unifrac distances were calculated based on all ZOTUs identified in samples and visualized with a multidimensional scaling plot.

Difference in microbial taxa abundances between diets. For non-sterilized larvae we found that the abundance of one phylum, two classes, 11 orders, 24 families, 72 genera, and 262

ZOTUs were significantly different between RHF and PHF foods (S3.3.01-06). Comparing larvae raised on RHF and PHFA diets, we found that the abundance of zero phyla, zero classes, two orders, four families, 11 genera, and 67 ZOTUs were significantly different (S3.3.01-06).

Lastly, comparing larvae raised on PHF and PHFA diets, we found that the abundance of one phylum, two classes, eight orders, 20 families, 51 genera, and 185 ZOTUs were significantly different (S3.3.01-06).

Larvae raised on HF diets in the absence of parental microbiota (S) produced the following results. The abundance of two phyla, four classes, seven orders, 12 families, 29 genera, and 136 ZOTUs were significantly different between RHF and PHF foods (S3.3.01-06).

Comparing larvae raised on RHF and PHFA diets, we found that the abundance of one phylum, two classes, one order, seven families, 10 genera, and 58 ZOTUs were significantly different

(S3.3.01-06). Between PHF and PHFA the abundance of one phylum, three classes, five orders, nine families, 34 genera, and 134 ZOTUs were significantly different (S3.3.01-06).

For non-sterilized larvae raised on HS diets we observed the following results. The abundance of five phyla, five classes, six orders, 15 families, 39 genera, and 155 ZOTUs were significantly different between RHS and PHS raised larvae (S3.3.01-06). Comparing larvae raised on RHS and PHSA11 diets, we found that the abundance of zero phyla, zero classes, one

108 order, five families, 12 genera, and 50 ZOTUs were significantly different (S3.3.01-06). One phylum, one class, one order, two families, 10 genera, and 36 ZOTUs were significantly different between RHS and PHSA6 food raised larvae (S3.3.01-06). Comparing larvae raised on

PHS and PHSA11 diets, we found that the abundance of one phylum, one class, six orders, 11 families, 27 genera, and 70 ZOTUs were significantly different (S3.3.01-06). The abundance of three phyla, four classes, nine orders, 16 families, 37 genera, and 113 ZOTUs were significantly different between PHS and PHSA6 (S3.3.01-06). Lastly, comparing larvae raised on PHSA11 and PHSA6 diets, we found that the abundance of zero phyla, zero classes, two orders, six families, 11 genera, and 32 ZOTU were significantly different (S3.3.01-06).

Larvae subjected to sterilization and raised on HS diet produced the following results.

The abundance of three phyla, two classes, six orders, 15 families, 29 genera, and 119 ZOTUs were significantly different between RHS and PHS (S3.3.01-06). Comparing larvae raised on

RHS and PHSA6 diets, we found that the abundance of zero phyla, zero classes, one order, one family, three genera, and 30 ZOTUs were significantly different between RHS and PHSA6 food

(S3.3.01-06). The abundance of one phylum, two classes, six orders, nine families, 18 genera, and 65 ZOTUs were significantly different between PHS and PHSA6 (S3.3.01-06).

2.2 Alpha diversity. Evaluating the effect that parental microbiota produced on the alpha diversities of larvae symbiotic bacterial community, we observed that sterilized larvae raised on a RHS diet exhibited significantly higher Shannon’s index, Simpson’s index, and species richness values compared with non-sterilized larvae raised on the same diet (p=0.0079, p=0.0120, and p=0.0101, respectively) (Table 3.9). All other comparisons were not significant

(Table 3.9).

109

Comparison Shannon index Simpson index Richness NS vs S PHFNS < PHFS PHFNS < PHFS PHFNS > PHFS NS vs S PHFANS > PHFAS PHFANS > PHFAS PHFANS < PHFAS NS vs S RHFNS > RHFS RHFNS > RHFS RHFNS > RHFS NS vs S PHSNS < PHSS PHSNS < PHSS PHSNS < PHSS NS vs S PHSA11NS < PHSA11S PHSA11NS < PHSA11S PHSA11NS < PHSA11S NS vs S PHSA6NS < PHSA6S PHSA6NS < PHSA6S PHSA6NS < PHSA6S NS vs S RHSNS < RHSS ** RHSNS < RHSS * RHSNS < RHSS *

Table 3.9: Influence of treatment on alpha diversity measurements of larval bacterial communities. Asterisks indicate the significance of comparisons p< 0.01 **, and p< 0.05 *.

Bray-Curtis Distances. Evaluating the impact of parental microbiota on the formation of microbial communities and accounting for all identified ZOTUs on each of the HF diets, we did not observe a significant difference in bacterial composition, between any diets (Table 3.10).

Comparing NS and S treatments on HS diets, we observed a significant difference in bacterial composition RHS p= 0.001, PHS p= 0.016, and PHSA6 p= 0.049 (PHSA11 raised larvae were not included due to insufficient sequencing results) (Table 3.10, Figure 3.9). p value Comparison Type 0.001 RHSNS vs RHSS HS 0.016 PHSNS vs PHSS HS 0.437 PHSA11NS vs PHSA11S HS 0.049 PHSA6NS vs PHSA6S HS 0.341 RHFNS vs RHFS HF 0.056 PHFNS vs PHFS HF 0.476 PHFANS vs PHFAS HF

Table 3.10: Influence of treatment on bacterial community composition measured with

Bray-Curtis distances. Bray-Curtis distances are calculated based on all ZOTUs identified in larval samples.

110

Figure 3.9: Parental microbiota significantly influences the formation of bacterial community of the larvae, only on the diets subjected to HS modification. A) Larvae raised on a RHS diet; B) Larvae raised on a PHS diet; C) Larvae raised on a PHSA diet. Bray-Curtis distances were calculated based on the abundance of all identified ZOTUs and visualized with a multidimensional scaling plot.

Evaluating the difference in Bray-Curtis distances between NS and S larvae raised on each diet at each identified taxonomic level, we observed the following results. At the phylum level, we observed a significant difference between NS and S larvae only on a RHS diet (S3.4).

At the class level, we found a significant difference between NS and S larvae on both PHF

(p=0.031) and RHS (p=0.002) diets (S3.4). At the order level, we observed a significant difference between NS and S larvae on both PHF (p=0.028) and RHS (p=0.002) diets (S3.4).

These same diets exhibited significant differences at the family level (PHF: p=0.017, RHS: p=0.001) (S3.4). At the genera level, we found significant differences on PHF (p=0.036), RHS

(p=0.001), PHS (p=0.037), and PHSA6 (p=0.039) diets (S3.4). Weighed Unifrac Distances.

Evaluating the difference between NS and S larvae on each diet, we observed a significant

111 difference between NS and S larvae on both PHF (p=0.007) and RHS (p=0.002) diets (Table

3.11). p value Comparison Type 0.002 RHSNS vs RHSS HS 0.181 PHSNS vs PHSS HS 0.579 PHSA11NS vs PHSA11S HS 0.172 PHSA6NS vs PHSA6S HS 0.847 RHFNS vs RHFS HF 0.007 PHFNS vs PHFS HF 0.504 PHFANS vs PHFAS HF 0.172 PHSA6NS vs PHSA6S HF

Table 3.11: Influence of treatment on phylogenetic diversity of larval bacterial community measured with Weighted Unifrac distances. Weighed Unifrac distances are calculated based on all ZOTUs identified in larval samples.

Abundance of microbial taxa. Comparing the microbial abundances at each taxonomic level between sterilized and non-sterilized larvae, raised on each diet produced the following results. On the RHF diet, we found that the abundance of one phylum, four classes, two orders, four families, 14 genera, and 87 ZOTUs were significantly different (S3.5.01-06). On the PHF diet, we found that the abundance of zero phyla, two classes, three orders, six families, 16 genera, and 77 ZOTUs were significantly different (S3.5.01-06). On the PHFA diet, we found that the abundance of zero phyla, two classes, one order, three families, seven genera, and 69

ZOTUs were significantly different between the sterilized and non-sterilized larvae (S3.5.01-06).

On the HS diets, the results were the following. On the RHS diet, we found that the abundance 11 phyla, 18 classes, 31 orders, 43 families, 71 genera, and 248 ZOTUs were significantly different between sterilized and non-sterilized treatments (S3.5.01-06). On the PHS

112 diet, we found that the abundance of four phyla, six classes, 14 orders, 16 families, 32 genera and

124 ZOTUs were significantly different (S3.5.01-06). Lastly, the abundances of zero phyla, zero classes, three orders, five families, 15 genera, and 61 ZOTUs were significantly different between the sterilized and non-sterilized treatments on the PHSA6 diet (S3.5.01-06).

2.3 Nutritional modifications influence larval microbial communities

Influence of nutritional modifications on alpha diversity. Comparing alpha diversities between larvae raised on normal and modified diets, we observed the following results. In a presence of parental microbiota, larvae raised on a RHF diet were significantly higher in

Shannon’s index and species richness measurements compared to larvae raised on a R diet

(p=0.0325 and p=0.0392, respectively) (Table 3.12). Larvae raised on a PHFA diet had significantly higher values for Shannon’s index and species richness than larvae raised on a PA diet, but only if parental microbiota were present (p=0.0382 and p=0.0165) (Table 3.12). Lastly,

NS larvae raised on a R diet had significantly higher Shannon’s diversity index value compared to larvae raised on a RHS diet (p=0.0156) (Table 3.12). All other comparisons were shown to be insignificant (Table 3.12).

113

Treatment Shannon index Simpson index Richness NS PHFA > PA * PHFA > PA PHFA > PA * NS PHSA11 < PA PHSA11 < PA PHSA11 < PA NS PHSA6 < PA PHSA6 < PA PHSA6 < PA NS PR < PHF PR < PHF PR < PHF NS PR < PHS PR < PHS PR < PHS NS RHF > R * RHF > R RHF > R * NS RHS < R * RHS < R * RHS < R S PHFA > PA PHFA < PA PHFA > PA S PHSA11 > PA PHSA11 > PA PHSA11 < PA S PHSA6 < PA PHSA6 < PA PHSA6 < PA S PR < PHF PR < PHF PR < PHF S PR < PHS11 PR < PHS PR < PHS S RHF < R RHF < R RHF < R S RHS < R RHS < R RHS < R

Table 3.12: Influence of nutritional modifications on alpha diversity measurements of larval bacterial communities. Asterisks indicate the significance of comparisons p< 0.05 *.

Influence of nutritional modifications on Bray-Curtis Distances. Comparing the microbial communities of normal and HF diets, we found that larvae raised on R and RHF diets exhibited a significant difference in their bacterial community composition for both treatments

(NS: p= 0.004, S: p= 0.009) (Table 3.13, Fig. 3.10A, D). Interestingly, the PR diet was significantly different from a PHF diet only if the parental microbiota were removed (S: p=

0.015) (Table 3.13, Fig. 3.10B, E). Larvae raised on PA and PHFA diets exhibited significantly different microbial communities for both treatments (NS: p=0.013, S: p= 0.02) (Table 3.13, Fig.

3.10C, F). Comparing the microbial communities of normal and HS diets, we found that larvae raised on R and RHS diets were significantly different only if the parental microbiota were removed (S: p= 0.001) (Table 3.13, Fig. 3.10A, D). The same pattern was observed when comparing PR and PHS diets (S: p= 0.001) (Table 3.13, Fig. 3.10B, E). On the opposite, microbial communities of the larvae raised on a PA diet were significantly different than those

114 raised on both PHSA6 and PHSA11diets, only in the presence of parental microbiota (NS: p=

0.037 and p= 0.007, respectively) (Table 3.13, Fig. 3.10C, F).

p value Comparison Type 0.004 RNS vs RHFNS N vs HF 0.189 PRNS vs PHFNS N vs HF 0.013 PANS vs PHFANS N vs HF 0.009 RS vs RHFS N vs HF 0.015 PRS vs PHFS N vs HF 0.02 PAS vs PHFAS N vs HF 0.994 RNS vs RHSNS N vs HS 0.257 PRNS vs PHSNS N vs HS 0.037 PANS vs PHSA6NS N vs HS 0.007 PANS vs PHSA11NS N vs HS 0.001 RS vs RHSS N vs HS 0.001 PRS vs PHSS N vs HS 0.069 PAS vs PHSA6S N vs HS 0.49 PAS vs PHSA11S N vs HS

Table 3.13: Influence of nutritional modifications on bacterial community composition measured with Bray-Curtis distances. Bray-Curtis distances are calculated based on all

ZOTUs identified in larval samples.

115

Figure 3.10: The response of microbial community to dietary modifications varied with the origin of a diet and presence of parental microbiota. A) Non-sterilized larvae raised on a R diet formed a distinct microbial community compared with larvae raised on a RHF but not a

RHS diet; B) Non-sterilized larvae raised on a PR diet did not form a distinct microbial community compared with larvae raised on PHF and PHS diets; C) Non-sterilized larvae raised on a PA diet formed a distinct microbial community compared with larvae raised on PHFA and

PHSA diets; D) Sterilized larvae raised on a R diet formed a distinct microbial community compared with larvae raised on RHF and RHS diets; E) Sterilized larvae raised on a PR diet formed a distinct microbial community compared with larvae raised on PHF and PHS diets; F)

Sterilized larvae raised on a PA diet formed a distinct microbial community compared with larvae raised on PHFA but not PHSA diets;

116

Evaluating the differences in Bray-Curtis distances between larvae raised on normal and high fat/high sugar diets at each identified taxonomic level, we observed the following results. At phylum and class level, there were no significant difference between all the comparisons. At the order level for NS larvae, we observed a significant difference between R and RHF diets

(p=0.045) and PA and PHSA6 diets (p=0.026) (S3.6). For S larvae, we observed a significant difference between PR and PHF diets (p=0.04). At the family level, NS larvae were significantly different between R and RHF diets (p=0.043), PA and PHSA6 (p=0.022), and PHSA11diets

(p=0.042) (S3.6). S larvae were significantly different between PR and PHF (p=0.043), as well as PHS (p=0.034) diets (S3.6). Additionally, larvae raised on the PA diet were significantly different from those raised on a PHSA6 diet (p=0.034) (S3.6). At the genera level, NS larvae were significantly different between R and RHF diets (p=0.049) diets (S3.6). Larvae raised on a

PA diet were significantly different from those raised on PHFA (p=0.028), PHSA6 (p=0.012), and PHSA11 (p=0.015) diets (S3.6). S larvae raised on a PR diet were significantly different from larvae raised on both PHF (p=0.027) and PHS (p=0.008) diets (S3.6). Larvae raised on a

PA diet were significantly different from those raised on both PHFA (p=0.023) and PHSA6

(p=0.026) diets. R food raised larvae were significantly different from those raised on an RHS diet (p=0.016) (S3.6).

Weighted Unifrac. Comparing the Weighted Unifrac distances between normal and modified diets, we observed that larvae raised on an R diet were significantly different from those raised on a RHS diet only in the absence of parental microbiota (S) (p=0.027) and had no significant difference with larvae raised on a RHF diet (Table 3.14). The microbial communities of larvae raised on PR diets were significantly different from those raised on PHF (p=0.006) and

PHS (p=0.017) diets only in the absence of parental microbiota (S) (Table 3.14). NS larvae

117 raised on PA diet had a significantly different symbiotic community from those raised on both

PHSA6 (p=0.001) and PHSA11 (p=0.003) diets (Table 3.14). S larvae raised on PA diet were significantly different from those raised on both PHFA (p=0.008) and PHSA6 (p=0.029) diets

(Table 3.14). p value Comparison Type 0.059 RNS vs RHFNS N vs HF 0.422 PRNS vs PHFNS N vs HF 0.07 PANS vs PHFANS N vs HF 0.094 RS vs RHFS N vs HF 0.006 PRS vs PHFS N vs HF 0.008 PAS vs PHFAS N vs HF 0.14 RNS vs RHSNS N vs HS 0.12 PRNS vs PHSNS N vs HS 0.001 PANS vs PHSA6NS N vs HS 0.003 PANS vs PHSA11NS N vs HS 0.027 RS vs RHSS N vs HS 0.017 PRS vs PHSS N vs HS 0.029 PAS vs PHSA6S N vs HS 0.144 PAS vs PHSA11S N vs HS

Table 3.14: Influence of nutritional modifications on phylogenetic diversity of larval bacterial community measured with Weighted Unifrac distances. Weighted Unifrac distances are calculated based on all ZOTUs identified in larval samples.

Abundance of the microbial taxa. Comparing the abundance of microbial taxa of larvae raised on both normal and high fat diets that had inherited parental microbiota (NS), we found following results. Larvae raised on R and RHF diets exhibited a significant difference in the abundance of seven phyla, 15 classes, 24 orders, 43 families, 94 genera, and 703 ZOTUs (S3.7).

Comparing larvae raised on PR and PHF diets, we found a significant difference in abundance of one phylum, six classes, 14 orders, 22 families, 45 genera, and 454 ZOTUs (S3.7). Lastly, four

118 phyla, six classes, 12 orders, 20 families, 65 genera, and 292 ZOTUs exhibited significant difference in abundance between larvae raised on PA and PHFA diets (S3.7).

Evaluating sterilized larvae, we observed the following results in the microbial abundances. Larvae raised on R and RHF diets exhibited a significant difference in the abundance of three phyla, six classes, 11 orders, 14 families, 57 genera, and 204 ZOTUs (S3.7).

Comparing larvae raised on PR and PHF diets, we found a significant difference in abundance of zero phyla, two classes, nine orders, 15 families, 57 genera, and 170 ZOTUs (S3.7). Lastly, PA and PHFA raised larvae were significantly different in abundances of two phyla, three classes, nine orders, 13 families, 39 genera, and 243 ZOTUs (S3.7).

Comparing the microbial abundances of larvae raised on both normal and high sugar diets that had inherited parental microbiota (NS), we observed the following differences. Larvae raised on R and RHS diets exhibited a significant difference in abundances of eight phyla, nine classes, 23 orders, 36 families, 62 genera, and 226 ZOTUs (S3.7). PR and PHS food raised larvae were significantly different in abundances of two phyla, three classes, 13 orders, 17 families, 36 genera, and 359 ZOTUs (S3.7). Larvae raised on PA diet were significantly different in the abundance of five phyla, five classes, 14 orders, 26 families, 52 genera, and 182 ZOTUs when compared to those raised on a PHSA6 diet, and in the abundances of three phyla, three classes, 17 orders, 33 families, 59 genera, and 194 ZOTUs when compared to larvae raised on a

PHSA11 diet (S3.7).

Evaluating sterilized larvae for the same diets, we observed the following results. Larvae raised on R and RHS diets exhibited a significant difference in the abundance of five phyla, six classes, 20 orders, 27 families, 81 genera, and 227 ZOTUs (S3.7). Comparing larvae raised on

119

PR and PHS diets, we found a significant difference in abundance of four phyla, eight classes, 23 orders, 35 families, 87 genera, and 257 ZOTUs (S3.7). Lastly, comparing larvae raised on PA and PHSA6 diets, we found a significant difference in abundance of one phylum, one class, five orders, 16 families, 33 genera, and 112 ZOTUs (S3.7).

Inter-microbial interactions. Evaluating the correlations between abundances of microbial taxa found in larvae raised on HF diets, we observed the following results. For the NS larvae on the RHF diet, we observed 45 significant correlations between microbial abundances at the phylum level, 138 at class level, 865 at the order level, 2,228 at the family level, 13,472 at the genera level, and 1,220,194 at the ZOTUs level (S3.8). For S larvae raised on the same diet, we observed 32 significant interactions at the phylum level, 129 at class level, 549 at the order level, 1,710 at the family level, 12,321 at the genera level, and 712,811 at the ZOTUs (S3.8). On the PHF diet there were 36 significant correlations at the phylum level, 87 at the class level, 466 at the order level, 1,328 at the family level, 10,808 at the genera level, and 1,156,977 at the

ZOTUs level for NS larvae (S3.8). For S larvae raised at the same diet we observed 43 significant interactions at the phylum level, 135 at the class level, 551 at the order level, 1,412 at the family level, 9,228 at the genera level, and 441,197 at the ZOTUs level (S3.8). NS larvae raised on PHFA diet exhibited 87 significant interaction at the phylum level, 207 at the class level, 936 at the order level, 2,657 at the family level, 17,735 at the genera level, and 1,181,711 at the ZOTUs level (S3.8). S larvae raised on PHFA displayed 79 significant interactions at the phylum level, 232 at the class level, 1,058 at the order level, 2,869 at the family level, 21,627 at the genera level, and 1,608,687 at the ZOTUs level (S3.8).

Considering larvae raised on HS diets, we observed the following results. On RHS diet, we found 54 significant interactions between microbial abundances of NS larvae at the phylum

120 level, 104 at the class level, 411 at the order level, 1,074 at the family level, 6,611 at the genera level, and 186,261 at the ZOTUs level (S3.8). S larvae on the same diet showed 47 significant interactions at the phylum level, 106 at the class level, 448 at the order level, 1,312 at the family level, 10,391 at the genera level, and 529,204 at the ZOTUs level (S3.8). NS larvae raised on

PHS diet exhibited 44 significant interactions at the phylum level, 115 at the class level, 397 at the order level, 1,202 at the family level, 8,748 at the genera level, and 1,207,912 at the ZOTUs level (S3.8); while S larvae exhibited 41 significant interactions at the phylum level, 108 at the class level, 430 at the order level, 1,238 at the family level, 9,384 at the genera level, and

486,978 at the ZOTUs level (S3.8). For NS larvae raised on PHSA6 diet, we observed 33 significant interactions at the phylum level, 91 at the class level, 509 at the order level, 1,498 at the family level, 7,485 at the genera level, and 143,642 at the ZOTUs level (S3.8). S larvae raised on the same diet exhibited 41 significant interactions at the phylum level, 79 at the class level, 437 at the order level, 1,539 at the family level, 8,708 at the genera level, and 164,649 at the ZOTUs level (S3.9). NS larvae raised on PHSA11 diet exhibited 30 significant interactions at the phylum level, 92 at the class level, 348 at the order level, 880 at the family level, 5,501 at the genera level, and 273,872 at the ZOTUs level (S3.8).

2.4 Dominant bacteria taxa are associated with diets and nutritional modifications

Dominant bacterial taxa associated with diets. We applied a discriminant analysis to evaluate which of the dominant bacteria taxa would be the most influential for defining each of the tested diets. At the phylum level, NS larvae raised on a RHF food were defined by

Epsilonbacteraeota (Fig. 3.11A). Larvae raised on a PHF diet were discriminated by the abundance of Cyanobacteria, and those raised on a PHFA diet by the abundance of

Planctomycetes (Fig. 3.11A). For S larvae, we observed the same results for those raised on PHF

121 and PHFA diets (Fig. 3.11B). Both Epsilonbacteraeota and Verucomicrobia produced a strong effect for discrimination of larvae raised on an RHF diet (Fig. 3.11B). The microbial community of NS larvae raised on a RHS diet was defined by Fusobacteria, and on a PHS diet by

Cyanobacteria (Fig. 3.11C). Microbial communities of PSHA6 and PHSA11 diets were similar, and both were defined by Tenericutes and Verrucomicrobia (Fig. 3.11C). S larvae raised on a

RHS food were primarily defined by Epsilonbacteraeota, on a PHS diet by Tenericutes and

Proteobacteria, and on a PHSA6 diet by Verrucomicrobia (Fig. 3.11D).

122

Figure 3.11: With any of the nutritional modifications, microbial communities of larvae raised on non-autoclaved peach diets can be differentiated by the abundance of dominant bacteria phyla. Linear discriminant analysis of 10 dominant bacterial taxa. A) Non-sterilized larvae raised on high fat diets; B) Sterilized larvae raised on high fat diets; C) Non-sterilized larvae raised on high sugar diets; D) Sterilized larvae raised on high sugar diets. The intensity of vector rays’ color corresponds to their length. Confidence ellipses are filled based on the color of the diet. Normal data ellipses are unfilled and leveled to include 50% of the samples.

At the class level for NS larvae, Erysipelotrichia defined a RHF diet, Oxyphotobacteria defined a PHF food, and Bacteroidia defined a PHFA diet (Fig. 3.12A,). RHS, PHSA6, and

PHSA11 diets were not clearly separated from each other, but as a cluster were differentiated by

Verrucomicrobiae and Actinobacteria (Fig. 3.12B). Larvae raised on a PHS diet were defined by

Oxyphotobacteria (Fig. 3.12B). For S larvae, Verrucomicrobiae differentiated a RHF diet,

Oxyphotobacteria differentiated a PHF diet, and Actinobacteria differentiated a PHFA diet (Fig.

3.12C). RHS diet was differentiated by Negativicutes, PHS diet was defined by

Oxyphotobacteria, and PHSA6 diet was separated by Verrucomicrobiae (Fig. 3.12D).

123

HS, NS

HF, S

Figure 3.12: With any of the nutritional modifications, microbial communities of larvae raised on non-autoclaved peach diets can be differentiated by the abundance of dominant bacteria classes. Linear discriminant analysis of 10 dominant bacterial taxa. A) Non-sterilized larvae raised on high fat diets; B) Sterilized larvae raised on high fat diets; C) Non-sterilized larvae raised on high sugar diets; D) Sterilized larvae raised on high sugar diets. The intensity of vector rays’ color corresponds to their length. Confidence ellipses are filled based on the color of the diet. Normal data ellipses are unfilled and leveled to include 50% of the samples.

124

For NS larvae at the order level, the RHF diet was defined by Erysipelotrichales, while the PHF diet was defined by Acetobacterales, Betaproteobacteriales, and Rickettsiales (Fig.

3.13A). The PHFA diet was discriminated by Clostridiales (Fig. 3.13A). RHS, PHSA6, and

PHSA11 diets were not clearly separated from each other, but as a cluster were defined by

Pseudomanades, Erysipelotrichales, and Corynebacteriales (Fig. 3.13B). The PHS diet was defined by Rickettsiales, Lactobacillales, and Enterobacteriales (Fig. 3.13B). S larvae raised on a

RHF diet were defined by Lactobacillales (Fig. 3.13C). PHF raised larvae was discriminated by

Rickettsiales and Erysipelotrichales (Fig. 3.13C). PHFA diet was separated by Clostridiales and

Betaproteobacteriales (Fig. 3.13C). RHS was separated by Pseudomonadales. Enterobacteriales and Acetobacterales defined PHS diet, and Lactobacillales differentiated PHSA6 food (Fig.

3.13D).

125

HS,HS, NS NS

HHF,F, SS

Figure 3.13: With any of the nutritional modifications, microbial communities of larvae raised on non-autoclaved peach diets can be differentiated by the abundance of dominant bacteria orders. Linear discriminant analysis of 10 dominant bacterial taxa. A) Non-sterilized larvae raised on high fat diets; B) Sterilized larvae raised on high fat diets; C) Non-sterilized larvae raised on high sugar diets; D) Sterilized larvae raised on high sugar diets. The intensity of

126 vector rays’ color corresponds to their length. Confidence ellipses are filled based on the color of the diet. Normal data ellipses are unfilled and leveled to include 50% of the samples.

At the family level, NS larvae raised on RHF and PHFA diets were not clearly separated from each other, with Bacteroidaceae and Lachnospiraceae mostly defining a PHFA diet, and

Corynebacteriaceae mostly separating RHF food (Fig. 3.14A). The PHF diet was differentiated by Acetobacteraceae and Enterobacteriaceae (Fig. 3.14A). RHS, PHSA6, and PHSA11 diets were not clearly separated from each other but were differentiated by Burkholderiaceae,

Corynebacteriaceae, and Bacteroidaceae (Fig. 3.14B). PHS food was defined by

Enterobacteriaceae and Muribaculaceae (Fig. 3.14B). S larvae raised on RHF were discriminated by Lactobacillaceae and Lachnospiraceae. PHF food was separated by

Erysipelotrichaceae and Acetobacteraceae (Fig. 3.14C). PHFA diet was defined by

Bacteroidaceae (Fig. 3.14C). RHS was discriminated by Burkholderiaceae. PHS was separated by Enterobacteriaceae and Acetobacteraceae (Fig. 3.14D). PHSA6 was differentiated by

Lactobacillaceae and Muribaculaceae (Fig. 3.14D).

127

HS, NS

HF, S

Figure 3.14: With any of the nutritional modifications, microbial communities of larvae raised on non-autoclaved peach diets can be differentiated by the abundance of dominant bacteria families. Linear discriminant analysis of 10 dominant bacterial taxa. A) Non-sterilized larvae raised on high fat diets; B) Sterilized larvae raised on high fat diets; C) Non-sterilized larvae raised on high sugar diets; D) Sterilized larvae raised on high sugar diets. The intensity of

128 vector rays’ color corresponds to their length. Confidence ellipses are filled based on the color of the diet. Normal data ellipses are unfilled and leveled to include 50% of the samples.

At the genera level RHF and PHF food raised NS larvae exhibited distinct microbial communities (Fig. 3.15A). However, PHFA raised larvae was not clearly separated from RHF and PHF diets (Fig. 3.15A). Acetobacter primarily differentiated PHF diet (Fig. 3.15A).

Streptococcus and Lactobacillus differentiated PHFA diet, and Blautia separated RHF diet (Fig.

3.15A). RHS, PHSA6, and PHSA11 were not clearly separated from each other (Fig. 3.15B).

Acetobacter was influential in separating PHSA6 diet and Pseudomonas and Alistipes influenced separation of RHS and PHSA11 diets (Fig. 3.15B). Gluconobacter and Lachnospiraceae separated PHS raised larvae (Fig. 3.15B). S larvae raised on RHF diet was separated by

Acetobacter and Streptococcus (Fig. 3.15C). RHF was separated by Lactobacillus (Fig. 3.15C).

PHFA was differentiated by Gluconobacter and Corynebacterium (Fig. 3.15C). RHS and

PHSA6 diets were not fully separated from each other and were differentiated by Blautia,

Lactobacillus, and Acetobacter (Fig. 3.15D). Gluconobacter differentiated PHS diet (Fig.

3.15D).

129

HS, NS

HF, S

Figure 3.15: With any of the nutritional modifications, microbial communities of larvae raised on non-autoclaved peach diets can be differentiated by the abundance of dominant bacteria genera. Linear discriminant analysis of 10 dominant bacterial taxa. A) Non-sterilized larvae raised on high fat diets; B) Sterilized larvae raised on high fat diets; C) Non-sterilized larvae raised on high sugar diets; D) Sterilized larvae raised on high sugar diets. The intensity of

130 vector rays’ color corresponds to their length. Confidence ellipses are filled based on the color of the diet. Normal data ellipses are unfilled and leveled to include 50% of the samples.

Dominant bacterial taxa defining diet modifications. With the discriminant analysis, we also evaluated which of the dominant bacteria taxa influenced the differentiation of the normal diet and its HF and HS modifications. At the phylum level, for NS larvae, RHF and RHS diets were separated by the highest distance with R diet placed between them (Fig. 3.16A). The

RHF diet was discriminated by Cyanobacteria and Actinobacteria (Fig. 3.16A). The RHS diet was differentiated by Firmicutes (Fig. 3.16A). A similar pattern was observed for peach diets

(Fig. 3.16B). PHF was differentiated by Cyanobacteria and Actinobacteria. Firmicutes discriminated the PHS diet (Fig. 3.16B). Among the autoclaved diets, PA was defined by

Epsilonbacteraeota and Verrucomicrobia (Fig. 3.16C). The PHFA diet was defined by

Fusobacteria, Bacteroidetes, and Actinobacteria (Fig. 3.16C). PHSA6 and PHSA11 were separated by Planctomycetes (Fig. 3.16C).

For the S treatment, larvae raised on either a RHF or RHS diet were closer to each other than to the R diet (Fig. 85). The R food was defined by Cyanobacteria and Verrucomicrobia (Fig.

3.16D). RHF was defined by Actinobacteria while RHS was differentiated by Planctomycetes and Tenericutes (Fig. 3.16D). For the peach diets, PR and PHF were closer to each other than to

PHS (Fig. 3.16E). PR was defined by Verrucomicrobia, PHF was primarily separated by

Actinobacteria, and PHS was defined by Planctomycetes and Tenericutes (Fig. 3.16E). For the autoclaved diets, PHFA and PHSA6 were closer to each other than to PA, with PHFA defined by

Tenericutes and PHSA6 separated by Planctomycetes (Fig. 3.16F). PA diet was defined by

Fusobacteria and Verrucomicrobia (Fig. 3.16F).

131

Figure 3.16: Nutritional modifications cause shifts in bacterial composition and are associated with changes in abundance of the dominant bacteria phyla. Linear discriminant analysis of 10 dominant bacterial taxa. A) Non-sterilized larvae raised on lab-based diets; B)

Non-sterilized larvae raised on peach-based diets; C) Non-sterilized larvae raised on autoclaved peach-based diets; D) Non-sterilized larvae raised on lab-based diets; E) Non-sterilized larvae raised on peach-based diets; F) Non-sterilized larvae raised on autoclaved peach-based diets. The intensity of vector rays’ color corresponds to their length. Confidence ellipses are filled based on the color of the diet. Normal data ellipses are unfilled and leveled to include 50% of the samples.

At the class level, NS larvae raised on R and RHS diets were closer to each other than those raised on a RHF diet (Fig. 3.17A). The RHS diet was differentiated by Bacilli, the R diet

132 was separated by Alphaproteobacteria, and RHF food was defined by Oxyphotobacteria (Fig.

3.17A). PR and PHS diets were closer to each other than to PHF with PR separated by

Negativicutes and PHS by Bacteroidia (Fig. 3.17B). PHF food was primarily differentiated by

Oxyphotobacteria (Fig. 3.17B). The PHFA diet was differentiated by Oxyphotobacteria (Fig.

3.17C). The PA diet was defined by Verrucomicrobiae (Fig. 3.17C). PHSA6 and PHSA11 were differentiated by Bacilli and Alphaproteobacteria (Fig. 3.17C).

Sterilized larvae raised on RHF and RHS diet were closer to each other than to the R food, with RHS was differentiated by Negativiutes and RHF was differentiated by

Actinobacteria and Erysipelotrichia (Fig. 3.17D). The R diet was separated by Verrucomicrobiae

(Fig. 3.17D). PR and PHF dies were closer to each other than to PHS diet, with PR separated by

Negativicutes, PHF differentiated by Erysipelotrichia and Actinobacteria, and PHS diet was defined by Bacilli and Gammaproteobacteria (Fig. 3.17E). PHFA and PHSA were close to each other, with PHFA differentiated by Actinobacteria and Erysipelotrichia while PHSA6 was differentiated by Bacilli (Fig. 3.17F). PA was separated by Oxyphotobacteria and Negativicutes

(Fig. 3.17F).

133

Figure 3.17: Nutritional modifications cause shifts in bacterial composition and are associated with changes in abundance of the dominant bacteria classes. Linear discriminant analysis of 10 dominant bacterial taxa. A) Non-sterilized larvae raised on lab-based diets; B)

At the order level, NS larvae raised on RHS were differentiated by Enterobacteriales, and

RHF was defined by Corynebacteriales (Fig. 3.18A). The R food community resembled the

134 intermediate state between RHS and RHF diets (Fig. 3.18A). PHS was defined by Rickettsiales, and PHF was separated by Corynebacteriales and Betaproteobacteriales (Fig. 3.18B). Microbial communities of larvae raised on PR food resembled an intermediate state between PHF and PHS communities and were influenced by Pseudomonadales (Fig. 3.18B). PHSA6 and PHSA11 diets were differentiated by Enterobacteriales and Lactobacilalles (Fig. 3.18C). PA and PHFA diets were similar to each other, with PHFA distinguished by Corynebacteriales while PA was separated by Acetobacterales (Fig. 3.18C).

S larvae microbial communities raised on RHF and RHS were not fully separated from each other with Enterobacteriales defining RHS and Corynebacteriales and Eysipelotrichales discriminating the RHF diet (Fig. 3.18D). R food was defined by Rickettsiales (Fig. 3.18D). PR and PHF communities were closer to each other than to PHS diet (Fig. 3.18E). PR food was differentiated by Betaproteobacteriales and Acetobacterales; PHF was defined by

Corynebacteriales, and the PHS diet was separated by Rickettsiales (Fig. 3.18E). PA and PHFA diets were closer to each other and defined by Acetobacterales and Erysipelotrichales, respectively (Fig. 3.18F). The PHSA6 diet was separated by Enterobacteriales (Fig. 3.18F).

135

Figure 3.18: Nutritional modifications cause shifts in bacterial composition and are associated with changes in abundance of the dominant bacteria orders. Linear discriminant analysis of 10 dominant bacterial taxa. A) Non-sterilized larvae raised on lab-based diets; B)

At the family level, NS larvae raised on RHS were differentiated by Enterobacteriaceae and Lactobacillaceae (Fig. 3.19A). RHF raised larvae were discriminated by Bacteroidaceae and

136

Corynebacteriaceae (Fig. 3.19A). R food larvae communities were intermediate between RHF and RHS and primarily influenced by Acetobacteraceae (Fig. 3.19A). PHF and PHS diets were the most separated from each other (Fig. 3.19B). PHS diet was differentiated by

Enterobcteriaceae and Lachnospiraceae (Fig. 3.19B). PHF food was separated by

Erysipelotrichaceae and Corynebacteriaceae (Fig. 3.19B). PR food community was influenced by Ruminococcaceae (Fig. 3.19B). PHFA diet was differentiated by Ruminococcaceae and

Corynebacteriaceae (Fig. 3.19C). PA microbial community was differentiated by

Lachnospiraceae (Fig. 3.19C). PHSA6 diet was not clearly separated from PHSA11 and PA communities but was mostly influenced by Lactobacillacelae, while PHSA11 was discriminated by Erysipelotrichaceae (Fig. 3.19C).

S larvae raised on R diets were primarily defined by Ruminococcaceae and

Burkholderiaceae (Fig. 3.19D). RHF food was differentiated by Erysipelotrichaceae and

Muribaculaceae (Fig. 3.19D). RHS was separated by Enterobactericeae and Lachnospiraceae

(Fig. 3.19D). The PR diet was differentiated by Lachnospiraceae, Burkholderiaceae, and

Acetobacteraceae (Fig. 3.19E). PHF was separated by Muribaculaceae and Corynebacteriaceae

(Fig. 3.19E). PHS was defined by Ruminococcaceae, Enterobacteriaceae, and Lactobacillaceae

(Fig. 3.19E). For the autoclaved diets, PA was defined by Acetobacteraceae, Muribaculaceae, and Burkholderiaceae (Fig. 3.19F). PHFA was defined by Corynebacteriacea and

Erysipelotrichaceae (Fig. 3.19F). PHSA6 was separated by Enterobacteriaceae and

Lactobacillaceae (Fig. 3.19F).

137

Figure 3.19: Nutritional modifications cause shifts in bacterial composition and are associated with changes in abundance of the dominant bacteria families. Linear discriminant analysis of 10 dominant bacterial taxa. A) Non-sterilized larvae raised on lab-based diets; B)

At the genera level, NS larvae raised on RHF diet were defined by the Lachnospiraceae

NK4A136 group (Fig. 3.20A). RHS diet was defined by Alsitipes and Blautia (Fig. 3.20A). R

138 food samples had a microbial community that resembled the intermediate state between the RHF and RHS diets and were influenced by Pseudomonas (Fig. 3.20A). Comparing peach diets, we observed that Pseudomonas and Gluconobacter were associated with PR diet, Lachnospiraceae

NK4A136 and Alistipes were associated with PHF diet, and Streptococus was associated with the

PHS diet (Fig. 3.20B). For the autoclaved diets, PA was defined by Pseudomonas, PHFA was separated by Gluconobacter, and PHSA6 and PHSA11 were separated by Acetobacter and

Lactobacillus (Fig. 3.20C).

For S larvae, RHF food was defined by Alistipes and Streptococcus (Fig. 3.20D). RHS and R diets were not clearly separated from each other (Fig. 3.20D). Blautia influenced RHS, and Acetobacter was influential in separation of R food community (Fig. 3.20D). PHS diet was separated by Lactobacillus and Gluconobacter (Fig. 3.20E). PHF diet was differentiated by

Corynebacterium 1 (Fig. 3.20E). Acetobacter defined the PR diet (Fig. 3.20E). Among the autoclaved diets, PA was defined by Streptococcus and Acetobacter (Fig. 3.20F). PHFA was separated by Corynebacterium 1 and Lachnospiraceae NK4A136 (Fig. 3.20F). PHSA6 was differentiated by Lactobacillus and Alistipes (Fig. 3.20F).

139

Figure 3.20: Nutritional modifications cause shifts in bacterial community composition and are associated with changes in abundance of the dominant bacteria genera. Linear discriminant analysis of 10 dominant bacterial taxa. A) Non-sterilized larvae raised on lab-based diets; B) Non-sterilized larvae raised on peach-based diets; C) Non-sterilized larvae raised on autoclaved peach-based diets; D) Non-sterilized larvae raised on lab-based diets; E) Non- sterilized larvae raised on peach-based diets; F) Non-sterilized larvae raised on autoclaved peach-based diets. The intensity of vector rays’ color corresponds to their length. Confidence

140 ellipses are filled based on the color of the diet. Normal data ellipses are unfilled and leveled to include 50% of the samples.

2.5 Genotype and interaction between genotype and environmental variables influence abundance of bacterial taxa

Influence of genetic and environmental variables on microbial abundances. For the larvae raised on HF food, diet significantly influenced one phylum, four classes, 13 orders, 27 families, 78 genera, and 455 ZOTUs (S3.10). Genotype produced a significant effect on 15 phyla, 25 classes, 56 orders, 81 families, 231 genera, and 1,197 ZOTUs (S3.10). Treatment significantly affected three phyla, five classes, six orders, eight families, 28 genera, and 241

ZOTUs (S3.10). For the larvae raised on HS diet, D produced a significant effect on six phyla, eight classes, 21 orders, 38 families, 89 genera, and 411 ZOTUs (S3.10). G significantly influenced 12 phyla, 19 classes, 45 orders, 78 families, 192 genera, and 932 ZOTUs (S3.10). T significantly affected eight phyla, 16 classes, 39 orders, 72 families, 139 genera, and 578 ZOTUs

(S3.10).

Influence of variables interactive effect on microbial abundances. On the HF diets, the interactive effect of D*G significantly influenced the abundance of seven phyla, nine classes,

17 orders, 32 families, 97 genera, and 522 ZOTUs (S3.10). D*T influenced two phyla, four classes, six orders, 15 families, 24 genera, 205 ZOTUs (S3.10). G*T influenced two phyla, four classes, eight orders, 16 families, 32 genera, and 299 ZOTUs (S3.10). D*G*T influenced one phylum, two classes, five orders, eight families, 14 genera, and 284 ZOTUs (S3.10). On the HS diet, the interactive effect of D*G significantly influenced the abundance of one phylum, three classes, 14 orders, 23 families, 67 genera, and 489 ZOTUs (S3.10). D*T influenced two phyla, eight classes, 16 orders, 26 families, 79 genera, and 502 ZOTUs (S3.10). G*T produced a

141 significant effect on two phyla, four classes, 11 orders, 13 families, 58 genera, and 403 ZOTUs

(S3.10). D*G*T significantly influenced 0 phyla, four classes, 13 orders, 20 families, 70 genera, and 533 ZOTUs (S3.10).

3.1 The effect that symbiotic bacterial taxa produce on larval phenotypes varies with diet, genotype and their interactive effect

Influence of symbiotic bacterial abundance on larvae phenotypes. Across NS larvae, at the phylum level, there were 18 significant interactions between larvae phenotypes and microbial abundances on a RHF diet, nine on PHF, 12 on a PHFA diet, seven on RHS, 17 on

PHS, 13 on PHSA6, and 10 on PHSA11 diet (S3.11 & S3.12). At the same taxonomic level, S larvae had seven significant interactions on a RHF diet, 18 on PHF, six on PHFA, two on RHS, ten on a PHS diet, and three on PHSA6 diet (S3.11 & S3.12). For NS larvae, there were 32 significant interactions on the RHF diet, 11 on PHF diet, 25 on PHFA, 10 on RHS, 26 on PHS,

19 on PHSA6, 15 on PHSA11 diet at the class level (S3.11 & S3.12). At the same taxonomic level for S larvae there were 12 significant interactions on the RHF diet, 25 on PHF, 12 on

PHFA, three on RHS, 16 on PHS diet, and eight on PHSA6 diet (S3.11 & S3.12). At the order taxonomic level for NS larvae there were 62 significant interactions on a RHF diet, 30 on a PHF diet, 58 on PHFA, 18 on RHS, 51 on PHS, 32 on PHSA6, 28 on PHSA11 (S3.11 & S3.12). For S larvae at the same taxonomic level there were 25 on a RHF diet, 49 on PHF, 20 on PHFA, 20 on

RHS, 34 on a PHS diet, and 17 on a PHSA6 diet (S3.11 & S3.12).

At the family taxonomic level, across the NS larvae there were 105 significant interactions on RHF diet, 54 on PHF, 91 on PHFA, 37 on RHS, 76 on PHS, 54 on PHSA6, and

50 on PHSA11 (S3.11 & S3.12). Considering S larvae at the same taxonomic level, there were

61 significant interactions on RHF diet, 85 on PHF, 50 on PHFA, 33 on RHS, 46 on PHS, and 29

142 on PHSA6 diet (S3.11 & S3.12). For NS larvae at the genus taxonomic level there were 290 significant interactions on the RHF diet, 145 on PHF diet, 230 on PHFA, 101 on RHS, 158 on

PHS, 104 on PHSA6, and 156 on PHSA11 (S3.11 & S3.12, Fig. 3.21-3.22). Across S larvae at the same taxonomic level there were 151 significant interactions on the RHF diet, 209 on PHF,

164 on PHFA, 84 on RHS, 139 on PHS, and 76 on a PHSA6 diet (S3.11 & S3.12, Fig. 3.21-

3.22). At the level of individual ZOTUs, for NS larvae there were 1,415 significant interactions on RHF diet, 644 on PHF, 1,092 on a PHFA diet, 314 on RHS, 620 on PHS, 272 on PHSA6, and

551 on a PHSA11 diet (S3.11 & S3.12). For S larvae at the same level there were 677 significant interactions on the RHF diet, 626 on PHF, 843 on a PHFA diet, 386 on RHS, 472 on a PHS, and

245 on PHSA6 diet (S3.11 & S3.12). Our results suggested that both parental and environmental microbes produce effects on the formation of larvae phenotype with possible interactive effects.

143

Figure 3.21: On high fat diets, the influence of bacterial taxa on larval phenotype varies with diet, treatment, and their interactive effect. A) Spearman’s rank correlation between the abundance of dominant bacterial taxa and larval survival; B) Spearman’s rank correlation between the abundance of dominant bacterial taxa and median number of days required for larvae to reach the 3rd instar stage; C) Spearman’s rank correlation between the abundance of dominant bacterial taxa and larval weight; D) Spearman’s rank correlation between the abundance of dominant bacterial taxa and larval triglyceride concentrations; E) Spearman’s rank correlation between the abundance of dominant bacterial taxa and larval protein concentrations;

F) Spearman’s rank correlation between the abundance of dominant bacterial taxa and larval glucose concentrations. Bacteria genera from left to right: Acetobacter, Alistipes, Bacteroides,

144

Blautia, Corynebacterium 1, Gluconobacter, Lachnospiraceae NK4A136 group, Lactobacillus,

Pseudomonas, and Streptococcus. Symbols indicate the significance of comparisons p< 0.001

***, p< 0.01 **, p< 0.05 *, p< 0.1 ·, and p>0.1 _.

Figure 3.22: On high sugar diets, the influence of bacterial taxa on larval phenotype varies with diet, treatment, and their interactive effect. A) Spearman’s rank correlation between the abundance of dominant bacterial taxa and larval survival; B) Spearman’s rank correlation between the abundance of dominant bacterial taxa and median number of days required for larvae to reach the 3rd instar stage; C) Spearman’s rank correlation between the abundance of dominant bacterial taxa and larval weight; D) Spearman’s rank correlation between the

145 abundance of dominant bacterial taxa and larval triglyceride concentrations; E) Spearman’s rank correlation between the abundance of dominant bacterial taxa and larval protein concentrations;

F) Spearman’s rank correlation between the abundance of dominant bacterial taxa and larval glucose concentrations. Bacteria genera from left to right: Acetobacter, Alistipes, Bacteroides,

Blautia, Corynebacterium 1, Gluconobacter, Lachnospiraceae NK4A136 group, Lactobacillus,

Pseudomonas, and Streptococcus. Symbols indicate the significance of comparisons p< 0.001

***, p< 0.01 **, p< 0.05 *, p< 0.1 ·, and p>0.1 _.

3.2 Interactive effect of microbial abundance and environmental/genetic variables on formation of larvae phenotypes

For larvae raised on HF food, A*D displayed 29 significant interactions with larvae phenotypes at the phylum level, 35 at the class level, 80 at the order level, 112 at the family level, 319 at the genera level, and 2,197 at the ZOTUs level (S3.13). A*G had 18 significant interactions at the phylum level, 37 at the class level, 60 at the order level, 120 at the family level, 295 at the genera level, and 1,963 at the ZOTUs level (S3.13, Fig. 3.23). A*T exhibited 11 significant interactions at the phylum level, 14 at the class level, 25 at the order level, 56 at the family level, 177 at the genera level, and 1,595 at the ZOTUs level (S3.13).

146

Figure 3.23: The influence of bacteria abundance on larval phenotypes, varies with diet, genotype, and their interactive effect. A) Spearman’s rank correlation between the abundance of Acetobacter and Lactobacillus and triglyceride concentrations of larvae from five DGRP lines raised on nutritionally modified diets. B) Spearman’s rank correlation between the abundance of

Acetobacter and Lactobacillus and glucose concentrations of larvae from five DGRP lines raised on nutritionally modified diets .Symbols indicate the significance of comparisons p< 0.001 ***, p< 0.01 **, p< 0.05 *, p< 0.1 ·, and p>0.1 _.

For larvae raised on HS food, we observed that A*D exhibited eight significant interactions with larvae phenotypes at the phylum level, 20 at the class level, 74 at the order level, 125 at the family level, 285 at the genera level, and 1,384 at the ZOTU’s level (S3.13).

A*G had four significant interactions at the phylum level, nine at the class level, 23 at the order level, 57 at the family level, 160 at the genera level, and 805 at the ZOTUs level (S3.13, Fig

147

3.23). A*T exhibited nine significant interactions with larvae phenotypes at the phylum level, 17 at the class level, 30 at the order level, 48 at the family level, 157 at the genera level, 1,011 at the

ZOTUs level (S3.13).

DISCUSSION

1.1-1.4 Survival. Consistent with our previous work, a larger number of larvae survived on lab diets no matter the modification of the food. This pattern is likely to be explained by the fact that protein is the limiting nutrient on a natural diet and not lipids or sugars (Pais et al. 2018, Bing et al. 2018). Also consistent with our previous findings, on the natural diets the presence of parental and environmental bacteria increased the survival of the larvae, no matter the food modification

(Chapter two). Interestingly, larvae raised on lab diets were less effected by food modifications than larvae raised on natural diets. On the HS peach diets, presence of parental microbiota was sufficient to reduce the mortality of larvae to the level of unmodified diets. This is consistent with previous research showing that the parental microbiota helps larvae adapt to nutritionally unfavorable conditions (Bing et al. 2018, Berger et al. 2005, Henry, Tarapacki, and Colinet

2020). Similarly, Wong et. al (2014) reported high pre adult mortality of flies on a high sugar low-yeast diet, especially for sterilized larvae.

Development. The influence of the diets on the development rate was also largely consistent with our previous work. For example, larvae raised on lab food developed faster than larvae raised on peach-based diets (Chapter two). On the peach-based diets, both environmental and parental microbiota were capable of increasing the development rate. Interestingly, the presence of parental microbiota could compensate for the absence of environmental microbes on

HS diet, but only with lower sugar concentrations. This suggests that consumption of excess

148 sugars may be one of the mechanisms through which microbes express the protective qualities on larvae growth on the HS diet. Such explanation is consistent with previous research which indicated that common Drosophila symbiont Acetobacter tropicalis can reduce sugars concentrations in the food (Huang and Douglas 2015). In addition, our previous observations showed that presence of the environmental microbes, during incubation period, decreases sugar concentrations in the peach food, even in the absence of larvae (Chapter two). Previous research indicated that high sugar low protein diets can increase the development time of Drosophila, especially in the absence of parental microbiota (Wong, Dobson, and Douglas 2014, Henry,

Tarapacki, and Colinet 2020). We observed similar results on our non-autoclaved peach diets.

Interestingly, the development rate of larvae raised on lab diets was less responsive to addition of

HF or HS than that of larvae raised on peach-based diets. The environmental microbiota on the peach diet were sufficient to eliminate the effects of HF modification but not HS.

Weight. Consistent with our previous work, diets that supported higher development and survival rates also generally supported higher weights of the larvae. Interestingly, parental microbiota did not produce a strong effect on weight of the larvae raised on HF diets. For HS diets, the positive effect of parental microbiota was present only in the absence of environmental microbes and with higher concentrations of sucrose. This once again suggests that reduction of sugar concentrations might be responsible for this positive effect. The increase in sugar and fat concentrations lead to decrease in larvae weights, which is consistent with previous research

(Reed et al. 2010, Musselman et al. 2011). Interestingly, the decrease of weight due to modification of the peach diets was noticeable only if parental microbiota were present. This is consistent with our previous conclusions that larvae weights on peach diets are primarily limited by protein availability (Chapter two). However, on autoclaved peach diets, especially with

149 harmful modifications, sterilized larvae exhibited extremely low survival rate and mostly were sampled from two genetic lines. That complicated the evaluation of a general pattern for metabolic phenotypes and suggested a presence of genetic adaptations in those Drosophila lines.

Triglyceride. Consistent with our previous findings, on HF and HS food, we observed that diets that were positively correlated with fitness traits were negatively correlated with triglyceride amounts in larvae (Chapter two). Interestingly, the presence of parental microbiota on lab diets was sufficient to eliminate the difference in triglyceride concentrations caused by any modification of the food. Previous research indicated that HS diets can increase triglyceride concentrations in Drosophila and vertebrate models (Galenza et al. 2016, Do et al. 2018, Jena et al. 2016, Jehrke et al. 2018, Wong et al. 2015). This pattern matched our findings for larvae raised on lab food and on the non-autoclaved peach diet. However, interactive effects of diet

(with environmental microbes) and parental microbiota seemed to strongly affect how larval triglyceride concentrations were impacted by the addition of extra fats or sugars.

Glucose and Protein. Consistent with our previous findings, the glucose and protein concentrations in larvae did not directly correlate with life history traits (Chapter two). In general, for sterilized larvae, both protein and glucose concentrations were similar in their response to changing diets compared to our previous work. For example, larvae raised on lab and autoclaved diet types had higher protein and glucose concentrations than larvae raised on peach diets (Chapter two). For the glucose phenotype, diet was a more impactful factor than presence of parental microbes alone. Previous research indicated that a high sugar diet may cause increased glucose concentrations in Drosophila and vertebrate models (Galenza et al. 2016, Do et al. 2018, Jehrke et al. 2018, Wong, Dobson, and Douglas 2014). We found similar results for sterilized larvae raised on peach diet and for non-sterilized larvae raised on autoclaved peach

150 diets. However, Henry et. al (2020) did not observe a positive correlation between additional sugar in a balanced diet and Drosophila glucose concentrations. Consistently, on the lab

(balanced) diet, we did not see an increase in larvae glucose concentrations, in response to consumption of additional sugar. Evaluating the influence of diet on fly protein concentrations,

Galenza et. al (2016) and Jehrke et. al (2018) reported no significant correlations with increased sugars consumption. While Wong (2014) also did not observe a significant increase in protein concentrations in response to increased dietary sugars, the authors reported a significant diet-by- bacteria interaction for male flies. In our work, interaction of nutritional, microbial and genetic variables appeared to be very influential in controlling the amount of glucose and protein in the larvae.

Genotype and interaction between genotype and environmental variables influence formation of larval phenotypes. Previous research indicated that genotype could significantly influence metabolic phenotypes of D. melanogaster such as weight, triglyceride and glucose concentrations, glycogen storage, and metabolic rate on normal as well as on modified diets

(Jumbo-Lucioni et al. 2010, Dew-Budd, Jarnigan, and Reed 2016, Reed et al. 2010). In our work, we observed that genotype had a significant influence on all phenotypes except glucose concentration on HF and HS diets. We also observed variation between genetic lines for the response of adding fats and sugars to diets. Previous research also indicated that phenotypic response to diet modification may vary with genotype and be influenced by a diet-by-genotype interaction (James et al. 2020, Reed et al. 2010, Reed et al. 2014, Zhu, Ingelmo, and Rand 2014).

We found that diet-by-genotype interactions on the HF and HS diets produced a significant effect on all tested phenotypes except glucose concentrations.

2.1-2.3 Microbial community composition varies with larval diets

151

Alpha Diversities. The alpha diversities of HS food communities vary between diets more than HF food communities, but the variation largely depended on parental microbiota presence. The change in alpha diversities between normal and modified diets suggested that addition of extra lipids may increase the diversity of the bacterial community unless the established natural bacterial community is present, while addition of extra sugars largely does not influence alpha diversities. However, this pattern also varied with the presence of parental microbiota. Our results were not entirely consistent with the previous research. Galenza et al

(2016) observed an increase in Shannon index of flies raised on a holidic diet, with known concentrations of all the nutrients, developed by Piper et. al (2014), supplemented with extra glucose (without a test for statistical significance). In our work, we observed a similar but non- significant pattern comparing PHS and PR diets. However, in contrast to the findings of Galenza et al (2016), we found that non-sterilized larvae raised on RHS exhibited a microbial community with a significantly lower Shannon’s index than larvae raised on R diet. Previous studies showed that increased nutrients and protein concentrations were negatively associated with bacterial diversity (Seganfredo et al. 2017, Kim et al. 2016, Erkosar et al. 2018). In our work, comparing lab modified and peach diets, we found that only RHS diet supported a microbial community with lower diversity, which partially might be explained by its higher protein content than natural diets. In mice, Do et al (2018) showed that increased fat, glucose, and fructose diets can reduce the bacterial richness and diversity, which is consistent with the patterns that we observed on HS but not on HF diets.

Beta diversities. While microbial communities did not alter much in alpha diversities, the composition of the microbial communities between peach-based diets and lab-based diets were distinct, as indicated by Bray-Curtis distances. Consistent with our previous work, we

152 observed that microbial communities raised on autoclaved peach diets are more similar to lab- based diets than to natural peach food, no matter which modification was applied (Chapter two).

Similar to alpha diversities, the presence of parental microbiota did play a more noticeable role on community structure of HS raised larvae than HF raised larvae. Interestingly, we observed that the differences in microbial communities within HF raised larvae were most noticeable at higher taxonomic levels. Within HS diets and comparing normal and modified diets, it was possible to observe major patterns only at lower taxonomic levels, such as family and genera.

A very limited number of studies that compared whole Drosophila gut bacterial communities in response to HF or HS modifications of the diet are currently available. Von

Frieling et. al (2020) suggested that HF microbial community is distinct from the normal diet.

Jehrke et. al (2018) did not observe substantial differences in the structure of the whole microbial communities of normal and HS diets. In our work, largely, standard and modified diets exhibited a significant difference in the community structurers as was indicated by Bray-Curtis distances, especially if larvae were sterilized. The phylogenetic diversity distances also indicated a significant difference between lab food-based diets and peach-based diets, especially if the parental microbiota were not present. Comparing normal and modified diets, we observed that

HF diets were more similar in their taxonomic diversity to normal diets than were HS diets, particularly for sterilized larvae. Considering vertebrate models, previous research indicated that mice and rats fed on HF and HS diets formed microbial communities that are distinct from ones feeding on a normal diet (von Frieling et al. 2020, Do et al. 2018). As described above, we observed similar patterns in our data, especially for sterilized larvae.

Abundance of individual taxa. On HF diets, the significant difference between the abundances of the highest number of taxa was observed between PHF and RHF diets for NS

153 larvae and between PHF and PHFA diets for S larvae. The pattern was similar on HS diets, where the greatest number of taxa that were significantly different in their abundances was observed between PHS and PHSA6 for non-sterilized larvae and PHS and RHS for sterilized larvae. Once again, parental microbiota produced more impact on the microbial community of larvae raised on HS diets than on the community of HF raised larvae. In general, the abundance of more taxa was influenced by diet modifications on lab diets, compared with peach diets.

2.4. Abundances of the dominant bacteria taxa are associated with diet types

Interestingly, across the diet comparisons, non-autoclaved peach diets tended to consistently associate with at least one dominant bacteria taxon, such as Cyanobacteria at the phylum level, Oxyphotobacteria at the class level, Enterobacteriales at the order level and

Enterobacteriaceae at the family level. Burkholderiaceae tended to associate with all lab-based diets but not as strongly.

Among the dominant bacteria groups, we also noticed several taxa that were consistently associated with certain food modifications. HF diets were associated with Actinobacteria and

Bacteroidetes at the phylum level, Corynebacteriales and Erysipelotrichales at the order level,

Corynebacteriaceae and for sterilized larvae, additionally Erysipelotrichaceae at the family level, and Corynebacterium 1 and Lachnospiraceae NK4A136 group at the genera level. Several representatives of Corynebacterium are known to be members of symbiotic microbiota of vertebrates, exhibit lipophilic qualities, and produce lipolytic enzymes (Hahne et al. 2018). In our work, Corynebacteriaceae was negatively correlated with the glucose concentrations on

RHF and PHF diets. Corynebacterium 1 was negatively correlated with glucose concentrations, on the same diets. Corynebacterium (a different taxon from Corinabacterium 1) was positively

154 correlated with glucose on PHFA and negatively correlated with glucose on PHF diet. Previous research observed an increase in the abundance of Erysipelotrichaceae in mice fed a high fat diet

(Kaakoush 2015). Erysipelotrichaceae was also shown to be associated with host lipid metabolism and dyslipidemic phenotype (Martínez et al. 2013). In our work, we observed positive correlations between Erysipelotrichaceae and triglyceride concentrations and the development rate on PHF diet. In addition, we observed a positive correlation with total larvae and glucose concentrations on PHFA diet and a negative correlation with glucose, protein concentrations and development time on RHF diet. Lachnospiraceae was also shown to be correlated with high-fat diets, altered lipid metabolism and obesity in humans and animal models

(Vacca et al. 2020). In our work, Lachnospiraceae was negatively correlated with glucose, protein concentrations, and development rate on a RHF diet, and glucose and development time on PHF diet. Positive correlations were observed with total larvae numbers on RHF and PHFA food types and glucose concentrations on PHFA food. These findings suggest that several symbiotic taxa have similar associations with multiple host organisms. Therefore, these relations might be evolutionarily conserved.

HS diets were associated with Tenericutes and Fermicutes at the phylum level,

Lactobacillales and Enterobacteriales at the order level, Enterobacteriaceae and

Lactobacillaceae at the family level, and Lactobacillus at the genus level. Lactobacillus is known to produce glycoside hydrolases and polysaccharide lyases which can degrade carbohydrates (Wang et al. 2020). Previous research showed that Lactobacillaceae abundances were increased in honeybees that were fed a sucrose solution; the same carbohydrate that we used in order to make HS diets (Wang et al. 2020). Lactobacillus strains are known to be a part of human microbiota and were correlated with a decrease in fasting glucose sugar and insulin

155 resistance (Azad et al. 2018, Khalili et al. 2019). Mice that were fed by high sucrose diet were shown to have higher abundances of Lactobacillus than mice fed on a normal diet (Magnusson et al. 2015). The survival of ingested Lactobacillus plantarum was shown to be improved upon feeding on high fat high sugar diets (Yin et al. 2017). In our work, on HS diets, we observed that

Lactobacillaceae was positively correlated with total number of larvae and development rate on

PHS diet and negatively correlated with protein and triglyceride concentrations on RHS diet.

Lactobacillus was positively correlated with the total number of larvae and development rate on

PHS food. In addition, a negative correlation was observed with weight on a PHS diet and protein and triglyceride concentrations on a RHS diet. Enterobacteriaceae and other members of

Protobacteria are known for their ability to utilize simple carbohydrates rapidly and for being associated with a high sugar diet (Satokari 2020, Volynets et al. 2017, Park et al. 2013,

Magnusson et al. 2015). In addition, Enterobacteriaceae were shown to be correlated with the obese phenotype (Xiao et al. 2014). In our work, Enterobacteriaceae was negatively correlated with triglyceride concentrations on PHS diet, positively correlated with development rate on

PHS and larvae survival on PHS and PHSA6 diets.

Surprisingly, unmodified diets are also consistently associated with certain microbial taxa, such as Virrumicrobia and Epsilonobacteria at the phylum level, Pseudomanales and

Acetobacterales at the order level, Acetobacteraceae and Burkholderiaceae at the family level, and Pseudomonas and Acetobacter (mostly for sterilized larvae). Members of Acetobacteraceae family are known to be a part of normal Drosophila microbiota in the lab and wild flies

(Vandehoef, Molaei, and Karpac 2020, Han et al. 2017, Bost et al. 2018, Winans et al. 2017). In our previous work, we observed that on unmodified diets, Acetobacteraceae was negatively correlated with triglyceride concentrations on PA and PR diets, positively correlated with weight

156 on PA and PR diets and positively correlated with development time on PR diet. Acetobacter was negatively correlated with triglyceride on PR and PA diets and glucose on PA diet.

Moreover, Acetobacter was positive correlated with weight on PA and PR diets and development time on a PR diet. Several members of Burkholderriaceae and Pseudomonas are known to be pathogens in Drosophila (Vodovar et al. 2005, Pilátová and Dionne 2012, D'Argenio et al.

2001). Our previous work found a negative correlation between Burkolderriaceae and weight on

PR diet and larvae development time on a R diet. Additionally, we observed a positive correlation with protein concentration on a PR diet. Pseudomonas was negatively correlated with the total number of larvae on R diet as well as development time and weight on a PR diet.

Triglyceride concentrations on PR diet and development time on PA diet were positively correlated with the abundance of Pseudomonas.

2.5 Bacterial communities are influenced by larval diets and interactions between environmental variables

Interactive effects between microbial abundances in larvae raised on different diets.

Previous research indicated that during colonization of the new environment, microbial taxa rely on interactions with each other (Matamoros et al. 2013, Li et al. 2004). Consistent with our expectations, among HF diets, we observed the highest number of significant microbial interactions on autoclaved peach diet. Surprisingly, on the HS diets, the pattern varied between diets at different taxonomic levels and it was not possible to make any general conclusions.

Larval genotype and interaction between environment and genotype influence microbial abundances. For both HF and HS diet types, genotype influenced the abundance of most taxa, which is consistent with our previous work (Chapter two) and research done by Jehrke et. al (2018). However, D*G was the second most influential variable on the HF food type while

157 treatment was the second most influential factor on HS food type. Kreznar et. al (2017) also reported a significant diet-by-genotype effect on the gut bacterial community composition. This supports the conclusion that although diet plays an important role in the formation of symbiotic bacterial community, it rarely can be considered the most influential factor in shaping a symbiotic microbiome. Overall, all tested variables and their interactive effects produced a significant influence on the abundance of symbiotic microbial taxa.

3.1 Abundances of individual microbial taxa are correlated with larval phenotypes

Previous studies have observed a correlation between the abundance of gut microbial taxa and measured phenotypes, as well as variation in the abundances of microbial taxa between obese and lean phenotypes (including accumulation of lipids and elevated glucose concentrations in the blood) (Chaston, Newell, and Douglas 2014, von Frieling et al. 2020, Martyn, Kaneki, and

Yasuhara 2008, Hasan and Yang 2019). Using a Drosophila model, von Frieling et. al (2020) observed that on HF diet, Enterobacteriales and Caulobacterales were significantly enriched.

While we did not observe a significant increase in the former on the high fat diets, the later was significantly higher in abundance on the RHF diet for non-sterilized larvae (abundance was also higher on peach diet but not to the level of significance). On RHF food, Caulobacterales was negatively correlated with triglyceride concentrations and development rate. Von Frieling (2020) also noted that Lactobacillaceae was significantly enriched on the normal diet. We also observed a significant increase in Lactobacillaceae abundance on a regular lab diet compared to a HF diet.

In addition, on the R diet Lactobacillaceae was positively associated with larvae survival.

Vandehoef et. al (2020) reported an increase in Lactobacillus on HS diets. Our findings also indicated higher abundances of Lactobacillus and Lactobacillaceae on peach based high sugar diets comparing with regular peach diets.

158

Cani et al. (2007) described a significant decrease in the abundance of

Enterobacteriaceae, Bacteroides, and Bifidobacteria in the obesity induced high fat fed mice.

Multiple studies have also described a link between obesity and members of the phylum

Bacteroidetes (Murphy et al. 2010, Hildebrandt et al. 2009, Zhang et al. 2010, Guo et al. 2017).

We did not observe a significant difference in the abundance of Enterobacteriaceae or

Bacteroides between any of our HF or R diets. However, in our work, Bacteroides was found to be negatively correlated with glucose concentrations on a RHF diet and on a PHF diet.

Bacteroides was also found to be positively correlated with glucose on a PHFA diet. Comparing

HF and normal diet, we observed that larvae raised on RHF diet had increased abundance of

Bifidobacterium, Bifidobacteriaceae, Bifidobacteriales, Actinobacteria and Bacteroidetes.

Moreover, larvae raised on PHFA had increased abundances of Bifidobacteriaceae,

Bifidobacteriales, and Actinobacteria. We also observed that Bifidobacterium were positively correlated with triglyceride on PHF diet and glucose concentrations on PHFA diet. Additionally, the family Bifidobacteriaceae, the order Bifidobacteriales, the class Actinobacteria, and the phylum Bacteroidetes were found to be negatively correlated with glucose concentrations in larvae raised on a PHF diet and on RHF diet. In larvae fed a PHFA diet, Bifidobacteriaceae,

Bifidobacteriales, and Bacteroidetes were positively correlated with glucose.

Previous research also identified a decrease in Bacteroidaceae, Prevotellaceae, and

Rikenellaceae in the obese mice fed on HF diet (Hildebrandt et al. 2009, Daniel et al. 2014). We did not find a significant difference in the abundance of any of these families between high fat and regular diets, however all three families were found to be negatively correlated with glucose in larvae raised on a RHF diet and positively correlated with glucose in larvae raised on PHFA food. Bacteroidaceae was also negatively correlated with glucose on a PHF diet.

159

In addition, Daniel et al. (2014) found that mice fed a high fat diet exhibited an increase in body weight and induced obesity as well as increased abundance of Rikenellaceae and decreased amount of Ruminococcaceae. In our work we did not observe an increase in

Rikenellaceae or Ruminococcaceae on any of the HF diets. However, we observed that

Rikenellaceae was positively correlated with triglyceride concentrations on RHF diet.

Ruminococcaceae was negatively associated with glucose on RHF and PHF diets. Additionally, it was positively correlated with glucose on PHFA diet and with triglyceride concentrations on

RHF diet.

Park et al. (2014) found that on a high-fructose diet, rodents developed characteristics of the metabolic syndrome; however, probiotic treatment with either L. plantarum or L. curvatus lowered plasma glucose, insulin, triglycerides, and oxidative stress levels. A similar study by

Zubiria et al. (2017) found comparable preventative effects using L. kefiri as a probiotic in high fructose fed mice. Comparing HS and normal diets, we observed that larvae raised on PHS had significantly higher abundances of Lactobacillus and Lactobacillaceae than larvae raised on PR diet. However, we did not observe a significant difference on lab diets.

Previous research indicated that mice fed on HS diets had higher weight, blood glucose concentrations, fat mass and triglyceride concentrations comparing with the control mice and had lower relative abundance of Firmicutes and increase abundance Proteobacteria, in particular

Desulfovibrio vulgaris and Akkermansia muciniphila (Do et al. 2018, Liu et al. 2016).

Comparing larvae raised on normal and HS diets, we observed that RHS raised larvae showed a decreased abundance of Bacteroidetes, Proteobacteria, and Firmicutes. PHS raised larvae had decreased abundance of Proteobacteria. Bacteroidetes was also decreased in larvae raised on

PHSA6 food. While there was no significant difference in the abundance of Desulfovibrio

160 between any high sugar and normal diet, there was a decrease in the abundance of Akkermansia in larvae raised on an RHS diet compared to regular diet and PHSA6 diet compared to PA food.

Furthermore, Bacteroidetes and Firmicutes were associated with increased triglyceride concentrations in larvae raised on PHS and PHSA11 food. Interestingly, Proteobacteria was found to be positively associated with triglyceride concentrations in non-sterilized larvae raised on the PHS diet, but negatively associated with triglyceride concentrations in sterilized larvae raised on the same diet. Additionally, Desulfovibrio was found to be positively correlated with triglyceride concentrations on PHS food raised larvae. Akkermansia was also found to be positively correlated with triglyceride concentrations in larvae raised on PHS, PHSA11, and

PHSA6 diets.

Org et al. (2015) indicated that two taxa from the family Lachnospiracea, Roseburia spp. and Ruminococcus gnavus, were positively correlated with obesity and metabolic phenotypes, including body fat increase on a high-fat/high-sucrose diet in mice. In our work, we observed that Roseburia spp. was positively correlated with triglyceride increase on Drosophila larvae fed on PHS diet. However, we observed that Ruminococcus gnavus had negative correlations with the triglyceride concentrations on RHF diet raised larvae. Kreznar et al. (2017) found that

Bacteroidaceae had the most negative correlation with several metabolic traits, including body weight and fasting plasma insulin on high-fat/high-sucrose diet raised mice. In our work,

Bacteroidaceae had negative correlations with body weight on PHF and RHS diets raised larvae.

In contrast, Rikenellaceae was among the most positively correlated with plasma insulin concentrations (Kreznar et al. 2017). Although we did not measure insulin concentrations directly, we did observe positive correlations between Rikenellaceae and glucose concentrations

161 on RHS diet raised larvae. In larvae raised on PHF diet, Rikenellaceae was negatively correlated with glucose.

3.2 Interactive effect of symbiotic bacteria with environmental and genetic variables in forming host phenotype

We observed a significant interactive effect between environmental and genetic variable and microbial abundances in forming phenotypes of the larvae. On HF diets, A*D and A*G were the most influential interactive effects in forming larvae phenotypes, while on HS diets, A*D was the dominant interactive effect. These results are consistent with the previous research that found a significant influence of A*D and A*D*G interactive effect in forming phenotypes of

Drosophila and mice models on standard and modified diets (Wong, Dobson, and Douglas 2014,

Kreznar et al. 2017, Henry, Tarapacki, and Colinet 2020, Zhang et al. 2010, Org et al. 2015).

Consumption of high fat and high sugar diets is often associated with a development of unhealthy metabolic phenotype and obesity (Seganfredo et al. 2017, Patterson et al. 2016, Al-

Goblan et al. 2014). With this work, I identified that larval metabolic response, to nutritional modification of the diet may vary between lab-based diets and peach-based diets. Microbiota acquired from the environment or inherited maternally are capable of reducing negative metabolic effects, caused by nutritional modification of a diet, especially on peach-based diets. I observed that nutritional modifications of the diets are associated with shifts in gut bacterial community composition. Interestingly, the dominant bacteria taxa associated with particular nutritional modification (high fat or high sugar), in our Drosophila model are similar with bacterial community members associated with westernized diet, in vertebrate models, and human populations. However, independent of the nutritional modification, genotype of the host influences abundance of more microbial taxa than any other tested variable, which is consistent

162 with the results observed in the second chapter of this work. In addition, the effects that microbial taxa produce on a host vary with genotype, diet, and treatment, which is also consistent with the results observed in the second chapter.

163

CHAPTER FOUR

SYNTHESIS

The symbiotic microbiota community often helps the host to adapt to nutritional challenges, which is consistent with the holobiont theory of evolution (Leitão-Gonçalves et al. 2017, Erkosar et al. 2015, Neis, Dejong, and Rensen 2015, Bordenstein and Theis 2015). However, many studies that evaluate the interaction between gut microbiota and their host use standard lab diets

(Chaston et al. 2016, Dobson et al. 2015, Wong, Dobson, and Douglas 2014). This approach may be disadvantageous for several reasons. First of all, it was shown that the gut microbiota composition of lab organisms and their natural populations exhibit significant differences

(Martínez-Mota et al. 2020, Chandler et al. 2011, Vacchini et al. 2017). The lab environment applies different selective pressure on the microbiota community compared with natural conditions, thus even the qualities of microbiota taxa may change during lab adaptation (Pais et al. 2018, Henry, Tarapacki, and Colinet 2020). It was shown that, relative to lab adapted microbiota, Drosophila symbiotic microbial taxa from natural populations are more efficient in colonizing the host and maintaining their population inside of the digestive tract (Pais et al. 2018,

Henry, Tarapacki, and Colinet 2020). On the other side, microbial taxa that are adapted to lab environment are more efficient in metabolizing ammonium-based compounds, which is likely the result of significantly higher concentrations of nitrogen wastes in dense lab colonies of

Drosophila (Pais et al. 2018, Henry, Tarapacki, and Colinet 2020). In addition, the nutritional

164 values of the food that organisms consume in the lab and natural environments are often different

(Martínez-Mota et al. 2020, Tefit et al. 2017, Douglas 2018). Considering the above issues studying the effect of lab adapted microbiota population on the host phenotype, using standard lab diets may take away a lot of evolutionary context from host-symbiont interactions. This can result in findings that are not necessarily applicable for the natural populations of the organisms.

From another point of view, it is not always possible to work with natural populations to address all experimental goals. The natural environment is extremely variable, and the symbiotic microbiota composition was shown to shift with seasonality, latitude, temperature, and many other environmental factors (Behar, Yuval, and Jurkevitch 2008, Ferguson et al. 2018,

Moghadam et al. 2018, Walters et al. 2020). Thus, it is easy to see how the interaction of all these variables might be overwhelming if experiments were designed to work only with natural populations. In addition, it might be difficult to differentiate between closely related species in the field. For example, the major morphological traits that allows scientists to recognize closely related Drosophila melanogaster and Drosophila simulans are present only in adult male flies

(Markow and O'Grady 2005, Strickberger 1962). In our work, we tried to address the issues described above and combine the advantages of working in the controlled lab environment. We experimented with inbred Drosophila lines of known genotype and a naturally rotted peach diet that was inoculated with symbiotic microbiota by wild Drosophilids and other natural vectors.

With our approach, we were able to evaluate the difference in larval phenotypic response to natural and lab diets, as well as to the nutritional modifications applied to them. We tested the effects that diet, nutritional modifications, and seeding parental microbiota produced on the formation of symbiotic bacterial community of the larvae. We studied the effects that individual bacterial taxa produced on the host life history traits and metabolic phenotype. Lastly, we

165 evaluated the effect that genotype and its interaction with environmental variables produced on the formation of Drosophila phenotype and bacterial community.

Influence of lab and natural diets and their modification on phenotypes of Drosophila larvae

Our work showed that every tested phenotype was significantly different between larvae raised on peach and lab diets. As such, Drosophila that were feeding on the peach diets had significantly lower survival and development rates, lower weight, protein, and glucose concentrations but higher triglyceride content. Based on the previous research, most of these phenotypic changes might be explained by the nutritional and pathogenic stresses that are associated with a natural nutritional environment (Klepsatel, Procházka, and Gáliková 2018, Pais et al. 2018, Bing et al. 2018, Staubach et al. 2013, Sang 1956). Furthermore, we observed that differences in survival, development rate, weights, and triglyceride concentrations were further enhanced if the symbiotic bacterial community was reduced by autoclaving the peach diet or sterilization of the embryos. Previous studies also observed that the absence of symbiotic microbiota can produce similar effects in Drosophila especially on low protein diets, which matched our experimental conditions (Pais et al. 2018, Bing et al. 2018, Wong, Dobson, and

Douglas 2014, Shin et al. 2011, Dobson et al. 2015).

Some of the described differences between larvae raised on lab and peach diets were preserved even when the diets were modified with the addition of extra fats or sugars. As such, no matter the food modification larvae, survived better, developed faster, weighed more, and had lower triglyceride concentrations on the lab-based diets compared with the peach-based diets. In general, the relation between modified lab-based diets and peach-based diets in glucose and protein concentrations was also similar with unmodified diets but only if the parental microbiota

166 was absent. The similarities that we observed on normal and modified diets further suggested that major nutritional limitation for larvae survival and normal development on the natural diets is the amount of available protein. Previous research showed that symbiotic microbiota are capable of suppressing Drosophilas’ appetites for yeast and is beneficial on fresh fruit diets, low in protein values (Leitão-Gonçalves et al. 2017, Pais et al. 2018, Bing et al. 2018). These results are consistent with our observations that showed the critical importance of the symbiotic bacteria for larvae survival and development on natural peach diets. Yeasts are the major source of protein for Drosophila that feed on a fruit and often on the lab diet, as well (Wong, Dobson, and

Douglas 2014, Leitão-Gonçalves et al. 2017, Shin et al. 2011). It would be interesting to test if symbiotic bacteria would still produce such strong effect on larvae raised on the natural diets, if we supply them with extra yeast or other protein sources such as whey or casein proteins.

We did not observe a consistent pattern in larval phenotypic response due to nutritional modifications across lab and peach diets. In some phenotypes, such as larval weights and triglyceride concentrations, larvae raised on the lab diets were more responsive to dietary changes. For other phenotypes, such as survival and development rate, the pattern was the opposite. It has to be noted that in experimenting with modified natural diets we observed an exceptional importance of the symbiotic bacteria for larval ability to withstand harmful dietary additions. As such, in the absence of environmental and parental bacteria, the survival of the larvae raised on peach high sugar diet varied between 0% and 12%, depending on the tested genetic lines. These results provide further support to conclusions in previous studies on the critical importance of symbiotic microbiota for adaptation to nutritional challenges and for the holobiont theory of evolution, in general (Pais et al. 2018, Bing et al. 2018, Henry et al. 2020,

Bordenstein and Theis 2015).

167

We observed very interesting results for the influence of natural diet and dietary modification on metabolic phenotype of the larvae, many of which also find support in the previous research. However, it is impossible to ignore some of the possible alternative explanations of our results that were not accounted for and will need further experiments to eliminate these gaps. One of the additional explanations for the difference in the metabolic phenotypes, due to feeding on different diets may be the amount of the food that larvae consume.

For example, it is possible to hypothesize that lower weight of the larvae raised on autoclaved peach diet comparing with peach diet may not be only explained by the activity of live microbes but also by the reduced appetite and therefore lower food consumption of the larvae. It was shown that Drosophila is capable of exhibiting the preference for the food that contains its symbiotic microbiota (Leitão-Gonçalves et al. 2017, Wong et al. 2017).

Several techniques have been developed to quantify the amount of food ingested by

Drosophila. The food source might be labeled with radioactive tracer (RT) or colorimetric dye and the amount of consumed food will be correlated with the intensity of color or radioactive tracers inside of the organism (Deshpande et al. 2014, Carvalho, Kapahi, and Benzer 2005).

Other techniques are capillary feeder (CAFE) and proboscis extension assays. Among these methods, CAFE and RT were shown to provide the most consistent results across the studies

(Deshpande et al. 2014). The CAFE assay requires a capillary filled with liquid food and one can measure the amount of food that is consumed by the animal via change in the volume of liquid within the capillary (Deshpande et al. 2014). Therefore, this assay is applicable only with the liquid types of nutrition. As we used solid and semisolid types of the media, the CAFE assay is not applicable in our case. In addition, proboscis/mouth part food extension assay will also be inefficient in our work with larvae, since they burrow in the food making them difficult to

168 observe. Therefore, to measure the amount of food consumed by larvae in our work, it would reasonable to apply the RT or colorimetric dye labeling of the diet.

Another complication in evaluating the difference in phenotypes of the larvae raised on autoclaved and peach diets is in terms of controlling for the actions of only live microbes is the potential changes in nutritional values of the food. That might be caused by a thermal effect produced by an autoclave machine. It is well known that heat application can result in the decomposition of sugars and denaturation of proteins (Urbaniec, Zalewski, and Zhu 2000, Hillier and Lyster 1979). In addition, from the sequences of larval microbial communities, we can see that autoclaved diets can be successfully recolonized with bacterial community, even if the parental microbiota is removed (Sterilized larvae raised on a PA diets formed microbial communities distinct from larvae raised on R or PR diets). It is possible to address the described issues with two approaches. First of all, we can recolonize freshly autoclaved food with the peach food microbiota before we place first instar larvae on it. Tefit et., al (2017) demonstrated the possibility of transferring natural microbiota on the food source by seeding it through adult male flies raised on the natural diet. Thus, the lack of difference between metabolic phenotypes of the larvae raised on natural peach diet and recolonized autoclaved peach diet would demonstrate that the difference in metabolic phenotypes of PA and PR raised larvae is due to the symbiotic bacterial activity and not shifts in the nutritional values of the food. Although this method can serve as a useful control, its application will not be able to prevent recolonization of the intact autoclaved food by bacterial community from lab environment.

An alternative approach would be to avoid usage of the autoclaving to eliminate microbiota community. Previous research indicated that it is possible to drastically reduce the fly microbiota community and potentially even create axenic condition via application of antibiotics

169 such as tetracycline (Griffin, Reed, and Evolution 2020, Heys et al. 2018). Applying antibiotics, we could create peach food with extremely low bacterial composition. In addition, it will allow us to preserve eukaryotic microbiota organisms. However, one of the disadvantages of this approach is that antibiotic treatment may affect not only the prokaryotic microbiota community but influence the metabolic phenotype of Drosophila as well, especially through altering mitochondrial metabolism (Ballard and Melvin 2007, Lardy, Connelly, and Johnson 1964). Thus, in order to control for the variety of possible interactive effects it might be better to combine both described techniques, expanding the variety of control samples.

The influence of natural diet, nutritional modifications, and parental microbiota on formation of Drosophila larval gut bacterial community

In evaluating the microbial communities of larvae raised on unmodified diets, we observed a significant difference between larvae raised on all three diets, especially if the parental microbiota was removed. If larvae were not deprived of their parental microbiota, samples raised on a PA diet were more similar in their bacterial composition with larvae raised on a R diet, compared with ones raised on a PR diet. This pattern is very interesting, since it shows that nutritional values of the food may not be the most definitive factor in shaping gut bacterial community. A similar conclusion was made by Jehrke et. al (2018), in their work on diet modifications and gut bacterial composition. Interestingly, we also observed a variation in inter-microbial interactions between diets. Thus, we found more medium to strong correlations, between the dominant bacteria taxa on a peach diet comparing with a regular diet. The pattern in samples raised on the PA diet was once again surprising. In the microbial community of non- sterilized larvae raised on a PA diet, we observed few correlations between the abundances of dominant bacteria taxa, which is similar to the microbial community of the larvae raised on the R diet. However, in the absence of parental microbiota, we observed more positive correlations

170 between the members of the microbial community. The observed results are consistent with the previous research that reported the dependency of some of the microbial taxa on each other during colonization of new environments (Matamoros et al. 2013, Li et al. 2004).

In addition, we observed that several dominant bacteria taxa were strongly associated with particular diets. As such, Lactobacillus exhibited highest abundances on the regular diet.

Previous studies identified Lactobacillus as one of the dominant bacterial taxa that is often associated with lab populations of Drosophila, which is consistent we our findings (Chandler et al. 2011, Douglas 2018, Chaston et al. 2016). In our work the members of Acetobactereceae were associated with peach-based diets. Previous research reported the presence of symbiotic relations between insects feeding on sugar-rich diets and members of Acetobactereceae family

(Leitão-Gonçalves et al. 2017, Crotti et al. 2010, Corby-Harris et al. 2007). Our peach diet provides a sugar-rich environment which may explain the observed results. Lastly, we saw a strong association between Leuconostoc and PR raised larvae. Leuconostoc is a lactic acid bacteria that was shown to be associated with symbiotic microbiota of natural Drosophila populations (Wright et al. 2017). In addition, it is commonly found in fermenting plant material such as kimchi or mango juice, with which our peach diet has obvious similarities (Yang et al.

2019, Sun et al. 2020).

High fat and high sugar dietary modifications did not change the difference between peach-raised larval and lab diet-raised larval bacteria. For difference in bacterial composition measured with Bray-Curtis distances and phylogenetic diversity measured with Weighed Unifrac distances, PHF and PHS diets were significantly different from RHF and RHS, diets respectively. PHFA and PHSA diets were still closer to lab diets than to peach diets. However, we cannot conclude that dietary modifications do not influence the gut bacterial composition. We

171 observed a significant difference between normal and modified diets in Bray-Curtis distances, especially for sterilized larvae. In terms of phylogenetic diversity, larvae raised on a R diet were significantly different only with RHS raised larvae while PR and PA raised larvae exhibited a significant difference with larvae raised on all of the tested nutritional modifications. Previous research in vertebrate models identified that high fat and high sugar dietary modifications may significantly alter gut bacterial composition, which is consistent with our results in Drosophila

(Do et al. 2018, von Frieling et al. 2020). Unfortunately, a very limited number of studies addressing the same question in Drosophila is available. Thus, further research will be required to confirm our findings.

Surprisingly, we observed that several of the dominant bacterial taxa were consistently associated with dietary modifications independent of the origin of the diet (peach or lab based).

For example, diets with extra lipids were associated with the increase abundances of

Corynebacteriaceae and Erysipelotrichaceae. Both of these families were reported to be members of the symbiotic microbiota in vertebrate hosts and were shown to be correlated with the consumption of high fat diets (Kaakoush 2015, Martínez et al. 2013). In our work, the consumption of high sugar diets was associated with the increased abundances of

Lactobacillaceae and Enterobacteriaceae. Previous studies also identified positive correlations between these microbial families and elevated sugars consumption, in various invertebrate and vertebrate models, as well as in humans (Satokari 2020, Volynets et al. 2017, Park et al. 2013,

Wang et al. 2020, Magnusson et al. 2015). These results may provide insights for evolutionary conserved associations between the host and its symbiotic microorganisms, in response to certain nutritional variations. In addition, our results show that these associations could be studied in a

Drosophila model.

172

We also observed that dietary modifications interacted with the effect that parental microbiota produced on the formation of a bacterial community. On all of the unmodified diets, presence of parental microbiota produced distinct microbial communities. The described results are consistent with Wong et.al (2015), who also observed a distinct effect of parental microbiota on the formation of the bacterial community. However, on the modified diets the pattern was only conserved for HS food, and primarily for a microbial community composition measured by

Bray-Curtis distances.

We recognized that some of the observed results may be subjected to the experimental bias. As we discussed above, the natural microbiota community may be very variable and can be altered with seasonality, temperature changes, geographic locations, and other environmental factors (Behar, Yuval, and Jurkevitch 2008, Ferguson et al. 2018, Moghadam et al. 2018,

Walters et al. 2020). Similarly, microbial community structure, in our experiments could also be impacted by the unique interaction of the environmental factors, during the acquisition of environmental microbiota. To address this issue, it would be favorable to repeat sampling of the naturally associated peach bacterial communities through different seasons and geographic locations. This will help to observe broader patterns in bacterial communities’ composition associated with a Drosophila’s natural diet.

In addition, all of the studies applying 16S rRNA sequencing techniques to identify the composition of microbial taxa suffer from several disadvantages of this method. First of all, the amplification of the sequences relies on a PCR reaction and the usage of degenerate primers.

Primers for nine hypervariable 16S rRNA regions V1-V9 are broadly used for phylogenetic analysis (Kumar et al. 2011). However, it was suggested that no single region can differentiate across all bacteria, which can lead to the bias and difficulties in the comparison of studies

173

(Chakravorty et al. 2007). In addition, it was shown that usage of different regions of 16S may result in different taxonomic diversity, as various regions are differentially conserved between the microbial taxa (Poretsky et al. 2014). 16S rRNA sequencing requires gene amplification, therefore results might be biased due to PCR artifacts (Poretsky et al. 2014). Incompletely amplified sequences might serve as primers during the PCR reaction and create chimera sequences (Jovel et al. 2016, Shah et al. 2011). 16S microbiome analysis might also have lower confidence in determining deep taxonomic levels such as genus due to the limited segment size of the region such as V4 and biases caused by a variation in microbial reads composition and identification across the databases (Chakravorty et al. 2007, Poretsky et al. 2014).

In order to address many of the described issues, it is possible to apply Whole Genome sequencing (WGS). WGS is an unrestricted sequencing technique that may provide the information about the metagenome of all microorganisms that are present in a sample and does not suffer from amplification biases. However, the application of WGS has some of its own challenges, such as the higher cost and the requirements for more complex bioinformatics analysis, which also involves the requirement for higher computational powers (Forbes et al.

2017). As WGS assembles the reads into larger contigs, similar genomic regions from different microbial taxa might be assigned to one contig (due to identical sequence), shifting the abundance of microbial species (Forbes et al. 2017). In addition, as metagenomic approach is relatively young comparing with 16S rRNA sequencing, it suffers from higher variability in

DNA preparation methods and differences between sequencing platforms, therefore is higher impacted by mechanistic biases (Forbes et al. 2017, Poretsky et al. 2014, Jovel et al. 2016).

As major Drosophila symbiotic taxa are aerobic and culturable, the described methods of assessing symbiotic microbiota community composition might be supplemented with culturing

174 an inoculate from the Drosophila digestive system (Leulier et al. 2017, Pais et al. 2018, Douglas

2018). Selective and differentiative media might be applied for this purpose, with further characterization of the isolated microbial taxa with morphological analysis and Sanger sequencing (Leulier et al. 2017, Douglas 2018, Morgan and Huttenhower 2012, Leboffe and

Pierce 2012). Unlike NextGen sequencing, this approach would allow one to identify the symbiotic microbiota taxa that were not previously described. However, due to laboriousness it might not be applicable for large experimental designs with many variables, such as mine.

Host’s genotype and genotype-by-environment interactive effect influence larvae phenotype, gut bacterial composition, and interactions between the host and its symbiotic bacteria

Although we were able to observe a common pattern in larval response to changing dietary conditions, the genetic factor and its interaction with other tested variables also produced a significant effect. As such, genotype played a significant role in formation of all larval phenotypes but glucose concentrations on the unmodified and high sugar diets, and on all of the tested phenotypes in larvae raised on high fat diets. Our results are consistent with the previous research that indicated the importance of accounting for genotype, in evaluation of the impact that diets produce on metabolic phenotypes (Reed et al. 2010, Reed et al. 2014, Jumbo-Lucioni et al. 2010, Dew-Budd, Jarnigan, and Reed 2016, James et al. 2020). We also consistently observed that interactions between diet, genotype, and treatment can significantly influence the development of larval phenotype and larval response to dietary modifications (Zhu, Ingelmo, and

Rand 2014, Reed et al. 2014, Reed et al. 2010, Wong, Dobson, and Douglas 2014, Henry et al.

2020).

Genotype influenced the abundance of more microbial taxa than any other tested variable. Similar results were reported by Jehrke et. al (2018) and combined with our previously

175 described observations may indicate that nutritional composition of the food cannot be considered the most influential factor in shaping symbiotic bacterial community. These findings may also imply that the influence of genotype on the bacterial composition must not be ignored during the development of treatments designed to modify gut bacterial composition. Obtaining healthy gut microbiota composition was shown to improve metabolic health and resist

Clostridium difficile infection (Borbet and Blaser 2019, Krautkramer et al. 2016, Org et al. 2015,

Zhang et al. 2009). Therefore, Drosophila model may provide tools that are necessary for the foundation of such studies.

Lastly, with our work, we demonstrated that the abundance of multiple bacterial taxa might be correlated with the development of larval phenotypes. Many of the identified correlations were also reported in previous works from Drosophila and vertebrate models. For example, in agreement with Chaston et. al (2014) and Dobson et. al (2016), we identified that the abundance of Acetobactereceae was negatively correlated with larval glucose concentrations.

Also, Coprococcus that was shown to be positively correlated with obesity in humans was positively correlated with glucose amounts in our larvae (Ignacio et al. 2016, Murugesan et al.

2015).

We observed similar results for the larvae raised on the diets modified with the addition of extra fats or sugars. As such, our results indicated that Erysipelotrichaceae, which as described above was associated with the HF nutritional modifications, was positively correlated with larvae triglyceride concentrations. Previous research demonstrated that Erysipelotrichaceae was associated with dyslipidemic syndrome and host’s lipids metabolism (Martínez et al. 2013).

Members of the Lactobacillaceae family that are commonly connected with HS diets, were shown to be correlated with the improvement of metabolic phenotype such as a decrease in

176 insulin resistance and fasting glucose concentrations. In our work, we observed negative correlations between abundance of Lactobacillaceae and triglyceride and glucose concentrations in larvae raised on HS diets (Azad et al. 2018, Khalili et al. 2019, Magnusson et al. 2015).

It is important to note that, we observed a significant variation in the effects that symbiotic bacteria produce on larvae phenotype among diets and genotypes. For example,

Bacteroides was positively correlated with larvae glucose concentrations on PHFA diet and negatively correlated with it on RHF diet, which represents a diet-by-abundance interactive effect. An example of diet-by-abundance-by-genotype interactive effect can be observed with

Lactobacillus that was positively correlated with triglyceride concentrations of the DGRP153 larvae, raised on RHF diet but was negatively correlated with the triglyceride concentrations of

DGRP 787 larvae raised on RHS food. Our results are consistent with previous research that indicated the importance of considering the interactive effects between diet, genotype, and abundance of microbial taxa (Zhu, Ingelmo, and Rand 2014, Jehrke et al. 2018). Which is especially important for studies that aim to identify the effects that symbiotic bacteria produce on the formation of hosts phenotype (Wong, Dobson, and Douglas 2014, Henry et al. 2020, Zhang et al. 2009, Org et al. 2015).

The last part of our research is also not free from the possibility to be improved by further experiments. One of the possible problems with the approach of correlating the abundance of bacterial taxa identified with 16S rRNA sequencing (besides of the problems with the sequencing approach described above) is that we sampled the results only in one point of time.

However, metabolism is a continuous process (Morowitz et al. 2000). In addition, gut microbiota communities are dynamic and often change with age (Odamaki et al. 2016, O’Toole and Jeffery

2015). Therefore, it is possible to miss important connections, especially if they play a major role

177 before the 3rd instar larvae stage. In addition, although we can determine which microbial taxa are associated with diets and phenotypes, our approach does not give the opportunity to evaluate the functionality of these interactions. It is especially important since previous studies indicated that lab adapted and naturally occurring microbiota taxa may possess different qualities (Pais et al. 2018, Henry et al. 2020). In order to develop understanding about the microbiome function and its communication with the host, it would be advantageous to sequence meta-transcriptomes

(Bashiardes, Zilberman-Schapira, and Elinav 2016). The application of transcriptomics helps us to understand the interactions in metagenome that influence the host, through the evaluation of which genetic pathways are up or down regulated (Bashiardes, Zilberman-Schapira, and Elinav

2016). In addition, the majority of microbiota induced signaling involved production of short chain fatty acids as well as other non-protein organic molecules and metabolites (Shin et al.

2011, Tilg and Moschen 2016). However, changes in transcription levels are not always directly correlated with changes in a metabolome of the organism (Hildreth et al. 2020). Therefore, an additional adoption of metabolomics approach may supply us with further information on the microbiota functions (Marcobal et al. 2013).

As described above, it is also possible to culture major taxa of Drosophila symbiotic microbiota. Which can help to evaluate their influence on the host via mono-associations and production of gnotobiotic larvae (Leulier et al. 2017, Leitão-Gonçalves et al. 2017, Consuegra et al. 2020). However, our own work and previous research indicated that the effect of some bacterial taxa on the host may be visible only in the presence of other microorganisms, which may be a restrictive factor due to complexity of interaction in the whole microbiota community

(Leitão-Gonçalves et al. 2017).

CONCLUSIONS

178

With this work, we demonstrated that peach food can be efficiently used in the controlled lab environment and provides nutritional conditions similar to the natural diet, as well as maintains key microbial taxa necessary for larvae survival and development. We identified that bacterial communities of the larvae raised on peach-based and lab-based diets are different and vary with the nutrition modifications. We observed that parental bacteria plays an important role in the development of the symbiotic bacterial community and provides a beneficial effect on larvae, especially on peach diets. We observed that genotype produces a noticeable effect on larvae metabolic phenotype and gut bacterial composition. Lastly, we identified that the interactions between genetic and environmental variables can produce a significant effect on larvae phenotype, gut bacterial composition and interactions between a host and its symbiotic microbiota.

FINAL NOTES

Previous research showed that symbiotic microbiota can help its host to overcome nutritional challenges and/or mitigate the effect of harmful nutrients (Ceja-Navarro et al. 2015, Shin et al.

2011, Kohl et al. 2014). These findings are consistent with the holobiont theory of evolution, which states that a host organism can acquire beneficial traits via symbiosis with microbiota

(Bordenstein and Theis 2015, Gordon et al. 2013). This work showed that symbiotic microbial community, acquired from environment and/or inherited maternally produces beneficial effect on

Drosophila larvae that has to develop on their natural diet. In addition, microbiota often can reduce the negative impact associated with harmful nutritional modification of the diet, which is especially evident on a high sugar food. Thus, results observed in this study provide further evidence for beneficial effect of gut microbiota on its host and for a holobiont theory.

179

REFERENCES

Adair, Karen L, Marita Wilson, Alyssa Bost, and Angela E Douglas. 2018. "Microbial community assembly in wild populations of the fruit fly Drosophila melanogaster." The ISME Journal:1.

Akter, R, A Nessa, MF Husain, F Wahed, N Khatun, M Yesmin, S Nasreen, and T %J Mymensingh medical journal: MMJ Tajkia. 2017. "Effect of Obesity on Fasting Blood Sugar." 26 (1):7-11.

Al-Goblan, Abdullah S, Mohammed A Al-Alfi, Muhammad Z %J Diabetes Khan, metabolic syndrome, obesity: targets, and therapy. 2014. "Mechanism linking diabetes mellitus and obesity." 7:587.

Allen-Vercoe, Emma. 2013. "Bringing the gut microbiota into focus through microbial culture: recent progress and future perspective." Current opinion in microbiology 16 (5):625-629.

Archie, Elizabeth A, and Kevin R Theis. 2011. "Animal behaviour meets microbial ecology." Animal Behaviour 82 (3):425-436.

Armougom, Fabrice, Mireille Henry, Bernard Vialettes, Denis Raccah, and Didier %J PloS one Raoult. 2009. "Monitoring bacterial community of human gut microbiota reveals an increase in Lactobacillus in obese patients and Methanogens in anorexic patients." 4 (9).

Ashburner, Michael. 1989. Drosophila. A laboratory handbook: Cold Spring Harbor Laboratory Press.

Azad, Md, Abul Kalam, Manobendro Sarker, Tiejun Li, and Jie %J BioMed research international Yin. 2018. "Probiotic species in the modulation of gut microbiota: an overview." 2018.

Baker, Keith D, and Carl S Thummel. 2007. "Diabetic larvae and obese flies—emerging studies of metabolism in Drosophila." Cell metabolism 6 (4):257-266.

180

Ballard, JWO, and Richard G %J Insect molecular biology Melvin. 2007. "Tetracycline treatment influences mitochondrial metabolism and mtDNA density two generations after treatment in Drosophila." 16 (6):799-802.

Bashiardes, Stavros, Gili Zilberman-Schapira, and Eran Elinav. 2016. "Use of metatranscriptomics in microbiome research." Bioinformatics and biology insights 10:BBI. S34610.

Behar, A, E Jurkevitch, and B Yuval. 2008. "Bringing back the fruit into fruit fly–bacteria interactions." Molecular Ecology 17 (5):1375-1386.

Behar, Adi, Boaz Yuval, and Edward Jurkevitch. 2008. "Community structure of the Mediterranean fruit fly microbiota: seasonal and spatial sources of variation." Israel Journal of Ecology and Evolution 54 (2):181-191.

Berger, Zdenek, Evangelia K Ttofi, Claire H Michel, Matthieu Y Pasco, Sean Tenant, David C Rubinsztein, and Cahir J %J Human molecular genetics O'Kane. 2005. "Lithium rescues toxicity of aggregate-prone proteins in Drosophila by perturbing Wnt pathway." 14 (20):3003-3011.

Bharucha, Kamal N. 2009. "The epicurean fly: using Drosophila melanogaster to study metabolism." Pediatric Research 65 (2):132-137.

Bing, XiaoLi, Joseph Gerlach, Gregory Loeb, and Nicolas Buchon. 2018. "Nutrient-dependent impact of microbes on Drosophila suzukii development." MBio 9 (2):e02199-17.

Bing, XiaoLi, Joseph Gerlach, Gregory Loeb, and Nicolas %J MBio Buchon. 2018. "Nutrient- dependent impact of microbes on Drosophila suzukii development." 9 (2):e02199-17.

Birse, Ryan T, Joan Choi, Kathryn Reardon, Jessica Rodriguez, Suzanne Graham, Soda Diop, Karen Ocorr, Rolf Bodmer, and Sean Oldham. 2010. "High-fat-diet-induced obesity and heart dysfunction are regulated by the TOR pathway in Drosophila." Cell metabolism 12 (5):533-544.

181

Bolger, Anthony M, Marc Lohse, and Bjoern %J Bioinformatics Usadel. 2014. "Trimmomatic: a flexible trimmer for Illumina sequence data." 30 (15):2114-2120.

Borbet, Timothy C, and Martin J %J Mucosal immunology Blaser. 2019. "Host genotype and early life microbiota alterations have additive effects on disease susceptibility." 12 (3):586-588.

Bordenstein, S. R., and K. R. Theis. 2015. "Host Biology in Light of the Microbiome: Ten Principles of Holobionts and Hologenomes." Plos Biology 13 (8). doi: ARTN e1002226

10.1371/journal.pbio.1002226.

Bost, Alyssa, Soeren Franzenburg, Karen L Adair, Vincent G Martinson, Greg Loeb, and Angela E Douglas. 2018. "How gut transcriptional function of Drosophila melanogaster varies with the presence and composition of the gut microbiota." Molecular ecology 27 (8):1848-1859.

Bradford, Marion M. 1976. "A rapid and sensitive method for the quantitation of microgram quantities of protein utilizing the principle of protein-dye binding." Analytical biochemistry 72 (1-2):248-254.

Bruce-Keller, Annadora J, J Michael Salbaum, Meng Luo, Eugene Blanchard IV, Christopher M Taylor, David A Welsh, and Hans-Rudolf %J Biological psychiatry Berthoud. 2015. "Obese-type gut microbiota induce neurobehavioral changes in the absence of obesity." 77 (7):607-615.

Brucker, Robert M, and Seth R Bordenstein. 2012. "Speciation by symbiosis." Trends in ecology & evolution 27 (8):443-451.

Bruno, Daniele, Marco Bonelli, Francesca De Filippis, Ilaria Di Lelio, Gianluca Tettamanti, Morena Casartelli, Danilo Ercolini, and Silvia %J Appl. Environ. Microbiol. Caccia. 2019. "The intestinal microbiota of Hermetia illucens larvae is affected by diet and shows a diverse composition in the different midgut regions." 85 (2):e01864-18.

182

Cani, Patrice D, Audrey M Neyrinck, F Fava, Claude Knauf, Rémy G Burcelin, Kieran M Tuohy, GR Gibson, and Nathalie M %J Diabetologia Delzenne. 2007. "Selective increases of bifidobacteria in gut microflora improve high-fat-diet-induced diabetes in mice through a mechanism associated with endotoxaemia." 50 (11):2374-2383.

Caporaso, J Gregory, Kyle Bittinger, Frederic D Bushman, Todd Z DeSantis, Gary L Andersen, and Rob %J Bioinformatics Knight. 2010. "PyNAST: a flexible tool for aligning sequences to a template alignment." 26 (2):266-267.

Caporaso, J Gregory, Justin Kuczynski, Jesse Stombaugh, Kyle Bittinger, Frederic D Bushman, Elizabeth K Costello, Noah Fierer, Antonio Gonzalez Pena, Julia K Goodrich, and Jeffrey I %J Nature methods Gordon. 2010. "QIIME allows analysis of high-throughput community sequencing data." 7 (5):335.

Carvalho, Gil B, Pankaj Kapahi, and Seymour Benzer. 2005. "Compensatory ingestion upon dietary restriction in Drosophila melanogaster." Nature methods 2 (11):813.

Ceja-Navarro, Javier A, Fernando E Vega, Ulas Karaoz, Zhao Hao, Stefan Jenkins, Hsiao Chien Lim, Petr Kosina, Francisco Infante, Trent R Northen, and Eoin L Brodie. 2015. "Gut microbiota mediate caffeine detoxification in the primary insect pest of coffee." Nature communications 6:7618.

Chakravorty, Soumitesh, Danica Helb, Michele Burday, Nancy Connell, and David Alland. 2007. "A detailed analysis of 16S ribosomal RNA gene segments for the diagnosis of pathogenic bacteria." Journal of microbiological methods 69 (2):330-339.

Chandler, James Angus, Jenna Morgan Lang, Srijak Bhatnagar, Jonathan A Eisen, and Artyom Kopp. 2011. "Bacterial communities of diverse Drosophila species: ecological context of a host–microbe model system." PLoS genetics 7 (9):e1002272.

Chaston, John M, Adam J Dobson, Peter D Newell, and Angela E Douglas. 2016. "Host genetic control of the microbiota mediates the Drosophila nutritional phenotype." Applied and environmental microbiology 82 (2):671-679.

183

Chaston, John M, Peter D Newell, and Angela E %J MBio Douglas. 2014. "Metagenome-wide association of microbial determinants of host phenotype in Drosophila melanogaster." 5 (5):e01631-14.

Clark, AG, and LE %J Genetics Keith. 1988. "Variation among extracted lines of Drosophila melanogaster in triacylglycerol and carbohydrate storage." 119 (3):595-607.

Consuegra, Jessika, Théodore Grenier, Patrice Baa-Puyoulet, Isabelle Rahioui, Houssam Akherraz, Hugo Gervais, Nicolas Parisot, Pedro da Silva, Hubert Charles, and Federica %J PLoS biology Calevro. 2020. "Drosophila-associated bacteria differentially shape the nutritional requirements of their host during juvenile growth." 18 (3):e3000681.

Corby-Harris, Vanessa, Ana Clara Pontaroli, Lawrence J Shimkets, Jeffrey L Bennetzen, Kristin E Habel, and Daniel EL Promislow. 2007. "Geographical distribution and diversity of bacteria associated with natural populations of Drosophila melanogaster." Applied and environmental microbiology 73 (11):3470-3479.

Cortines, Erika, and Ricardo Valcarcel. 2009. "Influence of pioneer-species combinations on restoration of disturbed ecosystems in the Atlantic Forest, Rio de Janeiro, Brazil." Revista Árvore 33 (5):927-936.

Crotti, Elena, Aurora Rizzi, Bessem Chouaia, Irene Ricci, Guido Favia, Alberto Alma, Luciano Sacchi, Kostas Bourtzis, Mauro Mandrioli, Ameur %J Applied Cherif, and Environmental Microbiology. 2010. "Acetic acid bacteria, newly emerging symbionts of insects." 76 (21):6963-6970.

D'Argenio, David A, Larry A Gallagher, Celeste A Berg, and Colin %J Journal of bacteriology Manoil. 2001. "Drosophila as a Model Host forPseudomonas aeruginosa Infection." 183 (4):1466-1471. da Veiga Moreira, Igor Magalhães, Maria Gabriela da Cruz Pedrozo Miguel, Whasley Ferreira Duarte, Disney Ribeiro Dias, and Rosane Freitas Schwan. 2013. "Microbial succession and the dynamics of metabolites and sugars during the fermentation of three different cocoa (Theobroma cacao L.) hybrids." Food research international 54 (1):9-17.

184

Daniel, Hannelore, Amin Moghaddas Gholami, David Berry, Charles Desmarchelier, Hannes Hahne, Gunnar Loh, Stanislas Mondot, Patricia Lepage, Michael Rothballer, and Alesia %J The ISME journal Walker. 2014. "High-fat diet alters gut microbiota physiology in mice." 8 (2):295-308.

Davison, James M, Colin R Lickwar, Lingyun Song, Ghislain Breton, Gregory E Crawford, and John F Rawls. 2017. "Microbiota regulate intestinal epithelial gene expression by suppressing the transcription factor Hepatocyte nuclear factor 4 alpha." Genome Research:gr. 220111.116.

De Luca, Maria, Nengjun Yi, David B Allison, Jeff Leips, and Douglas M %J Obesity research Ruden. 2005. "Mapping quantitative trait loci affecting variation in Drosophila triacylglycerol storage." 13 (9):1596-1605.

Deshpande, Sonali A, Gil B Carvalho, Ariadna Amador, Angela M Phillips, Sany Hoxha, Keith J Lizotte, and W Ja William. 2014. "Quantifying Drosophila food intake: comparative analysis of current methodology." Nature methods 11 (5):535.

Dew-Budd, Kelly, Julie Jarnigan, and Laura K Reed. 2016. "Genetic and sex-specific transgenerational effects of a high fat diet in Drosophila melanogaster." PloS one 11 (8):e0160857.

Do, Moon Ho, Eunjung Lee, Mi-Jin Oh, Yoonsook Kim, and Ho-Young %J Nutrients Park. 2018. "High-glucose or-fructose diet cause changes of the gut microbiota and metabolic disorders in mice without body weight change." 10 (6):761.

Dobson, Adam J, John M Chaston, Peter D Newell, Leanne Donahue, Sara L Hermann, David R Sannino, Stephanie Westmiller, Adam C-N Wong, Andrew G Clark, and Brian P Lazzaro. 2015. "Host genetic determinants of microbiota-dependent nutrition revealed by genome-wide analysis of Drosophila melanogaster." Nature communications 6:6312.

Douglas, Angela E. 2018. "The Drosophila model for microbiome research." Lab animal 47 (6):157-164.

185

Douglas, Angela E %J Lab animal. 2018. "The Drosophila model for microbiome research." 47 (6):157-164.

Early, Angela M, Niroshan Shanmugarajah, Nicolas Buchon, and Andrew G Clark. 2017. "Drosophila Genotype Influences Commensal Bacterial Levels." PloS one 12 (1):e0170332.

Edgar, Robert C %J Bioinformatics. 2010. "Search and clustering orders of magnitude faster than BLAST." 26 (19):2460-2461.

Erkosar, Berra, Gilles Storelli, Mélanie Mitchell, Loan Bozonnet, Noémie Bozonnet, and François Leulier. 2015. "Pathogen virulence impedes mutualist-mediated enhancement of host juvenile growth via inhibition of protein digestion." Cell host & microbe 18 (4):445- 455.

Erkosar, Berra, Erika Yashiro, Felix Zajitschek, Urban Friberg, Alexei A Maklakov, Jan R van der Meer, Tadeusz J %J Ecology Kawecki, and Evolution. 2018. "Host diet mediates a negative relationship between abundance and diversity of Drosophila gut microbiota." 8 (18):9491-9502.

Everard, Amandine, Clara Belzer, Lucie Geurts, Janneke P Ouwerkerk, Céline Druart, Laure B Bindels, Yves Guiot, Muriel Derrien, Giulio G Muccioli, and Nathalie M Delzenne. 2013. "Cross-talk between Akkermansia muciniphila and intestinal epithelium controls diet-induced obesity." Proceedings of the National Academy of Sciences 110 (22):9066- 9071.

Ferguson, Laura V, Pranav Dhakal, Jacqueline E Lebenzon, David E Heinrichs, Carol Bucking, and Brent J Sinclair. 2018. "Seasonal shifts in the insect gut microbiome are concurrent with changes in cold tolerance and immunity." Functional Ecology 32 (10):2357-2368.

Ferguson, Laura V, Pranav Dhakal, Jacqueline E Lebenzon, David E Heinrichs, Carol Bucking, and Brent J %J Functional Ecology Sinclair. 2018. "Seasonal shifts in the insect gut microbiome are concurrent with changes in cold tolerance and immunity." 32 (10):2357-2368.

186

Flint, Harry J, Sylvia H Duncan, and Petra Louis. 2017. "The impact of nutrition on intestinal bacterial communities." Current Opinion in Microbiology 38:59-65.

Forbes, Jessica D, Natalie C Knox, Jennifer Ronholm, Franco Pagotto, and Aleisha Reimer. 2017. "Metagenomics: the next culture-independent game changer." Frontiers in microbiology 8:1069.

Force, Allan, Michael Lynch, F Bryan Pickett, Angel Amores, Yi-lin Yan, and John %J Genetics Postlethwait. 1999. "Preservation of duplicate genes by complementary, degenerative mutations." 151 (4):1531-1545.

Furet, Jean-Pierre, Ling-Chun Kong, Julien Tap, Christine Poitou, Arnaud Basdevant, Jean-Luc Bouillot, Denis Mariat, Gérard Corthier, Joël Doré, and Corneliu Henegar. 2010. "Differential adaptation of human gut microbiota to bariatric surgery–induced weight loss." Diabetes 59 (12):3049-3057.

Galenza, Anthony, Jaclyn Hutchinson, Shelagh D Campbell, Bart Hazes, and Edan Foley. 2016. "Glucose modulates Drosophila longevity and immunity independent of the microbiota." Biology open 5 (2):165-173.

Gordon, J, N Knowlton, DA Relman, F Rohwer, and M Youle. 2013. Superorganisms and holobionts. Microbe 8: 152–153.

Griffin, Logan H, Laura K %J Ecology Reed, and Evolution. 2020. "Effect of gut microbiota on α‐amanitin tolerance in Drosophila tripunctata."

Guo, Xiulan, Jinchao Li, Renyong Tang, Guodong Zhang, Huawei Zeng, Richard J Wood, and Zhenhua %J Mediators of inflammation Liu. 2017. "High fat diet alters gut microbiota and the expression of paneth cell-antimicrobial peptides preceding changes of circulating inflammatory cytokines." 2017.

Haag, Lea-Maxie, André Fischer, Bettina Otto, Rita Plickert, Anja A Kühl, Ulf B Göbel, Stefan Bereswill, and Markus M Heimesaat. 2012. "Intestinal microbiota shifts towards elevated commensal Escherichia coli loads abrogate colonization resistance against Campylobacter jejuni in mice." PloS one 7 (5):e35988.

187

Hahne, Julia, Tabea Kloster, Sandra Rathmann, Mareike Weber, and André %J PloS one Lipski. 2018. "Isolation and characterization of Corynebacterium spp. from bulk tank raw cow's milk of different dairy farms in Germany." 13 (4):e0194365.

Han, Gangsik, Hyo Jung Lee, Sang Eun Jeong, Che Ok Jeon, and Seogang Hyun. 2017. "Comparative Analysis of Drosophila melanogaster Gut Microbiota with Respect to Host Strain, Sex, and Age." Microbial ecology 74 (1):207-216.

Han, Thang S, and Mike EJ Lean. 2016. "A clinical perspective of obesity, metabolic syndrome and cardiovascular disease." JRSM cardiovascular disease 5:2048004016633371.

Hasan, Nihal, and Hongyi %J PeerJ Yang. 2019. "Factors affecting the composition of the gut microbiota, and its modulation." 7:e7502.

Henry, Y, J Overgaard, H %J Comparative Biochemistry Colinet, Physiology Part A: Molecular, and Integrative Physiology. 2020. "Dietary nutrient balance shapes phenotypic traits of Drosophila melanogaster in interaction with gut microbiota." 241:110626.

Henry, Y, P Tarapacki, and H %J FEMS Microbiology Ecology Colinet. 2020. "Larval density affects phenotype and surrounding bacterial community without altering gut microbiota in Drosophila melanogaster." 96 (4):fiaa055.

Heys, Chloe, Anne Lizé, Frances Blow, Lewis White, Alistair Darby, Zenobia J %J Ecology Lewis, and evolution. 2018. "The effect of gut microbiota elimination in Drosophila melanogaster: a how‐to guide for host–microbiota studies." 8 (8):4150-4161.

Hildebrandt, Marie A, Christian Hoffmann, Scott A Sherrill–Mix, Sue A Keilbaugh, Micah Hamady, Ying–Yu Chen, Rob Knight, Rexford S Ahima, Frederic Bushman, and Gary D %J Gastroenterology Wu. 2009. "High-fat diet determines the composition of the murine gut microbiome independently of obesity." 137 (5):1716-1724. e2.

Hildreth, Sherry B, Evan E Foley, Gloria K Muday, Richard F Helm, and Brenda SJ %J Scientific Reports Winkel. 2020. "The dynamic response of the Arabidopsis root metabolome to auxin and ethylene is not predicted by changes in the transcriptome." 10 (1):1-15.

188

Hillier, Robyn M, and Richard LJ %J Journal of Dairy Research Lyster. 1979. "Whey protein denaturation in heated milk and cheese whey." 46 (1):95-102.

Hoffmann, Ary A. 1985. "Effects of experience on oviposition and attraction in Drosophila: comparing apples and oranges." The American Naturalist 126 (1):41-51.

Hoffmann, Ary A, Rebecca Hallas, Chantelle Sinclair, and Linda %J Evolution Partridge. 2001. "Rapid loss of stress resistance in Drosophila melanogaster under adaptation to laboratory culture." 55 (2):436-438.

Huang, Jia-Hsin, and Angela E Douglas. 2015. "Consumption of dietary sugar by gut bacteria determines Drosophila lipid content." Biology letters 11 (9):20150469.

Huang, Wen, Andreas Massouras, Yutaka Inoue, Jason Peiffer, Miquel Ràmia, Aaron M Tarone, Lavanya Turlapati, Thomas Zichner, Dianhui Zhu, and Richard F Lyman. 2014. "Natural variation in genome architecture among 205 Drosophila melanogaster Genetic Reference Panel lines." Genome research 24 (7):1193-1208.

Ignacio, A, MR Fernandes, VAA Rodrigues, FC Groppo, AL Cardoso, MJ Avila-Campos, V %J Clinical microbiology Nakano, and infection. 2016. "Correlation between body mass index and faecal microbiota from children." 22 (3):258. e1-258. e8.

Ignacio, Amigo, Miriam R Fernandes, Viviane AA Rodrigues, Francisco C Groppo, Ary Lopes Cardoso, MJ Avila-Campos, Viviane %J Clinical microbiology Nakano, and infection. 2016. "Correlation between body mass index and faecal microbiota from children." 22 (3):258. e1-258. e8.

Inglis, Timothy JJ. 2007. "Principia aetiologica: taking causality beyond Koch's postulates." Journal of medical microbiology 56 (11):1419-1422.

James, David E, Deanne Francis, Shila Ghazanfar, Essi Havula, James R Krycer, Alistair Senior, Annabel Y Minard, Thomas A Geddes, Fiona Weiss, and Jacqueline %J bioRxiv Stoeckli. 2020. "Genome-wide analysis in Drosophila reveals diet by gene interactions and uncovers diet-responsive genes."718304.

189

Jandhyala, Sai Manasa, Rupjyoti Talukdar, Chivkula Subramanyam, Harish Vuyyuru, Mitnala Sasikala, and D Nageshwar Reddy. 2015. "Role of the normal gut microbiota." World journal of gastroenterology: WJG 21 (29):8787.

Janos, David P. 1980. "Mycorrhizae influence tropical succession." Biotropica:56-64.

Jefferson, R. 1994. "The Hologenome." Agriculture, Environment and the Developing World: A Future of PCR. Cold Spring Harbor, New York.

Jehrke, Lisa, Fiona A Stewart, Andrea Droste, and Mathias Beller. 2018. "The impact of genome variation and diet on the metabolic phenotype and microbiome composition of Drosophila melanogaster." Scientific reports 8 (1):1-15.

Jehrke, Lisa, Fiona A Stewart, Andrea Droste, and Mathias %J Scientific reports Beller. 2018. "The impact of genome variation and diet on the metabolic phenotype and microbiome composition of Drosophila melanogaster." 8 (1):1-15.

Jena, Prasant Kumar, Bhumika Prajapati, Pradyumna Kumar Mishra, and Sriram %J Immunome Res Seshadri. 2016. "Influence of gut microbiota on inflammation and pathogenesis of sugar rich diet induced diabetes." 12 (1):109-119.

Jovel, Juan, Jordan Patterson, Weiwei Wang, Naomi Hotte, Sandra O'Keefe, Troy Mitchel, Troy Perry, Dina Kao, Andrew L Mason, and Karen L Madsen. 2016. "Characterization of the gut microbiome using 16S or shotgun metagenomics." Frontiers in microbiology 7:459.

Jumbo-Lucioni, Patricia, Julien F Ayroles, Michelle Moses Chambers, Katherine W Jordan, Jeff Leips, Trudy FC Mackay, and Maria De Luca. 2010. "Systems genetics analysis of body weight and energy metabolism traits in Drosophila melanogaster." BMC genomics 11 (1):297.

Kaakoush, Nadeem O 2015. "Insights into the role of Erysipelotrichaceae in the human host." Frontiers in cellular infection microbiology

190

5:84.

Kaur, Jaspinder. 2014. "A comprehensive review on metabolic syndrome." Cardiology research and practice 2014.

Khalili, Leila, Beitullah Alipour, Mohammad Asghari Jafar-Abadi, Ismail Faraji, Tohid Hassanalilou, Mehran Mesgari Abbasi, Elnaz Vaghef-Mehrabany, and Mahmood Alizadeh %J Iranian biomedical journal Sani. 2019. "The effects of lactobacillus casei on glycemic response, serum sirtuin1 and fetuin-a levels in patients with type 2 diabetes mellitus: a randomized controlled trial." 23 (1):68.

Kim, Eunjung, Dan-Bi Kim, Jae-Yong %J Preventive nutrition Park, and food science. 2016. "Changes of mouse gut microbiota diversity and composition by modulating dietary protein and carbohydrate contents: a pilot study." 21 (1):57.

Kimura, Ikuo, Kentaro Ozawa, Daisuke Inoue, Takeshi Imamura, Kumi Kimura, Takeshi Maeda, Kazuya Terasawa, Daiji Kashihara, Kanako Hirano, and Taeko Tani. 2013. "The gut microbiota suppresses insulin-mediated fat accumulation via the short-chain fatty acid receptor GPR43." Nature communications 4:1829.

Kimura, Masahito T. 1980. "Evolution of food preferences in fungus-feeding Drosophila: an ecological study." Evolution:1009-1018.

King, Elizabeth G, Stuart J Macdonald, and Anthony D Long. 2012. "Properties and power of the Drosophila Synthetic Population Resource for the routine dissection of complex traits." Genetics 191 (3):935-949.

Klepsatel, Peter, Emanuel Procházka, and Martina Gáliková. 2018. "Crowding of Drosophila larvae affects lifespan and other life-history traits via reduced availability of dietary yeast." Experimental gerontology 110:298-308.

Klepsatel, Peter, Emanuel Procházka, and Martina %J Experimental gerontology Gáliková. 2018. "Crowding of Drosophila larvae affects lifespan and other life-history traits via reduced availability of dietary yeast." 110:298-308.

191

Kohl, Kevin D, Robert B Weiss, James Cox, Colin Dale, and M %J Ecology letters Denise Dearing. 2014. "Gut microbes of mammalian herbivores facilitate intake of plant toxins." 17 (10):1238-1246.

Krautkramer, Kimberly A, Julia H Kreznar, Kymberleigh A Romano, Eugenio I Vivas, Gregory A Barrett-Wilt, Mary E Rabaglia, Mark P Keller, Alan D Attie, Federico E Rey, and John M Denu. 2016. "Diet-microbiota interactions mediate global epigenetic programming in multiple host tissues." Molecular cell 64 (5):982-992.

Kreznar, Julia H, Mark P Keller, Lindsay L Traeger, Mary E Rabaglia, Kathryn L Schueler, Donald S Stapleton, Wen Zhao, Eugenio I Vivas, Brian S Yandell, and Aimee Teo %J Cell reports Broman. 2017. "Host genotype and gut microbiome modulate insulin secretion and diet-induced metabolic phenotypes." 18 (7):1739-1750.

Kumar, Purnima S, Michael R Brooker, Scot E Dowd, and Terry Camerlengo. 2011. "Target region selection is a critical determinant of community fingerprints generated by 16S pyrosequencing." PloS one 6 (6):e20956.

Kumar, Ranjit, Peter Eipers, Rebecca B Little, Michael Crowley, David K Crossman, Elliot J Lefkowitz, and Casey D Morrow. 2014. "Getting started with microbiome analysis: sample acquisition to bioinformatics." Current protocols in human genetics:18.8. 1-18.8. 29.

Langille, Morgan GI, Jesse Zaneveld, J Gregory Caporaso, Daniel McDonald, Dan Knights, Joshua A Reyes, Jose C Clemente, Deron E Burkepile, Rebecca L Vega Thurber, and Rob Knight. 2013. "Predictive functional profiling of microbial communities using 16S rRNA marker gene sequences." Nature biotechnology 31 (9):814-821.

Lardy, Henry A, Jerald L Connelly, and Diane %J Biochemistry Johnson. 1964. "Antibiotics as tools for metabolic studies. II. Inhibition of phosphoryl transfer in mitochondria by oligomycin and aurovertin." 3 (12):1961-1968.

Leboffe, Michael J, and Burton E Pierce. 2012. Microbiology: laboratory theory and application: Morton Publishing Company.

192

Leitão-Gonçalves, Ricardo, Zita Carvalho-Santos, Ana Patrícia Francisco, Gabriela Tondolo Fioreze, Margarida Anjos, Célia Baltazar, Ana Paula Elias, Pavel M Itskov, Matthew DW Piper, and Carlos Ribeiro. 2017. "Commensal bacteria and essential amino acids control food choice behavior and reproduction." PLoS biology 15 (4):e2000862.

Lemaitre, Bruno, and Irene Miguel-Aliaga. 2013. "The digestive tract of Drosophila melanogaster." Annual review of genetics 47:377-404.

Leulier, François, Lesley T MacNeil, Won-jae Lee, John F Rawls, Patrice D Cani, Martin Schwarzer, Liping Zhao, and Stephen J Simpson. 2017. "Integrative physiology: at the crossroads of nutrition, microbiota, animal physiology, and human health." Cell Metabolism 25 (3):522-534.

Ley, Ruth E, Peter J Turnbaugh, Samuel Klein, and Jeffrey I %J nature Gordon. 2006. "Human gut microbes associated with obesity." 444 (7122):1022-1023.

Li, J, EJ Helmerhorst, CW Leone, RF Troxler, T Yaskell, AD Haffajee, SS Socransky, and FG Oppenheim. 2004. "Identification of early microbial colonizers in human dental biofilm." Journal of Applied Microbiology 97 (6):1311-1318.

Liu, Jian-Ping, Wen-Li Zou, Shui-Jiao Chen, Hong-Yun Wei, Ya-Ni Yin, Yi-You Zou, and Fang-Gen %J World journal of gastroenterology Lu. 2016. "Effects of different diets on intestinal microbiota and nonalcoholic fatty liver disease development." 22 (32):7353.

Ma, DALI, GILLES STORELLI, MÉLANIE MITCHELL, and FRANÇOIS LEULIER. 2017. "STUDYING HOST-MICROBIOTA MUTUALISM USING GNOTOBIOTIC DROSOPHILA." Old Herborn University Seminar Monograph 30.

Mackay, Trudy FC, Stephen Richards, Eric A Stone, Antonio Barbadilla, Julien F Ayroles, Dianhui Zhu, Sònia Casillas, Yi Han, Michael M Magwire, and Julie M Cridland. 2012. "The Drosophila melanogaster genetic reference panel." Nature 482 (7384):173-178.

193

Mackie, Roderick I, Abdelghani Sghir, and H Rex Gaskins. 1999. "Developmental microbial ecology of the neonatal gastrointestinal tract." The American journal of clinical nutrition 69 (5):1035s-1045s.

Magnusson, KR, L Hauck, BM Jeffrey, V Elias, A Humphrey, R Nath, A Perrone, and LE %J Neuroscience Bermudez. 2015. "Relationships between diet-related changes in the gut microbiome and cognitive flexibility." 300:128-140.

Mair, William, Matthew DW Piper, and Linda Partridge. 2005. "Calories do not explain extension of life span by dietary restriction in Drosophila." PLoS biology 3 (7):e223.

Marcobal, A, PC Kashyap, TA Nelson, PA Aronov, MS Donia, A Spormann, MA Fischbach, and JL Sonnenburg. 2013. "A metabolomic view of how the human gut microbiota impacts the host metabolome using humanized and gnotobiotic mice." The ISME journal 7 (10):1933.

Markow, Therese A, and Patrick O'Grady. 2005. Drosophila: a guide to species identification and use: Elsevier.

Martin, Tara L, Shannon R Balser, Gregory S Young, and Stephanie D %J Journal of the American Association for Laboratory Animal Science Lewis. 2016. "Cost and effectiveness of commercially available nesting substrates for Deer Mice (Peromyscus maniculatus)." 55 (4):412-418.

Martínez-Mota, Rodolfo, Kevin D Kohl, Teri J Orr, and M Denise %J The ISME Journal Dearing. 2020. "Natural diets promote retention of the native gut microbiota in captive rodents." 14 (1):67-78.

Martinez-Porchas, Marcel, Enrique Villalpando-Canchola, Luis Enrique Ortiz Suarez, and Francisco Vargas-Albores. 2017. "How conserved are the conserved 16S-rRNA regions?" PeerJ 5:e3036.

Martínez, Inés, Diahann J Perdicaro, Andrew W Brown, Susan Hammons, Trevor J Carden, Timothy P Carr, Kent M Eskridge, Jens %J Applied Walter, and environmental

194

microbiology. 2013. "Diet-induced alterations of host cholesterol metabolism are likely to affect the gut microbiota composition in hamsters." 79 (2):516-524.

Martyn, JA Jeevendra, Masao Kaneki, and Shingo %J Anesthesiology Yasuhara. 2008. "Obesity- induced insulin resistance and hyperglycemia: etiological factors and molecular mechanisms." 109 (1):137.

Matamoros, Sebastien, Christele Gras-Leguen, Françoise Le Vacon, Gilles Potel, and Marie- France de La Cochetiere. 2013. "Development of intestinal microbiota in infants and its impact on health." Trends in microbiology 21 (4):167-173.

Maturin, Larry, JT %J Bacteriological Analytical Manual US Food Peeler, and New Hampshire Avenue Silver Spring USA Drug Administration. 2001. "BAM aerobic plate count."

Maurice, Corinne F, Sarah CL Knowles, Joshua Ladau, Katherine S Pollard, Andy Fenton, Amy B Pedersen, and Peter J %J The ISME journal Turnbaugh. 2015. "Marked seasonal variation in the wild mouse gut microbiota." 9 (11):2423-2434.

Mendez, Sean, Louis Watanabe, Rachel Hill, Meredith Owens, Jason Moraczewski, Glenn C Rowe, Nicole C Riddle, and Laura K Reed. 2016. "The TreadWheel: a novel apparatus to measure genetic variation in response to gently induced exercise for Drosophila." PloS one 11 (10):e0164706.

Meyer, Folker, Daniel Paarmann, Mark D'Souza, Robert Olson, Elizabeth M Glass, Michael Kubal, Tobias Paczian, A Rodriguez, Rick Stevens, and Andreas Wilke. 2008. "The metagenomics RAST server–a public resource for the automatic phylogenetic and functional analysis of metagenomes." BMC bioinformatics 9 (1):386.

Million, M, M Maraninchi, M Henry, Fabrice Armougom, H Richet, P Carrieri, R Valero, D Raccah, B Vialettes, and Didier %J International journal of obesity Raoult. 2012. "Obesity-associated gut microbiota is enriched in Lactobacillus reuteri and depleted in Bifidobacterium animalis and Methanobrevibacter smithii." 36 (6):817-825.

195

Moghadam, Neda N, Pia Mai Thorshauge, Torsten N Kristensen, Nadieh de Jonge, Simon Bahrndorff, Henrik Kjeldal, and Jeppe Lund Nielsen. 2018. "Strong responses of Drosophila melanogaster microbiota to developmental temperature." Fly 12 (1):1-12.

Moghadam, Neda N, Pia Mai Thorshauge, Torsten N Kristensen, Nadieh de Jonge, Simon Bahrndorff, Henrik Kjeldal, and Jeppe Lund %J Fly Nielsen. 2018. "Strong responses of Drosophila melanogaster microbiota to developmental temperature." 12 (1):1-12.

Moghadam, Neda Nasiri, Martin Holmstrup, Tommaso Manenti, Marie Brandt Mouridsen, Cino Pertoldi, and Volker Loeschcke. 2015. "The role of storage lipids in the relation between fecundity, locomotor activity, and lifespan of drosophila melanogaster longevity-selected and control lines." PloS one 10 (6):e0130334.

Morais, Paula B, Marlucia B Martins, Louis B Klaczko, Leda C Mendonça-Hagler, and Allen N Hagler. 1995. "Yeast succession in the Amazon fruit Parahancornia amapa as resource partitioning among Drosophila spp." Applied and environmental microbiology 61 (12):4251-4257.

Morgan, Xochitl C, and Curtis Huttenhower. 2012. "Human microbiome analysis." PLoS computational biology 8 (12):e1002808.

Morowitz, Harold J, Jennifer D Kostelnik, Jeremy Yang, and George D %J Proceedings of the National Academy of Sciences Cody. 2000. "The origin of intermediary metabolism." 97 (14):7704-7708.

Murphy, EF, PD Cotter, S Healy, Tatiana M Marques, O O'sullivan, F Fouhy, SF Clarke, PW O'toole, Eamonn Martin Quigley, and C %J Gut Stanton. 2010. "Composition and energy harvesting capacity of the gut microbiota: relationship to diet, obesity and time in mouse models." 59 (12):1635-1642.

Murugesan, S, M Ulloa-Martínez, H Martínez-Rojano, FM Galván-Rodríguez, C Miranda-Brito, MC Romano, A Piña-Escobedo, ML Pizano-Zárate, C Hoyo-Vadillo, J %J European Journal of Clinical Microbiology García-Mena, and Infectious Diseases. 2015. "Study of the diversity and short-chain fatty acids production by the bacterial community in overweight and obese Mexican children." 34 (7):1337-1346.

196

Musselman, Laura Palanker, Jill L Fink, Kirk Narzinski, Prasanna Venkatesh Ramachandran, Sumitha Sukumar Hathiramani, Ross L Cagan, and Thomas J Baranski. 2011. "A high- sugar diet produces obesity and insulin resistance in wild-type Drosophila." Disease models & mechanisms 4 (6):842-849.

Neis, Evelien PJG, Cornelis HC Dejong, and Sander S %J Nutrients Rensen. 2015. "The role of microbial amino acid metabolism in host metabolism." 7 (4):2930-2946.

Newell, Peter D, and Angela E Douglas. 2014. "Interspecies interactions determine the impact of the gut microbiota on nutrient allocation in Drosophila melanogaster." Applied and environmental microbiology 80 (2):788-796.

Newell, Peter D, and Angela E %J Appl. Environ. Microbiol. Douglas. 2014. "Interspecies interactions determine the impact of the gut microbiota on nutrient allocation in Drosophila melanogaster." 80 (2):788-796.

Nunney, Leonard. 1990. "Drosophila on oranges: colonization, competition, and coexistence." Ecology 71 (5):1904-1915.

O’Toole, Paul W, and Ian B %J Science Jeffery. 2015. "Gut microbiota and aging." 350 (6265):1214-1215.

Odamaki, Toshitaka, Kumiko Kato, Hirosuke Sugahara, Nanami Hashikura, Sachiko Takahashi, Jin-zhong Xiao, Fumiaki Abe, and Ro %J BMC microbiology Osawa. 2016. "Age-related changes in gut microbiota composition from newborn to centenarian: a cross-sectional study." 16 (1):1-12.

Oksanen, J, R Kindt, P Legendre, B O’hara, GL Simpson, P Solymos, MHH Stevens, and H Wagner. 2009. vegan: community ecology package. R package version 1.15-4. R Foundation for Statistical Computing, Vienna, Austria.

Org, Elin, Brian W Parks, Jong Wha J Joo, Benjamin Emert, William Schwartzman, Eun Yong Kang, Margarete Mehrabian, Calvin Pan, Rob Knight, and Robert %J Genome research Gunsalus. 2015. "Genetic and environmental control of host-gut microbiota interactions." 25 (10):1558-1569.

197

Pais, Inês S, Rita S Valente, Marta Sporniak, and Luis Teixeira. 2018. "Drosophila melanogaster establishes a species-specific mutualistic interaction with stable gut-colonizing bacteria." PLoS biology 16 (7):e2005710.

Pais, Inês S, Rita S Valente, Marta Sporniak, and Luis %J PLoS biology Teixeira. 2018. "Drosophila melanogaster establishes a species-specific mutualistic interaction with stable gut-colonizing bacteria." 16 (7):e2005710.

Park, Do-Young, Young-Tae Ahn, Chul-Sung Huh, Robin A McGregor, and Myung-Sook %J World Journal of Gastroenterology: WJG Choi. 2013. "Dual probiotic strains suppress high fructose-induced metabolic syndrome." 19 (2):274.

Parks, Brian W, Elizabeth Nam, Elin Org, Emrah Kostem, Frode Norheim, Simon T Hui, Calvin Pan, Mete Civelek, Christoph D Rau, and Brian J %J Cell metabolism Bennett. 2013. "Genetic control of obesity and gut microbiota composition in response to high-fat, high- sucrose diet in mice." 17 (1):141-152.

Patterson, Elaine, Paul M Ryan, John F Cryan, Timothy G Dinan, R Paul Ross, Gerald F Fitzgerald, and Catherine Stanton. 2016. "Gut microbiota, obesity and diabetes." Postgraduate medical journal:postgradmedj-2015-133285.

Paulson, Joseph N, O Colin Stine, Héctor Corrada Bravo, and Mihai Pop. 2013. "Differential abundance analysis for microbial marker-gene surveys." Nature methods 10 (12):1200.

Pawitan, Yudi, Stefan Michiels, Serge Koscielny, Arief Gusnanto, and Alexander %J Bioinformatics Ploner. 2005. "False discovery rate, sensitivity and sample size for microarray studies." 21 (13):3017-3024.

Pilátová, Martina, and Marc S %J PLoS One Dionne. 2012. "Burkholderia thailandensis is virulent in Drosophila melanogaster." 7 (11):e49745.

Piper, Matthew DW, Eric Blanc, Ricardo Leitão-Gonçalves, Mingyao Yang, Xiaoli He, Nancy J Linford, Matthew P Hoddinott, Corinna Hopfen, George A Soultoukis, and Christine %J Nature methods Niemeyer. 2014. "A holidic medium for Drosophila melanogaster." 11 (1):100-105.

198

Poretsky, Rachel, Luis M Rodriguez-R, Chengwei Luo, Despina Tsementzi, and Konstantinos T Konstantinidis. 2014. "Strengths and limitations of 16S rRNA gene amplicon sequencing in revealing temporal microbial community dynamics." PLoS One 9 (4):e93827.

Price, Morgan N, Paramvir S Dehal, Adam P %J Molecular biology Arkin, and evolution. 2009. "FastTree: computing large minimum evolution trees with profiles instead of a distance matrix." 26 (7):1641-1650.

Puhl, Rebecca M, and Chelsea A Heuer. 2010. "Obesity stigma: important considerations for public health." American journal of public health 100 (6):1019-1028.

Quast, Christian, Elmar Pruesse, Pelin Yilmaz, Jan Gerken, Timmy Schweer, Pablo Yarza, Jörg Peplies, and Frank Oliver %J Nucleic acids research Glöckner. 2012. "The SILVA ribosomal RNA gene database project: improved data processing and web-based tools." 41 (D1):D590-D596.

Read, Mark N, and Andrew J Holmes. 2017. "towards an integrative Understanding of Diet– Host–Gut Microbiome interactions." Frontiers in Immunology 8.

Reed, Laura K, Kevin Lee, Zhi Zhang, Lubna Rashid, Amy Poe, Benjamin Hsieh, Nigel Deighton, Norm Glassbrook, Rolf Bodmer, and Greg Gibson. 2014. "Systems genomics of metabolic phenotypes in wild-type Drosophila melanogaster." Genetics 197 (2):781- 793.

Reed, Laura K, Stephanie Williams, Mastafa Springston, Julie Brown, Kenda Freeman, Christie E DesRoches, Marla B Sokolowski, and Greg Gibson. 2010. "Genotype-by-diet interactions drive metabolic phenotype variation in Drosophila melanogaster." Genetics 185 (3):1009-1019.

Ridaura, Vanessa K, Jeremiah J Faith, Federico E Rey, Jiye Cheng, Alexis E Duncan, Andrew L Kau, Nicholas W Griffin, Vincent Lombard, Bernard Henrissat, and James R Bain. 2013. "Gut microbiota from twins discordant for obesity modulate metabolism in mice." Science 341 (6150):1241214.

199

Ridley, Emma V, Adam CN Wong, Stephanie Westmiller, and Angela E Douglas. 2012. "Impact of the resident microbiota on the nutritional phenotype of Drosophila melanogaster." PloS one 7 (5):e36765.

Robinson, David %J arXiv preprint arXiv:. 2014. "broom: An R package for converting statistical analysis objects into tidy data frames."

Rosenberg, Eugene, Gil Sharon, Ilil Atad, and Ilana Zilber‐Rosenberg. 2010. "The evolution of animals and plants via symbiosis with microorganisms." Environmental microbiology reports 2 (4):500-506.

Rosenberg, Eugene, Gil Sharon, and Ilana Zilber‐Rosenberg. 2009. "The hologenome theory of evolution contains Lamarckian aspects within a Darwinian framework." Environmental microbiology 11 (12):2959-2962.

Rosenthal, Nadia, and Steve %J Nature cell biology Brown. 2007. "The mouse ascending: perspectives for human-disease models." 9 (9):993-999.

Rulifson, Eric J, Seung K Kim, and Roel %J Science Nusse. 2002. "Ablation of insulin- producing neurons in flies: growth and diabetic phenotypes." 296 (5570):1118-1120.

Russell, Trinity, Richard %J Staff Kurtz, and Reviewers-July. 2012. "A Comparison of Laboratory-Reared Stock and Captured Fruit Flies (Drosophila melanogaster) using Upward Movement, Phototaxic, and Starvation Assays Reveals Significant Behavioral Differences."6.

Ryu, Ji-Hwan, Sung-Hee Kim, Hyo-Young Lee, Jin Young Bai, Young-Do Nam, Jin-Woo Bae, Dong Gun Lee, Seung Chul Shin, Eun-Mi Ha, and Won-Jae Lee. 2008. "Innate immune homeostasis by the homeobox gene caudal and commensal-gut mutualism in Drosophila." science 319 (5864):777-782.

Sang, James H. 1956. "The quantitative nutritional requirements of Drosophila melanogaster." Journal of Experimental Biology 33 (1):45-72.

200

Sang, James H %J Journal of Experimental Biology. 1956. "The quantitative nutritional requirements of Drosophila melanogaster." 33 (1):45-72.

Satokari, Reetta. 2020. High Intake of Sugar and the Balance between Pro-and Anti- Inflammatory Gut Bacteria. Multidisciplinary Digital Publishing Institute.

Schwiertz, Andreas, David Taras, Klaus Schäfer, Silvia Beijer, Nicolaas A Bos, Christiane Donus, and Philip D %J Obesity Hardt. 2010. "Microbiota and SCFA in lean and overweight healthy subjects." 18 (1):190-195.

Seganfredo, FB, CA Blume, M Moehlecke, A Giongo, DS Casagrande, JVN Spolidoro, AV Padoin, BD Schaan, and CC Mottin. 2017. "Weight‐loss interventions and gut microbiota changes in overweight and obese patients: a systematic review." Obesity Reviews.

Sgro, Carla M, and Linda %J The American Naturalist Partridge. 2000. "Evolutionary responses of the life history of wild-caught Drosophila melanogaster to two standard methods of laboratory culture." 156 (4):341-353.

Shah, Neethu, Haixu Tang, Thomas G Doak, and Yuzhen Ye. 2011. "Comparing bacterial communities inferred from 16S rRNA gene sequencing and shotgun metagenomics." In Biocomputing 2011, 165-176. World Scientific.

Shanbhag, Shubha R, Abraham T Vazhappilly, Abhay Sane, Natalie M D'silva, and Subrata Tripathi. 2017. "Electrolyte transport pathways induced in the midgut epithelium of Drosophila melanogaster larvae by commensal gut microbiota and pathogens." The Journal of physiology 595 (2):523-539.

Shin, Seung Chul, Sung-Hee Kim, Hyejin You, Boram Kim, Aeri C Kim, Kyung-Ah Lee, Joo- Heon Yoon, Ji-Hwan Ryu, and Won-Jae Lee. 2011. "Drosophila microbiome modulates host developmental and metabolic homeostasis via insulin signaling." Science 334 (6056):670-674.

Signorell, A, K Aho, A Alfons, N Anderegg, T Aragon, A Arppe, A Baddeley, K Barton, B Bolker, and HW Borchers. 2020. DescTools: tools for descriptive statistics. R package version 0.99. 34.

201

Skorupa, Danielle A, Azra Dervisefendic, Jessica Zwiener, and Scott D Pletcher. 2008. "Dietary composition specifies consumption, obesity, and lifespan in Drosophila melanogaster." Aging cell 7 (4):478-490.

Staubach, Fabian, John F Baines, Sven Künzel, Elisabeth M Bik, and Dmitri A Petrov. 2013. "Host species and environmental effects on bacterial communities associated with Drosophila in the laboratory and in the natural environment." PloS one 8 (8).

Staubach, Fabian, John F Baines, Sven Künzel, Elisabeth M Bik, and Dmitri A %J PloS one Petrov. 2013. "Host species and environmental effects on bacterial communities associated with Drosophila in the laboratory and in the natural environment." 8 (8).

Stilling, Roman M, Seth R Bordenstein, Timothy G Dinan, and John F Cryan. 2014. "Friends with social benefits: host-microbe interactions as a driver of brain evolution and development?" Frontiers in cellular and infection microbiology 4.

Storelli, Gilles, Arnaud Defaye, Berra Erkosar, Pascal Hols, Julien Royet, and François %J Cell metabolism Leulier. 2011. "Lactobacillus plantarum promotes Drosophila systemic growth by modulating hormonal signals through TOR-dependent nutrient sensing." 14 (3):403-414.

Strickberger, Monroe W. 1962. "Experiments in genetics with Drosophila." Experiments in genetics with Drosophila.

Strigini, Maura, and François Leulier. 2016. "The role of the microbial environment in Drosophila post-embryonic development." Developmental & Comparative Immunology 64:39-52.

Sturtevant, AH. 1916. "Notes on North American Drosophilidae with Descriptions of Twenty- Three New Species." Annals of the Entomological Society of America 9 (4):323-343.

Sun, Mengzhen, Qiya Wang, Maomao Zhang, Guohua Zhang, Tao Wu, Rui Liu, Wenjie Sui, Jiaojiao Zhang, Jinjin Yin, Min %J Food Zhang, and Function. 2020. "Leuconostoc pseudomesenteroides improves microbiota dysbiosis and liver metabolism imbalance and

202

ameliorates the correlation between dihydroceramide and strains of Firmicutes and Proteobacteria in high fat diet obese mice." 11 (8):6855-6865.

Team, R Core, and R %J URL https://www. R-project. org DC. 2019. "A language and environment for statistical computing. Vienna, Austria: R Foundation for Statistical Computing; 2012."

Tefit, Melisandre, Benjamin Gillet, Pauline Joncour, Sandrine Hughes, and Francois Leulier. 2017. "Stable association of a Drosophila-derived microbiota with its animal partner and the nutritional environment upon transfer between populations and generations." bioRxiv:111492.

Thompson, Warren G, David A Cook, Matthew M Clark, Aditya Bardia, and James A Levine. 2007. "Treatment of obesity." Mayo Clinic Proceedings.

Tilg, Herbert, and Alexander R Moschen. 2016. "Gut Microbiome, Obesity, and Metabolic Syndrome." Metabolic Syndrome: A Comprehensive Textbook:447-459.

Tong, Qing, and Jian-tao %J Frontiers in microbiology Zhang. 2019. "Effects of captivity and season on the gut microbiota of the brown frog (Rana dybowskii)." 10:1912.

Turnbaugh, Peter J, Ruth E Ley, Michael A Mahowald, Vincent Magrini, Elaine R Mardis, and Jeffrey I Gordon. 2006. "An obesity-associated gut microbiome with increased capacity for energy harvest." nature 444 (7122):1027-131.

Ubeda, Carles, Vanni Bucci, Silvia Caballero, Ana Djukovic, Nora C Toussaint, Michele Equinda, Lauren Lipuma, Lilan Ling, Asia Gobourne, and Daniel No. 2013. "Intestinal microbiota containing Barnesiella species cures vancomycin-resistant Enterococcus faecium colonization." Infection and immunity 81 (3):965-973.

Urbaniec, K, P Zalewski, and XX %J Applied thermal engineering Zhu. 2000. "A decomposition approach for retrofit design of energy systems in the sugar industry." 20 (15-16):1431- 1442.

203

Vacca, Mirco, Giuseppe Celano, Francesco Maria Calabrese, Piero Portincasa, Marco Gobbetti, and Maria %J Microorganisms De Angelis. 2020. "The Controversial Role of Human Gut Lachnospiraceae." 8 (4):573.

Vacchini, Violetta, Elena Gonella, Elena Crotti, Erica M Prosdocimi, Fabio Mazzetto, Bessem Chouaia, Matteo Callegari, Francesca Mapelli, Mauro Mandrioli, and Alberto Alma. 2017. "Bacterial diversity shift determined by different diets in the gut of the spotted wing fly Drosophila suzukii is primarily reflected on acetic acid bacteria." Environmental microbiology reports 9 (2):91-103.

Vandehoef, Crissie, Maral Molaei, and Jason %J Cell Reports Karpac. 2020. "Dietary Adaptation of Microbiota in Drosophila Requires NF-κB-Dependent Control of the Translational Regulator 4E-BP." 31 (10):107736.

Venables, WN, and BD Ripley. 2002. "Random and mixed effects." In Modern applied statistics with S, 271-300. Springer.

Ver Hoef, Jay M, and Peter L %J Ecology Boveng. 2007. "Quasi‐Poisson vs. negative binomial regression: how should we model overdispersed count data?" 88 (11):2766-2772.

Vodovar, Nicolas, Marisa Vinals, Peter Liehl, Alan Basset, Jeril Degrouard, Paul Spellman, Frédéric Boccard, and Bruno Lemaitre. 2005. "Drosophila host defense after oral infection by an entomopathogenic Pseudomonas species." Proceedings of the National Academy of Sciences of the United States of America 102 (32):11414-11419.

Volynets, Valentina, Sandrine Louis, Dominik Pretz, Lisa Lang, Maureen J Ostaff, Jan Wehkamp, and Stephan C %J The Journal of nutrition Bischoff. 2017. "Intestinal barrier function and the gut microbiome are differentially affected in mice fed a western-style diet or drinking water supplemented with fructose." 147 (5):770-780. von Frieling, Jakob, Muhammed Naeem Faisal, Femke Sporn, Roxana Pfefferkorn, Stella Solveig Nolte, Felix Sommer, Philip Rosenstiel, and Thomas %J PLoS Genetics Roeder. 2020. "A high-fat diet induces a microbiota-dependent increase in stem cell activity in the Drosophila intestine." 16 (5):e1008789.

204

Wahba, Ihab M, and Robert H Mak. 2007. "Obesity and obesity-initiated metabolic syndrome: mechanistic links to chronic kidney disease." Clinical Journal of the American Society of Nephrology 2 (3):550-562.

Walters, Amber W, Rachel C Hughes, Tanner B Call, Carson J Walker, Hailey Wilcox, Samara C Petersen, Seth M Rudman, Peter D Newell, Angela E Douglas, and Paul S %J Molecular Ecology Schmidt. 2020. "The microbiota influences the Drosophila melanogaster life history strategy." 29 (3):639-653.

Wang, Hongfang, Chunlei Liu, Zhenguo Liu, Ying Wang, Lanting Ma, and Baohua %J BMC microbiology Xu. 2020. "The different dietary sugars modulate the composition of the gut microbiota in honeybee during overwintering." 20 (1):1-14.

Watanabe, Louis Patrick, and Nicole C Riddle. 2017. "Characterization of the Rotating Exercise Quantification System (REQS), a novel Drosophila exercise quantification apparatus." PloS one 12 (10):e0185090.

Wickham, Hadley. 2016. ggplot2: elegant graphics for data analysis: springer.

Williams, Stephanie, Kelly Dew-Budd, Kristen Davis, Julie Anderson, Ruth Bishop, Kenda Freeman, Dana Davis, Katherine Bray, Lauren Perkins, and Joana Hubickey. 2015. "Metabolomic and gene expression profiles exhibit modular genetic and dietary structure linking metabolic syndrome phenotypes in Drosophila." G3: Genes, Genomes, Genetics 5 (12):2817-2829.

Winans, Nathan J, Alec Walter, Bessem Chouaia, John M Chaston, Angela E Douglas, and Peter D %J Molecular ecology Newell. 2017. "A genomic investigation of ecological differentiation between free‐living and Drosophila‐associated bacteria." 26 (17):4536- 4550.

Wong, Adam C-N, Adam J Dobson, and Angela E Douglas. 2014. "Gut microbiota dictates the metabolic response of Drosophila to diet." Journal of Experimental Biology 217 (11):1894-1901.

205

Wong, Adam C-N, Yuan Luo, Xiangfeng Jing, Soeren Franzenburg, Alyssa Bost, and Angela E Douglas. 2015. "The host as the driver of the microbiota in the gut and external environment of Drosophila melanogaster." Appl. Environ. Microbiol. 81 (18):6232-6240.

Wong, Adam C-N, Yuan Luo, Xiangfeng Jing, Soeren Franzenburg, Alyssa Bost, and Angela E %J Appl. Environ. Microbiol. Douglas. 2015. "The host as the driver of the microbiota in the gut and external environment of Drosophila melanogaster." 81 (18):6232-6240.

Wong, Adam Chun-Nin, Qiao-Ping Wang, Juliano Morimoto, Alistair M Senior, Mathieu Lihoreau, G Gregory Neely, Stephen J Simpson, and Fleur %J Current biology Ponton. 2017. "Gut microbiota modifies olfactory-guided microbial preferences and foraging decisions in Drosophila." 27 (15):2397-2404. e4.

Wong, Adam CN, John M Chaston, and Angela E Douglas. 2013. "The inconstant gut microbiota of Drosophila species revealed by 16S rRNA gene analysis." The ISME journal 7 (10):1922.

Wong, Chun Nin Adam, Patrick Ng, and Angela E Douglas. 2011. "Low‐diversity bacterial community in the gut of the fruitfly Drosophila melanogaster." Environmental microbiology 13 (7):1889-1900.

Wright, Sage M, Courtney Carroll, Amber Walters, Peter D Newell, and John M %J Genome Announcements Chaston. 2017. "Genome sequence of Leuconostoc citreum DmW_111, isolated from wild Drosophila." 5 (24).

Wu, Gary D, Jun Chen, Christian Hoffmann, Kyle Bittinger, Ying-Yu Chen, Sue A Keilbaugh, Meenakshi Bewtra, Dan Knights, William A Walters, and Rob %J Science Knight. 2011. "Linking long-term dietary patterns with gut microbial enterotypes." 334 (6052):105- 108.

Xiao, Shuiming, Na Fei, Xiaoyan Pang, Jian Shen, Linghua Wang, Baorang Zhang, Menghui Zhang, Xiaojun Zhang, Chenhong Zhang, and Min %J FEMS microbiology ecology Li. 2014. "A gut microbiota-targeted dietary intervention for amelioration of chronic inflammation underlying metabolic syndrome." 87 (2):357-367.

206

Yang, Jinho, Andrea McDowell, Eun Kyoung Kim, Hochan Seo, Kyujin Yum, Won Hee Lee, Young-Koo Jee, Yoon-Keun %J Experimental Kim, and molecular medicine. 2019. "Consumption of a Leuconostoc holzapfelii-enriched synbiotic beverage alters the composition of the microbiota and microbial extracellular vesicles." 51 (8):1-11.

Yin, Xiaochen, Bokyung Lee, Jose Zaragoza, and Maria L %J Scientific reports Marco. 2017. "Dietary perturbations alter the ecological significance of ingested Lactobacillus plantarum in the digestive tract." 7 (1):1-13.

Zhang, Chenhong, Menghui Zhang, Shengyue Wang, Ruijun Han, Youfang Cao, Weiying Hua, Yuejian Mao, Xiaojun Zhang, Xiaoyan Pang, and Chaochun %J The ISME journal Wei. 2010. "Interactions between gut microbiota, host genetics and diet relevant to development of metabolic syndromes in mice." 4 (2):232-241.

Zhang, Husen, John K DiBaise, Andrea Zuccolo, Dave Kudrna, Michele Braidotti, Yeisoo Yu, Prathap Parameswaran, Michael D Crowell, Rod Wing, and Bruce E Rittmann. 2009. "Human gut microbiota in obesity and after gastric bypass." Proceedings of the National Academy of Sciences 106 (7):2365-2370.

Zhu, Chen-Tseh, Paul Ingelmo, and David M %J PLoS Genet Rand. 2014. "G× G× E for lifespan in Drosophila: mitochondrial, nuclear, and dietary interactions that modify longevity." 10 (5):e1004354.

Zilber‐Rosenberg, Ilana, and Eugene Rosenberg. 2008. "Role of microorganisms in the evolution of animals and plants: the hologenome theory of evolution." FEMS microbiology reviews 32 (5):723-735.

Zubiría, María Guillermina, Sabrina Eliana Gambaro, María Amanda Rey, Paula Carasi, María de los Ángeles Serradell, and Andrés %J Nutrients Giovambattista. 2017. "Deleterious metabolic effects of high fructose intake: the preventive effect of Lactobacillus kefiri administration." 9 (5):470.

207